Strategische Managementtheorie 9783110803402, 9783110157871

Citation preview

Bresser · Strategische Managementtheorie

Rudi Κ. F. Bresser

Strategische Managementtheorie

W DE G Walter de Gruyter · Berlin · New York 1998

Prof. Dr. Rudi Κ. F. Bresser Freie Universität Berlin, Fachbereich Wirtschaftswissenschaft, Institut für Management, Berlin

Das Buch enthält zahlreiche Abbildungen und Tabellen

® Gedruckt auf säurefreiem Papier, das die US-ANSI-Norm über Haltbarkeit erfüllt.

Die Deutsche Bibliothek —

CIP-Einheitsaufnahme

Bresser, Rudi K. F.: Strategische Managementtheorie / Rudi K. F. Bresser. — Berlin ; New York : de Gruyter, 1998 ISBN 3-11-015786-1 brosch. ISBN 3-11-015787-X Gb.

© Copyright 1998 by Walter de Gruyter G m b H & Co., D-10785 Berlin Dieses Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere f ü r Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Printed in Germany. Datenkonvertierung: Knipp Medien und Kommunikation O H G , Dortmund. - Druck: Gerike G m b H , Berlin. — Buchbinderische Verarbeitung: Lüderitz & Bauer G m b H , Berlin. Einbandentwurf: Johannes Rother, Berlin.

Danksagung

Sowohl der Autor als auch der Verlag danken den folgenden Verlagen und/oder Wissenschaftlern und Wissenschaftlerinnen für die freundliche Erteilung von Wiederabdruckgenehmigungen der folgenden Originalbeiträge: Der Academy of Management für die Artikel: Deephouse, D. L. „Does Isomorphism Legitimate?", Academy of Management Journal, Jg. 39, 1996, S. 1024-1039; Hambrick, D. C. und P. A. Mason. „Upper Echelons: The Organization as a Reflection of Its Top Managers", Academy of Management Review, Jg. 9, 1984, S. 193-206; Hill, C. W. L. „Differentiation Versus Low Cost or Differentiation and Low Cost: A Contingency Framework", Academy of Management Review, Jg. 13, 1988, S. 401-412; Hill, C. W. L. „Cooperation, Opportunism, and the Invisible Hand: Implications for Transaction Cost Theory", Academy of Management Review, Jg. 15, 1990, S. 500-513; Jacobson, R. „The ,Austrian' School of Strategy", Academy of Management Review, Jg. 17, 1992, S. 782807; Kelly, D. und T. L. Amburgey. „Organizational Inertia and Momentum: A Dynamic Model of Strategic Change", Academy of Management Journal, Jg. 34, 1991, S. 591612; Thomas, J. B., Clark, S. M. und D. A. Gioia. „Strategic Sensemaking and Organizational Performance: Linkages Among Scanning, Interpretation, Action, and Outcomes", Academy of Management Journal, Jg. 36, 1993, S. 239-270; Zajac, E. J. und M. H. Bazerman. ,31ind Spots in Industry and Competitor Analysis: Implications of Interfirm (Mis)perceptions for Strategic Decisions", Academy of Management Review, Jg. 16, 1991, S. 37-56. Der Zeitschrift Administrative Science Quarterly für die Artikel: Finkelstein, S. und D. C. Hambrick. „Top-Management-Team Tenure and Organizational Outcomes: The Moderating Role of Managerial Discretion", Administrative Science Quarterly, Jg. 35, 1990, S. 484-503; Haunschild, P. R. ,Jnterorganizational Imitation: The Impact of Interlocks on Corporate Acquisition Activity", Administrative Science Quarterly, Jg. 38, 1993, S. 564-592; Howell, J. M. und C. A. Higgins. „Champions of Technological Innovation", Administrative Science Quarterly, Jg. 35, 1990, S. 317-341; Hrebiniak, L. G. und W. F. Joyce. „Organizational Adaptation: Strategic Choice and Environmental Determinism", Administrative Science Quarterly, Jg. 30, 1985, S. 336-349; Tosi, H. L. und L. R. Gomez-Mejia. „The Decoupling of CEO Pay and Performance: An Agency Theory Perspective", Administrative Science Quarterly, Jg. 34, 1989, S. 169-189. Dem Gabler Verlag für den Aufsatz: Bresser, R. K. F. „Kollektive Unternehmensstrategien", Zeitschrift für Betriebswirtschaft, Jg. 59, 1989, S. 545-564. Dem Institute of Management Sciences für die Artikel: Cool, K. O. und D. Schendel. „Strategie Group Formation and Performance: The Case of the U.S. Pharmaceutical Industry, 1963-1982", Management Science, Jg. 33, 1987, S. 1102-1124. Hill, C. W. L„ Hitt, M. A. und R. E. Hoskisson. „Cooperative Versus Competitive Structures in Related and Unrelated Diversified Firms", Organization Science, Jg. 3, 1992, S. 501-521.

VI

Danksagung

Dem Institute of Operations Research and Management Sciences für den Artikel: Eisenhardt, Κ. M. und C. B. Schoonhoven. „Resource-based View of Strategie Alliance Formation: Strategie and Social Effects in Entrepreneurial Firms", Organization Science, Jg. 7, 1996, S. 136-150. Dem Verlag John Wiley & Sons, Ltd. für folgende Beiträge aus dem „Strategie Management Journal": Amit, R. und R Schoemaker. „Strategie Assets and Organizational Rent", Strategie Management Journal, Jg. 14, 1993, S. 33-46; Ansoff, H. I. „Critique of Henry Mintzberg's ,The Design School: Reconsidering the Basic Premises of Strategic Management'", Strategic Management Journal, Jg. 12, 1991, S. 449-461; Chatterjee, S. und Β. Wernerfeit. „The Link Between Resources and Type of Diversification: Theory and Evidence", Strategic Management Journal, Jg. 12, 1991, S. 33-48; Goold, M. und J. J. Quinn. „The Paradox of Strategic Controls", Strategic Management Journal, Jg. 11, 1990, S. 43-57; Lieberman, M. B. und D. B. Montgomery. ,first-Mover Advantages", Strategic Management Journal, Jg. 9, Summer 1988 (Special Issue), S. 41-58; Mintzberg, H. „The Design School: Reconsidering the Basic Premises of Strategic Management", Strategic Management Journal, Jg. 11, 1990, S. 171-195; Peteraf, M. A. „The Cornerstones of Competitive Advantage: A Resource-Based View", Strategic Management Journal, Jg. 14, 1993, S. 179-191; Reger, R. K. und A. S. Huff. „Strategic Groups: A Cognitive Perspective", Strategic Management Journal, Jg. 14,1993, S. 103-124; Weigelt, K. und I. MacMillan. „An Interactive Strategic Analysis Framework", Strategic Management Journal, Jg. 9, Summer 1988 (Special Issue), S. 27-40; Zajac, E. J. „CEO Selection, Succession, Compensation and Firm Performance: A Theoretical Integration and Empirical Analysis", Strategic Management Journal, Jg. 11, 1990, S. 217-230. Karl E. Weick für seinen Aufsatz: Weick, Κ. E. „Substitutes for Strategy", in: Teece, D. (Hrsg.), The Competitive Challenge. Strategies for Industrial Innovation and Renewal, Ballinger Publishing Company, Cambridge, 1987, S. 221-233.

Vorwort

Zu Beginn der Entwicklung des Strategischen Managements vor dreißig Jahren entstammten die meisten Impulse für Theorie und Praxis der Unternehmungs- und Consultingpraxis. Inzwischen hat sich die Beeinflussungsrichtung aufgrund der wissenschaftlichen Entwicklungen umgekehrt. Die weltweiten Bemühungen von Wirtschaftswissenschaftlern haben dazu geführt, daß sich die theoretischen Ansätze zum Strategischen Management sehr stark vermehrt haben. Führende Consultingfirmen sehen sich seit der Mitte der neunziger Jahre dazu veranlaßt, internationale Arbeitsgruppen zu bilden, um den Stand der theoretischen Entwicklung aufzuarbeiten. Die Boston Consulting Group hat z.B. zu diesem Zweck die „Strategy Practice Initiative" gestartet, und McKinsey & Company riefen die „Strategy Theory Initiative" ins Leben. Dieses Buch wendet sich an Theoretiker und Praktiker, indem es die gegenwärtig vorherrschenden Diskussionen zur Theorie des Strategischen Managements darstellt. Dabei werden unterschiedliche theoretische Bezugsrahmen aus mehreren sozialwissenschaftlichen Disziplinen aufgearbeitet und auf insgesamt vierzehn Problembereiche angewandt. Ohne die tatkräftige Unterstützung einer Reihe von freundlichen Helfern hätte dieses Buch in der vorliegenden Form nicht erstellt werden können. Danken möchte ich meinen Mitarbeitern, Herrn Dr. Wing Hin Chung, Herrn Dipl.-Kfm. Erik Eschen und Herrn Dipl.Kfm. Klemens Millonig, die den Text zur Korrektur gelesen und durch ihre Kommentare zu inhaltlichen Klarstellungen beigetragen haben. Mein besonderer Dank gilt meiner Sekretärin, Frau Sybille Graf, deren Sorgfalt sich nicht nur auf ihren persönlichen Einsatz im Hinblick auf die Erstellung des Manuskripts beschränkte. Sie hat darüber hinaus auch noch ihren Ehemann, Herrn Jürgen Graf, dafür gewinnen können, bei den redaktionellen Feinarbeiten behilflich zu sein. Mein Dank gilt deshalb auch Herrn Graf. Frau cand. rer.pol. Silke Strehlow danke ich für die gewissenhafte Überarbeitung des Literaturverzeichnisses. Andrea Laar hat mir durch ihre Liebe und ihr Verständnis während der vergangenen fünfzehn Monate den Mut gegeben, dieses Projekt ohne große Abschweifungen abzuschließen. Mein Dank hierfür läßt sich nicht in Worte fassen. Februar 1998

Rudi K. F. Bresser

Inhaltsverzeichnis

Danksagung Vorwort

V VII

Teil I Einleitung

1

1

3

Anliegen und Aufbau des Buches

Teil II Perspektiven zum Planungsprozeß 2

Der Streit zwischen Planern und Inkrementalisten

Mintzberg, H. „The Design School: Reconsidering the Basic Premises of Strategic Management", Strategic Management Journal, Jg. 11, 1990, S. 171-195 Ansoff, H. I. „Critique of Henry Mintzberg's ,The Design School: Reconsidering the Basic Premises of Strategic Management'", Strategic Management Journal, Jg. 12, 1991, S. 449-461 Weick, K. E. „Substitutes for Strategy", in: Teece, D. (Hrsg.), The Competitive Challenge. Strategies for Industrial Innovation and Renewal, Ballinger Publishing Company, Cambridge, 1987, S. 221-233

9 11

14

39

52

Teil III Strategische Entscheidungsträger

69

3

71

Der Konflikt zwischen Prinzipalen und Agenten

Tosi, H. L. und L. R. Gomez-Mejia. „The Decoupling of CEO Pay and Performance: An Agency Theory Perspective", Administrative Science Quarterly, Jg. 34, 1989, S. 169-189. . Zajac, E. J. „CEO Selection, Succession, Compensation and Firm Performance: A Theoretical Integration and Empirical Analysis", Strategic Management Journal, Jg. 11, 1990, S. 217-230 4

Strategische Entscheidungsträger und Unternehmungsperformance

Hambrick, D. C. und P. A. Mason. „Upper Echelons: The Organization as a Reflection of Its Top Managers", Academy of Management Review, Jg. 9, 1984, S. 193-206 Finkelstein, S. und D. C. Hambrick. „Top-Management-Team Tenure and Organizational Outcomes: The Moderating Role of Managerial Discretion", Administrative Science Quarterly, Jg. 35, 1990, S. 484-503

75

96 Ill 115

129

X

Inhaltsverzeichnis

Thomas, J. Β., Clark, S. M. und D. A. Gioia. „Strategie Sensemaking and Organizational Performance: Linkages Among Scanning, Interpretation, Action, and Outcomes", Academy of Management Journal, Jg. 36,1993, S. 239-270

149

Teil IV Externe Umweltanalysen

181

5

183

Branchenstruktur-und strategische Gruppenanalysen

Cool, K. O. und D. Schendel. „Strategie Group Formation and Performance: The Case of the U.S. Pharmaceutical Industry, 1963-1982", Management Science, Jg. 33, 1987, S. 1102-1124 Reger, R. K. und A. S. Huff. „Strategic Groups: A Cognitive Perspective", Strategic Management Journal, Jg. 14, 1993, S. 103-124

213

6

235

Die Konkurrentenanalyse

190

Weigelt, K. und I. MacMillan. „An Interactive Strategic Analysis Framework", Strategic Management Journal, Jg. 9, Summer 1988 (Special Issue), S. 27-40 Zajac, E. J. und M. H. Bazerman. ,31ind Spots in Industry and Competitor Analysis: Implications of Interfirm (Mis)perceptions for Strategic Decisions", Academy of Management Review, Jg. 16, 1991, S. 37-56 Jacobson, R. „The , Austrian' School of Strategy", Academy of Management Review, Jg. 17, 1992, S. 782-807

276

Teil V Interne Umweltanalysen

303

7

305

Ressourcen und Fähigkeiten der Unternehmung

Peteraf, M. A. „The Cornerstones of Competitive Advantage: A Resource-Based View", Strategic Management Journal, Jg. 14, 1993, S. 179-191 Amit, R. und P. Schoemaker. „Strategic Assets and Organizational Rent", Strategic Management Journal, Jg. 14, 1993, S. 33-46

242

256

312 325

Teil VI Strategieauswahl

339

8

341

Konvergierende Strategien

Haunschild, P. R. „Interorganizational Imitation: The Impact of Interlocks on Corporate Acquisition Activity", Administrative Science Quarterly, Jg. 38, 1993, S. 564-592 Deephouse, D. L., .Does Isomorphism Legitimate?", Academy of Management Journal, Jg. 39, 1996, S. 1024-1039

347

9

393

Gesamtunternehmungsstrategien

Hill, C. W. L. „Cooperation, Opportunism, and the Invisible Hand: Implications for Transaction Cost Theory", Academy of Management Review, Jg. 15, 1990, S. 500-513. . . .

376

403

Inhaltsverzeichnis

XI

Chatteijee, S. und Β. Wernerfeit. „The Link Between Resources and Type of Diversification: Theory and Evidence", Strategic Management Journal, Jg. 12, 1991, S. 33-48

417

10 Geschäftsbereichsstrategien

433

Lieberman, M. B. und D. B. Montgomery. .first-Mover Advantages", Strategic Management Journal, Jg. 9, Summer 1988 (Special Issue), S. 41-58 Hill, C. W. L. „Differentiation Versus Low Cost or Differentiation and Low Cost: A Contingency Framework", Academy of Management Review, Jg. 13,1988, S. 401-412. . . .

459

11 Kollektive Strategien

471

Bresser, R. K. F. „Kollektive Unternehmensstrategien", Zeitschrift fur Betriebswirtschaft, Jg. 59,1989, S. 545-564 Eisenhardt, Κ. M. und C. B. Schoonhoven. „Resource-based View of Strategic Alliance Formation: Strategic and Social Effects in Entrepreneurial Firms", Organization Science, Jg. 7,1996, S. 136-150

441

482

502

Teil VII Strategieimplementierung

517

12 Strukturen, Führungspersonal, Kulturen und Mikropolitik

519

Hill, C. W. L., Hitt, M. A. und R. E. Hoskisson. „Cooperative Versus Competitive Structures in Related and Unrelated Diversified Firms", Organization Science, Jg. 3,1992, S. 501-521. Howell, J. M. und C. A. Higgins. „Champions of Technological Innovation", Administrative Science Quarterly, Jg. 35, 1990, S. 317-341

532 553

Teil VIII Strategische Kontrolle

579

13 Bewertung und Systeme

581

Goold, M. und J. J. Quinn. „The Paradox of Strategie Controls", Strategie Management Journal, Jg. 11, 1990, S. 43-57 Eschen, E. „Finanzkennzahlen im Strategischen Management", Arbeitspapier des Lehrstuhls für Unternehmungsplanung und -politik, Freie Universität Berlin, 1998

587 602

Teil IX Strategischer Wandel

627

14 Strategische Wahlfreiheit versus strukturelle Trägheit

629

Kelly, D. und T. L. Amburgey. „Organizational Inertia and Momentum: A Dynamic Model of Strategic Change", Academy of Management Journal, Jg. 34, 1991, S. 591-612 Hrebiniak, L. G. und W. F. Joyce. „Organizational Adaptation: Strategic Choice and Environmental Determinism", Administrative Science Quarterly, Jg. 30, 1985, S. 336-349.

637 659

XII

Inhaltsverzeichnis

Teil Χ Epilog

673

15 Perspektiven für eine einheitliche Theorie des Strategischen Managements ..

675

Literaturverzeichnis Stichwortverzeichnis

683 711

Teil I Einleitung

1 Anliegen und Aufbau des Buches

Dieses Buch wendet sich an Studierende des Strategischen Managements. Es ist konzipiert als ein Buch für .fortgeschrittene", d.h. für Studierende, die sich die Grundlagen des Strategischen Managements bereits auf der Basis eines konventionellen Lehrbuchs angeeignet haben. Neben Studierenden im Hauptstudium ist das Buch auch von Interesse für Doktoranden, Praktiker und Wissenschaftler aus Nachbardisziplinen, die sich über die wichtigsten Forschungsfragen des Strategischen Managements informieren wollen. Ganz allgemein lassen sich die Zielsetzungen des Strategischen Managements mit der Schaffung nachhaltiger Wettbewerbsvorteile sowie der Planung und Steuerung der längerfristigen Unternehmungsentwicklung beschreiben (Welge & Al-Laham, 1992: 15; Wheelen & Hunger, 1995: 4; Mintzberg & Quinn, 1996: 3). Der Versuch, solche allgemeinen Definitionen oder Zweckbestimmungen des Strategischen Managements aufzustellen, ist allerdings riskant, da in diesem betriebswirtschaftlichen Forschungsgebiet nur sehr wenige allgemein akzeptierte Ansichten bestehen. Wenn man Igor Ansoffs Buch Corporate Strategy und die erste Ausgabe des klassischen Harvard Textes Business Policy: Text and Cases (Learned et al.), beide aus dem Jahre 1965, als die ersten geschlossenen Beiträge des Strategischen Managements betrachtet, so ist die Einsicht frappierend und fast beängstigend, daß wir nach einer über dreißigjährigen Entwicklungsgeschichte vor einer Disziplin stehen, die sich durch eine große Uneinheitlichkeit und Meinungsvielfalt auszeichnet.

Ein Abbild der Meinungsvielfalt Die Meinungsvielfalt innerhalb des Strategischen Managements begründet die Gestaltung des Buches als detailliert kommentierten Reader. Da es in einem fortgeschrittenen Stadium des Studiums des Strategischen Managements darauf ankommt, unterschiedliche theoretische Ansätze und Perspektiven eigenständig beurteilen zu können, ist es sinnvoll, die Autoren mit ihren unterschiedlichen Ansichten selbst zu Wort kommen zu lassen und nicht ausschließlich durch den Perzeptionsfilter eines Lehrbuchautors. Neben der akkurateren Wiedergabe der Meinungsvielfalt hat das Medium Reader aber noch andere Vorteile. Originalbeiträge regen den Leser zur kritischen Reflexion und selbständigen Meinungsbildung an. Anders ausgedrückt minimieren sie die Gefahr der voreiligen Übernahme der Ansichten eines Autors, der verschiedene Meinungen zusammenfaßt und seinen Präferenzen entsprechend interpretiert. Da kontroverse Originalbeiträge oft pointiert und provokant geschrieben sind, ist ein Reader interessanter und spannender als ein traditionelles Lehrbuch. Die meisten Beiträge des vorliegenden Readers entstammen wissenschaftlichen Fachzeitschriften, denn in diesen werden, zumindest im angloamerikanischen Raum, aktuelle wissenschaftliche Kontroversen auf hohem Niveau

4

R. Κ. F. Bresser

ausgetragen. Ein interessanter und problemorientierter Aufbau eines Readers kann das Interesse der Studierenden an wissenschaftlichen Fachzeitschriften fördern und sie animieren, diesen Journalen ihre regelmäßige Aufmerksamkeit zu schenken. Es ist meine Erfahrung, daß Studierende, die sich im Verlauf ihres Studiums insbesondere mit konventionellen Lehrbüchern auseinandersetzen müssen, die Lektüre von Fachzeitschriften (und die sich darin widerspiegelnde Aktualität) vernachlässigen.

Impulse zur Integration Neben der Wiedergabe der unterschiedlichen Forschungsansätze des Strategischen Managements soll dieses Buch auch einen Beitrag zur Integration leisten. Obwohl zahlreiche behavioristisch ausgerichtete, sozialwissenschaftliche Disziplinen (z.B. die Politikwissenschaften, die Psychologie, die Geschichtswissenschaften, die Soziologie) sich dem Phänomen der Unternehmungsstrategie zugewandt haben, verläuft die derzeitige Entwicklung des wissenschaftlichen Diskurses kontrovers. Die theoretischen Bemühungen im Strategiebereich werden von zwei Ansätzen dominiert: Der erste Ansatz wird von der jüngeren, behavioristisch ausgerichteten Managementtheorie inspiriert.1 Der zweite Ansatz argumentiert vor dem Hintergrund verschiedener ökonomischer Theorien, wobei insbesondere die Ansätze der neuen Institutionenökonomie bedeutungsvoll sind (Fredrickson, 1990). Leider ist zu beobachten, daß diese beiden Ströme der strategischen Managementforschung sich weitgehend losgelöst voneinander entwickelt haben und entwickeln.2 Eine Zielsetzung des Buches besteht darin, Ansätze zur Entwicklung einer geschlosseneren, integrierteren Theorie des Strategischen Managements vorzustellen und anzuregen. Es wird deshalb wiederholt die Frage aufgeworfen, inwieweit Erkenntnislücken vor dem Hintergrund ökonomischer Theorien durch verhaltenswissenschaftliche Fragestellungen (zumindest partiell) geschlossen werden können und vice versa. Ein Vorurteil des Autors, das sich in den Versuchen, Integrations- und Erweiterungspotential aufzuzeigen, durchgängig durch dieses Buch zieht, betrifft die Dynamisierung theoretischer Ansätze. Es wird davon ausgegangen, daß ein umfassendes, erklärungskräftiges Theoriegebäude des Strategischen Managements dynamisch sein muß.

1

2

Die behavioristisch angelegte Managementtheorie hat bisher noch kein geschlossenes Theoriegebäude entwickeln können. Sie umfaßt Ansätze, die betriebswirtschaftliche Fragen zum Phänomen Strategie um verhaltenswissenschaftliche Überlegungen anreichern und die ihre Ursprünge insbesondere in der Psychologie, der Soziologie und in den Politikwissenschaften haben (Bowman, 1974; Zajac, 1992). Einige Wissenschaftler begrüßen diese Separation verhaltenswissenschaftlicher und ökonomischer Ansätze (z.B. Montgomery et al., 1989), andere sehen in dem Anwachsen ökonomischer Ansätze sogar eine Gefahr (vgl. Hirsch et al., 1990). In diesem Buch wird demgegenüber von der Notwendigkeit einer Integration der beiden Ansätze ausgegangen (ähnlich auch Zajac, 1992). Einen umfassenden Uberblick zum Stand der Strategieforschung aus managementtheoretischer (verhaltenswissenschaftlicher) Sicht geben Mintzberg & Quinn (1996). Die ökonomische Sichtweise wird z.B. von Besanko et al. (1996) und partiell ebenso von Milgrom & Roberts (1992) dargestellt. Die beste Würdigung der verschiedenen Ansätze im deutschsprachigen Raum stammt von Knyphausen (1995).

Anliegen und Aufbau

5

Konzeptionelle und empirische Strategieforschung Seit dem Ende der 70er Jahre hat sich die strategische Managementforschung verstärkt um eine empirische Stützung und Präzisierung ihrer präskriptiven und konzeptionellen Aussagen bemüht. Dieser Prozeß wurde maßgeblich durch die Herausgeber des Strategie Management Journals, insbesondere Dan Schendel, beeinflußt. Schendel hat nicht nur als einer der ersten die Forderung nach einem stärkeren Empiriebezug des Strategischen Managements aufgestellt (Schendel & Hofer, 1979: 394), er hat dieses Ziel auch aktiv durch die Gestaltung des Journals umgesetzt. Der empirischen Ausrichtung des Strategischen Managements wird mitunter mit Skepsis begegnet (Daft & Buenger, 1990; Schreyögg, 1984: 229, 274f.; 1992). Schreyögg argumentiert zum Beispiel, daß es in einem essentiell sozialwissenschaftlichen Forschungsbereich nicht möglich und auch nicht sinnvoll sein kann, nach dem Muster der Naturwissenschaften generell gültige Gesetzmäßigkeiten der strategischen Unternehmungsführung ergründen zu wollen. Dieser Ansicht ist voll zuzustimmen, der Wert der empirischen Forschung bleibt von diesen Bedenken allerdings unberührt. Die kritische Einstellung gegenüber der Empirie im Strategiebereich ist zum Teil selbstverschuldet. Frühe, insbesondere der Praxis entstammende, strategische Konzepte traten unter dem heroischen Anspruch an, Gesetzmäßigkeiten empirisch nachweisen zu wollen. So bemühten sich z.B. die Begründer des PIMS-Projektes, „laws of the market place" zu ergründen (Schoeffler et al., 1974), und sowohl das Erfahrungskurven-Konzept als auch die frühen Portfolio-Modelle empfahlen das unkritische Befolgen von „Normstrategien" (Henderson, 1974; Scholz, 1987: 183-195). Darüber hinaus war ein großer Teil der empirischen Untersuchungen von einer Rentabilitätsorientierung determiniert, die einer theoretischen und methodischen Vielfalt im Wege stand (Daft & Buenger, 1990). In der empirischen Forschung der Gegenwart spielen derartige überzogene Erwartungen und einseitige theoretische Ausrichtungen allerdings keine Rolle mehr. Es geht heute primär um die empirische Erforschung von Regelmäßigkeiten (nicht Gesetzmäßigkeiten) und von ungewöhnlichen Fällen, um zu der Entwicklung von aussagekräftigen Hypothesen, Theorien und praktischen Gestaltungsempfehlungen des Strategischen Managements beitragen zu können. Obwohl von der„Natur" des Phänomens Strategie her Untersuchungen von Rentabilitätswirkungen nach wie vor eine große Rolle spielen (und auch spielen sollen), fließen inzwischen eine Vielzahl von theoretischen Perspektiven in den Theoriebildungsprozeß ein.3 Aus der Gruppe der möglichen empirischen Forschungsdesigns sind insbesondere zwei Vorgehensweisen zur Theoriebildung im Strategiebereich geeignet: Querschnittsund/oder Längsschnittsanalysen, die auf der Grundlage größerer Stichproben den Vergleich mehrerer Unternehmungen anstreben und (qualitative) Fallstudien, die sich auf die Sammlung und Auswertung von Daten einzelner Unternehmungen konzentrieren (Kubicek, 1975: 58-67).4 3

4

Man findet z.B. behavioristisch angelegte Studien, die populationsökologische (Hannan & Freeman, 1977) oder soziologisch-institutionalistische (Powell & DiMaggio, 1991) Perspektiven als Ausgangspunkt des Studiums strategischer Phänomene verwendet haben. Die Rentabilität einzelner Unternehmungen spielt in diesen Ansätzen oft nur eine untergeordnete Rolle. Forschungsdesigns, die aktive, gestaltende Eingriffe seitens der Forscher ermöglichen, wie z.B. Laborexperimente, Feldexperimente und die Aktionsforschung, dürften zur Theoriebildung im Strategischen Management weniger einsetzbar sein. Dies liegt insbesondere daran, daß es bei Laborexpe-

6

R. Κ. F. Bresser

Querschnitts- und Längsschnittsuntersuchungen sind zur Theoriebildung geeignet, denn sie geben Anhaltspunkte über generelle und durchschnittliche Tendenzen. Hierzu ein Beispiel: Eine auf einer repräsentativen Stichprobe von Unternehmungen basierende Studie über die strategischen Erfolgsfaktoren innerhalb einer bestimmten Branche wird (unter Verwendung von korrelations- und regressionstechnischen Auswertungsverfahren) vor allem Aussagen über das Mittelmaß innerhalb der Branche zulassen. Die Kenntnis dessen, was sich durchschnittlich als am erfolgreichsten erwiesen hat, ist sowohl für den Theoretiker als auch für den Praktiker eminent wichtig, denn nur vor dem Hintergrund der Kenntnis des strategischen Mittelmaßes ist es möglich, besonders erfolgreiche (bzw. besonders erfolglose) Unternehmungen und Strategien zu analysieren. Da es bei der Schaffung nachhaltiger Wettbewerbsvorteile oft darauf ankommt, idiosynkratische Kompetenzen differenzierungsfördernder und/oder kostensenkender einzusetzen als die Konkurrenz (Starbuck & Nystrom, 1981; Hamel & Prahalad, 1994), kann der strategische Gestaltungsspielraum der Unternehmung erst dann sinnvoll genutzt werden, wenn Kenntnisse über das strategische Mittelmaß gewonnen wurden. Auch Fallstudien dienen der Theoriebildung, da sie einerseits während der ersten Phasen der Beschäftigung mit einem strategischen Problem in der Lage sind, eine Fülle von Anregungen für die Vorbereitung umfassenderer Quer- oder Längsschnittsuntersuchungen bereitzustellen (Heydebrand, 1973). Andererseits dienen Fallstudien der Vertiefung der Ergebnisse vorangehender Feldstudien mit größerem Stichprobenumfang. Fallstudien bauen dann gewissermaßen auf den statistischen Ausreißern früherer Quer- oder Längsschnittsuntersuchungen auf, indem sie z.B. den besonderen strategischen Erfolg (oder Mißerfolg 5 ) einzelner Unternehmungen einer detaillierten Maßnahmen- und Wirkungsanalyse unterziehen. Fallstudiendesigns wurden früher oft wegen ihrer mangelnden Systematik und Rigorosität kritisiert (Kubicek, 1975: 60). Die Weiterentwicklung dieses Forschungsdesigns während der letzten beiden Jahrzehnte zeigt jedoch, daß systematisches Vorgehen beim Einsatz von Fallstudien nicht nur wünschenswert, sondern auch möglich ist (Eisenhardt, 1989a; Yin, 1984). Insbesondere die als „grounded theory" bekannte Methodologie (Glaser & Strauss, 1967) hat zu einer großen Zahl von qualitativen Studien geführt, die der Theoriebildung im Strategischen Management wichtige Impulse gegeben haben.6

5

6

rimenten schwierig sein wird, den verhaltensrelevanten betriebsintemen und -externen Kontext im Labor nachzubilden, und für Feldexperimente und Aktionsforschung wird es schwierig sein, Probanden zu finden, die bereit sind, die Bedingungen in ihren Unternehmungen versuchsweise zu verändern (Kubicek, 1975: 68f.). Diese Überlegung spricht jedoch nicht gegen die grundsätzliche Eignung dieser aktiven Forschungsdesigns zur Theorieentwicklung im Strategiebereich, sie stellt lediglich ihre praktische Realisierbarkeit in Frage. Vgl. hierzu auch Schwenk (1982) sowie Knyphausen (1995: 229-234). So konnten z.B. Bresser et al. (1994) in einer Längsschnittsstudie über die Zusammenhänge kompetitiver und kollektiver Strategien in der US-Sparkassenbranche zwei statistische Ausreißer identifizieren, die im Nachhinein eine besondere Behandlung erfuhren. Die beiden fraglichen Firmen gingen einige Jahre nach dem Ende der Untersuchungsperiode der Studie in Konkurs. Die (späteren) Schwierigkeiten der Firmen waren mit dem von Bresser et al. verwendeten Instrumentarium durchaus prognostizierbar. Dieser Umstand führte zu einer Revision des Frühwarnsystems des „Office of Thrift Supervision", d.h. der Regierungsbehörde, die das Geschäftsgebaren der amerikanischen Sparkassen überwacht. Vgl. z.B. die Studie von Browning et al. (1995) über das Entstehen von Kooperation in einer wettbewerbsintensiven Branche oder die Arbeiten von Eisenhardt (1989b) zur strategischen Entscheidungsfindung in turbulenten Umwelten.

Anliegen und Aufbau

7

Aus dem bisher Gesagten folgt, daß die Beiträge dieses Buches sowohl konzeptionelle als auch empirische Aufsätze umfassen. Das wichtigste Auswahlkriterium für die Einzelbeiträge war deren jeweiliger Beitrag zur Theorieentwicklung. Um das Phänomen Unternehmungsstrategie zu verstehen, sind vielfaltige theoretische und methodische Ansätze vonnöten (Daft & Buenger, 1990; Fredrickson, 1990) und in diesem Reader berücksichtigt. Generell gilt, daß ausgewählte konzeptionelle Beiträge aufzeigen, wie die erarbeiteten Überlegungen und Modelle empirisch überprüft werden bzw. bisher vorherrschende empirische Forschung ergänzen können. Den empirischen Beiträgen ist zu entnehmen, wie sie auf bestehender Theorie aufbauen, diese überprüfen und durch das Aufwerfen neuer Forschungsfragen erweitern. Die Auswahl der Beiträge dieses Buches und die zusammenfassenden Diskussionen unterliegen einer inhaltlichen Eingrenzung. Es wird versucht, die „MainstreamDiskussion" der strategischen Managementforschung nachzuzeichnen. Diese Fokussierung erfolgt aufgrund der Flut der Beiträge und Ansätze zum Strategischen Management und auch aufgrund der fehlenden Integration innerhalb der „Mainstream-Diskussion". Ein Forschungsgegenstand oder -ansatz wird dabei als dem „mainstream" zugehörig angesehen, wenn er bereits zu einer Serie von wissenschaftlichen Veröffentlichungen in anerkannten englischsprachigen Zeitschriften geführt hat.7

Ein prozeßorientiertes Ordnungsschema Zur Realisierung der Integrationsaufgabe dieses Buches ist es vonnöten, ein geeignetes Ordnungsschema auszuwählen. Eine der wenigen Übereinstimmungen in der Strategieliteratur betrifft die fundamentalen Dimensionen, die berücksichtigt werden müssen, um das Strategiephänomen zu erforschen. Drei Dimensionen gelten als bedeutsam: der Prozeß der Strategiebildung, der Inhalt von Strategien und der Kontext von Strategien (Pettigrew, 1987; Mintzberg, 1990a; Montgomery et al., 1989). Der Strategieprozeß betrifft die Art und Weise, wie in Unternehmungen Strategien entwickelt werden. Dieser Prozeß kann eher strukturiert (formal) oder eher unstrukturiert (informal) ablaufen: Auf jeden Fall muß eine Organisation mehrere Aktionsstufen durchlaufen, bevor sie eine Strategie inhaltlich determiniert hat (De Wit & Meyer, 1994: XI). Der Strategieinhalt ist somit das Resultat oder Produkt des Strategiebildungsprozesses; die Inhalte legen fest, welche Aktionsprogramme verfolgt werden sollen, um die Wettbewerbsfähigkeit der Unternehmung sicherzustellen. Der Kontext von Strategien bezieht sich auf jene organisatorischen und externen Umweltbedingungen, die sich auswirken, während Strategiebildungsprozesse ablaufen und zu bestimmten strategischen Inhalten führen. Aus dieser Bestimmung der drei Strategiedimensionen ergibt sich eine deutliche Dominanz der Prozeßdimension. Während es möglich (aber oft nicht sehr sinnvoll) ist, die Inhalte oder Kontexte von Strategien isoliert zu betrachten, macht dies bei der Prozeßdi7

Spezielle Fragestellungen des Strategischen Managements wie z.B. die Ökologisierung oder die Internationalisierung von Unternehmungsstrategien werden ebenfalls nicht gesondert thematisiert, da sich diese Fragestellungen gut in die „Mainstream-Ansätze" eingliedern lassen. Aus diesem Grunde werden Internationalisierungs- oder Ökologisierungsgedanken immer dann berücksichtigt, wenn sie im Rahmen einer inhaltlich-theoretischen Fragestellung dieses Buches zu Forschungsergebnissen geführt haben.

8

R. Κ. F. Bresser

mension wenig Sinn (Pettigrew, 1992). Für das Verständnis eines sich entfaltenden Strategiebildungsprozesses ist es notwendig, sowohl die kontextuellen Rahmenbedingungen als auch die möglichen Resultate des Prozesses zu verstehen. Aus diesem Grunde wird als Ordnungsschema für dieses Buch eine Prozeßperspektive gewählt, die Strategieinhalte und -kontexte impliziert. Welche konkreten Abstufungen und Einteilungen dieses prozeßbezogene Ordnungsschema haben wird, soll erst am Ende des zweiten Kapitels expliziert werden. Das Schema ergibt sich aus der Logik der dort vorgetragenen Kontroverse zwischen den „Planern" und den, Jnkrementalisten".

Teil II Perspektiven zum Planungsprozeß

2 Der Streit zwischen Planern und Inkrementalisten

Es wäre sicherlich hilfreich, wenn man zu Beginn eines Strategiebuches zunächst eine eindeutige Definition des untersuchten Phänomens „Unternehmungsstrategie" anbieten könnte. Eine solche klare Definition könnte nicht nur den weiteren Verlauf des Buches strukturieren helfen, sie könnte auch dazu dienen, konfligierende Auffassungen und Interpretationen einer Lösung zuzuführen. Zwar wird ein allgemeiner Definitionsversuch, und damit eine Strukturierungshilfe, das Resultat dieses Kapitels sein, Aufgabe dieser Strukturierungshilfe ist es dann jedoch nicht, konfligierende Auffassungen aufzulösen. Die Divergenzen über das, was man unter Unternehmungsstrategie und Strategischem Management verstehen sollte, sind fundamental, und es ist gerade ein Ziel dieses Readers, diese Meinungsunterschiede unverblümt offenzulegen. Das angestrebte prozeßbezogene Ordnungsschema (die Strukturierungshilfe) soll lediglich ermöglichen, das Strategiephänomen vor dem Hintergrund konfligierender Ansätze zu diskutieren und zu erforschen. Das Schema soll der Integration dienen, indem es dazu anhält, eine eklektische Theorie8 des Strategischen Managements zu entwickeln.

Prozeßmodelle der Strategieformation Die Art und Weise, in der sich Strategien in Unternehmungen herausbilden, gehört zu der ersten und historisch gesehen ältesten Kontroverse der strategischen Managementliteratur. Es gibt kein einheitliches Prozeßmodell der Strategieformation. Chaffee (1985) unterscheidet drei Modelle, Bailey und Johnson (1992) sechs, und Mintzberg (1990a) differenziert sogar zwischen zehn Schulen und Prozeßperspektiven des Strategischen Managements. Diese jeweils unterschiedliche Anzahl ist allerdings ein Resultat unterschiedlich breit gewählter Abgrenzungskategorien. Werden diese sehr breit ausgelegt, so lassen sich alle Unterscheidungen unter zwei Modellen subsumieren: einem präskriptiven Modell der synoptischen Planung und einem deskriptiven Modell des Inkrementalismus (Mintzberg, 1990a; DeWit & Meyer, 1994: 37). Das Planungsmodell entstammt der frühesten Entwicklungsphase des Strategischen Managements und ist eng mit dem Namen Igor Ansoff (1965) verbunden. Das Modell ist 8

Die Bezeichnung „eklektisch" kennzeichnet den Versuch, einen Forschungsgegenstand auf der Basis verschiedener theoretischer Ansätze und Paradigmen zu erklären (Gioia & Pitre, 1990; Daft & Buenger, 1990). Wie in der Einleitung expliziert, lassen sich die unterschiedlichen Ansätze des Strategischen Managements in zwei Gruppen einteilen: behavioristische und ökonomische Ansätze. Durch eine eklektische Vorgehensweise können Erklärungsdefizite ausgefüllt werden, die sich vor dem Hintergrund eines einzigen, dominierenden theoretischen Ansatzes ergeben. Eklektische Ansätze erlauben somit, die unterschiedlichsten Aspekte eines Phänomens zu verstehen, und sie können zu durchaus originellen Weiterentwicklungen dominierender Paradigmen führen (Kuhn, 1962; 1974).

12

R. Κ. F. Bresser

kennzeichnend für den größten Teil der heute veröffentlichten Lehrbücher (vgl. z.B. Welge & Al-Laham, 1992; Wheelen & Hunger, 1995). Dieser Ansatz ist präskriptiv, denn er macht Gestaltungsempfehlungen für den Prozeß der strategischen Planung. Ein Befolgen der Gestaltungsempfehlungen soll ein Höchstmaß an Effizienz bei der Durchführung der Planungsaufgabe ermöglichen. Als wesentliche Charakteristika aller synoptischen Planungsprozeßmodelle hebt Schreyögg (1984: 133f.) fünf Tatbestände hervor: • Die Strategieformulierung ist durch einen systematischen Prozeß gekennzeichnet, der mehrere aufeinanderfolgende Schritte umfaßt (Zielformulierung, SWOT-Analyse, Bewertung und Auswahl von Strategiealternativen). • Der so formulierte strategische Plan bildet die Grundlage für die Erstellung detaillierter Programme, Kurzfristpläne und Budgets, durch die der Plan implementiert werden soll. • Die Organisation und das Führungssystem werden auf die Strategie und ihre Implementierungserfordernisse zugeschnitten. • Der strategische Plan und daraus abgeleitete Maßnahmenpläne sind umfassend, d.h. sie sollten sämtliche als wichtig erkannte Faktoren und Aktivitätsbereiche berücksichtigen. • Strategische Planungsentscheidungen sind eine originäre, nichtdelegierbare Aufgabe der Unternehmungsleitung. Das Inkrementalmodell entwickelte sich aus der empirischen Analyse strategischer Entscheidungsprozesse, in denen eine eindeutige, phasendeterminierte Aktivitätenfolge, wie von den Planungsansätzen unterstellt, nicht nachgewiesen werden konnte (Braybrooke & Lindblom, 1963; Mintzberg et al., 1976; Pascale, 1984; Mintzberg & Waters, 1990). Hauptvertreter der inkrementalen Planungsprozeßmodelle sind gegenwärtig Henry Mintzberg (1987; 1990a) und James Brian Quinn (1980). Frühe Vertreter des Inkrementalismus argumentierten, daß aufgrund von Faktoren, wie begrenzter Informationsverarbeitungskapazität der Planungsträger und begrenzter Verfügbarkeit von Informationen, ein Planungsprozeß eher den Charakter des sich „Durchwursteins" (muddling through) und der Strategie der kleinen Schritte (disjointed incrementalism) habe, als den eines systematisch ablaufenden Prozesses (Lindblom, 1969). Diese Überlegungen wurden von Quinn (1980) zu einem „logischen Inkrementalismus" weiterentwickelt, der den Auffassungen der gegenwärtigen Vertreter von Inkrementalmodellen entspricht (Mintzberg & Quinn, 1996). Als wichtige Merkmale des logischen Inkrementalismus lassen sich wieder fünf Faktoren hervorheben (Schreyögg, 1984: 239ff.; Quinn, 1978; 1980): • Strategische Initiativen und die Vorformulierung von Strategien ergeben sich dezentral in den strategischen Subsystemen (z.B. Divisionen, Funktionsbereichen), und zwar unregelmäßig, nicht nach einem strengen Muster ablaufend. • Die zentrale Aufgabe des Top-Managements besteht in der Steuerung und Konkretisierung der Vorschläge, die in den Subsystemen entstehen, um diese zu einer Gesamtstrategie auszuformen. Obwohl diese Vorgehensweise nicht streng geordnet abläuft, folgt sie doch einer klaren inneren Logik, denn die Steuerungsfunktion wird vom TopManagement bewußt und proaktiv wahrgenommen. • Zur Vorsteuerung entwickelt das Top-Management strategische Globalziele, die an die Subsysteme kommuniziert werden. • Das formale System der strategischen Planung dient primär dazu, das TopManagement bei seiner integrierenden und intervenierenden Steuerungsaufgabe zu un-

Planer und Inkrementalisten

13

terstützen und Innovationsimpulse zu setzen. Zu diesen Funktionen zählen z.B. die Sensibilisierung für visionäre strategische Möglichkeiten und die Konkretisierung von Zukunftsprojekten, die von den Subsystemen vernachlässigt werden. • Die strategischen Pläne fungieren nur als grobe Richtlinie, eine detaillierte Jahresplanung obliegt den strategischen Subsystemen. Das Top-Management toleriert somit Ambiguität und Unbestimmtheit im Prozeßablauf. Eine umfassende Würdigung der frühen Kontroverse zwischen den „Planern" und den „Inkrementalisten" liefert Schreyögg (1984). Diese Kontroverse hat sich in den 80er Jahren fortgesetzt; es entstanden zusätzliche Variationen der beiden Grundmodelle, die ihr Augenmerk auf besondere Tatbestände, wie z.B. Kognitionen, Macht oder Organisationskulturen richteten (Mintzberg, 1990a). Diese Variationen konnten die grundlegende Kontroverse über das Wesen des strategischen Planungsprozesses nicht beenden. Die Kontroverse scheint allerdings gegenwärtig aufgrund jüngerer Entwicklungen etwas entschärft, was am Ende dieses Kapitels verdeutlicht werden soll. Im folgenden werden zunächst zwei Artikel (von Mintzberg und von Ansoff) vorgestellt, die die wesentlichen Argumente der Kontroverse zusammenfassen. Im Anschluß folgt ein sich zur Integration anbietender Beitrag von Weick. Den Abschluß dieses Kapitels bildet die Vorstellung des Prozeßmodells, das als Ordnungsschema dieses Buches Verwendung findet.

Die Artikel Der erste Artikel, „The Design School: Reconsidering The Basic Premises Of Strategic Management" von Henry Mintzberg, greift grundlegende Prämissen der Planungsmodelle an, und zwar auf der Basis der Erkenntnisse der Inkrementalisten. Zentral ist z.B. die Kritik an der Loslösung des (rationalen) Denkens von den Handlungen der Organisationsmitglieder oder an der Trennung von Strategieformulierung und -Implementierung. Zwar bezieht sich Mintzberg explizit auf die Design School der Harvard Business School, er macht jedoch deutlich, daß die vorgetragenen Kritikpunkte für alle präskriptiven Planungsmodelle gelten. Igor Ansoff verteidigt die Planungsansätze in einem zweiten Artikel, „Critique Of Henry Mintzberg's ,The Design School: Reconsidering The Basic Premises Of Strategic Management'", indem er versucht, die Prämissenkritik Mintzbergs zu entkräften. Er hält Mintzberg vor, in seiner Kritik weder methodisch sauber vorgegangen zu sein noch die faktische Evidenz richtig wiedergegeben zu haben. Darüber hinaus kritisiert Ansoff das Modell des organisationalen Lernens der Inkrementalisten als ein antiquiertes, existentielles Modell, das den Anforderungen moderner Unternehmungen nicht gerecht wird. Der dritte Artikel, „Substitutes For Strategy" von Karl Weick, fällt eher in die Argumentationslinie der Inkrementalisten als in die der Planer. Dennoch trägt er Gedanken vor, die zu einer friedlichen Koexistenz beider Ansätze beitragen können. Zentral in Weicks Argumentation ist die menschliche Handlung. Nur durch Handlungen kann Sinn erschlossen und können Ziele und Strategien mit Inhalt gefüllt werden. Handlungen in Unternehmungen können von formalen Plänen ausgelöst werden, aber auch durch Improvisation (just-in-time strategy) im Sinne der Inkrementalisten. Neben der Improvisation kann auch die Vision oder die Zuversicht (confidence) ein effektives, handlungsinduzierendes Substitut für einen formalen Plan abgeben.

14

H. M i n t z b e r g

Strategic

Management

Journal,

Vol.

11, 171-195

(1990)

THE DESIGN SCHOOL: RECONSIDERING THE BASIC PREMISES OF STRATEGIC MANAGEMENT HENRY MINTZBERG

Faculty of Management,

McGill University,

Montreal, Quebec,

Canada

Among the schools of thought ort strategy formation, one in particular underlies almost all prescription in the field. Referred to as the 'design school', it proposes a simple model that views the process as one of design to achieve an essential fit between external threat and opportunity and internal distinctive competence. A number of premises underlie this model: that the process should be one of consciously controlled thought, specifically by the chief executive; that the model must be kept simple and informal; that the strategies produced should be unique, explicit, and simple; and that these strategies should appear fully formulated before they are implemented. This paper discusses and then critiques this model, focusing in particular on the problems of the conscious assessment of strengths and weaknesses, of the need to make strategies explicit, and of the separation between formulation and implementation. In so doing, it calls into question some of the most deep-seated beliefs in the field of strategic management, including its favorite method of pedagogy.

T h e literature that can be subsumed u n d e r 'strategy f o r m a t i o n ' is vast, diverse a n d , since 1980, has b e e n growing at an astonishing rate. T h e r e has b e e n a general tendency to date it back to t h e mid-1960s, although some important publications precede that d a t e , such as N e w m a n ' s initial piece 'to show t h e n a t u r e and importance of strategy' (p. iii) in t h e 1951 edition of his textbook Administrative Action (1951: 110-118). Of course the literature o n military strategy goes back much f u r t h e r , in the case of Sun Tzu probably to t h e f o u r t h century B.C. (Griffith, in Sun T z u , 1971: ix). A good deal of this literature naturally divides itself into distinct schools of thought. In a n o t h e r publication (Mintzberg, 1989), this author has identified ten of these. T h r e e are prescriptive in orientation, treating strategy formation as a process of conceptual design, of formal planning, and of analytical positioning (the latter including much of the research on the content of competitive strategies). Six o t h e r schools deal with specific aspects of t h e process in a descriptive way, and 0143-2095/90/030171-25$12.50 © 1990 by John Wiley & Sons, Ltd.

are labeled the entrepreneurial school (concerned with strategy formation as a visionary process), the cognitive school (a mental process), the learning school (an e m e r g e n t process), and t h e environmental school (a passive process). A final school, also descriptive, but integrative and labeled configurational, by seeking to delineate the stages and sequences of the process, helps to place the findings of these o t h e r schools in context. This p a p e r addresses itself to the first of these schools, in some ways the most entrenched of the ten. Its basic f r a m e w o r k underlies almost all prescription in this field a n d , accordingly, has had e n o r m o u s impact o n how strategy and t h e strategy-making process are conceived in practice as well as in education and research. H e n c e our discussion, and especially critique, Of this school can in some ways b e taken as a c o m m e n t a r y on the currently popular beliefs in the field of strategic m a n a g e m e n t in general. O u r intention, however, is not to dismiss so important a school of t h o u g h t , but rather to understand it better Received 14 December 1987 Revised 22 May 1989

T h e Design School

172

15

H. Mintzberg

and so place it into its natural context, and thereby open up thinking in the held in general. This paper probes first into the basic model of the design school, then into the basic premises that underlie it. That leads to a critique of this school, which gives rise to an attempt to place it into its own viable context—the types of organizations and of situations most suited to it. In conducting this investigation we draw widely on the literature of this school, but use one text in particular—almost certainly this school's best known. In this sense this paper can also be viewed as a rather extensive review of a book that has had a major impact on the field of strategic management.

ORIGINS OF THE DESIGN SCHOOL Ostensibly the simplest and most fundamental view of strategy formation is as a process of informal conception—the use of a few essential concepts to design 'grand strategy.' Of these concepts the most essential is that of congruence or match. In the words of the design school's best-known proponents: 'Economic strategy will be seen as the match between qualification and opportunity that positions a firm in its environment' (Christensen, Andrews, Bower, Hamermesh, and Porter, 1982: 164) 1 'Capture success' seems to be the motto of the design school; 'find out what you are good at and match it with what the world wants and needs.' These capabilities or qualifications have been variously referred to as 'distinctive competence,' 'differential,' 'competitive,' or 'comparative advantage' (the latter more commonly used in the context of public policy), or more simply (and broadly) an organization's 'strengths and weaknesses.' The design school has generally been associated with the Business Policy group at the Harvard Business School. That group has pursued its own strategy for, as we shall see later, there is a clear congruence between the view of strategy formation that it has promoted for several decades and its own pedagogical requirements in using the case study method. The Christensen et al. book quoted from above, entitled Business Policy: Text and Cases, first

appeared in 1965 (by Learned, Christensen, Andrews, and Guth) and quickly became the dominant textbook in the field, as well as the dominant voice for this school of thought. 2 Certainly its text portion, attributed in the various editions to co-author Kenneth Andrews (who also published this material separately (Andrews, 1971,1980a, 1987)) stands as the most outspoken and one of the clearest statements of this school, although claims that this school, or even the concept of business strategy itself, originated with this group at Harvard (e.g. Bower, 1986: vii) do not stand up to scrutiny. Some of the basic concepts that underlie the design school, at least as published, would (as we shall see) appear to have been first stated in the academic world by a Berkeley sociologist named Philip Selznick, in his book Leadership in Administration, published in 1957. Even earlier, though less specific, is a 1955 article by Reilley, possibly the first reflection of this approach. Another key publication in 1962, Strategy and Structure, by historian Alfred Chandler (then at MIT), really established this school's concept of business strategy and its relationship to structure, although mention also has been made of the sophisticated discussion of 'Managerial Strategies' in David G . Moore's paper of that title in 1959. There followed an article by Seymour Tilles in 1963 (then a Harvard Business School lecturer) entitled 'How to evaluate corporate strategy', and a textbook chapter by William Newman of the Columbia University Business School in the same year (see especially 1963: 95-98; the passage noted earlier in the 1951 edition of the Newman textbook might make him the real father of the concept of business strategy in academe, although in private correspondence with this author, Newman has expressed the belief that the overall ideas may have originated in the McKinsey consulting practice, as reflected in the Reilley piece of 1955 (see also McKinsey, 1932, for early suggestions of this thinking)). The Andrews text followed in 1965, the same year that Igor Ansoff published his highly successful book Corporate Strategy, based on many of the same concepts (but more in the spirit of the planning school). 3

2

' Thus Lindgren and Spangberg (1981: 26) refer to this as the 'fit school.'

Undoubtedly encouraged by the fact that in the early years this group trained by far the largest number of doctoral students in business policy. 3 Porter (1981: 610; 1983: 173), a co-author in the 1982 and

16

H. Mintzberg

The Design School Subsequently these ideas embedded themselves in the management literature. Indeed, by the 1980s the Christensen et al. textbook was one of the few left that represented them in their pure form, most others favoring the more elaborated renditions of the planning or positioning schools. 4 Accordingly, and given the impact that this rendition of the design school has had over the years—as well as its clarity and forcefulness of expression—we shall use it as a primary source in the discussion that follows, referring to it as the 'Andrews' text'. We shall draw primarily on the 1982 edition of this textbook, but shall also reference relevant variations in its earlier editions as well as the latest one, published in 1987, although the changes from 1965 to 1987 were relatively minor. 3

Continued. 1987 editions of the Harvard textbook, writes of how the ideas in the original text (the ' L C A G paradigm,' after the names of the four original authors) were 'subsequently]' translated and extended by others, citing in particular A n s o f f s book Corporale Strategy. In fact, Ansoff went to press with his similar ideas in the same year (1965) as Porter's coauthors' originally did, and neither book references any work by authors of the other (although Edmund Learned, the senior author of the first edition of the Harvard textbook, did himself note the similarities in a book published with Sproat one year later: 'Significantly, [ A n s o f f s ] work offers numerous parallels with Harvard thinking that should not be obscured by differences in terminology, definitions, emphasis, and coverage' [1966: 94]). In the Preface to the first edition of the Harvard book, the authors write that the content of (he book 'is the outcome of about ten years of case and course development' (Learned et at., 1965: vii), although in the 1982 version they refer to the core idea having developed in the early 1960s (Christensen, Andrews, and Bower, 1982: viii; co-author Bower is more precise in a 1986 publication: 'The problem of corporate strategy was first phrased as a research question in 1959 when Kenneth Andrews reported his study of the Swiss watch industry in a note and a series of cases' (p. vi) ) Ansoff published a rough version of his approach in article form two years earlier (Ansoff, 1964), although he referred there to an initial unpublished paper of 1958. Note should also be made of comments by Chester Barnard in a 1948 book (p. 169) which seem to be in the spirit of the design school, of the discussion of 'administrative strategy' (pp. 10-18) in the Hardwick and Landuyt textbook by that title in 196-1, and of an article by Gilmorc and Brandenburg in 1962 entitled 'Anatomy of corporate planning'. Although its detail and elaborated steps place this last paper clearly in the planning school, underlying these steps is the same model as that of the design school. (Gilmore and Brandenburg note in a footnote that 'we are indebted to Dr. H . Igor Ansoff for introducing the concept of synergy to us and for his assistance in clarifying a number of steps in our planning framework' (1962: 61).) 4

To this could be added Tregoe and Zimmerman's book Top Management Strategy (1980), although not a textbook. The latest Newman text (Newman, Warren, and McGill, 1987) remains largely in the spirit of this school (in chapter 4 at least), although it also reflects increased attention to the planning school.

173

THE BASIC DESIGN SCHOOL MODEL In his 1957 book, Selznick wrote that: Leadership sets goals, but in doing so takes account of the conditions that have already determined what the organization can do and to some extent what it must do . . . In defining the mission of the organization, leaders must take account of (1) the internal state of the policy: the strivings, inhibitions, and competences that exist within the organization, and (2) the external expectations that determine what must be sought or achieved if the institution is to survive (pp. 62, 67-68). Selznick also coined the term 'distinctive competence' (pp. 43ff.) and noted that 'the task of leadership is not only to make policy but to build it into the organization's social structure' (pp. 62-63), an aspect of the process that came to be called implementation. Andrews summarizes the essence of his model 5 as the intellectual processes of ascertaining what a company might do in terms of environmental opportunity, of deciding what it can do in terms of ability and power, and of bringing these two considerations together in optimal equilibrium. . . . what the executives of a company want to do must also be brought into the strategic decision [as must] what a company should do. Finally, there is 'the implementation of strategy . . . comprised of a series of subactivities which are primarily administrative' (p. 98). The Andrews' text of 1982 splits into two 'books,' the first on 'determining', the second on 'implementing corporate strategy.' Our depiction of the basic design school model (similar to Andrews' own figure of the development of 'economic strategy' (p. 187), but with other elements of his discussion added (see also his figure on p. 99)), is shown here in Figure 1. Consistent with the attention accorded in the text, the model places primary emphasis on the appraisals of the external and internal situations, the former uncovering threats and opportunities in the environment, the latter revealing strengths and weaknesses of the organization. Secondary emphasis is placed on understanding the values of the management, as well as its social responsi5

We should point out that Andrews himself rejects the word 'model' (p. 12), a point we shall return to later.

The Design School

174

17

H. Mintzberg

External Appraisal

Internal Appraisal

Threats and opportunities in environment

Strengths and weaknesses of organization

distinctive competences

key success

Figure. 1.

Basic design school model

bilities. T h e match between these elements leads to the creation of strategies, which are then evaluated, with t h e chosen o n e subsequently implemented. A n d r e w s does not provide extensive discussion of any of these issues (the whole text portion of the 1982 b o o k n u m b e r s 114 pages, the rest of the 838-page book being cases), although others have developed some of these themes m o r e extensively. O n external appraisal, A n d r e w s ' section on 'The nature of the company's e n v i r o n m e n t ' totals 20 pages, but 12 of these come f r o m Michael Porter's work o n Competitive Strategy (1980), literally spliced into t h e A n d r e w s ' text, initially

in the 1982 edition. 6 Likewise, the section on internal appraisal, 'identifying c o r p o r a t e competences and resources,' is brief, touching on a variety of points, such as the difficulty 'for organizations as well as for individuals to know themselves' (p. 183) and the idea that 'individual and u n s u p p o r t e d flashes of strength are not as d e p e n d a b l e as the gradually accumulated productand market-related fruits of experience' (p. 185). This ties back to an important theme in Selznick's " T h e sentences immediately preceding and following this new material (pp. 167 and 179) are identical to those that appeared next to each other in the previous edition of the book (Christensen et al., 1978: 251).

18

H. Mintzberg

The Design School book, that 'commitments to ways of acting and responding are built into the organization,' intrinsic to its very 'character' (1957: 67). Figure 1 shows two other factors considered by this school to be taken into consideration in strategy-making. These are organizational values—the beliefs and preferences of those who formally lead the organization—and social responsibilities—specifically the ethics of the society in which the organization is embedded, at least as perceived by its managers. With the notable exception of Selznick (1957), however, most authors in this school accord values and ethics secondary attention. Andrews offers his two brief chapters well after he has developed the framework dealing with external and internal appraisals. On the actual generation of strategies, little is written in this school, besides an emphasis on this being a 'creative act,' to quote Andrews (1982: 186). Indeed, if this were true, what more could be said, short of trying to use cognitive psychology to probe inside the strategist's mind. In lieu of describing the process, however, a number of writers associated with this school do try to characterize the result, in particular seeking to distinguish some core or dominant element of the strategy (e.g. Tregoe and Zimmerman, 1980: 43 and Ohmae, 1982). This has an important implication, because it replaces strategic within strategy. In effect, rather than considering an organization's intentions as a set of distinct, if coupled, strategies (as tends to be done in the planning and even the positioning schools), it treats them as an integrated concept. Once strategies are created, the next step in the model is to evaluate them and choose the best one. The assumption, in other words— usually implicit—is that several alternative strategies have been produced, and one is to be selected. There is an ambiguity here, however, because even writers such as Andrews, who clearly view strategy formation as a custom-made process of design (for which there is evidence that organizations tend to produce only a single solution; Mintzberg, Raisinghani and Theoret, 1976), assume that alternate strategies (in other words, alternate conceptions of the business) will be evaluated to select a single one (Andrews' text, pp. 105, 109). Tilles published first on this subject in an article entitled 'How to evaluate corporate

175

strategy' (1963). Andrews (pp. 105-108) followed by combining Tilles's list of six criteria with other elements of the model, while Rumelt, a graduate of the Harvard doctoral program in policy, elaborated these ideas in a most succinct (1980) and sophisticated (1979) way, nevertheless retaining the spirit of the design school. Finally, virtually all of the literature of this school makes clear that, once strategy is designed and agreed upon, it is then implemented. We show implementation on the diagram as flaring out from formulation, as if the process draws on a variety of data to narrow down to convergent choice before it diverges again to ensure implementation across the entire organization. Andrews, for example, is clear on the subordinate role of these elements (e.g. p. 543). Interestingly, here also is the one place where he becomes rather specific. He lists 12 steps in the implementation process (backed up by a fair amount of text), a list that seems to encompass any aspect of the strategy process not considered in formulation. 7 The same tends to be true of other design school publications as well.

PREMISES UNDERLYING THE DESIGN SCHOOL Running through all of the literature that we identified with this school are a number of fundamental premises about the process of strategy formation. Some of these tend to be explicit, others implicit, but they are always evident. This is especially so in the Andrews' text, although all are at one time or another qualified in his discussion. But it is the central themes of a work that form the impression left with the reader, not the secondary qualifications. Below we discuss seven basic premises that underlie this school. Premise 1: Strategy formation should be a controlled, conscious process of thought It is not action that receives the greatest attention from the design school so much as reason— strategies formed through a tightly controlled process of conscious human thought. Action 7

In their book implementing Strategy, Hrebiniak and Joyce indeed refer to implementation as 'ail the remaining components' (1984 : 29).

The Design School

176

H.

19

Minlzberg

follows, once the strategies have been fully formulated. This theme runs through all of Andrews' writings, for example in the comment that managers 'know what they are really doing' only it they make strategy as 'deliberate as possible' (1981: 24), or more simply, in reference to his 'thesis' about 'conscious strategy' that should be 'consciously implemented' (Andrews' text, p. 543). But this is perhaps made most clear in his comments that, while the model may be simple, it is not necessarily natural (p. 185)—it must be learned, formally (e.g. Andrews' text p. 6). Andrews is careful to position his view of this process clear of intuition on one side (nonconscious thought) and emergent strategy on the other (where action drives reflection). On intuition, for example, Andrews comments that 'if [strategy] is implicit in the intuition of a strong leader, the organization is likely to be weak and the demands the strategy makes upon it are likely to remain unmet' (pp. 105-106). Likewise, he writes of the need to change 'intuitive skill' into 'conscious skill' (p. 6). And as for his view of emergent strategy, which means pattern in action over time that is not driven by central intention (Mintzberg, 1978; Mintzberg and Waters, 1985), while Andrews presents a definition of strategy in his 1982 text that makes reference to pattern (p. 93), in a new passage in his 1987 text, he makes clear that he means pattern among and across 'goals and policies,' not over time (p. 15; see also Premise 5). In this edition, as in all the others, Andrews clearly means to associate strategy with intentionality. Corporate strategy, for example, 'defines products and markets—and determines the company's course into the almost indefinite future' (1987: xi). In fact, in his 1982 text, Andrews equates emergent strategy with 'erosion:' Strategy will evolve over time, no matter what. It will be affected by the consequences of its implementation. But the elucidation of goals can transcend incrementalism to m a k e it a series of forays and experiments evaluated continuously against stated goals to result in the deliberate a m e n d m e n t of strategy or in the curtailment of strategic erosion (pp. 553-554)

Likewise, Andrews contrasts 'purpose' with 'improvisation,' 'planned progress' with 'drifting' (p. 20). At the end of his text he claims that 'a

strategy may suddenly be rationalized to mean something very different from what was originally intended because of the opportunism which at the beginning of this book we declared the conceptual enemy of strategy' (pp. 828-829)." Premise 2: Responsibility for that control and consciousness must rest with the chief executive officer: that person is THE strategist To the design school, ultimately there is only one strategist, and that is the manager who sits at the apex of the organizational hierarchy. In Hayes' terms, 'this "command-and-control" mentality allocates all major decisions to top management, which imposes them on the organization and monitors them through elaborate planning, budgeting, and control systems' (1985: 117). Again, the origins of this can be found in Selznick: 'it is the function of the leaderstatesman—whether of a nation or a private association—to define the ends of group existence, to design an enterprise distinctively adapted to these ends, and to see that that design becomes a living reality' (1957: 37). Once more Andrews reiterates the point most clearly. On page 3 of his text, he associates the whole field with the 'point of view' of the 'chief executive or general manager'; on page 19, he entitles a section 'the president as architect of organizational purpose' (hence Zand (1981: 125) refers to this school as the 'rational architect' model); and on page 545 he writes that 'the general manager is principally concerned with determining and monitoring the adequacy of strategy, with adapting the firm to changes in its enviroment, and with securing and developing the people needed to carry out the strategy or to help with its constructive revision or evolution.' As we shall soon discuss, in the 1987 text Andrews widens the participation of others in the strategy formulation process, especially in the 'innovative' corporation, but not at the expense of the chief executive's central role. It might be noted that this premise not only relegates other members of the organization to 8

A n d r e w s ' w o r d s a r e r e m i n i s c e n t of t h o s e of Selznick: ' W h e n institutional l e a d e r s h i p fails, it is p e r h a p s m o r e o f t e n by d e f a u l t t h a n by positive e r r o r o r sin. L e a d e r s h i p is lacking w h e n it is n e e d e d ; a n d t h e institution d r i f t s , e x p o s e d t o v a g r a n t p r e s s u r e s , readily i n f l u e n c e d by s h o r t - r u n o p p o r t u n istic t r e n d s ' (1957: 25).

20

H. Mintzberg

The Design School subordinate roles in strategy formation, but it also precludes external actors from the process altogether (except for the directors, who Andrews believes must review strategy (1980b, 1981)). This is most clearly reflected in Andrews' discussion of the ethics discussion in terms of the social responsibility of the managers rather than the sheer power of outsiders. This, in fact, is just one aspect of a larger problem associated with the design school—the relegation of the environment to a minor role, input to strategy formation but not an intrinsic part of the process, to be accounted for and then navigated through but not interacted with. Premise 3: The model of strategy formation must be kept simple and informal The Preface to the Harvard textbook contains a quotation by Andrews that 'the idea of corporate strategy constitutes a simple practitioner's theory, a kind of Everyman's conceptual scheme' (p. 14). Later, he adds in the text that this 'is not a "theory" attended in the rigorous sense by elegance and vigor', nor is it 'really a "model," for the relationships designated by the concept are not quantifiable;' rather it serves as an 'informing idea' (p. 12), or, as Rumelt puts it, 'a set of constructs' (1984: 558). Fundamental to (what we nonetheless prefer to call) the model is the belief that elaboration and formalization will sap it of its essence. This premise in fact goes with the last: one way to ensure that strategy can be controlled in one mind is to keep the process simple. As Andrews writes: 'When the variety of what must be known cannot be reduced by a sharply focused strategy to the capacity of a single mind and when the range of a company's activities spans many industries and technologies, the problems of formulating a coherent strategy begin to get out of hand', (p. 182). This premise, together with the first, forces Andrews to tread a fine line throughout his text, between nonconscious intuition on one side and formal analysis on the other, a position he seems to characterize as 'an act of judgment' (p. 108). This also seems to differentiate clearly the design school from the entrepreneurial school on one side and the other prescriptive schools of planning and positioning on the other (one emphasizing elaboration of the same basic model, with a

177

vengeance, the other, formal analysis of certain of its components). We have already noted Andrews' stand on intuition; on planning he writes that 'this book . . . virtually ignores the mechanisms of planning on the grounds that, detached from strategy, they miss their mark' (p. 10). 9 Of course, if elaboration is the problem, then even theory and research can pose a threat. Thus, Andrews adopts a position in the text that is not just atheoretical but decidedly anti-theory. For example, all of the research on organization structure is dismissed with the comment that 'the literature of organization theory is by itself . . . of very little use in managing a live orgnaization' (p. 554). The introduction to the first edition of the textbook contained the following comment: A considerable body of literature purporting to make general statements about policy-making is in existence. It generally reflects either the unsystematically reported experience of individuals or the logical projection to general management of concepts taken from engineering, economics, psychology, sociology, or mathematics. Neither suffices (Learned el al., 1965: 6). On the latter Andrews added: 'The disciplines cited have much to do with business, but their purposes are not ours. Knowledge generated for one set of ends is not readily applicable to another' (p. 6). The text went on to note that 'research has been for some time under way, but is not yet advanced enough to make more than a modest claim on our attention. . . . the most valid literature for our purpose is not that of general statements but case studies' (p. 6). These comments survived virtually intact to the 1982 edition, the most significant change being that research now 'begins to make a claim on our attention' (p. 6). Moreover, added is the statement that 'the books referred to [in the footnotes] comprise a relevant but incidental source of knowledge'. (It is, in fact, instructive to consider these references. In all, there are 39 of them to theoretical works in the footnotes of the 1982 edition of the text, of which 31 are by * Interestingly, in so dismissing planning at this point, Andrews resurrects intuition: All the knowledge, professional attitudes and analytical and administrative skills in the world cannot fully replace the intuitive genius of some of the natural entrepreneurs you will encounter in this book' (p. 10).

The Design School

178

H.

Mintzberg

faculty m e m b e r s or doctoral students at the H a r v a r d Business School.) 1 0 It might be n o t e d that this t r e a t m e n t of theory extends even to the work of a co-author whose ideas a r e contained in t h e same b o o k . A s noted earlier, Michael Porter's views (centrally located in the positioning school) were literally spliced into that portion of t h e text on assessing the environment. T h e text on either side of it, however, questions assumptions f u n d a m e n t a l to Porter's work (1980, 1985). T o take just o n e obvious example: The choice of objectives and the formulation of policy to guide action in the attainment of objectives depend upon many variables unique to a given organization and situation. It is not possible to make useful generalizations about the nature of these variables or to classify their possible combinations in all situations (Andrews' text, p. 5). Thus the P o r t e r graft does not take, a n d , m o r e importantly, o t h e r theory has not been allowed to infíltrate t h e model.

Premise 4: Strategies should be unique: the best ones result from a process of creative design A s suggested above, it is the specific situation that m a t t e r s , not any system of general variables. It t h e r e f o r e follows that strategies have to be tailored to t h e individual case: 'In each c o m p a n y , the way in which distinctive c o m p e t e n c e , organizational resources, and organizational values are combined is or should be unique' ( A n d r e w s ' text, p. 187). Stronger words a r e offered o n page 109: 'sometimes the companies of an industry run like sheep all in o n e direction' although imitation 'does not constitute t h e assurance of soundness.' 10

21

For the record, these are all the references found in the text portion of the book, as well as the Preface. References to cases, or references within the cases themselves, were not included. References were counted rather than sources, so that in a few cases the same source was referenced more than once. A source was considered to emanate from Harvard if at least one author was on the staff or was a doctoral student there. Lest this criticism be extended unfairly to all of the co-authors, or even the claims about the literature itself, it should be noted that Edward P. Learned, the senior author of the original edition of the textbook, in his book published together with Sproat in 1966, and entitled Organization Theory and Policy, contained perhaps half the amount of text, yet twice the number of references, only a small proportion of those emanating from the Harvard Business School. The 1987 edition of the Andrews' book contains 24 such references, 18 from Harvard.

A s a result of this premise, the design school says little a b o u t the content of strategies themselves, but instead concentrates o n the process by which they should be developed. A n d that process above all should be a 'creative act' ( A n d r e w s ' text, p. 186), to build on distinctive c o m p e t e n c e . Writing in support of the positioning school, H o f e r and Schendel refer to what we are calling the design school as the 'situational philosophy' (1978: 203), at o n e e x t r e m e of the field, in contrast with the 'principles of m a n a g e m e n t ' approach at t h e other.

Premise 5: Strategies emerge from this design process fully formulated A s n o t e d in passages cited above, this school offers little room to incrementalist views or e m e r g e n t strategies. It is the big picture that results f r o m the process—the grand strategy, an overall concept of the business. This is no Darwinian view of strategy formation but the Biblical version, with strategy the final conception! T h e r e is, in o t h e r words, a strong implication that strategy as perspective appears at a point in time, fully f o r m u l a t e d , ready to be implemented. H o w else could A n d r e w s have assumed that the process reduces to 'choice,' a word he uses o f t e n , referring also in the Preface to the 1987 edition to 'this decision,' 'the strategic decision,' 'the entrepreneurial decision . . . [once] identified' (p. xiv, italics a d d e d ) . In o t h e r words, the assumption is that the strategist is able to line u p alternative strategies b e f o r e him to be evaluated so that o n e can b e definitively chosen.

Premise 6: These strategies should be explicit and, if possible, articulated, which also favors their being kept simple While A n d r e w s accepts various reasons for not articulating strategy (such as confidentiality or difficulty of updating (see pp. 9 6 - 9 7 ) ) , he clearly views these as necessary evils. In c o m m o n with virtually all the writers of this school, he believes that strategies should at least be explicit to those w h o m a k e them a n d , if at all possible, articulated so that o t h e r s in the organization can understand them: ' T h e unstated strategy cannot be tested or contested and is likely t h e r e f o r e to be weak . . . . A strategy must be explicit to be effective and

22

H. Mintzberg

The Design specific enough to require some action and exclude others' (pp. 105-106). If strategies are to be so articulated, it also follows that they have to be kept rather simple: to the point, easily stated to be easily understood. 'Simplicity is the essence of good art,' writes Andrews, 'a conception of strategy brings simplicity to complex organizations' (p. 554). To him, strategy helps organizations make better decisions 'by reducing the world of detail to be considered to those central aspects of external environment and internal resources that affect the company and bear on the definition of its business' (p. 835). Of particular interest to Andrews is the role of outside directors in strategy formation: he believes that they must be actively involved at least in the evaluation and review processes. But this can only happen if strategies are explicit, so that they can be articulated to the directors, and simple, so that they can be understood by people who have only brief time to devote to the organization: 'The power of strategy as a simplifying concept enabling independent directors to know the business (in a sense) without being in the business will one day be more widely tested at board level' (p. 834). Premise 7: Finally, only after these unique, fullblown, explicit, and simple strategies are fully formulated can they then be implemented We have already noted the sharp distinction this school makes between the formulation of strategies on one hand and their implementation on the other. Consistent with classical notions or rationality—diagnosis, prescription, then action— the design school clearly separates thinking from acting (see Bourgeois and Brodwin on their 'change model.' 1984: 246). (It is of interest that the word used is 'implement,' not 'achieve,' the assumption being that given proper implementation, achievement is a foregone conclusion.) Central to this distinction is the associated premise that structure must follow strategy. As Andrews puts it: 'Corporate strategy must dominate the design of organizational structure and processes' (p. 543). The assumption seems to be that each time a new strategy is formulated, the state of structure and everything else organizational is to be considered anew: 'Until we know the strategy we cannot begin to specify the

School

179

appropriate structure,' writes Andrews (p. 551), as if the existing structure does not bear on the new strategy. Andrews' qualifications While these seven premises are clearly evident in Andrews' text, as noted earlier he does qualify virtually all of them, tucking into his text here and there either nuances that soften their character or else comments that acknowledge the unfortunate reality, as compared with the preferred ideal. A number of them have also been added to the 1987 text, for example: False hope, oversimplification, and naivete, as well as zest for power, have often led . . . to the assumption that the chief executive officer conceives strategy single-mindedly, talks the board of directors into pro forma approval, announces it as fixed policy, and expects it to be promptly executed by subordinates under conventional command and control procedures (p. 82).

Andrews rejects this view for all but 'the entrepreneurial startup stage,' while we see it as a not unreasonable caricature of his own text! In his 1982 text Andrews wrote that 'strategy formulation is itself a process of organization, rather than the masterly conception of a single mind' (p. 827), in other words, at least in 'technically or otherwise complex organizations,' 'an activity widely shared in the hierarchy of management' (p. 828). In 1987 he even contrasted 'constructive engagement' with 'archaic notions of authority, responsibility, hierarchy, status, and centralized decision making ' (p. 86). Yet his major justification for this seemed to be the generation of commitment to the strategy emanating from the apex of the organization's hierarchy. 'Commitment . . . is a simple reason for the involvement of whatever number of people is required to make a success of whatever is intended' (p. 120; see also pp. 55-56, 59). Andrews also accepted that 'in real life the processes of formulation and implementation are intertwined' (exist in a 'reciprocal relationship' (1987: 853)), that 'the formulation of strategy is not finished when implementation begins' even though the cases in the book have been arranged around these two topics 'for the sake of orderly presentation' (1982 text, p. 541). He also acknowledges that 'we should look first at the logical

T h e Design School

180

23

H. Mintzberg

proposition that structure should follow strategy in order to cope with the organizational reality that strategy also follows structure' (p. 99), in other words, that 'the structure and processes in place will in fact affect the strategy' (p. 552). In the 1982 text Andrews acknowledges emergent strategy as well (e.g. p. 553), going further in the 1987 text to discuss 'a balance between focus and flexibility, between a sense of direction and responsiveness to changing opportunities .... Corporate strategy need not be a straitjacket. Room for variation, extension, and innovation must be provided' (p. 84). Yet he is careful to avoid association with what he calls extreme incrementalism, understood as reactive improvisation, muddling through, or following one's nose' (p. 83). And elsewhere in the 1987 edition, new sections also make clear the continuing commitment to deliberateness: 'It is not only possible but also essential to plot a course into a future that cannot be foretold . . .' (p. xiii, see also the comments on Japanese management on pp. vi-vii).

themselves as, of course, is natural if he is not to undermine his own position. The fact is that the premises of the design school combine to form their own tightly integrated strategy—the whole thing really is a 'model' after all. By ultimately remaining true to its premises, Andrews positions the design school in its own niche, distinguishing it particularly from the planning and positioning schools on one side, which by elaborating the model shift it from the realm of judgement to that of analysis, and the entrepreneurial school on the other, which by mystifying the whole process locks it into the inaccessible (and unteachable) realm of intuition. The outstanding question is how large is that niche: how much of the viable strategic behavior of organizations, whether for purposes of description or prescription, is it reasonably able to encompass?

Adding all these qualifications together, one can easily come up with quite a different model of strategy formation. But of course, no reader can doubt for which model Andrews stands, and not just 'for the sake of orderly presentation'!" One obvious question that arises from a number of these qualifications is why practice seems to differ from the prescribed model, at least some of the time. Andrews does not address the question—research of course does, but he precludes the results of research from his text. Nor does he pursue these qualifications at any length. Most are presented as asides or afterthoughts, while his real commitment remains to the premises

The writings of the design school can be critiqued on a number of levels. In perhaps the most general sense, the school has denied itself the chance to adapt. Research results that have put parts of it under suspicion were not considered; indeed, there was no reason to, if the model could not be elaborated upon. This problem is well illustrated by the book by Norman Berg (1984), also of Harvard, who provided, two decades after the original Andrews' text, almost the same chapter headings (save an application at the end to the divisionalized firm), with the same points made in the same ways, including even the same qualifications (see for example, pp. 28-33).

11 Or 'temporary conceptual convenience', as Andrews put it in a memo to his colleagues in response to comments this author made in a talk given at the Harvard Business School in 1976:

A s Andrews so keenly argued, the source of data and inspiration for the model was to be the concrete case—the description of one firm in one situation. Ironically, however, his beliefs about theory kept him from using even this rich data base to build better theory. Certainly after 1965, if not before, if there was a relationship, then the model had to drive the writing of cases (for example, if you want to find out how strategy is made, go interview the strategist), not vice-versa, since the model has barely changed since then.

whatever our preferences, let us avoid the allegation that the central conceptualization of Business Policy as a field separates formulation from implementation for anything except temporary conceptual convenience. The interrelationships of a complex interdependency cannot be intelligently discussed all at once. What is being related can usefully be stopped and examined before reinstalling it conceptually in a dynamic process (Andrews, 1976: 4). In a personal reply to Andrews, I concluded that with respect to this dichotomy, 'Aside from headings, we may be doing the same thing. The question is: do headings matter?!' We shall return to this later in our discussion.

CRITIQUE OF THE DESIGN SCHOOL

Of course, its supporters might contend that the model was good enough in 1965 and remains

24

H. Mintzberg The Design School

so today. We shall not promote this contention, however, its contribution has been profound, as we shall point out later, but it has never been good enough, indeed no one model can be. It describes but one approach to strategy formation, and even that one sometimes exhibits a level of generality and a tone of inevitability that seems overly simple in places and, at times, dogmatic. Indeed one sometimes wonders whether, like the testament of a religious prophet, it comprises a set of profound truths subtly buried in simple prescriptions, or else if 'the whole idea is just one big fat platitude' (as the Harvard authors, to their credit, quote a critical company executive in the final case of their 1982 edition (p. 821)). In the Preface to this 1982 edition the authors wrote that 'our teaching focus then [when the core idea of the book was developed in the early 1960s], as today, emphasizes the determination of corporate strategy (Book One) and the implementation of corporate strategy (Book Two). This format has stood the test of time' (p. viii). But it has not, even in their own school, as attested to by recent difficulties and changes in the Harvard MBA Policy course. That course still splits into two on these lines, but formulation has been moved into the first year of the program, and refashioned in the spirit of the positioning school as Porter articulates it, while the second half on implementation remains (at the time of this writing) in a state of flux after several years of searching for a new formula. Strategy can locate a system in a niche, but in so doing narrow its perspective. This is what seems to have happened to the design school itself with respect to strategy formation. As noted, the premises of the model deny certain important aspects of strategy formation, including incrementalism and emergent strategy, the influence of existing structure on strategy, and the full participation of actors other than the chief executive. We wish to elaborate on these shortcomings in this critique, to indicate how they narrow its perspectives to certain contexts (as indeed, do the premises of most of the other schools). One point should be made before we probe into the details. Andrews might well argue that we are interpreting his writings too literally, that it is unfair to take apart a model—a specified sequence of prescriptive steps—when all he meant

181

to propose was a framework. Leaving aside the ambiguities in Andrews' own writings on this point, 12 even leaving aside the fact that an author must be interpreted by his central thrust rather than his secondary qualifications, in the most fundamental sense the two interpretations are not really different. Both are underlaid by some powerful assumptions, a critique of which will underlie our own argument. These concern the central role of conscious thought in strategy formation, that such thought must not only take precedence over action but must precede it in time, and correspondingly that the organization must separate the work of thinkers from that of doers. In our view, these assumptions often prove false, both descriptively and prescriptively. In other words, often not only don't organizations do these things, but by all accounts, they should not. This suggests that while the design school framework, if not the model, may never go out of date, it can easily go out of context. We develop our critique by considering specific aspects of the model—first, the belief about the need for a conscious assessment of strengths and weaknesses, then the assumed sequence of strategy followed by structure, after that the premise that all strategies should be made explicit, and finally the assumed dichotomy between formulation and implementation. We shall conclude the critique by considering the relationship between the design school model and case study teaching, before closing the paper with a delineation of the contexts we believe to be most appropriate for this school. The reader is asked to bear in mind that although the other prescriptive schools of planning and positioning have broken with certain of the premises of the design school (notably in keeping the process simple and strategies unique, to a lesser extent also in introducing the planner and analyst into the process alongside the chief executive), the fact that they have accepted the most basic ones renders most of the following a critique of those schools as well.

12 Or even his support for the specific model: for example, 'the text is dispersed throughout the book so as to permit a step-by-step consideration of what is involved in corporate strategy and in the subactivities required for its formulation and implementation* (p. 11; italics added).

The Design School 182

H.

25

Mintzberg

Assessment of strengths and weaknesses: thinking vs. learning Our critique of the design school revolves around one central theme: its promotion of thought independent of action, strategy formation above all as a process of conception, rather than as one of learning. We can see this most clearly in a fundamental step in the formulation process, the assessment of strengths and weaknesses. How does an organization know its strengths and weaknesses? On this, the design school is quite clear—by consideration, assessment, judgement supported by analysis, in other words, by conscious thought. One gets the image of executives sitting around a table discussing the strengths, weaknesses, and distinctive competences of an organization, much as do students in a case study class. Having decided what these are, they are then ready to design strategies. Some writers offer specific lists of potential strengths and weaknesses for all organizations, while others, though not offering such lists, do assume that types of strengths and weaknesses exist in general. Andrews, on the other hand, would only associate strengths and weaknesses with a particular organization—its competences are distinctive to itself. But does even that specify them precisely enough? In his article on 'strategic capability', Lenz (1980) critiques the use of an 'organizational frame of reference'—usually based on some abstract ideal or a comparison with the situation of the past—with an external frame of reference. In other words, internal capability has to be assessed with respect to external context. But as we have already mentioned, there is a tendency in the design school to slight the environment in favor of a focus on the organization itself (which may manifest itself in a tendency to overstate strengths and under emphasize weaknesses (e.g. Katz, 1970: 350; see also Dimma, 1985: 251). But the problem of assessing strengths and weaknesses may go deeper still. Might competences not also be distinctive to time, even distinctive to application (Radosevich, 1974: 360; see also Hofer and Schendel, 1978: 148-150)? And can any organization really be sure of its strengths before it tests them, empirically? The point we wish to make came out most clearly, if inadvertently, in a study carried out at Harvard by Howard Stevenson (1976), published

under the title 'Defining corporate strengths and weaknesses'. Starting out with a conventional design school view of these (see p. 53), Stevenson asked managers to assess their companies' strengths and weaknesses in general. Overall, 'the results of the study brought into serious question the value of formal assessment approaches.' In general, 'few members of management agreed precisely on the strengths and weaknesses exhibited by their companies' (p. 55). The overall impression left by this study is that the detached assessment of strengths and weaknesses may be unreliable, all bound up with aspirations, biases, and hopes. In fact, Stevenson's managers seemed to understand the problem, their 'most common single complaint' being that strengths and weaknesses have 'to be defined in the context of a problem,' or to quote one of his subjects, 'As I see it, the only real value in making an appraisal of the organization's capabilities comes in the light of a specific deal— the rest of the time it is just an academic exercise' (p. 65). Every strategic change involves some new experience, a step into the unknown, the taking of some kind of risk. Therefore, no organization can ever be sure in advance whether an established competence will prove to be a strength or a weakness. In its retail diversification efforts, a supermarket chain we studied (Mintzberg and Waters, 1982) was surprised to learn that discount stores, which seemed so compatible with its food store operations, did not work out well, while fast-food restaurants, ostensibly so different, did. The similarities of the discount store business— how products were displayed, moved about by customers, and checked out, etc.—were apparently overwhelmed by the subtle but different characteristics of merchandising—styling, obsolescence, etc. On the other hand, the restaurants may have looked very different, but they moved simple, basic, perishable, commoditylike products through an efficient chain of distribution much like the supermarket business did. The point we wish to emphasize is: how could the firm have known ahead of time? The discovery of what business it was to be in could not be undertaken on paper, but had to benefit from the results of testing and experience. (And the conclusion suggested from such experiences is that strengths generally turn out to be far narrower than expected and weaknesses,

26

H. Mintzberg

The Design consequently, far broader (see Mintzberg and McHugh, 1985)). Nowhere does this come through more clearly in practice than in attempts at related diversification by acquisition. Obviously no organization can undertake such activity without a prior assessment of its strengths and weaknesses. Yet the vast majority of experiences reported in the popular press and published research suggests that related diversification is above all a learning process, in which the acquiring firm has to make a number of mistakes until it gradually learns what works for it, if it ever does (see Miles, 1982; also Quinn, 1980: 28). And in the writings of academe, the problem is perhaps best illustrated by Levitt's (1960) popular 'marketing myopia' conception, that firms should define themselves by broad mission rather than narrow product or technology (e.g. transportation instead of railroad). The idea was enticing, but in many applications too easy, a cerebral exercise that could detach managers from the realities of the businesses they managed. What, in a few words on a piece of paper, would enable railroads to fly airplanes? Levitt, a marketing professor but here arguing in the spirit of the design school, wrote that 'once it genuinely thinks of its business as taking care of people's transportation needs, nothing can stop it from creating its own extravagantly profitable growth' (p. 33; our italics). Nothing except the limitations of its own distinctive competences! Structure follows strategy . . . as the left foot follows the right While the design school tends to promote the dictum, first articulated by Chandler (1962), that structure should follow strategy and be determined by it, in fact its model also accepts the opposite. Since the assessment of organizational strengths and weaknesses is an intrinsic part of the model, a basic input to strategy formulation, and since structure is a key component of this, housing the organization's capabilities, then structure must play a major role in determining strategy too, by constraining and conditioning it as well as guiding it. While this may be an obvious point, hardly disputed even within the design school, it does have a broader implication, an important one in our critique of this school's model of strategy

School

183

formation. No ongoing organization ever wipes the slate clean when it changes its strategy. The past counts, just as does the environment, and the structure is a significant part of that past. Claiming that strategy must take precedence over structure amounts to claiming that strategy must take precedence over the established capabilities of the organization, clearly an untenable proposition. By overemphasizing strategy, and the ability of the stategist to act rather freely, the design school slights, not just the environment, but also the organization itself. Structure may be malleable, but it cannot be altered at will just because a leader has conceived a new strategy. Many organizations have come to grief over just such a belief. We conclude, therefore, that structure follows strategy as the left foot follows the right in walking. In effect, strategy and structure both support the organization. None takes precedence; each always precedes the other, and follows it, except when they move together, as the organization jumps to a new position. Strategy formation is an integrated system, not an arbitrary sequence. Making strategy explicit: promoting inflexibility Once strategies have been created, via the conscious assessment of strengths and weaknesses among other things, then the model calls for their articulation. While recognizing some reasons for not making strategy explicit, this school generally considers an unwillingness to articulate strategy as evidence of fuzzy thinking, or else of political motive. But there are other, often more important, reasons not to articulate strategy, which strike at the basic assumptions of the design school. The reasons generally given for the need to articulate strategy are, first, that only an explicit strategy can be discussed, investigated, and debated (e.g. Andrews, 1981: 24); second, that only by making strategy explicit can it serve its prime function of knitting people together to 'provide coherence to organizational action' (Rumelt, 1980: 380); and third, that an articulated strategy can generate support—can rally the troops, so to speak, and reassure outside influencers. These all sound like excellent reasons for articulating strategy. And they are—so long as all the conditions are right. The most important

The Design School 184

H.

27

Mintzberg

of these is that the strategist is sure—knows where he or she wants to go, and has few serious doubts about the viability of that direction. In other words, the design school implicitly assumes conditions of stability or predictability. But organizations have to cope with conditions of uncertainty too. How can Andrews' company come 'to grips with a changing environment' when its 'strategy is [already] known' (1981: 24)? And how can its managers promote the necessary changes when its own board of directors uses that articulated strategy 'to prevent the company from straying off its strategic course'(1980b: 32)? Our point is that organizations must function not only with strategy, but also during periods of the formulation and reformulation of strategy, which cannot happen instantaneously. 'It is virtually impossible for a manager to orchestrate all internal decisions, external environmental events, behavioral and power relationships, technical and informational needs, and actions of intelligent opponents so that they come together at a precise moment' (Quinn, 1978: 17). Indeed sometimes organizations also need to function during periods of unpredictability,' when they cannot possibly hope to articulate any viable strategy. The danger during such periods is not the lack of explicit strategy but exactly the opposite—'premature closure,' the reifìcation of speculative tendencies into firm commitments. When strategists are not sure, they had better hot articulate strategies, for all the reasons given above. Moreover, even when it makes sense to articulate strategies, because they appear to be viable well into the future, the dangers of doing so must still be recognized. Explicit strategies, as implied in the reasons for wanting them, are blinders designed to focus direction and so to block out peripheral vision. Thus, they can impede strategic change when it does become necessary: to put this another way, a danger in articulating strategy is that while strategists may be sure for now, they can never be sure forever. The more clearly articulated the strategy, the more deeply imbedded it becomes in the habits of the organization as well as the minds of its strategists. There is, in fact, evidence from the laboratories of cognitive psychology that the explication of a strategy—even having someone articulate what he or she is about to do anyway— locks it in, breeding a resistance to later change (Kiesler, 1971).

Another reason not to articulate strategy is that pronouncements of it, often necessarily superficial, can engender a false sense of understanding. Andrews argues that 'a conception of strategy brings simplicity to complex organizations' (p. 554). True enough. But at what price? The potential danger of a little knowledge needs to be recognized: the possible trivialization and distortion of the subtle needs of a complex organization. As Wrapp has noted, sometimes it is impossible to articulate direction 'clearly enough so that everyone in the organization understands what they mean' (1967: 95). And the problems can magnify when outsiders are involved in the process, even board members. Perhaps that is why Andrews finds such strong managerial resistance to the inclusion of outside board members in strategy-making. To summarize, the problems of making strategy explicit essentially bring us back to the need to view strategy formation as a learning process, at least in some contexts. Sure strategies must often be made explicit, for purposes of investigation, coordination, and support. The questions are: when? and how? and when not? There is undoubtedly a need for closure at certain points in an organization's history, moments when the process of strategy formation must be suspended temporarily to articulate clear strategies. But this need should not lead us to believe that it is natural for strategies to appear fully developed all of a sudden, nor should it allow us to ignore the periods during which strategies must evolve. Separation of formulation from implementation: detaching thinking from acting

The formulation-implementation dichotomy is central to the design school—whether taken as a tight model or a loose framework. But is the distinction a valid one for conceptual and analytical, even pedagogical, purposes? In other words, should people concerned with strategy (including students learning about it) think, let alone behave, in terms of formulation and implementation? How can anyone really question this distinction, or even the assumption that formulation must precede implementation? After all, this is just another version of the basic form of rationality that underlies western thinking—in its simplest form, that to act you must first know what you want to accomplish. Think first, then do.

28

H. Mintzberg The Design School

The organizational form that corresponds to this dichotomy is the classical hierarchy, what we prefer to call 'machine bureaucracy' (Mintzberg, 1979). It above all emphasizes the distinction between the few people on top who are allowed to think and the many below who are supposed to act. Machine bureaucracy is common in mass production and the mass provision of services. It dominates thinking in the consulting profession (most of whose techniques promote this form of structure), in big business (outside of high technology), and in government, including the military. In his article on the dysfunctions of traditional military organization, Feld (1959), noted the sharp distinction that is made between the officers in the rear, who have the power to formulate plans and direct their execution, and the troops on the front, who, despite their first-hand experience, can only implement the plans given them. These 'organizations place a higher value on the exercise of reason than on the acquisition of experience, and endow officers engaged in the first activity with authority over those occupied by the second' (p. 15). This 'is based on the assumption that their position serves to keep them informed about what is happening to the army as a whole . . . [which] is supported by the hierarchical structure of military organization which establishes in specific detail the stages and the direction of the flow of information' (p. 22). This, in fact, is the assumption fundamental to the formulation-implementation dichotomy: that data can be aggregated and transmitted up the hierarchy without significant loss or distortion. It is an assumption that fails often, destroying carefully formulated strategies in the process. To use a quotation Feld (p. 15) meant for the military, in how many contemporary organizations do 'the conditions most favorable to rational activity, calm and detachment, stand in direct antithesis to the confusion and involvement' of the factory floor, the salesman's call, the government clerk's service? In how many does detached formulation render the organization ineffective? In how many is critical information ignored because it is deemed 'tactical'? Speaking from Japan, Ohmae goes so far as to suggest that 'separation of muscle from brain may well be a root cause of the vicious cycle of the decline in productivity and loss of international competitiveness in which U.S. industry seems to be caught' (1982: 226).

185

In recent years there has been a spate of books and articles on implementation (such as Hrebiniak and Joyce, 1984). Noting that few intended strategies are successfully realized—the figure cited by Fortune writer Walter Kiechel (1984: 8) that 'fewer than 10 percent of American corporations' implement their intended strategies, was deemed 'wildly inflated' by Tom Peters!— they call for more attention to the implementation process. 'Manage culture,' executives are advised, or 'pay more attention to your control systems.' If one side of the formulation-implementation dichotomy does not work, then effort must be invested in the other. Majone and Wildavsky point out that to study implementation is to raise 'the most basic question about the relation between thought and action: How can ideas manifest themselves in a world of behavior?'(1978:103). As they characterize the 'planning-and-control model of implementation,' which sounds to us like the design school model in the public sector, 'good implementation is the irresistible unfolding of a tautology,' or to translate their terms into ours, the transformation of intended strategy into realized strategy through a 'suitable . . . "production function" ', meaning goals, plans and controls, 'and—to take care of the human side of the equation—incentives and indoctrination' (p. 106); 'the perfectly pre-formed policy idea . . . only requires execution, and the only problems it raises are ones of control' (p. 114). All that would be fine were only the world cooperative. Unfortunately, often it is not, in many cases for good reason, whether the resistance to the intended strategy comes from the environment in which it is to be implemented, the organization that is supposed to do the implementing, or even from the strategy itself. Sometimes the 'implementors' who make up the rest of the organization are perfectly willing to proceed as directed from the center, but the environment simply renders the strategy a failure. It may change unpredictably, so that the intended strategy becomes useless, or it may remain so unstable that no specific strategy can be useful. Despite implications to the contrary, the external environment is not some kind of pear to be plucked from the tree of external appraisal, but a major and sometimes unpredictable force to be reckoned with. In other cases it is not the environment but the implementors within the organization who

The Design School 186

H.

29

Mintzberg

resist. They may, of course, be narrow-minded bureaucrats too wedded to their traditional ways to know a good new strategy when they see one, or small-minded ones who do not understand the new strategy, or bloody-minded ones who prefer to go their own way (e.g. Thoenig and Friedberg, 1976 and Scheff, 1961). But sometimes they are right-minded people who do what they do to serve the organization despite its leadership. They may resist implementation because they know the intended strategies to be unfeasible— that the organization will not be capable of realizing them or, once realized, they will fail in an unsuitable external environment. Implementational failure can also occur without inhospitable environments and resistant organizations. The problem can lie in the strategy itself; indeed, in part at least, it almost always does. For one thing, no intended strategy can ever be so precisely defined that it covers every eventuality. Moreover, while the formulators may be few, the implementors are typically many, functioning at different levels and in different units and places (Rein and Rabinovitz, 1979: 327-328), each with their own values and interpretations. They are not robots, nor are the systems that control them airtight. The inevitable result is some slipping between formulation and implementation. 'Slippage' is a term used in the public sector to mean that strategic intentions get distorted on their way to implementation; 'drift' is another term used there for realized strategies that differ from intended ones, but within their context (Majone and Wildavsky, 1978: 105; Kress, Koehler and Springer, 1980; Lipsky, 1978). Here, however, we would like to take a position beyond both concepts. Certainly much formulation is ill-conceived, just as much implementation is badly executed. But often the fundamental difficulty lies not in either side, but in conceiving a distinction between formulation and implementation in the first place. Behind the premise of the formulationimplementation dichotomy lies a set of very ambitious assumptions: that environments can always be known, currently and for a period well into the future, in one central place, at least by the capable strategists there. To state this more formally, by distinguishing formulation from implementation, the design school draws itself into two questionable assumptions in particular:

first, that the formulator can be fully, or at least sufficiently, informed to formulate viable strategies, and second that the environment is sufficiently stable, or at least predictable, to ensure that the strategies formulated will remain viable after implementation. Under some conditions at least, one or the other of these assumptions proves false. In an unstable environment, or one too complex to be comprehended in a single brain, thé dichotomy has to be collapsed, in one of two ways. If the necessary information can be comprehended in one brain, but the environment is unpredictable—or perhaps more commonly, takes time to figure out after an unexpected shift— then the 'formulator' may have to 'implement' him or herself. In other words, thinking and action must proceed in tandem, closely associated: the thinker exercises close control over the actions. The leader—here Andrews' one strategist (or a small group)—develops some preliminary ideas, tries them out tentatively, modifies them, tries again, and continues until a viable strategy emerges, much as Quinn (1980) described the process, or continues to act even if one does not. Such close control of a leader over both formulation and implementation is characteristic of the entrepreneurial mode of strategy-making, where power is highly centralized in a flexible organization (Mintzberg, 1973). But, as noted earlier, that mode, because it is rooted in the vagaries of intuition, tends to be dismissed by the design school. True, it may sometimes be 'opportunistic,' as Andrews claims, but such opportunism can be necessary, perhaps in and of itself or more productively perhaps, as a means to experiment and learn. Pascale (1984) provides a marvelous example of the latter in his description of how the Honda Motor Company executives in the United States backed their way into their highly successful motorcycle strategy of the 1960s, in contrast to the Boston Consulting Group's (1975) inference of that strategy as brilliantly deliberate. (In the 1987 textbook (p. vi), Andrews comments that 'Japanese management appears to be more truly strategic than improvisatory.' Perhaps for him, as for the Boston Consulting Group, believing is seeing.) Where there is too much information to be comprehended in one brain—for example, in organizations dependent on a great deal of sophisticated expertise, as in high-technology

30

H. Mintzberg

The Design School

187

firms, hospitals, and universities—then the strategy may have to be worked out on a collective basis. Here, then, the dichotomy collapses in the other direction: the 'implementors' become the 'formulators' (Hardy et al., 1984; Mintzberg and McHugh, 1985). As Lipsky puts it, implementation is 'turned on [its] head' (1978: 397), and actions in good part determine thoughts, so that strategies also emerge. Both situations—'formulators' implementing and 'implementors' formulating—amount to the same thing in one important respect: the organizations are learning. Andrews' great mistake was dismissing organizational learning by considering it opportunism. Even though he recognized the intertwining of formulation and implementation in practice, his making of the distinction conceptually led him to underestimate the important role of such learning, individually and especially collectively, over time, in strategy formation. More generally, the design school, by implicitly assuming that strategic learning somehow takes place in one head for a limited period of time and then stops, so that strategies can be articulated and implementation can begin, denied processes that have often proved critical to the creation of novel and effective strategies.

McHugh (1985) on a 'grass roots model,' and Mintzberg (1987) on 'crafting strategy.') And then, perhaps most common, are a whole range of possibilities in between—'implementation as evolution,' as Majone and Wildavsky (1978) put it—where there is thought, then there is action, this produces learning which alters thought, followed by adjustments to action, and so on. Intended strategies exist, but realized strategies have emergent as well as deliberate characteristics. Here words like 'formulation' and 'implementation' should be used only with caution, as should the design school model of strategy formation. To conclude this critique, this seemingly innocent model—for Andrews, just an 'informing idea'—in fact contains some ambitious assumptions about the capabilities of organizations and their leaders, assumptions that break down in whole or in good part under many common circumstances.

Out of all this discussion comes a whole range of possible relationships between thought and action. There are times when thought does, and should, precede action and guide it primarily, so that, despite some inevitable slippage, the dichotomy between formulation and implementation does hold up, more or less. In other words, while it may be true that 'literal implementation is literally impossible' (Majone and Wildavsky, 1978: 116), sometimes what is achieved is close enough. And here is where we might expect viable application of the design school model. Other times, however, especially during or immediately after a major unexpected shift in the environment, thought must be so bound up with action in an interactive and continuous process that 'learning' becomes a better label, and concept, for what happens then is 'formulation-implementation'. The organization may be groping its way toward a new strategy, or may simply be coping until things settle down so that it can then do so. (For models of strategymaking as a learning process, see Quinn (1980) on 'logical incrementalism,' Burgelman (1983) on 'corporate entrepreneurship,' Mintzberg and

We believe that the relationship between the design school model of strategy formation and the traditional method of case study teaching may help to explain why there has been so much reluctance in certain quarters to adapt the model to other views of strategy-making. The design school model matches perfectly the pedagogical requirements of the case study method, as Andrews and his colleagues note repeatedly. The students are handed a document of 20 or so pages that contains all the available information on the organization in question. They study it the evening before class (alongside the other cases they must prepare for that day), and then appear all ready to argue what it is that General Motors or the John F. Kennedy High School should do. Bear in mind that time is short: the external environment must be assessed, distinctive competences identified, alternate strategies proposed, and these evaluated, all before class is dismissed in 80 minutes. Two days later it's on to Xerox or Texas Instruments. Here is how the process is described by the senior author of the Harvard textbook:

THE DESIGN SCHOOL AND THE CASE STUDY METHOD

T h e Design School

188

31

H. Mintzberg

how d o those of us interested in management education strive to contribute to the development of future general mangers? W e d o this first by disciplined classroom drill with the concept of strategy. Drill in the formal and analytic—what is the current strategy of the firm? What are its strengths and weaknesses? W h e r e , in the firm's perceived industry, are profit and service opportunity? A n d , h o w can those corporate capacities and industry opportunities be effectively related? Moreover, this analytic classroom process focuses attention on a key administration skill— the process of selecting and ordering data so that management asks the critical questions appropriate to a particular situation (Christensen, in Christensen et al., 1982: i x - x ) .

But how can a student who has read a short resumé of a company possibly know these things? How can words and numbers on paper possibly substitute for the intimate knowledge of a complex organization? Can the 'critical questions' really be asked through the process of 'selecting and ordering' this kind of data? And what effect does this 'drill in the formal and analytic' have on the students when they finally do enter the executive suite? Given the requirements of case study teaching, how else can the faculty proceed but to keep the model simple, especially to presume that organizations can be quickly and easily understood, and to assume the necessity for fully developed and explicit but nonetheless simple strategies. And even if it is accepted that formulation and implementation are intertwined in practice, what good is that in the classroom where formulation (thinking) is possible while implementation (acting) is not? 1 3 Of course, proponents of this school might argue that this is a small price to pay for bringing reality into the classroom, enabling the students to gain exposure to many different organizations. True enough. But need the reality—even the 'reality' of the 20-page case—be dealt with in only this way? Is there not another option, which is to open up the students' perspective beyond the design school model, indeed even to use cases themselves to do so, but written and taught from a broader point of view? 13 In his 1987 book, Andrews acknowledges that 'How to get results is harder to teach and to learn in a classroom than on the scene. This difficulty may explain the neglect in business education of the art of implementation in favor of the analysis of potentially ideal strategies' (p. ix).

What effect has such case study teaching had on practice, on the generations of managers who have graduated from schools that rely on this pedagogy? If that has left managers with the impression that, to make strategy, they can remain in their offices with documents summarizing the situation and think—formulate so that others can implement—then it may well have done them and their organizations a terrible disservice, encouraging superficial strategies that violate the distinctive competences of their organizations. T o quote Livingston (1971), a Harvard professor at the time himself, in his classic article 'The myth of the well-educated manager,' the problem of management education is its 'secondhandedness:' 'Managerial aspirants are required only to explain and defend their reasoning, not to carry out their decisions or even to plan realistically for their implementation;' they 'are rarely exposed to "real" people or to "live" cases,' but rather to 'problems or opportunities discovered by someone else, which they discuss, but do nothing about.' Thus, many 'are not able to learn from their own firsthand experience . . . . Since they have not learned how to observe their environment firsthand or to assess feedback from their actions, they are poorly prepared to learn and grow as they gain experience' (pp. 79, 83, 84, 89). The fact is that the design school model dominates not only the world of pedagogy, either in its pure form, or as the foundation of the thinking behind the planning and positioning schools; it dominates beliefs in practice too. In other words, 'one best way' thinking is alive and well in the practice of strategic management, and it dictates that formulation must precede implementation, that this formulation must be conscious and controlled, by the chief executive as the architect of strategy, and that the resulting strategies must be deliberate and explicit. Here is how Robert McNamara, also formerly of the Harvard Business School, spelled out his approach to military strategy as Secretary of Defense: 'We must first determine what our foreign policy is to be, formulate a military strategy to carry out that policy, then build the military forces to successfully conduct this strategy' (quoted in Smalter and Ruggles, 1966: 70). H e did just this in Vietnam, distant from the realities of the rice paddies and for too long deaf to the calls to learn from the devastating

32

H. Mintzberg The Design School

results. Or consider the comment of one manager about an earlier chief executive of the General Electric Company: 'Borch had a sense that he wasn't looking for lots of data on each business unit, but really wanted 15 terribly important and significant pages of data and analysis' (quoted in Hamermesh, 1986: 191). As noted earlier, the problem may be most acute in diversification by acquisition, which often appear to have been undertaken by detached executives sitting up in executive suites designing strategies quite independently of any intimate understanding of the organization's real strengths and weaknesses. This 'one best way' thinking applies also to many of the consulting firms that specialize in this field—the so-called 'strategy boutiques.' Called in with limited knowledge of the industry in question, and limited time to find out, the design school provides a most convenient model. To quote from a popular book by two consultants: 'Four or five working days over a two-month period are required to set strategy. Two or three working days are required for the review and oneyear update of strategy' (Tregoe and Zimmerman, 1980: 120). There is not a lot of money to be made in saying: 'It's too complicated for us; go back and do your own homework; learn about your industry and your own distinctive competences by immersing yourself in the details and trying things; get many people involved; maybe over a few years you'll be able to develop an effective strategy. It's your responsibility; no one can do it for you." 4 As for Andrews' proposals about directors, his claim about 'the power of strategy as a simplifying concept enabling independent directors to know the business (in a sense) without being in the 14 In the early 1980s, frustration with the planning school and technique in particular, seems to have driven a number of practitioners and consultants back to the simpler design school model. Typical is the article by Walker Lewis (1984), founder of Strategic Planning Associates, and entitled 'The C E O and corporate strategy in the 1980s: back to basics.' It rediscovers all the elements of that model; for example, 'the C E O must be an informed generalist;' 'he must foster the building of comparative advantage;' 'good strategic management requires taking the wide view . . . it means setting a corporate direction based on . . . a comprehensively developed strategy;' 'In the end, it is the C E O who must serve as the force behind a return to basic integrated strategies in the 1980s;' and 'he must prod the corporation along the path to implementation' (pp. I, 2, 6). Ironically, Lewis concludes his article with the claim that 'coming to terms with these changes requires more than the old answers' (p. 6), although that is precisely what he offers.

189

business' (p. 834) might be more of a problem than a solution. Can anyone, director or student, even manager, really know an organization without being in it? The time of directors is limited; they must be briefed through short documents and snappy presentations that articulate strategies clearly and simply, so that they can be evaluated on the spot. Case study discussions in the boardroom. But at what cost in strategic thinking? And strategic action? Andrews claims that 'graduates of a demanding Policy course feel at home in any management situation and know at once how to begin to understand it' (p. 6). But that may be the very essence of the problem. Mary Cunningham is a graduate of the course Andrews had in mind. She may not be typical, but her experience does reveal the problem in its extreme. With a great deal of publicity, Cunningham leaped from the Harvard MBA program to the personal assistantship of William Agee, chief executive of the Bendix Corporation, himself a Harvard MBA. Later she wrote a book on those experiences, entitled Powerplay (Cunningham and Schumer, 1984). Kinsley published a scathing review of it in Fortune magazine, at one point hitting precisely on the issue under discussion here: There is nothing in Powerplay to support Cunningham's contention that she is a business genius. Her chapter about learning curves and other B-school buzzwords seems infantile. What little discussion there is of actual business consists mainly of genuflecting in front of a deity called The Strategy. The Strategy is what Mary and Bill were up to when nasty-minded people thought they were up to something else. Near as I can tell, it consisted of getting Bendix out of a lot of fuddy-duddy old-fashioned products and into glitzy high tech. What makes this a terribly ingenious idea, let alone a good one, she does not say. But she became very attached to it. 'How's The Strategy going?' she asked Agee the first time they met after her departure from Bendix. And at the book's emotional climax, as Agee realizes he's going to lose control of Bendix to Allied Corp., he says: ' "Of course, you know what this means? . . . The Strategy that we've worked on so hard"—and here he nodded at me—"won't be in our hands." ' And they cry (1984: 142).

If the case study method, based on the design school model, has encouraged leaders to oversimplify strategy, if it has given them the impression that 'you give me a synopsis and I'll

The Design School 190

H.

33

Mintzberg

give you a strategy,' if it has denied strategy formation as a long, subtle, and difficult process of learning, if it has encouraged managers to detach thinking from acting, remaining in their offices instead of getting into factories and meeting customers where the real information may have to be dug out, then it may be a major cause of the problems faced by contemporary organizations. This critique may sound extreme. We do not believe it is; as we shall discuss below, there is much in the design school to recommend it, at least under certain circumstances (indeed much in using cases as pedagogical devices too). But not when it is applied without a depth of understanding of what a particular organization is, and how it must sometimes learn.

THE DESIGN SCHOOL: CONTEXT AND CONTRIBUTION Our critique has not been intended to dismiss the design school model, but rather the assumption of its universality, that it somehow represents the 'one best way' to make strategy. In particular, we reject the model where we believe strategy formation must above all emphasize learning, notably in circumstances of considerable uncertainty and unpredictability, or ones of complexity in which much power over strategy-making has to be granted to a variety of actors deep inside the organization. We also reject the model where it tends to be applied with superficial understanding of the issues in question. Andrews thought it sufficient to delineate one model and then add qualifications to it. The impression left was that this was the way to make strategy, although with nuance, sometimes more, sometimes less. But that had the effect of associating strategy-making with deliberate, centralized behavior and of slighting the equally important needs for emergent behavior and organizational learning. Another extreme—what we have elsewhere presented under the label of the 'grass roots model' (Mintzberg and McHugh, 1985)—makes no more sense, since it overstates equally. But by positioning these two at ends of a continuum, we can begin to consider real-world needs along it. In other words it is not Andrews' qualifications that will hold the model in check so much as an alternate depiction of the process.

That is why the field of strategic management has need for these different schools of thought, so long as each is considered carefully in its own appropriate context. Accordingly, we can begin to delineate the conditions that should encourage an organization to tilt toward the design school model end of the continuum. We see a set of four in particular. 1. One brain can, in principle, handle all of the information relevant for strategy formation. The assumption of the single strategist sometimes does hold up: a chief executive (perhaps teamed up with other top managers), albeit one who is rather clever and especially adept at synthesis, can take full charge of the process for creating strategy. Here the situation must be relatively simple, involving a base of knowledge that can be comprehended in one brain. 2. That brain has full, detailed, intimate knowledge of the situation in question. The potential for centralizing knowledge must be backed up by sufficient access to, and experience of, the organization and its situation to enable the strategist to understand in a deep sense what is going on. We might add that he or she can only know the organization by truly being in the organization. This precludes the image of the case study classroom, the detached CEO with a pithy report, the 'quick-fix' consulting contract, the quarterly directors' meeting, even the weekend retreat of executives (although this may culminate the process). Rather it describes the strategist who has developed a rich, intimate knowledge base over a substantial period of time. 3. The relevant knowledge is established and set before a new intended strategy has to be implemented—in other words, the situation is relatively stable or at least predictable. Not only must the strategist have access to the relevant knowledge base, but there must also be some sense of closure on that base: at some point in time, the strategist must know what needs to be known to conceive an intended strategy that will have relevance well beyond the period of implementation. The world must, in other words, hold still, or— what amounts to a much more demanding assumption—the strategist must have the capability to predict the changes that will

34

H. Mintzberg

The Design come about. What this means is that individual learning must come to an end before organizational action taking can begin. And that can happen effectively only when the future can, in fact, be known. 4. The organization in question is prepared to cope with a centrally articulated strategy. For one thing, others in the organization must be willing to defer to a central strategist. For another, they must have the time, the energy, and the resources to implement a centrally determined strategy. And, of course, there has to be the will to do that implementation. These conditions suggest some clear contexts in which the design school model would seem to apply best—its own particular niche, so to speak, related to time as well as situation. In other words, this is a model to be applied only in certain kinds of organizations, and even there only in certain circumstances. Above all is the organization that needs a major reorientation, a new conception of its strategy. Newman recognized this early, referring to the 'quick reversal,' the 'sharp break' (1967: 117). Or, as Rumelt has put it, 'a good strategy does not need constant reformulation. It is a framework for continual problem solving, not the problem solving itself (1980: 365; see also Henderson, 1979: 38). Two conditions would seem to characterize what we call this period of reconception. First, there was a major change in the situation that previously supported the existing strategy, so that it has been seriously undermined. And second, there has developed the beginnings of a new stability, one that will support a new conception of strategy. In other words, the design school model would seem to apply best at the junction of major shift for an organization, coming out of a period of changing circumstances and into one of operating stability. We would normally expect the provoking change to be one of a crisis or problem in the external condition of the organization, for example a major realignment of competition, a key shift in market demand, a technological breakthrough. Yavitz and Newman also suggest that what they refer to as 'total reassessment' can be proactive too, triggered, for example, by 'milestones in major programs,' periods when 'large commitments of resources must be made' or 'key uncertainties are resolved,' or simply 'a

School

191

maximum period since the last full review' (1982: 215-216). Such strategic reassessments may also result from the introduction to the organization of fresh strategic thinking on the part of new leaders. There is another context where the design school model might apply, and that is the new organization, since it must have a clear sense of direction in order to compete with its more established rivals (or position itself in a niche free of their direct influence). This period of initial conception of strategy is, of course, often the product of an entrepreneur with a vision who created the organization in the first place. Context describes structure as well as time and situation. In the context described above, the structure tends to be simple—flexible, nonelaborated, very responsive to the dictates of a single leader (Mintzberg, 1979: chapter 17). Once under way, however, even simple structures with entrepreneurial leaders may not follow the design school model, even in times of reconception, because the leader's considerable personal discretion (including personal control of 'implementation') allows him or her to change strategy gradually, even continuously, without any need to articulate it. In a way, Andrews recognized this when he sought to distance his model from the entrepreneurial context and its reliance on intuition and 'opportunism.' But in so doing he also distanced it from some of the most creative strategy-making behavior found in organizations. The structural context Andrews seemed to favor for his model (although he would hardly use the label we are about to apply to it), and the one that appears to be most appropriate for the period of reconception of strategy in an existing organization, is what we call 'machine bureaucracy' (Mintzberg, 1979: chapter 18). This is structure characterized by a centralization of authority and a relatively stable context of operations, typically used in mass production and the mass delivery of services. Machine bureaucracies commonly pursue highly articulated and stable strategies. They therefore require in periods of reconception much of what the design school has to offer: a process whereby someone in central command somehow pulls the new conception together—defines it if not actually creates it—and then articulates it fully at a point in time so that everyone else can implement it and then pursue it.

The Design School 192

H.

35

Mintzberg

But there is an interesting anomaly here. The call from the design school for a personalized and creative form of strategic management (one strategist, strategies as novel conceptions) is not really compatible with machine bureaucracy, which tends to rely on standardized procedures and detached forms of control. In other words, machine bureaucracies are not mobiles to effect strategic change but stabiles for the continued pursuit of given strategies. For example, our own research on strategy formation (Mintzberg, 1978; Mintzberg et al., 1986) suggests that chief executives of machine bureaucracies tend to be caretakers of existing strategies—fine-tuners of set directions rather than champions of radically new ones—in part because of the constraints imposed by their own standardized procedures. These organizations are, after all, machines dedicated to the pursuit of efficiency in very specific domains. Indeed, the whole array of mechanisms proposed in the design school's own model of implementation—performance measures, incentive systems, various other control procedures, not to mention the articulation of strategy itself, as noted earlier—once in place act not to promote change in strategy but to resist it. Formal implementation, ironically, impedes reformulation. Our own evidence (Mintzberg, 1978), as well as that of Miller and Friesen (1984), suggests that major reformulation in machine bureaucracy typically occurs through a form of revolution; power is centralized around a single leader who acts personally and decisively to unfreeze existing practices and impose a new vision. In other words, in such 'turnarounds,' the organization tends to revert to the more flexible simple structure, and to its more entrepreneurial mode of strategy-making, at least until it has developed a new realized strategy, after which it tends to settle back down to its old machine bureaucratic way of functioning. The implication of this is that while the machine bureaucracy may occasionally require a period of reconception as provided for by the design school model, its own procedures impede the faithful use of that model. In a sense, implementation fits, formulation does not. Indeed, initial use of the model itself discourages later use of it: by articulating strategy and implementing it, as prescribed, the machine bureaucracy finds it difficult to change its strategy later, to reformu-

late. Thus, the design school model tends here to get 'caught in the middle,' to use Porter's phrase, tilts toward the personalized intuition of entrepreneurship for major reformulation and toward the analysis of planning for the more routine pursuit of strategy. Can we conclude, therefore, that by trying to position the design school model free of intuition on one side and planning on the other, Andrews left it little room for real application, perhaps mainly marginal strategic change in the machine bureaucratic type of organization where leaders can exercise 'judgement' but not rely on intuition or analysis?15 As for more complex types of organizations, which depend on expertise for their functioning, as we have argued elsewhere, 'professional bureaucracies' and 'adhocracies' cannot rely on the conventional prescriptive approaches to strategy-making, whether design, planning, or positioning school oriented, but must instead tilt toward the learning end of the continuum, developing strategies that are more emergent in nature through processes that have more of a grass roots orientation (Hardy et al., 1984; Mintzberg and McHugh, 1985).16 To conclude, should we take the design school model literally? In assessing the real contribution of this school, perhaps we should not. For while the model (even the framework) may have restricted application and often be overly simplified, this school's contribution as an 'informing idea' has been profound. The design school has provided important basic vocabulary by which we discuss grand strategy, and it has provided the central notion that underlies all prescription in this field, namely that strategy represents a fundamental congruence between external opportunity and internal capability. These important contributions will stand no matter how many of this school's specific premises may fall away.

15 A study of the cases favored by the design school may be instrumental in this regard. Our own suspicion is that there is probably a predisposition toward mass production or mass service organizations, typically machine bureaucratic in nature, although the role of the intuitive leader in trying to effect turnaround in them in a personalized way may be more evident in the cases than in the theory (e.g. in the J. I. Case case, in Learned el al., 1965: 82-102). 16 Note that, in describing the strategy-making process lavored in different types of organizations, we are further making the case for the impact of structure on strategy (see also Normann, 1977: 9, 19 and Bower, 1970: 286-287).

36

H. Mintzberg The Design School

ACKNOWLEDGEMENTS My thanks to one very thorough and considerate reviewer, also to Bill Newman for his comments on some of the early history of the use of the strategy concept in business.

REFERENCES Andrews, K. R. Memorandum: 'The formulation— implementation dichotomy' in the Concept of Corporate Strategy. Harvard Business School, 21 October 1976. Andrews, K. R. The Concept of Corporate Strategy, Irwin, Homewood, IL, 1971; second edition, 1980a; Third edition, 1987. Andrews, K. R. 'Directors' responsibility for corporate strategy' Harvard Business Review, NovemberDecember 1980b, pp. 30-44. Andrews, K. R. 'Replaying the board's role in formulating strategy', Harvard Business Review, May-June 1981, pp. 18-28). Ansoff, H. I. Ά quasi-analytical approach to the business strategy problem', Management Technology, June 1964, pp. 67-77. Ansoff, Η. I. Corporate Strategy, McGraw-Hill, New York, 1965. Barnard, C. I. Organization and Management: Selected Papers, Harvard University Press, Cambridge, MA, 1948. Berg, N. General Management: An Analytical Approach, Irwin, Homewood, IL., 1984. Boston Consulting Group Inc. Strategy Alternatives for the British Motorcycle Industry (A report prepared for the Secretary of State for Industry, Government of Great Britain, 1975). Bourgeois, L. J. Ill and D. R. Brodwin. 'Strategic implementation: Five approaches to an elusive phenomenon', Strategic Management Journal, 5, 1984, pp. 241-264. Bower, J. L. Managing the Resource Allocation Process: A Study of Corporate Planning and Investment. Harvard Business School, 1970; revised edition, 1986. Burgelman, R. Α. Ά process model of internal corporate venturing in the diversified major firm', Administrative Science Quarterly, XX'Vili, 1983, pp. 223-244. Chandler, A. D. Jr. Strategy and Structure: Chapter in the History of the Industrial Enterprise, Massachusetts Institute of Technology Press, Cambridge, MA, 1962. Christensen, C. R., K. R. Andrews and J. L. Bower. Business Policy: Text and Cases, fourth edition, Irwin, Homewood, IL, 1978. Christensen, C. R., K. R. Andrews, J. L. Bower, R. G. Hamermesh and M. E. Porter. Business Policy: Text and Cases, fifth and sixth editions, Irwin, Homewood, IL, 1982 and 1987.

193

Cunningham, M. and F. Schumer. Powerplay: What Really Happened at Bendix, Simon & Schuster, New York, 1984. Dimma, W. A. 'Competitive strategic planning', Business Quarterly, Spring 1985, pp. 22-28. Feld, M. D. 'Information and authority: The structure of military organization', American Sociological Review, XXIV, 1959, pp. 15-22. Gilmore, F. F. and R. G. Brandenburg. 'Anatomy of corporate planning', Harvard Business Review, November-December 1962, pp. 61-69. Hamermesh, R. G. Making Strategy Work, Wiley, New York, 1986. Hardwick, C. T. and B. F. Landuyt. Administrative Strategy, Simmons-Boardman, New York, 1961. Hardy, C., A. Langley, H. Mintzberg and J. Rose. 'Strategy formation in the university setting.' In J. Bess (ed.), College and University Organization: Insights for the Behavioral Sciences, New York University Press, New York, 1984. Hayes, R. H. 'Strategic planning—forward in reverse?', Harvard Business Review, November-December 1985, pp. 111-119. Henderson, B. D. On Corporate Strategy, Abt Books, Cambridge, MA, 1979. Hofer, C. W. and D. Schendel. Strategy Formulation: Analytical Concepts, West, St. Paul, MN, 1978. Hrebiniak, L. G. and W. F. Joyce. Implementing Strategy, Macmillan, New York, 1984. Katz, R. L. Cases and Concepts in Corporate Strategy, Prentice Hall, Englewood Cliffs, NJ, 1970. Kiechel, W. Ill 'Sniping at strategic planning' (Interview with himself), Planning Review, May 1984, pp. 8-11. Kiesler, C. A. The Psychology of Commitment: Experiments Linking Behavior to Belief, Academic Press, New York, 1971. Kinsley, Μ. Ά business soap opera', Fortune, June 1984, pp. 141-144. Kress, G., G. Koehler and J. F. Springer. 'Policy drift: An evaluation of the California Business Enterprise Program', Policy Studies Journal, Special Issue no. 3, 1980, pp. 1101-1108. Learned, E. P., C. R. Christensen, Κ. R. Andrews and W. D. Guth. Business Policy: Text and Cases, Irwin, Homewood, IL, 1965. Learned, E. P. and A. T. Sproat. Organization Theory and Policy: Notes for Analysis, Irwin, Homewood, IL, 1966. Lenz, R. T. 'Strategic capability: A concept and framework for analysis', Academy of Management Review, 5, 1980, pp. 225-234. Levitt, T. 'Marketing myopia', Harvard Business Review, July-August 1960, pp. 45-56. Lewis, W. W. 'The C E O and corporate strategy: Back to basics'. In A. C. Hax (ed.), Readings on Strategic Management, Ballinger, New York, 1984. Lindgren, U. and K. Spangberg. 'Corporate acquisitions and divestments: The strategic decisionmaking process', International Studies of Management and Organization, XI, 1981, pp. 24-47. Lipsky, M. 'Standing the study of public policy

The Design School 194

37

H. Mintzberg

implementation on its head'. In W. D. Burnham and M. W. Weinberg (eds), American Politics and Public Policy, MIT Press, Cambridge, MA, 1978. Livingston, J. S. 'The myth of the well-educated manager', Harvard Business Review, J a n u a r y February 1971, pp. 79-89. Majone, G. and A. Wildavsky. 'Implementation as evolution', Policy Studies Review Annual, U, 1978, pp. 103-117. McKinsey, J. O. Adjusting Policies to Meet Changing Conditions. American Management Association General Management Series, AM116, New York, 1932. Miles, R. H. Coffin Nails and Caporale Strategies, Prentice-Hall, Englewood Cliffs, NJ, 1982. Miller, D. and P. H. Friesen. Organizations: A Quantum View, Prentice-Hall, Englewood Cliffs, NJ, 1984. Mintzberg, H. The Nature of Managerial Work, Harper & Row, New York, 1973. Mintzberg, H. 'Patterns in strategy formation', Management Science, XXIV, 1978, pp. 934-948. Mintzberg, H. The Structuring of Organizations, Prentice-Hall, Englewood Cliffs, NJ, 1979. Mintzberg, H. 'Crafting strategy', Harvard Business Review, July-August 1987, pp. 66-75. Mintzberg, H. 'Strategy formation: Ten schools of thought', in J. Fredrickson (ed.), Perspectus on Strategic Management, Ballinger,"New York, forthcoming. Mintzberg, H. and A. McHugh. 'Strategy formation in an adhocracy', Administrative Science Quarterly, XXX, 1985, pp. 160-197. Mintzberg, H. and J. A. Waters. 'Tracking strategy in an entrepreneurial firm', Academy of Management Journal, XXV, 1982, pp. 465-499. Mintzberg, H. and J. A. Waters. 'Of strategies, deliberate and emergent', Strategic Management Journal, 6, 1985, pp. 257-272. Mintzberg, H., J. P. Brunet and J. A. Waters. 'Does planning impede strategic thinking?', Advances in Strategic Management, vol. 4, JAI Press, Greenwich, CT, 1986, pp. 3-41. Mintzberg, H., D. Raisinghani and A. Theoret. 'The structure of "unstructured" decision processes', Administrative Science Quarterly, XXI, 1976, pp. 246-275. Moore, D. G. 'Managerial strategies'. In W. L. Warner and N. H. Martin (eds), Industrial Man, Harper & Row, New York, 1959, pp. 219-226. Newman, W. H. Administrative Action: The Techniques of Organization and Management, Prentice-Hall, Englewood Cliffs, NJ, 1951; second edition, 1963. Newman, W. H., C. E. Summer and E. K. Warren. The Process of Management Concepts, Behavior, and Practice, Prentice-Hall, Englewood Cliffs, NJ, 1951; second edition, 1967. Newman, W. H., E. K. Warren and A. R. McGill. The Process of Management: Strategy, Action, Results, Prentice-Hall, Englewood Cliffs, NJ, sixth edition, 1987. Normann, R. Management for Growth, Wiley, New

York, 1977. Ohmae, K. The Mind of the Strategist, McGraw-Hill, New York, 1982. Pascale, R. T. 'Perspectives on strategy: The real story behind Honda's success', California Management Review, Spring 1984, pp. 47-72. Porter, M. E. Competitive Strategy: Techniques for Analyzing Industries and Competitors, Free Press, New York, 1980. Porter, M. E. 'The contributions of industrial organization to strategic management', Academy of Management Review, VI, 1981, pp. 609-620. Porter, M. E. 'Industrial organization and the evolution of concept for strategic planning: The new learning'. Managerial and Decision Economics, IV, 3, 1983, pp. 172-180. Porter, M. E. Competitive Advantage: Creating and Sustaining Superior Performance, Free Press, New York, 1985. Quinn, J. B. 'Strategic change: "Logical incrementalism" ', Sloan Management Review, Fall 1978, pp. 7-21. Quinn, J. B. Strategies for Change: Logical Incrementalism, Irwin, Englewood Cliffs, NJ, 1980. Radosevich, R. Ά critique of "comprehensive managerial planning" '. In J. W. McLuin (ed.), Contemporary Management, Prentice-Hall, Englewood Cliffs, NJ, 1974. Reilley, E. W. 'Planning the strategy of the business', Advanced Management, December 1955, pp. 8-12. Rein, M. and F. F. Rabinovitz. 'Implementation: A theoretical perspective'. In W. D. Burnham and M. W. Weinberg (eds), American Politics and Public Policy, MIT Press, Cambridge, MA, 1979. Rumelt, R. P. 'Evaluation of strategies: Theory and models'. In D. E. Schendel and C. W. Hofer (eds), Strategic Management, Little, Brown, Boston, MA, 1979. Rumelt, R. P. 'The evaluation of business strategy'. In W. F. Glueck (ed.), Business Policy and Strategic Management, McGraw-Hill, New York, third edition, 1980. Rumelt, R. P. 'Towards a strategic theory of the firm'. In R. B. Lamb (ed.), Competitive Strategic Management, Prentice-Hall, Englewood Cliffs, NJ, 1984. Scheff, T. J. 'Control over policy by attendants in a mental hospital', Journal of Health and Human Behavior, 1961, pp. 93-105. Selznick, P. Leadership in Administration: A Sociological Interpretation, Harper & Row, New York, 1957. Smalter, D. J. and R. L. Ruggles. Jr 'Six business lessons from the Pentagon', Harvard Business Review, March-April 1966, pp. 64-75. Stevenson, Η. H. 'Defining corporate strengths and weaknesses', Sloan Management Review, Spring 1976, pp. 51-68. Sun Tzu. The Art of War (S. B. Confitt, translator), Oxford University Press, New York, 1971. Thoenig, J. C. and E. Friedberg. 'The power of the field staff: The case of the Ministry of Public Works and Urban Affairs and Housing in France'. In A.

38

H. Mintzberg The Design School

F. Leemans (ed.). The Management of Change in Government, Martinus Nijoff, The Hague, 1976. Tilles, S. 'How to evaluate corporate strategy', Harvard Business Review, July-August 1963, pp. 111-120. Tregoe, Β. B. and J. W. Zimmerman. Top Management Strategy, Simon & Schuster, New York, 1980. Wrapp, Η E. 'Good managers don't make policy

195

decisions', Harvard Business Review, S e p t e m b e r October 1967, pp. 91-99. Yavitz, B. and W. H. Newman. Strategy in Action: The Execution, Politics, and Payoff of Business Planning, Free Press, New York, 1982. Zand, D. E. Information, Organization, and Power, McGraw-Hill, New York, 1981.

Critique of Henry Mintzberg

39

Strategic Management Journal, Vol. 12, 449-461 (1991)

CRITIQUE OF HENRY MINTZBERG'S THE DESIGN SCHOOL: RECONSIDERING THE BASIC PREMISES OF STRATEGIC MANAGEMENT' H. IGOR ANSOFF United States International University, San Diego, California, U.S.A.

Mintzberg's (1990) critique of the 'design school' of strategic management is evaluated on two criteria: methodological soundness and factual veracity. The critique is found to be deficient on both criteria. Mintzberg's own proposal for the basic principles of strategic management is critiqued using the same criteria. It is found that the exposition is deficient methodologically and that Mintzberg's descriptive and prescriptive assertions are at variance with facts observable in the current practice of strategic management. The variance is found to be due to several factors: lack of coherence in Mintzberg's presentation; his use of a definition of strategy which is at variance with the current practice of management, his failure to differentiate between prescriptive and descriptive statements; and his failure to define the context for his prescriptions. Using recent empirical research results on strategic success behaviors, Mintzberg's model is placed in a limited but important context in which it is a valid prescription for successful strategic behavior.

INTRODUCTION The key conclusions of Mintzberg's (1990) paper are the following: 1. The 'Design School' at The Harvard Business School, having enunciated in the 1960s a set of prescriptive concepts for strategy formulation, 'denied itself the opportunity to adapt these concepts ever since. 2. The 'other' prescriptive schools of strategy formulation (which are vaguely named, but not described by Mintzberg) shared the basic concepts of The Harvard Business School (HBS). 3. Like the Design School, the other prescriptive schools remained frozen in time. 4. The design principles shared by the design

Key words: tion.

Design School/critique, strategy forma-

0143-2095/91/070449-13S06.50 © 1991 by John Wiley & Sons, Ltd.

schools were, and still are, generally invalid except in a narrow specific context. 5. Interspersed with the critique of The Design School, are Mintzberg's own descriptions of the nature of strategy formation and prescriptions for the use of the 'emerging strategy' formation process, based on 'trial and experience'. Mintzberg argues that in unpredictable environments it is impossible to formulate an explicit strategy before the trial and experience process has run its course; and that it is not necessary to make strategy explicit in predictable environments.

Thus, according to Mintzberg, for all intents and purposes, all of the prescriptive schools for strategy formulation should be committed to the garbage heap of history, leaving the field to the 'emerging strategy' school which he represents. Many readers will recognize that the author of this paper is a 40-year-long card-carrying member of one of the schools which Henry confines to

Received 22 October 1990 Revised 21 February 1991

40

450

H. I. Ansoff

H. Igor

Ansoff

obscurity. These readers are also likely to know that my entire professional career has been focused on helping organizations manage their strategic behavior in unpredictable environments. Thus, if I am to accept Henry's verdict, I have spent 40 years contributing solutions which are not useful in the practice of strategic management. Therefore, it should not be surprising that I rise in defense of at least one prescriptive school (the one to which I belong) in an effort to set the record straight and thus salvage a lifetime of work which has received a modicum of acceptance by practicing managers. In situations like the present, it is easy to fall prey to a game of polemic charge-countercharge in the hope that the louder voice will carry the day. I will attempt to avoid this trap in two ways. First, I will show that the methodology by which Mintzberg disposes of the prescriptive school will hardly stand its day in the court of logic, and persuasiveness. Second, I will offer evidence of repeated instances in which Mintzberg's key assertions are factually wrong. Thirdly, I will fault Henry on the fact that, having confined the prescriptive schools to a narrow context, he does not place his own in an appropriate context. Finally, I will identify the context which is appropriate for Henry's prescriptions. It is ironic that this context will appear very similar to the context to which he confínes the prescriptive schools, but is somewhat larger in scope. Thus, to borrow a phrase which Henry uses in his critique of Professor Kenneth Andrews, his paper emerges as 'a caricature of his own model.'

and of conflicts which are typical of academic life. Therefore, Henry's generalization from a sample of one requires factual support. Such support is not offered. Instead, Mintzberg attempts to minimize evidence to the contrary. Since world-wide visibility of Michael Porter cannot be left unnoticed, Mintzberg tries to minimize his influence on The Design School on the grounds that the HBS classic text on policy devotes only one chapter to Porter. Thus the reader is asked to believe that Porter's influence in the Harvard Business School has been confined to one chapter in a book!

MINTZBERG'S PROOF THAT THE DESIGN SCHOOL DENIED ITSELF THE CHANCE TO ADAPT

Responsibility for (strategy formulation) must rest with the chief executive officer (CEO): that person is T H E strategist.

Generalization from a sample of one The writings of Professor Kenneth Andrews (1971) are the only source used in construction of this proof, and the Harvard Business School is made to appear to be solidly united behind him as the School's idealogue and spokesman. Any reader who spent time in the halls of academe would automatically suspect this assumption of absence of differences in viewpoints

Proof by implied intent Having chosen Andrews as the 'mouthpiece' of the Design School, Mintzberg uses Andrews' own writings to prove that the school 'refused itself the chance to adapt' over time. This is done by challenging Andrews' statements which suggest that the School's original design principles should be enlarged and modified. The methodology is simple. First, having quoted a paragraph from Andrews, which suggests to an intelligent reader that the Design School did indeed continue to elaborate the original principles, Henry asserts (without any further evidence) that Andrews did not really mean what he said! An example of one of several such 'proofs' should suffice to illustrate this 'methodology'. According to Mintzberg the second design principle advanced by the Design School (1990: 176) is as follows:

In discussing Andrews' qualifications of this premise, Mintzberg quotes the following paragraph from Andrews' writings: False hope, oversimplification, and naiveté, as well as zest for power, have often led. . . .to the assumption that the chief executive officer conceives strategy single-mindedly, talks the board of directors into pro forma approval, announces it as a fixed policy, and expects it to be promptly executed by subordinates under

Critique of Henry Mintzberg Critique of 'The Design School' conventional command and control procedure (Andrews, 1987: 82).

Admittedly, the paragraph is turgid and elliptical, but a careful reading makes clear the author's intent: 'It is an improper assumption that the CEO should be THE only strategist.' Mintzberg arrives at the same interpretation and then summarily and flippantly dismisses it in a half sentence. ' we see it (the quotation), as a not unreasonable caricature of his own text (Mintzberg, 1990: 179).

A reader will find in Mintzberg's paper several other such 'proofs' by assertion that, whenever Andrews tries to enlarge the original principle, he really does not mean what he says.

PROOF THAT OTHER PRESCRIPTIVE SCHOOLS HAVE ALSO REMAINED •FROZEN IN TIME' 'Proof by sweeping assertion As mentioned before, Mintzberg offers no description nor discussion of 'the other' prescriptive schools. However, this does not prevent him from making the following sweeping assertion: The reader is asked to bear in mind that although the other prescriptive schools of planning and positioning have broken with certain of the premises of the design school the fact that they have accepted the most basic ones renders most of the following a critique of those schools as well (Mintzberg, 1990: 181). (italics added for emphasis).

In scientific practice, sweeping assertions, such as the preceding one, are not accepted as proofs and must remain suspect until proven to be true or false. I will use two generally accepted proofs to show that the above assertion is false. The first is an epistemologica! proof suggested by Alfred North Whitehead (1962), who states that sweeping assertions should be tested for credibility against common experience. Here is what Mintzberg expects his readers to accept as credible: That a sizeable group of idiosyncratic individ-

41 451

uals who derive a substantial part of their living by selling their intellectual capital to practicing managers, would forego their idiosyncracy and their competitive advantage, for the privilege of following intellectual leadership of The Harvard Business School. To this author the above picture of academia is just as ludicrous as the earlier picture painted by Mintzberg of monolithic ideological unity within the HBS.

Contradictory factual evidence The credibility test is subjective. A more persuasive proof is a factual one. In such proof a single fact which contradicts the assertion is sufficient to falsify it. In mathematics this is known as the Gegenbeispiel principle of testing theoretical propositions. Presented below are two facts which contradict Mintzberg's assertion that in the 1960s all prescriptive schools were basically alike. The first fact may not have been available to Mintzberg. It is derived from a three-way meeting which took place at the Harvard Business School in 1962. The participants were two senior faculty members from each of the following major business schools: Sloan School of Management at MIT, Harvard Business School, and Graduate School of Industrial Administration at CarnegieMellon University. This writer was one of the participants. During an intensive 2 days of discussion the participants explored two basic questions about strategy formation. The first was whether strategy has a distinctive content of its own or whether it was simply on integration of functional inputs, such as marketing, R&D, etc. The second question was: if one assumed that strategy was a distinctive subject, is it possible to describe it in a structured manner, or must it of necessity remain an ephemeral concept which defies structuring and must, therefore, be studied by the verbal case method 'without writing anything down' (as was advocated in an early version of Harvard's classic case book on policy formation). For the purpose of the present concern, suffice it to say that, at the end of 2 days, the three participant schools enunciated fundamentally

42

452

H. I. Ansoff

H. Igor Ansoff

different views which led to different 'design principles,' thus denying Mintzberg's assertion that all prescriptive schools were alike. The second fact which contradicts Mintzberg's assertion should have been known to him, because it is discussed at length in a book published in 1965 (Ansoff, 1965), which he references in his paper. This fact is that the concept of strengths and weaknesses, ascribed by Mintzberg to the Design School, was conceptually criticized in this book, and a detailed alternative method was proposed for identifying future strengths and weaknesses of an organization. Incidentally, this method met (in 1965) many of the objections which Mintzberg makes in 1990 to the strengths/weaknesses concept of The Design School.

2.

3.

4.

5. Factual contradiction of assertion that all of the prescriptive schools denied themselves a chance to adopt with times One factual counterexample will suffice to prove this assertion false. In this example, I will briefly trace the evolution of one of the prescriptive schools, which through the years, has stayed in close touch with the changing practice of strategic management, adopted many prescriptions which have emerged in practice, and in recent years made several original contributions to the practice of management. I will refer to this School as the School of Holistic Strategic Management. (Because of his off-handed dismissal of 'the other' prescriptive schools, it is not possible to tell whether Mintzberg is aware of the existence of this school.) However, as shown below, its origins and its progress are well documented. The extent of progress of The School of Strategic Management between 1965 and 1990 can be assessed by comparing two books by this author: Corporate Strategy, first published in 1965 (Ansoff, 1965) and Implanting Strategic Management, which first appeared in 1984 (Ansoff, 1984). Following are the milestones of the School's Evolution: 1. As already discussed in a book published in 1965 (Ansoff, 1965), this School enunciated a concept of strengths and weaknesses which

6.

7.

8.

9. 10.

was drastically different from that of the Design School. The same book presented a structured method for analytic strategy formulation (which was a codification of its author's practical experience), a procedure which at the time was being used in practice but was considered impossible at The Harvard Business School (Ansoff, 1965). In 1978, the concept of strengths and weaknesses was replaced by a comprehensive concept of Organizational Capability, (Ansoff, 1978). The original concept that strategy formulation should be centralized in the hands of the C E O was replaced by the concept strategic bi-centralization (Ansoff, 1984). The concept of Strategic Myopia of key strategic managers and of resistance to strategic change were formulated and a practical procedure developed for overcoming both of them during strategy formulation and implementation (Ansoff, 1984). A diagnostic procedure was developed for sequencing strategy/structure development, according to the degree of urgency of strategic response being experienced by a firm. (Ansoff, Declerck and Hayes, 1974). In 1972 the overall perspective of the subject was broadened from strategy formulation to the overall process by which organizations adapt and succeed in turbulent environments, and the concept of Strategic Management was introduced (Ansoff, 1972). The concept of real time response was developed, as an alternative to periodic strategy planning, and three practical real time response procedures were proposed: (i) Strong Signal Issue Management·, (ii) Weak Signal Issue Management·, and (iii) Surprise Management (Ansoff, 1984; Ansoff, Kirsch and Roventa, 1980). In 1979 an applied theory of strategic behavior was developed and published (Ansoff, 1979). A Strategic Success Hypothesis, which is a keystone of this theory, was repeatedly tested and validated in a variety of organizational types and several countries. (Hatziantoniou, 1986; Salameh, 1987; Sullivan, 1987; Chabane, 1987; Lewis, 1989; Jaja, 1990; Ansoff and McDonnell, 1990; Ansoff, Sullivan et al., 1990.)

Critique of Henry Mintzberg

Critique of 'The Design School' 11. Based on the findings of this research a practical Strategic Diagnosis procedure was developed for determining the strategy and capability changes which an organization will have to make in order to succeed in the future (Ansoff, 1984; Ansoff and McDonnell, 1990). 12. Interactive Computer Software for strategy formulation in turbulent environment was developed (Ansoff, 1986) and marketed. In summary, at least one prescriptive school cannot be accused of having been a carbon copy of The Design School, either at its inception, nor during its subsequent evolution. Thus Mintzberg's assertion that all prescriptive Schools 'have accepted the premises' of The Design School and that they 'denied themselves the chance to adapt' is demonstrated to be false. Many additional counterexamples can be found in the bibliography attached to Mintzberg's paper. One of these deserves particular attention because it occurred within the Harvard Business School. It is found in the work of Michael Porter. Having banished Porter from the design school, Mintzberg totally ignores his massive and distinctive contribution to the literature on strategy formulation which certainly does not qualify for inclusion among the original design school principles at the Harvard Business School. Items 10 and 11 above show that The Holistic Strategic Management School, not only contributed new prescriptive principles, but also empirically identified the types of strategic behavior and their appropriate contexts which lead to organizational success. These findings will be used later in this paper for defining the appropriate context for Mintzberg's Model.

43

453

The critique is not confined to proving that The Design School's and other prescriptive Schools' principles are wrong. Interwoven with the critique are Mintzberg's own descriptive assertions about the real world, which he proceeds to convert into prescriptions for the manner in which strategy formation should take place in organizations. These prescriptions are sprinkled throughout the text and they are not summarized, nor logically connected. Therefore, the summary given below is this writer's attempt at a faithful summary of Henry's proposals. 1. The central prescription is that, with minor exceptions, all organizations should use what Mintzberg calls the 'emergent strategy' approach to strategy formation, using trial and experience process. 2. The output of this process is an observable strategy which is the logic pattern underlying the historical sequence of successful trials. 3. Except for minor exceptions, this strategy should not be made explicit:

Explicit Strategies are blinders designed to focus direction and so to block out peripheral vision (1990: 184). 4. It is not possible to formulate strategy in unpredictable environments:

. . .during periods of unpredictability. . . .(organizations) cannot possibly hope to articulate any viable strategy (1990: 184). 5. Nor is it possible to formulate a viable strategy in predictable environments:

MINTZBERG'S MODEL OF STRATEGY FORMATION Mintzberg leaves the reader in no doubt about his central theme: Our critique of the Design School revolves around one central theme: its promotion of thought independent of action, strategy formation above ail as a process of conception, rather than as one of learning (Mintzberg, 1990: 182).

The point we wish to emphasize is: how could the firm have known ahead of time? The discovery of what business it (firm) was to be in could not be undertaken on paper, but had to benefit from the results of testing and experience (1990: 182). The same quotation logically gives rise to the following conclusion, which is not articulated by Mintzberg:

44

H. I. Ansoff

454

H. Igor

Ansoff

6. It is not possible to forecast the future with complete confidence.

With these two exceptions recognized, we can infer the following prescription implied by Mintzberg:

Mintzberg's concern with managers' need 'to be sure,' and his assertion that they 'cannot' act before they are 'sure,' permeates the paper and is used as a basis for several descriptions and prescriptions, including the following:

10. The 'emerging strategy approach' should be used in all situations with the exception of the two specified above.

7. Managers should not make statements about the future if they are not totally sure of what they are saying. 8. Managers should not evaluate their organization's strengths and weaknesses until they become evident from the trial and error experience. 9. In complex organizations it is not possible to plan and coordinate an organization-wide process of strategy formulation. This assertion is contained in the following quotation from Brian Quinn, used and approved by Mintzberg: It is virtually impossible for a manager to orchestrate all internal decisions, external environmental events, behavioral and power relationships, technical and informational needs, and actions of intelligent opponents so that they come together at a precise moment (Quinn, 1978: 184).

Mintzberg makes no direct reference to the context in which his prescriptive principles should be used. But in his concern with what to do with The Design School, after he has demolished it, he does identify two contexts in which the explicit strategy formulation championed by the prescriptive schools may be applicable. One of these contexts is: a new organization. . . .(during). . . .the period of initial conception of strategy. . . . (1990: 191).

(In this case Mintzberg implicitly suspends his earlier claim that in unpredictable environments strategy cannot be formulated and allows the founding entrepreneur to have a 'vision'.) The other context is one in which: the design school model would seem to apply best. . .(is when) an organization (is) coming out of a period of changing circumstances and into one of operating stability (1990: 191).

In summary, Henry's prescription can be named as one of implicil strategy formation, under which strategy need not be a part of manager's concern, except under special circumstances. Managers should allow strategy and capabilities to evolve organically, through trial and experience, and focus their attention on the operating efficiency of the organization. Thus, Mintzberg prescribes a world free of explicit strategy formulation and free of strategic managers.

CRITIQUE OF MINTZBERG'S MODEL While reading the first part of the paper, one wonders why Mintzberg went to such length to prove that the prescriptive schools were identical and have jointly 'denied themselves' the opportunity to adapt to the changing times. The reason becomes clear in the second part: Mintzberg is now free to criticize all of the prescriptive schools as if they were still adhering to their original design principles of 1965. In the light of the methodological and factual deficiencies pointed out earlier in this paper, it is hardly worthwhile to challenge Mintzberg's criticisms of the original design principles, since they have been outstripped by developments, both in the practice of strategic management and in the writings of the prescriptive schools of thought. But Henry's own model of reality summarized in the preceding pages cries out for a critical appraisal. It is to this task that we now turn our attention. As a person who has spent over 40 years of his life as manager, consultant, educator, and a close observer of the business scene, I have difficulty accepting Henry's model as description of strategic management reality. And yet, Henry is an intellectually outstanding person, globally respected, and recognized as one of the leading contributors to the literature on strategic management. As I studied his paper several explanations of

Critique of Henry Mintzberg Critique of 'The Design School' this apparent paradox became clear. In the following pages I will present these explanations. As before, I will base my critique on methodological deficiencies and on factual contradictions between Henry's claims and the real world of strategic management. 2. Self-denial of a chance to study business environment It is strange how in his paper Mintzberg repeatedly commits sins of which he accuses the Design and the other prescriptive schools. One of these is the accusation directed at the Design School that it 'slight(s) the environment in favor of a focus on the organization' (1990: 182). Henry's paper shows that he commits the same sin. Below is the sum total of his references to the environment. One learns that managers: cannot be sure of the future. Sometimes organizations need to function during periods of unpredictability. Sometimes organizations come out of a period of changing circumstances into a period of operating stability.

3.

4.

5.

6. Nothing is said about how often is 'sometime', what is meant by 'unpredictability', by 'changing circumstances' or how long and how prevalent are the 'periods of operating stability.' The only complete sentence devoted to the environment does not help very much: . . . .environment is not some kind of pear to be plucked from the tree of external appraisal, but a major and sometimes unpredictable force. . . (1990: 185).

This cryptic statement begs all kinds of questions: whose environment is being discussed, what kind of influence does the force exert on organizations; under what circumstances is it exerted; what impact does it have on strategic behavior, etc? This slight of the environment is unfortunate. If Henry had taken the minimum trouble to peruse the cover pages of Business Week for the past 4-5 years, he would have easily found answers to most of the above questions. In brief, he would have found the following information. 1. In today's world, different types of organizations have different environments. Thus, since

45 455

the 1940s the environment of many business firms has progressively become more and more turbulent, unpredictable, and surpriseful. On the other hand, the not-for-profit organization had enjoyed a relatively placid environment until the 1970s (Ansoff, 1984). Within the two classes of organizations, the environments of different industries became differentiated. At one extreme, some organizations continue to enjoy a relatively placid existence and at the other extreme are organizations which are experiencing very high turbulence (Ansoff, 1984). The level of environmental turbulence has become a driving force which dictates strategic responses necessary for success (Ansoff and Sullivan, 1990). In high turbulence environments success comes to firms which use strategies which are discontinuous from their historical strategies (Ansoff and Sullivan, 1990; Ansoff et al., 1990). In low turbulent environments success comes to firms which use strategies of incremental development of their historically successful product-development, (op cit.). The final characteristic of the environment neglected by Mintzberg is the acceleration of the speed of change in the environment which has occurred during the past 30 years (Drucker, 1980).

The latter aspect of the environment puts in doubt the major prescription which Mintzberg offers in his paper. In turbulent environments, the speed with which changes develop is such that firms which use the 'emerging strategy formation' advocated by Mintzberg endanger their own survival. The reason is that when they arrive on a market with a new product/service, such firms find the market pre-empted by more foresightful competitors, who had planned their strategic moves in advance. Thus, the first reason for the contradictions between Mintzberg's picture of reality and the observable real world is his failure to observe the current business environment. Failure to meet validity tests for prescriptive and descriptive observations To be valid, a descriptive observation must meet a single test: it must be an accurate observation

46

456

H. I. Ansoff

H. Igor Ansoff

of reality. A prescription must pass a much more rigorous test: it must o f f e r evidence that use of the prescription will enable an organization to meet the objective by which it judges its success. Mintzberg seems to be oblivious to the need for evidence to support his descriptive statements, and he converts descriptions into prescriptions without any offering evidence that they will bring success to organizations using them. A n example of such conversion is offered by Mintzberg's treatment of experience with related diversifications. H e starts with a descriptive statement about the 'vast majority of experiences reported in the popular press' which shows that firms make a number of mistakes in their diversification programs and, without batting an eyelash, converts it into a prescriptive statement: 'acquiring firm has to make a number of mistakes until it gradually learns what works for it, if it ever does' (1990: 183) (italics added for emphasis). Thus a described pattern of successive failures is automatically transformed into a prescription for success. I am not sure that Henry appreciates the consequences of advocating use of trial and error in diversification programs. Having been in charge of a diversification department of a major American firm, I can testify to the fact that trial and error diversification is enormously expensive. The successive acquisitions require major investments by the acquirer, and disinvestment from mistakes multiplies the costs, because an acquisition cannot be sold-off overnight as one would sell a portfolio of poorly performing shares. But, even more importantly, the mere fact that 'the vast majority' of experiences has led to repeated mistakes is not a valid basis for recommending that others should follow the same path. What is being reported by Mintzberg are cases of failure and the fact that there are many of them does not mean that success seeking firms should follow their example.

. . . .sometimes organizations. . . .need to function during periods of unpredictability, when they cannot possibly hope to articulate any viable strategy (1990: 184) (italics added for emphasis).

Having stated the description, Henry offers the following prescription, again without any supporting evidence: When strategists are not sure, they had better not articulate strategies, for all the reasons given above (1990: 184) (italics added for emphasis).

However, a careful and multiple rereading of the proceeding text fails to reveal any 'reasons' unless it is the unarticulated conviction of Mintzberg's, which permeates the paper, that strategy formulation is impossible unless the environment is 'stable and predictable.' W e must now deal with the origin of this conviction.

Descriptive definition of strategy If Henry had taken the trouble to acquaint himself with the history and current practice of strategic management, he would have found widespread use of explicit a priori strategy formulation. Furthermore he would have found that explicit strategy formulation is typically used in environments in which managers are not 'sure' about the future (Steiner and Schollhammer, 1975). Thus, once more, Henry's assertion is contradicted by facts. In this case the explanation is twofold.

In fact, a major research study of mergers and acquisitions has shown that it is the planned approach to diversification, and not the trial and error approach, that produces better financial results (Ansoff et al., 1971).

The first is the black and white picture of the environment painted by Mintzberg: managers are either 'sure' or totally 'unsure' about the future. In the real world of management these two extremes are rarely observable (Schwartz, 1990). In practice managers are typically partially 'unsure' (see concept of partial ignorance in Ansoff, 1965). A n d they formulate strategy precisely because being 'unsure' makes it dangerous to assume that the firm's future will be an extrapolation of the past.

A second example is of critical importance to Mintzberg's model of strategic management. Without any prior evidence Henry offers the following description:

The second explanation is found in the difference between Henry's definition of the concept of strategy and the definition used in practice. His definition is descriptive since, in

Critique of Henry Mintzberg Critique of 'The Design School' order to identify the strategy, it is necessary to wait until a series of strategic moves has been completed. But the concept used in practice is prescriptive and it stipulates that strategy should be formulated in advance of the events which make it necessary. Thus Henry's failure to differentiate between descriptive and prescriptive statements once again places him in the position of contradicting observable reality.

Use of existential model of learning The model of organizational learning advocated by Mintzberg consists of a sequential trial and error process, neither preceded nor interrupted, nor followed by cognitive strategy formulation. To be sure, under special circumstances, he allows the possibility of postexperience strategy diagnosis. But nowhere in the paper does he suggest that the diagnosed strategy should in any way affect the choice of subsequent strategic moves. In fact, as cited before, Mintzberg considers explicit strategies to be 'blinders designed to block out peripheral vision.' This model of learning is the oldest one in human history. It was the model of the prehistoric man when he ventured from his cave in search for food. It was also the model of the master builders in The Middle Ages who created glorious cathedrals by repeating lessons learned from past successes, without understanding of what made the cathedrals stand or fall. This was also the model which was used to train new apprentices by putting them to work under direct guidance of experienced master builders. We shall refer to it as the existential model of learning. Henry's insistence on exclusive use of this most rudimentary model of learning in formation of strategy is ironic because it is the model on which The Harvard Business School Case method, which he criticizes at length, was originally built. The age of enlightenment ushered a new model which recognized importance of cognition in the affairs of man. In this model decision-making is the first stage, followed by implementation of the decision. It became the standard model of the natural sciences, and it was the model used in the early prescriptions for strategic planning. We shall call this model the rational model of learning. The rational model has several advantages over the existential:

47 457

1. In cases in which decision-making is less timeconsuming than trial and error, the rational model saves time by selecting action alternatives which are most likely to produce success. This time saving is of great importance in organizations which find themselves in rapidly changing environments. 2. It permits additional savings of time through starting strategic response in anticipation of need to act—a process called strategic planning. 3. It reduces the number of strategic errors and reduces costs by eliminating the probable "non-starters' from the list of possible strategic moves. Thus, the rational model becomes particularly important when the cost of a failed trial is very high, as in the case of diversification by business firms. Mintzberg makes no mention of the fact that the rational model is a legitimate alternative to the existential model. But he does devote a great deal of energy to proving that the existential model should be the only one used in strategic management. To support this claim, he makes a number of descriptive assertions which, as we have shown, are in conflict with factual evidence. First, he declares that cognitive strategy formulation is not possible in unpredictable environments, a claim which is contradicted by the fact of habitual strategy formulation in business firms. Second, he argues that, even in environments which are predictable, managers should not formulate a strategy unless they are sure of its consequences. He does this in the face of factual evidence that strategy formulation is typically found in firms whose managers are unsure about the future. Thirdly, he claims that explicit strategy makes strategic action rigid and forecloses opportunities which were not anticipated by the strategy. In making this claim, Henry neglects two facts which are readily available in the literature of the prescriptive schools (Ansoff, 1965). The first is that the strategy concept used in practice does not specify alternatives. On the contrary, it sets guidelines for the kinds of opportunities the firm wants to develop through search and creativity. The second fact is that successful practitioners of strategy typically use a strategic control mechanism which periodically reviews and, if

48

H. I. Ansoff

458

H. Igor Ansoff

necessary, revises the strategy in the light of experience. Thus, use of explicit strategy in successful practice is not rigid and does not foreclose attention to new opportunities which are outside the scope of strategy. But use of explicit strategy does control erratic deviations from the strategy. This point was well made in a quotation from Andrews used and rejected by Mintzberg: Strategy will evolve over time, no matter what. . . . .But the elucidation of goals can transcend incrementalism (and). . .result in the deliberate amendment of strategy or in curtailment of strategic erosion (Christensen et al., 1982: 553-554).

Use of strategic control converts the rational learning mode into a more sophisticated one. The model becomes a chain of cognition-trialcognition-trial etc. We will refer to it as strategic learning model (See Chapters 2.6, 2.9, 5.3 in Ansoff and McDonnell, 1990). Finally, Mintzberg attacks the rational model of learning by pointing out that it decouples strategy formulation from implementation, which causes organizational resistance and even failure of implementation. This point underlines the irony of Mintzberg's insistence on criticizing outdated original principles of the Design School without acquainting himself with their subsequent evolution. As discussed earlier in this paper, the problem of resistance to change has been recognized and treated back in the 1980s without abandoning explicit strategy formulation (Ansoff, Part 6, 1984). In summary, Mintzberg's 'proofs' that the rational model of learning does not apply to strategic management are contradicted by facts of management practice. And his insistence on universal use of the existential model invites managements to abdicate their role as strategic thinkers, and to confine their attention to optimizing the operating behavior of their organizations.

model is valid. It is curious because, as already discussed in this paper, Mintzberg does identify the context for the Design School Model. And in his other work he was one of the first researchers to call attention to the importance of contextual view of organizational structures (Mintzberg, 1979). His failure to identify the context for his own work is damaging because it exposes his model to counterexamples from the entire field of 'organizationatives' and from the complete range of organizational settings. As a result, in the absence of contextual limits, Mintzberg inadvertently ventures to make comments on contexts to which he has had little exposure. And yet, it is the opinion of this writer that, if streamlined and put into proper context, Mintzberg's model of strategy has demonstrable validity, both descriptively and preemptively, and represents an insightful and important contribution to Strategic Management. In the remainder of this paper I will describe the appropriate descriptive and prescriptive contexts for Mintzberg's model.

VALID CONTEXT FOR MINTZBERG'S PRESCRIPTIVE MODEL Modification of Mintzberg's Model A complete description of Mintzberg's Model was presented in this paper. From this model we abstract the following core concepts which can be shown to be valid in specified contexts. 1. To succeed, an organization should use the 'emergent strategy' trial and experience process of strategy formation. 2. No attempt should be made to formulate the firm's strategy in advance of the trial and experience process. 3. No formal organization-wide strategic planning should be used. 4. Except under special circumstances, the strategy which is implicit in the historical sequence of successful trials should not be made explicit.

Failure to identify relevant context

Description of the relevant research

The most curious and damaging aspect to Mintzberg's Model of strategy formation lies in his failure to identify the context in which his

The relevant empirical research which makes it possible to identify the context within which the above Model is a valid prescription, was briefly

Critique o f Henry Mintzberg

Critique of 'The Design School'

49

459

referred to earlier in this paper. A somewhat more detailed description follows. The research was addressed to testing the Strategic Success Hypothesis which was proposed by Ansoff in 1979. The Hypothesis states that an organization will optimize its success when the aggressiveness of its strategic behavior in the environment and its openness to the external environment are both aligned with the turbulence level of the organization's external environment.

identical with the type of strategic aggressiveness which was found to optimize firms' success in the extrapolative environment. Thus, empirical research described above shows that Mintzberg's Prescriptive Model is a valid prescription for organizations which seek to optimize their performance in environments in which strategic changes are incremental and the speed of the changes is slower than the speed of the organizational response.

The key contextual variable in this research was the concept of environmental turbulence which is an enlargement of the concepts of unpredictability and uncertainty used by Mintzberg. In the research, five distinctive levels of observable environmental turbulence were identified, ranging from stable to creative. For the purpose of identification of context it is useful to aggregate turbulence levels into two categories: (1) Incremental turbulence in which environmental changes are a logical evolution of the historical change process, and the speed of the changes is slower than the response time of the organizations; and (2) Discontinuous turbulence in which successive changes are discontinuous from the preceding ones, and speed of change is greater than the speed of the organizations' response.

It should be noted that, except for difference in the language (academic vs. practical), Mintzberg's model is identical to the injunction to firms to 'stick to their strategic knitting' which was offered in a world famous book The Search of Excellence by Peters and Waterman (Peters and Waterman, 1982). (It should further be noted that, while recommending conservative strategic behavior, Peters and Waterman recommend very aggressive competitive behavior by firms which aspire to succeed in extrapolative environments, a matter not mentioned by Mintzberg.) The size of the domain of applicability of Mintzberg's model to the business sector can be determined from an extensive unpublished survey by this author (which was briefly described in the introduction of this paper). According to the survey, roughly 20 percent of the firms in developed economies will need to use the Mintzberg/Peters/Waterman model in order to succeed in the 1990s.

To date the Strategic Success Hypothesis has been empirically tested in six different settings: 1. A cross-section of U.S. firms (Hatziantoniou, 1986) 2. Banks in United Arab Emirates (Salameh, 1987) 3. Public Service Organizations in the U.S. (Sullivan, 1987) 4. Parastatal firms in Algeria (Chabane, 1987) 5. Banks in San Diego County (Lewis, 1989) 6. Major U.S. banks (Jaja, 1990) In all six settings, the hypothesis was statistically sustained in all settings at 0.05 or better confidence level. And the levels of success in organizations which are aligned with the environment were substantially higher than in organizations which were out of alignment (Ansoff and Sullivan, 1990; Ansoff, Sullivan et al., 1990). The relevance of the research results to Mintzberg's model lies in the fact that Mintzberg's prescription for strategy formation is virtually

It must be mentioned that, in discontinuous environments, which constitute the remaining 80 percent of the sample, the research described above (and the aftermath of the Peters-Waterman research) both show that firms which persist in 'sticking to their strategic knitting' will not be among the successful performers and may jeopardize their own survival. Finally it is necessary to recognize that the context of the descriptive validity of Mintzberg's is much larger than the prescriptive. This context includes firms which are successful in the extrapolative business environments (in the business jargon those are called market driven firms); firms in discontinuous environments which are suffering from loss of competitiveness; and, in 1990, it includes a majority of the not-for-profit organizations in the U.S. Thus the paradox of a world-famous researcher opening himself to criticism could have been

50 460

H. I. Ansoff H. Igor

Ansoff

avoided if Henry had stuck to his own strategic knitting which is a deep knowledge of descriptive strategic behavior, particularly in not-for-profit organizations.

the 'emerging strategy' model is a valid prescription for success in incremented environments, a valid description of poorly performing firms in discontinuous environments, and a valid description of the behavior of a majority of not-for-profit organizations.

CONCLUSIONS In this paper, the thrusts of critique of Mintzberg's proofs and concepts were two: methodological weakness of the arguments, and contradiction to factual evidence. The conclusions of this critique are the following: 1. Mintzberg's proof that the Design School failed to adapt with times is methodologically unsound. 2. The assertion that other prescriptive schools shared their design principles with the Design School is factually inaccurate. 3. The assertion that the other prescriptive schools failed to adapt is factually inaccurate. 4. Because of the above conclusions, it is unproductive to address Mintzberg's specific criticisms of the Design School principles. 5. However, it is productive to critique the alternative to the Harvard Business School's design principles which is advanced by Mintzberg. 6. This critique finds that Mintzberg's proofs of his design principles are deficient on the following points: His 'self-denial' of knowledge of practice of strategic management in the business sector, which leads him to many assertions that are in direct contradiction to observable facts. Failure to meet validity tests for prescriptive and descriptive observations, which leads to unsupported claims for descriptions and arbitrary announcement of prescriptions. Use of a descriptive definition of strategy, which is different from the definition used in practice, which makes Mintzberg's conclusions appear contradictory to facts. Insistence on universal applicability of the existential learning model, which leads to assertions which contradict observable reality. Failure to specify the relevance context for his own model. By abstracting a set of coherent concepts from Mintzberg's model it is possible to show that

REFERENCES Andrews, K. R. The Concept of Corporate Strategy, Irwin, Homewood, IL, 1971, 2nd edn, 1980; 3rd edn, 1987. Ansoff, H. I. Corporate Strategy, McGraw-Hill, New York, 1965. Ansoff, H. I., R. J. Brandenburg, F. E. Portner and H. R. Radosevich. Acquisition: Behavior of US Manufacturing Firms 1946-65, Vanderbilt University Press, Nashville, TN, 1971. Ansoff, H. I. 'The concept of strategic management', Journal of Business Policy, 2(4), 1972, pp. 3-9. Ansoff, H. I., Roger P. Declerck and Robert L. Hayes. From Strategic Planning to Strategic Management, Wiley, New York, 1974. Ansoff, H. I. 'Corporate capability for managing change', SRI Business Intelligence Program, Research Report, No. 610, 1978. Ansoff, H.I. Strategic Management, MacMillan Press Ltd., London and Basingstoke, 1979. Ansoff, H. I., W. Kirsch and P. Roventa. 'Dispersed positioning in strategic portfolio analysis', EIASM Working Paper, 1980. Ansoff, H. I. Implanting Strategic Management, Prentice/Hall International, Englewood Cliffs, NJ, 1984. Ansoff, H. I. 'Competitive strategy analysis on the personal computer', Journal of Business Strategy, 6, Winter 1986, pp. 28-36. Ansoff, H. I. and P. Sullivan. 'Competitiveness through strategic response'. In Ralph Gilman (ed.), Making Organizations More Competitive: Constantly Improving Everything Inside and Outside the Organization, Jossey-Bass, San Francisco, CA, 1990. Ansoff, H. I., P. Sullivan, P. Hatziarttoniou, H. Chabane, R. Jaja, T. Salameh, A. Lewis, P. Wang and S. Djohar. 'Empirical validation of the strategic success hypothesis', paper in progress, 1990. Ansoff, H. I. and E. McDonnell. Implanting Strategic Management, 2nd edn. Prentice Hall, New York, 1990. Chabane, H. 'Restructuring and performance in Algerian state-owned enterprises: A strategic management study', Unpublished D.B.A. dissertation. United States International University, San Diego, 1987. Christensen, C. R., K. R. Andrews, J. L. Bower, R. G. Hamermesh and M. E. Porter. Business Policy: Text and Cases, Irwin, Homewood, IL, 6th edn, 1987. Drucker, P. F. Managing In Turbulent Times, Heinemann, London, 1990.

Critique of Henry Mintzberg Critique of 'The Design School' Hatziantoniou, P. 'The relationship of environmental turbulence, corporate strategic profile,and company performance. Unpublished D . B . A . dissertation. United States International University, San Diego, 1986. J a j a , R. M. 'Technology and banking: The Implications of technology myopia on banking financial performance, A strategic management analysis'. Unpublished D . B . A . dissertation. United States International University, San Diego, 1990. Lewis, A. 'Strategic posture and financial performance of the banking industry in California: A strategic management study', Unpublished D . B . A . dissertation, United States International University, San Diego, 1989. Mintzberg, H . The Structuring of Organizations, Prentice Hall, Englewood Cliffs, NJ, 1979. Mintzberg, H. 'Crafting strategy', Harvard Business Review, July-August 1987, pp. 66-75. Mintzberg, H. 'The Design School: Reconsidering the basic premises of strategic management'. Strategic

51 461

Management Journal, 11(3), 1990, pp. 171-195. Peters, T. and R. Waterman, In Search of Excellence. Harper and Row, New York, 1982. Salameh, T. T. 'Analysis and financial performance of the banking industry in United Arab Emirates: A strategic management study', Unpublished D . B . A . dissertation. United States International University, San Diego, 1987. Schwartz, P. 'Accepting the rise in forecasting', The New York Times Forum, September 2, 1990, p. 13. Steiner, G. A. and Η. Schòllhammer. 'Pitfalls in multi-national long range planning'. Long Range Planning, 8 (2), 1975, pp. 2-12. Sullivan, P. A. The relationship between proportion of income derived from subsidy and strategic performance of a federal agency under the Commercial Activities Program', Unpublished D . B . A . dissertation. United States International University, San Diego, 1987. Whitehead, Alfred North. The Function of Reason, Beacon Press, Boston, M A , 1962.

52

Κ. E. Weick

10

SUBSTITUTES FOR STRATEGY

Karl E. Weick

A little strategy goes a long way. Too much can paralyze or splinter an organization. That conclusion derives from the possibility that strategylike outcomes originate from sources other than strategy. Adding explicit strategy to these other tacit sources of strategy can be self-defeating and reduce effectiveness (Bresser and Bishop 1983). Thus, the focus of this chapter is substitutes for strategy. The model for this exercise is the concept in the leadership literature of substitutes for leadership (Kerr and Jermier 1978). Substitutes are conditions that either neutralize what leaders do or perform many of the same functions they would. Substitutes include characteristics of subordinates (ability, knowledge, experience, training, professional orientation, indifference toward organizational rewards), characteristics of the task (unambiguous, routine, provides its own feedback, intrinsically satisfying), and characteristics of the organization (high formalization, highly specified staff functions, closely knit cohesive groups, organizational rewards not controlled by leaders, spatial distance between subordinates and superiors). Leadership has less impact when one or more of these conditions obtains. It is not that the situation is devoid of leadership; rather, the leadership is done by something else. It seems reasonable to work analogically and investigate the extent to which it is possible to create substitutes for strategies. If pressed to define strategy, I am tempted to adopt DeBono's (1984: 143) statement that "strategy is good luck rationalized in hindsight," but I am also comfortable with a definition much like Robert Burgelmann's (1983)—namely, "strategy is a theory about the reasons for past and

Substitutes for Strategy 222

THE THEORETICAL CONTEXT OF STRATEGIC MANAGEMENT

current success of the firm." Both of my definitional preferences differ sharply from Chandler's (1962) classic definition of strategy—"The determination of the basic long-term goals and objectives of an enterprise, and the adoption of courses of action and the allocation of resources necessary for carrying out these goals." Definitions notwithstanding, I can best show what I think strategy is by describing an incident that happened during military maneuvers in Switzerland. The young lieutenant of a small Hungarian detachment in the Alps sent a reconnaissance unit into the icy wilderness. It began to snow immediately, snowed for two days, and the unit did not return. The lieutenant suffered, fearing that he had dispatched his own people to death. But the third day the unit came back. Where had they been? How had they made their way? Yes, they said, we considered ourselves lost and waited for the end. And then one of us found a map in his pocket. That calmed us down. We pitched camp, lasted out the snowstorm, and then with the map we discovered our bearings. And here we are. The lieutenant borrowed this remarkable map and had a good look at it. He discovered to his astonishment that it was not a map of the Alps, but a map of the Pyrenees. This incident raises the intriguing possibility that when you are lost, any old map will do. Extended to the issue of strategy, maybe when you are confused, any old strategic plan will do. Strategic plans are a lot like maps. They animate people and they orient people. Once people begin to act, they generate tangible outcomes in some context, and this helps them discover what is occurring, what needs to be explained, and what should be done next. Managers keep forgetting that it is what they do, not what they plan, that explains their success. They keep giving credit to the wrong thing—namely, the plan— and having made this error, they then spend more time planning and less time acting. They are astonished when more planning improves nothing. Kirk Downey has suggested that the Alps example is a success story for two quite specific reasons. First, the troops found a specific map that was relevant to their problem. Had they found a map of Disneyland rather than a map of the Pyrenees their problem would have deepened materially. Second, the troops had a purpose—that is, they wanted to go back to their base camp—and it was in the context of this purpose that the map took on meaning as a means to get them back. These conditions, however, do not negate the basic theme that meaning lies in the path of the action. A map of Disneyland makes it harder to develop a shared understanding of what has happened and where we have been, but if it does not inhibit action and observation, some clearer sense of the situation may emerge as action proceeds.

53

54

Κ. E. Weick SUBSTITUTES FOR CORPORATE STRATEGY

223

When I described the incident of using a map of the Pyrenees to find a way out of the Alps to Bob Engel, the executive vice president and treasurer of Morgan Guaranty, he said, "Now, that story would have been really neat if the leader out with the lost troops had known it was the wrong map and still been able to lead them back." What is interesting about Engel's twist to the story is that he has described the basic situation that most leaders face. Followers are often lost and even the leader is not sure where to go. All the leader knows is that the plan or the map he has in front of him is not sufficient by itself to get them out. What he has to do, when faced with this situation, is instill some confidence in people, get them moving in some general direction, and be sure they look closely at what actually happens, so that they learn where they were and get some better idea of where they are and where they want to be. If you get people moving, thinking clearly, and watching closely, events often become more meaningful. For one thing, a map of the Pyrenees can still be a plausible map of the Alps because in a very general sense, if you have seen one mountain range, you have seen them all (readers can test this assertion for themselves by examining " A Traveler's Map of the Alps" in the April 1985 issue of National Geographic Magazine). The Pyrenees share some features with the Alps, and if people pay attention to these common features, they may find their way out. For example, most mountain ranges are wet on one side and dry on the other. Water flows down rather than up. There is a prevailing wind. There are peaks and valleys. There is a highest point, and then the peaks get lower and lower until there are foothills. Just as it is true that if you have seen one mountain range you have seen them all, it also is true that if you have seen one organization you have seen them all. Any old plan will work in an organization because people usually learn by trial and error, some people listen and some people talk, people want to get somewhere and have some general sense of where they now are, 20 percent of the people will do 80 percent of the work (and vice versa), and if you do something for somebody, they are more likely to do something for you. Given these general features of most organizations, any old plan is often sufficient to get this whole mechanism moving, which then makes it possible to learn what is going on and what needs to be done next. The generic process involved is that meaning is produced because the leader treats a vague map or plan as if it had some meaning, even though he knows full well that the real meaning will come only when people respond to the map and do something. The secret of leading with a bad map is to create a self-fulfilling prophecy. Having predicted that the group will find its way out, the leader creates the combination of optimism

Substitutes for Strategy 224

THE THEORETICAL CONTEXT OF STRATEGIC MANAGEMENT

and action that allows people to turn their confusion into meaning and find their way home. There are plenty of examples in industry where vague plan and projects provide an excuse for people to act, learn, and create meaning. The founders of Banana Republic, the successful mail order clothier, started their business by acting in an improbable manner. They bought uniforms firom overthrown armies in South America and advertised these items in a catalog, using drawings rather than photographs. All of these actions were labeled poor strategy by other mail order firms. When these three actions were set in motion, however, they generated responses that no one expected (because no one had tested them) and created a belated strategy as well as a distinct niche for Banana Republic. Tuesday Morning, an off-price retailing chain that sells household and gift items, opens its stores when they have enough merchandise to sell and then closes them until they get the next batch. As managers followed this pattern, they discovered that customers love grand openings and that anticipation would build between closings over when the store would open again and what it would contain. These anticipations were sufficiently energizing that stores that opened intermittently for four to eight weeks sold more than equivalent stores that were open year round. The Microelectronics and Computer Technology Corporation (MCC) consortium in Austin, Texas, is a clear example of the sequence in which vague projects trigger sufficient action that vagueness gets removed. A key Texas state official described MCC as "an event, not a company." Bidding for MCC to locate in Texas became a vehicle to pull competing Texas cities together. It also became a vehicle to tell out-of-state people, "We are a national and an international force, not just a regional force, and not just a land of cowboys and rednecks." MCC became a tangible indication that Texas was growing, maturing, and on its way up. MCC's criteria for a good site became defining characteristics of what Austin was as a city, though Austinites did not realize they had this identity before. MCC said in its specifications that it did not want to locate where everyone thinks they know how high-tech R&D should be done. Texas thus "discovered" that its backwardness was in fact one of its biggest assets. Acquiring MCC became a strategy to strengthen Texas, but only quite late, when more and more problems were seen to be solved if it landed in Austin. The action of bidding for MCC fanned out in ways that people had not anticipated. The point is, if action is decoupled from strategy, then people have a better chance to be opportunistic, to discover missions and resources they had no idea existed. So far three themes have been introduced: (1) that action clarifies meaning; (2) that the pretext for the action is of secondary importance;

55

56

Κ. E. Weick S U B S T I T U T E S FOR CORPORATE STRATEGY

225

(3) and that strategic planning is only one of many pretexts for meaninggeneration in organizations. To clarify some ways in which action can substitute for strategy, we will look more closely at the dynamics of confidence and improvisation.

Confidence as Strategy In managerial work, thought precedes action, but the kind of thought that often occurs is not detailed analytical thought addressed to imagined scenarios in which actions are tried and options chosen. Instead, thought precedes action in the form of much more general expectations about the orderliness of what will occur. Order is present, not because extended prior analysis revealed it but because the manager anticipates sufficient order that she wades into the situation, imposes order among events, and then "discovers" what she had imposed. The manager "knew" all along that the situation would make sense. This was treated as a given. Having presumed that it would be sensible, the manager than acts confidently and implants the order that was anticipated. Most managerial situations contain gaps, discontinuities, loose ties among people and events, indeterminacies, and uncertainties. These are the gaps that managers have to bridge. It is the contention of this argument that managers first think their way across these gaps and then, having tied the elements together cognitively, actually tie them together when they act and impose covariation. This sequence is similar to sequences associated with self-fulfilling prophecies (see Snyder, Tanke, and Berscheid 1977). Thus presumptions of logic are forms of thought that are crucial for their evocative qualities. The presumption leads people to act more forcefully, the more certain the presumption. Strong presumptions (such as, "I know that these are the Pyrenees") lead to strong actions that impose considerable order. Weaker presumptions lead to more hesitant actions, which means either that the person will be more influenced by the circumstances that are already present or that only weak order will be created. Presumptions of logic are evident in the chronic optimism often associated with managerial activity. This optimism is conspicuous in the case of companies that are in trouble, but it is also evident in more run-of-the-mill managing. Optimism is one manifestation of the belief that situations will have made sense. William James (1956) described the faith that life is worth living that generates the action that then makes life worth living. Optimism is not necessarily a denial of reality. Instead it may be the belief that makes reality possible.

Substitutes for Strategy 226

THE THEORETICAL CONTEXT OF STRATEGIC MANAGEMENT

Presumptions of logic should be prominent among managers because of the climate of rationality in organizations (Staw 1980). Presumptions should be especially prominent when beliefs about cause and effect linkages are unclear (Thompson 1964: 336). Thompson labels the kind of managing that occurs when there are unclear preferences and unclear cause/effect beliefs inspiration. It is precisely in the face of massive uncertainty that beliefs of some sort are necessary to evoke some action, which can then begin to consolidate the situations. To inspire is to affirm realities, which then are more likely to materialize if they are sought vigorously. That sequence may be the essence of managing. Examples of the effect of presumptions are plentiful. A male who believes he is telephoning an attractive female speaks more warmly, which evokes a warm response from her, which confirms the original stereotype that attractive women are sociable (Snyder, Tanke, and Berscheid 1977). A new administrator, suspecting that old-timers are traditional, seeks ideas from other sources, which increases the suspicion of old-timers and confirms the administrator's original presumption (Warwick, 1975). People who presume that no one likes them approach a new gathering in a stiff, distrustful manner, which evokes the unsympathetic behavior they presumed would be there (Watzlawick, Beavin, and Jackson 1967: 9 8 - 9 9 ) . A musician who doubts the competence of a composer plays his music lethargically and produces the ugly sound that confirms the original suspicion ( Weick, Gilfillan, and Keith 1973 ). In each case, an initial presumption (she is sociable, they are uncreative, people are hostile, he is incompetent) leads people to act forcibly (talk warmly, seek ideas elsewhere, behave defensively, ignore written music), which causes a situation to become more orderly (warmth is exchanged, ideas emerge, hostility is focused, music becomes simplistic), which then makes the situation easier to interpret, thereby confirming the original presumption that it will have been logical. This sequence is common among managers because managerial actions are almost ideally suited to sustain self-fulfilling prophecies (Eden 1984). Managerial actions are primarily oral, face-to-face, symbolic, presumptive, brief, and spontaneous (McCall and Kaplan 1985). These actions have a deterministic effect on many organizational situations because those situations are less tightly coupled than are the confident actions directed at them. The situations are loosely coupled, subject to multiple interpretations, monitored regularly by only a handful of people, and deficient in structure. Thus a situation of basic disorder becomes more orderly when people overlook the disorder and presume orderliness, then act on this presumption, and finally rearrange its elements into a more meaningful arrangement that confirms the original presumption. It is suggested that typical

57

58

Κ. E. Weick SUBSTITUTES FOR CORPORATE STRATEGY

227

managerial behavior is more likely to create rather than disrupt this sequence. Thus, a manager's preoccupation with rationality may be significant less for its power as an analytic problemsolving tool than for its power to induce action that eventually implants the rationality that was presumed when the sequence started. The lesson of self-fulfilling prophecies for students of strategy is that strong beliefs that single out and intensify consistent action can bring events into existence (see Snyder 1984). Whether people are called fanatics, true believers, or the currently popular phrase idea champions, they all embody what looks like strategy in their persistent behavior. Their persistence carries the strategy; the persistence is the strategy. True believers impose their view on the world and fulfill their own prophecies. Note that this makes strategy more of a motivational problem than a cognitive forecasting problem. An argument can be made that the so-called computer revolution is an ideal exhibit of confidence as strategy. The revolution is as much vendor-driven as it is need-driven. The revolution can be viewed as solutions in search of problems people never knew they had. Vendors had more forcefulness, confidence, and focus than did their customers, who had only a vague sense that things were not running right, although they could not say why. Vendors defined the unease as a clear problem in control and information distribution, a definition that was no worse than any other diagnosis that was available. To say that it was IBM's strategy to be forceful is to miss the core of what actually happened. The key point is that IBM's strategy worked after it became self-confirming, when it put an environment in place. A common error is that the strategic plan is valued because it looks like it correctly forecast a pent-up demand for computers. Actually, it did no such thing. Instead, the plan served as a pretext for people to act forcefully and impose their view of the world. Once they imposed, enacted, and stabilized that view and once it was accepted, then more traditional procedures of strategic planning could be made to work because they were directed at more predictable problems in a more stable environment. What gets missed by strategy analysts is that proaction precedes reaction. Strategic planning works only after forceful action has hammered the environment into shape so that it is less variable and so that conventional planning tools can now be made to work. Because the constrained environment contains demands, opportunities, and problems that were imposed during proaction, proaction, not planning, predicts what the organization has to contend with. To see how self-fulfilling prophecies can mimic strategy and affect the direction of behavior, consider the problem of regulation. Although companies groan about the weight of regulation, data (McCaffrey 1982)

Substitutes for Strategy 228

THE THEORETICAL CONTEXT OF STRATEGIC MANAGEMENT

suggest that regulators do not have their act together and are loosely coupled relative to the tightly coupled organizations and lawyers they try to regulate. Thus, many organizations ironically create the regulators who control them. The way they do this is a microcosm of the point being made here about confidence. If a firm treats regulators as if they are unified and have their act together, then the firm gets its own act together to cope with the focused demands that are anticipated from the regulators. As the firm gets its act together it becomes a clearer target that is easier for the regulators to monitor and control. Concerted action undertaken by the firm to meet anticipated action from regulators now makes it possible for regulators to do something they could not have done when the firms were more diffuse targets. A confident definition of regulatory power, confidently imposed, stabilizes the regulation problem for a firm. The irony is that the faulty prophecy brings the problem into existence more sharply than it ever was before confident behavior was initiated. The firm has become easier to regulate by virtue of its efforts to prevent regulation. Environments are more malleable than planners realize. Environments often crystallize around prophecies, presumptions, and actions that unfold while planners deliberate. Guidance by strategy often is secondary to guidance by prophecies. These prophecies are more likely to fulfill themselves when they are in the heads of fanatics who work in environments where the definition of what is occurring can be influenced by confident assertions. Thus, presumptions can substitute for strategy. We assume co-workers know where they are going, they assume the same for us, and both of us presume that the directions in which we both are going are roughly similar. A presumption does not necessarily mean that whatever is presumed actually exists. We often assume that people agree with us without ever testing that assumption. Vague strategic plans help because we never have to confront the reality of our disagreements. And the fact that those disagreements persist undetected is not necessarily a problem because those very differences provide a repertoire of beliefs and skills that allow us to cope with changing environments. When environmental change is rapid, diverse skills and beliefs are the solution, not the problem.

Improvisation as Strategy Much of my thinking about organizations (such as Weick 1979) uses the imagery of social evolution, but there is a consistent bias in the way I use that idea. I consistently argue that the likelihood of survival goes up when variation increases, when possibilities multiply, when trial and

59

60

Κ. E. Weick SUBSTITUTES FOR CORPORATE STRATEGY

229

error become more diverse and less stylized, when people become less repetitious, and when creativity becomes supported. Notice that variation, trial and error, and doing things differently all imply that what you already know, including your strategic plan, is not sufficient to deal with present circumstances. When it is assumed that survival depends on variation, then a strategic plan becomes a threat because it restricts experimentation and the chance to learn that old assumptions no longer work. Furthermore, I assume that whatever direction strategy gives can be achieved just as easily by improvisation. Improvisation is a form of strategy that is misunderstood. When people use jazz or improvisational theater to illustrate improvising, they usually forget that jazz consists of variations on a theme and improvisational theater starts with a situation. Neither jazz nor improvisational theater are anarchic. Both contain some order, but it is underspecified. To understand improvisation as strategy is to understand the order within it. And what we usually miss is the fact that a little order can go a long way. For example, we keep underestimating the power of corporate culture because it seems improbable that something as small as a logo, a slogan, a preference (Geneen's one unshakable fact), a meeting agenda, or a Christmas party could have such a large effect. The reason these symbols are so powerful is that they give a general direction and a frame of reference that are sufficient. In the hands of bright, ambitious, confident people who have strong needs to control their destinies, general guidelines are sufficient to sustain and shape improvisation without reducing perceived control. If improvisation is treated as a natural form of organizational life, then we become interested in a different form of strategy than we have seen before. This newer form I will call a jwst-in-time strategy. Just-in-time strategies are distinguished by less investment in front-end loading (try to anticipate everything that will happen or that you will need) and more investment in general knowledge, a large skill repertoire, the ability to do a quick study, trust in intuitions, and sophistication in cutting losses. Like improvisation, a just-in-time strategy glosses, interprets, and enlarges some current event, gives it meaning, treats it as if it were sensible, and brings it to a conclusion. This form of activity looks very much like creating a stable small win (Weick 1984). And once an assortment of small wins is available, then these can be gathered together retrospectively and packaged as any one of several different directions, strategies, or policies. Strategies are less accurately portrayed as episodes where people convene at one time to make a decision and more accurately portrayed as

Substitutes for Strategy 230

THE THEORETICAL CONTEXT OF STRATEGIC MANAGEMENT

small steps (writing a memo, answering an inquiry) that gradually foreclose alternative courses of action and limit what is possible. The strategy is made without anyone realizing it. The crucial activities for strategymaking are not separate episodes of analysis. Instead they are actions, the controlled execution of which consolidate fragments of policy that are lying around, give them direction, and close off other possible arrangements. The strategy-making is the memo-writing, is the answering, is the editing of drafts. These actions are not precursors to strategy; they are the strategy. Strategies that are tied more closely to action are more likely to contain improvisations (Weiss 1980: 401): Many moves are improvisations. Faced with an event that calls for response, officials use their experience, judgment, and intuition to fashion the response for the issue at hand. That response becomes a precedent, and when similar— or not so similar—questions come up, the response is uncritically repeated. Consider the federal agency that receives a call from a local program asking how to deal with requests for enrollment in excess of the available number of slots. A staff member responds with off-the-cuff advice. Within the next few weeks, programs in three more cities call with similar questions, and staff repeat the advice. Soon what began as improvisation has hardened into policy.

Managers are said to avoid uncertainty, but one of the ironies implicit in the preceding analysis is that managers often create the very uncertainty they abhor. When they cannot presume order they hesitate, and this very hesitancy often creates events that are disordered and unfocused. This disorder confirms the initial doubts concerning order. What often is missed is that the failure to act, rather than the nature of the external world itself, creates the lack of order. When people act, they absorb uncertainty, they rearrange things, and they impose contingencies that might not have been there before. The presence of these contingencies is what is treated as evidence that the situation is orderly and certain.

Conclusion The thread that runs through this chapter is that execution is analysis and implementation is formulation. The argument is an attempt to combine elements from a linear and adaptive view of strategy, with a largely interpretive view (Chaffee 1985: 95). Any old explanation, map, or plan is often sufficient because it stimulates focused, intense action that both creates meaning and stabilizes an environment so that conventional analysis now becomes more relevant. Organizational culture becomes influential in this scenario because it affects what people expect will be orderly. These expectations, in turn, often become self-fulfilling. Thus

61

62

Κ. E. Weick SUBSTITUTES FOR CORPORATE STRATEGY

231

the adequacy of any explanation is determined in part by the intensity and structure it adds to potentially self-validating actions. More forcefiilness leads to more validation. Accuracy becomes secondary to intensity. Because situations can support a variety of meanings, their actual content and meaning is dependent on the degree to which they are arranged into sensible, coherent configurations. More forcefulness imposes more coherence. Thus those explanations that induce greater forcefulness often become more valid, not because they are more accurate but because they have a higher potential for self-validation. Applied to managerial activity, substitutes for strategy are more likely among executives because their actions are capable of a considerable range of intensity, the situations they deal with are loosely connected and capable of considerable rearrangement, and the underlying explanations that managers invoke (such as, "This is a cola war") have great potential to intensify whatever action is underway. All of these factors combine to produce self-validating situations in which managers are sure their diagnoses are correct. What they underestimate is the extent to which their own actions have implanted the correctness they discover. What managers fail to see is that solid facts are an ongoing accomplishment sustained as much by intense action as by accurate diagnosis. If managers reduce the intensity of their own action or if another actor directs a more intense action at the malleable elements, the meaning of the situation will change. What managers seldom realize is that their inaction is as much responsible for the disappearance of facts as their action was for the appearance of those facts. Gene Webb often quotes Edwin Boring's epigram: "Enthusiasm is the friend of action, the enemy of wisdom." Given the preceding arguments we can see reasons to question that statement. Enthusiasm can produce wisdom because action creates experience and meaning. Furthermore, enthusiasm can actually create wisdom when prophecies become selffulfilling and factual. One final example of a vague plan that leads to success when people respond to it and pay close attention to their response involves a religious ritual used by the Naskapi Indians in Labrador. Every day they ask the question, "Where should we hunt today?" That question is no different from, "Where is the base camp?" or "What should we do with these uniforms?" or "Should we open today?" or "Could this conceivably be the Silicon prairie?" The Naskapi use an unusual procedure to learn where they should hunt. They take the shoulder bone of a caribou, hold it over a fire until the bone begins to crack, and then they hunt wherever the cracks point. T h e surprising thing is that this procedure works. The Naskapi almost always find game, which is rare among hunting bands.

Substitutes for Strategy 232

THE THEORETICAL CONTEXT OF STRATEGIC MANAGEMENT

Although there are several reasons why this procedure works, one is of special interest to us: The Naskapi spend most of each day actually hunting. Once the cracks appear, they go where the cracks point. What they do not do is sit around the campfire debating where the game are today based on where they were yesterday. If the Naskapi fail to find any game, which is rare, they have no one in the group to blame for the outcome. Instead, they simply say that the gods must be testing their faith. The cracks in the bone get the Naskapi moving, just as the mountain paths drawn on the map get the soldiers moving, and just as high-tech backwardness gets Texans moving. In each case, movement multiplies the data available from which meaning can be constructed. Because strategy is often a retrospective summary that lags behind action, and because the apparent coherence and rationality of strategy are often inflated by hindsight bias, strategic conclusions can be misleading summaries of what we can do right now and what we need to do in the future. I do not suggest doing away with strategic plans altogether, but people can take a scarce resource, time, and allocate it between the activities of planning and acting. The combination of staffs looking for work, high-powered analytic MBAs, unused computer capability, the myth of quantitative superiority, and public pressure to account for everything in rational terms tempts managers to spend a great deal of time at their terminals doing analysis and a great deal less time anyplace else (Weick in press). It seems astonishing that one of the hottest managerial precepts to come along in some time (MBWA, management by walking around) simply urges managers to pull the plug on the terminal, go for a walk, and act like champions. One reason those recommendations receive such a sympathetic reception is that they legitimize key aspects of sensemaking that got lost when we thought we could plan meanings into existence. As we lost sight of the importance of action in sensemaking, we saw situations become senseless because the wrong tools were directed at them. Strategic planning is today's pretext under which people act and generate meanings and so is the idea of organizational culture. Each one is beneficial as long as it encourages action. It is the action that is responsible for meaning, even though planning and symbols mistakenly get the credit. The moment that either pretext begins to stifle action meaning will suffer, and these two concepts will be replaced by some newer management tool that will work, not for the reasons claimed but because it restores the fundamental sensemaking process of motion and meaning.

63

64

Κ. E. Weick S U B S T I T U T E S FOR C O R P O R A T E S T R A T E G Y

233

REFERENCES Bresser, R.K., and R.C. Bishop. 1983. "Dysfunctional Effects of Formal Planning: Two Theoretical Explanations." Academy of Management Review 8: 588-99. Burgelman, R.A. 1983. "A Model of the Interaction of Strategic Behavior, Corporate Context, and the Concept of Strategy. " Academy of Management Review 8:61 - 7 0 . Chaffee, E.E. 1985. "Three Models of Strategy." Academy of Management Review 10: 89-98. Chandler, A.D. 1962. Strategy and Structure. Cambridge, Mass.: MIT Press. de Bono, E. 1984. Tactics: The Art and Science of Success. Boston: Little, Brown. Eden, D. 1984. "Self-Fulfilling Prophecy as a Management Tool: Harnessing Pygmalion." Academy of Management Review 9: 64-73. James, W. 1956. "Is Life Worth Living?" In The Will to Believe, edited by W. James, pp. 32-62. New York: Dover. Kerr, S., and J.M. Jermier. 1978. "Substitutes for Leadership: Their Meaning and Measurement." Organizational Behavior and Human Performance 22: 375-403. McCaffrey, D.P. 1982. "Corporate Resources and Regulatory Pressures: Toward Explaining a Discrepancy." Administrative Science Quarterly 27: 398-419. McCall, M.W., Jr., and R.W. Kaplan. 1985. Whatever It Takes: Decision Makers at Work. Englewood Cliffs, N.J.: Prentice Hall. Snyder, M. 1984- "When Belief Creates Reality." In Advances in Experimental Social Psychology, Vol. 18, edited by L. Berkowitz, pp. 247-305. New York: Academic Press. Snyder, M., E.D. Tanke, and E. Berscheid. 1977. "Social Perception and Interpersonal Behavior: On the Self-Fulfilling Nature of Social Stereotypes." Journal of Personality and Social Psychology 35: 656-66. Staw, B.M. 1980. "Rationality and Justification in Organizational Life." In Research in Organizational Behavior, Vol. 2, edited by B.M. Staw and L.L. Cummings, pp. 45-80. Greenwich, Conn.: JAI. Thompson, J. D. 1964. "Decision-making, the Firm, and the Market. " I n New Perspectives in Organization Research, edited by W . W . Cooper, H.J. Leavitt, and M.W. Sheely II, pp. 334-48. New York: Wiley. Warwick, Donald P. 1975. A Theory of Public Bureaucracy: Politics, Personality, and Organization in the State Department. Cambridge, Mass. : Harvard University Press. Watzlawick, P., J.H. Beavin, and D.D. Jackson. 1967. Pragmatics of Human Communi' cation. New York: Norton. Weick, K.E. 1979. The Social Psychology of Organizing, 2d ed. Reading, Mass.: AddisonWesley. . 1984. "Small Wins: Redefining the Scale of Social Problems." American Psychologist 39: 40-49. . 1985. "Cosmos vs. Chaos; Sense and Nonsense in Electronic Contexts." Organizational Dynamics 14 (Autumn): 50-64. Weick, K.E., D.P. Gilfillan, and T. Keith. 1973. "The Effect of Composer Credibility on Orchestra Performance." Sociometry 36: 435-62. Weiss, C.H. 1980. "Knowledge Creep and Decision Accretion." Knowledge: Creation, Diffusion, Utilization 1(3): 381-404.

Planer und Inkrementalisten

65

Das Ordnungsschema Die Auseinandersetzung zwischen Ansoff und Mintzberg entbehrt nicht einer gewissen Schärfe und Polemik. Dieser Umstand mag dazu beigetragen haben, daß beide Autoren die Ansichten der jeweils anderen Seite etwas vereinfacht dargestellt haben. Mintzbergs Vorwurf beispielsweise, die Planer hätten es versäumt, ihre Modelle weiterzuentwickeln, ist faktisch unrichtig, wie ein Vergleich der beiden Planungsbücher Ansoffs von 1965 und 1984 schnell verdeutlicht. Ebenso vernachlässigt Ansoffs Darstellung des Mintzbergschen Inkrementalismus als einen Prozeß des „trail and error" die Tatsache, daß sich Mintzberg eher dem logischen Inkrementalismus Quinns verpflichtet fühlt als dem „disjointed incrementalism" Lindbloms (Quinn et al., 1988). Wichtiger als diese Mißverständnisse sind die Hinweise auf eine mögliche Konvergenz der Ansätze. Interessant ist, daß beide Autoren den jeweils angegriffenen Modellen eine begrenzte Gültigkeit zuerkennen, und zwar für essentiell gleichartige Situationen. Mintzberg argumentiert, daß synoptisches Planen in relativ stabilen und wenig komplexen Kontexten sinnvoll sein kann, während Ansoff gerade diese Kontexte als das Hauptanwendungsgebiet inkrementalen Planens ansieht. Es wäre eine lohnende Aufgabe zukünftiger Forschung, diese Aussagen wiederholten empirischen Tests zu unterziehen. Die Kontextbedingungen für eher synoptisches oder eher inkrementales Planen sind nicht ausreichend geklärt (Bresser & Bishop, 1983).9 In einer (hier nicht abgedruckten) Replik auf Ansoffs Kritik unterstreicht Mintzberg (1991) den Konvergenzgedanken. Er argumentiert, daß zum Verständnis des Phänomens Strategie sowohl das Intendierte (Formale, Synoptische) als auch das Emergente (Inkrementale) vonnöten sei.10 Als Bezugspunkt für die Einordnung der weiteren Beiträge dieses Buches wird ein synoptisches Planungsprozeßmodell verwendet. Hierfür sind vier Gründe ausschlaggebend: (1) Wie oben dargestellt, zeichnet sich eine Konvergenz zwischen den Planungs- und den Inkrementalismusansätzen ab. (2) In der Unternehmungspraxis hat sich der Planungsansatz weitgehend durchgesetzt (Esser et al., 1984; Hahn et al., 1990; Welge & Al-Laham, 1992: 25-31; Wheelen & Hunger, 1995: 41).

9

10

Einige empirische Studien liegen inzwischen vor, die zwar vom Ergebnis her widersprüchlich sind, insgesamt aber eher den synoptischen Planungsansatz zu unterstützen scheinen: Fredrickson und Mitchell (1984) und Fredrickson und Iaquinto (1989) konnten ermitteln, daß die synoptische Planung, wie von Mintzberg prognostiziert, in stabilen Umwelten positiv mit der Unternehmungsperformance zusammenhängt, während in instabilen Umwelten das Inkrementalmodell vorteilhafter ist. Demgegenüber stellt Eisenhardt (1989b; 1990) fest, daß sich erfolgreiche Unternehmungen in der sehr dynamischen Mikrocomputerbranche ebenfalls eines synoptischen Planungsmodells bedienen. Slevin & Covin (1997) ermitteln (in einer methodisch sehr sauber konzipierten Studie) positive Zusammenhänge zwischen der synoptischen Planung und der Unternehmungsperformance für instabile Umwelten jedoch für stabile Umweltsituationen eine Überlegenheit der inkrementalen Planung. Dean und Sharfman (1996) berichten, daß Manager, die rationale Planungstechniken einsetzen, effektivere strategische Entscheidungen fällen als Manager, die auf analytische Verfahren verzichten und statt dessen einen politischen, inkrementalen Entscheidungsprozeß verwenden. Vgl. hierzu auch Mintzberg und Waters (1985).

66

R. Κ. F. Bresser

(3) Der Grund für die praktische Verbreitung synoptischer Planungsmodelle dürfte nicht darin liegen, daß sich durch solche Verfahren intendierte Pläne hundertprozentig, d.h. ohne Veränderungen umsetzen lassen. Der Wert der Modelle liegt vielmehr darin, daß sie es erlauben, komplexe Problemstellungen zu strukturieren und zu rationalisieren (Schreyögg, 1984: 280f.). Als Orientierungshilfen zur Komplexitätsbewältigung kommt ihnen die wichtige Funktion zu, Handlungen auszulösen, deren Ergebnisse kontextabhängig interpretiert werden. Diese Interpretationen führen eventuell zu einer Revision ursprünglicher Pläne, auf jeden Fall aber zu weiteren Handlungen und Interpretationen, die alle dazu beitragen, die Inhalte von Unternehmungsstrategien zu präzisieren und fortzuentwickeln. Niemand hat die handlungsinduzierende Wirkung formaler Planungsmodelle anschaulicher dargestellt als Karl Weick: Sein Beispiel der Militärmanöver in der Schweiz zeigt, daß selbst ein falscher, schlechter Plan ausreichen kann, um einen Handlungs- und Interpretationsprozeß in Gang zu setzen, durch den sich eine erfolgreiche Strategie entwickeln und implementieren läßt. (4) Als Ordnungsschema ist ein synoptisches Planungsprozeßmodell in der Lage, alle Fragestellungen und Kontroversen der kontemporären (den „mainstream" betreffenden) Strategieforschung zuzuordnen. Dies gilt auch für die primären Forschungsinteressen der Vertreter des Inkrementalismus, wie z.B. die Auswirkungen von Macht, Politik und Organisationskulturen auf die Strategieformation. Für die weiteren Ausführungen dient eine modifizierte Fassung des strategischen Entscheidungsprozeßmodells von Wheelen und Hunger (1995: 41ff.) als (Zu)Ordnungsschema. Das Modell ist ein Klassiker der strategischen Managementlehre, und es ist in ähnlicher Fassung in fast allen Lehrbüchern des Strategischen Managements anzutreffen (vgl. z.B. Hofer & Schendel, 1978: 47; Schreyögg, 1984: 85). Das hier verwendete (modifizierte) Modell ist mit der Unterscheidung von sieben miteinander verbundenen Stufen differenziert genug, um die wesentlichen Forschungsfragen des Strategischen Managements behandeln zu können. Die sieben Stufen des Modells (Abb. 1) beinhalten: 1. Die Bewertung des Unternehmungserfolges, und zwar im Hinblick auf (a) relevante Finanzkennzahlen und (b) die Verwirklichung der Mission, längerfristiger Ziele und der intendierten Strategien. 2. Die Bewertung der strategischen Entscheidungsträger der Unternehmung. 3. Die Analyse der externen Umwelt zur Identifikation strategischer Faktoren, die als Chancen und Risiken klassifiziert werden (OT-Teil der S WOT-Analyse). 4. Die Analyse der internen Umwelt zur Identifikation strategischer Faktoren, die besondere Stärken und Schwächen der Unternehmung darstellen (SW-Teil der SWOTAnalyse). 5. Die Entwicklung, Evaluation und Auswahl der besten Strategiealtemativen im Lichte einer SWOT-Analyse. 6. Die Implementierung der ausgewählten Strategien. 7. Die Kontrolle der implementierten Strategien. Der weitere Verlauf des Buches beginnt mit der Stufe 2 des Ordnungsschemas, da Fragen der Bewertung des Unternehmungserfolges (Stufe 1) sinnvollerweise zusammen mit Fragen der strategischen Kontrolle (Stufe 7) behandelt werden. Kapitel 3 und 4 stellen

Bewertu Unternel erfolgs

Bewertung der strategischen Entscheidungsträger

Entwicklung, Bewertung, Auswahl der Strategien

Implementierung der Strategien

Strategische Kontrolle

Planer und Inkrementalisten

iί

1ί

¡

1 (Λ κ 60 iä 3c •O »? e

ν Ό

I

t-i υ .c o m OS σ\

on J3

Size (X2) Slack (X3) Performance (X4)

SP1 (Λ/ = 60)

SP2 (/V = 175)

SCI (/V = 60)

SC2 (/V = 175)

PC (/V = 175)

.33 (.41) .36 (.59) -.20 (.77) -.12 (.32)

.18* (.10) .12 (.17) .07 1.16) .08 (.09)

.01 (.17) .56* (.32) -.52* (.27)

.04 (-09) .38" (.17) -.18 (.12)

-.35· (.19) .34 (22) -.21 (.20)

-

-

-

Natural-Gas Distribution Industry (/V = 150) SP1 Tenure (X1 )

_

Size (X2)

-

Slack (X3)

-

Performance (X4)

-

SP2 -.06 (.13) -.34 (1.01) -.09 (.45) .17 (.24)

SCI

_ -

_ -

SC2

PC

.39" (.17) .05 (.24) -.10 (.12)

.34 (.21) -.17 (.24) -.30· (.18)

-

-

* ρ < .10; " p < .05; ~ p < .01. * Unstandardized regression coefficients; standard errors are in parentheses. Y = a + Ò1X1 + b2X2 + b3X3 + Ò4X4 + u.

Results generally support these predictions. Tenure coefficients in the computer industry were significant in the hypothesized direction in four of five cases. Tenure coefficients in the chemical industry were significant in the hypothesized direction only once, as was the case in the natural-gas industry. Computer-industry coefficients were significantly greater than those in the chemical industry in all regressions except SC1 (p < .01 for PC; ρ < .05 for SP1 and SP2; ρ < .10 for SC2; by Arnold's, 1982, "split groups" test). In addition, coefficients in the computer industry were greater than those in the natural-gas industry in two of three cases, significantly for SP2 (p < .01). Hence, the relationship between top-team tenure and firm persistence and conformity was strongest in the high-discretion computer industry. Estimates for the control variables indicate some interesting patterns. In both the chemical and natural-gas industries, slack was typically negative and in two of eight cases, significant. In contrast, slack was positive and significant (in three 496/ASQ, September 1990

142

S. Finkelstein/D. C. Hambrick Managerial Discretion

of five cases) in the high-discretion computer industry. This suggests a possible complementary effect, such that organizational sources of discretion, like slack, become important when environments confer limited latitude, while in higherdiscretion contexts they may actually constrain choice. As expected, size was positively associated with the dependent variables, and significant in three of 13 cases. An effective way to summarize the role of managerial discretion is to examine its incremental effect in consecutive regressions on tenure. Table 4 reports R 2 s of three regressions for each dependent variable: (1) on tenure (and the control variables), (2) on tenure (and the control variables) and the interaction of tenure and t w o industry dummy variables, and (3) on tenure (and the controls), industry χ tenure interactions, and the interactions of tenure with both size and slack. These results confirm those reported above for GLS models. Environmental and organizational sources of discretion improved explained variance significantly in all five models, providing strong support for hypotheses 2a and 2b. Regressions run on subsamples of high and low size and slack firms substantiated these results. Tenure coefficients were significant in the hypothesized direction in four out of five cases for both low-size and high-slack groups, while significant results were found in only one of five regressions in the high-size and lowslack subsamples. The results thus indicate considerable support for the hypotheses. In general, the longer a top team's tenure, (a) the more persistent, or unchanging, the firm's strategy, (b) the more the firm's strategy conformed to central tendencies of the industry, and (c) the more the firm's performance aligned with industry averages. However, the strength of these patterns differed depending on how much discretion the setting conferred on top managers. For example, the association between team tenure and organizational outcomes was substantially stronger in the high-discretion computer industry than in the lower-discretion chemical and natural-gas distribution industries. Similarly, team tenure and organizational outcomes bore a stronger correspondence in small and high-slack firms—those with substantial managerial discretion—than in large and low-slack firms.

Table 4 Results of OLS Regressions on Team Tenure in Firm*

Model I fl2 Model II R2 F for II vs. I Model III R2 F for III vs. II

SP1 (/V = 155)

SP2 (/V = 500)

SC1 (/V = 155)

SC2 (/V = 500)

PC (/V = 500)

.30 .33 7.70"· .38 5.61·"

.25 .31 19.37·" .33 6 26"·

.16 .19 5.26" .22 2.95·

.11 .12 3.57" .14 1.99·

.03 .06 6.20·" .09 3.64"·

• p < .10; -p < .05; · ~ ρ < .01. • Model I: Y = a + b1X1 + ¿>2X2 + b3X3 + M X 4 + b5X5 + Ü6X6 + u; Model II: Y = a + b1X1 + b2X2 + b3X3 + Ò4X4 + Ò5X5 + b6X6 + b7X1X5 + b8X1X6 + u; Model III: Y = a + b1X1 + b2X2 + b3X3 + b4X4 + b5X5 + b6X6 + b7X1X5 + b8X1X6 + b9X1X2 + Û10X1X3 + u; where X1 = tenure, X2 = size. X3 = slack, X4 = performance; X5 = 1 if computer industry, = 0 if other industry; X6 = 1 if chemical industry, = 0 if other industry; and u = error term.

497/ASQ, September 1990

Top-Management-Team Tenure and Outcomes

143

DISCUSSION Effects of Team Tenure As a straightforward upper-echelons test, this paper sheds new light on the possible effects that top-team tenures may have on the strategies and performance of organizations. Our results reveal the organizational outcomes of escalating commitment, risk aversion, and restrictions in information processing that develop as top-managerial tenure increases. Namely, long-tenure teams do not engage in as much strategic experimentation and change as short-tenure teams. Long-tenure teams tend to pursue imitative strategies directly in line with industry trends, whereas short-tenure teams tend to pursue novel strategies that deviate widely from industry patterns. Correspondingly, organizations with long-tenure teams exhibit organizational performance that closely adheres to industry averages, while short-tenure teams are associated with performance levels that deviate—being either much higher or lower—from industry tendencies. These findings, particularly for persistence, are consistent with prior findings that outside successors tend to undertake substantial strategic change (e.g., Helmich and Brown, 1972). In fact, one could ask whether our results are due strictly to extreme behaviors by brand-new, short-tenure teams. W e explored this possibility by examining the data by grouped gradations of team tenures. As Figure 1 (a,b,c) illustrates, the results strikingly indicate that the organizational attributes vary systematically across the full spectrum of tenure lengths. A new interpretation can thus be placed on the conclusions of prior researchers regarding outsiders. Namely, what is most important theoretically about outsiders is not that they are outsiders but that they have very short tenures. Outsiders are simply at an extreme on a continuum; they are not qualitatively special cases. As their tenures increase, almost with each passing year, they gradually tend to make fewer strategic changes, to imitate or match strategic tendencies of the industry, and, accordingly, to be associated with firm performance that rises and falls in line with industry fortunes. Topteam tenure appears to be a construct of fundamental and widespread importance to organizations. Discretion as a Moderator We went beyond the basic upper-echelons model (Hambrick and Mason, 1984) by examining managerial discretion as a potentially important moderator of the association between executive characteristics and organizational outcomes. Results were strengthened, at times substantially, when managerial discretion was accounted for. Explained variance was significantly increased in all ten regressions by adding either industry or organizational indicators of discretion. Major differences were found across different levels of both environmental and organizational discretion. The characteristics of top-management teams were most strongly reflected in organizational outcomes in the computer industry, which provided significant latitude of action. Conversely, the natural-gas distribution industry was more restrictive—a commodity-like product, capital intensive, mature, highly regulated—and executive-team tenure was typically unrelated to strategic persistence and conformity. This finding suggests that choice of 498/ASQ, September 1990

144

S. Finkelstein/D. C. Hambrick Managerial Discretion Figure 1. Graphs of organizational outcomes at different gradations of team tenure.*

0-5

5-10

10-15

15-20

20-25

25-30

30-35

35-40

Team Tenure (in Years)

-2 Γ

LL 0-5

J_ 5-10

10-15

_L 15-20

•

·

_L 20-25

_L 25-30

30-35

_L 35-40

_L 25-30

_1_ 30-35

J_ 35-40

Team Tenure (in Years}

£ E

o 8 0} (O o — ç Qα α _

- 2

-3

-, 0-5

5-10

10-15

_L 15-20

_L 20-25

Team Tenure (in Years)

• For Strategic Conformity and Perform a n c e Conformity, larger negative scores indicate less conformity.

499/ASQ, September 1990

Top-Management-Team Tenure and Outcomes industry in tests of upper-echelons theory is a critical decision that requires careful consideration. Moreover, it implies that top-managerial staffing decisions may have very different consequences depending on industry. In high-discretion contexts, as in the computer industry, managers seem to matter greatly. In contrast, top teams in low-discretion contexts, such as in the natural-gas distribution industry, may have much less bearing on organizational outcomes. Managerial discretion was also assessed at the organizational level of analysis. Again, results supported the hypotheses. When the organization allowed significant latitude to top managers, as evidenced by abundant slack or small size, the firm's degree of strategic persistence and conformity typically reflected the tenure of the top team. Conversely, in more restrictive organizations—those that were large or had little slack—managerial profiles showed little association with organizational outcomes. The importance of organizational discretion in explaining strategic decision patterns suggests that it does have a role to play in upper-echelons theory. Some organizations afford their top managers more opportunity for impact than others. As found here, and as argued by Hambrick and Finkelstein (1987), organization size and slack seem to be among the critical determinants of discretion. Limitations Several important limitations of the study warrant discussion. First, we restricted our focus to top managers when, in fact, lower-level employees may be influential in professional firms or in those with emergent strategies. However, the highest managers and the CEO retain considerable symbolic influence that can convey their preferences for lower-level initiatives (Hage and Dewar, 1973; Pfeffer, 1981). Additionally, agenda setting at the top serves as an important guide for lower-level managers to follow (Kotter, 1982). While influence may emanate from numerous parts of the organization, in this study we have at least expanded our scope beyond the predominant focus on the CEO alone. Second, our sample was restricted to large firms, and our findings therefore may not be generalizable to all firms in an industry. Managerial effects may be more intense in smaller firms because they are less constrained by organizational inertia (Miller, Kets de Vries, and Toulouse, 1982). Another important limitation of the study concerns causality. While hypotheses were stated in associational terms, the logic behind them implies that managerial tenure "causes" organizational outcomes. Unfortunately, even with longitudinal data it was not possible to completely disentangle causal direction. While the temporal ordering of team tenure and the strategic persistence measures was consistent with theory, hypotheses on conformity were tested at one point in time. Accordingly, it seems plausible that both managerial tenures and persistence and conformity are mutually reinforcing, a view very much in line with Miles and Snow (1978). Long-tenured teams follow stable strategies, and firms that follow stable strategies do not make many changes in executive leadership. 500/ASQ, September 1990

145

146

S. Finkelstein/D. C. Hambrick Managerial Discretion

A fourth limitation concerns our measures of managerial discretion. While w e used industry to represent environmental sources of discretion, the construct may require more subtle operationalization. For example, industries may not be consistently high or low on every dimension that makes up environmental discretion (Hambrick and Finkelstein, 1987). Nevertheless, industry has been considered an adequate measure of environment by many other researchers (e.g., Lawrence and Lorsch, 1967; Hrebiniak and Snow, 1980; Porter, 1980), and the industries studied in this paper do differ substantially across several key indicators of environmental discretion. In a related vein, this paper did not address individual sources of discretion. Finally, to study environmental sources of discretion more effectively, a greater number of industries will be needed. While important patterns emerged in our examination of three widely disparate industries, our industry sample size was only three, leaving considerable opportunity for other researchers. CONCLUSIONS Our results allow conclusions at two levels. First, it appears that managerial team tenure has a profound influence on organizational outcomes, such as strategic persistence, strategic conformity, and performance conformity. The apparent effect of tenure occurs over its full range of values and is not just an artifact of the radical changes brought on by top managers newly arrived from the outside. At a second level, our results indicate that upper-echelons theory should be extended to include a moderating role for managerial discretion. As Pfeffer and Salancik (1978: 245) have argued, "it is necessary to develop a model that suggests how much constraint an administrator faces in formulating action." This paper illustrates that such a position has strong empirical support, at least in the context of managerial tenure. There is clearly opportunity for further research to investigate both managerial discretion and upper-echelons theory. The results reported here addressed managerial tenure; other important demographic characteristics such as age, functional background, industry experience, and education may yield significant findings as well. This paper also suggests that a more careful examination of the strategic choice paradigm is required. Such research will greatly improve our understanding of the role of executive leadership in organizations.

REFERENCES Aldrich, Howard 1979 Organizations and Environments. Englewood Cliffs, NJ: Prentice-Hall. Arnold, Hugh J. 1982 "Moderator variables: A classification of conceptual, analytic, and psychometric issues." Organizational Behavior and Human Performance, 29: 143-174.

Bantel, Karen, and Susan Jackson 1989 "Top management and innovations in banking: Does the composition of the top team make a difference?" Strategic Management Journal, 10: 107-124. Barnard, Chester 1938 Functions of the Executive Cambridge, MA: Harvard University Press.

501/ASQ, September 1990

Bourgeois, L. J. 1981 "On the measurement of organizational slack." Academy of Management Review, 6: 29-39. Brown, M. Craig 1982 "Administrative succession and organization performance: The succession effect." Administrative Science Quarterly, 27: 1 - 1 6 .

Top-Management-Team Tenure and Outcomes Chandler, Alfred D. 1962 Strategy and Structure. Cambridge, M A : M.l.T. Press. Coffee, J o h n C. 1988 "Shareholders versus managers: The strain in the corporate w e b . " In John C. Coffee, Louis Lowenstein, and Susan Rose-Ackerman (eds), Knights, Raiders, and Targets: 77-134. New York: Oxford University Press. Cyert, Richard M „ and James G. March 1963 A Behavioral Theory of the Firm. Englewood Cliffs, NJ: Prentice-Hall. Dearborn, D. C „ and Herbert A. Simon 1958 "Selective perceptions: A note on the departmental identification of executives." Sociometry, 21: 140-144. Finkelstein, Sydney 1988 "Managerial orientations and organizational outcomes: The moderating roles of managerial discretion and power." Unpublished Ph.D. dissertation, Columbia University. Fox, Frederick V., and Barry M. Staw 1979 "The trapped administrator: Effects of job insecurity and policy resistance upon commitment to a course of action." Administrative Science Quarterly, 24: 449-471. G r i m m , Curtis M., and Ken G. Smith 1986 "The organization as a reflection of its top managers: An empirical test." Paper presented at the annual meeting of the Academy of Management, Chicago. Gupta, Anil K., and V. Govindarajan 1984 "Business unit strategy, managerial characteristics, and business unit effectiveness at strategy implementation." Academy of Management Journal, 27: 2 5 - 4 1 . Hage, Jerald, and Robert Dewar 1973 "Elite values versus organizational structure in predicting innovation." Administrative Science Quarterly, 18: 279-290. Hambrick, Donald C„ and Sydney Finkelstein 1987 "Managerial discretion: A bridge between polar views on organizations." In L. L. Cummings and Barry M. Staw (eds ). Research in Organizational Behavior. 9: 369-406. Greenwich, CT: JAI Press.

Hambrick, Donald C., and Phyllis Mason 1984 "Upper echelons: The organization as a reflection of its top managers." Academy of Management Review, 9: 193-206. Hannan, Michael T., and John H. Freeman 1977 "The population ecology of organizations." American Journal of Sociology, 82: 929-964. 1984 "Structural inertia and organizational change." American Sociological Review, 49: 149-164. Hannan, Michael T., and Alice A. Young 1977 "Estimation in multi-wave panel models: Results on pooling cross-section and time series." In D. Heise (ed.). Sociological Methodology: 5 2 - 8 3 . San Francisco: Jossey-Bass. Helmich, Donald, and W. B r o w n 1972 "Successor type and organizational change in the corporate enterprise." Administrative Science Quarterly, 17: 371-381 Hrebiniak, Lawrence G „ and W i l l i a m F. Joyce 1985 "Organizational adaptation: Strategic choice and environmental determinism." Administrative Science Quarterly, 30: 336-349. Hrebiniak, Lawrence G „ and Charles G. S n o w 1980 "Strategy, distinctive competence, and organizational performance." Administrative Science Quarterly, 25: 307-335. Katz, Ralph 1982 "The effects of group longevity on project communication and performance." Administrative Science Quarterly, 27 : 81-104. Kets de Vries, Manfred F. R„ and Danny Miller 1986 "Personality, culture and organization." Academy of Management Review, 11 : 266-279. Kimberly, J o h n R., and Michael J. Evanisko 1981 "Organizational innovation: The influence of individual, organizational, and contextual factors on hospital adoption of technological and administrative innovations." Academy of Management Journal, 24: 689-713.

502/ASQ, September 1990

147

Kotter, J o h n P. 1982 The General Managers. New York: Free Press. Lawrence, Paul R., and Jay W. Lorsch 1967 Organization and Environment. Homewood, IL: Irwin. Levine, Sol, and Paul E. White 1961 "Exchange as a conceptual framework for the study of interorganizational relationships." Administrative Science Quarterly, 5: 583-601. Lieberson, Stanley, and James F. O'Connor 1972 "Leadership and organizational performance: A study of large corporations." American Sociological Review, 37: 117-130. Mace, M. L. 1971 Directors: Myth and Reality. Boston: Harvard University Graduate School of Business Administration M addala, G. S. 1977 Econometrics. New York: McGraw-Hill. March, James C., and James G. March 1977 "Almost random careers: The Wisconsin school superintendency, 1940-1972." Administrative Science Quarterly, 22: 377-409. March, James G., and Herbert A. Simon 1958 Organizations. New York: Wiley. Masters, Marick F., and J o h n T h o m a s Delaney 1985 "The causes of union political involvement: A longitudinal analysis." Journal of Labor Research, 6: 341 - 3 6 2 . Miles, Robert H „ and Kim S. Cameron 1982 Coffin Nails and Corporate Strategies. Englewood Cliffs, NJ: Prentice-Hall. Miles, Robert E„ and Charles C. Snow 1978 Organization Strategy. Structure and Process. New York: McGraw-Hill Miller, Danny 1988 "Stale in the saddle: CEO tenure and adaptation." Working paper, McGill University. Miller, Danny, and Cornelia Droge 1986 "Psychological and traditional determinants of structure." Administrative Science Quarterly, 31: 539-560.

148

S. Finkelstein/D. C. Hambrick Managerial Discretion

M i l l e r , D a n n y , a n d Peter H. Friesen 1980 " M o m e n t u m and revolution in organizational adaptation." Academy of Management Journal, 23: 5 9 1 - 6 1 4 . M i l l e r , Danny, M a n f r e d F. R. Kets de Vries, a n d J e a n - M a r i e T o u l o u s e 1982 "Top executive locus of control and its relationship to strategy-making, structure, and environment." Academy of Management Journal, 25: 237-253. Miller, Danny, and Jean-Marie Toulouse 1986 "Strategy, structure, CEO personality and performance in small f i r m s . " American Journal of Small Business, 10(3): 47-62. M i n t z b e r g , Henry 1973 The Nature of Managerial Work. N e w York: Harper & Row. 1978 "Patterns in strategy formation." Management Science, 24: 9 3 4 - 9 4 9 . Pfeffer, J e f f r e y 1981 " M a n a g e m e n t as symbolic action: The creation and maintenance of organizational paradigms." In L. L. Cummings and Barry M . Staw (eds.l, Research in Organizational Behavior, 3: 1 - 5 2 . Greenwich, CT: JAI Press. 1983 "Organizational demography." In L. L. Cummings and Barry M . Staw leds.). Research in Organizational Behavior, 5: 2 9 9 - 3 5 7 . Greenwich, CT: JAI Press. Pfeffer, J e f f r e y , a n d Gerald R. Salancik 1978 The External Control of Organizations: A Resource Dependence Perspective. N e w York: Harper & Row. Porter, M i c h a e l E. 1980 Competitive Strategy: Techniques for Analyzing Industries and Competitors. Glencoe, IL: Free Press. Ranson, S t e w a r t , B o b H i n i n g s , a n d Royston Greenwood 1980 "The structuring of organizational structures." Administrative Science Quarterly, 25: 1-17. Salancik, Gerald R. 1977 " C o m m i t m e n t and the control of organizational behavior and belief." In Barry M. Staw and Gerald R. Salancik (eds), N e w Directions in Organizational Behavior: 1 - 5 4 . Chicago: St. Clair Press.

Salancik, G e r a l d R „ a n d J e f f r e y Pfeffer 1977 "Constraints on administrator discretion: The limited influence of mayors on city budgets." Urban Affairs Quarterly, 12: 4 7 5 - 4 9 8 .

Szilagyi, A n d r e w D., a n d D a v i d M . Schweiger 1984 " M a t c h i n g managers t o strategies: A review and suggested f r a m e w o r k . " Academy of Management Review, 9: 626-637.

S c h e n d e l , Dan E „ a n d G. Patton 1978 " A simultaneous equation model of corporate strategy." Management Science, 24: 1611-1621.

T h o m p s o n , J a m e s D. 1967 Organizations in Action. N e w York: McGraw-Hill.

Scherer, F. M. 1970 Industrial Market Structure and Economic Performance. Chicago: Rand McNally. Schoeffler, S y d n e y , Robert Buzzell, a n d D. Heany 1974 "The impact of strategic planning on profit performance." Harvard Business Review, 52: 137-145. S c h w e n k , Charles, a n d M i n g - J e Tang 1987 "Economic and psychological explanations for strategic persistence." Working paper, Graduate School of Business, Indiana University. Singh, Jitendra 1986 "Performance, slack, and risk taking in organizational decision making." Academy of Management Journal, 29: 562-585. S t a w , Barry M . 1976 "Knee-deep in the Big Muddy: A study of escalating commitment to a chosen course of action." Organizational Behavior and Human Performance. 16: 2 7 - 4 4 S t a w , Barry M., a n d Frederick V. Fox 1977 " S o m e determinants of commitment to a previously chosen course of action." Human Relations, 30: 431 - 4 5 0 . S t a w , Barry M., a n d J e r r y Ross 1987 "Behavior in escalating situations: Antecedents, prototypes, and solutions." In L. L. Cummings and Barry M. Staw (eds.), Research in Organizational Behavior, 9: 3 9 - 7 8 . Greenwich, CT: JAI Press. Stearns, T i m o t h y M., A l a n N. H o f f m a n , a n d J a n B. Heide 1987 "Performance of commercial television stations as an outc o m e of interorganizational linkages and environmental conditions." Academy of Management Journal, 30: 7 1 - 9 0 .

503/ASQ, S e p t e m b e r 1990

T u s h m a n , M i c h a e l L „ a n d Elaine Romanelli 1985 "Organization evolution: A metamorphosis model of convergence and reorientation." In L. L. Cummings and Barry M. Staw (eds.), Research in Organizational Behavior, 7: 1 7 1 - 2 2 2 . Greenwich, CT: JAI Press. T u s h m a n , M i c h a e l L., Beverly V i r a n y , a n d Elaine R o m a n e l l i 1985 "Executive succession, strategy reorientation, and organization evolution." Technology in Society, 7: 2 9 7 - 3 1 4 . Vancil, Richard F. 1987 Passing the Baton: Managing the Process of CEO Succession. Boston: Harvard Business School Press. V e n k a t r a m a n , N,, a n d V a s u d e v a n Ramanujam 1986 " M e a s u r e m e n t of performance in strategy research: A comparison of approaches." Academy of Management Review, 11: 8 0 1 - 8 1 4 . W a l s h , J a m e s P. 1988 "Selectivity and selective perception: An investigation of managers' belief structures and information processing." Academy of Management Journal, 31: 8 7 3 - 8 9 6 . W a n o u s , J. P. 1980 Organizational Entry: Recruitment, Selection, and Socialization of Newcomers. Reading, M A : Addison-Wesley. Weick, Karl E. 1979 The Social Psychology of Organizing, 2d ed. Reading, M A : Addison-Wesley.

Strategie Sensemaking and Performance

149

° Academy of Management Journal 1993, Vol. 36, No. 2, 239-270.

STRATEGIC SENSEMAKING AND ORGANIZATIONAL PERFORMANCE: LINKAGES AMONG SCANNING, INTERPRETATION, ACTION, AND OUTCOMES JAMES B. THOMAS SHAWN M. CLARK DENNIS A. GIOIA Pennsylvania State University This study investigated the strategic "sensemaking" processes of scanning, interpretation, and action and how those activities are linked to organizational performance. Using path analyses on data from 156 hospitals, we tested the direct and indirect effects among these sensemaking processes and performance outcomes and developed a model of their relationships. In a more general sense, the research represents an attempt to provide insight not only into relationships between cognition and action, but also into the links between those fundamental processes and organizational performance outcomes.

The exploration of relationships between cognition and action has gained increasing prominence in organizational studies in recent years. If cognition and action are linked, however, it is intuitively apparent that both should be related to performance. Performance linkages to cognition and action, although clearly important, have received considerably less study than those between cognition and action, and the existing work in this domain has been directed mainly at individual and group-level outcomes. An important area of concern that has received even less empirical attention is the investigation of the relationships among cognition, action, and organizational performance. Studies dealing with cognitive processes have explored the effects of antecedent and contextual factors on decision makers' interpretations of information (e.g., Bateman & Zeithaml, 1989; Thomas & McDaniel, 1990) and the effects of strategic interpretation on organizational action (e.g., Meyer, 1982; Smart & Vertinsky, 1984). Other relevant work has focused mainly on the relationship between group schémas and decision makers' attributions of causality and their subsequent decisions (e.g., Clapham & Schwenk, 1991) or

Thanks to Charles Snow, David Ketchen, Glenn Firebaugh, Reuben McDaniel, and this journal's reviewers for their helpful comments and guidance. We gratefully acknowledge the research support of Roger Harvey and the Healthcare Financial Management Association. 239

150

J. Β. Thomas/S. M. Clark/D. A. Gioia

240

Academy of Management Journal

April

has used laboratory settings to test hypothesized interpretation-outcome linkages at an individual level (e.g., Fredrickson, 1985; Gioia & Sims, 1986; Walsh, Henderson, & Deighton, 1988). There has been, however, a notable lack of empirical work seeking to link "sensemaking"—which subsumes the key cognition-action processes of environmental scanning, interpretation, and associated responses (Gioia & Chittepeddi, 1991; Weick, 1979)—with organizational performance. In the strategic context, increasing emphasis has been placed on understanding the link between how top managers make sense of information and how they act to influence organizational outcomes (Dutton, Fahey, & Narayanan, 1983; Dutton & Jackson, 1987; Nystrom & Starbuck, 1984). Because modern organizational environments are complex and dynamic, a key role of top management has become providing meaningful interpretations for patterns of ambiguous information. Indeed, the imposition of meaning on issues characterized by ambiguity has become a hallmark of the modern top manager (cf. Smircich & Stubbart, 1985; Weick, 1979). Those interpretations are often seen as critical to the success and even the survival of organizations, mainly because of their implications for influencing action alternatives and subsequent outcomes (Dutton & Duncan, 1987). This study was intended to articulate and empirically investigate linkages between sensemaking processes and organizational performance in the strategic arena to provide insight into what distinguishes organizations that flourish from those that flounder within the same competitive environment. We tested and refined hypothesized relationships among scanning, interpretation, action, and organizational performance. Overall, we posed the question, How are key cognitive processes and the associated actions of top managers linked to organizational performance? CONCEPTUAL FRAMEWORK Researchers have developed a number of models to describe the way managers and organizations deal with potentially significant information (e.g., Dutton & Duncan, 1987; Kiesler & Sproull, 1982; Meyer, 1982). Daft and Weick (1984) and Milliken (1990) proposed that organizational adaptation entails three key processes—scanning, interpreting, and responding. These are all important aspects of the more general notion of sensemaking, which involves the reciprocal interaction of information seeking, meaning ascription, and action (cf. Gioia & Chittipeddi, 1991; Weick, 1979). Each element of this sensemaking process is presumed to have some relationship to performance. Thus, the basic model underlying this study can be represented as a scanning-interpretation-action-performance sequence. Scanning Scanning involves information gathering; it usually is considered an antecedent to interpretation and action (Daft & Weick, 1984; Hambrick,

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

151 241

1982). Top decision makers typically have access to far more information than they can actually use (Mintzberg, 1973); therefore, they are necessarily selective in attending to specific information employed to develop and interpret strategic issues (Huber & Daft, 1987). Researchers have most often defined scanning as searching the external environment to identify important events or issues that might affect an organization (Daft & Weick, 1984; Kiesler & Sproull, 1982; Milliken, 1990). From a strategic perspective, however, scanning also involves searching an organization's internal environment to identify important elements that might bear on future performance (e.g., Cowan, 1986). Dutton and Duncan (1987) viewed these search activities as activators of strategic issue interpretation and action. Interpretation Interpretation involves the development or application of ways of comprehending the meaning of information; it entails the fitting of information into some structure for understanding and action (Gioia, 1986; Taylor & Crocker, 1981). Researchers have usually viewed interpretation as an individual-level process wherein people attend to and ascribe meaningful labels to incoming information (cf. Cantor & Mischel, 1979; Smart & Vertinsky, 1984; Taylor & Fiske, 1978). At a different level of conceptualization, however, organizations themselves can be viewed as interpretation systems (Daft & Weick, 1984). In this sense, different areas and levels of an organization may be involved in the scanning activities associated with sensemaking, but it is top managers who have primary influence over which strategic issues are attended to and how they are labeled (Bartunek, 1984; Hambrick & Mason, 1984). Meanings attached to strategic issues are often the result of the categories that decision makers use. Rosch (1978) defined categories as cognitive classifications that group objects, events, and the like with similar perceived attributes. When decision makers use particular labels to describe a given issue (cf. Dutton & Jackson, 1987), the labels initiate a categorization process that affects the subsequent cognitions and motivations of the decision makers. Two of the most salient strategic issue labels are "opportunity" and "threat" (Dutton & Duncan, 1987; Mintzberg, Raisinghani, & Theoret, 1976; Nutt, 1984; Perrow, 1970). Jackson and Dutton (1988) showed that three dimensions differentiate those strategically relevant interpretation labels: (1) whether decision makers evaluate an issue in positive or negative terms, (2) whether they see it as representing potential gain or loss for their organization, and (3) whether they see it as controllable or uncontrollable. Thomas and McDaniel (1990) verified the relevance of these labels for describing strategic issues but also noted that because the positive-negative and gainloss dimensions were operationally indistinguishable, and were in fact highly correlated (r = .90), they should be collapsed into a single dimension. We followed that suggestion in formulating hypotheses in this study and labeled the composite variable "positive-gain." A basic premise of research on interpretation in organizations is that the

152

J. Β. Thomas/S. M. Clark/D. A. Gioia

242

Academy of Management Journal

April

way a strategic issue is labeled or framed mobilizes action in a particular direction (Dutton, Fahey, & Narayanan, 1983). Dutton and Jackson (1987) noted that chosen labels can affect the level of risk taking, involvement, and commitment associated with strategic issues (cf. Kahneman & Tversky, 1984). Similarly, Fredrickson (1985) found that the decision labels "opportunity" and "problem" affected the type of recommendation participants in a business simulation made. Consequently, we assumed that interpretation is important in developing and sustaining the adaptive cognitive frameworks necessary for strategic action and change (Bartunek, 1984; Gioia & Chittipeddi, 1991). Action Effective organizational action in response to strategic issues often depends on an ability to implement decisions based on scanning strategies and subsequent interpretations of strategic information (cf. Pfeffer & Salancik, 1978). Indeed, Jackson and Dutton (1988) suggested that interpretation might hold the key to organizational adaptation models (e.g., Tushman & Romanelli, 1985) that attempt to link changes in strategic action to changes in top managers' readings of an environment. This interpretation-action link is also reflected in the work of both Whetten (1988) and Ranson, Hinings, and Greenwood (1980), who argued the importance of the link between managers' understanding of their environment and firm action. Adaptive actions involve some change; they can range from small-scale forms, such as changes in procedures, to larger-scale forms, such as productservice changes, revisions in overall strategy, and the redesign of organizational structures (Dutton & Duncan, 1987). For the purposes of this research, we defined action as any significant change in ongoing organizational practices, such as a substantive alteration in product or service offerings (Ginsberg, 1988). Performance As noted earlier, a body of work now exists exploring internal linkages among the sensemaking processes of scanning, interpretation, and action. Thomas and McDaniel (1990) argued that a "critical mass" of such research now exists and that one of the key remaining research issues is the extension of those relationships to account for organizational performance. Specifically, researchers need to examine how differences in the performance of similar organizations are related to differences in their ability to carry out the three sensemaking tasks. It is apparent that without a performance referent, neither researchers nor executives have a good way of evaluating the effectiveness of their scanning and interpretation processes or the associated actions. Additionally, examining the performance implications of the sensemaking process can provide a basis for understanding how the informationprocessing structures of firms might be designed to facilitate this vital set of activities.

Strategic Sensemaking and Performance

1993

Thomas, Clark, and Gioia

153

243

HYPOTHESES Our interest in investigating the links between the scanninginterpretation-action-performance variables drove hypothesis generation. However, to learn more about the intricacies of the process, we also compared the direct and indirect relationships among the variables. Scanning and Interpretation Top decision makers usually scan according to their perceptions of the necessity for information (Pfeffer & Salancik, 1978). For instance, they might differ with respect to the amount of information used and to their emphasis on attending to internal or external information. If decision makers across firms give unequal emphasis to the amount and type of information available to them, they are likely to differ in their interpretations and responses (Dutton & Duncan, 1987). The greater the level of information use, the more raw material top managers have for constructing their interpretations (Knight & McDaniel, 1979). More specifically, the more complete the information decision makers have about presumed cause and effect relationships (Thompson, 1967), the greater the likelihood they will perceive causes as controllable. A sense of mastery and a feeling that no stone has been left unturned emerges (Eisenhardt, 1989). From an information-processing perspective, top managers who use much information are more likely to emphasize the positive aspects of an issue (Thomas & McDaniel, 1990). Not only are they able to cope with ambiguity (Eisenhardt & Bourgeois, 1988) and uncertainty (Milliken, 1990), but also, over time, they will limit ambiguity through the informationprocessing structures that evolve, which might include teams, task forces, and reporting systems (Daft & Lengel, 1986). These arguments suggest the following hypotheses concerning information-rich scanning behaviors: Hypothesis la: High levels of information use among top managers will be positively related to their interpretation of strategic issues in positive-gain terms. Hypothesis lb: High levels of information use among top managers will be positively related to their interpretation of strategic issues as controllable. Smith, Grimm, Gannon, and Chen (1991) argued that top managers in externally oriented firms are more likely to sense competitors' actions and perceive themselves as more capable of implementing competitive responses than those in internally focused firms. This argument supplements Aldrich's (1979) observation that externally oriented managers will collect a richer array of information and be more confident that they can create a fit between their environment and organization when data from external scanning is added to their natural familiarity with internal factors. For those reasons, they will tend to interpret strategic issues as generally under their

154 244

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy

of Management Journal

April

control. Similarly, increased external scanning should enhance early detection of environmental disturbances before threat interpretations emerge (Jackson & Dutton, 1988), thus heightening managers' sense of control because they feel they can predict long-range outcomes (White, Duttrich, & Lang, 1980). Therefore, Hypothesis 2a: An external scanning orientation among top managers will be positively related to their interpretation of strategic is sues in positive-gain terms. Hypothesis 2b: An external scanning orientation among top managers will be positively related to their interpretation of strategic issues as controllable. Interpretation and Action Staw, Sandelands, and Dutton (1981) argued that the perception of an issue as a threat restricts the number of alternatives for action decision makers consider. In response to threat, managers tend to emphasize information that is consistent with current, conservative interpretive frames and can thus be more easily assimilated than new information. Managers also tend to rely on fewer sources of information in these circumstances (Smart & Vertinsky, 1977) and hence tend to use simplified assessments of situations and choose limited action-response patterns. Perceptions of threat also intensify concerns about efficiency (Williams, 1957), which are manifested in cost cutting, budget tightening, and the restriction of activities (Starbuck & Hedberg, 1977). Organizational actions associated with conservation of resources and tight control mechanisms are likely to be manifested in maintenance of the status quo, which in turn favors existing interpretation-action patterns. In sum, a threat interpretation, like a crisis, may distort information processing (Smart & Vertinsky, 1977), generate pathologies associated with action restriction (Rubin, 1977), and cause rigidity in decision processing that limits organizational responses (Staw et al., 1981). Issues labeled as opportunities are associated with projected positive outcomes and expectations of gain (Jackson & Dutton, 1988). In general, the opportunity label provides an issue with a "positive gloss" (Dutton, 1990), making it seem attractive and increasing decision makers' sense of control (Taylor, 1989). Further, the reaction to opportunity is likely to be low uncertainty (Milliken, 1990) and feelings of external control (Walsh et al., 1988). Those interpretations will in turn enhance the potential for top managers' taking strategic change actions, such as altering product-service offerings, because their confidence in achieving desired outcomes is high (Ginsberg, 1988; Milliken, 1987). Therefore, Hypothesis 3a: Top managers' labeling of strategic issues in positive-gain terms will be positively related to product and service changes.

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

155 245

Hypothesis 3b: Top managers' labeling of strategic issues as controllable will be positively related to product and service changes. Action and Performance The link between effective action and successful performance is a fundamental presumption in the strategic management literature. In the turbulent modern environment facing most competitive organizations, "domain offensive" actions in the form of new products and services (Miles, 1982) might be expected to engender competitive advantage. Smith and colleagues (1991) found that high numbers of organizational responses were linked to the profitability of domestic airlines. Similarly, Zajac and Shortell (1989) found that hospitals that stressed a commitment to new product and service development performed better than hospitals relying on current products and services. Shortell, Morrison, and Friedman (1990) made more specific findings showing that to be successful, hospitals' managers had to recognize the need to constantly differentiate and increase their product-service offerings to match the changing and growing needs of their patient bases. Snow and Hrebiniak (1980) suggested that such action-performance linkages might come about because of the development of necessary distinctive competencies in changing environments. Thus, Hypothesis 4: Organizations characterized by growth in their product and service offerings will exhibit higher performance than organizations characterized by little or no growth. Indirect and Direct Effects of Sensemaking Processes on Performance Strategic cognition-action models are generally predicated on an assumption that rational thought is closely linked to chosen actions. Likewise, the basic scanning-interpretation-action-performance model that undergirds this study also implies active information processing and systematic, sequential execution of the constituent stages. In this view, one way organizations compete is by acquiring superior strategic information via decision makers' effective scanning of the internal and external organizational environments and interpreting that information into a form they can use to implement appropriate actions that will lead to effective performance. However, several authors have questioned this portrayal of sensemaking and learning as deliberate. Issues such as beliefs, perceptions, politics, and goals (Staw, 1980) and the outcomes of prior actions and performance (Milliken & Lant, 1991; Weick, 1979) all conspire to complicate the process and give rise to nonlinear effects. O'Reilly (1982) and Beach and Mitchell (1978) suggested that because of decision makers' resistance to the expenditure of personal resources, some sensemaking antecedents might affect performance more directly or indirectly than the linear theoretical models predict. Accordingly, Daft, Sormunen, and Parks (1988) recommended that researchers

156

J. Β. Thomas/S. M. Clark/D. A. Gioia

246

Academy of Management Journal

April

explore the detailed linkages among scanning, interpretation, and strategic adjustment to understand their performance implications for organizations. It is apparent that the overall configuration of pathways and the comparison of the direct and indirect links between sensemaking and performance deserve closer examination. Because of the lack of coherent or compelling theoretical or empirical research evidence concerning those linkages, we framed our exploration as a research question: What is the relative strength of the direct and indirect paths between the sensemaking activities of scanning and interpretation, and organizational performance? METHODS Overview Industry studied. We decided to focus on a single industry to control for the impact of industry characteristics on the perceptions of strategic issues (cf. Yasai-Ardekani, 1986) and chose the health care industry. As Zajac and Shortell (1989) noted, in the 1980s health care organizations nationwide experienced an "environmental jolt" (Meyer, 1982) because of changes in regulation, clients' expectations, costs, and reimbursement mechanisms. Given that turbulence, the industry was a fertile ground for examining how top managers process information about issues associated with such environmental change and the action and performance consequences of that information processing. Past research also gave us confidence that top managers from different hospitals would be likely to focus on different information pertaining to the same environment (Ashmos, McDaniel, & Duchon, 1990; Meyer, 1982; Thomas & McDaniel, 1990) and respond strategically in different ways (Shortell & Zajac, 1990). Data sources. To obtain data on the scanning and interpretation behaviors of hospitals' decision makers, we sent 545 questionnaires to hospital chief executive officers (CEOs) in the state of Texas in 1987. We mailed the questionnaire only to public-access hospitals, excluding university health centers, prison hospitals, and hospitals located on military bases. Mailings were conducted in a single state to control for the possible impact of state hospital regulations. We used only those questionnaire responses indicating that the informant was the CEO of a hospital. Using the top executive as the sensemaking focal point for a hospital is consistent with the work of Daft and Macintosh (1981), Hambrick (1981), Zajac and Shortell (1989), and Thomas and McDaniel (1990), who noted that the CEO represents the most knowledgeable person regarding a hospital's strategic position and is the individual most responsible for taking actions intended to align the hospital's strategy, structure, processes, and environment. Additionally, this focus on CEOs seemed appropriate in light of the findings of Provan (1991), who found that hospitals' CEOs receive twice as much decision-making information as their boards and three times as much as their medical staffs. Those findings, coupled with the earlier finding of

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

157 247

Ritvo, Salipante, and Notz (1979) that the CEO is the primary interpreter of the issues faced by a hospital, suggested that CEOs were the appropriate people to inform us on the scanning foci and interpretations of hospitals. In addition, the members of a panel of experts (discussed fully later in this section) confirmed that CEOs would be the best informants for this study. Thus, although CEOs are not the only decision makers-sensemakers at hospitals, previous research and expert advice supported their use as informants. A total of 210 managers responded to the questionnaire, a response rate of 38.5 percent. Of those responses, 54 were unusable, 15 because the informant failed to complete parts of the questionnaire, 2 because the informant was not the CEO, 31 because the informant did not perceive the questionnaire as strategically relevant to his or her hospital (we discuss those cases in detail below), and 6 because product-service or performance data were not available. Chi-square analysis indicated no significant differences in size, type, and ownership between the remaining 156 hospitals and the nonresponding hospitals. 1 The average age of the CEO informants was 44; they had an approximate average of 12 years of managerial work experience and 5 years of strategic decision-making experience in their respective hospitals. Most of the informants (approximately 75 percent) had degrees in hospital administration. The average size of the 156 hospitals represented by the CEOs was 194 beds. We classified approximately 94 percent of these hospitals as "general hospitals" rather than specific types like psychiatric or rehabilitation facilities. The hospitals ranged in location from major metropolitan areas, such as Dallas and Houston, to rural areas. There was an approximately equal distribution across five ownership categories: church, nonprofit-nonchurch, county, hospital authority-district, and corporation. Questionnaire. We employed a case-scenario methodology to study the scanning and interpretation processes (cf. Fredrickson, 1986). Specifically, we provided the informants with two different case scenarios and then asked a series of questions regarding scanning and interpretation; the Appendix gives the scenarios and questions. To assure that there was no order bias, we used two versions of the questionnaire differing only in terms of the presentation order of the scenarios. An important assumption of this method is that strategic decision processes constitute patterns of decision behavior (Mintzberg et al., 1976; Weick, 1979). As Fredrickson (1986) noted, the characteristics of those decision processes—for example, scanning and interpretation—are consistent across decisions that are perceived as strategic.

1

For the analysis of respondent-nonrespondent differences, we divided size, defined as the number of licensed beds in a hospital, into three categories: small (500). We defined type as "general" and "other" and used five ownership categories. Chi-squares were as follows: size, 2.79 (p < .25); type, 2.52 (p < .63); ownership, 4.60 (p < .59).

158 248

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy of Management Journal

April

Before constructing the scenarios, we compiled a list of topics by tapping several sources: Interviews with health care executives, hospital strategic planning documents, articles in leading journals in the field, cases that had been written for hospital administration classes, and the popular press. Topics included malpractice insurance, the effects of Acquired Immune Deficiency Syndrome (AIDS) on long-term care, hospital affiliation agreements, health maintenance organizations (HMOs), outpatient surgical centers, and satellite care centers. We then formed an expert panel consisting of the president of the Texas State hospital association, the president of the state's physicians' association, two hospital CEOs, a chief operating officer, a chief financial officer, a marketing director, a general counsel (most members of which were employees of different hospitals), and two university faculty members whose primary research interest was hospital administration. We felt this selection would encompass any different perspectives or insights we might encounter. Our primary goal with the panel was to select topics and draft scenarios that hospital administrators in the state would consider strategic and were facing. As a result of this process, the HMO and satellite care center topics were selected. The next step was to draft scenarios that would provide a realistic yet balanced presentation of informational elements across the two scenarios. To assure realism, we gleaned much of the information used in the scenarios from strategic planning documents that we had obtained from several hospitals in the state. To assure balance and minimize framing effects, we constructed both scenarios using a two-by-two-by-two matrix. The three dimensions of the matrix were (1) whether the information was from a formal or informal source (written report or discussion, for instance); (2) whether the information came from a source within the hospital (an internal report) or outside it (general knowledge in the field); and (3) whether the information would generally be considered good (a reputation for quality care provision) or bad (outpatients are declining in number). Two pertinent information elements were placed in each of the resulting eight cells. Thus, each scenario consisted of 16 information cues that were identical in type but different in specific content. We then conducted multiple interviews with panel members to refine and cross-check the type and source of the information items to assure that they were accurate depictions of what would be encountered in the hospital environment. We also had them assess the overall content validity of each scenario as we proceeded through this process, asking "Is the scenario on the whole representative of the type of strategic situation being faced by hospital managers?" Finally, to assure that the hospital CEOs perceived the scenarios as strategically relevant, we included a question to that effect (see the Appendix). Informants who gave a scenario a strategic relevance rating of less than five on a seven-point scale were not included in the analysis.

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

159 249

Variables The questionnaire included multiitem scales with seven-point Likert response formats for all scanning and interpretation variables. Cronbach alphas for all scales were greater than .71. Scanning. Items for this scale were constructed from the information in the case scenario. The focus was on whether the information was in regard to, and was generated from, something external to the hospital (for instance, a rumor that a new hospital was moving into the area) or internal (for instance, an internal report indicating ambulatory care service was down; see the Appendix). Ten items consisting of five external and five internal information cues from each scenario were used initially to measure the extent to which the CEOs used internal or external information for strategic interpretation purposes. However, we found that one item for the second scenario had a negative interitem correlation and dropped it from the scale. Items were scaled so that high scores indicated high use of external information for interpretation. The Cronbach alpha for this scale was .73. We also asked the informants to indicate on a seven-point Likert scale the extent to which they would use the information cues in a scenario to clarify and define the strategic issues identified in the applicable case. Collectively, these questions about the information cues served as a measure of the amount of information in each case that was used to interpret strategic issues (information use). High scores on the 16-item scale for each scenario indicated a more comprehensive use of available information for developing a strategic issue. The Cronbach alpha for the information use scale was .88. Interpretation. We identified the measures of interpretation from Dutton and Jackson's (1987) three strategic interpretation dimensions (positivenegative, gain-loss, and controllable-uncontrollable). However, as noted earlier, we collapsed the positive and gain dimensions into a single scale prior to analysis. We posed five questions to assess the extent to which a CEO would use the controllable label to describe the two case scenarios and ten items to assess the positive-gain dimension of interpretation. All items were adapted from the questionnaire used by Jackson and Dutton (1988). For example, after respondents read a case scenario, they were asked: "To what extent would you label the situation as something positive?" and "To what extent do you feel the hospital can manage the situation instead of the situation managing it?" The continua were respectively scaled so that high ratings indicated that respondents labeled the case situation as positive or controllable (see the Appendix). The Cronbach alpha for the positive-gain scale was .91; the alpha for the controllability scale was .72. Organizational action. We measured the strategic changes (Ginsberg, 1988) that occurred in each hospital's product-service offerings over the three-year period (1987-89) that followed collection of the scanning and interpretation data. This measure assessed the extent to which hospitals

160 250

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy of Management Journal

April

added or deleted products and services. The Annual Guide Book of the American Hospital Association reports 54 product-service or facility categories for all U.S. hospitals each year (see the Appendix). The categories range from acquiring cutting-edge technologies (for instance, magnetic resonance imaging) through marketing actions (health promotion services) to specialization (women's centers). Using the list, we constructed a profile of product-service offerings for each of the hospitals analyzed for the years 1987 and 1989. The extent of product-service change was then measured in three different ways. The first was a weighted-differences method used by Hambrick (1981): a service added in 1989 that one-third or fewer of the hospitals offered in 1987 was assigned a weight of three, a service added in 1989 that existed in between one-third and two-thirds of the hospitals in 1987 was assigned a weight of two, and a service added by 1989 that existed in two-thirds or more of the hospitals in 1987 was assigned a weight of one. The difference between the number of services offered in 1987 and the summed weights of the products and services offered in 1989 indicated the extent of growth in offerings. Weighted changes in offerings between 1987 and 1989 ranged from - 2 to 41 for the hospitals analyzed. We then looked at product-service change from two other perspectives: (1) the extent to w h i c h hospitals had added high-technology products and services (for instance, open-heart surgery facilities) as opposed to lowtechnology ones (health promotion services) and (2) the extent to w h i c h they had added noncore products and services (like recreational therapy) versus core-related ones (intensive care units). We asked five health care executives and researchers w h o were not members of the expert panel to identify the services that hospital administrators would consider high technology and those they w o u l d consider core (see the Appendix). We then defined the action measure in two ways, as the difference between the numbers of hightechnology a n d noncore services offered in 1987 and the numbers of those offerings in 1989. We found that these three measures of product-service growth were highly correlated, with the lowest correlation among the three measures being .86; thus, the data demonstrated ample convergent validity. Because of its grounding in the literature, we retained only the measure based on Hambrick's method for the analyses. Performance. Three measures of performance were used: (1) occupancy was defined as hospital inpatient days divided by the total number of days beds were available; (2) profit per discharge was defined as the net inpatient revenue m i n u s inpatient cost divided by total discharges; and (3) admissions was the number of admissions divided by total beds to standardize for size. As Fottler (1987) noted, measures of performance outcomes for hospitals should reflect both efficiency indicators, such as productivity, and effectiveness indicators, such as financial viability (cf. Chakravarthy, 1986). Efficiency measures, such as the occupancy and admissions measures we used in this study, are important because competition for patients has been intensifying among hospitals (Fottler & Lanning, 1986). Financial viability,

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

161 251

measured here as profit, is crucial to survival in the dynamic health care environment (Shortell et al., 1990). The occupancy and profit measures were collected from the Healthcare Financial Management Association's Medicare Cost Report data base, an archival source of hospital operating and financial performance data. Admissions data were obtained from the American Hospital Association's Annual Guide Book. To guard against one-year outlier performance, we averaged performance data from 1989 and 1990 to form the measures. Control Variables Our basic perspective was an "upper echelons" one (cf. Hambrick & Mason, 1984). A number of authors have argued that top executives' characteristics, such as observable background attributes, affect their perceptions, and therefore, strategic choices (cf. Hitt & Tyler, 1991). Because CEOs are such a dominant strategic force in hospitals, we were concerned that their individual characteristics might affect the scanning, interpretation, and even strategic action variables (Norburn, 1989). Accordingly, in the regression analyses we controlled for those characteristics Hitt and Tyler (1991) found affected strategic decisions significantly, adapting them to the hospital environment where appropriate. These included age, type of education (1 = hospital administration degree and 0 = other), amount of managerial work experience, and years of experience in the strategic decision making of a given hospital. We obtained information on CEOs' individual characteristics from the American College of Healthcare Executives: A Biographical Dictionary of Membership. As in past studies, age and managerial experience were highly correlated (r = .75). We used only age after preliminary analyses revealed it made no difference which variable was entered in the regression models. Data Analysis We tested the hypothesized relationships among variables using path analysis. This analytical technique allowed us to identify the relative magnitudes of the direct and indirect effects of the three sensemaking activities, scanning, interpretation, and action, on organizational performance. We assessed the effect of case content on the scanning and interpretation variables by means of a paired comparison t-test, comparing the means for the responses to the first case to those of the second. We tested for differences in responses associated with the version of the questionnaire a respondent read (case presentation order) by means of univariate analysis of variance (ANOVA), where version was the main effect. We employed multivariate regression analysis to identify the overall relationship between the performance measures and the sensemaking process measures. To identify significant changes in product-service provision, it was necessary for us to have at least a two-year time lag between measurement points (Tushman, Virany, & Romanelli, 1985). We used the 1 9 8 7 - 8 9 difference in product-service provision to represent change in accordance with Tushman and colleagues' precedent. Because we used the average of the two years

162

J. Β. Thomas/S. M. Clark/D. A. Gioia

252

Academy of Management Journal

April

1989 and 1990 for the performance measures, the working assumption was a one-year lag time between product-service change and performance. RESULTS Descriptive Statistics Results of the t-tests and ANOVA indicated no significant differences in responses based on the case content or version of the questionnaire used. These results gave us confidence that there was a pattern of scanning and interpretation behavior across strategic issues. Because of the lack of differences in the variables, we calculated statistics on the basis of averaged responses for each of the variables from both cases. Table 1 presents means, standard deviations, and Pearson zero-order correlations. Multivariate statistics indicated that the set of performance measures—admissions, occupancy, and profit—was significantly related to the sensemaking process variables (Wilks's lambda = .678, F 1 S 3 7 6 = 3.77, ρ = .0001). Tests of Hypotheses The path analyses required us to analyze six regression equations: Admissions

Occupancy

Pro/it

Product-service

Positive-gain

= a + + +

+ b¡ product-service change b2 positive-gain interpretation b3 controllability + b4 information use b5 information source + e,

(1)

= a + + +

+ b1 product-service change b2 positive-gain interpretation b3 controllability + b4 information use b5 information source + e,

(2)

= a + + +

+ b1 product-service change b2 positive-gain interpretation b3 controllability + b4 information b5 information source + e.

(3)

use

change = a + control variables + bj positive-gain interpretation + b2 controllability + b3 information use + b4 information source + e, = a + control variables + b1 information + b2 information source + e,

(4) use

(5)

and Controllability

= a + control variables + b1 information use + b2 information source + e. (6)

Strategie Sensemaking and Performance 1993

253

Thomas, Clark, and Gioia

a U9 C O •a JS "öS b o υ «M 0) Ό (η Ο• Ο U » Ν Ό U S (Β 3 β Ο «α > ν Ρ -β

2

9 •3

•3 α

83 Η ili

b α

34)

.3 —

C ο>· οs fib ΒS g' 2ΟJ ^I üΒ S . o ä ^ b Ο. a ο ο s « Ο. VS U ι^ οο

163

S

** υ Λ ν 3υ S τ) £·β| Ο £ οι

Π

η ® >< Vν ν II 53 α. α. α. ζ > · : :

164

J. Β. Thomas/S. M. Clark/D. A. Gioia

254

Academy

of Management Journal

April

The beta coefficients obtained from these regression equations represent the path coefficients of the model and the direct effects of the antecedents on the relevant dependent variable. Table 2 presents results of this aspect of the analysis. These multiple regression results show a significant, positive relation-

TABLE 2 Results of Regression Analysis Variables

ß*

Profit Information use Information source Positive-gain interpretation Controllability Product-service change Occupancy Information use Information source Positive-gain interpretation Controllability Product-service change Admissions Information use Information source Positive-gain interpretation Controllability Product-service change Product-service change Experience Education Age Information use Information source Positive-gain interpretation Controllability Controllability Experience Education Age Information use Information source Positive-gain interpretation Experience Education Age Information use Information source * The * ρ ** ρ *** ρ

t

.06 .15 -.18 .14 .28

0.66 1.87* -2.08* 1.40 3.42***

-.06 .06 -.06 .08 .32

-0.70 0.79 -0.70 0.88 4.03***

.11 -.17 -.00 .00 .30

1.20 -2.11* -0.01 0.03 3.71***

-.08 -.05 .12 -.06 .19 -.03 .18

-0.08 -0.67 1.52 -0.67 2.35* -0.44 1.89*

-.00 -.08 .05 .52 .10

-0.08 -1.22 0.71 8.24*** 1.63

.01 -.10 -.01 .35 .07

0.09 -1.39 -0.21 5.01** 0.98

Λ2

F

df

0.17

5.39***

5,125

0.13

4.35**

5,149

0.11

3.62***

5,149

0.09

2.01*

7,147

0.29

15.25***

5,149

0.13

5.61***

5,149

reported beta coefficients represent path coefficients. < .05 < .01 < .001

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

165 255

ship between high information use during scanning and positive-gain interpretations of strategic issues (β = .35, t = 5.01, ρ < .01). More specifically, increased attention to a wide array of information during the scanning stage of the sensemaking process was tied to the interpretation of strategic issues as positive and as implying potential gains. Similarly, there was a significant relationship between information use and interpretations of issues as controllable (β = .52, t = 8.24, ρ < .001). Thus, results supported Hypotheses l a and lb. However, although the amount of information used was a predictor of interpretation, the type of information collected during scanning was not (see Table 2). Thus, results did not support Hypotheses 2a and 2b, although their direction was as predicted. There also was no support for the hypothesized relationship between positive-gain interpretations of an issue and subsequent action (Hypothesis 3a). However, there was a positive and significant interpretation-action linkage when we used perceptions of controllability as an antecedent variable to product-service change (β = .18, t = 1.89, ρ < .05). When top managers interpreted strategic issues as controllable, they tended to act upon that perception by adding products and services to their hospitals' offerings. This result supports Hypothesis 3b. For all the hypothesized relationships discussed to this point, individual characteristics were not significantly linked to the interpretation and action variables. The predicted relationship between action and performance was also supported (Hypothesis 4). Hospitals that added more products and services than others performed better across all three measures: for occupancy, the beta coefficient was .32 (t = 4.03, ρ < .001); for profit, beta was .28 (t = 3.42, ρ < .001); and for admissions, beta was .30 (t = 3.71, ρ < .001). Table 2 also shows a number of other significant and revealing links to performance. For example, the information source variable was directly, negatively linked to admissions and directly, positively linked to performance. 2 Specifically, scanning for internal information seemed to enhance admissions, and scanning for external information appeared to improve profits. Similarly, the positive-gain variable was directly and negatively linked to profit, suggesting that profit was enhanced when CEOs interpreted strategic issues negatively, in terms of potential loss. Figure 1 presents these findings, coupled with the results of the hypothesis testing, as a revised model of sensemaking and performance.

2 In a subsequent analysis, we included hospital ownership, defined as a dummy variable with 0 for nonprofit and 1 for for-profit, as a control in the regression equations. After reanalyzing the six equations involved in the path analysis, we found a positive relationship between ownership and profit (β = .21, ρ < .002). However, including ownership in the third model rendered the information source variable's relationship to profit nonsignificant. This was the only change across all six regression equations analyzed earlier. The significant correlation between ownership and information source (r = .26, ρ < .001) suggests that the top managers of for-profit hospitals tend to focus more on external than on internal information. Ownership, then, may explain some of the variance previously explained by scanning behavior.

166 256

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy of Management Journal

April

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

167 257

TABLE 3 Direct and Indirect Relationships Compared Performance Profit Information use Information source Positive-gain interpretation Controllability Occupancy Information use Information source Positive-gain interpretation Controllability Admissions Information use Information source Positive-gain interpretation Controllability " Beta t ρ *ρ ** ρ *** ρ

r

Direct Effect*

.07 .27*** -.10 .16*

.06 .15* -.18* •14t

.01 .12t .01 ,13t

-.06 .06 -.06 .08

.10 -.09 .01 .07

.11 -.17* -.01 .01

Indirect Effect .01 .06 .01 .05 .03 .07 -.01 .06 .01 .06 .01 .05

coefficients are reported, < .10 < .05 < .01 < .001

To examine our research question concerning the relative strengths of direct and indirect paths between sensemaking activities and organizational performance, we first calculated the indirect effects of the scanning and interpretation variables on the performance measures by multiplying the coefficients of all relevant paths (Duncan, 1971). For example, we calculated the indirect effect of a positive-gain interpretation on profit through product-service change by multiplying the coefficients for the path between the positive-gain and product-service change variables and the path between the product-service change and profit variables. We calculated the total indirect effect of the scanning variables by summing their indirect effects through the positive-gain interpretation, controllability, and product-service change variables (cf. Trevino & Youngblood, 1990). The next step involved examining the relative effects attributable to the indirect and direct effects on performance for each variable. Table 3 presents those results. This analysis revealed that although there were also indirect paths from information source to performance, the direct path dominated in terms of the effect of the variable on admissions and profit. The indirect path, however, dominated when occupancy was the dependent variable. In the absence of a significant, indirect path from positive-gain interpretation to profit (as depicted in Figure 1), it was not surprising that calculations showed that the significant, direct path dominated.

168

J. Β. Thomas/S. M. Clark/D. A. Gioia

258

Academy of Management Journal

April

DISCUSSION There has been long-standing interest in the relationship between cognition and action in organizations. More recently, researchers have shown increasing interest in the relationships among the more general sensemaking processes, which include not only cognitive processing but also considerations of associated action (Weick, 1979), and their relationship collectively and individually to performance. A number of researchers have noted that sensemaking activities, such as scanning an external environment (Nystrom & Starbuck, 1984} and interpretation (Ford, 1985), are key to organizational performance. Although sensemaking processes have been cited for their substantial theoretical importance, relatively little empirical research has investigated their proposed internal linkages (Milliken, 1990) or their relationships to organizational performance. As a way of articulating, testing, and refining the specific relationships involved, we framed this research in terms of the links among the sensemaking processes of scanning, interpretation, and action, and measures of organizational performance. In response to the fundamental research question of whether there are demonstrable links between sensemaking and variation in organizational performance, the results of this study indicate clearly that there are. However, this overarching result has perhaps even more important implications in terms of the constituent linkages among the sensemaking component processes. Using path analysis, we were able to construct a detailed model of those relationships, which is shown in Figure 1. Emergent Model Scanning and interpretation. Top managers' attention to high levels of information during scanning was related to their interpretation of strategic issues as positive and as implying potential gains. Such information use during scanning was also associated with a heightened interpretation of strategic issues as controllable. These results, taken in tandem, imply that high information use strongly influences strategic interpretation; attention to a wide array of information tends to influence the interpretation of strategic issues positively. This implication is consistent with the findings of Thomas and McDaniel (1990) and suggests that when organizations put mechanisms in place to increase information use, they can increase the likelihood that managers will interpret issues as potential gains and as controllable. Such mechanisms might include decreasing the use of formalized scanning procedures and increasing the use of "boundary spanners" (Daft & Weick, 1984), reducing internal communication barriers (Roberts & O'Reilly, 1974), and increasing interaction among top management team members (Gioia & Chittipeddi, 1991). Interpretation and action. Although we found no link between positivegain interpretations of strategic issues and subsequent action, there was a significant interpretation-action link when CEOs interpreted issues as con-

Strategie Sensemaking and Performance 1993

Thomas. Clark, and Gioia

169 259

trollable. More specifically, the interpretation of a strategic issue as controllable had a positive effect on product-service change. This result suggests that despite the inherently risky nature of strategic changes, top managers nonetheless increased the number and sophistication of their hospitals' product-service repertoires when they felt they could control strategic issues confronting the hospital—regardless of interpretations focusing on the positive-gain dimension. Positive change in product-service offerings was also linked to scanning efforts being focused outward, suggesting that, in the hospital industry, such scanning and interpretation together facilitated strategic action in the form of product-service changes. Action and performance. Hospitals that implemented more productservice changes performed better on all three performance measures: occupancy, profitability, and admissions. Finding such linkages was consistent with previous theory and research; our contribution to earlier work was in showing that the associations between external scanning behaviors and interpretations of controllability serve as sensemaking antecedents to strategic actions, which are in turn associated with effective organizational performance. These findings have implications for research that focuses on the effects of top management intervention on strategic change (e.g., Greiner & Bhambri, 1989). Direct and indirect links. Our results also suggest that in cases resembling the scenarios examined in this study, decision makers do not necessarily execute the sensemaking process in a normative, linear fashion. For example, although the amount of information attended to during scanning affected the interpretation of strategic issues, and through those interpretations the subsequent action taken, the focus of information search during scanning—internal or external—had direct linkages to both action and performance (see Figure 1). Specifically, we found that attention to external information was associated with higher profit, and attention to internal information was associated with higher hospital admissions levels. Those findings, coupled with the other results, suggest that although a focus on external information facilitates product-service change and increased profit, an eye on internal information enhances a hospital's ability to use strategic change to attract clients. These findings do not necessarily imply that a scanning focus circumvents the interpretation process involved in sensemaking. It is possible that the direct link between a scanning focus and performance reflects the inherent cyclicity of the sensemaking process. As Mintzberg and his colleagues (1976) discovered, decision processes often lose steps as decision cycles cascade over time. Decisions made in the past that generated effective outcomes do not always require reinterpretation and altered action; those stages often can be bypassed as a consequence of organizational learning (Walsh & Ungson, 1991). The multiyear time frame of this study may have provided a view of the sensemaking process that tapped this skipping of stages. It is possible, and perhaps even likely, that the hospitals were captured in the middle of a cycle in which top managers had taken previous

170 260

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy of Management Journal

April

actions that had affected performance. On the next scanning cycle, managers might have been affirming the pattern of information that led to the improved performance, thus bypassing any need for reinterpretation or revised action; or they might have been seeking types of information to support the actions that generated the observed profit increase—a possibility that is consistent with the "prior hypothesis bias" Levine (1971) observed and the "action justification efforts" Starbuck (1983) suggested.3 This nonlinear aspect of sensemaking is further evidenced in the revealing observation that positive-gain interpretations were also linked directly to profitability, although in an apparently counterintuitive fashion: perceptions of strategic issues as representing potential gains appeared to affect profit negatively. Again, knowledge of prior actions and outcomes would probably best explain the direct link between interpretation and performance in this instance. It appears, however, that the effects of such previous actions on profit influenced the hospital managers' interpretive orientations; a concern with avoiding failure rather than the expected concern with pursuit of profit seemed to dominate the process. Hospital managers, especially those of not-for-profit institutions, are perhaps highly sensitive to the potential for failure and therefore apt to be more responsive to potential problems or threats than to opportunities. Perhaps orienting strategic issue diagnosis toward threat interpretations heightens the sense of vigilance in identifying important strategic issues not previously noticed. Positive-gain interpretations may curtail decision makers' desire to understand a strategic issue of interest (Mintzberg et al., 1976) because they may assume that they can simply exploit opportunities without conducting extended analyses (Fredrickson, 1985). In the hospital industry, such a stance might lead managers to act in an overly optimistic manner toward strategic issues (Stevenson, 1976) even though they are ill-equipped to capitalize on possible opportunities (Thomas, McDaniel, & Anderson, 1991). The conservative stance found here also tends to encourage consideration of the implications of maintaining the status quo and works against a hospital's relying on preconceptions that might be obsolete in the face of a changing environment. In general, the present findings suggest that the managers of successful hospitals might begin from a position that strategic issues represent potential losses and then may adjust their product-service offerings through perceptions of controllability. Control variables. The findings also indicate that the individual characteristics (age, education, and organizational experience) used as control variables in the regression equations had no significant effect on interpretation or action. The absence of individual difference effects might be ex3 It is possible that the dimensions used in this study, positive-gain interpretation and controllability, did not fully capture the complexity of the interpretation process when scanning focus was treated as the antecedent. Dimensions such as legitimacy (Dutton, 1988), uncertainty (Milliken, 1990; Dutton & Webster, 1988), and importance (Dutton, Stumpf, & Wagner, 1990) might also need to be included in the model as possible issue characteristics.

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

171 261

plained by the research procedure used. In the questionnaire CEOs were asked to act as informants for their organizations rather than as individual decision makers. Future investigations may wish to examine the impact of these control variables on strategic sensemaking through the interactions they have with the kind of perceptual variables we used. Limitations and Normative Implications Some possible limitations of the study should be noted. First, the hospital industry is a regulated, high-technology industry characterized by high competition, low growth, and a mix of for-profit and nonprofit organizations; it remains to be seen if our findings will generalize to other settings. Also, because the measures of the scanning and interpretation constructs were drawn from a single source, some possibility of common methods bias exists. Although we examined an extensive array of actions that directly affect health care, the primary mission of the organizations studied, other organizational actions, such as reorganization, the installation of new management information systems, and changes in staffing that go beyond productservice change, were not represented in this study. Future research might be designed to specify how strategic action is related to sensemaking activities. For example, are interpretations of controllability linked to product-, market-, and technology-related actions differently? Do positive-gain interpretations affect actions surrounding such different elements of task environments as customers, suppliers, competitors, and regulators (Thompson, 1967) in different ways? We should also note that we tested the relationship between scanning and interpretation through the use of hypothetical, though quite realistic, cases, not through ongoing observations of scanning and interpretation in hospitals. Given the need to build our understanding of the cycles of the sensemaking-organizational outcome relationship, multiple field research methods (cf. Eisenhardt & Bourgeois, 1988) may be needed in light of the complex, temporal dimensions involved in such research. A number of possible implications for hospital CEOs engaged in actual sensemaking processes can be considered. For example, for the hospitals we studied, product-service growth was positively related to all three measures of performance. This finding does not imply that the quantity of change should be the dominant guideline, but it does suggest that aggressive development and implementation of product-service offerings is critical for hospital success. The sensemaking processes that help drive that growth are also worth considering as decision makers attempt to manage the change process in hospitals, institutions that have traditionally been rather static. Increased boundary-spanning behavior, coupled with a sense of control, appear to be positively linked to the propensity of a hospital to engage in strategic action with respect to product-service growth. Overall, the results suggest that hospital managers may need to alter the internal information-processing environments of their organizations to maximize the probability of successful

172 262

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy of Management Journal

April

change. For example, mechanisms that increase exposure to a wide range of strategic information, such as enhancing information-processing structures for top management teams (Thomas & McDaniel, 1990) and strengthening relationships with hospital stakeholders (Shortell et al., 1990), may facilitate the positive relationship between information use and controllability we found to be a precursor of strategic change. CONCLUSIONS Overall, the main contribution of this study is its specification of links between the cognitive tasks of scanning and interpreting and organizational actions and performance. More generally, the study established an empirical basis for exploring and understanding linkages among cognition, action, and outcomes and provided some insight into sensemaking processes that contribute to variations in organizational performance. Although significant headway has been made in understanding sensemaking processes and their links to organizational outcomes, a significant research agenda remains. For example, the cyclical link between performance outcomes and scanning activities has been, with some exceptions (e.g., Milliken & Lant, 1991), ignored in the literature. Because cognition often begins with action (Weick, 1979), the boundaries and sequence of the sensemaking and performance constructs become blurred, especially when we consider cyclical or historical influences (cf. Cook & Campbell, 1979; Keats & Hitt, 1988). Studies that incorporate scanning, interpretation, action, and performance outcomes from previous or overlapping time periods would further enhance researchers' understanding of sensemaking, especially in terms of providing insights into causal directions over time. Such research would enable specific feedback and learning "loops" to be incorporated into the model based on the present findings (Figure 1). Future integrative research might also include systematic examination of the relative impact on sensemaking of individual characteristics (Hitt & Tyler, 1991), cross-level effects such as organizational strategy (Daft & Weick, 1984) and top management team structure (Thomas & McDaniel, 1990), and multilevel constructs such as image and identity (Dutton & Dukerich, 1991). Moving beyond single-level approaches to sensemaking research and adopting methods that facilitate longitudinal observations may help researchers address the key question of how organizations can perform differently despite facing the same environment. Beyond the specific findings, this work encourages an overarching appreciation of the complexity of linkages between strategic sensemaking and organizational performance. Simon (1979) characterized complex issues, such as strategic issues, as mazes. Once defined, each maze consists of a set of possible paths between perceptions of where an organization is vis-à-vis its environment and where it wants to be in terms of performance outcomes. Some approaches to understanding strategic management imply that organizations need to focus on how to "run the maze" to maximize performance. Such portrayals amount to oversimplifications of sensemaking-performance

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

173 263

relationships and do not do justice to the intricate processes involved. They also suggest a kind of determinism that clearly does not characterize complex organizational settings. Our findings not only provide some specific insights into these intricacies, but also hint that the mazes might be constantly changing as a result of the very processes used to negotiate them. Variations in scanning orientations, interpretation modes, actions taken, and performance outcomes all conspire to alter strategic issue mazes over time. REFERENCES Aldrich, H. E. 1979. Organizations and environments. Englewood Cliffs, NJ: Prentice-Hall. Ashmos, D. P., McDaniel, R. R„ & Duchon, D. 1990. Differences in perception of strategic decision-making processes: The case of physicians and administrators. Journal of Applied Behavioral Science, 26: 201-218. Bartunek, ]. M. 1984. Changing interpretive schemes and organizational restructuring: The example of a religious order. Administrative Science Quarterly, 29: 355-372. Bateman, T. S., & Zeithami, C. P. 1989. The psychological context of strategic decisions: A model and convergent experimental findings. Strategic Management Journal, 9: 7 1 - 7 8 . Beach, L. R., & Mitchell, T. R. 1978. A contingency model for the selection of decision strategies. Academy of Management Review, 3: 439-449. Cantor, N., & Mischel, W. 1979. Prototypes in person perception. In L. Berkowitz (Ed.), Advances in experimental social psychology, vol. 12: 4 - 5 2 . Orlando, FL: Academic Press. Chakravarthy, B. S. 1986. Measuring strategic performance. Strategic Management Journal, 7: 437-458. Clapham, S. E., & Schwenk, C. R. 1991. Self-serving attributions, managerial cognition, and company performance. Strategic Management Journal, 12: 219-229. Cook, T. D., & Campbell, D. T. 1979. Quasi-experimentation: field settings. Boston: Houghton-Mifflin.

Design and analysis issues for

Cowan, D. Α. 19B6. Developing a process model of problem recognition. Academy of Management Review, 11: 763-776. Daft, R. L., & Lengel, R. H. 1986. Organizational information requirements, media richness, and structural design. Management Science, 32: 554-571. Daft, R. L., & Macintosh, N. B. 1981. A tentative exploration into the amount and equivocality of information processing in organizational work units. Administrative Science Quarterly, 26: 207-224. Daft, R. L., Sormunen, )., & Parks, D. 19Θ8. Chief executive scanning, environmental characteristics, and company performance: An empirical study. Strategic Management Journal, 9: 123-139. Daft, R. L., & Weick, K. E. 1984. Toward a model of organizations as interpretive systems. Academy of Management Review, 9: 284-295. Duncan, O. D. 1971. Path analysis: Sociological examples. In H. M. Blalock (Ed.), Causal models in the social sciences: 115-138. Chicago: Aldine-Atherson. Dutton, ]. E. 1988. Understanding strategic agenda building and its implications for managing change. In L. R. Pondy, R. ]. Boland, & H. Thomas (Eds.), Managing ambiguity and change: 127-144. Wiley. Dutton, ). E. 1990. The making of organizational opportunities: Context and construction in the interpretation of strategic issues. Paper presented at the Texas Conference on Organizations, Austin.

174 264

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy

of Management Journal

April

Dutton, ). E., & Dukerich, ). M. 1991. Keeping an eye on the mirror: Image and identity in organizational adaptation. Academy of Management Journal, 34: 517-554. Dutton, J. E., & Duncan, R. B. 1987. The creation of momentum for change through the process of strategic issue diagnosis. Strategic Management Journal, 8: 279-295. Dutton, J. E., Fahey, L., & Narayanan, U. K. 1983. Toward understanding strategic issue diagnosis. Strategic Management Journal, 4: 307-323. Dutton, ]. E., & Jackson, S. E. 1987. Categorizing strategic issues: Links to organizational actions. Academy of Management Review, 12: 7 6 - 9 0 . Dutton, J. E., Stumpf, S., & Wagner, D. 1990. Diagnosing strategic issues and the investment of resources. In R. Lamb & P. Shrivastava (Eds.), Advances in strategic management, vol. 6: 143-167. Greenwich, CT: JAI Press. Dutton, J. E., & Webster, J. 1988. Patterns of interest around issues: The role of uncertainty and feasibility. Academy of Management Journal, 31: 663-675. Eisenhardt, Κ. M. 1989. Making fast strategic decisions in high-velocity environments. Academy of Management Journal, 32: 543-576. Eisenhardt, Κ. M., & Bourgeois, L. J., III. 1988. Politics of strategic decision making in highvelocity environments: Toward a midrange theory. Academy of Management Journal, 31: 737-770. Ford, J. D. 1985. The effects of causal attributions on decision makers' responses to performance downturns. Academy of Management Review, 10: 770-786. Fottler, M. D. 1987. Health care organizational performance: Present and future research. Journal of Management, 13(2): 367-391. Fottler, M. D., & Lanning, J. A. 1986. A comprehensive incentive approach to employee health care cost containment. California Management Review, 29(1): 7 5 - 9 4 . Fredrickson, J. W. 1985. Effects of decision motive and organization performance level on strategic decision processes. Academy of Management Journal, 28: 821—843. Fredrickson, J. W. 1986. An exploratory approach to measuring perceptions of strategic decision process constructs. Strategic Management Journal, 7: 473-483. Ginsberg, A. 1988. Measuring and modelling changes in strategy: Theoretical foundations and empirical directions. Strategic Management Journal, 9: 559-575. Gioia, D. A. 1986. Symbols, scripts, and sensemaking: Creating meaning in the organizational experience. In H. P. Sims, Jr., & D. A. Gioia (Eds.), The thinking organization: 4 9 - 7 4 . San Francisco: Jossey-Bass. Gioia, D. Α., & Chittipeddi, K. 1991. Sensemaking and sensegiving in strategic change initiation. Strategic Management Journal, 12: 433-448. Gioia, D. Α., & Sims, H. P., Jr. 1986. Cognition-behavior connections: Attribution and verbal behavior in leader-subordinate interactions. Organizational Rehavior and Human Decision Processes, 37: 197-229. Greiner, L. E., & Bhambri, A. 1989. New CEO intervention and dynamics of deliberate strategic change. Strategic Management Journal, 10: 6 7 - 8 6 . Hambrick, D. C. 1981. Environment, strategy and power within top management teams. Administrative Science Quarterly, 26: 253-275. Hambrick, D. C. 1982. Environment scanning and organizational strategy. Strategic ment Journal, 3: 159-174.

Manage-

Hambrick, D. C., & Mason, P. A. 1984. Upper echelons: The organization as a reflection of its top managers. Academy of Management Review, 9: 193-206.

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

175 265

Hitt, Μ. Α., & Tyler, Β. Β. 1991. Strategic decision models: Integrating different perspectives. Strategic Management Journal, 12: 327-352. Huber, G. P., & Daft, R. L. 1987. The information environments of organizations. In F. M. Joblin, L. L. Putnam, Κ. H. Roberts, & L. W. Porter (Eds.), Handbook of organizational communication: 130-164. Newbury Park, CA: Sage Publications. Jackson, S. E., & Dutton, ]. E. 1988. Discerning threats and opportunities. Administrative Science Quarterly, 33: 370-387. Kahneman, D., & Tversky, A. 1984. Choice, values, and frames. American Psychologist, 39: 341-350. Keats, B. W., & Hitt, M. A. 1988. A causal model of linkages among environmental dimensions, macro organizational characteristics, and performance. Academy of Management Journal, 31: 570-598. Kiesler, S., & Sproull, L. 1982. Managerial response to changing environments: Perspectives on problem sensing from social cognition. Administrative Science Quarterly, 27: 548-570. Knight, K., & McDaniel, R. 1979. Organizations: An information systems perspective. Belmont, CA: Wadsworth. Levine, M. 1971. Hypothesis theory and nonlearning despite ideal S-R reinforcement contingencies. Psychological Review, 78: 130-140. Meyer, A. D. 1982. Adapting to environmental jolts. Administrative Science Quarterly, 27: 515-536. Miles, R. H. 1982. Coffin nails and corporate strategies. Englewood Cliffs, NJ: Prentice-Hall. Milliken, F. J. 1987. Three types of uncertainty about the environment: State, effect, and response uncertainty. Academy of Management Review, 12: 133-143. Milliken, F. J. 1990. Perceiving and interpreting environmental change: An examination of college administrators' interpretation of changing demographics. Academy of Management Journal, 33: 4 2 - 6 3 . Milliken, F. ]., & Lant, T. K. 1991. The effects of an organization's recent performance history on strategic persistence and change: The role of managerial interpretations. In P. Shrivastava, A. Huff, & J. Dutton (Eds.), Advances in strategic management, vol. 7: 129-156. Greenwich, CT: JAI Press. Mintzberg, H. 1973. Strategy making in three modes. California Management Review, 16(2): 44-53. Mintzberg, H., Raisinghani, D., & Theoret, A. 1976. The structure of "unstructured" decisions. Administrative Science Quarterly, 21: 246-275. Norburn, D. 1989. The chief executive: A breed apart. Strategic Management Journal, 10: 1-16.

Nutt, P. C. 1984. Types of organizational decision processes. Administrative Science Quarterly, 29: 414-450. Nystrom, P. C., & Starbuck, W. H. 1984. To avoid organizational crises, unlearn. Organizational Dynamics, 12(4): 53-65. O'Reilly, C. Α., III. 1982. Variations in decision makers' use of information sources: The impact of quality and accessibility of information. Academy of Management Journal, 25: 7 5 6 771. Perrow, C. 1970. Organizational analysis: A sociological view. Belmont, CA: Wadsworth. Pfeffer, J., & Salancik, G. R. 1978. The external control of organizations: A resource dence perspective. New York: Harper & Row.

depen-

176 266

J. Β. Thomas/S. M. Clark/D. A. Gioia Academy of Management Journal

April

Proven, Κ. G. 1991. Receipt of information and influence over decisions in hospitals by the board, chief executive officer, and medical staff. Journal of Management Studies, 28: 281-298. Ranson, S., Hinings, B., & Greenwood, R. 1980. The structuring of organizational structures. Administrative Science Quarterly, 25: 1 - 1 4 . Ritvo, R. Α., Salipante, R., & Notz, W. W. 1979. Environmental scanning and problem recognition by governing boards: The response of hospitals to declining birth rates. Human Relations, 3: 227-235. Roberts, K., & O'Reilly, C. 1974. Failures in upward communication in organizations: Three possible culprits. Academy of Management Journal, 17: 205-215. Rosch, E. 1978. Principles of categorization. In E. Rosch & B. Lloyd (Eds.), Cognition and categorization: 2 7 - 4 7 . Hillsdale, NJ: Erlbaum. Rubin, I. 1977. Universities in stress: Decision making under conditions of reduced resources. Social Science Quarterly, 58: 242-254. Shortell, S. M., Morrison, Ε. M., & Friedman, B. 1990. Strategic choices for America's hospitals: Managing change in turbulent times. San Francisco: Jossey-Bass. Shortell, S. M., & Zajac, E. J. 1990. Health care organizations and the development of the strategic management perspective. In S. Mick (Ed.), Innovations in health care delivery: 144-180. San Francisco: Jossey-Bass. Simon, H. A. 1979. The sciences of the artificial (2d ed.). Cambridge, MA: MIT Press. Smart, C., & Vertinsky, 1.1977. Designs for crisis decision units. Administrative Science Quarterly, 22: 640-657. Smart, C., & Vertinsky, I. 1984. Strategy and the environment: A study of corporate response to crises. Strategic Management Journal, 5: 199-214. Smircich, L., & Stubbart, T. 1985. Strategic management in an enacted world. Academy of Management Review, 10: 724-736. Smith, K. G., Grimm, C. M., Gannon, M. J., & Chen, M. 1991. Organizational information processing, competitive responses, and performance in the U.S. domestic airline industry. Academy of Management Journal, 34: 6 0 - 8 5 . Snow, C. C., & Hrebiniak, L. G. 1980. Strategy, distinctive competence, and organizational performance. Administrative Science Quarterly, 25: 317-335. Starbuck, W. H. 1983. Organizations as action generators. American Sociological Review, 48: 91-102. Starbuck, W. H., & Hedberg, B. L. T. 1977. Saving an organization from a stagnating environment. In H. Thorelli (Ed.), Strategy + structure = performance: 249-258. Bloomington: Indiana University Press. Staw, Β. M. 1980. Rationality and justification in organizational life. In B. M. Staw & L. L. Cummings (Eds.), Research in organizational behavior, vol. 2 : 4 5 - 8 0 . Greenwich, CT: JAI Press. Staw, B. M., Sandelands, L. E., & Dutton, J. E. 1981. Threat-rigidity effects in organizational behavior: A multilevel analysis. Administrative Science Quarterly, 26: 501-524. Stevenson, H. W. 1976. Defining corporate strengths and weaknesses. Sloan Management Review, 17(3): 5 1 - 6 8 . Taylor, S. 1989. Positive illusions. New York: Basic Books. Taylor, S. E., & Crocker, J. 1981. Schematic bases of social information processing. In E. T. Higgins, C. P. Herman, & M. P. Zanna (Eds.), Social cognition, vol. 1:89-133. Hillsdale, NJ: LEA.

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

177 267

Taylor, S. E., & Fiske, S. T. 1978. Salience, attention and attribution: Top of the head phenomena. In L. Berkowitz (Ed.), Advances in experimental social psychology, vol. 11: 249-288. New York: Academic Press. Thomas, J. B., & McDaniel, R. R., Jr. 1990. Interpreting strategic issues: Effects of strategy and the information-processing structure of top management teams. Academy of Management Journal, 33: 286-306. Thomas, ]. B., McDaniel, R. R., Jr., & Anderson, R. 1991. Hospitals as interpretation systems. Health Services Research, 25: 859-880. Thompson, J. D. 1967. Organizations in action. New York: McGraw-Hill. Trevino, L. K., & Youngblood, S. A. 1990. Bad apples in bad barrels: A causal analysis of ethical decision-making behavior. Journal of Applied Psychology, 75: 378-385. Tushman, M. L., Virany, B., & Romanelli, E. 1985. Executive succession, strategic reorientations, and organization evolution. Technology in Society, 7: 297-313. Walsh, J. P., Henderson, C. M., & Deighton, J. 1988. Negotiated belief structures and decision performance: An empirical investigation. Organizational Behavior and Human Decision Processes, 42: 194-216. Walsh, J. P., & Ungson, G. R. 1991. Organizational memory. Academy of Management 16: 57-91.

Review,

Weick, K. E. 1979. The social psychology of organizing. Reading, MA: Addison-Wesley. Whetten, D. A. 1988. Sources, responses, and effects of organizational decline. In K. S. Sutton & D. A. Whetten (Eds.), Readings in organizational decline: 151-174. Cambridge, MA: Ballinger. White, S., Duttrich, J. E., & Lang, J. R. 1980. The effects of group decision-making process and problem situation-complexity on implementation attempts. Administrative Science Quarterly, 25: 428-440. Williams, H. B. 1957. Some functions of communication in crisis behavior. Human tion, 16: 15-19. Yasai-Ardekani, M. 1986. Structural adaptations to environments. Academy of Review, 11: 9-21.

Organiza-

Management

Zajac, E. J., & Shortell, S. M. 1989. Changing generic strategies: Likelihood, direction, and performance implications. Strategic Management Journal, 10: 413-430.

APPENDIX Case Scenarios and Assessment Information Scenario One "The make-up of the area population, along certain dimensions, appears to be changing. For example, a recent study by the hospital reveals that nearly Ά of the area population has no regular physician and, considering that the number of medical indigents in the area is dramatically increasing, this proportion may continue to increase quite steeply. "At a recent executive committee meeting, a colleague remarked that nearly 10% of the local population is ineligible for health insurance coverage or government-funded health cost assistance. Additionally, the trend in the area towards more service-oriented jobs and selfemployment may lead to more and more people being uninsured. The rumor of a new, and supposedly quite large firm coming to the area is also discussed. "The hospital has been contemplating adding satellite centers to its operations. The mix of services offered by the hospital has seemed right, but the occupancy rate has continued downward. This has suggested to some that the hospital needs to reach out into the perimeters of the area to see more patients. With the hospital's reputation for quality care and with its capacity

178 268

J. Β . Thomas/S. M. Clark/D. A. Gioia Academy of Management /ournal

April

to handle increased service provision, your colleagues feel this possible expansion is attractive. However, there is a general concern that there may be some difficulty in attracting needed physicians. Further, with the general shortage of nurses in the area there is concern that nursing support for the centers may be a problem. " A page in the recent hospital financial report shows that DRG outliers are increasing." Given that nearly 40% of reimbursement for services to the hospital is from Medicare (with about 30% from commercial insurers) this information on outliers could have an effect on the financial performance of the hospital." Scenario Two " T h e role of HMO's in serving the medical needs of the area is changing. Information to support this includes a survey performed by the hospital which shows that HMO's have penetrated nearly 20% of the market for those under 65 years old. It has been suggested that this percentage could easily grow to 3 0 - 3 5 % in the next 15 years if the HMO option is made more available as expected. "Additionally, a local marketing firm hired by your hospital to track consumer trends in the area reports that the population in the area will be increasing through the year 2000. Currently, 75% of the population is under 44 years old. It is expected that this percentage will remain constant during the period. Executive staff feels that the medical needs of this growing population will not only change, but show dramatic growth in certain areas. "However, a recent internal operations report circulated to staff indicates that ambulatory care utilization of the hospital has fallen off. Some feel this downward direction in ambulatory care utilization may continue in light of the increase in the number of physician group practices and the in-house services that many of these groups are providing (a trend that will probably continue). " A n HMO has approached your hospital to negotiate a contractual agreement for the provision of certain services to its members. Your hospital was chosen, according to the HMO representatives, because of its good name recognition and location—two factors that scored very high in a recent survey of HMO users who were asked why they would choose a particular hospital. For some of the services requested by the HMO your hospital is presently unable to meet expected demand. However, top management has always maintained that it would be capable of bringing about needed expansion or change, though many feel a major reorganization of the hospital may be necessary. Attracting additional and/or specialty medical staff for any expansion program would not be difficult." Strategic Assessment of Scenarios " A situation is said to be strategic when it could alter the position of the hospital in the market, could significantly affect the whole hospital, and could have an effect on the hospital's purposes and goals. To what extent would your hospital consider the situation to be strategic?" (1 = small extent, 7 = great extent) Positive-Gain and Controllability All 15 items were repeated after each scenario was presented. The notations " P G " (positivegain) and " C " (controllability) were not on the questionnaire. "Using the scale provided, circle the number that is the best indicator of how your hospital would generally perceive the situation presented above:" (1 = small extent, 7 = great extent) " T o what extent would your hospital . . . Perceive that benefits will come from the situation? (PG) Label the situation as something negative? (PG) Have a choice about whether or not to address the situation? (C) Feel the future will be better because of the situation? (PG) Label the situation as a potential gain? (PG)

" A DRG is a diagnostic related group. DRG outliers are third-party reimbursements that fall outside of a standard payment schedule for that group.

Strategie Sensemaking and Performance 1993

Thomas, Clark, and Gioia

179 269

Feel it has the capability to address the situation? (C) See the situation as having positive implications for the future? (PG) Feel that there is a high probability of losing a great deal? (PG) Feel it can manage the situation instead of the situation managing it? (C) Be constrained in how it could interpret the situation? (C) Feel that how the situation is resolved will be a matter of chance? (C) Feel that there is a high probability of gaining a great deal? (PG) Label the situation as a potential loss? (PG) Label the situation as something positive? (PG) See the situation as having negative implications for the future?" (PG)

Information Use

The scale consisted of all 16 information cues in each scenario. Examples of information cues from scenario one are "the percent ineligible for health insurance" and "difficulty in attracting physicians." "Below is a listing of the information that is available from the situation description. Using the scale provided, please indicate for each piece of information the extent to which your hospital would use it to clarify and define the issue(s):"

Information Source

An example of the items derived from scenario one follows: External information elements "Report that 33% of the population is without a physician. "Same report indicates that the number of medical indigents is increasing. "Rumor that a new firm is moving into the area. "Report that self-employment and service-oriented jobs in the area are increasing. "High reputation in the community for providing quality care." Internal information elements "Occupancy rate of the hospital is down. "Hospital report indicates that DRG outliers are increasing. "Internal discussion suggests that hospital may need to increase patient volume. "Staff feels that hospital has capacity to meet changing demands of the community. "Hospital report indicates that the mix of hospital services seems right."

Product-Service Change Categories The notations "H," for a high-technology product or service, and " C " for a core product or service, did not appear on the questionnaire. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17.

Outpatient surgery services (H,C) Intensive care unit (cardiac care only) (H) Intensive care unit (mixed or other) (H,C) Open-heart surgery facilities (H) Trauma center (certified) (H) Ultrasound (H,C) X-ray radiation therapy (H) Megavoltage radiation facility (H) Radioactive implants (H) Diagnostic radioisotope facility (H) Therapeutic radioisotopic facility Histopathology laboratory (C) Organ/tissue transplant (H) Blood bank (C) Health promotion services Respiratory therapy services (C) Magnetic resonance imaging (MRI) (H)

180

J. Β. Thomas/S. M. Clark/D. A. Gioia

270

Academy of Management Journal 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. 46. 49. 50. 51. 52. 53. 54.

April

Alzheimer's diagnostic-assessment services Skilled nursing or other long-term care Hemodialysis (C) Hospice Burn care unit (H) Physical therapy services (C) Occupational therapy services Rehabilitation inpatient unit Rehabilitation outpatient Unit Psychiatric inpatient unit Psychiatric outpatient services (C) Psychiatric partial hospitalization (H) Psychiatric emergency services Psychiatric consultation-liaison services Psychiatric education services Women's center Organized outpatient department (C) Emergency department (C) Birthing room, labor, delivery, recovery room Reproductive health services (H) Genetic counseling-screening services (H) Extracorporeal shock wave lithotripter (ESWL) (H) Obstetrics (C) Home health services Recreational therapy Comprehensive geriatric assessment Speech pathology services Hospital auxiliary (C) Volunteer services department (C) Patient representative services (C) Alcoholism-chemical dependency inpatient unit Alcoholism-chemical dependency outpatient services Geriatric clinics Neonatal intensive care unit (H) Pediatric inpatient unit (C) CT scanner (H,C) Cardiac catheterization laboratory (H)

James B. Thomas is an assistant professor in the Smeal College of Business Administration, Pennsylvania State University. He received his Ph.D. degree from the University of Texas at Austin. His current research interests include strategic issue interpretation, top management team information processing, and strategic alignment. Shawn M. Clark received a master's degree from Brigham Young University. He is currently a doctoral candidate in management and organization at the Pennsylvania State University. His research interests include strategic sensemaking, the relationship between business strategy and information technology, and the role of trust in information processing. Dennis A. Gioia is a professor of organizational behavior in the Department of Management and Organization, the Smeal College of Business Administration, Pennsylvania State University. He received his doctorate from Florida State University. His primary research and writing focuses on the nature and uses of complex cognitive processes by organization members and the ways that these processes affect sensemaking, communication, influence, and organizational change.

Teil IV Externe Umweltanalysen

In der präskriptiven, traditionellen Strategieliteratur beziehen sich externe Umweltanalysen auf den OT-Teil der SWOT-Analysen, und sie umfassen zwei Beobachtungsfelder (vgl. z.B. Welge & Al-Laham, 1992: 83-108; Wheelen & Hunger, 1995: 82-113; Staehle, 1994: 594-602): Zum einen sollen die Trends in der generellen Umwelt analysiert werden. Hierzu zählen Gegebenheiten in den ökonomischen, technologischen, politischrechtlichen und sozio-kulturellen Bereichen. Zum anderen gilt es, die Aufgabenumwelt zu analysieren. Die Aufgabenumwelt einer Unternehmung wird oft mit der Branche gleichgesetzt, in der die Unternehmung tätig ist (Wheelen & Hunger, 1995: 85), denn von Interesse sind Entwicklungen bei brancheninternen Umweltelementen und -gruppierungen, wie z.B. den Konkurrenten, den Kunden, den Lieferanten, den Kreditoren usw. Der Sinn dieser Analysen besteht darin, Unternehmungsstrategien auszuwählen, die in einem .Strategic Fit' mit der Untemehmungsumwelt stehen und somit besonders gut geeignet sind, bestehende Umweltchancen auszunutzen. Die theoretische Durchdringung der externen Umweltanalyse bezieht sich bisher insbesondere auf die Aufgabenumwelt bzw. die Branche. Wichtige Impulse zur Theorieentwicklung gingen und gehen von der traditionellen und der neuen Industrieökonomik aus. Es haben sich drei Schwerpunkte der strategischen Forschung herausgebildet: die Branchenstrukturanalyse, die strategische Gruppenanalyse und die Konkurrentenanalyse. Die beiden ersten Forschungsgebiete stehen in engem Zusammenhang mit der traditionellen Industrieökonomik, und sie werden in Kapitel 5 behandelt. Einen wichtigen theoretischen Bezugsrahmen für die Konkurrentenanalyse liefert u.a. die neue Industrieökonomik·, diese Zusammenhänge sind Gegenstand des Kapitels 6.

5 Branchenstruktur- und strategische Gruppenanalysen

Die Branchenstrukturanalyse Besonders einflußreich für das Verständnis der relevanten Aufgabenumwelt des strategischen Planers war und ist die von Porter (1980) entwickelte Branchenstrukturanalyse. Porters Bezugsrahmen basiert auf dem von Mason (1939) und Bain (1956) entwickelten „Structure-Conduct-Performance-Paradigma" der traditionellen Industrieökonomik (Scherer, 1980; Teece, 1984). Diesem Ansatz zufolge ergibt sich der Erfolg einer Unternehmung und das Erfolgspotential einer Branche als Funktion der vorherrschenden Branchencharakteristika. Letztere bilden die Struktur, die das Unternehmungsverhalten und das Marktergebnis weitgehend determinieren. Da die Rentabilitäten (performance) der Branchenmitglieder durch die Branchenstruktur vorgezeichnet sind, spielt das Unternehmungsverhalten (conduct) in der Forschungstradition der traditionellen Industrieökonomik nur eine unwesentliche Rolle (Kreps & Spence, 1984; Knyphausen, 1995: 53). Porters Branchenstrukturanalyse (Porter, 1980: 3-33) unterscheidet fünf Wettbewerbskräfte, deren Stärke nicht nur die Wettbewerbsintensität einer Branche, sondern auch deren Rentabilitätspotential determinieren. Die fünf Wettbewerbskräfte umfassen (1) die Rivalität zwischen den etablierten Branchenanbietern, (2) die Bedrohung durch neu eintretende Wettbewerber, (3) die Bedrohung durch Substitutionsprodukte oder -dienstleistungen, (4) die Verhandlungsmacht der Zulieferer und (5) die Verhandlungsmacht der Käufer.18 Für die Stärke der einzelnen Wettbewerbskräfte sind jeweils mehrere Branchenstrukturmerkmale verantwortlich. In den Studien der traditionellen Industrieökonomik (wie auch in Porters Bezugsrahmen) gilt z.B. die Anbieterkonzentration als eine wichtige Determinante der Rivalität in einer Branche, und verschiedenartige Markteintrittsbarrieren gelten als strukturelle Merkmale, deren Vorhandensein die Wahrscheinlichkeit des Markteintritts abschwächt.19 Nach einer Analyse der jeweils relevanten Branchenstruktur empfiehlt Porter (1980: 34-46), daß Unternehmungen sich für eine von drei generischen Strategien (Kostenführerschaft, Differenzierung, Fokus) entscheiden, um sich vor den Wettbewerbskräften ihrer Branche zu schützen und um sich nachhaltige Wettbewerbsvorteile im Vergleich zu ihren Konkurrenten aufzubauen. Durch das Verfolgen einer generischen Strategie soll also Marktmacht für eine fokale Unternehmung begründet werden, die von an18

19

Obwohl Porter seine Analyse auf diese fünf Kräfte beschränkt, hat sich in der Praxis durchgesetzt, gemäß der Empfehlung von R. Freeman (1984) eine sechste Wettbewerbskraft, die „weiteren Stakeholder", in Branchenstrukturanalysen zu berücksichtigen. Dabei handelt es sich um Kräfte wie Gewerkschaften, Behörden, Umweltschutzorganisationen und andere Gruppierungen aus der Aufgabenumwelt, die über potentielle Verhandlungsmacht und die Möglichkeit verfügen können, in die Geschehnisse innerhalb einer Branche einzugreifen. Einen guten Überblick über den Stand der diesbezüglichen industrieökonomischen Studien gibt Knyphausen (1995: 53-61). Vgl. auch Minderlein (1990).

184

R. Κ. F. Bresser

deren etablierten und potentiellen Konkurrenten nicht so leicht in Frage gestellt werden kann. Der von Porter postulierte Zusammenhang zwischen dem Verfolgen von generischen Strategien und der Abschwächung der vorherrschenden Wettbewerbskräfte stellt eine interessante Erweiterung oder gar eine Umkehr (Teece, 1984) der Aussagen der traditionellen Industrieökonomik dar. In der Industrieökonomik gilt der vollkommene Wettbewerb als ein anstrebenswerter Zustand, der allerdings von Unternehmungen mit monopolistischer Macht, d.h. von Firmen, die sich z.B. durch das Errichten von Eintrittsbarrieren vor den Wettbewerbskräften geschützt haben, oft verhindert wird (Scherer, 1980). Deshalb ist für die klassischen Industrieökonomen das Studium der Beziehungen zwischen „Structure", „Conduct" und „Performance" letztendlich der Versuch, zu der Entwicklung einer optimalen, auf Wettbewerb ausgerichteten Wirtschaftspolitik beizutragen (Jacobson, 1992). Demgegenüber versuchen Porter und die von ihm beeinflußten strategischen Managementforscher in einem Akt, der einem „Reverse-Engineering" der klassischen Industrieökonomik gleichkommt, Unternehmungen dazu anzuhalten, sich machtvolle Marktpositionen aufzubauen, indem sie Marktunvollkommenheiten ausnutzen und/oder diese entstehen lassen. Ohne Zweifel hat Porters Branchenstrukturanalyserahmen die strategische Managementforschung und -praxis inspiriert. Allerdings kann man in der gegenwärtigen Literatur mehrere kritische Überlegungen finden, die dazu geeignet sind, das theoretische Fundament der externen Umweltanalyse zu erweitern: (1) Wohlfahrtsreduzierung. Wie oben angedeutet, kann strategischen Managementforschern vorgehalten werden, ihre Empfehlungen liefen auf eine Beschränkung der ökonomischen Wohlfahrt der Konsumenten hinaus, denn letztendlich propagieren sie das Verfolgen von Strategien, die die Wirkungen bestehender Wettbewerbskräfte beeinträchtigen (Rumelt, 1988; Jacobson, 1992). Dieser Vorwurf hat dazu beigetragen, daß nur eine unbequeme Symbiose zwischen der strategischen Managementforschung und der traditionellen Industrieökonomik entstanden ist. Wir werden im letzten Beitrag des sechsten Kapitels auf diese Problematik zurückkommen und einen ökonomischen Ansatz vorstellen, der eventuell besser als die Industrieökonomik dazu geeignet ist, eine theoretische Grundlage des Strategischen Managements zur Verfügung zu stellen (Stichwort: „Österreichische Schule"). (2) Variablenvielfalt und Indeterminiertheit der Interaktionen. Der Bezugsrahmen der Porterschen Branchenstrukturanalyse enthält eine große Zahl von Variablen, deren direkte und interaktive Wirkungen auf Zielgrößen wie die Branchenrentabilität weitgehend ungeklärt bleiben. Diese Kritik gilt in gleicher Weise für die (vom Ergebnis her enttäuschenden) multivariaten und großzahligen empirischen Querschnittsuntersuchungen der traditionellen Industrieökonomik und hat dazu geführt, daß sich die neue Industrieökonomik stärker auf die theoretisch stringentere Spieltheorie konzentriert hat (Tiróle, 1988: 2f.; Knyphausen, 1995: 61). Porter (1991: 97ff.) sieht diese Kritik, hält aber an seinem umfassenden Bezugsrahmen fest, da dieser mehr Komplexität als spieltheoretische Modelle einfangt und damit über einen praktischheuristischen Wert verfügt, d.h. der umfassende Bezugsrahmen hält den strategischen Planer dazu an, ein Entscheidungsproblem mit all seinen Verästelungen zu durchdenken. Was die Theorieentwicklung angeht, so ist Porter (1991) der Ansicht, daß sich spieltheoretische Modelle und umfassende Bezugsrahmen gegenseitig befruch-

Branchenstruktur- und strategische Gruppenanalysen

185

ten können. Die formalen Modelle können für eine begrenzte Zahl von Variablen subtile Interaktionen erforschen und dazu beitragen, relevante Variablen und ihre Beziehungen für die Erstellung von umfassenden Bezugsrahmen zu präzisieren.20 Die umfassenden Bezugsrahmen können hingegen dazu beitragen, daß formale Modelle die Diversität kompetitiver Situationen nicht zu vereinfacht darstellen und wichtige Variablen nicht ignorieren. (3) Mangelnde Dynamik. Die Fokussierung auf strukturelle Gegebenheiten der Branche hat dazu geführt, daß dynamische Aspekte (prozeßhafte Entwicklungen) einer Branche vernachlässigt werden (Mintzberg, 1990a: 124-137; Jacobson, 1992; Miles et al., 1993). Das Verhalten der Wettbewerber wird als weitgehend stabil angenommen, wie auch die Branchenstruktur insgesamt als der relativ stabile ökonomische und technische Kontext angesehen wird, in dem sich der Wettbewerb abspielt (Porter, 1980: 6f.; Garud, 1994: 388). Demgegenüber haben z.B. Nelson und Winter (1982) herausgestellt, daß der von Unternehmungen forcierte technologische Wandel dazu beiträgt, daß sich Branchenmerkmale sehr schnell und sehr grundlegend ändern können. D'Aveni (1994) argumentiert ähnlich, indem er für viele Branchen „hyperkompetitive" Zustände konstatiert. Es ist deshalb verfehlt, von einem relativ stabilen Kontext auszugehen (Ilinitch et al., 1996; Thomas, 1996). Eine partielle Dynamisierung der Branchenstrukturanalyse wird durch das Konzept der strategischen Gruppen erreicht (Caves, 1984).21 Bezogen auf die Konkurrentenanalyse erfolgt eine Dynamisierung durch die Hinwendung der neuen Industrieökonomik zur Spieltheorie. (4) Vernachlässigung kognitiver Prozesse. Die vorrangige Beschäftigung mit strukturellen Gegebenheiten einer Branche hat von der Wichtigkeit der Kognitionen bei der Durchführung von Branchenstrukturanalysen abgelenkt. Der Portersche Bezugsrahmen suggeriert die Fiktion einer objektiven, eindeutig determinierbaren Branchenstruktur. Diese Annahme wurde durch mehrere Studien widerlegt. So haben z.B. Jackson und Dutton (1988) ermittelt, daß bei der Interpretation von Umweltfaktoren ein „threat bias" besteht. Bei der Identifikation von externen Bedrohungen und Chancen laufen offensichtlich unterschiedliche kognitive Prozesse ab und zwar insofern, als Manager eher dazu bereit sind, Umweltinformationen als Bedrohung denn als Chance aufzufassen. Dieser „threat bias" sowie die Rolle, die Persönlichkeitsmerkmale (z.B. die Toleranz für Ambiguität) bei der Identifikation relevanter Umweltmerkmale spielen, legen die Vermutung nahe, daß es keine eindeutig erkennbaren und klassifizierbaren Branchenstrukturmerkmale gibt. Zu ähnlichen Ergebnissen kommen Fahey und Narayanan (1989). Diese Studie belegt, daß die kognitiven Vorstellungen von Top-Managern über ihre Branchenstrukturen signifikant von sogenannten „objektiven" Gegebenheiten abweichen, d.h. von Umweltbeschreibungen, die auf der Grundlage von Expertenbeschreibungen angefertigt wurden. Die Studie von Porac et al. (1995) hebt die soziale Konstruktion von Branchenstrukturen hervor. Demnach sind Branchenstrukturen keine objektiven, unverrückbaren Fakten und Beschränkungen, innerhalb derer sich der Wettbewerb abspielen muß. Branchenstrukturen erweisen sich vielmehr als soziale Konstrukte, die von den Managern einer Branche erst entdeckt und als existent angesehen werden müssen. Solche sozialen 20

21

Zu einer eingehenden Würdigung der Spieltheorie für die Entwicklung einer Theorie des Strategischen Managements vgl. Kapitel 6. Hierauf wird im nächsten Abschnitt näher eingegangen.

186

R. Κ. F. Bresser

Konstrukte stellen dann nur insofern Beschränkungen oder Bedrohungen dar, als die Manager an solche Beschränkungen glauben. Vor diesem Hintergrund erscheinen Termini wie Eintritts- oder Mobilitätsbarrieren plötzlich in einem anderen Licht. Unzweifelhaft ist, daß den kognitiven Aspekten der Umweltfindung und -interpretation (Weick, 1979) auch bei der Theoriebildung hinsichtlich des Einflusses der externen Branchenumwelt auf die Strategien der Unternehmung mehr Aufmerksamkeit geschenkt werden muß.

Strategische Gruppenanalysen Das Konzept der strategischen Gruppen hat inzwischen eine fünfundzwanzigjährige Geschichte, nachdem es von Hunt (1972) eingeführt und von Caves und Porter (1977) und Porter (1980) weiterentwickelt wurde. Das Konzept gilt als dynamisches Anhängsel des „Structure-Conduct-Performance-Paradigmas", denn es beschäftigt sich mit den Strategien von Firmengruppen und deren Veränderungen im Zeitablauf (Caves, 1984).22 Porter definiert eine strategische Gruppe als eine Gruppe von Unternehmungen, die innerhalb einer Branche die gleichen oder ähnliche Strategien verfolgt (Porter, 1980: 129). Eng verbunden mit dem strategischen Gruppenkonzept ist das Konzept der Mobilitätsbarriere. Dieses ist eine Erweiterung des Konzeptes der Eintrittsbarriere, und es umfaßt Branchenstrukturcharakteristika, die Wanderungsbewegungen von Unternehmungen aus verschiedenen strategischen Gruppen innerhalb einer Branche behindern (Caves und Porter, 1977). Das Konzept der Mobilitätsbarriere ist zentral für die Existenz von strategischen Gruppen, denn ohne diese Barrieren würden alle Unternehmungen im Zeitablauf die erfolgreichsten Strategien einer Branche adoptieren, was dann jegliche Rentabilitätsunterschiede verschwinden lassen würde (Wiggins & Ruefli, 1995). Mobilitätsbarrieren sind also Branchenstrukturmerkmale, die den Firmen einer strategischen Gruppe im Vergleich zu anderen Gruppierungen der Branche einen relativen Wettbewerbsvorteil verschaffen (Harrigan, 1985a). In der Literatur wird die Höhe der Mobilitätsbarrieren dazu verwendet, unterschiedliches Wettbewerbsverhalten und unterschiedliche Rentabilitäten von Unternehmungen in verschiedenen strategischen Gruppen zu erklären (Cool & Schendel, 1987; Fiegenbaum & Thomas, 1990; Harrigan, 1985a; Porter, 1980). Im Verlauf der Entstehungsgeschichte der strategischen Gruppenforschung haben sich zwei Forschungsrichtungen herausgebildet: eine strukturelle und eine kognitive. Die erste und historisch älteste Richtung entwickelte sich in enger Anlehnung an die traditionelle Industrieökonomik, und sie verwendet insbesondere ökonomische Argumente und Variablen, um die Verhaltens- und Performancekonsequenzen strategischer Gruppen zu untersuchen (z.B. Cool & Schendel, 1987; Fiegenbaum & Thomas, 1990; Ketchen et al., 1993). Die zweite, jüngere Forschungsrichtung ist sozialpsychologisch ausgerichtet und versucht zu ergründen, wie sich die Kognitionen und Perzeptionen von Managern auf die Erscheinungsformen von strategischen Gruppen auswirken (z.B. Porac et al., 1989; Reger & Huff, 1993; Reger & Palmer, 1996). 22

Nicht übersehen werden soll, daß es in der frühen Phase der strategischen Gruppenforschung auch einige deskriptive Studien gab, die eher in der strategischen Managementforschung als in der Industrieökonomik verankert waren. Vgl. z.B. Schendel und Hatten (1972) und Hattan et al. (1978).

Branchenstruktur- und strategische Gruppenanalysen

187

Der kognitive Ansatz der strategischen Gruppenforschung entwickelte sich als Reaktion auf die Kritik am strukturellen Ansatz. Einige Autoren bemängelten die methodische Vielfalt und die uneindeutigen Ergebnisse hinsichtlich zentraler Forschungshypothesen, die die strukturell angelegte empirische Forschung zeitigte (McGee & Thomas, 1986; 1992; Thomas & Venkatraman, 1988). Eindeutige Performance- und Verhaltenskonsequenzen der Gruppenzugehörigkeit ließen sich z.B. nicht ermitteln. Andere Forscher warfen die Frage auf, ob strategische Gruppen überhaupt existierten oder ob sie vielleicht nicht viel mehr als ein künstliches, analytisches Konstrukt der Industrieökonomen seien (Hatten & Hatten, 1987; Barney & Hoskisson, 1990). Bresser et al. (1994) beklagten eine mangelnde Dynamik der strategischen Gruppenforschung, die sich in einer Überbetonung des Strukturellen und der Vernachlässigung der Untersuchung von Gruppenveränderungen manifestiere. Die Vertreter des kognitiven Ansatzes argumentieren demgegenüber, daß sich viele ungelöste Probleme der strategischen Gruppenforschung lösen ließen, wenn der klassische, strukturelle Ansatz um kognitive Betrachtungen ergänzt würde (Fombrun & Zajac, 1987; Bogner & Thomas, 1993). Obwohl sich inzwischen die Ansicht durchgesetzt hat, daß sowohl ökonomische, strukturelle Daten als auch perzeptorische Daten bei dem Studium strategischer Gruppen verwendet werden sollten (vgl. z.B. McGee & Thomas, 1992; Tang & Thomas, 1992; Bogner & Thomas, 1993), ist das Verhältnis zwischen dem strukturellen und dem kognitiven Ansatz zum heutigen Zeitpunkt in keiner Weise geklärt. Bogner und Thomas (1993) und Reger und Huff (1993) gehen beispielsweise von einer gegenseitigen Beeinflussung aus; kognitive Gruppen beeinflussen die Entwicklung struktureller Gruppen und vice versa. Empirische Untersuchungen, die eine solche gegenseitige Beeinflussung oder andere Beziehungszusammenhänge zwischen den beiden Ansätzen nachweisen, gibt es bis heute allerdings nicht. Solche Studien sind dringend erforderlich. Wenn der kognitive Ansatz ernst zu nehmen ist, dann müssen aus ihm andere Gruppierungen und andere Verhaltens- und Performancekonsequenzen resultieren als aus dem strukturellen Ansatz. Ansonsten hätte der kognitive Ansatz keinen „value added". Was die Performance- und Verhaltenskonsequenzen der beiden Ansätze angeht, so soll im folgenden kurz aufgezeigt werden, welche Forschungsfragen von einer zukünftigen empirischen Forschung behandelt werden müßten: Aufgrund der bisher vorliegenden Evidenz ist anzunehmen, daß der strukturelle Ansatz eher als der kognitive dazu in der Lage ist, Performanceunterschiede zu erklären. Daß Unternehmungen aus verschiedenen strategischen Gruppen unterschiedlich hohe Rentabilitäten haben müßten, ist besonders für die strukturellen Ansätze von Bedeutung. Denn es sollen ja gerade stabile Mobilitätsbarrieren sein, durch die sich eine Branche in strategische Gruppen unterschiedlicher Rentabilitäten segmentieren läßt (Porter, 1979). Obwohl die Ergebnisse bezüglich der Performancewirkungen nicht eindeutig sind, werden direkte oder indirekte Zusammenhänge zwischen der Mitgliedschaft in bestimmten strategischen Gruppen und der Unternehmungsperformance von einer wachsenden Zahl von Studien nachgewiesen (vgl. z.B. Cool & Schendel, 1987; Fiegenbaum & Thomas, 1990; Cool & Dierickx, 1993). Demgegenüber hat der kognitive Ansatz das Verhältnis zwischen Gruppenstrukturen und Gruppenrentabilitäten generell vernachlässigt, denn dieser Ansatz ist insbesondere an dem Verhältnis zwischen kognitiven Strukturen und dem strategischen Verhalten von Gruppenmitgliedern interessiert (Reger & Huff, 1993). Die Dynamik dieses Ansatzes begründet sich ja gerade darin, daß man untersucht, wie sich wandelnde Perzeptionen von Top-Managern auf ihr strategisches Wahlverhalten auswir-

188

R. Κ. F. Bresser

ken (Bogner & Thomas, 1993; Reger & Palmer, 1996). Die Rentabilitäten oder andere Performancekriterien spielen in den kognitiven Modellen eine nachgeordnete Rolle: Sie werden erst indirekt durch die strategischen Wahlentscheidungen (und nicht durch die wahrgenommenen Gruppenstrukturen) beeinflußt. Interessant ist, daß das Verhalten von Gruppenmitgliedern im strukturellen Ansatz nur durch Mutmaßungen behandelt wird. Obwohl der strukturelle Ansatz davon ausgeht, daß das Verhalten zwischen den Mitgliedern einer strategischen Gruppe nach anderen Mustern abläuft, als das Verhalten zwischen Mitgliedern aus verschiedenen Gruppen (Porter, 1979; McGee & Thomas, 1992), wird diese Mutmaßung nicht explizit untersucht. Dies ist auch nicht sehr verwunderlich, denn letztendlich fehlt dem strukturellen Ansatz das methodische „Rüstzeug", um Verhalten empirisch zu erforschen.23 Es liegt deshalb die Hypothese nahe, daß sich der kognitive Ansatz besser dazu eignet, die Verhaltenskonsequenzen strategischer Gruppen zu untersuchen als der strukturelle Ansatz.

Die Artikel Die Kritik an der Branchenstrukturanalyse und der strategischen Gruppenanalyse verläuft weitgehend parallel. So sind z.B. die vier Kritikpunkte der Branchenstrukturanalyse (Wohlfahrtsreduzierung, Variablenvielfalt, mangelnde Dynamik und Vernachlässigung kognitiver Aspekte) ebenso auf den strukturellen Ansatz der strategischen Gruppenforschung anwendbar. Da sich die kontemporäre Forschung insbesondere mit dem Phänomen der strategischen Gruppe beschäftigt, sind die nachfolgenden Artikel diesem Forschungsbereich entnommen. Der erste Artikel, „Strategie Group Formation and Performance: The Case of the U.S. Pharmaceutical Industry, 1963-1982", von Cool und Schendel ist ein Klassiker der strategischen Managementforschung mit einer strukturellen Perspektive. Unter Anwendung großer methodischer Sorgfalt können die Autoren theoretisch fundierte Gruppenstrukturen identifizieren und deren Veränderungen im Zeitablauf untersuchen. Für insgesamt vier Mehljahreszeitperioden existieren relativ stabile strategische Gruppenstrukturen, und die Performance dieser Gruppen ist hinsichtlich der relativen Marktanteile in jeder Periode signifikant voneinander verschieden. Für ein rechnungswesenbasiertes Rentabilitätsmaß können hingegen keine Performanceunterschiede ermittelt werden. In dieser Studie offenbaren sich auch die Grenzen der Untersuchung dynamischer Prozesse, die für die strukturellen Ansätze typisch sind. Zwar werden Veränderungen der Gruppenstrukturen über die vier Perioden hinweg untersucht, damit endet dann allerdings auch die explizite Untersuchung dynamischer Aspekte. Tatsächliches strategisches Wahlverhalten einzelner Firmen, das zu einer Repositionierung und zu einer Veränderung von Performanceintensitäten zu späteren Zeitpunkten beiträgt, wird 23

Eine (wenn auch sehr eingeschränkte) Ausnahme von dieser Regel bildet die Studie von Cool und Dierickx (1993). Die Autoren untersuchen, wie Gruppenstrukturen sich auf die Intensität der Rivalität innerhalb einzelner und zwischen verschiedenen Gruppen auswirken und wie diese Rivalitätsintensitäten in der Folge mit den Unternehmungsrentabilitäten zusammenhängen. Allerdings werden gruppeninterne und -übergreifende Rivalitäten indirekt durch modifizierte Branchenkonzentrationsindices (also durch strukturelle Variablen) gemessen. Diese Verfahrensweise verdeutlicht auf eindrucksvolle Weise die Grenzen, innerhalb derer der strukturelle Ansatz sich bewegt, wenn es darum geht, das Verhalten von Unternehmungen und von Unternehmungsleitungen zu untersuchen.

Branchenstruktur- und strategische Gruppenanalysen

189

nicht erforscht. Allerdings werden die veränderten Gruppenstrukturen, die sich über die vier Perioden hinweg ergeben haben, verhaltenswissenschaftlich interpretiert. Diese Interpretationen sind jedoch nicht viel mehr als informierte Mutmaßungen. Der zweite Artikel, „Strategie Groups: A Cognitive Perspective", von Reger und Huff zeigt, daß sich strategische Gruppen auf der Basis von Managerperzeptionen eindeutig klassifizieren lassen. Darüber hinaus regen die Ergebnisse der Studie die Autorinnen an, mehrere Erweiterungen des strategischen Gruppenkonzeptes vorzuschlagen. Da eine strategische Gruppenzugehörigkeit nicht für alle Unternehmungen einer Branche zu jedem Zeitpunkt eindeutig gegeben sein mag, schlagen Reger und Huff vor, drei Typen von Unternehmungen zu unterscheiden: „core firms", „secondary firms" und „transient firms". Diese Differenzierung erlaubt es, wesentlich nuanciertere Gruppenanalysen durchzuführen, als es auf der Basis des strukturellen Ansatzes möglich ist. Ferner regen die Autorinnen an, ähnlich wie von Bresser et al. (1994) durchgeführt, Gruppenvariationen während instabiler strategischer Perioden zu untersuchen, denn es wird in vielen Branchen vermutlich immer wieder Perioden ohne klare Gruppenstrukturen geben. Eine Untersuchung dieser Instabilitätsperioden mit der hier vorgeschlagenen erweiterten Typologie von Unternehmungen kann ein besseres Verständnis von strategischen Reorientierungsprozessen ermöglichen. Obwohl die Untersuchung von Reger und Huff eine Querschnittsuntersuchung darstellt und somit das dynamische Potential des kognitiven Ansatzes nicht ausschöpfen kann, ist dieses Dynamisierungspotential ein zentraler Gegenstand der Diskussion in dem Artikel. Ergebnisse über dynamische Fragestellungen, z.B. wie sich ein Wandel von kognitiven Gruppierungen vollzieht und auf strategisches Wahlverhalten auswirkt, liegen bisher nur ansatzweise vor (vgl. z.B. Reger & Palmer, 1996). Die verhaltenswissenschaftliche Forschung schuldet uns hier noch den Beweis ihrer Ergiebigkeit.

190

Κ. Ο. Cool/D. Schendel MANAGEMENT SCIENCE Vol. 33, No. 9. September 1987 Primed in U.S.A.

STRATEGIC GROUP FORMATION AND PERFORMANCE: THE CASE OF THE U.S. PHARMACEUTICAL INDUSTRY, 1963-1982·" KAREL O. COOL AND DAN SCHENDEL INSEAD, Fontainebleau, France Krannert Graduate School, Purdue University, West Lafayette, Indiana 47907 The focus of this paper is on three major questions: ( 1 ) what is the theoretical rationale for the strategic group concept?; (2) does strategic group membership have performance implications?; and (3) are strategic groups and membership in strategic groups stable characteristics of industries? A statistical procedure is proposed to longitudinally identify strategic groups. The empirical setting is the U.S. pharmaceutical industry over the period 1963-82. While performance differences are found in terms of market share, profitability differences between groups are not observed. Neither are differences found in terms of risk and risk-adjusted performance. These results are attributed to the existence of performance variation within each strategic group. A dynamic model of competitive repositioning is proposed which helps integrate the findings from the study. (STRATEGIC GROUPS; PERFORMANCE; PHARMACEUTICAL; LONGITUDINAL)

Introduction Since introduced by Hunt (1972), the concept of "strategic groups" has received increasing attention in the strategic management and industrial organization economics literature. Its application to many different industry settings has illustrated that the strategic group concept has potential to provide insight into the nature of strategyperformance relationships and into the analysis of competition in general. Nevertheless many ambiguities still surround the "strategic group" concept. Prominent among these ambiguities are the issues pertaining to the definition of the concept itself, the lack of convincing evidence that different strategic groups exhibit differing performance results, and the absence of empirical analyses of how and why strategic group structures change over time. The purpose of this paper is to explore and expand the theoretical foundations of the strategic group concept, to develop a general procedure for both identifying strategic groups within an industry and following such groups over time, and to examine the strategy-performance consequences of strategic group membership for a firm. The U.S. pharmaceutical industry during the 20-year period 1963-1982 provides the setting for the study. A data base whose scope and content have not yet been available for a study of strategic groups was used. The paper begins with a brief literature review intended to focus the issues surrounding the strategic group concept. Next, the analytical framework, including hypotheses, variable definition, sampling approaches, and the data base employed, is developed. Finally, the results of the procedure to identify and track strategic groups over time, and an examination of the strategy-performance characteristics of strategic groups are presented. Development of the Strategic Group Concept Three central questions emerge from an extensive review of the rapidly growing research on strategic groups (Cool 1985, pp. 11-70): (1) what is the theoretical rationale * Accepted by Arie Y. Lewin, former Departmental Editor; received January 21, 1986. This paper has been with the authors 4 } months for 2 revisions. 1102 0025-1909/87/3309/1102Î01.25 Copyright © 1987, The Institute of Management Sciences

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1103

for the strategic group concept and is the concept capable of defining unequivocally the variables upon which group identification procedures are based?; (2) are the previous, conflicting findings on the strategy-performance consequences of strategic group membership due to an inadequate definition of the concept, or to an inadequate operationalization of it?; and (3) how stable are strategic group structures and strategic group membership? Theoretical Foundations of the Strategic Group Concept In the early 1970s, the strategic group concept developed from two separate viewpoints, one emanating from industrial organization (IO) economics, the other from the business policy discipline. Business policy research emerged in the early 1970s from essentially a descriptive, clinical orientation to one more concerned with the objectivity and methods of science. At the same time, the central concept of strategy was attracting both those interested in empirically-based research, and managers concerned with coping with change and competition. Models of the strategy-performance relationship were proposed as testable propositions (Schendel and Hatten 1972) that were related to the paradigm that drove industrial organization economics research. The argument flow went like this: Performance = /(controllable, noncontrollable variables)

(1)

where the noncontrollable variables can be related to the notion of environment, and controllable variables can be related to managerial decision variables. Some of these are related to efficiency or operations management, and the remainder are related to effectiveness or strategic management (Ackoff 1970, p. 10). The argument proceeded as: Goals = /(efficiency, effectiveness, environmental variables),

(2)

Performance = /(operations, strategy, environmental variables). (3) This formulation was compared to the structure-conduct-performance (S-C-P) paradigm of IO economics in the work of Schendel, Hatten and Cooper (1975); Hatten, Cooper, and Schendel (1978); Hatten and Schendel (1978). Using the brewing industry as the setting, firm performance was postulated to be influenced by strategic conduct (controllable or effectiveness variables) and industry structure (noncontrollable or environmental variables) as: Firm Performance = /(strategic conduct; industry structure).

(4)

Tests of statistical homogeneity of the brewing industry led to an examination of statistically homogeneous clusters of firms. These clusters were later related to the concept of strategic groups. One of the major conclusions of this early work was that, within the same industry, similar actions were associated with different performance consequences for firms belonging to different groupings. This finding has been corroborated by later research (Schendel and Patton 1978; Frazier and Howell 1983; Primeaux 1983a, b). While strategic management research work focused on the determinants of firm performance, other studies came to the strategic group notion from the industrial organization economics concern of explaining and evaluating industry performance. In the Mason (1939) and Bain (1956) tradition, industry performance was thought to depend primarily on an industry's structural (S) characteristics. Conduct (C) was postulated to be forced by structure, did not represent independent managerial action, and therefore could be ignored. Hunt (1972) departed from this "structuralist" perspective to explain the intense rivalry observed in the highly concentrated, but nevertheless, highly competitive U.S. home appliance industry in the 1960s. He observed stable

191

192

Κ. Ο. Cool/D. Schendel 1104

KAREL O. COOL AND DAN SCHENDEL

conduct differences such as product diversification policies and distribution arrangements, and suggested that these differences or "asymmetries" prevented the development of an industry-wide oligopolistic consensus as prior IO work would predict. Hunt coined the term "strategic groups" to refer to firms displaying similar conduct along these key strategic dimensions. Of course, the implication was that an "industry" is more likely composed of several competing subsets, or, strategic groups. This notion was more formally developed by Newman ( 1972) in his study of strategic groups in the U.S. chemical process industry. Defining strategic groups by their degree of vertical integration within the market in question, he demonstrated that the existence of strategic groups impairs expected (tacit) collusion among firms, reducing the explanatory power of the "structuralist" model of IO economic theory. Porter ( 1976) and Caves and Porter ( 1977) suggested that industry-wide inferences of market power cannot be made when strategic groups characterize competition. Since "mobility bariers", i.e., group-specific entry barriers, differentially protect strategic groups, entry conditions and the scope for collusive agreements differ between strategic groups. From this, it is inferred that industry participants have sustained performance differences (Porter 1979; pp. 226-227). It is interesting to note that the orientation toward industry performance has remained in 10 research, while the focus on firm performance continues to characterize research in strategic management. IO research typically is conducted on cross-sectional industry samples, with strategic groups identified on the basis of a single variable such as size (Porter 1979; Caves and Pugel 1980), advertising intensity (Oster 1982), geographic origin (Donsimoni and Leoz-Arguelles 1981), or with a small set of variables, including advertising and R&D intensity, vertical integration, and size (Tassey 1983; Hergert 1983). On the other hand, strategic management research on strategic groups typically is confined to a single industry, with the variables used to identify groups selected on the context of the industry under study (Ramsler 1982; Frazier and Howell 1983; Harrigan 1983; Dess and Davis 1984). The difference in focus between IO economists and strategists is owed to the orientation of the former to welfare or efficiency issues from a public policy viewpoint, while the latter is oriented to management decision-making in an individual firm context. These two different orientations or purposes have led to examination of different types of groupings or subsets of firms in a given industry, creating a confused picture on the significance and value of the strategic group concept, whether it be used for resolving public policy issues, or for advice to management. In addition, the wide variety of methodologies applied to the study of strategic groups has added to the confusion. Some approaches have been descriptive, interpretive, even anecdotal, and others quite elaborate in terms of the methods of science. Both the IO and strategic management literature have a predominantly empirical basis in their orientation, but the work is not, in general, of a theory-building, theory-testing kind, except for whatever guidance the S-C-P and strategy paradigms provide. Lacking is attention to the theoretical definition of the strategic group concept, which could guide empirical work and provide a common thread for research. Since Hunt coined the term "strategic group" in 1972, little theoretical work has been performed to establish what is really "strategic" about a strategic group. Porter (1980, pp. 127-128) identified a list of key strategic dimensions along which firms could position themselves differently. His approach is not tested theory however, and it gives little guidance as to just what are the key similarities among strategic group members. The lack of a clear definition of the notion of a strategic group is conspicuous and needs to be addressed. If a cumulative stream of research is to be built, the strategic group concept needs to be more carefully defined. In addition, a systematic procedure to operationalize the concept needs to be developed.

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

Performance Consequences of Strategie Group

1105

Membership

A central theme in the literature is that strategic group membership has performance consequences. However, the few empirical findings on this differential performance hypothesis are conflicting. Porter (1979), comparing the performance of his "leader" and "follower" strategic groups, stated that leader groups outperform followers. However, the difference found was not statistically significant. Neither did Caves and Pugel (1980) find a difference in profitability between smaller and larger firms. Oster (1982), on the other hand, found that high advertisers outperformed low advertisers in those industries where advertising spending has lasting effects. Frazier and Howell (1983) found no difference in performance among their strategic groups in the medical supply and equipment industry, while Dess and Davis (1984) observed that their "generic" strategic groups in the paint and allied products industry differed on some performance measures while not on others. Clearly, there is no consistent, uniform support from empirical studies for the differential performance hypothesis between strategic groups. Unfortunately, whether this lack of support is attributable to the wide variety of procedures used to identify strategic groups, or whether it is a reflection of actual strategic group performance cannot be concluded. Support or rejection of the value of strategic groups will require the clarification of the strategic group concept itself, and proper procedures for identifying the groups in the first place. Without these tasks being completed, further empirical research on performance differences is unlikely to provide unequivocal answers to the question of whether strategic group membership has performance consequences. Stability of Strategic Group

Structures

A final issue considered here is the stability of strategic group structures within an industry, and within each strategic group. The majority of previous studies have been limited to the identification of strategic groups at a given point in time. In other words they have been static analyses. The observed groupings are taken as evidence that strategic groups identified at a given point hold over time. To establish the general validity of this proposition, static analyses need to be expanded to longitudinal analyses to establish whether groupings identified in one point in time hold in another. Unless some degree of stability is observed, or some pattern to shifts can be identified, the very concept of a strategic group will prove random and of little value. A fortiori, as long as the validity of the construct is not established, research on performance differences among strategic groups is premature.

Research Framework and Hypotheses In this section, the definition of strategic groups and the measures of performance used in this study will be defined. Then the three basic hypotheses tested in the work will be presented. Defining the Strategic Group Concept If strategic groups are really strategic, then the concept of strategy that has been developing in the field should provide the basis for a sound definition of what variables can be used to isolate and study strategic groups. Prior work has not always been careful, and indeed much research in strategic management has not always been careful, with the definition of strategy. Following Ansoff (1965), Katz (1970), Hofer and Schendel (1978), Day (1984), and others, it is postulated here that business level strategy, the level relevant to strategic

193

194

Κ. Ο. Cool/D. Schendel 1106

KAREL O. COOL A N D DAN SCHENDEL

groups, consists minimally of two sets of activities1: (1) those dealing with business scope2 commitments, and (2) those dealing with resource commitments, both intended to achieve a competitive advantage. Included in the scope commitment are those decisions involving3: (1) the range of market segments that are targeted, (2) the types of products and/or services offered in the market segments selected, and (3) the geographic reach of strategy. Resource commitments are defined to include business-level deployments of resources (cash, human, materials, etc.) to those functional areas that are key to obtaining and maintaining a competitive advantage in target product-market segments. The distinctive combination of scope and resource commitments defines business strategy for a segment and underlies the pursuit of intended competitive advantage (Porter 1985; Miles 1982; Abell 1980;Rumelt 1979; Hofer and Schendel 1978; and Thompson 1967).4 Based on the two strategy components defined above, a more precise definition of the strategic group concept can be given. A strategic group is defined as: A set offirms competing within an industry on the basis of similar combinations of scope and resource commitments. Note that most prior studies (Cool 1985, pp. 86-93) have excluded either scope or resource commitment decisions in their operationalization of the strategic group concept. These exclusions lead to incomplete model specification and inconclusive empirical results. While this definition of strategic groups is still general, it does specify, however, the major components that need to be considered at a minimum. Since the pertinent scope and resource commitments are industry specific by definition, the actual determination of those variables that define strategic groups must be industry specific. Employing similar variables to identify strategic groups across industries, the common practice in ΙΟ-based studies, necessarily entails trade-offs, and inconsistencies that compromise accurate identification of strategic groups. Thus, the research worker is compelled by these considerations to have a working knowledge of the industry studied. Measuring Performance Further complicating the debate about strategic groups and their performance differences is the issue of how to measure performance. The deficiencies of accounting information to determine economic rates of return are well documented (Bernstein 1974, pp. 466-509; Winn 1975; Schwartzman 1975; Stauffer 1975; Fisher and McGowan 1983; Salamon 1985). Yet, unadjusted accounting measures, such as return on equity, and return on total firm assets have been employed in previous studies to evaluate strategic group performance differences. 1 It is recognized that strategy formation includes organizational components. These were not considered here to simplify the analysis and because they are difficult to study. Yet, because they deal with the usage of strategy they have influence on performance, if not on strategic group formation. 2 Scope commitments are also referred to as " d o m a i n " commitments (Thompson 1967, Miles 1982). Only the term "scope" will be used here. ' This scope description parallels Abell's ( 1980) "business definition" approach, in which he uses "customer groups", "customer functions" and "technology" as components. 4 These concepts will be illustrated later for the context of the U.S. pharmaceutical industry. As another example, one can consider the international financial services industry. Based on Walter's ( 1986) research, the components of scope include: (1) range of market segments or customer groups (sovereign states, corporate clients, correspondent institutions, private clients, retail clients; (2) types of products/services across the segments (credit products, financial engineering products, risk management products, market access products, arbitrage and positioning); (3) geographic scope (onshore, offshore, global, regional, national, sub-regional, location-specific). For resource commitments, Walter identifies the following activities as key contributors to competitive advantage: ( 1 ) adequacy of an institution's capital base and risk base, (2) human resources, (3) access to information and markets, (4) technology base.

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1107

Other confounding effects are introduced by taking total firm performance indicators to assess business strategy performance, a practice which is particularly troublesome, given the increasing diversification of firms (Montgomery 1979). A further limitation of previous analyses, with the exception of Oster ( 1982) and Dess and Davis (1984), is the exclusive reliance on single performance indicators to draw inferences about intra-industry performance differences. From a behavioral viewpoint at least, performance is clearly a multi-dimensional concept, which implies that multiple performance indicators need to be employed. Finally, even if multiple, unbiased performance measures can be obtained, their simple comparison across strategic groups may not be justified if realized performance is affected by different levels of risk-taking. While different strategies might result in different levels of performance, these strategies may entail different risk exposure. If so, consideration of risk-adjusted performance measures is required, a conclusion supported by finance theory, and in line with prior strategic management research (Rumelt 1974; Christensen and Montgomery 1981; Bettis and Hall 1982; Bowman 1980, 1982; Hambrick et al. 1982). Hypotheses As argued earlier, valid inferences about the relationship between strategic group membership and performance require longitudinal studies in addition to proper strategic group identification, and proper definition of performance indicators. The present study is concerned with using proper strategic groups, studying them longitudinally, and examining three dimensions of performance: (1) levels of performance, (2) levels of risk exposure, and (3) levels of risk-adjusted performance. The hypotheses are stated in null form. Hi : Strategic groups demonstrate the same level of performance, for all groups. Hi is tested against the alternative hypothesis that levels of economic performance difTer between strategic groups. Several performance measures will be considered to reflect the multi-dimensional nature of performance. The second hypothesis relates to the risk postures of strategic groups. It was suggested that different strategies may entail different levels of risk-taking, rendering the simple comparison of performance levels inadequate. The extent of different risk-postures of strategic groups will be tested with the following null hypothesis: H 2 : Strategic groups are characterized by similar levels of risk. H 2 is to be compared to the alternative hypothesis that risk exposures differ between groups. Finally, to complete the performance comparison from a risk-return perspective, performance will be compared across the strategic groups with the following hypothesis: H 3 : Strategic groups demonstrate the same level of risk-adjusted performance. H 3 will be tested against the alternative hypothesis that strategic groups have different risk-adjusted performance. Methodology,

Variable and Sample Selection, and Data Bases

In this section, the methodological approaches used in this study are outlined in sufficient detail to show how they attempt to overcome objections raised about prior studies concerning the definition of strategy, adequacy of the data bases employed, and the use of judgement versus objective evidence to support conclusions formed. Methodology Before the three hypotheses can be tested, strategic groups for a specific industry need to be identified. Also, the evolution of the industry's strategic groups in terms of

195

196

Κ. Ο. Cool/D. Schendel 1108

KAREL O. COOL AND DAN SCHENDEL

numbers and membership needs to be developed. In developing the groups, an objective and replicable procedure capable of showing différences in business strategies between industry participants is required. This procedure must be capable of identifying strategy differences at any point in time, and strategy changes over time. The procedure used to determine an industry's strategic group structure over time is based on testing the variance-covariance matrix of strategic variables for a given firm over time. Let X„ = [Xm, X^i, XhnJ denote the vector of observations at time t on the set of variables describing the strategic scope and resource commitments of firm i in the industry considered, where, i = 1 , . . . , η the number of sampled firms, j = 1 , . . . , m the number of variables describing business strategy, t = 1 , . . . , Τ the number of time periods for which strategy observations are made. Then, for any period t, an η by m matrix can be constructed describing the strategic position of the sampled firms. One way to determine whether the set of firms change their relative position in the industry over time is to calculate from the matrix of observations the m by m variance-covariance matrix S, for each period t, and to test whether successive covariance matrices differ statistically. The rationale for this method is that when the industry's firms alter their commitments along the identified strategy variables, the covariances between these variables should reflect this repositioning. By determining at what point in time the covariance structure has changed from previous periods, in a statistically significant way, it is possible to construct distinct periods of time within which the configuration of strategic positions of firms in the industry is more stable than between periods. In other words, the statistical pooling procedure makes it possible to identify transition points separating subperiods with distinct strategic group structures. The test procedure begins with an evaluation of the stability of the strategic group structure over the period, t. The null hypothesis of equality of the covariance matrices of the first two time periods: Ho: Σι = Σ 2 . When, for a chosen significance level, both matrices are not statistically different, the data on both periods is pooled and the test procedure is repeated for data over the first three periods. The following test is then performed: Ho: 2,2 = Σ 3 where Ση denotes the covariance matrix of the data pooled over the first two periods. Since the pooling of data over the first two periods might impede the detection of patterns of change occurring over the last two periods, an additional test needs to be performed, viz., Σ, , Σ^,. When both tests do not reject Ho, then the data over the first three periods are pooled and the test procedure is continued. In general, then, the test procedure is performed for period t as: Ho: Σ|2· . . t -l = Σ,, Ho: Σ Ι 2 .. , t _ 2 = Σ,_ι, Ho: Σ, , Σ23. . .,, against Hj. „,,.,)Σ are equal (for each H 0 ) where Σ 12 ...,_ι denotes the population covariance matrix for the period spanning subperiods 1 through t-l. The test statistic used for evaluating the equality of covariance matrices is a generalization of the Bartlett test for the homogeneity of m variances. For a description of the test, see e.g., Timm (1975, pp. 250-260) and Morrison (1967, pp. 152-153). Potentially, the determination of transition points is affected by the composition of the sample used in the pooling procedure. To verify the sensitivity of pooling results to sample composition, a complementary analysis is needed. The approach followed here

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1109

is to determine the transition points on the basis of a sample of q firms where q < n, and to repeat the analysis on samples where in each step one firm is added until the total sample of η is obtained. The procedure above permits identification of subperiods with relatively stable strategic group structures. Within each period, cluster analysis can be applied to determine to which strategic group each firm belongs. For a given subperiod, the following sequence of steps was used: if the subperiod spanned k years, then the strategy variables Xy, were averaged over the k years for each sampled firm. Upon standardization of the data, the "Error Sum of Squares" cluster algorithm (Anderberg 1973, pp. 142-149) was applied to uncover a strategic group structure. Large increases in the criterion value were postulated to signify inappropriate grouping, suggesting where to stop the aggregation of firms into successive clusters. This heuristic decision rule was supplemented with a multivariate analysis of variance (MANOVA) on the centroids defined over the averaged strategy variables for each strategic group. This was done to determine whether statistically different clusters existed. That cluster structure was selected where MANOVA-testing pointed to significant differences in the cluster centroids and where subsequent levels of aggregation resulted in nonsignificant differences between the cluster means. 5 Variable Selection and

Measurement

Two major sets of variables need to be developed. For the strategic grouping work, a rich set of variables capable of describing the competitive dimensions of the industry in question is required. To conduct the strategy-performance study, different performance measurement variables are required. Strategy Variables. The specification of particular strategy variables depends on the industry selected, in this case upon the U.S. pharmaceutical industry during the period 1963-1982. This industry was selected for the following reasons: ( 1 ) preliminary research indicated different firms pursued different strategies; (2) the 1962 amendments to the 1938 Food, Drug and Cosmetics Act significantly altered the environmental context in which drug firms had to compete; (3) detailed data bases appeared to be present, and, finally, (4) this industry had not yet received significant attention in strategic management research. The 20-year period was chosen to permit evaluation of the temporal stability of any observed strategic groupings. Actual selection of strategy variables was performed in two stages. In the first stage, a detailed study of the drug industry (Cool 1985, pp. 195-298), coupled with discussions with industry executives and experts, was conducted. A second stage dealt with the actual selection of the specific variables chosen to represent strategic scope and resource commitments. Table 1 provides a summary of the variables chosen. Scope Commitment Variables. Scope commitments in the U.S. drug industry can be described along three major dimensions: (1) the range of market segments that are targeted, (2) the types of products to compete in selected market segments, and (3) geographic scope. Decisions about the range of market segments were measured by establishing to what extent firms compete in a smaller or larger number of the eleven "therapeutic categories" 6 (FOCUS) used in this study, and to what degree firms target 5 The Ward criterion was chosen because it has been established that it uncovers best the "natural structure" in the data among the class of hierarchical cluster routines (see Punj and Stewart 1983). To evaluate the stability of the results, the "diameter" method was also used. Substantially identical cluster results were obtained. Finally, note that the MANOVA-procedure to determine the appropriate level of clustering is similar to other recently proposed "stopping rules" (see, e.g. Arnold 1979). 4 Eleven therapeutic categories were defined: cardiovascular^, nutritional products, pain control, internal medicine, mental health, topicals, anti-infectives, respiratory, IV fluids, cancer therapy, and a residual category. The reader can obtain a detailed overview of these categories from the authors.

197

198

Κ. Ο. Cool/D. Schendel 1110

KAREL O. COOL AND DAN SCHENDEL TABLE 1 Variables Describing Strategy in the U.S. Pharmaceutical Industry Strategy Dimension

SCOPE COMMITMENTS·. Range of Market Segments 1. Breadth of Scope (FOCUS) 2. Commitment to Ambulatory Care Market (DRUGST) Types of Products 3. Commitment to the Ethical Drug Market (Rx) Commitment to the Generic Drug Market 4. Branded Generics (BRANGEN) 5. Commodity Generics (COMMGEN) 6. Commitment to the maintenance drug market Geographic Scope. 7. Spatial Reach (FOREIGN) RESOURCE COMMITMENTS: Research and Development Commitments: 8. Current R&D spending (RDINTEN) 9. R&D Capital Stock 10. R&D Orientation (RDORIENT) Marketing Commitments: 11. Product Strategy (PRODSTR) Promotion Strategy 12. Promotion to the medical profession (PROFPROM) 13. Advertising to the consumer (CONSADV) 14. Distribution Strategy (DISTR) Size: 15. Scale of drug operations (SIZE)

Measure

1. (Rx sales in 3 largest therapeutic categories)/ (Total domestic Rx Sales) 2. % Drug store sales in total domestic drug sales

3. % Rx sales in total domestic drug sales

4. % Branded generic Rx sales in total domestic Rx sales 5. % commodity generic Rx sales in total domestic Rx sales 6. % maintenance drug sales in total domestic Rx sales 7. % total firm sales generated abroad

8. (Total firm R&D)/(Worldwide health care sales) 9. (Cumulative Number of NDAs submitted)/(Cumulative number of INDs submitted) 10. (Cumulative number of NCEs approved)/(Cumulative number of NDAs submitted) 11. (Cumulative number of NCEs introduced)/(Cumulative number of all products introduced) 12. (Total domestic professional promotion)/(Total Rx sales) 13. (Total domestic PTY drag advertising)/(Total domestic Rx sales) 14. % of total domestic drag sales shipped directly to drag stores and hospitals 15. Ln (Total domestic drag sales)

the hospital versus the ambulatory care (physician based) market (DRUGST). The second set of variables concerning the types of products selected by the competing firms to offer to market segments was measured by a set of proportions: (1) prescription relative to non-prescription drugs, (Rx); (2) branded drugs with patent7 protection versus branded drugs without patent protection, i.e., branded generics (BRANGEN); (3) branded drugs on which patents have expired, (off-patent drugs) versus those that are unbranded, i.e., commodity generics (COMMGEN); and (4) drugs for maintenance use (chronic) versus drugs for acute use (MAINT). Finally, the spatial or geographic reach of market activity was determined by the proportion of foreign to total firm sales8 (FOREIGN). ' Because they offer more protection, product rather than process patents are considered here. * Owing to a lack of reliable data on the split between U.S. and non-U.S. drug sales, the ratio of total foreign to total sales was used as a proxy value.

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1111

Resource Commitment Variables. Measures of resource commitments were based on an analysis of the key requirements for establishing a competitive advantage in the pharmaceutical industry. The primary functional areas from which competitive advantage is likely to come were determined to be R&D and marketing. Three variables were used to measure R&D resource commitments (see Table 1 ). The R&D-to-sales ratio measures the intensity of current R&D spending (RDINTEN). The second measure, the ratio of the cumulative number of New Drug Applications (NDAs) to the cumulative number of Investigational New Drugs (INDs), both submitted to the U.S. Food and Drug Administration, is a proxy for accumulated R&D competence in developing a potential drug from its initial discovery stage to a stage closer to approval for marketing. 9 This reflects the R&D capital stock (RDCAPIT) resulting from past commitments to R&D. Finally, the third measure, the ratio of the cumulative number of New Chemical Entities (NCEs) approved to the cumulative number of N D A s filed, was included as a measure of R&D orientation (RDORIENT) towards genuinely new drugs as opposed to mere modification or combination products. Three variables were considered to measure marketing commitments: ( 1 ) promotion outlays, (2) product strategy, and (3) distribution strategy. Promotional efforts were divided into promotion to the medical and paramedical professions for prescription drugs10 (PROFPROM), and direct media advertising to the consumer for nonprescription drugs (CONSADV). Product strategy is related to the reliance on new versus modification or combination products, and as such relates to competitive advantage sought through innovation rather than through followership. Product strategy is measured by the ratio of the cumulative number of NCEs introduced to the cumulative number of all drugs introduced" (PRODSTR). A key marketing strategy decision is whether the distribution system is direct, captive, or depends upon outside agents. Here, the commitment to distribution strategy was measured by computing the percentage of total domestic drug shipments distributed directly to drug stores and hospitals (DISTR). The size of the firm influences the ability to allocate different amounts of resources. An absolute measure of size, total domestic drug sales, (SIZE), was used to complement the relative measures of R&D and promotion. Performance Variables. Three different sets of performance variables were constructed: (1) drug market share (MS), (2) weighted segment share (WSS), and (3) inflation-adjusted return on sales (AROS). Market share, MS, is defined as the firm's share of prescription drug sales of total domestic market volume. Weighted segment share, WSS, is constructed as follows: WSS

where

Wj = Sj/S,, and, SSj = Sj/Sj, Sj = drug sales of firm i in therapeutic category j, S¡ = total prescription drug sales of firm i, Sj = total market volume of therapeutic category j. 9 Cumulative rather than annual data were employed to smooth out short-term fluctuations, as well as to account for the lags between IND and NDA filing, and to record inter-firm differences in the total amount of INDs and NDAs filed. IND filing became mandatory in 1963. In that year, firms also had to submit INDs for drugs under development during the period preceding the 1962 Amendments. To reduce the bias resulting from this retrospective filing of INDs in 1963, the average of the number of INDs submitted during 1964-1966 was taken as an estimate of the 1963 number of INDs. 10 Promotion expenses for prescription drugs include outlays for detailing, journal advertising and direct mailing to the medical and paramedical profession. 11 New dosage and package forms were excluded from the total number of new drugs introduced.

199

200

Κ. Ο. Cool/D. Schendel 1112

K A R E L O. C O O L A N D D A N S C H E N D E L

W S S measures the extent to which firms dominate the market segments in which they compete. This indicator was found to be very closely related to profitability by Bond and Lean (1977), and Schwartzman (1976). Finally an attempt was made to construct a profitability measure for the pharmaceutical operations of the firms competing in the drug industry. Since most firms are diversified, and since the asset or investment base for the drug operations is not available for a large sample of firms over a long period of time, this study had to rely on a return on sales measure. AROS was defined as the ratio of net income before interest and taxes for the pharmaceutical operations to total drug sales for the firm. This measure was adjusted for inflationary changes by restating depreciation charges, a correction necessary in longitudinal analyses. 12 While sales margin is usually not the best profitability measure that could be used, objections to it are lessened because of the single industry studied and the generally comparable environmental conditions for all competitors. Moreover, more direct measures are not available. For each of the three performance indicators selected, the temporal mean for each firm was computed over each time period for which a stable strategic group structure was observed. Risk measures were obtained by calculating the standard deviation about the temporal mean for each of the three performance measures for each period of stable strategic group structure. Risk-adjusted measures of performance were obtained by dividing average firm performance for each of the three indicators by their respective standard deviation. Data Sources This study used a much more extensive and detailed data base than any prior study has used. Most studies do not go beyond the PIMS or COMPUSTAT™ data bases and therefore must accept limitations as to firm anonymity and the consequent impossibility of matching specific market and environmental information to specific firm performance. This study used detailed and complete information that could measure strategy components and match them to time and performance. The detailed data were made possible by the cooperation of many different individuals associated with the pharmaceutical industry, and because the drug industry is so heavily monitored by governmental regulation resulting in a good deal of publicly available information. The data bases of IMS America constituted the major source of information on sales and marketing expenditures. Other data bases used included the Paul de Haen New Drug Analysis and New Product Survey, FDA data on INDs and NDAs, annual reports and 10-K statements, reports by Frost and Sullivan, Leading National Advertisers, Drug Topics, Advertising Age, Chemical Abstracts and the Merck Index'*. For the performance variables, the IMS data bases and 10-K line of business reporting provided the majority of the data. 14 A full description of the reliability and use of the data bases can be found in Cool (1985, pp. 325-337).

12 Because of inflation, asset book values and depreciation charges tend to be understated, leading to an overstatement of operating income. T o alleviate this bias the following procedure was used. First, an estimate was made of average fixed asset life. This estimate was used to restate depreciation expenses in current dollars of each base year. The difference between the computed and reported depreciation charges was then deducted from operating income to obtain the "adjusted" net income. This adjustment, as well as a restatment of R & D expenses (from expense to a depreciation of capitalized R&D) was made. 13 Without access to all of these data bases, the construction of the variables would not be possible. Almost all variables had to be transformed or combinations of data from different sources had to be formed. Without extensive information it would not be possible to use the type of data necessary to form strategic groups across time. 14 Line-of-business reporting has been available only since 1966, and then its nature is frequently changed. The computation of a divisional ROS measure poses significant problems, but these measures are superior to overall company figures.

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1Í13

Sample Selection Several criteria guided selection of the firms to be included. First, non-U.S. firms were excluded because of the lack of data. Second, firms needed to exist as separate legal entities over most of the 1963-1982 period to avoid biases resulting from a changing sample composition. Third, all firms whose business was exclusively in the generic drug segment were excluded because they operate on a different basis and have been an industry factor only since the mid-seventies. Fourth, firms were required to have significant commitments to the ethical drug industry. In total, 22 firms were sampled (see Table 3). Throughout, care was taken to account for acquisitions, mergers and divestments because these activities can represent changes in strategy. Strategic Groups in the U.S. Pharmaceutical Industry, 1963-1982 In this section application of the methodology outlined above for identifying strategic groups over time is reported along with the results obtained. It is the strategic groups rather than individual firms comprising them that are of interest, because it is these groups that are of interest to the later study of performance differences. Identification of Transition Points The methodology described earlier was applied sequentially to 19 randomly selected firms from the total of 22. The procedure was then repeated on a sample of 20,21, and then 22 firms to establish whether a changing sample composition would alter the observation of transition years. Table 2 summarizes the test results on the equality of the variance-covariance matrices, the central feature of the procedure. All tests were conducted at a significance level of one percent (a = 0.01 ). The test results were robust for all sample sizes, and most importantly, yield similar transition points over the 20-year period for all samples tested. The identified periods of four stable strategic group structures are respectively, 1963-69,1970-74,1975-79, and 1980-82. No significant F- value was found for the last period, implying that the group structure emerging since 1980 had not yet changed by 1982." TABLE 2 Identification of Periods of Stable Strategic Group Structure Sample Size

Periods

No. Years

η = 19

1963-1969 1970-1974 1975-1979 1980-1982 1963-1969 1970-1974 1975-1979 1980-1982 1963-1969 1970-1974 1975-1979 1980-1982 1963-1969 1970-1974 1975-1979 1980-1982

7 5 5 3 7 5 5 3 7 5 5 3 7 5 5 3

η = 20

π = 21

π = 22

F* 1.53 1.56 1.66 •No Test 1.44 1.63 1.70 *No Test 4.62 1.57 1.67 *No Test 1.39 1.42 1.40 *No Test

Probability 0.0009 0.0005 0.0001

Significant 0.0038 0.0001 0.0000 Significant 0.0000 0.0004 0.0001 Significant 0.0077 0.0050 0.0091 Significant

* Value of the F-statistic when a year is added to the periods specified in the same row. 15 Given a certain sample size, many tests had to be performed according to Arnold (1979) to establish whether a transition point was reached. Significant F-values were never found in isolation. Each time, several significant F-values were observed, pointing to the existence of a transition point.

201

202

Κ. Ο. Cool/D. Schendel 1114

KAREL O. COOL AND DAN SCHENDEL

The previous statistical procedure alone, however objective, cannot establish whether the transitions observed are owed to exogenous shocks in the environment, are triggered by autonomous firm actions, or are due to some combination of both. However, study of the U.S. pharmaceutical industry (Cool 1985, pp. 209-270) indicates that major changes have taken place over the twenty year period that can be associated with the observed strategic group changes. The 1962 Amendments, requiring firms to prove that newly developed drugs were effective as well as safe, provided a first major exogenous change. One effect of this change in regulation was that firms had to be more selective in their R&D commitments and product strategy, and be more explicit about their strength in R&D. This change also affected the range of attainable scope actions for firms. Since product-market changes take time to materialize in the pharmaceutical industry, it should not be surprising that the effects of this legal change were observable only after several years. During the late sixties and early seventies, other major environmental changes took place, including these: increasing importance of generic prescriptions following the patent expirations of many major drugs; the 1968 enactment of the Drug Efficacy Study Implementation (DESI) program to examine the efficacy of drugs marketed in the pre-1962 period, which led to the withdrawal of a large number of drugs; the 1972 OTC Review evaluating the efficacy of over-the-counter drugs; the expanded use of Abbreviated New Drug Applications (ANDAs) making the marketing of rival products for drugs coming off-patent easier; and the general decline in the development of New Chemical Entities. This set of changes also influenced the conduct of firms in the first half of the seventies and probably relates to the observation of a different strategic group structure in the second half of the seventies. Finally, repeal of State Anti-Substitution Laws and the institutionalization of the Maximum Allowable Cost (MAC) Program in 1975 both encouraged greater price competition in the U.S. drug industry. Coupled with the increasing participation of non-U.S. drug firms in the development and marketing of new drugs, these changes may also have contributed to the emergence of new strategies seeking new ways of creating competitive advantage and finally to yet a different strategic group structure in the early eighties. A more extensive analysis would be needed to establish causal relationships between these changes and the observed patterns of strategic group formation. This overview illustrates, however, that a meaningful interpretation may be attached to the statistical finding that periods with a distinct strategic group structure occurred in the 1963-82 period. Identification and Description of Strategic Groups The previous analysis cannot determine how firms position themselves. To determine those competitive alignments, a cluster analysis was undertaken to identify the strategic groups in each of the four time periods (Table 3). The observed patterns of strategic group formation are summarized in Table 4. This table depicts the changes in group composition and lists the cluster centroids relative to the industry averages for all strategic groups in each period. Patterns of Strategic Group Formation In the sixties, six strategic groups were found to characterize competition in the U.S. drug industry. The first group (SGli) consisted of large, R&D-intensive prescription drug firms competing in many market segments with a broad range of products. The second group (SG2i) included large firms, but differed from the first group on several key dimensions: (1) they were advertising-intensive and not R&D intensive; (2) they participated in the nonprescription as well as the prescription market segments; and (3) they participated in fewer market segments in the prescription drug market and offered

Strategic G r o u p Formation and P e r f o r m a n c e STRATEGIC GROUP FORMATION AND PERFORMANCE

1115

TABLE 3 Strategic Groups in the Period 1963-1982: Group Membership and MANO VA Test Results PERIOD I: 1963-1969. i^WILKS) = 3.105 (p = 0.028) SG2

SGI Abbott Lederle Lilly Merck Squibb Upjohn

American Home Bristol-Myers SmithKline Sterling Drug

SG3

SG5

SG4

Johnson & Johnson Searle Carter-Wallace Morton-Norwich Warner-Lambert Robins Pfizer Rorer Richardson-Vicks Schering-Plough Syntex

SG6 Marion

PERIOD II: 1970-1974. flWILKS) = 5.476 (p = 0.005) SGI

SG2

Abbott American Home Lederle Lilly Squibb Warner-Lambert

Bristol-Myers Carter-Wallace Johnson & Johnson Morton-Norwich Richardson-Vicks SmithKline Syntex

SG3 Merck Pfizer Schering-Plough Searle Sterling Drug Upjohn

SG4 Robins Rorer

SG5 Marion

PERIOD III: 1975-1979. ÍIW1LKS) = 6.887 (p = 0.000) SGI Abbott American Home Bristol-Myers Lederle Warner-Lambert

SG2 Lilly Merck Pfizer Schering-Plough Squibb Sterling Drug Upjohn

SG3

SG4

Johnson & Johnson Carter-Wallace Morton-Norwich Marion Richardson-Vicks Rorer Robins Searle SmithKline Syntex

PERIOD IV: 1980-1982. Í1(WILKS) = 2.623 (p =• 0.049) SGI Lilly Abbott American Home Merck Bristol-Myers Upjohn Pfizer SmithKline Warner-Lambert

SG2

SG3

SG4

Johnson & Johnson Searle Schering-Plough Syntex Squibb Sterling Drug

SG5

SG6

Carter-Wallace Lederle Marion Morton-Norwich Richardson-Vicks Robins Rorer

a less comprehensive product range. T h e third strategic group ( S G 3 | ) was c o m p o s e d o f medium-sized firms pursuing a predominantly " m e - t o o " strategy, indicated by their emphasis o n m e - t o o drug d e v e l o p m e n t rather than original drug research, and by their heavy c o m m i t m e n t to p r o m o t i o n to professionals and consumers. S G 3 i firms were c o m p e t i n g in both the nonprescription and prescription drug markets, but in the latter segment, with a narrow product range. T h e fourth strategic group ( S G 4 0 also consisted o f medium-sized firms, and were like the third group, except they lacked any real R & D competence. T h e fifth strategic group (SG5i) c o m b i n e d small prescription drug firms with a narrow product range, a selective participation in market segments, and m e - t o o drug development. Their heavy emphasis o n professional p r o m o t i o n was an outstanding characteristic. Strategic group six ( S G 6 | ) consisted o f only o n e member, and its very

203

204

Κ. Ο. Cool/D. Schendel

Si SA S Sw Sκι S I< U -Q 5

S 8.

8à. 8u>8K)8 »Î & sUt s^ sUsMs ~§• Q.

+++ II

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1117

small size, very focused product line, narrow scope and negligible R&D effort, left Marion Laboratories by itself. What was observed in the initial time period is major grouping by size, with large, medium and small components, each with two groupings. The large firms seem to lead with R&D and innovation in products, while all of the other firms seem to rely more on marketing emphasis, through imitating products, or by using selective products with selective markets and intensive marketing to them. These groupings bear remarkable similarities to those found in the studies of the brewing industry (Hatten and Schendel 1977; Hatten, Schendel and Cooper 1978). These groups went through various changes in the three subsequent time periods found, as Tables 3 and 4 indicate. By comparing the sixties with the first half of the seventies, time period I with period II, some interesting changes are observable: (1) the number of strategic groups declined from six to five, reflecting some reduction in strategic asymmetry between firms; (2) groups SG2i and SG4| had no counterparts in the second time period, suggesting that their strategies were not viable; (3) SGI· showed an expanded scope to include nonprescription market segments, with some decrease in R&D intensity; (4) a new strategic group, SG2u, in terms of component firms, was formed; and (5) about half of the firms who changed their strategy opted to imitate other firms more closely, the other half chose to develop a new posture entirely (SG2H). Comparing time period II to period III, even more firms altered their strategic actions, reducing the number of strategic groups to four, with these characteristics: (1) the first group, S G l m , leaned toward nonprescription market segments even more, with increased consumer advertising, still less R&D emphasis, and fewer original drugs; (2) SG2|n, on the other hand, moved more toward the prescription market, maintained its R&D concentration, and decreased its relative emphasis on marketing; and (3) SG3in and SG4m made few changes from the prior period. There was a reduction in asymmetry as firms tried to compete more like their competitors rather than in different ways. From period III to period IV the number of groups moved from four to six as firms again started to diverge. These changes are notable: (1) SGlm maintained its earlier profile and SGliv increased its membership, while losing one firm to a new group SG6i V ; (2) SG2|v split into two strategic groups, with some firms altering their scope commitments to that resembling SGli and others expanding their product range with me-too drug development and intensive advertising support; (3) SG3m showed the most dramatic change as its member firms moved to four different strategic groups, with SG5iv looking much like SG4m; and (4) the new SG4i V comprised firms with R&D emphasis competing in a small number of market segments with a narrow product range.

Discussion of Empirical Findings Overall, the pharmaceutical industry appears to have gone through profound structural changes in the 1963-1982 period. The sixties showed differentiation, where a clear strategic group structure with high asymmetry is observable. In the early seventies, firms experimented with new strategies for them, perhaps motivated by environmental change and their own lack of success. In the late seventies, a consolidation occurred in terms of strategic groups, only to be broken apart in the early eighties as new groups formed. While it is very premature, one is tempted to conclude that there is a cycle that seems to be followed of experimentation, imitation, and then new experimentation, with stable successful firms reinforcing what they do, while the less successful or more ambitious experiment through either imitation or innovation in R&D and/or marketing.

205

206

Κ. Ο. Cool/D. Schendel 1118

KAREL O. COOL AND DAN SCHENDEL

The prior discussion has been based only on the findings of the statistical work and does not capture the full range of changes that have occurred in the U.S. drug industry over the 1963-82 period. However, this longitudinal analysis does indicate that it is possible to objectively work with the dynamics of industry competition and structure. Static analyses and inferences drawn from them are valid only when an industry is in equilibrium and under conditions where asymmetries are protected in some fashion. Study of the structure of the drug industry at any one of the four periods uncovered in this analysis would prove invalid during the next period, however valid it might be at a given point in time. To make inferences about the future state of competition and alternative strategic postures, one of the more attractive, useful, and practical applications of the strategic group concept, would seem to require longitudinal, or dynamic analyses. Longitudinal analyses should also shed light on the relative difficulty of implementing a genuine shift in strategic posture. Strategic repositioning does not appear to be a matter of mere "choice". The analysis of the drug industry attests to the apparent difficulty of implementing sustained group shifts. Tables 3 and 4 show that while many firms attempted to significantly alter their strategic commitments, some could or did not sustain it. In several instances, reversals of earlier scope and resource commitments were observed. Apparently, the development of new competences necessary to successfully compete in a new mode, i.e. to perform a group shift or form a new strategic group, appears to be significantly constrained by past commitments and carries with it risks that differ depending upon which firm attempts the shift. Performance Consequences of Strategic Group Membership Different strategic groups use different strategies and because the scope and resource commitments differ, the competitive advantages achieved by each group should differ as postulated earlier. Within each group the firms may differ in performance as well, owing mainly to the efficiency with which they are able to execute the strategy. An empirical analysis of firm differences is beyond the scope of the study, however, three different performance indicators (MS, WSS, AROS) were defined for each group and three performance dimensions (levels of performance, risk and risk-adjusted performance) were considered. Since the strategic group structure differed over time, an analysis was performed for each time period. Performance Level Differences Between Strategic Groups A one-way analysis of variance was used to test whether average performance levels differ among strategic groups (H,). The results are summarized in Table 5. For MS, the results show significant differences at the 95 percent confidence level between the strategic groups in all time periods. Observed over two decades, this finding provides strong evidence for existence of inter-strategic group performance differences. The second performance indicator, WSS, shows a less unequivocal pattern. In the first and fourth time periods, periods showing high asymmetry, the WSS indicator attests to highly significant inter-strategic group differences in performance (a = 0.05). However, in the two intermediate periods, where more symmetry exists, no such differences in WSS were observed. This finding that performance differences are not significant in periods with lower strategic asymmetry suggests that this structural characteristic needs to be explicitly considered in strategy-performance studies. Possibly, the conflicting results of previous strategic group studies may have been affected by this structural phenomenon. AROS, the remaining performance indicator, follows yet a different pattern. Only in the last time period is there some support for profitability differences (a = 0.094).

Strategic Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1119

TABLE 5 Comparison of Performance Between Strategic Groups (ANOVA F-value and Significance of F) Period

WSS

MS

AROS

Performance Levels 1963-1969 1970-1974 1975-1979 1980-1982

6.8 4.3 4.2 9.6

(0.002) (0.019) (0.020) (0.001)

6.7 (0.002) 1.7(0.198) 2.9 (0.067) 4.2(0.018)

0.4 (0.797) 0.7 (0.562) 1.0(0.398) 2.4 (0.094)

Risk Exposure 1963-1969 1970-1974 1975-1979 1980-1982

1.9 (0.158) 4.3 (0.020) 1.1 (0.370) 0.8 (0.524)

1963-1969 1970-1974 1975-1979 1980-1982

0.03 3.8 2.1 0.8

2.4 (0.098) 0.1 (0.962) 1.3(0.320) 0.5 (0.705)

0.8 (0.524) 0.1 (0.996) 0.9 (0.469) 2.0(0.154)

Risk-Adjusted Performance (0.998) (0.031) (0.138) (0.560)

1.1 (0.400) 1.1 (0.392) 0.6 (0.598) 1.3 (0.298)

1.2(0.337) 0.6 (0.650) 1.4(0.273) 0.9 (0.474)

Overall, some support exists for the claim that strategic group membership has performance implications, especially in terms of market share. However, those market share differences do not appear to result in profitability differences.16 Various explanations are possible. First, the profitability indicator may be affected by nonrandom measurement error. Although possible, maximum care was taken in the construction of the indicator to avoid this bias. Second, members of the same strategic group may realize different profitability levels. These within-group differences might prevent the observation of between-group profitability differences. As indicated above, not all group members may be equally efficient in carrying out their current strategy, leading to performance variation. Finally, different strategies (strategic groups) may have different risk attributes, rendering the unadjusted profitability measures insufficient to evaluate performance differences. Differences in risk characteristics and risk-adjusted performance therefore need to be considered. Risk Differences Between Strategic Groups The test results regarding the differences in risk characteristics between strategic groups (H 2 ) are summarized in Table 5. The hypothesis that strategic groups display different risk characteristics is not borne out by the test results. Only for MS in one period (1970-74) is a difference with a 95 percent confidence interval observed. In all other cases, the test results support the view that different strategic groups are characterized by similar risk profiles, contradicting the findings of Oster (1982). Oster's procedure, however, used only one variable, advertising intensity, to identify strategic groups, which may have introduced some confounding effects. It is significant that no difference was found in risk for different strategic profiles. Risk may be firm-related, rather than a function of the type of strategy employed, and may have to do with the match between past commitments of scope and resources and " Nor does pairwise comparison of group means, using t-tests, point to profitability differences between the strategic groups in any of the four periods (see Cool 1985, pp. 388-429).

207

208

Κ. Ο. Cool/D. Schendel 1120

KAREL O. COOL AND DAN SCHENDEL

currently pursued strategic actions, rather than just current commitments alone. This suggests that firms within a strategic group face different levels of risk. Thus, even though differences in risk exposure between strategic groups may not be significant, differences within a given strategic group may be significant. All this implies that if a firm attempts to reorient its scope and resource commitments in ways incompatible with prior commitments, it is likely to have a larger risk exposure than other strategic group incumbents with a more harmonious correspondence or transition between past and present strategic commitments. Risk-Adjusted Performance Differences Among Strategic Groups A one-way analysis of variance was performed to test H 3 that strategic groups would have similar risk-adjusted performance. No significant differences in average risk-adjusted performance were found using a measure of the ratio of mean performance over the standard error of the performance indicator. (See Table 5.) This finding is postulated to stem from differences in risk exposure among strategic group members rather than from the existence of a positive risk-return relationship among strategic groups.17 More research is needed however to verify this relationship. Discussion and Conclusions Despite the fact that strategic groups have been studied since the early seventies, little empirical attention has been given to the central question of whether strategic groups really are structural phenomena in industries. Yet, strategic group analysis is meaningless as long as it has not been determined whether these groupings are more than random events. It is therefore imperative that longitudinal analyses be performed. The findings presented here constitute a first substantiation that strategic groups are a relatively stable phenomena. During the 20-year period only four subperiods with a distinct strategic group structure could be identified. These periods ranged in duration from five to seven years." Distinguishing the various strategic group structures from each other were the number of strategic groups in each period, the varying degree of strategic asymmetry between the groups, and the nature of the strategic commitment changes in each period. Even though strategic groups were observed to be relatively stable, the observation that they change raises the question of what forces triggered these changes. Evidence was presented that major environmental changes prompted firms to alter their scope and resource commitments. The analysis of inter-strategic group performance differences also suggests a second source of instability in industries. Firms may have been prompted to imitate the strategic commitments of rivals by the observation that these have a higher performance in terms of market dominance. These attempts at imitation, in the absence of fast-responding feedback mechanisms about the effectiveness of these actions, may have further destabilized strategic group structures. The combination of exogenous discontinuities (environment) and endogenous imitation activities probably functioned as powerful forces in upsetting structural equilibria. Nevertheless, structural transformation did not come about quickly. The relative length of each period with a distinct strategic group structure reflects this. Substantial mobility barriers seemed to affect the transformation process. Based on the findings reported here as well as in other studies, several hypotheses about barriers in structural

" When the relationship between risk and performance is negative or nonlinear, the ratio measuring risk-adjusted performance would indicate differences between the strategic groups. " The last period (1980-1982) is not considered because no statistical evidence was found in support of the hypothesis that a new strategic group structure already existed.

Strategie Group Formation and Performance STRATEGIC GROUP FORMATION AND PERFORMANCE

1121

evolution processes can be formed. A major barrier appears to be the degree of incongruity between past and intended strategic commitments. To the extent that input markets for required resources is imperfect and incomplete (Arrow 1974, pp. 22-23), the less feasible it will be and the longer it will take to build new capabilities. In addition, risk exposure is likely to increase as strategic actions are pursued while the needed capabilities have not been developed. Another mobility barrier is suggested by the work of Lippman and Rumelt (1982). They argue that it may not be possible to specify fully the set of inputs and the input-output transformation function underlying given strategies. Even if the input market for resources were perfect and complete, uncertainty about the way these resources are combined to yield competitive advantage prevents fast imitation. To the extent input markets are imperfect and incomplete, and uncertainty characterizes input-output functions, imitation of observed sets of strategic commitments poses formidable problems. Further affecting the imitation process is the feedback mechanism by which responses from the output markets are translated into organizational actions. If fast feedback about the effectiveness of new scope and resource commitments can be obtained, corrective actions can be taken quickly. Consequently, a disturbed equilibrium may move faster to a new equilibrium. In the absence of fast responses, or a slow organizational processing of these responses, sustained periods of disequilibrium may occur. In the pharmaceutical industry, slow feedback mechanisms appear to be present. Probably, these slow market feedback processes, or the slow organizational processing of these responses, has contributed to the rather long period where "experimentation" occurs followed by a new structure. Combining the various elements of exogenous discontinuities, endogenous imitation, input market imperfection and incompleteness, and feedback mechanisms, a dynamic framework of competition, strategy and performance can be built. Central in this framework is the strategic group concept enabling a tracing of changes in competition, strategy and performance in a systematic way. The framework opens up many opportunities for research. A first opportunity relates to the analysis of risk differences among strategic group incumbents. Why risk would differ needs to be further explored theoretically as well as empirically. The results reported here strongly suggest that an empirical analysis of risk differences among strategic group members will find significant differences. An interesting extension of this would be to relate risk to return, and analyze what relationship exists. Finance theory suggests that positive relationship should be found. The finance argument rests however on conditions of equilibrium, while the present analysis points to disequilibrium conditions in the drug industry. Combining these observations with Bowman's (1980, 1982) arguments on the risk-seeking behavior of what he called "troubled" firms, one might expect that in any strategic group risk-seeking firms (with negative risk-return relationships) and risk-averse firms (with positive risk-return relationships) will be found. The reasons were indicated earlier: some firms change their scope and resource commitments far beyond their accumulated competences. However, once the overcommitment is recognized it cannot be quickly adjusted because of input market imperfections and incompleteness and phenomena of "uncertain imitability". Thus, while in equilibrium conditions risk-return relationships of strategic actions should be positively related, negative relationships may be observed in states of disequilibrium. The analysis of risk-return relations may provide important insight into the nature of strategy-performance relationships. A further extension of research relates to the explicit consideration of the competitive effects of strategic groups in industries. This paper considered only the hypothesis that performance is different across strategic groups. If the concept of strategic groups is to

209

210

Κ. O . C o o l / D . S c h e n d e l

1 122

KAREL O. COOL AND DAN SCHENDEL

be integrated in a theory of competition, the competitive effects of strategic group structures will need to be developed and tested, going beyond the previous analyses. In particular, whether rivalry indeed increases, and performance decreases, when firms from different strategic groups compete in the same market segment needs to be tested (Porter 1979). Likewise, the incidence of rivalry among group incumbents needs to be studied further. Are some groups characterized by high rivalry, and is this the source of low performance despite the presence of high mobility barriers? Alternatively, are some groups characterized by tacit collusion, preventing rent destruction despite high segment interactions? These are important issues that need to be addressed empirically to provide greater insight into the competitive effects of strategic groups. The focus of strategic group analysis need not be restricted to rivalry and performance analysis, however. Another important extension rests in the prediction of conduct on the basis of strategic group membership. For example, one might study the foreign direct investment decision in terms of strategic group affiliation: are incumbents characterized by similar expansion strategies abroad? Alternatively, one might study how foreign firms enter a domestic market. Caves and Porter (1977) suggested that a "circuitous path" might characterize the entry process in markets. Groups with lower mobility barriers were postulated to be the first target, to be followed by shifts to groups with higher mobility barriers and potential rents. These research questions are important and empirical analyses would expand our understanding of the impact of strategic groups on conduct and rivalry. To address these issues, however, longitudinal analyses are needed to provide meaningful results. The procedures used in the present paper can, however, be employed to approach these research questions. In summary, strategic group analysis has significant potential to shed light on important issues of performance, rivalry, and conduct. The above avenues for research are only a small subset of the questions that can be addressed. If empirical analyses are executed carefully, strategic group study should prove to be a valuable domain for future research in strategic management. 19 " Thanks are due to IMS America and Paul de Haen Int. for permitting access to their data bases, and to Dr. Douglas Cocks, Mr. Ronald Matricaria and Mrs. Mary-Lou Mason of Eli Lilly for their help in making the data collection feasible. Professor Stephen W. Schondelmeyer of Purdue University's School of Pharmacy was essential to understanding the technical aspects of the pharmaceutical industry. The comments of Dr. Ph. Naert (Dean) and Dr. Ingemar Dierickx of INSEAD, as well as two anonymous reviewers, helped to improve the paper.

References ABELL, DEREK, Defining the Business, Prentice-Hall, Englewood Cliffs, N.J., 1980. ACKOFF, RUSSELL L., A Concepì of Corporate Planning, Wiley-Interscience, New York, 1970. ANSOFF, H. IGOR, Corporale Strategy, McGraw-Hill, New York, 1965. ANDERBERG, MICHAEL, Cluster Analysis for Applications, Academic Press, New York, 1973. ARNOLD, MICHAEL, "A Test for Clusters," J. Marketing Res., (November 1979), 545-551. ARROW, KENNETH, The Limits of Organization, W. W. Norton, New York, 1974. BAIN, JOE, Barriers to New Competition, Harvard University Press, Cambridge, 1956. BERNSTEIN, LEOPOLD, Financial Statement Analysis, Richard D. Irwin, Homewood, II., 1974. BETTIS, R. AND W. HALL, "Diversification Strategy, Accounting Determined Risk, and Accounting Determined Return," Acad. Management J., (1982), 254-264. BOND, R. AND D. LEAN, Sales, Promotion and Product Differentiation in Two Prescription Drug Markets, FTC, Washington, D.C., 1977. BOWMAN, EDWARD, "A Risk/Return Paradox for Strategic Management," Sloan Management Rev., Spring 1982a, 33-42. , "Risk Seeking by Troubled Firms," Sloan Management Rev., (Summer 1982b), 33-42. BUZZELL, ROBERT D . , BRADLEY, T . GALE, AND RALPH SULTAN, " M a r k e t S h a r e — A K e y t o P r o f i t a b i l i t y , " Harvard

Business

Rev.,

( J a n u a r y - F e b r u a r y 1975), 9 7 - 1 0 6 .

CAVES, R. AND M. PORTER, "From Entry Barriers to Mobility Barriers," Quart. J. Economics, (May 1977), 421-441.

Strategic Group Formation and Performance STRATEGIC G R O U P F O R M A T I O N A N D PERFORMANCE

1123

AND T. PUGEL, "lntraindustry Differences in Conduct and Performance: Viable Strategies in U.S. Manufacturing Industries," New York University, Salomon Brothers Center for the Study of Financial Institutions, Monograph 1970—No. 2, 1980. CHRISTENSEN, H. KURT AND CYNTHIA MONTGOMERY, "Corporate Economic Performance: Diversification Strategy versus Market Structure," Strategic Management J., (1981), 327-344. COCKS, DOUGLAS, "Product Innovation and the Dynamic Elements of Competition in the Ethical Pharmaceutical Industry," in Drug Development and Marketing, R. Helms (ed.), ΑΕΙ, Washington, D.C., 225-254. COOL, KAREL, "Strategic Group Formation and Strategic Group Shifts: A Longitudinal Analysis of the US Pharmaceutical Industry, 1963-1982," Ph.D. dissertation, Purdue University, 1985. DAY, GEORGE, Strategic Market Planning, West Publishing Co., St. Paul, 1984. DESS, GREGORY AND PETER DAVIS, "Porter's (1980) Generic Strategies as Determinants of Strategic Group Membership and Organizational Performance," Acad. Management J., (1984), 467-488. DONSIMONI, M. P. AND V. LEOZ-ARGUELLES, "Strategic Groups: An Application to Foreign and Domestic Firms in Spain," Recherches Economiques de Louvain, (September 1981 ), 291 -306. FISHER, FRANKLIN AND JOHN MCGOWAN, " O n the Misuse of Accounting Rates of Return to Infer Monopoly Profits," Amer. Economic Rev., (March 1983), 82-97. FKAZIER, GARY AND ROY HOWELL, "Business Definition and Performance," J. Marketing, (Spring 1983), 59-67. FROST AND SULLIVAN, "The US Generic Drug Market," Report No. 391, New York 1976. , "Generic Drug Market," Report No. 808, New York, 1980. HAMBRICK, DON, IAN MACMILLAN AND DIANA DAY, "Strategic Attributes and Performance in the BCG Matrix—A PIMS-Based Analysis of Industrial Product Businesses," Acad. Management J., (September 1982), 510-529. HARRIGAN, KATHRYN, " A n Application for Clustering for Strategic Group Analysis," Working Paper, Columbia University, 1983. HATTEN, KENNETH AND DAN SCHENDEL, "Heterogeneity Within an Industry: Firm Conduct in the US Brewing Industry," J. Industrial Economics, (December 1977), 97-113 , AND ARNOLD COOPER, "A Strategic Model of the US Brewing Industry," Acad. Management J.,( 1978), 784-810. HERGERT, MICHAEL, "The Incidence and Implications of Strategic Grouping in US Manufacturing Industries," Ph.D. dissertation, Harvard University, 1983. HOFER, CHARLES AND DAN SCHENDEL, Strategy Formulation: Analytical Concepts, West Publishing, St. Paul, 1978. HUNT, MICHAEL, "Competition in the Major Home Appliance Industry, 1960-1970," Ph.D dissertation, Harvard University, 1972. KATZ, R., Cases and Concepts in Corporate Strategy, Prentice-Hall, Englewood Cliffs, N.J., 1970. LIPPMAN, S. AND RICHARD RUMELT, "Uncertain Instability: An Analysis of lnterfirm Differences in Efficiency Under Competition," Bell J. Economics and Management Sci., (August 1982), 418-438. MASON, EDWARD, "Price and Production Policies of Large-Scale Enterprises," Amer. Economic Rev., (March 1939), 61-74. MILES, ROBERT, Coffin Nails and Corporate Strategy, Prentice-Hall, Englewood Cliffs, N.J., 1982. MONTGOMERY, CYNTHIA, "Diversification, Market Structure and Firm Performance: An Extension of Rumelt's Model," Ph.D. dissertation, Purdue University, 1979. MORRISON, DONALD, Multivariate Statistical Methods, McGraw-Hill, New York, 1967. NEWMAN, HOWARD, "Strategic Groups and the Structure-Performance Relationship: A Study with Respect to the Chemical Process Industries," Ph.D. dissertation, Harvard University, 1972. OSTER, SHARON, "lntraindustry Structure and the Ease of Strategic Change," Rev. Economics and Statist., (August L982), 376-383. PORTER, MICHAEL, Interbrand Choice, Strategy, and Bilateral Market Power, Harvard University Press, Cambridge, 1976. , "The Structure within Industries and Companies' Performance," Rev. Economics and Statist., (May 1979), 214-227. , Competitive Strategy, Free Press, New York, 1980. , Competitive Advantage, Free Press, New York, 1985. PRIMEAUX, WALTER, "A Method for Determining Strategic Groups and Life Cycle Stages of an Industry," Working Paper, University of Illinois, Champaign, 1983a. , "The Interdependence of the Life Cycle and Strategic G r o u p Concepts: Theory and Evidence," Working Paper, University oflllinois, Champaign, 1983b. PUNJ, G. AND D. STEWART, "Cluster Analysis in Marketing Research: Review and Suggestions for Application," Marketing Res., (May 1983), 134-148.

211

212

Κ. O. Cool/D. Schendel

1124

KAREL O. COOL AND DAN SCHENDEL

RAMSLER, MARTIN, "Strategie Groups and Foreign Market Entry in Global Banking Competition," Ph.D. dissertation, Harvard University, 1982. RUMELT, RICHARD, "Evaluation of Strategy: Theory and Models," in Strategic Management: A New View of Business Policy and Planning, D. Schendel and C. Hofer(Eds-), Little, Brown, Boston, 1979,196-212. SALAMON, GERALD, "Accounting Rates of Return," Amer. Economic Rev., (June 1985), 495-504. SCHENDEL, DAN AND KENNETH HATTEN, "Business Policy or Strategic Management: A Broader View of an Emerging Discipline," Proceedings, Academy of Management National Meeting, Minneapolis, MN, 1972. , AND A. COOPER, "Can Corporate Strategy be Modelled?," Purdue University, Institute for Research in the Behavioral, Economic and Management Science, Paper No. 517, 1975. AND R. PATTON, "A Simultaneous Equation Model of Corporate Strategy," Management Sci., (November 1978), 1611-1621. SCHWARTZMAN, DAVID, "Pharmaceutical R&D Expenditures and Rates of Return," in Drug Development and Marketing, R. Helms (ed.), ΑΕΙ, Washington, D.C., 1975, 63-80. — , Innovation in the Pharmaceutical Industry, John Hopkins University Press, Baltimore, 1976. STAUFFER, THOMAS, "Profitability Measures in the Pharmaceutical Industry," in Drug Development and Marketing, R. Helms (ed.), ΑΕΙ, Washington, D.C., 1975, 97-119. TASSEY, GREGORY, "Competitive Strategies and Performance in Technology-Based Industries," J. Economics and Business, ( 1983), 21 -40. THOMPSON, JAMES, Organizations in Action, McGraw-Hill, New York, 1967. TIMM, NEIL, Multivariate Analysis, Brooks/Cole, Monterrey, CA, 1975. UYTERHOEVEN, H . , R . W . ACKERMAN AND J . W . ROSENBLUM, Strategy

and Organization,

Richard D.

Irwin, Homewood, IL, 1977. WALTER, INGO, "Competitive Performance and Strategic Positioning in International Financial Services," paper prepared for a conference at the International University of Japan, Tokyo, 1-2 September 1986. WINN, DARYL, Industrial Market Structure and Performance, Graduate School of Business, University of Michigan, Ann Arbor, 1975. Woo, CAROLYN, "Strategies of Effective Low Share Businesses," Ph.D. dissertation, Purdue University, 1979.

Strategie Groups: A Cognitive Perspective

Strategic

Management

Journal,

Vol. 14, 103-124

213

(1993)

STRATEGIC GROUPS: A COGNITIVE PERSPECTIVE R H O N D A K. REGER Department of Management, Arizona State University, Tempe, Arizona, U.S.A. A N N E S I G I S M U N D HUFF College of Commerce and Business Administration, University of Illinois, Champaign, Illinois, U.S.A.

The strategic group concept provides an attractive middle ground between firm and industry for both theory development and empirical analysis. To date, this concept has been defined by researchers in terms of secondary accounting and financial data, and a number of critics have questioned the validity of this work. Our research shows that industry participants share perceptions about strategic commonalities among firms, and that participants cluster competitors in subtle ways not reflected in extant academic research on strategic groups. Decision makers' perceptions and cognitions are phenomena that can be expected to influence industry evolution. They are of research interest as an additional source of data on firm commonalities which helps address concerns about previous strategic group research.

In 1972 Hunt introduced the term 'strategic group' to describe the 'symmetry of operations' he observed in the appliance industry. Hunt noted significant differences in characteristics and strategies among the firms in his study, while at the same time he found that many firms followed similar strategies. Grouping firms clarified understanding of the apparently viable strategic options in the industry. Although no formal definition is universally accepted, a 'strategic group' continues to be commonly defined as a group of firms within the same industry making similar decisions in key areas (Porter, 1980: 129). If these possibilities are thought of as an n-dimensional graph, strategic groups identify clusters of firms in 'strategic space,' and group membership defines the essential characteristics of a firm's strategy. The identification of strategic groups has been used primarily to explore systematic differences in profitability among firms in the same industry Key words: Strategic groups, managerial cognition, cognitive maps, industry evolution, repertory grid technique 0143-2095/93/020103-22$16.00 © 1993 by John Wiley & Sons, Ltd.

(McGee and Thomas, 1986). The strategic groups area is quite active and continues to generate numerous studies. Table 1 summarizes the main studies published since the McGee and Thomas (1986) review. Recent reviewers, however, have voiced several frustrations with the state of strategic group theory and empirical research (Barney and Hoskisson, 1990; Cool, 1985; Hatten and Hatten, 1987; McGee and Thomas, 1986; Thomas and Venkatraman, 1988). Dissatisfactions include insufficient theoretic underpinnings for the construct itself, inadequate model specification, haphazard selection of strategic dimensions used to form groups, and inconclusive results from empirical research. Given these serious concerns, it is essential that we step back from research in this area to examine fundamental questions about the strategic group concept. The most recent and potentially most devastating attack on strategic group research has been made by Barney and Hoskisson (1990), who claim that two critical assertions of strategic group theory remain untested: '(1) that strategic groups exist and (2) that a firm's performance Received 1 June 1990 Final revision received 15 September 1992

214

R. Κ. Reger/A. S. Huff

104

R. Κ. Reger and A. S.

Huff

Table 1. Recent strategic groups studies Study Cool and Schendel (1987) Fombrun and Zajac (1987) Hatten and Hatten (1987) Johnson and Thomas (1987) Cool and Schendel (1988) Lawless, Bergh and Wilsted (1989) Lawless and Finch (1989) Mascarenhas (1989) Mascarenhas and Aaker (1989) McNamee and McHugh (1989) Zajac and Shortell (1989) Barney and Hoskisson (1990) Fiegenbaum and Thomas (1990) Kumar, Thomas and Fiegenbaum (1990) Lewis and Thomas (1990) Porac and Thomas (1990) Schroeder (1990) Ulrich and McKelvey (1990) Boeker (1991) Lawless and Tegarden (1991) Nohria and Garcia-Pont (1991) Sudharshan, Thomas and Fiegenbaum (1991) Tallman (1991) Caves and Ghemawat (1992)

Industry pharmaceutical financial services U.K. brewing pharmaceutical manufacturing firms 52 manufacturing firms oil-well drilling oil-well drilling Northern Ireland clothing industry hospitals insurance

Primary data type archival survey questionnaire validated with archival conceptual archival archival archival archival archival archival archival questionnaire conceptual archival conceptual

Japanese electronics brewing 47 four-digit SIC automobile pharmaceutical

archival conceptual interviews multiple case studies archival archival archival archival archival

automobile variety from PIMS NRDB database

archival archival

U.K. retail grocery foundry

depends upon strategic group membership' (1990: 187). Further, the assumption '[t]hat there are groups of firms suggests that firms are not idiosyncratic in strategically relevant ways' (Barney and Hoskisson, 1990: 188). Strategic group research is also suspect, since, to date, groups have been defined using 'an almost standard method. . .[which] employs some form of cluster or factor analysis' (Barney and Hoskisson, 1990: 189). Hatten and Hatten support this basic challenge with the assertion that strategic groups are merely analytical conveniences used by researchers, artifacts of our theories and techniques, without any objective analogue in the natural environment (1987: 329). Our view is, first, that strategic groups can be defined in a way that allows some strategically important variance among firms within each group. This view is supported by results presented in this paper. Second, although the explanation of performance differences among

firms is clearly elusive (Cool and Schendel, 1988; Fiegenbaum and Thomas, 1990; McGee and Thomas, 1986), we believe a modified theory of group structure, along lines developed later in this paper, may still make a contribution to the understanding of profitable strategic choices. Predicting firm profitability, however, is not the key contribution of a cognitive approach to strategic groups. Even if strategic groups are found to be uncorrected with performance outcomes, our position is that managerial perception of similarities and differences among competitors influence strategic decision making, and thus are worthy of study. In fact, strategists' grouping schemes may prove to be more significant than researcherdefined groupings for understanding competition and performance because, through enactment processes (Weick, 1979), the way firms see themselves and their competitors (Porac, Thomas, and Emme, 1987) is expected to have tangible

Strategie Groups: A Cognitive Perspective Strategic Groups effects on strategy reformulation and subsequent industry structure. It seems logical that strategists will think in terms of clusters of competitors to cognitively simplify a complex environment. Commonalities among firms should also be expected, independent of individual precepts, as a result of broader pressures creating institutional isomorphism (DiMaggio and Powell, 1983) and because of the many inertia factors inhibiting strategic change (Tushman and Romanelli, 198S). Recent work on isomorphism and inertia (Powell and DiMaggio, 1991) have bolstered our understanding of why firms do not always achieve the kind of idiosyncratic positions Barney and Hoskisson (1990) rightly indicate are often associated with competitive advantage. Finally, though we will make the argument that perceptual and economic factors can reinforce each other, fundamental questions about the real and the perceived cannot be answered definitively by any research endeavor. The demise of logical positivist standards make even simple assertions such as 'the sun will rise tomorrow' unprovable (Kuhn, 1970). As all proofs become suspect, no research project can satisfactorily say it has identified anything that is objectively 'real.' The study reported here focuses on determining if strategists perceive differentiating commonalities among firms and if so, to outline the nature of those perceptions. If groups of firms are 'real' for strategists, then research at the strategic group level is much more important than if groups are only the result of researchers' analytical exercises. Empirical evidence using strategists as data sources also provides multimethod confirmation (Huff, 1982; Jick, 1979) for extant research using archival sources. In this paper, we summarize research from cognitive psychology, organization theory and strategy that supports two propositions. These propositions, that strategists will perceive similarities among subgroups of firms in an industry and that strategists in the same industry will group participants in that industry in similar ways, are examined with data from the Chicago banking industry. The data show a shared sense of group structure in the industry, but also reveal a range of agreement on the similarity of firm strategies. These subtleties in executive perceptions add important dimensions to the

215 105

strategic group concept and increase its potential utility for understanding competitive positioning. To foreshadow points elaborated upon in the concluding sections of the paper, we believe those interested in describing and explaining strategic group structure might well amplify results from archival studies with cognitive studies. Those interested in predicting change in strategic position may be even more dependent on cognitive data to extend retrospective financial and accounting data.

STRATEGIC GROUPS AS THE RESULT OF PERCEPTION AND COGNITION Strategic management research is just beginning to assess the importance of cognitive frameworks in strategic decision processes (Dutton, Fahey, and Narayanan, 1983; Huff, 1990; Mason and Mitroff, 1981; Stubbart, 1986, 1987). Porac and Thomas' (1990) recent theoretical paper examines many of the cognitive rationales for mental models of competition. Two reasons why managers in demanding competitive situations might focus on groups of firms rather than individual competitors deserve special note. Simplification Cognitive research suggests that simplification is a cognitive necessity in cases of information overload. Research also shows that human beings can be easily overloaded. Miller (1956), for example, found that people cannot hold more than about seven bits of data in mind at a time. Work on cognitive simplification processes among managers (e.g., Schwenk, 1984; Simon, 1945) suggests that strategists simplify the complex cognitive problem of independently analyzing a large number of competitors by grouping them. In fact, executives, journalists and industry analysts often refer to groups of firms in public statements. Elaboration Cognitive elaboration involves filling in gaps (often unconsciously) when interpreting stimuli. The theoretical rationale for this activity is precisely the opposite of that offered for simplification. Sometimes data are not rich or varied

216

R. Κ. Reger/A. S. Huff

106

R. Κ. Reger and A. S. Huff

enough to allow interpretation without further elaboration. Often unconsciously, the unknowns about a particular case are likely to be filled in with information consistent with beliefs about other, better known examples that are believed to be similar (Rosch, 1981). Prahalad and Bettis (1986) have given a compelling example of this kind of embellishment. They propose that managers make acquisitions and divestments on the basis of a 'dominant logic,' or dominant paradigm developed through past experience. The result is a consistency in firm decisions over time, despite a wide variety of acquisition opportunities. We submit that executive perceptions may similarly 'gel' with respect to the dominant logic ascribed to surrounding firms. It is possible that these attributes go beyond the deliberate intentions of the companies involved, but the practical result is increased facility with the difficult problems of complex analysis. While simplification and elaboration are very different responses to competitive conditions, an important result of both processes, we suggest, is that strategists will perceive similarities among firms. Our first proposition, then, based on both cognitive simplification and elaboration is that: Strategists' perceptions of their competitors' strategies will be characterized by a group structure, rather than each firm's strategy being perceived as unique or all firms' strategies being perceived as similar.

Interaction While cognitive simplification and cognitive elaboration could lead to totally idiosyncratic groupings, strategists who work in the same industry environment are expected to develop shared perceptions of the competitive environment over time. Company executives interact with each other at industry associations and other gatherings; they share similar sources of information such as trade publications; they hire from the same professional labor pool and frequently employ the same consultants. As events shape the industry, such commonalities in information sources encourage shared interpretation of the present and shared expectations for the future, including shared perspectives on industry groupings (Huff, 1982; Porac, Thomas, and Baden-

Fuller, 1989). Thus, based on arguments of shared information sources and interaction, we further propose that: Strategic group structure will be widely shared by strategists within an industry, rather than each strategist holding unique perceptions of strategic group structure.

COGNITIVE STRUCTURES Given the potential importance of participant perspectives on the competitive environment, we believe that inquiry on strategic groups should include research based on cognitive data. This work appropriately begins with tests of the existence of competitor groupings in the minds of strategists. Previous research has made important first steps in this direction (Dess and Davis, 1984; Fombrun and Zajac, 1987). These researchers' purpose was to measure managerial perceptions of theoretically significant constructs, not to uncover cognitive structure perse. Questionnaires were used to gather data in these studies, and therefore it is not clear if the competitor groups found represent the respondents' cognitive structure of the industry or a structure inherent in the instruments used (Fransella and Bannister, 1977). Two streams of research in cognitive psychology are particularly appropriate for overcoming this problem in studies of cognitive structure including the study of cognitive strategic groups. Classification theories concentrate on categories of concepts and the hierarchical relationships among them (Johnson-Laird and Wason, 1977:169-253; Lakoff, 1987; Rosch, 1978). Alternatively, personal construct theory (Kelly, 19SS; Fransella and Bannister, 1977) focuses on the dimensions of concepts. The research reported here draws primarily on personal construct theory, which is especially germane given that extant strategic group theory assumes variation among firms along key dimensions of decision making (Porter, 1980; Cool and Schendel, 1987; Porac and Thomas, 1990) just as personal construct theory concentrates on the key dimensions, or constructs, of human cognition (Kelly, 1955). In addition, Thomas and Venkatraman (1988) advocated the use of this approach in their review of strategic groups research.

Strategie Groups: A Cognitive Perspective Strategic Groups Personal construct theory was first postulated as a theory of personality (Kelly, 1955); but later adherents assign it a more limited role as a theory of cognition (Fransella and Bannister, 1977). Briefly, the theory posits that bipolar constructs (or dimensions) are the primary mechanism individuals use to organize, simplify, and interpret the mass of stimuli that constantly confront them. Constructs are defined in terms of similarities and differences and are organized into systems of meaning which individuals use to develop theories about the environment, to make predictions and guide action. Personal construct theory is predicated on the belief that individuals act on their perceptions of the objective world filtered through their constructive system. The individual does not passively perceive the environment, rather he or she actively construes (attaches meaning to) perceptions. Kelly suggests that it is relatively easy for an individual to move an object along a cognitive construct, but much more difficult to think of objects in terms that are not part of an existing construct system. Constructs are thus seen to form a somewhat flexible yet structured network that both facilitates and restricts an individual's perceptions and actions (Kelly, 1955: 49). Hundreds of studies have been published in psychology and related fields using personal construct theory (Fransella and Bannister, 1977), but application of the theory and its related methodology, repertory grid technique, is quite new to strategic management. Ginsberg (1987) and his colleagues (Dunn et al. 1986; Dunn and Ginsberg, 1986) have recommended the theory and methodology for the study of business and public policy issues. Empirical studies are limited, but initial studies illustrate the flexibility of its application. Eden, Jones, and Sims (1979; 1983) combined personal construct theory with causal mapping to help organization members clarify and alter their perceptions of management situations. Walton (1986) built profiles of prototypical successful and unsuccessful firms using the repertory grid technique and Dutton (1987) studied the dimensions managers use to categorize issues with the same method.

RESEARCH DESIGN We used personal construct theory to address the research questions of this study. First, do

217 107

strategists group competitors in an industry? Second, are perceptions of competitor groupings widely shared or are they idiosyncratic? The two propositions were examined in the context of U.S. banking industry—a quite demanding test site, given that recent changes in regulation and increased competition might be expected to make clusters of competitors less stable and more difficult to assess. The research focused on the competitive strategies of the 18 largest bank holding companies (BHCs) headquartered in the Chicago area between 1982 and 1985 (Table 2). This time frame was chosen because it was bracketed by major regulatory changes in Illinois. In 1982, Illinois bank holding companies were allowed to acquire additional banks. Previous to this date, Illinois was a unit banking state, meaning that a bank holding company could only own one bank. Beginning in July, 1985, Illinois passed a regional, reciprocal banking pack with other midwestem states. Both of these changes were expected to increase the rate of environmental change and facilitate major strategic repositioning in the banking industry. While all 18 firms were not necessarily direct competitors in all markets when interviews were conducted in 1986, we expected, and found, that most strategists would be aware of the total population of major firms under these circumstances. Interviews were conducted at 6 of the 18 BHCs in the study population. The six were chosen for their diversity, relying on industry observers not in the study and secondary sources. Two firms were widely touted as well run, two were financially troubled and two had mixed performance reviews from industry insiders. Two of the banks were located 'in the loop,' two were in other downtown locations, and two were thought of as suburban banks. Two of the BHCs owned a single large 'flagship' bank, one was clearly organized as a multibank confederation. Interestingly, at least four of the six banks were rumored to be acquisition oriented, while five of the six were rumored to be takeover candidates. All six firms chosen for the sample agreed to participate in the study. The CEO at each bank identified five strategic decision makers within the firm who were especially familiar with competition in the Chicago market from 1982 through 1985. Four of the six CEOs personally participated in the study. In all, 30 strategists

218

R. Κ. Reger/A. S. Huff

108

R. Κ. Reger and A. S.

Huff

were contacted and 23 provided usable data for a 77% participation rate. All informants for the study were experienced bankers. They were employed in the banking industry for an average of 18.5 years, with banking experience ranging from 10 to 35 years. Respondents had worked for their current firm for an average of 10.25 years. Their experience varied widely, with seven (30%) in general management, five (22%) in finance, three (13%) in strategic or corporate planning and one (4%) in marketing. Seven of the informants worked in major product areas, including two (9%) in commercial loans, two (9%) in retail or private banking and one each in correspondent banking, financial institutions and trust (4% each).

study and to make similarity and difference judgments about individual examples of that phenomenon along those dimensions only. Thus, researcher-imposed structure on subject cognition is minimized. Each informant in the study was interviewed separately. Following standard procedures for the minimum context format, the 18 BHCs' names were presented to each informant three at a time. The sequence of presentation of triads was random and the same for each informant. Informants were asked the way or ways in which two of the BHCs were more similar strategically and how the third was different. In this way, bipolar dimensions used by strategists to differentiate among the strategies of competitors were elicited. Each informant personally identified dimensions, which were then used to elicit judgement about competitors' strategies. Ideally, informants would rate each focal BHC along all the dimensions they personally use to distinguish competitors. However, pretests indicated that asking an individual to rate all 18 BHCs along all defining dimensions was infeasible because of the time involved; pretest subjects reported their

Data collection Data were collected in Fall 1986 via semistructured interviews using the minimum context form of the repertory grid technique (Dunn and Ginsberg, 1986; Fransella and Bannister, 1977; Kelly, 1955; Reger, 1990b). This technique requires each informant to generate personal dimensions for describing the phenomenon of

Table 2. Number and percentage of informants who assigned each focal bank holding company to each cluster Cluster 1 No. %

Ouster 2 No. %

Cluster 3 No. %

First Chicago Northern Trust Continental 111. Harris

18 17 17 16

100 100 100 100

0 0 0 0

0 0 0 0

ABN-LaSalle Exchange

11 8

79 67

0 2

17

3 2

15 16 9

100 100 100

0 0 0

14 13 9 10

93 93 90 90

0 0 1 1

10 9

21 17

First Evergreen Gary Wheaton Affiliated

0 0 0

First United USAmeribancs State National First Illinois

1 1 0 0

17 7

First Colonial Cole-Taylor

1 2

10 13

8 11

80 73

1 2

10 13

Boulevard Lake Shore Unibancorp

4 4 3

29 24 27

6 6 2

43 35 18

4 7 6

29 41 55

Strategie Groups: A Cognitive Perspective Strategic Groups boredom level rose substantially when so many ratings were required. Therefore, each informant was randomly assigned to one of four groups to rate an overlapping subsample of 12 of the 18 focal BHCs. This sampling procedure, commonly used in marketing (Green, 1977), allowed for comparisons across informants without each informant rating all BHCs. Each informant rated the BHCs along his (all subjects were male) self-supplied dimensions using 11 point scales. The procedure yielded 23 η χ m rating grids, one for each informant, where η is the number of dimensions elicited for each of the 23 informants and m represents the number of firms the informant felt knowledgeable assessing. The Appendix provides a sample of the elicited dimensions. Detailed dimension-level results of this research also are reported elsewhere (Reger and Palmer, 1990). Data analysis It is possible to ask informants to directly form groups by sorting the population into subsets, a variant of repertory grid technique called the full context form (Fransella and Bannister, 1977:15). This sorting task was rejected for two reasons. First, extant strategic group theory is both theoretically and empirically concerned with the dimensionality of competitive strategy; that is, how firms are positioned in multidimensional competitive space. Therefore, we employed procedures which mirrored current research practice. Second, most applications of personal construct theory have used the minimum context form and, thus, reliability and validity measures have been better established for this procedure (Fransella and Bannister, 1977). Instead of directly asking informants for their perceived groupings, we cluster analyzed each informant's rating grid using SAS CLUSTER. Cluster analysis particularly fits the objectives of this study, since it places objects into groups suggested by the data. Clustering techniques can be criticized, however, since by their nature they will break the data available into subsets, however weak the association among data points. Therefore, three clustering methods were employed, each of which has been found to have somewhat different strengths and weaknesses. If all three clustering methods give similar results using multiple criteria, confidence that the groups

219 109

are an inherent part of the data and not an artifact of the particular clustering algorithm is increased. The first clustering technique used, the average linkage algorithm, forms clusters on the basis of the average distance between pairs of observations. This method 'tends to join clusters with small variances and is slightly biased toward producing clusters with the same variance' (SAS, 1985: 263). The second, centroid method maximizes the distance between group means, recomputing the centroids each time a new observation is included in a cluster. This method 'is more robust to outliers than most other hierarchical methods' (SAS, 1985 : 264). Ward's method was used as a third alternative. Groups are formed by minimizing the within cluster ANOVA sum of squares. With this technique Ward's cluster analysis 'tends to join clusters with a small number of observations and is strongly biased toward producing clusters with roughly the same number of observations' (SAS, 1985: 267). Cluster analysis is not an inferential statistical technique in which the researcher can link a specified confidence level with a set of clusters (Everitt, 1980: 66); instead, the researcher must specify rules for choosing the number of clusters that will be used for description and analysis of the data. Three decision rules determined the number of clusters from each algorithm's output for each informant's data: adopt cluster solutions at large breaks in the dendrograms, avoid cluster solutions that produce one firm groups, and choose solutions with high interpretability based on qualitative comments made by informants (Everitt, 1980; Hartigan, 1975; Romesburg, 1984). As a further assurance that the clusters represented 'real' groups in the data and not simply artifacts of the methods, the first author applied the decision rules to the average linkage output, while the second author independently evaluated the centroid and Ward's results. Use of these decision rules was considered preferable to blind application of any one decision rule that might misrepresent the data (Everitt, 1980: Romesburg, 1984). As a test of the correspondence between strategic groups identified by industry participants and economic 'reality,' we then looked at two outcome indicators. Data on profitability and survival, independent variables of longstanding interest in strategy research, were gathered on

220

R. Κ. Reger/A. S. Huff

110

R. Κ. Reger and A. S. Huff

the 18 firms in the study. Profitability data were gathered in 1986. Though mergers made it impossible to compare profitability in subsequent years, the merger pattern itself provides an interesting application of strategic groupings.

RESULTS Individual results The decision criteria outlined above yielded either two or three cluster solutions in the data from all but one of the 23 respondents, for whom only one of the three clustering algorithms yielded a cluster solution, the other two did not. This informant, however, only felt comfortable ranking 6 of the 12 BHCs presented, and these BHCs were most often clustered together by other respondents. For 17 of the other informants, the cluster results from the three methods converged. In one of the remaining five cases, one method created a group where the other two did not. In a second instance, one procedure did not create a group where the other two did. Two other cases involved one mismatch among the three methods out of the 12 firms the respondents answered, and a third involved 2 out of 12 discrepancies among the three methods. The fact that three algorithms yielded exactly the same clusters in data from 17 of 23 informants (74%) and that the worst case involved only a 16% discrepancy argues for accepting the first proposition of the study—that strategists cognitively group their competitors—and rejecting the alternative explanation that the clusters formed are mere artifacts of a clustering technique. Additional support for this proposition can be found in the strong similarities among clusters in the six cases that showed any discrepancy at all among the methods explored. Commonality across informants There is also strong commonality across respondents in the firms assigned to each cluster, which supports our second proposition: grouping schémas are not idiosyncratic, but widely shared across strategists. Since the results were similar across the three clustering algorithms, for clarity of exposition analyses from the average linkage method only are presented below. Table 2 summarizes the number and percentage

of informants who assigned each BHC to each of the three clusters commonly observed in the data. Both Cluster 1 and Cluster 2 include a core group of BHCs assigned to the same cluster by all informants rating that company and a secondary group of companies clustered with this core by more than 65% of the informants. Firms in Ouster 1 generally followed a 'money center' strategy concentrating on wholesale and commercial activities in a national or even international market whereas firms in Cluster 2 generally were stronger in middle market commercial, small business and consumer banking in a more limited geographic area. Ouster 3 does not exhibit a core, nor as strong a secondary group, and appears to be more of a catch-all cluster that cannot be as meaningfully compared across informants. For example, Boulevard Bankcorp (BOU) and Unibancorp (UNI) are assigned most often to this cluster, but are also found in each of the other clusters as well. Robustness of results Since cluster analysts requires researcher interpretation and is not a statistical methodology, it is prudent to examine the robustness of results under various assumptions. Thus, we took three approaches to evaluating the data shown in Table 2. First, we assumed that a three cluster solution represented common opinion about group structure in the industry, and asked: How often did a given individual assign a given firm to the 'right' cluster? Table 3, which summarizes the 251 competitor assessments shown in Table 2, indicates (on the diagonal) that in 209 instances, or 83% of the total, individual assessments resulted in assigning the firm to the cluster expected if group assignments are assumed a priori to be shared across all industry participants. If each informant in this study had randomly assigned firms to groups based on the relative sizes of the three groups, assignments would be 'successful' only 40% of the time. The difference between the observed result and the random expectation (which is statistically significant at the 0.001 level) provides strong support for the proposition that cognitive groups are shared among participants in an industry. It seems reasonable to suspect, given the low level of common assignments to Cluster 3, that this is a residual group which does not mirror

Strategie Groups: A Cognitive Perspective

221

Strategic Groups

111

Table 3. Assignment of Bank Holding Companies to Clusters: Three Cluster Solution

3 groups of firms

Cluster

A

1

87

2

3

Β

C

Total

5

11

103

2

105

14

121

5

5

17

27

94

115

42

251

strategic commonalities. The second test of shared strategic groups assumed that firms assigned to the first two clusters followed especially clear cut strategies. The data found in the upper left square in Table 3 summarize the number of firms assigned to these clusters. This inner square shows only 7 'misassignments' out of 199 judgements made about firms belonging to the two clear strategic groups in this industry; two in which afirmthat common wisdom said belonged in Cluster 1 was instead assigned to Cluster 2, and five in which a firm common wisdom assigned to Cluster 2 was instead placed in Cluster l. 1 This test shows even stronger support for the idea that group assignments are shared. Further examination of the data shows more agreement about the two frequently 'misassigned firms' than at first appears. The approach here was to relax the expectation that industry participants will share one view of competitor groupings. Tables 4 and 5 divide respondents by the number of clusters judged to best portray their assessments of competitors, and distinguish 1 Given the strength of results showing assignments in both the three group case and the two group case, no further evidence is really necessary to support proposition two. In more complex cases, it might be appropriate to compare the observed assignment with assignment to any other cluster. (In our case, the number of assignments to Ouster 1 vs. assignments to Clusters 2 or 3; and then the number of assignments to Cluster 2 vs. assignment to Clusters 1 or 3). A phi square, which corrects for the size of the sample (Reynolds, 1977: 27), or Cramer's V, which is particularly useful for assessing symmetric arrays (Reynolds, 1977: 32), can be used to describe data that are less clear cut.

'outlier'firmsfrom those in clusters. These tables show that ambiguity about the strategies of Boulevard Bankcorp (BOU) and Unibancorp (UNI) occurs almost exclusively among those who see the industry as falling into two strategic groups. There is little ambiguity among those who cluster the industry into three groups. They unanimously agree that BOU belongs in Group 3, and only one individual disagrees on the placement of UNI in Group 3. This third view of the data suggests even more strongly that there is considerable agreement on the strategic location of the BHCs in the population we studied. Commonality of group membership is supported under three different approaches to analyze the data, thus, proposition two is conclusively supported. Cognitive groups, performance and survival The strong results of this study show that cognitive data can be used to form strategic groups. Are these groups useful for understanding performance or other outcomes of interest? Table 6 shows the 1986 ROA for each focal BHC, as well as whether the BHC has remained independent or was acquired subsequent to the study. Again, since the status of Cluster 3 as a cohesive strategic group is in doubt, we concentrated our analysis on thefirsttwo clusters. Using a /-test of group mean differences for small samples, average ROA for the two groups was significantly different in 1986 (alpha = 0.005). Since many of the focal BHCs were acquired in 1987, and thus, did not report independent ROA figures, it is impossible to meaningfully compare profitability differences for later years. However, substantial differences exist between the two groups identified with cognitive data as to whether the members remained independent, or if acquired, what type of firm made the acquisition up to S years after executive assessments were made. In Cluster 1, both Harris and LaSalle had been purchased by international bank holding companies when data were collected in 1986. Subsequent to the interviews, LaSalle's parent, ABN, also acquired Exchange. The three other members of this cluster have remained independent. None of the core BHCs in Cluster 1 have changed ownership since the research was conducted.

222

R. Κ. Reger/A. S. Huff

112

R. Κ. Reger and A. S. Huff

Table 4. Assignment of bank holding companies to clusters by informants whose data resulted in a two cluster solution

Bank holding companies ABN-LaSalle Affiliated Boulevard Cole-Taylor Continental 111 Exchange First Chicago First Colonial First Evergreen First Illinois First United Gary Wheaton Harris Lake Shore Northern Trust State National Unibancorp USAmeribancs

A2

A3 A4

A5

A6 A7

1 2

1 2

1

2

1 2

1

1

2

X

2

1 1

1 1

1 1

1 1

1 1

2

2

2

2

2

2 2 1 X 1

2 2 1 1 1

2 2 1 1 1

2 2 1 2 1

2 2 1 2 1

Informants B1 B4 CI

1 X 1

2 2 1 2 1

2 2 1 1 1

2 2

2 2

2 1 2 1 2

2

1

2

2

C5

C6

C7

D1

D2

1 2 2

1 2 1

X 2 1

1

1

2

1 2 1

1

1

1

1

1

1 2 1

2 2 1

1

1 2 2

1 2 2

1 1 X

1 2 2

2

2

2

2

1

2

2 2 2

2 2 2

1 1

1 X

X

1 1

2

2

2

2

2

2

2

2

1 2 1 X

1 2 1 2

1

X

1 1

C2

2 1 2 1

X 2

1 X

2

χ - outlier firm

Table 5. Assignment of bank holding companies to clusters by informants whose data resulted in a three cluster solution

Bank holding companies ABN-LaSalle Affiliated Boulevard Cole-Taylor Continental IU Exchange First Chicago First Colonial First Evergreen First Illinois First United Gary Wheaton Harris Lake Shore Northern Trust State National Unibancorp USAmeribancs

B2

B3

BS

Informants B6 C3

3 2 1 2 1

3 2 1 1 1

3 2 1 3 1

3 2 1 3 1

2 2

2 2

2 2

2 2

2 1 3 1

2 1 2 1

2 1 3 1

2 1 3 1

D2

D4

D5

3

3

3

3 2 1

3 2 1

2 3 1

1 2 2

2

3

2

2

3 2 2

2 2 2

2 2 2

1 3 3 2

1 2 3 2

1 2 3 2

1 2 3 1

1 3 2

3

3

2

3 2

Strategie G r o u p s : A C o g n i t i v e P e r s p e c t i v e

Strategic Table 6.

Groups

223

113

Profitability and ownership changes 1986 ROA

Ownership1

Cluster 1 Core Firms First Chicago Northern Trust Continental Illinois Harris

0.71 0.57 0.S 0.7

Independent Independent Independent Bank of Montreal

Secondary Firms ABN-LaSalle Exchange

0.52 0.53

Algemene Bank Nederland Algemene Bank Nederland

Mean R O A , Cluster 1

0.591

Cluster 2 Core Firms First Evergreen Gary Wheaton Affiliated First United USAmeribancs State National First Illinois

1.3 1.1 1.0 0.85 1.2

Secondary Firms First Colonial Cole-Taylor

0.82

Mean R O A , Cluster 2

1.05

Cluster 3 Boulevard Lake Shore Unibancorp

1.0 1.0 1.1

Independent First Chicago Manufacturers (Michigan) First Chicago NBD Bancorp NBD Bancorp BancOne (Ohio)

Independent Independent

Independent Independent Old Kent Financial

'Mean ROA for Ouster 1 and Ouster 2 are significantly different at the 0.005 level of significance using a (-test of differences between group means for small sample sizes. ^Ownership changes which occurred after the interviews were conducted and announced as of October 1, 1991 are indicated in bold. All of the focal BHCs were independent at the time of the study except for the following: LaSalle was purchased by Algemene Bank Nederland (ABN) on August 14, 1979; Harris became a wholly-owned subsidiary of the Bank of Montreal on Sept. 4, 1984; NBD Bancorp (headquarter in Michigan) announced the purchase of USAmeribancs on October 23, 1986. Interviews were conducted in November 1986.

The B H C s in Ouster 2 have faced a quite different fate. Five of the nine B H C s have been acquired, all by B H C s headquartered in the midwestern United States. Only one of the core B H C s in this cluster, Evergreen, has remained independent. The other two banks which have remained independent in this group, First Colonial and Cole Taylor, were the members most often assigned to other clusters.

Performance at the time of data collection and survival 5 years later thus support the cognitive data on strategic group assignment. Firms in Cluster 1, especially those assigned unambiguously, did not change ownership. In fact, the only member of this group that was acquired, Exchange, followed the pattern set by other group members—it was acquired by a foreign bank holding company. AH except one of

224

R. Κ. Reger/A. S. Huff

114

R. Κ. Reger and A. S. Huff

the firms assigned by at least 90 percent of respondents to Cluster 2 were acquired by regionally-based U.S. bank holding companies within 5 years of our analysis. These results are supportive of the other findings on the homogeneity of firms within cognitive groups, however, the small sample sizes and the correlational nature of the analysis precludes the assertion of a causal link between these outcomes and strategic group membership.

DISCUSSION OF RESULTS Given the amount of change in the Chicago banking market from the early 1980s on, the support for the basic propositions of this study is surprisingly strong, especially since respondents used their own dimensions to describe firm strategies rather than a common set of researchersupplied categories. Nonetheless, idiosyncratic assessments in the data deserve further explanation. Why were some firms unambiguously assigned to a strategic group across informants, while placement of other firms is more uncertain? Further examination of the data suggest six arguments which may explain disagreement about strategic group membership for some firms. These arguments address the nature of strategic positioning as well as the ability of researchers and competitors to perceive subtleties in the strategic choices firms make. 1. Fuzzy Groups. Some BHCs that were not uniformly assessed by informants may have exhibited strategies essentially in line with those of one or more 'core' BHCs in a particular group, but these firms also may have shared secondary attributes with other groups or possessed idiosyncratic secondary traits. Zadeh (1965), a leading categorization theorist, has suggested that many conceptual categories are 'fuzzy' on their boundaries, such that some members of the category are better examples of the category than are others. Thus, a firm that shares many commonalities with the category is likely to be widely perceived as a good example of the category. A firm that shares fewer characteristics is likely to be associated with a wider variety of firms. Some informants might have concentrated on core aspects of this firm's strategy, while other informants focused on secondary characteristics the firm happens to share with firms in other clusters.

2. Obscured Strategy. Disagreement among this group of expert observers about the placement of a few BHCs also may be due to the difficulty of observing some firms' strategic actions. Porter (1980) and others have suggested that firms are wise to obscure their intentions in order to gain competitive advantage. Thus, it is likely that the strategies of some firms, either intentionally or unintentionally, will be difficult to interpret. Given the theoretical importance of surprise in establishing competitive advantage, our results are interesting not because a few firms' strategies were difficult to identify, but because respondents agreed on the basic strategy of almost all firms in the population. Obscured strategy appears to have played, at most, an extremely limited role in this industry. 3. Firm Repositioning. Even during the most stable periods of industry evolution, the strategies of some BHCs may be in flux due to firm specific conditions such as new management, poor past performance or firm specific innovation. Respondents in our study might have varied in their perceptions of such firms. Some observers may have focused on the BHC's old strategic position, some on current position and some may have been oriented toward anticipated future position. 4. Industry Realignment. Another possible explanation for variation in the categorization of some firms is tied to the structure of the industry as a whole. An industry in transition, such as the one we observed, is likely to be moving toward a larger or smaller number of viable group 'recipes' (Spender, 1980, 1989), or, if the key dimensions of strategy are themselves in the process of change, some observers may have been more aware of this transformation than others. Inconsistencies in the data may thus be due to some informants reporting an 'old' structure, while others anticipate a new industry structure. Though similar in effect to internally motivated repositioning of individual firms, industry realignment is conceptually different because it is more closely tied to external environmental change. 5. Reactors and Misfits. Some companies do not follow clear, constant strategies. Such firms have been described as 'stuck in the middle' (Porter, 1980) or 'reactors' (Miles and Snow, 1978); they change strategic direction often or simply react to environmental events as they

Strategie Groups: A Cognitive Perspective Strategic Groups occur. These firms might be best described as misfits in that they do not exhibit a coherent strategic pattern over time. It is not surprising that industry observers vary in the way they describe such firms. 6. Strategic ldiosyncracy. Some focal BHCs may have adopted strategies which are so dramatically different from all others in the industry that it is not easy to position them visà-vis other firms along common dimensions. These firms are not aligned with other firms; they do not compete in similar ways. Such firms may have a coherent strategy and that strategy may fit the industry, but it does not fit the cognitive map that usefully describes almost all other competitors within the industry. When Timex first entered the watch industry, it was misaligned in this way. Its strategy was so radically different from existing watch manufacturers that it could not be meaningfully mapped on dimensions which differentiated other competitors. New dimensions, and hence a new map, were needed. Lack of consensus among industry observers may indicate analogous outliers in the Chicago banking market. These arguments are complementary, not mutually exclusive. Multiple explanations appear to account for disagreements about the placement of all six equivocal firms in our study. For example, the banking industry was undergoing a major realignment at the time data were collected for this study (McKinsey and Company, 1985). Several informants indicated both Lake Shore and Boulevard were pursuing strategies that were radically different from other bank holding companies at this time. ABN-LaSalle, on the other hand, was often described as both a misfit and a firm with a 'fuzzy' strategic position. This firm had been acquired twice in the 7 years preceding the study. It had long operated under a consistent strategy which aligned it with firms in Cluster 1, and at the time of the study it retained strengths associated with that strategy. New ownership had moved the company toward strategies more distant from either Cluster 1 or 2, but decisions in the 1986 time frame were interpreted by some observers as reasserting a strategy in line with Cluster 1. The lack of agreement about the association of ABN-LaSalle across informants appears to be a good indicator of genuine uncertainty about the strategic direction of this firm.

225 115

A BROADENED THEORY OF STRATEGIC GROUPS While there is some ambiguity in the current literature about the ontological status of strategic groups, with the skeptic position expressed especially by Hatten and Hatten (1987) and Barney and Hoskisson (1990), our study suggests that strategic groups are readily perceived by strategists. The study thus provides evidence that strategic groups are more than analytical conveniences used by researchers; they are part of the way strategists organize and make sense of their competitive environment. The performance differences between Cluster 1 and Cluster 2 at the time of the study as well as the utility of these groups for differentiating acquisition experience 5 years after data collection suggest promising avenues for future research. Explaining performance was not the focus of the current research and these results, although suggestive, do not establish a causal link between group membership and outcomes. Nonetheless, the strategic groups derived from the cognitive structure of strategists in this industry demonstrate that executive perceptions have real potential for competitive strategy research. Further, the results contribute to evidence from previous empirical studies using different methods (Porac et al., 1989; Porac et al., 1987). Accumulating evidence suggests that strategic groups are phenomena with an ontological status; they are not the contrivance of a single method. This is particularly important as industry level groupings, such as those based on SIC code assignments, become more and more tenuous. Disagreement on strategic group assignment for some firms suggests, however, that future research might benefit from a revised view of strategic group structure. Existing research implicitly assumes that all firms follow a strategy, that this strategy is knowable by outsiders, and that all firms can be assigned unambiguously to a strategic group (even if the group has only one member). Few of the studies reviewed by Cool (1985) or by McGee and Thomas (1986) removed even a single firm from the analysis due to interpretability problems (Hatten, Schendel and Cooper, 1978, is an exception). The results of this study support the existence of strategic group structure, but suggest several ways in which current conceptualizations could be enriched.

226

R. Κ. Reger/A. S. Huff

116

R. Κ. Reger and A. S. Huff

1. Every firm does not have to have a strategic position. The overwhelming majority of informants in our study classified two-thirds of the BHCs into one of two strategic clusters. Agreement was lower for the other 6 firms in the population. Three firms (Boulevard Bankcorp, Lake Shore Bankcorp and Unibancorp) appear in all three clusters and, at the least, do not appear to be following a widely agreed upon strategy. A distinction must be made between strategic situation, which characterizes the unique environmental and internal circumstances challenging every firm, and strategic position, which characterizes the firm's response to those circumstances. Every firm faces a strategic situation. These data, however, suggest that the largely unexamined assumption of existing strategic group theory, that every firm has a strategic position, may be true for only some firms in the industry. The competitive positions of at least some firms are not easily assessed and/or readily assigned to a strategic group. When strategy is conceptualized as a pattern over time (Mintzberg, 1978), it is quite likely that some firms do not have a coherent strategy. Including such firms will diminish the usefulness of studies observing strategic position. One particularly interesting implication of this argument for future research is that past correlational studies might be usefully re-examined. For example, the ambiguous results of studies attempting to link strategic group membership to profitability seem to have brought this line of inquiry to an impasse. If problematic firms are removed from group assignments, it is possible that more conclusive results could be achieved. 2. Group membership is a matter of degree. Strategic groups were 'fuzzy sets' when aggregated across informants in this study. A more finegrained view of membership within groups might be profitable for many strategic group studies. Our results suggest that a strategic group might be best conceptualized as a core group of firms that define the group position and secondary firms that are aligned with core firms in many essential respects, though they also make some unique strategic decisions. Porter (1980: 154) implicitly acknowledges the possibility of ambiguously defined groups when he advises analysts to draw many strategic group maps using different dimensions depending upon the analyst's pur-

poses. However, no published empirical work has yet taken this approach. The general point worth considering is that some firms are prototypical of a group's position, while others are not. Categorization theorists label this phenomenon 'membership gradience' (Berlin and Kay, 1969; Zadeh, 1965). However, it is also probable that the firms included in a particular analysis will determine which firms are core and which are secondary (Lakoff, 1987). A map of the financial services industry (Fombrun and Zajac, 1987) provides different results than one which includes commercial banks only (Passmore, 1985). Reger (1990a) pursues the ramifications of this point for the design of competitive positioning research. 3. Some groups may be tightly associated while others are more diverse. One interesting observation on the pattern of cluster formation from our data analysis is that some firms are rapidly linked by clustering algorithms while others lie at greater distance and are incorporated into a group much later in the clustering procedure. The implication is that some firms in an industry are quite similar while others, though associated, are more disparate. Studies of performance, and other potential correlates of group membership, again might benefit from distinguishing 'tight' and 'loose' groups. 4. Overlapping group membership may characterize some industries. Our data suggest that strategic groups may be overlapping at their periphery, with a small set of firms sharing some strategic characteristics with one group and other strategic characteristics with another. While this possibility muddies the ideas of symmetry and simplification that are basic attractions of the strategic groups idea, allowing overlap may be necessary if we wish to more realistically map complex industries. The existence of overlap also raises some interesting research questions. Firms at the intersection of two existing groups may indicate a potentially viable location for new group formation, for example, or they may represent the problem of literally being 'stuck in the middle' (Porter, 1980). 5. For periods of time, group structure may not exist or may not be apparent. Cool (1985) and Fiegenbaum (1987) identify stable strategic time periods in an industry by stable strategic group membership. Taking the previous suggestions (1 through 4) to an extreme, it may be

Strategie Groups: A Cognitive Perspective Strategic Groups necessary to allow not only for shifting group membership between stable time periods, but for periods in which an industry can not be usefully defined in terms of strategic group structure at all. Mintzberg (1978) suggests that firms often go through periods of inchoate search before establishing a strategic direction or redirection. The same may be true for industries as a whole. Economic conditions can be in such flux that firms will not be able to follow consistent strategies, and, as firms change position this movement will tend to further destabilize the industry. However, we expected less stability in this study of bank holding companies than we found. Even in a period of uncertainty and search for new strategies—like the one that currently characterizes the banking industry—it may be that similar firms muddle in similar ways and the strategic groups idea remains useful. In summary, based on our findings from the bank study, we propose that strategic group structure be reconceptualized as graphically depicted in Figure 1. This figure provides a general picture of strategic groups composed of the following kinds of firms: - Core Firms that are tightly associated and define the basic 'recipe' of a strategic group. - Secondary Group Members that implement the strategic group recipe less consistently than core firms. - Transient firms whose strategies are changing from one strategic position to another, but along dimensions common to other firms in the industry. Left out of this schematic are two kinds of firms: - Misfit firms whose strategies are inconsistent over time. - Idiosyncratic firms whose strategies cannot be easily expressed in terms of the key strategic dimensions used to define the competitive positions of most firms in the industry.

DIRECTIONS FOR FUTURE RESEARCH If strategic group membership is reconceptualized as a matter of degree, a number of interesting research possibilities become available.

227 117

Transient Firms Misfits and Idiosyncratic Firms (inappropriate to map)

Figure 1. A reconceptualization of strategic group relationships 1. Strategic groups research provides an opportunity to look more closely at viable strategic alternatives. If groups are defined solely on dimensions associated with the core members of the group, clearer insights into strategic alternatives may be forthcoming. Deviation from the core might indicate either inability to implement the group recipe completely or deliberate attempts to differentiate the firm from core members. The distinction between imitability and differentiation might be addressed more directly by distinguishing core and secondary firm characteristics. 2. Intragroup dissimilarities may be clarified by exploring the roles different types of firms play in industry dynamics. For instance, the extent to which core firms serve as referents for other members or tend to be first movers are promising research avenues. We hypothesize that core firms will vary in terms of innovativeness, but that they will serve as key referents and tend to dominate external views of the industry. Transients, misfits and idiosyncratic firms may provide important clues as to future strategic options or necessities in an industry, and are therefore important to those interested in industry evolution and competitive dynamics (Lewin and Minton, 1986). From a population ecology perspective, these firms may represent the 'gene

228

R. Κ. Reger/A. S. Huff

118

R. Κ. Reger and A. S. Huff

pool' for future variation in response to new conditions; they provide some evidence as to what may be more or less attractive positions in the future. These firms provide the counterbalance to what can be learned from core firms. Core firms reflect the current consensus of opinion about viable modes of operating within an industry context; outlier firms suggest future possibilities. This evidence is especially important as exogenous forces impinge upon the industry or powerful competitors change the nature of competitive dynamics within the industry. 3. Attention should be directed at more valid methods for determining group membership and structure. Valid strategic groups should account for outcomes of interest in strategic management such as strategic choice and performance. Our results tentatively suggest that cognitively-derived strategic groups are promising for exploring performance outcomes and independent survival in a consolidating industry. More research should focus on the usefulness of strategic groups for predicting other strategic outcomes. One focus of this inquiry might be on the relationships between 'competitive' groups (firms that compete against each other) and 'strategic' groups (firms that compete using similar strategies). The interaction between changes in strategic positioning as perceived by managers and changes evident in 'objective' data is also an alluring research area. Under what conditions do managerial perceptions lead, or lag, economic and financial indicators of changing positions? 4. The study of cognitive strategic groups can enrich the theory of how analytically identified groups form and why they tend to remain stable. We suggest that the 'analytic' strategic groups formed by observable variations in accounting and financial data (Cool, 1985; Fiegenbaum, 1987) owe their regularities, at least in part, to the process of enactment (Weick, 1979). It is reasonable to expect that the groups managers perceive will have real effects on strategy reformulation, strategic action and subsequently on industry structure. Economic realities make strategic similarities likely, but cognitive processes also may be expected to reinforce economic influences and thus help maintain the existing group structure. As Pfeffer and Salancik note, 'environments are not given realities, but created through a process of attention and interpretation' (1978: 13). Enactment of strategic

groups may be aided by several processes including simplification, elaboration, interaction, borrowed experience, and expectations. In aggregate, these processes provide the mechanisms whereby cognitive and economic realities converge. The effect of cognitive simplification, we suggest, is not just that strategists will tend to think of competitors in clusters, but that they will tend to recognize similarities between their own firm and a set of fellow firms and act accordingly. While the link between thought and action remains problematic in social science research, most strategy researchers assume some connection. It then follows that cognitive simplification should contribute to simplification of action. As firms recognize links between themselves and others, and act based on these apparent simplifications, groupings of firms within an industry will tend to become even more distinct. Cognitive elaboration, the process which creates information about competitors, may also lead to the enactment of strategic groups. To the extent strategists find similarities between their firm and others, and act upon these similarities, they can create further alignment in economic realities. For example, if firm A believes its competitor Β faces similar production capacity constraints (whether this is true or not), it may monitor and react to B's capacity expansion decisions in a number of ways that work to fulfill that expectation. If Β begins to act as if it might increase its capacity (by initiating exploratory studies, for example), A may begin to contemplate a similar move that it had not here-to-fore considered. Firm A interprets B's actions in a way that leads it to imitate B's strategy. A's assumption of the need for competitive reaction thus serves to bring the two firms' strategies into closer alignment. Our point, again, is that perceived similarities, whether or not they accurately reflect reality initially, channel actions into a smaller number of alternatives than might otherwise be the case and strategic groups are likely to further coalesce. Likewise, interaction among strategists and with industry analysts, distributors and other third parties tends to result in similar thought patterns (Porac et al., 1989). Executive fraternization makes it more likely that firms will react in similar ways to threats and opportunities within the industry—further clustering firms.

Strategie Groups: A Cognitive Perspective Strategic Groups While shared information and interpretation are expected to promote similar decisions, cognitive clustering is even more likely when strategists 'borrow experience' through observing other firms in the industry as a substitution for direct experimentation with strategic alternatives (Huff, 1982). As firms experience success and failure, their actions will tend to be copied and/or avoided by others in the industry. Over time we expect this process of replication and circumvention to make a major contribution to clustering firms within an industry. As strategists scrutinize other firms within an industry and develop mental frameworks for interpreting what they see, they come to expect certain behavior and act on these expectations. Expectations thus are another force for channeling firms into a limited repertoire of behavior. If a firm is classified as aggressive, for instance, small changes in its pricing policy will be given increased importance and retaliatory responses are more likely. Similar actions by another firm, perceived to be following a less aggressive path, will be taken less seriously. Such expectations exist not only among competitors, but among buyers, suppliers and potential entrants. The result, particularly given limited ability to make distinctions among firms due to cognitive overload, is to further homogenize both the perception of strategic options within an industry and the repertoire of strategic actions taken.

LIMITS TO COGNITIVE GROUP RESEARCH In our view, these are strong arguments for assuming that strategists' cognitive structures will have a material effect on strategic choices. As choices channelled by managerial perception contribute to the economic structure of an industry, the external stimulus for future perceptions of similarity are further reinforced. In other words, cognitive groups tend to reinforce economic groups. They combine with forces at other levels that create institutional isomorphism (DiMaggio and Powell, 1983). The spiral of mutual reinforcement should be balaneed by evolutionary changes in technology, buyer demographics and so on, along with the declining profitability of overpopulated strategic positions. These evolutionary agents force firms

229 119

to change their strategies and break groups apart. Research shows, however, that established mental maps lead individuals to ignore contradictory data (e.g., Prahalad and Bettis, 1986). Thus a problem with cognitive associations is they may not reflect evidence from a changing world. Cognitive structures also are inevitably based on incomplete knowledge, and even the simplest inferences are frequently biased (Schwenk, 1984). These limitations on the accuracy of managerial perceptions limit the research utility of managerial observations in general, and cognitive strategic groups in particular, but they do not justify abandoning cognitive data. In our view, strategy researchers interested in competitive interaction should treat participants in an industry as expert witnesses and important sources of data. Many biases and limitations can be mitigated, if not canceled, by broadly seeking opinions among industry participants. Furthermore, accounting and financial data have their own categorization and reporting biases that might be more easily seen when compared with the categorizations of industry participants. Our results must be interpreted with caution since the study focuses on only one geographic area of one industry. The banking industry in Chicago was facing unique environmental challenges stemming primarily from state deregulation. Thus, future research will be necessary to determine the generalizability of these results to other industries facing other environmental conditions. Another limitation has to do with the small number of informants; the intensive interviewing needed to collect rich data necessarily limits cognitive studies. Future research might be usefully directed to the optimal number of subjects necessary to establish a 'cognitive community' (Porac et al., 1989). Again, further research is required to examine these generalizability issues.

CONCLUSION Strategic groups offer a promising way to summarize competitors' strategies in industries populated by so many competitors that individual consideration of each firm is impossible. This study theoretically argues and finds empirically that even in a changing industry strategists cognitively group their competitors' strategies.

230

R. Κ. Reger/A. S. Huff

120

R. Κ. Reger and A. S.

Huff

barriers', Strategic Management Journal, 13, 1992, pp. 1-12. Cool, K. O. .'Strategic group formation and strategic skills: A longitudinal analysis of the U.S. pharmaceutical industry, 1963-1982'. Unpublished doctoral dissertation, Purdue University, May 1985. Cool, K. O. and D. E. Schendel. 'Strategic group formation and performance: The case of the U.S. pharmaceutical industry, 1963-1982'. Management Science, 33(9), 1987, pp. 1102-1124. Cool, K. O. and D. E. Schendel. 'Performance differences among strategic group members', Strategic Management Journal, 9, 1988, pp. 207-223. Dess, G. G. and P. S. Davis. 'Porter's (1980) generic strategies as determinants of strategic group membership and organizational performance', Academy of Management Journal, 27, 1984, pp. 467-488. DiMaggio, P. J. and W. W. Powell. 'The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields', American Sociological Review, 48(2), 1983, pp. 147-160. Dunn, W. N., A. G. Cahill, M. J. Dukes and A. Ginsberg. 'The policy grid: A cognitive methodology for assessing policy dynamics'. In W. N. Dunn (ed.). Policy Analysis: Perspectives, Concepts, and Methods, JAI Press, New York, 1986, pp. 355-375. Dunn, W. N. and A. Ginsberg. Ά sociocognitive network approach to organizational analysis', Human Relations, 40(11), 1986, pp. 955-976. Dutton, J. E. 'Separating the wheat from the chaff: The important dimensions of strategic issues'. Conference on Managerial Thinking in Business Environments, Boston University, 1987. Dutton, J. E., L. Fahey and V. K. Narayanan. 'Toward understanding strategic issue diagnosis', Strategic Management Journal, 4,1983, pp. 307-324. Eden, C., S. Jones and D. Sims. Thinking in Organizations, Macmillan, London, 1979. ACKNOWLEDGEMENTS Eden, C., S. Jones and D. Sims. Messing About in Problems, Pergamon, Elmsford, NY, 1983. The authors wish to thank Irene Duhaime, Luis Everitt, B. Cluster Analysis, second edition, Heinemann Educational Books, London, 1980. Gomez-Mejia, Richard Gooding and anonymous Fiegenbaum, A. 'Dynamic aspects of strategic groups Strategic Management Journal reviewers for their and competitive strategy: Concepts and empirical comments. examination in the insurance industry'. Unpublished doctoral dissertation, University of Illinois at Urbana-Champaign, 1987. Fiegenbaum, A. and H. Thomas. 'Strategic groups REFERENCES and performance: The U.S. insurance industry, 1970-1984', Strategic Management Journal, 11,1990, Barney, J. B. and R. E. Hoskisson. 'Strategic pp. 197-215. groups: Untested assertions and research proposals'. Managerial and Decision Economics, 11, 1990, Fombrun, C. J. and E. J. Zajac. 'Structural and perceptual influences on intraindustry stratification', pp. 187-198. Academy of Management Journal, 30, 1987, Berlin, B. and P. Kay. Basic Color Terms: Their pp. 33-50. Universality and Evolution. University of California Fransella, F. and D. Bannister. A Manual for Repertory Press, Berkeley, CA, 1969. Grid Technique, Academic Press, New York, 1977. Boeker, W. Organizational strategy: An ecological perspective', Academy of Management Journal, 34, Ginsberg, A. 'A cognitive methodology for assessing strategic diversity', Conference on Managerial 1991, pp. 613-635. Thinking in Business Environments, Boston UniverCaves, R. E. and P. Ghemawat. 'Identifying mobility sity, 1987.

In fact, there is strong uniformity across our sample of informants in the grouping of almost all 18 firms studied. These groups accurately predicted subsequent performance differences and acquisition patterns. Managers group firms in subtle ways not captured by economically oriented research. If more subtle distinctions in strategic group membership along the lines shown in Figure 1 are identified, we believe new utility in the strategic group concept may emerge. Even more interesting, in our view, is the potential of a refined concept of strategic groups for further understanding competitive positioning. What are the differences between core and secondary group members? Are there barriers to mobility even within groups? How will firms perceived as following aspects of more than one strategy fare? Are they stuck in the middle or are they developing a new and viable niche? Once questions of positioning like these become the focus for research, it becomes more and more interesting to include practitioners as a source of data on strategic groups. The fact that we found such strong consensus among informants, even in a rapidly changing industry, suggests that practitioners can be a reliable and valid source of data. The apparent ability of these observers to capture nuances in association among competitors over time suggests that this data can considerably enrich archival studies.

Strategie Groups: A Cognitive Perspective Strategic Groups Green, P. Ε. Ά new approach to market segmentation', Business Horizons, February 1977, pp. 61-73. Hartigan, J. A. Clustering Algorithms, Wiley, New York, 1975. Hatten, Κ. J. and M. L. Hatten. 'Strategic groups, asymmetrical mobility barriers and contestability', Strategic Management Journal, 8,1987, pp. 329-342. Hatten, Κ. J., D. E. Schendel and A. Cooper. Ά strategic model of the U.S. brewing industry: 1952-1971', Academy of Management Journal, 21, 1978, pp. 592-610. Huff, A. S. 'Industry influences on strategy reformulation', Strategic Management Journal, 3, 1982, pp. 119-131. Huff, A. S. Mapping Strategic Thought, Wiley, Chichester, 1990. Hunt, M. S. 'Competition in the major home appliance industry, 1960-1970'. Unpublished doctoral dissertation, Harvard University, 1972. Jick, T. D. 'Mixing qualitative and quantitative methods: Triangulation in action', Administrative Science Quarterly, 24, 1979, pp. 602-611. Johnson, G. and H. Thomas. "Hie industry context of strategy, structure and performance: The U.K. brewing industry', Strategic Management Journal, 8, 1987, pp. 343-361. Johnson-Laird, P. N. and P. C. Wason. Thinking, Cambridge University Press, Cambridge, M A , 1977. Kelly, G. A. The Psychology of Personal Constructs, Vols. 1 and 2, Norton, New York, 1955. Kuhn, T. The Structure of Scientific Revolutions, University of Chicago Press, Chicago, IL, second edition, 1970. Kumar, K. R., H. Thomas and A. Fiegenbaum. 'Strategic groupings as competitive benchmarks for formulating future competitive strategy: A modelling approach', Managerial and Decision Economics, 11, 1990, pp. 99-109. Lakoff, G. Women, Fire, and Dangerous Things, University of Chicago Press, Chicago, IL, 1987. Lawless, M. W., D. D. Bergh and W. D. Wilsted. 'Performance variations among strategic group members: An examination of individual firm capability', Journal of Management, 15, 1989, pp. 649-661. Lawless, M. W. and L. K. Finch. 'Choice and determinism: A test of Hrebiniak and Joyce's framework on strategy-environment fit', Strategic Management Journal, 10, 1989, pp. 351-365. Lawless, M. A. and L. F. Tegarden, Ά test of performance similarity among strategic group members in conforming and non-conforming industry structures', Journal of Management Studies, 28, 1991, pp. 645-664. Lewin, A. Y. and J. W. Minton. 'Determining organizational effectiveness: Another look, and an agenda for research', Management Science, 32, 1986, pp. 514-538. Lewis, P. and H. Thomas. 'The linkage between strategy, strategic groups, and performance in the U.K. retail grocery industry', Strategic Management Journal, 11, 1990, pp. 385-397. Mascarenhas, B. 'Strategic group dynamics', Academy of Management Journal, 32, 1989, pp. 333-352.

231 121

Mascarenhas, B. and D. A. Aaker. 'Mobility barriers and strategic groups', Strategic Management Journal, 10, 1989, pp. 475-485. Mason, R. O. and I. I. Mitroff. Challenging Strategic Planning Assumptions, Wiley, New York, 1981. McGee, J. and H. Thomas. 'Strategic groups: A useful linkage between industry structure and strategic management', Strategic Management Journal, 7, 1986, pp. 141-160. McKinsey and Company. Bankers' Merger and Acquisition Choices: Mapping a Course through Consolidation, Bank Administration Institute, Rolling Meadows, IL, 1985. McNamee, P. and M. McHugh. 'Mapping competitive groups in the clothing industry (Part 2)', Long Range Planning, 22, 1989, pp. 89-97. Miles, R. E . and C. C. Snow. Organization Strategy, Structure, and Process, McGraw-Hill, New York, 1978. Miller, G. A. 'The magic number seven plus or minus two: Some limits to our capacity for processing information', Psychology Review, 50,1956, pp. 81-97. Mintzberg, H. D. 'Patterns in strategy formation', Management Science, 24, 1978, pp. 934-948. Nohria, N. and C. Garcia-Pont. 'Global strategic linkages and industry structure', Strategic Management Journal, 12(Special Issue), 1991, pp. 105-124. Passmore, S. W. 'Strategic groups and the profitability of banking'. Research paper No. 8501, Federal Reserve Bank of New York, New York City, 1985. Pfeffer, J. and G. R. Salancik. The External Control of Organizations, Harper & Row, New York, 1978. Porac, J. F. and H. Thomas. 'Taxonomic mental models in competitor definition', Academy of Management Review, 15, 1990, pp. 224-240. Porac, J. F., Η. Thomas and C. Baden-Fuller. 'Competitive groups as cognitive communities: The case of Scottish knitwear manufacturers', Journal of Management Studies, 15, 1989, pp. 397-416. Porac, J., H. Thomas and B. Emme. 'Understanding strategists' mental models of competition.' In G. N. Johnson, (ed.), Business Strategy and Retailing. Wiley, New York and Chichester, 1987, pp. 59-79. Porter, M. E . Competitive Strategy. Free Press, New York, 1980. Powell, W. W. and P. J. DiMaggio (eds.). The New Institutionalism of Organizational Analysis. University of Chicago Press, Chicago, IL, 1991. Prahalad, C. K. and R. P. Bettis. 'The dominant logic: A new linkage between diversity and performance', Strategic Management Journal, 7,1986, pp. 485-501. Reger, R. K. 'Managerial thought structures and competitive positioning'. In A. S. Huff (ed.), Mapping Strategic Thought, Wiley, Chichester, 1990a, pp. 71-88. Reger, R. K. 'The repertory grid technique for eliciting the content and structure of cognitive constructive systems'. In A. S. Huff (ed.), Mapping Strategic Thought, Wiley, Chichester, 1990b, pp. 301-309. Reger, R. K. and T. B. Palmer. 'Cognitive schémas of competition: Understandings of strategy in a turbulent industry'. Paper presented at the Academy of Management Meetings, August, 1990.

232

R. Κ. Reger/A. S. Huff

122

R. Κ. Reger and A. S.

Huff

Reynolds, H. T. Analysis of Nominal Data, Sage, Beverly Hills, CA, 1977. Romesburg, H. C. Cluster Analysis for Researchers, Lifetime Learning Publications, Belmont, CA, 1984. Rosch, E. 'Principles of categorization'. In E. Rosch and B. Lloyd (eds), Cognition and Categorization, Erlbaum, Hillsdale, NJ, 1978, pp. 27-48. Rosch, R. 'Prototype classification and logical classification: The two systems'. In E. Scholnick (ed.). New Trends in Cognitive Representation: Challenges to Piaget's Theory, Lawrence Erlbaum Associates, Hillsdale, NJ, 1981, pp. 73-86. SAS Institute, Inc. SAS Users Guide: Statistics, Version 5 Edition, SAS Institute, Cary, NC, 1985. Schroeder, D. Μ. Ά dynamic perspective on the impact of process innovation upon competitive strategies', Strategic Management Journal, 11, 1990, pp. 25-41. Schwenk, C. R. 'Cognitive simplification processes in strategic decision-making', Strategic Management Journal, 5, 1984, pp. 111-128. Simon, H. A. Administrative Behavior, Macmillan, New York, 1945. Spender, J.-C. 'Strategy making in business'. Unpublished doctoral dissertation. Manchester Business School, University of Manchester, 1980. Spender, J.-C. Industry Recipes, Basil Blackwell, Oxford, England, 1989. Stubbart, C. I. 'The cognitive psychology of strategic planning: Research opportunities', National Academy of Management Meetings, August 1986. Stubbart, C. I. 'Cognitive science and strategic management: Theoretical and methodological

issues', National Academy of Management Meetings, August 1987. Sudharshan, D., H. Thomas and A. Fiegenbaum. 'Assessing mobility barriers in dynamic strategic groups analysis', Journal of Management Studies, 28, 1991, pp. 429-438. Tallman, S. B. 'Strategic management models and resource-based strategies among MNEs in a host country', Strategic Management Journal, 12(Special Issue), 1991, pp. 69-82. Thomas, H. and N. Venkatraman. 'Research on strategic groups: Progress and prognosis', Journal of Management Studies, 25, 1988, pp. 537-555. Tushman, M. L. and E. Romanelli. 'Organizational evolution: A metamorphosis model of convergence and reorientation', Research in Organizational Behavior, vol. 7, JAI Press, Greenwich, CT, 1985, pp. 171-222. Ulrich, D. and B. McKelvey. 'General organizational classification: An empirical test using the United States and Japanese electronics industries', Organization Science, 1, 1990, pp. 99-118. Walton, E. 'Managers' prototypes of financial terms', Journal of Management Studies, 23, 1986, pp. 679-699. Weick, K. E. The Social Psychology of Organizing, Addison-Wesley, Reading, MA, 1979. Zadeh, L. 'Fuzzy sets', Information and Control, 8, 1965, pp. 338-353. Zajac, E. and S. M. Shorten. 'Changing generic strategies: Likelihood, direction, and performance implications', Strategic Management Journal, 10, 1989, pp. 413-430.

APPENDIX

dimensions and results of the dimension level analysis are available from the first author.

This appendix lists examples of the dimensions elicited in the interviews. In all, 331 dimensions were elicited, or approximately 13 per informant. The dimensions presented here have been grouped together by categories agreed upon by three raters. The numbers in parentheses indicate the percentage of informants who provided dimensions in that category (only those categories where at least 25 percent of informants gave dimensions are presented). Every informant provided at least one idiosyncratic dimension (92 of the 331 dimensions or 28 percent of the total). Some examples of dimensions provided by only one informant are 'ethnic clientele vs. broad based clientele,' 'supports liberalized branching laws vs. unit banks,' and 'interested in buying savings and loans vs. not interested in buying savings and loans.' Additional information on procedures for categorizing the

Geographic Scope (88% of informants) national market vs. local strategy international presence vs. no international presence Target Market: Lending (88%) money center, wholesale vs. retail operation middle market vs. retail, small business Growth Strategies (79%) planned expansion through acquisition vs. internal development growth vs. retrenchment and survival were major concerns Location (71%) downtown area vs. suburban operation loop area competitors vs. suburban banks

Strategie Groups: A Cognitive Perspective Strategic Groups Management (63%) ability to implement strategic focus vs. nonstrategically focused average management capabilities vs. high management capabilities

HC Structure and Management (58%) unit bank vs. multibank holding company 'federal approach': collection of banks vs. centralized

Trust (54%) strong trust department vs. no trust department trust department important (commercial and retail) vs. personal trust only

233 123

Product!Market Scope (42%) broad range of financial services vs. focused on a niche full service banks vs. limited services Successful Company (29%) good earnings performers vs. poor earnings performance control overhead vs. not careful with overhead Ownership and Control (25%) public ownership vs. private ownership independently owned and managed by owners vs. managers don't control Asset Based Lending (25%) asset based lenders vs. unsecured lending asset based lending important to strategy vs. asset based lending not important to strategy

6 Die Konkurrentenanalyse

Die Konkurrentenanalyse wird von vielen Autoren als das Kernstück der externen Umweltanalyse oder sogar der Strategieformulierung angesehen (Amit et al., 1988: 431; Porac & Thomas, 1990). Eine detaillierte Analyse der wichtigsten Konkurrenten ist für jede Unternehmung auch sehr sinnvoll, da die Konkurrentenanalyse eine der Voraussetzungen für die Auswahl vorteilhafter Strategien ist. Die Konkurrentenanalyse ermöglicht eine Beurteilung möglicher Strategien im Lichte der Konkurrentenintentionen sowie der Stärken- und Schwächenprofile sowohl der Konkurrenten als auch der eigenen Unternehmung. Die Praxis der Konkurrentenanalyse beinhaltet oft das Erstellen von Konkurrentenprofilen entlang relevanter Variablen, um einen Vergleich mit der eigenen Unternehmung zu ermöglichen. Verschiedene Verankerungen solcher Profile sind denkbar. Die Profile können beispielsweise einen externen Bezug haben, indem sie die Marktpositionen der Konkurrenten abbilden. Oft werden auch interne Aspekte bezüglich der vorhandenen Ressourcen und Fähigkeiten zur Profilbildung herangezogen. Populär ist in diesem Zusammenhang die Erstellung und der Vergleich von Wertketten (Porter, 1985: 62-118; Hax & Majluf, 1996: 125-127). Ein alternativer Ansatz der Konkurrentenanalyse ist die Klassifizierung von Wettbewerbern mit Hilfe von Typologien. Durch die Zuordnung von Konkurrenten zu bestimmten Typen kann man deren generelle strategische Ausrichtung im Sinne typischer oder wahrscheinlicher Verhaltensmuster identifizieren. In der Praxis beliebt (Wheelen & Hunger, 1995: 102) und von der empirischen Forschung bestätigt (Zahra & Pearce, 1990; Doty et al., 1993) sind Klassifizierungen auf der Basis der Typologie von Miles und Snow (1978). Miles und Snow unterscheiden vier Typen von Unternehmungen: Verteidiger, Prospektoren, Analysierer und Reaktoren. Jeder dieser Typen verläßt sich auf unterschiedliche Strategien, um die Herausforderungen der externen Umwelt zu beantworten, und jeder Typ ist durch eine spezifische Konfiguration von Struktur, Prozeß- und Organisationskulturmerkmalen gekennzeichnet. Klassifikationen dieser Art lassen nicht nur erkennen, warum Unternehmungen, die ähnlichen Umweltbedingungen ausgesetzt sind, unterschiedlich reagieren, sie helfen auch, Szenarien für zukünftige Marktentwicklungen und Wettbewerbshandlungen zu erstellen. Ein weiteres Mittel der Konkurrentenanalyse ist das Erstellen von Karten strategischer Gruppen (Porter, 1980: 126-155; Hax & Majluf, 1996: 97-104). Der Wert dieser Gruppenkarten für die Konkurrentenanalyse ist allerdings umstritten (Chen, 1996). Es ist z.B. weder theoretisch noch empirisch geklärt, ob die Mitglieder einer strategischen Gruppe durch höhere oder niedrigere Rivalitäten gekennzeichnet sind, als die Mitglieder aus verschiedenen strategischen Gruppen (Cool & Dierickx, 1993; Kotler & Armstrong, 1989: 496).24 24

Vgl. hierzu auch die Ausführungen in Kapitel 5.

236

R. Κ. F. Bresser

Ein besonders interessantes, neues Verfahren der Konkurrentenanalyse wird von Chen (1996) propagiert. Unter Rückgriff auf die Literatur zur Mehrpunktkonkurrenz (Karnani & Wernerfeit, 1985; Porter, 1985: 353-361) und zum ressourcenbasierten Ansatz in der strategischen Managementforschung (siehe Kapitel 7) entwickelt und operationalisiert Chen zwei Dimensionen, die eine dyadische Klassifizierung von Wettbewerbern ermöglichen: die Marktgemeinsamkeit und die Ressourcenähnlichkeit. Die Marktgemeinsamkeit mißt das Ausmaß, in dem zwei Firmen in einem Markt direkte Wettbewerber sind. Die Ressourcenähnlichkeit ist als die Ausstattung von zwei Wettbewerbern mit vergleichbaren strategischen Ressourcen definiert. Das Interessante an diesem Ansatz ist, daß er die Aufmerksamkeit auf die Dynamik von Wettbewerbsinteraktionen zwischen Konkurrenten lenkt, und zwar im Sinne des Wechselspiels von Wettbewerbsvorstößen (Attacken) und Verteidigungsmaßnahmen (Vergeltungen). Chen geht davon aus, daß die Intensität der Variablen Marktgemeinsamkeit und Ressourcenähnlichkeit Prognosen darüber zuläßt, welcher Konkurrent sich zu einer Wettbewerbsattacke und welcher sich zu einer Verteidigungsmaßnahme entschließen wird. Das Instrumentarium ist dabei nicht nur auf die Konkurrenten einer Branche beschränkt, d.h. Prognosen über wahrscheinliches Wettbewerbsverhalten sind möglich, unabhängig davon, ob die Konkurrenten sich auf der Gesamtunternehmungsebene (corporate level), der Geschäftsbereichsebene (business level) oder im internationalen Wettbewerb gegenüberstehen (Chen, 1996: 124). Chen (1996) verankert seine Überlegungen auch in einigen empirischen Studien aus den frühen 90er Jahren, die allesamt als Studien des strategischen Managements mit industrieökonomischen und verhaltenswissenschaftlichen Wurzeln gelten können, und die sich mit den Auswirkungen von Wettbewerbsattacken und Verteidigungsmaßnahmen auf die Unternehmungsrentabilitäten auseinandersetzten. Die Evidenz dieser Studien unterstreicht insbesondere die Vorteilhaftigkeit eines aggressiven Wettbewerbsverhaltens. Young et al. (1996) stellen z.B. fest, daß Unternehmungen um so bessere Rentabilitäten erzielen, je häufiger sie Wettbewerbsvorstöße unternehmen. Chen und MacMillan (1992) berichten, daß attackierende und schnell nachfolgende Unternehmungen sich Marktanteile zu Lasten der spät folgenden Unternehmungen sichern können. Smith et al. (1991) belegen, daß Unternehmungen mit einer hohen Neigung, auf Wettbewerbsvorstöße mit Vergeltungsmaßnahmen zu reagieren, höhere Rentabilitäten erzielen als Firmen mit einer geringen Reaktionsneigung. Demgegenüber ermitteln Chen und Miller (1994), daß die Rentabilitäten insbesondere jener Unternehmungen leiden, die viele Verteidigungsmaßnahmen ihrer Konkurrenten provozieren. Auf einer generellen Ebene betrachtet, belegen diese Studien auch folgendes: Die Art und Intensität von Wettbewerbsinitiativen und Vergeltungsmaßnahmen beeinflußt die Rentabilität der Akteure. Damit bestätigen diese Studien die Grundstruktur der meisten spieltheoretischen Modelle. Ein Spiel wird in der Spieltheorie definiert als eine Gruppe von Akteuren, die bestimmte Spielzüge (Initiativen und Reaktionen) hintereinander durchführen, wobei die Art der Spielzüge Konsequenzen für die „pay-offs" der Akteure hat (Camerer, 1991: 139).25 Während sich in den oben aufgeführten Studien vereinzelt Hinweise auf die spieltheoretische Literatur finden lassen, haben sich andere Bausteine 25

Diese Definition eines Spiels impliziert, daß es sich um ein Wiederholungsspiel oder ein dynamisches Spiel handelt, bei dem bestimmte Spielzüge begrenzt oder unbegrenzt wiederholt werden. In solchen Kontexten ist es sinnvoll, die Spielzüge der Akteure als Initiativen oder Reaktionen zu interpretieren. Derartige Interpretationen von Spielzügen sind bei sogenannten „One-Shot"-Spielen nicht möglich,

Konkurrentenanalyse

237

der modernen Spieltheorie, z.B. Begriffe wie Gleichgewichtslösungen, Rationalisierbarkeit, ,3attle of the Sexes", usw. bisher nicht oder doch nur sehr vereinzelt auf die strategische Managementliteratur ausgewirkt (Camerer, 1991: 140). Woran liegt das?26

Die neue Industrieökonomik Die Spieltheorie stellt das Studium mathematischer Modelle von Wettbewerbssituationen zwischen rationalen Akteuren dar (Shubik, 1984; Camerer, 1991). Sie ist in ihren modernen Ausprägungen kennzeichnend für die „neue" Industrieökonomik. Nachdem die Querschnittsstudien der traditionellen Industrieökonomik hinsichtlich der Entwicklung theoretischer Kausalmodelle wenig befriedigend waren, entwickelte sich in den 70er Jahren ein Forschungszweig der Industrieökonomik, der die nicht-kooperative Spieltheorie27 als Standardmethode der Analyse von Wettbewerbssituationen einsetzt. Durch diesen Methodenwechsel erhofft man sich Fortschritte bei der Entwicklung dynamischer, logisch konsistenter Modelle und Theorien zur Erklärung von Wettbewerbsverhalten (Tiróle, 1988: 2f.). Durch diesen Methodenwechsel verlagert die industrieökonomische Forschung auch eindeutig ihren Analyseschwerpunkt, und zwar auf die Ebene der einzelnen Unternehmung und weg von der Branchenebene. Damit wird die Industrieökonomik/Spieltheorie an sich sehr interessant für eine Theorie des Strategischen Managements, denn sie analysiert Wettbewerbsverhalten von Einzelunternehmungen und damit die Ausübung von „strategic choice" (Knyphausen, 1995: 62). Leider hat die moderne Spieltheorie sich bisher nur recht begrenzt auf die strategische Managementforschung ausgewirkt, zumindest was die Zahl der explizit spieltheoretisch ausgerichteten Aufsätze zum Strategischen Management betrifft: Amit et al. (1988) nutzen beispielsweise die Logik der Spieltheorie, um „conjecural variations", d.h. Vermutungen der Konkurrenten über das eigene Wettbewerbsverhalten, in der strategischen Planung zu verankern. Weigelt und Camerer (1988) verwenden die Spieltheorie, um das Konzept der Unternehmungsreputation zu formalisieren und um Verhaltensimplikationen ihrer formalen Modelle zu diskutieren. Weigelt und MacMillan (1988) entwickeln einen spieltheoretisch fundierten Analyserahmen, der

26

27

denn hier können Akteure sich lediglich einmal, und zwar simultan, für einen Spielzug entscheiden (Saloner, 1991: 122; Camerer, 1991: 139; Knyphausen, 1995: 70). Es gibt wohl keine ökonomisch inspirierte Theorie, deren Wert für das Strategische Management umstrittener ist als die Spieltheorie. Während spieltheoretisch ausgerichtete Ökonomen (z.B. Saloner, 1991; Camerer, 1991), den potentiellen Beitrag der Spieltheorie zur Entwicklung des Strategischen Managements hoch veranschlagen, sind führende Vertreter des Strategischen Managements, die gegenüber ökonomischen Theorien ansonsten sehr aufgeschlossen sind, sehr skeptisch, wenn nicht gar ablehnend (vgl. z.B. Teece, 1990; Rumelt et al., 1991). Teece (1990: 54) urteilt z.B., die Spieltheorie „has almost nothing to offer at this time to key issues in strategic management". Die in der Spieltheorie getroffene Unterscheidung zwischen kooperativen und nicht-kooperativen Spielen differenziert danach, ob zwischen Spielern bindende Vereinbarungen getroffen werden können. Bei kooperativen Spielen ist dies der Fall. Hier besteht die Möglichkeit, Abmachungen zwischen den Spielern exogen durchzusetzen, z.B. unter Rückgriff auf das geltende Rechtssystem. Bei nicht-kooperativen Spielen können die Spieler zwar Abmachungen über das vorzunehmende strategische Verhalten treffen, diese Abmachungen sind aber nicht durchsetzbar. Klar ist, daß nichtkooperative Spiele die Realität der strategischen Unternehmungsführung wesentlich besser einfangen als kooperative Spiele (Holler & Illing, 1991: 183ff.; Knyphausen, 1995: 68).

238

R. Κ. F. Bresser

Managern bei der Beurteilung strategischer Alternativen helfen soll. Lieberman (1987) verwendet ein spieltheoretisches Modell, um die Implikationen von Lernkurveneffekten auf das Wettbewerbsverhalten von Unternehmungen zu simulieren. Auf die bereits aufgeworfene Frage nach den Gründen für die beschränkte Integration der modernen Spieltheorie in die strategische Managementforschung gibt es mehrere Antworten, die mit der Entwicklung der modernen Spieltheorie zusammenhängen.28 Seit der Entwicklung der mathematischen Spieltheorie durch Neumann und Morgenstern (1944) hat diese, inspiriert durch die Arbeiten von Selten (1965) und Harsanyi (1967), erhebliche Fortschritte gemacht. Während ursprünglich nur Spiele mit vollständiger Information als lösbar galten, können heute auch für dynamische, nicht-kooperative Spiele mit unvollständiger Information Gleichgewichte berechnet werden (Knyphausen, 1995: 69). Allerdings sind auch die heutigen Modelle noch mit Problemen behaftet, die ihren theoretischen Wert sowie ihre praktische Anwendbarkeit beeinträchtigen. Insbesondere drei Problembereiche behindern die Übertragung spieltheoretischer Erkenntnisse auf das Strategische Management: (1) Das Rationalitätsproblem. Obwohl die moderne Spieltheorie unvollkommene und unvollständige Informationen berücksichtigt, stellt sie sehr hohe Rationalitätsanforderungen an die Akteure. Aus der Sicht einer „realistischen" Theorie des Strategischen Managements ist dieser Umstand ein erhebliches Hemmnis (Zajac & Bazerman, 1991; Zajac, 1992). Camerer (1991) und Saloner (1991) verweisen zwar darauf, daß viele spieltheoretische Modelle keine extrem hohen Rationalitätsanforderungen stellen und dennoch Gleichgewichtslösungen herbeiführen können: Bei genauerer Betrachtung verbleiben aber immer noch erhebliche Anforderungen an die Rationalität und die Informationsverarbeitungskapazität der Akteure, so daß die Realismuskritik nach wie vor berechtigt bleibt (Knyphausen, 1995: 71f.). (2) Das Problem der Gleichgewichtslösungen. Spieltheoretische Modelle basieren auf bestimmten Gleichgewichtskonzepten, z.B. Nash-Gleichgewichten (Knyphausen, 1995: 73). Eine Nash-Gleichgewichtslösung ist ein theoretisch vorgegebener Optimalzustand, von dem abzuweichen kein Spieler einen Anreiz hat, da beide Spieler unter Berücksichtigung der Strategien der jeweils anderen Seite ihr jeweils bestmögliches Resultat erzielen (Besanko et al., 1996: 3If.). Aus der Sicht einer Theorie des Strategischen Managements stellt sich die Frage, ob solche Gleichgewichte plausibel sind und ob nicht vielmehr Ungleichgewichte in der ökonomischen Realität gang und gäbe sind (Jacobson, 1992). Außerdem ist die moderne Spieltheorie auch mit dem Problem multipler Gleichgewichte behaftet. Vor allem dyna-

28

Für Mißverständnisse sorgt sicherlich auch der Umstand, daß die Spieltheorie und das Strategische Management den Strategiebegriff unterschiedlich definieren. Während das Wort Strategie sich in der Spieltheorie auf die Planung einer bestimmten Folge von Spielzügen (Handlungen) eines rationalen Akteurs bezieht (Camerer, 1991: 139; Holler & Illing, 1991: 34), hat der Begriff im Strategischen Management eine umfassendere, komplexere Bedeutung. Strategien sind im Strategischen Management zwar auch Handlungspläne, jedoch sind diese im Kontext einer sehr vieldeutigen, unsicheren Umwelt abgeleitet, und ihre Inhalte verändern sich in der Regel während ihrer Implementierung (Knyphausen, 1995: 75f.; Mintzberg & Waters, 1985). Um die Anpassungsfähigkeit einer Unternehmung zu bewahren, werden strategische Pläne oft sogar bewußt vage formuliert. Demgegenüber ist eine Strategie im spieltheoretischen Sinn eine vollständige Beschreibung der Handlungen, die ein Spieler auszuführen plant.

Konkurrentenanalyse

239

mische Spiele mit unvollständiger Information sind für eine Vielzahl von Gleichgewichtslösungen anfällig (Saloner, 1991: 129f.; Knyphausen, 1995: 74). (3) Das ,¿Collage "-Problem. Die spieltheoretischen Modelle sind zu fragmentiert, denn jedes Modell konzentriert sich auf ein enges Spezialproblem. Was fehlt, ist eine zusammenhängende Theorie, die dieses Mosaik an Modellen integriert. So können Spieltheoretiker z.B. das Phänomen der Abschreckung neuer Konkurrenten ganz unterschiedlich modellieren, etwa auf der Basis einer Reputation für aggressive Vergeltungsmaßnahmen, als Resultat von Faktorspezifität, auf der Basis von Economies of Scope, usw. Diese separaten Modelle sind allerdings wenig informativ im Hinblick auf die relative Bedeutung der aufgeführten Faktoren in spezifischen Situationen. Die fehlenden Qualifizierungen erschweren die empirische Überprüfung der Modelle und damit ihren möglichen Nutzen für die Entwicklung einer Theorie des Strategischen Managements (Teece, 1990; Rumelt et al., 1991; Camerer, 1991). Diese Beschränkungen der neuen Industrieökonomik verdeutlichen, daß der praktische Wert spieltheoretischen Denkens sich darauf beschränkt, eine Heuristik zu sein, die zu besseren strategischen Entscheidungen führen kann. Damit ist der praktische Beitrag der Spieltheorie nicht höher zu veranschlagen, als der anderer Heuristiken des Strategischen Managements, z.B. der SWOT-Analyse, der Szenariotechnik oder der Portfolioanalyse. Die Spieltheorie stellt Werkzeuge zur Konstruktion logischer Modelle über Rationalverhalten zur Verfügung. Der strategische Manager kann durch solche Modelle erfahren, was Akteure tun würden, wenn sie sich in bestimmten Situationen wirklich rational verhielten (Postrel, 1991). In diesem Sinne kann die Spieltheorie dazu beitragen, das Denken der Praktiker zu schärfen (Saloner, 1991: 126ff.). Hinsichtlich des Beitrags der Spieltheorie zur weiteren Entwicklung einer Theorie des Strategischen Managements ist zu erwarten, daß sie bisher vernachlässigte theoretische Konstrukte und Variablen für die empirische, verhaltenswissenschaftlich ausgerichtete Strategieforschung identifiziert, z.B. Begriffe wie „Vermutungen von Akteuren", „Länge der Rivalitätsbeziehungen", usw. (Camerer, 1991: 147; Porter, 1991: 98). Darüberhinaus wäre auch durch eine verhaltenswissenschaftliche Anreicherung spieltheoretischer Modelle, etwa im Hinblick auf realistischere Rationalitätsprämissen, ein Beitrag zur Theorieentwicklung denkbar (Zajac & Bazerman, 1991; Zajac, 1992; Kadane & Larkey, 1982; 1983).

Die Artikel Von den folgenden drei Beiträgen haben die beiden ersten einen direkten Bezug zum Verhältnis zwischen der Spieltheorie und der strategischen Managementforschung. Der dritte Beitrag ist umfassender. Er spricht Probleme an, die sowohl für die traditionelle als auch für die neue Industrieökonomik kennzeichnend sind. Die Lösung der Probleme soll durch die Fokussierung auf eine alternative ökonomische Theorie ermöglicht werden. Der erste Artikel, „An Interactive Strategie Analysis Framework", von Weigelt und MacMillan stellt einen spieltheoretischen Analyserahmen vor, der Managern bei der systematischen Bewertung und Auswahl strategischer Alternativen helfen soll. Der Analyserahmen wird anhand eines Zwei-Konkurrentenfalles exemplifiziert, der sich tatsächlich

240

R. Κ. F. Bresser

in der Praxis abgespielt hat. Beide Unternehmungen sind in hohem Maße interdependent, und beide wollen ihre Marktanteile erhöhen. Die eine (für diesen Fall fokale) Unternehmung erwägt, zum Zwecke der Marktanteilssteigerung ihren Kunden längere Zahlungsziele einzuräumen. Für die andere Unternehmung werden unterschiedliche Reaktionen für wahrscheinlich gehalten, sie könnte z.B. mit möglichen Zahlungszielverlängerungen der fokalen Unternehmung gleichziehen, oder sie könnte statt dessen die Preise um zehn Prozent senken. Für verschiedene Szenarien werden Payoff-Matrizen erstellt, und eine Nash-Gleichgewichtslösung wird berechnet. Die Lösung impliziert, daß die fokale Unternehmung auf jeden Fall ihre Zahlungsziele von 60 auf 90 Tage ausweiten soll. Der Beitrag zeigt, daß einige wenige Autoren offensichtlich ganz enthusiastisch an einer Übertragung spieltheoretischer Erkenntnisse auf das Strategische Management arbeiten. Dem Beitrag ist allerdings auch zu entnehmen, daß die Anwendbarkeit des Modells in der Praxis eher beschränkt sein dürfte. Die Anwendung ist nicht nur kompliziert, sie stellt auch sehr hohe Informations- und Rationalitätsanforderungen an die Akteure, wie die Ausführungen über das, was als „common knowledge" zu gelten hat, belegen. Der zweite Beitrag, „Blind Spots in Industry and Competitor Analysis: Implications of Interfirm (Mis)perceptions for Strategic Decisions", von Zajac und Bazerman versucht, für drei strategische Entscheidungen, die oft Gegenstand spieltheoretischer Modelle sind, realistischere Rationalitätsannahmen zu entwickeln. Die Entscheidungen betreffen: Kapazitätsausweitungen, Markteintritt durch interne Entwicklungen und Markteintritt qua Akquisition. Unter Rückgriff auf die sozialpsychologische Forschung zeigen die Autoren, daß strategische Akteure in Wettbewerbskontexten regelmäßig und häufig Fehlbewertungen vornehmen, die auf „blind spots" in ihren Wahrnehmungen zurückzuführen sind. Derartige „blind spots" der Wahrnehmung beziehen sich z.B. auf Phänomene wie „übertriebene Zuversicht" oder eine „Eskalation der eingegangenen Verpflichtungen". Durch ihre verhaltenswissenschaftlichen Überlegungen können Zajac und Bazerman hartnäckige und nur schwer verständliche Marktzustände und -geschehnisse erklären, wie z.B. Überkapazitäten, schnelle Konkurse junger Unternehmungen und überhöhte Akquisitionsprämien. Die Autoren regen an, die Instrumente der Konkurrentenanalyse um die diskutierten typischen Wahrnehmungsverzerrungen von Konkurrenten zu erweitern. Dies gilt auch für spieltheoretische Modelle, in denen, wie dargestellt, Vermutungen über das wahrscheinliche Verhalten der Konkurrenz einen wichtigen Stellenwert haben. In dem Ausmaß, in dem spieltheoretische Modelle von realistischeren Annahmen über Konkurrentenverhalten ausgehen, sind sie eher dazu geeignet, zur weitern Entwicklung einer Theorie des Strategischen Managements beizutragen. Der dritte Artikel, „The ,Austrian' School of Strategy", von Robert Jacobson bricht mit grundlegenden Prämissen der traditionellen und der neuen Industrieökonomik, indem er einen alternativen Ansatz der Ökonomie favorisiert. Die sogenannte „Österreichische Schule" der Ökonomie begann mit Carl Menger und setzte sich fort über Ludwig Mises, Friedrich Hayek und Joseph Schumpeter bis zu ihren modernen Vertretern, z.B. Israel Kirzner. Jacobson argumentiert, die Ansätze der „Österreicher" eigneten sich besser zur Entwicklung einer Theorie des Strategischen Managements als die durch die neoklassische MikroÖkonomie inspirierten industrieökonomischen Ansätze, da die ersteren von kontinuierlicher Innovation und einer auf Anpassung an veränderte Umweltbedingungen ausgerichteten, flexiblen strategischen Planung ausgingen. Im Brennpunkt der Betrachtungen der „Österreicher" ist der „Entrepreneur", der nach supranormalen Gewinnen trachtende Unternehmer. Der Gewinnanreiz der Unternehmer ist der Motor der Innovati-

Konkurrentenanalyse

241

on, der Realisierung von Gewinnchancen in einer sich kontinuierlich verändernden Welt. Da die Konkurrenten eines Marktes erfolgreiche Innovationen imitieren, sind supranormale Gewinne immer nur temporär, was einen weiteren Anreiz zu kontinuierlicher Innovation darstellt. Die Welt der Ökonomie befindet sich also typischerweise im Ungleichgewicht. Das für die traditionelle Industrieökonomie und die Spieltheorie so wichtige Konzept des Gleichgewichts ist somit eine recht unglückliche Annahme, denn es setzt einen ungerechtfertigt hohen Wissensstand bei allen Akteuren voraus. Die neoklassische Vorstellung der vollkommenen Information vernachlässigt die Bedeutung des nach Innovationen trachtenden Unternehmers, und insofern ist auch die Gleichgewichtsannahme unrealistisch, denn sie impliziert, daß in einem Markt keine Innovationen mehr stattfinden. Gegenüber den industrieökonomischen Ansätzen hat die „Österreichische Schule" zumindest vier Grundannahmen, die der Entwicklung einer Theorie des Strategischen Managements entgegenkommen: (1) Das strategische Ziel ist die Förderung von Innovationen anstatt die Begrenzung der Marktkräfte, die eine generelle Wohlfahrtsreduzierung herbeiführt. (2) Die neoklassische Vorstellung von Marktgleichgewichten wird zugunsten von Ungleichgewichten als Normall aufgegeben. (3) Anstelle von empirischen Gesetzmäßigkeiten oder formalen Optimalzuständen rückt die Heterogenität des Unternehmungserfolges in den Vordergrund der Analysen. (4) Hinsichtlich der Determinanten des Unternehmungserfolges verschiebt sich der Fokus von beobachtbaren auf nichtbeobachtbare (intangible) Faktoren und Ressourcen. Die Grundannahmen des „Österreichischen Ansatzes" lassen ihn als geeignet erscheinen, zur Entwicklung einer dynamischen Theorie des Strategischen Managements beizutragen. Obwohl ein nicht unerheblicher Teil der strategischen Managementforschung „Österreichische" Qualitäten aufweist (Jacobson, 1992: 802f.), u.a. auch der im nächsten Kapitel behandelte „ressourcenbasierte Ansatz", wird ein expliziter Zusammenhang mit diesem Theoriegebäude erst seit jüngster Zeit und bisher nur vereinzelt hergestellt (vgl. z.B. Young, 1995; Young et al., 1996). Die zukünftige Entwicklung des Strategischen Managements wird zeigen müssen, inwieweit die „Österreichische Schule" die industrieökonomischen Ansätze ablösen kann und inwieweit sie zur theoretischen Untermauerung externer (und auch interner) Umweltanalysen beizutragen in der Lage ist.

242

Κ. Weigelt/I. M a c M i l l a n

Strategic

Management

Journal,

Vol.

9, 27-40

(1988)

AN INTERACTIVE STRATEGIC ANALYSIS FRAMEWORK KEITH WEIGELT Graduate School of Management,

New York University, New York, New York,

IAN MACMILLAN Wharton School, University of Pennsylvania, Philadelphia, Pennsylvania,

U.S.A.

U.S.A.

A framework is presented that helps managers systematically evaluate strategic alternatives. We empirically illustrate how lo use the framework, and show how changes in beliefs about environmental factors affect the payoffs of specified strategic alternatives.

INTRODUCTION Choosing a m o n g strategic alternatives is usually difficult. M a n a g e r s e n c o u n t e r this difficulty because they lack perfect foresight. T h e y must choose a course of action today, whose success depends on f u t u r e conditions, without knowing exactly what t h e f u t u r e looks like. T h e r e f o r e , in selecting a strategic alternative, the m a n a g e r is actually placing a ' b e t ' : h e is using his beliefs about t h e f u t u r e to choose a plan that best fulfills his firm's objectives. W e present a f r a m e w o r k which systematically evaluates and compares the effectiveness of strategic alternatives given different scenarios of t h e f u t u r e . M a n a g e r s should find t h e f r a m e w o r k helpful in placing their strategic bets for a n u m b e r of reasons. O n e , given t h e m a n a g e r ' s beliefs about t h e f u t u r e , t h e f r a m e w o r k shows which strategy produces t h e optimal results and why. It compares the p e r f o r m a n c e of t h e optimal strategy vis-à-vis that of alternative strategies, thereby indicating to m a n a g e r s how costly t h e use of non-optimal strategies can be. Since t h e f r a m e w o r k systematically evaluates strategies it can also illustrate to managers how changes in beliefs a b o u t t h e f u t u r e affect the p e r f o r m a n c e of a strategy. T w o , t h e business environment is becoming m o r e competitive and complex. Various factors have increased competitive pressures in m a r k e t s , and m a d e relationships between environmental 0143-2095/88/050027-14$07.00 © 1988 by John Wiley & Sons, Ltd.

factors m o r e complex (e.g. global markets, faster technology transfers). W e a k e r firms are being driven f r o m markets, with the m o r e aggressive survivors having significant resources to exploit any strategic opportunity. In such an environment a competitive advantage is the ability to quickly and correctly interpret changes in the environm e n t , a n d d e t e r m i n e what actions, if any, the firm should t a k e . Because of its systematic n a t u r e , t h e f r a m e w o r k first helps managers m a k e statistically correct inferences f r o m available d a t a , and t h e n explicitly shows how environmental changes affect strategic p e r f o r m a n c e . Skeptics may argue the worth of such systematic evaluation; after all won't managers get the same results f r o m using 'gut feelings'? Such reasoning ignores t h e mounting evidence showing the inability of h u m a n s to intuitively calculate probabilities, and draw correct inferences (see Kahnem a n , Slovic and Tversky, 1982). Because of b o u n d e d rationality, h u m a n s use heuristics to process information and these heuristics can cause decision biases which lead to p o o r j u d g e m e n t s (i.e. incorrect inference f r o m available data). F o r e x a m p l e , in a randomly selected group of 24 p e o p l e , what is the probability that any two p e o p l e have t h e s a m e birthday? T h e correct answer is 50 p e r c e n t , but few people give this reply even though most know the relevant d a t a ; namely that each person has a 1/365 chance of being b o r n on any given day.

Interactive Strategie Analysis Framework 28

Κ. Weigelt and I.

243

MacMillan

Three, recent articles have viewed the competitive strategic process as being dynamic in nature (Ghemawat, 1986). In such an environment, managers need to view strategic planning not as a series of static one-shot decisions, but as an ongoing dynamic process. Competitive advantages are generally short-lived; the manager's strategic goal cannot be simply to gain a competitive advantage, but to recognize that every advantage is eventually eroded by competitive counterattack and must be replaced by a 'second round' advantage. Gaining the competitive advantage is not enough; one must sustain it. Managers will find the framework useful in such a planning context. The framework can help identify not only those advantages that are more sustainable, but also what actions competitors must take to overcome a specified advantage, and whether the competitor presently has the resources to do so. Additionally, the framework can help a manager better anticipate the future actions of competitors. A manager can generate a series of scenarios and use the framework to identify the most likely actions of competitors in each scenario. This can help a manager maintain strategic control of the market since he can better anticipate the future actions of competitors, which gives him the time to do more detailed contingency planning. The remainder of the paper is as follows. In the next section we use an illustrative case to show how the framework works, so the reader can get a better understanding for the framework's underlying theoretical foundations. A discussion of model uses and limitations then follows. Finally, we present conclusions and avenues for future research.

THE FRAMEWORK Before formally presenting the components of the framework, we discuss how the framework explicitly acknowledges two key attributes of the strategic problem, and why this gives the framework greater analytic power. Payoffs of strategic decisions are interactive in nature. This means that a firm's payoff for any given strategy is dependent on the firm's action and the actions of competitors. For example, the payoff to a firm that increases its prices will

depend on whether its competitors did or did not raise their prices. Thus, in choosing between strategic alternatives, managers need to consider (and anticipate) the future actions of competitors. Selecting among strategic alternatives without considering the possible moves of competitors is like trying to determine who won a football game and only hearing the score of one team. Unfortunately, most strategic analysis frameworks are myopic in nature; the interactive nature of payoffs is not explicitly recognized. However, our framework does explicitly recognize that no independent optimal strategic choice exists: strategies are optimal conditional on the actions of competitors. Thus, the framework shows managers how a strategic alternative will fare against a range of future actions by competitors. This helps managers assess the riskiness of a strategic alternative (using variance as a risk measure), and relative to non-interactive frameworks, it gives them better information on which to make a decision. Correctly anticipating the future moves of competitors is a formidable, but unavoidable, task. It requires that managers examine the feasible choices of competitors, and then select the most likely ones. The task is difficult for two primary reasons. First, competitors usually can choose from a large set of options. Second, managers lack key data about their competitors (i.e. preferences, competitive attributes). Therefore they assess competitors in an environment of uncertainty, and must use personal beliefs. For example, while managers at firm X have some idea of firm Y's cost structure, they do not know it with certainty. Firm X's managers want to formulate a new pricing strategy, and in considering the possible future pricing moves of firm Y, they look at firm Y's cost structure. Because they don't know firm Y's costs, X's managers form beliefs based on known information. For instance: W e k n o w what o u r manufacturing costs are, so let's use that information along with any additional information we have. W e know they are using the same manufacturing techniques as us, and that their labor costs are roughly the same since the s a m e union represents workers at b o t h plants. O u r valve supplier told us they are using the same valves as us, but they are vertically integrated so they produce their own plastic while we buy ours . . .

244

Κ. Weigelt/I. MacMillan An Interactive Strategic Analysis Framework

Thus, a manager's beliefs influence his decision choice. Our framework recognizes this relationship and systematically shows managers how changes in beliefs should influence choices. Because the framework asks managers to quantify beliefs, it highlights whether a small change in beliefs should or should not change his decision choice. The theory underlying our framework is game theory. We realize that many strategic planning researchers regard game theory as 'a theory of undelivered promises'—though highly touted in the mid-1950s, it failed to make meaningful contributions to the field. Today, most strategic planning researchers are exposed to only one branch of game theory (games with complete information) where simple games, such as the prisoner's dilemma and the battle of the sexes, are illustrated with a 2 x 2 matrix. Because a basic assumption of complete information models is that options, outcomes, and preferences are clearly specified and known by all players, the models have limited applicability. Few (if any) business strategic situations fulfill such restrictive assumptions. While early game theorists were overly optimistic in touting the theory's applicability, recent theoretical advances have reduced the gap between promised and actual results. Our framework is based upon a model of incomplete information—the restrictive assumption of complete information is relaxed. Harsanyi (1967-68) formally introduced this model with a series of papers on Bayesian games. The incomplete information model recognizes that players may make strategic decisions without knowing all relevant information. Since players are not equally informed about parameters that define payoffs and strategies, players form beliefs based on known information. Important for our purposes is that the theory formally recognizes both the interactive nature of payoffs, and that players must make decisions without having complete information about the strategic setting. The competitive situation Our illustrative case is based on an actual (though disguised) event. Firms A and Β are both divisions of large firms, and are the dominant players in their industry. The essence of firm A's strategy against Β is to gain market share via

29

'salami slicing': that is to say, to make a series of modest strategic moves that do not provoke a serious response. (This is not unlike the entry strategies of several Japanese firms into the U.S. market.) Thus, while each individual move might appear trivial, in the long run firm A expected to gain market share precisely because firm Β would not see any individual move as threatening. The challenge facing firm A's managers was to identify such moves and their likelihood of a reaction. Managers at firm A heard a rumor that B's parent company was concerned about the possible downgrading of its bond rating. Firm Β was a mature industrial company whose corporate role was suspected to be that of a 'cash cow'. The managers at firm A wanted to know how to capitalize on this situation. Firm A's managers considered offering distributors longer credit terms, from the current 60 days to 90 or 120 days. Their reasoning was as follows: if firm Β matched the offer, its available cash would decrease because of the slower payment of receivables. This might adversely affect key financial ratios, thereby putting greater pressure on the bond rating of firm B's parent corporation. Rather than match firm A's offer, firm Β might sacrifice a modest amount of market share to protect the bond rating. If firm A did not appear 'greedy' (in terms of market share gain), and firm B's losses were modest then Β would probably not retaliate: however, if Β thought A was greedy, then Β would institute price cuts thereby starting a price war.

F R A M E W O R K COMPONENTS Player types Information that is known to one player but not to anyone else is called that player's private information. We summarize the relevant private information with a device called the player's 'type'. In business strategic situations this relevant private information (i.e. type) generally consists of competitive attributes. Private information is relevant if it meets two conditions: one, the information is not known with certainty by other players; two, the private information affects payoffs. In our previous example, firm Y's type consisted of its cost structure. Since firm X's

Interactive Strategie Analysis Framework 30

Κ. Weigelt and I.

245

MacMillan

managers did not know Y's cost structure (i.e. type) they had to form a probability distribution over possible types. Note how these beliefs about Y's cost structure will affect the strategic choice decision by X's managers. If X's managers think it is highly likely that Y has a lower cost structure than X, they will probably choose a different pricing scheme than if they think X's cost structure is lower than Y's. Player types, their relation to strategic choice, and the use of subjective probability distributions to assess possible types is naturally appealing. In discussing competitive analysis, Porter (1980) uses a similar, but more qualitative, concept. He develops competitors' 'profiles' to predict future moves. Additionally, managers act as if knowing a rival's type is strategically important. Firms spend millions of dollars on intelligence gathering, and spend equal amounts protecting their confidential information (like costs). Below, we show why—knowing a competitor's type reduces uncertainty, therefore making it easier to predict its future moves. Types can present a wide range of attributes, and can include both physical and non-physical resources (since non-physical resources also affect behavior). For example, R&D programs, cost structures, information-gathering networks, plant capacity, cash reserves, management values, and firm objectives. From a practical perspective if the number of attributes grows too large, the model becomes unwieldy. 1 Though model limitations are discussed in the next section, we would like to make two points here: first, a player's type need only consist of those attributes that are relevant (i.e. affect payoffs and are privately known); second, attributes can be bundled, thus reducing the number of possible types. For expository purposes we limit the possible attribute types of our players to two. If firm Β is concerned with the possible downgrading of its bond rating designate it as type 1 (b 1 ), if it is not concerned, designate the firm as type 2 (b 2 ). Similarly, if firm A is satisfied with a market share increase of l.S percent, designate it as not being greedy (a 1 ). If it wants a 3 percent increase, label it as greedy (a 2 ). 1 Though from a mathematical viewpoint, Mertens and Zamir (1983) show that type-sets can capture all the different information and beliefs players may bring into a game.

We can now differentiate between possible environments by using a combination of players' types. Our game will be played in one of four different environments: firm A is not greedy and firm Β is worried about its bond rating (a',b'); firm A is not greedy and firm Β is not worried about its bond rating (a',b 2 ); firm A is greedy and firm Β is worried about its bond rating (a 2 ,b'); and firm A is greedy and firm Β is not worried about its bond rating (a 2 ,b 2 ). We might expect firms to use different strategies and have different payoffs conditional on the true environment. For example, if firm B's managers are actually worried about B's bond rating they may choose a different strategy than if they are not worried. Additionally, the managers at firm Β may prefer one result if they are worried about their bond ratings, and another result if they are not worried. Conversely, if managers at firm A believe that B's managers are worried about B's bond ratings then they might choose a different strategy than if they don't think B's managers are worried. Though each firm can narrow the possible environments down to two (since each knows its type), each must consider all four because of the incomplete information and the interactive nature of payoffs. Consider the manager of firm A. He does not know whether B's managers are worried about its bond ratings, although this factor will influence their decision choice, and thus the payoff from his choice (because payoffs are interactive). Since he cannot observe whether B's managers are worried or not, he must form conjectures about what actions the managers would take if they were worried. That is, firm A's manager must consider what actions each possible type of firm Β would take. Currently, firm B's managers must recognize that A's manager performs this process because an action for Β is optimal only conditional on A's action, which in turn is conditional on A's beliefs. The same reasoning holds true for firm B's managers with respect to the situation facing A's manager. This process underscores the strategic importance of knowing others' types, because you reduce the number of possible environments, and thus the uncertainty. For example, if firm A's managers know, with certainty, that B's managers are worried about B's bond ratings, then they know the exact environment in which the game is being played.

246

Κ. Weigelt/I. MacMillan An Interactive

Common knowledge and joint probabilities We have shown that because of the interactive nature of payoffs, managers, in choosing a strategic alternative, must consider the future actions of competitors. In incomplete information situations, what a manager thinks competitors will do is dependent on what he thinks they believe. For instance, in our previous example, the managers at firm X were trying to choose a new pricing strategy. Their choice would certainly be influenced by whether or not they thought Y's cost structure was higher than theirs. Conversely, if Y's managers thought that X's managers believed that Y's cost structure was higher, then the beliefs of Y's managers about what actions X will take would certainly be different than if they thought that X's managers believed the opposite. Therefore, a manager must consider the first-order beliefs of competitors: what do competitors think my true type is? Strategically, however, we cannot stop at firstorder beliefs, but must consider higher-order beliefs (i.e. what do I think others think my type is, what do others think I think others think my type is . . .). In lieu of some ad hoc stopping point, the regression is infinite in scope. What we would like is a device that cuts the regression chain after its first link. After all, it is difficult enough to think about first-order beliefs. Common knowledge is the device we use to stop the infinite regression of expectations. Common knowledge exists when two necessary conditions are met: one, each player observes some event occurring; two, each player realizes that all other players observed the event. 2 An event that is common knowledge is public in the fullest sense—not only does each player observe it, but each player also recognizes that all others observed it. If first-order beliefs are common knowledge then all players know the first-order beliefs of others, and each player realizes that others know this. Therefore, we do not need to specify a probability distribution for a player over what others are thinking. Common knowledge does not mean that all

Strategic Analysis

31

players have the same beliefs. It does signify that at one point in the past all players shared the same information space,- that is all relevant information was common knowledge. Since that time, players may have acquired some private information. The individual beliefs of players will differ because in forming beliefs players use their private information. That is, beliefs are conditional on the player's private information. The framework will assume that there is an underlying probability distribution about the types of firm A and Β that is common knowledge. Our game is played in one of four environments. We want to capture the beliefs of managers about which environment they think the game is actually being played in. Note that no 'correct' ex ante beliefs exist, in the sense that one belief is right and other beliefs are wrong. The key point is that different beliefs can influence which strategic alternative a manager chooses. This is the relationship that we need the framework to capture. Beliefs about each environment are captured by assessing the subjective probability distributions of managers. Since we are trying to intuitively explain the basic framework, we will not address the assessment procedure here. For those interested readers, a companion paper (Weigelt, 1985) does illustrate, in detail, various assessment procedures. Table 1 illustrates possible beliefs about our strategic situation. The numbers represent the joint subjective beliefs that a specific environment is the 'correct' one. For example, 0.35 represents the joint beliefs that firm Β is type 1 (managers are concerned with the possible downgrading of its bond rating), and firm A is type 1 (managers are not greedy). Since we must consider all possible type combinations, the probabilities always add to 1. Marginal probabilities indicate the beliefs that a firm is a certain type. For example, the beliefs that firm A is type 1 (not greedy) is 0.35 +0.25 or 0.60. We will come back and use this table when calculating our expected unconditional payoffs. Table 1. Joint probabilities

2

Formally, we represent common knowledge as (Aumann, 1976; Milgrom, 1981): Lei (Ω,ρ) be a finite probability space with ρ and ζ as partitions representing the information of two players. Let R be the meet of Ρ and Ζ (the finest common coarsening of two partitions). Then define common knowledge as an event A at w (νν6Ω), if R(w)CA.

Framework

b1 b2

a1

a2

0.35 0.25 0.60

0.15 0.25 0.40

0.50 0.50

Interactive Strategie Analysis Framework

32

Κ. Weigelt and I.

MacMillan

Feasible strategy set We define a feasible strategy set as those strategies that are available to players. A strategy is generally not included in the set if players do not realize it exists, or because its expected payoff is too low. The set is commonly known to players. Again, there is no 'correct' set. If managers believe that a competitor could use a strategy it should be included in the set. We elicited the beliefs of firm A's managers to determine set members. The managers discussed various strategies rejecting many because of market structure, the existence of market norms, or the past history of firm B. We summarized their beliefs into three possible choices for each firm. Firm A could choose from the following: retain their present policy of a credit for 60 days; extend credit for 90 days; or extend credit for 120 days. Firm Β could also choose between three strategies: remain with their present credit terms of 60 days; match any credit policy changes of firm A, or reduce their prices by 10 percent. 3 Payoff matrices A firm's payoff from implementing any given strategy is dependent on the actions of competitors, and the firm's true type. The framework should explicitly recognize this, and show how payoffs change when competitors use different actions, or when the firm's type changes. We assume that, while the managers do not know the true type of the opposing firm, the mathematical form of the payoff function is commonly known. That is, if the firm's managers know, with certainty, the true type of the other firm, then for any given strategy combination they could calculate both their own payoffs, and the payoffs of the other firm. In our previous example this would mean that if the managers at firm X know firm Y's cost structure, then they can calculate firm Y's profit conditional on any pricing strategy that Y selects. The procedure outlined in the Appendix was used to assess the beliefs of firm A's managers about payoff functions (again, for those interested 3

247

The managers believed that this was the most drastic action that firm Β would take. A 'market norm 1 had been established, and a price cut of this amount might set off a market price war. They thought firm Β might do this to discipline firm A.

readers the assessment procedure is given in detail in Weigelt, 1985). Note how during the assessment procedure we form an explicit link between payoff functions and types. For example, if the managers at firm Β are actually worried about their bond rating, then firm A's managers believe that they will be more concerned with cash flow than RONI. Conversely, if they are not worried about their bond rating, then they will be more concerned about RONI than cash flow. After assessing the beliefs of firm A's managers, we asked them to consider scenarios using various strategy combinations. Table 2 shows these scenarios. Although the managers knew the financial data of their own firm, they were not sure of B's costs. In a strict technical sense we should have enlarged the type space to include this information. However, because the two firms had been rivals for so long, and the product they were producing was fairly standardized, the managers of firm A were highly confident in their estimates of B's costs (see the model usage section for further comments). An examination of Table 2 reveals the logic underlying the scenarios. For example, if firm A extends their credit terms to 90 days, and firm Β matches this extension, then we would not expect the market share of either firm to change, although we would expect a negative cash flow effect and a negative effect on RONI. In column 2 of Table 2 we see that this is what occurs. If firm A extends their credit, and firm Β do not match, then we would expect firm A to gain market share, improve RONI, and show a decrease in cash flow. Column 3 shows this is what firm A's managers believe will happen. Constructing such a table makes managers systematically think about the data, and consider how their beliefs change as the scenario changes. Though managers find such an exercise useful, they also find it difficult because it makes them think about what they really believe and why they believe it. Based on this assessment procedure, we can represent the payoffs for each possible environment in Tables 3-6. As explained in the Appendix, payoffs were normalized by labelling a firm's worst possible outcome as 0, and best possible outcome as 1. Firm B's payoff is represented by the first number in each cell, firm A's payoff is the second number. Note how the tables explicitly

248

Κ. Weigelt/I. MacMillan An Interactive Strategic Analysis Framework

33

Table 2. Competitive situation—Firm A

Current situation

Col. 2 We offer 90 days; Β matches

Price Credit days Invoice days Total market units Market share Revenues COGS Gross profit SG&A EBIT and depreciation

$10.00 60 60 1000 15.0% $1,500 $1,125 $375 $250 $125

$10.00 90 60 1000 15.0% $1,500 $1,125 $375 $250 $125

$10.00 90 60 1000 16.5% $1,650 $1,238 $413 $250 $163

$10.00 120 60 1000 15.0% $1,500 $1,125 $375 $250 $125

$10.00 120 60 1000 18.0% $1,800 $1,350 $450 $250 $200

$9.00 60 60 1050 15.0% $1,418 $1,181 $237 $250 ($13)

Receivables Inventory Fixed assets Payables Investment Cashflow effect RONI

$247 $185 $900 $185 $1,147

$407 $203 $900 $203 $1,307 ($123) 12.4%

$493 $185 $900 $185 $1,393 ($247) 9.0%

$592 $222 $900 $222 $1,492 ($270) 13.4%

$233 $194 $900 $194 $1,133 $00

10.9%

$370 $185 $900 $185 $1,270 ($123) 9.8%

Current situation

A offers 90 days; Β matches

A offers A offers 90 days; 120 days; no match Β matches

A offers 120 days; no match

Β cuts price

$10.00 60 60 1000 42.0% $4,200 $2,730 $1,470 $600 $870 $690 $449 $2,000 $449 $2,690 ($56) 32.3%

$9.00 60 60 1050 45.0% $4,253 $3,071 $1,182 $600 $582 $699 $509 $2,000 $509 $2,699 ($41) 21.6%

Col. 1

Col. 3 Col. 4 We offer We offer 90 days; 120 days; no match Β matches

Col. 5 We offer 120 days; no match

Col. 6 Β cuts price

—

Competitive situation—Firm Β

Price Credit days Invoice days Total market units Market share Revenues COGS Gross profit SG&A EBIT and depreciation Receivables Inventory Fixed assets Payables Investment Cashflow effect RONI

$10.00 60 60 1000 45.0% $4,500 $2,925 $1,575 $600 $975 $740 $481 $2,000 $481 $2,740 35.6%

show the interactive nature of payoffs: payoffs are conditional on type and the actions of others. Look at Table 3: this represents the payoffs for both firms in an environment where firm A is actually not greedy and firm Β is concerned about the downgrading of its bond ratings (i.e. if a ' , b ' ) . If firm A keeps its credit policy at 60 days, and firm Β cuts its price, then A suffers its worst possible consequence (represented as

$10.00 90 60 1000 45.0% $4,500 $2,925 $1,575 $600 $975 $1,110 $481 $2,000 $481 $3,110 ($370) 31.4%

$10.00 60 60 1000 43.5% $4,350 $2,828 $1,523 $600 $923 $715 $465 $2,000 $465 $2,715 ($28) 34.0%

$10.00 120 60 1000 45.0% $4,500 $2,925 $1,575 $600 $975 $1,479 $481 $2,000 $481 $3,479 ($740) 28.0%

0.000). Presumably, this occurs because when firm A cuts its price, many of firm B's customers defect to firm A. Unconditional payoff matrix Tables 3 - 6 represent the payoffs for each possible environment; that is they represent the payoffs to firms conditional on the true types of each

Interactive Strategic Analysis Framework 34

K. Weigelt and I.

249

MacMillan

Table 3. If a'.b 1 Firm A

Firm Β Status Match Price cut

60 days

90 days

120 days

1.000, 0.424 1.000, 0.424 0.975 , 0.000

0.923, 0.514 0.543, 0.348 0.874, 0.232

0.801, 0.409 0.102, 0.242 0.864, 0.183

Table 4. If a',b 2 Firm A

Firm Β Status Match Price cut

60 days

90 days

120 days

1.000, 0.424 1.000, 0.424 0.943, 0.000

0.795, 0.514 0.674, 0.348 0.714, 0.232

0.527, 0.409 0.409, 0.242 0.691, 0.183

T a b l e S . If a 2 ,b' Firm A

Firm Β Status Match Price cut

60 days

90 days

120 days

1.000, 0.424 1.000, 0.424 0.975, 0.000

0.803, 1.00 0.543, 0.348 0.826, 0.292

0.643, 0.923 0.102, 0.186 0.789, 0.158

Table 6. If a 2 ,b 2 Firm A

Firm Β Status Match Price cut

60 days

90 days

120 days

1.000, 0.424 1.000, 0.424 0.943, 0.00

0.534, 1.00 0.674, 0.348 0.603, 0.292

0.161, 0.923 0.409, 0.186 0.514, 0.158

firm. However, managers do not know, with certainty, the true type of competitors, and thus which environment they are playing in. They do have beliefs about which environment they are playing in (Table 1), and since these beliefs will affect their strategic choice, we need to link beliefs and payoffs. We do this by deriving an unconditional payoff matrix from Table 1 and Tables 3-6. This unconditional payoff matrix is represented in Table 7.

Because a firm's type can affect its strategic choice and payoff, we need to explicitly represent a firm's possible types, the strategies it can use, and its payoff: this payoff is conditional on its true type, the strategy it selected, and the actions of competitors. Thus, represent firm A's strategies by y'' (i,j = 1,2,3), and firm B's strategies by z'1 (i,j = 1,2,3). For example, y2' means that firm A uses strategy 2 (90 days credit) if its true type is 1 (not greedy), and strategy 1 (60 days credit)

250

Κ. WeigeM. MacMillan An Interactive

Strategic Analysis

Framework

35

Table 7. U n c o n d i l i o n a l n o r m a l i z e d e x p e c t e d p a y o f f s

y"

y"

y2'

y-2

y2>

0.78 0.71

0.66 0.68

0.81 0 . 4 2

0.67 0 . 6 5

0 . 5 5 0.61

z"

1 0.79 0.50 1 0.69 0 . 4 5

0 . 7 8 0.37

0.67 0.44

0 . 5 8 0.39

* 12

y"

z"

1.00 0 . 4 2

0.85 0.65

0.74 0 . 6 2

0.92 0.48

z 12

1.00 0.42

0.89 0.49

0.80 0.44

0.89 0.44

2 13

0.98 0.21

0.86 0.37

0.81 0.33

0.90 0.30

0.77 0.46

0 . 7 3 0.42

0.85 0 . 2 5

0.72 0.41

0.68 0 . 3 7 z "

z 21

1.00 0 . 4 2

0.81 0 . 5 6

0.66 0.51

0 . 7 2 0.41

0.60 0.55

0 . 4 4 0.51

0.57 0 . 3 6

0.38 0 . 4 9

0.22 0.45 z2'

z22

1.00 0 . 4 2

0.85 0.45

0.72 0 . 3 3

0.76 0.38

0.61 0 . 3 5

0 . 4 8 0.28

0 . 5 4 0.31

0.39 0 . 2 8

0.26 0.22 Z "

23

0 . 9 8 0.21

0.82 0 . 2 7

0 . 7 3 0.22

0.76 0.24

0 . 6 0 0.31

0.51 0.25

0.60 0.19

0.44 0 . 2 6

0.35 0.20 2 - '

31

0.99 0.21

0.85 0.40

0.75 0.36

0.90 0.32

0.76 0.50

0 . 6 6 0.46

0 . 8 3 0.27

0.69 0.46

0 . 5 9 0.42 2·"

32

0 . 9 9 0.21

0.88 0 . 2 4

0.81 0 . 1 8

0.87 0 . 2 7

0.77 0 . 3 0

0 . 7 0 0.24

0.80 0.23

0.70 0 . 2 6

0 . 6 3 0 . 1 9 Z>2

z33

0.97 0.00

0.85 0.12

0.82 0.06

0.87 0.14

0 . 7 6 0.26

0.73 0.20

0.87 0.11

0.75 0 . 2 3

0 . 7 2 0.17 z "

2

Z z

if its true type is 2 (greedy), z " means that no matter what firm B's true type is, it will use strategy 1 (status quo). Cell entries represent the expected payoff for firms A and Β with respect to alternative strategy pairs. That is, since a manager does not know which environment the game is being played in, he takes the expected value over all environments. This value is merely his payoff in any environment multiplied by his beliefs that the game is being played in that environment. Thus, for strategy pair (z",y'2), firm B's expected normalized payoff is 0.85, while firm A's expected normalized payoff is 0.65. These payoffs are derived in the following way:

actions. Since one goal of our framework is to identify optimal actions, we want to specify criteria for discriminating between payoffs. We choose a solution concept based on the premises that players act according to their incentives; and in interactive environments they condition their actions on their expectations of others' actions. The solution concept is called the Nash equilibrium. It represents the most important concept for bridging the chasm between simply describing a situation and predicting the actions of players. The Nash equilibrium captures the essence of both economic theory and interactive decisionmaking. A Nash equilibrium is composed of strategies such that no player has an incentive to change his strategy given he believes that all others will choose their best corresponding Firm Β = 1.00(0.35)+0.795(0.25)+1.00(0.15) strategy. It is a solution of mutual best response: +0.534(0.25) = 0.85 Firm A = 0.424(0.35)+0.514(0.25)+0.424(0.15) my choice is optimal given that others choose their optimal choices. Solutions are stable since + 1.00(0.25) = 0.65 no player can improve his payoff by unilaterally Thus, these payoffs are simply the belief that a changing his strategy choice. specified environment is the true one (Table 1), Look at Table 7: the joint strategy (z 12 ,y 22 ) times the payoff from that environment (Tables is both a Nash equilibrium and a Bayesian equilibrium. 4 Look across row z 12 : even if the 3-6). managers at firm A know that firm B's managers will choose strategy z 12 , they will choose y 22 , because this will give them the highest expected FINDING EQUILIBRIUM SOLUTIONS payoff. Look down column y22: even if managers Since there were four possible environments and three strategy choices for each firm, Table 7 4 H a r s a n y i shows ( t h e o r e m I) t h a t this p o i n t is b o t h sufficient shows the expected payoffs for 81 joint strategic and necessary in t h e N a s h e q u i l i b r i u m s e n s e (1968: 321).

Interactive Strategie Analysis Framework 36

Κ. Weigelt and I.

251

MacMillan

at firm Β know that firm A's managers will choose strategy y22, they will use strategy z 12 , because this gives them the highest expected payoff. If either player deviated from their Nash strategies, their payoffs would decrease. Given both the competitive attributes and beliefs of firms A and Β the joint strategy point (zl2,y22) prescribes the optimal strategy for the firms. What this says is the following: if firm Β is truly concerned about a possible downgrade in its bond rating (type 1) then it will not counteract the actions of firm A. However if firm Β is not concerned about a possible downgrade (type 2), then it will match any action of firm A. On the other hand, regardless of whether firm A is greedy or not, it should extend its credit terms from 60 to 90 days.

players believe that there is a greater probability that firm A is not greedy, and that firm Β actually is concerned about the downgrading of its credit, firm Β should not change its present policy, no matter what its true type is. In contrast, let Table 9 represent the joint beliefs of firms A and B. The Nash equilibrium now shifts to (z 32 ,y 22 ) because the probability of firm A being greedy increased from 0.40 to 0.85. Intuitively, this means that firm Β is less willing to sacrifice any market share to firm A, even if it is worried about its credit rating. The framework now recommends that firm Β should cut its price if it is concerned about its credit rating. If it is not concerned about its rating, it should match any move firm A makes. Table 9. Joint probabilities

The effect of beliefs Beliefs about a competitor's true type can affect which strategy managers select. A viable framework should reflect revised beliefs in a manner that is congruent with underlying causal relationships. This is important, because if a manager does not intuitively understand a recommended change in strategy, then maybe he has not realized all the relevant relationships. We test how well the framework reflects changes in beliefs with a series of simulations. We altered the joint probabilities assigned to the different environments (Table 1), thus simulating revised beliefs about players' types. For example, if we change the probabilities in Table 1 to those in Table 8, these changes in beliefs will be reflected in the new unconditional payoff matrix. In fact this change in beliefs causes the Nash equilibrium to change from (z l2 ,y 22 ) to (zn,y22). The shift is caused by the increased marginal probability (0.60 to 0.85) that firm A is not greedy (a1), and that firm Β is worried about the downgrading of its credit (the probability of b1 shifts from 0.35+0.15 or 0.50 to 0.50+0.10 or 0.60). In terms of strategic actions what does this mean? While the optimal strategy for firm A remains to extend its credit to 90 days, the Table 8. Joint probabilities

b' b2

b1 b2

a1

a2

0.10 0.05

0.65 0.20

Note the stability of firm A's optimal strategy. Even though we changed beliefs, A's optimal strategy remains the same (extend credit terms for 90 days). In one sense, revising beliefs can act as a form of sensitivity analysis to test the stability of an equilibrium solution. One might place more confidence in a strategy that remains optimal even when beliefs undergo radical changes. On the other hand, if equilibrium solutions change with a small change in beliefs (say 0.05), then managers should consider gathering more information and re-assessing beliefs. Postscript The managers of Firm A did extent credit for 90 days, and for a period of a year secured and maintained a gain in market share of nearly 2 percentage points. By the time firm Β had recovered from its economic ills, firm A had moved to another strategy which focused on a new packaging technology. MODEL USAGE AND LIMITATIONS

a'

a2

0.50 0.35

0.10 0.05

Benefits Though the framework can identify optimal actions, we view its primary purpose as providing

252

Κ. WeigeM. MacMillan

An Interactive Strategic Analysis Framework a systematic way for managers to evaluate different courses of action in a manner that is consistent with their basic beliefs and preferences. We do not claim that firm A's managers would not have chosen the same strategy without using the framework. However, we believe that using the framework helps managers construct models in a systematic manner, which forces them to answer key questions, such as: What are the important factors and why? What information do players have? What are the payoffs of players? In answering such questions managers must think carefully about the relevant causal model; identifying key environmental factors, and evaluating their position relative to others. Similarly, the task of quantifying beliefs (i.e. gut feelings) causes them to focus on their supporting data, and sometimes shows inconsistencies in their judgement. Additional benefits include the following: 1. The task is helpful in a dynamic sense; it increases managers' sensitivity to changes in relevant factors, which can result in quicker responses to the actions of others, or changes in environmental factors. This not only promotes a proactive attitude, but the framework provides consistency for re-evaluating a situation in a multi-period setting. 2. Better communication between decision-makers is promoted. Managers have a common ground for evaluating different rationales, and showing what information supports their position. If the sensitivity analysis shows managers that their different beliefs do not change the optimal solution, then managers will avoid needless arguments. 3. Managers can quickly evaluate new alternative strategies. 4. Opinions of experts can be systematically incorporated into the analysis. The framework also gives experts a better understanding of the relevant causal model, and how their opinions will interact with other variables. The question then becomes—is it worth the time and effort? We think that since hundreds of thousands of dollars (if not millions) are involved in many business-level strategic decisions, the relative cost of systematically thinking about alternatives is rather small. Our framework does not require managers to radically change their thought processes; it just requires that they think

37

a little harder about the situation and their beliefs. In today's business environment firms are required to 'bet' large amounts of money on their strategic plans. Our framework will not eliminate bad bets, but it will certainly reduce their frequency, and improve firm performance. Limitations Our framework cannot be universally applied; correct use requires an understanding of the strategic situation, and model limitations. 5 We want to review some of the model's major limitations, and discuss them within the context of business-level strategic planning. First, the number of relevant players should be small. The greater the number of players, the less mutually dependent are payoffs. Therefore, the framework is best suited for markets with few players. This is true not only for theoretical reasons, but also for model tractability purposes. While the model can accommodate η players, the time and effort required for calculating an equilibrium solution gets greater as η increases. This limitation is not as constraining as it first seems, since modelling techniques can help reduce the number of players. For example, under certain conditions it is possible to bundle players into groups (i.e. strategic groups). Or a market may have a few dominant players, and many 'fringe' players. A major assumption of the framework is that certain parameters are common knowledge, especially the subjective probability distributions over possible types. Therefore one should apply the framework only to those situations where the common knowledge structure is large enough to support the notion that relevant private information can be summarized with a small set of types. Business strategic situations do have characteristics that are conducive to this requirement. First, players generally play for long periods against a stable set of rivals. Not only does this long learning period increase the probability that beliefs will converge (assuming equal likelihood functions), but players get to know more about each other and usually manage s

We wanted to intuitively show how one might apply the model to a business-level planning decision. In so doing we had to neglect some of the subtle areas of model construction. Interested readers are referred to articles that more fully explain the model's basic technical properties (see Harsanyi, 1967-68; Myerson, 1983 and Weigelt, 1985).

Interactive Strategie Analysis Framework 38

Κ. Weigelt and 1.

253

MacMillan

to establish 'market norms' (e.g. price levels, standardized systems and accounting practices). Experimental evidence has shown that as an environment gets more familiar, subjects tend to act more rationally. For example, Camerer and Weigelt (forthcoming) report experiments where subjects were able to reach equilibriums vastly more complex than the Nash. Second, market firms generally use the same data sources. Weigelt (1985) found that the three firms in the market received their market data from the same marketing firm, and all knew (and knew the others knew) that a firm would buy the product of others and have its engineers dismantle the product to estimate product cost. Results do show that very small changes in player beliefs can alter the nature of the equilibrium (see Milgrom and Roberts, 1982; Kreps and Wilson, 1982). The framework does require management to document beliefs and preferences, and commit themselves to specific strategies. For various reasons (e.g. the intra-firm political situation), management may prefer to operate within a 'fuzzy' setting, and retain strategic flexibility. The use of formal decision analysis The process of thinking strategically is difficult: gathering good information is difficult, synthesizing it is more difficult, and choosing a good strategic response is very difficult. In the early 1960s a group of Harvard Business School researchers helped develop a framework to help a manager organize and systematically evaluate information with respect to beliefs and preferences. The researchers called their framework decision analysis (see Pratt, Raiffa and Schlaifer, 1965; Raiffa, 1970). Our framework modestly extends their work not only by helping managers evaluate information with respect to beliefs and preferences, but also by recognizing that consequences are interactive in nature and evaluating them on that basis. While most firms initially ignored formal decision analysis, the number of firms employing the technique has grown in recent years. We expect this trend to continue, primarily because of the increasing availability of personal computers. Recent articles in popular business journals illustrate how the use of personal computers is making it easier for managers to use formal

decision techniques. In a recent Fortune, Kupfer (1987) reports how firms are increasingly using expert systems to support the managerial decisionmaking process. He cites one example where a decision-maker just did not think he could quantitatively express his beliefs: Take the case of an expert cheesemaker. Knowledge engineers from Teknowledge of Palo Alto, followed him around for days trying to figure out how he decided when a batch of cheese was ready. Ί just feel it', he said. No one else could detect the difference between a ripe batch and a raw one—until someone got the idea of clogging the expert's nose and discovered that he smelled when the cheese was done (p. 78).

The engineers then built their expert system around the smell of cheese. In another recent Fortune article, O'Reilly (1987) reports how planners at Atlantic Richfield use personal computers to produce bar graphs which help geologists quantify their beliefs. The geologists change the size of the bars until they feel the bar matches their intuitive feelings. We think that such advances in computer software and hardware will increase the use of formal decision analysis. Not only do computers help planners quantify their beliefs, they also greatly reduce computational time. Additionally, computers encourage sensitivity analysis, since it becomes relatively easy to evaluate how changes in beliefs or preferences affect payoffs.

CONCLUSIONS Avenues for future research The simple incomplete information model we used is the first step in building a dynamic competitive framework. Though the framework captured some dynamic aspects of the strategic situation, it was not a true dynamic framework. Given the time-dependent nature of actions, a truly dynamic framework must consider future payoffs, and their effect on players' expectations. For example, a dynamic model might have considered that when firm A showed they were not greedy (by only taking a small amount of market share), they were sending a signal to firm Β that 'not greedy' was their true type. This

254

Κ. Weigelt/I. MacMillan An Interactive Strategic Analysis Framework

signal should have impacted the beliefs of firm B, and thus their actions in future periods. We note that incomplete information models are found in all branches of the social sciences. Within business disciplines the models have recently been used to look at signalling between firms, quality levels in products, corporate culture, auditing practices, entry deterrence, the underwriting of new stock issues, and debt repayment (see Weigelt and Camerer, forthcoming). However, incomplete information models use relatively new technology, and much work, both theoretical and empirical, remains. For example, one problem with multi-period models is that the Nash solution sometimes indicates a wealth of equilibria. Theorists are presently working on more refined equilibrium concepts that formalize intuitive 'reasonable' notions for eliminating many of the equilibria.

CONCLUSIONS Most business-level strategic planning consists of situations where payoffs are mutually dependent and interactive, and where managers must form beliefs because they lack complete information. Given this environment, a framework that explicitly recognizes these characteristics can help both researchers and managers better understand strategic situations. The framework's mathematical discipline enables managers to organize their beliefs and helps them gain a better understanding of the strategic situation. And it does so with a sparsity of exogenous variables so managers can explicitly see how and why changes in assumptions alter payoffs. We hope our message has been strong enough—one does not use the framework by 'plugging-in' numbers and 'cranking' out optimal solutions. The complexity of the environment (e.g. incomplete information, interactive payoffs), requires any analytic model to be balanced on a set of assumptions. Fortunately, managers using personal computers can easily simulate changes in beliefs and preferences, and evaluate their resulting impact on payoffs. The framework's purpose is to help analyze a strategic situation, use it as a means of achieving that goal, not as the goal itself. Finally, using a rigorous framework does not imply that we believe the business environment

39

is totally deterministic. Random occurrences will always be part of the environment. Managers using this framework will make bad 'bets', as they will using any framework. However, we believe the framework's ability to generate and evaluate many alternative strategies and scenarios will allow managers to move quickly when the unexpected occurs; both in recognizing changes and adapting to it. As Robert Waterman (1987) said in a recent Fortune article: T h e main strategic weapon of those who succeed in the f u t u r e will he information. Their main strategic advantage will be the ability to move fast (p. 182).

APPENDIX: P A Y O F F MATRICES Determining payoff functions is a multi-step task. For our purposes it can be divided into the following steps: 1. attributes, 2. relevant consequence range, 3. determination of scalars. Attributes Payoff functions are mathematical representations of preference orderings. If decisions are governed by a mechanism that selects the highest expected payoff of any available—then payoff functions can be used to select strategic choices. Con I ra ry to widespread belief, payoff functions are rarely single-attribute (i.e. the omnipresent profit-maximizing firm), but consist of multi-attributes. Beliefs about which attribute should be included in the payoff function are assessed using various techniques, and then validated with supporting evidence (e.g. economic theory, market behavior, express goals, etc.). Since firm Β was concerned about its bond rating, primary attributes in its payoff function were expected loss in cash flow, and return on net investment. Firm A's managers defined their primary attributes as being return on net investment and market share gain. Relevant consequence space The primary reason for selecting a relevant consequence space is that preference orderings

Interactive Strategic Analysis Framework

40

Κ. Weigelt

and I.

255

MacMillan

are not equal over the entire consequence space. Therefore, you elicit preferences for the relevant (between the best and worst possible outcomes) consequence space. In so doing you also can compare interjacent outcomes through normalization (the worst outcome is labelled 0, and the best outcome is labelled 1).

Determination of scalers A two attribute payoff function can take the following form: u(x,y) = atu, + a2u2 The coefficients a„ are called the scaling constants of the payoff function. The scalars represent a pseudo-preference ordering indicating that if a strategist has to choose between increasing χ from its worst outcome to its best outcome, or increasing y in a similar fashion, that attribute y would be selected over attribute χ (Keeney and Raiffa, 1976). Weightings are derived by assessing the beliefs of managers. A related procedure is the assessment of the payoff's functional form. Since it was found that the attributes were mutually utility-dependent, an additive equation was used (Keeney and Raiffa, 1976). Scalars can be used to represent player types. For example, we represent a firm Β that is not worried about its bond rating to have a utility function of: u{b2) = 0.8 R O N I + 0.2 cash flow W e assume that a firm Β that is worried about its bond rating has a payoff function of: «(6 1 ) = 0.2 R O N I + 0.8 cash flow

ACKNOWLEDGEMENTS Thanks to Colin Camerer, Cynthia Montgomery, Ruth Raubitschek, two anonymous referees, participants at the 1987 Academy of Management Meetings, and especially Marlene Frisella.

REFERENCES Aumann, Robert. 'Agreeing to disagree', Annals of Statistics, 4, 1976, pp. 1236-1239. Camerer, Colin and Keith Weigelt. 'Experimental test of a sequential equilibrium reputation model', Econometrica (forthcoming). Ghemawat, Pankaj. 'Sustainable advantage', Harvard Business Review, 64, September-October 1986, pp. 53-58. Harsanyi, John. 'Games with incomplete information played by Bayesian players', Management Science, 14, 1967-68, pp. 159-189, 320-334, 486-502. Kahneman, Daniel, Paul Slovic, and Amos Tversky. Judgment Under Uncertainty: Heuristics and Biases, Cambridge University Press, New York, 1982. Keeney, Ralph and Howard Raiffa. Decisions with Multiple Objectives: Preferences and Value Tradeo f f s , John Wiley and Sons, New York, 1976. Kreps, David and Robert Wilson. 'Reputation and imperfect information', Journal of Economic Theory, 27, 1982, pp. 863-894. Kupfer, Andrew. 'Live experts on a floppy disk', Fortune, 116, 12 October 1987, pp. 69-77. Mertens, J. and S. Zamir. 'Formalization of Harsanyi's notion of "type" and "consistency" in games with incomplete information'. CORE discussion paper, Université Catholique de Louvain, 1983. Milgrom, Paul. 'An axiomatic characterization of common knowledge', Econometrica, 49, 1981, pp. 219-222. Milgrom, Paul and John Roberts. 'Prédation, reputation, and entry deterrence', Journal of Economic Theory, 27, 1982, pp. 280-312. Myerson, Roger. 'Bayesian equilibrium and incentivecompatibility: an introduction'. Northwestern University Center for Mathematical Studies Working Paper No. 548, 1983. O'Reilly, Brian. 'Apple finally invades the office', Fortune, 116, 9 November 1987, pp. 69-77. Porter, Michael. Competitive Strategy, Free Press, New York, 1980. Pratt, John, Howard Raiffa and Robert Schlaifer. Introduction to Statistical Decision Theory, McGrawHill, New York, 1965. Raiffa, Howard. Decision Analysis, Addison-Wesley, Menlo Park CA, 1970. Waterman, Robert. 'Strategy in a more volatile world', Fortune, 116, 21 December 1987, pp. 181-182. Weigelt, Keith. 'An empirical application of Harsanyi's incomplete information game model'. Unpublished doctoral dissertation, Northwestern University, 1985. Weigelt, Keith and Colin Camerer. 'Reputations and corporate strategy: a review of recent theory and applications', Strategic Management Journal (forthcoming).

256 e

E. J. Zajac/M. H. Bazerman

Academy oí Management 1991, Vol. 16. No. 1, 3 7 - 5 6

Review

BLIND SPOTS IN INDUSTRY AND COMPETITOR ANALYSIS: IMPLICATIONS OF INTERFIRM (MISjPERCEPTIONS FOR STRATEGIC DECISIONS EDWARD J. ZAJAC MAX H. BAZERMAN Northwestern University This article bridges the literatures on competitor analysis and strategic decision making by (1) introducing the notion oi competitive decision making into the strategic decision-making literature and (2) embedding this notion into a framework oi industry and competitor analysis. The article shows that decision makers typically have specific "blind spots" when they consider the contingent decisions of competitors. The article identifies these blind spots and discusses how they may explain persistent, commonly observed phenomena such a s industry overcapacity, new business entry failures, and acquisition premiums.

Two major areas of strategy research are industry and competitor analysis (e.g., Porter, 1980) and strategic decision making (e.g., Fredrickson, 1984; Mintzberg, Raisinghani, & Théorêt, 1976; Schwenk, 1984). The former—often studied from an economic perspective—is seen as dealing with decision outcomes, whereas the latter—often studied from a behavioral perspective—is seen as dealing with decision making. Clearly, decision making and decision outcomes are not independent topics, yet there is a general lack of conceptual integration of these areas. Our ability to analyze, understand, and predict firms' strategic behaviors may be unnecessarily limited by the fact that the literature on industry and competitor analysis has been largely silent on decision-making processes, whereas the literature on strategic decision making has typically neglected competitive decision-making dynamics. This article links these literatures by ( 1 ) introducing the notion of competitive decision making into the strategic decision-making literature and (2) embedding this decision-making perspective into a framework of industry and competitor analysis. The article shows that, in competitive situations,

The authors benefitted from comments made by Gregory Dess, Moshe Farjoun, Brian Golden, Stephen Shortell, and Howard Thomas. Support from Northwestern University's Dispute Resolution Research Center is gratefully acknowledged. 37

Blind Spots in Industry Analysis 38

Academy

oí Management

Review

257

January

strategic decision makers typically do not sufficiently consider the decisions of competitive others and that this deficiency leads to a variety of specific judgmental mistakes, or "blind spots." This article is concerned with identifying and analyzing those blind spots that are specifically created and exacerbated by competitive decision making and, thus, are likely to be relevant in competitive situations. Our objectives are therefore somewhat different from those of Schwenk (1984), Barnes (1984), Bateman and Zeithaml (1989) and others, whose work focuses on applying cognitive/psychological phenomena to the strategic decisionmaking area. These authors do not emphasize competitive decision making or attempt to link the strategic decision-making literature with the industry and competitor analysis literature. This article's focus on competitive decision making permits an analysis of the implications of competitive blind spots in the context of two major strategic decisions relating to industry and competitor analysis: the capacity expansion decision and the new business entry decision (through internal development and through acquisition). The analysis shows how competitive blind spots can explain commonly observed phenomena such as industry overcapacity, new business entry failures, and acquisition premiums. The article concludes with a discussion on the value of a revised framework of industry and competitor analysis that relies not only on structural factors to explain firm and market conduct, but also on cognitive/perceptual factors. Such a revised framework would help researchers to more explicitly consider the issue of interfirm (mis)perceptions and incorporate a more realistic competitive decision-making perspective. COMPETITIVE DECISION MAKING Competitive decision making is defined here as strategic decision making that requires the focal actor to consider the contingent decisions of competitive actors (the term contingent is used to emphasize that the competitor's decision is a function of the focal actor's decision). The major research traditions in strategic decision making (cf. Fredrickson, 1984; Mintzberg, Raisinghani, & Théorêt, 1976; Schwenk, 1984) have made important contributions to our understanding of a number of decision-making issues. However, they have not sought to address those decision-making issues arising specifically in competitive decision problems (i.e., those requiring a firm to consider the contingent actions of competitors). Such is the domain of competitive decision making. We view competitive decision making as an important, but neglected, dimension of strategic decisión making that can provide the necessary conceptual link betweeen the behavioral literature on decision making and the economics-based literature on industry and competitor analysis. A basis for discussing competitive decision making can be found in the game theory literature, which studies mathematical models of competitive situations between rational actors (Myerson, 1990; Shubik, 1984). Not sur-

258 1991

E. J. Zajac/M. H. Bazerman Zajac and Bazerman

39

prisingly, industrial organization economists (e.g., Spence, 1979; Salop, 1979; Tiróle, 1988) have increasingly taken a game theoretic approach to the study of strategic behavior of firms in oligopolies, where each firm's strategic decisions are interdependent with those of the other firms in the industry. In fact, the use of game theory is the distinguishing feature of the "new industrial organization" literature (Jacquemin, 1987; Tiróle, 1988). A major potential contribution of game theory to the strategic management area is the recognition that competitive actors need to fully consider the contingent decisions that other parties will be making. Game theoiy appears to offer the ideal perspective for understanding competitive environments and making recommendations to competitive actors. In fact, strategic management researchers (cf. Weigelt & MacMillan, 1988) are beginning to invoke the logic of game theory to suggest optimal competitive strategies (and, implicitly, prescriptions for top management). However, important limitations emerge from this perspective. Game theory specifies the solution that should be observed in competitive situations if all actors behaved in a completely rational manner. Thus, game theory prescribes how fully rational competitors should behave against other fully rational competitors. However, ample evidence exists that fully rational competitors simply do not exist (Kahneman, Slovic, & Tversky, 1982). Several authors have explicitly recognized some of the limitations of a game theoretic perspective. For example, Raiffa (1982), a n early contributor to game theory, recognized the importance of considering competitors' deviations from rationality when constructing models of competitive action. He argued against simply assuming rational behavior by a competitor, as suggested by the game theoretic solution and, instead, proposed the use of decision analytic techniques to prescribe the rational course of action for a competitive actor based on a realistic description of the contingent decisions of the opponent(s). Interestingly, the need to consider a more realistic competitive decisionmaking perspective is also implicit in Porter's (1980) prescriptive framework for industry and competitor analysis. One "crucial" component in his framework is the identification of each competitor's assumptions about itself and about the other companies in the industry. Porter (1980: 59) implies that these assumptions may be strongly influenced by biases or "blind spots, " defined as "areas where a competitor will either not see the significance of events at all, will perceive them incorrectly, or will perceive them very slowly." Knowing a competitor's blind spots, he argues, will help the firm to identify competitor weaknesses. If one seeks to make prescriptions, it would seem more appropriate to do so based on a realistic assessment of a competitor, rather than a n oversimplification created for the convenience of allowing mathematical analysis. However, Raiffa (1982) stopped short of identifying the nature of the deviations from rationality that affect a competitive actor (the same can be said about Porter's [1980] discussion of blind spots). This is a n important

Blind Spots in Industry Analysis 40

Academy

oí Management

Review

259

January

limitation. For example, if actors deviated from rationality, but the errors were randomly distributed around what would constitute a rational decision, then the game theoretic assumption of a rational opponent would be the best approximation available to the competitive actor. However, behavioral decision research suggests a number of systematic ways in which judgment deviates from rationality (Kahneman, Slovic, & Tversky, 1982; Kahneman & Tversky, 1979). It should also be noted that both Raiffa (1982) and Porter (1980), in their prescriptions, focus only on the competitive blind spots of the opponent and do not address the blind spots that affect the focal actor (as if blind spots would apply to competitor firms, but not the firm doing the competitor analysis). However, it also is necessary to address the competitive blind spots of the focal actor that may inhibit the utilization of prescriptive advice. Thus, the next section not only identifies specific blind spots emerging from competitive behavior, but also emphasizes that strategic decision makers often do not follow the prescriptive advice to consider the perspective of competitors. COMPETITIVE BLIND SPOTS To show that competitors tend to insufficiently consider the contingent decisions of their competitive others, this section draws mainly from behavioral decision research. Such research is very relefvant for the present discussion because (1) it examines competitively interdependent situations that parallel the game theoretic perspective used by economists to analyze the competitive behavior of firms in an industry and (2) it tests whether normative game theoretic predictions are, in fact, empirically observed in the actual decision-making processes of the participants in the competitive situation. In other words, it offers a useful point of departure from the classical game theoretic perspective on competitive decision making. The important conclusion emerging from this body of research is that despite the obvious importance of considering the strategy of a competitor, there is a fundamental observed impediment to rational decision making in actual competitive situations: namely, the failure of the competitive actor to sufficiently consider the contingent decisions of the opponent (cf. Bazerman & Carroll, 1987). The manifestations of this failure are discussed in this article in terms of competitive blind spots, such as the "winner's curse," nonrational escalation of commitment, overconfidence in judgment, and limited perspective and problem framing. Research on the winner's curse problem illustrates how competitive decision makers insufficiently consider the contingent decisions of others and, thus, make judgmental mistakes. This research has shown that under asymmetric information, competitive actors systematically fall prey to a winner's curse (i.e., they consistently, and voluntarily, enter into loss-making purchases). Samuelson and Bazerman (1985) described this nonintuitive result us-

260 1991

E. J. Zajac/M. H. Bazerman Ζajac and Bazerman

41

ing a n example of acquiring a company. In their example, potential acquirers know only that the target company is equally likely to be worth any value between $0 and $100 per share and that, whatever its value, it is worth 50 percent more to the buyer than to the seller. The seller knows the exact value and will accept any bid at or above that value. The study showed that most of the bids were in the $50-$75 range, and they were based on the following simple (naive) logic: namely, that while the exact value of the target firm is not known to the bidder, the expected value is $50 per share for the seller and $75 per share for the bidder. Thus, the bidder believes that he or she can make a reasonable profit by offering some price slightly in excess of $50 per share (e.g., $60 per share, or $51 per share). This logic would be rational if the target was also uninformed about the value of the firm. However, a n informed target will only accept offers if they are profitable, which an acquirer could assess through the appropriate conditional logic or the use of the following simple example. The bidder considers making an offer of $60 per share. If it is accepted, the firm must be worth between $0 and $60 per share. The average value of the firm to the target when the offer is accepted is $30 per share to the seller and $45 per share to the bidder. The bidder's profit has an expected value of $45 minus $60, or - $ 1 5 per share when the offer is accepted. Thus, when a n offer is accepted, the bidder will have acquired a company that is worth (on average) 25 percent less than what the bidder paid for it. It is obvious that the best decision is to make no offer at all (i.e., $0/share). However, fewer than 10 percent of the competitive actors make this decision, even though it requires minimal analytical skill. This tendency to insufficiently consider information that is available by considering the contingent decisions of others has also been observed in competitive situations with multiple bidders (i.e., the winner's curse in an auction context). In this case, winning bidders often find that they have overpaid for the acquired commodities (Capen, Clapp, & Campbell, 1971; Cassing & Douglas, 1980; Kagel & Levin, 1986; Roll, 1986). This occurs in the auction context because the highest bid is likely to be from an acquirer with one of the more optimistic estimates of the commodity's value. If this adverse selection problem is not accounted for by the bidders, winning bids will frequently result in negative returns. In fact, Bazerman and Samuelson (1983) showed that bidders typically fail to incorporate either the relevance of the adverse selection issue or factors that exacerbate the adverse selection problem, such as the number of bidders and commodity value uncertainty. One might argue that competitors will correct their judgments by learning from feedback regarding past decisions. For example, Kagel and Levin (1986: 917) viewed the winner's curse as a disequilibrium phenomenon and maintained that bidders will eventually learn to avoid the winner's curse after receiving sufficient feedback. Kagel and Levin (1986) did show a reduction in the winner's curse in an auction context as the market learns over time. Some learning occurs through the observation of consistent losses

Blind Spots in Industry Analysis 42

Academy

of Management

Review

261

January

being suffered by winners in the auction, and additional learning is through the disappearance of the most aggressive bidders from the market (note that the latter does not suggest individual competitor learning, but rather natural selection in a market context). Tversky and Kahneman (1986: 274-275) argued that basic judgmental biases are unlikely to be corrected in the real world, because learning requires accurate and timely feedback. Such feedback, they suggested, is rarely available because (1) outcomes are commonly delayed and not easily attributable to a particular action, (2) variability in the environment degrades the reliability of feedback, (3) there is often no information about what the outcome would have been if another decision had been made, and (4) many important decisions are unique and therefore provide little opportunity for learning. In fact, Ball, Bazerman, and Carroll (in press) observe little learning from experience in the winner's curse situation, despite ideal conditions for learning (real money was used, the game was played 20 times, and competitors were provided immediate and full feedback after each iteration). In summary, the persistence of the winner's curse problem suggests that, in fact, decision makers typically either do not consider the contingent decisions of competitive others or make simplifying and often false assumptions. Several other dysfunctional judgmental effects also manifest themselves as a result of ignoring or oversimplifying the contingent decisions of competitive others. These are discussed briefly in the next sections. Nonrational escalation oi commitment. Deteriorating competitive relationships often begin with a small dispute that each side continually escalates to a higher level. Individual and organizational decision makers tend to make decisions to justify their earlier charted directions (Staw, 1976, 1981), and competitive situations aggravate this tendency (Rubin, 1980). An interesting aspect of escalatory traps is that they are very difficult to disengage. Once a decision maker is caught up in a trap, the course of action needed to get out is not obvious. A central reason why competitive decision makers get trapped is that they fail to consider the contingent decisions of the competitor before engaging in competitive behaviors. If we would put ourselves in the perspective of the competitor, we might see the likely pattern of escalation, anticipate the escalatory response of the competitor, and devise a more appropriate competitive response. However, research suggests that most actors fail to adequately consider the other side in competitive situations. Overconfidence in judgment. Why can an advertising war continue even though neither side will win? In addition to the tendency to escalate commitment, competitive actors tend to be overconfident in their fallible judgments (Einhorn & Hogarth, 1981; Fischhoff, 1982). Farber (1981) discussed this problem in terms of divergent expectations and (using a competitive bargaining example) found that each side in a dispute is usually inappropriately optimistic that a neutral third party would agree with his or

262 1991

E. J. Zajac/M. H. Bazerman Zajac and Bazerman

43

her perspective. The key point is that competitive actors are likely to be overconfident, and they should expect similar overconfidence in the behavior of their opponents. The problem of overconfidence has also been widely studied using a number of types of decisions. Schwenk (1986: 302) reviewed this literature and noted that overconfidence, although not yet tested in corporate strategic decisions, has been "observed in other decisions of consequence." Even though Schwenk (1986) emphasized overconfidence in noncompetitive situations, our competitive bargaining example (consistent with our overall emphasis on competitive situations) suggests that overconfidence can result from competitive actors not considering the perspective of opponents. Limited perspective and frame to the problem. Creativity in competitive situations requires looking at the problem from new and different perspectives or frames (Winklegreen, 1974). Competitors, however, often develop perspectives for understanding a problem in self-centered ways. A firm that considers the decisions that competitors are facing enhances the likelihood of developing a creative strategic position. One example of such a creative strategic position is the $1.3 billion joint venture between IC Industries and PepsiCo for purposes of bottling Pepsi products. The joint venture agreement emerged after a legal battle between the two companies over IC Industries' acquisition of a Pepsi bottling operation and Pepsi's subsequent refusal to give the needed consent for the franchise transfer. The two firms were facing an impasse that seemed resolvable only through a costly legal process. However, when the two firms recognized that each one wanted to expand its share of Pepsi bottling operations, the joint venture idea, which enabled both firms to fulfill their goals, became a creative solution to the competitive situation. The IC Industries' CEO stated that "this is the best kind of agreement—everybody wins" (Chicago Tribune, 1987), and the PepsiCo CEO called the agreement " a strategic breakthrough." The fact that the firms viewed the decision as a "breakthrough" suggests that such decisions are, in fact, rare. In general, then, there is a tendency for competitors to ignore or insufficiently consider the decisions of competitive others. Manifestations of this tendency include the winner's curse, nonrational escalation of commitment, overconfidence of judgment, and a limited perspective and frame to the problem. The implications that such manifestations can have for specific strategic decisions relating to industry and competitor analysis are examined in the following section. IMPLICATIONS FOR STRATEGIC DECISIONS In keeping with the article's emphasis on competitive decision making in industry and competitor analysis, this section focuses on two specific strategic decisions: (1) the capacity expansion decision and (2) the new business entry decision (through internal development and through acqui-

Blind Spots in Industry Analysis 44

Academy oí Management Review

263

January

sition). Though it may be that no decision is inherently strategic (Mintzberg, 1979), these two decisions are typically strategic, given that they involve substantial capital, are considerably complex, and have the potential to influence competition and profitability (Porter, 1980). Capacity expansion. Porter (1980: 324) described the capacity expansion decision as "one of the most significant strategic decisions faced by firms, measured both in terms of the amount of capital involved and the complexity of the decision-making problem." The complexity stems from the need for accurate expectations with respect to future demand and competitors' reactions. For example, where there is low uncertainty with respect to future demand (e.g., demand will increase moderately), Porter (1980: 327) noted that "the capacity expansion process becomes a g a m e of préemption." In a rational scenario, those firms most able to quickly meet the demand will do so, and other rational firms will recognize that there is no gain in their also adding capacity. However, Porter (1980: 328) made the descriptive point that "too many firms try to preempt," whereby firms "mistake each others' intentions, misread signals, or misjudge their relative strengths and staying power." Porter (1980: 325) viewed this as a major cause of the overbuilding of industry capacity, and he noted that "undercapacity in an industry is rarely a problem," but that "industry overbuilding is a chronic problem." These descriptive observations, although very important in documenting the errors that real-world competitors commonly make, simply reveal the frequency with which firms "mistake . . . misread . . . or misjudge" when making capacity expansion decisions (Porter, 1980: 328). The present article's perspective, however, suggests that these judgmental errors and the chronic problem of industry overcapacity could be better understood if analyzed in terms of competitor blind spots. For example, a firm's insufficient consideration of competitors' contingent decisions may manifest itself in terms of overconfidence in judgment about the capacity expansion decision. In this case, although there can be only one preempting firm, competitors are overconfident in their belief that they will be the preempting firm, failing to sufficiently consider the likelihood that another competitor will be the preempting firm. This leads to greater investments in capacity expansion (creating overcapacity), and is consistent with Porter's observation that when facing a capacity expansion decision, too many firms try to preempt. Also, a firm investing in the race to be the preemptive firm in expanding capacity, but failing to recognize that some competitors, for whatever reason, place a higher value on expanding capacity and will expand regardless of who wins the preemption race, may find itself in a winner's curse situation. In this case, the winning firm may discover that preemption has not deterred competitors from also expanding capacity, or that the price of winning the race to be the preempting firm exceeds the actual benefit (because both cost and benefit were calculated without sufficient consideration

264 1991

E. J. Zajac/M. H. Bazerman Zajac and Bazerman

45

of competitors' reactions). This is also consistent with the observation noted above that firms facing a capacity expansion decision often mistake e a c h other's intentions. Similarly, a firm that does not adequately consider competitors' reactions may be viewing the capacity expansion situation only a s a gain opportunity, while failing to also consider the perspective of the losses that could accrue if competitors reacted aggressively to the focal firm's capacity expansion decision. Such a limited frame is consistent with Porter's (1980) observation that firms facing a capacity expansion decision often misjudge their "staying power" and their strength relative to competitors. Finally, a firm that insufficiently considers the contingent decisions of competitors may (when other firms in the industry begin to credibly move to expand capacity) become overly committed to its course of action and nonrationally escalate its commitment, despite competitors' reactions. This is consistent with the observation noted above that firms facing a capacity expansion decision often misread e a c h other's signals. It is also interesting to note that although Porter (1980: 325) acknowledged that firms often make capacity expansion mistakes and that overcapacity is a "chronic problem," Porter and Spence's (1982: 266) simulation analysis of capacity expansion in the corn wet milling industry hinges on the assumption of consistency, that is, that "firms do not have expectations that a r e inconsistent with the likely behavior of rivals." While this assumption is needed in Porter and Spence's (1982: 266) "searching for equilibrium," the likelihood that the assumption of consistency is descriptively correct may b e quite low. In fact, Porter and Spence (1982: 265) noted that "there is no logical reason why firms could not have expectations that were inconsistent with the likely behavior of rivals." They also acknowledged that their equilibrium result for capacity expansion would c h a n g e substantially if firms made mistakes in assessing their competitors' preferences, or if firms simply did not analyze rivals sufficiently. The present discussion of competitor blind spots suggests that such insufficient attention to competitors' contingent decisions is to b e expected and that the Cournot solution discussed by Porter and Spence m a y not b e a reasonable depiction of the capacity expansion decision process or outcome. (Winter [1982] drew a similar conclusion about the Porter a n d S p e n c e result, but for different reasons.) Empirical research on actual industry capacity expansion decisions is exceedingly rare; however, a study by Yoon and Lilien (1989) examined a set of firms in the titanium dioxide industry and observed that capacity expansion decisions were made largely a s a function of company-specific objectives. Their results suggest that firms focused primarily on their own growth, margin, and plant utilization data, more than on the interdependence of their competitive decision situation. Even though the study did not specifically address the issue of industry overcapacity, the results seem consistent with our emphasis on competitor blind spots.

Blind Spots in Industry Analysis 46

Academy

of Management

Review

265

January

The preceding discussion of competitive blind spots in capacity expansion decisions can be summarized with the following proposition: Proposition 1: All else being equal, strategic decision makers considering capacity expansion will pay insufficient attention to competitors' contingent decisions, resulting in industry overcapacity. New business entry (through internal development). The industrial organization economics literature on new business entry is voluminous, and a game theoretic perspective has been typically used to address the issue (Dixit, 1982; Salop, 1979; Spence, 1979; Tiróle, 1988). Porter (1980) also devoted considerable attention to the strategic decision to enter a new business—either through internal development or external acquisition. With respect to the entry decision through internal development, Porter (1980) noted that the firm must confront whatever structural entry barriers exist and must also gauge the expected effect of competitor retaliation (likelihood times magnitude). The rational firm will assess both types of entry barriers and decide accordingly. However, Porter (1980: 341-342) also noted that "often neglected in the entry decision is the probable reactions of existing firms" and that "too often [an entrant's] financial analysis assumes the industry prices and costs prevailing before entry." Here again, the descriptive observation, though important in documenting the mistakes that real-world competitors commonly make, simply reveals the frequency of judgmental errors in new business entry decisions. The emphasis in this article on competitive decision making, however, suggests that these judgmental errors may be better understood if analyzed in terms of competitor blind spots. For example, a prospective entrant's insufficient consideration of competitors' contingent decisions may manifest itself in terms of overconfidence in judgment about the new business entry decision (i.e., underestimating both the likelihood and magnitude of competitor retaliation). This is consistent with Porter's observation that entrants often neglect the probable reactions of existing firms in the industry. In addition, insufficient attention to competitor's contingent decisions can manifest itself in the limited frame problem. For example, a firm contemplating entry may be focusing only on the ex ante attractiveness of the market, failing to recognize that its entry into the market, in inviting competitor retaliation, would make the entire market less attractive ex post. This idea is consistent with Porter's observation that an entrant's financial analyses (and expectations) are often based on industry prices and costs prevailing before entry, rather than after entry. Also, if an entrant plans to gain a foothold by entering with low prices but fails to recognize that competitors, for whatever reasons (e.g., irreversible industry-specific resource commitments), place a higher value on maintaining their existing market shares, this can create a winner's curse situation. In other words, a firm that succeeds in establishing a position in the industry may find that the cost of entry has exceeded the actual benefit

266 1991

E. J. Zajac/M. H. Bazerman Ζajac and Bazerman

47

(because both cost and benefit were calculated without sufficient consideration of competitors' reactions). This idea is consistent with Porter's (1980: 341) observation that new entrants often neglect the effect of their entry on the supply/demand balance (and price cutting) in the industry. Finally, once the market entry process has begun and competitors begin to retaliate, a new entrant that is insufficiently considering the contingent decisions of competitors also may nonrationally escalate its commitment to this strategic direction (rather than withdraw from the market), even if competitors' actions are a signal of continued effective retaliation. In general, then, the implications of competitive blind spots for the new entrant's viability can be serious. The key point is that these blind spots emerge from the same basic problem: insufficiently considering the contingent decisions of competitors. It is also important to note that, relative to Porter (1980), game theorists representing the "new industrial organization" literature have a much more complex view of what a rational prospective entrant should be considering vis-à-vis competitors (which requires very strong rationality assumptions in terms of fully considering competitors' contingent decisions). For example, Jacquemin (1987: 124) has analyzed the new business entry decision in terms of sequential—or perfect Bayesian—equilibrium in an incompleteinformation game (the entrant does not know the exact cost structure of an incumbent firm). Fortunately, Jacquemin also offers a precise discussion of the level of rationality assumed in such a competitive game (which has been "simplified" because it has only one entrant and one incumbent firm): The entry candidate is assumed to be capable of imagining all possible cost structures of its adversary and of attributing a subjective probability to all eventualities; of calculating the likelihood of the price strategy used by its rival as a function of each cost structure c, c¡, . . . , cn; of estimating the joint probability to have at the same time a cost structure c¡ and a price policy ρ for each possible cost structure; of evaluating the marginal probability of having this price policy p, whatever the cost structure; and finally of assessing the posterior probability of the cost structure affecting its rival. It is only after consideration of these points that the entrant can discover its best response.

Although Jacquemin acknowledged the problems in requiring superrationality for such games, Weigelt and Camerer (1988: 446) argued that "the study of sequential equilibria in incomplete-information games may begin to finally fulfill the potential for applications of game theory to strategy research." It is interesting (if not ironic) that as strategy research touts the potential usefulness of sequential equilibria in incomplete-information games, Jacquemin, an industrial organization economist, bemoans the unrealistic rationality assumptions of such games and calls for a collaboration with a "behaviorist" perspective on the new business entry decision. The competitive decision-making approach outlined in this article offers one such collaborative perspective. (Note that our "behaviorist" discussion has

Blind Spots in Industry Analysis 48

Academy

of Management

Review

267

January

never presumed complete or perfect information games; game theorists assume that even in incomplete and imperfect information games, all available information is used properly by the [rational] players in the game.) The discussion of competitive blind spots in new business entry decisions can be summarized with the following proposition: Proposition 2: AU else being equal, strategic decision makers considering entry into a new business (through internal development) will pay insufficient attention to competitors' contingent decisions, resulting in new business entry failures. New business entry (through acquisition). In discussing new business entry through the acquisition of an existing firm, Porter (1980: 351, 355) observed that "it is quite difficult [for the buyer] to win at the acquisition game" and that there are "irrational bidders" who will overpay to acquire a firm. Roll (1986), a financial economist, noted that firms pay excessively high premiums over market value for acquisitions, a phenomenon he attributed to the "hubris" of managers in acquiring firms. Similarly, Salter and Weinhold (1979) suggested that bidding firms typically overestimate the value of acquisition target firms (see also Barney, 1988). These observations about irrational bidders and acquisition premiums are important in noting the frequency of judgmental errors firms make in actual acquisition decisions. This article suggests that persistent phenomena of acquisition premiums can be analyzed usefully in terms of competitive blind spots that emerge from not fully weighing the information that is available by considering the contingent decisions of competitive others. For example, according to a competitive decision-making perspective it is difficult to win at the acquisition game because there is usually an asymmetry of information in acquisition situations (favoring the seller) that the buyer does not take into sufficient account. Thus, the acquiring firm overpays for the acquisition because it does not adequately consider the fact that the target firm can selectively accept those acquisition offers that the target knows are profitable for itself. The winner's curse results discussed previously support this scenario, which is also consistent with research on actual acquisitions that showed that firms typically pay a substantial premium over market value for their acquisitions (Roll, 1986). Also, recall that the acquisition overpayments observed in the "acquiring a company" problem discussed previously were obtained despite acquirers knowing that their targets would be worth 50 percent more under the acquirer's management than under the target's management (suggesting that the acquirer, in overemphasizing the gain opportunity, has a limited frame). This idea is consistent with research suggesting that in acquisitions, the value created by the synergy of the two firms commonly goes to the seller (through the sales price), not to the buyer (Porter, 1980; Rappaport, 1986). In addition to judgmental errors between a single buyer and seller, another aspect of the acquisition decision involves competitive blind spots:

268 1991

E. J. Zajac/M. H. Bazerman Zajac and Bazerman

49

namely, the issue of competing bids, which are becoming more common in the market for companies. In the case of multiple bids, the acquiring firm also may be paying premiums because of the auction context of the market. As discussed in the earlier section on the winner's curse in a n auction context, higher bids are likely to come from a n acquirer who has one of the more optimistic estimates of the target firm's value. Failure to consider this issue can manifest itself in a nonrational escalation of commitment in competitive bidding for the target firm, resulting in the overpayment observed in many acquisition situations (Roll, 1986). This idea is also consistent with Porter's (1980: 355) observation that there are "irrational bidders" in acquisition situations. Similarly, if a bidding firm believes that it has some distinctive ability to improve the acquired firm, but fails to sufficiently consider the likelihood that other bidders also see themselves as having similar abilities, this insufficient consideration can manifest itself in terms of overconfidence in the judgment of the competing potential acquirers. Such overconfidence in a competitive bidding situation can lead to excessively high bids, and it is consistent with Roll's observation that top management "hubris" can lead to acquisition premiums (this situation might also affect later divestment decisions [Duhaime & Schwenk, 1985]). In general, the discussion of competitive blind spots in acquisition decisions can be summarized with the following proposition: Proposition 3: All else being equal, strategic decision makers considering an acquisition will pay insufficient attention to competitors' contingent decisions, resulting in acquisition premiums. Figure 1 shows how the analysis of specific strategic decisions and their outcomes can benefit from a competitive decision-making perspective that emphasizes the fundamental problem of firms' insufficiently considering the contingent decisions of competitive others. Figure 1 depicts how the manifestations of this fundamental problem (i.e., competitive blind spots) can lead to the commonly observed outcomes corresponding to each of the strategic decisions addressed above. CONCLUSIONS This article has shown that a behaviorally based, competitive decisionmaking perspective can help to explain persistent but poorly understood strategic phenomena such as industry overcapacity, new business entry failures, and acquisition premiums. The article's approach is thus consistent with that of other researchers who have used behavioral decision research to explain publicly observable, real-world decisions at the organizational and even governmental levels (e.g., Duhaime & Schwenk, 1985; Staw, 1981; Steinbruner, 1974). However, it is important to note that the article's competitive decisionmaking perspective draws not only on empirical findings from behavioral

Blind Spots in Industry Analysis 50

Academy of Management

§ 88 •a υ

J

w ΆΓ* ·Η M

O

Review

January

£G

ao « S S

3 m te 3 J S u-,

S3

il

a IO •O .S

I

a w a s f § m a

® E Β ο) "> 3 -Se CQ 0 * ζφ —

269

cφ β a o "Β > α> -α

c ω S Φ e g ,3 m S

'

Ê φ * ® ο

-.Βα a 3 σ

8

270 1991

E. J. Zajac/M. H. Bazerman Zajac and Bazerman

51

research, but also on the observations made by economists such a s Porter and Roll regarding actual firms' behavior in competitive situations. The persistence of such findings in the real world should not be considered surprising because the complexity and ambiguity surrounding the capacity expansion, new business entry, and acquisition decisions that firms must make are much more complex and difficult decision problems than those used in experimental settings. Additional problems in organizational decision making of a sociopolitical nature (cf. Cyert & March, 1963) can also contribute to the presence of firms' competitive blind spots. Taken together, these issues suggest that phenomena such as industry overcapacity, new business failures, and acquisition premiums may be the norm, not the exception. This raises the question as to whether a n alternative explanation for these phenomena can be constructed that would be more plausible than the perspective offered in this article. An obvious basis for a competing perspective would be a game theoretic approach that seeks to show some underlying rationality behind these (seemingly) nonrational phenomena. Given the superrationality assumptions underlying the game theoretic perspective (recall the earlier Jacquemin quotation), however, most strategy and organizational researchers would find it implausible that firms use such a rational decision process. An economist/game theorist might concur but also might claim that the actual decision process is irrelevant because the intensity of competitive pressures allows one to assume that in the long run firms will act "as if" they are fully rational. Note, however, that this "as if" argument (regardless of its accuracy) does not imply individual firm rationality (in process or outcome), but rather the "natural selection" of competitors at the industry level (Friedman, 1953). Therefore, if the research objective is to analyze the long-run equilibrium of markets, a rational, economic, game theoretic approach may be highly useful. However, if the research objective is to explain and understand the actual behaviors of individual firms as they interact with competitors, a behaviorally based, competitive decision-making approach may have distinct advantages in shaping future research questions in competitor analysis. For example, although it might appear desirable to offer clear-cut suggestions on how to avoid the competitive blind spots discussed here, empirical studies suggest that avoiding judgmental mistakes, or learning from them, is not easy (Schwenk, 1984; Tversky & Kahneman, 1986). Indeed, to propose that firms follow prescriptive frameworks without first "unfreezing" the tendency toward competitive blind spots may be counterproductive. Thus, although the powerful logic of game theory seems well-suited for suggesting optimal strategies and prescriptions for top management, the observed persistence of competitive blind spots suggests that game theory prescriptions should be viewed with caution. These cautionary comments about game theory prescriptions in strategic management are not intended to minimize the usefulness of game the-

Blind Spots in Industry Analysis

52

Academy

oí Management

Review

271

January

ory. Rather, game theory could be considered as a useful conceptual framework with which competitive blind spots in strategic decision making can be discussed. Harrigan (1985: 73-74), for example, discussed interfirm "misunderstandings" by making reference to a situation in which two parties fail to anticipate or recognize their divergent payoff matrices. Also, Weigelt and Camerer (1988), in discussing sequential equilibria in incomplete-information games, suggested in a footnote that an indicator of a firm's commitment to a particular strategy might be the probability of that firm deviating from rationality. A related issue involves the contingencies surrounding decisionmaking biases or blind spots. While Bateman and Zeithaml (1989) examined several possible mediators or interaction effects of decision biases, Schwenk (1984: 124) suggested that it may be premature to try and specify all the possible conditions under which decision-making biases will or will not apply in strategic decisions. We believe that the biases discussed in this article are generally strong tendencies, and that establishing interaction effects may be an important objective for future research (e.g., examining the role of top management composition and compensation, prior firm performance, and industry characteristics). This article, however, has focused on (1) developing the argument that there are biases or blind spots that are specifically created and exacerbated by competitive decision making and (2) linking this competitive aspect of strategic decision making to the competitor analysis literature by showing how blind spots can explain persistent, empirically observable phenomena relating to industry and competitor analysis. The discussion of behavioral issues in competitive decision making also has direct relevance for the emerging new industrial organization literature on industry and competitor analysis (which is more oriented toward game theory than the traditional structure-conduct-performance approach) (Jacquemin, 1987; Tiróle, 1988). As Waterson (1984: 18-19) noted, "the firm's conjecture about interdependence is . . . at the heart of the oligopoly problem. " This article gives explicit consideration to the behavioral realities underlying such conjectural, competitive decision-making issues. More generally, strategic management research may benefit from taking an integrated behavioral/economic perspective toward specific topics from the industrial organization economics literature. For example, Fombrun and Zajac (1987) explained how strategic groups can be more fully understood by complementing the usual structural approach with an approach that emphasizes interfirm perceptions (they viewed mobility barriers as partly perceptual in nature, with groups of organizations sharing common environmental perceptions). Similarly, perhaps our understanding of exit barriers can be enhanced when the usual emphasis on structural variables is complemented with an emphasis on cognitive/perceptual variables. Also, the important concept of committed competition (Schelling, 1960) can be better explained when concepts such as sunk costs are complemented with decision-making concepts such as the escalation of com-

272 1991

E. J. Zajac/M. H. Bazerman Zajac and Bazerman

53

mitment. Finally, an understanding of competitive blind spots could contribute to a theory of how conjectural variations (Bowley, 1924) are formulated, which would enable game theorists to use conjectural variations in dynamic competitive contexts (Jacquemin, 1987; Tiróle, 1988). Future research exploring each of these issues could provide a richer and more realistic assessment of a variety of strategic decisions. Indeed, the specific strategic decisions examined in this paper represent only a subset, albeit an important one, of decisions that can benefit from the decision-making/decision outcome link offered here. For example, a competitive decision-making perspective could be used to discuss other current topics in industry and competitor analysis, such as market signaling (e.g., to what extent are sent signals not "received" by competitor firms, and what are the competitive implications of unintentional signals?), contestable markets (e.g., when is new market entry actually anticipated by incumbents?), and the choice of optimal research and development or advertising levels (e.g., to what extent are levels chosen with competitors in mind?). Even the issue of oligopolistic pricing could be reexamined. For example, Shepherd (1979) noted that oligopolists often may ignore rivals' reactions in pricing decisions. Given the importance of interfirm perceptions in industry and competitor analysis, researchers who are sensitive to the role of blind spots in competitive decision making will be in a favorable position to predict and explain the actual strategic decisions made by firms. Furthermore, other economics-based theories and topics that have become increasingly relevant to strategy and organizational research may also benefit from attempts to integrate behavioral/perceptual and economic perspectives (Zajac & Olsen, forthcoming). For example, addressing the behaviorally based, subjective perceptions (and misperceptions) of top managers in organizations can play an important role in bridging the gap between descriptive and normative analyses of principal-agent relationships (Gaynor & Kleindorfer, 1987; Zajac, 1990a,b). The article also has implications for the nature of empirical research on topics relating to industry and competitor analysis. Much of the existing research in this area has relied solely on secondary and archival data sources, consistent with the usual preferences of the economists who have shaped this literature. However, primary data collected from key strategic decision makers (whether through interviewing, observational field research, or survey methods) would seem to have some advantage in terms of capturing the behavioral decision-making issues facing firms as they consider major strategic decisions relating to competitor analysis. Such alternative methods could provide a useful complement to traditional archival approaches, and they could enable comparative tests of behavioral and economically rational models of strategic decision making and decision outcomes. Finally, this article can be viewed as suggesting that the content versus process distinction, which is often used to characterize strategic management research, may actually hinder researchers' understanding of strategic

Blind Spots in Industry Analysis

54

Academy

ol Management

Review

273

January

decisions (Miller, Droge, & Toulouse, 1988; Zajac & Shortell, 1989). Characterizing strategy research as strictly process or content implicitly downplays the potential opportunities of integrating the strategic decision-making and competitor analysis literatures. As noted, decision outcomes and decision making are not independent topics, and they are in need of conceptual integration. This article has shown how these related topics can be brought together and used to explain strategic phenomena and to suggest directions for future research on strategic decision making and industry and competitor analysis. REFERENCES Ball, S. B., Bazerman, M. H., & Carroll, J. S. in press. An evaluation of learning in the Bilateral Winner's Curse. Organizational Behavior and Human Decision Processes. Barnes, J. H. 1984. Cognitive biases and their impact on strategic planning. Strategic Management Journal. 5: 129-138. Barney, J. B. 1988. Returns to bidders in mergers and acquisitions: Reconsidering the relatedness hypothesis. Strategic Management Journal. 9: 71-78. Bateman, T. S., & Zeithami, C. P. 1989. The psychological context of strategic decisions: A model and convergent experimental findings. Strategic Management Journal. 10: 59-74. Bazerman, M. H., & Carroll, J. S. 1987. Negotiator cognition. In B. M. Staw & L. L. Cummings (Eds.), Research in Organizational Behavior. 9: 247-288. Greenwich, CT: JAI Press. Bazerman, M. H., & Samuelson, W. F. 1983.1 won the auction but don't want the prize. Journal oí Conflict Resolution. 27: 618-634. Bowley, A. 1924. The mathematical groundwork of economics. Oxford: Clarendon Press. Capen, E. C., Clapp, R. V., & Campbell, W. M. 1971. Competitive bidding in high risk situations. Journal o/ Petroleum Technology. 23: 641-653. Cassing, J., & Douglas, R. W. 1980. Implication of the auction mechanism in baseball's free agent draft. Southern Journal of Economics. 47: 110-121. Chicago Tribune. 1987. IC, Pepsi settle suit by joining forces. November 10: 32. Cyert, R. M., & March, J. G. 1963. A behavioral theory of the firm. Englewood Cliffs, NJ: Prentice-Hall. Dixit, A. K. 1982. Recent developments in oligopoly theory. American Economic Review. 72: 12-17. Duhaime, I. M., & Schwenk, C. R. 1985. Conjectures on cognitive simplification in acquisition and divestment decision making. Academy of Management Review. 10: 287-295. Einhorn, H., & Hogarth, R. 1981. Behavioral decision theory: Processes of judgments and choice. Annual Review of Psychology. 32: 53-88. Färber, H. S. 1982. Splitting-the-difference in interest arbitration. Industrial and Labor Relations Review. 35: 70-77. Fischhoff, Β. 1982. Debiasing. In D. Kahneman, P. Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases: 422-444. New York: Cambridge University Press. Fombrun, C. J., & Zajac, E. J. 1987. Structural and perceptual influences on intraindustry stratification. Academy of Management Journal. 30: 33-50. Fredrickson, J. W. 1984. The comprehensiveness of strategic decision processes: Extension, observations, future directions. Academy of Management Journal. 27: 445-466.

274

1991

E. J. Zajac/M. H. Bazerman

Ζajac and Bazerman

Friedman, M. 1953. Essays in positive economics.

55

Chicago: University of Chicago Press.

Gaynor, M., & Kleindorfer, P. R. 1987. Misperceptions, equilibrium, and incentives in groups and organizations. In G. Bamberg & K. Spremann (Eds.), Agency theory, information, and incentives. Berlin: Springer-Verlag. Harrigan, Κ. R. 1985. Strategies for joint ventures. Lexington, MA: Lexington Books. Jacquemin, A. 1987. The new industrial organization: Market forces and strategic behavior. Cambridge, ΜΑ: ΜΓΓ Press. Kagel, J. H., & Levin, D. 1986. The winner's curse and public information in common value • auctions. American Economic Review. 76: 894-920. Kahneman, D., Slovic, P., &Tversky, A. (Eds.). 1982. Judgement and biases. New York: Cambridge Press.

under uncertainty: Heuristics

Kahneman, D., & Tversky, A. 1979. Prospect theory: An analysis of decision under risk. Econometrica. 47: 263-291. Miller, D., Droge, C., & Toulouse, J. M. 1988. Strategic process and content as mediators between organizational context and structure. Academy of Management Journal. 31: 544569. Mintzberg, H. 1979. The structuring of organizations. Englewood Cliffs, NI: Prentice-Hall. Mintzberg, H., Raisinghani, D., & Théorêt, A. 1976. The structure of "unstructured" decision processes. Administrative Science Quarterly. 21: 247-275. Myerson, R. 1990. Game theory: Analysis of conflict. Boston, MA: Harvard Business School Press. Porter, M. E. 1980. Competitive strategy. New York: Free Press. Porter, M. E., & Spence, A. M. 1982. The capacity expansion process in a growing oligopoly: The case of corn wet milling. In J. I. McCall (Ed.), The economics of information and uncertainty; 259-309. Chicago: University of Chicago Press. Raiffa, H. 1982. The art and science of negotiation. Cambridge, MA: Harvard University Press. Rappaport, A. 1986. Creating shareholder

value. New York: Free Press.

Roll, R. 1986. The hubris hypothesis of corporate takeovers. Journal of Business. 59: 197-216. Rubin, J. 1980. Experimental research on third party intervention in conflict: Toward some generalizations. Psychological Bulletin. 87: 379-391. Salop, S. C. 1979. Strategic entry deterrence. American Economic Review. 69: 335-338. Salter, M., & Weinhold, W. 1979. Diversification through acquisition: Strategies for creating economic value. New York: Free Press. Samuelson, W. F., & Bazerman, M. H. 1985. Negotiating under the winner's curse. In V. Smith (Ed.), Research in experimental economics, vol. 3: 105-137. Greenwich, CT: JAI Press. Schelling, T. C. 1960. The strategy of conflict. Cambridge, MA: Harvard University Press. Schwenk, C. R. 1984. Cognitive simplification processes in strategic decision making. Strategic Management Journal. 5: 111 128. Schwenk, C. R. 1986. Information, cognitive biases, and commitment to a course of action. Academy of Management Review. 11: 298-310. Shepherd, W. G. 1979. The economics of industrial organization. Englewood Cliffs, NJ: Prentice-Hall. Shubik, M. 1984. Game theory in the social sciences. Cambridge, ΜΑ: ΜΓΓ Press. Spence, A. M. 1979. Investment strategy and growth in a new market. Bell Journal of Economics. 10: 1-19. Staw, B. M. 1976. Knee-deep in the big muddy: A study of escalating commitment to a chosen course of action. Organizational Behavior and Human Performance. 16: 27-44.

Blind Spots in Industry Analysis 56

Academy of Management

Review

275

January

Staw, Β. M. 1981. The escalation oí commitment to a course of action: A review and analysis. Academy of Management Review. 6: 577-587. Steinbruner, J. D. 1974. The cybernetic

theory of decision. Cambridge, ΜΑ: ΜΓΓ Press.

Tiróle, ]. 1988. The theory of industrial organization. Cambridge, MA: MIT Press. Tversky, Α., & Kahneman, D. 1986. Rational choice and the framing of decisions. The Journal of Business. 59: 251-284. Waterson, M. 1984. Economic theory of the industry. Cambridge, ΜΑ: ΜΓΓ Press. Weigelt, K., & Camerer, C. 1988. Reputation and corporate strategy: A review of recent theory and applications. Strategic Management Journal. 9: 443-454. Weigelt, K., & MacMillan, I. C. 1988. An interactive strategic analysis framework. Strategic Management Journal. 9: 27-40. Winklegren, W. A. 1974. How to solve problems.

San Francisco: Freeman.

Winter, S. G. 1982. Comment on the capacity expansion process in a growing oligopoly: The case of corn wet milling. In I. ]. McCall (Ed.), The economics of information and uncertainty: 310-316. Chicago: University of Chicago Press. Yoon, E. S., & Lilien, G. 1989. An oligopoly market model when capacity and price are decision variables: A case study of the titanium dioxide industry. Working paper, Pennsylvania State University, University Park, PA. Zajac, E. ]. 1990a. Economic and behavioral perspectives on the structuring of CEO compensation: Evidence from CEOs. Working Paper, Northwestern University, Evanston, IL. Zajac, E. J. 1990b. CEO selection, succession, compensation, and firm performance: A theoretical integration and empirical analysis. Strategic Management Journal. 11: 217-230. Zajac, Ε. I., & Olsen, C. P. forthcoming. From transaction costs to transactional value analysis: Implications for the study of interorganizational strategies. Journal of Management Studies. Zajac, Ε. I., & Shortell, S. M. 1989. Changing generic strategies: Likelihood, direction, and performance implications. Strategic Management Journal, 10: 413-430. Edward I. Zajac received his Ph.D. in organization and strategy from the Wharton School, University of Pennsylvania. He is an assistant professor at the J. L. Kellogg Graduate School of Management, Northwestern University. His research focuses on integrating economic and behavioral perspectives on strategic adaptation, corporate governance, and interorganizational strategies. Max H. Bazerman received his Ph.D. in organization behavior from Carnegie-Mellon University. He is the I. L. Kellogg Distinguished Professor of Dispute Resolution and Organizations at Northwestern University. His research focuses on the decision processes of negotiators and dispute resolution mechanisms.

276

R.Jacobson

Academy oí Management Review 1992. Vol. 17. Mo. 4. 782-807.

e

THE "AUSTRIAN" SCHOOL OF STRATEGY ROBERT JACOBSON University of Washington Although traditional industrial organization continues to serve as one oi the conceptual foundations for strategic thinking and research, many oi Its premises have come under widespread criticism. Industrial organization largely Ignores, despite their importance, change, uncertainty, and disequilibrium in the business environment. Because these fundamental characteristics are cornerstones of the Austrian School of Economics, this doctrine offers unique strategic perspectives. The Austrian emphasis on "the market process" and entrepreneurial discovery establishes a framework for both strategy formulation and research.

According to neoclassical microeconomic theory, profit-maximizing firms in competitive environments earn zero economic profits. They earn a return just sufficient to maintain capital investment. The notion of "perfect competition" sets the standard for traditional industrial organization and provides the foundation for the premise that firms earn supranormal returns primarily by exercising monopoly power (Bain, 1951). Monopoly power, in turn, exists to the extent that the firm or industry has erected barriers to entry that restrict competitive forces. Industrial organization (IO) economists study the linkages among industry structure, conduct, and performance in order to derive public policies that promote competition. Porter (1981: 617) concluded that "there is gold to mine in applying IO concepts to strategy formulation." He viewed the Bain/Mason paradigm of industrial organization as offering strategic management a systematic model for assessing competition and for strategy formulation. Indeed, theoretical perspectives drawn from IO have exerted a substantial and pervasive influence on strategic thinking. Mainstream strategy topics such as Porter's generic strategies for coping with competitive forces, strategic groups, and mobility barriers have their theoretical base in traditional IO (Caves & Porter, 1977). Indeed, entire research streams (e.g., the PIMS approach, Buzzell & Gale, 1987) build off this same base. To a large extent, business policy researchers reverse the intent of IO concepts to form strategic recommendations. Much of the current thinking I thank John Butler, Gary Hansen, and Dennis Quinn for their comments and suggestions. 782

The .Austrian' School of Strategy 1992

Jacobson

277 783

about strategic management focuses on ways that firms can create imperfectly competitive product markets in order to obtain greater than normal profits. Whereas industrial organization economists seek to determine what can be done to promote competition, many strategy researchers seek to determine what managers can do to limit competition. Porter (1980: 4) defined the strategic objective of a business unit as to position itself in an industry where it can best defend itself against competitive forces, or at least influence them in its favor. Despite its widespread influence on strategic thinking, some scholars have questioned the utility of IO concepts because of an inadequate theoretical substructure. For example, Nelson and Winter (1982) criticized IO because of the lack of attention its theorists give to the dynamic environment brought about by technological change. A substantial amount of technical advance results from profit-oriented investment on the part of firms. The profits from successful innovation are disequilibrium phenomena stemming from the lead time over competition that innovation affords. The equilibrium analysis of IO provides insufficient insights into either the motivation for or the consequences of innovation. Nelson (1976) contended that if change, uncertainty, disequilibria, and institutional complexity are important parts of what is going on, then implications derived from the traditional theory based on different assumptions must be viewed with a jaundiced eye. An inconsistency with public policy concerns has also led to a questioning of the role of industrial organization as a basis for business policy. Perfect competition sets the standard for traditional IO because it leads, at least according to neoclassical economic theory, to economic welfare maximization. Strategy researchers, who under this same theoretical framework emphasize restricting competitive forces, advocate, therefore, restricting economic/customer welfare. Customers pay higher prices or accept inferior quality goods as a consequence of strategies that restrict competition. Rumelt (1988) summarized this market power perspective on strategy a s involving in its most basic form coi/uding behind strategically erected entry barriers in order to make money. This implication has made it difficult for strategy researchers who are concerned with public policy to embrace or build upon the IO framework as part of their theoretical base. Weaknesses in the industrial organization-based strategy literature suggest that other economic perspectives may warrant consideration as a framework for strategic analysis. Austrian economics is one perspective that has considerable potential to enhance strategic thinking and research. Though the term Austrian economics suggests different ideas to different people, it is most widely used to refer to perspectives that emphasize "the market process," that is, market dynamics. This school of thought originated in the 1870s with Carl Menger in Vienna and has continued through his direct and indirect disciples, in particular, Ludwig Mises, Friedrich Hayek, and Israel Kirzner. Although differences exist,

278 784

R.Jacobson

Academy

of Management

Review

October

the work of Joseph Schumpeter (α student of Austrian economists von Weiser and Bohm-Bawerk) is also often linked to the Austrian School. Even though relatively few strategy researchers explicitly attribute or link their analysis to Austrian economics (Wensley, 1982, is a notable exception), the influence of Austrian thinking is more widespread than this lack of attribution might suggest. Austrian concepts shaped the work of a number of economists whose theories have had a great impact on strategy research, in particular, Joseph Schumpeter. Rumelt (1984: 560) noted that the chief concern of business policy researchers (i.e., profit seeking through corporate entrepreneurship) "is most closely associated with Schumpeter's vision of competition as the process of 'creative destruction' rather than as a static equilibrium condition." Thus, a substantial component of strategy research has implicitly incorporated and built upon concepts closely associated with Austrian economics. Interestingly, Porter (1990), noting the need for a new paradigm, advocated a number of views that move away from the static concepts of traditional IO to those that more closely resemble Austrian thinking. For example, he stated that competitive advantage grows fundamentally out of improvement, innovation, and change and that it is sustained only through relentless improvement. Austrian economics warrants consideration as a framework for strategy development because it emphasizes perspectives, courses of action, and research topics that, although regarded as fundamental to business success, currently receive relatively little attention in the strategy literature. In particular, strategic issues relating to (a) continuous innovation, (b) flexibility, (c) intertemporal heterogeneity, and (d) unobservable influences of business performance, all of which have received rather cursory treatment in the strategy literature, are of central importance in Austrian-based strategy. OVERVIEW TO AUSTRIAN ECONOMICS No definitive "Austrian litmus test" exists. Austrian views are not completely orthogonal to views arising from neoclassical economics or traditional industrial organization. Indeed, Kirzner (1981) believes that the Austrians view the foundation of neoclassical economic theory as sound. This is not surprising because many of the principles of neoclassical economics have originated with the Austrian economists. Such concepts as marginalism, opportunity costs, and diminishing marginal utility are tied closely to the founders of the Austrian school. As such, the distinction between Austrian-based strategy and industrial organization-based strategy becomes, in some instances, a matter of degree or emphasis. However, certain perspectives tend to differentiate Austrians from nonAustrians. Table 1 summarizes these perspectives and contrasts them with strategic perspectives drawn from traditional industrial organization.

The .Austrian' School of Strategy 1992

279 785

Jacobson

TABLE 1 Differences in Perspective

Strategic Objective Market Conditions Profitability Modeling Nature of Success Factors

Industrial Organization-Based Strategy

Austrian Economics-Based Strategy

Restricting competitive forces Equilibrium Empirical regularities Observed strategic factors

Entrepreneurial discovery Disequilibrium Heterogeneity Unobservable factors

The notion of the market process, in particular, tends to distinguish Austrians from non-Austrians. Though neoclassical theorists concentrate on equilibrium, with a static notion of the nature of competition, Austrian economists view markets as processes of discovery that mobilize dispersed information. The Austrians contend that firms earn profits through entrepreneurial discovery. They focus on the entrepreneur, motivated by the desire for supranormal profits, as a vehicle for promoting discovery and for realizing opportunities in a constantly changing (disequilibrium) marketplace. Because competitors imitate strategies known to generate above-normal returns until their return premium is eliminated, the abnormal returns associated with this discovery are only temporary. This suggests that empirically modeling business performance to find systematic strategies (regularities) that firms can implement to earn supranormal returns, as is widely done in strategy research, will be largely unsuccessful. Indeed, business success is depicted as depending critically on many time- and firm-specific unobservable factors. ENTREPRENEURIAL DISCOVERY AND PROFITS Perhaps the primary limitation of the neoclassical theory of competition and profit is that it fails to provide a motive for the search for new products and methods (i.e., innovation). The Austrian school highlights profits not as the result of monopoly power but rather as the consequence and the incentive for discovery and innovation. Under this view, the goal of strategy formulation centers not on limiting competitive forces but rather on entrepreneurial discovery.1 Gluck, Kaufman, and Walleck (1980) maintained that the essence of strategy involves avoiding competition through an indirect approach. This indirect approach is entrepreneurial discovery, which these authors believe includes, for example, a reformulation of a product's function, the development of new manufacturing 1 This is not to suggest that competition is an unimportant strategic consideration. Rumelt (1987) suggested that isolating mechanisms (i.e., impediments to imitation by competition) need to be present so that the time lag for imitation will be sufficient so as to encourage innovation. In this way the innovator can reap some of the benefits of the innovation in terms of supranormal profit before imitators complete the competitive process. Coping with competitive forces, however, is clearly of secondary importance to discovery.

280 786

R.Jacobson Academy of Management

Review

October

methods or distribution channels, or the discovery oí dimensions of competition that competitors have overlooked. Profits in an Efficient Market As a parallel to the Austrian view, consider the behavior of efficient markets in equilibrium. Efficiency characterizes a number of markets (e.g., stocks, bonds, commodity futures, foreign exchange, sports gambling). The price of goods in an efficient market represents the consensus opinion of their present value. This valuation is determined through market participants' incorporating all available information. As soon as a new piece of information becomes available, the market participants analyze it, and the results are reflected immediately in the price of the good. Any profit opportunity resulting from this information dissipates quickly. For example, weather information influences the supply and, therefore, the price of orange juice. Market participants incorporate pertinent weather information into the price of orange juice futures once it becomes available (Roll, 1984). This information provides no profit opportunity if others in the market also are aware of it. The current price already reflects its effect. The basic assumptions underlying the efficient markets/rational expectations framework preclude persistent profit opportunities and downplay the role of the entrepreneur. Profits exist only in a world of uncertainty and disequilibrium. When subjective probability distributions used by decision makers are the same as the probability distribution of the relevant variables (i.e., economic agents form their expectations as if they had knowledge of the "true" model linking the relevant variables), profits take on a purely random character. In this framework, a firm will earn economic profits only by chance and only for a very brief period of time. Barney (1986b) noted that when firms that seek to control resources to implement a strategy and firms that currently control these resources have accurate expectations about their future value, then the price paid for resources approximates their value once they are implemented. In this efficient market, all profits that can be obtained from the implementation of a strategy will be anticipated (i.e., reflected in the price of the resource) and, therefore, competed away. Once a profit opportunity becomes obvious, market participants respond by appropriately adjusting prices. This does not mean that firms cannot earn what might be perceived as an abnormal return. Those firms possessing unique resources, whether it be by design or chance, are rewarded (Barney, 1986b). Neoclassical theorists view these rewards not as profits but rather as returns on scarce resources. Further, by pure chance, an unexpected event may arise that generates abnormal returns for a firm following a strategy that has an expectation under anticipated conditions of normal returns. However, neoclassical theorists concentrate on firms earning supranormal profits primarily by restricting competitive forces and, therefore, the efficient functioning of markets.

The ,Austrian' School of Strategy

1992

Jacobson

281

787

Entrepreneurial Profits

Dissatisfied with the neoclassical conceptualization of business profitability, Schumpeter (1934, 1942) discussed the importance of the entrepreneur and innovation as fundamental to business success. He argued, in a distinctly Austrian manner, that economic development takes place when firms implement new products, production processes, and organizational techniques (i.e., engage in entrepreneurship). In the Schumpeterian system, the entrepreneur disrupts the market and moves it away from its equilibrium. When the new innovations are completed and the new products enter the market or when new production processes are in place, the innovator outcompetes the other firms and earns economic profits. Realization of supranormal profits provides the incentive for innovation, but these profits are short lived. As innovations are imitated, economic profits dissipate and finally disappear. The market returns to equilibrium until another innovation occurs. Each innovation is imitated and then replaced by yet another innovation. This is the process of "creative destruction." The gains realized from the innovation give the firm only the means to pursue new innovations. The forces of dynamic competition doom any firm that merely attempts to maintain its present position. Schumpeter's notion of the market being, at times, in equilibrium separates him from the "mainstream" Austrian viewpoint. 2 However, similarities also exist. As with Schumpeter, Austrians focus on the entrepreneur and see profits as the incentive that stimulates entrepreneurship. Mises (1949) commented that entrepreneurs are the first to understand that a discrepancy exists between what is currently done and what could be done. According to Mises (1949), entrepreneurship is an action that successfully directs the flow of resources toward the fulfillment of consumer needs. Alertness to opportunities is the hallmark of entrepreneurs. Entrepreneurs discover errors or inefficiencies and try to eliminate them. Although opportunities for profit are always present, businesses may not discern them. Many potentially advantageous opportunities go unnoticed. The role of the entrepreneur is to see economic opportunities that have been overlooked by others. Kirzner (1973, 1979) argued that at any given time, an enormous amount of ignorance stands in the way of the complete coordination of the actions and decisions of the many market participants. As such, innumerable opportunities for mutually beneficial exchange are likely to exist unperceived. Even though some profit opportunities are uncovered by pure chance, certain firms have more information than others, and this knowledge gives them an advantage in ascertaining market inefficiencies. The existence of true entrepreneurial profit

2 Austrians argue that ior economy to be in equilibrium, innovations must be discontinuous (i.e., appearing only in discrete clusters). They view innovation as a continuous process. Therefore, the market is never in equilibrium.

282 788

R.Jacobson Academy of Management

Review

October

depends on the possession of superior information. This is the entrepreneurial role: to gather, evaluate, and utilize information. Resources flow toward the firms that are most competent in using information, and the least efficient firms are forced out of business. Market imperfections or inefficiencies allow a market to be in disequilibrium and are responsible for profit opportunities. In this sense, the entrepreneur acts as an arbitrager. The entrepreneur sees a mismatch between what the resource market has to offer and what customers will be willing to pay. By exploiting this market imperfection, the entrepreneur receives the residue from the arbitrage (i.e., economic profits). Similar to Mises and Kirzner, Rumelt (1987: 143) defined "entrepreneurial rent as the difference between a venture's ex post value (or payment stream) and the ex ante cost (or value) of the resources combined to form the venture." The possibility of earning these profits sustains the entrepreneur in a state of alertness. Engaging in entrepreneurial discovery by bringing resources, costs, and prices further in the direction of equilibrium generates profits. Entrepreneurial Discovery Entrepreneurial discovery involves a wide range of activities. Schumpeter (1942: 132) viewed the function as "to reform or revolutionalize the pattern of production by exploiting an invention, or more generally, an untried technological possibility for producing a new commodity or producing an old one in a new way, by opening up a new source of supply of materials or a new outlet for products, by reorganizing an industry." A variety of researchers have highlighted the fundamental role of discovery in influencing business success. Some innovation can be termed scientific in nature (e.g., the result of research and development). Schumpeter (1934, 1942) emphasized the crucial role of R&D in determining economic well-being and business success. Solow (1957) highlighted technological change (the advancement of knowledge) as the primary determinant of economic growth, and Hayes and Abernathy (1980: 468) argued that "success in most industries today requires an organizational commitment to compete in the marketplace on technological grounds." However, Hayek (1945) noted that there is a body of very important but unorganized information that cannot be classified as scientific in the sense of knowledge of general laws. Rather, this information is knowledge of particular circumstances of time and place. Knowledge of people, local conditions, and special circumstances is just as important to business success as scientific facts. It is crucial both in ascertaining profit opportunities and in implementing strategies. Kirzner (1979) agreed that entrepreneurial discovery includes shortrun activities fully as much as it does long-run developmental changes. Imitators, attempting to exploit the opportunities generated by the actions of the innovators, exercise entrepreneurship as much as the innovators themselves. Levitt (1986) asserted that no company can survive on its own

The ,Austrian' School of Strategy 1992

Jacobson

283 789

innovations. In a world of eager competitors, there will always be others who, on some dimensions, take the innovator's lead. He maintained that even though a company must work hard at being a leader, it must work equally hard at being a systematically eager imitator. Entrepreneurship occurs whenever a market participant does something even a little different from what is currently being done, so that some market imperfection can be profitably exploited. Bhide (1986) stated that opportunities to gain lasting advantage through blockbuster strategic moves are rare in any business. In the current business environment where innovations are quickly imitated or quickly become obsolete, he argued that the theory of sustainable comparative advantage has had its day. Accordingly, what counts most is a firm's "nimbleness" to exploit rapidly dissipating and changing opportunities. DISEQUILIBRIUM The Austrian emphasis on the role of knowledge and learning in dynamic competitive markets explains differences between the neoclassical and the Austrian views on equilibrium. Hayek (1937) regarded the neoclassical use of the concept of equilibrium as unfortunate because it presupposes an unwarranted degree of knowledge. The neoclassical assumption of perfect knowledge does not recognize the significance of entrepreneurial discovery in an uncertain and changing world. As such, equilibrium is unrealistic because it implies that discovery has ceased. The changing of plans as a result of the acquisition of new information constantly disrupts the market. As such, an unending stream of knowledge keeps the market in perpetual motion. Because knowledge is continually changing, the market is continually changing. Continuous changes in the state of knowledge produce new disequilibrium situations and, therefore, new profit opportunities. When the competitive process eliminates one opportunity, changes in the stream of knowledge produce other opportunities. Thus, there are constant profit opportunities resulting from ever-changing sources. The Market Process As in neoclassical theory, Austrian theory stresses the beneficial role of competition. However, Austrians treat competition as a process rather than as a static notion. Hayek (1948) noted that "perfect competition" in neoclassical economics is a market that is already in equilibrium, that is, a state of affairs in which there is no opportunity "to compete." Adherents of neoclassical economics assume away the fundamental dynamic characteristics of competition and concentrate primarily on specifying the effects of competition after the process of competition has reached its limits. Even though they describe equilibrium, their analysis does not explain the competitive process that led to this outcome. The Austrians focus on this process and, therefore, on the role of the entrepreneur. For

284

790

R.Jacobson

Academy

of Management

Review

October

them, the entrepreneur is the mechanism that drives the competitive process to its fruition. Indeed, competition and entrepreneurship are viewed as inseparable. Hayek (1945) stressed that economic problems arise always and only as a consequence of change. As long as events continue to be the same, or at least continue as they were expected to, there will be no new problems that require a decision. But such a description does not characterize the business situation. The economic problem of business is mainly to adapt to rapid changes in the environment. Some commentators on business performance have expressed the same sentiments. For example, Peters (1987) argued that violent, accelerating change and chaos characterize the business environment. Thus, companies win by constantly adapting and innovating. Peters (1987: 45) admires management styles and organizational structures that respond quickly to shifting circumstances. He stated, "Today, loving change, tumult, even chaos is a prerequisite for survival, let alone success." Profits as a Disequilibrium Phenomenon Abnormal profits will not exist indefinitely: Profits will decay to the competitive level as competitors imitate successful practices and market conditions change. Because firms must trade in the marketplace to realize the benefits of its information, they cannot profitably exploit knowledge without conveying hints to other firms. If the information is revealed, competitors will respond and, perhaps, will dissipate the advantage before the firm has had an opportunity to capitalize on it. A number of firms. Frito Lay, for example, have dispensed with some types of test marketing because it gives competitors too much information prior to the launch of the product. Japanese firms are especially reluctant to engage in market research for this reason—when Kao researched its laundry product. Attack, it attempted to limit exposure of the product to potential competitors by giving test samples to only a small number of geishas. Teece (1987) commented that quite often imitators profit more from an innovation than does the firm first to commercialize it. He believes that many science- and engineering-driven firms labor under the illusion that developing new products that meet customer needs will ensure fabulous success not only for the product but also for the innovating firm. There is a long list of companies, for example, EMI (CAT scanner), Xerox (office computer), and Ampex (VCR), that failed to profit from their innovations. Mansfield, Schwartz, and Wagner (1981) reported that about 60 percent of patented successful innovations were imitated within four years. In addition, an imitator's development costs were approximately 35 percent lower than those of the innovator. In their survey of R&D managers. Levin, Klevorik, Nelson, and Winter (1987) found that even major patented innovations could be imitated within three or fewer years in well over half of the 129 lines of business they covered. Approximately 65 percent of "typical" unpatented innovations could be imitated in less than one year.

The .Austrian' School of Strategy

1992

Jacobson

285

791

These authors noted that managers place relatively little faith in the ability of patents to prevent rivals from imitating either their product or process innovations. Lead time was regarded as far more effective than patents in protecting a firm's comparative advantage. Competitors are delayed from imitating innovators' actions when, for example, knowledge available to them about these actions is ambiguous. Demsetz (1973: 2) noted that "it is not easy to ascertain just why GM or IBM perform better than their competitors. The complexity of these firms defies easy analysis." 3 Lippman and Rumelt (1982) discussed how firms are able to earn long-term profits when uncertainty surrounds the factors and actions responsible for superior performance and, as such, competitors are unsure what actions to imitate. Winter (1987) noted that some types of process R&D, because lack of observability may delay imitation, can be expected to give rise to longer lasting advantages than those obtained through new product R&D. The existence of ambiguity (i.e., imperfect information) means that successful strategies may not be imitated and the price of effective resources may not be bid up. Ambiguity surrounds not just the strategy itself, but also its implementation. Generic strategies are often not detailed in the specificity required for use by managers. For example, the suggestion to raise product quality may be inadequate for managers to act upon. Product quality is a multidimensional construct. Managers may not know which elements of product quality have an impact on profitability. Further, even if these elements were known, it is often unclear how to implement the techniques necessary to improve product quality. General Motors' failed efforts with automation is a classic example (Economist, 1991). GM spent $60 billion from 1979 to 1986 to equip its factories with "21st-century" automation to improve quality. Its joint venture with Toyota, however, uses less automation than the average U.S. car plant, yet is said to produce higher quality cars than GM's most highly automated factory (Business Week, 1987). MODELING THE DETERMINANTS OF BUSINESS SUCCESS

Entrepreneurial discovery and innovation (i.e., doing what is currently being done in a different way to better or more efficiently meet customer needs) induces profits in the Austrian framework. This perspective generates a sharp split between ΙΟ-based strategy and Austrian perspectives with respect to the role of empirical methods. In contrast to IO researchers, Austrians contend that empirical models cannot uncover strategies that yield supranormal profits. Undertaking strategies whose

3 Demsetz's example also serves to highlight the transitory nature of profits. Despite their earlier successes, in 1991 GM lost $4.452 billion and IBM lost 2.827 billion, which are the largest and fifth largest annual losses in U.S. corporate history.

286

R.Jacobson

792

Academy ot Management Review

October

value and implementation are known involves little (if any) entrepreneurial discovery and, therefore, will not generate abnormal returns.4 Profitability Models In 1987, the Harvard Business Review reprinted a passage from its first issue (1922), which stated: "The theory of business . . . must develop to such a point that the executive, who will make the necessary effort, may learn effectively from the experiences of others in the past what to avoid and how to act under the conditions of the present. Otherwise, business will continue [to be] unsystematic, haphazard, and for many . . . a pathetic gamble." 5 Strategy research has been responsive to this call. Empirical models of business performance, based on past data and experiences, have been developed under the belief that they depict strategies that managers can follow to achieve superior business performance. Porter (1981) noted that industrial organization researchers have developed a strong empirical tradition built around the statistical analysis of the populations of firms and industries. A stream of research, both in strategy and IO, is based on the premise that underlying regularities (sometimes labeled strategic laws) govern business behavior and determine profit performance. Schoeffler (1977) noted that the implication for research is that the process of formulating a business strategy is becoming an applied science. Business situations can be understood by an empirical scientific approach. Heterogeneity Perhaps no view is a greater antithesis of the Austrian doctrine than the notion of interfirm and/or intertemporal homogeneity that is imbedded in empirical models of business profitability. Austrian economics rejects econometrics as a tool of theory development. Mises (1949: 55-56) defined the Austrian position most clearly. He stated: There are, in the field of economics, no constant relations, and consequently no measurement is possible. If a statistician determines that a rise of 10 per cent in the supply of potatoes in Atlantis at a definite time was followed by a fall of 8 per cent in the price, he does not establish anything about what happened or may happen with a change in the supply of potatoes in another country or in another time. He has not "measured" the "elasticity of demand" of potatoes. He has established a

* Emphasis on unobservable firm-specific factors as primary determinants of business performance further reduces the value that Austrians place on attempts to explicitly model the determinants of business performance. 5 Interestingly, next to this passage is a cartoon in which an executive indicates that his success was due to a chance occurrence (i.e.. having "found forty million dollars under a rock").

The .Austrian' School of Strategy 1992

Jacobson

287 793

unique and individual historical fact. . . . Statistical figures referring to economic events are historical data. They tell us what happened in nonrepeatable historical cases. The sole excuse that the econometrician advances is that his hypotheses are "saying only that these unknown numbers remain reasonably constant through a period of years." Now whether such a period of supposed constancy of a definite number is still lasting or whether a change in the number has already occurred can only be established later on.

Absence of constants in economic life makes any attempt at econometric determination of such constants a misleading enterprise. Mises pointed out that each historical event involves a complex interaction among a variety of variables, none of which ever remains in constant relationship with the others. Every historical event is heterogeneous. Therefore, historical events cannot be used either to test or to construct laws. Mises claimed that, given the ever-changing conditions of human will, knowledge, and values, it is inconceivable that econometrics will ever be able to do so. The exact role of econometrics in the Austrian tradition is unclear. In fact, Littlechild (1978: 22) noted that "no two Austrians have ever completely agreed on methodology." A number of Austrians have asserted that econometrics has no role in the advancement of knowledge. Rizzo (1978) commented that although this view may serve as a tolerable first approximation of the truth, statistical regularities can serve as a starting point for an investigation. The central role of econometrics involves the use of statistical methods in providing insights into historical events. The econometrician is an economic historian. However, these historical relationships are not interpreted as constants applicable to all situations in all times. Extrapolating the future from the past is an inductive leap that Austrians are very reluctant to make. This view is not unique to the Austrian school. The rational expectations critique of econometric models contends that econometric models estimated from past data may not be relevant for policy evaluation because the parameters of such models can be expected to change along with the policy. Lucas (1981: 126) stated, "Given that the structure of an econometric model contains optimal decision rules of economic agents, and that optimal decision rules vary systematically with changes in the structure of series relevant to the decision maker, it follows that any change in policy will systematically alter the structure of econometric models." Kristol (1988: 18) observed that econometric models assume "all other things being equal." However, he argued that "human beings are not things, they are purposive creatures shaped by history and expectations. So, when one deals with human behavior, 'other things being equal' is a premise that is conditional, not absolute, and has a transiency as well as an indeterminacy that makes large economic theories less solid than economists would have us believe." The criticism of econometric models is especially pertinent with re-

288 794

R.Jacobson Academy oi Management Review

October

gard to models of business profitability. If these regressions have detected regularities in business performance, why don't all businesses follow those strategies that are known to lead to supranormal profits? Managers should be expected to continue to invest in activities or to follow strategies that are known to increase returns until the return premium is exhausted. Profitability, or the lack thereof, is a signal that directs resources toward their most advantageous use. Techniques are imitated and the price of those resources that contribute to supranormal profits are bid up. The adjustment of resources and output into areas earning excess profits and away from areas earning below-average profits brings returns back to competitive levels. The very nature of competition suggests that no replicable strategy will allow businesses to earn long-run supranormal profits. It is the very uncertainty of a given strategy or its implementation that makes it a candidate for determining business success. The return premium on easyto-implement strategies dissipates quickly. Few would argue that product markets adjust at, say, the same speed as the stock market. But, product markets will react to, and eliminate, abnormal profit opportunities. Therefore, factors identified in the past as generating supranormal returns should not be expected to have this effect in the future. Consider, for example, empirical analysis that indicated that advertising expenditures systematically generated supranormal profits (e.g., $1 spent on advertising generated incremental profits of 500). The very fact that some firms have exploited the opportunity of increased advertising spending brings its benefits to the attention of other market participants. These other firms respond by increasing their own spending on advertising until the return on advertising expenditures tends to equilibrate with other uses of funds. The return premium from advertising expenditures, which existed previously, dissipates. The profit opportunity from advertising cannot be expected to exist forever. In fact, the only question is whether the empirical analysis brought the benefits of advertising to the attention of the firms so that the return premium will be subsequently competed away or whether the firms were aware of the benefits of advertising prior to the empirical analysis so that the return premium has already been competed away. Wensley (1982) questioned the assumption of some scholars that the marketplace ignores underlying regularities governing business performance uncovered in profitability models. He asserted that the observed regularities that serve as the basis for, or confirmation of, strategic theory also form the basis for action. Wensley emphasized the importance of private, as opposed to widely available (public), information as the primary source of comparative advantage and profit opportunities. He indicated that researchers should expect that effects of first-order significance will already be acted upon by the market participants. Having found a statistically significant relationship in a profitability model, Wensley contended that theorists need to be aware of possible misinter-

The .Austrian' School of Strategy 1992

Jacobson

289 795

pretation before concluding that this relationship is evidence of blatant market inefficiency. UNOBSERVABLE FACTORS Critics of the Austrian approach often argue that because its proponents place great emphasis on unobservable factors they are inherently incapable of saying anything about observable factors. 6 Austrians believe that the truth is the other way around. Their methods provide insight about the real world because they take into account unobservables. Kirzner (1976: 7) noted, "The real world includes a whole range of matters beyond the scope of the measuring instruments of the econometrician. Economic science must be able to encompass this realm." Rizzo (1978) indicated that not all issues or variables of interest are quantifiable. Explaining complex phenomena only by reference to quantifiable variables neglects available information, which Austrians believe must be considered in order to obtain insights into business performance. Invisible Assets According to Itami and Roehl (1987: 1), "Analysts have tended to define assets too narrowly, identifying only those that can be measured, such as plant and equipment." They noted that "invisible assets" (e.g., a particular technology, accumulated consumer information, brand name, reputation, corporate culture, and management skill) are the real source of comparative advantage. How strategies affect and are affected by these invisible assets influences a firm's competitive success. Similarly, Bonoma (1985) suggested that certain areas of interest to marketers simply defy counting approaches; that is, they are not amenable to quantification. He indicated, for example, the nature of good practice in marketing management and the coordination of marketing activities with other business functions are currently unquantifiable. They are so complex at this time it is impossible to know what to count. These invisible assets are key success factors because they are difficult to obtain. Accumulation of these assets requires ongoing, conscious, time-consuming, and uncertain efforts. For example, there is no 6 The general criticism of Austrian economics is that although it highlights shortcomings in traditional analysis, it fails to provide a framework for making a positive contribution (Klein, 1975). In particular, disequilibrium analysis does not have the degree of precision. completeness, or lack of ambiguity characterized by equilibrium models. Even though Austrians counter that this uncertainty characterizes "the real world." traditional economists view equilibrium models as a valuable first step or organizing framework. The reluctance to use analytical tools or to develop testable propositions is also viewed as a deficiency in the Austrian approach. Research that has embraced the Austrian framework (e.g.. Nelson & Winter, 1982) lessens the validity of these criticisms as inherent flaws of adopting an Austrian perspective.

290

R.Jacobson

796

Academy of Management Review

October

easy way to obtain a desired corporate culture (Barney, 1986a). This is something that competitors, even those with substantial resources, cannot buy or readily obtain. Therefore, it is these invisible assets that are likely to have the greatest and longest lasting impact on performance (Reed & DeFillippi, 1990; Winter, 1987). While acknowledging the existence of unobservable factors and their effects on profitability, strategy researchers contend that modeling the linkages among observed strategic factors provides reasonable approximations of business performance. The assumption is that the effect of observed strategic factors can be assessed irrespective of the firm's unobserved strategic factors. Buzzell and Gale (1987: 50-51) stated, "If we knew literally all the factors that influence profits, we could explain 100% of the variation. How close can we get to this standard of perfection? The Par ROI [return on investment] model explains over 70%. " 7 However, profitability models, such as the Par ROI model, that ignore unobservable factors should be viewed cautiously. In a world with imitation and competition dissipating abnormal profits, adapting to a changing environment is crucial for a firm. Invisible assets help to position a firm to exploit new opportunities. Given a firm's invisible assets, some strategy choices will prove more effective than others. Strategies developed in the absence of an appreciation of these invisible assets are unlikely to prove successful. As such, studies that fail to control for a firm's invisible assets are unlikely to assess accurately the influence of observed strategic factors on business performance. Yet, most empirical studies of business performance, both in industrial organization and in strategy, ignore the role of the unobserved factors. The Extent of Influence of Unobservable Factors Making use of statistical techniques designed to help control for unobservable factors, Jacobson (1990) finds that a serial correlation model of the form: ROI¡,

=

ßXit

+ €¡( with €i( = peiM + η„

approximates the manner in which ROI is associated with strategic factors (Xi() and unobserved factors (eit). The fact that the unobservable factors are autocorrelated induces serially correlated residuals, as depicted by the coefficient ρ in the structural equation. As discussed by Hsiao (1986), for example, the lagged value y it .! is informative in the structural equation's reduced form solution, that is, the estimating equation Yu = PYu-i

+ ß*i» - ßP*i(-j + η ί(

7 The Par ROI is a multivariable regression equation developed by the Strategic Planning Institute to ascertain what the expected ROI of a business should be if its market environment, competitive position, and marketing strategy are considered.

The .Austrian' School of Strategy 1992

Jacobson

291 797

because it helps predict the effect of the autocorrelated unobservable factors. The fact that the unobserved factors are correlated with some of the observed strategic factors results in biased estimates of the effect of strategic factors and an overstatement of their explanatory power. The influence of these unobservable factors is so pervasive as to invalidate many of the conclusions drawn from studies failing to control for their effects. In fact, Jacobson (1990) showed that failure to control for firm-specific latent factors can result in a bias so substantial as to reverse the sign of the estimated effect of strategic factors. For example, the negative correlation between ROI and marketing intensity reported in some PIMS-based studies (e.g., Buzzell & Gale, 1987) results not from an adverse impact of marketing expenditures but rather from the failure to control for unobservable effects. Unobservable effects that shift the demand curve, for example, jointly influence ROI and sales (the denominator in the marketing intensity measure) and so induce a spurious negative association. Once these unobservable effects are controlled, marketing intensity is estimated to have a positive influence on ROI. Further, although observed strategic factors have a statistically significant influence on profitability, it is the unobservable factors that for the most part determine business performance. The failure of observed strategic factors to accurately predict future business performance can be seen by assessing the predictive performance of the Strategic Planning Institute's Par ROI estimate. Contrary to the assertions of the ability of the Par ROI equation to explain 70 percent of the variation in ROI, when researchers take into account the role of unobservable factors, the Par ROI model is able to forecast about 1 percent of the year-to-year variance of business performance. The explanatory power of the Par ROI model is largely attributable to the failure to account for serially correlated residuals generated by autocorrelated unobservable factors influencing business performance.8

8 Failing to deal with the potential problems highlighted by serially correlated residuals has led to a number of classic articles that warn of and illustrate the misuse of econometric methods (e.g.. Hendry. 1980; Plosser & Schwert. 1978; Yule. 1926). The consequences of ignoring the warning signs provided by serially correlated residuals are perhaps best illustrated by the work of levons (1884). levons developed what is known as the quantity theory of sunspots, that is, that sunspot activity causes economic activity. The empirical evidence for this relationship seems overwhelming. A 1 percent increase in sunspots is associated with a .72 percent increase in income. The association is "significant" at the 99.9 percent confidence level. The fl2 value suggests that sunspot activity explains 82 percent of the variation in income. But. of course, this correlation is spurious. It merely reflects the fact that both income and cumulative sunspot activity are increasing over time. The correlation of income with sunspots vanishes once a serial correlation correction is made to the model. Many empirical regularities reported in the strategy literature seem to be based on a similar misuse of statistical methods.

292 798

R.Jacobson Academy oí Management Review

October

AN AUSTRIAN ECONOMICS-BASED STRATEGY AGENDA Austrian economics raises perspectives and, therefore, managerial and strategic research directions that are different from those of traditional industrial organization. With respect to the dimensions delineated in Table 1, continuous innovation, strategic flexibility, factors influencing the speed of the market process, and unobservable factors useful in adapting to a dynamic environment, for example, are agenda items for Austrian-based strategy. Each of these topics deals with the distinct feature of Austrian economics of the market process. Entrepreneurial Discovery: Continuous Innovation The Austrian School highlights entrepreneurial discovery/innovation as central to business success. Although it includes more than R&D, this perspective puts great emphasis on R&D. Interestingly, some scholars have commented that it seems that the more R&D-based innovations U.S. firms develop, the further they ultimately end up falling behind foreign, in particular, Japanese, imitators because of the efficiency and speed of these competitors (Mansfield, 1988a,b). Each new innovation provides another opportunity for a foreign rival to out-imitate a U.S. company in producing it. The success of the Japanese firm has shown that profitable innovation is obtained not just through major breakthroughs but also through continuous improvements. Indeed, the limitation of attempting to compete on the basis of major breakthroughs is that it confers a comparative advantage but only until this breakthrough is imitated by competitors. This is the Schumpeterian notion of the market equilibrium being disrupted by innovations and then returning to equilibrium. To regain an advantage, the firm must develop a new breakthrough product, which is an uncertain, time-consuming, and expensive process. Some firms have realized the benefits of incremental processoriented R&D (Business Week. 1989). NEC, for example, is said to view innovation as the result of tiny improvements in a thousand places. As long as the firm continues to undertake incremental improvements, it (because of lags in competitive response) continues to have a comparative advantage. However, even this comparative advantage will dissipate a s major breakthroughs circumvent the incremental processgenerated gains. Still, managers at Ford Motor Co. contend that the cumulation of a large number of small improvements in products and processes is the surest path, in most industries, to competitive advantage. The importance of continuous innovation is in keeping with the Austrian tradition of entrepreneurial discovery and its characterization of markets as disequilibrium processes. Companies, even when they develop a major breakthrough innovation, often credit their success to the ability to continually refine the original product and production processes. For example, when Sony in-

The .Austrian' School of Strategy 1992

Jacobson

293 799

troduced the Walkman in 1980. creating a new market, imitators quickly reacted. Not only did Sony's market share steadily decline to a low of approximately 18 percent in 1983, but its sales actually declined during that year. Sony was able to reverse this trend, increasing its market share to approximately 40 percent and tripling its sales, because of a stream of new and better models and improved process techniques. In particular, it cut the number of parts and the assembly time for a Walkman in half. Disequilibrium: Adapting to Change Even though strategy researchers have emphasized the effect of strategy on the level of return, as if a successful strategy ensured constant supranormal returns over time, studies of the time series properties of ROI report that it can be characterized as a mean-reverting process (Beaver, 1972; Lookabill, 1976). That is, firms' abnormal returns dissipate over time. This is an expected consequence of the market process, and it supports the characterization of profits as a disequilibrium phenomenon. Investors tend to move funds toward areas generating supranormal profits and away from areas generating less than normal profits. Because the returns to a given strategy dissipate, firms must adapt and respond to changing conditions. As such, flexibility becomes a critical strategic factor. Indeed, a number of firms have effectively used flexibility as a key strategy. Motorola is one such firm: for example, its customized pocket pagers go into production 17 minutes from the time of order, are shipped within 2 hours, and are received by the customer the next day. Much of Compaq Computer's success can be traced to its ability to respond quickly to and take advantage of others' innovations. One of Honda's advantages over other auto makers is its ability to bring a new model to market within three years, as opposed to four to five years required for its competitors. Generally, because flexibility allows small firms to respond more quickly to changes in industry demand, they can often compete effectively with large firms, even in the face of higher minimum average cost. Taiwanese firms, for example, use flexibility as a strategy when competing with much larger competitors, in particular, Korean firms. Although Korean firms tend to focus on high volume, high productivity, and standardized products, Taiwanese firms emphasize the advantages brought about by flexibility in responding quickly to changing market conditions and in marketing nonstandardized products. Indeed, Taiwanese companies provide a good illustration of Austrian economics' depiction of entrepreneurial firms dissipating profits. Levy (1988) noted, "Taiwanese firms nip at the heels of innovators, each firm content with a small market share of some new, highly profitable market even as they drive down returns for the industry as a whole." De Meyer, Nakane, Miller, and Ferdows (1989) asserted that flexibility will be the next competitive battleground as manufacturers strive to overcome the trade-offs that have existed in the past between flexibility and

294 800

R.Jacobson Academy oí Management Review

October

cost efficiency. Their survey of Japanese manufacturers indicated an emphasis on cost-e/íicienf flexibility. Japanese respondents cite rapid design changes as their second highest competitive priority (behind low prices) and rate flexible manufacturing systems as the most important action plan. De Meyer and his colleagues concluded that U.S. and European manufacturers that thought they would be on par with their Japanese competitors once they were able to compete on product quality will come to find that the next form of competition will be waged on overcoming the efficiency-flexibility trade-off. Obtaining superior business performance will require new entrepreneurial discovery (e.g., efficiently implementing flexibility), as the market process continues to dissipate the benefits firms realize by offering high-quality products. Empirical Regularities: The Persistence of Returns The usefulness of empirical models of business performance, even those controlling for unobservable effects, is limited because the market process leads firms to adjust to and, therefore, to dissipate the effects of strategies shown to lead to supranormal profits. These models describe what happened in the past, not what will happen in the future. The models have value to the extent that they point out regularities that may have been overlooked by firms or suggest how a firm might use its unique resources or skills. Empirical methods also might be used to assess the aspects of the market process (e.g., how strategic factors influence the speed of dissipation of abnormal returns). Miller and Rock (1985) noted that the effect of an earnings announcement on stock price (i.e., shareholder value, a primary performance criterion measure) depends not only on the magnitude of the return, but also on its persistence. The greater the persistence (i.e., the slower abnormal profits dissipate over time), the greater the effect. Therefore, the market process determining the persistence of return is an important strategic consideration. Jacobson (1988) found that market share and marketing expenditures, for example, tend to slow the decay of ROI. A number of other factors can be expected to influence this process. For example, even though attention has been directed at the effect of order-or-entry (pioneering) on the level of return, its effect might be additionally, or even exclusively, hypothesized to influence the persistence of return. Jacobson (1991) reported, based on analysis of PIMS data, that the first-order autoregressive process that characterizes the time series behavior of ROI for pioneers is more persistent than that of nonpioneers, .81 versus .71, respectively. 9 Further, once the statistically significant difference in the autoregressive coefficient for pioneers versus nonpioneers was taken into account, the 9 An AR(1) coefficient of 1.00 indicates no dissipation of ROI; a coefficient of 0.00 indicates immediate (within-year) dissipation.

The , Austrian' School of Strategy

1992

Jacobson

295

801

coefficient indicating the direct effect of pioneering on ROI became statistically insignificant. How strategies and tactics influence the persistence of returns is an area for future research. The types of R&D used after the initial innovation (e.g., product- versus process-oriented innovation) might be one possibility. Unobeervable Factors: Monitoring and Control Griliches (1974) noted that unobservable factors can be viewed as including unobserved and hard to proxy, as well as truly unobservable, variables. Peters (1987: 488) maintained that "our fixation with financial measures leads us to downplay or ignore less tangible nonfinancial measures, such as product quality, customer satisfaction, order lead time, factory flexibility, the time it takes to launch a new product, and the accumulation of skills by labor over time. Yet these are increasingly the real drivers of corporate success." Some of these essential yet typically "unobservable" factors can be measured to some extent. Hansen and Wernerfelt (1989) reported that their measures of organization factors (e.g., corporate culture) explained twice the amount of variation in ROI as did economic factors. Narver and Slater (1990) constructed a measure of market orientation and found that it exhibited a significant positive association with business performance. Latent factors such as corporate culture and market orientation can be monitored over time. Increased attention to these unobservable factors, which can be expected to influence long-term business success because they are difficult to copy or imitate, can lead to reduced reliance on measures such as ROI that encourage a short-term orientation. Further, the use of such factors can provide an indication of the firm's ability to adjust to a dynamic environment and to exploit market opportunities. It is reasonable to conclude, however, that important strategic factors influencing business performance will never be measured completely or without error. Such important factors as corporate culture are fuzzy and ill-defined. The difficulties in quantifying the unobserved strategic factor luck seem insurmountable. Yet, it can have important and persisting effects on business performance and on observed strategic factors (Alchian, 1950; Mancke, 1974). Therefore, the question remains as to how a researcher might control for unobservable factors in empirical models in order to assess the previous effect of strategic factors. Hausman and Taylor (1981: 137), articulating current econometric practices, suggested the benefits of panel data. They noted: An important benefit from pooling time-series and crosssectional data is the ability to control for individual specific effects—possibly unobservable—which may be correlated with the other included variables in the specification of an economic relationship. Analysis of cross-sectional data alone can neither identify nor control for such effects.

296 802

R.Jacobson Academy of Management Review

October

This current perspective places into question many oí the conclusions drawn from empirical strategy research, which is based almost exclusively on cross-sectional analysis. This stream of research has not included the effects of unobservable factors. Further, it suggests that future empirical strategy research should emphasize the use of panel data and statistical methods that help control for the influence of unobservable factors. CONCLUSION Perhaps the closest correspondence between Austrian economics and "established" strategic perspectives pertains to the concept of strategic windows. Advocating dynamic rather than static analysis, Abell (1978) noted that frequently market changes are so far-reaching that the competence of the firm to continue to compete effectively is called into question. There are only limited periods (i.e., a strategic window) during which the fit between the key requirements of a market and the particular competencies of a firm competing in that market is at an optimum. Investment activities should coincide with periods in which such a strategic window is open, and divestment should be contemplated if what was once a good fit has eroded. Broadly speaking, Austrian economic theory depicts the business environment as a continual process of strategic windows.10 Opportunities constantly arise and then dissipate. Although it includes a wider view of the concept, Austrian economics can be regarded as the theoretical foundation for strategic windows. In addition, some more recent strategy research, though seemingly disconnected, is distinctly Austrian in nature. Rumelt's (1984) and Levitt's (1986) emphasis on entrepreneurial activities, Peters' (1987) and Bhide's (1986) depictions of market dynamics and disequilibrium. Itami and Roehl's (1987) and Jacobson's (1990) concern for unobservable factors, Wensley's (1982) view on learning dissipating empirical regularities in profitability models, for example, fit squarely into the Austrian school of thought. To a large extent, these analysts are articulating the Austrian perspective as it pertains to strategic issues. Thus, their research can be seen as forming an "Austrian School of Strategy." Because Austrian concepts have been incorporated by other schools of thought, a variety of strategic frameworks share commonalities with Austrian perspectives. This occurrence is consistent with the perspective of Milton Friedman that "there is no Austrian economics—only good economics and bad economics" (cited in Dolan, 1976:4).11 Friedman indicated 10 Peters (1987) expounded on this notion of continuous strategic windows. He contended that chaos and uncertainty are—will be—market opportunities for the wise and that capitalizing on fleeting market anomalies will be the successful business's greatest accomplishment. 11 Kirzner (1989), however, discussed the advantage of the Austrian label.

The ,Austrian' School of Strategy 1992

Jacobson

297 803

that what is good about Austrian economics would be incorporated in generally accepted mainstream analysis. So it is not surprising that Austrian thought and the resource-based theory of the firm, in particular, exhibit a great many similarities. For example, theorists from both schools of thought place similar emphasis on firm heterogeneity and invisible assets. They differ from Schumpeter in noting the role of incremental improvements in generating profits (Conner, 1991). However, differences also exist. In particular, the resource-based perspective does not focus on the market process but rather makes use of equilibrium concepts and analysis. For example, Barney (1986a, 1991) emphasized sustained competitive advantage (where current and potential competitors are unable to duplicate a strategy) as opposed to temporary (i.e., dissipating) competitive advantage, whose importance has generally been downplayed in the strategy literature. Austrians do not make this distinction. Strategies that can be duplicated have value in that lags in imitation allow the firm to earn supranormal profits. Indeed, Austrians emphasize the gradual dissolution of all entrepreneurial profit as more and more people discover the knowledge needed to imitate a strategy. A further integration of Austrian perspectives into resource-based theory would involve placing greater emphasis on, for example, organizational learning (Stata 1989), the importance of imitable strategies as sources of profits, and analysis of the market processes influencing the erosion of supranormal profits. As more markets become subject to intense foreign competition, rapid technological change, and shorter product life cycles, even many former advocates of ΙΟ-based strategy have begun to articulate views that move closer to Austrian thinking. However, inconsistencies can arise when attempting to integrate other frameworks with Austrian paradigms. For example, Porter failed to reconcile his continued advocacy of restricting competitive forces as a means of achieving higher profits (1990: 35) with his conclusion that a company should actively seek out pressure and challenges, not try to avoid them, to provide sufficient pressure for innovation (1990: 586).12 Even though in his early view he advocated attempting to limit the power of customers and competition, he later maintained that a home base containing demanding buyers with stringent needs and able competition is a distinct advantage to a firm. Though it has a number of different dimensions, the strategic implications from Austrian economics are closely related, if not saying the same thing. Rumelt (1988) commented that if a phenomenon is understood well enough to model it, it is too late to make money from it. Opportunities for profit are rooted in private information, ambiguity, special situations, and entrepreneurial insights. Thus, no general laws of business exist.

12 As discussed previously, Austrians believe the possibility of earning supranormal profits provides the incentive ior innovation.

298 804

R.Jacobson Academy of Management

Review

October

Rather, business success is a "science oi the specific" that depends on private information and specialized resources. The key strategic resources of a firm are the special knowledge and skills of its employees and the perceptions of its customers. These strategic insights from Austrian economics offer perspectives that need to be more fully incorporated into future strategic thinking and research. REFERENCES Abell, D. F. 1978. Strategic windows. Journal of Marketing. 42(3): 21-26. Alchian, A. A. 1950. Uncertainty, evolution, and economic theory. Journal of Political Economy. 58: 211-221. Bain, I. S. 1951. Relation of profit rate of industry concentration: American manufacturing, 1936-1940. Quarterly Journal of Economics. 65: 293-324. Barney, I. 1986a. Organizational culture: Can it be a source of sustained competitive advantage? Academy of Management Review, 11: 656-665. Barney, I. 1986b. Strategic factor markets: Expectations, luck and business strategy. Management Science. 32: 1231-1241. Barney, J. 1991. Firm resources and sustained competitive advantage. Journal of Management. 17(1): 99-120. Beaver. W. H. 1970. The time series behavior of earnings. Journal of Accounting Research. [Supplement: Empirical Research in Accounting: Selected Studies] 8: 62-69. Bhide, A. 1986. Hustle as strategy. Harvard Business Review. 64(5): 59-65. Bonoma, T. V. 1985. Case research in marketing: Opportunities, problems, and a process. Journal of Marketing Research. 22: 199-208. Business Week. 1987. Detroit is trying, but its image lags. June 8: 138-139. Business Week. 1989. Innovation in America. [Special 1989 Bonus Issue]. Buzzell, R. D., & Gale, Β. T. 1987. The PIMS principles. New York: Free Press. Caves, R. E., & Porter, M. E. 1977. From entry barriers to mobility barriers: Conjectural decisions and contrived deterrence to new competition. Quarterly Journal of Economics, 91: 421-441. Conner, K. R. 1991. A historical comparison of resources-based theory and five schools of thought within industrial organization economics: Do we have a new theory of the firm? Journal of Management. 17(1): 121-154. De Meyer, Α., Nakane, J., Miller, J. G., & Ferdows, K. 1989. Flexibility: The next competitive battle. The manufacturing futures survey. Strategic Management Journal. 10: 135-144. Demsetz, H. 1973. Industry structure, market rivalry, and public policy. Journal of Law and Economics. 16: 1-9. Dolan, E. G. 1976. Austrian economics as extraordinary science. In E. G. Dolan (Ed.), The foundations of modern Austrian economics; 3-15. Kansas City, MO: Sheed & Ward. Economist. 1991. When GM's robots ran amok. August 19: 64-65. Gluck, F. W., Kaufman, S. P., & Walleck, A. S. 1980. Strategie management for competitive advantage. Harvard Business Review. 58(4): 154-161. Griliches, Z. 1974. Errors in variables and other unobservables. Econometrica, 42: 971-998.

The .Austrian' School of Strategy 1992

Jacobson

299 805

Hansen. G. S., & Wernerfeit. B. 1989. Determinants of firm performance: The relative importance of economic and organizational factors. Strategic Management Journal, 10: 399411. Harvard Business Review. 1987. The pathetic gamble. 65(3): 90. Hausman, J. Α., & Taylor, W. E. 1981. Panel data and unobservable individual effects. Econometrica. 49: 1377-1397. Hayek, F. A. 1937. Economics and knowledge. Economica. 3: 33-54. Hayek, F. A. 1945. The use of knowledge in society. American Economic Review. 35:519-530. Hayek, F. A. 1948. The meaning of competition. In Individualism and economic order: 92106. Chicago; University of Chicago Press. Hayes, R. H., & Abernathy, W. I. 1980. Managing our way to economic decline. Harvard Business Review. 58(4): 67-77. Hendry, D. F. 1980. Econometrics—Alchemy or science? Economica. 47: 387-406. Hsiao, C. 1986. Analysis oí panel dala. Cambridge: Cambridge University Press. Itami. H. with Roehl, T. W. 1987. Mobilizing invisible assets. Cambridge, MA: Harvard University Press. lacobson, R. 1988. The persistence of abnormal returns. Strategic Management Journal. 9: 415-430. lacobson, R. 1990. Unobservable effects and business performance. Marketing Science, 9: 74-85. Jacobson, R. 1991. A critique ot empirical models of business performance. Paper presented at the 1991 American Marketing Association Doctoral Consortium, Los Angeles. Jevons, W. S. 1884. Commercial crisis and sunspots. In H. S. Foxwell (Ed.), Investigations in currency and finance: 221-243, London: Macmillan. Kirzner, I. M. 1973. Competition and entrepreneurship. Press.

Chicago: University of Chicago

Kirzner, I. M. 1976. On the method of Austrian economics. In E. G. Dolan (Ed.), The foundations of modern Austrian economics: 40-51, Kansas City. MO: Sheed & Ward. Kirzner, I. M. 1979. Perception, opportunity, and profit. Chicago: University of Chicago Press. Kirzner. I. M. 1981. The "Austrian" perspective. In D. Bell & I. Kristol (Eds.). The crisis in economic theory: 111-122. New York: Basic Books. Kirzner. I. M. 1989. The use of labels in doctrinal history. Cato Journal. 9: 231-235. Klein. B. 1975. Book review: Competition and entrepreneurship. Journal of Political Economy. 83: 1305-1306. Kristol, I. 1988. Voodoo economics or voodoo economists? Wall Street Journal. October 18: 28. Levin, R. C., Klevorik. A. K„ Nelson, R. R„ & Winter, S. G. 1987. Appropriating the returns from industrial research and development. Brookings Papers on Economic Activity: 783820.

Levitt, T. 1986. The marketing imagination. New York: Free Press. Levy, B. 1988. Korean and Taiwanese firms as international competitiors: The challenges ahead. Columbia Journal of World Business, 23(1): 43-51. Lippman, S. Α., 8c Rumelt, R. P. 1982. Uncertainty imitability: An analysis of interfirm differences in efficiency under competition. Bell Journal of Economics. 13: 418-453.

300 806

R.Jacobson Academy

of Management

Review

October

Littlechild, S. C. 1978. The fallacy of (he mixed economy. Sussex. England: The Institute oi Economic Affairs. Lookabill, L. L. 1976. The time series properties of accounting earnings. The Accounting Review. 51: 724-738. Lucas. R. E. 1981. Econometric policy evaluation: A critique. In R. E. Lucas (Ed.), Studies in business cycle theory: 104-130, Cambridge, MA: MIT Press. Mancke, R. B. 1974. Causes of interfirm profitability differences: A new interpretation of the evidence. Quarterly Journal of Economics, 88: 181-193. Mansfield. E. 1988a. Industrial R&D in lapan and the United States: A comparative study. American Economic Review Papers and Proceedings. 78: 223-228. Mansfield, E. 1988b. The speed and cost of industrial innovation in lapan and the United States: External vs. internal innovations. Management Science. 34: 1157-1168. Mansfield, E., Schwartz, M., & Wagner, S. 1981. Imitation costs and patents: An empirical study. Economic Journal. 91: 907-918. Miller, M. H.. 8c Rock, K. 1985. Dividend policy under asymmetric information. Journal Finance. 40: 1031-1051.

of

Mises, L. 1949. Human action; A treatise on economics. New Haven. CT: Yale University Press. Narver, J. C.. 8c Slater, S. F. 1990. The effect of a market orientation on business profitability. Journal of Marketing. 54: 20-36. Nelson, R. 1976. Goldschimd, Mann, and Weston's industrial concentration: The new learning. Bell Journal of Economics. 7: 729-732. Nelson, R.. 8c Winter, S. 1982. An evolutionary Belknap Press. Peters, T. 1987. Thriving on chaos: Handbook

theory of economic change. Cambridge, MA: for a management

revolution. New York: Knopf.

Plosser, C. I., 8c Schwert, G. W. 1978. Money, income, and sunspots: Measuring economic relationships and the effects of differencing. Journal of Monetary Economics. 4: 637-660. Porter, M. E. 1980. Competitive

strategy.

New York: Free Press.

Porter. M. E. 1981. The contributions of industrial organization to strategic management. Academy of Management Review. 6: 609-620. Porter. M. E. 1990. The competitive advantage

of nations. New York: Free Press.

Reed, R.. 8c DeFillippi, R. 1990. Causal ambiguity, barriers to imitation, and sustainable competitive advantage. Academy of Management Review, 15: 88-102, Rizzo, M. 1978. Praxeology and econometrics: A critique of positive economics. In L. M. Spadro (Ed.). New directions in Austrian economics: 40- 56. Kansas City. MO: Sheed, Andrews. 8c McMeel. Roll, R. 1984. Orange juice and weather. American

Economic

Review.

74: 861-880.

Rumelt, R. 1984. Towards a strategic theory of the firm. In B. Lamb (Ed.). Competitive tegic management: 556-570. Englewood Cliffs. NJ: Prentice Hall.

stra-

Rumelt, R. 1987. Theory, strategy, and entrepreneurship. InD. ]. Teece(Ed.), The competitive challenge: Strategies for industrial innovation and renewal: 137-159. Cambridge, MA: Ballinger. Rumelt. R. 1988. Competitive marketing strategy. Panel discussion, ORSA/TIMS Marketing Science Conference, Seattle, WA. Schoeffler, S. 1977. Nine basic findings on business strategy. The P1MSLETTER on Business Strategy. (1): 1 - 7 .

The , Austrian' School of Strategy 1992

lacobson

301 807

Schumpeter, J. A. 1934. The theory of economic development. Cambridge, MA: Harvard University Press. Schumpeter, I. A. 1942. Capitalism, socialism a n d democracy. New York: Harpers. Solow, R. 1957. Technical c h a n g e a n d the a g g r e g a t e production function. Review of Economics and Statistics. 39: 312-320. Stata, R. 1989. Organizational learning—The key to management learning. Sloan Management Review. 30(3): 63-74. Teece, D. 1987. Profiting for technological innovation: Implications for integration, collaboration, licensing, a n d public policy. In D. Teece (Ed.), The competitive challenge: Strategies for industrial innovation a n d renewal; 185-219. Cambridge, MA: Ballinger. Wensley, R. 1982. PIMS a n d BCG: New horizons or false dawn. Strategic Management Journal. 3: 147-158. Winter, S. G. 1987. Knowledge a n d competence a s strategic assets. In D. Teece (Ed.), The competitive challenge: Strategies for industrial innovation and renewal: 159-184. Cambridge. MA: Ballinger. Yule, G. U. 1926. Why do we sometimes get nonsense correlations between time series? Study in sampling a n d the nature of time series. Journal of the Royal Statistical Society. 89(1): 1-64. Robert Jacobson received his Ph.D. in economics from the University of California, Berkeley. He is a professor of marketing at the School of Business Administration, University of Washington. His research interests a r e in the a r e a of business strategy.

Teil V Interne Umweltanalysen

Während sich externe Umweltanalysen auf den OT-Teil der SWOT-Analyse beziehen, beschäftigen sich interne Umweltanalysen mit dem SW-Teil. Nach Maßgabe der präskriptiven Strategieliteratur sollen die Stärken und Schwächen der internen Umwelt der Unternehmung untersucht werden, bevor (nach einem Vergleich mit den Ergebnissen der externe Umweltanalyse) Strategien selektiert werden (Welge & Al-Laham, 1992: 109-131; Wheelen & Hunger, 1995: 114-140). Ein .Strategic Fit' ist herzustellen zwischen den Gegebenheiten der internen Umwelt und der Unternehmungsstrategie. Die Suche nach internen Stärken und Schwächen hat in der Praxis der strategischen Planung eine lange Tradition. Am geläufigsten sind fiinktionsbezogene Analysen der Ressourcen und Fähigkeiten einer Unternehmung, d.h. für wichtige Unternehmungsfunktionen (z.B. F&E, Produktion, Marketing, Finanzierung) werden Kompetenzprofile erstellt (Ansoff, 1988: 66-71). Ein weiteres, oft verwendetes Verfahren zur Stärken- und Schwächenanalyse verwendet Checklisten, deren Inhalte sich aus den Ergebnissen der PIMS-Untersuchungen ableiten (Rowe et al. 1989: 129f.). PIMS-basierte Analysen sind aufgrund der diesen Untersuchungen inhärenten Probleme allerdings etwas in Ungnade gefallen (vgl. z.B. Chussil, 1991; Schwalbach, 1991b). Schließlich haben sich auch Wertkettenanalysen zur Untersuchung interner Stärken und Schwächen durchgesetzt. Dieses von Porter (1985: 33-61) eingeführte Verfahren untersucht die von einer Unternehmung durchgeführten Aktivitäten, die sich mit Hilfe des Konzepts der Wertkette klassifizieren und anordnen lassen. Stellt man die Frage nach einem theoretischen Fundament der internen Umweltanalyse, so stößt man sehr schnell auf den ressourcenbasierten Ansatz (RBA) des Strategischen Managements. Der gegenwärtige Stand der Diskussion hinsichtlich des ressourcenbasierten Ansatzes wird in Kapitel 7 dargestellt.

7 Ressourcen und Fähigkeiten der Unternehmung

Während die Bedeutung der internen Ressourcen einer Unternehmung schon frühzeitig von der ökonomischen Theorie (Penrose, 1959) und der strategischen Managementliteratur erkannt wurde (Christensen et al., 1982; Rumelt, 1984), war es Birger Wernerfeit (1984), der dem RBA als potentielle Grundlage für die Forschungen des Strategischen Managements den Weg bereitete.29 Wernerfeit bezieht sich in seinem grundlegenden Artikel, , A Resource-based View of the Firm", nicht so sehr auf die Traditionen der präskriptiven strategischen Managementliteratur, als vielmehr auf industrieökonomische Arbeiten (Caves, 1980; Porter, 1980), die er durch eine Ressourcenperspektive ergänzen will. Der RBA wird heute oft als eine Alternative zu industrieökonomischen Konzepten im Strategischen Management angesehen, da er von einer grundsätzlichen Unterschiedlichkeit aller Unternehmungen ausgeht und diese nicht, wie die traditionelle industrieökonomische Forschung, als gleich behandelt (Rumelt, 1991; Knyphausen, 1993). Einige Autoren fordern sogar, den RBA als eine neue, realistischere Theorie der Unternehmung anzusehen (Conner, 1991; Kogut & Zander, 1992; Mahoney & Pandian, 1992).30

Argumentationslogik des ressourcenbasierten Ansatzes Die Vertreter des RBA versuchen, den Erfolg einer Unternehmung und die Performanceunterschiede zwischen mehreren Unternehmungen einer Branche durch das Vorhandensein firmenspezifischer, einzigartiger Ressourcen zu erklären. Inzwischen liegen mehrere empirische Untersuchungen vor, die einen Zusammenhang zwischen spezifischen Ressourcen und dem Unternehmungserfolg bestätigen (Barnett et al., 1994; Henderson & Cockburn, 1994; Mehra, 1996; Miller & Shamsie, 1996). Durch den Aufbau und die Nutzung solcher einzigartigen Ressourcen sollen Unternehmungen in der Lage sein, sich dauerhafte Wettbewerbsvorteile zu sichern. Die Nutzung des Ressourcenpotentials wird somit zur Quelle von Wettbewerbsvorteilen, und sie ersetzt Maßnahmen zur Wettbewerbsbeschränkung (z.B. Eintritts- und Mobilitätsbarrieren), die in den industrieökonomischen Ansätzen Wettbewerbsvorteile begründen (Barney, 1991; Teece et al., 1994).

29

30

Der RBA hat inzwischen auch die Praxis des Strategischen Managements durchdrungen. Hierfür waren insbesondere Prahalad und Hamel (1990) mit ihrem Harvard Business Review-Artikel, „The Core Competence of the Corporation", verantwortlich (Wemerfelt, 1995). Der Anspruch, eine neue „theory of the firm" darzustellen, ist natürlich nicht unumstritten, da die Ökonomie mit dem Transaktionskostenansatz (Coase, 1937; Williamson, 1975) ein vielbeachtetes Theoriegebäude zur Verfügung stellt. Diese Kontroverse ist „ongoing", wie einige jüngere Veröffentlichungen in der Zeitschrift Organization Science belegen. Vgl. Foss, 1996a; 1996b; Conner & Prahalad, 1996; Kogut & Zander, 1996.

306

R. Κ. F. Bresser

Der Ressourcenbegriff selbst ist durchaus noch schillernd und von der Literatur nicht eindeutig definiert. Für Wernerfeit (1984: 172) sind Ressourcen „anything which could be thought of as a strength or weakness of a given firm". Eine wichtige Untergliederung besteht in der Unterscheidung zwischen materiellen und immateriellen Ressourcen (Wernerfeit, 1984; Rasche & Wolfrum, 1994). Beispiele für materielle Ressourcen sind Produktionsanlagen, der Standort, EDV-Systeme, usw. Immaterielle Ressourcen umfassen u.a. die Organisationskultur, Patente, die Reputation, technisches Know-how und organisatorische Fähigkeiten, z.B. Koordinations-, Integrations- und Lernfähigkeiten. Für viele immaterielle Ressourcen versagen die Faktormärkte, denn solche Ressourcen sind oft intangibel oder „tacit", d.h., sie sind nicht handelbar, sondern können nur in Unternehmungen selbst entwickelt werden und sind somit unternehmungsspezifisch (Dierickx & Cool, 1989; Knyphausen, 1993). Das Vorhandensein dieser nichthandelbaren, unternehmungsspezifischen Ressourcen erklärt, warum Unternehmungen sich voneinander unterscheiden. Als zwei grundlegende Prämissen des RBA gelten somit die Ressourcenheterogenität und die Ressourcenimmobilität (Barney, 1991). Die Prämisse der Ressourcenheterogenität besagt, daß Unternehmungen durch asymmetrische Ressourcenausstattungen gekennzeichnet sind, denn ein Großteil der Ressourcen einer Unternehmung ist spezifischer Natur. Die Prämisse der Ressourcenimmobilität besagt, daß wichtige, insbesondere immaterielle Ressourcen der Unternehmung nicht handelbar und damit immobil sind. Vor dem Hintergrund der Annahmen der Ressourcenheterogenität und -immobilität stellt sich die Frage, welche Bedingungen erfüllt sein müssen, damit Unternehmungen aus ihren Ressourcen dauerhafte, nachhaltige Wettbewerbsvorteile generieren können. Unter dauerhaften Wettbewerbsvorteilen versteht die RBA-Literatur solche Vorteile, die nicht durch die Imitationen der Konkurrenten zunichte gemacht werden können (Barney, 1991: 103). Offensichtlich enthalten nicht alle Unternehmungsressourcen das Potential für die Schaffung nachhaltiger Wettbewerbsvorteile. In der Literatur werden häufig vier Bedingungen für das Entstehen dauerhafter Wettbewerbsvorteile diskutiert (vgl. z.B. Barney 1991: 105-112; Grant 1991a: 11 Iff.; 1991b: 123ff.; Knyphausen, 1993; Rasche & Wolfrum, 1994). Erstens, Ressourcen müssen einen wertstiftenden Charakter besitzen, d.h. sie müssen es der Unternehmung erlauben, Strategien zu entwickeln und zu implementieren, die die Effizienz und die Effektivität der Unternehmung erhöhen. Zweitens, Ressourcen müssen knapp sein: Wenn jeder Wettbewerber über die gleichen wertstiftenden Ressourcen verfügt, kann keiner daraus einen nachhaltigen Wettbewerbsvorteil ableiten. Drittens, Ressourcen dürfen nicht substituierbar sein, d.h., es darf keine ähnlichen oder alternativen Ressourcen geben, die die zum Aufbau von Wettbewerbsvorteilen notwendigen Leistungen gleichwertig erbringen können. Die vierte Bedingung, die Nichtimitierbarkeit von Ressourcen, ist ex definitione die wichtigste Quelle dauerhafter Wettbewerbsvorteile. Die Ressourcen einer Unternehmung dürfen durch andere Unternehmungen nicht ohne weiteres kopierbar sein. Je weniger Unternehmungen sich vor den Imitationsbemühungen der Konkurrenz schützen können, desto geringer ist das strategische Potential ihrer Unternehmungsressourcen. Die Literatur hebt drei „Imitationsbarrieren" hervor: die Historizität der Unternehmung, die kausale Ambiguität und die soziale Komplexität. Was die Historizität betrifft, so ist jede Unternehmung durch eine idiosynkratische Entwicklung gekennzeichnet, die zu spezifischen Ressourcen (z.B. zu einer einzigartigen Unternehmungskultur) geführt hat. Ressourcen sind typischerweise durch eine „Pfadabhängigkeit" gekennzeichnet, d.h., ihre jeweilige Beschaffenheit ist durch sämtliche

Ressourcen und Fähigkeiten

307

vergangenen Entscheidungen des Managements beeinflußt, ein Umstand der ihre Einzigartigkeit begründet (Dierickx & Cool, 1989; Barney, 1991). Historisch gewachsene Ressourcen sind aufgrund ihrer Einzigartigkeit nicht reproduzierbar oder imitierbar. Die kausale Ambiguität betrifft das Ausmaß der Unklarheit über vermutete Kausalzusammenhänge zwischen den unternehmungsspezifischen Ressourcen und den daraus ableitbaren Wettbewerbsvorteilen (Lippmann & Rumelt, 1982; Barney, 1992). Wenn diese Kausalitäten nicht klar beschreibbar und verständlich sind, fällt es einem Konkurrenten nicht leicht, eine erfolgreiche Strategie zu kopieren, denn er weiß nicht, welche Ressourcen er imitieren soll. Diffuse Kausalzusammenhänge sind insbesondere bei immateriellen Ressourcen mit einem hohen Grad an „tacitness" anzutreffen, z.B. bei organisatorischen Fähigkeiten (Teece et al., 1994). Die soziale Komplexität schließlich wirkt sich ebenfalls oft als Imitationsbarriere aus. Viele Ressourcen (z.B. Organisationskultur, Reputation) bestehen aus zahlreichen Elementen, die interagieren und kaum isoliert voneinander analysiert werden können. Die Komplexität dieser aggregierten Ressourcen erschwert Duplikationsversuche.

Stand der Forschung Der RBA hat zahlreiche empirische Studien und auch weiterführende theoretische Arbeiten inspiriert, z.B. über die Zusammenhänge zwischen der bestehenden Ressourcenbasis und (1) dem Diversifikationsverhalten von Unternehmungen (Montgomery & Hariharan, 1991; Chatterjee & Wernerfeit, 1991; Markides & Williamson, 1994), (2) der Bildung von strategischen Allianzen (Hamel, 1991; Eisenhardt & Schoonhoven, 1996), (3) der Implementierung von Internationalisierungsstrategien (Chang, 1995; Roth, 1995) und (4) der Ökologisierung des Strategischen Managements (Hart, 1995). Darüber hinaus wurden organisatorische Faktoren und ihr Stellenwert als eigenständige Quelle von Wettbewerbsvorteilen untersucht, z.B. hinsichtlich vorhandener Organisationsstrukturen (Powell, 1992; Markides & Williamson, 1996) und Organisationskulturen (Barney, 1986; Fiol, 1991) sowie der Fähigkeiten des Managements (Castanias & Helfat, 1991). Gegenwärtig sind die Vertreter des RBA vornehmlich um die Klärung der Frage nach der Generierung und dem Erlernen neuer Ressourcen bemüht (Black & Boal, 1994). Dieser dynamische Blickwinkel gilt dabei insbesondere den organisatorischen Fähigkeiten (capabilities, routines), denn in der jüngeren Literatur wird diesen „organizational capabilities" ein herausragender Stellenwert für die Entwicklung langfristiger Wettbewerbsvorteile zugesprochen (Teece et al., 1994; 1997; Knyphausen, 1995: 88-114; Grant, 1996). Durch den Einsatz organisatorischer Routinen werden nicht nur alle wichtigen Funktionen der Unternehmung immer wieder erfolgreich durchführbar, der dynamische Charakter solcher Routinen beeinflußt auch die Innovationsfähigkeit der Unternehmung und die Generierung immer neuer Fähigkeiten (Amit & Schoemaker, 1993). Bei der Veränderung von Fähigkeiten ist zum einen zu berücksichtigen, daß neue Fähigkeiten sich in der Regel aus vorhandenem Wissen fortentwickeln (Cohen & Levinthal, 1990; Bresser & Bishop, 1983). Zum anderen beschränkt vorhandenes Wissen (z.B. aufgrund seiner Historizität) die Generierung neuer Fähigkeiten, denn „Capabilities" sind immer auch „Rigidities" (Henderson & Clark, 1990; Leonard-Barton, 1992; Kogut & Zander, 1992; Zander & Kogut, 1995). Zu den von der Literatur als besonders wichtig angesehenen dynamischen

308

R. Κ. F. Bresser

Fähigkeiten zählen die Lern- und die Anpassungsfähigkeit der Unternehmung (Teece et al., 1994; Knyphausen, 1995: 99-107; Sanchez, 1995; Grant, 1996). Das Interesse der Literatur am Studium organisatorischer Fähigkeiten darf nicht über die Schwierigkeiten hinwegtäuschen, diese im konkreten Fall zu bestimmen. Collis (1994) warnt davor, die Suche nach wettbewerbsvorteilgenerierenden Fähigkeiten gewissermaßen im luftleeren Raum vorzunehmen. Da die Anzahl möglicher organisatorischer Fähigkeiten unendlich groß ist, wird sich ohne klare Grenzziehungen kaum eine Fähigkeit nachweisen lassen, die anderen gegenüber eindeutig überlegen ist. Nach Collis (1994) sind die Vorteile, die sich aus den meisten organisatorischen Fähigkeiten ableiten lassen, nur kurzlebig, denn sie unterliegen oft, insbesondere in hyperkompetitiven Umwelten (D'Aveni, 1994), der Erosion und dem Ersatz durch bessere Fähigkeiten. Die Vorteile organisatorischer Fähigkeiten sind somit kontext- und zeitabhängig. Die Forschungen über wettbewerbsvorteilstiftende Fähigkeiten sollten sich deshalb auf bestimmte Kontexte (z.B. Branchen) konzentrieren und auf jene Fähigkeiten, von denen Vorteile in der nahen Zukunft erwartet werden können. Trotz des fraglos großen Potentials des RBA für die Entwicklung einer Theorie des Strategischen Managements weist auch dieser Ansatz mehrere Problembereiche auf, deren Lösung für eine Weiterentwicklung des Ansatzes als geboten erscheint: (1) Terminologische Probleme. Die RBA-Literatur ist durch eine Begriffevielfalt und Uneinheitlichkeit relevanter Bezeichnungen gekennzeichnet, die die Fokussierung der Forschungsbemühungen gefährdet (Collis, 1994; Rasche & Wolfrum, 1994). Mit dem Begriff Ressource häufig synonym verwendete Begriffe sind z.B. organisatorische Fähigkeit (capability, routine, skill), Kompetenz (competence) oder Aktivposten (asset). Eine klarere Abgrenzung und Festlegung der Begrifflichkeit ist dringend geboten. (2) Vernachlässigung der externen Umwelt. Für viele (nicht alle) Beiträge des RBA gilt, daß sie ausschließlich auf die „Unternehmungsspezifität" einer Ressource als Begründung für dauerhafte Wettbewerbsvorteile abstellen. Diese Betrachtung ist zu eng und kontextlos. Es wird in zukünftigen Forschungen wichtig sein, die Argumente und Konzepte der extern orientierten Forschungsansätze (z.B. industrieökonomischen Ansätze) deutlicher mit dem RBA zu verbinden (Collis, 1994; Conner, 1994; Rühli, 1994; Rasche & Wolfrum, 1994; Bamberger & Wrona, 1996). (3) Die Betonung dauerhafter statt temporärer Wettbewerbsvorteile. Für viele hyperkompetitive Hochtechnologiebranchen ist der Gedanke eines dauerhaften (sustainable) Wettbewerbsvorteils unrealistisch, da sich die technologisch bedingten Grundlagen des Wettbewerbs zu schnell verändern (D'Aveni, 1994) und, wenn nicht Imitationen, so doch sehr schnelle Substitutionen von kritischen Ressourcen gang und gäbe sind (Ilinitch et al., 1996; Thomas, 1996). Der RBA sollte somit die Möglichkeiten und Implikationen temporärer Wettbewerbsvorteile stärker in den Mittelpunkt der Analysen rücken, eine Empfehlung, die sich auch aus der Sicht der „Österreichischen Schule" der Ökonomie ergibt (Jacobson, 1992; Nault & Vandenbosch, 1996). (4) Nichtimitierbarkeit statt Imitation. Eng zusammenhängend mit dem vorangehenden Kritikpunkt ist der Stellenwert der Nichtimitierbarkeit. Bei schnellebigen Produktlebenszyklen spielt die Nichtimitierbarkeit von Ressourcen oft nur eine beschränkte Rolle für die Wettbewerbsfähigkeit der Unternehmung. Wenn der Wert einer Ressource durch schnelle Erosion, Substitution und Imitation gefährdet ist (Collis,

Ressourcen und Fähigkeiten

309

1994), dann kann es zur Generierung von temporären Wettbewerbsvorteilen sinnvoll sein, Imitationen zu erleichtern. Selbst wenn Imitationen möglich sind, können einer Unternehmung, die eine wertvolle Ressource entwickelt hat, noch Wettbewerbsvorteile entstehen, da Imitationen oft mit „time-lags" verbunden sind, die einer Unternehmung vorübergehend die Möglichkeit geben, „supranormale" Rentabilitäten zu erzielen (Jacobson, 1992: 803). Garud und Kumaraswamy (1993) haben anhand der Sun Microsystems Corporation aufgezeigt, wie eine Unternehmung Wettbewerbsvorteile aufbauen kann, indem sie ihren Konkurrenten die Imitation der eigenen Ressourcen erleichtert. Sun hat Konkurrenten einen leichten Zugang zu den eigenen Technologien gewährt und dadurch eine Multisystemkompatibilität auf dem Computer-Workstation-Markt durchgesetzt. Gleichzeitig hat Sun sich als Experte für Kompatibilitätstechnologien etabliert und sich zum führenden Anbieter im Workstation-Markt fortentwickelt. Es erscheint somit dringend geboten, daß der RBA sich stärker mit der Wichtigkeit imitierbarer Ressourcen für den Aufbau von Wettbewerbsvorteilen auseinandersetzt. Als ergänzende theoretische Grundlage kann hierbei auf den institutionalistischen Ansatz der Organisationssoziologie zurückgegriffen werden (DiMaggio & Powell, 1983; Zucker, 1987), der die Imitation für die Entwicklung institutioneller Strukturen in den Brennpunkt der Betrachtung rückt.31 (5) Verhaltenswissenschaftliche Anreicherung. Trotz der zunehmenden Beschäftigung mit der Genese von Ressourcen hat der RBA sein verhaltenswissenschaftliches Potential noch nicht voll ausgeschöpft. Die Frage, „wie" wettbewerbsvorteilbegründende Ressourcen entstehen und adaptiert werden, wurde bisher zwar oft diskutiert aber nicht hinreichend geklärt. Es ist notwendig, einen Handlungs- und Entscheidungsaspekt in die Analyse einzuführen und empirisch zu untersuchen (Porter, 1991:108f.). In Anlehnung an das Modell von Hambrick und Mason (1984) wäre es z.B. sinnvoll, eine Top-Management-Perspektive der Generierung kritischer Ressourcen zu entwickeln, denn der systematischen Entwicklung spezifischer Ressourcen (z.B. im F&E-Bereich) gehen oft Top-Management-Entscheidungen voraus. Zur Untersuchung dieser behavioristischen Fragestellungen erscheinen folgende Variablenklassen und Verbindungen bedeutsam zu sein (Abb. 2:)

Die Artikel Beide Artikel dieses Kapitels wurden gewählt, da sie u.a. einen Beitrag zur Verringerung der Begriffevielfalt leisten. Der erste Artikel, „The Cornerstones of Competitive Advantage: A Resource-Based View", von Margaret Peteraf integriert verschiedene Ansätze über die Bedingungen, durch die Ressourcen zu Wettbewerbsvorteilen führen sollen, und stellt ein knappes, aus vier Bausteinen bestehendes Modell vor. Die vier Bedingungen für das Entstehen dauerhafter Wettbewerbsvorteile sind Ressourcenheterogenität, ex postBeschränkungen des Wettbewerbs, imperfekte Mobilität und ex ante-Beschränkungen des Wettbewerbs. Die Autorin untersucht darüber hinaus auch generelle Implikationen 31

Vgl. hierzu Kapitel 8.

310

R. Κ. F. Bresser

υ

Pi u

υ •o c 2υ 'S e

υ

Iu £

Uj a

s J3 o 3 a

Ρ

203 003 Û0 σ\ 3 β o C tafl

£

Τ3 e s υ 'Π ta Χ

-a J3 m -O -O

Ressourcen und Fähigkeiten

311

einer ressourcenbasierten Sichtweise für das Verfolgen von Konzentrations- und Diversifikationsstrategien. Der zweite Artikel, „Strategie Assets and Organizational Rent", von Amit und Schoemaker reduziert die Vielfalt der Begriffe im RBA auf drei Klassen: „resources", „capabilities" und „strategic assets". „Resources" sind die materiellen und immateriellen Komponenten des Unternehmungsoutputs (Produkte und Dienstleistungen). Die Konversion der Ressourcen in Outputs erfolgt durch den Einsatz von „capabilities" (organisatorische Fähigkeiten). Eine Untergruppe von Unternehmungsressourcen und -fähigkeiten ist hochspezifisch, kaum handelbar und nicht einfach zu imitieren. Diese Untergruppe stellt die „Strategie assets" (strategischen Aktiva) der Unternehmung dar, durch die dauerhafte Wettbewerbsvorteile erzielt werden können. Amit und Schoemaker verbinden diese ressourcenbasierten Überlegungen mit der externen Umweltanalyse: Die Unternehmung muß „strategic industry factors" (strategische Branchenfaktoren) berücksichtigen, bevor sie sich für den Aufbau bestimmter strategischer Aktiva entscheidet. Strategische Branchenfaktoren sind Ressourcen und Fähigkeiten, die sich als wichtige Determinanten des Wettbewerbs einer Branche herausgestellt haben. Durch die Verbindung interner Ressourcen mit externen Branchenfaktoren liefern Amit und Schoemaker ein theoretisches Fundament sowohl für die SWOT-Analyse als auch für ein integriertes Strategic-FitKonzept.

312

M. A. Peteraf

Strategie Management Journal, Vol. 14, 179-191 (1993)

THE CORNERSTONES OF COMPETITIVE ADVANTAGE: A RESOURCE-BASED VIEW MARGARET A. PETERAF J. L. Kellogg Graduate School of Management, Illinois, U.S.A.

Northwestern

University,

Evanston,

This paper elucidates the underlying economics of the resource-based view of competitive advantage and integrates existing perspectives into a parsimonious model of resources and firm performance. The essence of this model is that four conditions underlie sustained competitive advantage, all of which must be met. These include superior resources (heterogeneity within an industry), ex post limits to competition, imperfect resource mobility, and ex ante limits to competition. In the concluding section, applications of the model for both single business strategy and corporate strategy are discussed.

INTRODUCTION In recent years, a model of how firms compete, which is unique to the field of strategic management, has begun to emerge. Known as the 'Resource-Based View', it is regarded by some as having momentous potential as a paradigm for our field. Others wonder whether this emergent model provides much additional insight over traditional understandings. Admittedly, resource-based work is consistent with and rooted squarely in the policy research tradition. The notion that firms are fundamentally heterogeneous, in terms of their resources and internal capabilities, has long been at the heart of the field of strategic management. The classic approach to strategy formulation, for example, begins with an appraisal of organizational competencies and resources (Andrews, 1971). Those which are distinctive or superior relative to those of rivals, may become the basis for competitive advantage if they are matched appropriately to

environmental opportunities (Andrews, 1971; Thompson and Strickland, 1990). Those ideas may be thought of as the basic principles upon which resource-based research continues to build. While the model is still in the developmental stage, it has deepened our understanding regarding such topics as how resources are applied and combined, what makes competitive advantage sustainable, the nature of rents, and the origins of heterogeneity. 1 The work of Penrose (1959) is considered a very influential force. Other notable contributions include Lippman and Rumelt (1982), Teece (1980, 1982), Nelson and Winter (1982), Rumelt (1984, 1987), Wernerfelt (1984), Barney (1986, 1991), Dierickx and Cool (1989), Castanias and Helfat (1991), Conner (1991), and Mahoney and Pandian (1992). This research stream is an impressive one. And while many agree that there is a need for greater rigor and richness of detail, the work that has been done provides a strong foundation and an inspiration for work to come. In reviewing this work, one encounters numer-

Key words: Resources, rents, competitive advantage, single-business strategy, corporate strategy 0143-2095/93/030179-13J11.50 © 1993 by John Wiley & Sons, Ltd.

1 t h i s is not meant to suggest that the contributions of resource-based work have been limited to these topics.

Received 27 April 1992 Final revision received 9 November 1992

Cornerstones of Competitive Advantage 180

M. A.

313

Peteraf

ous strands of research on a series of closely related topics. While each paper offers a distinct contribution, there is also considerable overlap of ideas. To the uninitiated this may be confusing. In part, this is because subtle variations in terminology across papers have made communication more difficult. But in addition, the underlying model seems somewhat disjoint, as if the ideas of these disparate authors have not fully coalesced into an integrated whole. While there is general agreement as to the basic insights of the model, there are small disagreements over minor points. The purpose of this paper is to develop a general model of resources and firm performance which at once integrates the various strands of research and provides a common ground from which further work can proceed. My aim is to build consensus for a parsimonious model, clarify basic issues, suggest possible implications, and, in so doing, facilitate the continuing dialogue among scholars. In the first section, a resource-based model of the theoretical conditions which underlie competitive advantage is presented. There are four such conditions, all of which must be met. The first of these is resource heterogeneity, from which come Ricardian or monopoly rents. Ex post limits to competition are necessary to sustain the rents. Imperfect resource mobility ensures that the rents are bound to the firm and shared by it. Ex ante limits to competition prevent costs from offsetting the rents. Each of these conditions is described in turn. The model is intended to aid our theoretical understanding of superior firm performance as well as to inform management practice. In the final section, some applications and implications of the model are described. In particular, the application of resource-based work to single-business strategy, as well as to multibusiness corporate strategy, in all of its forms, is discussed.

firms (Barney, 1991).2 One might describe productive factors in use as having intrinsically differential levels of 'efficiency.' Some are superior to others. Firms endowed with such resources are able to produce more economically and/or better satisfy customer wants. Heterogeneity implies that firms of varying capabilities are able to compete in the marketplace and, at least, breakeven. Firms with marginal resources can only expect to breakeven.3 Firms with superior resources will earn rents.4 Ricardian rents Heterogeneity in an industry may reflect the presence of superior productive factors which are in limited supply. They may be fixed factors which cannot be expanded. More often, they are quasi-fixed, in the sense that their supply cannot be expanded rapidly. They are scarce in the sense that they are insufficient to satisfy demand for their services. Thus, inferior resources are brought into production as well. This is the familiar Ricardian argument.5 It may be understood most clearly by assuming that firms with superior resources have lower average costs than other firms.6 (See Figure 1.) These low cost firms have somewhat inelastic supply curves, in that they cannot expand output rapidly, regardless of how high the price may be. High prices, however, do induce other less efficient firms to enter the industry. Such firms will enter and produce so long as price exceeds their marginal cost (MC). In equilibrium, industry demand and supply are in balance, high-cost firms breakeven (P = AC), and low-cost firms earn supranormal profits in the form of rents to their scarce resources (P > AC). Note that this model is consistent with competitive behavior in the product market. Firms are price takers and produce at the point where 2

A MODEL OF COMPETITIVE ADVANTAGE Heterogeneity A basic assumption of resource-based work is that the resource bundles and capabilities underlying production are heterogeneous across

See Nelson (1991) and Williams (1992) for discussions on why firms are different. 3 In equilibrium, industry demand and supply conditions determine the minimum efficiency level required to breakeven. 4 Earnings in excess of breakeven are called rents, rather than profits, if their existence does not induce new competition. ' See Ricardo (1817) and Rumelt (1987). ' Note, however, that superior resources do not necessarily lead to a low cost position. This is simply the most tractable example.

314

M. A. Peteraf

Cornerstones

Industry

High Cost Firm

of Competitive

Advantage

181

Low Cost Firm

Figure 1. Scarcity rents with heterogeneous factors Key: P* = Equilibrium Price, SS = Rents to Efficient Producer

Industry

Efficient Firms

Inefficient Firm

Figure 2. Imitation (expansion) of low cost firms causes rents to disípate and high-cost firms to exit Key: P** = New Equilibrium Price

price equals marginal cost. The high returns of efficient firms cannot be attributed to an artificial restriction of output or to market power. Neither do they depend upon uniqueness or even rarity in the absolute sense. It is theoretically possible for rents to be earned by a number of equally efficient producers, so long as an efficiency differential remains between them and other producers. What is key is that the superior resources remain limited in supply. Thus, efficient firms can sustain this type of competitive advantage only if their resources cannot be expanded freely or imitated by other firms. Consider what happens if this is not so. (See Figure 2.) Increased production by additional efficient producers will shift the supply curve out. This will drive down the equilibrium price, forcing marginal firms to leave the market. Remaining firms will produce at the point where

price equals both marginai cost and average cost. As a result, rents will be dissipated and only normal returns will be earned by efficient (now homogeneous) producers. The Ricardian model is often thought of with respect to resources which are strictly fixed in supply. But it may be applied as well to quasifixed resources, which are of much greater importance. These are resources which, while limited in the short run, may be renewed and expanded incrementally within the firm that utilizes them. 7 Utilization of such resources may in fact augment them. Prahalad and Hamel (1990) describe how core competencies, particularly those which involve collective learning and are knowledge-based, are

7

See Nelson and Winter (1982) and Wernerfelt (1989).

Cornerstones of Competitive Advantage 182

M. A. Peteraf

enhanced as they are applied. Such resources may provide both the basis and the direction for the growth of the firm itself. For example, there may be a natural trajectory embedded in a firm's knowledge base. 8 Current capabilities may both impel and constrain future learning and investment activity.9 Incremental growth and renewal of such limited resources, however, is not inconsistent with a Ricardian view of rent and competitive advantage. Monopoly rents The condition of heterogeneity is equally consistent with models of market power and monopoly rents as it is with the Ricardian story. What distinguishes monopoly profits from Ricardian rents is that monopoly profits result from a deliberate restriction of output rather than an inherent scarcity of resource supply. In monopoly models, heterogeneity may result from spatial competition or product differentiation.10 It may reflect uniqueness and localized monopoly. It may be due to the presence of intra-industry mobility barriers which differentiate groups of firms from one another (Caves and Porter, 1977.) It may entail size advantages and irreversible commitments or other first mover advantages.11 There are numerous such models. What they all have in common is the supposition that firms in favorable positions face downward sloping demand curves. These firms then maximize profits by consciously restricting their output relative to competitive levels. These are models of market power. Unlike Ricardian models, many are 'strategic' in that firms take into account the behavior and relative position of their rivals. Apparently homogeneous firms may also earn monopoly rents. Cournot behavior exhibited by identical rivals, for example, may yield prices in excess of marginal costs. So may collusive behavior, tacit or otherwise. But these kinds of behaviors are facilitated by fewness of numbers and therefore depend on barriers to entry. Asymmetries must exist between incumbent 8

315

This is a notion attributable to organizational economics. See Teece (1990). * See Dosi, Teece, and Winter's (1990) discussion of core capabilities, path dependencies and learning. 10 See Schmalensee (1978). 11 See Ghemawat (1986) and Lieberman and Montgomery (1988). Consider also models of dominant firm behavior.

firms and potential entrants. In this case, the heterogeneity occurs across these two groups of firms.

Ex post limits to compétition Regardless of the nature of the rents, sustained competitive advantage requires that the condition of heterogeneity be preserved. If the heterogeneity is a short-lived phenomenon, the rents will likewise be fleeting. Since strategists are primarily concerned with rents over a longer term, the condition of heterogeneity must be relatively durable to add value. This will be the case only if there are in place ex post limits to competition as well. By this I mean that subsequent to a firm's gaining a superior position and earning rents, there must be forces which limit competition for those rents. Competition may dissipate rents by increasing the supply of scarce resources. Alternatively, it might undermine a monopolist's (or oligopolists') attempts to restrict output. Figure 2 illustrates how ex post competition makes the industry supply curve more elastic and erodes Ricardian rents. Ex post competition erodes monopoly rents as well, by increasing output or by making individual demand curves more elastic. Resource-based work has focused on two critical factors which limit ex post competition: imperfect imitability and imperfect substitutability.12 Substitutes reduce rents by making the demand curves of monopolists or oligopolists more elastic. This is one of Porter's (1980) classic 'five forces.' Much greater attention, however, has been given to the condition of imperfect imitability. Rumelt (1984) coined the term 'isolating mechanisms' to refer to phenomena which protect individual firms from imitation and preserve their rent streams. These include property rights to scarce resources and various quasi-rights in the form of lags, information asymmetries, and frictions which impede imitative competition (Rumelt, 1987). Of particular interest is the notion of causal ambiguity (Lippman and Rumelt, 1982). This refers to uncertainty regarding the causes of efficiency differences among firms. Causal ambiguity prevents would-be-imitators

12

See Barney (1991) and Dierickx and Cool (1989).

316

M. A. Peteraf merstones of Competitive Advantage

from knowing exactly what to imitate or how to go about it. Coupled with nonrecoverable costs, such uncertainty may limit imitative activity, thus preserving the condition of heterogeneity. Other isolating mechanisms include producer learning, buyer switching costs, reputation, buyer search costs, channel crowding, and economies of scale when specialized assets are required (Rumelt, 1987).13 Rumelt (1984) describes isolating mechanisms as an analog of Caves and Porter's (1977) mobility barriers, which are themselves an extension of Bain's (1956) concept of entry barriers.14 Mobility barriers, however, serve to isolate groups of similar firms in a heterogeneous industry, while entry barriers isolate industry participants from potential entrants. Yao (1988) has distilled a set of factors more basic than the list of entry barriers suggested by Porter (1980) and Bain (19S6). He contends that failures of the competitive market are due more fundamentally to production economies and sunk costs, transaction costs, and imperfect information. Ghemawat (1986) suggests a different categorization, with more of a firm than a market orientation. He argues that inimitable positions derive from size advantages, preferred access to either resources or customers, and/or restrictions on competitors' options. Dierickx and Cool (1989) offer a unique perspective on the topic of limits to imitation. They focus on factors which prevent the imitation of valuable but nontradeable asset stocks. They maintain that how imitable an asset is depends upon the nature of the process by which it was accumulated. They identify the following characteristics as serving to impede imitation: time compression diseconomies, asset mass efficiencies, interconnectedness of asset stocks, asset erosion, and causal ambiguity. Dierickx and Cool's (1989) paper is a particularly important piece of work because it focuses precisely on those kinds of resources and capabilities which are of central concern to resourcebased theory: nontradeable assets which develop and accumulate within the firm. Such assets tend to defy imitation because they have a strong 13 These topics and other related ones have received much attention in modern industrial organization literature, as well. 14 For further discussion, see Mahoney and Pandian (1992).

183

tacit dimension and are socially complex. They are born of organizational skill and corporate learning. Their development is 'path dependent' in the sense that it is contingent upon preceding levels of learning, investment, asset stocks, and development activity.11 For such assets, history matters. Would-be-imitators are thwarted by the difficulty of discovering and repeating the developmental process and by the considerable lag involved. Importantly, assets of this nature are also immobile and thus bound to the firm. Factor immobility, or imperfect mobility is another key requirement for sustainable advantage. Imperfect mobility Resources are perfectly immobile if they cannot be traded. Dierickx and Cool (1989) discuss several examples of this sort. Resources for which property rights are not well defined or with 'bookkeeping feasibility' problems fall into this category (Dierickx and Cool, 1989; Meade, 1952; Bator, 1958). So do resources which are idiosyncratic to the extent that they have no other use outside the firm. (See Williamson, 1979). Other kinds of resources may be described as imperfectly mobile. These are resources which are tradeable but more valuable within the firm that currently employs them than they would be in other employ. Resources are imperfectly mobile when they are somewhat specialized to firm-specific needs.16 Montgomery and Wernerfelt (1988) use the concept of switching costs to discuss how firmspecific investments may cement the trading relationship between a firm and the owners of factors employed by the firm. These investments by the resource owners may be regarded as a sunk cost (nonrecoverable cost) which may inhibit the factor's exit from a firm. These costs give the firm a greater claim on the resource in question. Cospecialized assets may be another case in point (Teece, 1986). These are assets which must be used in conjunction with one another or which have higher economic value when 19

See, Barney (1991) and Dosi, Teece, and Winter (1990). 14 Williamson (1985) discusscs such assets and their implications for efficient firm boundaries extensively.

Cornerstones of Competitive Advantage 184

M. A.

Peteraf

employed together. To the extent that they have no other equivalent uses (they are transaction specific) and to the extent that at least one of the assets is firm-specific, their mobility is limited. Other resources may be imperfectly mobile simply because the transactions costs associated with their transfer are exceedingly high (Williamson, 1975; Rumelt, 1987.) Because immobile or imperfectly mobile resources are nontradeable or less valuable to other users, they cannot be bid away readily from their employer. They remain bound to the firm and available for use over the long run. Thus, they can be a source of sustained advantage.17 Furthermore, the opportunity cost of their use is significantly less than their value to the present employer. This is an important point and one which will be developed further in the next section. It implies that any Ricardian or monopoly rents generated by the asset will not be offset entirely by accounting for the asset's opportunity cost. I use opportunity cost, here, in a sense slightly different from the conventional use of the term. Conventionally, it refers to the value of a resource in its next best use. Here, I mean it to refer to the value of the resource to its second-highest valuing potential-user. (See Klein, Crawford, and Alchian, 1978.) The use to which the potential user may wish to put it may be exactly the same. This difference between the value of a resource to a firm and its opportunity cost is also a form of rent. Pareto rents, also called quasi-rents, are the excess of an asset's value over its salvage value or its value in its next best use. Following Klein et al. (1978), I use the term 'appropriable quasi-rents' or Ά - Q rents' to refer to the excess of an asset's value over its value to the secondhighest valuing potential user or bidder for the resouce. Klein et al. (1978) demonstrate that it is entirely possible for a resource to generate AQ rents in the absence of either Ricardian or monopoly .rents. Resources need not be rare or inimitable for them to be differentially valuable to possible users. Thus the presence of A-Q

17

317

On the other hand, such assets may make a firm less responsive and flexible in the face of environmental or technological changes which upset a previously held advantage. Specialization is a two-edged sword.

rents is not a sufficient indicator of competitive advantage. There must be monopoly or Ricardian rents generated as well. A-Q rents are appropriable in the sense that they need not be paid out to the resource for the user to retain its services (Klein et al., 1978). Were the user to appropriate the whole of the A-Q rents, the resource could earn no more elsewhere.18 It may be more accurate, however, to recognize that the rents will be shared between the factor owners and the firm employing them. First, one might as easily view the firm as tied to the use of specialized factors, since it cannot substitute generic factors at equal cost. This implies that the situation might be characterized best as a bilateral monopoly, in which the distribution of rents is indeterminate. Secondly, it should be recognized that the rents are in fact jointly produced and are as much due to the firm as to the factor. A specialized factor cannot be so productive apart from the firm. Therefore, its super-productivity is attributable as much to the context and other elements of the firm as to the factor itself. The firm and the factor are, in essence, a team. Caves (1980) states that rents are not entirely passed on to factors which are not traded on the open market. In a similar vein, Rumelt (1987) has argued that 'the rent on (specialized) factor(s) is not logically or operationally separable from the profits of the firm' (p. 143). These two facts—that imperfectly mobile resources will remain available to the firm and that the rents will be shared by the firm—are the key features of imperfect factor mobility (see Wernerfelt, 1989). They, in turn, make imperfect factor mobility a necessary condition for sustainable competitive advantage. In addition, imperfect factor mobility is a particularly important component of the model because such resources are less likely to be imitable than other kinds.19 Furthermore, the opportunity cost of such assets, as defined above, does not offset the rents. But even together with heterogeneity

,a Note that, in a multiperiod model, human resources would be reluctant to invest in firm-specific attributes if they expected the firm to appropriate the rents generated. 19 Dierickx and Cool (1989) contend that nontradcability is required to ensure that an asset remains fixed in supply.

318

M. A. Peteraf merstones of Competitive

and ex post limits to competition, imperfect factor mobility is not yet sufficient for sustained competitive advantage. Ex ante limits to competition One last condition must be met for a firm to have competitive advantage. There must be ex ante limits to competition as well. By this I mean that, prior to any firm's establishing a superior resource position, there must be limited competition for that position. This may be best explained by illustration. Suppose it is perceived, a priori, by equally endowed firms that by occupying certain choice locations they can gain an inimitable resource position over their rivals. What will ensue is fierce competition for those locations to the point that the anticipated returns are, in essence, competed away. A superior location could only be a source of above normal returns if some firm had the foresight or good fortune to acquire it in the absence of competition. This is the point brought out by Barney (1986) in arguing that the economic performance of firms depends not only on the returns from their strategies but also on the cost of implementing those strategies. Without imperfections in strategic factor markets, where the resources necessary to implement strategies are acquired, firms can only hope for normal returns. Rumelt (1987) makes a similar point in noting that unless there is a difference between the ex post value of a venture and the ex ante cost of acquiring the necessary resources, the entrepreneurial rents are zero. Profits come from ex ante uncertainty. While only tradeable resources can be acquired in strategic factor markets, the argument can be extended to immobile and imperfectly mobile resources as well, as both Dierickx and Cool (1989) and Barney (1989) have noted. Ex ante competition to develop imperfectly mobile resources, such as the good will of clients, can also dissipate expected returns. While it is less likely that the full value of such resources will be anticipated or that firms will be equally efficient in accumulating such resources, it is important to recognize that imperfect resource mobility is not sufficient unto itself. There must be limits to ex ante competition as well.

Advantage

185

The cornerstones of competitive advantage In sum, four conditions must be met for a firm to enjoy sustained above-normal returns. Resource heterogeneity creates Ricardian or monopoly rents. Ex post limits to competition prevent the rents from being competed away. Imperfect factor mobility ensures that valuable factors remain with the firm and that the rents are shared. Ex ante limits to competition keep costs from offsetting the rents. The model is summarized in Figure 3. This model is intended to highlight the importance of each of these conditions, as distinct from one another, and to explicate the particular role that each plays in creating and sustaining rents. It is not meant to imply, however, that these four conditions are entirely independent of one another. They are, in fact, related conditions. Heterogeneity is the most basic condition. It is the sine-qua-non of competitive advantage and has long been a fundamental concept of strategic management. For these reasons it deserves special emphasis. The model tells us that heterogeneity is necessary for sustainable advantage, but not sufficient. For rents to be sustained, we required ex post limits to competition as well. One can imagine heterogeneity without ex post limits to competition. Firms may have short-lived and unsustainable readily-imitated differences. It takes a greater stretch of the imagination to conceive of ex post limits to competition without heterogeneity. (Perhaps a regulator enforcing a pricing cartel among numerous homogeneous trucking firms.) For the most part, ex post limits to competition imply heterogeneity, although heterogeneity does not imply ex post limits to competition. Heterogeneity underlies the condition of imperfect mobility as well. Again heterogeneous resources need not be imperfectly mobile. But it is hard to imagine any imperfectly mobile resources which are not also heterogeneous in nature. Resources which are immobile because of their idiosyncratic or firm-specific nature are certainly heterogeneous. Resources which are immobile due to ill-defined property rights or the lack of a market might possibly be homogeneous (pollution rights, for example?) Once again, however, imperfect mobility, for the most part, implies heterogeneity as well. Finally, it is important to recognize that the

Cornerstones of Competitive Advantage 186

M. A.

319

Peteraf

Ex Post Limits to Competition

Heterogeneity

Competitive Advantage Rents Sustained within the Arm

imperfect Mobility

Rents not offset by costs

Ex Ante Limits to Competition

Figure 3. The cornerstones of competitive advantage

productivity of superior resources depends upon the nature of their employment and the skill with which a strategy based on resource superiority is implemented.

APPLICATIONS OF THE RESOURCEBASED MODEL A major contribution of the resource-based model is that it explains long-lived differences in firm profitability that cannot be attributed to differences in industry conditions. Indeed, there is considerable evidence to show that such differences are not well explained by industry participation (Schmalensee, 1985; Mueller, 1986; Wernerfelt and Montgomery, 1988; Hansen and Wernerfelt, 1989; Rumelt, 1991). There is less

agreement on the relative magnitude of firm effects, but several studies have indicated that these effects are substantial (Mueller, 1986; Hansen and Wernerfelt, 1989; Rumelt, 1991). The resource-based model is a theoretical complement to this work. On the practical side, the model may prove useful to managers seeking to understand, preserve, or extend their competitive advantage. While the model itself is freely available to all, its strategic implications depend on a firm's specific resource endowment. Barney (1986) argues that a firm may gain expectational advantages by analyzing information about the assets it already controls. So long as its assets are imperfectly mobile; inimitable, and nonsubstitutable, other firms will not be able to mimic its strategy. Thus, application of the model will not

320

M. A. Peteraf rnerstones of Competitive Advantage

increase competition for available rents. It will only ensure that each firm optimizes the use of its own specialized resources. Because of its focus on imperfectly mobile resources, for which the transactions cost of market exchange are high, resource-based theory has important implications for corporate strategy and issues regarding the scope of the firm as well as single business strategy. Some applications in each of these areas are discussed in turn. Single business strategy At the single business level, the model may help managers differentiate between resources which might support a competitive advantage from other less valuable resources (Barney, 1991). For example, a brilliant, Nobel prize winning scientist may be a unique resource, but unless he has firm-specific ties, his perfect mobility makes him an unlikely source of sustainable advantage. Managers should ask themselves if his productivity has to do, in part, with the specific team of researchers of which he is a part. Does it depend on his relationship with talented managers who are exceptionally adept at managing creativity? Does it depend on the spirit of the workers or the unique culture of the firm? A resource-based perspective may also help a firm in deciding whether to license a new technology or whether to develop it internally. If the technology is imperfectly mobile in the sense that its potential value cannot be well communicated to others because of the risk of revealing proprietary information, it might best be developed internally. Alternatively, its marketability might depend upon cospecialized assets such as long established relationships with vendors who are reluctant to switch to other suppliers. If the cospecialized assets are held by the firm and are themselves immobile, internal development may still make sense. If the innovation is perfectly mobile, the innovators could do no better than to license the technology. Decision-making would also be enhanced by considering how imitable the innovation is. If the innovation is no more than a clever and complex assembly of relatively available technologies, then no wall of patents could keep opponents out. Recognizing this vulnerability, a manager might want to think more carefully about the length of the expected entry lag and

187

whether or not there may be some advantage possible due to firm-specific learning or asset mass efficiencies. He might consider trying to use his head start to build other cospecialized resources that are less available (say a reputation for service on the new technology). This might be possible if the secondary resource is time path dependent or if his expectational advantage inhibits competition from developing the secondary resource. The general point is that by analyzing his resource position, a manager would have a clearer understanding of whether his situation meets necessary conditions for a sustainable advantage. Fewer strategic mistakes would be made. But in addition, it might help him to utilize his expectational advantage in looking ahead. Amit and Schoemaker (1993) draw upon resource-based theory in developing a behavioral view of strategic assets and offer some prescriptive advice on how to target, develop and deploy them. Wernerfelt (1989) proposes some guidelines to help managers identify their critical resources and decide how to apply them. In some cases causal ambiguity may make it impossible for a firm to evaluate its resources or even to identify them. (See Lippman and Rumelt, 1982). While such resources may be the basis for competitive advantage, the causal ambiguity involved leaves little room for strategy. Firms owning the resources have no informational advantage over other firms and little ability to leverage these resources further since there is uncertainty regarding their dimensions and/or their value. Other resources can more easily be identified as value-creating resources, but their reproduction may be highly uncertain. Resources which are strongly time-path dependent or which are socially complex fit this category. (See Barney, 1991.) While these resources may be difficult to reproduce or extend, the firm owning the assets is likely to have a strong advantage in extending them over other firms. In part, this advantage is informational, based on complex and tacit understandings, not easily accessible to outsiders. But also it's because the production of a socially complex resource is likely to require firm specific cospecialized assets which cannot be duplicated in other settings. The resource-based view would help managers to understand that such resources

Cornerstones of Competitive Advantage 188

M. A.

Peteraf

can be an important basis for competitive advantage. And, by highlighting the value of these resources, it might help managers see that, despite the difficulty, they should consider leveraging these resources further.

Corporate strategy The resource-based model is fundamentally concerned with the internal accumulation of assets, with asset specificity, and, less directly, with transactions costs. TTius it lends itself naturally to the consideration of questions regarding boundaries of the firm. A number of researchers have utilized a resource-based view to analyze issues regarding the scope of the firm. Barney (1988), for example, has addressed the issue of whether bidding firms may realize abnormal returns from strategically related acquisitions. His resource-based framework provides the answer that it depends upon how rare and inimitable is the resulting combination of resources. Montgomery and Hariharan (1991), have shown that firms with broad resource bases tend to pursue diversification. (See Penrose, 1959, as well). In doing so, firms tend to enter markets where the resource requirements match their resource capabilities. More generally, the prevailing theory of diversification can be characterized as resourcebased. (See, for example, Teece, 1982; Wernerfelt, 1984; Williamson, 1985; Wernerfelt and Montogmery, 1986; Montgomery and Wernerfelt, 1988). This theory characterizes the kinds of resources which support diversification as quasifixed, yet inherently fungible: that is, they can support a variety of products. Other resources may possess a property of public goods, in that their use in one application does not diminish their availability for other uses. A brand name, for example, may be used without being 'used up' in the process. The crux of the theory is that diversification is the result of excess capacity in resources which have multiple uses and for which there is market failure.20 Without market failure, due to high transactions costs or imperfect mobility, the firm could simply sell the services of their redundant resources. In that case, single 20

321

For some empirical evidence on this point, see Chatterjee and Wernerfelt (1991).

business firms could operate more efficiently than a diversified firm, even if there are economies of scope (Teece, 1980, 1982). One issue, which has been inadequately addressed, is the paradox of how 'excess capacity' in resources may lead to 'scarcity rents' for resource holders. Certainly, these notions are incompatible if the resource has but a single use, since inferior resources would be driven from the market21 (see Figure 2). Recall, however, that the price of a resource is determined by the condition of supply and demand in the factor market. Factor demand, in turn, is derived from the demands of all products which it can be used to produce. If, at the equilibrium price, heterogeneous factors are employed across the markets, then superior factors will earn rents, regardless of whether their availability surpasses the needs of a single-product market. They are still scarce relative to total demand for their usage. In this way, excess capacity of a resource in a single-product market is compatible with its ability to command scarcity rents. Similarly, resources with pubjic good characteristics may earn rents, despite their availability for multiple employment. Since, after some point, there are limits to the expansion of these resources, perhaps because of a fixed supply of «specialized assets within the firm, such resources may still be scarce relative to total demand for their services. Eastman Kodak is an example of a firm that has diversified on the basis of excess capacity in its core capability in photographic technology. Its ability to expand in certain markets was limited by its high market share and antitrust considerations. In the mid-70s, its market share for film was estimated at 90 percent; it was estimated at 85 percent for cameras.22 In order to more fully utilize its prodigious R&D capabilities Kodak had to seek opportunities outside its original markets. This was possible because the potential for photographic technology applications was quite broad, encompassing movie films and equipment, medical and industrial Xray films and equipment, audiovisual products, microfilm, etc. In 1975, Kodak had a market

21

In this country, antitrust constraints typically limit market share. For this reason, inferior resources may well remain in the market despite excess capacity in single-use superior resources. 12 See 'Polaroid-Kodak,' HBS case # 376-266.

322

M . A. Peteraf

of Competitive share of just 38 percent of the total U.S market for amateur photographic products. In this sense, its resources were 'scarce' relative to total demand for their use over all applications, despite excess capacity relative to particular markets. A second issue which needs further attention is the question of why firms do not expand more fully in initial markets before they enter additional ones. It may be that the competitive model is inadequate to characterize product markets. Or it may be that, in general, both resources and market conditions may be better represented in a dynamic model, changing incrementally over time (Montgomery and Hariharan, 1991). Montgomery and Wernerfelt (1989), employ a framework which characterizes resources by their 'specificity' or range of application. Diversification is viewed as a result of matching a firm's resources to the set of market opportunities. These two conditions together determine both the range of strategic options and the profitability of a firm. For example, the high specificity of expertise in glass technology would constrain a firm from diversifying far afield on the basis of this resource. And, since specialized resources also tend to be relatively scarce, the model would predict higher rents for narrow diversifies. In contrast, firms with generalizable resources may face a much wider opportunity set. So, for example, a firm with expertise in cost cutting, embodied in a team of managers and firmspecific routines, might diversify quite widely. Lower rents would be expected, however, since these skills might be in greater supply. This does not imply that there is no scarcity value to such resources, but simply that they are relatively less scarce than more specialized resources. What is important is that heterogeneous managerial resources are heterogeneous and superior managers are less than perfectly mobile. Although the authors do not say so, the model also implies an optimal extent of diversification. Since the returns in each added market diminish due to resource efficiency loss, diversification will cease when rents in the final added market are zero. See Figure 4. Dosi, Teece, and Winter (1990) address the issue of the degree of relatedness among a firm's products—what they term 'coherence' in its business activities. The authors draw on concepts from organizational economics to explain the connection between a firm's core competencies

Advantage

189

Diversification

Figure 4. The determination of the extent of diversification Key: D" = Extent of Diversification, l»im>.= Accumulated Rents

and the degree of coherence among its parts. According to this theory, variations in the speed of learning, the breadth of the path dependencies, the degree of asset specialization and the nature of the selection environment explain the nature and extent of the scope of the firm. This work, although it is preliminary, appears to make a very fruitful start. In addition, it highlights the rich use that may be made of evolutionary economics, in particular, toward explaining phenomena of central interest to researchers taking a resource-based view of strategy. As these examples demonstrate, resourcebased theory, clearly, has power and implications for many important questions regarding corporate scope. It is a unifying theory which allows us to view both related and unrelated diversification through a common lens. It addresses diversification extent as well as type. It goes further than competing theories in simultaneously explaining the differences in profitability which are observed across firms, while also offering an explanation about why all firms do not and cannot pursue strategies which in the aggregate offer the highest returns. Instead, firms are seen as adopting strategies which their resources can support. Just as all resources supporting single business strategies do not have equal profit generating potential, neither do the resources supporting various diversification strategies. For an individual firm, whether it is a single-line business or widely diversified, the critical task is to use its available resources to the greatest end they can support. In sum, this emerging theory may prove to be a paradigm capable of elucidating and integrating

Cornerstones of Competitive Advantage

190

M. A.

323

Peteraf

Castanias, R. and C. Helfat (1991). 'Managerial resources and rents', Journal of Management, 17, pp. 155-171. Caves, R. E. (1980). 'Industrial organization, corporate strategy and structure', Journal of Economic Literature, 18, pp. 64-92. Caves, R. E. and M. Porter. (1977). 'From entry barriers to mobility barriers: Conjectural decisions and contrived deterrence to new competition', Quarterly Journal of Economics, 91, pp. 241-262. Chatterjee, S. and B. Wernerfelt. (1991). "The link between resources and type of diversification: Theory and evidence'. Strategic Management Journal, 12, pp. 33-48. Connor, K. (1991). Ά historical comparison of resource-based theory and five schools of thought within industrial organization economics: Do we have a new theory of the firm?', Journal of Management, 17, pp. 121-154. Dierickx, I. and K. Cool. (1989). 'Asset stock ACKNOWLEDGEMENTS accumulation and sustainability of competitive advantage', Management Science, 35, I would like to thank Connie Helfat, Yair pp. 1504-1511. Aharoni, Kurt Christensen, Joe Mahoney and Dosi, G., D. Teece, and S. Winter. (1990). 'Toward a theory of corporate coherence: Preliminary Ruth Raubitschek, for helpful comments. Raffi remarks', Working paper. Amit, Jay Barney, Anne Huff, Bruce Kogut, Ghemawat, P. (Sept-Oct 1986). 'Sustainable advanCynthia Montgomery, and Birger Wernerfelt tage', Harvard Business Review, pp. 53-58. gave me constructive criticism on an earlier Hansen, G. and B. Wernerfelt, (1989). 'Determinants version of this paper. I am grateful to David of firm performance: The relative importance of economic and organizational factors', Strategic Besanko and Jeff Williams for their encourageManagement Journal, 10, pp. 399-411. ment and support. Thanks are due as well to the Klein, B., R. Crawford, and A. Alchian. (1978). SMJ editors and reviewers. Remaining errors are 'Vertical integration, appropriable rents, and the my own. competitive contracting process', Journal of Law and Economics, 21, pp. 297-326. Lieberman, M. and D. Montgomery. (1988). 'First mover advantage', Strategic Management Journal, 9, Special issue, pp. 41-58. REFERENCES Lippman, S. A. and R. P. Rumelt. (1982). 'Uncertain imitability: An analysis of interfirm differences in Amit, R. and P. J. Schoemaker. (1993). 'Strategic efficiency under competition', The Bell Journal of assets and organizational rent', Strategic ManageEconomics, 13, pp. 418-438. ment Journal, 14, pp. 33-46. Andrews, K. R. (1971). The Concept of Corporate Mahoney, J. and J. R. Pandian. (1992). 'The resourcebased view within the conversation of strategic Strategy, Irwin, Homewood, IL. management', Strategic Management Journal, 13, Bain, J. (1956). Barriers to New Competition, Harvard pp. 363-380. University Press, Cambridge, MA. Barney, J. B. (1986). 'Strategic factor markets: Meade, J. (1952). 'External economies and diseconomies in a competitive situation', Economic Journal, Expectations, luck and business strategy', Managepp. 56-67. ment Science, 42, pp 1231-1241. Barney, J. B. (1988). 'Returns to bidding firms Montgomery, C. A. and S. Hariharan. (1991). 'Diversified expansion by large established firms', in mergers and acquisitions: Reconsidering the Journal of Economic Behavior, pp. 71-89. relatedness hypothesis', Strategic Management JourMontgomery, C. A. and B. Wernerfelt. (1988). nal, 9, pp. 71-78. 'Diversification, Ricardian rents, and Tobin's q', Barney, J. B. (1989). 'Asset stocks and sustained Rand Journal, pp. 623-632. competitive advantage: A comment', Management Mueller, D. (1986). Profits in the Long Run. Cambridge Science, 35,-pp. 1511-1513. University Press, Cambridge, MA. Bamey, J. B. (1991). 'Firm resources and sustained competitive advantage', Journal of Management, Nelson, R. (1991). 'Why do firms differ and how does it matter', Strategic Management Journal, 12, 17, pp. 99-120. pp. 61-74. Bator, F. (1958). 'The anatomy of market failure', Nelson, R. R. and S. G. Winter. (1982). An Quarterly Journal of Economics, pp. 351-379. research in all areas of strategy. Despite the need for further work, it has already shown itself to be a robust and integrative tool. It has strong implications for single-business strategy, for corporate strategy, for theorists and practitioners alike. Importantly, it is the only theory of corporate scope which is capable of explaining the range of diversification, in all its richness, from related constrained to the conglomerate form. This is the crucial mark of a robust theory of diversification (Teece, 1982). It is an area ripe for research, which has already demonstrated its fruitfulness and deserves the concentrated efforts of this community of scholars.

324

M. A. Peteraf Cornerstones of Competitive

Advantage

191

Thomas, III. (ed.), The Economics of Strategic Evolutionary Theory of Economic Change. Belknap Planning, Lexington, Lexington, MA, pp. 187-199. Press, Cambridge, MA. Penrose, Ε. T. (1959). The Theory of Growth of the Teece, D. (1990). 'Contributions and impediments of economic analysis to the study of strategic Firm, Basil Blackwell, London. Porter, M. E. (1980). Competitive Strategy: Techniques management'. In J. W. Fredrickson (ed.), Perspectives on Strategic Management, Harper Business, for Analyzing Industries and Competitors, The Free New York. pp. 39-80. Press, New York. Prabalad, C. K. and G. Hamel. (May-June 1990). Thompson, A. A. and A. J. Strickland. (1990). Strategic Management: Concepts and Cases, Irwin, 'The core competence of the corporation', Harvard Homewood, IL. Business Review, pp. 79-91. Ricardo, D. (1965, Original 1817). The Principles of Wernerfeit, Β. (1984). Ά resource based view of the firm', Strategic Management Journal, 5, pp. 171-180. Politicai Economy and Taxation. Reprinted, J. M. Wernerfelt, B. (1989). 'From critical resources to Dent and Son. London. corporate strategy', Journal of General ManageRumelt, R. P. (1984). Toward a strategic theory of ment, 14, pp. 4-12. the firm'. In R. Lamb (ed.), Competitive Strategic Management, Prentice Hall, Englewood Cliffs, NJ, Wernerfelt, B. and C. A. Montgomery. (1986). 'What pp. 556-570. is an attractive industry?', Management Science, 32, pp. 1223-1229. Rumelt, R. P. (1987). "Theory, strategy, and entrepreneurship'. In D. Teece, (ed.), The Competitive Wernerfelt, B. and C. A. Montgomery. (1988). 'Tobin's q and the importance of focus in firm Challenge. Ballinger, Cambridge, MA, pp. 137-158. performance', American Economic Review, 78, Rumelt, R. P. (1991). 'How much does industry pp. 246-250. matter?', Strategic Management Journal, 12, Williamson, O. (1975). Markets and Hierarchies. Free pp. 167-186. Press, New York. Schmalensee, R. (1978). 'Entry deterrence in the ready-to-eat breakfast cereal industry', Bell Journal Williamson, Ο. E. (1979). 'Transaction-cost economics: The governance of contractual relations', Journal of Economics, 9, pp. 305-327. of Law and Economics, 22, pp. 233-261. Schmalensee, R. (1985). 'Do markets differ much?', The American Economic Review, 75, pp. 341-350. Williamson, Ο. E. (1985). The Economic Institutions of Capitalism. Free Press, New York. Teece, D. J. (1980). 'Economies of scope and the scope of the enterprise', Journal of Economic Williams, J. (1992). 'Strategy and the search for rents: The evolution of diversity among firms,' Working Behavior and Organization, 1, pp. 223-247. Paper, Carnegie Mellon University. Teece, D. J. (1982). 'Toward an economic theory of the multiproduct firm', Journal of Economic Yao, D. (1988). 'Beyond the reach of the invisible Behavior and Organization, 3, pp. 39-63. hand: Impediments to economic activity, market failures, and profitability, Strategic Management Teece, D. J. (1986). 'Firm boundaries, technological Journal, 9, pp. 59-70. innovation, and strategic management'. In L. G.

Strategie Assets and Rent

325

Strategic Management Journal, Vol. 14, 33-46 (1993)

STRATEGIC ASSETS AND ORGANIZATIONAL RENT RAPHAEL AMIT Faculty of Commerce and Business Administration, University of British Columbia, Vancouver, British Columbia, Canada PAUL J.H. SCHOEMAKER Graduate School of Business, University of Chicago, Chicago, Illinois, U.S.A.

We build on an emerging strategy literature that views the firm as a bundle of resources and capabilities, and examine conditions that contribute to the realization of sustainable economic rents. Because of (1) resource-market imperfections and (2) discretionary managerial decisions about resource development and deployment, we expectfirmsto differ (in and out of equilibrium) in the resources and capabilities they control. This asymmetry in turn can be a source of sustainable economic rent. The paper focuses on the linkages between the industry analysis framework, the resource-based view of the firm, behavioral decision biases and organizational implementation issues, it connects the concept of Strategic Industry Factors at the market level with the notion of Strategic Assets at the firm level. Organizational rent is shown to stem from imperfect and discretionary decisions to develop and deploy selected resources and capabilities, made by boundedly rational managers facing high uncertainty, complexity, and intrafirm conflict.

INTRODUCTION However they phrase them, executives often examine such questions as, 'What makes us distinctive or unique?'; 'Why do some and not other customers buy from us?'; 'Why are we profitable?'. Typical answers might refer to the firm's 'technical know-how,' 'responsiveness to market needs,' 'design and engineering capability,' or 'financial resources.' The common theme among these responses is that management deems some firm-specific resources and capabilities to be crucial in explaining a firm's performance. While empirical models may, ex post, point to a limited set of resources and capabilities that explain some of the firm's past performance, ex ante such models offer limited insight into the Key words: Bounded rationality, heuristics, organizational rents, resource view, strategic assets, strategic industry factors 0143-2095/93/010033-14$12.00 © 1993 by John Wiley & Sons, Ltd.

dimensions of competition that will prevail in the future. For managers, the challenge is to identify, develop, protect, and deploy resources and capabilities in a way that provides the firm with a sustainable competitive advantage and, thereby, a superior return on capital. Managerial decisions concerning such resources and capabilities are ordinarily made in a setting that is characterized by: (1) Uncertainty about (a) the economic, industry, regulatory, social, and technological environments, (b) competitors' behavior, and (c) customers' preferences; (2) Complexity concerning (a) the interrelated causes that shape the firm's environments, (b) the competitive interactions ensuing from differing perceptions about these environments; and by (3) Intraorganizational conflicts among those who make managerial decisions and those affected by them. These conditions of uncertainty, complexity, and conflict are usually difficult to articulate or model. For example, the exact relationships between the firm's bundle of capaReceived 18 June 1990 Revised 18 August 1992

326 34

R. Amit/P. Schoemaker R. Amit and P.J.H.

Schoemaker

bilities and its performance may be unclear in the present, let alone the future.1 By explicitly addressing these dimensions of the managerial challenge, our paper attempts to link the 'industry analysis framework' with the 'resource view of the firm' and highlight the human limitations in crafting firm strategy. We start by briefly reviewing the existing literature on the resource-based view and defining the terms we use. We proceed by articulating our view and contribution to the theory. We end by examining the theory in the context of multiple dimensions and emphasizing the heuristic nature of organizational rent creation. LITERATURE AND DEFINITIONS A growing body of empirical literature points to the importance of firm-specific factors in explaining variations in economic rent2 (Jacobson 1988; Hansen and Wemerfelt, 1989). For example, Cool and Schendel (1988) reported significant and systematic performance differences among firms belonging to the same strategic group within the U.S. pharmaceutical industry. Additionally, Rumelt (1991) found that business units differ far more within than across industries. Theorists have long recognized the importance of firm differences and distinctive competencies (Selznick, 1957; Ansoff, 1965; Andrews, 1971; Hofer and Schendel, 1978). Current managerial writings such as Irvin and Michaels (1989), Wemerfelt (1989), Prahalad and Hamel (1990), Grant (1991), or Stalk, Evans, and Shulman (1992) further evidence a continuing interest in core skills and capabilities as a source of competitive advantage. Vasconcellos and Hambrick (1989) recently conducted an empirical, ex post test of the longstanding strategy premise that an organization's success depends on the match between its

1

Lippman and Rumelt (1982) refer to this as 'causal ambiguity.' 2 Economists commonly distinguish among three types of rent: Ricardian rents are extraordinary profits earned from resources that are in fixed or limited supply. Pareto rents (or quasi rents) refer to the difference between the payments to a resource in its best and second best use. Lastly, Monopoly rents stem from collusion or government protection. Klein, Crawford and Alchian (1978) examine quasi-rents in the context of vertical integration.

strengths and the Key Success Factors (KSF)3 in its environment. Using a range of mature industrial product industries, their empirical findings showed that organizations which rated highest on industry KSF clearly outperformed their rivals. Although this analysis provides an important test of a core thesis in strategy, it also raises further questions. First, the Vasconcellos and Hambrick (1989) study considers the industry as the primary unit of analysis, whereas managers operate from a firm perspective. Second, the empirical analysis is ex post, whereas managers need to make resource deployment decisions ex ante, which involves uncertainty, complexity and organizational conflict. Third, it should be recognized that if all firms score high on the presumed KSF, these factors will cease to be KSF. Thus, we need to introduce sustainable asymmetry into the analysts, possibly stemming from mobility barriers, organizational inertia, heterogeneous expectations, failures in resource markets, and so forth. The use of KSF as a core concept in strategy was recently critiqued by Ghemawat (1991a) as lacking: (1) identification (there may be many success factors, making it hard to decide which ones to focus on); (2) concreteness (ambiguity about the causal processes that tie the firm's success factors to its performance); (3) generality (to be success factors they must be undervalued; i.e., the cost benefit ratio associated with their development must be less than one); and (4) necessity (the failure of the success factor approach to account for dynamic aspects of strategy). Whereas we agree with Ghemawat (1991a) about these challenges, it should be pointed out that without uncertainty, complexity, and conflict, there would be no room for discretionary managerial decisions on strategy crafting. Only differences in initial endowments, or luck, could underlie asymmetric performance in that case. Since KSF notions are commonly used by strategy scholars and managers alike, they need to be related more carefully to strategy theory. An emerging theoretical perspective—that of the firm as a collection of resources and capabilities s

There are numerous interpretations in the Marketing and Strategic Management literature concerning the meaning of KSF. See for example Thompson and Strickland (1990).

required for product/market competition—provides one such undeipinning. This Resource View of the firm (Coase, 1937; Penrose, 1959; Nelson and Winter, 1982; Teece, 1982; Rumelt, 1984; Wernerfelt, 1984; Barney, 1986a, 1986b, 1989, 1991; Dierickx and Cool, 1989a, 1989b, 1990; Teece, Pisano, and Shuen, 1990; Conner, 1991; Ghemawat, 1991b; Peteraf, 1991) focuses on factor market imperfections and highlights the heterogeneity of firms, their varying degrees of specialization, and the limited transferability of corporate resources. The resource perspective complements the industry analysis framework (Porter, 1980; Schmalensee, 1985). The latter focuses on product markets; it views the sources of profitability to be the characteristics of the industry as well as the firm's position within the industry. The resource view holds that the type, magnitude, and nature of a firm's resources and capabilities are important determinants of its profitability. In developing the theoretical foundations, we shall build on both perspectives: The resource view of the firm and the industry analysis framework. In addition, we introduce a third perspective, that of Behavioral Decision Theory (BDT). This new field explicitly acknowledges that managers often make suboptimal choices, be it in personnel selection or in crafting their firm's strategy. BDT can shed light on how boundedly rational managers cope with the kinds of uncertainty and complexity referred to above. Unlike the resource view, which focuses on failures in resource markets, the BDT perspective highlights cognitive imperfections that, while internal to the firm (e.g., internal conflict, cognitive biases of managers, etc. 4 ), have a great impact on the firm's approach to its external environment. To date, few links have been drawn between the BDT literature, the industry analysis framework and the resource view of the firm (for an exception, see Zajac and Bazerman, 1991). Before proceeding to the theory section, where these perspectives are examined and integrated, we clarify below the key terms and concepts we use.

Strategie Assets and Rent

327

Strategic Assets

35

the firm. Resources are converted into final products or services by using a wide range of other firm assets and bonding mechanisms such as technology, management information systems, incentive systems, trust between management and labor, and more. These Resources consist, inter alia, of knowhow that can be traded (e.g., patents and licenses), financial or physical assets (e.g., property, plant and equipment), human capital, etc. 3 Capabilities, in contrast, refer to a firm's capacity to deploy Resources, usually in combination, using organizational processes, to effect a desired end. They are information-based, tangible or intangible processes that are firmspecific and are developed over time through complex interactions among the firm's Resources. They can abstractly be thought of as 'intermediate goods' generated by the firm to provide enhanced productivity of its Resources, as well as strategic flexibility and protection for its final product or service. Unlike Resources, Capabilities are based on developing, carrying, and exchanging information through the firm's human capital. Itami (1987) refers to information-based Capabilities as 'invisible assets.' He notes that some of the firm's invisible assets are not carried by its employees but rather depend on the perceptions of the firm's customer base (as brand names may do). Capabilities are often developed in functional areas (e.g., brand management in marketing) or by combining physical, human, and technological Resources at the corporate level. As a result, firms may build such corporate Capabilities as highly reliable service, repeated process or product innovations, manufacturing flexibility, responsiveness to market trends, and short product development cycles.

The firm's Resources will be defined as stocks of available factors that are owned or controlled by

Some of the firm's Resources, but especially its Capabilities, may be subject to market failure; that is, an inability to trade these factors in perfect markets. Multiple sources of market failure have been suggested: Williamson (1975) points to small numbers, opportunism, and information impactedness; Klein, Crawford and Alchian (1978) focus on factor specialization in terms of use or location; Caves (1984) highlights sunk costs, and suggests that a factor's value is inversely related to the extent of its specialization

* Penrose's (1959) seminal work also addresses some of these intrafìrm issues.

' See Grant (1991) for a detailed description of various types of both tangible and intangible resources of the firm.

Definitions

328 36

R. Amit/P. Schoemaker R. Amit and P.J.H.

Schoemaker

Table 1. General characteristics of strategic industry factors (SIF)* a. b. c. d. e. f. g. h. i. j.

Stock type Resources and Capabilities that ex post are shown to be key determinants offirmprofitability in an industry; Determined at the market level through complex interactions among industry rivals, new entrants, customers, regulators, innovators, suppliers, and other stakeholden; Strategic in that they are subject to market failures and may be the basis for competition among rivals; The bundle of SIF changes over time and is not known ex ante; Their development takes time, skill, and capital; they may be specialized to particular uses; Investments in them are largely irreversible (i.e., entail sunk costs); Their values deteriorate or appreciate, over time, at varying rates of change; Their pace of accumulation may be affected by a range of managerial actions (policy levers) and by the magnitude of other Resources and Capabilities that are controlled by industry rivals. One cannot easily speed up their development (e.g., doubling the investment will not usually halve the time); Their value to any particular firm may depend on its control of other factors—the complementarity property. For instance, the value of afirm'sproduct design capability may depend upon the effectiveness of its distribution network; Not all aspects of their development and interactions will be known or controllable.

This table synthesizes notions from Penrose, 1959; Nelson and Winter, 1982; Teece, 1982; Rumelt, 1984; Wernerfelt, 1984; Barney, 1989, 1991; Dierickx and Cool, 1989a, 1989b, 1990; Teece el. al., 1990; Conner, 1991; Ghemawat, 1991b; Peteraf, 1991.

for a particular use or industry setting.6 We thus define the firm's Strategic Assets as the set of difficult to trade and imitate, scarce, appropriable and specialized Resources and Capabilities that bestow the firm's competitive advantage. When the industry (or product market) is the unit of analysis, one may observe that, at a given time, certain Resources and Capabilities which are subject to market failures, have become the prime determinants of economic rents. These will be referred to as Strategic Industry Factors (SIF). For instance, Ghemawat (1991b) suggests that one may classify industries in terms of the 'strategic factors that drive competition in them by virtue of dominating the structure of sunk costs incurred in the course of competition.' Strategic Industry Factors, in this context, are characterized by their proneness to market failures and subsequent asymmetric distribution over firms. By definition, Strategic Industry Factors are determined at the market level through complex interactions among the firm's competitors, customers, regulators, innovators external to the industry, and other stakeholders. Their main characteristics are articulated in Table 1. It is important to recognize that the relevant

set of Strategic Industry Factors changes and cannot be predicted with certainty ex ante.7 The challenge facing a firm's managers is to identify, ex ante, a set of Strategic Assets (SA) as grounds for establishing the firm's sustainable competitive advantage, and thereby generate Organizational Rents. These are economic rents that stem from the organization's Resources and Capabilities, and that can be appropriated by the organization (rather than any single factor). This requires managers to identify the present set of Strategic Industry Factors (SIF) as well as to assess the possible sets of SIF that may prevail in the future. Also, decisions on the further development of existing and new Strategic Assets—those that are most likely to contribute to the creation and protection of economic rents—need to be made. Not every firm will succeed with its targeted set of SA, as their applicability and relevance ultimately hinges on the complex interaction referred to above. Examples of possible SA include: Technological capability; fast product development cycles; brand management; control of, or superior access to, distribution channels; a favorable cost structure; buyer-seller relationships; the firm's installed user base; its R&D capability; the firm's service

6

7

The roles of factor specialization and sunk costs in a firm's ability to earn economic rents have been examined by Klein el al. (1978), as well as by Baumol, Panzar, and Willig (1982).

While it may not be possible to identify ex ante the relevant set of strategic assets, one can screen out those assets that are not strategic.

Firm

Strategic Assets and Rent

329

Strategic Assets

37

Industry

Resources

Capabilities

• externally available & transferable • owned or controlled by the finn • convertible

• information based organizational processes • firm specific t tangible or intangible • intermediate goods

Rivals

Customers

\

/ Strategic Industry Factors

Substitutes

• industry • R&C subjectto m a t e ! failure • affect industry profitabity

Strategic Assets

• change & • a subset of the • firm's R&C subject • to market failure • • overlap with strategic • industry factors • • uncertain ex-ante • form the basis of the firm's competitive strategy • determine organizational rents

Entrants

subject to ex-ante uncertainty

non-tradable complementary scarce appropriable firm specific

/

Environmental

\

Suppliers

Factors (e.g. technology, regulation)

Figure 1.

Key constructs

organization; its reputation and so forth. The Resources and Capabilities (R&C), whose economic relationships between industry determined Strategic returns are appropriable by the firm. The basic Industry Factors, and firm level Resources, Capabili-idea that underlies this perspective, cited earlier ties, and Strategic Assets, are depicted in Figure l. 8 as the Resource-Based View Of The Firm, is that marshalling a set of complementary and specialized Resources and Capabilities which are scarce, durable, not easily traded, and difficult to imitate, A RESOURCE VIEW OF STRATEGIC may enable the finn to earn economic rents. Thus, ASSETS according to the resource perspective, the value of By focusing on the firm as the relevant unit of a firm's Strategic Assets extends beyond their analysis, managers are concerned with the creation contribution to the production process. It depends of a bundle of tangible as well as intangible on a wide range of characteristics (see Figure 2), and varies with changes in the relevant set of Strategic Industry Factors, as depicted by Figure 1. ' Note that we abandon from here on the term Key Success The supposition is that, even in equilibrium, firms Factors, because of its many possible interpretations and may differ in terms of the Resources and Capabilities uses.

330

38

R. Amit/P. Schoemaker

R. Amit and P. LH. Schoemaker

Figure 2. Desired characteristics of the firm's resources and capabilities

they control, and that such asymmetric firms may coexist until some exogenous change or Schumpeterian shock occurs (Schumpeter, 1934).9 Economic rents, in this setting, derive from asymmetry in initial resource endowments, resource 9

The assumption of heterogeneous firms controlling resources that are not perfectly mobile (i.e., that cannot be easily bought, sold or imitated) is essential to the existence of such an equilibrium. Lippman and Rumelt (1982) and Barney (1986a, 1986b) articulate some of the reasons for imperfect imitabUity. These include unique historical conditions, causal ambiguity; and complexity. Ghemawat (1991b) refers to these conditions as intrinsic inimitability and therefore the firm's factor combinations are viewed as intrinsically heterogeneous. He suggests that less stringent conditions (e.g., imitation being costly but not infeasible) may be sufficient for sustainability. Relatedly, Peteraf (1991) equates resource heterogeneity to differential levels of factor efficiency.

scarcity, limited transferability of Resources, imperfect substitutability, and appropriability.10 Barney (1986a, 1986b, 1989, 1991), Dierickx and Cool (1989a, 1989b, 1990), and Ghemawat (1991b) provide incisive discussions of desired attributes of such firm Resources. Figure 2 summarizes the primary determinants of the rent producing capacity of a firm's Strategic Assets. In general, the strategic value of a firm's Resources and Capabilities is enhanced the 10 Whereas Industrial Organization economics often looks outside the firm to explain sustained superior performance by examining, for example, various market structures, alternative regulatory settings, collusive relationships, or substitute technologies, the source of rents according to the resource perspective is internal.

more difficult they are to buy, sell, Imitate or substitute. For example, invisible assets such as tacit organizational knowledge or trust between management and labor cannot be traded or easily replicated by competitors since they are deeply rooted in the organization's history. Such firm-specific and often tacit assets accumulate slowly over a period of time (i.e., they are history-dependent state variables. See Dierickx and Cool 1989a, 1989b, 1990). The focus here is not just on the material aspects of Resources and Capabilities, but especially on their transformational characteristics. These are often specific to a firm and/or to a particular industry at a given point in time. This idiosyncracy makes them difficult to imitate and their development time cannot be easily compressed. In addition, the applicability of the firm's bundle of Resources, and Capabilities to a particular industry setting (i.e., the overlap with the set of Strategic Industry Factors), will determine the available rents. Managers influence the development and deployment of Strategic Assets by adopting a process perspective (in contrast to an input-output model). This perspective recognizes distinct phases of development, the importance of feedback, and the need for vision. It also entails careful scripting of how Resources, information and people are combined and sequenced over time in order to evolve specific Capabilities. In this sense, the viewpoint is essentially an institutional one (de Gregori, 1987). Dierickx and Cool (1989a, 1989b) especially highlight the importance of processes for asset accumulation and their impact on inimitability of the firm's Resources. The firm's Strategic Assets may further exhibit complementarity in deployment or application (Barnard, 1938); that is, the strategic value of each asset's relative magnitude may increase with an increase in the relative magnitude of other Strategic Assets (also known as positive externalities; see Dierickx and Cool, 1990). An example is Teece's (1986) notion of co-specialized assets— those for which there is a bilateral dependence in application. Under complementarity, the combined value of the firm's Resources & Capabilities may be higher than the cost of developing or deploying each asset individually. Conversely, the strategic value of the firm's Resources ά

Strategic Assets and Rent

331

Strategic Assets

39

Capabilities declines to the extent that they are substitutes.11 The more firm-specific, durable and scarce Strategic Assets are, the more valuable to the firm can be their deployment, for at least three reasons. First, if few other firms have R&C that are in high demand and are difficult to imitate, fewer firms will pursue market strategies based thereon, since others would find these strategies too costly and time consuming.12 Second,firm-specificityand the presence of transaction costs suggest that the value of some Resources and Capabilities will be lower for certain firms. Third, the more durable they are, the smaller will be the investment required to offset their depreciation, if any.13 These characteristics of the firm's assets emphasize the trade-off between the specialization of assets (a necessary condition for rent) and the robustness of these assets across alternative futures (see Schoemaker, 1992a). The trade-off between specialization and robustness is only partial, as specialization can be of two kinds: (1) limited use or (2) unique use. Limited use entails reduced robustness in that the asset is of little value in particular states of nature. Uniqueness, in contrast, is defined relative to other players (rather than to states of nature) and need not be restricted in scope or by circumstance. Due to competitive pressures, the kinds of specialization that can yield positive rents tend to entail limited use (and hence, risk). Uniqueness, in contrast, may reflect historical accident or heterogeneous expectations as the primary reasons for non-imitation. In essence, firms develop specialized assets to enhance profits at the price of reduced flexibility in the face of Schumpeterian shocks. This tradeoff is, in our view, a core issue in deciding which R&C to develop. Sustainable advantage is obtained when existing and potential competitors

11 Diericlu and Cool (1989b, 1990) have introduced the notion of complementarity in asset accumulation (or interconnectedness) which refers to economies of scope in asset accumulation. This distinction highlights the dynamic nature of asset accumulation, whereas complementarity in asset deployment is a static notion. 13 The strategic value of R&C may not lie merely in the scarcity of natural resources such as land and oil reserves, but also in the ability to deploy concurrently in multiple uses such invisible firm-specific assets as culture, reputation, and relationships with suppliers and buyers. 13 Unlike physical capital, most capabilities are enhanced with use as more experience is gained.

332

40

R. Amit/P. Schoemaker

R. Amit and P.J.H. Schoemaker

(new entrants) lack either the ability or desire to imitate the rent-producing R&C. A firm's managers can lessen the incentives of competitors to imitate or develop close substitutes by, for example, erecting entry or mobility barriers or by building 'isolating mechanisms' (Rumelt, 1984). Like Ghemawat (1986), we focus here on aspects that relate to the firm's superior access to Resources. (Of course, competitive advantage may also arise from size and scope, as well as legal or other restrictions on competition.) Given the competitive and changing context in which managers must decide which R&C to develop as their firm's basis for competition, it is doubtfùl that decisions about which SA to develop and deploy can be optimally deduced from a general normative theory. More likely, continually changing heuristics will emerge that strive to better incorporate the uncertainty, complexity and organizational conflicts confronting managers.14 As such, our view extends that of Porter's (1980) by emphasizing not only the industry environment in determining future profit but especially the importance of managerial discretion and innovation in SA decisions. The latter are by no means foregone conclusions; the external environment is only one part of the economic rent story.

DECISIONS ABOUT STRATEGIC ASSETS In making investment decisions about Strategic Assets, managers face the daunting tasks of (1) anticipating possible futures, (2) assessing competitive interactions within each projected future, and (3) overcoming organizational inertia and internal dispute in order to realign the firm's bundle of SA. Recent psychological literature (Kahneman, Slovic, and Tversky, 1982) suggests that managers will approach this uncertainty, complexity, and conflict with considerable bias, illusion, and suboptimality. Even if highly simplified and abstracted, the associated SA decisions may not be solvable in closed-form equilibrium terms (although, see Camerer, 1991).15 14 Economic rent may accrue to firms with superior or more timely heuristics, thereby capitalizing on variable as well as bounded rationality (see Schoemaker, 1990). 19 For example, when modeled as a differential game, the problem will probably not be tractable. Closed or even openloop solutions are generally unattainable when confronted with

Uncertainty Under rational expectations, the SA challenge will largely vanish as managers will hold the same expectations about the set of SIF that will prevail in the future. Since they will maximize the expected value of returns, their initial SA endowments are the only source of variance regarding their behavior. In reality, however, managers face considerable uncertainty and ambiguity, stemming from new proprietary technologies, economic and political trends, competitive actions, changes in societal values, and corresponding shifts in consumer preferences. Pervasive uncertainty and ambiguity make it probable that managers will hold diverse expectations about such key variables as demand growth, price levels, costs, and consumer tastes. Further, their judgements and choices are likely to exhibit idiosyncratic aversions to risk and ambiguity (Kahneman and Tversky, 1979; Einhorn and Hogarth, 1986).16 The joint effects of heterogeneous beliefs and manager-specific decision processes (and biases) make equilibrium analyses hard to conduct for both managers and researchers. Coupled with overconfidence (Lichtenstein, Fischhoff, and Phillips, 1982) and a penchant for confirming over disconfirming evidence (Klayman and Ha, 1987), Strategic Assets choices under uncertainty may entail opposing biases whose net effects are hard to assess. For example, ambiguity aversion and underweighting of medium and high probabilities will normally lead to risk aversion. However, this tendency may be countered or mitigated by overconfidence and ambitious targets, either of which can induce strong risk-seeking.17 Consequently, the final SA investment decisions are a multiplicity of state and control variables in noncooperative multiplayer games. An added complication in our case arises from the difficulty of specifying the game in terms of the number of players, as well as the state, action, and pay-off spaces. 16 When gambles entail well-defined probabilities, most people exhibit risk aversion (except for low probability and pure loss gambles). If probabilities are ill defined (the case of ambiguity), even greater risk-aversion is encountered due to people's dislike to unknown risk. Most managerial decisions entail risk as well as ambiguity. 17 The predicted bias is toward risk-seeking for R&C that are deemed to be below some chosen reference point and toward risk-aversion for those that exceed this aspiration level (see Kahneman and Tversky, 1979). Thus, unrealistic goals or ambitious targets will likely result in unduly risky R&C decisions. For additional biases and indeterminacies in risk-taking see MacCrimmon and Wehrung (1986).

hard to predict without detailed micro-level knowledge of mánagers' reference points, problem framing, degrees of overconfidence, nonlinear weighting of probabilities, etc. (see Schoemaker, 1992b). A bounded rationality view (Simon, 1979) may nonetheless predict some overriding biases. For example, managers will probably over-emphasize past Strategic Industry Factors, and the SA associated therewith. People generally tend to repeat what was rewarded before. Consequently, managers might be too focused on past competitors and pay too much attention to recent experience. The latter is known as the recency effect which is closely linked to the more general notion of the availability heuristic (Tversky and Kahneman, 1974). If perceptions about strategy are unduly anchored on past SA, rent opportunities arise for firms that approach the future more flexibly and imaginatively. These may be new firms or incumbent ones that vigorously challenge their own beliefs. Past success may especially bias managers toward an illusion of control (Langer, 1975). Recent emphasis on the strategic importance of continual organizational learning (de Geus, 1988; Senge, 1990) underscore the special challenges posed by uncertainty and complexity, whether the firm has been successful or not. Complexity To keep SA decisions within cognitive bounds, managers must often and extensively simplify (Russo and Schoemaker, 1989). The kinds of simplification they engage in may lead to additional biases. Tversky and Kahneman (1981) offer persuasive examples of how simplified framing (such as isolating alternatives or expressing outcomes relatively) can lead to inconsistent decisions. Specifically, frames may (1) bound out important futures, competitors, or new technologies; (2) dictate the reference point relative to which SA are measured (e.g., Chrysler comparing its quality control capability to GM's rather than to Japan's Honda); and, (3) specify the yardsticks or metric used to measure SA (e.g., measuring quality in terms of defective parts per thousand vs. number and type of consumer complaints). Managers' attempts to understand present and past SIF may be hampered by additional biases.

Strategie Assets and Rent

333

Strategic Assets

41

In hindsight, chance and skill are often confused (Fischhoff and Beyth, 1975). Judgments about correlation or relative importance frequently miss important cues and interactions (Jennings, Amabile, and Ross, 1982; Hammond, 1955; Hogarth, 1987), especially if not driven by a causal theory. Imputations about causality, in turn, may be overly sensitive to temporal and spatial contiguity, covariation, and similarity of cause and effect (Einhorn and Hogarth, 1986). Unless aided by formal analyses, managers may easily misconstrue the industry's success factors and persist in erroneous beliefs about their firm's SA until proven wrong by competitors. Lindblom (1959) and Quinn (1980), among others, have highlighted the incremental way in which managers usually deal with complexity. Writers on policy formation have, in general, emphasized the contextual and labile nature of organizational decision making (Mintzberg, 1978; Isenberg, 1987, MacCrimmon and Wehrung, 1986). An example is Cohen, March, and Olsen's (1972) garbage can model, in which problems, solutions, hidden agendas, coalitions and so on mesh in complex ways to yield decisions. Mintzberg (1978) and Mintzberg and Waters (1983) further highlight the role of the firm's unconscious past. They view a firm's realized strategy (e.g., its SA decisions) to be a blend of rational, or at least intentional choices, and implicit or tacit forces within organizations (see also Hamel and Prahalad, 1989). The litany of biases mentioned above serves to underscore our main point here: Discretionary managerial decisions that relate to Strategic Assets are affected by a wide range of cognitive biases about the handling of uncertainty and complexity. This, in turn, creates suboptimality, imperfect imitability, and organizational rents for some firms.

Conflict Intraorganizational conflict is another serious challenge encountered by management in making SA decisions. Any change in the existing bundle of SA may benefit some employees and hurt others. Not only do complex agency problems (Jensen and Meckling, 1976; Fama and Jensen, 1983a, 1983b) exist in obtaining the necessary information and judgments concerning SA selection, but also issues of cooperation, trust, and

334 42

R. Amit/P. Schoemaker R. Amit and P.J.H.

Schoemaker

competence must be factored into the decision of which Resources and Capabilities to develop and how. Allison's (1971) classic treatment of the Cuban Missile Crisis illustrates clearly the importance of organizational and political dimensions, in addition to rational ones, for setting policy. The key point is that organizations are complex social entities with their own inertia and constraints. The issue is not simply to select the subset of Resources and Capabilities that is most likely to yield high rents, but to make organizational participants an integral part of such decisions. Among other things, this poses problems of nestedness; for example, SBU level choices impact divisional as well as corporate Capabilities and vice versa. The convenient view that organizations have carefully solved their principal-agent problems and need only select from the implicit market for Resources and Capabilities, which and how much of each to buy, denies the crucial role of asymmetric Resources and Capabilities as well as the complex decisions managers face.

sheds a different light on the Strategic Assets challenge as captured below. Industry Analysis excels in assessing the profit potential of various industry participants by focusing on the external competitive forces and barriers that prevail in different product/market segments. Further, it is essential in deriving a set of Strategic Industry Factors. It is incomplete, however, in that it treats the firm largely as a black box (i.e., a faceless, unitary actor), while deemphasizing the role of managerial discretion. Assuming high rationality and substitutability of executive talent, industry analysis logically deduces the end-game consequences of differences in participants' initial conditions (for a particular industry structure, technology, and action space). Thus, the focus is on rent distribution in equilibrium, given initial firm asymmetries, industry structure, and known rules of the game.

The Resource View, in contrast, highlights imperfections in factor markets, resulting in systematic firm differences. Limited transferability of Resources, scarcity, complementarity In sum, as the firm's environment changes, and appropriability in turn give rise to rent different sets of Strategic Assets may have to opportunities. Economic rents, in this view, be developed by firms. Core Capabilities, by derive from properties unique to the firm's definition, cannot be purchased off the shelf but Resources and Capabilities. The focus is thus require strategic visions, development time, and more internal and institutional, recognizing the sustained investment. Decisions about Strategic often slow and evolutionary path by which firmAssets (i.e., the subset of Resources and Capabili- specific Capabilities develop (e.g., see Nelson ties that bestows sustainable competitive and Winter, 1982.) These Capabilities may include advantage) are among the most complex that executive talent, culture and other less tangible managers encounter. They are characterized by dimensions that in standard models of rational 19 high uncertainty, complexity, and conflict, to an behavior have received limited attention. Also, extent that defies optimization. Indeed, this lack the exclusive focus on equilibrium and structural of solvability is a necessary condition for their dimensions is absent. Instead, disequilibrium and strategic importance and positive rent potential. process dynamics loom primary. Behavioral Decision Theory (BDT) complements the resource perspective in explicitly acknowledging bounded rationality and, in parSTRATEGIC ASSETS DEVELOPMENT: ticular, the crucial roles of problem framing and A MULTIDIMENSIONAL VIEW heuristic decision-making. Differences in decision The above analysis of Strategic Assets underscores frames and heuristics give rise to 'variable the need for a multidimensional approach; one rationality' among and within players over time that includes internal and external elements, (see Schoemaker, 1990). A rational end-game static and dynamic aspects, and rational as well as behavioral considerations.18 Each perspective sociological, political, anthropological). To integrate these 18

While we hold that these dimensions need to be reflected in any comprehensive analyses of firm's Strategic Assets, there may well be other relevant dimensions (e.g., ecological,

additional dimensions, however, is beyond our present scope. " Some of this is changing. For instance rational models have been developed concerning the role of culture (Camerer and Vepsalainen, 1988) and reputation (Weigelt and Camerer, 1988).

analysis would largely ignore such factors since it generally assumes constant rationality.20 In actuality, however, managers are hardly playing a well-defined end game. Logical consequences of moves are seldom ascertainable and equilibrium solutions are not usually transparent in complex strategy decisions. Because the rules of the game, the number of players, and the action space are seldom fixed, creative changes and innovations are permitted, which makes predictions of outcomes especially difficult. Reliance on heuristics and on a limited repertoire of responses, punctuated by occasional bold or creative moves, introduces complexities whose net effects are hard to assess. Players generally harbor imperfect comprehensions of the deeper relationships operative in the industry or, indeed, within their firm. In this view, strategy becomes partly a shot in the dark and partly an exercise in heuristic creativity aimed at overcoming biases and blind spots (Zajac and Bazerman, 1991). These biases will not be just individual or cognitive; many concern group biases (e.g., groupthink) and may be affective in nature, such as wishful thinking, dissonance reduction, etc. (see Russo and Schoemaker, 1989). The BOT perspective is especially important in light of the pervasive uncertainty and complexity surrounding SA decisions. Any industry or market segment will undergo Schumpeterian shocks such that most equilibria (if computable at all) will have finite lives. Robust strategies thus must pay attention to disequilibrium, uncertain futures and ambiguous relationships. Without ambiguity and complexity, the SA question would perhaps be reducible to a rational end-game analysis. In practice, however, it is about the fashioning and deployment of firm-specific Capabilities whose rents depend partly on unfathomable futures. In terms of theoretical underpinnings, various attempts have been made to model the effects of uncertainty or ambiguity on individual decision making (Einhorn and Hogarth, 1986) as well as markets (Kleindorfer and Kunreuther, 1982). The dimension of complexity has yet to see 30

Variable rationality refers to actors differing in the degree to which they exhibit bounded rationality. A rational endgame analysis is one in which all possible moves and countermoves are identified and evaluated.

Strategie Assets and Rent

335

Strategic Assets

43

significant formal treatment (although, see Rosenhead, 1980). In psychology, however, various models and techniques exist to depict how people represent complex problem situations, ranging from scripts and schema to cognitive maps (for a review see Klayman and Schoemaker, 1992). Also, numerous heuristic guidelines exist for managers on how to cope with and manage complexity, such as scenario analysis (Wack, 1985a, 1985b; Schoemaker, 1991). Our further emphasis on conflict and organization inertia brings to the fore implementation and other intraorganizational problems in the development and deployment of Strategic Assets. The resource and behavioral perspectives refer to these organizational issues but do not develop them. Principal-agent theory provides a highly rational treatment of incentive problems, with abstract links to the origin, scope and organizational form of firms (e.g., partnerships vs. corporations). Transaction cost economics focuses more generally on organizational structure (e.g., U- vs. M-form) and scope, while placing greater emphasis on bounded rationality and internal firm complexity. Organization theory, in contrast, has been more descriptive and process oriented in seeking to understand how firms control and coordinate activities. Rather than making conflict or transactions the unit of analysis, organization theory focuses on systemic aspects, in particular the interactions among such subsystems as the firm's structure, processes, rewards, culture, people and technology. These can explain firm inertia and the adaptation difficulties encountered when the environment changes and managers attempt to redirect their firm's Strategic Assets.

CONCLUSION We have sought to replace the strategy field's concept of Key Success Factors with the notions

of: (1) Strategic Industry Factors, the set of Resources & Capabilities that has become the prime determinant of economic rents for industry participants; and (2) Strategic Assets, a firm level construct, referring to the set of firm specific Resources and Capabilities developed by management as the basis for creating and protecting their firm's competitive advantage. The rent producing capacity of these Strategic Assets depends, in part, on their own unique character-

336 44

R. Amit/P. Schoemaker R. Amit and P.J.H. Schoemaker

istics as well as on the extent to which they as the transaction cost view, an emphasis on the overlap with the industry-determined Strategic uniqueness and limited mobility of Resources and Industry Faeton. Capabilities. However, it is not market power Building on insights from the Resource View (IO view) per se, or greater operating efficiency of the firm, and Behavioral Decision Theory, (neoclassical and Chicago school views) that we identified important theoretical features of produces organizational rents, although these Strategic Assets and the conditions under which may be consequences. In this paper uniqueness they could produce organizational rents. The and low mobility of Resources and Capabilities managerial difficulty of identifying, developing, stem from imperfect and hard to predict decisions and deploying an appropriate mix of SA was by boundedly rational managers facing high highlighted in the discussion. Owing to uncer- uncertainty (a la Schumpeter), complexity, and tainty, complexity, and conflict (both in and intrafirm conflict. We thus strengthen the outside the firm), different firms will employ resource view by adding behavioral decision different Strategic Assets, without any one set making biases and organizational implementation being provably optimal or easily imitated. At aspects as further impediments to the transferbest, managers can devise heuristic solutions that ability or imitability of a firm's Resources and navigate between the numerous cognitive and Capabilities. affective biases characteristic of humans and organizations. We articulated a multidimensional view for the crafting of Strategic Assets, in relation ACKNOWLEDGEMENTS to market-determined Strategic Industry Factors. Its dimensions consist of (1) industry analysis, Karel Cool, Ingemar Dierickx, James Emery, (2) the resource perspective and (3) behavioral Robin Hogarth, Margaret Peteraf, Birger Werdecision theory. The latter perspective emphasizes nerfelt and two anonymous reviewers are the pervasive uncertainty and complexity faced acknowledged for their helpful comments on by managers, often resulting in suboptimal earlier versions. Insightful discussions with Jay Strategic Assets decisions. In this context, the Barney and Chaim Fershtman contributed to the role of intraorganizational conflict and inertia development of this paper and are gratefully were identified as important barriers to acknowledged. implementing changes to the firm's bundle of Strategic assets. Throughout, Strategic Assets decisions were examined in light of resource market imperfections, bounded and variable rationality within and across firms. If optimal solutions were derivable for a firm's Strategic Assets, the latter would largely vanish. Barring market or cognitive imperfections, allfirmswould envision and pursue an optimal strategy with zero expected rents. As such, the existence of Strategic Assets and presence of bounded rationality are closely linked. A normative Strategic Assets theory that could systematically lead to the creation of sustainable rents is implausible due to competitive pressures. Our paper instead sought to develop a behavioral view of Strategic Assets, with limited prescriptive advice on how to target, develop and deployfirm-specificStrategic Assets. In concluding, it may be useful to place our view of organizational rent creation by firms within the larger framework articulated by Conner (1991). We share with the resource view, as well

REFERENCES Allison, G. Essence of Decision: Explaining the Cuban Missile Crisis, Little, Brown, Boston, MA, 1971. Andrews, K. R. The Concept of Corporate Strategy, Irwin, Home wood, IL, 1971. Ansoff, H. I. Corporate Strategy, McGraw Hill, New York, 1965. Barnard, C. I. The Functions of the Executive, Harvard University Press, Cambridge, MA, 1938. Barney, J. B. 'Strategic factor markets: Expectations, luck, and business strategy', Management Science, 42, 1986a, pp. 1231-1241. Barney, J. B. Organization culture: Can it be a source of sustained competitive advantage?', Academy of Management Review, 11, 1986b, pp. 656-665. Barney, J. B. 'Asset stocks and sustained competitive advantage: A comment', Management Science, 35, 1989, pp. 1511-1513. Barney, J. B. 'Firm resources and sustained competitive advantage', Journal of Management, 17(1), 1991, pp. 99-120. Baumol, W. I., J. C. Panzar and R. D. Willig. Contestable Markets and the Theory of Industry

Structure, Haicourt Brace Jovanovich, New York, 1982. Camerer, C. F. 'Dose strategy research need game theory?', Strategic Management Journal, 12, 1991, pp. 137-152. Camerer, C. F. and A. Vepsalainen. 'The economic efficiency of corporate culture', Strategic Management Journal, 9, 1988, pp. 115-126. Caves, R. E. 'Economic analysis and the quest for competitive advantage', American Economic Review, 74(2), 1984, pp. 127-132. Co ase, R. H. 'The nature of the firm', Economica, 4, 1937, pp. 331-351. Cohen, M. D., J. G. March and J. P. Olsen. Ά garbage can model of organizational choice', Administrative Science Quarterly, 17, 1972, pp. 1-25. Cool, K. and D. Schendel. 'Performance differences among strategic group members', Strategic Management Journal, 9(3), 1988, pp. 207-224. Conner, K. R. Ά historical comparison of resourcebased theory and five schools of thought within industrial organization economics: Do we have a new theory of the firm?', Journal of Management, 17(1) 1991, pp. 121-154. de Geus, A. 'Planning as learning', Harvard Business Review, March-April, 1988, pp. 70-74. de Gregori, T. R. 'Resources are not; they become: An institutional theory', Journal of Economic Issues, 21, 1987, pp. 1241-1263. Dierickx, I. and K. Cool. 'Asset stock accumulation and sustainability of competitive advantage', Management Science, 35, 1989a, pp. 1504-1511. Dierickx, I. and K. Cool. 'Competitive strategy, resource accumulation and firm performance,' Mimeo, INSEAD, May 1989b. Dierickx, I. and K. Cool. Ά resource based perspective on competitive strategy', Mimeo, INSEAD, September, 1990. Einhorn, Η. and R. Hogarth. 'Judging probable cause', Psychological Bulletin, 99, 1986, pp. 3-19. Fama, E. F. and M. C. Jensen. 'Agency problems and residual claims', Journal of Law and Economics, 26, 1983a, pp. 327-349. Fama, E. F. and M. C. Jensen. 'Separation of ownership and control', Journal of Law and Economics, 26, 1983b, p. 301. Fischhoff, Β. and R. Beyth. Ί knew it would happen— Remembered probabilities of once-future things', Organization Behavior and Human Performance, 13, 1975, pp. 1-16. Ghemawat, P. 'Sustainable advantage,' Harvard Business Review, (4, 1986, pp. 53-58. Ghemawat, P. Commitment, The Free Press, New York, 1991a. Ghemawat, P. 'Resources and strategy: An IO perspective', Mimeo, Harvard Business School, May 1991b. Grant, R. Β. Ά resource based theory of competitive advantage: Implications for strategy formulation', California Management Review, 33(3), 1991, pp. 114-135. Hamel, G. and C. K. Prahalad. 'Strategic intent', Harvard Business Review, β9(3), 1989, pp. 63-76.

Strategie Assets and Rent

337

Strategic Assets

45

Hammond, K. R. 'Probabilistic functioning and the clinical method', Psychological Review, 62, 1955, pp. 255-262. Hansen, G. S. and B. Wemerfelt. 'Determinants of firm performance: The relative importance of economic and organizational factors', Strategic Management Journal, 10, 1989, pp. 399-411. Hofer, C. W. and D. Schendel. Strategy Formulation: Analytical Concepts, West Publishing Company, St. Paul, MN, 1978. Hogarth, R. M. Judgment and Choice, Wiley, Chichester, 1987. Irvin, R. A. and E. G. Michaels III. 'Core skills: Doing the right things right', The McKinsey Quarterly, Summer 1989, pp. 4-19. Isenberg, D. J. 'The tactics of strategic opportunism', Harvard Business Review, 65(2), March 1987, pp. 92-97. Itami, Η. Mobilizing Invisible Assets, Harvard University Press, Boston, MA, 1987. Jacobson, R. 'The persistence of abnormal returns', Strategic Management Journal, 9, 1988, pp. 41-58. Jennings, D. L., Τ. M. Amabile and L. Ross. 'Informal covariation assessment: Data-based versus theorybased judgments'. In D. Kahneman, P. Slovic and A. Tversky (eds), Judgment Under Uncertainty: Heuristics and Biases, Cambridge Press, Cambridge, MA, 1982, pp. 211-230. Jensen, M. C. and W. Meckling. 'Theory of the firm: Managerial behavior, agency costs and ownership structure', Journal of Financial Economics, 1976, pp. 305-360. Kahneman, D. and A. Tversky. 'Prospect theory', Econometrica, 47(2) 1979, pp. 263-292. Kahneman, D., P. Slovic and A. Tversky. Judgment Under Uncertainty: Heuristics and Biases, Cambridge Press, Cambridge, MA, 1982. Klayman, J. and P. J. H. Schoemaker. 'Thinking about the future: A cognitive perspective', Journal of Forecasting, 1992 (in press). Klayman, J. and Y. Ha. 'Confirmation, disconfirmation, and information in hypothesis testing', Psychological Review, 94(2), 1987, pp. 211-228. Klein, B., R. Crawford and A. Alchian. 'Vertical integration, appropriable rents, and the competitive contracting process', Journal of Law and Economics, 21, 1978, pp. 297-326. Kleindorfer, P. R. and H. Kunreuther. 'Misinformation and equilibrium in insurance markets'. In J. Finsinger (ed.), Issues in Pricing and Regulation, Lexington Books, Lexington, MA, 1982, pp. 67-90. Langer, E. 'The illusion of control', Journal of Personality and Social Psychology, 32, 1975, pp. 311-328. Lichtenstein, S., Β. Fischhoff and L. D. Phillips. 'Calibration of probabilities: The state of the art to 1980'. In D. Kahneman, P. Slovic and A. Tversky (eds), Judgment Under Uncertainty: Heuristics and Biases, Cambridge Press, Cambridge, MA, 1982, pp. 306-334. Lindblom, C. E. 'The science of muddling through', Public Administration Review, 19, 1959, pp. 79-88. Lippman, S. and R. Rumelt. 'Uncertain imitability:

338 46

R. Amit/P. Schoemaker R. Amit and P.J.H. Schoemaker

An analysis of interfìrm difference in efficiency under competition', Bell Journal of Economics, 1982, pp. 41&-438. MacCrimmon, K. R. and D. A. Wehrung. Taking Risks, Free Press, New York, 1986. Mintzberg, H. 'Patterns in strategy formation', Management Science, 24, 1978, pp. 934-948. Mintzberg, H. and J. A. Waters. 'The mind of the strategist(s)'. In S. Srivastva, (ed)., The Executive Mind, Jossey-Bass, San Francisco, CA, 1983, pp. 58-83. Nelson, R. and S. Winter. An Evolutionary Theory of Economic Change, Harvard University Press, Cambridge, MA, 1982. Penrose, E. T. The Theory of Growth of the Firm, Wiley, New York, 1959. Peteraf, M. A. "The cornerstone of competitive advantage: A resource based view', Northwestern University, J. L. Kellogg Graduate School of Management, General Motors Research Center for Strategy in Management, Discussion Paper No. 90-29, March, 1991. Porter, M. Competitive Strategy, Free Press, New York, 1980. Prahalad, C. K. and G. Hamel, 'The core competence of the corporation', Harvard Business Review, 68(3), 1990, pp. 79-91. Quinn, J. B. Strategies for Change: Logical Incrementaiism, Irwin, Homewood, IL, 1980. Rosenhead, J. 'Planning under uncertainty (II): A methodology for robustness analysis', Journal of Operational Research Society, 31,1980, pp. 331-341. Rumelt, R. P. 'Towards a strategic theory of the firm'. In R. B. Lamb, (ed.), Competitive Strategic Management, Prentice Hall, Engelwood Cliffs, NJ, 1984, pp. 556-570. Rumelt, R. P. 'How much does industry matter?', Strategic Management Journal, 12(3), 1991, pp. 167-185. Russo, J. E. and P. J. H. Schoemaker. Decision Traps: Ten Barriers to Brilliant Decision Making and How To Overcome Them, Doubleday, New York, 1989. Schmalensee, R. 'Do markets differ much?' American Economic Review, 75, 1985, pp. 341-351. Schoemaker, P. J. H. 'Strategy, complexity and economic rent', Management Science, 36(10), 1990, pp. 1178-1192. Schoemaker, P. J. H. 'When and how to use scenario planning', Journal of Forecasting, 10,1991, pp. 549-564. Schoemaker, P. J. H. 'Developing strategic vision: A core capabilities approach applied to Apple Computer', Sloan Management Review, 1992a, (in press). Schoemaker, P. J. H. 'Determinants of risk-taking:

Behavioral and economic views', Journal of Risk and Uncertainty, 1992b, (in press). Schumpeter, J. The Theory of Economic Development, Harvard University Press, Cambridge, MA, 1934. Selznick, P. Leadership in Administration: A Sociological Interpretation, Harper and Row, New York, 1957. Senge, P. "The leader's new work: Building learning organizations', Sloan Management Review, Fall, 1990, pp. 7-23. Simon, H. A. Models of Thought, Yale University Press, New Haven, CT, 1979. Stalk, G., P. Evans and E. Shulman. 'Competing on capabilities: The new rules of corporate strategy', Harvard Business Review, March-April, 1992, pp. 57-69. Teece, D. J. 'Towards an economic theory of the multiproduct firm', Journal of Economic Behavior and Organization, 3, 1982, pp. 39-63. Teece, D. J. 'Profiting from technological innovation', Research Policy, 15, 1986, pp. 285-305. Teece, D. J., G. Pisano and A. Shuen. 'Firm capabilities, resources, and the concept of strategy', Mimeo, University of California at Berkeley, Haas School of Business, September, 1990. Thompson, A. A. and A. J. Strickland. Strategic Management: Concepts and Cases, Irwin, Homewood, IL, 1990. Tversky, A. and D. Kahneman. 'Judgment under uncertainty: Heuristics and biases', Science, 185, 1974, pp. 1124-1131. Tversky, A. and D. Kahneman. 'The framing of decisions and the psychology of choice', Science, 211, 1981, pp. 453-458. Vasconcellos, J. A. and D. C. Hambrick. 'Key success factors: Test of a general framework in the mature industrial-product sector', Strategic Management Journal, 10(4), 1989, pp. 367-382. Wack, P. 'Scenarios: Uncharted waters ahead', Harvard Business Review, 63(5), 1985a, pp. 72-89. Wack, P. 'Scenarios: Shooting the rapids', Harvard Business Review, 63(6), 1985b, pp. 139-150. Weigelt, K. and C. Camerer, 'Reputation and corporate strategy', Strategic Management Journal, 9, 1988, pp. 443-454. Wernerfelt, Β. Ά resource based view of the firm', Strategic Management Journal, 5,1984, pp. 171-180. Wernerfelt, B. 'From critical resources to corporate strategy', Journal of General Management, 5, 1989, pp. 171-180. Williamson, O. Markets and Hierarchies, Free Press, New York, 1975. Zajac, E. J. and Μ. H. Bazerman. 'Blind spots in industry and competitor analysis', Academy of Management Review, 16, 1991, pp. 37-56.

Teil VI Strategieauswahl

Zur Generierung, Bewertung und Auswahl angemessener Unternehmungsstrategien hält die präskriptive Strategieliteratur eine große Palette analytischer Instrumente bereit. Hierzu zählen die bereits in den Teilen IV und V erwähnten Techniken zur Durchführung externer und interner Umweltanalysen (z.B. Branchenstruktur-, Wertketten- und SWOTAnalysen). Darüber hinaus kommen in der Praxis zahlreiche Instrumente zur Prognose zukünftiger Konsequenzen von Strategiealternativen zur Anwendung (Klein und Linnemann, 1984). Diese Instrumente sind entweder quantitativer oder qualitativer Natur. Bei quantitativen Instrumenten (z.B. Trendextrapolationen oder finanzwirtschaftlichen Verfahren wie der Kapitalwert- oder der internen Zinsfußmethode) sind die Ergebnisse der Prognosen intersubjektiv eindeutig determinierbar, denn sie ergeben sich als Resultate mathematisch-statistischer Manipulationen. Qualitative Instrumente (z.B. das Cognitive Mapping oder die Szenariotechnik) beruhen hingegen oft auf subjektiven Bewertungen, die von Individuum zu Individuum variieren können. Die oben erwähnten analytischen Instrumente können noch durch eine große Zahl weiterer Verfahren ergänzt werden (Hax & Majluf, 1996). Die analytischen Instrumente der strategischen Planung sind allerdings nicht Gegenstand dieses Teils, und sie spielen auch insgesamt in diesem Buch nur eine untergeordnete Rolle, denn als Heuristiken der strategischen Entscheidungsfindung haben sie nur einen geringen Stellenwert für die Entwicklung einer Theorie des Strategischen Managements. Wichtiger ist im sechsten Teil eine Diskussion verschiedener Strategien aus inhaltlicher Sicht. Unterschiedliche Arten von Strategien haben auf den Ebenen der Gesamtunternehmung, der Geschäftsbereiche, der Funktionsbereiche und auf der kollektiven Ebene eine umfangreiche theoretische und empirische Auseinandersetzung ausgelöst. Diese Diskussionen über Strategieinhalte gehören zum „mainstream" der Strategieforschung, und sie werden in ihren wichtigsten Ausprägungen in den Kapiteln 9 bis 11 aufgearbeitet. Kapitel 9 thematisiert Gesamtunternehmungsstrategien, insbesondere die Strategien der Vertikalen Integration und der Diversifikation. Kapitel 10 wendet sich Geschäftsbereichsstrategien zu und greift die Diskussion über die generischen Strategien Porters auf.32 Kapitel 11 behandelt die inzwischen sehr umfangreiche Literatur zum Thema kollektive oder kooperative Strategien. Was den tatsächlichen Auswahlprozeß der in den Kapiteln 9 bis 11 behandelten Strategien angeht, so wird unterstellt, daß die Strategien bereits ausgewählt wurden, und zwar entweder im Rahmen eines (begrenzt)rationalen Prozesses der synoptischen Planung oder

32

Gesamtunternehmungs- und Geschäftsbereichsstrategien werden aus analytischen Gründen in diesem Buch getrennt voneinander behandelt, obwohl in der Praxis vielfältige Interaktionen zwischen diesen Strategieebenen bestehen. Darüber hinaus wird darauf verzichtet, den Funktionsbereichsstrategien ein gesondertes Kapitel zu widmen. Die grundlegenden theoretischen Zusammenhänge von Funktionsbereichsstrategien verlaufen parallel zu den Überlegungen, die für Gesamtunternehmungsund insbesondere Geschäftsbereichsstrategien gelten.

340

R. Κ. F. Bresser

im Rahmen eines Inkrementalprozesses, der, wie in Kapitel 1 beschrieben, davon ausgeht, daß eine realisierte Strategie sowohl intendierte als auch emergente Bestandteile enthält. Was interessiert, ist somit der Output eines Strategieentwicklungsprozesses. Inwieweit strategische Entscheidungsträger bei der Strategieauswahl mehr dem synoptischen Planungsmodell oder dem Inkrementalmodell folgen, ließe sich z.B. im Rahmen des in Kapitel 4 vorgestellten verhaltenswissenschaftlichen Forschungsmodells von Hambrick und Mason (1984) darstellen und untersuchen. Diese Kontroverse zwischen Planern und Inkrementalisten soll in diesem Teil allerdings nicht noch einmal aufgegriffen werden, um Redundanzen und Überschneidungen (insbesondere mit Kapitel 1) zu vermeiden.33 Während die Kapitel 9 bis 11 eine inhaltliche Ausrichtung haben, beginnt der Teil VI mit einer Prozeßperspektive. Kapitel 8 thematisiert das Phänomen der „Imitation" von bereits existierenden Unternehmungsstrategien. Strategieimitationen sind in der Unternehmungspraxis weit verbreitet, und es ist wichtig für eine Theorie des Strategischen Managements, diese Verhaltensweise einer Erklärung zuzuführen. Das Imitationsverhalten von Unternehmungen hat im „institutionalistischen Ansatz" der Organisationstheorie ihren wichtigsten theoretischen Begründungszusammenhang gefunden.

33

Obwohl sich die Ausführungen des Teils VI primär mit Strategieinhalten auseinandersetzen, sei hier nochmals an die in Kapitel 1, S. 14, vorgenommene Relativierung erinnert, wonach sich die Inhalte von Strategien in der Realität nur bedingt von den Prozeßaspekten und Kontexten der Strategiebildung trennen lassen. Eine solche Trennung erfolgt hier lediglich aus didaktischen Gründen, und der Leser bleibt aufgefordert, nachfolgende Ausführungen zu bestimmten inhaltlichen Gegebenheiten immer auch im Hinblick auf ihre Prozeß- und Kontextimplikationen zu durchdenken, und zwar insbesondere auch dann, wenn diese Querverbindungen von den jeweiligen Autoren nicht deutlich hergestellt werden.

8 Konvergierende Strategien

Während die Nichtimitierbarkeit von Ressourcen im RBA als eine zentrale Bedingung für die Entwicklung von Strategien gilt, die die Herstellung dauerhafter Wettbewerbsvorteile ermöglichen, sind Imitationen in der Unternehmungspraxis dennoch populär.34 Dieser Umstand steht nicht notwendigerweise im Widerspruch zum RBA, denn Unternehmungsleitungen können (und müssen) sich mitunter mit nur suboptimalen aber dennoch akzeptablen Strategien zufriedengeben. Die Tatsache, daß Strategieimitationen weit verbreitet sind, weist aber ebenfalls auf die Möglichkeit bestimmter Vorteile der Nachahmung hin, Vorteile, die von den Vertretern des RBA eventuell übersehen oder zumindest nicht zutreffend eingeschätzt werden. In der Managementliteratur gibt es einige theoretische Ansätze, die die Vorteilhaftigkeit der Imitation behandeln (Haunschild, 1993). Die Vertreter des Strategic-ChoiceAnsatzes sehen in der Imitation z.B. eine strategische Reaktion auf die Vorstöße von Wettbewerbern. Imitatoren ziehen Vorteile daraus, daß Pioniere bereits das Risiko der Entwicklung neuer Produkte oder Technologien übernommen haben (Lieberman & Montgomery, 1988). Vertreter der Theorien des organisationalen Lernens argumentieren ähnlich. Imitatoren erzielen Kostenvorteile, denn sie überlassen es anderen, die Kosten der Forschung und Entwicklung zu absorbieren (Dutton & Freedman, 1985; Levitt & March, 1988). Der institutionalistische Ansatz der Organisationstheorie argumentiert, daß Organisationen die Aktivitäten und Strukturen anderer Organisationen kopieren, um dadurch Legitimität zu erwerben (DiMaggio & Powell, 1983).

Der institutionalistische Ansatz (IA) Der IA hat in der strategischen Managementforschung die größte Aufmerksamkeit bei der Erklärung des Imitationsphänomens gefunden. Obwohl er seine Ursprünge in der soziologischen Organisationstheorie hat (Meyer & Rowan, 1977; DiMaggio & Powell, 1983), zählt der Ansatz heute zum „mainstream" des Strategischen Managements.

34

Ein von der Praxis zu Zwecken der Effizienzsteigerung stark genutztes Verfahren, das benchmarking", kann als eine Anleitung zur Imitation angesehen werden. Das benchmarking" hält Unternehmungen dazu an, die eigenen Aktivitäten und Leistungen kontinuierlich mit denen wichtiger Konkurrenten und anerkannter Branchenführer zu vergleichen. Schneidet eine Unternehmung dabei schlecht ab, werden Programme entwickelt, durch die die besten Praktiken der Branchenführer imitiert werden sollen (Main, 1992).

342

R. Κ. F. Bresser

DiMaggio und Powell (1991: 63f.)35 verwenden die Metapher vom „eisernen Käfig", um auf die allgegenwärtigen konvergierenden Kräfte aufmerksam zu machen, die sich auf moderne Institutionen auswirken. Einerseits sind Organisationen (z.B. Unternehmungen) in einen institutionellen Kontext36 eingebunden, der sich steuernd auf das Verhalten und die Strukturen von Organisationen auswirkt, und andererseits haben einzelne Organisationen selbst den Charakter von Institutionen, denn sie haben sich gemäß den Anforderungen ihres institutionellen Kontextes herausgebildet und stellen somit nachahmenswerte Modelle dar (Knyphausen, 1995: 159). Die institutionelle Strukturierung eines Feldes impliziert, daß Organisationen sich in ihren zentralen Verhaltensweisen und Strukturen im Zeitablauf aneinander angleichen, und zwar selbst dann, wenn sich durch eine solche Angleichung die technische Effizienz der einzelnen Organisationen nicht erhöhen läßt (DiMaggio & Powell, 1991: 64). Institutionelle Prozesse führen somit zu Konvergenz beziehungsweise Isomorphic. DiMaggio und Powell (1991: 67-71) unterscheiden drei Mechanismen, die zur Herausbildung institutioneller Isomorphic beitragen: Zwang, normativer Druck und Imitation. Ein Zwang zur Angleichung von Verhaltensweisen und Strukturen ergibt sich insbesondere aufgrund allgemein gültiger rechtlicher Rahmenbedingungen. Der Staat normiert durch seine Gesetze und Erfüllungsorgane, z.B. in den Umwelt- und Verbraucherschutzbereichen, im Arbeitsrecht und hinsichtlich geltender Rechnungslegungsvorschriften. Unter normativem Druck zu isomorphem Wandel verstehen DiMaggio und Powell die Auswirkungen der Professionalisierung. Durch die Professionalisierung bestimmter Berufssparten (das bedeutet unter anderem auch eine weitgehende Standardisierung der Ausbildungsgänge) werden Anwälte, Ärzte und auch Diplom-Kaufleute ähnliche kognitive Prozesse und Verhaltensweisen in der Praxis vollziehen, um anstehende Probleme zu lösen. Ein weiterer Impuls zur Strukturierung von Institutionen ist somit unvermeidbar. Der für dieses Kapitel wichtigste Mechanismus für das Entstehen institutioneller Isomorphic ist die Imitation. Wenn Organisationen sich in komplexen und unsicheren Umwelten bewegen, Umwelten, die sich rasch verändern, in denen die verwendeten Technologien nur schwer verständlich sind (March & Olsen, 1976) und in denen generell ein hohes Maß an Ambiguität hinsichtlich relevanter Ursache-Wirkungszusammenhänge besteht, dann kann das Suchen nach nachahmenswerten Modellen kostengünstige Problemlösungen ermöglichen (Cyert & March, 1963). Die Imitation z.B. besonders erfolgreicher anderer Unternehmungen wird dann zu einer rational begründbaren Entscheidung. So haben beispielsweise während der letzten Jahrzehnte viele amerikanische und europäische Automobilfirmen versucht, die Praktiken japanischer Unternehmungen zu imitieren, um die eigenen Produktivitätsprobleme zu lösen. Imitationen anderer Organisationen sind aber nicht nur im Angesicht unsicherer Organisationsumwelten sinnvoll. Sie erfolgen auch immer dann, wenn wichtige Kräfte der Gesellschaft (z.B. Konsumenten) ein bestimmtes Verhalten von Organisationen erwarten, da es ihren kulturellen Wer35

36

Hier wird auf die 1991er Version des Artikels Bezug genommen, der zuerst 1983 im American Sociological Review erschien. Die 1991er Version enthält mehrere Modifikationen gegenüber der Ursprungsfassung. DiMaggio und Powell (1991: 64f.) sprechen in diesem Zusammenhang von einem „organisatorischen Feld". Hierunter verstehen sie alle Organisationen, die in ihrer Gesamtheit ein anerkanntes Gebiet institutionellen Lebens konstituieren, also z.B. wichtige Lieferanten, die Konsumenten von Ressourcen und Produkten, Behörden sowie andere Organisationen, die ähnliche Dienstleistungen oder Produkte herstellen.

Konvergierende Strategien

343

ten entspricht. Imitationen haben dann nichts mit technischer Effizienz zu tun, sondern mit kulturell bedingten Erwartungshaltungen. Generell läßt sich somit festhalten, daß imitierende Organisationen durch ihr nachahmendes Verhalten in der Lage sind, die eigene Legitimität37 zu erhöhen (Suchman, 1995). Kopieren Unternehmungen erfolgreichere andere Unternehmungen, gewinnen sie Legitimiät zumindest gegenüber ihren Anteilseignern und anderen Kapitalgebern. Kopieren sie kulturell wünschenswerte (aber nicht notwendigerweise rentabilitätsfördernde) Praktiken, gewinnen Organisationen ebenfalls Legitimität, und zwar sowohl bei Anteilseignern und Fremdkapitalgebern als auch bei anderen gesellschaftlichen Kräften, denn sie können ihr Verhalten als ethisch und „der Norm entsprechend" darstellen.38 Die empirische Evidenz über das Entstehen institutioneller Isomorphic ist inzwischen recht umfangreich. Die frühesten Studien beschäftigen sich mit der Konvergenz von Organisationsstrukturen und Organisationspraktiken in organisatorischen Feldern (Meyer & Rowan, 1977; Tolbert & Zucker, 1983; Meyer et al., 1983; Singh, Tucker & House, 1986). Einige Untersuchungen fokussieren auf Top-ManagementEntscheidungen, die zwar wichtig sind, aber keinen direkten Bezug zu Unternehmungsstrategien aufweisen. So untersuchen z.B. Galaskiewicz und Wassermann (1989) das philantropische Verhalten von Unternehmungen und kommen zu dem Schluß, daß dieses weitgehend durch mimetische (auf Imitation beruhende) Isomorphic gekennzeichnet ist. Mezias (1990) kann für den Zeitraum von 1962 bis 1984 eine starke Konvergenz des Publizitätsgebarens bei den 200 größten US-Firmen (Fortune 200) nachweisen. Inzwischen liegen auch mehrere Untersuchungen zu Unternehmungsstrategien vor. Fligstein (1991) ermittelt, daß das Diversifikationsverhalten von Unternehmungen durch Imitationsverhalten geprägt ist. Havemann (1993) weist für eine Stichprobe amerikanischer Sparkassen nach, daß der Prozeß der Expansion in neue Märkte zu einer durch Imitation geprägten strategischen Isomorphic führt. Greve (1995) führt Nachweis darüber, daß selbst die Aufgabe bisheriger Unternehmungsstrategien durch Prozesse der Imitation gekennzeichnet ist. In der Gegenwart hat es den Anschein (ohne daß dieser Umstand bisher empirisch rigoros untersucht wurde), als ob „corporate spinoffs", also Desinvestitionen, institutionellen Imitationsprozessen unterlägen (Eaton, 1997). Während das Vorhandensein institutioneller Isomorphic sowie deren Bestimmungsfaktoren durch ein breites Feld empirischer Untersuchungen gekennzeichnet sind, wurde das Verhältnis zwischen Isomorphic und Legitimität bisher kaum erforscht. Dies ist verwunderlich, denn gemäß dem IA ist die Erzielung von Legitimität ein wichtiger Grund für konvergierendes Verhalten von Unternehmungen und für das Entstehen institutioneller Isomorphic. Eine Erklärung für die bisherige Vernachlässigung der Zusammenhänge zwischen Isomorphic und Legitimität liegt in den nicht unerheblichen Schwierigkeiten, das Konzept der Legitimität zu operationalisieren (Suchman, 1995; Deephouse, 1996). Der IA hat zweifellos erhebliches Potential, die Entwicklung einer eklektischen Theorie des Strategischen Managements zu inspirieren. Der Ansatz bedarf allerdings einiger Erweiterungen, um sein volles Potential entfalten zu können: 37

38

,legitimacy is a generalized perception or assumption that the actions of an entity are desirable, proper, or appropriate within some socially constructed system of norms, values, beliefs, and definitions" (Suchman, 1995: 574). Die Quellen der Legitimiät beruhen selbstverständlich nicht nur auf durch Imitation hervorgerufener Isomorphic. Konformität mit institutionellen Praktiken aufgrund von Zwang oder normativem Druck wirken sich ebenfalls legitimitätserhöhend aus (DiMaggio & Powell, 1991).

344

R. Κ. F. Bresser

(1) Strategische Reaktionen auf institutionelle Zwänge. Der bisherigen Forschung auf der Basis des IA kann vorgehalten werden, daß sie zu einseitig auf Konformität ausgerichtet ist. Es wird unterstellt, daß ein passives Anpassen und „sich Fügen" (z.B. qua Imitation) die wichtigste, wenn nicht einzige Verhaltensweise ist, die der Unternehmung zur Verfügung steht, wenn sie mit institutionellen Zwängen konfrontiert wird. Oliver (1991) schlägt deshalb vor, daß der IA auch andere, aktivere Möglichkeiten integriert, auf die Unternehmungen im Umgang mit institutionellen Zwängen zurückgreifen können. Unter Bezug auf den Strategic-Choice-Ansatz und die damit verbundene Resource-Dependence-Perspektive (Pfeffer & Salancik, 1978) entwickelt Oliver eine Typologie von fünf strategischen Reaktionen (Konformität, Kompromiß, Vermeidung, Mißachtung, Manipulation), die entlang eines Kontinuums zwischen passiver Konformität und aktivem Widerstand angesiedelt sind. (2) Der Wandel institutioneller Strukturen. Der IA hat Schwierigkeiten, den Wandel existierender institutioneller Strukturen zu erklären (Barley & Tolbert, 1997; Greenwood & Hinings, 1996). Durch die fast vollständige Ausrichtung auf den Prozeß der Strukturierung eines organisatorischen Feldes vernachlässigen die Vertreter des LA die Frage: Wie kommt es zu einem endogenen Wandel institutioneller Strukturen und Praktiken, nachdem diese sich aufgrund des Einwirkens verschiedener institutioneller Kräfte herausgebildet haben? Erst in jüngster Zeit wenden sich einige Erklärungsmodelle dieser Fragestellung zu, so z.B. die Ansätze von Greenwood und Hinings (1996), Vadlamani (1996) und Garud und Kumaraswamy (1996). Greenwood und Hinings (1996) integrieren die interne Dynamik von Organisationen in ihre Überlegungen, und sie stellen ihr Erklärungsmodell für institutionellen Wandel auf die Interaktionen zwischen internen (organisatorischen) und externen (institutionellen) Faktoren ab. Vadlamani (1996) spricht von einem „Paradoxon der Isomorphic", das darin besteht, daß Isomorphic einerseits die Konvergenz institutioneller Praktiken fördert. Andererseits ist in der Isomorphic aber auch gleichzeitig die Saat für divergierende Kräfte enthalten, die die institutionellen Strukturen zu einem späteren Zeitpunkt destabilisieren. Durch eine anwachsende Institutionalisierung werden Organisationen, die sich den institutionellen Zwängen nicht beugen wollen, zunehmend marginalisiert. Um zu überleben, werden sich an den Rand gedrängte Organisationen aktiv darum bemühen, alternative Strukturen aufzubauen, die die Schwächen der alten Institutionen überwinden können. Es ergibt sich somit ein zyklisches Modell des endogenen Wandels institutioneller Strukturen (Vadlamani, 1996): Während der Anfangsphase der Institutionalisierung kommt es zu einer kontinuierlich anwachsenden Konvergenz kognitiver, sozialer und technologischer Strukturen. Während der Reifephase eines organisatorischen Feldes werden demgegenüber divergierende Kräfte, die eine alternative Struktur favorisieren, immer stärker aktiviert. Dieser Widerstreit zwischen konvergierenden und divergierenden Kräften setzt sich fort bis die alternative Struktur die bisherige ablöst. Es ist offensichtlich, daß dieses zyklische Modell des institutionellen Wandels starke Ähnlichkeiten mit dialektischen Modellen des organisationalen Wandels (Heydebrand, 1977; Bresser & Bishop, 1983) und dem Konzept des Paradigmenwechsels (Kuhn, 1962; 1974) aufweist. Ein Vadlamani ähnelndes Modell des Widerstreits zwischen konvergierenden und divergierenden Kräften entwickeln Garud und Kumaraswamy (1996) am Beispiel von Netzwerkindustrien im Bereich der Mikroelektronik: In der Computerindustrie entwickeln sich technologische Systeme, indem zunächst ein Wettbewerb um die Etablierung von branchen-

Konvergierende Strategien

345

weiten Standards stattfindet. Dieser Wettbewerb führt zu einer Institutionalisierung von Branchenstandards, d.h. zu einer gleichartigen Branchenarchitektur. Die Branchenstandards sind als Basistechnologien sämtlichen Wettbewerbern einer Branche zugänglich. Der Wettbewerb verlagert sich nun auf die technische Ebene der Produktinnovationen. Erfolgreiche Produktinnovationen führen im Zeitablauf dazu, daß sich auch die Branchenarchitektur vorherrschender Standards weiter verändert, denn Unternehmungen werden danach trachten, erfolgreiche technische Veränderungen auf der Produktebene zu einer Erweiterung der bestehenden Branchenarchitektur zu verwenden. Diese durch Produktinnovationen ausgelösten Variationen der Branchenarchitektur sind nach Garud und Kumaraswamy (1996) charakteristisch für einen endogenen Wandel institutioneller Strukturen in Netzwerkindustrien. (3) Integration des RBA. Eine Belebung der strategischen Managementforschung kann von einer Integration des RBA und des IA erwartet werden. Die Zusammenhänge zwischen den beiden Ansätzen sind offensichtlich, sie wurden bisher von der Literatur aber so gut wie nicht aufgegriffen (Oliver, 1997). In beiden Ansätzen kommt der Imitation ein zentraler Stellenwert zu, allerdings aus entgegengesetzten Gründen. Während im RBA Imitation als eine Verhaltensweise gilt, die zu vermeiden und zu verhindern ist, damit dauerhafte Wettbewerbsvorteile aufgebaut werden können, geht der IA von der Funktionalität der Imitation aus, da diese Legitimität erzeugt. Eine Integration der Ansätze läßt sich eventuell durch eine Rückbesinnung auf die „Österreichische Schule" der Nationalökonomie bewirken, denn letztere modelliert neben temporären Wettbewerbsvorteilen auch die Vorteilhaftigkeit der Imitation (Jacobson, 1992).

Die Artikel Der erste Beitrag,, Jnterorganizational Imitation: The Impact of Interlocks on Corporate Acquisition Activity", von Pamela Haunschild stellt unter Beweis, daß Imitation auch bei sehr wichtigen und weitreichenden strategischen Entscheidungen ein verbreitetes Phänomen ist. Auf der Basis einer Stichprobe von η = 327 untersucht die Verfasserin das Akquisitionsverhalten von amerikanischen Unternehmungen, die durch Aufsichtsratsmandate mit anderen Unternehmungen, die als Modelle fungieren können, verbunden sind. Die Evidenz für das Vorliegen von Imitation ist überwältigend, und die Studie zeichnet sich insbesondere auch dadurch aus, daß sie methodisch sauber vorgeht und alternative Erklärungen für das beobachtete Akquisitionsverhalten in die empirischen Tests einbezieht. Was die Studie nicht untersucht, ist die Frage nach dem „warum" der Imitation, konkret also die Frage: Imitieren Unternehmungen Akquisitionsverhalten, um ihre Legitimität gegenüber der Außenwelt zu erhöhen oder gibt es hierfür andere Gründe? Der zweite Artikel,„Does Isomorphism Legitimate?", von David Deephouse thematisiert demgegenüber das Thema Legitimitätserzielung. Die Studie überprüft die zentrale Aussage des IA, wonach Isomorphic zu erhöhter Legitimität führen soll. Die Ergebnisse bestätigen die Annahme des IA. Für eine Stichprobe von amerikanischen Banken weist Deephouse nach, daß strategische Isomorphie mit einer erhöhten Legitimität sowohl aus der Sicht einschlägiger Regulierungsstellen als auch in der Wahrnehmung der Medien zusammenhängt. Die Studie von Deephouse stellt eine der wenigen Versuche dar, die Zu-

346

R. Κ. F. Bresser

sammenhänge zwischen Isomorphie und Legitimität zu erforschen. Da die Untersuchung von bereits bestehender Isomorphie ausgeht, können von ihr allerdings keine Aufschlüsse über die Gründe für das Entstehen von gleichartigen Strategien abgeleitet werden. Obwohl im vorliegenden Fall die Imitation als ein wahrscheinlicher Mechanismus für die entstandene Isomorphie gelten kann, ist ihre relative Bedeutung unklar, denn der Einfluß anderer Mechanismen (Zwang, normativer Druck) wurde nicht konstant gehalten.39

39

Die Interpretation der vorliegenden Studie im Sinne kausaler Beziehungszusammenhänge ist nicht nur aufgrund der Vernachlässigung der zu Isomorphie führenden Mechanismen problematisch. Kausalitätsrückschlüsse sind auch aufgrund des Forschungsdesigns nur sehr vorsichtig vorzunehmen: Die Studie ist (wie die meisten Untersuchungen der empirischen Sozialforschung) eine Querschnittsuntersuchung und verwendet Regressionstechniken zur Datenanalyse.

Interorganizational Imitation

Interorganizational Imitation: The Impact of Interlocks on Corporate Acquisition Activity Pamela R. Haunschild University of Wisconsin-Madison

347

In this study, I examine direct evidence for the influence of interorganizational imitation on a voluntary, substantive strategic action that affects the economic core of the firm: corporate acquisitions. I argue that firm managers are exposed to the acquisition activities of other firms when they sit on those firms' boards. The acquisition activities of the firms they are tied to serve as models to be imitated. Hypotheses are developed and tested on 1981-1990 acquisition data for a sample of 327 firms. Results show that firm managers are imitating the acquisition activities of those other firms to which they are tied through directorships. Competing rival interpretations of positive evidence for imitation are examined, and the imitation interpretation is found to hold* Interorganizational imitation of practices and structures plays a central role in several theories of organizational action. Theories of organizational learning, for example, argue that organizations copy other organizations, letting others absorb the costs of experimentation or discovery (Dutton and Freedman, 1985; Levitt and March, 1988; Lant and Mezias, 1990; Bolton, 1992). Strategic choice theories suggest that imitation can be a strategic response to competitor activities, whereby second-movers take advantage of the fact that the risk associated with product development has been absorbed by the first-movers (Lieberman and Montgomery, 1988). Institutionalization theory argues that organizations copy practices adopted by others in an effort to acquire legitimacy (DiMaggio and Powell, 1983). Because it suggests social mechanisms that facilitate one firm doing what other firms are doing, the research on interorganizational networks and director interlocks is also relevant to interorganizational imitation. Some of the network research, for example, suggests that interorganizational networks function as a mechanism for the diffusion of innovative practices (Rogers, 1983; Granovetter, 1985). Recent research on director interlocks suggests that these entities function as important conduits of information about business practices (Useem, 1984; Davis and Powell, 1992). The dissemination of information through interlocks may result in firms doing the same things their interlocked partners are doing (Davis, 1991; Palmer, Jennings, and Zhou, 1993).

© 1993 by Cornell University. 0001-8392/93/3804-0564/Î1.00.

I would like to thank my dissertation committee: Mark Fichman, Gerald Salancik, Doug Wholey, and Steve Klepper for their generous support and assistance. Thanks also to Alison Davis-Blake, Anne Miner, Craig Olson, Jim Walsh, Marshall Meyer, and three anonymous reviewers for providing helpful comments and insights. Earlier versions of this paper were awarded the 1992 Academy of Management, OMT Division, Louis R. Pondy Award for the best paper based on a dissertation, and the 1991 TIMS College on Organization, Best Dissertation Proposal Award.

Not surprisingly, early work on interorganizational imitation and diffusion focused on theoretical rationales and exploring empirical evidence that these effects occur. Several studies showed, for example, that the number of prior adopters in an organizational field seems to affect chances of later adoption (Tolbert and Zucker, 1983; Fligstein, 1985, 1990a, 1990b). But evidence for imitation in such studies is still indirect and open to varied interpretations. Appropriately, more recent work has explored the mechanisms of imitation more directly. Galaskiewicz and Wasserman (1989) showed that intercorporate networks had a direct effect on decisions involving the charitable recipients of corporate philanthropy, but while such decisions are important, they do not affect the core economic activities of the firm. One could even 564/Administrative Science Quarterly, 38 (1993): 5 6 4 - 5 9 2

348

P. R. Haunschild Interorganizatlon*! Imitation

argue that imitation should affect symbolic actions like charitable recipients, but other mechanisms should drive choice in voluntary strategic actions involving major resources and the operations of the firm. The first goal of this study was to explore evidence concerning the direct role of imitation of other firms on a major strategic action, corporate acquisitions. I argue that firm managers are exposed to the acquisition activities of other firms when they sit on those firms' boards. For these managers, the acquisition activities of tied-to firms serve as models, examples to imitate or emulate. Therefore, I examined the impact of interlocks with firms that have previously made acquisitions on current acquisitions by other firms. The study's second goal was to examine competing rival interpretations of positive evidence for imitation, exploring the potential impact of tied firms experiencing similar conditions and the effect of private information about acquisition activity that might be transmitted through direct interlocks. Corporate Acquisitions Corporate acquisitions are an interesting and important firm activity with which to explore the substantive impact of interorganizational imitation. During the 1980s, the American economy was significantly restructured by a wave of mergers and acquisitions. Approximately 24,000 firms acquired others or merged with each other, in an exchange of $1.3 trillion in assets. Many of the U.S.'s largest firms were among those acquired, including 28 percent of the 1980 Fortune 500 (Shleifer and Vishny, 1988). Acquisitions represent a serious strategic choice and often involve the commitment of substantial resources. Ninety-five of the 100 largest acquisition transactions that ever occurred occurred during the early 1980s (Grimm, 1987). The acquisition of one company by another often involves significant firm change and can negatively affect the lives of many employees (e.g., Hirsch, 1987; Walsh, 1988; also see review by Schweiger and Walsh, 1990). Despite the prevalence and importance of acquisitions, our understanding of what influences whether a firm will acquire another firm is incomplete. Most of the research on the motives for acquisitions is financial or economic. This research has received mixed support, and no clear consensus has emerged. As Ravenscraft (1987: 20) said " . . . our understanding of the basic determinants of merger motives reflects a large degree of ignorance or at least disagreement." There are three classes of theories that address acquisition motives: (1) financial theories; (2) resource dependence theory; and (3) managerial and agency theories. Most of the research on acquisition motives falls into the first class, the financial theories, and is efficiency-based. Many of these theories propose that acquisitions are driven by the search for synergy, yet whether acquisitions are driven by synergies remains unclear, as mixed and contradictory findings have been found (Jensen, 1984; Chatterjee, 1986). Further, the relationship between synergies and their realization is equivocal (e.g., Chatterjee, 1986; Lubatkin, 1987; Ravenscraft and Scherer, 1987; Barney, 1988). One of the 565/ASQ, December 1993

Interorganizational Imitation

349

most well-supported findings in the acquisition literature is that only target-firm, not acquiring-firm shareholders benefit from an acquisition. Some observers ascribe the value that accrues to target-firm shareholders as the result of anticipated synergies. Others ascribe it to the anticipated gains from shifting control of the target's assets to the more efficient acquiring-firm managers (an agency theory prediction). Neither explanation has been conclusively supported. Further, long-term studies that focus on the accounting and investor returns of the acquiring firm for several years around the acquisition date show that acquisitions (especially conglomerate acquisitions) are not, on average, profitable for the acquiring firm (e.g., Hogarty, 1970; Herman and Lowenstein, 1988; Fowler and Schmidt, 1988). This means that if synergy is the motive for acquisitions, then the anticipated gains are often not realized. This may not be surprising, given the many other factors that affect the relationship between motives and outcomes of acquisitions (Schweiger and Walsh, 1990). More promising are the non-efficiency-based managerial and resource dependence theories. According to resource dependence theory, mergers and acquisitions are a response to the constraints imposed by organizational interdependence (Pfeffer, 1972; Pfeffer and Salancik, 1978). When organizations are dependent on other organizations for resources, such constraints can be managed through mergers. Relationships between various forms of interdependence and mergers are predicted by the theory. Horizontal mergers are used to reduce competitive or commensalistic interdependence. Vertical mergers are used to reduce symbiotic (buyer-seller) interdependence. Conglomerate mergers are a response to dependence on other organizations (1) that constitute a large portion of the organization's exchanges (2) for which few substitutes exist, and (3) in which a vertical merger with the other organization is not feasible, e.g., as in the case of dependence on the government. Support for resource dependence theory as an explanation for industry merger patterns can be found in Pfeffer (1972), Pfeffer and Salancik (1978), and Burt (1983). Some theorists suggest that the third class, the managerial and agency theories, are much more important as explanations of acquisition motives than the financial theories (Ravenscraft and Scherer, 1987; Trautwein, 1990). The managerial theories propose that acquisitions are driven by managerial desire for the prestige, power, salary, and job security that comes with managing large companies (Williamson, 1963; Marris, 1964; Baumol, 1967). Managerial theorists argue that the lack of profitability from acquisitions is no surprise, because managers of acquiring companies are paying for benefits to themselves that are of no value to shareholders (Amihud and Lev, 1981; Shleifer and Vishny, 1988; Morck, Shleifer, and Vishny, 1990). The managerial theories apply mainly as a motive for conglomerate (unrelated) acquisitions. Self-interested managers are motivated to engage in conglomerate acquisitions for two reasons. First, managers of conglomerates are less threatened with job loss due to poor performance, since they can offset poor performance in one business with good 566/ASQ, December 1993

350

P. R. Haunschild Interorganizational Imitation

performance in another (Amihud and Lev, 1981). Second, managers in poorly performing firms have incentives to acquire new businesses that they might be better at running than their current business (Shleifer and Vishny, 1990). Support for managerial theories is provided by studies showing that the alignment of managerial and shareholder interests results in acquisitions that are more profitable for shareholders (Amihud and Lev, 1981; Morck, Shleifer, and Vishny, 1990). Our understanding of corporate acquisitions is incomplete in that most existing theories do not consider the social context that firms are embedded in. We know that firms and firm managers exist in a social world, and it is likely that social embeddedness (Granovetter, 1985) affects even such major activities as acquiring another firm. This study provides social context by investigating whether director interlocks are a source of acquisition models to be imitated. This provides a social explanation for acquisition activity that is quite different from existing explanations. It also extends the scope of imitation to include a major, voluntary strategic action that was responsible for reshaping the American economy during the 1980s. Imitation Theory and Hypotheses As mentioned earlier, most studies of imitation provide only indirect evidence that imitation is occurring. Fligstein found, for example, that one predictor of engaging in financial restructuring (Fligstein, 1990a, 1990b) and adopting an M-form structure (Fligstein, 1985) was the percentage of other large firms that had done so. Fligstein interpreted these results as imitation, but other explanations exist. These and other studies vary in the extent to which their evidence clearly documents that an imitation process was responsible for adoption. At a minimum, demonstrating that one firm's practice is the result of imitation of another firm's practice requires that three conditions be satisfied: (1) a model firm exhibits the practice at time f, (2) representatives from the imitating firm are exposed to the model, and (3) the imitating firm exhibits the practice at time t + x, where χ is some positive but unknown period of time. The reason for specifying these conditions is to be able to demonstrate that imitation, rather than some alternative factor, is responsible for the firm's adoption of a practice. Since I focus on director ties as a source of models to be imitated, evidence for imitation requires that (1) the model (the tied-to firm) engages in an acquisition at time t, (2) the imitator (the focal firm) is exposed to the model through director ties, and (3) at some later time, the focal firm engages in an acquisition. While there are many ways that models to be imitated may come to the attention of firm managers, there are several reasons why director ties might be one of them. Several recent studies have shown that director interlocks provide a mechanism for the diffusion of organizational practices and structures. These include Mizruchi's (1992) study, showing similarity in political campaign contributions among firms tied to each other through director ties. Also, O'Reilly, Main, and Crystal (1988) found chief executive officers' salaries to be positively related to the salaries of their firms' outside 567/ASQ, December 1993

Interorganizational Imitation

351

directors. T w o recent studies directly measured the relationship between interlocks and the adoption of firm practices and structures. First, Davis (1991) showed that intercorporate director ties affected the adoption of "poison pills," a firm-level defense against unwanted corporate takeovers. The likelihood of a firm adopting a poison pill w a s increased w h e n the firm w a s tied to other firms that had adopted. Second, Palmer, Jennings, and Zhou (1993) showed that interlocks affected the adoption of the multidivisional form. Ties to firms that adopted an M - f o r m influenced adoption, while ties to firms that did not adopt did not. While these studies do not discuss director ties as a source of models, they s h o w that director ties are important sources of information and influence on firm structures and practices. Director ties provide managers w i t h information from a familiar source, and familiarity has been s h o w n to affect imitation among individuals (e.g., Bandura, 1977) and firms (Galaskiewicz and Wasserman, 1989). Director ties render events that are otherwise distant more proximate. Sitting on a board and watching another firm make an acquisition creates an immediate, concrete example that may encourage imitation. Useem's (1984) study supports the idea that imitation occurs through director ties, which provide a source for models. Based on interviews w i t h 129 American and British executives and directors of companies, Useem (1984) proposed that the primary function of director interlocks is to act as a mechanism through which managers can achieve an optimal "business scan" of the latest business practices and the overall business environment. Useem provided several examples in which managers describe sitting on a board and then deciding that w h a t that firm is doing might be relevant for their o w n firm. Director interlocks are an important source of personal contacts among those managers w i t h the power to affect organizational merger and acquisition (M&A) activity. Inside directors, because they are also managers, are more likely to influence firm M & A activities than outside directors. Inside directors are also more likely to sit on outside boards than are lower-level managers. Any one inside director might have multiple ties, and each tied-to firm might complete multiple acquisitions. The effectiveness of modeling is also increased by multiple models (Kazdin, 1976), since the more models exhibiting behavior X one is exposed to, the more likely one is to imitate behavior X . This suggests that the number of acquisitions completed by all tied-to firms affects the likelihood of imitation by the focal firm and leads to the following hypothesis: Hypothesis 1 (H1): The number of prior acquisitions completed by firms that are tied to a focal firm through director ties is positively associated with the number of current acquisitions by the focal firm. It is reasonable to expect that the relationship between the number of acquisitions by the focal and tied-to firms will not be linear w h e n the tied-to firms complete large numbers of acquisitions, because there is only so much information that the focal firm can obtain from observing the acquisition behavior of the tied-to firms. At some point, the impact of additional models on imitation behavior should begin to 568/ASQ, December 1993

352

P. R. Haunschild Interorganizational Imitation

increase at a decreasing rate. After seeing several models, some of this information is likely to be duplicated, producing less impact on the focal firm. This leads to the hypothesis that the relationship in H1 will be nonlinear, with diminishing effects at high levels of acquisition activity by the tied-to firms: Hypothesis 2 (H2): The relationship between the number of current acquisitions by the focal firm and the number of prior acquisitions by the tied-to firms will be positive but will increase at a decreasing rate.

Existing literature on imitation is unclear about exactly which dimensions of the behavior exhibited by a model will be imitated. If the focal firm is imitating tied-to firm acquisitions, it is unclear whether the focal firm will imitate only the fact of the acquisition, in which case they may make a different type of acquisition than the tied-to firm, or whether they will imitate the specific type of acquisition that the tied-to firm completed, in which case they will make the same type of acquisition as the tied-to firm. A stronger case for imitation can be made if the latter is true, if there is some similarity between the type of acquisition undertaken by the tied-to firm and the acquisition undertaken by the focal firm. Several classification schemes exist for determining whether acquisitions are similar (e.g., Salter and Weinhold, 1979). One commonly used criterion is whether the acquisition is horizontal, vertical, or conglomerate. Horizontal acquisitions occur when firms acquire their competitors, vertical acquisitions occur when firms acquire their suppliers or distributors, and conglomerate acquisitions occur when firms acquire unrelated firms. The logic behind director ties serving as models also supports the idea that firms will imitate the specific type of acquisition, because part of the information and examples transmitted through director ties is likely to be the general strategies that are involved in making vertical, horizontal, or conglomerate acquisitions. Decisions about substantive acquisitions are often justified to the board of directors, and part of this justification is likely to be the strategy involved in the specific type of acquisition being proposed. For example, if a firm is contemplating buying a supplier in a vertical acquisition, then the logic for a vertical acquisition is likely to be explicitly laid out at the board meeting. The rationale for a particular form of acquisition and its proposed benefits are likely to influence managers of other firms that are sitting in these board meetings. Accordingly, we should see a relationship between the type of acquisition (horizontal, vertical, or conglomerate) completed by the tied-to and focal firms: Hypothesis 3a (H3a): The number of prior horizontal acquisitions by tied-to firms is positively associated with the number of current horizontal acquisitions by the focal firm. Hypothesis 3b (H3b): The number of prior vertical acquisitions by tied-to firms is positively associated with the number of current vertical acquisitions by the focal firm. Hypothesis 3c (H3c): The number of prior conglomerate acquisitions by tied-to firms is positively associated with the number of current conglomerate acquisitions by the focal firm. 569/ASQ, December 1993

Interorganizational Imitation Alternative Explanations Results consistent with the above hypotheses are subject to two general types of alternative explanations. This first is that the relationship between acquisitions by the tied-to and focal firms may be due to the tied-to and focal firms being subject to similar internal or external conditions that cause the observed levels of acquisition activity by both firms. The second alternative explanation is that the relationship between acquisitions by the tied-to and focal firms may be due to the director ties acting as a conduit of "private" information about acquisitions, and this information, not imitation, is causing the observed relationship. Similar conditions. The first alternative explanation is that director ties may occur among firms subject to similar conditions. If this is true, then both focal firms and tied-to firms could be acquiring or not acquiring in response to these conditions, and not because the focal firm is imitating the tied-to firm. This is an omitted-variable problem: Some variable related to both acquisition levels and interlocks is causing the observed relationships. One way of testing for similar conditions is to see if the relationship between acquisitions by the focal and tied-to firms still holds when the focal and tied-to firms are very dissimilar. I thus hypothesize: Hypothesis 4 (H4): The relationship between focal and tied-to firm acquisitions is restricted to cases in which the focal and tied-to firms are similar. If the relationship between acquisitions by the focal and tied-to firms holds when the firms are dissimilar, the similarity explanation is not supported. It could be argued that even if no support is found for hypothesis 4, the reason is that the proper dimensions of similarity were not captured in this test. This argument would propose the existence of another factor common to both focal and tied-to firms, and also to the likelihood of interlocks, that is causing similar levels of acquisition activity by both firms. Some managerial and agency theories outlined earlier could be proposed to work in this way. For example, an argument could be advanced that proposes (1) managers complete acquisitions for personal motives, and (2) managers who have the motivation and ability to do this tend to sit on each other's boards. Any such factor that is associated with both ties and acquisitions (but not imitation) would operate simultaneously in both the focal and the tied-to firms. This means that the requirement for demonstrating imitation, that some time elapse between the acquisitions by the tied-to firms and the acquisitions by the focal firms, does not hold. When managers are sitting on boards and observe a firm making an acquisition, they cannot go back to their own firms and immediately execute a deal. Therefore it is likely that any acquisition completed by tied-to firms in a given year cannot be imitated by a focal firm in that same year. Yet if some common condition, like self-interested managers, exists in both firms, then acquisitions by the tied-to firms in one year should be related to the number of acquisitions by the focal firm in that same year: 570/ASQ, December 1993

353

354

P. R. Haunschild Interorganizational Imitation Hypothesis 5 (H5): The number of current-year acquisitions by tied-to firms is related to the number of current-year acquisitions by the focal firm.

If w e find that this simultaneous relationship does not hold, it suggests that no common condition is simultaneously influencing the acquisition behavior of the focal and tied-to firms. Finally, if some condition common to both firms is causing the acquisition relationship, we should find both focal and tied-to firms acquiring or not acquiring. This means that prior acquisitions by the focal firm should be related to current acquisitions by the tied-to firms. If this relationship cannot be reversed, then the similarity explanation is not supported: Hypothesis 6 (H6): The number of prior acquisitions by the focal firm is related to the number of current acquisitions by the tied-to firms.

Private information. The second alternative explanation relates to ambiguity about what interorganizational director ties represent. Although they were discussed above as a source of models, they could be simply a source of information about mergers and acquisitions: managers who sit on boards of firms that are making acquisitions are exposed to information about acquisitions. This interpretation is consistent with imitation, which requires exposure to examples of other firms, but it is also consistent with some financial theories known as private-information theories. Private-information theories say that acquisitions are executed by managers who have "better" information about the target, or the target's value, than managers of other firms (e.g., Bradley, Desai, and Kim, 1983). Presumably, the more boards managers sit on, the greater their chances of getting this "better" information. There are three ways of exploring this private-information perspective: (1) measuring and controlling for access to private information; (2) focusing on what types of private information might be communicated through director ties; and (3) examining imitation over time. Access to information has commonly been measured by network centrality, i.e., having many director ties to other firms (e.g., Mariolis and Jones, 1982; Useem, 1984; Davis, 1991). Firms that are tied to many other firms are more likely to have access to private information about acquisitions than firms that are tied to few other firms. According to an imitation perspective, network centrality is not a requirement for imitation. Thus, finding a relationship between network centrality and acquisitions would support a private-information explanation: Hypothesis 7 (H7): There is a positive relationship between focal-firm network centrality and the number of acquisitions completed in the current year.

The second way of exploring the private-information explanation is to focus in more detail on what types of private information might be communicated through these interorganizational director ties. If evidence is found that private information is related to acquisition activity, it would be useful to see what form this information takes, which is not clear in the literature discussing private information. 571/ASQ, December 1993

Interorganizational Imitation

355

Discussions of private information usually focus on the effects of asymmetric distributions of information, without considering its content. For purposes of this study, it is useful to distinguish among types of private information that might be communicated through director ties. There are three general types of private information that might be communicated: (1) information about acquisition opportunities, e.g., special opportunities for buying particular firms, especially undervalued firms; (2) general know-how information, e.g., what investment bankers to use or how to negotiate a low purchase price; and (3) normative information, e.g., acquisitions are "the thing to do." Information about acquisition opportunities is probably closest to a financial argument for private information, while normative information is probably furthest from it. The idea of firms responding to normative information is very similar to the idea of firms imitating other firms, however, and the effects of the impact of normative information and imitation on a firm therefore cannot be disentangled. It is useful, nevertheless, to attempt to disentangle information about acquisition opportunities from know-how and normative information, and this can be done with the data used in this study. The alternative explanation that information about special acquisition opportunities is causing the relationship between acquisitions by the tied-to and focal firms can be dealt with by determining in which industries the focal and tied-to firms are acquiring. Firms that acquire other firms in a particular industry are more likely to have information about opportunities in that industry than nonacquirers in an industry. If they pass on this information, there should be a relationship between industries in which the tied-to firms are acquiring and the industries in which the focal firms are acquiring. If they are not acquiring in the same industries, then it is less likely that private information about opportunities is being communicated through these director networks and causing the relationship between acquisitions by the tied-to and focal firms. Hypothesis 8 (H8): There is a positive relationship between the industries of the tied-to firm's acquisitions and the industries of the focal firm's acquisitions. Finally, the third way of testing whether private information is causing the relationship between acquisitions by the tied-to and focal firms is to examine differences in the impact of imitation variables over time. Private-information and institutional theories of imitation make different predictions about effects over time. Empirical research investigating the institutionalization of various practices and structures has documented that early adoption of these practices and structures is due to the practice being functional for the adopting firm. This is not true for later adoption, when the practice is widely understood to be something that "rational" firms do. Then the practice is followed whether or not it makes sense for the adopting firm. Tolbert and Zucker (1983) found this pattern in the diffusion of civil service reform, and Armour and Teece (1978) and Fligstein (1985) found support for similar predictions in the spread of the M-form. Historically, mergers 572/ASQ, December 1993

356

P. R. Haunschild Intel-organizational Imitation

have occurred in waves, w i t h peaks in merger activity in about 1899, 1929, 1967, and 1986. The wave of mergers in the 1980s began in 1975, peaked about 1986, and the number of mergers decreased thereafter. Applying the above theory to the 1980s wave, it may be that early in the wave, acquisitions occur w h e n they are functional for the acquiring firm, but later in the wave, w h e n many firms are making acquisitions and acquisitions are normative, social processes dominate. This leads to the hypothesis that the imitation relationships are more strongly associated w i t h later rather than earlier acquisition activity: Hypothesis 9 (H9): The imitation variables are more strongly associated with acquisitions during the later part of the 1980s merger wave. METHOD The design compares firms w i t h different numbers of acquisitions (including no acquisitions) during the 1981-1990 period. This period was chosen for t w o reasons. First this was a period in w h i c h great numbers of acquisitions occurred. Second, it is the period covered by the best available database of mergers and acquisitions, the database covered by the Journal of Mergers and Acquisitions (the M & A database). The M & A database includes all acquisitions during this period that exceeded a purchase price of $1,000,000 and was used to code the acquisition variables. Sample The sample consists of all medium and large-sized firms (over $35 million in assets) listed in the 1981-1990 COMPUSTAT databases for four industries. Small firms w e r e not studied because they make very f e w acquisitions and have very f e w interlocks. COMPUSTAT historical files w e r e used, so that firms that w e r e acquired by other firms during the 1981-1990 period w e r e included in the sample. The sample is not restricted to survivors. Sampling from COMPUSTAT means that leveraged buyouts (LBOs) and other forms of going-private transactions in which individuals rather than firms are making the acquisition w e r e not included in this study. Only "traditional" acquisitions, in which one existing firm is buying another existing firm, w e r e sampled. Because traditional acquisitions accounted for 85 percent of all acquisitions completed from 1981 to 1987 (Blair, Lane, and Schary, 1991), this sample includes a large portion of the acquisition activity during the 1980s. I restricted the focal firms I sampled to four industries, because acquisition activity varies by industry, and industry has to be controlled for in all analyses. Having a f e w firms in many industries w o u l d restrict controlling for industry differences. The four Standard Industry Classification (SIC) industries selected w e r e electrical equipment manufacturing (SIC 36), transportation equipment (SIC 37), wholesale trade (SIC 50), and business services (SIC 73). The electrical and transportation equipment industries w e r e chosen because manufacturing industries like these have been the subject of almost all of the acquisition research to date (e.g., Golbe and White, 1988; Fowler and Schmidt, 1988). The wholesale and business services industries w e r e chosen because they w e r e important, growing industries during the 1980s. 573/ASQ, December 1993

Interorganizational Imitation Selecting all medium and large-sized firms (over $35 million in assets) from these four industries resulted in a sample of 327 focal firms. These 327 focal firms were tied to 622 other firms, so data were collected on a total of 949 firms. Data were not collected for all ten years during this 1981-1990 period, because collecting acquisition and director tie data is a laborious task. Rather, for each focal firm, one year from the 1981-1990 period was randomly selected and acquisition data were collected for the selected year and three years before the selected year. This sampling approach minimized data collection while still providing variance across firms. The three-year period was used because it is reasonable to think that events in the distant past will have less impact than events in the recent past. Three years was assumed to be a reasonable period for an acquisition to serve as a model for other firms. Dependent Variables The number of acquisitions completed by the focal firms during the randomly selected year and three years before the selected year was obtained from the Journal of Mergers and Acquisitions. The number of acquisitions completed by the focal firms in the selected year ranged from zero to nine. Because the industry of almost all acquired firms was available from either the M&A database or COMPUSTAT, each acquisition by the focal firm could be classified according to whether it was vertical, horizontal, or conglomerate. A classification scheme that has been used in several other studies of acquisition activity (e.g., Blair, Lane, and Schary, 1991) was used to classify acquisitions as horizontal, vertical, or conglomerate. An acquisition was coded as horizontal when the two-digit industry code of the acquiring firm matched that of the acquired firm. An acquisition was coded as vertical when the industry of the acquiring firm either (1) sold more than 5 percent of its output to or (2) received more than 5 percent of its input from the industry of the acquired firm. The input-output numbers were obtained from the input-output tables published annually by the Survey of Current Business (U.S. Department of Commerce). Finally, acquisitions that were neither horizontal nor vertical were coded as conglomerate. This coding scheme assumes that firms can be uniquely classified into 2-digit industry codes. Although the primary industry of these firms was used in the coding scheme, very large firms may have operations in several industries other than the primary one. This means that this coding scheme will result in classifying some acquisitions as conglomerate that are really horizontal or vertical. Obtaining completely accurate industry codes requires line-of-business acquisition data, which is not publicly available. Any misclassifications by this coding scheme work against finding effects, however, so any effects found would be stronger with more precise industry data. Independent Variables Director ties. Imitation is more likely to have an effect when directors of the focal firms who sit on the boards of tied-to firms are inside directors (i.e., managers) of their own firms. 574/ASQ, December 1993

357

358

P. R. Haunschild Interorganizational Imitation

This is because inside directors are more likely to influence their own firms' decisions about acquisitions than outside directors. The ties of inside directors are a subset of all possible interlock ties between two firms. The ties in this study are almost all unidirectional, i.e., a focal firm manager sits on the tied-to firm's board, but no tied-to firm managers sit on the focal firm's board. They are also direct ties. Other studies have classified interlocks into those that are direct (created by people affiliated with the two firms that are connected) and those that are indirect (created by people not affiliated with those two firms). Several researchers have suggested that direct interlocks are more influential than nondirectional interlocks (Palmer, 1983; Mintz and Schwartz, 1985; Stearns and Mizruchi, 1987; but see Palmer, Jennings, and Zhou, 1993 for an exception). Interlock data were collected for this study by obtaining the names of all inside directors, their titles and director ties from the proxy statements of the focal firm. Then each firm for which these managers served as director (the tied-to firm) was checked against the M&A database. The number of acquisitions by these tied-to firms during the sampled year and three years before the sampled year was used to create the tied-to firm acquisition variables. The size of the firms acquired by these tied-to firms was obtained from COMPUSTAT and/or the M&A database. The type of acquisition (horizontal, vertical, or conglomerate) completed by the tied-to firm was coded the same way as that of the focal firm. Network centrality was calculated as the sum of ties for all inside directors of a firm, minus any duplicated ties, i.e., when two directors of the same firm sit on the same board. The number of ties to other firms ranged from zero to fifteen. To test for differences in imitation over time, the sample was split into two periods. The split occurred between 1986 and 1987, which is right after the peak of the 1980s merger wave. A dummy variable for period and an interaction variable (period by number of acquisitions by the tied-to firms) were created. Control Variables Some variables not considered in the hypotheses and alternative explanations may be related to the number of acquisitions by the focal firms, including the firms' free cash flow, past performance, previous acquisition activity, and size and factors associated with the economy and a firm's industry. Free cash flow. According to Jensen's (1987) free cash flow theory, high cash flow and low debt create agency costs associated with conflicts between managers and shareholders over the payout of this free cash, which is the cash left after the firm has invested in all available positive net present value projects. Jensen's theory says that managers have incentives to invest excess free cash in negative net present value projects (especially those related to firm growth), rather than pay it out to shareholders as dividends. Free cash flow theory predicts that free cash flow will be positively related to conglomerate mergers, because these mergers provide growth and are also generally not 575/ASQ, December 1993

Interorganizational Imitation

359

wealth-enhancing projects for acquiring firm shareholders. Thus, we might expect free cash flow, i.e., high levels of current free cash and low debt, to be related to the number of conglomerate acquisitions by the focal firm. Consistent with Lehn and Poulsen (1989), free cash flow was measured as follows: Current-year Cash Flow = (Operating Income Taxes - Interest Expense - Preferred Dividend Common Dividend)/Equity. All variables were obtained from COMPUSTAT. Two free cash flow measures were developed, one measuring free cash flow at the end of the year before the sampled year and one measuring average free cash flow over the prior three years. The second measure of free cash flow is the firm's debt (long- and short-term) to equity ratio. Low debt in relation to equity means that the firm has more free cash flow. Again, both debt/equity as of the end of the year before the sampled year and average debt/equity over the prior three years were measured. Past performance of the focal firm. Firm performance is likely to influence the number of acquisitions, but the direction of this effect is difficult to specify. On the one hand, increasingly good prior performance may result in managerial hubris (Roll, 1986), which in turn results in acquisitions, especially risky acquisitions (Morck, Shleifer, and Vishny, 1990). Managers with hubris will systematically overestimate their ability to make risky acquisitions work. An alternative interpretation of a positive relationship between performance and acquisitions is that good past performance results in easy access to acquisition financing, enabling firms to make more acquisitions. On the other hand, Morck, Shleifer, and Vishny (1990) hypothesized that managers in poorly performing firms have incentives to try something new, and they enact this by buying new businesses that they may be better at running than their current businesses. If this is true, then we may see more acquisitions by poorly performing firms. Three measures of the past performance of the focal firm were developed. Firm performance relative to industry was used because industries vary in performance measures, and the industry component is presumably not under managerial control. Performance not under managerial control is unlikely to affect the managerial motivations outlined earlier. The first measure is income growth relative to the industry over the three years before the selected year. Following Morck, Shleifer, and Vishny (1990), income growth was measured as log[/(t - 1 )] - log [/(f - 4)], where t is the selected year and I is the sum of net income, interest income, and deferred taxes. The mean income growth for all firms in the same industry was subtracted from this number. The second measure is return on assets (relative to industry) during the three years prior to the selected year. The third measure is return on equity, relative to the industry, during the three years prior to the selected year. All variables were obtained from COMPUSTAT. Other controls. Four other variables were included to control for factors known or expected to affect acquisition activity. One control is a firm's previous acquisition activity. 576/ASQ, December 1993

360

P. R. Haunschild Interorganizational Imitation

The probability of acquisition activity in one year is not likely to be completely independent of acquisition activity in the previous year (or years) (Amburgey and Miner, 1992). To control for firm-level tendencies to acquire or not acquire, independent of imitation, the number of acquisitions by the focal firm during the three years before the selected year was used as a control variable. Macroeconomic researchers suggest that some macroeconomic variables, such as annual gross national product (GNP), are associated with levels of acquisition activity (e.g., Becketti, 1986; Golbe and White, 1988). Evidence based on merger activity prior to the 1980s shows that low interest rates and high stock prices are associated with the number of mergers per year (e.g., Beckenstein, 1979), although this relationship appears to break down in the 1980s (Golbe and White, 1988). Other macroeconomic factors unique to the 1980s wave have been proposed, including (1) lax antitrust enforcement by the Reagan administration, which resulted in an increase in the number of related mergers and (2) the rise of the junk bond market and alternative forms and techniques for acquisition financing (Jarrell, 1987). In this study, the potential effect of macroeconomic conditions on acquisition activity was controlled in two ways: (1) by entering years as a set of dummy variables to the models and (2) by directly entering those macroeconomic variables that have been found to be associated with acquisition activity by prior studies. These variables are annual GNP, cost of capital, measured as the annual rating on Moody's Aaa bonds, and a stock market index measured as the annual volume traded on the New York Stock Exchange (NYSE) (e.g.. Nelson, 1959; Golbe and White, 1988). The third control accommodates the argument that industry conditions can affect levels of acquisition activity. Several theories of acquisition activity that operate at the industry level have been proposed (e.g., Jensen, 1987). Some of these explanations are unique to the 1980s, including (1) increased industry deregulation, which has the potential to increase competition (e.g., airlines) and/or relax former prohibitions on acquisitions by firms in certain industries (e.g., banking), and (2) the rise of foreign competition, which affected manufacturing industries more than service industries (Jarrell, 1987). These explanations have not been tested, but they are still important to control for. Therefore, differences in acquisition activity for the four industries in this study were controlled by entering industry (as a set of dummy variables) into the model. Finally, it has been shown in previous studies that acquiring firms tend to be larger than nonacquiring firms. Therefore, the size of the focal firm (measured as total assets) was added as the fourth control variable. Asset size was obtained from COMPUSTAT. Because size is both a cause and an effect of acquisitions, firm size as of the end of the year before the sampled year was used in the analyses. Tables 1 and 2 present descriptive statistics and correlations among the study variables. 577/ASQ, December 1993

Interorganizational Imitation

361

Table 1 Descriptive Statistics for Study Variables* Variable

Mean

S.D.

Min.

Max.

Ν

85.57 .327 .156 .281 .235 1803 .551 11.84 .000 .000 .000 1.30

2.91 .470 .363 .450 .425 12727 .681 55.01 1.03 2.03 1.67 2.11

81 0 0 0 0 35 0 -2.27 -6.85 -4.44 -1.12 0

90 1 1 1 1 164063 6.81 365.53 5.29 30.57 25.82 16

327 327 327 327 327 327 327 327 327 327 327 327

.651 .286 .088 .250

1.20 .704 .354 .759

0 0 0 0

9 5 2 6

327 318 317 316

No. tied-to firm acquisitions Current yr. Prior 3 yrs. Current yr. (banks only) Prior 3 yrs. (banks only) Current yr. (nonbanks only) Prior 3 yrs. (nonbanks only) Current yr., horizontal Prior 3 yrs., horizontal Current yr., vertical Prior 3 yrs., vertical Current yr., conglomerate Prior 3 yrs., conglomerate

.997 2.05 .266 .593 .745 1.64 .461 .987 .155 .284 .238 .581

2.56 5.09 1.02 1.99 2.08 4.30 1.24 2.97 .72 1.30 .91 1.71

0 0 0 0 0 0 0 0 0 0 0 0

19 49 9 16 19 45 8 33 6 15 10 14

327 327 316 316 316 316 310 299 309 299 311 301

Other variables Network centrality

1.90

2.49

0

15

327

Control variables Year Elee, equip. Transportation Wholesale Business sves. Assets (in millions) Debt/equity Free cash flow Adjusted income Adjusted ROA Adjusted ROE No. focal acqsns. (prior 3 yrs.) No. focal firm acquisitions Current yr. Horizontal Vertical Conglomerate

* Except as noted, untransformed values are reported for variables.

Model Estimation The dependent variable (number of acquisitions by the focal firm) ranges from zero to nine, with 82 percent of the focal firms completing either zero or one. This means that the dependent variable is nonnegative, by definition, and that a number of observations have a value of zero. Because this is therefore a dependent variable with a limited range, standard multiple regression techniques are inappropriate. More appropriate is the Tobit method, which is designed explicitly to account for limited dependent variables (Tobin, 1958) and has been used in other studies in which the dependent variable is an acquisition count (Amihud and Lev, 1981). The Tobit method was used to estimate the parameters of all models, using the LIMDEP statistical package (Greene, 1990a). The Tobit model is derived from an underlying regression, y* = β' χ, + e,, where e, ~N(0, σ 2 ) and Ely) = 0 Prob [y = 0] + Ely*/y' > 0] Prob [y* > 0],

With Tobit, y* is not directly observed, and in some ranges the true value of y* is masked (Greene, 1990b). In this 578/ASQ, December 1993

362

P. R. Haunschild Interorganizational Imitation

study, y* represents the focal firm's (unobserved) propensity to merge. The method of estimation is maximum likelihood, and f-tests are used to assess the significance of individual coefficients (for more details on the Tobit method, see Maddala, 1983; Greene, 1990a, 1990b). The significance of the results does not vary when the dependent variable is collapsed into a 0, 1 and logistic regression is used instead of Tobit. Further, the results are approximately the same with Poisson regression. RESULTS Imitation Table 3 presents the results of the analysis for imitation. Model 1 shows the results of the test of H1. As predicted, the number of prior acquisitions completed by the tied-to firms is positively related to the number of current acquisitions completed by the focal firm. Model 2 shows the results of the test of H2. As predicted, the squared term for the number of acquisitions by the tied-to firms is negative and significant. The peak of the linear relationship occurs at 22 acquisitions by the tied-to firms during the prior three years. The model including both acquisitions by the tied-to firms and the square of this term significantly improves model fit over the base model (χ2 = 34.79, df = 2, ρ < .001). Table 3 also presents results for imitation of the type of acquisitions. Models 3-5 show the results of the tests of H3a, H3b, and H3c. As predicted by H3a, there is a positive relationship between the number of prior horizontal acquisitions completed by the tied-to firms and the number of current horizontal acquisitions completed by the focal firm. Yet there is no relationship between the number of prior vertical or conglomerate acquisitions by the tied-to firms and current horizontal acquisitions by the focal firm. This means that horizontal acquisitions by the tied-to firm result in horizontal acquisitions by the focal firm, but not in vertical or conglomerate acquisitions. As predicted by H3b, there is a positive relationship between the number of vertical acquisitions completed by the tied-to firms and the number of vertical acquisitions completed by the focal firm. There is no relationship between the number of horizontal and conglomerate acquisitions by the tied-to firms and vertical acquisitions by the tied-to firms. This means that vertical acquisitions by the tied-to firm result in vertical acquisitions by the focal firm, but not in horizontal or conglomerate acquisitions. As predicted by H3c, there is a positive relationship between the number of conglomerate acquisitions by the tied-to firms and the number of conglomerate acquisitions by the focal firm. There is no relationship between horizontal and vertical acquisitions by the tied-to firms and conglomerate acquisitions by the focal firm. This means that conglomerate acquisitions by the tied-to firm result in conglomerate acquisitions by the focal firm, but not in horizontal or vertical acquisitions. 579/ASQ, December 1993

Interorganizational Imitation

363

Table 2

Correlations among Key Study Variables* Variable Control 1. Elect, equip. 2. Transportation 3. Wholesale 4. Business svcs. 5. Assets 6. Debt/equity 7. Free cash flow 8. Adj. income 9. Adj. ROA 10. Adj. ROE 11. No. focal acqsns. (prior 3 yrs.) No. 12. 13. 14. 15.

focal firm acquisitions Current yr. Horizontal Vertical Conglomerate

No. 16. 17. 18.

tied-to firm acquisitions Current yr. Prior 3 yrs. Prior 3 yrs. (banks only) Prior 3 yrs. (nonbanks only) Horizontal Vertical Conglomerate

19. 20. 21. 22.

23. Network cent.

1

2

3

4

5

6

-.30 -.44 -.39 -.07 -.09 -.04 -.04 .01 .01

-.27 -.24 .16 -.06 -.13 .05 -.01 -.01

-.34 -.04 .09 -.08 -.23 .01 .01

-.03 .06 .25 .02 .01 .01

.01 -.05 .08 .14 .07

.10 -.23 -.71 .02

.25 .05 .04

.07 -.17

.33

-.05

.11

-.05

.02

.21

-.02

-.03

.03

.14

-.16 -.14 -.08 -.08

.10 -.05 .16 .14

.11 .04 .01 -.01

.18 .16 -.05 -.01

.16 .06 .07 .19

.07 .04 .03 .04

-.01 .10 -.06 -.06

.03 .01 -.03 .06

.29 .01 -.04 .51

-.07 -.07

.20 .19

-.04 -.03

-.04 -.05

.22 .26

-.03 -.01

.04 .03

-.06 -.07

.27 .25

-.03

.29

-.11

-.10

.21

-.05

-.07

.01

.36

-.04 -.03 -.01 -.06

.09 .18 .18 .18

-.02 -.12 -.08 -.02

-.02 .01 -.08 -.07

.24 .01 .16 .26

.03 -.01 -.02 -.03

.06 -.05 -.08 -.05

-.09 -.01 .02 -.07

.13 .24 .13 .32

-.03

.18

-.02

-.10

.29

-.01

-.06

-.01

.07

7

8

9

• The approximate cutoff for significance at the .05 level is any correlation greater than .11 or less than - . 1 1 .

Forty-one percent of the focal firms do not have director ties to other firms, though they may still make acquisitions. To determine whether the imitation relationships hold only for firms that have ties, the above analyses were repeated using only those firms with ties. Results of these analyses support the above hypotheses, e.g., prior acquisitions by the tied-to firms are significantly related to current acquisitions by the focal firm, [«193) = 3.24, ρ < .001], Similar Conditions To test H4, whether the relationship between acquisitions at the focal and tied-to firms still holds when the focal and tied-to firms are very dissimilar, the following analysis was performed. First, ties were classified by the industry of the tied-to firm. This classification shows that the largest proportion of ties are to commercial banks. Across firms, 22 percent of ties are to banks. Banks are in many ways dissimilar to firms in the sampled industries, and the financial services industry is quite different from the manufacturing and service industries sampled in this study, having undergone a great deal of deregulation during the 1980s. Given that banks represent such a large proportion of ties, and that banks are in many ways dissimilar to the focal firms, I performed an analysis to test whether the 580/ASQ, December 1993

364

P. R. Haunschild Intel-organizational Imitation

Table 2 (continued)

10

11

12

13

14

15

16

17

18

19

20

21

22

.05 -.01 -.01 -.04 -.02

.53 .32 .11 .50

.65 .48 .79

.04 .11

.25

.03 .01

.21 .32

.31 .37

.25 .24

.15 .18

.21 .29

.79

-.03

.29

.35

.10

.22

.34

.54

.61

.16 -.01 .01 .03

.24 .21 .04 .23

.31 .36 .32 .39

.24 .26 .11 .11

.16 .17 .42 .21

.22 .24 .29 .43

.69 .58 .54 .62

.90 .70 .60 .82

.28 .64 .30 .43

.53 .72 .80

.42 .46

.42

-.01

.17

.27

.09

.21

.24

.60

.58

.38

.51

.44

.44

relationship between acquisitions by the focal and tied-to firms varies by whether the tied-to firm is a bank. To do this, the number of acquisitions completed by the tied-to firms was separated into two variables, according to whether the tied-to firm was a bank or a nonbank, one variable representing the number of acquisitions completed by banks during the three years before the sampled year was created and the other representing the number of acquisitions completed by nonbanks during the three years before the sampled year. If the similarity explanation is true, the relationship between acquisitions by the tied-to and focal firms should hold only when the tied-to firm is not a bank. Table 4 presents the results. Model 1 shows that the relationship between acquisitions at the tied-to and focal firms does not differ by whether the tied-to firm is a bank. Both the number of acquisitions by tied-to banks and the number of acquisitions by tied-to nonbanks are significantly related to acquisitions by the focal firm. Thus H4 receives no support from this test. The relationship between acquisitions by the focal and tied-to firms holds even when the focal and tied-to firms are dissimilar. Another model, not reported here, shows that the squared terms for both banks and nonbanks are significant when included in this model. This also supports the hypothesis that both banks and nonbanks are a source of imitation. 581/ASQ, December 1993

Interorganizational Imitation

365

Table 3

Imitation Results* Model Variable

(1)

Intercept

— 1.53** (.535)

Ν Acqsns. by tied-to firms Prior 3 years (Prior 3 years)

2

.085(.026)

(2) -1.63(.502)

Adj. Income growth Adj. R O A Adj. ROE (Log) Free cash flow Year 1982 1983 1984 1985 1986 1987 1988 1989 1990 Transportation Wholesale Business svcs. Assetst Ν focal firm acqsns. prior 3 yrs. Log-likelihood Ν

1.24* (.511) .001 (.001) .085 (.063) -.303 (.221) -.173 (.143) -.731 (.591) -1.61* (.655) -.753 (.596) -.618 (.604) .311 (.571) .787 (.588) -1.97" (.714) -.768 (.568) -.791 (.607) .489 (.445) .707* (.353) .763* (.371) 2.99 (25.99) .453(.067) -324.71 327

(4) -1.96 (1.51)

(5) 2.96(.937)

.296(.050) - .007(.001)

Horizontal prior 3 yrs. Vertical prior 3 yrs. Conglomerate prior 3 yrs. Control variables (Log) Debt/equity

(3) -1.76* (734)

1.23(.476) .001 (.001) .069 (.058) -.305 (.200) -.213 (.136) -.909 (.553) -1.73(.615) -1.14* (.567) -.553 (.559) -.096 (.539) .643 (.547) -1.85(.660) -.848 (.529) -.854 (.564) .307 (.416) .820* (.332) .760* (.348) -17.71 (24.56)

.114* (.055) -.105 (.271) .129 (.108)

.177 (.181) .173* (.059) -.084 (1.76)

.715 (702) .001 (001) -.018 (.078) .019 (.142) -.345 (.208) -1.37 (.990) -.828 (.886) .005 (.771) 1.00 (.737) .626 (.746) 1.85* (-738) -1.59 (1.06) -1.56 (1.03) .521 (.757) -.259 (.606) .394 (.486) .920* (.478) 52.14 (40.30)

2.73* (1.18) .001 (.001) -.017 (.116) -.265 (.436) -1.18 (.999) -2.23 (.143) -.816 (.654) -.139 (.984) -.645 (1.13) -.083 (.997) .022 (1.04) -1.20 (1.14) -3.14* (1.59) -1.63 (1.23) 1.65* (.866) .154 (.766) -.258 (.888) 85.57 (60.67)

-.031 (.101) .300 (.269) .291* (.118) 1.78* (.819) .001 (.001) .227(.073) -.134 (.167) -.015 (.199) .637 (.914) -.543 (1.05) -.103 (1.01) -.121 (.970) 1.01 (.913) .478 (.972) -.594 (1.01) -.272 (.965) -.059 (1.01) .824 (.623) .350 (.571) .752 (.570) 126.67* (59.08)

.448(.062) -312.43 327

-156.04 289

-86.07 289

-152.19 289

* p < .05; ~ p < .01. • The dependent variable for models 1 and 2 is the number of current-year acquisitions by the focal firm. The dependent variable for model 3 is the number of current-year horizontal acquisitions by the focal firm, for model 4 it is the number of current-year vertical acquisitions by the focal firm, and for model 5 it is the number of current-year conglomerate acquisitions by the focal firm, t To correct for right skew, the negative reciprocal of the assets variable (originally measured in thousands) w a s used.

582/ASQ, D e c e m b e r 1993

366

P. R. Haunschild Interorganizational Imitation

Table 4 Alternative Explanation Results· Model Variable

(2)

(1)

Intercept

-1.56** (.529)

Ν A c q s n s . by tied-to firms Current year only

.056 (.072) .269** (.061) - .007(.001)

Prior 3 years (Prior 3 years)

2

Prior 3 years {banks only) Prior 3 years (nonbanks only) Network centrality

-1.68— (.507)

(3) -.713 (.9441

.5151.046)

(4) -1.99(.532)

.239(.058) -.006(.001)

.332(.124) -.025* (.011) -.154 (.316) .053 (.085)

Period Period χ tied-to acqsns.

Adj. income growth Ad¡. R O A Ad¡. R O E (Log) Free cash flow Year—1982 1983 1984 1985 1986 1987 1988 1989 1990 Transportation Wholesale B u s i n e s s services Assetst

Ν focal acqsns. prior 3 yrs. Log-likelihood

Ν

.2761.081) - .007(.002)

.149* (.072) 075~ (.029)

(Network centrality) 2

Control variables (Log) Debt/equity

(5) -2.18(.490)

1 23* (.504) .001 (.001) .061 (.066) -.249 (.217) -.160 (.142) -.741 (.582) -1.62* (.646) -.869 (.597) -.623 (.597) .235 (.566) .683 (.582) -1.98(.7041 - 815 (.562) -.853 (.598) .455 (.442) .768* (.349) .820* (.368) -.919 (25.73) ,449~ (.066) -323.31 316

1.25(.476) .001 1.001) .063 (.059) -.314 1.200) -.213 (.136) - 870 (.556) -1.67(.619) -1.15· (.569) -.538 (.560) -.107 (.540) .681 (550) -1.82(.662) -.794 (.534) -.809 (.567) .284 (.417) .817* (.332) 752* (.348) -17.17 (24.60) .455(.063) -312.13 327

-.568 (.987) 001 1.001) 173 (.105) .015 (.142) -.185 (.272) -.213 (.990) -2.75* (1.32) .205 (1.03) .342 (1.03) .595 (1.03) -1.23 (1.13) -2.25 (1.23) -1.90 (1.09) -.481 (1.08) .809 (.720) -.139 (.6431 .079 (.673) 148.88* (60.13) -.187 (.1261 -318.62 327

1.13* (.476) .001 1.001) .080 (.058) -.306 (.196) -.219 1.135) -.940 (.555) -1.701.611) -.964 (.567) -.612 (.562) .052 (.546) .743 (.550) -1.78(.656) -.628 (.533) - 746 (.564) .175 (.416) .807* (.331) .799* (.348) -25.80 (24.69) .440— (.061) -308.79 327

.833 (.4991 001 (.001) .056 (.079) -.331 (.2231 -.272 (.146)

.317 (.441) .903* (.358) .942(.373) -7.78 (26.39) .454(.066) -326.44 327

* p < .05; - p < .01. • The dependent variable for models 1 . 2 , 4 , and 5 is the number of focal-firm current-year acquisitions. The dependent variable for model 3 is the number of tied-to firm current-year acquisitions, t To correct for right skew, the negative reciprocal of the a s s e t s variable w a s used.

583/ASQ, December 1993

Interorganizational Imitation Model 2 in Table 4 presents the results of the test of H5, whether current acquisitions by the tied-to firms are related to current acquisitions by the focal firm. Results show no support for this hypothesis. Only prior acquisitions, not current acquisitions by the tied-to firms, are related to current acquisitions by the focal firm. Other results, not reported here, show that this relationship also holds for the number of horizontal, vertical, and conglomerate acquisitions by the focal firm. Further, this result holds for the number of acquisitions by banks and nonbanks. Only prior, not current acquisitions by both banks and nonbanks are related to current acquisitions by the focal firm. Model 3 in Table 4 presents the results of the test of H6, whether the relationship between the tied-to and the focal firm can be reversed. The dependent variable in this analysis is current-year acquisitions by the tied-to firms, and the main independent variable is prior year acquisitions by the focal firm, the reverse of the relationship tested in model 1 in Table 3. Prior acquisitions by the focal firm are not related to current acquisitions by the tied-to firms. Thus, the imitation relationship does not reverse. This result supports the idea that it is not some third variable common to both the focal and tied-to firms that is causing the observed acquisition relationship. Private Information Table 4 also presents results of tests for private-information explanations. Model 4 shows the results of the test of H7, whether access to private information accounts for the relationship between acquisitions by the tied-to and focal firms. Support for this hypothesis requires that network centrality be positively related to the number of acquisitions by the focal firm. Yet if the imitation relationship is just a proxy for private-information transmission, then including network centrality in this model should result in a nonsignificant effect for the number of acquisitions by the tied-to firms. As shown in model 4, independent of the number of acquisitions by the tied-to firms, those firms that have more director ties to other firms, and are therefore more central in the network, are completing more acquisitions in the current year, which supports the private-information explanations. Centrality squared is negative and significant, however, suggesting that there is a limit to the effect of private information. But even controlling for network centrality, imitation predictions are still supported. Because evidence of private-information transmission through director ties was found, I tested whether the information communicated is private information about opportunities. What we want to know is whether there is a relationship between the industries in which the tied-to firms are acquiring and the industries in which the focal firms are acquiring. If this relationship exists, then the number of acquisitions by the focal firm in a particular industry should be greater when the tied-to firms also acquired in that industry. To test this, the number of acquisitions by the focal firm were classified by whether the tied-to firms acquired in that industry or not. This analysis was performed for those 72 pairs of firms in which both the focal and tied-to firms 584/ASQ, December 1993

367

368

P. R. Haunschild Intel-organizational Imitation

completed one or more acquisitions. There were only twelve cases in which the focal and tied-to firms completed one or more acquisitions in the same industry. A f-test of the mean number of focal-firm acquisitions in an industry when the tied-to firms also acquired in that industry (mean = 1.60, SD = 1.24) and of the mean number of focal-firm acquisitions in an industry when the tied-to firms did not acquire in that industry (mean = 1.89, SD = 1.55) is not significant [f(250) = -1.11, ρ < .27], This suggests that director ties were not used to communicate private information about acquisition opportunities. The final test of the private-information explanation assesses whether the imitation relationship is stronger later in the 1981-1990 period. A model including the dummy variable for period and the interaction of period with tied-to firm acquisitions was run. As shown in model 5 in Table 4, the interaction of period by tied-to firm acquisitions is not significant, indicating there is no difference in the impact of imitation over the period covered by this study. Control Variables

Although there does not appear to be support for free cash flow as an explanation for either the number or types of acquisitions completed by the focal firm, it appears that firms with more debt relative to equity make more acquisitions, especially vertical and conglomerate acquisitions. This result is the opposite of Jensen's prediction and may reflect the rise of debt as a financing tool in the 1980s. Since prior acquisitions are associated with current acquisitions, the relationship between debt and current acquisitions may be the result of financing these earlier acquisitions. Measures of average free cash flow and debt over the prior three years were used in all analyses. Results do not change when prior year free cash flow and debt were used instead of averages. There is some support for the theory that prior firm performance is positively related to conglomerate acquisitions. High return on assets, relative to the rest of the firms in the industry, is positively related to the number of conglomerate acquisitions completed by the focal firms. This supports either a hubris or availability-of-financing explanation for conglomerate acquisitions. As expected, the inclusion of dummy variables for year shows that the number of acquisitions varies for some years. When annual GNP, cost of capital, and NYSE volume were included instead of the dummy variables for year, the coefficients on GNP and NYSE volume were significant. GNP was negatively related and NYSE volume was positively related to acquisition activity, although these models did not fit as well as the models with year. The dummy variables for year, being more global proxies for macroeconomic conditions, capture more variance and are more conservative controls. Therefore, the year dummies were included in all other analyses reported in this study. The number of acquisitions varies by the industry of the focal firm. Firms in the wholesale and business services industries completed more acquisitions than firms in the 585/ASQ, December 1993

Interorganizational Imitation

369

electrical equipment and transportation industries. Further, the type of acquisition varies by industry. Firms in the business services industry completed more horizontal acquisitions than firms in the electrical equipment and wholesale industries. Firms in the transportation industry completed more vertical acquisitions than firms in other industries. The relationship between the size of the sampled firm and the number of acquisitions completed is not significant. Yet size is positively related to the number of conglomerate (but not horizontal or vertical) acquisitions. Finally, the number of acquisitions by the focal firm in the prior three years has a positive effect on acquisitions in the current year. This is consistent with Amburgey and Miner's (1992) finding that firms exhibit repetitive momentum in merger activity, tending to repeat specific strategic actions. DISCUSSION AND CONCLUSIONS Overall, the results provide strong evidence that imitation plays a powerful role in corporate acquisition activities. There is a relationship between a focal firm's acquisition activity and acquisition activity by those firms that are tied to a focal firm through directorships. A predicted nonlinear relationship between the number of acquisitions by the tied-to firms and the number of acquisitions by the focal firm was found. When further broken down into acquisition types, the relationship also holds for the number of horizontal, vertical, and conglomerate acquisitions by the tied-to firms. These results hold independent of several controls for financial and managerial explanations for acquisitions. Analyses also showed that the imitation relationship is not restricted to those cases in which the focal and tied-to firms are subject to similar conditions. Three pieces of evidence help demonstrate that imitation, not some variable related to both firms being subject to similar conditions, is driving the relationship between acquisitions by the focal and tied-to firms: (1) the focal firms are imitating the acquisition activities of banks, and banks are quite dissimilar to the focal firms, (2) there must be a one-year lag before imitation occurs, and (3) focal firms are imitating their tied-to firms, but tied-to firms are not imitating focal firms. Results are also consistent with the idea that acquisitionrelated information may be communicated through director ties. Results show that firms with many ties to other firms are also making many acquisitions, suggesting that they are exposed to "better" acquisition information than firms with few ties to other firms. This may be especially true of bank ties. Because of their role as financial intermediaries, banks may be particularly good sources of private information. Although support for private information was found through this network centrality test, it is important to realize that the number of acquisitions by the tied-to firms is still related to the number of acquisitions by the focal firm. This relationship holds for both bank and nonbank ties and is independent of the effect of the number of ties (network 586/ASQ, December 1993

370

P. R. Haunschild Interorganizational Imitation

centrality). This suggests that director ties are not only a potential source of private information, they are also a source of models to be imitated. Further, the information that is being communicated does not appear to take the form of information about acquisition opportunities. The results showing no relationship between the industries in which the focal and tied-to firms are acquiring indicates that it is unlikely that private information about opportunities is being transmitted. This is closest to a financial or efficiency-based theory of private information. Rather, private information is likely to take the form of more general acquisition know-how or normative information about the appropriateness of acquisitions. The fact that the imitation relationship does not appear to vary over the period included in this study provides no support for an institutional interpretation of the imitation relationship. It may be that imitation is occurring for noninstitutional reasons. Another explanation for the lack of variation rests on the fact that the origins of the 1980s merger wave date back to the mid-1970s. If acquisitions were becoming normative and socially accepted during the 1975-1980 period, the imitation model would fit better during this earlier period than the period that was tested in this study. Tentative support for this explanation was found when the 1981 and 1982 acquisition data were dropped from the sample, and models were run on the 1983-1990 period alone. All imitation results hold for the 1983-1990 period but do not hold for the 1981-1982 period. It appears that the imitation results are stronger later in the 1981-1990 period, where later means after 1982. This suggests that had data for 1975-1980 been available, and had the sample been split between 1975-1982 and 1983-1990, the interaction term may have been significant. The above explanations and interpretations of results are quite conservative in that they gave the private-information explanation every possible chance to work. Yet most of the results that were obtained not only support privateinformation theory, they also support imitation. It was outlined above how the relationship between network centrality and acquisitions by the focal firm could support the private-information explanation. Yet two other interpretations of the relationship between network centrality and the number of acquisitions are possible. First, it could be that the tied-to firms are being exposed to some kind of normative information about the appropriateness of acquisitions, which is not related to the tied-to firms' completing acquisitions themselves. Being tied to more firms means being exposed to more of this normative information. The transmission of normative information is very different from the transmission of information about opportunities or know-how. Thus, it can be argued that this result does not support the private-information interpretation, at least not in the way that private information has been discussed in other literatures. The second alternative interpretation is that this relationship could be the result of second-order imitation or exposure to information. Firms 587/ASQ, December 1993

Interorganizational Imitation

371

with many ties are likely to be tied to firms that are tied to other firms completing acquisitions. Information about these acquisitions may filter back to the focal firm through director ties and cause imitation by the focal firms. These various interpretations cannot be disentangled in this study. One limitation of this study is a lack of direct indicators of what types of information are being communicated through director networks. Thus, only one form of private information, information about acquisition opportunities, could be tested. There are no other interlock studies that measure the information transmitted through the interlock. This study shows that this information may vary and may have important effects on the interlocked firms. Better measures of both private and nonprivate information transmission would be valuable for future studies. Another limitation of this study has to do with the causes of imitation behavior by these managers. We know that acquisitions are imitated, but we don't know why. Imitation may be the result of an effort to increase legitimacy, it may be a form of interorganizational learning or a strategic response to competitor activities, or it may also be that managers are responding to uncertainty and doing whatever those other firms they are familiar with are doing. But perhaps the most compelling explanation for imitative behavior lies at the intersection of various theories of social structure, resource dependence, and institutionalization. If a tied-to firm acquires another firm as a means to manage dependence, then the focal firm might do the same thing either because (1) the logic of such an action becomes clear, a variation of the private-information argument, or (2) an eat-or-be-eaten sense of urgency is transmitted through the tie, which is likely, given that the direct tie turns this event into something proximate and concrete.1 Or it may be that generalized beliefs about the efficacy of acquisitions are highlighted by direct ties. These ideas point to some limitations in the current state of organizational imitation theory. We know little about what dimensions of activities will be imitated under various conditions. Do firms imitate exact practices, or some variant of them? How much of the activity surrounding a transaction will a firm imitate? Evidence from this study suggests that firms are imitating the type of acquisition completed by the tied-to firms. But are they imitating other acquisition-related activities, like deal negotiation techniques or use of a particular investment banking firm? We also know little about why firms imitate other firms. Furthermore, we know little about when firms imitate and whether uncertainty is a necessary condition. Finally, there is more to be learned about who imitates. The results of this study, as well as of Galaskiewicz and Wasserman's (1989), show that imitation flows along the interpersonal networks of members of different firms. Presumably, these are people who know and trust each other and may also be similar to each other. But there are others that may be imitated. DiMaggio and Powell , (1983), for example, suggested that prestigious firms are Thanks to Marshall Meyer for suggesting m o r e likely to be imitated. These are all issues we need to these connections.

explore. 588/ASQ, December 1993

372

P. R. Haunschild Interorganizational Imitation

Implications This study enhances our understanding of the scope and potential mechanisms of interorganizational imitation. We know that director interlocks are a source of imitative behavior among firms. Most imitation studies, especially those that use the percentage of adopters as support for imitation, do not specify exactly how imitation takes place (see Galaskiewicz and Wasserman, 1989 for an exception). A major contribution of this study is that support for imitation is achieved through rigorous development and testing of alternative explanations for results consistent with imitation. This study also represents a contribution to the interlock literature. The results of this study are consistent with studies that show a relationship between interlocks and the adoption of firm practices and structures (O'Reilly, Main, and Crystal, 1988; Davis, 1991; Mizruchi, 1992; Palmer, Jennings, and Zhou, 1993). There are no other studies that investigate interlocks as a source of models to be imitated, however, and researchers are only beginning to look at interlocks as a source of information. This study is consistent with Useem's (1984) proposition that interlocks serve as a tool for firms to use in their scan of the business environment. While Useem and others have not directly measured information transmission through interlocks, the results of this study suggest that interlocks serve as a source of both acquisition models and acquisition information. Further, the information transmitted through interlocks appears to be inconsistent with predictions generated by financial theories of information transmission. Existing acquisition theories focus on financial and market conditions or managerial motives. To date, little consideration has been given to exploring the effect on acquisition activities of the social context surrounding firms and firm managers. The results of this study support a social embeddedness perspective on corporate acquisitions. Further, a specific mechanism for how social structure affects firm activities is specified. Social networks are an important source of acquisition models and acquisition information. This provides a social explanation in an area in which much disagreement about other explanations exists. Further research related to imitative processes seems promising. Director ties are not likely to be the only mechanism through which imitation occurs. The business press, which regularly reports on firm practices, may diffuse models. The media also provides an evaluation of the success of various practices, which may affect the likelihood of adoption. Professional firms are another mechanism through which practices and structures are spread from firm to firm (DiMaggio and Powell, 1983). Investment bankers, consultants, accounting firms, and other professionals may have influenced the spread of acquisition models from firm to firm in the 1980s and are likely to influence the adoption of other important practices. It is clear from this study that imitation is an important influence on a substantive strategic action: corporate acquisitions. Other major structures and practices are also likely to be subject to imitative pressures. For example, the recent proliferation of Total Quality 589/ASQ, December 1993

Interorganizational Imitation

373

Management practices is likely to be partially due to interorganizational imitation. Practices and structures are proliferating every day, and the same forces that led to imitation in the merger wave may also be fueling the spread of these other practices. REFERENCES Amburgey, Terry L., and Anne S. Miner 1992 "Strategic momentum: The effects of repetitive, positional, and contextual momentum on merger activity." Strategic Management Journal, 13: 335-348. Amihud, Yakov, and Baruch Lev 1981 "Risk reduction as a managerial motive for conglomerate mergers." Bell Journal of Economics, 12: 605-617. Armour, Henry, and David Teece 1978 "Organizational structure and economic performance." Bell Journal of Economics, 3: 106-122. Bandura, Albert 1977 Social Learning Theory. Englewood Cliffs, NJ: Prentice-Hail. Barney, Jay B. 1988 "Returns to bidding firms in mergers and acquisitions: Reconsidering the relatedness hypothesis." Strategic Management Journal, 9: 71-78. Baumol, William J. 1967 Business Behavior, Value and Growth. New York: MacMillan. Beckenstein, Alan R. 1979 "Merger activity and merger theories: An empirical investigation." The Antitrust Bulletin, Spring: 105-128. Washington, DC: Federal Legal Publications. Becketti, Sean 1986 "Corporate mergers and the business cycle." Federal Reserve Bank of Kansas City, Economic Review (May): 13-26. Blair, Margaret, Sarah Lane, and Martha Schary 1991 Patterns of Corporate Restructuring, 1955-87. Washington, DC: Brookings Institution.

Bolton, Michele 1992 "Organizational miming in the radio broadcasting industry: Reducing the risks of innovation." Working Paper, College of Business, San Jose State University. Bradley, Michael, Anand Desai, and E Han Kim 1983 "The rationale behind interfirm tender offers: Information or synergy?" Journal of Economics, 11: 183-206. Burt, Ronald S. 1983 Corporate Profiles and Cooptation: Networks of Market Constraints and Directorate Ties in the American Economy. New York: Academic Press. Chatterjee, Sayan 1986 "Types of synergy and economic value: The impact of acquisitions on merging and rival firms." Strategic Management Journal, 7: 119-140. Davis, Gerald F. 1991 "Agents without principles? The spread of the poison pill through the intercorporate network." Administrative Science Quarterly, 36: 583-613. Davis, Gerald F„ and Walter W. Powell 1992 "Organization-environment relations." In Marvin D. Dunette and Leaetta M. Hough (eds.), Handbook of Industrial and Organizational Psychology: 315-376. Palo Alto, CA: Consulting Psychologists Press. DiMaggio, Paul J „ and Walter W. Powell 1983 "The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields." American Sociological Review, 48: 147-160.

Dutton, John M „ and Richard D. Freedman 1985 "External environment and internal strategies: Calculating, experimenting and imitating in organizations." In R. Lamb and P. Shrivastava (eds.). Advances in Strategic Management, 3: 39-67. Greenwich, CT: JAI Press. Fligstein, Neil 1985 "The spread of the multidivisional form among large firms, 1919-1979." American Sociological Review, 50: 377-391. 1990a The Transformation of Corporate Control. Cambridge, MA: Harvard University Press. 1990b "The finance conception of the corporation and the causes of the financial reorganization of large American corporations, 1979-87." Paper presented at the Annual Meeting of the American Sociological Association, Washington, DC. Fowler, Karen, and Dennis Schmidt 1988 "Tender offers, acquisitions and subsequent performance in manufacturing firms." Academy of Management Journal, 31: 962-974. Galaskiewicz, Joseph, and Stanley Wasserman 1989 "Mimetic processes within an interorganizational field: An empirical test." Administrative Science Quarterly, 34: 454-479. Golbe, Devra, and Lawrence White 1988 "A time-series analysis of mergers and acquisitions in the U.S. economy." In Alan J. Auerbach (ed.). Corporate Takeovers: Causes and Consequences: 265-309. Chicago: University of Chicago Press. Granovetter, Mark 1985 "Economic action and social structure: The problem of embeddedness." American Journal of Sociology, 91 : 481-510.

590/ASQ, December 1993

374

P. R. Haunschild Interorganizational Imitation

Greene, William H. 1990a LIMDEP. New York: Econometric Software. 1990b Econometric Analysis. New York: Macmillan. Grimm, W. T., 8c Company 1987 Mergerstat Review. Chicago, W.T. Grimm. Herman, Edward S., and Louis Lowenstein 1988 "The efficiency effects of hostile takeovers." In John Coffee, Louis Lowenstein, and Susan Rose-Ackerman (eds ). Knights, Raiders and Targets: 211-240. New York: Oxford University Press. Hirsch, Paul M. 1987 Pack Your Own Parachute: How to Survive Mergers, Takeovers, and Other Corporate Disasters. Reading, MA: Addison-Wesley.

Lehn, Kenneth, and Annette Poulsen 1989 "Free cash flow and stockholder gains in going private transactions." Journal of Finance, 44: 771-787. Levitt, Barbara, and James G. March 1988 "Organizational learning." In W. Richard Scott and Judith Blake (eds.), Annual Review of Sociology. 14:' 319-340. Palo Alto, CA: Annual Reviews. Lieberman, Marvin, and David Montgomery 1988 "First-mover-advantages." Strategic Management Journal, 9: 41-58. Lubatkin, Michael 1987 "Merger strategies and stockholder value." Strategic Management Journal, 8: 39—53.

Hogarty, Thomas F. 1970 "Profits from mergers: The evidence of fifty years." St. John's Law Review, 44: 378-391.

Maddala, G. S. 1983 Limited Dependent and Qualitative Variables in Econometrics. New York: Cambridge University Press.

Jarrell, Gregg A. 1987 "Financial innovation and corporate mergers." In L. E. Browne and E. S. Rosengren (eds.), The Merger Boom: 52-73. Boston: Federal Reserve Bank.

Mariolis, Peter, and Maria H. Jones 1982 "Centrality in corporate interlock networks: Reliability and stability." Administrative Science Quarterly, 27: 571-584.

Jensen, Michael 1984 "Takeovers: Folklore and science." Harvard Business Review, 62 INov.-Dec.): 109-121. 1987 "The free cash flow theory of takeovers: A financial perspective on mergers and acquisitions and the economy." In. L. Browne and E. Rosengren (eds.). The Merger Boom: 102-143. Boston: Federal Reserve Bank.

Marris, Robin 1964 The Economic Theory of Managerial Capitalism. London: MacMillan.

Kazdin, A. E. 1976 "Effects of covert modeling, multiple models and model reinforcerrent on assertive behavior." Behavior Theory, 7: 211-222. Lant, Theresa K., and Stephen J. Mezias 1990 "Managing discontinuous change: A simulation study of organizational learning and entrepreneurship." Strategic Management Journal, 11 : 147-179.

Mintz, Beth, and Michael Schwartz 1985 The Power Structure of American Business. Chicago: University of Chicago Press. Mizruchi, Mark 1992 The Structure of Corporate Political.Action: Interfirm Relationships and Their Consequences. Cambridge, MA: Harvard University Press. Morck, Randall, Andrei Shleifer, and Robert W. Vishny 1990 "Do managerial objectives drive bad acquisitions?" Journal of Finance, 45: 31-48. Nelson, Ralph L. 1959 Merger Movements in American Industry, 1895-1956. Princeton: Princeton University Press. O'Reilly, Charles Α., Brian G. Main, and Graef S. Crystal 1988 "CEO compensation as tournament and social comparison: A tale of two theories." Administrative Science Quarterly, 33: 257-274. 591/ASQ, December 1993

Palmer, Donald 1983 "Broken ties: Interlocking directorates and intercorporate coordination." Administrative Science Quarterly, 28: 40-55. Palmer, Donald, Ρ Devereaux Jennings, and Xueguang Zhou 1993 "Late adoption of the muitidivisional form by large U.S. corporations: Institutional, political and economic accounts." Administrative Science Quarterly, 38: 100-131. Pfeffer, Jeffrey 1972 "Merger as a response to organizational interdependence." Administrative Science Quarterly, 17: 382-394. Pfeffer, Jeffrey, and Gerald Salancik 1978 The External Control of Organizations: A Resource Dependence Perspective. New York: Harper and Row. Ravenscraft, David 1987 "The 1980s merger wave: An industrial organization perspective." In L. Browne and E. Rosengren (eds.), The Merger Boom: 17-37 Boston: Federal Reserve Bank. Ravenscraft, David, and F. M. Scherer 1987 Mergers, Sell-Offs and Economic Efficiency. Washington, DC: Brookings Institution. Rogers, Everett 1983 Diffusion of Innovations, 3rd ed. New York: Free Press. Roll, Richard 1986 "The hubris hypothesis of corporate takeovers." Journal of Business, 59: 197-212. Salter, Malcolm, and Wolf A. Weinhold 1979 Diversification through Acquisition. New York: Free Press. Schweiger, David M „ and James P. Walsh 1990 "Mergers and acquisitions: An interdisciplinary view." In J. R. Ferris and Κ. M. Rowland (eds.). Research in Personnel and Human Resource Management, 8: 41-107. Greenwich, CT: JAI Press. Shleifer, Andrei, and Robert W. Vishny 1988 "Value maximization and the acquisition process." Journal of Economic Perspectives, 2: 7-20.

Interorganizational Imitation 1 9 9 0 " T h e takeover w a v e of the 1 9 8 0 s . " Science, 17: 745-749. S t e a r n s , Linda Brewster, a n d M a r k S . Mizruchi 1987 " B r o k e n tie reconstitution and the functions of interorganizational interlocks: A reexamination." Administrative S c i e n c e Quarterly, 3 1 : 5 2 2 - 5 3 8 . Tobin, J a m e s 1 9 5 8 " E s t i m a t i o n of relationships for limited dependent variables." Econometrica, 26: 24-36.

Tolbert, P a m e l a S., a n d L y n n G . Zucker 1 9 8 3 "Institutional s o u r c e s of c h a n g e in the formal structure of organizations: T h e diffusion of civil service reform, 1 8 8 0 - 1 9 3 5 . " Administrative Science Quarterly. 2 8 : 2 2 - 3 9 . Trautwein, Friedrich 1 9 9 0 " M e r g e r motives and merger prescriptions." Strategic M a n a g e m e n t Journal, 11 : 283-295. Useem, Michael 1 9 8 4 T h e Inner Circle. N e w York: O x f o r d University Press.

5 9 2 / A S Q , D e c e m b e r 1993

W a l s h , J a m e s P. 1 9 8 8 " T o p m a n a g e m e n t turnover following m e r g e r s a n d acquisitions." Strategic M a n a g e m e n t Journal, 9: 173-183. W i l l i a m s o n , Oliver E. 1 9 6 3 " M a n a g e r i a l discretion a n d b u s i n e s s behavior." A m e r i c a n E c o n o m i c Review, 1032: 1047-1051.

375

376

D. L. Deephouse

O Academy of Management fournal 1996, Vol. 39, No. 4, 1 0 2 4 - 1 0 3 9 .

DOES ISOMORPHISM LEGITIMATE? DAVID L. DEEPHOUSE Louisiana State University This study tests a central proposition of institutional theory, that organizational isomorphism increases organizational legitimacy. Results show that isomorphism in the strategies of commercial banks is related to legitimacy conferred by bank regulators and the media, even in the presence of organizational age, size, and performance.

Research in institutional theory has examined the causes of isomorphism, that is, the factors that lead organizations to adopt similar structures, strategies, and processes (Davis, 1991; DiMaggio & Powell, 1983; Mezias, 1990; Palmer, Jennings, & Zhou, 1993; Tolbert & Zucker, 1983). Isomorphism also has consequences that require examination (Jepperson, 1991; Zucker, 1987). A fundamental consequence of institutional isomorphism, according to institutional theory, is organizational legitimacy, the acceptance of an organization by its external environment (DiMaggio & Powell, 1983; Meyer & Rowan, 1977; Meyer & Scott, 1983). Like isomorphism, legitimacy is a crucial concept in institutional theory, serving as the "anchor-point of a vastly expanded theoretical apparatus" (Suchman, 1995: 571). Nevertheless, there have been few systematic efforts to test the isomorphism-legitimacy link because of continuing difficulties in defining and measuring legitimacy (Bozeman, 1993; Galaskiewicz, 1985; Suchman, 1995; Terreberry, 1968). This study addresses these gaps in institutional research by examining whether isomorphism in strategies is related to legitimacy conferred by regulators and the media. HYPOTHESIS DEVELOPMENT Organizational isomorphism (isomorphism, hereafter) is defined as the resemblance of a focal organization to other organizations in its environment (DiMaggio & Powell, 1983). Although DiMaggio and Powell discussed isomorphism as both a state and a process, I conceptualize it here as a state. That is, this article focuses on isomorphism as the similarity among a set of organizations at a given point in time.

I would like to thank Nathan Bennett, Timothy Coombs, David Ketchen, Matthew Kraatz, Richard Nelson, Craig Russell, and two anonymous reviewers for their assistance and helpful comments on earlier drafts of this article. I also appreciate the guidance of Philip Bromiley, Joseph Galaskiewicz, and especially my advisor, Andrew Van de Ven, on the dissertation from which this work was developed. 1024

Does Isomorphism Legitimate? 1996

Deephouse

377 1025

Researchers have identified several organizational characteristics that are subject to isomorphism. Initial concerns were with structures and practices (Meyer & Rowan, 1 9 7 7 ; Tolbert & Zucker, 1 9 8 3 ) . Recently, strategies have been examined. For example, Fligstein ( 1 9 9 1 ) , Haunschild ( 1 9 9 3 ) , Haveman ( 1 9 9 3 ) , and Abrahamson and Hegeman ( 1 9 9 4 ) demonstrated the importance of imitating other firms (i.e., mimetic isomorphism) in the choice of acquisition, diversification, and financial strategies. This study focuses on strategic isomorphism, the similarity of a focal organization's strategy to the strategies of other organizations in its industry. As was true regarding isomorphism, legitimacy can be conceptualized as both a process and a state. Here I emphasize the latter. Legitimacy also can be conceptualized from an evaluative perspective, signifying desirability and normativity, or from a cognitive perspective, signifying understandability and taken-for-grantedness (Aldrich & Fiol, 1994; Jepperson, 1991; Suchman, 1995). Here I emphasize the evaluative perspective. Organizational legitimacy (legitimacy, hereafter) is defined as a status conferred by social actors (Ashforth & Gibbs, 1 9 9 0 ; Pfeffer & Salancik, 1 9 7 8 ) . From the perspective of a particular social actor, a legitimate organization is one whose values and actions are congruent with that social actor's values and expectations for action (Galaskiewicz, 1 9 8 5 ; Pfeffer & Salancik, 1 9 7 8 ) . The social actor accepts or endorses the organization's means and ends as valid, reasonable, and rational (Ashforth 8c Gibbs, 1990; Baum & Oliver, 1991; Meyer & Scott, 1 9 8 3 ; Singh, Tucker, & House, 1 9 8 6 ; Stinchcombe, 1 9 6 8 ) . Given that legitimacy is the endorsement of an organization by social actors, a key step in defining it is identifying relevant social actors. In this research I follow Meyer and Scott (1983), Galaskiewicz (1985), and Baum and Oliver (1991) in arguing that only certain actors have the standing to confer legitimacy. One important set of actors includes the government regulators who have authority over an organization (Baum & Oliver, 1991; Galaskiewicz, 1985; Meyer & Scott, 1983). A second key actor is public opinion, which has the important role of setting and maintaining standards of acceptability (Elsbach, 1994; Galaskiewicz, 1985; Meyer & Rowan, 1977; Meyer & Scott, 1983). Thus, this article focuses on two types of legitimacy by examining the evaluations of two social actors, government regulators and the general public. Regulatory endorsement is the acceptance of an organization by the state agencies that formally regulate it. Public endorsement is the acceptance of an organization by the general public. Theoretically, strategic isomorphism increases regulatory endorsement and public endorsement in the following way. In most industries, particular strategies are not required. Instead, ambiguity and uncertainty make the choice of appropriate strategies unclear (Abrahamson & Hegeman, 1994; Haveman, 1993). Consequently, organizations create norms of strategic behavior that social actors also come to accept (DiMaggio & Powell, 1983; Edelman, 1992). Proper strategic behavior diffuses across an industry in at least two related ways. First, organizations imitate other successful organiza-

378

D. L. Deephouse

1026

Academy

of Management

Journal

August

tions in the face of uncertainty (DiMaggio & Powell, 1983; Haveman, 1993). Second, organizations learn about proper behavior through trade associations, director linkages, and other networks (DiMaggio & Powell, 1983; Galaskiewicz & Wasserman, 1989; Haunschild, 1993). On the one hand, an organization conforming to norms of strategic behavior demonstrates that it is acting in an acceptable manner and social actors should evaluate it as legitimate (Meyer & Rowan, 1977). On the other hand, organizations that innovate or have unique strategies suffer in terms of legitimacy—such behavior is questioned or even deemed unacceptable by external actors (Meyer & Rowan, 1977). In sum, this study tests whether strategic behavior deviating from the norm is related to negative evaluations of organizations made by regulators and the general public. Stated formally, Hypothesis 1 : Greater strategic isomorphism is associated with greater regulatory endorsement. Hypothesis 2: Greater strategic isomorphism is associated with greater public endorsement. METHODS The hypotheses were tested in the entire population of commercial banks in the Minneapolis-Saint Paul metropolitan area (the Twin Cities) from 1985 through 1992. Commercial banks are chartered by regulators and differ from bank holding companies, which can own several financial service businesses. Hypotheses derived from institutional theory should hold in this sample because commercial banks face strong institutional forces (Scott & Meyer, 1991). For instance, banks face periodic scrutiny from regulators (Spong, 1990); banks also have a high degree of public trust (Ashforth & Gibbs, 1990). Furthermore, commercial banks are in the for-profit sector of the economy, which Powell (1991) recommended as an area for expanded empirical study. Moreover, studying a single industry in a single area eliminates confounding influences of different regulators and publics. I chose 1985 as the initial year because regulators changed reporting requirements in 1984 for the financial data used. All banks are required to file financial statements known as call reports with bank regulators. I collected the sample of banks and their yearend financial data from these reports. The number of banks ranged from 152 in 1985 to 95 in 1992. The unit of analysis was the bank-year. Measures Regulatory endorsement. Regulators evaluate commercial banks in two important ways: by examining banks' financial capital and by examining them on-site (Spong, 1985, 1990). A bank's financial capital position reflects its ability to protect depositor savings. Regulators assess this ability by classifying the bank's capital position into three ordered discrete categories. Banks in lower categories are not fully endorsed by the regulators. These banks instead are subject to increased regulatory scrutiny. The classification of banks used the capital ratios and categories specified by regulatory agencies

Does Isomorphism Legitimate? 1996

Deephouse

379 1027

(Spong, 1985, 1990). A minor complication is that regulators changed the capital ratio and categories used for classification during the sample period. From 1985 through 1988, they used the "total capital ratio"; from 1989 through 1992, they adopted the "tier 1 leverage ratio." (The definitions of these ratios and classification schemes appear in Appendix A.) The resulting variables are called regulatory assessment of total capital, 1985-88 and regulatory assessment of tier 1 leverage capital, 1989-92. They are coded (0, 1, 2), with fully endorsed banks having the highest coding (2).1 Regulators make on-site examinations to ascertain the safety and soundness of a bank's assets. When their examinations reveal that a bank has lowquality assets and is following unsafe banking practices, regulators issue enforcement actions. Enforcement actions require the bank to take certain actions, such as curtailing lending to a certain industry or firing top management. Information about such enforcement actions only became publicly available in 1991, by congressional statute. I obtained the record of enforcement actions from LEXIS, a legal database, and created a dichotomous variable called absence of regulatory enforcement actions, 1991-92. Banks not subject to an enforcement action during a year were given a rating of 0; banks under an enforcement action were given a - 1 . Public endorsement. I measured public endorsement from articles in the print media using content analysis. Media influence and reflect the values of a culture (Chen & Meindl, 1991; Dowling & Pfeffer, 1975; Gans, 1979). When activities of an organization are illegitimate, comments and attacks will occur, and the media will report such comments (Dowling & Pfeffer, 1975; Pfeffer & Salancik, 1978: 194). Researchers are beginning to use the media to measure legitimacy. For example, Hybels, Ryan, and Barley (1994) content-analyzed business periodical abstracts to assess the legitimacy of the population of "dedicated" biotechnology firms. Coombs (1992) contentanalyzed the New York Times and the Washington Post to assess the legitimacy of President Reagan's Task Force on Food Assistance. The Twin Cities' two metropolitan daily newspapers, the Minneapolis Star Tribune and the Saint Paul Pioneer Press, were sampled from 1988 to 1992. In a national survey, Stempel (1991) found that 67.3 percent of the population got their news about local businesses from the local newspaper. In contrast, the percentages using television, other people, and radio were much lower (27.0, 22.2, and 10.1%, respectively). Furthermore, audience recall of information contained in newspapers exceeds the recall of information from television and radio (DeFleur, Davenport, Cronin, & DeFleur, 1992). These two papers have the largest circulations in the area and thus should represent Twin Cities' public opinion. I selected 1988 as the first year of data collection for two reasons related to measurement accuracy. First, coding

1

After collecting the data, 1 discovered that there was no variation in the regulatory assessment of tier 1 leverage capital, 1989-92, among Twin Cities banks. All banks were fully endorsed. Consequently, this measure received no further statistical attention.

380 1028

D. L. Deephouse Academy of Management Journal

August

First Bank and Norwest Bank, the two largest area banks, would have been problematic for the period before their consolidation of their numerous banking units, which occurred at the beginning of 1988. For instance, before 1988 "First Bank" could refer to First Bank Lake, First Bank Grand, or several other independent banks. Thus, measurement accuracy could be reduced for these banks. Second, costs in time and money are important influences on sampling design (Sudman, 1976). I chose to focus the sampling effort on ensuring accuracy within years rather than increasing breadth across years. The sample of articles included all letters to the editor, all editorials, all columns, and a stratified sample of news articles. All letters, editorials, and columns were included because they represent interpretations of organizations that are important for conferring legitimacy (Dowling & Pfeffer, 1975; Pfeffer & Salancik, 1978). News articles reflect daily events and often come from press releases. I selected all news articles for each bank with fewer than eight articles in a year. For banks with more than eight, I randomly selected a total of eight plus 25 percent of the remaining number of articles. A sampling fraction of 25 percent is well above that used in most communication research (e.g., Krishnaiah, Signorielli, & McLeod, 1993; Riffe, Aust, & Lacy, 1993). Such research usually examines one or two well-covered topics over time. In contrast, this study looked at over 95 banks, many of which had little or no press coverage. In total, this procedure yielded 1,277 articles. Coding the articles entailed identifying and rating recording units (Weber, 1990). A recording unit comprised an individual bank in a single article. Because several articles mention many banks, 2,150 recording units were identified. Each recording unit was rated as endorsing or challenging the subject bank's legitimacy (Ashforth & Gibbs, 1990; Hirsch & Andrews, 1984). I developed a coding scheme to rate each recording unit. Appendix Β contains the terms and activities that challenged a bank's legitimacy. All articles were coded by the author. A colleague was instructed to use the same coding scheme on 23 percent (52) of the articles from one year. The two raters agreed on 68 of the 71 recording units (95.8%), suggesting high levels of intercoder reliability (Weber, 1990). The next step was to transform the recording units into a measure suitable for statistical analysis. The Janis-Fadner coefficient of imbalance was used to create annual measures of public endorsement for each bank (Budd, Thorp, & Donohew, 1967; Coombs, 1995; Hurwitz, Green, & Segal, 1976; Janis & Fadner, 1965). As implemented here, the coefficient measures the relative proportions of endorsing and challenging recording units for each bank in a year. The formula for its calculation is in Appendix C. I labeled this variable the coefficient of media endorsement. This coefficient has many useful properties, such as (1) a meaningful zero point when there are equal numbers of endorsing and challenging recording units, (2) a decrease in the coefficient when the number of challenging recording units increases, and (3) an increase in the coefficient when the number of endorsing recording units increases (Budd et al., 1967; Janis & Fadner, 1965). The measure is bounded by 1 and - 1 .

Does Isomorphism Legitimate? 1996

Deephouse

381 1029

Strategic isomorphism. Strategic isomorphism was measured using strategic conformity, the extent to which an organization's strategies resembled the conventional, normal strategies in an industry (Abrahamson & Hegeman, 1994; Finkelstein & Hambrick, 1990). Bank asset strategies were the key strategy variables used to measure strategic conformity. An asset strategy is the allocation of resources to a certain market (Chandler, 1962). It is measured here as a proportion of total assets. For example, the commercial lending strategy is measured as the proportion of assets that a bank commits to commercial loans. Eleven bank asset strategy variables were included here: commercial loans, real estate loans, loans to individuals, agriculture loans, other loans and leases, cash, overnight money, securities, trading accounts, fixed assets, and other assets. Haveman (1993) and Reger, Duhaime, and Stimpert (1992) used similar categories. I computed strategic conformity following Finkelstein and Hambrick (1990). Each key asset strategy for each bank was compared to the industry mean value for that variable and expressed as a standard deviation. The absolute values of the standard deviations for all the strategy variables were totaled for each bank, giving a holistic and parsimonious measure of deviation. Multiplying by - 1 created a scale on which more positive numbers indicate greater conformity. Organizational Attributes: Age, Size, and Performance In addition to isomorphism, the organizational attributes of age, size, and performance have been suggested by researchers as potentially important determinants of legitimacy.2 Older organizations are more likely to (1) develop strong exchange relationships, (2) become part of a power hierarchy, (3) be endorsed by powerful social actors, and (4) have an "aura of inevitability" (Hannan & Freeman, 1984:158; Singh et al., 1986). "Nothing legitimates both individual organizations and forms more than longevity" (Hannan & Freeman, 1984:158). I obtained the founding year for each bank from Polk's Bank Directory, a semiannual standard reference of banks. Legitimacy may also be affected by an organization's size. Larger firms may have more contractual and social ties to and endorsements from actors in their external environments (Galaskiewicz, 1985; Pfeffer & Salancik, 1978; Singh et al., 1986). I measured size using total average assets from the call reports (cf. Haveman, 1993). Performance also might affect legitimacy. Firms performing well are efficient at converting resources into goods and services, and society values such efficiency (Dowling & Pfeffer, 1975; Meyer & Rowan, 1977). Consistent with bank regulatory practice, return on average assets (ROA) from the call reports was the measure of performance used here. Analysis There were two types of dependent variables, each requiring a different analytic technique. I tested the regulatory endorsement variables, which are 2

1 thank an anonymous reviewer for recommending the inclusion of performance.

382

D. L. Deephouse

1030

Academy of Management Journal

August

ordered categorical variables, using logistic regression analysis (Fienberg, 1980; Greene, 1993; Maddala, 1983). The coefficient of media endorsement is bounded and may have many observations at the boundaries. It was estimated with censored regression (the "tobit" model; Amemiya, 1984; Greene, 1993; Maddala, 1983). The following equation is the general statistical model used for testing the models: Endorsement

= b0 + b, x strategic conformity + b2 x age + b3 x size + b4 Χ performance + e. RESULTS

Table 1 displays the descriptive statistics for the variables, including the frequencies of the regulatory endorsement measures. Of the observations for the 1985-88 regulatory assessment of total capital, 84.1 percent scored a 2, the highest level. Of the observations for the absence of regulatory enforcement actions, 1991-92, 95.4 percent scored a 0, the higher level, indicating that the observed banks did not have actions against them. For the coefficient of media endorsement, only 269 of the observations (50.6%) received press coverage in 1988-92. Of these, 78.4 percent scored a 1, meaning these banks had only endorsing coverage. Table 2 shows the correlations among the variables. The number of observations for each correlation varies because of the different sample periods used for each legitimacy measure. The coefficient of media endorsement is positively correlated with the two regulatory measures. Table 3 displays the results of the hypothesis testing. Standardized coefficients are reported. The first two models examine regulatory endorsement. Model 1 estimates the regulatory assessment of total capital, 1985-88, for which there were 554 observations. Strategic conformity had a significantly positive coefficient (β = 0.371, ρ < .01), supporting Hypothesis 1. The coeffiTABLE 1 Descriptive Statistics Frequency Variables Regulatory assessment of total capital. 1985-88 Absence of regulatory enforcement actions, 1991-92 Coefficient of media endorsement, 1988-92 Strategic conformity, 1985-92 Age, 1985-92 Size, 1985-92' Performance, 1985-92

Ν

Mean

554

1.81

0.46

194

-0.05

0.21

269 969 969 969 969

0.87 -7.68 53.32 277.85 0.01

0.31 2.96 33.40 1,436.02 0.01

' Size is expressed in millions of dollars.

s.d.

- 1 0

1 16

9

185

72

2 466

D o e s

I s o m o r p h i s m

Legitimate?

Deephouse

1 9 9 6

1 0 3 1

O O S ) O O CO oí

ο ο

in

o ο fi

o o o

σι co O)

o o t-í

o o

o> o o

o o

o

Ò

ο ο

σι co

to ο 0 1

t ο

σι o c o CS ο σ> ö ò

σ> ΙΟ o a> r-< c o co ο σ> α 01 ο

σι co 05

σ> co O)

o o ó

ö

fH ò

o o o

ÍÜ φ >. α>

o> co o> •3

Ut

,Ρ

o o i-í

o o o

σ> CM c o co r * o C~4 O o

OI O CO O) CO o (O *H o CO rH o CM CM O o O o

OI CO O CM Ò

CM σ ι i n co CM ο

CO > CO

Λ

o β Û3

8

S3

.

o ο r-i

o o Ò

f m o CT> en o iH c> o

in O)

*H (O CM iH O O) O 0 0 Γ-» Ò Ò O o

O) t-t

r-t m 00 o o o

f r-t O a> CM O rH O o

O) tH

Λ

S

Sá S

I m oo σι

g £ 13 Ο

3 8 3

α> I m οο σ>

BO


-a ¿.TS ^ tí

Ό ®

.2

S O.

Ο

S S M •g

®

a 2, S3 13 > 03 E? Η

® «

IH O

CM O

ò

ö

«β ¿3 ÛJ

I

§

§ I

t a ν ο ι o Ό CO U f i ® ω A «η ce

η

Ό I

en 03 £

i

0) »

in Ν 's

I

s ï

u

1 3

¡

¿

S

fl o

in ^ Ο Ö

m ^ τΗ Ò I

oo o O Ò I

CO Ol Ν Ο

Μ

CM t o co I O CO ο ò

co ò

1

Ό Λ)

w 1-J

g M

sa S

Ο Ν io en CO

«

«

+• co IH

* i-t m co co

1

ύ

a (Β

o 1-9 «M 0

3 • PN o. β υ

1

2

«s

s

i

°

1· tn o

O) o ò

CO I - I Ν CO r< ν σι oi τ—
iH

•a

„

Ό

s

u

J3

I

e o

" Ό

s

Φ

ça ® r i ·!"• o 00 φ

ä _ CO

• a

J3

?

Í

F o

Τ

•s 5. «Λ

co o O Ò I

« S *

m — 1

Ò S

ο

'S | < ¿ U S « o ce .. s. U co co O)

co M fH

o

00 o CM CM c o ΐ-ι I H I - I

co o IH

•

Ν

o

Ο

ò

Ò

Φ Λ

Ò

Ò

«« ««

ca.

« «

«

* It •

'S

s •ο

s

S" •β a

il

ΛΝ

c O u

Β S « S S S ï a « M

s«

Ό ο

β

s a 'S

a

O. O . . H φ φ 00 O U I ) « - â < w

«

Ο ΐ 55 tu

ε ^ ϊ ο ο« Β

,o3 es I

ce fa ® >2 M . Φ .a o o n α, S

Β

a oc o , J c

Q t-l Φ ,Ο e 3 s

ε o

®

S3 . 2

S

ra

-S

φ

a

S

"2

s. I φ Η o *-· . i-» CO ö * φ o V

Φ »—< JO α . -a — 63-1 h « -o S o*

o o

io o V a.

V a.

a,

Does Isomorphism Legitimate? 1996

Deephouse

385 1033

cient for performance was also positive and significant [β = 0.549, ρ < .001). The coefficients for age and size were not significant. Model 2 estimates the absence of regulatory enforcement actions, 1 9 9 1 92, for which there were 194 observations. Strategic conformity had a positive coefficient (/3 = 0.543) that was significant at the ρ < .07 level, providing modest support for Hypothesis 1. Performance again had a significantly positive coefficient (β = 0.735, ρ < .05). Age and size again were not significant. Model 3 of Table 3 estimates the coefficient of media endorsement, 1988-92, for which there were 269 observations. Strategic conformity had a significantly positive coefficient (β = 0.045, ρ < .05), supporting Hypothesis 2. Age had significantly negative coefficient (β = - 0 . 1 4 5 , ρ < .001) as did size 03 = - 0 . 0 3 8 , ρ < .05). Performance had no effect. DISCUSSION AND CONCLUSION The results constitute the first systematic tests of a fundamental linkage in institutional theory. Evidence suggests a positive relationship between strategic isomorphism and multiple measures of legitimacy, even when age, size, and performance are included. The findings support the general proposition made by Meyer and Rowan (1977) and DiMaggio and Powell (1983) stating that organizational isomorphism increases organizational legitimacy. Organizations that conform to the strategies used by other organizations are recognized by regulators and the general public as being more legitimate than those that deviate from normal behavior. This study also demonstrated how organizational legitimacy could be operationally defined using regulators and the media as sources. Researchers have called for more empirical attention to legitimacy for decades (Bozeman, 1993; Galaskiewicz, 1985; Suchman, 1995; Terreberry, 1968). This lack of attention is especially disappointing because legitimacy is an "anchor-point" concept on which many propositions of institutional theory are based (Suchman, 1995: 571). The operational definitions developed here could be applied in other settings, with appropriate contextual modifications. The results also suggest that regulators and the media confer legitimacy in different ways. The correlations between the measures of regulatory endorsement and public endorsement were lower than .34. Moreover, the pattern of the regression coefficients for the independent variables differed. Future research should examine in more depth how regulators and the media confer legitimacy. At a more general level, institutional theorists should refine their propositions involving legitimacy to recognize that there are different types and sources of legitimacy (Galaskiewicz, 1985; Suchman, 1995). The specific results for the independent variables suggest avenues for future research into the other determinants of legitimacy. The differential regression results stemmed from organizational age, size, and performance.3 3 Statistical comparison of coefficients between regulatory endorsement and public endorsement estimations is inappropriate because the analysis techniques differed.

386 1034

D. L. Deephouse Academy

of Management

Journal

August

Age and size are discussed together because of the common patterns in their results. Age and size had no significant effect on regulatory endorsement. These findings suggest that regulators do not consider these factors important in judging bank safety and soundness. Although regulators had slightly more liberal capital standards for large banks in the 1 9 8 5 - 8 8 period, as illustrated in Appendix A, this differential was removed in the 1 9 8 9 - 9 2 period. Age and size may have little impact on evaluations by state regulators in other industries. However, larger and older banks had lower levels of media endorsement. This is somewhat surprising for at least two related reasons. First, larger and older organizations have more social and economic ties to their environment (Pfeffer & Salancik, 1978). Second, they have a longer history of interactions with their environment (Hannan & Freeman, 1984). One explanation is that the public may hold higher standards for the larger banks, because of their greater impact and visibility in the community. The larger banks tended to be the older ones as well, as evident in the 0.29 correlation between age and size in Table 2. An alternative explanation is that larger banks receive more newspaper coverage, and this increases the likelihood that challenging newspaper stories about them will appear. Only 11 percent of the recording units in this sample challenged banks' legitimacy. Banks that have more stories about them would be more likely to have legitimacy-challenging stories, ceteris paribus. Although larger and older banks had lower media endorsement scores, these banks still were endorsed by the public. Their coefficients of media endorsement were greater than zero, meaning they had more endorsing recording units than challenging ones. Overall, a greater understanding of the relationship between the media and business organizations will further theories of public endorsement. Performance had a positive relationship with regulatory endorsement. This is not surprising given regulators' interest in banks' solvency. More profitable banks tend to increase their capital. Moreover, regulatory examinations would find the banks had better-quality assets, so enforcement actions would be less likely. Lower performance did not, however, result in challenges to a bank by the media. A possible explanation is that the media did not attend to differences in bank performance unless a bank was losing money. I examined this possibility post hoc by splitting the sample into banks that were making money (i.e., had a positive ROA) and those that were not. A i-test showed no difference in the coefficient of media endorsement between the two groups (f = 0.51, df = 37, ρ = .62). Clearly, the relationship between performance and public endorsement requires further study. A limitation is that the sample population contained only commercial banks. Nevertheless, the inferences drawn here may apply to other organizations in strong institutional environments, such as hospitals and universities (Scott & Meyer, 1991). Another limitation is that only strategic isomorphism was examined. Future work should examine the effect on legitimacy of

Does Isomorphism Legitimate? 1996

Deephouse

387 1035

other types of isomorphism, such as structural and procedural isomorphism (Scott, 1987). Also, only two social actors conferred legitimacy, regulators and the media. Future work should examine other sources of legitimacy, such as intellectuals and funding agents (Galaskiewicz, 1985). Finally, the causal direction of the independent variables and legitimacy has been assumed to be in the direction predicted by institutional theory. Strong causal inference from these results would be inappropriate, given the cross-sectional design. In sum, this study empirically answers yes to the question raised in the title—Does isomorphism legitimate?—given the aforementioned caveats. This affirmation for isomorphism in strategies extends across two sources of legitimacy, regulators and the media. REFERENCES Abrahamson, E., & Hegeman, R. 1994. Strategic conformity: An institutional theory explanation. Paper presented at the annual meeting of the Academy of Management, Dallas. Aldrich, H. E., & Fiol, C. M. 1994. Fools rush in? The institutional context of industry creation. Academy of Management Journal, 19: 6 4 5 - 6 7 0 . Amemiya, T. 1984. Tobit models: A survey. Journal of Econometrics,

24: 3 - 6 1 .

Ashforth, B. E., & Gibbs, B. W. 1990. The double-edge of organizational legitimation. Organization Science, 1: 1 7 7 - 1 9 4 . Baum, J. A. C., & Oliver, C. 1991. Institutional linkages and organizational mortality. Administrative Science Quarterly, 36: 187-218. Bozeman, B. 1993. Understanding the roots of publicness. In B. Sutton (Ed.), The corporation: 6 3 - 8 1 . Cambridge, MA: Blackwell.

legitimate

Budd, R. W., Thorp, R. K., & Donohew, L. 1967. Content analysis of communications. York: Macmillan.

New

Chandler, A. D. 1962. Strategy and structure. Cambridge: MIT Press. Chen, C. C., & Meindl, ]. R. 1991. The construction of leadership images in the popular press: The case of Donald Burr and People Express. Administrative Science Quarterly, 36: 5 2 1 - 5 5 1 . Coombs, W. T. 1992. The failure of the task force on food assistance: A case study of the role of legitimacy in issue management. Journal of Public Relations Research, 4(2): 1 0 1 - 1 2 2 . Davis, G. F. 1991. Agents without principles? The spread of the poison pill through the intercorporate network. Administrative Science Quarterly, 38: 5 8 3 - 6 1 3 . DeFleur, M. L., Davenport, L., Cronin, M., & DeFleur, M. 1992. Audience recall of news stories presented by newspaper, computer, television, and radio. Journalism Quarterly, 6 9 : 1 0 1 0 1022. DiMaggio, P. J., & Powell, W. W. 1983. The iron cage revisited: Institutional isomorphism and collective rationality in organizational fields. American Sociological Review, 48: 147-160. Dowling, ]., & Pfeffer, J. 1975. Organizational legitimacy: Social values and organizational behavior. Pacific Sociological Review, 18: 122-136.

388 1036

D. L. Deephouse Academy of Management Journal

August

Edelman, L. 1992. Legal ambiguity and symbolic structures: Organizational mediation of civil rights law. American Journal of Sociology, 97: 1531-1576. Elsbach, Κ. D. 1994. Managing organizational legitimacy in the California cattle industry: The construction and effectiveness of verbal accounts. Administrative Science Quarterly, 39: 57-88. Fienberg, S. E. 1980. The analysis of cross-classified categorical data (2nd ed.). Cambridge, MA: MIT Press. Finkelstein, S., & Hambrick, D. C. 1990. Top-management-team tenure and organizational outcomes: The moderating role of managerial discretion. Administrative Science Quarterly, 35: 484-503. Fligstein, N. 1991. The structural transformation of American industry: An institutional account of the causes of diversification in the largest firms, 1919-1979. In W. W. Powell & P. J. DiMaggio (Eds.), The new institutionalism in organizational analysis: 311-336. Chicago: University of Chicago Press. Galaskiewicz, J. 1985. Interorganizational relations. In R. H. Turner & J. F. Short, Jr. (Eds.), Annual review of sociology, vol. 11: 281-304. Palo Alto, CA: Annual Reviews. Galaskiewicz, J., & Wasserman, S. 1989. Mimetic processes within an interorganizational field: An empirical test. Administrative Science Quarterly, 34: 454-479. Gans, H. J. 1979. Deciding what's news. New York: Pantheon. Greene, W. H. 1993. Econometric analysis (2nd ed.). New York: Macmillan. Hannan, M. T., & Freeman, J. 1984. Structural inertia and organizational change. American Sociological Review, 49: 149-164. Haunschild, P. 1993. Interorganizational imitation: The impact of interlocks on corporate acquisition activity. Administrative Science Quarterly, 38: 564-592. Haveman, H. A. 1993. Follow the leader: Mimetic isomorphism and entry into new markets. Administrative Science Quarterly, 38: 593-627. Hirsch, P., & Andrews, J. A. Y. 1984. Administrators' response to performance and value challenges: Stance, symbols, and behavior. In T. ]. Sergiovanni & J. E. Corbally (Eds.), Leadership and organizational culture: 170-185. Urbana: University of Illinois Press. Hurwitz, L., Green, B., & Segal, H. E. 1976. International press reactions to the resignation and pardon of Richard M. Nixon. Comparative Politics, 9: 107-123. Hybels, R., Ryan, Α., & Barley, S. 1994. Alliances, legitimation, and founding rates in the U.S. biotechnology field, 1971-1989. Paper presented at the annual meeting of the Academy of Management, Dallas. Janis, I. L., & Fadner, R. 1965. The coefficient of imbalance. In H. Lasswell, Ν. Leites, & Associates (Eds.), Language of politics: 153-169. Cambridge, MA: MIT Press. Jepperson, R. L. 1991. Institutions, institutional effects, and institutionalism. In W. W. Powell & P. ]. DiMaggio (Eds.), The new institutionalism in organizational analysis: 143-163. Chicago: University of Chicago Press. Krishnaiah, J., Signorielli, N., & McLeod, D. M. 1993. The evil empire revisited: New York Times coverage of the Soviet intervention in and withdrawal from Afghanistan. Journalism Quarterly, 70: 647-655. Maddala, G. S. 1983. Limited-dependent and qualitative variables in econometrics. New York: Cambridge University Press. Meyer, J. W„ & Rowan, B. 1977. Institutional organizations: Formal structure as myth and ceremony. American Journal of Sociology, 83: 340-363.

Does Isomorphism Legitimate? 1996

389 1037

Deephouse

Meyer, J. W., & Scott, W. R. 1983. Centralization and the legitimacy problems of local government. In J. W. Meyer & W. R. Scott (Eds.), Organizational environments: 199-215. Newbury Park, CA: Sage. Mezias, S. J. 1990. An institutional model of organizational reporting practice: Financial reporting at the Fortune 200. Administrative Science Quarterly, 35: 431-457. Palmer, D. P., Jennings, D., & Zhou, X. 1993. Late adoption of the multidivisional form by large U.S. corporations: Institutional, political, and economic activity. Administrative Science Quarterly, 38: 100-131. Pfeffer, J., & Salancik, G. R. 1978. The external Harper & Row.

control of organizations. New York:

Powell, W. W. 1991. Expanding the scope of institutional analysis. In W. W. Powell & P. J. DiMaggio (Eds.), The new institutionalism in organizational analysis: 183-203. Chicago: University of Chicago Press. Reger, R. K., Duhaime, I. M., & Stimpert, J. L. 1992. Deregulation, strategic choice, risk, and financial performance. Strategic Management Journal, 13: 189-204. Riffe, D., Aust, C. F., & Lacy, S. R. 1993. The effectiveness of random, consecutive day, and constructed week sampling in newspaper content analysis. Journalism Quarterly, 70: 133-139. Scott, W. R. 1987. Organizations: Rational, natural, and open systems (2nd ed.). Englewood Cliffs, NJ: Prentice-Hall. Scott, W. R„ & Meyer, J. W. 1991. The organization of societal sectors. In W. W. Powell & P. J. DiMaggio (Eds.), The new institutionalism in organizational analysis: 108-140. Chicago: University of Chicago Press. Singh, J. V., Tucker, D. )., & House, R. J. 1986. Organizational legitimacy and the liability of newness. Administrative Science Quarterly, 31: 171-193. Spong, K. 1985. Banking regulation: Its purposes, implementation, and effects (2nd ed.). Kansas City: Federal Reserve Bank of Kansas City. Spong, K. 1990. Banking regulation: Its purposes, implementation, and effects (3rd ed.). Kansas City: Federal Reserve Bank of Kansas City. Stempel, G. H. III. 1991. Where people really get most of their news. Newspaper Journal, 12(4): 2-9.

Research

Stinchcombe, A. L. 1968. Constructing social theories. New York: Harcourt, Brace, & World. Suchman, M. C. 1995. Managing legitimacy: Strategic and institutional approaches. Academy of Management Review, 20: 571-610. Sudman, S. 1976. Applied sampling. New York: Academic Press. Terreberry, S. 1968. The evolution of organizational environments. Administrative Quarterly, 12: 590-613.

Science

Tolbert, P. S., & Zucker, L. G. 1983. Institutional sources of change in the formal structure of organizations: The diffusion of civil service reform, 1880-1935. Administrative Science Quarterly, 28: 22-39. Weber, R. P. 1990. Basic content analysis (2nd ed.). Newbury Park, CA: Sage. Zucker, L. G. 1987. Institutional theories of organization. In W. R. Scott & J. F. Short, Jr. (Eds.), Annual review of sociology, vol. 13: 443-464. Palo Alto, CA: Annual Reviews.

390

D. L. Deephouse

1038

Academy

of Management

Journal

August

APPENDIX A Regulatory Rules for Classifying Banks' Capital Positions" Regulatory Assessment of Total Capital, 1 9 8 5 - 8 8 (total equity capital + limited life preferred stock + subordinated notes and debentures + minority interests in consolidated subsidiaries + allowance for loan and lease losses) (total assets + allowance for loan and lease losses) Categories 2 = Banks with a total capital ratio greater than 7.0 percent and total average assets less than $1 billion, or banks with a total capital ratio greater than 6.5 percent and total average assets of $1 billion or more. 1 = Banks with a total capital ratio between 7.0 percent and 6.0 percent inclusive and total average assets less than $1 billion, or banks with a total capital ratio between 6.5 percent and 5.5 percent inclusive and total average assets of $1 billion or more. 0 = Banks with a total capital ratio less than 6.0 percent and total average assets less than $1 billion, or banks with a total capital ratio less than 5.5 percent and total average assets of $1 billion or more. Regulatory Assessment of Tier 1 Leverage Capital, 1 9 8 9 - 9 2 Tier 1 leverage capital ratio

=

(total equity goodwill) (total assets - good will ^

Categories 2 = Banks 1 = Banks to 2.0 0 = Banks

with tier 1 leverage capital ratio greater than or equal to 3.0 percent. with tier 1 leverage capital ratio less than 3.0 percent and greater than or equal percent. with tier 1 leverage capital ratio less than 2.0 percent.

' Source is Spong (1985, 1990). b From an accounting perspective, goodwill is the excess cost of an acquired bank over the fair market value of its assets less its liabilities.

APPENDIX Β Classification of Recording Units Legitimacy-Challenging Terms Adjectives: Bad, clubby, confusing, disingenuous, hostile, misguided, reckless, speculative, unpopular. Nouns: Big Cigars, corruption, criticism, disappointment, dyspepsia, empire of Genghis Khan, extremely tight credit leash, failure, gamble, golden parachute, greed, iron triangle, moguls, Pharaohs, ploy, ransoms, trouble, woes. Verbs: Blame, chide, complicate, defended (the action against criticism), failed, justified, loses, obscure, oppose, placate, redlining, sour, tried to explain, unsure of what the bank would do. Activities That Challenge Legitimacy Court action in which the bank is a defendant. Failure of a bank and possible reasons.

Does Isomorphism Legitimate? 1996

391 1039

Deephouse

Public relations activity suggestive of spin doctors used with verbs such as "tried to explain" as listed above. Regulatory action denying a bank regulatory request. Regulatory action penalizing a bank, often a fine or an enforcement action.

APPENDIX C Formula for the Coefficient of Media Endorsement' (e2 — eel Coefficient of media endorsement = i if e > c,

(ec - c 2 ) . , . ρ if c > e. 0 if e = c, where e = number of endorsing recording units in a given year, c = number of challenging recording units in a given year, and t = e + c.

* Sources are Budd, Thorp, and Donohew (1967) and Janis and Fadner (1965). David L. Deephouse is an assistant professor of management at Louisiana State University. He received his Ph.D. degree from the University of Minnesota. His research interests include interorganizational relations, organizational heterogeneity, and organizational legitimacy.

9 Gesamtunternehmungsstrategien

Gesamtunternehmungsstrategien (corporate strategies) determinieren das Geschäftsbereichsportfolio, d.h. die Branchen und Märkte, innerhalb derer eine Unternehmung tätig ist und ihre Wachstumsziele verfolgt. Die Literatur verwendet unterschiedliche Klassifizierungen potentieller Gesamtunternehmungsstrategien. Allerdings werden zwei Arten von Gesamtunternehmungsstrategien besonders hervorgehoben: die Vertikale Integration und die Diversifikation (Wheelen & Hunger, 1995: 150-181; Welge & Al-Laham, 1992: 178-351).

Vertikale Integration (VI) Die Herstellung von Produkten und Dienstleistungen involviert eine Reihe von vertikal miteinander verbundenen Aktivitäten, die sich in ihrer Gesamtheit als eine Wertkette darstellen lassen (Porter, 1985: 33-61). Die Zahl der von einer Unternehmung durchgeführten Wertkettenaktivitäten kennzeichnet ihre Leistungstiefe, beziehungsweise das Ausmaß ihrer vertikalen Integration. VI kann vor- oder nachgelagert sein. Vorgelagerte VI (Rückwärtsintegration/ Backward Vertical Integration) bezeichnet die Durchführung von Wertkettenaktivitäten, die relativ weit vom Endverbraucher entfernt sind (z.B. die Fertigung von Vorprodukten). Bei der nachgelagerten VI (Vorwärtsintegration/ Forward Vertical Integration) werden Aktivitäten integriert, die die Unternehmung näher an die Endverbraucher heranrücken (z.B. die Übernahme der Distribution). Die Make-or-Buy-Entscheidung, d.h. die VI, gilt als das paradigmatische Problem des Transaktionskostenansatzes der Ökonomie (Walker & Weber, 1984: 373; Williamson, 1989:150ff.). Die Entscheidung zur vertikalen Integration kann als eine Entscheidung für eine bestimmte Beherrschungsstruktur zur Durchführung ökonomischer Transaktionen gelten.40 Der Transaktionskostenansatz unterscheidet drei grundsätzliche Beherrschungsstrukturen: den Markt, die Hybridform und die Hierarchie. Für einen großen Teil ökonomisch relevanter Transaktionen verwenden Unternehmungen den Markt. Die Transaktionspartner bleiben bei marktlich koordinierten Transaktionen weitgehend anonym, denn sie verlassen sich auf den Marktpreis als Koordinationsmechanismus. Können Unternehmungen für bestimmte Transaktionen nicht auf einen funktionstüchtigen Markt zurückgreifen, bleibt ihnen als alternative Beherrschungsstruktur die Hierarchie, d.h. die Integration der Transaktionen in die Grenzen der eigenen Unternehmung. An die Stel40

In diesem Kapitel werden nur einige grundlegende, für die vertikale Integrationsentscheidung relevante Überlegungen des Transaktionskostenansatzes behandelt. Im Vordergrund stehen dabei die Ausführungen Oliver Williamsons. Vgl. zu einer ausführlichen Darstellung dieses Ansatzes Williamson (1975; 1985) und im deutschsprachigen Raum Picot (1982; 1991).

394

R. Κ. F. Bresser

le des Marktpreises als Koordinationsmechanismus tritt nun die Anweisung innerhalb der Unternehmung. Falls die Integration einer Transaktion in die bestehende Unternehmungshierarchie dazu führt, daß die Unternehmung durch eine höhere Leistungstiefe gekennzeichnet ist und sich damit die Zahl der Wertkettenstufen erhöht, in der sie tätig ist, entspricht die Wahl der Beherrschungsstruktur „Hierarchie" der VI. Die sich zwischen Markt und Hierarchie befindende Hybridform ist ebenfalls eine Alternative zur Markttransaktion; sie umfaßt eine ganze Reihe von Kooperationsarrangements, z.B. Lizenzvereinbarungen oder Joint-ventures.41 Die Hybridform als Beherrschungsstruktur unterscheidet sich insofern von Markt und Hierarchie, als sie sich sowohl auf marktliche als auch auf hierarchische Koordinationsmechanismen verläßt. Von beiden Koordinationsmechanismen wird allerdings in einem geringeren Umfang Gebrauch gemacht, als dies bei den reinen Formen von Markt und Hierarchie der Fall ist (Williamson, 1991a). Eine bedeutende Antriebskraft für die Wahl der VI bzw. der Hierarchie als Beherrschungsstruktur ist die Gefahr des Opportunismus. Unter Opportunismus subsumiert Williamson (1975: 26-28; 1985: 47-52) verschiedene Formen des eigennützigen Verhaltens des einen Transaktionspartners zu Lasten des(r) anderen. Ein opportunistischer Transaktionspartner versucht die (oft nur vermeintlichen) Schwächen anderer auszubeuten, z.B. indem er Leistungen von geringerem Wert, als vertraglich vereinbart, erbringt. In vielen Fällen kann opportunistisches Verhalten von Transaktionspartnern nur schwer überprüft werden, da es ihnen an Erfahrungen und Informationen mangelt (Beispiel: Die Überprüfung der Qualität einer Pkw-Reparatur durch den Inhaber des Pkws).42 Für ökonomische Akteure, die sich darum bemühen, rational zu handeln, besteht ein wichtiges Kriterium für die Wahl einer Beherrschungsstruktur zur Durchführung bestimmter Transaktionen offensichtlich darin, Strukturen zu wählen, die die Gefahren opportunistischen Verhaltens minimieren. Der Transaktionskostenansatz geht davon aus, daß die Hierarchie in stärkerem Maße opportunismusreduzierend wirkt als die Markttransaktion, denn durch die Hierarchie gewinnt die integrierende Unternehmung ein hohes Maß an Kontroll- und Sanktionsmöglichkeiten (Barney & Ouchi, 1986: 23; Barney, 1997: 321).43 Dieser Umstand bedeutet aber nicht, daß die Hierarchie immer die präferierte Beherrschungsstruktur darstellt, denn jede Beherrschungsstruktur verursacht Transaktionskosten, die für unterschiedliche Transaktionen sehr stark variieren können. Marktliche Transaktionskosten umfassen z.B. die Kosten des Suchens geeigneter Transaktionspartner, die Kosten der Verhandlung und der Ausgestaltung von Verträgen, die Kosten der Überwachung hinsichtlich der Einhaltung von Vertragsbestimmungen, Nachverhandlungskosten sowie die Kosten der Konfliktlösung (etwa im Falle einer gerichtlichen Aus41 42

43

Vgl. hierzu die Ausführungen in Kapitel 11. Die durch asymmetrische Informationsverteilungen bedingten und in der ökonomischen Literatur am häufigsten diskutierten Ausprägungen opportunistischen Verhaltens sind „Cheating", „Shirking" und „Hold-up". Vgl. hierzu z.B. Barney und Ouchi (1986: 18-26) und Vogt (1996: 36-44). Die Annahme, opportunistisches Verhalten könne durch eine Hierarchisierung effektiv beherrscht werden, ist aus verhaltenswissenschaftlicher Sicht durchaus fragwürdig. Sowohl in der psychologischen als auch in der soziologischen Literatur existiert erhebliche Evidenz über Dysfunktionen der Bürokratisierung. Bürokratische Kontrollverfahren wirken oft demotivierend und entfremdend auf Organisationsmitglieder, und sie rufen in der Folge opportunistisches Verhalten hervor. An die Stelle eines exogenen, marktbedingten Opportunismus, der sich durch eine Hierarchisierung vermeiden läßt, kann leicht ein endogener, „hausgemachter" Opportunismus treten, wenn vertikale Integrationsmaßnahmen zu hochentwickelten Kontrollsytemen führen. Vgl. hierzu die Literaturübersichten bei Ghoshal & Moran (1996: 20-25) und bei Masuch (1985: 18).

Gesamtunternehmungsstrategien

395

einandersetzung). Hierarchische Transaktionskosten umfassen zum einen ähnliche Kostenkategorien wie die marktlichen Transaktionskosten.44 Zum anderen entstehen in der Hierarchie zusätzliche Bürokratiekosten, da die Transaktionen intern, d.h. innerhalb einer Unternehmung durchgeführt werden müssen. Zu diesen Bürokratiekosten zählen z.B. Koordinations- und Kommunikationskosten, Kosten für umfassende Anreiz- und Kontrollsysteme sowie die Kosten einer reduzierten strategischen Flexibilität45 (Mahoney, 1992; D'Aveni & Ravenscraft, 1994). Falls die VI die Entstehung eines hochentwickelten Kontrollapparates innerhalb der Unternehmung auslöst, kann die Beherrschungsstruktur Hierarchie erheblich höhere Kosten verursachen als vergleichbare Marktlösungen. Sind die antizipierten Transaktionskosten des (regelmäßigen) Marktbezuges von Gütern oder Dienstleistungen höher als die antizipierten Transaktionskosten der Hierarchie, sollten Unternehmungen - den normativen Ansprüchen des Transaktionskostenansatzes gemäß - die jeweiligen Transaktionen durch vertikale Integrationsmaßnahmen internalisieren. Die Höhe der mit alternativen Beherrschungsstrukturen verbundenen Transaktionskosten ist maßgeblich von den Maßnahmen abhängig, die getroffen werden müssen, um der Gefahr der opportunistischen Ausbeutung zu begegnen. Es stellt sich somit die Frage nach den wichtigsten Bestimmungsfaktoren dieser Bedrohung. Der Transaktionskostenansatz Williamsonscher Prägung geht davon aus, daß das Ausmaß einer Bedrohung durch Opportunismus insbesondere von zwei Determinanten abhängt: von der Höhe transaktionsspezifischer Investitionen und vom Ausmaß der Unsicherheit und Komplexität von Transaktionen (Williamson, 1991c; Barney, 1997: 322). Eine Investition ist transaktionsspezifischer Natur, wenn ihr Wert in einer ganz bestimmten Transaktionsbeziehung wesentlich höher ist als in allen alternativen Transaktionsbeziehungen (Williamson, 1975: 26-30; 1985: 52-56). Wenn es für die Aufnahme einer Geschäftsbeziehung notwendig ist, daß ein Transaktionspartner spezifische Investitionen tätigt (z.B. eine Spezialmaschine entwickelt, für die es keine andere Verwendung gibt), wächst die Gefahr der opportunistischen Ausbeutung durch den anderen Transaktionspartner stark an, falls dieser nicht ebenfalls Investitionen in spezifische Produktionsfaktoren vornehmen muß. Der Transaktionskostenansatz empfiehlt deshalb, Transaktionen, die von hohen spezifischen Investitionen abhängig sind, zu integrieren. Unsichere und komplexe Transaktionen erhöhen ebenfalls die Bedrohung durch opportunistisches Verhalten. Bei unsicheren und komplexen Transaktionen ist es nicht möglich, abzuschätzen, wie sich eine Transaktionsbeziehung (z.B. mit einem Lieferanten) in der Zukunft entwickeln wird. Insbesondere können potentielle Formen opportunistischen Verhaltens nicht hinreichend antizipiert werden, so daß sich ein alle wichtigen Eventualitäten abdeckender Vertrag weder entwickeln noch durchsetzen läßt (Williamson, 1991c: 79). Dieser Zustand ist gleichzusetzen mit einer erhöhten potentiellen Bedrohung durch opportunistische Transaktionspartner. Der Transaktionskostenansatz empfiehlt deshalb, sehr unsichere und komplexe Transaktionen ebenfalls zu integrieren, d.h. hierarchisch zu koordinieren. 44

45

Allerdings sind die markttransaktionstypischen Kosten in der Hierarchie oft geringer. Es ist z.B. in der Regel billiger, die Einhaltung bestimmter Fristen und Leistungen innerhalb der Hierarchie als am Markt bei verschiedenen unabhängigen Lieferanten zu überwachen. Die strategische Flexibilität einer vertikal integrierten Unternehmung leidet in der Regel, da die VI oft mit dem Aufbau spezifischer, langlebiger Aktiva verbunden ist. Die Aufgabe dieser Aktiva bei sich verändernden Umweltbedingungen ist kostspielig und kann deshalb notwendige Anpassungsmaßnahmen verzögern (Teece, 1992; Richardson, 1996).

396

R. Κ. F. Bresser

Die empirische Evidenz für die grundlegenden Aussagen des Transaktionskostenansatzes zur vertikalen Integrationsentscheidung ist beeindruckend. Zahlreiche Studien konnten positive Zusammenhänge zwischen verschiedenen Formen transaktionsspezifischer Investitionen und dem Ausmaß der VI nachweisen (z.B. Joskow, 1985; MacMillan et al., 1986; Caves & Bradburd, 1988; Masten et al., 1991). Dies gilt ebenfalls für die Zusammenhänge zwischen dem Ausmaß der Unsicherheit und Komplexität von Transaktionen und der VI, obwohl die Evidenz in diesem Fall nicht so eindeutig ist wie die Forschungsergebnisse hinsichtlich transaktionsspezifischer Investitionen (Joskow, 1988; Mahoney, 1992: 572f.). 4i Trotz der weitgehenden Bestätigung des Transaktionskostenansatzes hinsichtlich der vertikalen Integrationsentscheidung besteht die Möglichkeit, den Ansatz für die Entwicklung einer Theorie des Strategischen Managements zu erweitern. Wichtige Impulse für eine solche Erweiterung können dem RBA entnommen werden. Folgt man dem traditionellen Transaktionskostenansatz, sollte die VI nur durchgeführt werden, wenn der Wert der mit der Hierarchisierung verbundenen zusätzlichen Kontrolle die Kosten der Hierarchisierung überschreitet. Diese Situation ist typischerweise dann gegeben, wenn die Opportunismusgefahr hoch ist, bei Transaktionen also, die hohe spezifische Investitionen erfordern und die sich durch Unsicherheit und Komplexität auszeichnen. Genau diese Transaktionen sind aber gemäß dem RBA besonders wertvoll, denn es ist anzunehmen, daß durch eine Integration dieser Transaktionen seltene, nicht substituierbare und nur schwer imitierbare Ressourcen und Fähigkeiten erworben werden können (Reve, 1990; Langlois, 1992; Teece et al., 1994: 26; Barney, 1997: 337). Ein Problem des traditionellen Transaktionskostenansatzes besteht darin, daß er annimmt, der ökonomische Wert einer Transaktion sei konstant und unabhängig von der gewählten Beherrschungsstruktur: Das Entscheidungsproblem für einen Manager bestehe deshalb lediglich darin, eine Beherrschungsstruktur auszuwählen, die die Bedrohung durch Opportunismus bändigt und zwar zu möglichst niedrigen Kosten (Williamson, 1991c). Übersehen wird dabei, daß die Beherrschungsstruktur selbst eine wertvolle Quelle dauerhafter Wettbewerbsvorteile sein kann. Um die potentiellen Wettbewerbsvorteile, die sich aus unterschiedlichen Beherrschungsstrukturen ableiten lassen, bewerten zu können, muß in das Modellgebäude des Transaktionskostenansatzes mehr Heterogenität eingeführt werden. Gemäß dem RBA ist ja gerade die Heterogenität der Ressourcen und Fähigkeiten dafür verantwortlich, daß Unternehmungen sich unterschiedliche Wettbewerbsvorteile aufbauen können (Barney, 1997: 339). Mehr Heterogenität (oder Variabilität) im Transaktionskostenansatz ist in zweifacher Hinsicht vonnöten: bezüglich der Produktionstechnologien und hinsichtlich der Oppor-

46

Vergleicht man die empirischen Forschungsergebnisse transaktionskostentheoretisch inspirierter Studien zur VI mit jenen, die auf der Basis der traditionellen Industrieökonomik durchgeführt wurden, so ergibt sich eine eindeutige Überlegenheit des Transaktionskostenansatzes. Die Industrieökonomik hat versucht, das Phänomen der VI aus einer Marktmachtperspektive zu erklären. So wurde z.B. argumentiert, die VI bewirke eine Erhöhung der Markteintrittsbarrieren sowie der Marktaustrittsbarrieren. Außerdem sei die VI dazu geeignet, Kollusionskosten zu senken und Preisdiskriminierungen zu ermöglichen. Die empirische Bestätigung dieser Marktmachthypothesen ist wesentlich schwächer als die Evidenz für die zentralen Erklärungshypothesen des Transaktionskostenansatzes (vgl. Picot & Franck, 1993). Allerdings soll nicht übersehen werden, daß die Kausalitäten in den Studien des Transaktionskostenansatzes aufgrund der gewählten Forschungsdesigns oft nicht eindeutig sind und alternative Erklärungen zulassen (Masten et al., 1991; Moran & Ghoshal, 1996: 66-69).

Gesamtunternehmungsstrategien

397

tunismusannahme. Die Forderang, anzuerkennen, daß Unternehmungen unterschiedliche Fähigkeiten haben können, Werte aus bestimmten Beherrschungsstrukturen abzuleiten, heißt anzuerkennen, daß die Produktionstechnologien der Unternehmungen heterogen sind (Chung, 1997: 114-131).47 Heterogenität hinsichtlich der verwendeten Produktionstechnologien48 kann in den Transaktionskostenansatz eingeführt werden, indem man z.B. berücksichtigt, daß ökonomische Akteure über unterschiedliche Fähigkeiten verfügen, um (1) unsichere und komplexe Transaktionen zu analysieren und (2) bestimmte Beherrschungsstrukturen effizient auszubauen (Barney, 1997: 339-342). Diese heterogenen Analyse- und Organisationsfähigkeiten sind potentielle Quellen für Wettbewerbsvorteile, und sie bereichern die Diskussion über die Vorteilhaftigkeit vertikaler Integrationsentscheidungen. Darüber hinaus ist es ebenso sinnvoll, die Opportunismusneigung nicht als invariante Verhaltensannahme zu modellieren, die für alle Menschen in gleicher Weise Gültigkeit hat (Williamson, 1985: 64). Das Ausmaß der Opportunismusneigung einzelner Akteure ist variabel,49 und dieser Umstand kann wiederum die Wahl verschiedener Beherrschungsstrukturen unterschiedlich wertvoll machen (Barney, 1997: 342f.). Wenn es bestimmten Unternehmungen gelingt, mit Transaktionspartnern zu kontrahieren, die sich durch eine geringe Opportunismusneigung auszeichnen, dann können diese Unternehmungen in vergleichsweise kostengünstige Beherrschungsstrukturen investieren. Diese Selektionsfähigkeit läßt den fraglichen Unternehmungen Wettbewerbsvorteile gegenüber jenen Konkurrenten zukommen, die nicht in der Lage sind, Transaktionspartner mit geringer Opportunismusneigung und kostengünstige Beherrschungsstrukturen auszuwählen. Wenn man die Forderung, mehr Heterogenität in das Prämissengebäude des Transaktionskostenansatzes einzuführen, ernst nimmt, dann kommt man in letzter Konsequenz nicht daran vorbei, ein dynamischeres Unternehmerbild zu verwenden (Chung, 1997: 109-114). Es sind nach Wettbewerbsvorteilen strebende Unternehmer und TopManagement-Teams, die durch Prozesse der schöpferischen Zerstörung Innovationen und Marktungleichgewichte erzeugen (Schumpeter, 1926; 1946) und sich auf diese Weise, wenn keine dauerhaften, so doch temporäre Wettbewerbsvorteile verschaffen. Damit ist unsere Diskussion wieder bei der „Österreichischen Schule" angelangt (Jacobson, 1992). Die Forderung nach mehr Heterogenität ist nicht nur durch den RBA begründbar, sie ist auch aus verhaltenswissenschaftlicher Sicht notwendig. Durch eine Dynamisierung der Annahmen zu den Konzepten Produktionstechnologie und Opportunismus wäre der Transaktionskostenansatz in der Lage, ein realistischeres Modell der relevanten Entscheidungssituation vor der Wahl einer angemessenen Beherrschungsstruktur zu erstellen.

47

48

49

Der Transaktionskostenansatz unterstellt, daß die Produktionstechnologien von der Wahl der Beherrschungsstrukur unbeeinflußt bleiben (Williamson, 1981; Picot, 1982: 271; Walker & Weber, 1984: 377). Der Begriff Produktionstechnologie wird hier in einem sehr weiten Sinne verstanden. Er umfaßt nicht nur physische Aggregate, sondern das gesamte, einer Unternehmung zur Verfügung stehende Wissen, also auch sämtliche Fähigkeiten, die zur Erfüllung der Unternehmungsaufgaben eingesetzt werden (Perrow, 1970: 79-91). Die Opportunismusneigung einzelner Akteure ist genauso variabel wie das Ausmaß tatsächlich beobachtbaren opportunistischen Verhaltens. Dieser Tatbestand wird im traditionellen Transaktionskostenansatz nicht deutlich herausgearbeitet (Hill, 1990; Zajac & Olsen, 1993; Ghoshal & Moran, 1996).

398

R. Κ. F. Bresser

Diversifikationsstrategien Gemessen an der Häufigkeit einschlägiger Berichte und Analysen in der Wirtschaftspresse, ist die Diversifikation die populärste Wachstumsstrategie der Unternehmungspraxis. Unternehmungen, die verschiedenartige Produktprogramme oder Dienstleistungen in multiplen Märkten offerieren, verfolgen eine Diversifikationsstrategie (Ansoff, 1965: 108f.). Diversifizierte Unternehmungen unterscheiden sich in dem Ausmaß oder dem Grad, in dem sie diese Strategie verfolgen. Die Literatur differenziert zwischen zwei grundlegenden Formen der Diversifikation, die als Indikatoren für den Diversifikationsgrad herangezogen werden können: verbundene (related, mediale, concentric) und unverbundene (unrelated, laterale, conglomerate) Diversifikation (Rumelt, 1974). Bei der verbundenen Diversifikation werden komplementäre Produkte oder Dienstleistungen in das Leistungsspektrum der Unternehmung aufgenommen. Alle Geschäftszweige der Unternehmung haben einen inneren Zusammenhang. Diese Komplementarität kann z.B. auf ähnlicher Technologie, ähnlichem Konsumentenverhalten, ähnlichem Vertrieb oder ähnlichem Know-how beruhen. Bei der unverbundenen Diversifikation wird das Prinzip der Komplementarität aufgegeben; die Geschäftszweige der Unternehmung befinden sich in sehr unterschiedlichen Wirtschaftssektoren und weisen keinen inneren Verbund auf. Das Ausmaß der Diversifikation verschiedener Unternehmungen kann sowohl im Hinblick auf die jeweilige Intensität der verbundenen oder der unverbundenen Diversifikation gemessen werden als auch im direkten Vergleich der beiden Arten. Die unverbundene Diversifikation gilt als die umfassendere Art, da bei ihr kein innerer Zusammenhang besteht. Um ökonomisch sinnvoll zu sein, müssen (verbundene oder unverbundene) Diversifikationsstrategien einen ökonomischen Wert generieren.50 Diversifikation führt zu ökonomischem Wert, wenn zwei Bedingungen erfüllt sind (Barney, 1997: 359): erstens, die Diversifikation muß das Ausschöpfen von Synergievoñeilen ermöglichen. Finanzwirtschaftlich ausgedrückt liegt ein diversifikationsbedingter Synergiewert vor, wenn der Nettokapitalwert des Gesamtkonzerns höher ist als die Summe der Nettokapitalwerte der einzelnen Geschäftsbereiche. Kann die Unternehmung durch ihre Diversifikationsstrategie einen Synergiewert erzielen, dann liegt dies daran, daß sie „ E c o n o m i e s of Scope" ausschöpft. Durch die diversifikationsbedingte breite Anwendung von physischen, immateriellen und finanziellen Ressourcen entstehen Kostenvorteile oder Erlössteigerungen, die als Economies of Scope bezeichnet werden (Teece, 1982: 40). Zweitens, für das Entstehen eines ökonomischen Wertes ist das Vorliegen von Synergievorteilen alleine nicht ausreichend. Der diversifizierende Konzern muß diese Synergievorteile auch kostengünstiger realisieren können als Anteilseigner (Eigenkapitalgeber). Wenn Anteilseigner Diversifikationsvorteile günstiger realisieren können als eine Unternehmung, ist es offensichtlich wenig sinnvoll, wenn sie die Manager einer Unternehmung damit „beauftragen", Diversifikationsstrategien durchzuführen. Die Diversifikationsforschung hat sich am ausführlichsten mit dem Verhältnis zwischen Diversifikationsausmaß und Erfolg auseinandergesetzt (Ramanujam & Varadarajan, 1989). Dieser Schwerpunkt ist auch leicht verständlich, denn es besteht ein erhebli50

Den ökonomischen Wert der Diversifikation kann man z.B. als zusätzlichen „shareholder value" definieren, d.h. als die Summe der diskontierten Aktienkurssteigerungen und Dividendenerhöhungen, die den Anteilseignern aufgrund der Diversifikationsstrategie zufallen.

Gesamtunternehmungsstrategien

399

ches pragmatisches Interesse daran, festzustellen, ob die verbundene oder die unverbundene Diversifikation rentabler ist und unter welchen Bedingungen. Zahlreiche empirische Studien, die über drei Jahrzehnte hinweg durchgeführt wurden, unterstützen die These, daß die verbundene Diversifikation die rentablere, wertvollere Alternative ist (vgl. z.B. Rumelt, 1974; 1982; Varadarajan & Ramanujam, 1987; Bühner, 1990; Robins & Wiersema, 1995).51 Dennoch, die Kontroverse über den Wert der unverbundenen Diversifikation ist nicht verstummt, und sie wirft die Frage nach einer theoretischen Basis für die Vorteilhaftigkeit alternativer Diversifikationsstrategien auf. Auf der Suche nach theoretischen Erklärungen der Diversifikation stößt man auf ähnliche Ansätze wie im Fall der VI: die traditionelle Industrieökonomik, den Transaktionskostenansatz und den RBA. Darüber hinaus existieren finanzierungstheoretische Begründungen. Was die traditionelle Industrieökonomik betrifft, so sind die Ergebnisse recht unbefriedigend geblieben. Dieser Ansatz beschäftigt sich vornehmlich mit der unverbundenen Diversifikation und verwendet das Streben nach Marktmacht als zentrale Erklärungshypothese. Die empirischen Ergebnisse hinsichtlich der Fragen, ob und inwieweit Diversifikation antikompetitive Effekte zeitigt, sind (wie auch bei der VI) sehr widersprüchlich (Ramanujam & Varadarajan, 1989: 536f.). Die für das Strategische Management wichtigsten Erklärungsansätze sind der Transaktionskostenansatz und der RBA. Interessant ist, daß beide Ansätze im Rahmen der Diversifikationsforschung wesentlich enger miteinander verzahnt sind als dies in anderen Bereichen der Fall ist, z.B. bei der VI.52 Der Kern der transaktionskostentheoretischen Argumentation ist folgender: Unternehmungen diversifizieren, da sie über ungenutzte Ressourcenkapazitäten verfügen, die vielfältig eingesetzt werden können aber auf Marktversagen treffen (Teece, 1982). Da viele Ressourcen hochspezifisch sind, entstehen Transaktionsprobleme am Markt, ein Umstand, der zu einer internen Nutzung der ungenutzten Ressourcen qua Diversifikation führt. Die wichtigste Ergänzung dieser Argumentationslinie aus Sicht des RBA ist die Feststellung, daß das in einer Unternehmung vorhandene Ressourcenprofil ausschlaggebend ist für den Erfolg potentieller Diversifikationsstrategien (Chatterjee & Wernerfeit, 1991). Je flexibler (unspezifischer) die vorhandenen Ressourcen, desto höher kann das Ausmaß der verfolgten Diversifikationsstrategie sein. Die optimale Form der Diversifikation läßt sich für drei Ressourcenklassen bestimmen: physische, intangible (immaterielle) und finanzielle Ressourcen (Teece, 1982; Chatterjee & Wernerfeit, 1991). Durch jede der drei Ressourcenklassen können Economies of Scope verwirklicht werden, aber ihre jeweilige Eignung für verschiedene Diversifikationsarten variiert (Barney, 1997: 360-379). Physische Ressourcen ermöglichen eine Diversifikation durch die gemeinsame Durchführung von Aktivitäten der Wertkette (Porter, 1987). Allerdings sind physische Ressourcen recht inflexibel; Produktionsmaschinen sind z.B. oft nur für die Produktion bestimmter Produktarten geeignet. Aus diesem Grunde sind ungenutzte physische Ressourcen am besten in benachbarten Märkten einsetzbar, d.h. für die verbundene Diversifikation geeignet. Intangible Ressourcen sind auf Know-how

51

52

Aus einer von Barney (1997: 388f.) erstellten Übersicht ergibt sich, daß 19 Studien eine Überlegenheit der verbundenen Diversifikation ermitteln, drei Studien können keine signifikanten Unterschiede zwischen den beiden Diversifikationsarten feststellen, und nur eine Studie ermittelt höhere Erfolgsraten für die unverbundene Diversifikation. Vgl. zu dieser Einschätzung auch Peteraf (1993: 188) und Wernerfeit (1995: 172).

400

R. Κ. F. Bresser

und Erfahrung beruhende Fähigkeiten, und sie werden im RBA als die wahren Kernkompetenzen der Unternehmung angesehen (Prahalad & Hamel, 1990). Honda hat z.B. besondere Fähigkeiten in der Konstruktion effizienter Motoren entwickelt und dazu verwendet, nicht nur Motorräder, sondern auch Automobile, Generatoren und andere motorgetriebene Maschinen zu vertreiben. Da intangible Ressourcen Fähigkeiten sind, die den bisherigen Kernbereichen einer Unternehmung entstammen, und diese Fähigkeiten auch oft nicht exakt beschrieben werden können,53 sind auch sie recht inflexibel und am besten zur verbundenen Diversifikation geeignet. Finanzielle Ressourcen (z.B. verfügbare Liquidität oder unausgenutzte Kreditlinien) weisen ein hohes Maß an Flexibilität auf, und sie können sehr vielfältigen Verwendungen zugeführt werden. Von allen Ressourcenklassen sind sie am ehesten dazu geeignet, unverbundene Diversifikationsstrategien zu ermöglichen (Chatteijee & Wernerfeit, 1991). In vielen Fällen ist es allerdings fraglich, ob der Einsatz finanzieller Ressourcen zu Zwecken der Diversifikation einen ökonomischen Wert entstehen läßt, der nicht kostengünstiger durch Anteilseigner realisiert werden könnte. An dieser Stelle werden zwei finanzierungstheoretische Erklärungsansätze bedeutsam. Der erste Ansatz geht von einem möglichen Allokationsvorteil der Unternehmung gegenüber den Kapitalmärkten aus (Williamson, 1975: 162; Stiglitz, 1981). Es wird argumentiert, daß die Manager einer Unternehmung über bessere, detailliertere Informationen hinsichtlich lohnender Investitionsmöglichkeiten verfügen können als externe Quellen von Kapital (z.B. Banken oder Anteilseigner). In diesen Fällen kann die Verwendung überschüssiger Liquidität zur unverbundenen (und auch verbundenen) Diversifikation durchaus zu einer Ausnutzung finanzieller Economies of Scope führen, die einen ökonomischen Wert entstehen läßt, der von den Anteilseignern nicht kostengünstiger herbeigeführt werden kann (Barney, 1997: 373-377).54 Eine zweite finanzierungstheoretische Überlegung leitet sich aus der Portfoliotheorie ab und empfiehlt die unverbundene Diversifikation als Mittel zur Reduzierung des unsystematischen (firmenbezogenen) Risikos der Unternehmung (Smith & Schreiner, 1969; Joehnk & Nielsen, 1974). Zwar können Unternehmungen ihr unsystematisches Risiko reduzieren, indem sie ihre überschüssigen finanziellen Ressourcen für Diversifikationsstrategien verwenden, diese Risikoreduktion ist allerdings für Anteilseigner nicht wertvoll: Anteilseigner können durch den Kauf von Aktien oder Fondsanteilen Risiko wesentlich kostengünstiger reduzieren als eine diversifizierende Unternehmung (Jensen, 1986; Chang & Thomas, 1989). Empirische Studien bestätigen, daß Risikoreduktionsmotive fragwürdige Begründungen für Diversifikationsstrategien liefern; solche Motive zur 53

54

Die mangelnde Fähigkeit, intangible Ressourcen exakt zu beschreiben, rührt daher, daß Know-how oft den Charakter stillschweigenden Wissens (tacit knowledge) hat (Teece, 1982). Eine über besondere, erfahrungsbedingte Fähigkeiten verfügende Person kann diese Fähigkeiten zwar erfolgreich einsetzen, sie weiß aber oft nicht, warum die eigenen Aktivitäten zum Erfolg führen - Beispiel: Fahrradfahren (Polanyi, 1958: 49f.). Erwähnenswert sind an dieser Stelle die von der Consultingpraxis entwickelten, auf einfachen Matrizen beruhenden Portfolioplanungsmethoden. Diese sind auch heute noch in der Praxis weit verbreitet. Mitunter wird in der Literatur der mißverständliche Eindruck vermittelt, die Portfolioplanungsansätze könnten eine theoretische Begründung für den Erfolg verschiedenartiger Diversifikationsstrategien liefern (Haspeslagh, 1982; De Wit & Meyer, 1994: 263ff.). Dies ist nicht der Fall. Der Wert der Portfolioplanung besteht darin, dem diversifizierten Konzern ein einfaches, kommunikationsfreudiges Instrument zu liefern, das es erleichtert, strategische Analysen durchzuführen und strategische Entscheidungen vorzubereiten (Hax & Majluf, 1996: 301-313).

Gesamtunternehmungsstrategien

401

Diversifikation sind i.d.R. mit unterdurchschnittlichen Rentabilitäten der Diversifizierer verbunden (Amit & Livnat, 1988; Hill & Hansen, 1991). Zusammenfassend läßt sich über die theoretische Basis der Diversifikation konstatieren, daß die schlüssigsten Begründungen durch den RBA in Verbindung mit dem Transaktionskostenansatz und einigen finanzierungstheoretischen Erwägungen ermöglicht werden. Bisher haben wir nur vor dem Hintergrund einer Bedingung für das Entstehen dauerhafter Wettbewerbsvorteile argumentiert, dem wertstiftenden Charakter von Ressourcen. Es sei daran erinnert, daß aus dem Blickwinkel des RBA die zur Diversifikation verwendeten Ressourcen noch weitere Eigenschaften aufweisen sollen, um Wettbewerbsvorteile abzusichern: Sie sollen auch knapp, nicht substituierbar und nicht imitierbar sein (vgl. Kapitel 7). Diese weiteren Eigenschaften wurden bisher nur unzureichend in der Diversifikationsforschung berücksichtigt (Markides & Williamson, 1994). Allerdings ist die empirische Evidenz dafür, daß unterschiedliche Ressourcenprofile wichtige Determinanten für die Wahl angemessener Diversifikationsstrategien sind, inzwischen umfangreich und wachsend (vgl. z.B. Montgomery & Wernerfeit, 1988; Montgomery & Hariharan, 1991; Chatteqee & Wernerfeit; 1991; Farjoun, 1994; Robins & Wiersema, 1995). Für die zukünftige Diversifikationsforschung erscheinen drei Hinweise geboten: (1) Kontingenztheorie der Diversifikation. Bisherige Studien haben sich recht einseitig auf die Zusammenhänge zwischen Ressourcenprofilen, Diversifikationsarten und Diversifikationserfolg konzentriert. Vernachlässigt wurde der Umstand, daß die Beziehung zwischen der Diversifikationsalt und dem Diversifikationserfolg durch moderierende Variablen beeinflußt wird, z.B. durch Gegebenheiten der Branchenstruktur55 und der Organisation56 der diversifizierenden Unternehmung (Hoskisson, 1987; Ramanujam & Varadarajan, 1989: 537-545; Datta et al., 1991). In dem Maße, in dem moderierende Variablen Berücksichtigung finden, kann sich die theoretische Grundlage der Diversifikation zu einer Kontingenztheorie weiterentwickeln, die die Erfolgsdeterminanten der Diversifikation umfassender abbildet. (2) Dynamische Implikationen. Diese wurden ebenfalls weitgehend von der Forschung vernachlässigt. Zu klären wäre z.B., wie der Erfolg eingeschlagener Diversifikationsstrategien in hyperkompetitiven Umwelten (D'Aveni, 1994) aufrechterhalten werden kann. Aus verhaltenswissenschaftlicher Sicht interessiert insbesondere das zur Auswahl bestimmter Diversifikationsstrategien führende Entscheidungsverhalten von Top-Management-Teams (siehe Kapitel 4). Dieses ist nicht immer von ökonomischem Sachverstand geleitet, sondern kann Integrations- oder Diversifikationsentscheidungen z.B. aus Gründen der Machtausweitung herbeiführen (Barney, 1997: 398f.).57 (3) Methodische Herausforderungen. Ein großer Teil der Diversifikationsforschung basiert auf Querschnittsuntersuchungen sowie Korrelations- und Regressionsanalysen.

55

56

57

Ein wichtiges Branchenmerkmal, das sich moderierend auf den Diversifikationserfolg auswirken kann, ist z.B. das Branchenrisiko (Bettis & Mahajan, 1985; Wernerfeit & Montgomery, 1986). Der Erfolg der Diversifikation hängt stark von der Fähigkeit der Unternehmung ab, die Strategie effektiv zu implementieren. Insbesondere bei extern (qua Akquisition) durchgeführten Diversifikationen gehen entscheidende Einflüsse von der Fähigkeit der Unternehmung aus, verschiedenartige Organisationsstrukturen und -kulturen zu konsolidieren und zu harmonisieren. Vgl. zu dieser Implementierungsproblematik Haspeslagh und Jemison (1991) und Barney (1997: 396-475). Vgl. zu dieser Agency-Problematik auch die Ausführungen des Kapitels 3.

402

R. Κ. F. Bresser

Inwieweit die auf diese Weise generierten Ergebnisse eine längerfristige Stabilität aufweisen, läßt sich nur durch Längsschnittsstudien und durch detaillierte, langfristig konzipierte Fallstudien untersuchen (vgl. z.B. Hill, 1983; Ramanujam & Varadarajan, 1989: 540f.). Darüber hinaus gibt es nicht unerhebliche Meßprobleme in der empirischen Forschung. Diese Probleme betreffen die Messung der Diversifikationsarten, die Performance-Messung und auch die Messung von Ressourcenkategorien (vgl. z.B. Ramanujam & Varadarajan, 1989: 538ff.; Montgomery & Wernerfeit, 1988; Henderson & Cockburn, 1994; Robins & Wiersema, 1997). Für Fortschritte in der Diversifikationsforschung ist es erforderlich, daß die „Scientific Community" sich den methodischen Herausforderungen stellt.

Die Artikel Der erste der beiden folgenden Beiträge, „Cooperation, Opportunism, and the Invisible Hand: Implications for Transaction Cost Theory", von Charles Hill betrifft die transaktionskostentheoretische Begründung der VI-Entscheidung. Hill setzt sich kritisch mit der Opportunismusannahme Williamsons auseinander. Unter Rückgriff auf spieltheoretische und evolutionsökonomische Ansätze argumentiert er, daß die von Williamson unterstellte Gefahr der opportunistischen Ausbeutung, und damit das Hauptargument für die VI, übertrieben sei. Die unsichtbare Hand des Marktes würde im Zeitablauf opportunistische Akteure eliminieren, da letztere nicht nur in einem Markt (z.B. einem Endverbrauchermarkt) operieren, sondern in ein ganzes System von Märkten (z.B. auch in Zuliefererund Kapitalmärkten) eingebettet sind. Opportunistisches Verhalten in einem Markt hat nach Hill auch Auswirkungen auf andere Märkte, die es wahrscheinlich machen, daß sich langfristig kooperatives Verhalten durchsetzt. Das Bestechende an Hills Beitrag ist, daß er mit ökonomieimmanenten Ansätzen arbeitet und zur Entkräftung der Opportunismusannahme nicht einmal auf verhaltenswissenschaftliche Forschungsergebnisse zurückgreifen muß. Der zweite Artikel, „The Link Between Resources and Type of Diversification: Theory and Evidence", von Chatteijee und Wernerfeit ist eine ressourcenbasierte Begründung der Diversifikation. Die Ergebnisse dieser durch ihre klare Strukturierung auffallenden Studie stützen die vorherrschenden theoretischen Argumente. Intangible Ressourcen stellen sich als die dominierende Grundlage für eine erfolgreiche verbundene Diversifikation heraus. Was eine erfolgversprechende unverbundene Diversifikation betrifft, so erweisen sich langfristig zur Verfügung stehende interne finanzielle Ressourcen als die zentrale Ressourcenkategorie.

Cooperation, Opportunism, and the Invisible Hand

403

• Academy oí Management Hevlew, 1990, Vol. 15. No. 3, 500-513.

Cooperation, Opportunism, and the Invisible Hand: Implications for Transaction Cost Theory CHARLES W. L. HILL

University of W a s h i n g t o n Transaction cost theorists have generally neglected to consider the implications that the invisible hand oí the market mechanism can have lor the risk oí opportunism. In the long run, the invisible hand deletes actors whose behaviors are habitually opportunistic. Consequently, as markets move toward the state of competitive equilibrium, the risk of opportunism will be low, even for transactions supported by specific asset investments. Therefore, in many contexts the transaction cost rationale for internalization has been overstated. Over the last two decades transaction cost theory has emerged as a major paradigm in the academic literature. Perhaps the most influential statements about this theory have been made by Williamson (1975, 1985). Drawing on Williamson's work, authors have used transaction costs to explain the configuration of organizational form and a range of strategic phenomena, including diversification, vertical integration, foreign direct investment, joint ventures, and business-level strategy. (Such work includes but is hardly limited to Bowan and Jones [1986], Dundas and Richardson [1980], Hennart [1982, 1988], Hill and Kim [1988], Jones [1983, 1984, 1987], Jones and Hill [1988], Jones and Pustay [1988], Klein, Crawford, and Alchian [1978], Ouchi [1980], Provan and Skinner [1989], Teece [1977, 1982], Walker and Weber [1984], Williamson and Ouchi [1981].) In other words, transaction cost theory has been enormously influential in shaping our thinking. One of the central assumptions underlying this theory is the belief that the risk of opportun-

ism is inherent in many transactions. Opportunism is defined as "self-interest seeking with guile. This includes but is scarcely limited to more blatant forms, such as lying, stealing, and cheating. . . . More generally, opportunism refers to the incomplete or distorted disclosure of information, especially to calculated efforts to mislead, distort, disguise, obfuscate, or otherwise confuse" (Williamson, 1985, p. 47). When joined with bounded rationality and asset specificity, the risk of opportunism suggests that participants to an exchange may attempt to expropriate the composite quasi rent that persuaded others to enter the exchange in the first place (Alchian & Woodward, 1988). A quasi rent is the excess above the returns necessary to maintain a resource in current operation. It can be the means to recover sunk costs, such as those that result from investments in specific assets. A composite quasi rent is that proportion of the quasi rent generated by a resource that depends on continued association with the resources of others. Composite quasi rent is the 500

404

C. W. L. Hill

product of investment in specialized assets to support an exchange. It is also the amount an opportunistic party to an exchange could expropriate and still not force other participants to withdraw from the exchange. (The concept of composite quasi rent was developed by Marshall [1936] who used the example of a steel mill that locates near a power station and makes an investment, the profitability of which depends on buying power at a promised price. Once the steel mill incurs the sunk costs associated with establishing its operations, the power company could raise energy prices. It would make sense for the steel mill to stay in production as long as its new [higher] marginal cost did not exceed marginal revenue, even though sunk costs are not being recovered [i.e., even though it is not recovering its investment in dedicated assets].) The costs of negotiating, monitoring, and enforcing a contingent claims contract to ensure against this possibility are called transaction costs. If the transaction costs of undertaking an exchange through the market outweigh the bureaucratic costs of managing an exchange within a hierarchy, according to transaction cost theory it is more efficient to coordinate the exchange within a hierarchy (Jones & Hill, 1988; Williamson, 1985). Opportunism plays a key role in this choice because without it cooperation will be the norm between parties to an exchange and promise will suffice to safeguard market transactions (Williamson, 1985, p. 31). In other words, without opportunism the economic rationale for coordinating an exchange within a hierarchy would be substantially reduced. However, Williamson assumed that if asset specificity is high, the risk of opportunism is often great enough to warrant replacing the market with a hierarchy. Despite a growing body of work, theorists have neglected to consider the implications that the wider context has for the risk of opportunism. This wider context includes (a) the social context within which economic transactions are embedded and (b) the invisible hand of the market mechanism. One exception to this line of think-

ing has been by Granovetter (1985), who argued that the social context within which transactions are embedded has a profound impact on the behavior of economic actors. Granovetter described social networks and interpersonal relationships between parties to an exchange as mechanisms for attenuating opportunism in situations where Williamson would have suggested that opportunism was likely and the use of a hierarchy was necessary to hold opportunism in check. This article explores the implications that the other component of the wider context—the invisible hand—has for transaction cost theory. It is argued that the rent that is generated by specific asset investments is dissipated by (a) opportunism, (b) the safeguards adopted to attenuate opportunism, and (c) internalization of an exchange relationship within a hierarchy. Accordingly, the value of the composite quasi rent that is inherent in a transaction is maximized whenever actors who are prepared to cooperate meet each other. Consequently, in the long run, the invisible hand selects actors whose behaviors are biased toward cooperation. The selection of such actors leads to the conclusion that the transaction cost rationale for vertical integration has been overstated.

Probability oi Opportunism Consider an exchange between parties X and Y. Williamson (1985) argued that the probability, p, of X behaving opportunistically, p(Ox), is a function of the extent to which Y must invest in specialized assets (ky) to support the exchange relationship, bounded rationality, and uncertainty. If we assume that bounded rationality and uncertainty are givens and hold them constant, it is suggested that: p(Cg = Hky)

(1)

This function is graphed in Figure 1, which shows p(Ox) as an increasing function of ky. The greater the asset specificity, the greater the switching costs facing Y, and the greater the 501

Cooperation, Opportunism, and the Invisible Hand P(0.)

Figure 1. The risk of opportunism and asset specificity. ability of X to expropriate composite quasi rent from Y. However, Williamson did not claim that all actors will behave opportunistically; he merely suggested that the probability of opportunism occurring increases as asset specificity increases. Even if the asset specificity is high, there will be actors who, perhaps for reasons of principle, will never be the first to act opportunistically. Rather, they will choose to cooperate and trust others. Thus, in Figure 1 the function p(Ox) = f(ky) is shown as asymptotic to the line a a ' where a < 1. Whatever the level of asset specificity [k), some proportion of the population will not behave opportunistically. Williamson's argument, however, is that due to bounded rationality and uncertainty it is difficult to distinguish actors who will cooperate and trust from those who will behave opportunistically. Taking this as a starting point, Williamson's theory suggests that the optimal level of safeguards, s, for Y to have to ensure against the risk of opportunism is a function of p(Oz) and Y's share of the composite quasi rent generated by the specific asset investments, ry. Thus: s = Λρ(Οχ) · Tr]

(2)

For example, if p(Ox) = 0.2 and ry = $100, the expected loss due to opportunism would be $20. Thus, it would be to Y's benefit to have safeguards (insurance) in the amount of $20. Contin-

405

gent claims contracts and credible commitments are examples of safeguards. The costs of establishing safeguards include negotiating costs, bonding costs, monitoring costs, and enforcement costs. These are transaction costs. If transaction costs outweigh the bureaucratic costs of managing the exchange within a hierarchy, the theory suggests that internalization of the exchange is efficient. However, Williamson's arguments imply that the choice between market and hierarchy is ultimately determined by the invisible hand of the market mechanism, rather than by rational deliberation on the part of X or Y. Indeed, due to uncertainty and bounded rationality, neither Y nor X can know for sure what the optimal level of safeguards actually is. However, it is argued that the system of markets that surrounds individual economic actors selects those who, through luck or entrepreneurial inspiration, economize on transaction costs by adopting the appropriate mechanisms to govern their exchanges with others. The stress on the phrase system of markets is important. Even if the market in which X and Y interact is characterized by small numbers and a lack of competitive pressure (e.g., when asset specificity is high), competitive pressures in other markets in which X and Y are present (e.g., the capital market) require that these actors structure all exchanges as efficiently as possible so that they may continue to survive. Furthermore, the market in which X and Y interact may itself be contestable in the sense implied by Baumol, Panzer, and Willig (1982). According to these authors, if X and Y fail to structure their exchange relationship in the most efficient manner, the market may be taken over by other more efficient actors. This view is supported by a growing body of theory and evidence that suggests that, in the long run, entry barriers give way to contestable markets (Spence, 1983). This thesis neglects any discussion of the factors that determine the distribution of opportunistic and cooperative actors in the population 502

406

C. W. L. Hill

(i.e., what determines 0 - a in Figure 1). Yet logic suggests that the distribution is an important determinant of transaction costs. Specifically, if we signify the proportion of actors who will behave opportunistically by b0 and the proportion who will cooperate and trust by b c , Equation 1 can be reformulated as: p(Ox) = f(kr, bjbj

mechanism selects organizations on the basis of their repertoires of behavior, suggesting that the distribution of cooperative and opportunistic actors is not exogenous to the economic system. Rather, over the long run, the invisible hand deletes actors who are habitually opportunistic. Consequently, the ratio bjbc declines over time as markets move toward the state of competitive equilibrium. By extension, this decline in ratio implies a decline in the optimal level of safeguards, and in situations in which asset specificity is high, it implies a switch from hierarchy to market as the most efficient way of governing transactions. To demonstrate this concept, a game theory model of opportunism is developed in the following sections.

(3)

where b 0 + b c =. 1. Substituting Equation 3 for Equation 2 we get: s = my, bjbj

• ry]

(4)

Because transaction costs are synonymous with the costs of safeguards that are required to keep opportunism in check, the implication is that bj b c is a major determinant of transaction costs and, thus, of the optimal governance structure. If bjhc declines in value, so does the optimal level of safeguards that Y should have (i.e., the expected loss due to opportunism declines). In turn, this makes market contracting more attractive relative to hierarchy.

Game Theory Model oí Opportunism Opportunism and the Prisoner's Dilemma The starting point for this discussion is the prisoner's dilemma (Friedman, 1986; Luce & Raiffa, 1957). Widely used in the game theory literature, the structure of the prisoner's dilemma closely resembles the structure of an exchange relationship. Consider the prisoner's dilemma as applied to a bilateral exchange relationship that requires significant investments in specialized assets that generate a substantial composite quasi rent. This is the kind of situation that Williamson argued would result in high transaction costs due to the threat of opportunism. In the basic scenario two players have two choices: to cooperate and trust each other or to act opportunistically and expropriate the composite quasi rent. Each player must make a choice without knowing what the other will do. Table 1 describes how the game works. Player X (the row player) either cooperates or acts opportunistically. Similarly, player Y (the column player) either cooperates or acts opportunistically. These choices result in one of the four payoff outcomes, which are shown in the matrix. If both players cooperate and trust each other,

Williamson treated bjbc as exogenous to his model. It appears that he regarded the determination of this ratio as outside the frame of reference of economic models. However, he did recognize the importance of bjbc when he acknowledged that the social context within which transactions are embedded should be taken into account when moving from one culture to another (Williamson, 1985, p. 22). In this article it is argued that when asset specificity is high the probability of opportunism may be attenuated by factors other than that of a favorable social context. In particular, selection mechanisms favor actors whose repertoires oí behavior are biased toward cooperation. Repertoires of behavior are the explicit and tacit rules, routines, and procedures that shape the way in which organizations (or individuals) make decisions and generally "do business." As articulated by Alchian (1950) and Nelson and Winter (1982), such repertoires have important implications for the efficiency of an organization. Thus, the invisible hand of the market 503

Cooperation, Opportunism, and the Invisible Hand Table 1 The Prisoner's Dilemma Player! Player Y

Coopérât· and Trust

Cooperate and Trust

Payoff X Payoff Y

Act Opportunistically

Payoff X Payoff Y

= = = =

Act Opportunistically

$10 $10

Payoff X Payoff Y

$0 $15

Payoff X Payoff Y

= = = =

$15 $0 $5 $5

the payoff is $10 per player. The combined payoff of $20 is equivalent to the total composite quasi rent that can be generated from the exchange. The greater combined value of this payoff (compared to other outcomes) reflects the fact that the tactic of cooperation is superior to the tactic of individual action for achieving virtually all of the players' goals (Alchian & Demsetz, 1972; Maitland, Bryson, & Van de Ven, 1985). If one player cooperates and the other acts opportunistically, the opportunistic player receives $15 and the cooperating player receives the "sucker's" payoff of $0. Put another way, the opportunistic player should be able to expropriate much of the composite quasi rent. However, it is reasonable to assume that some of the composite quasi rent will be dissipated because of the lack of cooperation and goodwill that will ensue once the victim realizes what is happening. The victim of opportunism ends up with zero composite quasi rent. If both players act opportunistically, both receive $5, the punishment for lack of trust. It is assumed that mutual opportunism and the lack of cooperation and goodwill will dissipate much of the composite quasi rent. If this game is played only once, there is a unique equilibrium point. Suppose you are player X and you think player Y will cooperate. If you cooperate, you get a payoff of $10, but if you act opportunistically, you get a payoff of $15. Thus, it is beneficial for you to act opportunistically and get a payoff of $15. Conversely, suppose you think player Y will act opportunis-

407

tically. Now, you have a choice between cooperating, which would make you a sucker and give you a payoff of $0, or acting opportunistically, which would give you a payoff of $5. Again, it is beneficial for you to act opportunistically. Thus, it is better to act opportunistically if you think the other player will act opportunistically, and it is better to act opportunistically if you think the other player will cooperate. So, no matter what the other player does, it pays you to act opportunistically. This logic suggests that the level of transaction costs in such circumstances will be high. Before the exchange, each player will demand bonding costs from the other as an insurance against opportunism. In addition, it may be beneficial for each player to bear negotiating, drafting, and monitoring costs in order to ensure that the other player does not act opportunistically and expropriate the composite quasi rent. The total transaction costs that would be beneficial for both parties to bear in this situation would be equal to the difference between the payoff from mutual cooperation and the payoff from mutual opportunism ($10). Therefore, the benefits that might result from internalizing the focal exchange relationship could be substantial. Transaction cost logic could be used to justify integration in such circumstances. The conclusion that it is best to act opportunistically continues to hold for a repeated game that is played a known or finite number of times. This is true for the last move because there is no future to influence. On the next-to-the-last move neither player will have an incentive to cooperate and trust because each one can anticipate opportunistic action by the other on the final move. Thus, both players have an incentive to act opportunistically on the next-to-the-last move. As originally demonstrated by Luce and Raiffa (1957, pp. 94-102), this logic implies that opportunistic action will be taken on the first move of any sequence of plays that is of a known, finite length. Again, the suggestion is that opportunism may be prevalent in exchanges of this nature. 504

408

C. W. L. Hill

Limitations oí the Finite Game One problem with the argument that opportunism may be prevalent is the assumption that two (or more) players know precisely how many times they will interact in the future. In a world in which the only constant is change, the future is unknowable. However, it seems evident that across a majority of exchanges both parties enter with the expectation that they may interact again in the future, although neither party can predict or know exactly how many times this will occur. More generally, there is the issue of reputation. Reputation has an economic value, and its consideration may play an important part in determining the willingness of others to enter into an exchange with a given actor in the future (Fama, 1980). We can expect that actors will try to avoid entering into an exchange with another actor who has a questionable reputation. For example, if Y enters into an exchange with X and subsequently becomes the victim of opportunism by X, it is reasonable to suppose that Y will communicate this to others through a social network (e.g., gossip). In addition, Y may take legal action against X, and this action, even if it fails, may raise questions in the business community about how trustworthy X actually is. Such factors will tarnish the reputation of X, thereby reducing the probability that other actors will enter into future exchanges with X. If avoidance of such a party is not viable, other actors will demand that the potentially opportunistic party absorb bonding costs when they enter into an exchange with an actor of questionable reputation, such as X. In addition, the other actors will have to bear monitoring costs in order to detect opportunism. These bonding and monitoring costs will absorb much of the composite quasi rent that is inherent in the exchange, reducing the economic value generated from any investment in specialized assets. Thus, the importance of reputation suggests that even a one-time-only transaction has repercussions that go beyond the context and bounds

of that exchange. The potential parties to any future exchange with a given actor will take their cues about what behavior to expect and how to behave in response from that actor's previous behavior in exchanges with others. So long as behavior is visible after the exchange and information pertaining to previous exchanges is available to potential participants in future exchanges, the importance of reputation effects suggests that a one-time-only exchange is simply one in an infinite series of possible future exchanges. (Of course, it is possible that the victim might not be able to detect opportunism, particularly when outcomes are uncertain. The ramifications arising from this possibility are discussed in a later section of this article.) The effects of reputation suggest that Y in the current game need not be a single player. Rather, Y may be any member of a set of players, Yj (i = 1 to n), who are connected by a social network that enables them to share information about the reputation of X. Thus, the strategy adopted by Y2 with regard to X may be a function of how X behaved to Y! in the previous move. With regard to the total set of potential players, Y¡, this argument indicates that the value that can be created in all future exchanges involving X is influenced by X's conduct in present exchanges. Specifically, the value created by exchanges involving actors oí questionable reputation is significantly reduced by the need to set up safeguards to limit opportunism. An Iniinitely Repeated Game The previous discussion raises the question of how the prisoner's dilemma will develop if the game is repeated an unknown or infinite number of times. Consider the game based on Table 1 and suppose that each player discounts future payoffs using a discount parameter, d, of 0.9. The strategy of acting opportunistically provides an equilibrium solution, yielding a payoff of $5 χ (1 + 0.92 + 0.93 + 0.9" . . . + 0.900). Because the sum of any infinite series of the form 1 + d 2 505

Cooperation, Opportunism, and the Invisible Hand + d 3 + . . . d 0 0 is equal to 1/(1 - d), the payoff is equal to 5 x 1/(1 - 0.9) = $50. This is not the only equilibrium point. Contrast this with a strategy in which player Y cooperates during the first iteration and during subsequent iterations if player X does likewise. However, if player X acts opportunistically, even once, player Y will revert forever to the "safe" policy of acting opportunistically. If player X adopts the same strategy, both players can realize a payoff of $10 each per period, yielding a total of $100. If player X takes advantage of player Y's willingness to cooperate, he or she can receive $ 15 during the first period and will receive $5 in each subsequent period, for a total of $60. Clearly, it is not in player X's interest to follow this latter strategy. Therefore, the interest of both parties is best served by adopting a position of cooperation and trust. Referred to as a noncooperative equilibrium, this position can be achieved without the parties discussing their plans in advance. If each assumes the other will have the good sense to act in a cooperative manner, their assumptions will be correct and they will both be at a mutually beneficial noncooperative equilibrium point. Extensions of an Infinitely Repeated Game One limitation of the example of an infinitely repeated game is that it focuses on a restricted number of decision rules or behavioral repertoires. In theory, there are an infinite number of decision rules that can be used to play an infinitely repeated prisoner's dilemma. One wellknown decision rule is "tit for tat" or the strategy of cooperating on the first move and thereafter doing whatever the other player did on the previous move. Another decision rule might be characterized as "always cheat" or the strategy of always acting opportunistically, irrespective of the other's previous behavior. More generally, a decision rule is a specification of what course of action to pursue given the history of the game. Decision rules can be very sophisticated, for example, when a player uses the history of

409

the game to model the behavior of the other player and subsequently uses probability theory to select what should be the best long-run choice. Alternatively, decision rules can lack any rational component, for example, when a player acts opportunistically and cooperates on a random basis. Because of this potential for infinite variety, we might question whether there is one best decision rule. Theory indicates that as long as the discount parameter, d, is sufficiently high (i.e., the future is important) there is no one best strategy independent of the strategy used by the other player (Friedman, 1986). The payoff generated by one strategy is dependent on the strategy used by the other player. Indeed, the central feature of the prisoner's dilemma is the interdependence of outcomes. However, the possibility still exists that if we begin with a large number of different rules among a population of potential players, certain rules might achieve better outcomes in the long run. This is analogous to the situation investigated by Axelrod (1980, 1981, 1984). Using an iterative prisoner's dilemma, Axelrod examined rules that would achieve better outcomes for a population of players when the future was important and the number of future interactions between players was unknown. He used two mediums to examine this situation: (a) a series of round-robin computer tournaments between a set of players' decision rules devised by professional game theorists and (b) subsequent computer simulations. In the round-robin tournaments each player's entry was paired with every other entry, its own twin, and with "random," a nonrational decision rule that randomly cooperated and acted opportunistically with equal probability. The main findings were (a) the strategy of "tit for tat" was associated with the greatest payoff, (b) the high-scoring rules (including "tit for tat") had the property of never being the first to act opportunistically, and (c) the low-scoring rules were those that tried to exploit the other player via opportunism. 506

410

C. W. L. Hill cient market will delete actors whose behavioral repertoires (decision rules) depress performance (Alchian, 1950; Nelson & Winter, 1982). The game theory model discussed in the previous section suggests that opportunistic behaviors fall into such a category. The act of opportunism, by either eliciting a response in kind or by necessitating that actors bear monitoring and bonding costs during future exchanges, reduces the value that can be generated from investments in specialized assets. In other words, the value of the composite quasi rent that is generated from asset specificity is dissipated either by opportunistic behavior or by the safeguards adopted to attenuate the risk of such behavior. It is important to recognize that the invisible hand of the market will delete opportunistic actors even when the focal exchange is characterized by substantial asset-specific investments and high switching costs, that is, even when the focal market is not competitive. This is because the focal market is not the only mechanism through which the invisible hand deletes opportunistic actors. As noted earlier, every firm is surrounded by a system of markets, which includes the capital market, the various markets in which it purchases physical and human inputs, and those markets in which it sells its outputs. Opportunistic action by a firm in any of these markets has ramifications for its ability to prosper in other markets. For example, the firm that acts opportunistically toward a supplier might, in the long run, find its ability to compete in its end market limited by higher costs that were the direct result of a lack of cooperation with its suppliers. As an example, consider the automobile industry. Often, auto part suppliers are obliged to make specific asset investments in order to serve individual auto companies. The high switching costs that these investments entail imply that part suppliers are vulnerable to opportunistic action by the large auto makers (i.e., action designed to expropriate composite quasi rents). Historically, this has been the case for the auto industry in the United States. As recently as

Axelrod explained the robust success of "tit for tat" in terms of the combination of the players using it to be nice, retaliatory, forgiving, and clear. The player who used this rule was nice because he or she was never the first to act opportunistically. The player was retaliatory because he or she retaliated in kind to the opportunism of the other player. The player was forgiving because he or she reverted back to cooperation if the other player did. And the player was clear because he or she sent an unambiguous signal to the other player. However, Axelrod went to great lengths to point out that "tit for tat" was by no means the best rule. Indeed, he identified a number of rules that could have scored even higher. These rules offered players the same central features as "tit for tat" (e.g., niceness), but they were either even more forgiving and/or more sophisticated. The message that can be derived from the tournament (and subsequent computer simulations) is: Players who used rules that deliberately tried to exploit the other players (opportunistic rules) always faired poorly in the long run. Opportunism without provocation inevitably set up a response in kind from players who used the majority of rules and therefore depressed total payoffs. In addition, Axelrod concluded that players who used poorly scoring rules consistently underestimated the value of forgiving the opportunism of the other player in an attempt to establish future cooperation.

Selection Mechanisms and the Evolution of Behavior Axelrod extended his experiment to include the elimination of poorly performing rules by an evolutionary selection mechanism. He found that over time actors whose decision rules (behavioral repertoires) stressed cooperation and trust, rather than opportunism, came to dominate the population of players. The significance of this finding for the present topic is that the invisible hand of the market is an evolutionary selection mechanism. In the long run, an effi507

Cooperation, Opportunism, and the Invisible Hand

411

ism, also dissipates some of the composite quasi rent that is inherent in the exchange. Using hierarchy involves additional bureaucratic costs that do not have to be borne by actors who tacitly agree to cooperate and trust each other. (The sources of bureaucratic costs are discussed in detail by Williamson in chapter 5 of The Economic Institutions oI Capitalism. See also Granovetter [1985], Jones and Hill [1988], and Perrow [1986] for discussions of bureaucratic costs.) For example, General Motors and Ford have historically undertaken extensive vertical integration, apparently as a way of resolving contract problems that were the result of specific asset investments (Monteverde & Teece, 1982). Indeed, the classic example of vertical integration in the transaction cost literature concerns the merger of General Motors and Fisher Body in 1926 (Klein et al., 1978). However, vertical integration has proved to be an imperfect solution to the problem of structuring exchanges involving specific asset investments. Both General Motors and Ford found that vertical integration, by giving in-house suppliers a captive market, has attenuated incentives for in-house suppliers to maximize efficiency. Thus, vertical integration has resulted in substantial bureaucratic costs in the form of low productivity resulting from a lack of high-powered market incentives. As a response, both General Motors and Ford have pursued strategies of greater outsourcing, despite the fact that specific asset investments remain as important as ever in the auto component industry (Standard & Poor's Industry Survey, 1987). In sum, the example of the U.S. auto industry suggests that in the long run the invisible hand of the market will delete opportunistic actors or actors who bear the additional bureaucratic costs of vertical integration because both strategies result in lower payoffs than those generated from cooperation and trust. This deletion of actors will occur even though many of thfe intermediate product markets in which the threat of opportunism is high are not competitive (due to small numbers trading conditions and high

1986, for example, Chrysler instructed its parts suppliers to cut their prices by 2.5 percent, irrespective of prior pricing agreements. Chrysler forced its policy and weeded out suppliers who did not obey (The Next Act, 1986). Although opportunistic action of this type yields short-term benefits, a long-term cost must be borne in the form of a lack of trust and hostility between the company and its suppliers. This long-term cost may, for example, inhibit the future introduction of just-in-time inventory systems, make it difficult for suppliers and manufacturers to work together on designs that will improve the quality of parts and lower assembly costs, and make suppliers hesitant about committing to future cost reductions and/or quality enhancing specific asset investments. In the United States auto industry, the consequences of years of such conflict between the auto makers and parts suppliers are arguably "coming home to roost" in the form of higher costs, lower quality, and declining market share, relative to the Japanese competition. Unlike their U.S. counterparts, the Japanese auto companies have long recognized and nurtured cooperative long-term relationships with their parts suppliers. As a consequence of this policy, in 1985 U.S. manufacturers spent an average of $3,350 on parts, materials, and services for small cars ($6,000), whereas the average Japanese company spent $2,750—a cost saving of $600 that was achieved mainly through more efficient vendor relations (Behind the Hype, 1985). The point of this example is that a major reason for the success of Japanese auto companies is their cooperative relationship with suppliers. The logical consequence of such trends in the auto industry is that if the U.S. firms do not change their behavioral repertoires vis-à-vis their suppliers, they will continue to lose market share to the Japanese companies. In other words, the invisible hand of the market mechanism will delete actors whose opportunistic repertoires dissipate value. It is also important to note that the use of hierarchy, as a response to the threat of opportun508

412

C. W. L. Hill

switching costs). The invisible hand of the market works in the auto industry case because the end market is competitive and because opportunism in intermediate product markets has negative implications for the opportunistic organization's ability to compete in that end market. In terms of the theory discussed previously, trends in the auto industry suggest that the ratio of opportunistic to behavioral actors in the population (bjbc) is declining. This decline is occurring not so much through bankruptcy but through the imitation of successful behavioral repertoires (e.g., those of the Japanese). As this ratio declines, so does the probability of opportunism, p(OJ, and, by extension, the level of safeguards required to check opportunism. If there is no major change, this decline suggests that the competitive equilibrium condition in the auto industry will be one in which auto makers and parts suppliers structure their exchange relationships around behavioral repertoires that include cooperation and trust.

Occurrence oi Opportunism The argument that has been presented here implies that among a population of economic actors who are engaged in repeated transactions that require investments in specialized assets, behaviors that stress cooperation, trust, and forgiveness of isolated opportunism by others have an economic value. In general, the theory suggests that even in cases that are characterized by small numbers and high switching costs, in the long run opportunism does not pay. The act of opportunism itself, the safeguards adopted to guard against opportunism, or internalization as a response to the threat of opportunism, all ultimately dissipate some of the composite quasi rent that is inherent in a transaction supported by specific asset investments. From an ""evolutionary perspective, this theory suggests that over time cooperative, rather than opportunistic, actors will dominate a given population. Thus, in the state of competitive equilibrium we would not expect to observe opportu-

nistic actors. Given the basis for this theory, it is pertinent to ask whether there are circumstances in which opportunism does pay. More specifically, what are the limitations of the game theory model discussed earlier? Shadow oi the Future Opportunism may be a viable strategy if the future is noi important to the aggressor. But as already argued, the possibility that the participants to an exchange may interact again in the future, along with the importance of an actor's reputation for future exchanges with others, suggests that in general the future is important. Except in those rare cases when the actor is prepared to exit from the market, the impact of an actor's current behavior on his or her future business opportunities cannot be discounted. Limits of Reputation The game theory model assumes that the effects of reputation will be reasonably strong. However, there are limits to the efficacy of reputation. Reputation may be difficult to establish due to dispersion problems or a lack of communication among the population of potential participants to an exchange. The case of local traders who deal with mobile customers is an example of such problems. Geographical dispersion, along with the absence of mechanisms for achieving communication between mobile customers, may be the reason that a local trader continually charges mobile customers exorbitant prices. More generally, it is important to note that there is a cost involved in the dissemination of reputation. An actor may have to undertake a costly search to establish the reputation of a trading partner. The more dispersed the population from which the actor comes, the greater the cost of establishing reputation, and the more viable opportunism becomes as a value-maximizing strategy. However, there are also a number of mechanisms that work to improve the efficacy of repu509

Cooperation, Opportunism, and the Invisible Hand

413

might use that technology for purposes they have not agreed on. In such a context, the opportunistic actor expropriates more than the composite quasi rent that is inherent in the focal transaction. This actor also expropriates much of the rent that might be generated by the technology in all future transactions. Therefore, opportunistic behavior by licensees can result in the licensor's losing its firmspecific advantage in technology to its rivals. If the technology is of fundamental importance in establishing a competitive advantage, the payoff to licensees from opportunism might outweigh the present value of future cooperation that is put in jeopardy by opportunistic action. When this is the case, it would be better to expand through internalization, rather than licensing. Thus, it is with justification that transaction costs theorists suggest that it may be better to expand overseas by foreign direct investment, rather than licensing, when technological transfers are risky (Buckley & Casson, 1976; Hennart, 1982; Hill & Kim, 1988).

tation. As Granovetter (1985) noted, informal social networks are a surprisingly effective way of communicating information concerning an actor's reputation. These may be reinforced by more formal mechanisms such as trade associations. The net effect of these mechanisms is to reduce the cost of disseminating information about an actor's reputation. Undetected Opportunism So far we have assumed that opportunism can be detected after the exchange. This is not always possible. When there is significant uncertainty about the outcome of an exchange, opportunism may go undetected. If opportunism is undetected, the exploiting actor will not suffer a decline in reputation and may repeat such behavior in subsequent transactions with impunity. It may well be in the best interest of an actor to behave opportunistically in such circumstances. The conventional transaction cost interpretation of situations in which outcomes are highly uncertain is that given the inevitable contract and monitoring problems, unless a suitable credible commitment can be devised, internalization is the appropriate response (Akerlof, 1970; Alchian & Demsetz, 1972; Williamson, 1985).

However, examples of such cases are limited. For the majority of exchange situations, firmspecific resources are not in jeopardy of being expropriated by a party to the transaction. In most vertical integration contexts, for example, the rent in danger of expropriation is limited to the composite quasi rent that is generated from the focal exchange.

Returns from Opportunism Opportunism might pay when the returns from opportunism in a given time period outweigh the discounted present value of future cooperation that is put in jeopardy by such action. There are circumstances in which it might be worthwhile for an actor to bear the negative reputation effects of opportunistic behavior. However, for this to be the case the gains from opportunism must exceed the composite quasi rent that is inherent in the focal exchange. One such situation is when the focal transaction involves the licensing of firm-specific technological know-how. Licensing know-how can be regarded as an arm's length or market transaction. When a firm licenses its technology to another firm, it runs the risk that the licensee

Market Efficiency Opportunistic actors might survive and prosper when selection mechanisms fail. Although opportunistic behaviors may be associated with inferior returns, when markets are inefficient the inability of the invisible hand to punish opportunistic actors through bankruptcy or takeover implies their continued survival. For this to occur, however, all the major markets in which an actor participates must be characterized by failure to such an extent that the actor is not subjected to significant competitive pressures. Suffice it to say that the logic underlying economic theories of natural selection is accepted 510

414

C. W. L. Hill

here (e.g., Alchian, 1950; Nelson & Winter, 1982). In the long run the system of markets surrounding an actor are efficient at deleting unproductive behaviors (Jacobson, 1988; Spence, 1983). Although generalized market failure exists, it tends to be a short-term phenomenon. Thus, if there are efficient selection mechanisms, in the state of competitive equilibrium we would not expect to find opportunistic actors. They would have been deleted by product market or capital market pressures.

Conclusion If the wider context within which economic transactions are embedded is considered, it can be concluded that over time the invisible hand of the market favors actors whose behavioral repertoires are biased toward cooperation, rather than opportunism. Opportunism, the safeguards needed to check opportunism, and internalization as a response to opportunism dissipate the composite quasi rent that is generated from investments in specialized assets. The composite quasi rent is only maximized when actors who are prepared to cooperate and trust meet each other. Thus, over time the invisible hand of the market will favor cooperative actors. Because there will be at least a few actors who are initially prepared to cooperate and trust, this argument suggests that when the state of competitive equilibrium is reached the population of economic actors will contain only those whose behavioral repertoires are biased toward cooperation. This line of reasoning raises serious questions about the transaction cost rationale for integration. Whereas conventional transaction cost theorists see internalization as the equilibrium response to situations involving substantial specific asset investments, this article suggests the equilibrium response is the emergence of a cooperative and trusting relationship. The error of conventional transaction cost theory comes not

so much in the construction of the theory but from the failure of researchers to consider the implication of the wider context for the distribution of behavioral repertoires among a population of economic actors. In addition, conventional transaction cost theory is essentially static. Theorists have failed to consider what the ramifications of a dynamic revolutionary process would be for the distribution of behavioral repertoires. For example, Williamson's model, as outlined in chapter 4 of The Economic Institutions of Capitalism, is a single-period model, and its predictions do not hold for the long run. However, it is apparent that the cooperative behavior solution is unable to deal with opportunism when (a) outcomes are highly uncertain, (b) reputation is difficult to establish, and (c) the payoff from opportunism in the present period outweighs the discounted present value of future cooperation that is put in jeopardy by such action. Thus, the theory advanced here does noi dispense with economic arguments for hierarchy. Indeed, it confirms important aspects of transaction cost theory. For example, the argument presented here suggests there is value in the theory that hierarchical governance has efficiency properties when outcomes are highly uncertain or ambiguous. This position regarding hierarchy also can be found in the works of Akerlof (1970), Alchian and Demsetz (1972), Jones (1984), and Ouchi (1980). Akerlof s (1970) arguments suggest that hierarchy may be appropriate when outcomes are uncertain and measurement costs are high (e.g., with complex heterogeneous products). Building on this position, Anderson and Schmittlein (1984) found that problems associated with measuring individual performance are a strong predictor of whether firms use independent sales representatives (market contracting) or a direct sales force (hierarchy) to sell their products. They also found that outcome uncertainty was a much greater predictor of hierarchy than asset specificity, a finding that echoes some of the themes advanced in this article. 511

C o o p e r a t i o n , Opportunism, and the Invisible H a n d

A l c h i a n a n d D e m s e t z (1972), J o n e s (1984), a n d O u c h i (1980) s t r e s s e d t h e p r o b l e m of perform a n c e a m b i g u i t y that results from t e a m p r o d u c tion. T h e difficulty of identifying individual c o n tributions in c a s e s of t e a m production a l l o w s individual m e m b e r s of a t e a m to shirk a n d f r e e ride (act opportunistically) o n t h e contribution of others. In s u c h c i r c u m s t a n c e s , h i e r a r c h y a r g u a b l y h a s definite monitoring a n d e n f o r c e m e n t a d v a n t a g e s over market-mediated mechanisms for limiting opportunism.

415

c i a t e d with substantial a s s e t - s p e c i f i c investm e n t s . T h e a r g u m e n t a l s o h a s i m p l i c a t i o n s for p u b l i c policy a n d b u s i n e s s policy. In t h e c a s e of p u b l i c policy, t h e p e r m i s s i v e policy s t a n c e tow a r d m e r g e r s that is a d v o c a t e d b y t r a n s a c t i o n cost theorists c a n b e q u e s t i o n e d . In t h e c a s e of b u s i n e s s policy, t h e r e c o m m e n d a t i o n that integ r a t i o n is t h e b e s t w a y of d e a l i n g with t h e p r o b l e m s that result from s p e c i f i c a s s e t i n v e s t m e n t s c a n b e c h a l l e n g e d . R a t h e r , this a r t i c l e s u g g e s t s that t h e construction of a l o n g - t e r m r e l a t i o n s h i p b a s e d a r o u n d c o o p é r a t i o n a n d trust is optimal; therefore, r e s e a r c h e r s s h o u l d direct their attention t o w a r d identifying w a y s in w h i c h this c a n b e established.

Notwithstanding t h e s e qualifications, t h e theory a d v a n c e d h e r e raises serious questions a b o u t t h e g e n e r a l applicability of t h e t r a n s a c tion cost r a t i o n a l e for integration in c a s e s a s s o -

References Akerlof, G. A. (1970) The market lor "lemons": Quality uncertainty and the market mechanism. Quarterly Journal oí Economics, 84, 488-500.

Buckley, P. O., & Casson, M. C. (1976) The future of the multinational enterprise. London: Macmillan. Dundas, Κ. R . & Richardson, P. R. (1980) Corporate strategy and the concept of market failure. Strategic Management Journal, 1, 177-188.

Alchian, A. A. (1950) Uncertainty, evolution, and economic theory. Journal ol Political Economy, 58, 211-221. Alchian, Α. Α., & Demsetz, H. (1972) Production, information costs and economic organization. American Economic Review. 62, 777-795.

Fama, E. F. (1980) Agency problems and the theory of the firm, /ournai of Political Economy, 88, 288-307. Friedman, J. W. (1986) Game theory with applications to economics. Oxford, England: Oxford University Press.

Alchian, Α. Α., & Woodward, S. (1988) The firm is dead: Long live the firm. Journal oí Economic Literature. 26, 6579.

Granovetter, M. (1985) Economic action and social structure: The problem of embeddedness. American Journal ol Sociology, 91, 481-510.

Anderson, E., & Schmittlein, D. C. (1984) Integration of the sales force: An empirical examination. Rand Journal of Economics, 3, 385-395.

Hennart, J. F. (1982) A theory oí the multinational Ann Arbor: University of Michigan Press.

Axelrod, R. ( 1980) Effective choice in the prisoner's dilemma, /ourna/ of Conflict Resolution, 24, 3-25.

Hennart, J. F. (1988) A transaction cost theory of equity joint ventures. Statistical Management /ournai, 9, 361-374.

Axelrod, R. (1981) The emergence of cooperation among egoists. American Political Science Review, 75, 306-318. Axelrod, R. (1984) The evolution oí cooperation. Basic Books.

enterprise.

Hill, C. W. L., & Kim, W. C. (1988) Searching for a dynamic theory of the multinational enterprise: A transaction cost model. Strategic Management Journal. 9, 93-104.

New York:

Jacobson, R. (1988) The persistence of abnormal returns. Strategic Management Journal, 9, 415-430.

Baumol, W. I., Panzar, J. C., & Willig, R. D. (1982) Contestable markets and the theory oí industry structure. San Diego: Harcourt Brace Jovanovich. Behind the hype at GM's Saturn. (1985, November 11) Fortune, pp. 34-46.

Jensen, M. C. (1988), The takeover controversy: Analysis and evidence. In I. C. Coffee, L. Lowenstein, & S. RoseAckerman (Eds.), Knights, raiders and targets (pp. 314— 354). Oxford, England: Oxford University Press.

Bowen, D. E., 8t Iones, G. R. (1986) Transaction cost analysis of service organization-customer exchange. The Academy of Management Review, U, 428-441.

Jensen, M. C., & Ruback, R. (1983) The market for corporate control: The scientific evidence. Journal of Financial Economics, 11, 5-50.

512

416

C. W. L. Hill

Jones, G. R. (1983) Transaction costs, property rights, and organizational culture: An exchange perspective. Administrative Science Quarterly, 28, 454 -467. Jones, G. R. (1984) Task visibility, free riding, and shirking: Explaining the effect of structure and technology on employee behavior. Academy of Management Review, 9, 684-695. Jones, G. R. (1987) Organization-client transactions and organizational governance structures. Academy of Management Journal, 30, 197-218. Jones, G. R., 8c Hill, C. W. L. (1988) Transaction cost analysis of strategy-structure choice. Strategic Management Journal 9, 159-172. Jones, G. R., 8c Pustay, M. W. (1988) Interorganizational coordination in the airline industry. Journal of Management, 14, 529-546. Klein, B.( Crawford, R. Α., & Alchian, A. A. (1978) Vertical integration, appropriate rents, and the competitive contracting process. Journal of Law and Economics, 21, 297326. Luce, R. D., 8c Raiffa, H. (1957) Games and decisions. New York: Wiley. Maitland, I., Bryson, J., & Van de Ven, Α. (1985) Sociologists, economists, and opportunism. Academy of Management Review, 10, 59-65.

Perrow, C. (1986) Complex organizations: A critical essay. New York: Random House.

Nelson, R. R., 8c Winter, S. G. (1982) An evolutionary theory of economic change. Cambridge, MA: Harvard University Press. The next act at Chrysler. (1986, November 3) Business Week, pp. 66-72. Ouchi, W. G. (1980) Markets, bureaucracies, and clans. Administrative Science Quarterly, 25, 120-142.

Williamson, O. E. (1985) The economic institutions of capitalism. New York: Free Press.

Provan, K. G., 8c Skinner, S. J. (1989) Interorganizational dependence and control as predictors of opportunism in dealer-supplier relations. Academy of Management Journal, 32, 202-212. Spence, A. M. (1983) Contestable markets and the theory of industry structure: A review article. Journal of Economic literature, 21, 981-990. Standard 8c Poor's Industry Survey. (1987, April 23) AutosAuto Parts. Teece, D. I. (1977) Technology transfer by multinational firms. Economic Journal, 87, 242-261. Teece, D. J. (1982) Toward a n economic theory of the multiproduct firm, /ou m ai of Economic Behavior and Organization, 3, 39-64. Walker, G., 8c Weber, D. (1984) A transaction cost approach to make or buy decisions. Administrative Science Quarterly, 29, 373- 391. Williamson, O. E. (1975) Markets and hierarchies: Analysis and antitrust implications. New York: Free Press.

Williamson, O. E., 8c Ouchi, W. G. (1981) The markets and hierarchies program of research: Origins, implications, and prospects. In W. F. Joyce 8c A. Van de Ven (Eds.), Perspectives on organizational design and behavior (pp. 347-390). New York: Wiley.

Charles W. L. Hill (Ph.D., University of Manchester, England) is Assistant Professor of Management and Organization at the Universify of Washington. Correspondence regarding this article can be sent to him at Department of Management and Organization, DJ-10, Schools of Business, University of Washington, Seattle, WA 9Θ195. Thanks to Peter Mills, Tom iones, and Tom Thomas for their comments on an earlier draft of this manuscript.

513

R e s o u r c e s a n d T y p e of Diversification

Strategic

Management

Journal,

Vol. 12, 33-48

417

(1991)

THE LINK BETWEEN RESOURCES AND TYPE OF DIVERSIFICATION: THEORY AND EVIDENCE

(

SAYAN CHATTERJEE

Weatherhead School of Management, Case Western Reserve University, Cleveland, Ohio, U.S.A.

BIRGER WERNERFELT

Sloan School of Management, Massachusetts Institute of Technology, Cambridge, Massachusetts, U.S.A.

In this paper we theoretically and empirically investigate the idea that firms diversify in part to utilize productive resources which are surplus to current operations. Knowledge of these resources allows us to make predictions about the direction of a firm's expansion. In particular, we suggest that excess physical resources, most knowledge-based resources, and external financial resources are associated with more related diversification, while internal financial resources are associated with more unrelated diversification.

Perhaps the research question that has attracted the most attention in the strategic management discipline is the possible association between firm diversification and performance. While there are many studies which have supported R u m e l t ' s (1974) original finding that related diversified firms p e r f o r m better than those that are unrelated (Montgomery, 1979; Bettis, 1981; Rumelt, 1982; Palepu, 1985; V a r a d a r a j a n , 1986; V a r a d a r a j a n and R a m a n u j a n ) , 1987; Jose, Nichols, and Stevens, 1986; Lubatkin and Rogers, 1989) there is a growing n u m b e r that find the opposite (Michel and Shaked, 1984; R a j a g o p a l a n and Harrigan, 1986; Elgers and Clark, 1980; C h a t t e r j e e , 1986) or are indifferent (Lubatkin, 1987). This lack of consistent support for the relatedness hypothesis raises a question. Is unrelated diversification the better choice in specific instances, even though on average it seems to be inferior to related diversification? T o illustrate, Montgomery (1979) found that related diversifiers operate within high-profit industries. This could be because the same firms are better at diversification strategy as well as industry selection, or it could be because some underlying factors allow them to enter these industries and make related diversification their best strategy. If it can be shown that, under specific situations, unrelated 0143-2095/91/010033-16$08.00 © 1991 by John Wiley & Sons, Ltd.

diversification can also create value, then we will have taken the first step to explaining some of the apparent inconsistencies in the literature. T o the best of our knowledge there are no studies that have tried to answer this question. 1 It may well be that firms are indifferent to the type (related or unrelated) of diversification. H o w e v e r , if we can identify systematic factors that influence the type of markets e n t e r e d , this will be an important finding. For example, the relatedness hypothesis would need to be modified if it can be shown that both related and unrelated diversification can be justified depending upon certain ex-ante factors specific to individual firms. This p a p e r builds a theoretical foundation to identify systematic factors that influence the type of diversification, and empirically examines the validity of these factors to explain the type of diversification undertaken by a diverse group of firms between 1981 and 1985. This paper demonstrates that the resource profile of the

1 C h a t t e r j e e und W e r n e r f e i t (1988) d o link r e s o u r c e s to the type of e n t e r e d m a r k e t . H o w e v e r , they d o not c o n t r o l for o t h e r factors that can affect t h e decision to e n t e r certain types of m a r k e t s , n o r d o they d r a w the link b e t w e e n the selection of type of m a r k e t a n d p e r f o r m a n c e . T h e s e arc the m a j o r e x t e n s i o n s m a d e in this p a p e r over the C h a t t e r j e e and W e r n e r f e l t (1988) study.

Received 9 October 1989 Revised 20 April 1990

418 34

S. Chatteijee/B. Wernerfeit S. Chatterjee and Β. Wernerfeit

firms before 1981 can partially explain the type of diversification the firms in the sample engaged in from 1981 to 1985. The paper also demonstrates that high-performing firms conform more closely to the theoretical predictions than do lowperforming firms. The results have important implications for managing diversification and for research in the area of diversification.

THEORETICAL FOUNDATIONS This paper is not about why a firm diversifies, but rather the type of market that a firm chooses to enter. Once a firm decides to diversify, the type of market chosen for entry should be such that it provides the firm with a competitive advantage. Porter (1987) suggests that a firm can gain such competitive advantages if it has skills or resources that it can transfer into the new market. His suggestion is not new. Resources have long been recognized to be one of the key factors in explaining diversification (Penrose, 1959). Rumelt (1974) talks about 'core skills' which can be used in related markets. Pfeffer and Salancik (1978) and Burt (1983) view multimarket operations of diversified firms as a means of managing resource-dependent relationships. The same theme is echoed in the economics literature (Teece, 1982; Gorecki, 1975; Caves, 1982; Lecraw, 1984). The empirical evidence also suggests an association between diversification and the diversifying firm's resource position. At the aggregate industry level both Lemelin (1982) and Carleton, Harris, and Stewart (1984) find that firms tend to diversify into industries which use resources similar to their own. Studies by Lecraw (1984) and Montgomery and Hariharan (1990) corroborate this at the individual firm level. All these studies suggest there is a systematic relationship between the type of market a firm chooses to enter and its resource profile. To use this relationship in developing our hypotheses, we need to identify (a) a typology of resources which is generalizable across different firms, and, (b) the association between resources, type of markets, and the potential for value creation. If these can be identified, then we should be able to see systematic patterns across a cross-section of firms about the type of entered market. Put another way, a resource-based approach allows

us to adopt the perspective of the diversifying firm's managers. If our basic assumption is valid, then we would expect managers to deploy firm resources to markets they believe would lead to the most profits. Of course, resources may not be the only factors that can explain the type of entered market. The empirical tests shall attempt to control for the more important of these other factors. Hypotheses about type of entered markets. While Rumelt (1974) originally classified diversification as either related or unrelated, most recent literature considers the issue a matter of degree which is continuously variable (Montgomery, 1982; Caves, Porter, Spence, and Scott, 1980; Montgomery and Wernerfelt, 1988). We will therefore follow the latter approach, and both theoretically and empirically think of degrees of relatedness which can vary continuously from horizontal to unrelated. So the phrase 'type of diversification' should be understood in this context. The type of diversification that we would expect to result from a resource depends on its specificity within a particular industry (Montgomery and Wernerfelt, 1988; Gort, Grabowski, and McGuckin, 1985; Williamson, 1975; Gorecki, 1975). Clearly, if a resource can be used to produce only one product, it is not suitable for diversification. However, most resources can be used for more than one end-product. In the interest of brevity let us call this characteristic of resources 'flexibility.' If a firm owns resources which are fairly end-product-specific (inflexible), then such afirmwould be constrained to diversify in a relatively related fashion; whereas if a firm possesses resources which are flexible (regarding end-products), it would have the option of either more or less related diversification. Following a substantial tradition in the literature (Teece, 1982; Macdonald, 1984; Montgomery and Hariharan, 1990) we consider three classes of resources: (a) physical resources, (b) intangible assets, and (c) financial resources. The first two are fairly inflexible; therefore, they can be used only to enter closely related markets. Financial resources, being most flexible, are useful for any type of diversification. A testable hypothesis from the above arguments is: physical and intangible assets would lead to more related diversification, while

Resources and Type of Diversification Link between Resources and Type of Diversification financial resources can lead to any type of diversification. To complete the argument we need to consider the extent to which different resources can be leveraged. Some resources, such as physical and financial resources, can be used only to the point where they are physically exhausted. So the only excess capacity available for diversified expansion is the stock beyond the requirements of the current businesses. By contrast some intangible resources such as brand names can be repeatedly used with different products with little cost in the effectiveness of original operations. These intangible assets usually accrue to a firm over time, and reside in the human capital of the firm in the form of knowledge and expertise. A patented chemical formula is an extreme example of such a resource. Less extreme examples include innovative capability or marketing skills of research and marketing staffs (see Wemerfelt, 1989 for more details). We now present the specific hypotheses to be considered with the supporting reasoning for each. Physical resources Physical resources of a firm, such as plant and equipment, are characterized by fixed capacity. Also, they are usually useful in a few very similar industries (inflexible). So if excess physical capacity motivates diversification, it would be in industries closely related to those in which the capacity is being used. Barton (1988) and Bettis (1981) have shown that capital expenditures are associated with related diversification. Firms which have excess capacity of such resources are unlikely to use it for diversification far from their core businesses. HI: Excess physical capacity will lead to related diversification. Intangible assets Intangible assets include brand names or innovative capability. Unlike physical assets, intangible assets tend to have 'softer' capacity constraints. A brand name can be applied to several products with little or no adverse effects on existing applications. Similarly, a strong marketing team or innovative research department can success-

419 35

fully market or innovate new products in many different markets without affecting the original businesses. Intangible assets are also relatively inflexible and, therefore, can be used to most advantage in related industries. This expectation has also been suggested by others. Bettis (1981) suggests that related firms perform better because these (intangible) assets 'open up the possibility for differentiation and segmentation' (p. 381) and achieve high performance 'by early entry into (related) industries susceptible to entry barriers and then exploiting a "core skill" . . . to erect such barriers' (p. 390). Hill and Snell (1988) also suggest that in high research-intensive industries the best interest of stockholders would be served by limited and related diversification. Empirical evidence supports this expectation. There is evidence that firms operating in advertisingor research-intensive industries diversify into industries having high research or advertising intensity (Montgomery and Hariharan, 1990; Carleton, Harris, and Stewart, 1984; Lecraw, 1984; and Lemelin, 1982) which are related to their core markets (Bettis, 1981; Caves, Porter, Spence, and Scott, 1980). Taken together these studies suggest the testable hypothesis that intangible assets are used to enter related markets where they are most likely to generate a competitive advantage. In sum, we expect high levels of intangible assets to encourage related diversification. H2: Presence of intangible assets will lead to related diversification. Financial resources Financial resources in general are the most flexible of all resources because they can be used to buy all other types of productive resources. To arrive at hypotheses about how financial resources will be used we will break them up into two classes. The first class, internal funds, consists of liquidity at hand and unused debt capacity to borrow at normal rates. The second class, external funds, consists of new equity and possibly high-risk debts (such as junk bonds). Several theories suggest that lower levels of internal funds (relative to external funds) will lead to lower levels of unrelated diversification and vice-versa.

420 36

S. Chatteijee/B. Wernerfeit S. Chatterjee and Β.

Wernerfeit

If unrelated diversification is truly unprofitable then the only reason managers would pursue such diversification is to increase the size of the firm and the manager's power as predicted by agency theory. Under this scenario, Jensen's (1986) 'free cash flow' hypothesis suggests that only firms with low.leverage can pursue such unprofitable unrelated diversification. Firms with high leverage by necessity will have to go to the capital market for funds when 'the markets have an opportunity to evaluate the company, its management, and its proposed projects' (Jensen, 1986: 324). Since unrelated diversification is thought to be risky by the capital market (Montgomery and Singh, 1984; Rajagopalan and Harrigan, 1986; Lubatkin and O'Neill, 1987; Barton, 1988) external funds will not generally be available for unrelated projects. In other words, if agency behavior is widespread we would expect firms with low leverage pursuing unrelated diversification. What if, by and large, managers do try to maximize shareholder wealth (i.e. agency theory is not the prevalent predictor for management behavior)? The only reason then a manager would undertake a relatively unrelated diversification is if the manager is convinced that if (s)he can invest in the project then ex-post it will increase shareholder wealth. However, given the capital market's reluctance with unrelated diversification the manager is faced with a situation where the information about the project is either unknown to the capital market or is not acceptable exante. Myers and Majluf (1984) suggest that under this difference in information managers would utilize internal funds to fund such projects. Further, since related moves are looked at more favorably, external funds will be reserved for more related diversification.2 Both of the two theoretical perspectives lead us to expect that relatively more unrelated diversification will be associated with internal funds and relatively more related diversification will be associated with external funds, which leads to the following hypotheses. 1 While both arguments lead to the same behavioral predictions (the type of entered market), the performance implications are totally opposite. Under the free cash flow' scenario the unrelated moves will not increase profitability ex-post, while under the Myers and Majluf scenario it will. This point is discussed in detail when the findings are interpreted in the discussion section.

H3A : A variability of internal funds or unused debt capacity will favor more unrelated diversification. H3B: Availability of equity capital will favor more related diversification. The theoretical predictions are summarized in Figure 1. Hypotheses about performance Our theory is developed according to the basic assumption that firms undertake strategic moves with the expectation of improved performance. We do not claim that any one type of diversification will lead to higher performance, but it is the proper application of resources that will improve performance. Thus we would expect that high-performing firms will use intangible and physical assets to enter more related markets. We also allow for the possibility that managers may be able to identify profitable opportunities in unrelated markets. If our theory is descriptive of profit-maximizing behavior, the firms which follow the predictions more closely should perform better. H4: Firms which have higher performance, ex-post, will conform better to our model. Controls While the resources of a firm may provide a systematic explanation of the type of market decision there are other factors which may affect the type of entered market. The factors chosen have typically been found to influence diversification. To account for any possible systematic influences these factors are used as control variables. Risk One problem with empirical verification of our model lies with our basic assumption that managers act to benefit stockholders and not their own utility (agency theory). If indeed managers are trying to increase their own utility in a relatively large proportion of the firms studied, then diversification may be pursued to build empires or reduce personal risk. For example, Hill and Snell (1988: 580-581) suggest

Resources and Type of Diversification Link between Resources and Type of Diversification Unrelated

421 37

Financial Resources: (Internal Funds)

M A

(Low-Risk Debts)

R Intangible Assets

Κ E Τ

Related

Physical Resources

Intangible Assets

Financial Resources: (Equity Capital) (Junk Bonds)

Low

High

FLEXIBILITY O F R E S O U R C E C L A S S E S Figure 1. The relationship between the flexibility of resources and the type of market.

that in a high-risk/high-return environment of research-intensive industries, risk-averse managers may choose diversification (agency behavior) while the best interest of the stockholders will be served by limited and related diversification (our H2). Clearly, if agency behavior prevails we should not find support for our hypothesis and, therefore, needs to be controlled for. Since agency behavior is likely to surface when the risk of bankruptcy (and personal loss for the managers) is high (Amihud and Lev, 1981), we use the initial level of risk to control for agency costs.·1

Size While the resource-based approach does not allow us to make a prediction about the direction of association between size and the type of diversification, large initial size of the firm may be associated with unrelated diversification and, therefore, should be controlled for. Size is used as a control variable in practially all multivariate studies of this type.

Λ Some authors have tried to capture agency problems by looking at stock concentration (Hill and Snell. 1988). However, typically the dispersion in stock concentration is low (Hill and Snell, 1988: S87) and even then there are other factors such as compensation schemes that can reduce agency costs. Risk is a much more direct measure and has inherently more variation than stock concentration, making it more suitable for multivariate tests.

Capital

intensity

Both Barton (1988) and Bettis (1981) have found an association between capital intensity and related diversification. We will therefore control for the initial level of capital intensity. Initial level of

diversification

It is possible that the initial level of diversification may influence future diversification decisions. In a theoretical sense the initial level of diversification may indicate a level that the firm is comfortable with. Thus, for example, a firm with a related diversification status may be less inclined to undertake a relatively unrelated move. Empirically, when we try to measure a change, the initial level may be correlated with a future level.

DATA AND MEASURES We are interested in a quantifiable measure of the change in diversification profile for a sample of firms between 1981 and 1985, and explain this change as a function of the resources that the firm possessed in 1980, i.e. at the beginning of the period. The sample was compiled from two primary data sources, the Trinet Establishment data base and the Compustat Industrial Annual data base. The Trinet data contain employment, sales, and SIC code information for over 200, 000 plants having more than 10 employees, which

422 38

S. Chattel]ee/B. Wernerfeit S. Chatterjee and Β. Wernerfeit

are associated with more than 4000 parent firms. For these parent firms the Trinet data allow us to compute sales per four-digit SIC code in each of the two years considered. Of these parent firms 1203 are listed on the Compustat tapes. The sample reduced to 678 firms for which information on order backlogs, sales, capitalization and stock prices were available in the Compustat tapes, but data on advertising and research and development expense were only available for 167 firms. The need for stock return data from the CRSP tapes (to compute risk) further reduced the sample to 118. These 118 are listed in Compustat under 82 four-digit SIC codes covering 63 three-digit codes. The total assets per firm per SIC listing vary from 6 million dollars (computer disk and tape drives) to 12 billion dollars (motor vehicle and car bodies). Average assets per firm are roughly 900 million dollars.

Dependent variable We compute a diversification index DW at two points in time—1981 and 1985: DW = 2 d lhPli i

where dih equals distance of industry i from that of the firm's largest business (Λ). and p¡¡ equals the fraction of the firm /'s sales which are in industry j. Following Caves, Porter, Spence, and Scott (1980: 199-200), dih = 0 if i and h are in the same four-digit SIC code, d,h = 1 if they are in the same three-digit SIC code, and so on. DW measures diversification away from a core business (the largest business). Like all diversification measures of this type, it captures the reallocations of the firm's resources between different product markets, as well as entry into a new product market. Our dependent variable is DELTADW = DW(85) - DW(81). This variable thus measures a longitudinal change in the degree of diversification between two points in time. Note that large values of DELTADW indicate relatively unrelated diversification while small values indicate relatively related diversification. Several other continuous

measures of diversification, such as the entropy and the Herfindahl measures, have been used in the literature. As Caves et al. (1980: 201) have shown, most of these are highly correlated and lead to similar results.4 Physical resources A direct measure of the firm's excess capacity at any point was not available. However, we noted that a firm would have a relatively lower backlog of orders during a period when it has excess capacity. An economy-wide increase in backlog indicates a very high level of capacity utilization.5 The Compustat tapes provide data on backlog of orders for individual firms. Instead of using the absolute backlog in 1980, we use the ratio of the backlog of orders in 1980 to the 3-year moving average backlog going back to 1974. In dividing by the historical levels we partially correct for the industry and persistent firmdifferences in average backlogs, and by taking a moving average we take some account of different growth rates. We use the name BKLOG for this variable. A firm with a lot of excess plant and equipment is likely to have a low level of backlog. Given our hypothesis, the coefficient on BKLOG is expected to be positive. Intangible assets Intangible resources such as marketing and innovative skills are usually measured by absolute levels of spending intensity. Following Bettis (1981) and Lecraw (1984), we used R&D to sales and advertising to sales. So we define:

4

We estimated the model using several of the measures developed by Caves el at. (1980) and others (entropy). The results are virtually identical. We are presenting the findings with DELTADW because it is most likely to capture diversification away from a core business (which is more likely to contribute to the core skills). Note also that continuous measures such as DELTADW are different from the categorical measures used by Rumelt (1974). However. Montgomery (1982) demonstrated that continuous measures correlate very strongly with the categorical measures used by Rumelt (1974). So we feel that results obtained using these measures are robust. (It would of course be very difficult to construct a measure of change in diversification using a categorical measure.) * The Business Conditions Digest, published by the Department of Commerce, uses both "unfilled order of durable goods' and 'slower deliveries' of all goods as leading indicators along with capacity utilization.

Resources and Type of Diversification Link between Resources and Type of Diversification RSL = the ratio of R&D expenses to sales in 1980. ASL = the ratio of advertising expenses to sales in 1980.A Following our theoretical arguments we expect the coefficients on RSL and ASL to be negative, since lower values of DELTADW indicate more related moves. Financial resources Internal funds and debt The standard measures for liquidity are usually used as a proxy for availability of internal capital (Palepu, 1986). These are the debt to market value ratio and the current ratio. We decided to use both of them: DEMKT = ratio of long-term debt to market value in 1980. CR = ratio of current asset to current liabilities in 1980. Note that low values of DEMKT and high values of CR would also imply a low default risk. So these variables can be used to test for both availability of internal funds and/or 'low default risk' debt.7 According to H3A, we predict a negative sign on the coefficient on DEMKT and a positive sign is expected on the coefficient of CR. Equity capital To measure the firm's ability to raise external capital during the study period, we need to relate " Note that it is important to concentrate on absolue levels of these expenditures as Bettis (1981) does. The average firm in an industry characterized by a high level of research intensity will have a better chance of exploiting diversification opportunities than an average firm in an industry characterized by a lower level of research intensity. While the R&D spending of both firms relative to the level in the industry is likely to be of similar magnitude, the absolute level of R&D spending will be higher for the firm in the more research-intensive industry. 7 Practically speaking a firm with enough internal funds will not need to borrow. Hence it will almost always have a low leverage and low default risk. However, some firms may have been paying out their excess cash as dividends. In such a case a low leverage may not imply availability of internal funds. However, if the firm takes on debt it will still be of a low default risk because now the cash flow can be used to service the debt instead of dividends.

423 39

the average stock price in the study period to the 'normal' stock price for the firm. To be consistent with the other measures of resources we should use the stock price of the firm in 1980 and relate it to a 'normal' stock price. However, unlike leverage, stock prices are much more volatile and even a relatively temporary rise in stock prices provides opportunities to quickly raise external funds or even engage in a stockswap merger. To take this characteristic of stock prices into account we take the average stock price in 1980-84 divided by that of the preceding period 1975-79. We use 1980-84 instead of 1981-85 on the judgement that it takes roughly a year to translate expenditures into sales." If this ratio is high, the market is willing to supply capital below historical averages. Since the time periods are the same for all firms, this automatically corrects for the market variation in stock prices. So we define: RLSTK = the ratio of average stock price, 1980-84, to average stock price, 1975-79. According to H3B we expect that the coefficient on RLSTK to be negative.

Performance measures We use the average return on assets (ROA) over the period 1984-86 to measure ex-post performance. Since the diversification moves were observed between 1981 and 1985, we chose 1984 as the first point when the performance would show up. By starting measurement too early we would lose effects of late actions, and by starting too late we could lose effects of early actions. 1984 is a compromise. The mean ROA was used as the cut-off point for high and low performance.

Control Measures We control for the initial levels of risk, size, capital expenses, and diversification of the firms in the sample. " The model was also estimated using only the average stock price in 1980 divided by the previous period. The results were very similar to the ones reported.

424

S. Chatteijee/B. Wernerfeit

40

S. Chatterjee and Β.

Wernerfeit

Risk We use the variance of the firm's returns (TOTRISK) from 1978 to 1980 as a measure of the risk which managers may be tempted to reduce by unrelated diversification according to the agency theories. TOTRISK should, therefore, have a positive sign.

implying that these firms, on the average, have moved to more diversified postures over the study period. Second, there is a positive correlation between excess capacity (BKLOG) and equity capital (RLSTK), suggesting that unexpected increases in demand influence stock prices. Full sample

Size We use the natural log of total assets in 1980 to measure the impact of size (Bettis, 1981). SIZE = -l/log(total assets) While there are no rigorous theories that predict the influence of size on the type of diversification, since large firms are typically unrelated conglomerates, size may be associated with unrelated diversification. SIZE should have a positive sign. Capital expenses We use net fixed assets per unit of total assets in 1980 to proxy for capital expenses (CAPEXP) as has been done by Barton (1988). CAPEXP is expected to be associated with future related diversification and should have a negative sign. Level of previous diversification The initial levels of diversification are given by DW81. We have no a priori expectations of the sign of this variable. Summarizing, we use OLS to estimate the following model: DELTADW =

BKLOG(+), RSL(-), ASL(-), RLSTK(-), DEMKT(-),CR(+), TOTRISK(+), SIZE(+), CAPEXP(-), DW81(?)

The results of the regression models for the full and the stratified sample are given in Table 2. The overall findings about the resource variables are supportive of the theory. All coefficients have the expected signs and, except those of BKLOG, CR and RLSTK, all are significant. The coefficients on RSL (pcO.Ol), ASL (p — O w π h í o« o: esodo o —o o d o I S I II l i l i — »fi II II e·· + + I c·. +1 I I + I

% *u Ou ^ Ê & S ocxtu SO sS xΕ OÜO abed; profit has increased by an amount equal to aeig - abed. The major reason for increased profit is that the LRAC curve is shown to decrease significantly over the range of output considered due to scale economies. In short, differentiation allows the firm to attain a low-cost position. The situation discussed here will not always hold. The ability of differentiation to help the firm achieve a low-cost position depends on two factors: the extent to which expenditure on differentiation significantly increases demand, shifting the demand curve to the right, and the extent to which significant reductions in unit costs arise from increasing volume. Both of these factors are situation dependent, and they form the basis for the contingency framework discussed here.

nities for differentiation. Even a homogeneous product can be differentiated if the product has different uses, if it is sold to different user groups, if the psychosocial characteristics of consumers within or across user groups are diverse, or if a combination of these conditions exists. In short, the diversity of user characteristics is of critical importance. For example, it is common to find firms differentiating their product between industrial and domestic users. Similarly, it might be possible to successfully differentiate a relatively homogeneous product if the psychosocial characteristics of consumers within a given user group are diverse. Competitive Nature of the Product Market Environment The competitive nature of the product market environment moderates the relationship between differentiation expenditure and demand. There are two critical and significant contingent factors: market structure and the stage of product market evolution (Hofer, 1975). Industrial organization economics sees market structure as a key determinant of competition. Other things being equal, oligopolistic markets tend to be characterized by substantial nonprice competition (by attempts to differentiate the product). A major reason for this is the fear of price wars in circumstances in which conditions of oligopolistic interdependence apply. There is empirical support for this proposition. The evidence suggests that an inverted-U-shaped relationship exists between a major determinant of market structure, concentration ratios, and differentiation. Cable (1973) and Sutton (1974) both found an inverted-U-shaped relationship between advertising to sales ratios and concentration ratios for samples of consumer goods industries (in consumer goods industries, advertising tends to be the most obvious expression of differentiation). Advertising intensity reached a peak in oligopolistic and duopolistic markets, before declining in monopolistic markets. Thus, the evidence suggests that often differentiation intensity is greatest in oligopolistic markets.

Differentiation and Demand: Contingencies The impact that differentiation has on demand depends on three major contingencies: the ability of the firm to differentiate its product, the competitive nature of the product market environment, and the commitment of consumers to the products of rival firms. Ability of the Firm to Differentiate Its Product The ability of the firm to differentiate its product is itself a function of two contingent factors: product characteristics and user characteristics. Any product can be viewed as a bundle of different characteristics or attributes (Lancaster, 1966). Attributes can be varied in quantity and/or combined in different ways to differentiate a product. The number of attributes inherent in a product creates scope for differentiation. Relatively homogeneous products, such as bulk chemicals, have few attributes and offer little scope for differentiation. More complex products, such as motor cars, contain many attributes and offer greater scope for differentiation. However, there is not a direct linear relationship between number of attributes and opportu404

Differentiation and Low Cost However, although efforts to differentiate appear to be greatest in oligopolies, it does not follow that differentiation has the greatest impact on quantity demanded within oligopolistic markets. Indeed, within established oligopolies, the opposite may hold. Given an absence of significant price competition, differentiation may be necessary simply to maintain current levels of demand. Differentiation effort by major players may cancel each other's effort. Hence, differentiation will not increase market share enough to enable the firm to realize substantial cost economies. By way of contrast, in fragmented markets, differentiation can have a substantial impact on quantity demanded, thereby enabling the firm to realize cost economies. In fragmented markets, an absence of major players means that the canceling out effect does not apply. Thus, other things being equal, as we move from fragmented to oligopolistic markets, the impact differentiation expenditure has on demand declines. There is a difference, however, between an established oligopoly in which market shares are stable and opportunities for expanding volume are limited and an oligopoly that is experiencing rapid growth. This brings us to the second contingency, the stage of product market evolution. Although the stage of product market evolution and the product life cycle (PLC) are not the same (a product or a product group may have its own life cycle within a more broadly defined product market), the two concepts are closely related, and often they are treated as synonymous for analytical purposes. Hofer (1975) suggested that the product life cycle may be the most important contingency variable in formulating a business strategy. There is evidence to support his idea (Anderson & Ziethaml, 1984; Hambrick & Lei, 1985; Macmillan, Hambrick, & Day, 1982), although it is questionable whether the product life cycle is the most important contingency variable (Hambrick & Lei, 1985). PLC concepts suggest that competitive forces will be weaker in emerging or growth markets than in mature or declining markets, where 405

463

growth is either slow or negative. Therefore, in growth markets, successful differentiation is likely to have a positive impact on volume and, hence, on the ability of the firm to realize scale economies. Conversely, in slow or negative growth markets, one firm's expansion must take place at the expense of another firm's market share. This intensifies competitive forces, makes expansion more difficult, and attenuates the ability of the firm to increase volume and realize cost economies through differentiation. The impact that the product market environment has on the ability of the firm to expand demand through differentiation is summarized in Figure 2. This predicts that the impact that differentiation expenditure has on demand will be greatest in fragmented market structures characterized by high growth (Cell 1), and it will be least in low-growth oligopolistic markets (Cell 4). Commitment of Consumers to the Products of Rival Firms The costs of switching products and consumer brand loyalty for the products of rival firms are other factors that determine the extent to which differentiation can be used to increase demand. If the costs to users of switching to the firm's product are high, the impact that differentiation has on demand will be attenuated. The costs of switching products are likely to be greater when users have molded operations to a competitor's product and must bear reorganization or adaptation costs to switch to the firm's product. For example, Apple Computers faced a formidable problem switching personal computer users from the IBM PC to the Macintosh because of their high level of dependence upon PC-based software, applications, and networks. Consumer loyalty for the products of rival firms is likely to be a function of many of the factors already discussed. Of particular importance, however, is the stage of product market evolution. As noted, efforts to differentiate peak within mature oligopolistic markets; therefore, other things being equal, consumer loyalty for the

464

C. W. L. Hill

Stage of

Cell 1

Cell 2

High

High

Medium

Growth

Impact

Impact

Cell 3

Cell 4

Stage

Industry Evolution Low

Medium

Low

Growth

Impact

Impact

Fragmented

Oligopoly

Stage

Market Structure

Figure 2. The impact of the product market environment on the ability of differentiation to increase demand· products of rival firms will be greatest at this stage. In addition, if the industry's product is perceived by consumers as a commodity (it has few attributes), then brand loyalty is likely to be lower.

Economies Due to Learning Effects Researchers often assume that learning effects persist over time (Boston Consulting Group, 1971). However, evidence suggests that learning effects are greatest during the start-up period associated with a new plant or process (Alchian, 1963; Baloff, 1966; Hall & Howell, 1985) and that they decline and die out once a certain cumulative output is reached. Any observed cost decline in an experience curve after such a point, therefore, must be due to effects other than those associated with learning (e.g., economies of scale). Given the nature of learning effects, the two major determinants of their importance are the age and the complexity of the manufacturing or service process used by an organization. The potential to realize learning effects will be greater in the case of a new process than in the case of an established process. Similarly, the more complex or variable a process, the greater the learning effects. These factors are summarized in Figure 3. Learning effects will be most significant in the case of new and complex processes. In these circumstances, investment in differentiation to increase market share will have the greatest downward impact on unit costs. Learning effects also

Potential to Reduce Costs Apart from demand, for differentiation to be a means of establishing a low-cost position, there must be a significant decline in costs while output increases. Only when the increase in costs due to differentiation is outweighed by cost reductions associated with expanding volume can differentiation be seen as a way of achieving a low-cost position. In this connection, three sources of cost economies are relevant: economies due to learning effects, economies of scale, and economies of scope. Both economies due to learning effects and economies of scale have been identified as major underlying components of the experience curve phenomenon (Amit, 1986; Hall & Howell, 1985). To a large extent, the significance of cost leadership as an independent strategy is based on the presumption that the experience curve declines continually with accumulated output over time, a view popularized by the Boston Consulting Group (BCG) (1971). 406

Differentiation and Low Cost

New Process

Cell 1

Cell 2

Significant Learning Over Short Time Period

Significant Learning Over Long Time Period

Age of Process Established Process

Cell 3

Cell 4

No Significant Learning

No Significant Learning

Low Complexity

High Complexity

465

Complexity of Process

Figure 3. Learning effect a and proceii.

will be significant in the case of new processes, even if they are routine and involve low variability. However, in this case, the duration of the learning period will be shorter. If a process has been long established, learning effects will be negligible no matter how complex or variable the process.

than MES are slight. Third, in many industries, MES is reached at low levels of market share. For example, Table 1 summarizes some of the results of a study by Scherer et al. (1975). This shows that MES, as a percentage of the market, was less than 5 percent for all but two of the industries examined, suggesting that there are only limited plant-level scale economies associated with pursuing volume. Moreover, the cost disadvantages of operating at one-third of MES were less than 10 percent for the majority of industries that were considered. In order to put plant-level scale economies in perspective, consider that a 10 percent cost disadvantage often can be overcome just as easily through more efficient management as by increasing operating scale.

Economies of Scale After learning effects have been exhausted, further experience effects can be reaped only from economies of scale. There are two sources of scale economies: the plant level and the firm level. The concept of minimum efficient scale (MES) defines the minimum plant size necessary to realize plant-level scale economies. Empirical evidence suggests the following about the attainment of plant-level scale economies (Pratten, 1971; Scherer, Beckenstein, Kaufer, & Murphy, 1975; Silberston, 1972; Walters, 1963). First, once MES is reached, little in the way of additional cost reductions from plant-level scale economies is possible. Second, in many industries, the cost disadvantages of operating at substantially less

The limited extent to which plant-level scale economies can be used and the asymptotic nature of learning curves suggest that the experience curve, as defined by the BCG, does not continually decline: Instead, it bottoms out. After it bottoms out, expanding output will not produce further cost reductions. In this connection, Hall and Howell (1985) argued that the BCG's demonstrations of continually declining experi407

466

C. W. L. Hill

Table 1 Minimum Efficient Scale (MES). Market Share, and Cost Disadvantages of Operating at Less than MES

Industry Brewing Cigarettes Fabrics Paints Petroleum Refining Shoes Glass Bottles Cement Steel Bearings Refrigerators Storage Batteries

MES a · % of U.S. Market

% Increase In Cost· at 1/3 MES

3.5 6.5 0.2 1.4 1.9 0.2 1.5 1.7 2.6 1.4 14.1 1.9

5.0 2.2 7.6 4.4 4.8 1.5 11.0 26.0 11.0 8.0 6.5 4.6

Note. From The Economics oí Multiplant Operations (pp. 80, 94) by F. M. Sherer, A. Beckenstein, E. Käufer, and R. D. Murphy, 1975, Cambridge, MA: Harvard University Press. Copyright 1975 by Harvard University Press. Reprinted by permission.

enee curves can be largely explained by spurious correlations. If the experience curve does bottom out, the potential for using differentiation to achieve a low-cost position would be limited to situations in which new processes are being introduced. However, this view ignores the impact that volume has on firm-level scale economies. Firms can exploit firm-level scale economies in marketing, buying, distribution, finance, and so forth as well as economies from multiplant operations (Prais, 1976; Scherer et al., 1975). Unfortunately, there is little evidence about the significance of firm-level scale economies. In one such study, Prais (1976) showed that the cost of capital for large firms is markedly lower than for smaller enterprises, suggesting that substantial financial economies exist at the firm level. Prais attributed this to the reduced risk that investors attach to lending to very large firms. More generally, Scherer et al. (1975) compiled evidence (interviews) on the number of multiplant

operations at MES a firm must have in order to realize firm-level scale economies. This evidence indicates that when firm-level scale economies are accounted for, the requisite market share necessary to realize the majority of cost saving does increase markedly in some industries, but not at all in others. For example, in Table 1, the output necessary in the beer industry to reach MES was only 3.5 percent of the U.S. market. However, Scherer et al. found that for this industry to realize firm-level scale economies, between three and four MES plants were needed and a 10-14 percent market share. Given that the avera g e market share of the three leading firms at the time of the study was 13 percent, this suggests that scope existed for medium-sized firms to use differentiation as a means of expanding volume to realizing firm-level scale economies, thus matching dominant firms in terms of cost position. On the other hand, no significant firmlevel scale economies were identified for some of the industries in Table 1 (e.g., paints, cement, ordinary steel, batteries). This suggests that many firm-level scale economies are industry dependent. Hence, the ability of the firm to use differentiation to establish a low-cost position depends on the extent to which firm-level scale economies can be realized within that industry.

Economies Oi Scope For a range of goods, economies of scope imply a potential for sharing resources, which reduces the economic cost of producing them (Teece, 1980). This is relevant to the present debate because a common differentiation strategy involves the manufacture of a product line that serves several market segments. Economies of scope can reduce the costs of differentiation by product line. In terms of Figure 1, the rise in the LRAC curve due to differentiation would be attenuated. Indeed, in extreme cases, the LRAC curve might fall due to economies of scope. If this occurred, differentiation would then be the way of establishing a low-cost position. Unfortunately, although theory about economies of scope has been well developed, there 408

Differentiation and Low Cost have been no attempts to systematically test for economies of scope. It is not possible, therefore, to reach any conclusions concerning their overall significance. All that can be said at this juncture is that for firms whose differentiation strategy involves manufacturing a product line, economies of scope are a potentially important determinant of the extent to which differentiation can be used to establish a low-cost position.

467

some source, whether it is from learning effects, economies of scale, or economies of scope. It is also possible to identify specific industry environments within which the strategy is more likely to succeed. The importance of high market growth and a fragmented market structure on the demand side and learning effects on . the cost side suggest that differentiation can be a strategy for achieving a low-cost position in emerging industries. As markets mature, differentiation becomes less relevant for a firm to achieve a low-cost position. Typically, in mature markets, growth is low, the structure is oligopolistic, brand loyalty is well established, and processes are well developed. These factors attenuate the impact of the strategy. However, even in mature markets, opportunities may arise for a firm to use differentiation in order to achieve a low-cost position: Technological change that leads to new processes can make learning effects significant again. If this occurs, the firm may use differentiation to facilitate movement down the new experience curve to a position of competitive advantage.

Bringing the Concepts Together By pulling together the concepts discussed so far, some general conclusions can now be reached about the issues raised at the beginning of this article. First, the contingencies under which it might be feasible to use differentiation as a means for achieving a low-cost position can be identified. Second, it is now possible to show when a sustained competitive advantage might be based on the simultaneous and continuous implementation of both low-cost and differentiation strategies. Differentiation to Achieve Low Cost Differentiation is most consistent with achieving a low-cost position under the following circumstances: when the firm's ability to differentiate the product is high, when consumers' commitment to the products of rival firms is low, when market growth is high, when market structure is fragmented, when the production process is new and complex, when economies of scale (particularly firm-level) are present, and when economies of scope exist. It is not necessary that all of these contingencies exist concurrently for the strategy to succeed. However, some contingencies are critical if the strategy is to work. Specifically, if the firm's ability to differentiate the product is low, if switching costs are high, if the production process is well established, and if economies of scale and scope are negligible, the strategy will not work. Thus, at a minimum, it must be possible to differentiate the product, switching costs must be reasonable, and there must be the potential for cost reduction from 409

Simultaneous Emphasis on Differentiation and Low Cost When the firm uses differentiation to achieve a low-cost position, it is emphasizing both strategies. However, this is not the only situation in which differentiation and low cost are emphasized simultaneously. The firm may stress both differentiation and low cost even when the impact that differentiation has on demand is negligible and when the firm has already achieved a low-cost position. This occurs when the low-cost position gained by the firm is no/ unique. A presumption underlying the notion of a lowcost strategy is that it is possible for a firm to become the overall cost leader in an industry. In other words, that there is a unique cost leadership position that can be occupied by only one firm at any given time. This presumption is based upon the notion of a continually declining experi-

468

C. W. L. Hill

enee curve. But, in many situations, the continually declining experience curve is fiction: Learning effects die out, plant-level scale economies can be exhausted, a n d firm-level scale economies seem to be industry dependent. This suggests that in many industries there is no potential for a firm to gain a unique overall low-cost position. A more realistic view is that there is some threshold level of market share necessary in order for a firm to gain all substantial economies from learning a n d scale effects. Moreover, this threshold level might be relatively low, such that several firms in a n industry could have a minimum-cost position. Hence, there may be a group of firms in a n industry that have realized significant economies from scale a n d learning effects. If this is the case, then how can individual firms in the low-cost group establish a sustainable competitive advantage over other firms in this group? One way by which individual firms in the lowcost group could still gain additional cost economies, despite the exhaustion of scale a n d learning effects, is through differentiation by product line to achieve economies of scope. As noted above, in extreme cases, economies of scope may be substantial enough to produce a downward shift in the firm's LRAC curve. However, an overall low-cost position created in this way is still not unique. By similarly differentiating their product lines, other firms also could benefit from the same economies of scope. Soon the competitive advantage of the firm would be eroded, a n d the problem of how to sustain a competitive advantage would remain. A partial answer to this dilemma is found in Hambrick's (1983) suggestion that three dimensions underlie Porter's scheme: efficiency, differentiation, and scale/scope (the third dimension should probably b e scale/scope/learning). Even when economies from scale, scope, or learning have been exhausted, a firm can still become a cost leader simply through efficiency. However, efficiency is not so much a strategy as a function of the skill with which a firm m a n a g e s the process of converting inputs into outputs. Effi-

ciency is sustainable, but it is also imitable. Management skills a r e not firm-specific; they can be bought on the open market. Although a firm can gain a competitive edge by being more efficient, efficiency, over the long run, can be matched by other firms. By w a y of contrast, differentiation creates something that is specific to the firm. It creates brand loyalty for consumers that, once established, can take on the characteristics of a durable asset. Therefore, because differentiation is based upon firm-specific skills a n d creates a durable asset, it is more difficult to imitate. Hence, differentiation can form the basis of a sustainable competitive advantage when all significant cost economies have been exhausted. In sum, it is suggested that once a firm h a s achieved a minimum-cost

posilion, a n d efficiency

among competing firms is equal, it can gain a sustainable competitive advantage only through some form of differentiation. The extent to which it is possible for a firm to do this without simultaneously jeopardizing the advantages of a minimum-cost position depends on how price-sensitive consumers a r e (the extent to which brand loyalty effects outweigh price effects). For industries in which price sensitivity is low, successful differentiation, and the extra price it entails, will not jeopardize market share sufficiently to threaten the firm's minimum-cost position. Thus, for a group of firms that has achieved minimum costs, a sustained competitive advantaged might be based on both differentiation a n d maintaining a minimum-cost position. Anecdotal evidence suggests that this is the case for many mature oligopolistic industries. In the auto industry, for example, major firms have differentiated their products a n d at the same time have not jeopardized their low-cost position. Indeed, once differentiation becomes a n industry norm, then failure to differentiate by a firm may result in a declining market share and the loss of scale economies. Hence, differentiation may become the way a firm maintains its scale economies a n d safeguards its market share. Far from being incompatible, the simultaneous pursuit of both 410

Differentiation and Low Cost differentiation and low cost may be necessary to both establish and maintain a sustained competitive advantage.

469

the potential to differentiate the product. The second environment is within mature industries that are experiencing significant technological change because the implied change in process gives rise to new learning economies. More generally, the simultaneous pursuit of differentiation and low-cost strategies is most likely to be consistent with superior performance in mature industries where all experience curve economies have been exhausted and several firms have achieved a minimum-cost position. Some important implications for managers and researchers follow from these conclusions. Managers need to recognize that differentiation can be a way of achieving low cost and that a sustainable competitive advantage frequently involves the simultaneous pursuit of differentiation and low-cost strategies. The contingency framework that is developed here and summarized in Table 2 provides the basis for identifying when these two conditions hold. In addition, researchers need to consider how the simultaneous pursuit of both strategies might have an impact on the firm's profitability. Table 2 can provide the beginning point for such future empirical research.

Conclusion This article has developed a framework that identifies the contingencies under which differentiation and low-cost strategies are compatible. These contingencies and the relative importance they hold for firms are summarized in Table 2. Two main conclusions follow from the discussion presented here. First, differentiation can be a way of achieving a low-cost position. Second, because there often is no unique low-cost position, a firm may have to base its sustainable competitive advantage on the simultaneous and continuous pursuit of both low cost and differentiation. Table 2 suggests two major industry environments within which differentiation might be used to achieve a low-cost position. The first environment is within emerging industries that are characterized by high growth, that have significant learning and scale economies, and that have

Table 2 A Summary of the Important Contingencies Affecting (he Compatibility oí Differentiation and Low Coti Major Contingency

Secondary Contingencia·

Comment·

Ability to differentiate

User characteristics Product characteristics

Diversity of users necessary for differentiation.

Commitment of users to products of rival firms

Switching costs Brand loyalty

Brand loyalty greatest in mature oligopolies.

Product market environment

Market structure Product market evolution

High growth state of PLC most important secondary contingency.

Learning effects

Age of process Complexity of process

Age of process very important: New processes imply significant learning.

Economies of scale

Plant level Firm level

Plant level often exhausted at low market share; firm level may be more significant. Dependent upon breadth of product line. May be important.

Economies of scope

411

470

C. W. L. Hill

References MacMillan, 1. C., Hambrick, D. C., & Day, D. L. (1982) The product portfolio and profitability: A PIMS based analysis of industrial product businesses. Academy of Management Journal, 25, 733-755.

Alchlan, A. (1963) Reliability of progress curves in airframe production. Econometrica. 31, 679-693. Amit, R. (1986) Cost leadership strategy a n d experience curves. Strategic Management/ournoJ, 7, 281-292.

Phillips, L. W., Chang, D., & Buzzell, R. D. (1983) Product quality, cost position and business per/ormance.- A test of some key hypotheses. Unpublished manuscript, Harvard University, Harvard Business School, Boston.

Andereon. C. R . & Ziethaml, C. P. (1984) Stage of the product life cycle, business strategy, and business performance. Academy of Management Journal, 27, 5-24. Baloff, H. (1966) The learning curve: Some controversial issues. Journal of Industrial Economics, 14, 275-282. Boston Consulting Group (1972) Perspectives on Boston: Author.

Porter, M. E. (1980) Competitive Press.

experience.

strategy. New York: Free

Porter, M. E. (1985) Competitive advantage. New York: Free Press.

Cable, ). (1973) Advertising a n d market structure. In K. Cowling (Ed.), Market structure and corporate behavior (pp. 105-124). London: Macmillan.

Prais, S. Ï. (1976) The evolution of giant firms in Britain. Cambridge, England: University of Cambridge Press.

Dess, G. G.. & Davis, P. S. (1984) Porter's generic strategies a s determinants of strategie group membership and organizational performance. Academy of Management journal, 27, 467-488.

Pratten, C. F. (1971) Economies of scale in manu/acfuring industry. Cambridge, England: University of Cambridge Press.

Hall, G., 8t Howell, S. (1985) The experience curve from the economists' perspective. SfrafegJc Management Journal, 6, 197-212.

Scherer, F. M., Beckeristein, Α., Käufer, E., & Murphy, R. D. (1975) The economics of muItiplant operations. Cambridge, MA: Harvard University Press.

Hall, W. K. (1980) Survival strategies in a hostile environment. Harvard Business Review, 58(5), 75-85.

Silberston, Ζ. Α. (1972) Economies of scale in theory a n d practice. Economic Journal, 82, 369-391.

Hambrick, D. C. (1983) High profit strategies in mature capital goods industries: A contingency approach. Academy of Management Journal, 26, 687-707.

Sutton, C. J. (1974) Advertising, concentration a n d competition. Economic Journal, 84, 56-69.

Hambrick, D. C., & Lei, D. (1985) Towards a n empirical prioritization of contingency variables for business strategy. Academy of Management Journal, 28, 763-788.

Teece, D. ). (1980) Economies of scope and the scope of the enterprise. Journal of Economic Behavior a n d Organization, 3, 223-247.

Hofer, C. W. (1975) Towards a contingency theory of business strategy. Academy of Management Journal, 18, 784-810.

Walters, A. A. (1963) Production and cost functions. Econometrica, 31, 1-66. White, R. E. (1986) Generic business strategies, organizational context and performance: An empirical investigation. Strategic Management Journal, 7, 217-231.

Lancaster, K. J. (1966) A new approach to demand theory. Journal of Political Economy, 74, 132-157.

Charles W. L. Hill is Assistant Professor of Management, University of Washington, Seattle. Correspondence regarding this article can be sent to him at the Department of Management, Graduate School of Business Administration, University of Washington, Seattle, WA 98145. This article was written while he was Assistant Professor of Management at Michigan State University.

412

11 Kollektive Strategien

Eine kollektive Strategie ist eine von mehreren Unternehmungen gemeinschaftlich durchgeführte, systematische Vorgehensweise, die der Stabilisierung und Beherrschung interdependenter Aufgabenumwelten dient (Bresser, 1989: 545). Grundsätzlich können kollektive Strategien, genau wie unabhängig verfolgte Strategien, auf den drei Ebenen der Gesamtunternehmung, der Geschäftsbereiche und der Funktionsbereiche realisiert werden. In der Praxis besteht eine Häufung auf der Geschäftsbereichsebene, so daß eine kollektive Strategie als eine besondere, auf Kooperation beruhende Wettbewerbsstrategie angesehen werden kann. Kollektive Strategien werden insbesondere dann angestrebt, wenn Unternehmungen befürchten, daß sie sich durch im Alleingang verfolgte Wettbewerbsstrategien in ihren komplexen und dynamischen Umwelten nicht behaupten können (Astley & Fombrun, 1983; Bresser & Harl, 1986; Bresser, 1988; 1989). Zur Realisierung einer kollektiven Strategie dient die Beherrschungsstruktur der Unternehmungskooperation. Oft wird für den Begriff der Unternehmungskooperation der der „Strategischen Allianz" synonym verwendet (Ohmae, 1989; Backhaus & Plinke, 1990). Es gibt eine Vielzahl von Kooperationsformen und Versuchen, diese zu klassifizieren (Hammes, 1994: 25; Chung, 1997: 18-25). In der Literatur scheint sich die Unterscheidung zwischen „equity alliance" (Kapitalbeteiligungsallianz) und „non-equity alliance" (Sonstige Allianz) durchzusetzen (Hennart, 1988; Barney, 1997: 285). Eine „non-equity alliance" beruht auf einem Vertrag, in dem rechtlich selbständige Kooperationspartner vereinbaren, gemeinschaftlich Ressourcen aufzubringen und zu nutzen, um Produkte oder Dienstleistungen zu entwickeln, zu erstellen oder zu vertreiben oder andere strategische Ziele zu verfolgen.72 Diese Form der Allianz umfaßt Lizensierungen, langfristige Liefer- und Abnahmeverpflichtungen, die Bereitstellung technischer Assistenz, den Abschluß von Managementverträgen u.ä. (Hennart, 1988: 363). Eine „equity alliance" ist ebenso wie die „non-equity alliance" definiert; der Unterschied besteht darin, daß die kooperativen Arrangements zusätzlich durch Kapitalbeteiligungen gekennzeichnet sind. „Equity alliances" entstehen z.B., wenn sich eine Unternehmung an einer anderen beteiligt oder wenn beide (alle) Kooperationspartner sich gegenseitig untereinander beteiligen. Darüber hinaus zählen Joint-ventures zu den „equity alliances". Im Fall des Joint-ventures gründen Kooperationspartner eine rechtlich selbständige Unternehmung, an der alle Partner Kapitalbeteiligungen erwerben. Die Motive für das Verfolgen kollektiver Strategien und für das Eingehen von Unternehmungskooperationen sind vielfältig (Contractor & Lorange, 1988; Barney, 1997: 286294). Als übergeordnetes Motiv kann die Schaffung eines gemeinsamen Wertes angesehen werden (Zajac & Olsen, 1993), der höher ist als der Wert, der sich erzielen ließe, wenn die Kooperationspartner ihre Ressourcen im Alleingang zur Verfolgung strategi72

Andere strategische Ziele können sich z.B. auf die gemeinschaftliche Durchführung von Grundlagenforschung beziehen.

472

R. Κ. F. Bresser

scher Ziele einsetzen würden. Chung (1997: 31-34) führt die große Zahl der in der Literatur auffindbaren Argumente für das Eingehen von Unternehmungskooperationen auf vier Bereiche zurück: (1) die Realisierung von Skalenvorteilen und Lernkurveneffekten, (2) den Zugang zu und die gemeinsame Nutzung von technologischen Fachkenntnissen und sonstigen Fähigkeiten sowohl in bisherigen als auch in neuen, z.B. internationalen Märkten, (3) die Senkung (Teilung) der Investitionsrisiken und (4) die Einflußnahme auf die Intensität des Wettbewerbs73.74 Eine Gemeinsamkeit hinsichtlich der meisten in der Literatur vorfindbaren Aussagen über die Vorteilhaftigkeit von Unternehmungskooperationen fällt auf: Sie entfalten ihre Argumentationskraft insbesondere im Hinblick auf komplexe, dynamische und auch hyperkompetitive Umweltbedingungen, wie sie für viele Märkte der Gegenwart kennzeichnend sind (Chung, 1997: 34; D'Aveni, 1994: 333-341). Kooperationen und die durch sie angestrebten kollektiven Strategien sind Mittel des Managements der Interdependenzen in komplexen, dynamischen Aufgabenumwelten (Bresser & Harl, 1986: 408). Wie die Motive, so sind auch die theoretischen Ansätze zur Erklärung des Entstehens und des Erfolges von Unternehmungskooperationen vielfältig. Während der letzten zehn Jahre kam es zu einer explosionsartigen Zunahme an Kooperationsstudien und -abhandlungen, so daß manche Autoren den gegenwärtigen Stand der theoretischen Durchdringung des Forschungsgegenstandes als chaotisch bezeichnen (Osborn & Hagedoorn, 1997: 261). Um die Vielfalt der Abhandlungen etwas zu systematisieren, werden die für das Strategische Management bedeutenden theoretischen Ansätze in zwei Bereiche aufgeteilt: in transaktionskostentheoretische und managementtheoretische Erklärungen.

Transaktionskostentheoretische Erklärungen Auf der Basis des Transaktionskostenansatzes ist die Erklärung von Unternehmungskooperationen natürlich auf den Aspekt der Transaktionskosteneffizienz ausgerichtet. In dem Bemühen, die Gefahr der opportunistischen Ausbeutung zu minimieren, ist die Kooperation dann als Beherrschungsstruktur auszuwählen, wenn sie geringere Transaktionskosten verursacht als der Markt oder die Hierarchie. 73

74

Die traditionelle Industrieökonomik hat Unternehmungskooperationen gemeinhin unter dem Begriff der „Kollusion" subsumiert, d.h. sie als Versuche zur Einschränkung des Wettbewerbs abqualifiziert (vgl. z.B. Scherer, 1980: 169-197). Dabei wird übersehen, daß Kooperationen, auch zwischen Konkurrenten, erhebliche positive wohlfahrtsökonomische Effekte zeitigen können. Sinn und Zweck vieler Kooperationen ist nachgerade die Entwicklung innovativer Produkte und Produktionsprozesse, die die Konsumenten in Form ganz neuartiger Produkte und Dienstleistungen, einer verbesserten Produktqualität und/oder in Form niedrigerer Verkaufspreise erreichen (Teece, 1992; Hammes, 1994: 117; Shan et al., 1994; Young et al., 1996). Hagedoorn (1993) kann in einer mehrere tausend weltweite Kooperationen umfassenden Studie feststellen, daß im Zeitraum von 1980 bis 1989 die weitaus meisten Kooperationen durch das zweite Motiv des Zugangs zu und der Verwertung von technologischen Fähigkeiten bestimmt waren. Kostenund Risikoerwägungen (hier das erste und das dritte Motiv) spielten demgegenüber eine deutlich geringere Rolle. Über die Relevanz des vierten Motivs lassen sich kaum empirische Aussagen gewinnen, da dieses Motiv der Kollusion am nächsten steht und somit von den Beteiligten nicht publik gemacht wird.

Kollektive Strategien

473

Williamson hat sich erst Anfang der 90er Jahre darum bemüht, Unternehmungskooperationen in seine transaktionskostentheoretischen Analysen zu integrieren. Allerdings haben andere Autoren transaktionskostentheoretische Begründungen der Kooperation bereits in den 80er Jahren unternommen. Die Argumentationslinie dieser frühen Erklärungsversuche verläuft in etwa wie folgt: Wie bei der vertikalen Integrationsentscheidung wird explizit (Kogut, 1988) oder implizit (Hennart, 1988; Stuckey, 1983) davon ausgegangen, daß die Entscheidungssituation durch hohe Faktorspezifität und hohe Unsicherheit gekennzeichnet ist.75 An sich böte sich die Integration als Lösung an. Dennoch sind Joint-ventures oder andere Kooperationsformen unter folgenden Bedingungen effizienter und der Integration vorzuziehen: Erstens, technologische Unteilbarkeiten aufgrund von Economies of Scale behindern eine Integrationslösung (Stuckey, 1983).76 Zweitens, eine Unternehmung möchte nur einen Teil der Assets einer anderen Unternehmung erwerben. Wird in diesen Situationen eine Akquisition durchgeführt, müssen nicht gewollte Assets verwaltet oder veräußert werden. Die Verwaltung nicht benötigter Assets verursacht Kosten, die Veräußerung scheitert häufig an der Unternehmungsspezifität der Assets (Hennart, 1988; Kogut, 1988).77 Drittens, wenn gewünschte Assets intangibler Natur (durch „tacitness" gekennzeichnet) sind, kann der Wert dieser Assets oft ex ante nicht bestimmt werden. Die Kooperation ist auch in diesem Fall effizienter als die Integration, denn man vermeidet Opportunismusprobleme, z.B. einen zu hohen Preis für ein schwer zu bewertendes Akquisitionsobjekt zu entrichten, dessen Wert der Verkäufer des Assets zu inflationieren versucht (Hennart, 1988).78 Williamson (1991a; 1991b) integriert Unternehmungskooperationen als Hybridform(en) in sein Kontinuum zwischen Markt und Hierarchie. Er argumentiert, daß sich hybride Arrangements insbesondere dann als effizient erweisen, wenn die Entscheidungssituation durch mittlere Faktorspezifität79 und geringe bis mittlere Unsicherheit gekenn75 76

77 78 79

Ähnliche Argumente finden sich bei Weder (1990). Ein Beispiel hierfür ist die Produktion von Aluminium. Die Herstellung erfolgt in drei vertikal angeordneten Produktionsstufen: In Bauxitminen wird das Mineral Bauxit gewonnen, das in einer Raffinerie zu Aluminiumoxyd verarbeitet wird; schließlich wird das Aluminiumoxyd in Hütten zu Aluminium verschmolzen. Da Bauxit ein sehr heterogenes Mineral ist, müssen Raffinerien ihre Technologie genau auf die Eigenschaften einer bestimmten Bauxitmine abstimmen. Dieser Umstand macht Raffinerien für opportunistisches Verhalten seitens der Betreiber der Bauxitminen anfällig. Die hohe Faktorspezifität erfordert deshalb eine vertikale Integration. Allerdings sind die verschiedenen Produktionsstufen in der Aluminiumindustrie durch unterschiedliche mindestoptimale Betriebsgrößen gekennzeichnet, ein Umstand, der die Integrationslösung erschwert. Die optimale Betriebsgröße einer Bauxitmine ist etwa zweieinhalbmal so groß wie die einer Raffinerie und fünfmal so groß wie die einer Aluminiumhütte. Wollte eine Aluminiumhütte rückwärtsintegrieren, müßte sie gleichzeitig ihre eigenen Kapazitäten verfünffachen. In Zeiten langsamen Wachstums könnte die vertikale Integration deshalb zu Überkapazitäten und teuren Preiskämpfen führen. Um sich die Möglichkeit zu bewahren, die Kapazitäten schrittweise und maßvoll zu erhöhen, erschließen Aluminiumhersteller typischerweise gemeinsam eine Mine, und sie errichten gemeinsam Raffinerien und Hütten durch Joint-ventures (Stuckey, 1983: 150-156). Eine empirische Bestätigung dieses Arguments liefern Hennart und Reddy (1997). Ähnlich argumentieren auch Balakrishnan und Koza (1993). Ein Beispiel für ein Gut mittlerer Spezifität ist eine (geplante) Pipeline in einer Ölförderungsregion, in der bereits einige Pipelines existieren. Die neue Pipeline könnte die Transportkosten mehrerer Ölproduzenten erheblich reduzieren. Insofern wäre der Betreiber der neuen Pipeline nach dem Bau der Anlage nicht von nur einem potentiellen Kunden abhängig, wie dies oft bei hoher Faktorspezifität der Fall ist. Auch wären die potentiellen Benutzer der neuen Pipeline nur begrenzt von dem Pipelinebetreiber erpreßbar, denn sie könnten auf die Nutzung anderer Pipelines ausweichen, die allerdings

474

R. Κ. F. Bresser

zeichnet ist. Treten unvorhergesehene Ereignisse ein, dann kann die Hybridform sowohl auf die Anreize eines Marktes zu autonomen Anpassungen (vermittelt durch Preise) als auch auf den Anpassungsmechanismus der hierarchischen Koordination (die Anweisung) zurückgreifen. Die marktlichen und die hierarchischen Anpassungsanreize sind bei der Hybridform nur mittelstark ausgeprägt. Dieser Umstand versieht die Hybridform im Falle mittlerer Faktorspezifität mit Effizienzvorteilen (Williamson, 1991a: 279286); sie ermöglicht geringere Beherrschungsstrukturkosten als reine Markt- oder Hierarchielösungen. Was die Unsicherheit der Transaktionen angeht, so verweist Williamson zunächst darauf, daß Hybridformen durch eine besondere Art von Bürokratiekosten gekennzeichnet sind: die Kosten der Konfliktlösung (Williamson, 1991a: 291f.). Aufgrund der Unabhängigkeit der an Unternehmungskooperationen Beteiligten können Konfliktlösungen und Anpassungen an Störungen nur im beiderseitigen Einverständnis erfolgen, was, verglichen mit der Hierarchie, in der Hybridform schwierig und zeitaufwendig ist.80 Steigt der Unsicherheitsgrad stark an, können die notwendigen Anpassungen nicht mehr effizient durchgeführt werden, da die Konfliktlösungen zu viel Zeit beanspruchen würden. Aus diesem Grunde sind die Hybridformen nur für besonders stabile Umweltsituationen geeignet, in denen Störungen nur selten auftreten und einvernehmliche Einigungen eine hohe Erfolgs Wahrscheinlichkeit aufweisen (Williamson, 1991a: 291 ; Chung, 1997: 69). Das Williamsonsche Modell steht offensichtlich in zweifacher Hinsicht im Widerspruch zu anderen Erklärungsansätzen: Einerseits widerspricht die Annahme mittlerer Faktorspezifität und geringer Transaktionsunsicherheit den Annahmen der frühen transaktionskostentheoretischen Erklärungsversuche: Kogut (1988) und Hennart (1988) sehen hohe Faktorspezifität und hohe Unsicherheit als Bedingungen für das Entstehen der Unternehmungskooperation an. Andererseits widerspricht die Annahme geringer Transaktionsunsicherheit der empirischen Realität: Der sprunghafte Anstieg kooperativer Vereinbarungen zwischen Unternehmungen wird ja gerade durch das Entstehen komplexer, dynamischer oder gar hyperkompetitiver Umweltbedingungen begründet (Bresser & Harl, 1986; Backhaus & Plinke, 1990; D'Aveni, 1994: 333-341; Chung, 1997: 25-34). Insgesamt ergibt sich also ein recht unklares Bild hinsichtlich der transaktionskostentheoretischen Begründungen der Kooperation. Empirische Studien, die die Annahmen der frühen Erklärungsversuche mit denen Williamsons vergleichen, liegen nicht vor.81 Es

80

81

weiter von ihren Ölförderungsanlagen entfernt wären. Um sich die effiziente Nutzung der neuen Pipeline zu sichern, ohne mit irgendwelchen Holdup-Problemen konfrontiert werden zu können, könnten die potentiellen Nutzer der neuen Pipeline beschließen, diese gemeinschaftlich, ζ. B. in der Form eines Joint-ventures, zu betreiben. Vgl. hierzu auch Pearce (1997), der unter Rückgriff auf Milgrom und Roberts (1992: 192-194) die vergleichsweise hohen politischen Beeinflussungs- und Verhandlungskosten der Kooperation thematisiert. Allerdings gibt es einige Studien, die (trotz anderer Fragestellungen) Ergebnisse erzielten, die im vorliegenden Zusammenhang interpretiert werden können und die eher die Annahmen der frühen transaktionstheoretischen Ansätze unterstützen. Dickson und Weaver (1997) stellen für eine Stichprobe von 433 norwegischen Industrieunternehmungen fest, daß der Abschluß von strategischen Allianzen signifikant positiv mit der durch die Top-Manager empfundenen Unsicherheit zusammenhängt, und zwar für mehrere Unsicherheitsmaße. Zaheer und Venkatraman (1995) berichten für eine Stichprobe von 329 Versicherungsmaklern einen signifikant positiven Zusammenhang zwischen Faktorspezifität und Kooperationsintensität und keine Zusammenhänge für wahrgenommene Umweltunsicherheiten.

Kollektive Strategien

475

verwundert nicht, daß viele weiterführende Erklärungsansätze die Schwächen des Transaktionskostenansatzes als Ausgangspunkt verwenden (Sydow, 1992: 166-168).

Managementtheoretische Erklärungen Genau genommen ist der Begriff „managementtheoretische Erklärungen" eine Fehlbezeichnung für die Erklärungsversuche, die den Transaktionskostenansatz nicht als (primären) gedanklichen Bezugsrahmen gewählt haben. Diese Erklärungsversuche verwenden Argumente aus einer Vielzahl theoretischer Überlegungen, die ihre Ursprünge außerhalb des Managementbereichs haben. Dennoch zeichnen sich alle im folgenden dargestellten Ansätze durch zwei Gemeinsamkeiten aus: Erstens, sie wurden von Wissenschaftlern entwickelt, die das Phänomen Unternehmungskooperation im Kontext des Strategischen Managements erklären wollen. Zweitens, die Ansätze versuchen, dynamische Erklärungen zu entwickeln, die die Erkenntnisbeschränkungen statischer Gleichgewichtsanalysen, wie sie z.B. dem Transaktionskostenansatz immanent sind (Williamson, 1991a: 293; Ghoshal & Moran, 1996: 40), aufheben. Im folgenden werden sieben Ansätze skizziert, die zum Teil miteinander verbunden sind und insgesamt die gegenwärtige Breite der Diskussionen gut repräsentieren.82 Strategische Motive der Kooperation. Mehrere Autoren argumentieren, der Transaktionskostenansatz könne die wichtigsten Motive und Antezedenzien der Kooperation nicht erklären, nämlich die strategischen (Zajac & Olsen, 1993; Eisenhardt & Schoonhoven, 1996). Strategische Motive umfassen beispielsweise die Realisierung von FirstMover-Vorteilen, den schnellen Markteintritt, den Eintritt in ausländische Märkte und insbesondere das Erlernen von Fähigkeiten, durch die sich dauerhafte Wettbewerbsvorteile generieren lassen (Hamel et al., 1989; Hamel, 1991; Tyler & Steensma, 1995; Doz, 1996; Powell et al., 1996). Aufgrund strategischer Motive entstandene Kooperationen können hohe spezifische Investitionen verlangen und aus transaktionskostentheoretischer Sicht wesentlich ineffizienter sein als alternative Beherrschungsstrukturen; sie werden dennoch durchgeführt, da der erwartete Wert der Kooperation den Transaktionskostennachteil kompensiert. Konsequenterweise fordern Zajac und Olsen (1993), den Transaktionskostenansatz durch einen Transaktionswertansatz abzulösen. Ein Transaktionswertansatz macht den Prozeß der Entwicklung und Verteilung eines gemeinsamen Wertes zum Gegenstand der Analyse. Die Höhe der Transaktionskosten stellt in diesem Ansatz nur ein Entscheidungskriterium dar, das neben strategischen Überlegungen während der Initiierungsphase einer Untemehmungskooperation Berücksichtigung findet. Die Kooperation als Wachstumsoption. Ein sämtliche Motive der Kooperation umfassender, dynamischer Erklärungsrahmen kann aus der Optionspreistheorie abgeleitet werden (Kogut, 1991; Bowman & Hurry, 1993; Chung, 1997). Faßt man die Kooperation als Investitionsentscheidung auf, läßt sie sich mit den Instrumenten der Optionspreistheorie präzisieren und zu den Alternativen Integration und Markt abgrenzen. Vor 82

Ein der Kooperationsproblematik zuzurechnendes Thema, das der „Strategischen Netzwerke", wird hier aus Platzgriinden nicht behandelt. Insofern ist die folgende Übersicht nicht repräsentativ für diese Diskussion. Zur Organisation und Evolution „Strategischer Netzwerke" vgl. insbesondere Sydow (1992).

476

R. Κ. F. Bresser

dem Hintergrund heterogener Unternehmungen, dynamischer Umweltbedingungen mit hohen Verhaltens-, Absatz- und Prozeßunsicherheiten sowie der Irreversibilität von spezifischen Investitionen können Unternehmungskooperationen Effekte erzielen, die jenen voller Integrationsmaßnahmen deutlich überlegen sind. Versteht man Unternehmungskooperationen als Investitionen in Wachstumsoptionen, so läßt sich nicht nur erklären, daß durch Kooperationen Kosten und Risiken geteilt werden können und gemeinschaftlich neues technologisches Wissen entwickelt werden kann. Darüber hinaus verdeutlicht der Optionsgedanke, daß die Kooperation den beteiligten Partnern erlaubt, mit der Bildung umfangreicher spezifischer Investitionen abzuwarten und damit Unsicherheit abzubauen. Erst wenn sich im Verlauf einer Kooperation herausstellt, daß sich bestimmte, angestrebte Innovationen am Markt gewinnbringend realisieren lassen, werden die Kooperationspartner sich dazu veranlaßt sehen, spezifische Investitionen in einem Umfang zu tätigen, der bei der Integrationsentscheidung schon wesentlich früher angefallen und somit der vorzeitigen Ausübung einer Option gleichzusetzen gewesen wäre. Die Eleganz der optionspreistheoretischen Begründung der Kooperation besteht darin, daß die restriktiven, der Neoklassik entstammenden Annahmen des Transaktionskostenansatzes aufgegeben werden, ohne daß wichtige Bestandteile des Ansatzes selbst, z.B. Opportunismus, Faktorspezifität und Unsicherheit, unberücksichtigt bleiben. Stabilitätsbedingungen der Kooperation. Da opportunistisches Verhalten innerhalb einer Kooperation deren Erfolg gefährden kann (Parkhe, 1993a; 1993b; Park & Ungson, 1997), beschäftigen sich mehrere Abhandlungen mit den internen Stabilitätsbedingungen der Unternehmungskooperation. Anleihen werden dabei insbesondere bei der Spieltheorie gemacht, genauer gesagt, dem Aspekt, der als „shadow of the future" bezeichnet wird: Kooperationen, die zeitlich nicht begrenzt sind, sich durch eine hohe Frequenz von Interaktionen und eine generelle Verhaltenstransparenz auszeichnen, gelten als tendenziell stabil (Weder, 1990; Dollinger, 1990; Parkhe, 1993a; 1993b). Weitere in der Literatur diskutierte stabilitätsfördernde Faktoren (mit teilweise widersprüchlicher empirischer Evidenz) umfassen: ein partizipatives Kommunikations- und Konfliktlösungsverhalten (Mohr & Spekman, 1994), Ähnlichkeiten in den verfolgten strategischen Absichten (Saxton, 1997), Kooperationserfahrungen (Kogut, 1989; Park & Russo, 1996; Park & Ungson, 1997) sowie die kulturelle Nähe der Partner (Park & Ungson, 1997; Barkema et al., 1997). Als wichtigster Stabilitätsfaktor wird das Vertrauen zwischen den Kooperationspartnern hervorgehoben. Vertrauen statt Opportunismus. Hill (1990) weist bereits daraufhin, daß die Opportunismusannahme des Transaktionskostenansatzes vermutlich überzogen ist, und insbesondere Unternehmungskooperationen eher auf der entgegengesetzten Verhaltensannahme der vertrauensvollen Zusammenarbeit beruhen. Will man gemeinschaftlich einen Wert erstellen, ist gegenseitiges Vertrauen unabdingbar; die Partner müssen sich darauf verlassen können, daß alle Beteiligten von opportunistischen Verhaltensweisen Abstand nehmen und die erwarteten Leistungen erbringen (Ring & Van de Ven, 1992; Parkhe, 1993b; Zajac & Olsen, 1993). Sämtliche oben aufgeführten internen Stabilitätsbedingungen können als Voraussetzungen für das Entstehen von Vertrauen gelten (Zucker, 1986). Insofern kann Vertrauen als der überragende Stabilitätsfaktor einer Unternehmungskooperation angesehen werden. Mehrere Forschungstraditionen haben sich mit dem Phänomen Vertrauen auseinandergesetzt, z.B. die Sozialpsychologie (Deutsch, 1973), die Spieltheorie (Axelrod, 1984), die Soziologie (Granovetter, 1985; Browning et al., 1995) und die Organisationstheorie

Kollektive Strategien

477

(Ouchi, 1980). Für die Diskussion der Unternehmungskooperation im Rahmen des Strategischen Managements sind zwei Entwicklungen bedeutsam. Zum einen gibt es Versuche, den Transaktionskostenansatz um das Konzept Vertrauen zu erweitern, um so die Wahl für eine Untemehmungskooperation besser zu begründen. Vertrauen zwischen Kooperationspartnern tritt in diesem Kontext als drittes Koordinationsinstrument neben die Anweisung der Hierarchie und den Preis des Marktes (Bradach & Eccles, 1989; Zaheer & Venkatraman, 1995; Vogt, 1997). Zum anderen wird das Konzept Vertrauen benutzt, nicht um das „warum" der Kooperation zu erforschen, sondern die Frage nach dem „mit wem". Dabei wird Bezug genommen auf den sozialen Kontext der Kooperation. Unternehmungen, die Erfahrungen mit Unternehmungskooperationen sammeln konnten, eine positive Kooperationsreputation haben und in ein soziales Netz kooperierender Unternehmungen eingebunden sind, erweisen sich als vertrauenswürdige, präferierte Partner für neue Kooperationsvereinbarungen, und sie initiieren auch selbst am häufigsten weitere Kooperationen (Gulati, 1995a; 1995b; Powell et al., 1996; Dollinger et al., 1997; Nooteboom et al., 1997). Dysfunktionen der Kooperation. Einige Autoren analysieren potentielle Dysfunktionen oder unbeabsichtigte Konsequenzen kooperativer Strategien mit Hilfe des Ressourcenabhängigkeitsansatzes der Organisationstheorie (Pfeffer & Salancik, 1978). Diesem Ansatz gemäß erfolgen Unternehmungskooperationen, um den Zugang zu kritischen Ressourcen sicherzustellen bzw., allgemeiner ausgedrückt, um die Dynamik interdependenter Aufgabenumwelten zu managen. Bei den hervorgehobenen Dysfunktionen handelt es sich nicht um die verschiedenen Ausprägungen opportunistischen Verhaltens, sondern um mögliche Nebeneffekte einer Kooperation, deren Ursachen in der durch die Kooperation begründeten Veränderung der Umweltinterdependenz liegen. Eine mögliche Destabilisierung wird also nicht so sehr in den internen Gegebenheiten einer Kooperationsbeziehung festgemacht als im externen Kontext. Konkret umfassen solche Dysfunktionen eine Verringerung der strategischen Flexibilität der Kooperationspartner, die unkontrollierbare Preisgabe strategisch sensitiver Informationen, eine Verstärkung der Einwirkungen externer Störungen und die mögliche Verursachung neuen Markteintritts (Bresser, 1984; 1988; Bresser & Harl, 1986).83 Da die Dysfunktionen einer kollektiven Strategie dazu führen können, daß Unternehmungen ihre Kooperationsbeziehungen beenden und durch unabhängige Wettbewerbsstrategien substituieren, wird insbesondere das dynamische Wechselverhältnis zwischen kooperativen und unabhängigen Wettbewerbsstrategien zu einem interessanten Forschungsgegenstand (Bresser, 1988; Lado et al., 1997). Um ihre Wettbewerbsfähigkeit dauerhaft erhalten zu können, müssen Unternehmungen in der Lage sein, wiederholt zwischen mehr Kooperationsbeziehungen und mehr unabhängigem Wettbewerb wechseln zu können (Bresser & Harl, 1986: 425). Eine solche Flexibilität scheint insbesondere vor dem Hintergrund der gegenwärtigen Auseinandersetzung mit hyperkompetitiven Umwelten (D'Aveni, 1994) geboten zu sein. Ressourcen und Kooperation. Aus ressourcenbasierter Sicht stellt sich der Wert einer Kooperation als ein Synergievorteil dar, der sich daraus ergibt, daß unabhängige Unternehmungen einen Teil ihrer Ressourcen gemeinschaftlich nutzen. Kooperationen sind besonders dann attraktiv, wenn die zum Wettbewerb benötigten Ressourcen sehr heterogen zwischen verschiedenen Unternehmungen verteilt sind und als knapp und schwer 83

Ähnlich argumentieren auch Hamel et al. (1989) und Mitchell & Singh (1996), allerdings ohne explizite Bezüge zum Ressourcenabhängigkeitsansatz herzustellen.

478

R. Κ. F. Bresser

imitierbar gelten (Jorde & Teece, 1990; Teece, 1992; Barney, 1997: 300ff.). In diesen Situationen ermöglichen Kooperationen den schnellen Zugang zu kritischen Ressourcen. Der ressourcenbasierte Ansatz hat sich erst seit wenigen Jahren mit der Unternehmungskooperation beschäftigt; insofern sind empirische Studien noch selten. Da insbesondere intangible Ressourcen als Quellen nachhaltiger Wettbewerbsvorteile gelten (Barney, 1991), kann den bereits dargestellten Gedankengängen entnommen werden, welche Ressourcen für den Erfolg von Unternehmungskooperationen einen zentralen Stellenwert haben und eine verstärkte empirische Erforschung rechtfertigen. Die Lernfähigkeit der Kooperationspartner ist vermutlich eine besonders kritische Ressource (Hamel, 1991; Mowery et al., 1996), denn Kooperationen involvieren oft den Transfer von technologischen und anderen Fähigkeiten (Hagedoorn, 1993). Das Erlernen von Fähigkeiten kann nur dann für alle Beteiligten erfolgreich verlaufen, wenn die Kooperation interne Stabilität aufweist. Insofern wären Fähigkeiten, die die Stabilität der Kooperation fördern, ebenfalls kritische Erfolgsfaktoren. Hierzu zählen z.B. die Koordinations-, Kommunikations- und Konfliktlösungsfähigkeiten (Mohr & Spekman, 1994), die Fähigkeit, vertrauensvolle Beziehungen herzustellen (Barney & Hansen, 1994) und die Fähigkeit, an Kooperationsnetzwerken teilzuhaben (Gulati, 1995a). Die Institutionalisierung der Kooperation. Die in der Literatur gerne hervorgehobenen positiven Performancekonsequenzen der Unternehmungskooperation haben sich durch die Empirie nicht eindeutig bestätigen lassen (Smith et al., 1995; Singh, 1997). Es scheint vielmehr oft so zu sein, daß Kooperationspartner eine schlechte Performance in Kauf nehmen oder daß Unternehmungen mit unterdurchschnittlichen Rentabilitäten kollektive Strategien implementieren (Albach, 1991; Burgers et al., 1993). Dieser Sachverhalt deutet darauf hin, daß die starke Verbreitung der Untemehmungskooperation institutionelle Gründe haben könnte. Unternehmungen könnten beispielsweise institutionellen Zwängen nachgeben und das Kooperationsverhalten anderer Branchenmitglieder imitieren, auch wenn dies aus Rentabilitätsgründen nicht unbedingt das beste strategische Verhalten wäre (Venkatraman et al., 1994). Für die These, daß Unternehmungen Kooperationsverträge schließen und imitieren, um ihre Legitimität innerhalb eines relevanten institutionellen Kontextes zu erhöhen, liegt bisher nur begrenzte Evidenz vor (Sharfman et al., 1991; Baum & Oliver, 1991; 1992). Eine intensivere Erforschung dieser Zusammenhänge steht noch aus.

Weiterführende Kooperationsforschungen Es ist durchaus sinnvoll, wenn die dargestellten unterschiedlichen Forschungstraditionen zum Kooperationsphänomen weiter fortgeführt werden. Hierfür sprechen mindestens drei Gründe: Erstens, es gibt Widersprüche innerhalb einzelner Ansätze und zwischen den Ansätzen, die aufgelöst werden sollten. Zweitens, innerhalb der verschiedenen Ansätze besteht oft noch ein erheblicher empirischer Forschungsbedarf. Drittens, existierende Komplementaritäten in den Forschungsfragen und Argumenten machen eine stärkere Integration der Ansätze wünschenswert. Bei der Fortführung der Kooperationsforschung sollten insbesondere folgende Mängel bzw. Defizite geschlossen werden:

Kollektive Strategien

479

( 1 ) Vergleichende Forschungsansätze. Die empirische Forschung sollte sich verstärkt der Erforschung konkurrierender Hypothesen zuwenden, die sich vor dem Hintergrund unterschiedlicher Erklärungsansätze ergeben. Wir benötigen z.B. mehr Studien, die die zentralen Argumente des Transaktionskostenansatzes mit den Argumenten anderer Ansätze, z.B. denen des strategischen Erklärungsansatzes oder des ressourcenbasierten Ansatzes, vergleichen und einem konkurrierenden empirischen Test zuführen. (2) Bezüge zwischen Kooperationsformen und Strategieinhalten. Die Kooperationsforschung hat sich auf die Antezedenzien für die Wahl und die Performancekonsequenzen unterschiedlicher Kooperations/ormen konzentriert. Diese Konzentration führte zu einer weitgehenden Vernachlässigung der durch Unternehmungskooperationen verfolgten Strategiem/ia/ie. Es ist notwendig, klarere Bezüge zu den verfolgten Geschäfts- und Funktionsbereichsstrategien herzustellen. Inwieweit unterschiedliche Kooperationsformen sich z.B. besser zur Unterstützung von Differenzierungsstrategien als zur Realisierung von Kostenführerschafts- oder Nischenstrategien eignen, oder inwieweit verschiedene Kooperationsformen sich besser oder schlechter für die Verfolgung von First-Mover- oder Late-Follower-Strategien eignen, sind in der empirischen Literatur kaum behandelte Fragestellungen (Bresser et al., 1994; McGee et al., 1995). Die fatale Konsequenz der Loslösung der Diskussion der Kooperationsformen von den Strategieinhalten läßt sich durch eine Metapher veranschaulichen: Diese Loslösung ist vergleichbar mit dem Versuch, den Wert eines Kunstmuseums nur im Hinblick auf die vorhandenen Gebäude und Raumaufteilungen diskutieren zu wollen, ohne den Inhalt des Museums, die Kunstschätze selbst, zu berücksichtigen. (3) Vergleiche kooperativer mit unabhängigen Strategien. Überraschend ist auch die Vernachlässigung einer vergleichenden Beurteilung kooperativer Strategien mit von Unternehmungen im Alleingang durchgeführten Strategien. Bresser et al. (1994) können z.B. für eine Stichprobe amerikanischer Sparkassen nachweisen, daß kollektive und unabhängige Wettbewerbsstrategien in unterschiedlichen Kombinationen und mit unterschiedlichen Performancekonsequenzen durchgeführt werden. Eine weitere der wenigen Studien, die eine derartige vergleichende Betrachtung vornimmt, ist die Untersuchung von Mitchell und Singh (1996). Die Autoren untersuchen die relative Vorteilhaftigkeit kooperativer und unabhängiger Strategien in Abhängigkeit vom Grad der Umweltvariation.84 Offensichtlich besteht hier ebenfalls eine erhebliche zu schließende Forschungslücke. Wir benötigen Kenntnisse, die die Wahl zwischen kooperativen und unabhängigen Strategien erleichtern. (4) Die Performancekonsequenzen der Kooperation. Wie bereits erwähnt, spricht die empirische Evidenz gegen die Auffassung, Unternehmungskooperationen würden durchweg eine überdurchschnittliche Performance ermöglichen. Die vorliegende Evidenz unterstützt eher die These, daß positive Performancekonsequenzen limitiert und stark kontextabhängig sind (Kent, 1991; Hagedoorn & Schakenraad, 1994; 84

Mitchell und Singh (1996) stellen (für eine branchenspezifische Stichprobe) fest, daß kooperierende Unternehmungen in stabilen, sich kontinuierlich verändernden Umwelten bessere Überlebenschancen haben als Unternehmungen, die unabhängige Wettbewerbsstrategien verfolgen. Demgegenüber sind kooperierende Unternehmungen in Umwelten, die sich abrupt verändert haben, im Vergleich mit unabhängigen Konkurrenten benachteiligt, sofern die Umweltveränderungen die zentralen Aktivitäten der Kooperation betreffen. Bestehen zwischen den zentralen Kooperationsaktivitäten und den durch die abrupten Umweltveränderungen betroffenen Aktivitätsbereichen hingegen starke Unterschiede, so erhöht sich die Überlebensfähigkeit der kooperierenden Unternehmungen.

480

R. Κ. F. Bresser

Smith et al., 1995; Mitchell & Singh, 1996; Singh, 1997). Neben Studien zur Klärung der Frage nach der relativen Vorteilhaftigkeit von Kooperationsstrategien ist auch eine stärkere Thematisierung der verwendeten Performancekonzepte und -maße wichtig. Für Unternehmungskooperationen ist der Performancebegriff noch schillernder und umfassender als für unabhängige Unternehmungen, denn das Konzept wird von den Interessen und Zielen mehrerer Unternehmungen beeinflußt. Die Schwierigkeit, ein angemessenes Performancekonzept auszuwählen, kann am Beispiel der Kooperationsstabilität verdeutlicht werden. Die Stabilität und Langlebigkeit einer Kooperation wurde in vielen Studien als Performancekriterium ausgewählt. Intertemporale Stabilität ist aber ein fragwürdiges Kriterium, wenn die wichtigste Funktion der Kooperation darin besteht, bestimmte Fähigkeiten des jeweils anderen Partners zu erlernen (Hamel, 1991). (5) Kontingenztheoretische Präzisierung. Was der Kooperationsforschung derzeit fehlt (dies ergibt sich zum Teil auch aus den vorangehenden vier Kritikpunkten), ist die Weiterentwicklung zu einer differenzierten Kontingenztheorie der Kooperation (Osborn & Hagedoorn, 1997). Interne und externe Kontextfaktoren der Kooperation müssen differenziert für unterschiedliche Kooperationsstrategien und -formen und im Vergleich zu nicht-kooperativen Strategien auf ihre Performancekonsequenzen hin untersucht werden.85 Eine derartige kontingenztheoretische Präzisierung ist sicherlich nicht einfach und nur durch einen erheblichen Forschungsaufwand zu erreichen, sie kann aber dazu beitragen, daß das momentane Chaos der Kooperationsforschung überwunden wird.

Die Artikel Der erste Artikel, „Kollektive Unternehmensstrategien", von Rudi Bresser behandelt potentielle Risiken der Kooperation. Unter Rückgriff auf den Ressourcenabhängigkeitsansatz und andere organisationstheoretische Überlegungen werden drei Dysfunktionen abgeleitet: strategische Inflexibilität, erhöhte Störungseinwirkungen und Markteintritt. Der Versuch, diesen Mängeln einer kollektiven Strategie entgegenzuwirken, kann eine explosive Dynamik in Gang setzen, die letztendlich zur Aufgabe und Substitution der Kooperation durch unabhängige Wettbewerbsstrategien führt. Der Artikel behandelt auch die (begrenzten) Möglichkeiten, auf die Unternehmungen zurückgreifen können, um die Dysfunktionen ihrer Kooperationsstrategien auszugleichen. Darüber hinaus thematisiert der Artikel das Zusammenspiel zwischen kollektiven und unabhängigen Strategien. Der zweite Beitrag, Resource-based View of Strategie Alliance Formation: Strategie and Social Effects in Entrepreneurial Firms", von Eisenhardt und Schoonhoven ist einer der wenigen Versuche, strategische Motive der Unternehmungskooperation unter 85

Auffällig ist, daß die empirische Forschung sich insbesondere mit internen Kontextbedingungen (Stabilitätsfaktoren) beschäftigt und den externen Kontext vernachlässigt hat. Dies verwundert etwas, denn zum externen Kontext der Kooperation wurden bereits Mitte der 80er Jahre durch Harrigan (1985b: 103-128; 1986: 85-124; 1988) ausführliche konzeptionelle Überlegungen angestellt. Ein weiterer Schwerpunkt der empirischen Forschung liegt bei den „equity alliances", während „non-equity alliances" vergleichsweise selten untersucht werden. Zukünftige Studien sollten auch dieses Defizit ausgleichen.

Kollektive Strategien

481

Rückgriff auf den ressourcenbasierten Ansatz zu erforschen. Argumentiert wird vor dem Hintergrund der „strategischen Notwendigkeit" und der „sozialen Möglichkeit", die als kooperationsauslösende Faktoren gelten. Die Stichprobe bezieht sich auf junge Unternehmungen der amerikanischen Halbleiterbranche. Als Indikatoren für die Konzepte „strategische Notwendigkeit" und „soziale Möglichkeit" werden Branchen-, Unternehmungsund Top-Management-Team-Charakteristika herangezogen. Die wichtigsten Hypothesen der Längsschnittsuntersuchung finden Bestätigung. Insbesondere Unternehmungen, die sich in einer angreifbaren strategische Position befinden und darüber hinaus über TopManagement-Teams verfügen, die in ein kooperationsförderndes soziales Netz eingebunden sind, entschließen sich zur Durchführung kollektiver Strategien.

482

R. Κ. F. Bresser

Kollektive Untemehmensstrategien Von Rudi K.F. Bresser Kollektive Unternehmensstrategien können Wettbewerbsstrategien, insbesondere Geschäftsbereichs- und Konzernstrategien ergänzen oder ersetzen. Dieser Beitrag behandelt die Chancen und Risiken, die sich fiir Einzelunternehmungen ergeben, wenn sie sich an kollektiven Strategien beteiligen. Kollektive Strategien können Umweltvarianten und Entscheidungsunsicherheit abbauen oder durch dysfunktionale Entwicklungen erhöhen. Zu den Dysfunktionen kollektiver Strategien zählen Potentiale für strategische Inflexibilität, verstärkte Störungseinwirkungen und die Attraktion neuer Marktteilnehmer. Ein zusammenfassendes Modell der Konsequenzen kollektiver Strategien wird entwickelt, das eine Diskussion des Verhältnisses zwischen Wettbewerbsstrategien und kollektiven Strategien ermöglicht. Fallbeispiele aus den europäischen und amerikanischen Erfahrungsbereichen verdeutlichen die interkulturelle Bedeutsamkeit des Themas. Möglichkeiten der Beseitigung oder Abschwächung dysfunktionaler Entwicklungen werden diskutiert.

A. Enleitung Die amerikanische Managementliteratur beschäftigt sich seit Beginn der 80er Jahre mit einem Themengebiet, das als „Kollektive Untemehmensstrategien" bezeichnet wird (Astley und Fombrun, 1983 a; Bresser und Harl, 1986). Die Literatur unterscheidet für gewöhnlich zwei Ebenen strategischer Planung: Geschäftsbereichsstrategien und Konzernstrategien (Hofer und Schendel, 1978; Thompson und Strickland, 1987). Während Geschäftsbereichsstrategien darauf abzielen, einer Unternehmung Wettbewerbsvorteile innerhalb eines bestimmten Industriezweiges zu vermitteln, sind Konzernstrategien darauf bedacht, Geschäftsbereichsstrategien zu integrieren und auf die Zielsetzungen des Konzerns auszurichten. Diese Zweiteilung strategischer Planungsansätze vernachlässigt die vielfaltigen Interdependenzen, die in der Aufgabenumwelt einer bestimmten Unternehmung bestehen können. Zur Berücksichtigung solcher Interdependenzen wurde eine dritte Ebene strategischer Planung vorgeschlagen - die der kollektiven Strategie (Astley und Fombrun, 1983a; Fombrun und Astley, 1983a, b). Dieser Beitrag untersucht die Wirkungsweisen kollektiver Strategien sowie das Verhältnis zwischen kollektiven und Wettbewerbsstrategien. Kollektive Untemehmensstrategien repräsentieren systematische Vorgehensweisen, die von mehreren Organisationen gemeinschaftlich entwickelt und implementiert werden; sie dienen der Stabilisierung und Beherrschung interdependenter Aufgabenumwelten.1 Umweltinterdependenz innerhalb einer Organisationspopulation hat strategische Bedeutung, denn sie führt häufig zu turbulenten Entwicklungen, die von einzelnen Organisationen schwer antizipiert werden können und somit Entscheidungsunsicherheit hervorrufen (Emery und Trist, 1965; Hannan und Freemann, 1977). Kollektive Strategien können reaktiven oder proaktiven Charakter haben: Sie sind reaktiv wenn sie Umweltvariation absorbieren und proaktiv wenn sie eigenständiges, nicht-konzertiertes Verhalten von Einzelorganisationen verhindern. Eingegangen: 19. September 1988 Professor Dr. Rudi K. F. Bresser, Bernard M. Baruch College, The City University of New York, 17 Lexington Avenue, New York, New York, 10010.

ZfB 59. Jg. (1989), H. 5

545

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

Um die Wirkungsweisen kollektiver Strategien besser beurteilen zu können, ist ein Rückgriff auf die anglo-amerikanische Organisationstheorie sinnvoll. Im foglenden werden zunächst verschiedene Konzeptionen und Koordinationsformen kollektiver Strategien näher definiert. Abschnitt C diskutiert das Verhältnis zwischen Umweltinterdependenz und Unternehmensstrategie. Abschnitt D erläutert die Dimensionalität des Interdependenzbegriffs und Abschnitt E entwickelt ein integriertes Modell der Konsequenzen kollektiver Strategien.

B. Konzeptionen und Koordinationsformen kollektiver Strategien In der Literatur werden zwei Konzeptionen des Begriffs „Kollektive Unternehmensstrategie" verwendet (Astley und Fombrun, 1983 a; Bresser und Harl, 1986; Carney, 1987). Zum einen werden als kollektive Strategien jene interorganisationellen Netzwerke bezeichnet, die sich ohne Gesamtplan, gewissermaßen unbeabsichtigt entwickeln. In dem Maße, in dem sich die Verbindungen zwischen einzelnen Organisationen zu einem größeren Netzwerk aggregieren, entsteht eine unbeabsichtigte kollektive Strategie, die von keiner der im Netzwerk eingebundenen Organisationen geplant war oder vorausgesehen werden konnte. Die Entstehung der Telekommunikationsindustrie ist ein Beispiel (Astley und Fombrun, 1983 b; Fombrun und Astley, 1982). Zum anderen werden kollektive Strategien als bewußt geplante, systematische Vorgehensweisen aufgefaßt, die von mehreren Organisationen mit dem Ziel implementiert werden, die gemeinsame Umweltinterdependenz zu managen. Für die Zwecke dieses Beitrags ist lediglich die zweite Definition kollektiver Strategien bedeutsam. Von Interesse sind beabsichtigte, systematisch geplante und implementierte Unternehmensstrategien. Organisationen können sich unterschiedlicher Koordinationsformen bedienen, um eine (beabsichtigte) kollektive Strategie zu verwirklichen. Tabelle 1 präsentiert diese Koordinationsformen nach Maßgabe ihres Formalisierungsgrades. Staatliche Regulierungen (sofern diese durch kollektives Lobbyieren beeinflußt wurden) und vertragliche Vereinbarungen sind Koordinationsformen mit hohem Formalisierungsgrad. Kollektive Strategien, die auf Schachtelaufsichtsratsmandaten (interlocking directorates) beruhen oder von organisierten Interessenverbänden (Handelskammern, Industrieverbänden) implementiert werden, haben einen mittleren Formalisierungsgrad. Kollusion und Indur a i . 1:

546

Koordinationsformen für kollektive Strategien Koordinationsformen

Formalisierungsgrad

Staatliche Regulierungen Vertragliche Vereinbarungen Unternehmenszusammenschlüsse Joint Ventures

Hoch Hoch Hoch Hoch

Kooptation & Schachtelaufsichtsratsmandate

Mittlere Höhe

Interessenverbände

Mittlere Höhe

Kollusion & Industrieführerschaft

Gering

ZfB 59. Jg. (1989), H. 5

483

484

R. Κ. F. Bresser Kollektive Untemehmensstrategien

strieführerschaft gelten als Koordinationsformen mit geringem Formalisierungsgrad (Bresser, 1988; Fombrun und Astley, 1983 a).

C. Umweltinterdependenz und Untemehmensstrategien

Die Annahme, daß Umweltinterdependenzen turbulente Entwicklungen und Entscheidungsunsicherheit hervorrufen können, bedarf der Erläuterung. Umweltinterdependenz zwischen Organisationen besteht immer dann, wenn einzelne Organisationen nicht all jene Faktoren unter ihrer Kontrolle haben, deren Beherrschung notwendig ist, um den Erfolg einer bestimmten Aktion bzw. das Erlangen eines angestrebten Resultats sicherzustellen (Pfeffer und Salancik, 1978). Unternehmungen können sich in drei Arten von Umweltinterdependenz befinden: in horizontaler, vertikaler oder symbiotischer Interdependenz (Fombrun und Astley, 1982; Pennings 1981). Horizontale Interdependenz besteht zwischen Firmen, die innerhalb desselben Marktes in Wettbewerb stehen. Vertikale Interdepenz besteht zwischen Firmen, die zwar in unterschiedliche Märkten tätig sind, aber durch aufeinanderfolgende Arbeitsprozesse derartig verknüpft sind, daß der Output des einen zum Input des anderen wird. Symbiotische Interdependenz besteht zwischen Firmen, die sich in der Bereitstellung bestimmter Dienstleistungen und/oder Produkte ergänzen. Umweltinterdependenz führt i.d.R. dann zu stürmischen Entwicklungen innerhalb einer Organisationspopulation, wenn die Aktivitäten von Einzelorganisationen die strategischen Intentionen anderer Organisationen gefährden. Organisationen, die ihre strategischen Pläne in Gefahr sehen, werden zu Gegenmaßnahmen greifen, was weitere Gegenmaßnahmen nach sich zieht, usw. Für die Einzelunternehmung kann Umweltinterdependenz Entscheidungsunsicherheit hervorrufen, da der Erfolg ergriffener Akquisitions-, Transformations- oder Verkaufsmaßnahmen von den Aktivitäten anderer Organisationen abhängt. Die Unternehmung ist mit Entscheidungsunsicherheit konfrontiert, wenn sie sich über ihre aus Interdependenz entstehenden Probleme bewußt ist und darüber hinaus nicht in der Lage ist, die Aktivitäten anderer Organisationen gänzlich zu kontrollieren (Pfeffer und Salancik, 1978). Organisationen, die sich in vertikaler oder symbiotischer Interdependenz befinden, stehen meist einer nur geringen Entscheidungsunsicherheit gegenüber, da ihre Beziehungen sich auf der Basis vertraglich geregelter Kooperationsmuster vollziehen, was die Schwierigkeiten der gegenseitigen Verhaltenskontrolle reduziert. Demgegenüber ist die Entscheidungsunsicherheit für horizontal interdependente Organisationen hoch. Diese Organisationen kämpfen miteinander um Marktanteile und haben somit erhebliche Schwierigkeiten der gegenseitigen Verhaltenskontrolle. Entscheidungsunsicherheit unter horizontal interdependenten Organisationen tritt mit größter Wahrscheinlichkeit in oligopolitischen Märkten auf, denn Oligopolisten sind sich ihrer Interdependenz bewußt, und sie sind alleine nicht in der Lage, die Verhaltensweisen anderer Wettbewerber zu kontrollieren (Hall et al., 1977; Pennings, 1981; Scherer, 1980). Organisationen sind bemüht, ihre Umweltinterdependenzen zu beeinflussen und die daraus resultierende Entscheidungsunsicherheit zu reduzieren (Cyert und March, 1963; Hickson et al., 1971; Thompson, 1967). Zu diesem Zwecke können sie sowohl Wettbewerbs- als auch kollektive Strategien einsetzen. Wettbewerbsstrategien (ζ. B. ProduktdifZfB 59. Jg. (1989), H. 5

547

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

ferenzierungen) können zur Beherrschung von Umweltinterdependenzen führen, wenn sie der Einzelunternehmung eine vorteilhafte Marktposition verschaffen, durch die das Unternehmen weitgehend unabhängig von den Maßnahmen anderer Wettbewerber wird. Da relevante Interdependenzen sich im komplexen Umweltsituationen von Einzelunternehmungen häufig nicht hinreichend eindeutig identifizieren lassen, sind individuell initierte Wettbewerbsstrategien oft für das Management interdependenter Umwelten ungeeignet. In diesen Situationen können kollektive Strategien Wettbewerbsstrategien ergänzen oder ersetzen, um Umweltveränderungen und Entscheidungsunsicherheit zu reduzieren (Astley und Fombrun, 1983 a; Bresser und Harl, 1986).

D. Zur Dimensionalität von Umweltinterdependenz Unabhängig von der Art der Umweltinterdependenz lassen sich zwei Dimensionen des Umweltinterdependenzbegriffs unterscheiden: Umweltvariation und Umweltverkettung (Emery und Trist, 1965). Die Dimension Umweltvariation (movement) umfaßt sowohl die Häufigkeit als auch die Voraussagbarkeit von Veränderungen innerhalb einer Organisationsumwelt. Umweltvariation nimmt zu, wenn sich die Elemente einer Organisationsumwelt mit zunehmender Häufigkeit verändern und/oder wenn die Voraussagbarkeit dieser Veränderungen abnimmt. Zunehmende Umweltvariation erschwert das Management interdependenter Umwelten (Miles, 1980; Pfeffer und Salancik, 1978). Der Begriff Umweltverkettung (interconnectedness) ist definiert als die Regulierung von Interaktionen zwischen Umweltelementen durch Verhaltensmechanismen oder formale und informale Vereinbarungen. Der Grad der Umweltverkettung steigt an, wenn bestimmte Mechanismen oder Vereinbarungen existierende Verbindungen zwischen Umweltelementen verstärken oder neue Verbindungen entstehen lassen, und dadurch ein größeres Ausmaß an Umweltinteraktionen determinieren. Obwohl die Organisationsliteratur den Begriff Umweltverkettung eindimensional definiert (Emery und Trist, 1965; Miles, 1980; Pfeffer und Salancik, 1978), ist es für die Analyse der Wirkungsweise kollektiver Strategien zweckmäßig, zwei Arten von Umweltverkettung zu unterscheiden: Wettbewerbsverkettungen und Vertragsverkettungen. Wettbewerbsverkettungen beruhen auf Marktmechanismen, die das Verhalten konkurrierender Firmen beeinflussen. Vertragsverkettungen ergeben sich aus formalen und informalen interorganisationellen Vereinbarungen, die sich ebenfalls verhaltensregulierend auswirken. Kollektive Strategien können die Intensität der Wettbewerbs- und Vertragsverkettungen innerhalb eines Umweltsystems verändern. Diese Veränderungen beeinflussen die Qualität der Interdependenz zwischen den Elementen einer Organisationsumwelt, so daß sowohl Reduktionen als auch Zunahmen von Umweltvariation und Entscheidungsunsicherheit möglich werden. Der folgende Abschnitt behandelt diese alternativen Konsequenzen kollektiver Strategien. E. Funktionale und dysfunktionale Konsequenzen kollektiver Strategien Die Bewertung der Konsequenzen kollektiver Strategien ist nicht einheitlich. Sie ist natürlich abhängig von den Interessen und Zielsetzungen bestimmter Interessengruppen. 548

ZÍB59. Jg. (1989), H. 5

485

486

R. Κ. F. Bresser Kollektive Untemehmensstrategien

Während die an einer kollektiven Strategie beteiligten Organisationen die kollektiv verwirklichte Umweltstabilität als positiv bewerten mögen, kann dieselbe Stabilität von Vertretern des öffentlichen Interesses negativ bewertet werden, ζ. B. wenn eine kollektive Strategie Wettbewerbsbeschränkungen herbeiführt. Während die gesamtwirtschaftlichen Konsequenzen kollektiver Strategien ein wichtiges Forschungsgebiet darstellen (Caves, 1982; Scherer, 1980), fragt dieser Beitrag nach den Chancen und Risiken kollektiver Strategien aus der Sicht beteiligter Einzelorganisationen. Diese einzelwirtschaftlichen Fragestellungen sind bedeutsam, da sich kollektive Strategien aus gesamtwirtschaftlicher Sicht als problemlos oder akzeptabel erweisen können, während sich aus einzelwirtschaftlicher Sicht erhebliche Probleme ergeben. Darüber hinaus ist ein Verständnis insbesondere der Risiken kollektiver Strategien auch dann von Nöten, wenn kollektive Strategien wettbewerbspolitisch anrüchig sind. Neben das Risiko der Beanstandung solcher Strategien durch Aufsichtsbehörden können weitere, strategiespezifische Risikofaktoren treten, und die Kenntnis solcher Faktoren kann sich hemmend auf die Bereitschaft von Unternehmensleitungen auswirken, wettbewerbspolitisch problematische kollektive Strategien zu verfolgen. Darüber hinaus konzentrieren sich die nachfolgenden Überlegungen auf Unternehmen, die in oligopolitischen Märkten agieren. Diese Ausrichtung der Diskussion ist sinnvoll, da Oligopolisten sich ihrer Interdependenz bewußt sind und sich darum bemühen, die resultierende Umweltvariation und Entscheidungsunsicherheit durch Wettbewerbs- oder kollektive Strategien abzubauen (Bresser und Harl, 1986; Pennings, 1981). Aus der Sicht der an einer kollektiven Strategie beteiligten Unternehmungen (nachfolgend als Organisationskollektiv bezeichnet) ergeben sich die Hauptchancen kollektiver Strategien aus der Möglichkeit, zumindest zeitweilig eine relativ stabile und voraussagbare Organisationsumwelt herzustellen (Astley und Fombrun, 1983 a; Pennings, 1981). Sofern Organisationen bei zunehmender Umweltvariation eine kollektive Strategie erfolgreich implementieren (ζ. B. um ihre Produktentwicklungen zu harmonisieren), ist es ihnen gelungen, einige Quellen der gemeinsamen Umweltvariation und Entscheidungsunsicherheit zu beseitigen. Der Erfolg eines Organisationskollektivs, einen Teilbereich seiner Organisationsumwelt zu stabilisieren, beruht darauf, daß sich die Intensität der Umweltverkettungen aufgrund der implementierten kollektiven Strategie verändert. Wenn Unternehmungen ihre Marktaktionen harmonisieren, reduzieren sie den Grad der Wettbewerbsverkettungen, denn die durch den Marktmechanismus bewirkte Verhaltenssteuerung ist für einen Teilbereich des Marktes außer Kraft gesetzt. Abnehmende Wettbewerbsverkettungen verringern die Häufigkeit der Umweltvariation, da individuelles Wettbewerbsverhalten für den Zuständigkeitsbereich der kollektiven Strategie entfallt. Dies bewirkt den Abbau von Entscheidungsunsicherheit. Ähnliche Wirkungen ergeben sich durch die Veränderung der Vertrags Verkettungen. Eine kollektive Strategie erhöht die Intensität der Vertragsverkettungen innerhalb einer Organisationsumwelt, da die durch Vereinbarungen bewirkten verhaltensregulierenden Verbindungen zwischen Umweltelementen zunehmen. Die Zunahme der Vertragsverkettungen ermöglicht den Abbau von Umweltvariation und Entscheidungsunsicherheit, denn die getroffenen Vereinbarungen lassen Kommunikationskanäle entstehen, durch die sich das Verhalten der an der kollektiven Strategie beteiligten Firmen leichter voraussagen läßt. ZfB 59. Jg. (1989), H. 5

549

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

Abbildung 1 repräsentiert ein integriertes Modell der Konsequenzen kollektiver Strategien. Das Modell stellt beabsichtigte und unbeabsichtigte Konsequenzen schematisch dar. Die beschriebenen Chancen oder beabsichtigten Konsequenzen kollektiver Strategien sind in Abbildung 1 als vier Kausalschleifen dargestellt (Schleifen 1 bis 4), die die Variablen Umweltvariation, Entscheidungsunsicherheit, Kollektive Strategien, Wettbewerbsverkettungen und Vertragsverkettungen verbinden. 2

Abb. 1: Die Konsequenzen kollektiver Strategien

Umweltvariation (1)

H + Wettbewerbsverkettungen

(2)

Kollektive Strategien (3)

Vertragsverkettungen

(4)

(-) + Störungseinwirkung (5)

Markteintritt (7)

Anpassungsfähigkeit

7 V

(8)

Strategische Flexibilität (6)

.

Entscheidungsunsicherheit (9)

Kausalschleifen der beabsichtigten Konsequenzen:

Schleife Schleife Schleife Schleife

1 2 3 4

1-3-21-3-41-9-31-9-3-

1 1 2-1 4-1

Kausalschleifen der unbeabsichtigten Konsequenzen:

Schleife 5 Schleife 6 Schleife 7 Schleife 8 Schleife 9 Schleife 10 Schleifell Schleife 12 Schleife 13 Schleife 14 Schleife 15

3-4-63-4-63-4-63-4-63-4-53-4-53-4-53-4-53-2-73-2-73-2-7-

8-9 -3 8-7 -1 8-7 -9 8 - 7 -1 8-9 -3 8 - 7 -1 8-7 -9 8-7 -1 1-3 9-3 1-9 -3

550

-3 -3 -9- 3 -3 -3 -9- 3

ZfB 59. Jg. (1989), H. 5

487

488

R. Κ. F. Bresser Kollektive Unternehmensstrategien

Während Organisationskollektive durch kollektive Strategien Umweltvariation und Entscheidungsunsicherheit zweifellos reduzieren können, ist es nicht so klar, inwieweit diese gewünschten Effekte von dauerhafter Natur sind. Dieselben Veränderungen von Umweltverkettungen, die die beabsichtigten Konsequenzen kollektiver Strategien bewirken, sind auch in der Lage, dysfunktionale Entwicklungen auszulösen. Hier soll dargestellt werden, daß insgesamt drei dysfunktionale Entwicklungen auftreten können, und zwar isoliert oder in beliebigen Kombinationen, durch die sich die Variation innerhalb einer Organisationsumwelt und die damit verbundene Entscheidungsunsicherheit letztlich wieder erhöhen (Bresser, 1984; Bresser und Harl, 1986):

I. Strategische Inflexibilität Die erste dysfunktionale Konsequenz kollektiver Strategien ist ein Resultat höherer Vertragsverkettungen, die die strategische Flexibilität einzelner Organisationen beeinträchtigen können. Diese Inflexibilität kann sich durch drei Entwicklungen ergeben: durch Beschränkungen in der Strategieauswahl, durch „Sunk Costs" oder durch unkontrollierte Informationspreisgabe. Beschränkung in der Strategieauswahl. Organisationskollektive erzeugen vertragliche Verkettungen, indem sich die beteiligten Einzelorganisationen darauf verständigen, Marktverhalten zu koordinieren. In dem Maße, in dem Organisationen auf den Einsatz von Wettbewerbsstrategien verzichten, sind sie in ihrer Strategieauswahl eingeschränkt. Wenn ζ. B. ein Organisationskollektiv entscheidet, Produktentwicklungen zu harmonisieren, hat sich das Repertoire der verfügbaren Wettbewerbsstrategien verkleinert, d. h., die strategische Flexibilität der einzelnen Organisationen ist gesunken. Weitere Beschränkungen in der Strategieauswahl können von Kartellbehörden ausgehen, und zwar hinsichtlich zusätzlicher Kooperationen innerhalb eines Marktes. Bestehende kollektive Strategien verringern tendenziell die Bereitschaft der Kartellbehörden, weitere Kooperationsvereinbarungen zu tolerieren. Sunk Costs. Probleme strategischer Inflexibilität können auch dann entstehen, wenn Organisationen ihre kollektiven Strategien im Zeitablauf konsolidieren und somit ihre vertraglichen Verkettungen verstärken. Während der Konsolidierungsphasen einer kollektiven Strategie werden oft umfangreiche Investitionsmittel gebunden, und diese „sunk costs" verpflichten die Management-Teams der beteiligten Organisationen, an einer bestehenden Strategie festzuhalten, und zwar oft selbst dann, wenn veränderte Umweltbedingungen eine Aufgabe der Strategie nahelegen. Die Verwerfung einer kollektiven Strategie, die bereits zu erheblichen Investitionen geführt hat, kann von dritter Seite leicht als ein Indiz für Mißmanagement aufgefaßt werden. Während ein inflexibles Festhalten an einem einmal eingeschlagenen Weg auch bei Einzelunternehmungen beobachtet werden kann, die individuelle Wettbewerbsstrategien verfolgen (Harrigan, 1985), ist das Problem derartiger Inflexibilität von höherer Virulenz, wenn sich Gruppen von Organisationen darauf verständigen, eine gemeinschaftliche Strategie zu implementieren. Die gruppendynamische Literatur belegt, daß Gruppen sich wesentlich eher als Individuen Illusionen der Unverletzlichkeit hingeben, Informationen mißachten oder an dem Status quo festhalten, wenn sich konfligierende MeinunZfB 59. Jg. (1989), H. 5

551

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

gen nicht beilegen lassen (Gottfredson und White, 1981; Janis, 1972; Staw, 1982). Es ist somit anzunehmen, daß derartige Prozesse die Bereitschaft eines Organisationskollektivs erhöhen, an einer implementierten kollektiven Strategie festzuhalten. Unkontrollierte Informationspreisgabe. Bresser (1988) verweist darauf, daß kollektive Strategien zu einer unkontrollierbaren Preisgabe strategisch sensitiver Informationen führen können. Dies wäre der Fall, wenn die durch kollektive Strategie entstehenden Kommunikationskanäle nicht nur der Verhaltenskoordination in den beabsichtigten Bereichen dienen, sondern darüber hinaus auch noch die Preisgabe von Informationen erlauben, die die Wettbewerbsstrategien der beteiligten Einzelorganisationen betreffen. Firmen, die in Joint Ventures engagiert sind und zu diesem Zwecke regelmäßig Informationen austauschen, laufen ζ. B. Gefahr, daß sie ihre Wettbewerbsvorteile und -Strategien nicht voneinander verbergen können. Die potentielle Preisgabe wettbewerbsrelevanter Informationen erschwert die Kombination von Wettbewerbsstrategien und kollektiven Strategien, denn die Einzelunternehmung muß befürchten, daß sie ihre Wettbewerbsstrategien nicht geheimhalten kann. Diese Befürchtung kann strategische Inflexibilität nach sich ziehen. Wenn Unternehmensleitungen damit rechnen müssen, daß sie Wettbewerbsstrategien aufgrund vertraglicher Verkettungen nicht vor anderen Konkurrenten verbergen können, wirkt sich dies tendenziell verringernd auf ihre Bereitschaft aus, Wettbewerbsstrategien in den nicht-koordinierten Marktbereichen zu verfolgen. Das Risiko der unkontrollierten Informationspreisgabe läßt den Erfolg solcher Wettbewerbsstrategien als fragwürdig erscheinen. Ein allen Formen strategischer Inflexibilität gemeinsames Resultat betrifft die Anpassungsfähigkeit des Organisationskollektivs. Die Reduktion strategischer Flexibilität verringert die Kapazität des Organisationskollektivs, sich als Ganzheit oder individuell an veränderte Umweltgegebenheiten anzupassen. Sobald die Mitglieder eines Organisationskollektivs ihre verringerte Anpassungsfähigkeit wahrnehmen, löst dies Entscheidungsunsicherheit aus. Diee Annahme ist darin begründet, daß die Mitglieder eines Organisationskollektivs auf Anpassungsfähigkeit nicht verzichten können (Cyert und March, 1963), sich aber nicht darüber im Klaren sind, wann und in welchem Umfang die kollektiv bewirkte Reduzierung ihrer Anpassungsfähigkeit zum Problem wird. Darüber hinaus steht zu erwarten, daß innovative Außenseiter sich dazu angehalten fühlen, in einen kollektiv organisierten Markt einzutreten, wenn sie feststellen, daß die gegenwärtigen Marktteilnehmer unter strategischer Inflexibilität und mangelnder Anpassungsfähigkeit leiden. Markteintritt stellt Umweltvariation dar, die weitere Entscheidungsunsicherheit zur Folge hat. Abbildung 1 skizziert die mit strategischer Inflexibilität zusammenhängenden möglichen Entwicklungen als Kausalschleifen 5, 6, 7 und 8.

II. Erhöhte Störungseinwirkung Die zweite dysfunktionale Konsequenz kollektiver Strategien ist ebenfalls auf zunehmende Vertragsverkettungen zurückzuführen. Indem kollektive Strategien vertragliche Verkettungen begründen oder verstärken, errichten sie gleichzeitig Kanäle, durch die sich die Einwirkung externer Störungen rasch auf das gesamte Organisationskollektiv ausbreiten kann (Aldrich, 1979; Pfeffer und Salancik, 1978; Weick, 1979). Dieser Dominoef552

ZfB 59. Jg. (1989), H. 5

489

490

R. Κ. F. Bresser Kollektive Unternehmensstrategien

fekt vollzieht sich parallel zu Entwicklungen, die die strategische Flexibilität eines Organisationskollektivs hemmen, und er verschlimmert Flexibilitätsprobleme. Da vertragliche Verkettungen „tightly coupled systems" (Weick, 1979) entstehen lassen, die die Einwirkungskraft externer Störungen erhöhen, verringert sich die Fähigkeit eines Organisationskollektivs, erfolgreiche Anpassungen an Bedrohungen aus der Organisationsumwelt vorzunehmen. Die Effekte erhöhter Störungseinwirkungen sind somit ganz ähnlich wie jene, die durch strategische Inflexibilität gefördert werden: Abnehmende Anpassungsfähigkeit kann Entscheidungsunsicherheit, Markteintritt und Umweltvariation hervorrufen. Abbildung 1 verzeichnet diese Beziehungszusammenhänge als Kausalschleifen 9, 10, 11 und 12.

III. Markteintritt Die dritte dysfunktionale Konsequenz kollektiver Strategien ist das Resultat abnehmender Wettbewerbsverkettungen. Organisationskollektive reduzieren die Intensität der Wettbewerbsverkettungen innerhalb ihrer Märkte, indem sie von individuellem Wettbewerbsverhalten Abstand nehmen. Eine verringerte Wettbewerbsintensität ruft häufig quasi-monopolitisches Marktverhalten und Innovationsverringerungen hervor. Die resultierende Untätigkeit oder Trägheit eines Marktes liefert innovativen Außenseitern Anreize, in den Markt einzutreten und mit dem Organisationskollektiv zu wetteifern. Drüber hinaus führt die Marktträgheit einzelne Mitglieder des Organisationskollektivs in Versuchung, abtrünnig zu werden, um durch Wettbewerbsverhalten besondere Vorteile zu erzielen (Caves, 1982; Scherer, 1980). Diese Markteintrittsvorgänge erhöhen die Umweltvariation und die Entscheidungsunsicherheit unter den Mitgliedern eines Organisationskollektivs. Abbildung 1 stellt diese Prozesse als Kausalschleifen 13,14 und 15 dar.

F. Beispiele dysfunktionaler Konsequenzen Die beschriebenen Konsequenzen kollektiver Strategien sind als Hypothesen über mögliche Entwicklungen aufzufassen, die sich im Einzelfall einstellen können aber nicht müssen. Da nur sehr wenige empirische Untersuchungen zur Dynamik kollektiver Strategien vorliegen (Astley und Fombrun, 1983 b; Edström et al., 1984; Fombrun und Astley, 1983 b; Thakur und Hoffman, 1988), beruht das in Abbildung 1 abgebildete Modell auf theoretischer Deduktion. Die erwünschten Konsequenzen kollektiver Strategien ergeben sich aus der präskriptiven Managementliteratur. Die unerwünschten, dysfunktionalen Konsequenzen lassen sich aus der organisationstheoretischen Literatur ableiten. Darüber hinaus existieren mehrere Fallbeispiele, die das vorgeschlagene Modell unterstützen. Beispiele für die funktionalen und dysfunktionalen Konsequenzen kollektiver Strategien lassen sich für alle in Tabelle 1 dargestellten Koordinationsformen anführen (Bresser und Harl, 1986). Dieser Abschnitt präsentiert lediglich zwei Fallbeispiele (für die Koordinationsformen Joint Ventures und staatliche Regulierungen), die insbesondere ZfB 59. Jg. (1989), H. 5

553

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

die dysfunktionalen Konsequenzen kollektiver Strategien hervorheben. Um die interkulturelle Bedeutsamkeit des Themas zu verdeutlichen, beruht das erste Beispiel auf europäischen und das zweite auf spezifisch amerikanischen Erfahrungen. Das erste Fallbeispiel bezieht sich auf den französischen Produzenten technischer Gase, L'Air Liquide. Zwischen 1971 und 1986 bedienten sich die L'Air Liquide und der schwedische Konzern AGA mehrerer Joint Ventures, um einige kollektive Strategien innerhalb Europas zu implementieren. Durch gemeinsame Töchter in Deutschland und den Benelux-Ländern gelang es den Partnerunternehmen, potentielle Umweltdynamik und damit einhergehende Entscheidungsunsicherheit zu verringern und das gemeinschaftliche Ziel, eine bedeutende Stellung auf dem europäischen Markt für Industriegase zu erreichen, zu verwirklichen. Eine Stabilisierung der relevanten Organisationsumwelten gelang somit durch Prozesse, die in den Kausalschleifen 1 bis 4 (Abbildung 1) entsprechen. Der Fall der L'Air Liquide ist darüber hinaus ein Beispiel für das Problem strategischer Inflexibilität. Durch die Joint Ventures mit dem schwedischen Konzern AGA entstanden vertragliche Verkettungen, die die strategische Flexibilität der L'Air Liquide beeinträchtigen: Der französische Konzern fühlte sich Mitte der 80er Jahre in der Strategieauswahl gehemmt (die geplante Fusion mit der Agefko stieß auf den Widerstand der Kartellbehörden) und mußte ferner feststellen, daß er strategisch sensitive Informationen (z. B. technologisches Know-how) nur mit Mühe vor der AGA verbergen konnte. Die wahrgenommene strategische Inflexibilität rief Entscheidungsunsicherheit bei der L'Air Liquide hervor, und zwar ähnlich wie die Kausalschleife 5 der Abbildung 1 beschrieben. Im Jahre 1986 entschloß sich die L'Air Liquide, alle Partnerschaften mit der AGA aufzulösen (Le Monde, 1986; Frankfurter Allgemeine Zeitung, 1986). Das zweite Fallbeispiel nimmt Bezug auf Entwicklungen im amerikanischen Bankwesen. Während der 70er Jahre gelang es der amerikanischen Bankenindustrie durch Einflußnahme auf staatliche Regulierungen eine kollektive Strategie zu verwirklichen, die die Umweltdynamik und Entscheidungsunsicherheit innerhalb des Bankwesens in relativ engen Grenzen hielt. Insbesondere gelang es den Banken, die auf Einlagen zu zahlenden Zinssätze auf einem künstlich niedrigen Niveau zu halten (Parker, 1981). Trotz der anfänglichen Erfolge dieser kollektiven Strategie, lassen sich auch in diesem Falle wieder dysfunktionale Entwicklungen verzeichnen. Das Beispiel belegt, daß alle unerwünschten Konsequenzen kollektiver Strategien gleichzeitig auftreten können. Die kollektiv herbeigeführten staatlichen Regulierungen reduzieren die Intensität der Wettbewerbsverkettungen, was innovative Maklerfirmen (z.B. Merrill-Lynch mit seinem „Cash Management Account") dazu veranlaßte, in den Bankenmarkt einzutreten. Dieser Markteintritt repräsentierte Umweltvariation. Da die staatlichen Regulierungen darüber hinaus Vertragsverkettungen entstehen ließen, durch die einzelne Banken in ihrer Strategieauswahl gehemmt waren, hatten die Banken Mühe, flexibel und umgehend auf die Bedrohung durch die Maklerhäuser zu reagieren. Staatliche Regulierungen mußten zunächst verändert werden bevor die Banken ihre Depositenzinsen anheben konnten. Die strategische Inflexibilität der Banken war besonders gravierend, da die staatlichen Regulierungen die Banken zu einem eng verketteten System zusammenfügten, so daß sich Störungen rasch auf das gesamte Bankwesen auswirken konnten. Mehrere Bankzusammenbrüche waren die Folge. Die zu Umweltvariation und Entscheidungsunsicher554

ZfB 59. Jg. (1989), H. 5

491

492

R. Κ. F. Bresser

Kollektive Unternehmensstrategien

heit führenden dysfunktionalen Prozesse der Regulierungen im amerikanischen Bankwesen entsprechen den Kausalschleifen 5 bis 12 in Abbildung 1. Das Beispiel der amerikanischen Banken verdeutlicht, daß die dysfunktionalen Konsequenzen kollektiver Strategien die Existenz einzelner Unternehmungen gefährden können. Während die potentiellen Dysfunktionen kollektiver Strategien allen Koordinationsformen gemein sind, ist es nicht möglich, eine Rangreihe mehr oder minder „gefährlicher" Koordinationsformen aufzustellen. Sobald eine kollektive Strategie implementiert ist, ist die Basis für mögliche dysfunktionale Konsequenzen errichtet, und zwar unabhängig von der gewählten Koordinationsform. Die Schwere und Geschwindigkeit der Gefährdung durch dysfunktionale Entwicklungen kann innerhalb jeder Koordinationsform variieren. Sie ist von den spezifischen Gegebenheiten einer bestimmten kollektiven Strategie abhängig, wie z.B. der Aggressivität neuer Marktteilnehmer oder der Fähigkeit eines Organisationskollektivs, sich externer Störungen zu erwehren.

G. Die Explosivität dysfunktionaler Entwicklungen Abbildung 1 stellt die Variablenzusammenhänge, die die gewünschten Konsequenzen kollektiver Strategien widerspiegeln (Kausalschleifen 1 bis 4), als abweichungsreduzierende Zirkel dar (Maruyama, 1963; Weick, 1979). Ursprüngliche Steigerungen an Umweltvariation und Entscheidungsunsicherheit lösen kollektive Strategien aus, die durch Veränderungen der Wettbewerbs- und Vertragsverkettungen den Grad der Umweltvariation und die Entscheidungsunsicherheit der beteiligten Organisationen wieder verringern. Diese Dynamik steht im Gegensatz zu den ungewollten Konsequenzen kollektiver Strategien. Die dysfunktionalen Konsequenzen können leicht eine sich selbst verstärkende Dynamik entwickeln. Wenn sich Umweltvariation und Entscheidungsunsicherheit als Resultate einer kollektiven Strategie ergeben, werden kollektiv vorgehende Organisationen ihre Kooperationsvereinbarung als fehlerhaft ansehen. Sie stehen vor der Alternative, die kollektive Strategie aufzugeben oder sie zu verbessern, d. h. ihre Mängel zu beseitigen. Es ist wahrscheinlich, daß Organisationskollektive die zweite Alternative wählen, da die eingetretene strategische Inflexibilität, die reduzierte Anpassungsfähigkeit des Kollektivs und die Bedrohung durch neue Marktteilnehmer der Aufgabe einer kollektiven Strategie im Wege stehen. Rationalisierungsprozesse können erwartet werden, durch die die TopManagement-Teams der beteiligten Organisationen ihre Verpflichtung, den einmal eingeschlagenen Weg auch fortzusetzen, rechtfertigen und bekräftigen (Janis, 1972; Staw, 1981). Derartige Verpflichtungen auf den Erfolg einer kollektiven Strategie rufen Maßnahmen hervor, durch die wahrgenommene Mängel einer Vereinbarung beseitigt werden sollen. Dies hat zur Konsequenz, daß sich ein einmal initiierter Prozeß fortsetzt und es erwächst die Gefahr, daß der Prozeß und die Konsequenzen kollektiven strategischen Vorgehens eine explosive, abweichungsverstärkende Eigendynamik annehmen. Abbildung 1 stellt die explosiven Tendenzen der dysfunktionalen Konsequenzen kollektiver Strategien als elf abweichungsverstärkende Kausalschleifen dar (Maruyama 1963). Die Kausalschleifen 5 bis 15 sind in dem Sinne explosiv, als eine Zunahme an ZfB 59. Jg. (1989), H. 5

555

Kollektive Untemehmensstrategien Rudi K.F. Bresser

kollektiven Strategien Prozesse in Gang setzt, die das Ausmaß kollektiver Strategien letztendlich noch weiter erhöhen. Die in Abbildung 1 abgebildeten Kausalschleifen repräsentieren somit zwei einander entgegengesetzte Triebkräfte. Einerseits erhält Abbildung 1 (vier) abweichungsreduzierende Kausalschleifen, die den erwünschten Konsequenzen kollektiver Strategien entsprechen. Andererseits wirken diesen erwünschten Prozessen (elf) abweichungsverstärkende Kausalschleifen entgegen, die die dysfunktionalen Entwicklungen repräsentieren. Es besteht die Gefahr, daß die abweichungsverstärkenden Tendenzen kollektiver Strategien letztendlich die Oberhand gewinnen und die Existenz der beteiligten Organisationen gefährden, wobei der Aufbau der Dysfunktionen zu existenzgefährdenden Krisen im Einzelfall durchaus einen längeren Zeitraum beanspruchen kann.

H. Das Problem der Bewältigung dysfunktionaler Entwicklungen Die widerstreitenden Konsequenzen kollektiver Strategien erfordern einen Ausgleich, d. h. Aktivitäten, durch die die dysfunktionalen und abweichungsverstärkenden Tendenzen kollektiver Strategien gemeistert werden können. Organisationskollektive können grundsätzlich auf zwei Methoden zurückgreifen: Beseitigung und Abschwächung. Der erste Ansatz beseitigt Dysfunktionen durch die Auflösung kollektiver Strategien, während der zweite Stoßdämpfer errichtet, durch die die Auswirkungen dysfunktionaler Entwicklungen abgeschwächt werden.

I. Zur Beseitigung dysfunktionaler Entwicklungen Die in Abbildung 1 dargestellten abweichungsverstärkenden Tendenzen lassen sich durch eine Stabilisierung der explosiven Kausalschleifen beseitigen. Eine Stabilisierung ist möglich, wenn Organisationskollektive auf dysfunktionale Entwicklungen nicht mit Verbesserungsversuchen, sondern mit der Preisgabe ihrer kollektiven Strategie reagieren. Dieser Ansatz bedingt, daß die Top-Management-Teams der beteiligten Organisationen ein umfassendes Verständnis der Konsequenzen kollektiver Strategien haben und jenem gruppendynamischen Druck widerstehen, der zu einer stärkeren Verpflichtung auf die bereits implementierte Strategie führt. Konkret ist eine Stabilisierung explosiver Kausalzusammenhänge möglich, wenn Organisationskollektive auf Umweltvariation und Entscheidungsunsicherheit mit einer Reduktion ihrer kollektiven Vorgehensweisen reagieren und gleichzeitig in stärkerem Maße auf individuelle Wettbewerbsstrategien zurückgreifen. Dieses Verhalten ist in Abbildung 1 durch zwei in Klammern gesetzte Minuszeichen für jene Zusammenhänge angezeigt, die die Variablen Umweltvariation und kollektive Strategien sowie Entscheidungsunsicherheit und kollektive Strategie verbinden. Der Übergang von kollektiven zu Wettbewerbsstrategien verringert den Grad der Vertragsverkettungen, was höhere strategische Flexibilität, geringere Störungseinwirkungen und eine verbesserte Anpassungfahigkeit bewirkt. Eine verbesserte Anpassungsfähigkeit der beteiligten Unternehmungen wirkt sich dämpfend auf das Ausmaß der Entscheidungsunsicherheit aus und macht einen Markt weniger attraktiv für potentiellen Eintritt. Darüber 556

ZfB 59. Jg. (1989), H. 5

493

494

R. Κ. F. Bresser

Kollektive Unternehmensstrategien

hinaus erhöht der Wechsel- von kollektiven zu Wettbewerbsstrategien den Grad der Wettbewerbsverkettungen, ein Umstand, der potentiellen Markteintritt ebenfalls abschreckt und sich deshalb tendenziell reduzierend auf die Höhe der Entscheidungsunsicherheit und der Umweltvariation auswirkt. Durch die freiwillige Aufgabe kollektiver Strategien und ihren Ersatz durch Wettbewerbsstrategien können Organisationen Umweltstörungen besser bewältigen, denn sie schaffen mehr interne Varietät und Unberechenbarkeit für Außenseiter. Sie verhalten sich im Einklang mit dem kybernetischen Gesetz der erforderlichen Varietät, indem sie mehr interne Varietät entstehen lassen, um externer Varietät effektiver begegnen zu können (Conant und Ashby, 1970; Mirow, 1969). Für neue Marktteilnehmer wird es schwieriger, sich in einem Markt zu etablieren, wenn ein Organisationskollektiv Markteintritt mit der Aufgabe kollektiver Strategien und einer Vielzahl individueller Wettbewerbsstrategien beantwortet. 3 Ganz gleich welche Arten von Wettbewerbsstrategien von den Mitgliedern eines sich auflösenden Organisationskollektivs ergriffen werden, der Erfolg von Wettbewerbsstrategien im Management von Umweltinterdependenz begründet sich für die Einzelunternehmung in der Schaffung vorteilhafter Marktpositionen. Unternehmungen, die Wettbewerbsstrategien erfolgreich nutzen, um Umweltinterdependenz abzubauen, sind in der Lage, existierende Wettbewerbsverkettungen außer Kraft zu setzen; sie schaffen sich eine relativ geschützte Marktdomäne und koppeln sich von anderen Wettbewerbern ab (Pfeffer und Salancik, 1978). In einem durch allgemeines Wettbewerbsverhalten gekennzeichneten Markt kann allerdings davon ausgegangen werden, daß es nur einer relativ kleinen Zahl von Unternehmungen im Zeitablauf gelingt, geschützte Marktdomänen zu entwikkeln. Für den größten Teil der Unternehmungen ist der verschärfte Einsatz von Wettbewerbsstrategien nur kurzfristig vorteilhaft und zwar im Sinne wiedergewonnener Flexibilität und Anpassungsfähigkeit; auf lange Sicht ergeben sich stärkere Wettbewerbsverkettungen, mehr Umweltvariation und damit größere Interdependenzprobleme. Abbildung 1 sind diese Effekte den vier Kausalschleifen zu entnehmen, die ursprünglich eine abweichungsreduzierende Dynamik hatten. Wenn durch kollektive Strategien bedingte Umweltvariation und Entscheidungsunsicherheit zu einer Substitution kollektiver Strategien durch Wettbewerbsstrategien führen, steigt die Variation innerhalb einer Organisationsumwelt kontinuierlich an, da intensiveres Wettbewerbsverhalten die Folge ist. Umweltvariation erhöht die Entscheidungsunsicherheit der Einzelunternehmungen, was zusätzliche Wettbewerbsstrategien ( = Umweltvariation) auslöst, usw. Offensichtlich lassen sich die explosiven Entwicklungen kollektiver Strategien durch das Umsteigen von kollektiven auf Wettbewerbsstrategien stabilisieren. Allerdings bewirkt diese Stabilisierung der Kausalschleifen 5 bis 15 eine gleichzeitige Destabilisierung der Kausalschleifen 1 bis 4, die ursprünglich die gewünschten Konsequenzen kollektiver Strategien widerspiegelten. Kausalschleifen 1 bis 4 repräsentieren nun vier explosive Beziehungszusammenhänge; dies ist schematisch durch die (eingeklammerten) Minuszeichen dargestellt. Es ist somit zweifelhaft, ob die durch das Umsteigen von kollektiven auf Wettbewerbsstrategien ausgelöste Umweltdynamik langfristige Vorteile im Management von Umweltinterdependenz ermöglicht. Einerseits gelingt es i. d. R . nur einer kleinen Zahl von Unternehmungen, geschützte Marktpositionen aufzubauen. Andererseits steht der Einsatz von Wettbewerbsstrategien mit den Variablen Umweltvariation und EntscheidungsZfB 59. Jg. (1989), H. 5

557

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

Unsicherheit in einem abweichungsverstärkenden Beziehungszusammenhang. Im Zeitablauf ist es wahrscheinlich, daß der sich ausweitende Wettbewerb turbulente Umweltsituationen, ζ. B. Preiskriege, entstehen läßt. Umweltturbulenzen und die mangelnde Fähigkeit vieler Unternehmungen, sich in diesen Situationen zu behaupten, legen es nahe, daß Organisationen sich wieder auf die Vorteile der Kooperation besinnen und wieder auf kollektive Strategien zurückgreifen, um ihre Organisationsumwelt zu stabilisieren. Die vorangehende Diskussion legt zwei Folgerungen nahe. Erstens, sowohl kollektive als auch Wettbewerbsstrategien führen zu dysfunktionalen Entwicklungen, die das Management von Umweltinterdependenz erschweren. Stabilisierungseffekte im Sinne der Beherrschung interdependenter Umwelten sind von beiden Strategiearten nur auf begrenzte Sicht zu erwarten. Zweitens, kollektive Strategien und Wettbewerbsstrategien scheinen durch eine spezifische Dialektik gekennzeichnet zu sein (Benson, 1977; Bresser und Bishop, 1983; Cornforth, 1971). Ist eine Organisationsumwelt von Wettbewerbsstrategien geprägt, die Umweltturbulenzen und Entscheidungsunsicherheit hervorrufen, entsteht das Verlangen nach geordneter Kooperation, d.h. nach kollektiven Strategien. Die kollektiven Strategien repräsentieren dann für einzelne Organisationen eine Synthese zu dem Wiederspruch zwischen gewünschten und unerwünschten Konsequenzen des Wettbewerbs. Sobald eine Orgnisationsumwelt von kollektiven Strategien geprägt ist, kann sich im Zeitablauf leicht eine entgegengesetzte Entwicklung ergeben. Die dysfunktionalen Konsequenzen kollektiver Strategien lassen dann Wettbewerbsstrategien wiederum als die geeigneteren Mittel des Managements von Umweltinterdependenz erscheinen. Es kann also ein latentes Spannungsverhältnis zwischen kollektiven und Wettbewerbsstrategien angenommen werden, das zu wiederholtem Umsteigen von der einen zu der anderen Strategieart führt. Dies bedeutet, daß Organisationen ihre längerfristige Überlebensfahigkeit nicht durch ein starrer Festhalten an jeweils kollektiven oder Wettbewerbsstrategien sicherstellen, sondern durch einen Prozeß des Wechsels von kollektiven zu Wettbewerbsstrategien und vice versa.

II. Zur Abschwächung dysfunktionaler Entwicklungen Anstatt die Auflösung dysfunktionaler kollektiver Strategien zu bewirken, können Organisationen sich darum bemühen, Dysfunktionen abzuschwächen, so daß sich kollektive Strategien länger nutzen lassen. Dies kann einerseits durch ein Verbessern dysfunktionaler Vereinbarungen angestrebt werden. Obwohl Verbesserungsversuche anfanglich typischerweise erwartet werden können, ist dieser Ansatz symptomatisch für jene Prozesse, durch die die Dysfunktionen kollektiver Strategien explosiven Charakter annehmen. 4 Verbesserungsversuche sind reaktiv und über längere Zeiträume hinweg führen sie eher zu einer Verschärfung dysfunktionaler Entwicklungen als zu ihrer Begrenzung. Ein zweiter Weg der Abschwächung dysfunktionaler Entwicklungen ist proaktiv und zielt darauf ab, ein Organisationskollektiv vor überraschend einsetzender Umweltturbulenz abzuschirmen. Dieser Ansatz wird in der amerikanischen Literatur als „buffering" bezeichnet (Cyert und March, 1963; Thompson, 1967), und er beruht auf der Errichtung von Stoßdämpfern, die Markteintritt oder -variation entweder verhindern oder in seinen 558

ZfB 59. Jg. (1989), H. 5

495

496

R. Κ. F. Bresser Kollektive Unternehmensstrategien

negativen Auswirkungen auf ein Organisationskollektiv begrenzen. U m das vorbeugende Potential des Ansatzes voll zu nutzen, sollten Möglichkeiten des Buffering bereits in die Konzeption einer kollektiven Strategie einfließen und gleichzeitig mit ihr verwirklicht werden. Drei Arten des Buffering erscheinen geeignet, die Auswirkungen dysfunktionaler Entwicklungen abzuschwächen: Eintrittsbarrieren, Anwendungsvariationen und Umweltmunifizenz. Markteintrittsbarrieren (Caves, 1982) lassen Flexibilitäts- und Anpassungsverluste weniger problematisch erscheinen, da sie wie ein Polster wirken, das das Organisationskollektiv von Umweltvariation abschirmt. Der Schutz durch Eintrittsbarrieren ist allerdings begrenzt und oft nur von vorübergehender Natur, da große Unternehmungen sich leicht über Eintrittsbarrieren hinwegsetzen können, und technologische Innovationen der 70er und 80er Jahre den Markteintritt in vielen Industrien erleichtert haben (Abernathy et al., 1981; Goldhar und Jelinik, 1983). Abschwächungen in den Auswirkungen dysfunktionaler Entwicklungen sind auch durch intraindustrielle, interindustrielle oder internationale Anwendungsvariationen kollektiver Strategien möglich. Wenn Organisationskollektive nur in einigen Marktbereichen kooperieren und in anderen konkurrieren, oder wenn sie kollektive Strategien in einigen Industrien oder Nationen anwenden und in anderen Wettbewerbsstrategien verfolgen, können sie die Effekte dysfunktionaler Entwicklungen eingrenzen und die Notwendigkeit eines eventuellen Wechsels von kollektiven zu Wettbewerbsstrategien herauszögern. Für ein Organisationskollektiv, das beispielsweise nur im Bereich des Marketing kooperiert, ist Umweltvariation (etwa in der Form eines abtrünnig werdenden Mitglieds) tendenziell nicht so störend wie für ein Kollektiv, dessen kollektive Strategie jegliches Wettbewerbsverhalten ausschließt. Allerdings sind die durch Anwendungsvariationen möglichen Abschwächungseffekte häufig von nur vorübergehender Natur, da sie strategische Komplexität und einander widerstreitende Umweltdynamiken hervorrufen (Bresser und Harl, 1986). Für jene Wettbewerbsdimensionen, Märkte oder Nationen, die durch eine kollektive Strategie gekennzeichnet sind, lassen dysfunktionale Entwicklungen das Verlangen entstehen, die kollektive Strategie durch Wettbewerb zu ersetzen. Und für jene Dimensionen, Märkte oder Nationen, die durch ungehinderten Wettbewerb gekennzeichnet sind, lassen dysfunktionale Konsequenzen des Wettbewerbs das Bedürfnis nach mehr Kooperation entstehen. Diese einander entgegenlaufenden Effekte erzeugen Komplexität und machen es unwahrscheinlich, daß Anwendungsvariationen mehr als temporäre Abschwächungen dysfunktionaler Konsequenzen ermöglichen. Wenn Organisationen in munifizienten ( = freigebigen) Märkten tätig sind, reduziert dies die Virulenz dysfunktionaler Entwicklungen, denn munifizente Umwelten enthalten eine Fülle von Ressourcen, die es Organisationen erlauben, Reserven zu bilden und unerwartete Umweltvariation zu absorbieren (Cyert und March, 1963; Galbraith, 1973). Unter munifizenten Umweltbedingungen stellen sich Probleme der Marktanteilssicherung oder des Überlebens nicht in der gleichen Schärfe wie unter Konditionen allgemeiner Ressourcenknappheit, und Organisationen können es sich erlauben, mit Mängeln behaftete Strategien zu verfolgen (Aldrich, 1979; Pfeffer und Salancik, 1978). Die große Auswahl munifizenter Umwelten als Schaubühnen kollektiver Strategien stößt ebenfalls auf Grenzen, wenn es darum geht, Organisationsumwelten zu stabilisieZfB 59. Jg. (1989), H. 5

559

Kollektive Unternehmensstrategien Rudi K.F. Bresser

ren. Zwei Entwicklungen können erwartet werden. Erstens, da Umweltmunifizenz rar und attraktiv ist, werden viele Organisationen sich darum bemühen, in einen freigebigen Markt einzutreten. Verstärkter Markteintritt führt zu einer Erosion der Umweltmunifizienz, so daß die Vorteile munifizenter Umwelten meist nur recht kurzlebig sind (Aldrich, 1979). Zweitens, es besteht Gefahr, daß Organisationen ihre Sensitivität gegenüber Umweltveränderungen verlieren, wenn sie in munifizenten Umwelten agieren (Hedberg, 1981). Da Munifizienz sich als eine Situation darstellt, in der jeder einen Zuwachs an Ressourcen verwirklichen kann, verbreitet sich leicht allgemeine Selbstzufriedenheit (Hedberg et al., 1976). Wenn Organisationskollektive im Zeitablauf die Überzeugung gewinnen, sie seien unverletzlich, werden sie abnehmende Munifizienz nicht oder nur zu spät als Bedrohung wahrnehmen, so daß die Dysfunktionen ihrer kollektiven Strategien sie letztlich stark gegenüber innovativen Marktteilnehmern benachteiligen können. Alle drei Arten des Buffering ermöglichen eine Abschwächung der Virulenz dysfunktionaler kollektiver Strategien. Längerfristige Stabilisierungen sind jedoch aus zwei Gründen unwahrscheinlich. Zum einen werden durch die Errichtung von Stoßdämpfern die abweichungsverstärkenden Beziehungszusammenhänge nicht beseitigt, durch die kollektive Strategien und andere Variablen miteinander verbunden sind. Maßnahmen des Buffering können die Auswirkungen dysfunktionaler Entwicklungen lediglich aufhalten und abschwächen. Über längere Zeiträume hinweg steigt die Wahrscheinlichkeit, daß sich die abweichungsverstärkende Dynamik kollektiver Strategien durchsetzt und bestehende Abschwächungsverfahren versagen. Zum anderen sind die drei Arten der Abschwächung selbst von Widersprüchen gekennzeichnet. Sie können z. B. strategische Komplexität oder Intensivität gegenüber Umweltveränderungen hervorrufen und somit den Erfolg der jeweiligen Buffering-Maßnahme in Frage stellen oder beenden.

I. Schlußbemerkungen Die vorliegende Diskussion kollektiver Unternehmensstrategien stellt einen theoretischen Bezugsrahmen bereit, anhand dessen sich verschiedene Formen betrieblicher Kooperation integrieren und analysieren lassen. Aus organisationstheoretischer Sicht ergeben sich die Chancen kollektiver Unternehmensstrategien hauptsächlich aus den Möglichkeiten, interdependente Organisationsumwelten zu stabilisieren. Die Risiken kollektiver Strategien (Potentiale für strategische Inflexibilität, verstärkte Störungseinwirkungen, die Attraktion neuer Marktteilnehmer und abweichungsverstärkende Eigendynamiken) sind allerdings beträchtlich; sie können nicht nur zu Umweltvariation und Entscheidungsunsicherheit führen, sondern auch die Existenz beteiligter Organisationen gefährden. Organisationen können versuchen, die Dysfunktion ihrer kollektiven Strategien zu beseitigen oder abzuschwächen. Überlegungen zum Thema „Beseitigung dysfunktionaler Konsequenzen" legen nahe, daß zwischen kollektiven Strategien und Wettbewerbsstrategien ein Spannungsverhältnis besteht, das sich als ein Prozeß des Wechsels von der einen zu der anderen Strategieart entfalten kann. Obwohl reaktive und proaktive Abschwächungsversuche die Vorteile kollektiver Strategien mitunter über längere Zeiträume hinweg erhalten, verzögern Abschwächungsmaßnahmen i. d. R. nur die dysfunk560

ZfB 59. Jg. (1989), H. 5

497

498

R. Κ. F. Bresser Kollektive Unternehmensstrategien

tionalen Auswirkungen kollektiver Strategien. Die Notwendigkeit einer eventuellen Preisgabe kollektiver Strategien und ihre (vorübergehende) Verdrängung durch Wettbewerbsstrategien, d.h. das Spannungsverhältnis zwischen beiden Strategiearten, wird durch Abschwächungsversuche nicht aufgehoben. Organisationen müssen somit Kapazitäten entwickeln und bewahren, die einen Wechsel zwischen kollektiven und Wettbewerbsstrategien ermöglichen. Die Möglichkeiten der Entwicklung (und Bewahrung) von Kapazitäten für wiederholten Strategiewechsel haben psychologische, materielle und informationstechnische Wurzeln. Unternehmensleitungen müssen sich psychologisch auf strategische Impermanenz einstellen. Wenn Impermanenz als Normalfall angesehen wird, können Top-Management-Teams der Gefahr einer eskalierenden Verpflichtung auf bereits implementierte Strategien entgehen. Wiederholter Strategiewechsel läßt sich materiell durch die Bereitstellung von Ressourcen oder Rücklagen sicherstellen, die eventuelle negative Konsequenzen einer implementierten Strategie absorbieren und neue Strategien alimentieren können. Darüber hinaus benötigen Unternehmensleitungen hochentwickelte Management-Informations-Systeme, um dysfunktionale Entwicklungen implementierter Strategien beurteilen und antizipieren zu können. Für den Betriebswirt bietet das Thema „Kollektive Unternehmensstrategien" mehrere Ansatzpunkte für empirische Forschung. Die erwünschten und dysfunktionalen Konsequenzen kollektiver Strategien, die beschriebenen abweichungsreduzierenden und abweichungsverstärkenden Tendenzen, das Spannungsverhältnis zwischen kollektiven und Wettbewerbsstrategien und die Effekte des „Buffering" müssen alle als erste, vorläufige Hypothesen angesehen werden, die der empirischen Überprüfung bedürfen.5

Anmerkungen 1 Diese Definition verdeutlicht, daß kollektive Strategien aus wettbewerbspolitischer Sicht problematisch sein können. Während gesamtwirtschaftliche Gesichtspunkten vor allem für Ökonomen von Interesse sind, widmet die amerikanische Managementliteratur (wie auch dieser Beitrag) der einzelwirtschaftlichen Problematik des Themas ihre Aufmerksamkeit. Zur Wichtigkeit einer einzelwirtschaftlichen Bewertung kollektiver Strategien vgl. Abschnitt E. 2 Das Aggregationsniveau der in Abbildung 1 einbezogenen Variablen variiert zwischen zwei Stufen, der des Organisationskollektivs und der nächst höheren Stufe des Marktes (der Industrie). Obwohl dieser Beitrag kollektive Strategien aus der Sicht der beteiligten Organisationen (des Organisationskollektivs) analysiert, ist die Einbeziehung von Variablen unterschiedlichen Niveaus notwendig. Dies liegt daran, daß Aspekte der Interdependenz innerhalb eines Marktes (einer Umwelt) das Verhalten der Mitglieder von Organisationskollektiven beeinflussen, ebenso wie die Aktivitäten von Organisationskollektiven die Qualität der Umweltinterdependenz verändern. In Abbildung 1 repräsentieren die Variablen 1, 2, 4 und 7 das höhere Niveau der Organisationsumwelt (des Marktes oder der Industrie) und die Variablen 3, 5, 6, 8 und 9 das Niveau des Organisationskollektivs. 3 Kollektive Strategien, die sich durch Informalität, Kurzfristigkeit und geringe Teilnehmerzahlen auszeichnen, sind am besten zur Schaffung erforderlicher Varietät geeignet, denn diese Bedingungslagen erleichtern das Umsteigen von kollektiven auf Wettbewerbsstrategien. Während informale Vereinbarungen genauso bindend sein können wie formale (Galbraith, 1973), kann eine informale Vereinbarung im Falle dysfunktionaler Entwicklungen jedoch leichter aufgegeben werden. Die Preisgabe einer formal fixierten Vereinbarung ist zeitraubender und komplizierter, da zu ihrer Veränderung häufig zusätzliche formale Übereinkünfte notwendig sind. Darüber

ZfB 59. Jg. (1989), H. 5

561

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

hinaus haben mit der Auflösung nicht übereinstimmende Parteien oft die Möglichkeit, die Auflösung durch Rechtsmittel zu verzögern (Gottfredson und White 1981). Kurzfristig angelegte kollektive Strategien erleichtern Strategiewechsel, da sie organisatorische Verpflichtungen zeitlich begrenzen, und Strategien mit geringer Teilnehmerzahl lassen sich leichter verändern, da das Potential für konfligierende Interessen tendenziell geringer ist (Pfeffer und Salancik, 1978). Falls Organisationskollektive formal geregelte, langfristige kollektive Strategien implementieren wollen, die für eine größere Zahl von Mitgliedern maßgeblich sein sollen, ist es ratsam, in die Vereinbarung eine Option aufzunehmen, die es den beteiligten Organisationen erlaubt, die kollektive Strategie vorübergehend und unmittelbar aufzugeben und durch Wettbewerbsverhalten zu ersetzen, wenn Markteintritt stattfindet oder bevorsteht. 4 Vgl. Abschnitt G. 5 Der vorliegende Beitrag basiert auf der anglo-amerikanischen Management- und Organisationsliteratur. Die interkulturelle Bedeutsamkeit des Themas legt nahe, daß in einzelnen Ländern Forschungsarbeiten existieren, die eine weitere (kulturspezifische) Präzisierung des vorgeschlagenen Modells ermöglichen. Derartige Impulse könnten ζ. B. von den zahlreichen deutschen Arbeiten über Verbände und Verbandsbetriebe ausgehen (Blümle, 1987; Lehmann, 1980). Aus Platzgründen muß dieser Integrationsversuch einer eigenständigen Untersuchung vorbehalten bleiben.

Literatur Abernathy, W. ].; Clark, K. B.; Kantrow, Α. M. (1981): The new industrial competition, Harvard Business Review 59 (5), S. 68-81. Aldrich, H. E. (1979): Organizations and Environment, Prentice-Hall, Englewood Cliffs, NJ. Astley, W. G.; Fombrun, C. J. (1983 a): Collective strategy: Social ecology of organizational environments, Academy of Management Review 8, S. 576-587. Astley, W. G.; Fombrun, C. J. (1983 b): Technological innovation and industrial structure: The case of telecommunications, in: Lamb, R. (Hrsg.), Advances in Strategic Management, Bandi, S. 205-229, JAI Press, Greenwich, CT. Benson, J. K. (1977): Organizations: A dialectical view, Administrative Science Quarterly 22, S. 1-21. Blümle, Ε. Β. (1987): Verbände, Führung in, in: Kieser, Α.; Reber, G.; Wunderer, R. (Hrsg.), Handwörterbuch der Führung, S. 2004-2015, Poeschel, Stuttgart. Bresser, R. K. (1984): The captives of collective strategies, Proceeding of the American Institute for Decision Sciences, S. 383-395, Toronto. Bresser, R. K. (1988): Matching collective and competitive strategies, Strategic Management Journal 9, S. 375-385. Bresser, R. K.; Bishop, R. C. (1983): Dysfunctional effects of formal planning: Two theoretical explantions, Academy of Mangement Review 8, S. 588-599. Bresser, R. K.; Harl, J. E. (1986): Collective strategy: Vice or virtue?, Academy of Management Review 11, S. 408 -427. Carney, M. G. (1987): The strategy and structure of collective action, Organization Studies 8, S. 341-362. Caves, R. (1982): American Industry: Structure, Conduct, and Performace, 5. Aufl., Prentice-Hall, Englewood Cliffs, NJ. Conant, R. C.; Ashby, R. W. (1970): Every good regulator of a system must be a model of that system, International Journal of Systems Science 1 (2), S. 89-97. Cornforth, M. (1971): Materialism and the Dialectical Mind, International Publishers, New York. Cyert, R. M.; March, J. G. (1963): A Behavioral Theory of the Firm, Prentice-Hall, Englewood Cliffs, NJ. Edström, Α.; Högberg, Β.; Norbäck, E. (1984): Alternative explanations of interorganizational cooperation: The case of joint programs and joint ventures in Sweden, Organization Studies 5, S. 147-168.

562

ZfB 59. Jg. (1989), H. 5

499

500

R. Κ. F. Bresser Kollektive Unternehmensstrategien

Emery, F. E.; Trist, E. L. (1965): The causal texture of organizational environments, Human Relations 18, S. 2 1 - 3 2 . Fombrun, C. J.; Astley, W. G. (1982): The telecommunications community: An institutional overview, Journal of Communication 32 (4), S. 56-68. Fombrun, C. J.; Astley, W. G. (1983 a): Beyond corporate strategy, Journal of Business Strategy 4 (1), S. 4 7 - 5 4 . Fombrun, C. J.; Astley, W. G. (1983 b): Strategies of collective action: the case of the financial services industry, in: Lamb, R. (Hrsg.), Advances in Strategic Management, Band 2, S. 125—139, JAI Press, Greenwich, CT. Frankfurter Allgemeine Zeitung (1986): AGA und L'Air Liquide lösen Partnerschaft, Nov. 11, S. 16. Galbraith, J. (1973): Designing Complex Organizations, Addison-Wesley, Reading, MA. Goldhar, J. D.; Jelinek, M. (1983): Plan for economies of scope, Harvard Business Review 61 (6), S. 141-148. Gottfredson, L. S.; White, P. E. (1981): Interorganizational agreements, in: Nystrom, P. C.; Starbuck, W. H. (Hrsg.), Handbook of Organizational Design, Band 1, S. 471 - 4 8 6 , Oxford University Press, Oxford. Hall, R. H.; Clark, J. P.; Giordano, P. C.; Johnson, P. V.; Van Rökel, M. (1977): Patterns of interorganizational relationships, Administrative Science Quarterly 22, S. 457-474. Hannan, M.; Freeman, J. (1977): The population ecology of organizations, American Journal of Sociology 82, S. 929-964. Harrigan, Κ. (1985): Strategic Flexibility, Lexington Books, Lexington, MA. Hedberg, B. (1981): How organizations learn and unlearn, in: Nystrom, P. C.; Starbuck, W. H. (Hrsg.), Handbook of Organizational Design, Band 1, S. 3 - 2 7 , Oxford University Press, Oxford. Hedberg, B.; Nystrom, P. C.; Starbuck, W. H. (1976): Camping on seesaws: Prescriptions for a self-designing organization, Administrative Science Quarterly 21, S. 4 1 - 6 5 . Hickson, D.; Hinings, C.; Lee, C.; Schneck, R.; Pennings, J. (1971): A strategic contingencies theory of intraorganizational power, Administrative Science Quarterly 16, S. 216-227. Hofer, C. W.; Schendel, D. (1978): Strategy Formulation: Analytical Concepts, West, New York. Janis, I. (1972): Victims of Groupthink, Houghton-Mifflin, Boston. Lehmann, Η. (1980): Stand und Entwicklung der Organisationsforschung im Hinblick auf Verbände - ein theoretischer Abriß, Zeitschrift für Organisation 49, S. 265-272. Le Monde (1986): L'Air Liquide rompt ses accords europeens avec le groupe suédois AGA, Nov. 14, S. 38. Maruyama, M. (1963): The second cybernetics: Deviation-amplifying mutual causal processes, American Scientist, 51, S. 164-179. Miles, R. H. (1980): Macro Organizational Behavior, Boodyear, Santa Monica, CA. Mirow, H. M. (1969): Kybernetik - Grundlage einer allgemeinen Theorie der Organisation, Gabler, Wiesbaden. Parker, G. G. (1981): Now management will make or break the bank, Harvard Business Review 81 (6), S. 140-148. Pennings, J. M. (1981): Strategically interdependent organizations, in: Nystrom, P. C.; Starbuck, W. H. (Hrsg.), Handbook of Organizational Design, Band 1, S. 434-455, Oxford University Press, Oxford. Pfeffer, J.; Salancik, G. R. (1978): The External Control of Organizations, Harper and Row, New York. Scherer, F. M. (1980): Industrial Market Structure and Economic Performance, 2. Aufl., Rand McNally, Chicago. Staw, B. M. (1981): The escalation of commitment to a course of action, Academy of Management Review 6, S. 577-587. Thakur, M.; Hoffman, W. (1988): Cooperative strategizing: A study of selected firms in England, Holland and the United States, in: Millar, C. (Hrsg.) Proceedings of the Academy of International Business, U K Chapter, S. 119-136, Thames Polytechnic, London.

ZfB 59. Jg. (1989), H. 5

563

Kollektive Unternehmensstrategien

Rudi K.F. Bresser

Thompson, J. D. (1967): Organizations in Action, McGraw Hill, New York. Thompson, Α. Α.; Strickland, A. J. (1987): Strategie Management: Concepts and Cases, 4. Aufl., BPI, Dallas, TX. Weick, Κ. E. (1979): The Social Psychology of Organizing, 2. Aufl., Addison Wesley, Reading, MA.

Summary This paper discusses opportunities and risks of collective strategies which can supplement or supplant business and corporate level strategies. Some dysfunctional results of collective strategies include their tendencies to reduce strategic flexibility, to increase the impact of external disturbances, and to attract new entrants. A model is proposed which provides an integrated analysis of collective strategy consequences and which highlights the relationship between collective and competitive strategies. To demonstrate the intercultural relevance of the topic, case examples involving European and American firms are provided. Possibilities for avoiding and muting the dysfunctions of collective strategies are evaluated.

16: 21:

564

Unternehmenspolitik Unternehmensfiìhrung

ZfB 59. Jg. (1989), Η. 5

501

502

Κ. M. Eisenhardt/C. Β. Schoonhoven

Resource-based View of Strategic Alliance Formation: Strategic and Social Effects in Entrepreneurial Firms Kathleen M. Eisenhardt · Claudia Bird Schoonhoven Department of Industrial Engineering and Engineering Management, Stanford University, Stanford, California 94305 The Amos Tuck School of Business, Dartmouth College, Hanover, New Hampshire 03755

Abstract Why do firms form strategic alliances? The traditional theoretical answer has been transaction cost explanations. Yet, these explanations which center on transaction characteristics, static efficiency, and routine situations do not capture the strategic and social factors which propel many firms into alliance formation. In this study, however, we combine these alternative social and strategic explanations for alliance formation. Consistent with these explanations, we find that alliances form when firms are in vulnerable strategic positions either because they are competing in emergent or highly competitive industries or because they are attempting pioneering technical strategies. We also find that alliances form when firms are in strong social positions such that they are led by large, experienced, and well-connected top management teams. The underlying logic of alliance formation is, thus, strategic needs and social opportunities. We develop these findings by extending the resource-based view of the firm to alliance formation and then examining the resulting hypotheses using product development alliances. The study is longitudinal and focuses on entrepreneurial semiconductor firms. Overall, strategic and social explanations of organizational phenomena as well as industry, firm, and top management team factors emerge as central in the paper. This suggests that these factors are relevant for predicting alliance formation, especially in high-velocity industries such as semiconductors. We conclude that failure to include social and strategic explanations creates an impoverished view of alliance formation.

(Strategic Alliances·, Entrepreneurship; Top Management Teams; Product Innovation; Resource-based View of the Firm)

Establishing strategic alliances is a central strategy for many contemporary firms (e.g., Harrigan 1988, Parkhe

136

ORGANIZATION S C I E N C E /

Vol.

7,

No.

2,

March-April

1993, Powell 1990). Such cooperative relationships can help firms to conserve resources and share risks (e.g., Hamel et al. 1989, Ohmae 1989). Alliances can also serve as signals of enhanced legitimacy for firms (e.g., Baum and Oliver 1991) and as opportunities for gaining new competencies (e.g., Hagedoorn 1993, Hamel et al. 1989, Hennart 1991). In addition, alliances can help firms to gain market power (Hagedoorn 1993), move more quickly into new markets and technologies, and create options for future investments (Kogut 1991). Yet, despite their attractiveness, cooperative relationships with other organizations can be problematic. For example, alliance relationships can have high transaction costs (e.g., Hennart 1991, Williamson 1991). Moreover, alliances can lull firm managers into failing to develop important firm capabilities and can be conduits by which certain types of technology and other core competencies are easily siphoned from the firm (e.g., Hamel et al. 1989, Teece 1987). Alliances can also reduce revenue streams by forcing firms to share profits (e.g., Shan 1990). Because of this mix of advantages and disadvantages, firms are likely to adopt strategic alliances at varying rates. The principal theoretical approach for understanding when strategic alliances form is transaction cost economics (e.g., Hennart 1988,1991; Pisano and Teece 1989; Shan 1990; Williamson 1991). This theory emphasizes transaction cost efficiency as the motivation for such cooperation. It has been effective in predicting vertical integration among suppliers and buyers in mature industries such as automobile manufacturing, and the use of equity as a governance mechanism (e.g., Hennart 1991, Osborn and Baughn 1990). However,

1996

1047-7039/96/0702/0136/S01.25 Copyright Ο 1996. Institute for Operations Research and the Management Sciences

Strategie Alliance Formation

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

while helpful, the logic of transaction cost minimization does not capture many of the strategic advantages of alliances such as learning, creation of legitimacy, and fast market entry that we noted at the outset. As Ghosal and Moran (1995) observe, the theory is most relevant to static efficiency and routine situations. The surprisingly small empirical literature on alliance formation also often departs from transaction cost explanations. Shan (1990) emphasizes strategic factors. In his study of entrepreneurial firms in biotechnology, he found that high competition was associated with alliance formation. Similarly, Mitchell and Singh (1992) observed that late entrants in the medical diagnostics industry were likely to enter alliances. Kogut et al. (1992) and Gulati (1993) emphasize the social basis of alliance formation, finding current alliance formation to be related to past alliance relationships. Taken together, these studies suggest that both strategic and social factors are critical to alliance formation. Yet, to our knowledge, no theory or study of alliance formation combines these two sets of factors. The purpose of our article is to relate both strategic and social factors to strategic alliance formation. Specifically, we begin with the resource-based perspective in which firms are seen as bundles of resources (e.g., Wernerfelt 1984, Peteraf 1993). By resources we mean strengths or assets of the firm that may be tangible (e.g., financial assets, technology) or intangible (e.g., reputation, managerial skills). We then extend this view to alliances by arguing that strategic alliances arise when firms in vulnerable strategic positions need the resources that alliances bring or when firms in strong social positions capitalize on their assets to create alliance opportunities. Alliances are, therefore, cooperative relationships driven by a logic of strategic resource needs and social resource opportunities. In comparison with transaction-cost thinking, this perspective emphasizes (1) strategic and social factors, not transaction costs, (2) characteristics of the firm (e.g., strategy, top management), not the transaction, and (3) a theoretical logic of needs and opportunities, not efficiency. The findings highlight the importance of vulnerable strategic position (i.e., new markets, many competitors, and pioneering technology) and strong social position (i.e., large, well-connected, and high status top management team) in the formation of alliances, especially by entrepreneurial firms. They also reveal what is perhaps the most fundamental irony of alliancing: firms must have resources to get resources. Our identification of factors that influence alliance formation also

ORGANIZATION S C I E N C E / V O I .

7, No. 2, March-April 1996

Strategic Alliance

503

Formation

contributes to understanding of entrepreneurial organizations because of the importance of such relationships to the success of these firms (e.g., Baum and Oliver 1991, Miner et al. 1990).

Background and Theoretical Development Strategic Needs for Cooperation Why do some firms form strategic alliances while others do not? Two themes from the literature on cooperation seem relevant to this question. One theme emphasizes the strategic aspects of cooperation. The central premise is that people engage in cooperation because of appropriate payoff structures (e.g., Axelrod 1984, Parkhe 1993). That is, people cooperate when the payoff for cooperation exceeds that of proceeding alone. When the payoff shifts to favor solo endeavors, people will disengage from cooperative activities. Thus, this perspective takes an atomistic view of individuals (e.g., Hennart 1988, Pisano and Teece 1989, Williamson 1991), emphasizing the nature of the payoff structure and the role of self-interest in cooperation (e.g., Axelrod 1984, Parkhe 1993). In the case of alliances, a high payoff for cooperation is particularly likely when firms are in vulnerable strategic positions. Since strategic position depends on characteristics of both firm strategy and market (Conner 1994), vulnerable strategic positions occur when firms are in difficult market situations, or are undertaking expensive or risky strategies. In such situations, alliances can provide critical resources, both concrete ones such as specific skills and financial resources (e.g., Hamel et al. 1989, Pisano and Teece 1989) as well as more abstract ones such as legitimacy and market power (e.g., Weiwel and Hunter 1985, Hagedoorn 1993, Baum and Oliver 1991) that improve strategic position. For example, firms in highly competitive markets have vulnerable strategic positions because margins are low and product differentiation is difficult. Hie resources such firms acquire through alliances can enable them to share costs or to gain differentiable product technologies which outweigh the disadvantages of alliance formation. In contrast, if the market is munificent or the firm is pursuing a strategy for which it has extensive resource capabilities, there is much less incentive to cooperate. Firms are more likely to continue alone. Thus, alliance formation occurs when firms are in vulnerable strategic positions such that the payoffs to cooperation are high. The underlying logic is need.

137

504

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

Social Opportunities for Cooperation A second theme emphasizes the social aspects of cooperation. As Granovetter (1992) notes, all action, including economic action, is embedded in a social fabric of opportunities to interact. Interaction and ultimately cooperation are likely to happen among people who know one another or, as Heimer (1992) writes, are "friends". These personal relationships create opportunities for cooperation by deepening awareness, trust, and commitment among parties within the relationship (Larson 1992, Ouchi 1980). For example, Nohria (1992) linked interpersonal contacts within the Route 128 business community to implications for cooperative action among firms. In addition to direct interpersonal contact, status and reputation also enhance the likelihood of cooperation (Podolny 1994). These qualities signal the skill and trustworthiness of potential partners and so facilitate cooperation, particularly when there is high uncertainty. For example, D'Aveni (1990) found that creditors were more likely to support failing firms when the top managers of the firms had prestigious backgrounds. Overall, the evolution of awareness, mutual knowledge, and trust through status and reputational processes and, more directly, social interaction is central to the creation of cooperative relationships. In the case of alliances, awareness, status, mutual knowledge, and trust are likely to be present when firms have strong social positions. Strong social position in alliance formation, as in other interfirm relationships (e.g., Davis 1991, Galaskiewicz and Wasserman 1989), depends upon social advantages such as the personal relationships, status, and reputation of firms and key individuals within those firms. Strong social positions lead to alliance formation because extensive personal relationships create an awareness of opportunities for alliancing as well as knowledge and trust among potential partners. Personal relationships set the stage for alliance formation since "the key is who you know" (Larson 1992, p. 84). In addition, strong social positions lead to alliance formation because high status and reputation signal the quality of the firm and attract partners who want to associate with high-status others (Podolny 1994). Such signals mitigate the risks of alliance formation such as poor implementation, opportunism, and appropriation of knowledge (Larson 1992). In contrast, if firms and their managers lack the resources of a strong social position, they may be unaware of opportunities or may have little information with which to assess the trustworthiness and commitment of potential ) They may also lack the status and reputation necessary to attract good partners and

138

Strategic Alliance Formation

opportunities. They may be forced (or, given their choice of partners, prefer) to continue alone. In sum, alliance formation is enabled by firms' having strong social positions. The underlying theoretical logic is opportunity. Combining Strategic Needs and Social Opportunities The resource-based view of the firm can be extended to combine these two themes of alliance formation. The resource-based view assumes that firms are bundles of resources (e.g., Penrose 1959, Wernerfelt 1984). These resources can be thought of as strengths, advantages, or assets of the firm, and as Wernerfelt (1984) observes, can be either tangible or intangible. Examples of such resources are technical know-how, management skills, capital, and reputation. The resourcebased view can be extended to alliances by arguing that resources provide both the needs and the opportunities for alliance formation. That is, alliances form when firms are in vulnerable strategic positions for which they need additional resources that alliances can provide to compete effectively, or when firms are in strong social positions such that they have the resources necessary to know, attract, and engage partners.

Resource-based View of Strategic Alliances Strategic Position Hypotheses Competition plays an important role in determining strategic position. The argument is that when a firm faces many competitors, its strategic position is vulnerable. Resources are squeezed, profits are stressed, and even survival is threatened (Klepper and Graddy 1990, Shan 1990). For example, as Sahlman and Stevenson (1985, p. 23) describe in their study of the disk drive industry: "The greatest single industrywide problem that arose in 1983 and 1984 was the increased intensity of competition.... Given the high fixed costs, including R & D, margins fell sharply." In addition, because of many look-a-like firms and products in highly competitive markets, it is difficult for firms to set themselves apart from others. In contrast, in markets with fewer competitors, profits are higher (Klepper and Graddy 1990, Porter 1980) and survival is more likely (Carroll and Hannan 1989). Alliances improve the strategic position of firms in competitive markets by providing resources from other firms that enable them to share costs and risks. Such resources give firms a cushion to weather business downturns and other setbacks, and ensure more even and predictable resource flows (Baum and Oliver 1991,

ORGANIZATION S C I E N C E / V O I .

7, No. 2, March-April 1996

Strategie Alliance Formation

K A T H L E E N M . E I S E N H A R D T AND CLAUDIA B I R D S C H O O N H O V E N

Miner et al. 1990, Ohmae 1989). This buffering and cost sharing eases profit pressures, which are particularly intense in highly competitive industries (Dunne et al. 1988, Sahlman and Stevenson 1985), and gives firm partners the slack they need to ride out difficult times and to learn better ways to compete. Alliances also improve the strategic position of firms by enhancing legitimation. Cooperating with another organization can give a firm visibility and signal enhanced status to would-be buyers, suppliers, and employees (Baum and Oliver 1991, Weiwel and Hunter 1985). Such alliances help to distinguish firms from other competitors, which is particularly important in crowded markets. For example, in the early 1980s, executives at Sun Microsystems deliberately established alliances with major firms such as Kodak and AT&T to distinguish the firm from other microcomputer companies. In addition, alliances can often improve the market power of a firm, either because the alliance partner is a customer for the product or because the distribution channels and buying power of the partners can be combined. Also, the close interfirm relationships within alliances provide specific knowledge-based resources such as manufacturing or customer information (Hamel et al. 1989, Shan 1990, Teece 1987) that can give firms a further edge over competitors. Finally, to the extent that firms ally with competitors, highly competitive markets are also likely to have more potential partners. HYPOTHESIS 1. The greater the number of competitors, the greater the rate of alliance formation. Market stage also affects strategic position. As many authors have observed (e.g., Eisenhardt and Schoonhoven 1990, Hambrick et al. 1982, Porter 1980), markets evolve through several configurations of attributes or stages. Initially, emergent-stage markets arise and are small, new, and characterized by a lack of product clarity (e.g., Anderson and Zeithaml 1984, Porter 1980). Market share can shift dramatically. The winning technology and appropriate distribution channels are often unclear as is the eventual direction of the market (Klepper and Graddy 1990). Such markets may take a long time to develop and some never become viable. While some firms will be very successful in emergent markets, many will fail (Klein 1977, Tushman and Anderson 1986). With such competitive instability, strategic position is precarious. Alliances improve strategic position in emergent markets in several ways. They can provide financial

ORGANIZATION S C I E N C E / V O I . 7 , N o . 2 , M a r c h - A p r i l 1 9 9 6

Strategic Alliance

505

Formation

resources that enable cost- and risk-sharing with other firms (Miner et al. 1990, Ohmae 1989). Extra resources are particularly helpful because emergent markets may take a long time to become viable and the winning paths are often not clear at the outset (e.g., Porter 1980, Van de Ven and Polley 1992). In addition, alliances can help to legitimate a new market. By cooperating with an important potential customer or a competitor, a firm may be able to signal that the market will become established. Further, because winning paths are often not clear at the outset, an alliance may give a boost to the local firm's approach to the market by tying other firms and their resources to that approach. These arguments suggest that the rate of alliance formation will be high in emergent markets. Growth-stage markets are large markets characterized by rapid increases in demand (Anderson and Zeithaml 1984). There may be multiple options for how to compete, multiple candidates for the dominant design, and rapid turnover of competitors (e.g., Porter 1980, Tushman and Anderson 1986). Consequently, market share (Hambrick et al. 1982) as well as competitive structure and technology are volatile (Anderson and Zeithaml 1984, Tushman and Anderson 1986). As Eisenhardt (1989) found, fast pace and quick moves are keys to success in such high-velocity environments. Alliancing can improve the strategic position of firms in growth-stage markets by creating flexibility. In alliances, firms are less constrained by the fixed resources of vertical integration and thus are more able to make adjustments to resources as conditions change (Ohmae 1989). Alliances may also help firms to obtain needed skill-based resources quickly (Hamel et al., 1989, Shan 1990). This is especially true for tacit skills which may be slow to develop in-house and difficult to buy (e.g., Powell 1990, Hennart 1991). However, growth-stage markets are typically ones in which many finns prosper. Strategic position is usually strong, lessening the need for alliancing. In addition, although alliances may enable firms to adapt and leam, substantial evidence suggests that gaining resources from alliances can be slow and difficult (e.g., Larson 1992, Heikal 1992, Parkhe 1993). Because of this mix of pressures, the rate of alliance formation is likely to be lower in growth-stage markets than in emergent ones. Eventually, markets mature. In contrast to emergent and growth-stage markets, mature-stage markets are more stable as demand approaches saturation (e.g., Anderson and Zeithaml 1984, Klepper and Graddy 1990). Dominant product designs and process technologies are usually clear, and technology, market share, and competition are much slower to change

139

506

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

(Hambrick et al. 1982). Hence, uncertainty is relatively low, and speed and flexibility are not critical. Moreover, firms often have resources from past investments and knowledge which they have built up over time that they may be reluctant to share in alliances. Alliances, therefore, offer firms few advantages in mature markets (Hagedoorn 1993). In addition, to the extent that firms have already formed alliances in preceding market-stages, they may have less need to form new ones. HYPOTHESIS 2. The rate of alliance formation is higher in emergent-stage markets than in growth-stage markets, and higher in growth-stage markets than in mature-stage markets. Stategic position depends not only on the characteristics of the market, but also on firm strategy. Although strategy can be characterized in several ways (e.g., Miles and Snow 1978, Porter 1980, Rumelt 1974), the most relevant strategy difference across technologybased firms is the degree of innovation (Boeker 1989, Maidique and Patch 1982). Maidique and Patch (1982) describe how differences in strategy affect the needed level of resources. Technically innovative strategies demand a high level of competence in basic technology and so require substantial resources. Innovative technology rarely works initially, and often takes a long time, if ever, to become viable (Dosi 1988, Jelinek and Schoonhoven 1990). During this time, substantial resources can be consumed. An alliance partner can improve the strategic position of a firm with a pioneering strategy. If the alliance directly involves the creation of technical innovations (e.g., joint product development, technology exchange), it enables firms to gain financial and other resources for developing such technology (Hamel et al. 1989). If it does not (e.g., manufacturing or marketing alliances), the alliance may enable firms to concentrate resources on the difficult job of executing a highly innovative strategy while leveraging the resources of the partner in other functional areas. In addition, the commercial viability of a new technology is often unclear and technological standards can evolve for political or social reasons, not just superior technology (David 1985). For example, Betamax was considered to be technically superior to VHS as a video recording format, but, for a variety of nontechnical reasons including interfirm cooperation, VHS became the standard (Arthur 1990). This example suggests the importance of legitimation for pioneering technologies. Alliances can help firms to gain this legitimacy by tying other firms and their resources to the technology. Further, if the alliance

140

Strategic Alliance Foimalion

partner is a customer for the new technology or has customers who might be, the alliancing firm gains an additional edge. Therefore, innovative strategy is likely to be associated with alliance formation. In contrast, firms with less innovative strategies are less likely to need alliances. The technologies associated with such strategies may be new, but more evolutionary than revolutionary (Eisenhardt and Schoonhoven 1990). The time and resources needed to develop them are, therefore, more certain, less risky, and usually less substantial. Hence, less innovative strategy is less likely to be associated with alliance formation. HYPOTHESIS 3. The more innovative the strategy, the greater the rate of alliance formation. Social Position Hypotheses The above hypotheses emphasize vulnerable strategic position. That is, firms form alliances when they operate in emergent markets, in high competition, and with pioneering technology. However, alliance formation also relies on social position. Strong social position enhances alliance formation by capitalizing on advantages such as contacts, reputation, and status that create opportunities for alliance formation. Because top management team members are often the conceptualizes of alliancing strategy and the key sources of leads to potential alliancing partners, the social position of top managers is particularly relevant to opportunities for alliance formation. Given the power of social relationships (e.g., D'Aveni 1990, Granovetter 1992, Heimer 1992), teams with strong social resources are more likely to gain alliances if they need them and more likely to be offered attractive alliancing arrangements even if they do not. The social advantages of top managers are particularly critical for young firms because they have not had the time to establish firm-level networks. In addition, top managers usually serve as the principal negotiators of the alliancing agreement. The level of trust between negotiating principals and their negotiating skills are important resources that can make or break the alliance deal (Larson 1992). The characteristics of top managers can also serve as signals of the quality of the firm (e.g., D'Aveni 1990, Finkelstein 1994) and thus improve the firm's chances of being a sought-after alliance partner or a successful suitor. We, therefore, argue that the formation of strategic alliances depends greatly on the social position of top managers to create opportunities for alliance formation. Several attributes of the top management team are especially likely to contribute to social position. One is

ORGANIZATION S C I E N C E / V O I .

7, No. 2, March-April 1996

Strategic Alliance Formation

507

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN Stratégie Alliance Formation

team size. Large top management teams are likely to have more extensive connections and relationships with potential partnering organizations than small ones. These advantages enhance the firm's likelihood of either finding or being found as an alliancing partner. Given the typically extensive personal and business contacts possessed by most senior-level managers, the likelihood of having a friend, neighbor, former coworker, or business associate who works for or knows a potential alliance partner is greater for a large top management team than for a small one. Furthermore, these personal relationships lay the groundwork of trust and knowledge that is important for interfirm cooperation (Galaskiewicz and Wasserman 1989) and alliance formation (Larson 1992). Finally, since alliances are usually sought after, orchestrated and negotiated at the highest levels of the organization, there is a greater chance that one or more executives will have the time to devote to alliance formation when the team is large (Eisenhardt and Schoonhoven 1990). When the team is small, it is more likely that all of the top executives will be tied up with pressing, short-term operating matters, and so lack the time to develop alliance relationships. HYPOTHESIS 4. The greater the size of the top management team, the greater the rate of alliance formation. A second attribute that contributes to the social position of top managers is the number of previous focal industry employers that members of the top management team have had. Top management teams whose members have worked for many firms within an industry are more likely to have connections with firms that are potential alliance partners than teams whose members have worked for only a few firms within the industry. Executives who have worked for many firms probably have more leads to alliance partners (either directly to former employers or indirectly to past suppliers and customers) and more relationships to draw upon in assessing the commitment and trustworthiness of partners. They are also likely to be better known throughout the industry and thus have leads back to the focal firm. Such industiy connections make it easier for top managers to gauge the risk of opportunism from potential partners and vice versa. Thus, to the extent that a team has more connections to potential partners through previous industry employers, the rate of alliance formation is likely to be higher.

A third attribute that contributes to social position is previous employment in high-leuel jobs by top managers. Top managers who have held high-level jobs at other firms are likely to know the people who are making alliancing deals. They will, thus, be more aware of alliance opportunities, better able to assess the risks associated with specific partners, and more likely to have already built up relationships with members of the prospective partners. As members of the executive elite, their high-level connections are a particularly important resource (D'Aveni 1990). Top managers who have held high-level jobs are also more likely to have negotiation skills that enable them to consummate alliances. Similarly, such managers have particularly high status (e.g., Davis 1991, Finkelstein 1994), an important advantage creating an image of skill, trustworthiness, and substantial reputation that enhances the firm's chances of being sought after as a partner or accepted as a suitor. Their high status signals to others that they and their firm are successful or very likely to become so. HYPOTHESIS 6. The higher the level of previous jobs held by members of the top management team, the greater the rate of alliance formation.

Research Methods

Sample The research sample was the population of semiconductor firms that were launched in the United States between 1978 and 1985. A new semiconductor firm is defined as a single-business organization that is formed to participate in the merchant semiconductor industiy. This definition excludes captive suppliers, electronics distributors, and custom design houses whose sole business is subcontracting their expertise to other firms. The population of semiconductor firms was developed from industiy lists maintained by the Semiconductor Industiy Association (SLA); Dataquest and Integrated Circuit Engineering (ICE), two of the principal market research firms specializing in the semiconductor industiy; Electronic News; and Semiconductor Equipment and Materials Institute (SEMI), which follows the semiconductor industry for member companies. The final population consisted of 102 firms. All semiconductor companies founded in the United States between 1978 and 1985, regardless of whether or not the firm subsequently failed, were included. The 98 HYPOTHESIS 5. The greater the number of previous firms that participated in the study were located industry employers of top management team members, the throughout the United States and were involved in a variety of product areas. greater the rate of alliance formation.

ORGANIZATION S C I E N C E / V O I . 7 , N o . 2 , M a r c h - A p r i l

1996

141

508

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

This article focuses on product development alliances, which are both common and important in the semiconductor industry. They are defined as interorganizational arrangements in which the partnering firms combine engineering and other personnel for the joint design of new products that at least one partner will sell. A single type of alliance was studied to control for any extraneous effects due to variations in alliance type, an empirical issue in some studies (e.g., Auster 1992). Data Sources We gathered field data on firms founded in the United States between 1978 and 1985 (inclusive), and on alliance formations (year and month) and the covariates from the birth of the firm until 1988 (or until the firm failed). The primary data source was a structured, confidential interview conducted with CEOs, founders, or other key executives. These individuals were chosen because of their extensive knowledge of the firm's history. Most of the interviews were conducted on site and in pairs, with one researcher asking questions while the other took notes and recorded responses. Most interviews were also tape-recorded and lasted between one and four hours. The interviews were followed by phone calls to clarify any incomplete data. The firm's financial history was obtained using a supplementary questionnaire completed by the firm's chief financial officer. These data were supplemented with information from 10-K and annual reports, backgrounders, press releases, business plans, and market-research reviews. Finally, we gathered data on market constructs from SIA, Dataquest, and ICE. Measures Number of Competitors. We measured the number of competitors via several steps. First, using worldwide market research data from Dataquest and ICE as well as the advice of industry executives, we identified six broad markets (application-specific circuits, memory, logic, gallium arsenide, linear, and discrete) and segments within these markets. This step gave us a highly accurate and refined view of the industry. Second, we assigned firms to market segments on the basis of their products. The third step was to measure the number of competitors in each market segment by using worldwide industry data from Dataquest, ICE, and, in a very few cases, confidential market research reports from firms in the study. The data were updated annually. Market Stage. As in our previous research (e.g., Eisenhardt and Schoonhoven 1990), we conceptualized

142

Strategic AtUamx Formation Tablai

Categorization of Mirtnt Stag·«

Market

Emergent

Growth

Semi-custom (ASIC) Digital Gate Arrays Analog Gate Arrays Standard Cells Programmable Logic Devices Silicon Compilers Memory DRAM SRAM ROM EPROM EEPROM FERRAM Discrete Optoelectronic Components Power Field Effect Transistors Other Linear Gallium Arsenide Digital Analog Logic Microprocessors Microperipherals Standard Logic

1978-82 1978-80 1978-88 1978-82 1978-82 1981-88

1983-88 1981-88

1978-85 1985-88

Mature

1983-88 1983-88 1978-88 1978-88 1978-88 1978-82 1978-84 1986-88

1983-88 1985-88

1978-88 1978-88 1978-88 1978-88 1978-88 1981-88 1981-88 1982-88 1978-88 1978-88 1978-88

market stage as a configurational variable, not a continuous one. Therefore, using dimensions consistent with our previous theoretical development of each market-stage configuration, we categorized each market segment as emergent, growth, or mature for each year in the study. We defined emergent-stage markets as (1) new (less than 7 years old) and (2) small (less than $100 million in sales, adjusted to 1988 dollars). Growth-stage markets (1) were large (more than $100 million in sales, adjusted to 1988 dollars), (2) had high growth (growth rate greater than 20% a year), and (3) followed emergent markets. Mature-stage markets (1) were large (more than $100 million in sales, adjusted to 1988 dollars, (2) had low growth (growing at less than 20% a year, and (3) followed growth markets. Specific market designations are listed in Table 1. Market stage was updated annually. Following Cameron et al. (1987), we selected these cutpoints carefully on the basis of conversations with industry executives, market research reports, and natural breaks in the data. We confirmed these market-stage designations by matching qualitative market descrip-

ORGANIZATION SCIENCE/VOI.

7, N o . 2, M a r c h - A p r i l 1996

Strategie Alliance Formation

509

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN Strategic Alliance Formation

tions (e.g., commercial viability) in market research reports with the theoretical descriptions of each market stage. For example, the linear market in 1986, which we had typed as mature, was termed "very mature" by Integrated Circuit Engineering (1987). We empirically verified the choice of market stage by examining the formation of alliances using market size, a continuous variable related to the stages. The categorical operationalization of market stage provided a better fit with the data and is more consistent with the underlying theoretical argument. Strategy. As described previously, innovation-based approaches appropriately capture strategy in technology-based firms such as semiconductor ventures (Boeker 1989, Maidique and Patch 1982). Therefore, our hypothesis and measures of strategy are related to innovation. We used two measures. One was an objective measure, the minimum design feature size (in microns). Feature size is a physical measurement of the miniaturization of the line widths in the circuit designs used by semiconductor firms. Smaller feature sizes mean more innovative technology because packing more circuits into a smaller area is the fundamental technical challenge within the industiy. This information was provided by the CEO, chief technical officer, or other key respondent during the structured interviews. The second measure of strategy consisted of three questions that were answered by the CEOs or other key respondents during the structured interviews. The first question captured innovation through extensions to existing knowledge: "To what extent could you rely on existing knowledge to build the first product?" The second question captured the innovation that occurs when technologies are combined: "To what extent did you synthesize existing knowledge to produce your first product?" The third question measured the overall degree of technical innovation: "How difficult was it to produce your first product?" Respondents used a 0 to 10 scale to respond to the questions. A three-item index was developed by (1) reverse coding the first question (2) computing the mean response to the three questions. The Cronbach alpha is 0.75. Strategy was measured at founding because of the persistence of founding strategy, particularly for very new firms such as we have here (Boeker 1989, Eisenhardt and Brown 1996, Kimberly 1980, Mintzberg and Waters 1982). Top Management Team. Most semiconductor ventures are managed by groups of individuals (Boeker 1989, Eisenhardt and Schoonhoven 1990). Top man-

ORGANIZATION S C I E N C E / V O I .

7, No. 2, March-April 1996

agers were designated as those people identified as members of top management in the structured interviews. Such individuals typically included the CEO and the people who reported directly to the CEO (usually the functional heads of finance, sales, engineering, and manufacturing), all of whom worked full time for the firm in executive positions. Respondents easily identified the top management teams in their firms as this designation is clear and such individuals are frequently identified as members of the top management group in corporate documents such as annual business plans and press releases, and in corporate functions such as CEO staff meetings. Team Size. Team size was measured as the number of individuals who were designated by the company respondent in the structured interview as members of top management. This measure was updated monthly. Number of Previous Semiconductor Employers. This measure was computed as the number of semiconductor industiy firms for which members of the top management team had previous worked. The data on previous employers were obtained from the structured interviews. This measure was updated monthly. Mean Previous Job Title. We measured the mean previous job title by first recording the highest previous job title of each top management team member. Job titles were coded by hierarchical level, ranging from 7, president or chief executive officer of a firm, to 0, nonmanagerial. These data are updated monthly. Founding Date. Several authors have noted that alliance formation increased during the time period of the study (e.g., Harrigan 1988, Parkhe 1993). Therefore, we controlled for founding date by converting this date (month, year) to the number of months after Januaiy 1978, the beginning of the study. Size. We controlled for firm size by using total assets, a meaningful measure of firm size for new ventures. Total assets were obtained from the financial questionnaire completed by the firm's financial officer. Cumulative Alliances. The number of previous product development alliances formed by the firm was used in the analyses to control for possible inertial behavior by firms. This control was updated monthly. Analysis and Dependent M e a s u r e

The analytic technique was event history, which takes into account both the occurrence and timing of events

143

510

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN Strategie Alliance Formation

(Allison 1984, Tuma and Hannan 1984). We used a Cox proportional hazard model (Cox 1972) because our principal theoretical interest was in testing hypotheses related to firms, not to time-dependence patterns. We estimated the model by using the TDA statistical package. We identified product development alliances by asking CEOs or key respondents in the structured interviews: "Does your firm have any formalized alliancing agreements with other companies?". If yes, the respondent was then asked to confirm that the alliance was a product development one (using the previously given definition), that work was actually done together by the two partners, when the alliance was formed, and the name of the partner. The formation date was defined as the date (month, year) on which both parties agreed to the alliance. Almost all of the alliances involved the signing of a formal agreement and an associated press release. Therefore, a specific date was available and clear. We also took several steps to ensure the accuracy of the information. First, recall was enhanced by the facts that the alliances were major, objective events for these young firms, were often accompanied by press releases, and often involved our respondents among the top management negotiators. Thus, the information was important, objective, and well-known to the respondents. Second, when respondents were unsure of information such as the exact month of formation, they consulted other sources to ascertain the correct information. Third, alliance information was gathered chronologically to facilitate informant recall. Finally, we were able to check the informants' responses for most of the firms against various secondary data including press releases and business plans. In a small number of the alliances, there was a discrepancy surrounding the exact date of formation which was cleared up by further consultation and consulting corporate archives. Approximately 60% of the firms formed product development alliances. All firms were included in the analysis until the end of observations in 1988 or, for the few firms that failed, until their deaths. Such right censoring is routinely and robustly handled within event history analysis (e.g., Tuma and Hannan 1984). There is no left censoring as all firms were tracked from birth. We identified 82 product development alliances. The alliance partnérs were approximately 50% customers (including firms in the computer, automotive, telecommunications, defense, aircraft, and other industries), 34% major semiconductor merchant producers (e.g., Intel, National, Phillips, Toshiba), 13% other firms in

144

the sample, and 3% miscellaneous or unknown. In comparison with biotechnology (e.g., Powell and Brantley 1992), the range of industries represented by the partners is much wider, perhaps reflecting the greater range of applications of semiconductor products during the time period of this study.

Results Table 2 reports the results of the event history analyses. Column 1 gives the results for the baseline model that controls for the firm's founding date, size, and cumulative number of product development alliances. The results show that firms founded later in the study and firms with more past alliances have higher rates of alliance formation. Firm size is not significant. Column 2 provides the results for the hypothesized constructs. Overall, Model 2 is a statistically significant improvement over the baseline model. Consistent with HI, firms in markets with many competitors have higher rates of alliance formation. Consistent with H3, firms with more technically innovative strategies have higher rates of alliance formation as measured by both minimum feature size and the three-item innovation scale. Significant support is found for the top management hypotheses also. Firms with larger founding top management teams (H4) whose members had previously worked for many semiconductor firms (H5) in high management positions (H6) have significantly higher rates of alliance formation. Finally, H2 posits that firms in emergent-stage markets have higher alliance formation rates than firms in growth-stage markets which in turn have higher rates of alliance formation than firms in mature-stage markets. Because market stage is a categorical variable, there are several ways to test this hypothesis. In Model 2, we set the market-stage variable equal to 3 for emergent markets, 2 for growth, and 1 for mature to capture the hypothesized ranking. Using this approach, market stage is not significantly associated with higher rates of alliance formation although the sign is in the hypothesized direction. In addition, the founding date control is still significant, but cumulative alliances and firm size are not. In Model 3, we examined market stage further by recomputing the market-stage measure (named alternate market stage) so that firms in mature-stage markets were hypothesized to have higher rates of alliance formation than firms in growth-stage markets (i.e., emergent = 3, mature = 2, growth = 1). Our thinking was that product development alliances may involve such substantial time demands to develop collaborative working relationships that they are inappropriate for growth-stage markets where there is a premium on

ORGANIZATION S C I E N C E / V O I .

7, No. 2, March-April 1996

Strategie Alliance Formation KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN Table 2

511

Stmtcgic Alliance Formation

Cox Modal Predicting Rata* of AUIanca Formation

Variables Founding Date Firm Size Cumulative Alliances

1

2

0.02*** (0.01) 0.00 (0.00) 0.31** (0.15)

0.02*** (0.01) 0.00 (0.00) 0.15 (0.16) 0.01* (0.01)

Number of Competitors Market-stage Market-stage

3 0.02* * * (0.01) 0.00 (0.00) 0.12 (0.16) 0.01** (0.01)

Models 4

5

6

0.02*** (0.01) 0.00 (0.00) 0.14 (0.16) 0.01* (0.01)

0.02*** (0.01) 0.00 (0.00) 0.12 (0.16) 0.01 (0.01)

0.02* * * (0.01) 0.00 (0.00) 0.14 (0.16) 0.00 (0.01)

0.02*** (0.01)

0.01· (0.01)

0.16 (0.22)

Alternate Market-stage

0.37** (0.19)

Emergent-stage

0.37** (0.19) 0.60** (0.36)

Growth-stage

-0.51** (0.38)

Mature-stage

0.13 (0.32)

Technical Innovation Feature Size (Microns)

-0.18* (0.12) 0.11** (0.19)

Innovation Scale Top Management Team Size # Previous Semiconductor Employers Mean Previous Highest Job Level Log L

7

-460.1***

-0.24** (0.12) 0.11** (0.06)

-0.20** (0.12) 0.10** (0.06)

0.08* (0.05) 0.37*** (0.14) 0.15** (0.08)

0.09** (0.05) 0.38*** (0.14) 0.15** (0.08)

0.08** (0.05) 0.41*** (0.14) 0.16** (0.12)

-449.9***

-448.8***

-448.8***

-0.26** (0.13) 0.12** (0.06) 0.08** (0.05) 0.41*** (0.14) 0.16** (0.08) -448.7***

-0.22" (0.12) 0.12** (0.06)

-0.25** (0.12) 0.12** (0.06)

0.08* (0.05) 0.36*** (0.13) 0.15·* (0.08)

0.09** (0.05) 0.48*·· (0.14) 0.17·· (0.08)

-450.01***

-448.6···

One-tail test significance levels. < 0.05* < 0.01** < 0.001*** Standard errors in parentheses. 4113 spells.

speed. As indicated in column 3, the alternate marketstage variable is significantly associated with rates of alliance formation, indicating partial support for H2. In Models 4 through 6, we used different analyses to ensure analytic robustness for market stage by entering a market-stage dummy variable for each type of market in successive models. We used emergent (emergent » 1, others = 0), growth (growth - 1, others = 0), and mature (mature '= 1, others = 0) in Models 4, 5, and 6, respectively. In Model 4, emergent-stage market is

ORGANIZATION SCIENCE/VOI.

7, No. 2, March-April 1996

associated significantly and positively with alliance formation as hypothesized. In Model 5, growth-stage market is associated significantly and negatively with alliance formation. Mature-stage market is not significant in Model 6. Hence, consistent with the analysis in Model 3 using alternative market stage, emergent-stage markets are associated with the highest rates of alliance formation and growth-stage markets are associated with the lowest rates of alliance formation, indicating partial support for H2.

145

512

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN Strategie Alliance Formation

Model 7 is a parsimonious model which does not include the nonsignificant control variables. The results for the hypothesized variables are similar to those of the previous models. That fs, number of competitors, emergent-stage market, innovative strategy, and the three top management constructs are all associated significantly with higher rates of alliance formation.

Discussion This article is the first, we think, to combine strategic and social explanations of strategic alliance formation. We did so by extending the resource-based perspective to the issue of alliance formation through strategic and social resource arguments. The resulting hypotheses were examined using the product development alliances of entrepreneurial semiconductor firms. There are several results. Principal Results First, consistent with the strategic position hypotheses, difficult market conditions and risky firm strategies increased the rate of alliance formation. Specifically, semiconductor ventures formed product development alliances at higher rates when they faced emergentstage markets, markets with many competitors, or when they were attempting an innovative strategy. We argued that such situations would put firms in vulnerable strategic positions for which the additional resources (e.g., technical know-höw, cash, legitimacy) available through alliances would be beneficial. Overall, the results emphasize that vulnerable strategic position (i.e., emergent markets, innovative technologies, high competition) leads to the formation of strategic alliances. Second, consistent with the social position hypotheses, top management characteristics affected the rate of alliance formation. Firms with top management teams that were large, experienced, and well-connected through former employers and high-level previous jobs formed product development alliances at higher rates. We argued that such top management teams would have the resources (e.g., time, skills, connections, status) necessary to form alliances at higher rates. Such teams would be more likely to gain alliances if they needed them and to be offered attractive alliancing arrangements even if they did not. Overall, the results emphasize the importance of strong social position (i.e., large, well-connected, high-level top management teams) to the formation of strategic alliances. Moreover, while past research focusing on social determinants of alliance formation has centered on firm posi-

146

tion within a network of other firms (Gulati 1993, Kogut et al. 1992), the contribution here is to recognize the social advantages of senior executives in facilitating alliance formation, particularly for entrepreneurial firms. Taken together, these two sets of results affirm the view that strategic action takes place within a social context (Granovetter 1992). The rational calculus of strategic moves clearly captured some of the rationale behind alliance formation in this study. But alliance formation was also affected by a social calculus related to the skills, status, reputation, and past relationships of top managers. While this interplay between the strategic and social appears in studies of other phenomena (Haunschild 1994, O'Reilly et al. 1994, Podolny 1994), it becomes apparent here in alliance formation. Moreover, the interplay is particularly striking as settings become more uncertain (as we have here in emergent markets and pioneering technologies of semiconductor ventures) as executives come to rely more on social processes in determining critical firm behaviors and outcomes. Third, we discovered founding date to be an important predictor of alliance formation rate. That is, firms founded later in the 1978 to 1985 founding period formed product development alliances at higher rates. It is difficult to know whether economic factors such as the increasingly global and competitive business climate and rapid technical change created the higher rates, or whether the simply faddish popularity of the alliancing phenomenon or some other factors were at work. A combination seems most likely. Regardless of reason, the effect was highly significant.

Some Surprises Not all of our expectations, however, are confirmed by the data. We hypothesized that rates of alliance formation would be higher in growth-stage markets than mature ones because of the need for flexible resources in fast-paced market segments. In fact, we found the reverse: growth-stage markets have the lowest rates of alliance formation. Why? One reason may be that growth-stage markets are simply so munificient that firms have little need for the additional resources that alliances might bring. An alternative reason, however, was suggested by informal conversations with key executives. We learned that product development alliances involve close interaction and collaborative work among firm personnel, particularly design engineers. Rather than simply handing off design specifications as in a manufacturing relationship

ORGANIZATION SCIENCE/VOI.

7, No. 2, March-April 1996

Strategie Alliance Formation

513

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN Strategie Alliance Formation

or technological concepts as in licensing agreements, tbe partners in product development alliances must intensively interact with one another to gain the benefits of cooperation (Larson 1992, Heikal 1992). Therefore, such alliances may be slow to reach their potential and so less attractive in growth-stage markets where speed is a necessity (Eisenhardt 1989). If this explanation is valid, it suggests a need for sensitivity to alliance type in future research. The results also highlight the firms with low rates of alliance formation. Topically, these firms had few resources (i.e., mundane technologies and small, less well-connected, less experienced top management teams.) One reason for their low rates of alliance formation may be that their managers were simply not interested in product development alliances. Alternatively, these firms may have lacked the resources necessary for alliance formation. New firms in technology-based industries, such as we studied, have few resources other than novel technology and experienced management to lure partners. Consequently, although these particularly resource-poor firms may have needed alliances as much or perhaps more than other firms, they offered few attractions to potential partners. For these alliance "wallflowers", this explanation highlights an irony of strategic alliances; that is, cooperation requires resources to get resources. For the young firms here, alliance formation may well have simply allowed the "rich" firms to get "richer".

Beyond Alliance Formation Finally, our results extend previous findings. The perspective we develop contrasts with transaction cost explanations by emphasizing (1) strategic and social factors, not transaction costs, (2) characteristics of the firm (e.g., strategy, top management), not the transaction, and (3) a theoretical logic of needs and opportunities, not efficiency. Because we did not test explicitly for transaction cost explanations, we cannot claim that our approach is superior. However, our hypotheses surrounding firm-level strategic and social factors are supported. We also found, contrary to transaction cost arguments (e.g., Williamson 1991), that in highly uncertain situations (i.e., emergent markets, innovative technologies) firms seek, not avoid, alliances. The strategic benefits of alliance formation appear to outweigh transaction cost inefficiencies. Consistent with Ghosal and Moran (1995), this suggests that strategic and social factors may dominate transaction costs, especially in high-velocity industries and innovative processes.

ORGANIZATION S C I E N C E / V O I .

7,

No.

2,

March-April

19%

Our work also relates to the current debate on industry versus firm-level effects (e.g., Conner 1994, Peteraf 1993). On one side, critical firm actions and outcomes are seen as shaped by market factors such as competition, barriers to entry, and market stage. This view broadly characterizes the industrial organization and organizational ecology literatures. On the other side, the organizational view emphasizes firm-level differences or heterogeneity. In this view that broadly characterizes perspectives such as upper echelons and resource-based views, unique firm-level capabilities are seen as powerful determinants of firm actions and outcomes. This study suggests the relevance of both firm (e.g., top management team characteristics and firm strategy) and industiy factors (e.g., competition and market stage) in determining alliance formation. As in the work on managerial discretion (Finkelstein and Hambrick 1990, Hambrick and Finkelstein 1987) and our previous work on growth (Eisenhardt and Schoonhoven 1990) and innovation (Schoonhoven et al. 1990) in semiconductor ventures, the relevance of both sets of factors to key firm-level outcomes is supported. This suggests that the debate move from firm versus industry to when and how one dominates the other. In particular, both firm and industry factors appear powerfully.to shape the outcomes of entrepreneurial firms. Finally, our work relates to the upper-echelons perspective on top management teams. Traditionally (e.g., Hambrick and Mason 1984, Keck and Tushman 1993, Michel and Hambrick 1992, Smith et al. 1994, Wiersma and Bantel 1992), this research stream has focused on the internal, cognitive linkages of the demographic characteristics of top management teams to firm outcomes. We extend that reasoning to the external, social linkages of these characteristics to outcomes. The findings indicate that the demographic characteristics of top management teams may well link not only to internal, cognitive factors such as consensus, risk-taking, communication, social integration, and willingness to change (e.g., Smith et al. 1994, Wiersma and Bantel 1992), but also to external, social ones such as status, reputation, and contacts (c.f., D'Aveni 1990). These findings broaden the scope of the upper-echelons perspective within strategic management research, bolstering the relevance of top management to central firm outcomes.

Conclusions At one level, we conclude that cooperation through product development alliances by semiconductor ven-

147

514

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

Strategic Alliance Formation

tures is an increasingly popular strategy that large, experienced, and well-connected top management teams apparently use to cope with new or competitive markets and pioneering technologies. In brief, firms cooperated when they needed to, when they were able to, and perhaps when it was popular. At a more fundamental level, we conclude that strategic alliance formation is a complex phenomenon involving both strategic and social factors operating within a logic of needs and opportunities for cooperation. Firms in vulnerable strategic positions or strong social positions were more likely to cooperate with other firms through strategic alliances. The central conclusion is that failure to include either strategic or social explanations creates an impoverished picture of alliancing activity.

Carroll, G. and M. Hannan (1989), "Density Delay in the Evolution of Organizational Populations: A Model and Five Empirical Tests," Administrative Science Quarterly, 34, 411-430.

Acknowledgements

Dunne, T., M. Roberts and L. Samuelson (1988), "Patterns of Firm Entry and Exit in U.S. Manufacturing Industries," The RAND Journal of Economics, 19, 495-515.

Data collection was supported by the U.S. Department of Commerce, Economic Development Administration, Grant No. RED870-G-86-15 and data analysis by the Alfred P. Sloan Foundation, and the National Science Foundation, Decision Risk and Management Science Program, Award No. 8911370. We extend special thanks to Kathy Lyman and Lee Fleming for their expert analytic assistance. We also appreciate the enthusiastic guidance of seminar participants at UCLA, Washington University, ΜΓΓ, Stanford University, and University of Alberta, and the advice of Bill Barnett, Shona Brown, Jeff Dyer, John Freeman, Connie Gersick, Rebecca Henderson, Arie Lewin, Anne Miner, Mette Monsted, Charles O'Reilly, Laura Poppo, Dick Scott, Ken Smith, and the anonymous reviewers. A previous version of this paper won first place, 1994 Stern Competition for the Best Paper on Entrepreneurship and Innovation.

References Allison, P. (1984), Event History Analysis: Regression for Longitudinal Event Data, Beverly Hills, CA: Sage Publications. Anderson, C. and C. Zeithaml (1984), "Stage of the Product Life Cycle, Business Strategy, and Business Performance," Academy of Management Journal, 27, 5-24. Arthur, B. (1990), "Positive Feedbacks in the Economy," Scientific American, February, 92-99. Auster, E. (1992), "The Relationship of Industry Evolution to Patterns of Technological Linkages, Joint Ventures, and Direct Investment between U.S. and Japan," Management Science, 38, 6, 778-792. Axelrod, R. (1984), The Evolution of Cooperation, New York: Basic Books. Baum, J. and C. Oliver (1991), "Institutional Linkages and Organizational Mortality," Administrative Science Quarterly, 36, 187-218. Boeker, W. (1989), "Strategic Change: The Effects of Founding and History," Academy of Management Journal, 32, 489-515. Cameron, K., M. Kim, and D. Whetten (1987), "Organizational Effects of Decline and Turbulence," Administrative Science Quarterly, 32, 222-240.

148

Conner, K. (1994), "The Resource-based Challenge to the Industrystructure Perspective," in D. Moore (Ed.), Academy of Management Best Papers Proceedings 1994, Charleston, S.C.: Academy of Management. Cox, D. (1972), "Regression Models and Life-tables," Journal of the Royal Statistical Society, 34, 187-220. D'Aveni, R. (1990), " T o p Managerial Prestige and Organizational Bankruptcy," Organization Science, 1, 187-220. David, P. (1985), "Clio and the Economics of QWERTY," American Economic Review, 75, 332-337. Davis, G. (1991), "Agents without Principle? The Spread of the Poison Pill through the Intercorporate Network," Administrative Science Quarterly, 36, 583-613. Dosi, G. (1988), "Sources, Procedures, and Microeconomic Effects of Innovation," Journal of Economic Literature, 26, 1120-1171.

Eisenhardt, Κ. (1989), "Making Fast Strategic Decisions in High Velocity Environments," Academy of Management Journal, 32, 543-576. and S. Brown (1996), "Organizational Evolution," in J. Baum and J. Dutton (Eds.), Advances in Strategic Management, Greenwich, CT: JAI Press, forthcoming. and C. Schoonhoven (1990), "Organizational Growth: Linking Founding Team, Strategy, Environment and Growth among U.S. Semiconductor Ventures, 1978-1988," Administrative Science Quarterly, 35, 504-529. Finkelstein, S. (1994), Personal communication. and D. Hambrick (1990), "Top-management Team Tenure and Organizational Outcomes: The Moderating Role of Managerial Discretion," Administrative Science Quarterly, 35, 484-503. Galaskiewicz, J. and S. Wassennan (1989), "Mimetic and Normative Processes within an Interorganizational Field: An Empirical Test," Administrative Science Quarterly, 34, 454-479. Ghosal, S. and P. Moran, (1995), "Bad for Practice: A Critique of the Transaction Cost Theory," in D. Moore (Ed.), Academy of Management Best Papers Proceedings 1995, Charleston, S.C.: Academy of Management, pp. 12-16. Granovetter, M. (1992), "Problems of Explanation in Economic Sociology," in N. Nohria and R. Eccles (Eds.), Networks and Organization: Structure, Form, and Action, Boston, MA: Harvard Business School Press, pp. 25-56. Gulati, R. (1993), The Dynamics of Alliance Formation, Unpublished doctoral dissertation, Cambridge, MA: Harvard University. Hagedoorn, J. (1993), "Understanding the Rationale of Strategic Technology Partnering: Interorganizational Modes of Cooperation and Sectoral Differences," Strategic Management Journal, 14, 371-385. Hambrick, D. and S. Finkelstein (1987), "Managerial Discretion: A Bridge between Polar Views of Organizations," in L. Cummings and B. Staw (Eds.), Research on Organizational Behavior, 16, 171-213.

ORGANIZATION SCIENCE/VOI. 7, N o . 2, M a r c h - A p r i l 1996

Strategic Alliance Formation KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

515

Strategic Alliance Formation

, I. MacMiUan and D. Day (1982), "Strategic Attributes and Performance in the BCG Matrix—A PIMS-based Analysis of Industrial Product Businesses," Academy of Management Journal, 25, 510-531.

Maidique, M. and P. Patch (1982), "Corporate Strategy and Technological Policy," in M. Tushman and W. Moore (Eds.), Readings in the Management of Innovation, Marshfield, MA: Pitman, pp.

and P. Mason (1984), "Upper Echelons: The Organization as a Reflection of Its Top Managers," Academy of Management Review, 9, 193-206.

Michel, J. and D. Hambrick (1992), "Diversification Posture and Top Management Team Characteristics," Academy of Management Journal, 35, 9-37.

Hamel, G., Y. Doz, and C. Prahalad (1989), "Collaborate with Your Competitors and Win," Harvard Business Review, JanuaryFebruary, 133-139.

Miles, R. and C. Snow (1978), Organizational Strategy, Structure, and Process, New York: McGraw-Hill.

273-285.

Harrigan, Κ. (1988), "Joint Ventures and Competitive Strategy," Strategic Management Journal, 9, 141-158.

Miner, Α., T. Amburgey and T. Stearns (1990), "Interorganizational linkages and Population Dynamics: Buffering and Transformational Shields," Administrative Science Quarterly, 35, 689-713.

Haunschild, P. (1994), "How Much is That Company Worth? In· terorganizational Relationships, Uncertainty, and Acquisition Premiums," Administrative Science Quarterly, 39, 391-411.

Mintzberg, H. and J. Waters (1982), "Tracking Strategy in an Entrepreneurial Firm," Academy of Management Journal, 25, 465-499.

Heikal, A. (1992), "Evolutionary Processes in Strategic Alliances," Unpublished doctoral dissertation, Stanford, CA: Stanford University.

Mitchell, W. and K. Singh (1992), "Incumbents' Use of Pre-entry Alliances before Expansion into New Technical Subfields of an Industry," Journal of Economic Behavior and Organizations, 18, 347-372.

Heimer, C. (1992), "Doing Your Job and Helping Your Friends: Universalistic Norms about Obligations to Particular Others in Networks," in N. Nohria and R. Eccles (Eds.), Networks and Organizations'. Structure, Form, and Action, Boston, MA: Harvard Business School Press, pp. 118-142. Hennart, J. (1988), "A Transaction Costs Theory of Equity Joint Ventures," Strategic Management Journal, 9, 361-374. (1991), "The Transaction Costs Theory of Joint Ventures: An Empirical Study of Japanese Subsidiaries in the United States," Management Science, 37, 483-497. Integrated Circuit Engineering (1987), Status 1986: A Report on the Integrated Circuit Industry, Scottsdale, AZ: Integrated Circuit Engineering. Jelinek, M. and C. B. Schoonhoven (1990), The Innovation Oxford, U.K.: Basil Blackwell, Ltd.

Marathon,

Keck, S. and M. Tushman (1993), "Environmental and Organizational Context and Executive Team Structure," Academy of Management Journal, 36, 1314-1344. Kimberly, J. (1980), "Initiation, Innovation, and Institutionalization in the Creation Process," in J. Kimberly and R. Miles (Eds.), Organizational Life Cycle, San Francisco, CA: Jossey-Bass, pp. 134-160. Klein, B. (1977), Dynamic Economics, Cambridge, MA: Harvard University Press. Klepper, S. and E. Graddy (1990), "The Evolution of New Industries and the Determinants of Market Structure," The RAND Journal of Economics, 21, 27-44. Kogut, B. (1991), "Joint Ventures and the Option to Expand and Acquire," Management Science, 37, 19-33.

Nohria, N. (1992), "Introduction: Is a Network Perspective a Useful Way . of Studying Organizations?" in N. Nohria and R. Eccles (Eds.), Networks and Organizations'. Structure, Form, and Action, Boston, MA: Harvard Business School Press, pp. 1-22. Ohmae, K. (1989), "The Global Logic of Strategic Alliances," Harvard Business Review, March-April, 143-154. O'Reilly, C., G. Main and G. Crystal (1994), " C E O Compensation as a Tournament and Social Comparison," Administrative Science Quarterly, 33, 257-274. Osborn, R. and C. Baughn (1990), "Forms of Interorganizational Governance for Multinational Alliances," Academy of Manage• ment Journal, 33. 503-519. Ouchi, W. (1980), "Markets, Bureaucracies, and C a n s , " Administrafive Science Quarterly, 25, 124-141. Parkhe, A. (1993), "Strategic Alliance Structuring: A Game Theoretic and Transaction Cost Examination of Interfirm Cooperation," Academy of Management Journal, 36, 794-829. Penrose, E. (1959), The Theory of the Growth of the Firm, New York: Wiley. Peteraf, M. (1993), "The Cornerstones of Competitive Advantage: A Resource-based View," Strategic Management Journal, 14, 171-191. Pisano, G. and D. Teece (1989), "Collaborative Arrangements and Global Technology Strategy: Some Evidence from the Telecommunications Equipment Industry," in R. Rosenblum (Ed.), Research on Technological Innovation, Management and Policy, 4, Greenwich, CT: JAI Press, pp. 227-256. Podolny, J. (1994), "Market Uncertainty and the Social Character of Economic Exchange," Administrative Science Quarterly, 39, 458-483.

W. Shan and G. Walker (1992), "The Make-or-cooperate Decision in the Context of an Industry Network," in N. Nohria and R. Eccles (Eds.), Networks and Organizations: Structure, Form, and Action, Boston, MA: Harvard Business School Press, pp. 348-365.

Porter, M. (1980), Competitive Strategy : Techniques for Analyzing Industries and Competitors, New York: The Free Press.

Larson, A. (1992), "Network Dyads in Entrepreneurial Settings: A Study of the Governance of Exchange Relationships," Administrative Science Quarterly, 37, 76-104.

Powell, W. (1990), "Neither Market Nor Hierarchy "in L. Cummings and B. Staw (Eds.), Research on Organizational Behavior, Greenwich, CT: JAI Press.

ORGANIZATION

SCIENCE/VOI. 7, N o . 2, M a r c h - A p r i l

1996

149

516

Κ. M. Eisenhardt/C. Β. Schoonhoven

KATHLEEN M. EISENHARDT AND CLAUDIA BIRD SCHOONHOVEN

and P. Brantley (1992), "Competitive Cooperation in Biotechnology: Learning through Networks?" in N. Nohria and R. Eccles (Eds.), Networks and Organizations Structure, Form, and Action, Boston, MA: Harvard Business School Press, pp. 365-394. Rumelt, R. (1974), Strategy, Structure, and Economic Performance, Cambridge, MA: Harvard University Press. Sahlman, W. and H. Stevenson (1985), "Capital Market Myopia," Journal of Business Venturing, 1, 7-30. Schoonhoven, C , K. Eisenhardt and K. Lyman (1990), "Speeding Products to Market: Waiting Time to First Product Introductions in New Firms," Administrative Science Quarterly, 35, 177-207. Shan, W. (1990), "An Empirical Analysis of Organizational Strategies by Entrepreneurial High-technology Finns," Strategic Management Journal, 11, 129-139. Smith, K. G., K. A Smith, J. Otian, H. Sims, D. O'Bannon and J. Scully (1994), " T o p Management Team Demography and Process: The Role of Social Integration and Communication," Administrative Science Quarterly, 39, 412-438. Teece, D. (1987), "Profiting from Technological Innovation: Implications for Integration, Collaborating, Licensing, and Public Pol-

Strategic Alliance Formation

icy," in D. Teece (Ed.), The Competitive Challenge, New York: Harper Collins. Tuina, Ν. and M. Hannan (1984), Social Dynamics: Models and Methods, Orlando, FL: Academic Press. Tushman, M. and P. Anderson (1986), "Technological Discontinuities and Organizational Environments," Administrative Science Quarterly, 31, 439-465. Van de Ven, A. and D. Polley (1992), Learning while Innovating, Organization Science, 3, 92-116. Weiwet, W. and A. Hunter (1985), "The Interorganizational Network as a Resource: A Comparative Case Study on Organizational Genesis," Administrative Science Quarterly, 30, 517-547. Wernerfelt, B. (1984), " A Resource-based View of the Firm," Strategic Management Journal, 5, 171-180. Wiersma, M. and K. Bantel (1992), "Top Management Team Demography and Corporate Strategic Change," Academy of Management Journal, 35, 91-121. Williamson, O. (1991), "Comparative Economic Organization: The Analysis of Discrete Structural Alternatives," Administrative Science Quarterly, 36, 269-296.

Accepted by Arie Y. Lewin·, received December 1994. This paper has been with the authors for three revisions.

150

ORGANIZATION SCIENCE/VOI.

7, No. 2, March-April 1996

Teil VII Strategieimplementierung

12 Strukturen, Führungspersonal, Kulturen und Mikropolitik

Nachdem im Rahmen des Strategieformulierungsprozesses unter Berücksichtigung der externen Chancen und Risiken sowie der intern vorhandenen Kompetenzen ein Strategiebündel ausgewählt wurde, muß die Unternehmungsleitung organisatorische Maßnahmen zur Implementierung der intendierten Strategien ergreifen. Unter Strategieimplementierung kann man die Summe aller Aktivitäten und organisatorischen Arrangements verstehen, die in der Unternehmung durchgeführt bzw. bereitgestellt werden, damit strategische Pläne möglichst effizient und effektiv realisiert werden können (Miles et al., 1978; Hill & Jones, 1992: 312). Die wichtigsten Implementierungsaktivitäten beziehen sich auf die sachorientierte Umsetzung: Es werden operative Programme und Budgets spezifiziert, die die bereichs- und abteilungsbezogene Umsetzung der Strategien anleiten (Kolks, 1990: 83f.; Welge, 1992: 388; Wheelen & Hunger, 1995: 224f.). Neben umsetzungsorientierten Aktivitäten dienen auch Maßnahmen der Organisationsgestaltung der Implementierung einer Strategie. Durch die bewußte Gestaltung von zentralen Organisationsmerkmalen, wie z.B. der Organisationsstruktur oder der Organisationskultur, versucht das Management, ein kontinuierliches, strategiegeleitetes Handeln der Unternehmungsmitglieder sicherzustellen (Galbraith & Kazanjian, 1986). Die Anpassung wichtiger Organisationsmerkmale an die Implementierungserfordernisse einer Strategie bewirkt somit eine Erleichterung von Prozeßabläufen, wie z.B. der Kommunikation, der Problemidentifizierung oder der Konfliktlösung. Das Thema Strategieimplementierung führt uns zunächst zurück zu der im zweiten Kapitel skizzierten Kontroverse zwischen Planern und Inkrementalisten. Mintzberg (1990b) bevorzugt den Ausdruck „Strategieformation" anstelle der Unterscheidung zwischen Strategieformulierung und -Implementierung, da realisierte Strategien immer auch emergente, ungeplante Komponenten beinhalten. Aufgrund der in Kapitel 2 dargestellten Konvergenz zwischen diesen Ansätzen wird hier davon ausgegangen, daß sich die Strategieimplementierung sowohl aus sachlogischen als auch aus didaktischen Gründen von der Strategieformulierung unterscheiden läßt. Selbst wenn sich eine realisierte Strategie in starkem Maße inkremental herausgebildet hat, sind flankierende Implementierungsmaßnahmen im Sinne organisatorischer Anpassungen oft vonnöten, um aus der realisierten Strategie längerfristig Vorteile ziehen zu können. Durch die Systematisierung der emergenten, sich zufällig entwickelnden Bestandteile einer Strategie und das Treffen organisatorischer Vorkehrungen können zukünftige strategiefördernde Prozeßabläufe erleichtert werden. Darüber hinaus kann es sinnvoll sein, durch Maßnahmen der Strategieimplementierung organisatorische Arrangements aufzubauen, die die inkrementalen Aspekte des Strategieentwicklungsprozesses besonders fördern.86 Die Strategieimplementierung ist nicht nur ein logischer Bestandteil des strategischen Planungsprozesses, sie ist auch deshalb bedeutsam, weil sie in der Praxis erhebliche Pro86

Näheres hierzu wird im Abschnitt „Strategie und Managementqualifikationen" dargelegt.

520

R. Κ. F. Bresser

bleme aufwirft. 87 In einer von Alexander (1991) durchgeführten Studie werden zehn Probleme identifiziert, denen sich Unternehmungen bei der Implementierung einer strategischen Reorientierung häufig gegenübersehen: 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Man benötigt mehr Zeit für die Implementierung als ursprünglich veranschlagt war. Es treten schwerwiegende, nicht vorhergesehene Umsetzungsprobleme auf. Aktivitäten werden nicht ausreichend koordiniert. Krisen führen zur Vernachlässigung der Implementierungsaufgaben. Unzureichende Fähigkeiten des Personals. Unzureichendes Training und mangelnde Unterweisung des ausführenden Personals auf den unteren Hierarchieebenen. Unkontrollierbare Faktoren in der externen Umwelt. Unzureichende Führung durch die Abteilungsleiter. Unklare Festlegung der Hauptimplementierungsaufgaben. Unzureichende Überwachung der Implementierungsaufgaben.

Die diesen Ergebnissen zugrundeliegende Datenbasis beruht auf einer Zufallsstichprobe von 93 Fortune-500-Unternehmungen. Wenn man sich vor Augen hält, daß alle diese zehn Probleme bei mehr als der Hälfte der Unternehmungen aufgetreten sind, wird die Wichtigkeit der Implementierung offenbar. Die schlechte Implementierung einer angemessenen Strategie führt in aller Regel zum Scheitern der Strategie. Demgegenüber fördert eine gute Implementierung nicht nur den Erfolg einer angemessenen Strategie; sie kann darüber hinaus auch rechtzeitige Modifikationen unangemessener Strategien bewirken und somit einer zunächst fehlerhaften Strategie zum Erfolg verhelfen (Wheelen & Hunger, 1995: 221).

Kontingenztheorie und ressourcenbasierter Ansatz Der Teil der Strategieformulierung, der sich mit der sachbezogenen Umsetzung von Strategien durch Programme, Prozeduren, Budgets usw. beschäftigt, ist theoretisch wenig spannend. Für die im Rahmen einer Strategieimplementierung vorzunehmenden Modifikationen zentraler Organisationsmerkmale existieren demgegenüber interessante theoretische Begründungen. Maßnahmen der Organisationsgestaltung lassen sich kontingenztheoretisch und durch den ressourcenbasierten Ansatz (RBA) begründen. Die dominierende theoretische Grundlage der Strategieimplementierung ist die Kontingenztheorie (Galbraith & Kazanjian, 1986: 28-45). Die Managementforschung hat den Kontingenzgedanken ursprünglich nur auf das Verhältnis zwischen Organisationsstruktur und -umweit angewendet (Burns & Stalker, 1961; Pugh et al., 1963): Die angemessene Organisationsstruktur wurde als abhängig (kontingent) von bestimmten internen und externen Umweltgegebenheiten modelliert. Child (1972) führte die Unternehmungsstrategie in die kontingenztheoretische Diskussion ein, indem er argumentierte, daß die Wahl der angemessenen Organisationsstruktur nicht nur von Umweltfaktoren abhänge, sondern auch von der verfolgten Strategie. Das SWOT-Konzept ist heute das umfassendste Bei87

Berichten aus der Beratungspraxis zufolge scheitern ca. 70 bis 80 Prozent aller Restrukturierungsanstrengungen (Fischer et al., 1994; Scott-Morgan, 1994).

Strukturen, Personal, Kulturen, Mikropolitik

521

spiel für einen kontingenztheoretischen Analyserahmen i m Strategischen Management. Was die Strategieimplementierung betrifft, so konzentriert sich die A n a l y s e auf interne Kontingenzfaktoren: Interne Gegebenheiten (z.B. die Organisationsstruktur oder die Managementqualifikationen) m ü s s e n mit der Unternehmungsstrategie einen „Strategie Fit" bilden, damit überdurchschnittliche Rentabilitäten m ö g l i c h werden. Eine z w e i t e wichtige theoretische Grundlage der Strategieimplementierung liefert der R B A . D i e s e m z u f o l g e gelten die internen Kontingenzfaktoren als Ressourcen und Fähigkeiten der Unternehmung (Barney, 1991). D a sich der R B A gegenwärtig vornehmlich mit der Generierung und d e m Wandel kritischer Fähigkeiten (core competencies) beschäftigt (siehe Kapitel 7), erhält die strukturelle Orientierung der klassischen Kontingenztheorie durch den R B A potentiell eine prozeßorientierte Erweiterung. In der Literatur wird die Herstellung eines Einklangs bzw. Fits z w i s c h e n der Unternehmungsstrategie und insbesondere vier Organisationsmerkmalen diskutiert: der Organisationsstruktur, den Managementqualifikationen, der Organisationskultur und d e m mikropolitischen U m f e l d ( W h e e l e n & Hunger, 1995: 2 2 0 - 2 7 9 ; Hill & Jones, 1992: 3 0 9 - 4 2 3 ) . N e b e n der Herstellung eines Fits z w i s c h e n den einzelnen Merkmalen und der Unternehmungsstrategie gilt ferner, daß die Kontingenzfaktoren auch untereinander konsistent und widerspruchsfrei aufeinander abgestimmt sein sollen (Galbraith & Kazanjian, 1986: 108123). D i e relevanten Argumente dieser Diskussion werden i m folgenden sowohl aus der Sicht der klassischen Kontingenztheorie als auch aus der des R B A s dargestellt. 88

88

Das Strategic-Fit-Konzept hat unterschiedliche theoretische Bedeutungen und kann dementsprechend unterschiedlich definiert werden (Venkatraman & Camillus, 1984). Venkatraman (1989) unterscheidet z.B. sechs Perspektiven (vgl. zum folgenden auch Knyphausen, 1995: 209f.): (1) Fit as Moderation, d.h. der Einfluß der Strategie auf die Unternehmungsperformance ist von den Werten moderierender Kontextvariablen abhängig. Anders ausgedrückt, die Beziehung zwischen der Strategie und der Moderatorvariablen ist die primäre Determinante der Performance. (2) Fit as Mediation bedeutet, daß die Beziehung zwischen Strategie und Performance eine indirekte ist, die durch intervenierende Kontextvariablen vermittelt wird. (3) Fit as Matching geht davon aus, daß die Strategie und eine relevante Kontextvariable als zusammenpassend angesehen werden können, ohne daß dabei Bezug auf eine Kriteriumsvariable (wie z.B. Performance) genommen wird. (4) Fit as Gestalts impliziert, daß mehrere Variablen als zueinander passend angesehen werden. Dabei sind die Zusammenhänge zwischen den Variablen nicht sehr genau spezifiziert. (5) Fitas Profile Deviation geht von einem Nonnprofil theoretisch verbundener Variablen aus und mißt, inwieweit reale Profile von diesem abweichen. Schließlich (6) Fit as Covariation unterstellt, daß mehrere Kontextvariablen und die Strategie logisch-konsistent untereinander verbunden sein müssen, um die Performance positiv beeinflussen zu können. Während es für das generelle Verständnis des Strategic-Fit-Konzeptes ausreicht, davon auszugehen, daß zwei oder mehrere Variablen aufeinander abgestimmt (im Einklang) sein sollen, muß für die Entwicklung theoretischer Propositionen und die Durchführung empirischer Studien klar bestimmt werden, welche der obigen Fit-Definition Verwendung findet. In dieser Hinsicht hat die strategische Managementliteratur oft gesündigt. Für die nachfolgende Diskussion der vier Kontingenzfaktoren Organisationsstruktur, Managementqualifikation, Organisationskultur und Mikropolitik wird Fit als Moderation verstanden: Der Einfluß der Strategie auf die Untemehmungsperformance wird durch eine oder mehrere Kontextfaktoren moderiert. Diese Faktoren müssen mit der jeweiligen Strategie widerspruchsfrei verbunden sein, um positive Performancekonsequenzen zu ermöglichen. In dem Ausmaß, in dem unterstellt wird, daß alle relevanten Kontextfaktoren logisch-konsistent untereinander verbunden sein müssen, um die Performance positiv beeinflussen zu können, wird auch das Konzept des Fits als Kovariation bedeutsam.

522

R. Κ. F. Bresser

Strategie und Struktur Seit den frühen Arbeiten Chandlers (1962) und Rumelts (1974) geht die Literatur davon aus, daß Strategien und Organisationsstrukturen eng aufeinander abgestimmt sein müssen, um positive Performancekonsequenzen haben zu können (Miller, 1987). Unter Organisationsstrukturen werden nicht nur Spezialisierungs-, Zentralisierungs- und Koordinationsaspekte verstanden, sondern auch andere, mit der Organisationsstruktur zusammenhängende administrative Systeme, z.B. die Anreiz- und Kontrollsysteme. Was die Kausalität der Zusammenhänge angeht, so unterstützen die meisten Untersuchungen Chandlers zentrale These des „structure follows strategy" (Galbraith & Kazanjian, 1986: 13-27; Donaldson, 1987). Allerdings gibt es auch Beispiele, die belegen, daß die umgekehrte Beeinflussung ebenso existiert: Bestehende Strukturen können sich auf die Wahl der Unternehmungsstrategie auswirken.89 Wichtiger als die Richtung der Beeinflussungsbeziehung ist die generelle Notwendigkeit des Fits zwischen Strategie und Struktur.90 Die empirische Evidenz ist hier recht eindeutig. Die meisten Studien untersuchen Gesamtunteraehmungsstrategien, insbesondere das Verhältnis zwischen Diversifikationsart und Organisationsstruktur (Lorsch & Allen, 1973; Armour & Teece, 1978; Teece, 1981; Hill et al., 1992; Markides & Williamson, 1996). Die Ergebnisse lassen sich wie folgt zusammenfassen: Überdurchschnittliche Rentabilitäten ergeben sich, wenn folgende Strategie-Struktur-Kombinationen vorliegen: (a) Konzentrationsstrategie und funktionale Struktur, (b) verbundene Diversifikation und divisionale Struktur (mit Vorkehrungen zur Förderung der interdivisionalen Kooperation), (c) unverbundene Diversifikation und divisionale Struktur (mit Vorkehrungen zur Wahrung der Autonomie einzelner Divisionen). Auf der Ebene der Geschäftsbereiche gelten folgende Zusammenhänge: Performanceerhöhende Wirkungen ergeben sich, wenn Differenzierungsstrategien mit organischen Strukturmerkmalen gepaart werden und Kostenführerschaftsstrategien mit mechanistischen Strukturen (White, 1986; Miller, 1988). Aus der Perspektive des RBAs lassen sich Strategie-Struktur-Zusammenhänge durch besondere organisatorische Fähigkeiten begründen. Eine effizienz- und differenzierungsfördernde Abstimmung zwischen Unternehmungsstrategien und strukturellen Merkmalen erfordert organisatorische Fähigkeiten, die als „alignment skills" bezeichnet werden können (Powell, 1992). Unternehmungen, die ihre Strukturen immer wieder erfolgreich mit ihren Strategien in Einklang bringen, verfügen über dynamische Organisations- bzw. Implementierungsfähigkeiten (Teece et al., 1994), die durchaus zu Quellen nachhaltiger Wettbewerbsvorteile werden können. Wenn ein Konzern beispielsweise über Jahre hinweg eine auf Akquisitionen beruhende Wachstumsstrategie verfolgt, kann er besondere organisatorische Fähigkeiten hinsichtlich der Eingliederung von Übernahmekandidaten entwickeln. Zwischen der externen Wachstumsstrategie des Konzerns und den erforderlichen Integrationsfähigkeiten entwickelt sich somit ein strategischer Fit. 89

90

Die Ergebnisse der Studie von Fouraker und Stopford (1968) belegen z.B., daß diversifizierte Unternehmungen mit divisionaler Struktur häufiger den Weg der Internationalisierung ihrer Geschäftstätigkeiten beschreiten als Unternehmungen mit Konzentrationsstrategien und funktionaler Struktur. Das Erfordernis der Stimmigkeit zwischen Strategie und Struktur wird auch transaktionskostentheoretisch begründet (vgl. z.B. Williamson, 1975: 132-154; Jones & Hill, 1988). Da in diesen Arbeiten ebenfalls von einem Zusammenspiel zwischen Strategie und Struktur als Voraussetzung für überdurchschnittliche Rentabilitäten ausgegangen wird, ist der transaktionskostentheoretische Ansatz dem kontingenztheoretischen untergeordnet.

Strukturen, Personal, Kulturen, Mikropolitik

523

Sofern die Fähigkeiten, einen solchen Fit herzustellen, rar und nur schwer imitierbar sind, können sie über längere Perioden hinweg Wettbewerbsvorteile ermöglichen (Barney & Zajac, 1994).91 Die doppelte Rolle organisatorischer Fähigkeiten, sowohl Implementierungserfordernis als auch eigenständige Quelle von Wettbewerbsvorteilen darzustellen, wird in der empirischen Literatur durch Studien von Powell (1992), Henderson und Cockburn (1994) und Pisano (1994) belegt.

Strategie und Managementqualifikationen Die Bedeutung organisatorischer Fähigkeiten für die Strategieimplementierung ist eng verbunden mit dem für die Implementierung zuständigen Managementpersonal. Dieses Personal umfaßt die Träger besonderer Fähigkeiten. Insofern besteht die Notwendigkeit der Herstellung eines Fits zwischen der ausgewählten Unternehmungsstrategie und dem mit der Implementierung betrauten Führungspersonal (Schuler & Jackson, 1989; Welge & Al-Laham, 1992: 401-403). Die strategische Kontingenztheorie hat diese Zusammenhänge bereits früh erkannt und fordert in Abhängigkeit vom Strategietyp unterschiedliche führungsbezogene Persönlichkeitsprofile sowie Trainingsprogramme (Tichy, Fombrun & Devanna, 1982). Die empirische Evidenz über die Performancekonsequenzen eines Fits zwischen Strategien und Managementcharakterisika ist noch spärlich und uneindeutig (Govindarajan, 1989; Bryson & Bromiley, 1993). Daß ein solcher Zusammenhang besteht, gilt jedoch als gesichert, ebenso wie die Unterschiedlichkeit der Persönlichkeitsanforderungen bei verschiedenartigen Strategien (Galbraith & Kazanjian, 1986:100-107; Wheelen & Hunger, 1995: 256ff.)92.93 91

92

93

Vieles spricht dafür, daß besondere organisatorische Fähigkeiten den üblichen Desiderata des RBAs entsprechen: Aufgrund ihrer Historizität, kausalen Ambiguität und sozialen Komplexität dürften solche Fähigkeiten nicht nur rar sein, sondern oft auch erhebliche Imitationsbarrieren darstellen (Barney, 1991; Powell, 1992). Als Beispiele für diese Zusammenhänge sei an die bereits in Kapitel 10 angeführten Studien von Gupta & Govindarajan (1984; 1986) und Govindarajan (1989) erinnert: F&E-Expertise des Managements gilt z.B. als implementierungsfördernd bei der Verfolgung von Differenzierungsstrategien, und Produktionsexpertise gilt als Erfolgsdeterniinante von Kostenführerschaftsstrategien (Govindarajan, 1989). Neben den Funktionsbereichsqualifikationen werden als weitere implementierungsfördemde Persönlichkeitsmerkmale des Managements in der Literatur besonders oft die interpersonellen Fähigkeiten, die Risikobereitschaft und die Innovationsneigung hervorgehoben (Galbraith & Kazanjian, 1986: 107; Howell & Higgins, 1990). Wegen der Uneinheitlichkeit und noch geringen Zahl der Studien über die Implikationen eines Fits zwischen Strategie und Managementcharakteristika werden die Ergebnisse dieser Studien hier nicht detailliert dargestellt. Erwähnenswert ist auch der Unterschied zwischen der hier (in Anlehnung an die klassische strategische Kontingenztheorie) hervorgehobenen Bedeutung der Managementqualifikationen und den in Kapitel 4 thematisierten Sachverhalten. Die hier behandelte Fragestellung konzentriert sich auf die Auswahl des besten Managementpersonals für die Implementierung einer bereits formulierten Strategie. Die in Kapitel 4 skizzierte Thematik behandelt generelle Zusammenhänge zwischen den Charakteristika von Top-Management-Teams und der Auswahl von erfolgreichen Unternehmungsstrategien. Während in Kapitel 4 insbesondere die strategischen Wahlentscheidungen des TopManagements und damit Strategieformulierungsaspekte von Bedeutung sind, interessieren hier die implementierungsfördernden Qualifikationen des Managements. Trotz der bestehenden Unterschiede dieser Ansätze ist eine stärkere Integration der beiden Forschungsrichtungen anzuraten. Durch eine solche Integration kann z.B. der Frage nachgegangen werden, inwieweit systematische Unterschiede

524

R. Κ. F. Bresser

Bei vielen Implementierungsaufgaben ist die Fähigkeit des Managements, Widerstände gegenüber der verfolgten Strategie abzubauen, von entscheidender Bedeutung. Da durch Strategieveränderungen organisatorische Routinen und Machtverteilungen in Frage gestellt werden, treffen Implementierungsversuche häufig auf die Widerstände des mittleren Managements und anderer durch die Veränderungen betroffenen Gruppen. Derartige Widerstände und Konflikte können Implementierungen nicht nur verzögern, sondern insgesamt gefährden (Guth & MacMillan, 1986). Die Literatur hat sich deshalb mit der Fähigkeit des Managements, Verhaltenswiderstände abzubauen und eine strategiebezogene Akzeptanz zu vermitteln, intensiv auseinandergesetzt (Bourgeois, 1980; Bourgeois & Brodwin, 1984). Manager mit transformativen, visionären Führungsfähigkeiten94 gelten hierfür als besonders geeignet, denn sie übernehmen oft die Rolle des „champions", des Promotors strategischer Veränderungen (Schön, 1963; Witte, 1973). Generell wird davon ausgegangen, daß partizipative Führungsstile und Entscheidungsstrukturen in vielen Situationen widerstandsreduzierende Wirkungen haben können. In dem Ausmaß, in dem eine Unternehmung nachgeordnete Hierarchieebenen zunehmend in den Strategieformulierungsprozeß einbindet, fördert sie die inkrementale Entwicklung realisierter Strategien sowie deren breite Akzeptanz (Bourgeois & Brodwin, 1984). Der RBA hebt die Fähigkeiten des Managements als Quellen eigenständiger Wettbewerbsvorteile hervor. Dabei wird i.d.R. nicht zwischen Implementierungs- und Formulierungsfähigkeiten unterschieden; die Rede ist vielmehr von „managerial capabilities" schlechthin (Lado et al., 1992). Castanias und Helfat (1991) unterscheiden drei Arten von Managementfähigkeiten, die ökonomisch wertvoll95 sein können: generische (branchenunabhängige), branchenspezifische und unternehmungsspezifische Fähigkeiten. Wenn dem Management die Möglichkeit gegeben wird, sich einen Teil des aufgrund dieser Fähigkeiten herstellbaren Wertes anzueignen, besteht für das Management ein Anreiz, seine Fähigkeiten rentabilitätsfördernd und zugunsten sowohl des Managements als auch der Anteilseigner einzusetzen. Aus diesem Grunde existiert nach Auffassung der Autoren dann auch kein latenter Interessenkonflikt zwischen Anteilseignern und Management, wie er von der Agency-Theorie unterstellt wird. Da branchen- und insbesondere firmenspezifische Fähigkeiten nicht ohne weiteres übertragbar sind, können besondere Managementfähigkeiten der Unternehmung durchaus lange erhalten bleiben und zu Quellen dauerhafter Wettbewerbsvorteile werden (Castanias und Helfat, 1991; Knyphausen, 1993: 779).96

94

95

96

zwischen dem für die Formulierung und dem für die Implementierung zuständigen Managementpersonal bestehen bzw. bestehen sollten (Zaleznik, 1977). Transformative Führer sind in der Lage, Mitarbeiter von neuen Ideen zu begeistern, denn sie verfügen über Charisma und die Fähigkeit, sich mit den individuellen Wünschen ihrer Mitarbeiter auseinanderzusetzen und diese in Problemlösungen einfließen zu lassen (Bass, 1985; Tichy & Devanna, 1986; Westley & Mintzberg, 1989). Ökonomisch wertvoll sind Managementfähigkeiten, wenn sie der Erzielung von Ricardorenten und (aufgrund ihrer Knappheit und unzulänglichen Imitierbarkeit) dauerhaften Wettbewerbsvorteilen dienen, und wenn darüber hinaus zumindest Teile dieser Renten den Anteilseignern der Unternehmung zufließen. Die mangelnde Unterscheidung des RBAs zwischen Managementfähigkeiten zur Entwicklung innovativer Strategien einerseits und jenen zur Implementierung andererseits, rückt diese Diskussion in die Nähe der in Kapitel 4 dargestellten verhaltenswissenschaftlichen Ansätze zur Untersuchung der strategischen Entscheidungsträger. Dabei ist die ressourcenbasierte Diskussion zur Rolle des Top-

Strukturen, Personal, Kulturen, Mikropolitik

525

Strategie und Kultur Der strategische Kontingenzansatz geht - ebenso wie bei der Organisationsstruktur und dem Managementpersonal - davon aus, daß „corporate culture should follow strategy" (Wheelen & Hunger, 1995: 265). Unter dem Begriff Unternehmungskultur versteht man die Gesamtheit der in einer Unternehmung vorherrschenden Wertvorstellungen, die sich in den Traditionen, Anekdoten, Mythen, Normen und Denkhaltungen der Unternehmungsmitglieder widerspiegeln (Hinterhuber, 1989: 220; Mintzberg, 1990a: 167). Stark ausgeprägten Unternehmungskulturen kommt eine erhebliche Verhaltenssteuerungsfunktion zu.97 Durch die Internalisierung zentraler Werte und strategischer Zielsetzungen der Unternehmung wird das mit Implementierungsaufgaben betraute Personal dazu stimuliert, eigenständig kreative Lösungen für anstehende Implementierungsaufgaben zu entwickeln. Der im Verlauf einer Strategieimplementierung herzustellende Strategie-Kultur-Fit macht es erforderlich, die Ist-Kultur der Unternehmung zu erfassen und mit einer SollKultur zu vergleichen (Schwartz & Davis, 1981; Scholz, 1987: 90-102). Durch diesen Vergleich wird es möglich, die für einen gewünschten kulturellen Wandel notwendigen Maßnahmen zu determinieren. In der theoretischen Literatur existieren kritische Stimmen, die die Möglichkeit (und auch den Wert) der Veränderung von Unternehmungskulturen durch Managementinterventionen bezweifeln (Smircich, 1983; Schreyögg, 1991; Krell, 1994: 248-281).98 Demgegenüber belegen viele Beispiele aus der Praxis, daß kultureller Wandel plan- und implementierbar ist (Peters & Waterman, 1982; Bresser, 1986). Der Erfolg und die Schnelligkeit des Wandels einer Unternehmungskultur hängen erheblich von der Kompatibilität der bisherigen Kultur mit der intendierten neuen Strategie ab, d.h. von dem Ausmaß, in dem bestehende Wertvorstellungen durch strategische Reorientierungen in Frage gestellt werden. Besondere Schwierigkeiten können ungeschriebene, versteckte kulturelle Normen bereiten, denn diese gilt es im Rahmen der Ist-Kultur-Erfassung zu entschlüsseln, um sie bei den geplanten Veränderungsmaßnahmen berücksichtigen zu können (ScottMorgan, 1994). Auf jeden Fall muß davon ausgegangen werden, daß ein tiefgreifender kultureller Wandel durch einen langfristigen und oft schwierigen, schmerzvollen Prozeß gekennzeichnet ist. Starke Kulturen sind historisch gewachsen, und Organisationsmitglieder fühlen sich auf diese Kulturen verpflichtet und sind emotional an sie gebunden. Es verwundert insofern nicht, daß Kulturvariationen häufig von Widerständen begleitet sind und besondere Herausforderungen an die Führungseigenschaften des Managements stellen (Dunbar et al., 1982).

97

98

Managements aus behavioristischer Sicht jedoch noch zu steril, wie die Kritik des Ansatzes in Kapitel 7 (Kritikpunkt 5) hervorhebt. Von Interesse sind in diesem Abschnitt ausschließlich die dominierenden, von der Untemehmungsleitung sanktionierten Organisationskulturen. Neben dominierenden Kulturen unterscheidet die Literatur auch Subkulturen, die mit dominierenden Kulturen in einem angespannten Verhältnis stehen können (Martin & Siehl, 1983; Sackmann, 1992). Diese Kritik wird ausschließlich von verhaltenswissenschaftlich orientierten Wissenschaftlern vorgetragen. Die ökonomische Theorie argumentiert demgegenüber ähnlich wie der strategische Kontingenzansatz. So hat z.B. Kreps (1990) die Notwendigkeit eines Strategie-Kultur-Fits spieltheoretisch begründet und damit die generelle Gestaltbarkeit von Unternehmungskulturen vorausgesetzt.

526

R. Κ. F. Bresser

Die empirische Evidenz zur Bedeutung eines Strategie-Kultur-Fits für die Leistungsfähigkeit von Organisationen beruht fast ausschließlich auf Fallstudien.99 Dies liegt an der Schwierigkeit, ein schillerndes, ätherisches Konzept wie das der Unternehmungskultur den „harten" (quantitativen) Methoden der empirischen Sozialforschung zugänglich zu machen (Mintzberg, 1990a: 171). Die Fallstudien liefern jedoch vielfältige Belege dafür, daß Unternehmungsstrategien und -kulturen kompatibel sein sollten, um geplanten strategischen Wandel zu ermöglichen. Während die ressourcenbasierte Diskussion die Bedeutung implementierungsfördernder Unternehmungskulturen nicht in Frage stellt, konzentriert sie sich auf die Frage, ob und unter welchen Bedingungen Kulturen dauerhafte Wettbewerbsvorteile begründen können (Barney, 1986; Fiol, 1991). Barney (1986) nennt die drei Bedingungen des RBA: Die Kulturen müssen ökonomischen Wert entstehen lassen, knapp (bzw. einzigartig) und schwer imitierbar sein. Es läßt sich durchaus argumentieren, daß starke Kulturen, die sich z.B. durch Teamgeist und Innovativität auszeichnen, ökonomische Werte entstehen lassen können (Peters & Waterman, 1982). Aufgrund ihrer firmenspezifischen Historizität sind viele Kulturen auch einzigartig und nur unvollkommen imitierbar. Die Imitierbarkeit wird zusätzlich durch die unklaren Kausalzusammenhänge zwischen kulturellen Wertesystemen und dem Erfolg der Unternehmung erschwert (Knyphausen, 1993: 777). Auch der RBA erkennt die Schwierigkeiten (z.B. Widerstände), die für viele Versuche der Veränderung bestehender Unternehmungskulturen kennzeichnend sind. Fiol (1991) argumentiert, daß derartige Veränderungen durch weitgehend unstrukturierte Lernprozesse in die Wege geleitet werden sollten. Nur so könne eine neue Kultur entstehen, die wertvoll, einzigartig und vor erfolgreicher Imitation geschützt sei.

Strategie und Mikropolitik Obwohl bewußt gestaltete strategische Planungsprozesse dazu beitragen, strategische Entscheidungen zu versachlichen, finden diese in der Realität in einem Umfeld mikropolitischer Machtspiele und Konflikte statt (Mintzberg, 1983; Hill & Jones, 1992: 396423). Unter Mikropolitik (politics) versteht man sämtliche Versuche von Organisationsmitgliedern, Macht zu erlangen, um die Ziele sowie die strategischen und operativen Entscheidungsprozesse innerhalb der Unternehmung im Sinne eigener Interessen zu beeinflussen (Miles, 1980: 154; Mintzberg, 1983: 172). Um Macht und Einfluß (jenseits formaler Zuständigkeiten) zu gewinnen, bedienen sich Organisationsmitglieder verschiedener politischer Verhaltensweisen. Beliebt sind das Bilden von Koalitionen (March, 1962; Narayanan & Fahey, 1982) sowie Versuche, bestimmte Interessen durch Verhandlungen, Drohungen oder Kompromisse durchzusetzen (Mintzberg, 1983: 187-217; Falbe & Yukl, 1992). Die Quellen mikropolitischer Verhaltensweisen liegen oft in Meinungsverschiedenheiten und Konflikten über anstehende strategische Entscheidungen. Da strategische Ent-

99

Vgl. z.B. die Studien in dem von Jelinek et al. (1983) herausgegebenen Sonderheft der Administrative Science Quarterly sowie Dunbar et al. (1982) und Feldman (1986).

Strukturen, Personal, Kulturen, Mikropolitik

527

Scheidungen die Machtverteilungen100 und Interessen innerhalb der Unternehmung beeinflussen, sind solche Konflikte unvermeidbar. Durch anstehende Entscheidungen werden die Interessen einzelner Unternehmungsmitglieder und Abteilungen unterschiedlich stark tangiert. Deshalb versuchen verschiedene Akteure, durch politisches Verhalten die Unternehmungsentscheidungen im Sinne eigener Interessen zu beeinflussen (Hill & Jones, 1992: 399). Mikropolitik wird negativ beurteilt, wenn sie lediglich zum Ausbau individueller Machtbasen dient und sich unkontrolliert vollzieht. Politische Verhaltensweisen können aber auch positive Konsequenzen haben, wenn sie dazu verwendet werden, legitime Unternehmungsaufgaben besser durchzuführen. „Politics" können z.B. dazu führen, daß notwendige Veränderungen vorgenommen werden, um Unternehmungen durch innovative Problemlösungen aus einer Krise herauszuführen (Bresser, 1987; Hill & Jones, 1992: 400). In der Literatur werden die positiven, auf Wandel ausgerichteten Konsequenzen mikropolitischer Verhaltensweisen besonders hervorgehoben (Eisenhardt & Zbaracki, 1992). Durch den Einsatz mikropolitischer Verhaltensweisen kann ein rascher strategischer Wandel herbeigeführt werden, wenn es gelingt, die Verzögerungseffekte tradierter Organisationskulturen auszuschalten, und die Kulturen selbst in den Wandel einzubeziehen (Dunbar et al., 1982; Mintzberg, 1990a: 166). Das Thema Mikropolitik verweist auf den engen Zusammenhang zwischen Strategieformulierung und -implementierung.101 Die Allgegenwärtigkeit mikropolitischer Auseinandersetzungen impliziert, daß sowohl Strategieformulierungs- als auch implementierungsprozesse politischer Natur sind (Narayanan & Fahey, 1982; Gray & Ariss, 1985). Der Einfluß der Mikropolitik auf die Strategieformulierung läßt sich z.B. im Rahmen der von Hambrick und Mason (1984) entwickelten Top-Management-Perspektive modellieren und erforschen. Aus der Implementierungsperspektive interessiert insbesondere die Herstellung eines Strategie-Politik-Fits. Grundsätzlich kann davon ausgegangen werden, daß das mikropolitische Umfeld der Unternehmung durch Managementinterventionen beeinflußbar ist (Gray & Ariss, 1985). Pondy (1967) zeigt, wie durch Veränderungen der in diesem Kapitel behandelten Designvariablen (Struktur, Personal und Kultur) der Kontext latenter Konflikte und politischer Machtspiele verändert werden kann. Eine (Ent)politisierung kann durch die Beeinflussung struktureller Gegebenheiten, z.B. der Aufgabeninterdependenz, der Anreizsysteme, der organisatorischen Prozeduren oder des Wettbewerbs um knappe Ressourcen angestrebt werden. Durch die Versetzung, Entlassung oder Neueinstellung von Managementpersonal ist das mikropolitische Umfeld über die Managementqualifikationen und -verhaltensprädispositionen beeinflußbar. Auch

100

101

Unter Macht versteht man die Fähigkeit einer Person, Abteilung oder Division, bei anderen gewünschte Verhaltensweisen auch gegen deren Willen durchzusetzen (Miles, 1980: 164ff.). In Unternehmungen ergeben sich konkrete Machtverteilungen meist aufgrund früherer strategischer Entscheidungen, durch die bestimmten Individuen oder Abteilungen unterschiedlich wichtige Rollen zukommen. Unternehmungsentscheidungen können z.B. dazu führen, daß bestimmte Abteilungen oder Individuen, besser als andere Unsicherheiten in der Aufgabenumwelt oder der Unternehmung abbauen können, da sie über mehr Informationen und Know-how verfügen. Machtdifferentiale können auch entstehen, wenn bestimmte Funktionsträger nicht-substituierbare Aufgaben erfüllen oder sich im Entscheidungszentrum eines Netzes von interdependenten Unternehmungseinheiten befinden (Hickson et al., 1971). Dieser Zusammenhang ist selbstverständlich auch für die Wirkungen der Oganisationsstruktur, -kultur und der Managementqualifikationen gegeben. Er ist bei der Mikropolitik nur besonders evident.

528

R. Κ. F. Bresser

kann die Intensität der Mikropolitik durch Kulturveränderungen beeinflußt werden, z.B. über die Steuerung vorherrschender Gruppennormen. Die Herstellung eines Strategie-Politik-Fits bedeutet, daß je nach Umfang des intendierten strategischen Wandels die Machtbalance sowie die Art und das Ausmaß der mikropolitischen Aktivitäten variiert werden sollten (Gray & Ariss, 1985; Hill & Jones, 1992: 405f.). Bei umfassenden strategischen Reorientierungen, wie z.B. der Einführung technologischer Innovationen, sollten Koalitionen und andere politische Verhaltensweisen (sachliches Argumentieren, Verhandlungen usw.) in vergleichsweise hohem Ausmaß Verwendung finden, um Widerstände frühzeitig durch breite Unterstützung ersetzen zu können (Howell & Higgins, 1990). Die Relevanz mikropolitischen Verhaltens für die Adoption und Implementierung strategischer Entscheidungen ist durch mehrere Fallstudien recht gut abgesichert (z.B. Allison, 1971; Quinn, 1980; Eisenhardt & Bourgeois, 1988).102 Demgegenüber sind die Notwendigkeit der Herstellung eines Strategie-Politik-Fits und die dadurch zu erwartenden Performancekonsequenzen vornehmlich theoretisch-argumentativ begründet. Eisenhardt und Zbaracki (1992: 34) fordern deshalb, die implementierungsorientierte empirische Forschung zu verstärken. Der RBA hat sich der Mikropolitik bisher nicht zugewendet. Dies ist zumindest aus zwei Gründen bemerkenswert: Zum einen läßt sich das übliche Instrumentarium zur Begründung dauerhafter Wettbewerbsvorteile leicht auf die Ressource „Mikropolitik" übertragen. Zum anderen zeichnet sich vor dem Hintergrund negativer Erfahrungen mit mikropolitischem Verhalten (Eisenhardt & Bourgeois, 1988; 1992; Dean & Sharfman, 1996) zur Zeit eine Kontroverse über den Wert von „politics and conflict" in der Literatur ab (Eisenhardt & Zbaracki, 1992). Es wäre ein lohnendes Unterfangen, diese Diskussion durch ressourcenbasierte Analysen anzureichern.

Weiterführende Implementierungsforschung Für die Weiterentwicklung des Wissensstandes zu Fragen der Strategieimplementierung erscheinen zwei kritische Beobachtungen relevant: (1) Die statische Orientierung der Kontingenztheorie. Einige Autoren stehen der Kontingenztheorie und dem Konzept des Strategie Fit skeptisch gegenüber, da beide häufig eine statische Ausrichtung haben (Venkatraman & Camillus, 1984; Miller & Friesen, 1984; Ginsberg & Venkatraman, 1985). In vielen Untersuchungen zum Fit-Phänomen wird zumindest implizit unterstellt, daß durch die Herstellung einer Entsprechung zwischen oft nur zwei Variablen (z.B. Strategie und Struktur) die erfolgreiche Implementierung über längere Zeiträume hinweg ermöglicht wird. Dabei wird übersehen, daß die Herstellung eines Fits ein multidimensionaler, dynamischer Prozeß ist, in dem mehrere Dimensionen der Organisationsgestaltung kontinuierlich an sich verändernde Umweltgegebenheiten angepaßt werden müssen (Venkatraman & Camillus, 1984; Van de Ven & Drazin, 1985). Zajac et al. (1997) fordern daher,

102

Vgl. auch die Übersicht einschlägiger Studien bei Eisenhardt & Zbaracki (1992: 24f.).

Strukturen, Personal, Kulturen, Mikropolitik

529

Strategic-Fit-Modelle zu dynamisieren und als solche empirisch zu überprüfen. 103 Nicht zuletzt aufgrund der statischen Ausrichtung strategischer Konsistenztheorien hat sich eine Prozeßforschung des Strategischen Managements herausgebildet. Wie im folgenden gezeigt wird, war diese Prozeßforschung bisher für Fragestellungen der Strategieimplementierung wenig ergiebig. (2) Die Vernachlässigung der Implementierung in modernen Forschungsansätzen. Wenn man die Beiträge der strategischen Prozeßforschung insgesamt betrachtet, so fällt auf, daß diese sich vor allem mit Fragestellungen der strategischen Wahl, des Strategiewandels sowie der Entwicklung innovativer Strategien und Kompetenzen auseinandersetzen (Pettigrew, 1992). Demgegenüber werden die dynamischen Probleme der Strategieimplementierung weitgehend vernachlässigt. 104 Der schwache Fokus auf Implementierungsfragen könnte daher rühren, daß die Prozeßforschung oft hervorhebt, Implementierungsfragen seien nicht losgelöst vom Prozeß der Generierung bestimmter Strategieinhalte diskutierbar (Pettigrew, 1992; Barney & Zajac, 1994). Der enge Zusammenhang zwischen Prozessen der Strategieformulierung und -Implementierung wird heute nicht mehr ernsthaft in Frage gestellt: Wenn beispielsweise bestimmte Strategieinhalte durch einen partizipativen Entscheidungsprozeß determiniert werden, wird man im Rahmen der Strategieimplementierung weniger und andere Vorkehrungen zum Abbau von Widerständen treffen müssen, als bei ausschließlich vom Top-Management geprägten Inhalten. Dennoch ist der Prozeß der Implementierung einer Strategie eine separate Aufgabe, und die hierzu erforderlichen Maßnahmen sollten deshalb gesondert thematisiert werden. Wie bereits zu Beginn des Kapitels dargelegt, soll die separate Behandlung der Strategieimplementierung gewährleisten, daß die Umsetzung strategischer Intentionen effektiv und effizient erfolgt. Auch die Beiträge des RBAs behandeln die Strategieimplementierung „stiefmütterlich". Die Literatur zum RBA beschäftigt sich primär mit der Frage, wie Ressourcen und Fähigkeiten entstehen und zu Quellen dauerhafter Wettbewerbsvorteile werden. Ressourcen und Fähigkeiten, die zwar der Implementierung von Strategien sehr förderlich sein können, aber aus denen sich keine nennenswerten Wettbewerbsvorteile ableiten lassen, fallen weitgehend unter den Tisch. 103

104

Die Forderung, strategischen Kontingenztheorien eine dynamischere Ausrichtung zu geben, gilt nicht nur für den Bereich der Strategieimplementierung. Sie ist relevant, wo immer der Kontingenzgedanke strategische Forschungen anleitet, z.B. auch bei den in Kapiteln 9 bis 12 geforderten kontingenztheoretischen Präzisierungen. Bei der Untersuchung kontingenztheoretischer Zusammenhänge sowie ihrer Dynamisierung kann es auch immer nur um das Studium weniger aber wichtiger theoretischer Konstrukte gehen. Durch die Beschränkung auf wenige Konstrukte und generelle Beziehungszusammenhänge entstehen nicht nur aussagekräftige Theorien, die ein breiteres Spektrum an Phänomenen abdecken (Bacharach, 1989; Sutton & Staw, 1995), eine Beschränkung ist auch notwendig, um die Operationalität kontingenztheoretischer Modelle sicherzustellen (Schreyögg, 1992). Die Vernachlässigung der Strategieimplementierung wird z.B. in drei Sonderheften des Strategie Managementjournals evident, die sich mit strategischen Prozessen beschäftigen. Vgl. hierzu die Beiträge (a) in dem von Chakravarthy und Doz (1992) herausgegebenen Sonderheft über „strategy process research", (b) in dem von Pettigrew (1992) herausgegebenen Sonderheft über ,fundamental themes in strategy process research" und (c) in dem von Barney und Zajac (1994) herausgegebenen Heft über „competitive organizational behavior". Zu einem ähnlichen Ergebnis kommt Al-Laham (1997: 6, 5559) in seiner Würdigung der strategischen Prozeßforschung: Hinsichtlich der Gestaltung effizienter Prozeßverläufe besteht ein erhebliches Forschungsdefizit.

530

R. Κ. F. Bresser

Will man den Implementierungsgedanken stärker in der Prozeßforschung und dem RBA verankern, ist es vielleicht sinnvoll, ähnlich wie Porter (1996) zwischen operativen und strategischen Aktivitäten und Kompetenzen zu unterscheiden. Man könnte z.B. das Strategic-Fit-Konzept aufspalten in die Begriffe „Operational Fit" und „Competitive Fit". Operational Fit bezieht sich darauf, alle anstehenden Aufgaben und Tätigkeiten effektiv, effizient und möglichst besser als die Konkurrenten durchzuführen und außerdem auf die Konsistenz aller Tätigkeiten mit der Unternehmungsstrategie zu achten.105 Es handelt sich hierbei also essentiell um Tätigkeiten, die traditionell in den Bereich der Strategieimplementierung fallen. Nur muß damit gerechnet werden, daß effektive, effiziente und konsistente Aufgabenerfüllungen keine dauerhaften Wettbewerbsvorteile begründen: Rivalen können optimale Implementierungspraktiken in den meisten Fällen relativ leicht imitieren. Competitive Fit bedeutet, einzigartige Aktivitäten und Ressourcen zu mobilisieren. Durch die Vernetzung und Optimierung einzigartiger Aktivitäten und Ressourcen innerhalb der Unternehmung soll versucht werden, Kernkompetenzen und (relativ) dauerhafte Wettbewerbs vorteile aufzubauen.106 Aufgrund der Komplexität und Einzigartigkeit dieses vernetzten Aktivitätensystems wird es für Konkurrenten schwierig, erfolgreiche Imitationsversuche durchzuführen. Die zu einem Competitive Fit beitragenden Aktivitäten und Ressourcen beziehen sich somit auf die Kernaufgabe der Strategieformulierung: die Entwicklung innovativer, wettbewerbsvorteilgenerierender Unternehmungsstrategien.107

Die Artikel Zunächst folgt ein Beitrag zum Strategie-Struktur-Fit. Der Artikel, „Cooperative Versus Competitive Structures in Related and Unrelated Diversified Firms", von Hill, Hitt und Hoskisson ist in die Gruppe der Diversifikationsstudien einzuordnen. Die Autoren argumentieren, daß die ökonomischen Vorteile der verbundenen Diversifikation durch Economies of Scope und die der unverbundenen Diversifikation durch eine effiziente interne Organisation begründet sind. Beide Diversifikationsarten benötigen deshalb unterschiedliche organisatorische Arrangements, um die Vorteile der jeweiligen Strategie voll ausschöpfen zu können: Unternehmungen, die Economies of Scope ausschöpfen wollen, benötigen Strukturen, die die Kooperation zwischen Geschäftseinheiten erleichtern. Unternehmungen, die durch eine effiziente interne Organisation ökonomische Vorteile erzielen wollen, stellen sich besser, wenn ihre Strukturmerkmale den Wettbewerb zwischen den Geschäftseinheiten anfachen. Anhand einer Querschnittsuntersuchung von 184 Konzernen belegen die Autoren, daß ein Fit zwischen Diversifikationsstrategie und Organisationsstruktur mit überdurchschnittlichen Rentabilitäten verbunden ist. Der zweite Beitrag, „Champions of Technological Innovation", von Howell und Higgins liefert empirische Belege für die Bedeutung eines Fits der Untemehmungsstrategie 105

106 107

Diese Definition von Operational Fit entspricht dem, was Porter (1996) als „operational effectiveness" und „first-order-fit" bezeichnet. Porter (1996) spricht in diesem Zusammenhang von „second-order-fit" und „third-order-fit". „Operational Fit" und „Competitive Fit" im hier beschriebenen Sinne unterstellen ein Kovariationskonzept (Venkatraman, 1989). Durch die logisch-konsistente und sich gegenseitig fördernde Verbindung von Aktivitäten und Ressourcen entsteht ein Zusammenhang zwischen mehreren Variablen, der sich als Verbundeffekt auf die Untemehmungsperformance auswirken soll.

Strukturen, Personal, Kulturen, Mikropolitik

531

mit den Managementcharakteristika und dem mikropolitischen Umfeld. Das empirische Beispiel bezieht sich auf den Informationsbereich der Unternehmung und die Verfolgung einer Funktionsbereichsstrategie der Nutzung moderner Informationstechnologien. Informationstechnologische Neuerungen treffen oft auf intensive Widerstände des Personals, und ihre erfolgreiche Implementierung stellt deshalb besondere Ansprüche an das Führungspersonal. Für 25 Unternehmungen, die ein informationstechnologisches Innovationsprojekt implementiert haben, wird der Einfluß der mit der Implementierung betrauten Manager untersucht. Dabei wird für jedes Projekt zwischen Managern unterschieden, die die Rolle eines „Projektchampions" einnehmen und jenen, die zwar auch bei der Implementierung involviert sind, aber eher normales Führungsverhalten zeigen. In allen Fällen ist der Beitrag der .Projektchampions" nicht nur instrumental für den Erfolg der Implementierung, als Gruppe stechen die „Champions" auch mit besonderen Merkmalen hervor: Sie sind risiko- und innovationsfreudiger als „Nonchampions", und sie verwenden signifikant häufiger transformatives Führungsverhalten und mikropolitische Beeinflussungstaktiken.

532

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson ORGANIZATION SCIENCE Vol. 3, No. 4, November 1992 Printed in U.S.A.

COOPERATIVE VERSUS COMPETITIVE STRUCTURES IN RELATED AND UNRELATED DIVERSIFIED FIRMS* CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON Department of Management and Organization, Schools of Business, University of Washington, Seattle, Washington 98195 Department of Management, Texas A &M University, College Station, Texas 77843 Department of Management, Texas A & M University, College Station, Texas 77843 Herein we argue that different diversification strategies are associated with different sets of economic benefits. Firms that have diversified into related areas can realize benefits from economies of scope, while those that have diversified into unrelated areas can realize benefits from efficient internal governance mechanisms. We hypothesize that distinctly different internal organizational arrangements are required to realize these different benefits. Firms attempting to realize economies of scope need organizational arrangements that stress cooperation between business units. Firms attempting to realize economic benefits from efficient internal governance need organizational arrangements that stress competition between business units. If a diversified firm is to achieve high performance it must establish an appropriate fit between its diversification strategy on the one hand, and its organizational structure and control systems on the other. We test this thesis on 184 Fortune 1000 firms that participated in a survey of organizational arrangements. The results indicate that the appropriate fit between strategy, structure, and control systems is associated with superior performance. Firms attempting to realize economies of scope perform better if their organizational arrangements stress cooperation between business units, while firms attempting to realize economic benefits from efficient internal governance perform better if their organizational arrangements stress competition between business units. (DIVERSIFICATION STRATEGY; ORGANIZATIONAL STRUCTURE AND ECONOMIC P E R F O R M A N C E )

Introduction The relationship between diversification strategy and economic performance has been the focus of extensive research in the strategic management literature (e.g. Amit and Livnat, 1988; Bettis 1981; Christensen and Montgomery 1981; Montgomery 1985; Rumelt 1974, 1982; Wernerfelt and Montgomery 1988). Despite this research, a recent survey of the literature concluded that "the findings of studies attempting to demonstrate the effects of diversification on performance remain inconclusive (Ramanujam and Varadarajan 1989, pp. 539-541). Aside from methodological flaws, there is a plausible explanation. A significant amount of research on diversification has ignored the importance of implementation on the strategy-performance relationship. This is unfortunate because almost 30 years ago Chandler (1962) noted that the success of a diversification strategy depends on how it is implemented. Diversification alone will not produce superior performance; the firm must also adopt the appropriate internal organizational arrangements. By ignoring the effects of organizational characteristics, many prior studies may have produced erroneous or, at best, incomplete results. Accordingly, the objective of the current study is to explore how organizational factors influence the relationship between diversification strategy and economic "Accepted by Donald C. Hambrick; received March 1990. This paper has been with the authors for two revisions. 501 1047-7039/92/0304/0501 /$01.25 Copyright O 1992, The Institute οΓ Management Sciences

Structures in Diversified Firms 502

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON

performance. The paper builds on a stream of literature that relates differences in diversification strategy to differences in internal organizational arrangements and managerial reward systems (e.g. Ackerman 1970; Berg 1969, 1973; Govindarajan and Fisher 1990; Gupta and Govindarajan 1986; Hill and Hoskisson 1987; Hill 1988; Kerr 1985; Keats and Hitt 1988; Lorsch and Allen 1973; Pitts 1977, 1980; Vancil 1980). Drawing on this literature, we relate differences in the degree of decentralization, the degree of integration, internal control practices, and incentive structures, within multidivisional (M-form) firms to diversification strategy and performance. The paper begins by arguing that different diversification strategies are associated with different economic benefits. Realizing different benefits is hypothesized to require distinctly different internal organizational arrangements. This leads to the proposition that performance is not the consequence of a firm's diversification strategy per se, but of the degree of fit between corporate strategy and internal organizational arrangements. Diversification and Economic Benefits The economic benefits of related diversification have been argued to arise from the ability of such firms to exploit economies of scope (Jones and Hill 1988; Porter 1987; Teece 1982). Providing precision to the term "synergy", economies of scope simply means that for two outputs, XI and X2, the value created by their joint production is greater than the value created if they are produced separately. Within diversified firms, achieving joint production requires some degree of resource sharing and/or skill transfers between otherwise separate businesses (Porter 1987). For example, related businesses may utilize common distribution channels, engage in common advertising campaigns when their products are compatible, share marketing and technological information for mutual gain, transfer skills between activities, and share manufacturing facilities in order to realize economic benefits from the optimal utilization of capacity. As articulated by Porter (1987), resource sharing and skill transfers enable the diversified firm either to reduce overall operating costs in one or more of its divisions, and/or to better differentiate the products of one or more of its divisions (thus enabling a higher price to be charged). In both cases, economic wealth is created by increasing value added. The economic benefits of unrelated diversification have been argued to arise from the governance characteristics of such firms (Dundas and Richardson 1982; Jones and Hill 1988; Williamson 1975, 1985). We shall refer to these benefits as governance economies. The argument is built on the proposition that the stock market, as an external governance mechanism, suffers from two limitations which constrain its ability to efficiently allocate resources between firms, and to discipline the managers of poorly performing firms. The first limitation is that stockholders experience information disadvantages in their relationship with the firm (Black 1986; Hill 1988; Williamson 1985). The existence of information asymmetries gives managers the ability to pursue objectives and undertake expenditures that, while in their own interests, are not consistent with optimizing the efficiency of the firm. For example, resources may be absorbed in lavish executive suits, a fleet of executive jets, company cars, excess staff, expense paid trips to "conferences" that front for vacations, and so on. Alternatively, resources may be invested within the firm in a manner that reflects internal political considerations rather than actual investment needs. While such expenditures are not consistent with efficiency, they may occur because stockholders, who might object, cannot detect them.

533

534

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

503

A second stock market limitation is that even when stockholders detect that managers are not investing funds in the most efficient manner, they are limited in the disciplinary actions they can initiate to major discrete moves, such as selling their holdings in order to precipitate a takeover. They cannot intervene in corporate affairs to 'fine tune" operations. This is unfortunate, since takeovers represent an expensive adjustment mechanism due to the high premiums over market value paid for acquired firms (averaging over 60% in recent years) and post-takeover adjustment costs (Mueller 1989). Consequently, disciplinary takeovers are unlikely to occur unless efficiency shortfalls have become very serious.1 Thus, substantial firm inefficiencies may go uncorrected by the stock market. In contrast, it has been argued that the corporate office of the unrelated firm, as an internal governance mechanism, can overcome these limitations in its relationship with divisions (Dundas and Richardson 1980; Jones and Hill 1988; Hill 1988; Williamson 1975, 1985). Corporate managers can use internal audits to negate information asymmetries, identifying opportunistic behavior and inefficient expenditures by divisional managers. They can use performance monitoring systems, selection and termination policies, reward and incentive schemes, and direct intervention to "fine tune" operations and discipline divisional managers who fail to optimize the performance of the operations under their control. In addition, internal competition for capital resources between divisions can be established as a mechanism for enhancing control and as an incentive for greater performance. Clearly, this notion does not deny the fact that information asymmetries and political considerations also influence economic decision making within the unrelated firm.2 However, the aforementioned arguments are valid if these problems are relatively less serious between the corporate office and divisions within the firm, than between stockholders and free standing firms. The implication is that due to information and control advantages, unrelated firms that function in this manner should be able to achieve a more efficient allocation of capital resources between divisions, and police the efficiency of divisions more effectively, than the stock market could were each division an independent firm. Thus, an unrelated diversified firm may realize economic benefits by acquiring inefficient firms that have not been disciplined by the stock market, and exposing them to internal audits, tight internal financial controls, and incentive and reward structures that promote improvements in efficiency. Economic Benefits and Organizational Arrangements It has been argued that the multidivisional (M-form) structure is the appropriate organizational form for diversified firms (Chandler 1962) and the evidence suggests that most diversified firms do operate with an M-form structure (e.g. Hill and Pickering 1986; Rumelt 1974). However, M-form structures are not a homogeneous set of organizational arrangements. Studies have found substantial differences in internal arrangements among M-form firms with regard to centralization, integration, and internal control configurations (e.g. Allen 1978; Hill and Pickering 1986; Lorsch and Allen 1973; Vancil 1980). There is a need for researchers to identify how internal organizational arrangements should be configured within the M-form so that the economic benefits of pursuing a given strategy can be realized. 'The empirical evidence seems to indicate that this is the case. For example, Singh (1971) found that over considerable ranges of profitability the probability of being acquired was the same, and that only firms with the very worst profit records stood a greater than average chance of being acquired. 2 Bower (1970) has described in some detail how information asymmetries can re-emerge between corporate executives and divisional personnel in diversified firms.

Structures in Diversified Firms 504

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON

It has been proposed that related diversification is associated with economies of scope and unrelated diversification with governance economies. We argue that, within the context of an M-form structure, the realization of economies of scope requires organizational arrangements that stress cooperation between divisions while the realization of governance economies requires organizational arrangements that emphasize competition between divisions. These two sets of arrangements are suggested to be incompatible. Cooperative Organizations Within related diversified firms cooperation between divisions is necessary to realize economies of scope. There is a need to coordinate the activities of otherwise independent divisions so that skills can be transferred and resources shared (Porter 1985). Child (1984) has argued that some centralization is necessary to achieve such coordination. Similarly, Mintzberg (1983) noted that interdependencies between divisions in related diversified firms encourage the corporate office to retain some control over the functions common to the divisions to ensure coordination. Consistent with this view, both Berg (1973) and Pitts (1977) found evidence that the interdivisional sharing of technological resources was achieved through centralization of research activities. Also, as discussed by Sloan (1963) in the case of General Motors, centralized oversight is necessary to ensure that the strategy and investments of cooperating divisions are complementary. (At GM centralized oversight was used to ensure consistency between the pricing and product strategies of different divisions so that, to paraphrase Sloan, Cadillac didn't produce a car in the Chevrolet price range.) Thus, Ackerman (1970) found that the identification of and impetus for major capital investment decisions was more centralized within M-form firms with a high degree of interdivisional integration than M-form firms with a low degree of interdivisional integration. In addition, the input of senior executives may be needed to resolve conflicts between personnel arising from interdependent divisions (Boulding 1964). Therefore, some degree of centralized control over the strategic and operating decisions of interdependent divisions in related diversified firms is required to realize economies of scope. Thus: Hypothesis 1. The interaction between related diversification and centralization mil be positively related to firm performance. In addition to centralization, coordination between divisions also requires integrating mechanisms to achieve lateral communication between divisions. The complexity of these mechanisms will vary, depending on the extent of interdependence, from simple liaison roles and temporary task forces, to permanent teams (Child 1984; Galbraith 1977; Lawrence and Lorsch 1967). As noted by Luke, Begun, and Pointer (1989), in related diversified firms interdependent divisions need to be tightly coupled. Thus: Hypothesis 2. The interaction between related diversification and integration will be positively related to firm performance. One problem with attempts to achieve coordination between interdependent divisions is that it can create performance ambiguities (Govindarajan and Fisher 1990; Gupta and Govindarajan 1986; Hill and Pickering 1986; Lorsch and Allen 1973; Vancil 1980). When divisions lack complete autonomy with regard to operating and strategic decisions, objective financial criteria, which might be used to assess divisional performance, do not constitute an unambiguous signal of divisional efficiency. Poor financial performance of a certain division might be due to inefficiencies within that division, or inefficiencies within another division with which it is tightly coupled.

535

536

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson C O O P E R A T E VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

505

Alternatively, poor performance might be due to poor central input into key operating decisions. Without access to more information, it can be difficult to assign accountability. The firm can overcome performance ambiguity problems by increasing the amount of information it processes (Daft and Lengel 1986). More precisely, as interdivisional coordination increases, the firm may de-emphasize objective market-based performance measures of divisional performance, such as return on investment, and emphasize more subjective modes of evaluating performance (Govindarajan and Fisher 1990; Hill 1988; Ouchi 1980). That is, within cooperative organizations the corporate office needs to base its assessment of divisional performance on a wide range of criteria. These criteria might include subjective measures of divisional performance (e.g. ability to innovate, degree of cooperation among interdependent divisions) along with objective noniinancial measures (e.g. labor productivity, capacity utilization, market share, and growth). In addition, we would expect cash flows to be allocated by the corporate office between competing claims based on such criteria. As a result, within related diversified firms, the corporate office needs to rely on subjective and objective noniinancial criteria when evaluating divisional performance and allocating cash flows if it is to overcome performance ambiguities arising from interdependent divisions and realize maximum economic benefits. For example, Gupta and Govindarajan (1986) found that the greater the degree of resource sharing between divisions, the greater the reliance on subjective criteria when assessing the performance of divisional managers. Thus: Hypothesis 3. The interaction between related diversification and corporate office use of subjective and objective nonfinancial criteria when evaluating divisional performance will be positively related to firm performance. Finally, coordination may be enhanced if reward and incentive schemes emphasize interdivisional cooperation rather than the performance of each division as an independent unit (Gupta and Govindarajan 1986; Kerr 1985; Lorsch and Allen 1973; Pitts 1974; Salter 1973). This can be achieved if profit bonus schemes for divisional managers within related diversified firms are linked to corporate rather than divisional profitability. Since corporate profitability within a related diversified firm depends upon the success of interdivisional cooperation, such reward schemes provide divisional managers with an incentive to cooperate. It follows that in related diversified firms there is a need to emphasize incentive schemes based on corporate profitability if the economic benefits associated with the strategy are to be realized. Thus: Hypothesis 4. The interaction between related diversification and incentive schemes based on corporate profitability will have a positive effect on firm performance. Competitive Organizations As with related firms, the literature suggests that unrelated firms require an M-form structure (Chandler 1962; Rumelt 1974). However, the internal organizational arrangements of unrelated firms are different from those found in related firms (Hill and Hoskisson 1987; Mintzberg 1983; Pitts 1980; Vancil 1980). Jones and Hill (1988), Hill (1988), and Williamson (1975) have proposed that for an unrelated firm to realize governance economies, a number of organizational features must be present.3 First, each division must have autonomy with regard to operating 'Empirical work has confirmed the presence of several of these features in high performing M-form conglomerates—see Armour and Teece (1978), Chang and Choi (1988), Hill (1988), Steer and Cable (1978) and Teece (1981).

Structures in Diversified Firms 506

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON

decisions so that divisional managers can be held accountable for divisional profit performance (operating and strategic decisions should be decentralized to divisions). Second, to preserve autonomy the relationship between the corporate office and operating divisions should be arms length; the corporate office should not intervene in divisional affairs except to audit operations, discipline opportunistic or incompetent divisional managers, and correct performance shortfalls. Third, the corporate office should exercise control over divisions by setting rate of return targets and monitoring outcomes. That is, the corporate office should rely on objective financial criteria to measure divisional performance. Fourth, incentive systems for divisional managers should be linked to divisional returns. Fifth, cash flows should be allocated between divisions by the corporate office to high yield uses on a competitive basis, rather than returned to source divisions. Within the unrelated firm, these features replicate the relationship that exists between investors and freestanding firms in the stock market, while overcoming the limitations associated with the stock market (Jones and Hill 1988; Hill 1988; Williamson 1985). Each division is evaluated on the basis of its rate of return as a self-contained unit. Divisional managers are responsible for all relevant operating and strategic decisions. Consequently, they are accountable for the performance of the divisions under their control. As a result, they have a stronger incentive to maximize the efficiency of divisional operations than would be the case if accountability was weaker (as in cooperative organizations where decisions are shared with the corporate office and other divisions). Also, it is clear to divisional managers that promotional opportunities and job tenure are contingent on the attainment of specific rate of return targets. Gearing bonus pay to divisional returns, and allocating capital between divisions on the basis of relative yields, reinforces the incentive to maximize divisional performance. The system is predicted to produce competition among divisions for capital (Williamson 1975). Divisional managers may also be compared on the basis of their ability to achieve rate of return targets for their respective divisions. Thus, internal promotion opportunities may be determined by competitive criteria. Hence, the internal ethos of such organizations is explicitly competitive rather than cooperative. A number of hypotheses are suggested by these arguments. It has been argued that, to realize governance economies, unrelated diversified firms need to decentralize the responsibility for operating and strategic decisions to divisions. This is required if divisional executives are to be held accountable for the profit performance of the operations under their charge. Thus: Hypothesis 5. The interaction between unrelated diversification and decentralization will be positively related to firm performance. Second, these arguments suggest that, within unrelated firms, interdivisional integration is unnecessary. Integration serves only to increase performance ambiguities, information processing requirements, and corporate bureaucracy without producing a commensurate increase in economic benefits. Thus: Hypothesis 6. The interaction between unrelated diversification and integration will be negatively related to firm performance. Third, it has been argued that rate of return criteria are appropriate for control purposes in unrelated diversified firms. Reliance on subjective and objective nonfinancial criteria will increase information processing requirements without producing any commensurate increase in economic benefits. Moreover, as noted by Williamson (1975), reliance on subjective criteria within competitive organizations can introduce

537

538

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson C O O P E R A T E VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

507

subjective biases into the control process. This suggests that reliance on such criteria may be dysfunctional in unrelated firms. Thus: Hypothesis 7. The interaction between unrelated diversification and corporate office use of subjective and objective nonfinancial criteria for evaluating divisional performance will be negatively related to firm performance. Finally, unlike related diversified firms, within unrelated diversified firms each division is evaluated as an autonomous independent unit. Accordingly, incentive schemes should be linked to divisional as opposed to overall corporate performance. Thus: Hypothesis 8. The interaction between unrelated diversification and incentive schemes based on divisional profitability will be positively related to firm performance. Organizational Incompatibilities An assumption underlying the above argument is that it is difficult for diversified firms to simultaneously realize economic benefits from economies of scope and governance economies. The lack of commonalities between divisions in unrelated firms precludes economies of scope. Related firms find it difficult to simultaneously realize benefits from economies of scope and governance economies because of performance ambiguities that arise from substantial interdependencies between divisions. Earlier we argued that because of performance ambiguities, to control the divisions of a related firm, corporate executives must process more information and undertake more subjective performance evaluations than their counterparts in unrelated firms. As argued by Williamson (1975), subjective evaluations increase the probability that the biases of corporate executives will enter the performance evaluation process. As a consequence, compared to unrelated firms, there is a greater tendency in related firms for rewards to be determined and capital funds allocated by a political bargaining process, rather than by reference to objective financial criteria. Since, by definition, the realization of governance economies requires that rewards be determined and resources allocated on an arms length basis according to objective financial criteria, this makes it difficult for the related diversified firm to realize governance economies. Thus, a fundamental tra'de-off is involved here. While related diversified firms organized on a cooperative basis can realize economies of scope, the resulting performance ambiguities increase the probability that subjective biases will enter the control and resource allocation processes. Hence, it becomes more difficult for the firm to realize governance economies. While this partly explains the incompatibility between the two structures other differences are also evident. Competitive and cooperative organizations have different internal configurations with regard to centralization, integration, control practices, and incentive schemes. As a consequence, the internal management philosophies of cooperative and competitive organizations are incompatible. In cooperative organizations, cooperation between divisions is fostered and encouraged. In competitive organizations, competition between divisions is fostered and encouraged. It is exceedingly difficult to simultaneously encourage competition and cooperation between divisions. The notion that different structures and control systems are needed to implement related and unrelated diversification strategies is supported by the work of several authors. For example, Mintzberg (1983) made a similar distinction between the internal structure of "related-product" and "conglomerate" firms. Additionally, Pitts' (1977, 1980) distinction between the internal structure and incentive systems of internal diversifiers (largely related firms) and acquisitive diversifiers (largely unrelated firms) echoes similar themes. The distinction made by Lorsch and Allen (1973)

Structures in Diversified Firms 508

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON

between conglomerates and vertically integrated firms is also similar in many respects to that discussed here for related and unrelated firms. Vancil (1980) also found differences in decentralization and control practices among firms pursuing different diversification strategies. Moreover, several authors have observed that the appropriate incentive systems for divisional managers are a function of the degree of resource sharing (relatedness) between divisions (Gupta and Govindarajan 1986; Kerr 1985; Lorsch and Allen 1973; Pitts 1974; Salter 1973). In addition, recent empirical studies by Eccles (1985), Goold and Campbell (1987), and Hill (1988) all echo the incompatibility described above. Finally, note that we are not arguing that firms cannot mix strategies. It is possible for a firm to mix a related and unrelated diversification strategy (although our data suggest that only a minority of firms pursue such a hybrid strategy; see below). We are arguing that a firm finds it difficult to mix the structures required to implement each effectively. A firm that pursues both related and unrelated diversification equally may have to choose whether to implement a competitive or a cooperative structure. Clearly, this choice has implications for the value that a firm can create from its corporate strategy. If it chooses a competitive structure it is difficult to create value from economies of scope, while if it chooses a cooperative structure it may be unable to create value from governance economies. Thus, such hybrid firms face a conundrum. Of course, one can argue that for the firm which pursues both strategies the optimal choice of structure must be based upon a comparison of the relative economic benefits and bureaucratic costs of trying to realize governance economies and economies of scope respectively. Identifying the relative benefits and costs with precision, however, is clearly no easy task. As we discuss shortly, this has implications for our empirical approach. Methods Analytical Strategy The above hypotheses suggest that the functional form of the relationship between diversification strategy and firm performance is moderated by certain organizational arrangements and control systems. Following Stone and Hollenbeck (1984, 1989) and Venkatraman (1989), we use conventional moderated regression analysis to test these hypotheses. This involves utilizing a regression model of the general form: Y = a + />, · X + b2 • Ζ + ¿»J · W + f>4 · XZ + e where Y measures firm performance, X measures related or unrelated diversification, Ζ is one of the organizational characteristics discussed earlier, and W is a vector of other independent variables that might affect Y. Our hypotheses suggest that the impact of strategy (Λ') on performance (Y) is moderated by the effects of structure (Z). Our primary interest is in the interaction term XZ. A moderation hypothesis is supported if the unstandardized coefficient b4 differs significantly from zero and the sign is consistent with our hypothesis (Venkatraman 1989). Organizational Data Collection Testing the hypotheses required the collection of data on the internal organizational arrangements of diversified firms. Such data are not available from published sources and, therefore, were obtained through a survey. The survey was pilot tested using members of a regional branch of the Planning Forum (a nationwide association of corporate planning officers). The final version was mailed in 1986 to the CEOs of 780 of the largest United States firms that could be identified from the list of Fortune

539

540

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

509

1000 firms in 1985. The response rate for completed surveys was 24% (184 returns). These firms represented 15 separate 2-digit and 50 separate 4-digit industries. Possible nonresponse bias was examined by comparing the 1986 return on assets of the responding firms against the return on assets of all industrial firms listed by COMPUSTAT with 1986 sales of greater than $100 million (« = 1063). The mean return on assets for the sample firms was 4.11% against 3.7% mean return on assets registered for all COMPUSTAT firms. The difference was not statistically significant (t = 0.40). These results suggest that the responding firms did not differ materially from other large publicly traded U.S. firms. To obtain information regarding the survey's validity multiple respondents were sought from a subset of the sample firms. The degree of interrator reliability in the 20 firms with two independently completed questionnaires was encouraging. Significant agreement was observed between the independent evaluations of multiple respondents across the majority of survey items. For example, 80% of the respondents agreed on the integration complexity measure and 90% agreed on the basis for incentive systems. More generally, statistically significant positive correlations were observed between the independent evaluations of multiple respondents across the large majority of survey items. Organizational Variables To indicate a firm's incentive system, a dummy variable was calculated. The variable scored 1 if the firm based its incentive systems for divisional managers, at least in part, on corporate profitability, and 0 if incentives for divisional managers exclusively emphasized divisional profitability. The construction of the remaining organizational variables was more complex. Four main questions in the survey asked respondents to indicate their firm's policy with regard to; (a) decentralization (14 items), (b) control of divisions (13 items), (c) cash flow allocation between divisions (10 items), and (d) interdivisional integration (6 items). The complete questions are given in the Appendix.4 For each item a seven-point response scale was used. Underlying these questions was a desire to obtain three dimensions of internal organization; decentralization (the degree to which decisions are decentralized to divisions or centralized at the corporate office), integration (the degree of integration between divisions), and breadth of control (the extent to which the corporate office relied on comprehensive subjective and objective nonfinancial criteria when evaluating divisional performance and deciding how to allocate capital within the firm). To reduce the 43 items to these three dimensions for purposes of analysis we used a two-stage data reduction strategy. First, the responses to each of the four main questions were factor analyzed using a varimax rotation to identify response patterns (i.e. we did four factor analyses; one for each question). A loading of 0.4 or greater was used to assign an item to a particular factor. In two cases, a variable loaded greater than 0.4 on two factors. In these cases, we treated the variable as a member of the factor for which its loading was greatest. This exercise yielded 10 scales from the four analyses. Two of these scales summarized the degree to which operating and strategic decisions were decentralized to operating divisions. Two other scales summarized the degree of integration between divisions. Three scales summarized different aspects of the extent to which the head office used subjective and objective nonfinancial criteria when controlling divisions. The final three scales summarized different aspects of the extent to which the head office used subjective and objective nonfinancial criteria when deciding how to allocate cash flows between divisions. Cronbach's Alpha of the internal consistency 4

The Appendix is available upon request from the TIMS office in Providence, RI.

Structures in Diversified Firms 510

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON TABLE 1 Factor Analysis of Organizational Scales

Scale Decentralization (1) Decentralization (2) Integration (1) Integration (2) Control Criteria (1) Control Criteria (2) Control Criteria (3) Cash Allocation Criteria (1) Cash Allocation Criteria (2) Cash Allocation Criteria (3) Eigenvalue Percent of Variance

Factor 1

Factor 2

Factor 3

-0.056 0.002 0.164 0.117 0.569 0.058 0.815 0.605 0.685 0.858

0.283 0.020 0.651 0.578 0.354 0.795 -0.061 0.381 0.368 0.009

0.765 0.750 0.332 0.304 -0.156 -0.017 0.176 -0.271 -0.197 0.117

3.25 32.5

1.78 17.8

1.07 10.7

of each of these scales varied from a low of 0.61 to a high of 0.93. The constituent variables for each scale are given in the Appendix. The second stage of data reduction involved factor analyzing the 10 scales to delineate organizational dimensions that could be used in the analysis. The results are reported in Table 1. Factor 1 was labelled the breadth of control dimension. This factor loaded high on five of the six scales describing the extent to which the corporate office used subjective and objective nonfinancial criteria when assessing divisional performance and deciding how to allocate capital within the firm. Factor 2 was labelled the integration dimension. This factor loaded highly on both scales that summarized the importance of integration between divisions. It also loaded high on the remaining control scale. This was perhaps not surprising. As argued earlier, integration can give rise to performance ambiguity. Hence, the greater integration, the greater the tendency to utilize subjective and objective nonfinancial control criteria. Factor 3 loaded high on the two scales measuring decentralization of operating and strategic decisions and thus was labelled the decentralization dimension. We used standardized factor scores based on the factor score coefficient matrix for the analysis reported in Table 1 to summarize a firm's position on each of the three dimensions. A high negative score on the breadth of control dimension indicated that the corporate office relied heavily on subjective and objective nonfinancial criteria when assessing divisional performance and allocating cash flows. Thus, a high negative score indicated high information processing by the corporate office. A high negative score on the integration dimension indicated that a high importance was attached to interdivisional integration. A high negative score on the decentralization dimension indicates that the responsibility for operating and strategic decisions was relatively centralized. One problem with the measure of integration is that it tells us little about the complexity of integrating mechanisms between divisions (Galbraith 1977). Rather, it focuses on the importance that the CEO attaches to achieving integration between divisions. Although these two should be correlated, they are separate dimensions of integration. Accordingly, we used the survey data to construct a measure of the complexity of integrating mechanisms (integration complexity). We asked respondents whether they used (a) program or project managers, (b) temporary task forces or committees, (c) permanent teams or committees, and (d) formal group structures to achieve integration between divisions (Galbraith 1977). The responses were then

541

542

C. W. L. HilL/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

511

transformed into a 5-point scale where 1 signified that a firm used no integrating mechanisms, and a score of between 2 and 5 indicated the number of different integrating mechanisms used. Thus, if a firm used all four integrating mechanisms described above it received a score of 5.5 Diversification Strategy We based our measures of diversification strategy on the entropy measure of diversification (Jacquemin and Berry 1979; Palepu 1985). This index is being increasingly used in the strategy literature (e.g. Amit and Livnat 1988, Baysinger and Hoskisson 1989, Palepu 1985). The entropy measure of total diversification (DT) is defined as:

DT=

E/Vln(l/P,.), i-l

where Ρ is the share of the segment i and ln(l/P ( ) is the weight given to each segment (which is the logarithm of the inverse of its share). This measure takes into account the number of segments in which a firm operates and the relative importance of each segment in total firm revenues. The measure can be separated into unrelated and related components, such that DT = DR + DU where DT is total diversification and DR and DU are related and unrelated diversification. DR is defined as related diversification arising out of operating segments within an industry group. DU is defined as unrelated diversification between industry groups. The greater DR, the greater the extent of related diversification. The greater DU, the greater the extent of unrelated diversification. Standard Industrial Classification (SIC) codes were used to define industry segments and industry groups. Two-digit SIC industries were treated as industry groups. Four-digit SIC industries were treated as the industry segments. The use of SIC codes has been supported by Montgomery (1985) and Baysinger and Hoskisson (1989). Both of these studies found that SIC-based measures agreed closely with the more subjective measures developed by Rumelt (1974) and widely used in the literature. Data to calculate these measures were taken from Standard and Poor's COMPUSTAT business segment file for 1985 to 1987. We computed the three-year average indices of diversification. These variables could only be identified for 138 of the 184 firms in the sample. Thus, the subsequent performance analyses were conducted for 138 firms. For details of how the measure works, see Palepu (1985). We used DR and DU as our basic measures of related and unrelated diversification, respectively. Economic Benefits, Industry, and Firm Effects Since our arguments focus upon the efficiency effects of achieving a fit between strategy and structure, it is appropriate to measure economic benefits by the efficiency of resource use within a firm. This is best captured by a measure of profitability. Three measures of profitability were considered as possible candidates: return on assets (ROA), return on investment (ROI), and return on equity (ROE). ROE was ruled out because the value of this ratio is as much a function of the capital structure of the firm (i.e. the split between debt to equity) as it is of efficiency. Both ROA and 5 W e did not include this measure in the original factor analysis because it was based on a single-item response. The variables identified using the factor analysis, on the other hand, were based upon multiple items, so factor analysis was necessary to reduce those items to manageable proportions.

Structures in Diversified Firms 512

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON TABLE 2 Means and Standard Deviations

Variable

Mean

SD

Return on Assets (%) Industry Return on Assets (%) R & D Intensity (%) Advertising Intensity (%) Related Diversification Unrelated Diversification M-form

4.91 5.01 0.00 -0.03 0.59 0.79 0.78 0.73 -0.11 -0.07 0.02 3.18

4.01 2.38 2.09 2.21 0.40 0.50 0.47 0.45 0.98 0.96 0.95 1.14

Incentive Systems Control Dimension Integration Dimension Decentralization Dimension Integration Complexity

Range - 5 . 3 3 to - 5 . 5 8 to - 6 . 4 9 to - 9 . 8 3 to 0.00 to 0.00 to 0.00 to 0.00 to - 1 . 8 4 to - 2 . 7 2 to - 2 . 6 0 to 1.00 to

13.7 8.84 7.33 11.3 1.62 2.10 1.00 1.00 4.06 2.38 2.34 5.00

Note 1) R & D Intensity and Advertising Intensity adjusted for industry intensity.

ROI performed in a similar fashion and were highly correlated. We choose ROA, however, because it displayed greater year on year stability. We used average ROA for the 1986 to 1988 period inclusive as the dependent variable in our regressions. The three-year average smooths out annual fluctuations in the accounting data. The organization variables were measured in 1986, and diversification strategy is an average over 1985-87, thus, the performance data lag the organizational and strategic data. To control for industry effects, the average return on assets earned by all firms classified by COMPUSTAT into each firm's primary 2-digit industry over the 1986-88 time period was calculated. This variable was then used as a measure of primary industry profitability. The effects of firm level R & D and advertising expenditures on profitability were also controlled. The industrial organization literature suggests that R & D and advertising intensity have a positive effect on firm profitability (Hay and Morris 1979). Each firm's R & D intensity was measured by the ratio of R & D expenditure to sales. Using COMPUSTAT data, the ratio of each firm's average R & D expenditure to sales was computed for 1985 to 1987 inclusive. The figure was adjusted for industry R & D intensity by subtracting each firm's R & D to sales ratio from the average R & D to sales ratio reported by COMPUSTAT for all firms classified into that firms's primary 2-digit industry. Advertising intensity was measured by each firm's average advertising to sales ratio for 1985 to 1987. As with R & D intensity, this measure was then adjusted for primary 2-digit industry advertising intensity. Note that performance lags R & D and advertising intensity by one time period.6 Due to the adjustment for industry effects, we refer to these variables as relative R & D intensity and relative advertising intensity. A (0,1) dummy variable representing whether a firm had an multidivisional structure was used as an additional control. Williamson (1975) suggested that firms organized as multidivisionals outperform firms organized on a functional basis. In addition, the conceptual arguments presented earlier suggest that an Af-form structure is a basic pre-request for implementing a competitive or cooperative organization. These data were taken from the survey. The variable scored 1 if the firm had an

6 Due to incomplete reporting, R & D data were only available for 106 firms, and advertising data for 120 firms. We used mean substitution for missing data.

543

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

o ο 3 o I

—

d

—

o d

h ö

**

*

a ο

«Λ Ö

o

= 3 s

—

o

o o

S

S

« « 0.26"

**

2

ι

5

• . 0.02

o I

—

o ο—:

0.07

sο

5 * ί S

o

0.08

o

J S;

0.02

m β

— —

0.18"

o

0.10

544

513

Structures in Diversified Firms 514

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON TABLE 4 Regression Results : Base Regression (.Dependent Variable Return on Assets) Independent Variables

ROA

Industry Return on Assets Relative Advertising Intensity Relative R & D Intensity M-form Dummy Related Diversification Unrelated Diversification

0.63*** (6.21) 0.28** (2.21) 0.17 (1.17) 1.18** (2.17) -0.16 (0.21) -0.63 (1.10)

^•Statistic Adjusted A-squared

11.17*** 0.28

(1) *p < 0.1. **/> < 0.05. ***/> < 0.01. (2) 7-statistics in parentheses.

M-form structure. In total, 31 firms did not have an M-form structure. Of these, 12 firms were organized on a functional basis (all were undiversified). A further 19 firms were organized on a functional basis plus subsidiaries (halfway between functional and M-form structures). Some of these firms were, at least partly, diversified. Prior work suggests that these firms may have been inappropriately organized (Chandler 1962; Rumelt 1974). Means, standard deviations and ranges for all variables are reported in Table 2. An intercorrelation matrix is presented in Table 3. Results The base regression (main effects only without interaction effects or any of the organizational dimensions) is reported in Table 4. The equation was statistically significant (p < 0.001) and explained approximately 28% of the variance in ROA. It is of note that the coefficients for related and unrelated diversification were not statistically significant. This finding is consistent with our thesis that it is the interaction between strategy and structure that is important, rather than strategy alone. To test each hypothesis we first added the appropriate organizational dimension and then the interaction between diversification strategy and that organizational dimension to the equation reported in Table 4. To maintain simplicity, we do not report the full regressions here (that would involve the unnecessary and lengthy duplication of almost identical regression equations). Rather, in Table 5 we report the unstandardized coefficients for the various interaction terms, along with the change in R 2 following entry of the interaction term into the regression equation and the F-test for the significance of the change in R2 (which in this case is the same as the 7-statistic squared). Note that the R2 for each restricted model prior to entry of the interaction term was around 0.28. None of the organizational variables exhibited a statistically significant main effect with the exception of the breadth of control dimension. This variable was significant at the 10% level and had a negative

545

546

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson

COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

515

TABLE 5 Regression Results using Continuous Measures of Diversification ( Dependent Variable Return on Assets ) Interaction Term Related Diversification X Decentralization Dimension Related Diversification X Integration Dimension Related Diversification x Integration Complexity Related Diversification X Breadth of Control Dimension Related Diversification X Incentive Systems Unrelated Diversification X Decentralization Dimension Unrelated Diversification X Integration Dimension Unrelated Diversification X Integration Complexity Unrelated Diversification X Breadth of Control Dimension Unrelated Diversification X Incentive System

Coefficient

Change Ä 2

F-test

-1.45

0.018

4.22**

-0.96*

0.018

4.49**

0.16

0.000

0.08

-1.35

0.018

4.24**

3.35

0.031

7.71***

1.23

0.019

4.47**

0.52

0.004

0.85

0.36

0.003

0.78

0.87

0.017

3.88**

-1.79

0.016

3.88**

(1) *p < 0.10; **p < 0.05; ***p < 0.01.

sign, suggesting that, on average, greater reliance on subjective and objective nonfinancial controls had a beneficial effect on profitability. The statistically significant effect that the interaction between related diversification and the decentralization dimension had on ROA ( ρ < 0.05) provides support for Hypothesis 1. The negative sign suggests that the combination of related diversification and greater centralization was positively related to ROA (a high negative score on the decentralization dimension indicates greater centralization). The statistically significant effect that the interaction between related diversification and the integration dimension had on ROA ( ρ < 0.05) provides partial support for Hypothesis 2. The negative sign suggests that the combination of greater related diversification and greater emphasis on interdivisional integration was positively related to ROA (a high negative score on the integration dimension implied a high emphasis on interdivisional integration). However, the interaction between related diversification and integration complexity was not statistically significant. Thus, support for Hypothesis 2 is mixed. Support for Hypothesis 3 can be found in the statistically significant effect that the interaction between related diversification and the breadth of control dimension had on ROA ( ρ < 0.05). The negative sign suggests that the combination of related diversification and greater reliance on subjective and objective nonfinancial controls had a beneficial effect on profitability (a high negative score on the breadth of control dimension implied that the firm utilized subjective and objective nonfinancial criteria for evaluating divisional performance and reallocating cash flows). The statistically significant effect that the interaction between related diversification and incentive systems had on ROA ( ρ < 0.01) provides support for Hypothesis 4. The positive sign suggests that related diversified firms which included corporate performance in divisional profit bonus schemes had a higher ROA than those that did not.

Structures in Diversified Firms 516

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON

Support for Hypothesis 5 can be found in the statistically significant effect that the interaction between unrelated diversification and the decentralization dimension had on R O A ( ρ < 0.05). The positive sign suggests that the combination of unrelated diversification and greater decentralization was positively related to ROA. There was no support for Hypothesis 6 as shown in Table 5. Regardless of the integration measure, the interaction of unrelated diversification with a measure of integration did not produce the hypothesized negative effect on ROA. The statistically significant effect that the interaction between unrelated diversification and the breadth of control dimension had on R O A ( ρ < 0.05) provides support for Hypothesis 7. The positive sign suggests that the combination of greater unrelated diversification and less reliance on subjective and objective nonfinancial controls has a positive effect on performance. Finally, support for Hypothesis 8 is shown in the statistically significant effect that the interaction between unrelated diversification and incentive systems had on R O A ( ρ < 0.05). The negative sign suggests that the combination of unrelated diversification and incentive systems for divisional managers based solely upon divisional profitability had a positive effect on ROA.

Discussion The conceptual arguments articulated herein suggest that related diversification has the potential to realize economic benefits from economies of scope, while unrelated diversification has the potential to realize economic benefits from governance economies. Building on the work of Ackerman (1970), Berg (1969, 1973), Gupta and Govindarajan (1986), Hill and Hoskisson (1987), Kerr (1985), Lorsch and Allen (1973), Pitts (1977, 1980), and Vancil (1980), realizing economies of scope has been argued to require a cooperative organizational form and realizing governance economies a competitive organizational form. This perspective suggests that the combination of related diversification and cooperative organizational arrangements should be associated with greater economic performance, as should the combination of unrelated diversification and competitive organizational arrangements. The regression results provide support for this perspective. Of eight hypotheses, six were fully supported (at the 95% acceptance level or better) and one received partial support. However, although statistically significant, the strength of the observed interactions, in terms of their contribution to total variance explained, was not that great. Support for our perspective must be tempered by this fact. One explanation for why we didn't find still stronger support for the hypotheses is that our measures of related and unrelated diversification may have introduced some ambiguities into the results. Our analytical strategy implicitly assumes that most firms score high on one but not both dimensions of diversification. Thus, it is implicitly assumed that a firm which scores high on DR will not simultaneously score high on DU (and vice versa). However, if there are a large number of firms in our sample that score high on both DR and DU, the results may be more ambiguous. It is difficult to discern which structure such hybrid firms should adopt, a competitive or cooperative structure. The conceptual arguments articulated above really address the structures that pure related or pure unrelated firms should adopt. For firms that pursue both strategies, the optimal choice of structure may be based upon a comparison of the relative economic benefits and bureaucratic costs of trying to realize governance economies and economies of scope respectively. As we argued earlier, it would be difficult for firms to mix structures (although they may mix strategies). The optimal choice in such circumstances, however, is all but impossible to identify from published data alone.

547

548

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

517

TABLE 6 Regression Results using Categorical Measures of Diversification (Dependent Variable Return on Assets) Interaction Term Related Diversification X Decentralization Dimension Related Diversification x Integration Dimension Related Diversification x Integration Complexity Related Diversification χ Breadth of Control Dimension Related Diversification X Incentive Systems Unrelated Diversification X Decentralization Dimension Unrelated Diversification X Integration Dimension Unrelated Diversification x Integration Complexity Unrelated Diversification X Breadth of Control Dimension Unrelated Diversification X Incentive System

Coefficient

Change R2

-1.41

0.023

5.34"

-0.89

0.012

2.70*

0.16

0.020

4.62**

-1.34

0.016

3.92**

2.45

0.022

5.38**

2.32

0.040

9.73***

1.71

0.019

4.45**

-0.09

0.000

0.07

0.88

0.003

1.69

-2.61

0.026

6.45**

F- test

(1) *p < 0.10; **p < 0.05; ***p < 0.01.

To identify how many hybrid firms there were in our sample we plotted individual firm scores on DR and DU. No firm scores high on both DR and DU. However, a number of firms did have moderately high scores on both DR and DU. The presence of these firms in the sample may have partially biased our results. To determine whether this was the case we did a post-hoc analysis of the data. Our approach was to use the scores on DR and DU to separate firms into the following categories; (1) single and dominant businesses (firms that score low on DR and DU), (2) pure related firms (firms that score high on DR and low on DU), (3) pure unrelated firms (firms that score low on DR and high on DU), and (4) hybrid firms (firms that score moderate to high on both DR and DU). We used the mean values of DR and DU for the sample as the dividing points between high and low scores on a dimension. (This approach closely parallels that of Palepu 1985, and is similar to that employed by Baysinger and Hoskisson 1989 and Varadarajan and Ramanujam 1987.) The mean values were 0.59 for DR and 0.79 for DU. Using this method there were 21 single or dominant firms, 48 pure related diversified firms, 43 pure unrelated diversified firms, and 26 hybrid firms. A series of (0,1) dummy variables were then created to indicate whether a firm fell into a certain category. We then reanalyzed the data using the categorical measures (testing for interactions between the various organizational variables and the categorical measures of related, unrelated, and hybrid diversification). We observed no significant interaction effects for the hybrid firms. This may be because these organizations must make a choice between a cooperative and a competitive structure on a case by case basis. That is, we cannot generalize with regard to what is the best structure for a hybrid firm. However, we did observe a number of significant interaction effects for the pure related and pure unrelated firms. These are reported in Table 6. The pattern of results reported in Table 6 is broadly consistent with those discussed earlier. The most notable departure is that we now find evidence that the

Structures in Diversified Firms 518

CHARLES W. L. HILL, MICHAEL A. HITT AND ROBERT E. HOSKISSON

interaction between related diversification and integration complexity is statistically significant and positively related to ROA ( ρ < 0.05). This strengthens the mixed support for Hypothesis 2 noted earlier and further suggests that related diversified firms perform better when they adopt interdivisional integrating mechanisms. Furthermore, the interaction between the unrelated category and the integration dimension had a statistically significant effect on ROA ( ρ < 0.05). The positive sign suggests that unrelated firms that did not emphasize interdivisional integration performed better than those that did. This is consistent with the prediction of Hypothesis 5, the one hypothesis for which we found no support in the earlier analysis. Also, by focusing on pure related and pure unrelated firms we found stronger results for some, but not all, interactions. This was particularly true of those interactions involving the decentralization dimension. However, the amount of variance in ROA explained by the interactions, while generally stronger than in the previous analysis, was still moderate. In sum, this post-hoc analysis further supports our basic perspective. Were it not for a number of limitations of this study, we think that we may have found still stronger evidence in support of our perspective. Measuring organizational characteristics is a difficult and often complex process. The measures of organizational arrangements used herein were largely experimental. Although considerable care was taken in constructing and testing the measures, survey data are by its nature imperfect and may not fully capture all aspects of the phenomena under investigation. One concern is the tendency for people to lapse into their own pattern of responding to scaled items when answering survey questions. Obviously, such factors may have "blurred" the distinctions we were trying to examine. Also, our data reduction strategy, while necessary to reduce a large number of variables to a manageable number, may have compounded any noise in the data. Additionally, there are the normal problems associated with cross-sectional data. Cross-sectional data do not allow the researcher to make adjustments for the impact of unobservable factors on firm performance (Jacobson 1990). Unobservable factors, such as management quality, may account for much of the unexplained variance in ROA and, furthermore, may have been important moderators of the hypothesize fit relationships. Unfortunately, it is only possible to control for unobservable factors if the researcher has access to pooled cross-sectional time series data. Unobservable factors show up as serial correlation in such data sets and can be controlled for by appropriate autoregression techniques. However, the very nature of survey research obviously makes it extremely difficult to assemble such large sample, multi-time period data sets. Finally, our analytical strategy was reductionist. We assumed that it was possible to decompose elements of an organizational form into its constituent parts. As advocates of the systems approach to testing contingency relationships have noted, such an approach may not be able to detect effects of fit that are present at a holistic or gestalt level (Drazin and Van de Ven 1985). Unfortunately, problems associated with operationalizing the systems approach disallowed its application in the present study. Conclusion In conclusion, the research reported herein can be added to a growing body of work on the organizational arrangements needed to implement different diversification strategies (Ackerman 1970; Berg 1969, 1973; Govindarajan and Fisher 1990; Gupta and Govindarajan 1986; Hill and Hoskisson 1987; Hill 1988; Kerr 1985; Lorsch and Allen 1973; Pitts 1977, 1980; and Vancil 1980). Our research suggests that

549

550

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

519

generating economic value from different diversification strategies requires distinctly different internal organizational arrangements and control systems. Our results suggest that firms pursuing related diversification to realize economies of scope need to adopt cooperative organizational arrangements, while those pursuing unrelated diversification to realize governance economies need to adopt competitive arrangements. There are important messages for managers in these results. The conceptual arguments and empirical tests suggest that managers need to be sensitive to the organizational attributes necessary to implement diversification strategies. Different strategies are associated with different economic benefits. Unless managers adopt the appropriate set of organizational arrangements, a diversification strategy may reduce, or at least not improve, firm performance. The lack of any relationship between diversification and profitability found herein, and the contradictory results of so many studies, suggests that knowledge of appropriate linkages between organizational arrangements and diversification strategies may not yet be common among senior managers. However, the results herein suggest that senior managers may effectively implement diversification strategies that lead to higher performance if they adopt the appropriate control systems, distribute decision making authority in an appropriate manner, adopt mechanisms for integrating divisions when appropriate, use appropriate criteria for cash flow allocation, and base incentive systems on the type of diversification strategy being pursued. Therefore, the results of this study provide refinements to a stream of research examining the relationships between diversification strategy and structure. Acknowledgements This paper has benefited from the comments of many colleges including Brad Agel, Gray Hansen, Tom Thomas, and participants of research seminars hosted by the Management Departments of the University of Washington and the University of Alberta. An earlier version of this paper was presented at the 1988 Academy of Management meetings in California.

References Ackerman, R. W. (1970), "Influence of Integration and Diversity on the Investment Process," Administrative Science Quarterly, 15, 341-351. Allen, S. A. (1978), "Organizational Choices and General Management Influence Networks in Divisionalized Companies," Academy of Management Journal, 16, 341-365. Amit, R. and J. Livnat (1988), "Diversification Strategies, Business Cycles and Economic Performance," Strategic Management Journal, 9, 99-110. Armour, H. O. and D. J. Teece, (1978), "Organizational Structure and Economic Performance: A Test of the Multidivisional Hypothesis," Bell Journal of Economics, 9, 106-122. Baysinger, B. and R. E. Hoskisson (1989), " R & D Intensity in Corrupted Multidivisional M-Form Firms," Academy of Management Journal, 32, 310-332. Berg, N. A. (1969), "What's Different about Conglomerate Management? Harvard Business Review (November-December), 112-120. (1973), "Corporate Role in Diversified Companies," in Taylor, B. and MacMillan, K. (Eds.), Business Policy. Teaching and Research, New York: Wiley. Bettis, R. A. (1981), "Performance Differences in Related and Unrelated Diversified Firms," Strategic Management Journal, 2, 379-394. Black, F. (1986), "Noise," Journal of Finance, 41, 529-543. Boulding, Κ. E. (1964), " A Pure Theory of Conflict Applied to Organizations," In Fisk, G. (Ed.), The Frontiers of Management Psychology, New York: Harper & Row. Bower, J. L. (1970), "Planning within the Firm," American Economic Review, 59, 186-194. Chandler, A. D. (1962), Strategy and Structure·. Chapters in the History of the Industrial Enterprise, Cambridge, MA: MIT Press. Chang, S. J. and U. Choi (1988), "Strategy, Structure and Performance of Korean Business Groups," The Journal of Industrial Economics, 37, 141-158.

Structures in Diversified Firms

520

CHARLES W. L. HILL, MICHAEL A. H I T T A N D R O B E R T E. HOSKISSON

Child, J. (1984), Organization: A Guide ίο Problems and Practice (2nd ed.), London: Harper & Row. Christensen, Η. Κ. and C. A. Montgomery (1981), "Corporate Economic Performance: Diversification Strategy versus Market Structure," Strategic Management Journal, 2, 327-343. Daft, R. L. and R. H. Langel (1986), "Organization Information Requirements, Media Richness and Structural Design," Management Science, 32, 554-571. Drazin, R. and A. H. Van de Ven (1985), "Alternate Forms of Fit in Contingency Theory," Administrative Science Quarterly, 30, 514-539. Dundas, Κ. Ν. and P. R. Richardson (1980), "Corporate Strategy and the Concept of Market Failure," Strategic Management Journal, 1, 177-188. and (1982), "Implementing the Unrelated Product Strategy," Strategy Management Journal, 3, 287-301. Eccles, R. G. (1985), The Transfer Pricing Problem: A Theory for Practice, Lexington, MA: Lexington Books. Galbraith, J. R. (1977), Organizational Design, Reading, MA: Addison-Wesley. Gooid, M. and A. Campbell (1987), Strategies and Style, London: Blackwell. Govindarajan, V. and J. Fisher (1990), "Strategy, Control Systems, and Resource Sharing: Effects on Business-Unit Performance," Academy of Management Journal, 33, 259-285. Grant, R. M. (1988), "On Dominant Logic, Relatedness, and the Link between Diversity and Performance," Strategic Management Journal, 9, 639-642. Gupta, A. K. and V. Govindarajan (1986), "Resource Sharing among SBUs: Strategic Antecedents and Administrative Implications," Academy of Management Journal, 29, 695-714. Hay, D. A. and D. J. MorTis (1979), Industrial Economics: Theory and Evidence, Oxford: Oxford University Press. Hill, C. W. L. (1988), "Internal Capital Market Controls and Financial Performance in Multidivisional Firms," The Journal of Industrial Economics, 37, 67-83. and J. F. Pickering (1986), "Divisionalization, Decentralization, and Performance of Large United Kingdom Companies," Journal of Management Studies, 23, 26-50. and R. E. Hoskisson (1987), "Strategy and Structure in the Multiproduct Firm," Academy of Management Review, 12, 331-341. Jacobson, R. J. (1990), "Unobservable Effects and Business Performance," Marketing Science, 9, 74-85. Jacquemin, A. P. and C. H. Beny (1979), "Entropy Measure of Diversification and Corporate Growth," The Journal of Industrial Economics, 27, 359-369. Jones, G. R. and C. W. L. Hill (1988), "Transaction Cost Analysis of Strategy-Structure Choice," Strategic Management Journal, 9, 159-172. Keats, B. W. and M. A. Hitt (1988), " A Causal Model of Linkages Among Environmental Dimensions, Macro Organizational Characteristics, and Performance," Academy of Management Journal, 31, 570-598. KerT, J. L. (1985), "Diversification Strategies and Managerial Rewards: An Empirical Study," Academy of Management Journal, 28, 155-179. Lawrence, P. R. and J. W. Lorsch (1967), Organization and Environment: Managing Differentiation and Integration. Boston, MA: Harvard Business School Press. Lorsch, J. W. and S. A. Allen (1973), Managing Diversity and Inter Dependence: An Organizational Study of Multidivisional Firms. Boston, MA: Harvard University. Luke, R. D., J. W. Begun and D. D. Pointer (1989), "Quasi Firms: Strategic Interorganizational Forms in the Health Care Industry," Academy of Management Review, 14, 9-19. Mintzberg, H. (1983), Structure in Fives: Designing Effective Organizations, Englewood Cliffs, NJ: PrenticeHall. Montgomery, C. A. (1985), "Product Market Diversification and Market Power," Academy of Management Journal, 28, 789-798. Mueller, D. C. (1989), "Mergers: Causes, Effects, Policies," International Journal of Industrial Organization, 7, 1-10. Ouchi, W. G. (1980), "Markets, Bureaucracies, and Clans," Administrative Science (Quarterly, 25, 129-141. Palepu, K. (1985), "Diversification Strategy, Profit Performance, and the Entropy Measure of Diversification," Strategic Management Journal, 6, 239-255. Pitts, R. A. (1974), "Incentive Compensation and Organization Design," Personnel Journal, 53, 338-344. (1977), "Strategies and Structures for Diversification," Academy of Management Journal, 20, 197-208. (1980), "Towards a Contingency Theory of Multibusiness Organizational Design," Academy of Management Review, 5, 203-210. Porter, Μ. E. (1985), Competitive Advantage, New York: Free Press. (1987), "From Competitive Advantage to Corporate Strategy," Harvard Business Review, (May-June), 43-59.

551

552

C. W. L. Hill/M. A. Hitt/R. E. Hoskisson COOPERATIVE VS COMPETITIVE STRUCTURES IN DIVERSIFIED FIRMS

521

Ramanujam, V. and P. Varadarajan (1989), "Research on Corporate Diversification: A Synthesis," Strategic Management Journal, 10, 523-551. Rumelt, R. P. (1974), Strategy, Structure, and Economic Performance, Boston, MA: Harvard Business School Press. (1982), "Diversification Strategy and Profitability," Strategic Management Journal, 3, 359-369. Salter, M. S. (1973), "Tailor Incentive Compensation to Strategy," Harvard Business Review, 51, 3, 94-102. Singh, A. J. (1971), Takeovers: Their Relevance to the Stockmarket and the Theory of the Firm, Cambridge, UK: Cambridge University Press. Sloan, A. P. (1963), My Years with General Motors, New York: Doubleday & Co. Steer, P. O. and J. Cable (1978), "Internal Organization and Profit: An Empirical Analysis of Large U.K. Companies," Journal of Industrial Economics, 27, 13-29. Stone, E. F. and J. R. Hollenbeck (1984), "Some Issues Associated with the Use of Moderated Regression," Organizational Behavior and Human Performance, 34, 195-213. and (1989), "Clarifying Some Controversial Issues Surrounding Statistical Procedures for Detecting Moderator Variables: Empirical Evidence and Related Matters," Journal of Applied Psychology, 74, 3-10. Teece, D. J. (1981), "Internal Organization and Economic Performance: An Empirical Analysis of the Profitability of Principal Firms, Journal of Industrial Economics, 30, 173-199. (1982), "Towards an Economic Theory of the Multiproduct Firm" Journal of Economic Behavior and Organization, 3, 39-63. Vancil, R. F. (1980), Decentralization: Managerial Ambiguity by Design, Homewood, IL: Dow-Jones Irwin. Varadarajan, P. R. and V. Ramanujam (1987), "Diversification and Performance: A Re-examination Using a New Two-Dimensional Conceptualization of Diversity in Firms," Academy of Management Journal, 30, 380-393. Venkatraman, N. (1989), "The Concept of Fit in Strategy Research: Towards a Verbal and Statistical Correspondence," Academy of Management Review, 14, 423-444. Wernerfeit, B. and C. A. Montgomery (1988), "Tobin's Q and the Importance of Focus in Firm Performance," American Economic Review, 78, 246-250. Williamson, O. E. (1975), Markets and Hierarchies: Analysis and Antitrust Implications, New York: Free Press. (1985), The Economic Institutions of Capitalism, New York: Free Press.

Champions of Technological Innovation Champions of Technological Innovation Jane M. Howell Christopher A. Higgins The University of Western Ontario

This study investigated the personality characteristics, leadership behaviors, and influence tactics of champions of technological innovations. Analyses of questionnaires and interview transcripts of twenty-five matched pairs of champions and nonchampions revealed that champions reported using transformational leader behaviors to a significantly greater extent than did nonchampions. Champions exhibited higher risk taking and innovativeness, initiated more influence attempts, and used a greater variety of influence tactics than nonchampions. Regression analysis of a model of champion emergence, relating personality characteristics, transformational leader behaviors, and influence tactics, showed that champions were significantly higher than nonchampions on all paths in the model* The increased turbulence, complexity, and competitiveness of organizational environments have made the identification, evaluation, and adoption of technological innovations a critical determinant of organizational productivity, competition, and survival (Zaltman, Duncan, and Holbeck, 1973; Bigoness and Perreault, 1981; Morgan, 1988). As a result, a major research effort has focused on variables that facilitate or hinder the adoption of technological innovations (e.g., Rogers and Shoemaker, 1971; Kelly and Kranzberg, 1978; Kimberly and Evanisko, 1981; Pennings and Buitendam, 1987). One variable that has been strongly linked to the success of technological innovations is the presence of a champion. This is an individual who informally emerges in an organization (Schön, 1963; Tushman and Nadler, 1986) and makes "a decisive contribution to the innovation by actively and enthusiastically promoting its progress through the critical [organizational] stages" (Achilladelis, Jervis, and Robertson, 1971: 14). Twenty-six years ago, in a seminal article on radical military innovations, Schön (1963) identified the role of a champion. He contended that in order to overcome the indifference and resistance that major technological change provokes, a champion is required to identify the idea as his or her own, to promote the idea actively and vigorously through informal networks, and to risk his or her position and prestige to ensure the innovation's success. According to Schön (1963: 84), " t h e new idea either finds a champion or dies."

® 1990 by Cornell University. 0001 -8392/90/3502-0317/S1.00.

Funding for this study was supplied by Social Sciences and Humanities Research Council of Canada Grant 494-85-0018 and by the Plan for Excellence, The University of Western Ontario. The authors are indebted to Hugh Arnold. Bruce Avolio, Bernard Bass, Deborah Compeau, Martin Evans, Peter Frost. David Waldman, and three anonymous reviewers for their insightful comments. This paper was originally submitted for the ASQ special issue on Technology, Organizations, and Innovation (March 19901. Editorial work was begun by Richard R. Nelson and Michael N. Tushman and was completed by John H. Freeman.

553

A multitude of field and case studies have found strong support for Schön's contention that innovation success is closely linked with the presence of a champion (e.g., Roberts, 1968; Achilladelis, Jervis, and Robertson, 1971 ; Rothwell et al., 1974; Burgelman, 1983; Ettlie, Bridges, and O'Keefe, 1984). Yet, despite the important contribution attributed to champions in the innovation process, rigorous empirical investigation of these individuals is lacking. Prior studies examining champions are plagued by several conceptual and methodological problems, thereby casting doubt on the validity and interpretability of their findings. Four problems with previous research on champions are evident. First, prior research has paid little attention to the systematic measurement of individual attributes of champions such as personality and leadership. Most of what is reported about champions is anecdotal, reflecting the researcher's im317/Administrative Science Quarterly, 35 (1990): 317-341

554

J. M. Howell/C. A. Higgins pressions, rather than reliable and valid measurement using well-accepted instruments (e.g., Schön, 1963; Fernelius and Waldo, 1980; Delbecq and Mills, 1985; Dean, 1987). A second, related issue is that comparison groups for champions are not identified in any study. Therefore, it is unclear to what extent champions actually differ from the population of managers in general. A third problem is that previous studies suffer from methodological flaws in the identification of champions, which makes their results questionable. To illustrate, Table 1 summarizes the methods used to identify champions in fifteen prior studies. Most authors do not even discuss how champions are identified. Other authors use individual responses, uncorroborated by others, to identify the project champion. This latter method, where single individuals are polled as to the presence or absence of champions, is problematic, since bias may be introduced due to the tendency to report oneself as the champion, a socially desirable label. A final problem in the reliable identification of champions is the lack of specification of the various roles played by individuals in the innovation process. Several authors have identified a number of different roles associated with innovation (e.g., Roberts, 1968; Achilladelis, Jervis, and Robertson, 1971; Maidique, 1980; Curley and Gremillion, 1983; Katz and Tushman, 1983). For example, gatekeepers acquire, translate, and distribute external technological knowledge and advancements to their colleagues (Allen, 1977; Tushman and Scanlan, 1981 ; Katz and Tushman, 1983). Project champions, the focus of the present study, distill creative ideas from information sources and then enthusiastically promote them within the organization (Achilladelis, Jervis, and Robertson, 1971). Business innovators provide support, access to resources, and protection from organizational interference as innovations emerge (Achilladelis, Jervis, and Robertson, 1971). Technical innovators design and/or develop the innovation, while user champions implement the innovation by training and providing assistance to the users (Achilladelis, Jervis, and Robertson, 1971; Rothwell et al., 1974; Curley and Gremillion, 1983). In order to identify project champions reliably, different types of innovator roles need to be distinguished. To illustrate, while both project champions and gatekeepers are involved in communication and information-processing activities, gatekeepers gather and disseminate external information to project groups while champions seek out creative ideas from information sources and then enthusiastically sell them. Thus, many studies reportedly investigating champions may not be studying champions at all if they have been inappropriately identified. The conceptual and methodological problems of previous research call into question the adequacy of our current knowledge about champions. Accordingly, the purpose of the present study was to conduct a rigorous, empirical investigation of champions of technological innovations, focusing on three variables that influence the emergence of champions in organizations: personality characteristics, leadership behaviors, and influence tactics. Personality characteristics are the qualities that predispose individuals to engage in champion activities. Leadership and influence are also important. 318/ASQ, June 1990

Champions of Technological Innovation Champions of Innovation Table 1

Identification of Champions in Previous Studies Source Achilladelis, Jervis, and Robertson (1971)·

Organizations

Innovation type

Methodology

Identification of champions

58 firms in scientific instruments and chemical industries 6 divisions of a single firm 45 NASA innovations

Various product and process innovations

Case studies and comparative analysis

Not discussed

Various product innovations Various product innovations

Case studies

Not discussed

Case studies

Dean (1987)

5 manufacturing organizations

Interviews, both retrospective and real-time

Delbecq and Mills (1985)

Several hundred managers in high-technology firms and healthservices organizations 192 firms in meat. canning, and fish industries

Advanced manufacturing technology innovations Various

Investigators' judgment based on interviews. letters, sales brochures, and company memoranda Not discussed

Interviews, comparing successes and failures

Not discussed

Consumer Retort Pouch technology

Surveys, field interviews in 56 firms

78 member firms of the Industrial Research Institute 90 Research and Development managers

Unclear from paper

Case histories submitted by organizations

Single question: Is there a person in your firm w h o is currently advocating consumer retortable pouch technology? Not discussed

Not defined

Survey

Single electronics firm 9 organizations at different stages of development 1 division of an integrated electronics manufacturer 86 firms in the chemical and scientific instruments industries 25 military inventions 100 projects in 17 member firms of the Industrial Research Institute

New electronics product Mostly product innovations

Case history

Self-reports of the amount of time spent in "championing activities" Not discussed

Case studies

Not discussed

New business ventures

Case studies

Person who headed the new venture

Various product and process innovations

Case studies

Not discussed

Product innovations Product innovations

Case studies

Not discussed

Interviews, case studies

Not discussed

Burgelman (1983) Chakrabarti (1974)

Ettlie (1983); Ettlie, Bridges, and O'Keefe (1984)

Fernelius and Waldo (1980)

Frohman (1978)

Galbraith (1982) Maidique (1980)

Roberts (1968)

Rothwell et al. (1974)*

Schön (1963) Souder (1981)

* Two phases of the same investigation.

319/ASQ, June 1990

555

556

J. M. Howell/C. A. Higgins since innovation adoption is largely a process of influence (Burgelman, 1983; Dean, 1987), both with subordinates, as indicated by leadership behavior, and with peers and superiors, as indicated by influence tactics. While the capacity of champions to influence others has been widely discussed, there has been no attempt to integrate findings with the theoretical and empirical literature concerning social influence processes in organizations. Moreover, researchers have not systematically measured the frequency and variety of influence attempts used by champions or explored whether champions differ from other managers in the amount and quality of their influence. Although personality characteristics, leadership behaviors, and influence tactics are frequently discussed in the champion literature (e.g., Schön, 1963; Burgelman, 1983; Dean, 1987; Van de Ven, 1986), for the most part, results of different studies have been poorly integrated. In addition, little empirical evidence exists with respect to the contribution of these variables to champion emergence. What appears to be needed is the development and testing of a model that integrates the factors influencing how individuals emerge to assume the champion role. The relationships posited in the model used here were developed inductively from literature in the areas of entrepreneurship, leadership, and influence. We used the entrepreneurship literature to identify the important dimensions of the champion personality, since several authors have drawn parallels between champions (or intrapreneurs) and entrepreneurs (e.g., Collins and Moore, 1970; Maidique, 1980; Pinchot, 1985). Transformational leadership theory was examined, because champions serve as informal leaders, inspiring others with their vision of an innovation's potential (Van de Ven, 1986). Research on influence was reviewed, because champions act as influence agents to promote their ideas (Burgelman, 1983; Dean, 1987). BACKGROUND AND HYPOTHESES Leadership Behavior The literature on champions and innovation highlights the capacity of champions to inspire and enthuse others with their vision of the potential of an innovation, to persist in promoting their vision despite strong opposition, to show extraordinary confidence in themselves and their mission, and to gain the commitment of others to support the innovation (Maidique, 1980; Burgelman, 1983; Dean, 1987). In Schön's words (1963: 84), "it is characteristic of champions . . . that they identify with the idea as their own, and with its promotion as a cause, to a degree that goes far beyond the requirements of their job." These champion behaviors are similar to the qualities of transformational leaders, leaders who inspire their followers to transcend their own self-interests for a higher collective purpose (Burns, 1978). Drawing on Burns's (1978) analysis of political leadership, Bass (1985) developed a theory of transformational leadership in organizational settings. Based on both qualitative and quantitative procedures, he identified four transformational leadership factors: 320/ASQ, June 1990

Champions of Technological Innovation

557

Champions of Innovation charisma, inspiration, intellectual stimulation, and individualized consideration. 1 Charisma refers to the leader's ability to articulate a captivating vision, to inspire and encourage higher-order effort on the part of followers, and to instill respect, faith, loyalty, and trust in the leader. Inspiration involves the leader's use of emotional appeals, communication of vivid, persuasive images, and provision of examples to enhance followers' confidence and motivation to pursue elevated goals. Intellectual stimulation represents the leader's ability to suggest creative, novel ideas that challenge and refocus followers' conceptualization, comprehension, and discernment of the nature of problems and their solutions. Individualized consideration refers to the leader's developmental and individualistic orientation toward followers. The leader provides examples and assigns tasks to followers on an individual basis to help t h e m significantly alter their abilities and motivation. Several writers have conceptually linked transformational leadership to the innovation process (Oberg, 1972; House, 1977; Bass, 1985; Conger and Kanungo, 1987). Oberg (1972) discussed the change-agent function of the transformational leader. He contended that the transformational leader brings about radical change by espousing beliefs and values that are different from the established order. According to Bass (1985), w h o offered a consistent profile of the transformational leader as active innovator, transformational leaders use intellectual stimulation to enhance followers' capacities to think on their own, to develop new ideas, and to question the operating rules and systems that no longer serve the organization's purpose or mission. By using individualized consideration and intellectual stimulation, transformational leaders enhance followers' confidence and skills to devise and implement innovative responses to current problems facing their organization. Thus, in order to promote innovation in organizations, it is likely that champions exhibit transformational leadership behaviors: Hypothesis 1 : Champions will exhibit transformational leader behaviors, that is, charisma, inspiration, intellectual stimulation, and individualized consideration to a greater extent than nonchampions. Personality Characteristics

ι Bass's (1985) descriptive statements about transformational and transactional leadership w e r e derived from a survey of senior executives and from the leadership literature. These statements w e r e sorted by raters into transformational and transactional categories. Items that raters could reliably assign to each category w e r e retained and placed in a survey administered to senior army officers, w h o rated their superiors in terms of h o w frequently they exhibited each leadership characteristic. Principal components factor analysis of the final items resulted in the four transformational leadership factors.

The innovation literature describes champions as risk takers (Schön, 1963; Cox, 1976), socially independent (Cox, 1976), and politically astute (Schön, 1963; Burgelman, 1983; Dean, 1987). Moreover, champions are said to display persistence and dedication even in the face of frequent obstacles and imminent failures (Schön, 1963; Frohman, 1978). While these descriptors provide a useful starting point for understanding the champion's personality, they cannot be taken as conclusive, since they tend to be based on the researchers' impressions of the individuals rather than on scores obtained on validated personality tests. Furthermore, the observations are made in the absence of an explicit comparison group, and it is therefore unclear if champions actually differ from managers in general. While the personality characteristics of champions have not been systematically explored in the innovation literature, research focusing on the entrepreneurial personality offers 321/ASQ, June 1990

558

J. M . Howell/C. A . Higgins

some insights into potentially important dimensions. In a seminal study of the entrepreneurial personality, McClelland (1967), using Thematic Apperception Tests (TATs), discovered that entrepreneurs scored high on need for achievement. He also found that entrepreneurs desired to take personal responsibility for decisions, preferred decisions involving a moderate degree of risk, were interested in concrete knowledge of the results of decisions, and disliked routine work. Consistent results have been reported by other authors with respect to various personality dimensions of entrepreneurs. Roberts (1968) and Wainer and Rubin (1969), who investigated the relationship between successful research and development (R&D) entrepreneurs and company performance, found that the highest performing companies were led by entrepreneurs with high need for achievement and moderate need for power, as measured by projective TAT tests. Similarly, Hornaday and Aboud (1971), using objective personality measures, reported that entrepreneurs scored significantly higher on achievement and independence than the population in general. Based on interviews with 40 entrepreneurs and life-history analysis of 30 entrepreneurs, Kets de Vries (1977) observed that entrepreneurs are innovative, doing things not generally done in the ordinary course of business routine. Furthermore, he reported that they exhibit resilience in the face of setbacks and have the ability to start over again when disappointments occur. These findings are consistent with the general literature on innovators, which indicates that they are less conforming, value creativity, have more confidence in their abilities, and want opportunities to test their ideas (Stein, 1968; Maddi, 1976). Thus, w e hypothesize: Hypothesis 2: Champions will exhibit higher achievement, persistence, innovativeness, persuasiveness, and risk taking than nonchampions. The requisite personality characteristics of entrepreneurs are also common to transformational or charismatic leaders. From biographies, case studies, and theoretical statements a pattern of transformational personality characteristics can be discerned, including innovation, risk taking, achievement, persuasiveness, and endurance. Bass (1985: 176) contended that transformational leaders are high in social boldness, need for achievement, creativity, originality, and self-determination. Other theorists argue that transformational or charismatic leaders, through their passionate advocacy for change, create and take great personal risks, devise novel and unconventional plans for action, and engage in artful persuasion (Conger and Kanungo, 1987; Sashkin, 1988). From this, w e hypothesize: Hypothesis 3: There will be a more positive relationship between personality dimensions and transformational leader behaviors for champions than for nonchampions. Influence Tactics The use of influence by champions to promote their ideas has been documented in case studies of innovation (Schön, 1963, Burgelman, 1983; Dean, 1987). According to Schön (1963: 84), champions are "capable of using any and every means of informal sales and pressure in order to succeed." The results 322/ASQ, June 1990

Champions of Technological Innovation

559

Champions of Innovation of Burgelman's (1983) investigation of the internal corporate venturing process revealed that astute organizational champions actively influenced the dispositions of top managers toward a n e w corporate venture and made them see the strategic importance of it. Dean's (1987) study of the decision processes surrounding the adoption of advanced manufacturing technology demonstrated that champions relied on a variety of influence tactics, including rational justification, repeated informal expression of enthusiasm and confidence about the innovation, and sharing of information w i t h potential coalition members. The social influence literature contends that the decision to engage in influence is a function of several factors, including self-characteristics (e.g., risk-seeking propensity, persuasiveness, need for achievement), characteristics of the influence target (e.g., perceived relative power), and situational factors (e.g., reason for exercising influence) (Mowday, 1979; Porter, Allen, and Angle, 1981). W h e n he examined the personal characteristics of the influence agent, M o w d a y (1979) found that need for achievement and need for power are positively related to the propensity to engage in influence and to the methods of influence used in organizational decision situations (persuasive argument and manipulation). W e thus hypothesize: Hypothesis 4: There will be a more positive relationship between personality dimensions and influence tactics for champions than nonchampions. Theoretically, charismatic and transformational leaders are characterized as exercising "diffuse and intensive influence over the normative [ideological] orientations of other actors" (Etzioni, 1961: 203). Imbued w i t h self-confidence in their o w n capabilities, convinced in the Tightness of their beliefs and ideals, and strong in the need for power, these leaders are highly motivated to influence their followers (House, 1977; Bass, 1985). By engaging in impression management to bolster their image of competence and success, by vividly portraying for followers an attractive vision of the outcomes of their efforts, and by relating the mission to strongly held values and ideals shared by members of the organization's culture, charismatic and transformational leaders mobilize their followers w i t h moral inspiration and purpose to accomplish great feats (House, 1977; Bass, 1985). Substantial empirical evidence collected in a w i d e array of settings w i t h diverse methodologies has confirmed that charismatic and transformational leaders have a profound impact on follower performance, satisfaction, and effectiveness (Yukl and Van Fleet, 1982; Avolio, Waldman, and Einstein, 1988; Hater and Bass, 1988; Howell and Frost, 1989). Thus, w e hypothesize: Hypothesis 5: Transformational leader behaviors will be more positively related to influence tactics for champions than nonchampions. Characteristics of the influence target, particularly his or her relative power, play a role in determining the choice of influence strategies (Mowday, 1979; Kipnis, Schmidt, and Wilkinson, 1980). Different influence strategies are needed w h e n the target is a superior, peer, or subordinate. Rational presentation of ideas and informal exchange (ingratiation) are associated w i t h upward influence, while assertiveness and 323/ASQ, June 1990

560

J. M . Howell/C. A. Higgins sanctions are typical of downward-influence tactics (Kipnis, Schmidt, and Wilkinson, 1980; Schilit and Locke, 1982). Finally, situational factors also determine the choice of influence tactic. Kipnis, Schmidt, and Wilkinson (1980), in their study of intraorganizational influence tactics, reported that individuals rely on different influence strategies when they are seeking personal assistance, assigning work, attempting to improve others' performance, obtaining benefits, or initiating change. In particular, they found that when individuals try to convince others to accept new ideas, they use a variety of influence tactics, including rationality, assertiveness, ingratiation, coalition, and exchange. In terms of the champion role, where multiple stakeholders must be convinced of the champion's vision, this literature suggests that frequent influence attempts and a variety of influence tactics will be required: H y p o t h e s i s 6: Champions will attempt to influence others more frequently and will use a wider variety of influence tactics than nonchampions.

A GENERAL MODEL OF CHAMPION EMERGENCE The above theoretical arguments and supporting empirical evidence were combined into a single model of champion emergence, shown in Figure 1. According to this model, champion emergence is a function of personality characteristics, transformational leadership behaviors, and frequency and variety of influence tactics. Particular personality characteristics predispose individuals to engage in transformational leadership behaviors and influence tactics directed toward instigating innovations within organizations. Specifically, personality characteristics, including achievement, innovativeness, persistence, persuasiveness, and risk taking are associated with individuals' propensities to display transformational leadership behavior and to engage in frequent and varied influence tactics. As specified in the entrepreneurial literature, personality characteristics may also directly affect champion emergence, independent of leadership behavior. Finally, transformational leadership is related to the frequency and variety of influence tactics. As Bass (1985) Figure 1. A general model of champion personality characteristics, transformational leadership, and influence tactics.

324/ASQ, June 1990

Champions of Technological Innovation Champions of Innovation

theorized, a fundamental quality of transformational leaders is their ability to influence persuasively the needs and orientations of others. The relationship between transformational leadership and influence tactics may be reciprocal. Although w e expected that transformational leadership behaviors would lead one to engage in many and varied influence tactics, the reverse relationship might prevail. It is conceivable that an individual who exerts maximum influence over other group members may be perceived as a leader. However, the exercise of influence does not automatically result in the attribution of transformational leadership. It could, for example, produce the attribution of transactional leadership in instances in which an individual exerts influence by controlling organizational resources. Thus, given the uncertainty about whether influence tactics lead to transformational leadership, this relationship was excluded from the model. The model was tested with a study of information-technology innovations in 25 large Canadian companies. A matched sample of champions and nonchampions were interviewed and completed a questionnaire measuring their personality characteristics and leadership behaviors. Content analysis of the interview transcripts was conducted for the presence of themes related to leader behavior and influence tactics. METHOD Description of Innovations For the purpose of this study, an innovation was defined as the adoption of a new product or process that reflected the application of information technology (Pennings, 1987). Information-technology innovations that required hands-on interaction with a computer-based system through a computer keyboard were examined in this study. Three criteria for selecting these innovations were established in light of the importance of clearly specifying the type of innovation to be sampled, in order to overcome the empirical instability and theoretical confusion arising from research on innovation in complex organizations (Downs and Möhr, 1976; Bigoness and Perreault, 1981). The first criterion was that the technological innovations had to have been designed for use by managers and/or professionals. At management levels, adoption of information technology is typically voluntary rather than mandatory. Consequently, champions may be needed to promote the introduction and implementation of these technologies. To ensure more accurate recall of the innovation process by participants, the second criterion was that the innovation had to have been implemented within the 18 months prior to the study. The third criterion was that the innovation had to have represented a significant financial investment to the company. This criterion ensured that the innovation was visible in the organization and had a potentially important impact on managerial work. Sample Eighty-eight organizations that had recently implemented a technological innovation were identified through a survey of 325/ASQ, June 1990

561

562

J. M . Howell/C. A . Higgins 350 chief executive officers (CEOs) of Canadian firms listed on the Financial Post 500, an annual enumeration of the largest Canadian companies (in terms of sales). Based on the survey responses, 56 innovations appeared to meet the three selection criteria for inclusion in the study. Preliminary interviews, either in person or by telephone, were conducted with executives from these companies. During these interviews, the investigators explained to the executives the background and general purpose of the study and obtained from them an in-depth description of the innovation. Based on these interviews, 28 information-technology innovations were identified that clearly met our established criteria. All 28 organizations agreed to participate in the study. Procedure Individuals who played a major role in the adoption of the innovation were identified by a two-stage process. First, a company executive identified the central individuals involved in the introduction and implementation of the innovation. Interviews with these key individuals often revealed others who played an important role in the innovation. These individuals were then also interviewed, creating a snowball sample (Rogers and Kincaid, 1981). For each innovation, an average of five key individuals were interviewed. In total, 153 interviews were conducted and tape recorded. On average, the interviews took one and one-half hours. The interviews with key individuals were conducted using a structured protocol. The first part of the interview, described below, was used to identify champions and nonchampions. The remainder of the interview focused on (a) the key individual's personal involvement in the innovation; (b) the initial receptiveness and commitment of targeted users of the innovation; (c) the methods of influence used by the individual to initiate and implement the innovation; (d) the resistance encountered and methods used by the individual to overcome it; (e) the motivation of the individual to participate in the innovation and the recognition received; (f) the risk, both personal and organizational, associated with the innovation; (g) the individual's perception of his or her effectiveness in the process; (h) the identification of factors contributing to the success or failure of the innovation; and (i) the career history of the individual. Transcripts of the interviews were contentanalyzed for leadership behaviors and influence tactics used by champions and nonchampions. After the interview, all key individuals were requested to complete a questionnaire and return it to the investigators. The questionnaire contained measures, described below, of personality characteristics and leadership behaviors. The response rate for this questionnaire was 88 percent. Identification of Champions and Nonchampions The first part of the interview identified champions and nonchampions through peer nomination, which has been shown to be a highly reliable and valid technique (Kane and Lawler, 1978; Love, 1981). Initially, via an open-ended question, the respondent was asked to identify the key people associated with the innovation and to describe the roles each of them played. The respondent was subsequently given a set of five role definitions derived from the innovation literature. Four of 326/ASQ, June 1990

Champions of Technological Innovation

563

Champions of Innovation

these roles—project champion, technical innovator, business innovator, chief executive—were drawn from Achilladelis, Jervis, and Robertson (1971). The fifth role, user champion, was derived from the information-systems literature (Curley and Gremillion, 1983). Each key individual then identified the person or persons who clearly fit each of these roles. The respondents were allowed to name anyone associated with the innovation and were told to leave it blank if no one fit the role. Several steps were taken to minimize attributional bias in identifying champions. First, respondents were informed that the purpose of the study was to examine factors influencing the introduction of technological innovations. Thus, the explicit purpose of the study was disguised. Second, the role descriptions derived from the innovation literature were not identified as representing different champion roles. Thus, respondents were given a description of activities associated with various roles, but the actual roles were not labelled as "project champion," "technical innovator," and so on. Third, key individuals were not told who had been nominated to the various innovation roles. The criterion for accurately identifying project champions was complete (100 percent) agreement among the key individuals as to the person who played a particular role in championing the project. Using this criterion, 25 project champions were identified. We dropped from the analysis three projects in which there was disagreement about who was the project champion. To further verify that the 25 identified project champions actually played this role, we examined key individuals' responses to the open-ended question regarding the roles played by people associated with the innovation. In all cases, there was complete agreement between these two methods of identifying champions. In order to test the hypotheses of the present study, we used four criteria to establish a comparison group, labelled "nonchampions." First, nonchampions and champions had to have been involved in the same innovation within the same organization. Pairing a champion and a nonchampion for each innovation therefore controlled for the cost, type, importance, success, or complexity of the innovations, company size, and industry. Moreover, since champions and nonchampions were involved in the same innovations, the fact that the innovations were judged to be relatively successful by the users should have caused both to receive the same "success bias," as reflected in the interviews and self-report measures. Second, both nonchampions and champions had to have played active, informal roles in the promotion of the innovation. However, in comparison to champions, nonchampions could not have been consensually identified as playing a specific role in the innovation (i.e., project champion, technical innovator, business innovator, chief executive, user champion) in the peer-nomination procedure. This criterion ensured that the champions and nonchampions were highly comparable in technical knowledge (since champions and nonchampions were not identified as the technical innovator) and position power (since champions and nonchampions were not identified as the business innovator, chief executive, or user champion). Third, champions and nonchampions had to be peers at the same organizational level. That is, there was no formal 327/ASQ, J u n e 1990

564

J. M. Howell/C. A. Higgins reporting relationship between them. Finally, champions and nonchampions had to have been self-appointed to the project. It was not within their job mandate to seek out and promote new technological innovations. Thus, differences between champions and nonchampions could not be attributed to role requirements of assigned positions, such as project leader. To confirm that champions and nonchampions were similar on specific demographic variables, paired t-tests were conducted. The assumption underlying a paired f-test is that the sample of differences is normally distributed. This assumption was tested using the nonparametric Kolmogorov-Smirnov test. The assumption held for both age and salary data. The results indicated that champions and nonchampions were not significantly different with regard to age [f(24) = .61 ; ρ > .05] and salary (f(24) = .62;,p > .05]. In addition, chi-squared tests of significance revealed that champions and nonchampions were not significantly different in job level [χ2(3) = 2.39; ρ = .49], functional area [χ2(3) = 3.84; ρ = .28] and educational level [χ2(1) = 1.09; ρ = .29]. Thus, any differences found between the two groups could not be attributed to differences in these demographic characteristics. The final sample consisted of 25 pairs of champions and nonchampions. The 50 participants, all of whom were male, had an average age of 45 years and were at executive (52 percent) or middle management (48 percent) levels across a variety of functional areas, including accounting, engineering, finance, general management, information systems, and marketing. Content Analysis The transcripts of the champion and nonchampion interviews were content-analyzed for the presence of themes relating to leadership behavior and influence tactics, using the same procedure as House, Woycke, and Fodor (1988). Descriptions of leadership behaviors were developed using charismatic and transformational leadership theories as a guide (Bass, 1985; Conger and Kanungo, 1987; House, Woycke, and Fodor, 1988). Descriptions of influence tactics were adopted from Kipnis and Schmidt (1982). Transcripts separate from those to be used in the study were selected to clarify the operationalization of the leadership behaviors and influence tactics. Three individuals independently coded these practice passages, discussed their ratings, and then clarified the descriptions where necessary. Five iterations of rating and discussion were completed before we felt that there was an unambiguous operationalization of each leadership behavior and influence tactic. The final coding of these practice passages became the key on which the coders were trained and eventually tested (House, Woycke, and Fodor, 1988). University students were recruited to code the interview transcripts. They were given a test to ensure that their reading comprehension was adequate for the material they would have to code. Of twelve students who took the reading test, seven passed. Ultimately, six of these seven students were hired to code the transcripts. They were randomly assigned to code either leader behaviors or influence tactics. Using precoded practice material and a description of leader 328/ASQ, June 1990

Champions of Technological Innovation Champions of Innovation arid influence behaviors, the students w e r e trained so that they understood and w e r e able to code the interviews accurately. The six students trained to code leader behaviors or influence tactics met the established criterion of 75 percent agreement w i t h the key. In total, the training required 12 hours. The identification of champions and nonchampions w a s disguised so that coders could not determine the identity of the people to which the transcripts applied. In addition, coders w e r e unaware of the hypotheses of the study. In total, over 1000 pages of transcripts w e r e analyzed. No significant differences in the lengths of champion and nonchampion interview transcripts w e r e found. Measures Personality characteristics. Five scales in the questionnaire assessed the personality characteristics of champions and nonchampions. Three scales from the Jackson Personality Inventory (Jackson, 1976) w e r e used: risk taking, innovation, and social adroitness. In addition, w e used t w o scales from the Personality Research Form E (Jackson, 1967): achievement and endurance. For these personality measures a truefalse response format was used. Extensive empirical evaluation of these measures revealed high internal consistency and test-retest reliabilities, minimal acquiescence and social desirability response biases, and adequate convergent and discriminant validity (Jackson, 1967, 1976, 1977). The number of items per scale, reliabilities (Cronbach's alpha), and the conceptual definition for each scale are as follows: risk taking (17 items, α = .84): enjoys taking chances; innovation (17 items, α = .84): develops novel solutions to problems; social adroitness (11 items, α = .63): persuades others to achieve a particular goal; achievement (12 items, α = .73): aspires to accomplish difficult tasks; and endurance (11 items, α = .68): perseveres even in the face of great difficulty. Leadership behavior. Leadership behavior was assessed in t w o ways. First, Bass's (1985) Multifactor Leadership Questionnaire (Form 5-Self) was administered to the key individuals. This 34-item questionnaire tapped the four leadership scales of relevance to the present study: charisma, inspiration, intellectual stimulation, and individualized consideration. Key individuals w e r e asked to judge h o w frequently they engaged in different leadership behaviors measured by the questionnaire. Each behavior was rated on a 5-point frequency scale ranging from 0 = not at all, to 4 = frequently, if not always. Supportive reliability and validity data for the Multifactor Leadership Questionnaire are presented by Bass and Avolio (1989). Specifically, based on samples of military and industrial leaders, three separate factor analyses of the questionnaire have produced the four factors that constitute the behaviors associated w i t h transformational leadership. The number of items per scale, reliabilities (Cronbach's alpha), and the conceptual definition for each scale are as follows: charisma (10 items, α = .74): expresses an exciting vision and communicates confidence in followers' abilities to meet high performance expectations; inspiration (9 items, α = .75): gets others to excel by, for example, giving pep talks; intel329/ASQ, June 1990

565

566

J. M . HowelL/C. A . Higgins lectual stimulation (9 items, α = .71): questions the tried and true ways and solves problems creatively; and individualized consideration (6 items, α = .70): provides encouragement, support, and developmental opportunities for followers. The second measure of leadership behavior involved content analysis of the interview transcripts to determine if, in comparison to nonchampions, champions engaged more frequently in behaviors that differentiated transformational from nontransformational leaders theoretically. Each passage of the interview was coded separately for the presence or absence of the individual's display of self-confidence; strong conviction about ideological goals; communication of individualized consideration for others (i.e., superiors, peers, subordinates); expression of high expectations for others; communication of confidence in others' ability to accomplish tasks; demonstration of unconventional, innovative, or countercultural behavior; and assessment of environmental resources and constraints for bringing about change (Bass, 1985; Conger and Kanungo, 1987; House, Woycke, and Fodor, 1988). Influence tactics. The frequency and variety of influence tactics used by champions and nonchampions were assessed using the same content-analysis technique described earlier. Passages describing strategies were coded based on Kipnis and Schmidt's (1982) description of eight different influence tactics, including building coalitions, appealing to higher authority, bargaining, acting in a clandestine manner, presenting rational arguments, applying sanctions, using friendliness and ingratiation, and being assertive. Variety of influence was calculated by determining how many of the eight different influence tactics were used. Frequency of influence was the total count of the number of influence attempts. Both variety and frequency of influence were measured in order to distinguish individuals who used only one influence tactic repeatedly from individuals w h o used a wide range of influence tactics with differing frequencies. Intercorrelations of Measures Pearson correlation coefficients among the measurement scales are presented in Table 2. Many correlations were significant (p < .05). The measures of personality characteristics

Table 2 Intercorrelations for Personality, Leadership, and Influence Measures Variable 1. 2. 3. 4. 5. 6. 7. 8. 9. 10.

Achievement Risk taking Endurance Innovativeness Social adroitness Charisma Inspiration Intellectual stimulation Individualized consideration Frequency of influence tactics 11. Variety of influence tactics

1

2

3

4

5

6

7

8

9

10

_ 26· .44.32.26· .05 .13 .08 .29*

-

.21 .37.38.29· .32.21 .13

.59.21 -.04 .15 .14 .42-

-.03

.39-

.08

.40-

-

.04 .02 .31.18 .44-

.07 .14 .01 .08

.60.55.34-

.5842-

.29*

.04

.22

.12

.27·

.37-

.33-

.25·

.01

.28·

.14

26·

.39-

.41-

.29·

• ρ < .05; - ρ < .01. 330/ASQ, June 1990

-

_ -

_ -

.85-

Champions of Technological Innovation

567

Champions of Innovation

were highly intercorrelated, as were the measures for transformational leadership and influence. The relationships between constructs showed that the measures of transformational leadership and influence were highly correlated, whereas the measures of personality characteristics (with the exception of innovation and risk taking) tended to be weakly correlated with the other measures. Data Analysis Hypotheses 1, 2, and 6 postulated differences between champions and nonchampions on several personality, leadership, and influence variables. To test these hypotheses, a sample of differences was calculated by subtracting nonchampion scores from champion scores. Thus, a positive score indicated that the champion in a matched pair scored higher than a nonchampion on a particular variable. For each hypothesis, Hotelling's one-sample multivariate T2 statistic was used. This test, which is the simplest form of MANOVA, allows for the simultaneous testing of the hypothesis that several population means do not differ from a specified set of constants. Prior to conducting the test, we checked each sample of differences for normality (the primary assumption underlying Hotelling's Γ2) using a Kolmogorov-Smirnov test. All samples of differences met the assumption of normality. Individual F-tests were then used to identify which of the variables significantly differentiated between champions and nonchampions. Since the probability of committing Type I errors increases as the number of statistical tests increases, the significance level used to evaluate the comparisons was divided evenly among the comparisons actually being made (Kirk. 1982). Hypotheses 3, 4, and 5 postulated that the strength of the relationships between personality characteristics, leadership behaviors, and influence tactics would be significantly greater for champions in comparison to nonchampions. Regression analysis was used to test these hypotheses. This analysis assumes that the residuals are normally distributed and that variance of the residuals is constant. For each of the three regressions, the normality of the residuals was checked using a Kolmogorov-Smirnov test. Constant variance was assessed by examining plots of the residuals. These assumptions were met. In the regression analysis, indices were formed on each of the three constructs. Conceptually, the four scales of transformational leadership were viewed as tapping the same higher-order construct (i.e., transformational leadership) and were combined into a single index. In addition, the five scales measuring personality characteristics theoretically represented the higher-order construct "champion personality." Accordingly, these scales were formed into a single index. An index was also computed for influence, given the high intercorrelation between frequency and variety of influence tactics. Indices were calculated using a procedure recommended by Hotelling (1933), in which weights are obtained from the unrotated factor matrix of a principal components factor analysis. The weights produced can be considered an indicator of each scale's contribution to the higher-order construct. For 331/ASQ, June 1990

568

J. M. Howell/C. A. Higgins example, to compute the index for transformational leadership the following formula was used: Transformational Leadership Index = A χ Wa + Β χ Wb + C χ Wc + D χ Wd. w h e r e A, B, C, and D represent the standardized scale scores for charisma, inspiration, intellectual stimulation, and individualized consideration, and Wa, Wb, Wc, and Wd are their respective weights. Indices for personality characteristics and influence tactics w e r e computed similarly. Once the indices w e r e computed, regression analysis w a s conducted on the three paths in the model of champion emergence. This analysis used a d u m m y variable coded to reflect the champion and nonchampion groups. A significant f-value on the d u m m y variable would indicate that champions and nonchampions are significantly different w i t h respect to the path being tested. T w o supplemental analyses w e r e also conducted. First, in order to understand more fully the interrelationships among constructs in the model of champion emergence, w e used Partial Least Squares (PLS) analysis. PLS simultaneously estimates relationships between measures and constructs, as well as paths b e t w e e n constructs. Accordingly, it provides additional information regarding the relationships between personality characteristics, leadership behaviors, and influence tactics of champions. PLS is an extremely powerful multivariate analysis technique that is ideal for testing structural models w i t h latent variables (see Wold, 1985, for a comprehensive description). It has its roots in regression, path analysis, and principal components factor analysis. Sample sizes can be small and assumptions of normality are not necessary (Fornell, 1982), making the technique ideal in the early stages of theory building and testing. Because PLS requires that the data be analyzed within a theoretical model (Fornell, 1982), it was appropriate to test the model of champion emergence only with the sample of champions. Since relationships in the model are not hypothesized to hold for nonchampions, PLS analysis would be inappropriate for that sample. The path coefficients from a PLS analysis are standardized regression coefficients. The loadings of the items on the constructs are factor loadings. Thus, the results can easily be interpreted by considering t h e m in the context of regression and factor analysis. The second supplemental analysis used discriminant analysis in order to determine if champions and nonchampions could be distinguished using our hypothesized personality characteristics and leadership behaviors as discriminating variables. Discriminant analysis requires that the discriminating variables be drawn from a population with a multivariate normal distribution and that the covariance matrices of the t w o groups not be significantly different (Klecka, 1980). For each discriminant analysis, Kolmogorov-Smirnov tests indicated that the normality criterion was satisfied. Box's M test confirmed that the covariance matrices w e r e equal. 332/ASQ, June 1990

Champions of Technological Innovation

569

Champions of Innovation

RESULTS Hypothesis 1 posited that champions would score higher than nonchampions on measures of transformational leadership. Two sources of data were used to analyze this hypothesis: the self-ratings of transformational leadership from the Multifactor Leadership Questionnaire and the frequency of transformational themes in the interview transcripts. Table 3 presents the results of the Hotelling's T2 test for the Multifactor Leadership Questionnaire. The Hotelling's Γ 2 statistic was significant, indicating that champions and nonchampions can be distinguished on the basis of transformational leadership behaviors. Univariate F-tests, also summarized in Table 3, indicate that inspiration and intellectual stimulation significantly differentiated between champions and nonchampions. Charisma (p < .02) just failed to meet the conservative significance levels (p < .0125) adopted for these tests. Individualized consideration did not significantly distinguish between champions and nonchampions (p > .05). Content analysis was used to determine the frequency of transformational behaviors exhibited by champions and nonchampions. The results of a paired f-test comparing differences across the two samples revealed that champions reported transformational leadership behaviors to a significantly greater extent than nonchampions [f(24) = 4.12; ρ < .001]. Consistent with the analysis of the questionnaire measures, this finding lends further support to Hypothesis 1. Hypothesis 2 stated that champions would display greater achievement, risk taking, persistence, innovativeness, and persuasiveness than nonchampions. To test this hypothesis, Hotelling's T2 statistic was calculated on the sample of differences between champion and nonchampion pairs. As shown in Table 3, there was a significant difference between champions and nonchampions for personality characteristics. To Table 3

Hotelling's Ρ Test for the Personality Characteristics, Leadership Behaviors, and Influence Tactics of Champions and Nonchampions Variable

Champion Mean

Nonchampion Mean

d.f.

F

Personality characteristics Risk taking .71 Achievement .91 .84 Innovativeness Social adroitness .50 Endurance .74 Hotelling's 7* = .72, F(5,20) = 2.88, ρ < .05

.53 .82 .75 .43 .72

1.24 1.24 1.24 1.24 1,24

13.13— 4.42· 339· 2.66 .22

Leadership behavior Inspiration 2.66 Intellectual stimulation 3.19 Charisma 3.05 Individualized consideration 3.09 2 Hotelling's T = .85, F(4,21) = 4.48, ρ < .01

2.19 2.92 2.78 2.89

1.24 1,24 1,24 1,24

18.34— 9.394.88· 2.23

Influence tactics Frequency 9.16 Variety 2.95 2 Hotelling's T = .78, F(2,23) = 8.99, ρ < .001

4.64 1.79

1,24 1.24

17.92— 15.29—

• p < .05; " p < .01; — ρ < .001. 333/ASQ, J u n e 1990

570

J. M. HowelL/C. A. Higgins determine which of the personality characteristics distinguished between champions and nonchampions, univariate F-tests were computed. Using a conservative p-value of .01 (.05 divided by 5, the number of tests), risk taking was the only statistically significant variable. Achievement and innovativeness approached significance. Social adroitness and endurance, however, did not significantly differentiate between champions and nonchampions (p > .05). Hypothesis 3 stated that there would be a more positive relationship between personality characteristics and leadership behaviors for champions than for nonchampions. Hypothesis 4 postulated that there would be a more positive relationship between personality characteristics and influence tactics for champions in comparison to nonchampions. Hypothesis 5 posited that there would be a more positive relationship between transformational leadership behaviors and influence tactics for champions than for nonchampions. For hypothesis 3, the dummy variable distinguishing between champions and nonchampions was significant [t(47) = 3.81 ; ρ < .001 ; R2 = .29]. Similarly, for hypothesis 4 [t(47) = 3.96; ρ < .001 ; R2 = .26], and hypothesis 5 [f(47) = 2.87; ρ < .01 ; fî2 = .29], the dummy variables were significant. These results lend support to hypotheses 3, 4, and 5. Hypothesis 6, which postulated that champions would initiate more influence attempts and would use a greater variety of influence tactics than would nonchampions, was supported. The Hotelling's T2 value reached statistical significance and the univariate F-tests showed that both frequency and variety of influence attempts were significantly different across the champion and nonchampion pairs (see Table 3). It should be noted that given the high intercorrelation between frequency and variety of influence tactics, the univariate F-tests are essentially testing the same variable. Supplemental Analyses The results of the PLS analysis are shown in Table 4. Factor loadings greater than .7 are generally considered meaningful, since this implies that the construct explains more than 50 percent of the variance in the measure. An examination of the factor loadings clearly indicates that two variables drive the personality construct within the theoretical context of our model. They are risk taking and innovativeness, with loadings of .68 and .75, respectively. Since constructs obtain their meaning from the theoretical relationships in which they are imbedded (Fornell, 1982), these results imply that risk taking and innovativeness are the predominant personality characteristics of champions. The transformational leadership construct is derived mainly from charisma, inspiration, and intellectual stimulation. Individualized consideration is related to the construct but has a much lower loading than the other three items. These results provide partial support for Bass's (1985) conceptualization of transformational leadership. Finally, influence is jointly, and almost equally, defined in terms of frequency and variety of attempts. As shown in Table 4, the path coefficient between personality and leadership behaviors was large and significant. Similarly, significant relationships were found between transformational leadership and influence and between personality character334/ASQ, June 1990

Champions of Technological Innovation

571

Champions of Innovation Table 4 Partial Least Squares Analysis of Champion-Emergence Model Path

Path coefficient

Personality Leadership Personality - » Influence Leadership - » Influence

.58"· .19" .27-

Construct

Factor loading

Personality characteristics Achievement Risk taking Endurance Innovativeness Social adroitness

-.33 .68 .01 .75 .01

Leadership behavior Charisma Inspiration Intellectual stimulation Individualized consideration

.80 .86 .75 .38

Influence tactics Frequency Variety

.84 .99

— ρ < .01;

ρ < .001.

istics a n d influence. T h e m o d e l w a s able t o explain 3 4 perc e n t o f t h e v a r i a t i o n in t r a n s f o r m a t i o n a l l e a d e r s h i p a n d 17 p e r c e n t o f t h e v a r i a n c e in i n f l u e n c e t a c t i c s . Table 5 presents t h e results of t h e discriminant analysis using personality and leadership variables. T h e discriminant function w a s significant |χ2(9) =

22.96; ρ
> > > > -α • o 2 i ".3 α Ν È1 Ν S ΐί lg 15 cd Ν II . 2 v » 'Ξ 'S "3 "a "3 3 3 'θ 'S 3 ce β tí C C Z í S DC α. e a •ή » » g S g s ? j i a 3 s a 3 3 ff £ O O U ü U υ α O rtNn^intONno)

Organizational Inertia and Momentum 1991

Amburgey

K e l l y and

s i Β J 2

& S

¿ I "e βi g· 3

JS u a o

e o

, , o to ο IV CM ο Η «* Τ T -H σ> οο có 1

α

•2S 3s o « o& (Λ υ

Τ3 S (β » oc

*

ο σ> CM 03 CO •Ψ in 00 CM

, , T-t σ> ο σ> ο m ο co •—

£

rH o CM CO o Τ o Ν to o Γ 1

ν 1(Â 1 cd S a

e e

ScS β

b

*

03 tv f in I

,

tv 03 03 f

o o 1· co evi

, ,

, ,

,

,

O ο ο Ο co σι ο [Ν t CM 03 ΙΟ ο in CM CM *-< to co rH 1*-« CM r-i I —• ' — '

, O m to rH ftv ·"»· tv CM •—-

«

o o o o

*

CM o O o O o o o 1

«

,

o o to to ο

σι co *-Η tv o o CM I

«

m co rH ^ Γ

co oo m o 1

_IO «o

, ,

*

, ,«

en Ο) O o o 00 f-s CM O (O en f o o CJ3 to rH 03 co CO rH S3 o 03 f τ—' oo •3· in C» in to ÇO rH CÓ rH Τ Η C M 1 1 in

cd Ν

ω ì

Β

O to in 03 o tv CM in ^ 1ι

·? â M "3

l«' I

«

O to to rH

«

•

>