Advances in the Economic Analysis of Participatory and Labor-Managed Firms [1 ed.] 9780857244543, 9780857244536

Citation preview

Eriksson Editor

Advances in the Economic Analysis of Participatory and Labor-Managed Firms Volume 11

Advances in the Economic Analysis of Participatory and Labor-Managed Firms Volume 11 Advances in the Economic Analysis of Participatory and Labor-Managed Firms Volume 11

The eleventh volume of Advances in the Economic Analysis of Participatory and Labor-Managed Firms comprises nine original research chapters on a broad range of topics within the area of participatory organizations. The first chapter uses a novel data set from a large multinational firm to look at an important, yet not well understood question: what determines employees’ decisions to join employee share ownership plans? Three contributions deal with the adoption and consequences of new work organizations with a higher degree of involvement and decision-making authority of the employees. Another set of three chapters is concerned with the performance impact of stock-based compensation schemes and the impact of Japan’s “lost decade” on union wage permia. Two chapters report research on recent challenges facing two specific participatory organizations: the worker-owned cooperative and the kibbutz.

Advances in the Economic Analysis of Participatory and Labor-Managed Firms

Advances in the Economic Analysis of Participatory and Labor-Managed Firms

Advances in the Economic Analysis of Participatory and Labor-Managed Firms Tor Eriksson Editor

I S B N 978-0-85724-453-6

9 780857 244536

Emerald_AEAP-V011_Cover.indd 1

10/22/2010 5:06:28 PM

ADVANCES IN THE ECONOMIC ANALYSIS OF PARTICIPATORY AND LABOR-MANAGED FIRMS

ADVANCES IN THE ECONOMIC ANALYSIS OF PARTICIPATORY AND LABOR-MANAGED FIRMS Series Editor: Takao Kato Recent Volumes: Volume 8:

Employee Participation, Firm Performance and Survival – Edited by V. Perotin & A. Robinson

Volume 9:

Participation in the Age of Globalization and Information – Edited by Panu Kalmi & Mark Klinedinst

Volume 10:

Cooperative Firms in Global Markets: Incidence, Viability and Economic Performance – Edited by Sonja Novkovic & Vania Sena

ADVANCES IN THE ECONOMIC ANALYSIS OF PARTICIPATORY AND LABOR-MANAGED FIRMS VOLUME 11

ADVANCES IN THE ECONOMIC ANALYSIS OF PARTICIPATORY AND LABOR-MANAGED FIRMS EDITED BY

TOR ERIKSSON Department of Economics, Aarhus University, Denmark

United Kingdom – North America – Japan India – Malaysia – China

Emerald Group Publishing Limited Howard House, Wagon Lane, Bingley BD16 1WA, UK First edition 2010 Copyright r 2010 Emerald Group Publishing Limited Reprints and permission service Contact: [email protected] No part of this book may be reproduced, stored in a retrieval system, transmitted in any form or by any means electronic, mechanical, photocopying, recording or otherwise without either the prior written permission of the publisher or a licence permitting restricted copying issued in the UK by The Copyright Licensing Agency and in the USA by The Copyright Clearance Center. No responsibility is accepted for the accuracy of information contained in the text, illustrations or advertisements. The opinions expressed in these chapters are not necessarily those of the Editor or the publisher. British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN: 978-0-85724-453-6 ISSN: 0885-3339 (Series)

Emerald Group Publishing Limited, Howard House, Environmental Management System has been certified by ISOQAR to ISO 14001:2004 standards Awarded in recognition of Emerald’s production department’s adherence to quality systems and processes when preparing scholarly journals for print

CONTENTS LIST OF CONTRIBUTORS

vii

FOREWORD

ix

INTRODUCTION

xi

TO JOIN OR NOT TO JOIN? FACTORS INFLUENCING EMPLOYEE SHARE PLAN MEMBERSHIP IN A MULTINATIONAL CORPORATION Alex Bryson and Richard B. Freeman

1

DO INNOVATIVE WORKPLACE PRACTICES FOSTER MUTUAL GAINS? EVIDENCE FROM CROATIA Derek C. Jones and Srecko Goic

23

THE PERFORMANCE IMPACT OF FINANCIAL PARTICIPATION: SUBJECTIVE AND OBJECTIVE MEASURES COMPARED Panu Kalmi and Christina Sweins

69

EMPLOYEE DISCRETION AND THE LABOR-MARKET ENVIRONMENT Jaime Ortega

89

THE PRODUCTIVITY EFFECTS OF PROFIT SHARING, EMPLOYEE OWNERSHIP, STOCK OPTION AND TEAM INCENTIVE PLANS: EVIDENCE FROM KOREAN PANEL DATA Takao Kato, Ju Ho Lee and Jang-Soo Ryu v

111

vi

CONTENTS

STOCK OPTION SCHEMES AND FIRM TECHNICAL INEFFICIENCY: EVIDENCE FROM FINLAND Mikko Ma¨kinen

137

THE UNION WAGE PREMIUM, VOICE, AND NONUNION WORKERS’ ATTITUDES: BEFORE AND AFTER JAPAN’S LOST DECADE Tsuyoshi Tsuru

161

WORKER AND COMMUNITY COOPERATIVES: A MULTI-CRITERION MODEL Roger A. McCain

205

ANTECEDENTS AND CONSEQUENCES OF THE ADOPTION OF MARKET-BASED COMPENSATION BY ISRAELI KIBBUTZIM Raymond Russell, Robert Hanneman and Shlomo Getz FROM DESTRUCTIVE TO CREATIVE TRADE THROUGH ECONOMIC DEMOCRACY Jaroslav Vanek

225

247

LIST OF CONTRIBUTORS Alex Bryson

National Institute of Economic and Social Research, London, UK

Richard B. Freeman

Department of Economics, Harvard University, Cambridge, MA, USA

Shlomo Getz

Institute for Kibbutz Research, University of Haifa, Haifa, Israel

Srecko Goic

Department of Management, University of Split, Split, Croatia

Robert Hanneman

Department of Sociology, University of California, Riverside, CA, USA

Derek C. Jones

Department of Economics, Hamilton College, Clinton, NY, USA

Panu Kalmi

Aalto University School of Economics, Helsinki, Finland

Takao Kato

Colgate University, Hamilton, NY, USA

Ju Ho Lee

KDI School of Public Policy and Management, Seoul, South Korea

Mikko Ma¨kinen

Aalto University School of Economics, Helsinki, Finland

Roger A. McCain

Department of Economics and International Business, Drexel University, Philadelphia, PA, USA

Jaime Ortega

Department of Business Administration, Carlos III University, Madrid, Spain

Raymond Russell

Department of Sociology, University of California, Riverside, CA, USA vii

viii

LIST OF CONTRIBUTORS

Jang-Soo Ryu

Pukyong National University, Busan, South Korea

Christina Sweins

Aalto University School of Science and Technology, Helsinki, Finland

Tsuyoshi Tsuru

Institute of Economic Research, Hitotsubashi University, Tokyo, Japan

Jaroslav Vanek

Department of Economics, Cornell University, Ithaca, NY, USA

FOREWORD The series Advances in the Economic Analysis of Participatory & LaborManaged Firms was launched 25 years ago by Derek C. Jones and Jan Svejnar. Since then, Advances has been a leading forum for high-quality original theoretical and empirical research in the broad area of participatory and labor-managed organizations. Although general and specialized journals publish work in this field, many do so only occasionally. Advances has been the only annual periodical that presents some of the best chapters in the field in a single volume. It is my great pleasure to present Volume 11 of Advances in the Economic Analysis of Participatory and Labor-Managed Firms which marks the first volume in this series to be produced by Emerald. It is also the first volume since I took over the editorship from Derek Jones. I would like to take this opportunity to express my deepest debt of gratitude to Derek for his outstanding editorship. Advances has been making frequent use of guest editors. This volume is also ably edited by Tor Eriksson. The selection of Tor as this volume’s guest editor symbolizes profound changes in the nature and scope of mainstream labor economics in recent years with the rising influence of European labor economists, in particular in the growing fields of behavioral economics and new institutional economics (Tor is among a few labor economists in Europe who have been making significant contributions to mainstream labor economics consistently over the past two decades or so). It is my hope that Advances increasingly reflects such recent changes and welcomes chapters utilizing diverse methodologies ranging from conventional economic analysis (including both theoretical and econometric studies) to new institutional economics to behavioral economics. The scope of Advances will also reflect great changes in the realities of participatory organizations in the past two decades or so. Following the disintegration of the Former Republic of Yugoslavia, the principal systemic example of self-management was replaced with diverse forms of participatory systems. In advanced market economies, many firms have been experimenting with new and innovative work practices aimed at promoting employee participation in decision making in the workplace (sometimes even at the top corporate level) and alternative compensation systems ix

x

FOREWORD

designed to align the interest between labor and management. In addition, a number of significant examples of worker cooperatives have flourished. In transition economies, the collapse of the former USSR triggered widespread experimentation with diverse forms of participation, in particular employee ownership. I hope you will find this volume informative and stimulating and that you will consider contributing to the future volumes of Advances and sharing information about Advances with other interested colleagues. Takao Kato Series Editor

INTRODUCTION This volume of Advances in the Economic Analysis of Participatory and LaborManaged Firms comprises nine original research chapters and a short comment.1 The chapters cover a broad range of topics: from share ownership plan membership to determinants and performance outcomes of adoption of high performance work and pay practices, to changes in traditional participatory organizations. Geographically, the chapters span over a relatively wide range: from Northern Europe, a southeast European transition economy, Israel, Korea, Japan to an international corporation with operations in at least four continents. Employee share ownership has grown in importance in many countries in recent years as companies introduce them hoping that they will contribute to the alignment of the interests of the firm and its employees. The first chapter in this volume by Alex Bryson and Richard B. Freeman analyzes employees’ decisions to join the employee share ownership plan in a multinational firm. Although the plan is financially attractive to the great majority of the employees, far from all of them choose to join. The study finds that participation varies greatly with demographic and job characteristics (job tenure and job autonomy) and location (country). Differences in information do not explain the observed differences in participation. Overall, the analysis demonstrates that the participation decision is not chiefly driven by pure economic incentives, and that behavioral economics responses notably coworker effects, are helpful in providing a better understanding of its determinants. Common to the three next contributions is the focus on the adoption and consequences of new work organizations which imply a higher degree of involvement and decision-making authority of the employees in the firm. Following the rapid spread of innovative work practices like teams, quality circles, job rotation, employee ownership, and involvement schemes, a research literature concerned with the effects of these practices on worker and firm outcomes has been built up. The growing body of evidence is, however, almost exclusively based on studies of data from firms and employees in advanced market economies, whereas investigations from developing countries are rather thin on the ground. The chapter by Derek C. Jones and Srecko Goic makes a valuable contribution by examining worker as well xi

xii

INTRODUCTION

as firm outcomes exploiting a detailed data set from a Croatian manufacturing firm. The analysis of the Croatian firm is also interesting because it is one of the few studies to examine work organization in a former communist country. Furthermore, the specific Yugoslavian legacy with self-managed firms under communism is an additional reason for why a study of a Croatian firm should attract the attention of scholars in this field. The key results of their analysis is that of the several innovative work practices adopted by the firm, offline teams not only have an especially strong positive effect on workers’ effort, but also on the involvement and job satisfaction of the employees. The mutual gains for both the firm and the workers are particularly strong when offline teams are used together with financial incentives. An interesting finding of the study is the observed gender differences: women feel that they less empowered than their male colleagues, and women are also less willing to engage in peer monitoring. This is noteworthy as it is likely to be the main mechanism through which the new work practices give rise to enhanced firm performance. Empirical studies of the performance impact of financial participation programs make use of either objective or subjective impact measures. In their study, Panu Kalmi and Christina Sweins use both and contrast them using a matched firm-pair comparison methodology. Moreover, they examine the quality of subjective data collected from single, well-informed respondents with those from several, but less informed respondents. The two main findings of their analysis are that objective impact measures are smaller (and frequently insignificant) than the subjective impact measures, and that the correlation between the objective and subjective measures is in general positive for the well-informed respondent (personnel fund chairperson), whereas it is negative, albeit often insignificant, for the less informed respondents. The latter result indicates that it is better to use a single, well-informed respondent as data source, despite the potential single respondent bias associated with this approach. A distinguishing feature of many of the new high involvement work practices is delegation of decision-making authority to individual employees. Increased discretion of employees to organize their work is often considered to be a central mechanism through which the new work organizations yield performance gains. In his chapter in this volume, Jaime Ortega examines whether the firm’s decision to delegate authority to its workers is affected by the employees’ bargaining power and labor market regulations. The point of departure of the analysis are two theoretical perspectives yielding different predictions: the bargaining approach which implies a positive relationship between bargaining power and

Introduction

xiii

discretion, and an efficiency wage model which predicts that stronger labor regulations and employee bargaining power are associated with less discretion given to workers. The empirical analysis is based on data from the European Working Conditions Survey from 2000 (covering 27 EU countries) and information on labor regulations (at the country level) and regional/national unemployment rates from two additional sources. A key finding is that there is a negative relationship between unemployment levels and discretion. This contradicts the efficiency wage story but is consistent with a bargaining model where employees like to have more discretion. Although the results concerning the labor regulations are less unambiguous, the important message from them is that labor regulations do not only influence firms’ employment decisions but also how they organize work within firms. The three following chapters deal with remuneration of workers and consequences thereof. The first of these is concerned with the productivity effects of profit sharing, employee ownership, stock option and team incentive plans on firm productivity. Traditionally Korean firms have subscribed to the East Asian model of industrial relations, with long-term employment, seniority-based pay, and internal promotion systems. However, in recent years this appears to have been subject to changes, as a growing number of firms have adopted stock options and team incentive shemes. The study by Takao Kato, Ju Ho Lee and Jang-Soo Ryu use panel data on publicly traded Korean companies to examine the productivity impacts of the new as well as more traditional remuneration schemes. In short, they find that profit sharing and team incentives lead to considerable increases in productivity, whereas stock options and ESOPs seem not to be associated with improvements in productivity. Mikko Ma¨kinen’s chapter deals with the performance impact of a specific form of financial participation scheme: employee stock options. As in many other countries, their use has also increased significantly since the late nineties in the Finnish manufacturing and ICT sectors which he studies. Unlike financial participation programs like ESOPs or bonus schemes, there is relatively little research on how employee stock options or managerial equity ownership affect firm performance. Ma¨kinen contributes to this small literature by analyzing high-quality firm panel data (from 1992 to 2002), which allow him to estimate stochastic production frontiers and to examine the link between the firm’s use of stock options (distinguishing between broad-based programs and targeted to particular employees) and the technical efficiency of the firm.

xiv

INTRODUCTION

The statistical analysis provides little support for the notion that stock option schemes reduce technical inefficiency and hence contribute to improved firm performance. Rather, the estimation results show that broad-based option programs increase inefficiency in the manufacturing sector. In the information and communication technology sector the marginal effects of option programs do not differ statistically from zero. It should be noted, that in contrast to the few earlier studies, the current one is based on data that spans over a full business cycle and is comprehensive in terms of coverage (that is less likely to be plagued by selection problems). Tsuyoshi Tsuru’s contribution is concerned with industrial relations in Japan before and after the so-called lost decade, the prolonged recession the analysis and experiences of which have recently become of interest to a wider readership than only economists focusing on the Japanese economy. More specifically, Tsuru addresses three questions: what has happened to union wage premia and to union voice effects (job separation rates and job satisfaction), and has there been a change in non-union workers’ attitudes to unionization of their workplaces? For these purposes he carries out a series of econometric analyses on data from two identical surveys conducted in 1992 and 2007, respectively. The results of these exercises suggest that there have indeed been notable changes in Japanese industrial relations and that these are not merely due to countercyclicality in union effects. Thus, it is found that while there were no union wage premia in 1992, they do exist 15 years later, but only for male employees. Likewise, there was a decline in the propensity for voluntary quitting and in job dissatisfaction, and again exclusively for males. Finally, support for unionization has increased among nonunion workers. These findings imply that unions have exercised their bargaining power to maintain and increase their wage levels at the same as employment in the unionized firms has been declining, and that the Japanese unions seem to largely neglect the interests of the female employees. The next two chapters are concerned with recent challenges facing participatory organizations that were originally created in response to criticisms of capitalist corporate organizations: the worker-owned cooperative and the kibbutz. Roger A. McCain’s theoretical analysis extends models of participatory firms in a number of ways. First, it introduces multi-criterion decision making into an otherwise standard model of worker-owned cooperatives. Second, the analysis also accounts for ‘‘community cooperatives,’’ which are cooperatives with broader interests than those of the owners. Moreover,

xv

Introduction

McCain makes a distinction between layoffs and other separations in the models of the participatory firm. The chapter provides a discussion and comparison of three forms of cooperatives: worker-owned firms with closed and open membership, respectively, and the ‘‘community cooperative,’’ which is a hybrid cooperative/nonprofit firm. The latter differs from the two former ones in that it can have incentives to abate externalities. Whether it does, depends on the power of good intentions in politics. Probably the best known example of an organization that distributes resources according to the principle ‘‘from each, according to ability, to each according to need’’ is the kibbutz. The Israeli kibbutzim has, as is documented in the chapter by Raymond Russell, Robert Hanneman and Shlomo Getz, in recent times undergone fundamental changes: paying members different salaries, asking members to pay individually for their consumption, privatization of services and adoption of more business type governance. Their analysis has three aims: to examine the determinants of this transformation at the level of the kibbutz, to investigate the extent to which the changes have improved the economic and membership problems, and to test sociological theories of organizational transformation. The statistical analysis of the data from a panel of 249 kibbutzim during years 1995 to 2004 documents that the kibbutzim primarily adopted the new structure in response to economic and demographic problems. In accordance with theories of organizational inertia and deinstitutionalization, the transformations to new types of kibbutz were not, however, found to have been followed by improvements in the financial situation of the kibbutzim, nor in their membership development. Finally, one of the pioneers in the analysis of worker-owned firms, Jaroslav Vanek, in a short note discusses the differences democratic firms can make for globalization and international trade.

NOTE 1. The submitted chapters were evaluated by anonymous referees, for whose help and advice I am deeply grateful. Owing to their relatively small number, I have chosen not to list them in order to maintain their anonymity. Many thanks also to the Series Editor, Takao Kato, for his help and support in the production of this volume.

Tor Eriksson Guest Editor

TO JOIN OR NOT TO JOIN? FACTORS INFLUENCING EMPLOYEE SHARE PLAN MEMBERSHIP IN A MULTINATIONAL CORPORATION Alex Bryson and Richard B. Freeman 1. INTRODUCTION Ownership of shares by employees in their own firm has grown substantially in the advanced world. In the past two decades, it increased in Britain (Pendleton, Whitfield, & Bryson, 2009), the United States (Kruse, Freeman, & Blasi, 2010), and in many EU countries (Pendleton, Poutsma, van Ommeren, & Brewster, 2005; European Federation of Employee Share Ownership, 2009). By 2004, one-fifth of British workplaces had share ownership plans covering one-third of private sector employees (Bryson & Freeman, 2010). In the United States in 2006, an estimated 18% of workers had shares in their own firm, some held through collective employee stock ownership plans, some bought through employee stock purchase plans that give employees a discount on shares, and some through their 401k retirement savings Plan money. In addition to owning shares, 9% of US employees had stock options with the firm. Taking account of the overlap, 24% had an ownership stake through shares or options (Kruse, Blasi, & Park, 2010, Table 1). Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 1–22 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011005

1

2

ALEX BRYSON AND RICHARD B. FREEMAN

Table 1.

Rates of Joining Share Plan by Demographic and Personal Factors and by Job-Related Factors.

Whole sample Country UK USA South Africa Australia Demographic factors Age (years) o25 25–34 35–44 45–54 55þ Sex Male Female Ethnicity Black Not black Qualifications Degree No degree Professional qualifications Yes No Household circumstances Not married/living as married Married, no children living at home Married, children living at home Personal factors Risk scale (1,10) 1 5 10 Sociability scale (0–7) 0 3 7

Mean Membership

Monthly Contributions (US$), Members Only

56

153

56 45 34 75

152 155 71 169

28 56 63 64 58

101 134 179 161 159

61 51

179 121

32 60

64 160

63 51

182 130

58 56

180 147

47 62 62

136 150 174

54 55 43

96 137 195

51 57 50

131 183 257

3

Factors Influencing Employee Share Plan Membership

Table 1. (Continued ) Mean Membership

Job-related factors Occupation Senior manager Middle manager Lower manager Operational/delivery Support Technical Sales Payment method Hourly Salary only Salary plus bonus/commission Supervisory responsibilities Yes No Contracted weekly hours of work o35 35 W35o40 40þ Company tenure Z4 years Yes No Close supervision scale (1,10) 1 5 10 How easy to monitor others scale (1,10) 1 5 10 % family income from ShareCo earnings o80% 80%þ Worked for company acquired by ShareCo Yes No

Monthly Contributions (US$), Members Only

82 65 47 46 58 72 77

254 187 129 106 137 191 181

35 57 78

99 147 210

68 50

183 132

56 56 59 53

127 149 160 155

68 44

156 147

65 50 28

168 130 93

48 54 49

106 148 157

55 57

151 156

60 54

158 149

Note: N varies from 2,725 to 2,783. Contributions are converted to $US using exchange rates at the time of the survey. Monthly contributions are the mid-point in banded data.

4

ALEX BRYSON AND RICHARD B. FREEMAN

Firms introduce share ownership plans in the hope that ownership will align employee and employer objectives to increase productivity and profits. Surveys show that many employees desire some form of ownership in the firm in which they work, so providing this benefit ought to increase their loyalty and willingness to work hard for the firm. When firms subsidize the purchase of shares through a share ownership Plan, the financial deal is often so good that nearly all workers should join. But substantial numbers of eligible employees do not join share plans in the firms that offer them. What motivates the decision to invest in a subsidized share Plan? Opportunity to make money? The marketing of the Plan? Knowledge of how the Plan works? Do employees make their decision individually or are they influenced by the behavior of others in their work group?

2. THE PLAN AND THE SURVEY To see what affects the decision to join a share ownership Plan, in 2007– 2008 we surveyed 3,360 employees of a multinational firm ShareCo that offers a similar Plan to employees around the world. We asked employees about their knowledge of the share Plan (‘‘the Plan’’) and the reasons they had or had not joined it.1 The firm distributed the survey through the internet to employees in Australia and New Zealand, South Africa, the United Kingdom and Ireland, and the United States.2 The response rates for the survey were 65% in the United Kingdom and Ireland, 62% in Australia and South Africa but just 35% in the United States. The data covers 19 business divisions and 39 office locations. Dividing workers into groups based on the intersection of division and location to obtain a closer fix on likely ‘‘work groups’’ where employees may interact regularly, we obtain 81 work units with one or more person. For most employees, the Plan ought to be financially attractive. We illustrate this with details of the Plan in Australia. In Australia, all permanent employees resident for tax purposes could choose between the Exempt Plan, which seemed most suitable for low-paid employees and the Deferred Plan, which seemed more suitable for high-paid employees. The Exempt Plan allowed employees to contribute up to AU$500 per annum in before-tax salary to acquire shares, which ShareCo matched up to a maximum of AU$500 per annum. The shares were free from taxation on acquisition up to AU$1,000. An employee who held the shares for at least 3 years could sell or transfer them tax free. Thus a worker who bought AU$500 shares and held them for 3 years would double their money if the

Factors Influencing Employee Share Plan Membership

5

share price held steady. The price would have to drop by half before the employee would lose money. The Deferred Plan allowed employees to contribute between AU$1,500 and half of their before-tax annual salary to acquire shares. ShareCo matched contributions up to AU$3,000 per annum. The government deferred income tax on share acquisition and did not tax employee bought shares on sale/transfer for 1 year and did not tax shares given to employees by ShareCo matching their purchases for 2 years. A worker who bought AU$3,000 and held them for 2 years would double their money if the share price held steady and again break even if the price fell to one-half its purchase price. The pecuniary incentive to join the Plan differs in the other countries where ShareCo operates because the firm offers different matching rates to workers and because each country gives different tax advantages for ownership. In the United Kingdom, the company matches shares purchased by the employee one for one up to a limit. The scheme qualifies for tax advantages as a Share Incentive Plan (SIP), whose tax advantages are comparable to those in the Australian tax code.3 In South Africa, the matching scheme resembles the Australian Deferred Plan and thus seems more suited to higher wageworkers. Employee contributions come from after-tax income and the employer match is subject to income taxation (which the company pays), while the gains from sale are subject to capital gains tax.4 All of which makes the scheme less valuable to workers than the Australian and UK schemes. In the United States, the company has the smallest matching incentive – a 15% discount on the market price of every purchase up to a maximum of US$1,990, which is considerably below the 50% subsidy implicit in the matching schemes. A share price fall of 15% or more means that US members of the Plan lose money.5 Table 1 shows the rate of joining the Plan by country, demographic characteristics of the worker, attitudes toward risk and sociability, and by job characteristics of workers. Fifty-six percent of all surveyed employees join the firm’s Plan, with considerable variation by country and employee characteristics. Australian and UK employees have higher membership rates than US employees, while the lowest membership rate is in South Africa. Joining varies substantially by demographic and job-related characteristics, but a substantial number of employees do not join the Plan even in the groups with the highest participation rates. Consider, for example, rates of joining by age. Older workers are more likely to be members than younger workers. However, apart from the small number of workers aged under 25 years who have very low rates of participation, participation rates vary only modestly by age, peaking at 64% for 45–54 year olds.

6

ALEX BRYSON AND RICHARD B. FREEMAN

Among occupations, there are also large differences in the rates of joining but the 82% rate for the highest group, senior managers, is still far from universal joining. It is possible that some combination of personal and job characteristics explains much of the variation among individuals in joining. To examine this possibility, we estimated the effect of employees’ personal and job characteristics on the probability that the employees joined the Plan using probit regressions that link joining the Plan to the demographic and job characteristics shown in Table 1. The demographic characteristics are measured by dummy variables for age (five dummy variables): male; black; household status (three dummies for marital and child status); degree holder; holds a professional qualification. In addition, we have two measures of the people’s personality, their risk preferences and sociability (to be described shortly). We measure job characteristics by occupation (seven dummies): supervisory responsibility; contractual hours (four dummies); months’ tenure with the employer and its squared term; payment method (three dummies); whether working in a company previously acquired by ShareCo; and by measures relating to the degree of job autonomy and the ease with which one can monitor coworkers’ efforts. There are also dummy variables for the four countries surveyed. Column 1 of Table 2 presents the probit coefficients and the t-statistics for their impact, while column 2 transforms the coefficients into their marginal effects on the probability of joining the Plan. Mirroring the means in Table 1, the probability of Plan membership rises with age until employees reach their mid-50s and then falls. Compared to an employee aged under 25, a ‘‘like’’ employee who is aged 45–54 years has a 15 percentage point greater probability of being a Plan member. Interpreted as reflecting age, this suggests that many nonmembers will join the Plan in the future as they age. But it is also possible that the coefficient reflects cohort differences, in which case joining need not rise as person’s age. The probability of being a member is also significantly higher for men, for those with degree-level educational qualifications, and for married persons with children. The gender, degree, and marital status effects are of a similar size, raising the probability of membership by around 6–8 percentage points.6 Together, demographic characteristics account for 7% of the variance in Plan participation. The estimated coefficients and marginal effects of occupation and position in the firm in Table 2 show that these factors are more important in determining membership than the demographic factors.7 Salaried staff were more likely to join the Plan than hourly paid workers, while those paid salaries plus bonuses were the most likely to join. Being a Senior Manager

7

Factors Influencing Employee Share Plan Membership

Table 2. Estimates of the Effect of Characteristics, Risk Aversion, and Sociability on Plan Membership and Monthly Contributions ($US). (1a) Membership (2) Membership (3) Monthly Contributions (Only Probit Marginal Effects for Those Who Contribute) Age (ref.: o25 years) 25–34 years 35–44 years 45–54 years 55þ years Male Black Degree Professional qualification

0.209 (2.193) 0.243 (2.212) 0.387 (3.307) 0.313 (2.229)

0.081 (2.211) 0.094 (2.253) 0.146 (3.474) 0.118 (2.343)

12.302 (1.119) 44.908 (3.428) 52.846 (3.849) 63.535 (3.971)

0.151 (2.642) 0.043 (0.419) 0.206 (3.167) 0.087 (1.182)

0.059 (2.647) 0.017 (0.417) 0.080 (3.192) 0.034 (1.177)

27.315 (4.304) 39.309 (3.068) 12.867 (1.797) 1.852 (0.170)

Household status (ref.: not married/living as married) Married/living as married, 0.039 0.015 no children at home (0.567) (0.569) Married/living as married, 0.147 0.057 children at home (2.112) (2.129) Sociability scale Risk scale Risk scale squared

0.003 (0.135) 0.033 (0.572) 0.004 (0.770)

Occupation (ref.: operational/delivery) Senior manager 0.414 (2.487) Middle manager 0.315 (2.526) Lower manager 0.229 (2.068) Support 0.128 (1.440) Technical 0.199 (1.922) Sales 0.284 (1.933) Supervisory 0.122 responsibilities (1.821)

0.001 (0.135) 0.013 (0.572) 0.001 (0.770) 0.153 (2.709) 0.119 (2.648) 0.089 (2.110) 0.05 (1.458) 0.077 (1.968) 0.108 (2.027) 0.048 (1.831)

4.515 (0.578) 11.278 (1.276) 6.637 (1.851) 12.458 (1.719) 1.319 (2.124) 62.596 (3.285) 28.400 (1.744) 2.254 (0.157) 22.137 (2.431) 42.163 (3.806) 31.746 (2.310) 16.844 (1.868)

8

ALEX BRYSON AND RICHARD B. FREEMAN

Table 2. (Continued ) (1a) Membership (2) Membership (3) Monthly Contributions (Only Probit Marginal Effects for Those Who Contribute) Contractual weekly hours (ref.: 40þ hours) o35 hours 0.149 (1.320) 35 hours 0.101 (0.904) W35 hourso40 hours 0.228 (2.909)

0.059 (1.312) 0.040 (0.900) 0.090 (2.902)

13.387 (1.159) 3.907 (0.379) 2.342 (0.300)

0.013 (10.715) 0.000 (8.093)

0.005 (10.730) 0.000 (8.094)

0.09 (0.603) 0.000 (0.037)

Payment method (ref.: Salary only) Hourly 0.240 (3.076) Salary plus bonus/ 0.288 commission (2.932)

0.095 (3.064) 0.110 (3.049)

21.847 (2.280) 21.98 (1.793)

0.046 (4.493) 0.009 (0.804) 0.015 (0.231)

0.018 (4.491) 0.004 (0.804) 0.006 (0.231)

0.311 (0.211) 1.815 (1.289) 6.368 (0.799)

0.574 (4.152) 0.504 (4.005) 0.640 (7.085) 0.672 (2.928) 0.20 2706

0.226 (4.242) 0.199 (4.085) 0.240 (7.616)

Tenure (months with company) Tenure squared

Close supervision scale How easy to monitor others’ efforts scale Worked in company acquired by ShareCo Country (ref.: UK) USA South Africa Australia Constant r2 N

0.20 2706

23.062 (1.246) 69.953 (5.255) 0.685 (0.082) 83.529 (2.997) pWw2 ¼ 0.0000 1506

Notes: (1) Model 1 is a probit for membership. The marginal effects are in 1(b). Model 2 uses interval regression for contributions per month for current members where the dependent variable is banded contributions data converted into US$ using exchange rates at the time of the survey. The interval regression lnsigma 4.83t ¼ 82.66. Model based on 0 left-censored observations, 284 uncensored observations, 39 right-censored observations, and 1,183 interval observations. (2) Robust estimator used. (3) t-statistics are presented in parentheses. Significant at 10% level; Significant at 95% confidence interval; Significant at 99% confidence interval.

Factors Influencing Employee Share Plan Membership

9

raised the probability of Plan membership by 17 percentage points compared to a member of the Operations and Delivery Staff (e.g., in customer service or a communication center worker). Supervisory status was also positively associated with Plan membership. But Plan membership probabilities are not a simple reflection of occupational hierarchy: sales staff had similar Plan membership probabilities to those in middle and lower management. The probability of Plan membership is strongly associated with tenure but the relation is not a simple linear one; tenure squared has a negative coefficient in the probit analysis. Membership rises with tenure until a worker has 20 years of tenure and then falls. This pattern mirrors the pattern of membership in the age dummy variables. Lower membership in the Plan with age and tenure could reflect the desire of older and more senior workers to diversify their assets as they near retirement, but it could also reflect the fact that some of those workers joined the firm before the Plan ever came into being and never changed their status quo position. To drill deeper into personal factors that might be associated with joining the Plan, we asked workers about their attitude toward risk and their propensity to join organizations – what may be called sociability. For risk, we asked a question that Dohmen et al. (2005) have shown correlates with risk behavior in laboratory experiments. The question is: ‘‘Are you generally a person who is fully prepared to take risks or do you try to avoid taking risks?’’ The responses are scaled from 1 (‘‘Unwilling to take risks’’) to 10 (‘‘Fully prepared to take risks’’). The probit estimates in column 1 show that risk preferences are not significantly related to the person joining the Plan. For ‘‘sociability,’’ we asked: ‘‘Do you take part in the following activities, either as part of your job or outside work? Please select as many as apply to you y . Belong to a trade/professional body or association; working with schools, colleges, universities; being involved in charities or voluntary bodies; being a member of a social, sports, or arts club; being an active member of a political party; being an active member of a religious group; socialising with coworkers outside of work; none of these.’’ We counted the number of activities employees engaged in and entered it into the probit equation. This variable is not associated with membership. We also examined whether employees who felt that they had control of their work were more likely to join than those who worked under close supervision. We developed a scale based on the question: ‘‘Are you closely supervised, or do you work fairly independently of close supervision?’’ We coded responses from 1 representing ‘‘working independently of close supervision’’ to 10 ‘‘closely supervised.’’ There is a strong negative

10

ALEX BRYSON AND RICHARD B. FREEMAN

correlation between close supervision and the propensity to participate in the Plan. An increase of 1 point on the close supervision scale reduces Plan participation by 2 percentage points, other things equal. The last factor that substantially influences employee decisions to join the Plan is the location of workers. Addition of the country dummies raises the proportion of the variance in Plan participation accounted for by the model from 15% to 20%. The differences among countries shown in the Table 1 are barely affected by the covariates in the multivariate analysis. To illustrate the magnitude of the country effect, consider the difference in outcomes among observationally equivalent employees in different countries. As a base case, we take a 45- to 54-year-old married man with children, with a degree, in senior management, with supervisory responsibilities, with a contract for 40 or more hours per week, with 10 years tenure, paid a salary with bonus, who is not closely supervised, and has sample mean characteristics on all other variables in the model. The model in Table 2 gives this ‘‘base case’’ person an estimated probability of Plan membership in Australia of 90%, of membership in the United Kingdom of 84%, in the United States of 70% probability, and in South Africa of 69% probability. Column 3 of Table 2 uses linear regression to assess the determinants of the monthly contributions of workers to the Plan among those who made contributions. For simplicity, we transformed all of the contributions data into US dollars at the then prevailing exchange rate. These estimates show that most of attributes of persons and jobs associated with a greater probability of being a Plan member are also associated with greater contributions conditional on being a member. Again, more of the variation in amounts contributed is attributable to differences in job-related characteristics than in demographic or personal characteristics. In sum, the key finding in Table 2 is that job characteristics and the location of the job are more important in determining membership in the Plan than the demographic characteristics of workers or our measures of risk preferences or sociability. It is more what you do and where you do it than who you are.

3. HOMO ECONOMICUS AND HOMO BEHAVIORICUS EXPLANATIONS The economist’s model of Homo Economicus directs attention at pecuniary factors as likely determinants of joining the Plan. Homo Economicus presumably assesses the size of the subsidy and tax breaks, the time required

Factors Influencing Employee Share Plan Membership

11

to hold the stock before those benefits kick in, and the likely trend and variability in the share price in deciding whether to join a share Plan. At a big enough subsidy/tax break and expectation of staying with the firm long enough to gain the advantages, this model suggests that almost everyone would buy the shares. But with a modest subsidy/tax break and a short time horizon of staying with the firm, it predicts that many workers would reject holding assets in company stock.8 As noted, the cross-country differences in joining are consistent with differences in the company subsidy to purchase shares and the tax break for owning shares. But there are other reasons why pecuniary considerations might lead someone to take their money in cash rather than to invest in the Plan. A person paying high interest on credit card or other debt has a pecuniary incentive to pay the debt rather than to invest in the firm. The Economicus model also allows for nonpecuniary factors associated with the person’s preferences toward risk to affect the decision to join a share Plan. Someone who finds it painful to see the share price fluctuate significantly over time and who gains little additional utility from increases in the price ought to keep their money in some safer asset. Our measure of risk aversion did not help explain decisions, but perhaps a measure of loss aversion would help explain some of those who turn down the seemingly profitable investment opportunity. The psychology model of Homo Behavioricus directs attention at the imperfect way people actually make decisions. One factor that has received attention in analysis of responses to seemingly fruitful decisions is procrastination, a delay in changing a default position even when it is advantageous to make one’s choice quickly, that imply large internal transaction costs (Madrian & Shea, 2000; Engelhardt & Madrian, 2004). Another Behavioricus factor that has also received attention in decisions are peer effects, where someone’s decision depends critically on the decisions of others with whom they associate. To be sure, there can be rational reasons for peer effects – the wisdom of the crowd that often gives a better assessment of reality than individual judgment. But the traditional Economicus model does not treat them as a major factor in decision making. Finally, both models recognize that imperfect information can prevent an individual from rationally assessing the costs and benefits of investments. If you are uncertain of the consequences of an action, don’t act. The Economicus model treats lack of information as reflecting the costs of obtaining it. Given company efforts to inform employees about the Plan, it seems implausible that costs of information deter employees from learning the facts. The Behavioricus model raises the possibility that employees may tune out firm-provided information as just another bit of firm propaganda

12

ALEX BRYSON AND RICHARD B. FREEMAN

or sales pitch and procrastinate in addressing that information, though one could also easily see this as rational behavior. To find out the pecuniary factors and behavioral factors stressed by these two models that may underlie employee decisions on joining ShareCo’s share Plan we asked employees why they did or did not participate in the Plan. Then we examine the pattern of membership across business units and locales for evidence on one of the main factors they identified as important in decisions: discussions with coworkers.

3.1. What Workers Say Economists are often leery of what people say about their decisions on surveys, but it is usually better to obtain such information when possible than to speculate about why persons behaved in particular ways without any indication of what they believe affected their decision. We asked workers who had joined the Plan: ‘‘What made you join the Plan?’’ and asked those who did not join: ‘‘Why have you never joined the Plan?’’ We allowed them to give more than one reason. Table 3 displays the percentage of responses given to each of the questions (which sum to 100%) and the percentage of respondents who gave the answers (which can sum to over 100% because respondents could give multiple answers). The responses in Panel A show that the employees who joined the Plan deliberated over the decision. Just 10% of responses and 13% of respondents gave the response that the employee had joined automatically without thinking much about it. The most common reason for joining was that it had been a ‘‘good investment’’ given by 73% of Plan members. Country data (not shown in the table) reveal that the percentage motivated by good investment varied by country. Eighty-seven percent of UK Plan members cited ‘‘good investment’’ as the reason for joining whereas 73% of US Plan members cited good investment. This presumably reflects the fact that joining the Plan was a better investment in the United Kingdom than in the United States because the company sharing rate was much lower in the United States. In the total sample, 39% of respondents reported that their joining was because they felt good about the company, which implies that the decision was influenced by factors beyond the expectation of future financial rewards. Those in the United States (again data not given in the table) were significantly more likely to cite feeling good about the company as a reason for joining: one-half did so compared to around one-third in the other three countries.

13

Factors Influencing Employee Share Plan Membership

Table 3.

Reasons for Joining and Not Joining the Plan.

Panel A: Reason for Joining

Good investment Joined automatically without thinking much about it Felt good about the company Other reasons Panel B: Reason for Not Joining

Would take too much out of my salary/ can’t afford it Financial sense to invest outside the firm where you work Don’t want risk of investing in shares Don’t intend to be with the company very long I am about to join/will join shortly Features of the Plan I don’t like I don’t really understand the Plan Other reasons

Percent of Responses

Percent of Joiners (Can Answer More Than One Category)

56 10

73 13

30 5

39 7

Percent of Responses

Percent of Nonjoiners (Can Answer More Than One Category)

31

37

5

6

7 6

9 7

20 3 12 16

25 3 14 20

Note: Panel A: Unweighted N ¼ 1,776 employees and 2,320 responses. Panel B: Unweighted N ¼ 1,076 employees and 1,300 responses.

The responses in panel B show that the employees who did not join also paid attention to perceived pecuniary returns. Thirty-seven percent of nonmembers said their contribution would take too much out of their salary.9 Six percent of respondents thought it made sense to invest outside the firm in which they worked. Approximately twice as many nonjoiners gave that answer in the United States, where many employees invest substantial sums in their own businesses through 401k retirement plans. Nine percent of nonmembers didn’t want the risk of investing in shares per se. One-quarter of employees said they were ‘‘about to join’’ the Plan, which fits with the behavioral proposition that individuals often procrastinate in making a decision beneficial to them (Rabin, 1998). These nonmembers had lower tenure than other nonmembers – 28% had been with the company for 6 months or less compared to 18% of other nonmembers – suggesting that they had insufficient time to make their decision since joining the company.

14

ALEX BRYSON AND RICHARD B. FREEMAN

Seven percent of nonmembers said they chose not to join because they ‘‘Don’t intend to be with the company very long.’’ Almost two-thirds of these nonmembers expected to be working at ShareCo for under a year, compared with 7% of other nonmembers, which means they would gain less from the investment. Finally, 3% of nonmembers cited features of the Plan they did not like as a reason to avoid investing in it. A larger proportion (14%) said that one reason for not joining the Plan was that they did not understand it. Of the 20% who gave ‘‘other reasons,’’ 8% said they had never heard of the Plan. In a separate question, we also asked employees how well they understood aspects of the Plan. Consistent with the notion the insufficient information may have deterred some from joining, 27% of nonmembers answered ‘‘not very well’’ or ‘‘not at all well,’’ while just 4% of those who had joined the Plan gave those answers to the information question.

3.2. The Role of Coworkers We used data on the office locations and business units of the firm to estimate the proportion of employee respondents who worked in the same office and business unit. This enables us to estimate whether membership is more concentrated among employees likely to interact with each other than would happen if each employee decided to join independently of those of others in their location/unit. Greater concentration of membership than expected by independent choice would indicate that decisions were potentially subject to the influence of coworkers through some form of peer effects. To determine the expected level of Plan membership at the different locations, we used our Table 2 probit model to predict the determinants of the probability an individual employee would join the Plan. Then we averaged the probabilities for the employees at each location to get the expected level of joining in the location. Since the probabilities for individuals come from the same model, the predicted levels of joining vary across locations because of differences in the observable characteristics of workers across the locations. In offices with senior upper level managers, for instance, our model predicts higher membership than in workplaces with many less highly paid and skilled workers. Graph 1 is a scatter graph that plots the actual mean membership against the predicted rate of joining the Plan for each of the 88 location/ business unit categories for which we have data on more than a single

15

Factors Influencing Employee Share Plan Membership 1

(mean) prmemb1

0.8

0.6

0.4

0.2

0 0

0.2

0.4

0.6

0.8

1

(mean) member

Graph 1. The Percentage of Workers by Office who Join the Plan in our Sample Compared to the Percentage Predicted by Worker Characteristics at Each Office. Note: Each dot represents an office/business unit location. Locations with only a single respondent have been removed. N ¼ 88. Correlation of the actual rate of membership versus predicted rate across offices of 0.60.

person. The predicted and actual distributions are positively correlated at 0.60. If the decisions of workers at a particular office–business unit are influenced by the decisions of others, the dispersion of the rate of actual membership should be greater than the dispersion of the predicted rates, since the latter are based on a model that did not allow for peer or contagion type effects. Most models of peer or contagion effects predict greater dispersion in measures of behavior across groups than would occur based on the demographic characteristics of people because interactions in the group produce similar behavior (Glaeser, Sacerdote, & Scheinkman, 1996). The models allow the interactions to produce both more and less of the behavior. To see if this was true in our data, we calculated the dispersion of actual and predicted membership in the Plan across the sites. In fact, the standard deviation of the distribution of actual rates of joining across the 88 location/ business units was 0.24 compared to a standard deviation of just 0.17 for the

16

ALEX BRYSON AND RICHARD B. FREEMAN

predicted rates across the same location/business units. A variance ratio test for the equality of standard deviations confirms that the distributions are significantly different from one another.10 There is, however, a difficulty with this analysis. Some of our location/ business units have many employees and survey respondents while others have few employees and survey respondents. The range of responses was from 383 in the largest unit to less than 5 in 30 units. Consistent with a peer effect or contagion model, much of the greater variation in actual rates occurs in workplaces with few workers.11 But such a pattern could arise for reasons of sampling variability as well as for interactions. The smaller the number of observations, the more dispersed will any distribution be around its mean. One way of dealing with this problem is to compute the standard deviations from variances weighted by the number of persons in the location/ business unit. Weighted by the number of persons in each office, the standard deviation for the actual distribution is 0.19 while the standard deviation for the predicted distribution is 0.15 – a smaller but still noticeable difference.12 The standard deviation is, however, a crude measure of the way peer effects or interactions would produce a different shape of the distribution of actual outcomes than the distribution that would result absent peer effects. Graph 2 displays the histogram for the distribution of observed rates of joining the Plan among locations/business units, weighted by the number of employees and the histogram for the rates predicted by our model and the demographic distribution of the workplaces. The continuous curve shows the normal curve fit. It shows a more bifurcated distribution for the actual (left-hand panel) than for the predicted (right-hand panel) rates, which is what one would expect if peer effects induce more workers to join at some sites and fewer workers to join at others than would happen from choices that were not influenced by fellow workers. As a final test of the relation between location/business unit and joining the Plan, we added dummy variables for business unit/location to our Table 2 probit analysis. The addition of dummies raises the pseudo-R square that summarizes the fit of the model from 0.19 to 0.22, which is significant by a chi-square test. Thus, we are better able to predict which workers join the Plan and which do not upon addition of the information on business unit/ location data. Note, however, that this tells us only that units/locations differ in rates of joining from what one would expect on the basis of the characteristics of employees. It does not tell us at which locations the peer effects are likely to produce higher or lower rates of joining. Nor does it tell us whether in fact the observed patterns are truly attributable to the influence of coworkers on decisions or some other aspect of the workplace.

17

Factors Influencing Employee Share Plan Membership 3 2.5

2

1.5

Density

Density

2

1 1 0.5

0

0 0

0.2

0.4

0.6

(mean) member

0.8

1

0

0.2

0.4

0.6

0.8

1

(mean) prmemb1

Graph 2. Mean Membership Per Office/Business Unit and Predicted Mean Membership Per Office/Business Unit Weighted by Number of Unit Respondents (Continuous Line is the Normal Curve Fit).

To get more direct evidence on whether coworkers influence persons to join the Plan compared, we asked employees about the influence of fellow workers, management, and other persons on their decision regarding the Plan. We asked: ‘‘Have you/did you ever talk to any of the following people about membership of the Plan? y Fellow workers; My Supervisor; HR Manager/Department; Family or Friends Outside the Company; A Financial or Legal Adviser Outside the Company.’’ If they answered yes, we asked if the people were important in the decision that the worker made. Table 4 summarizes responses to these questions. Fifty-two percent of employees cited none of the five sources of information as important, but members reported speaking to more people than nonmembers and ascribed more importance to those discussions in their membership decision than did nonmembers. Employees were most likely to discuss Plan membership with fellow employees – 59% had done so – than with anyone else. In addition, more employees viewed these discussions as important in deciding whether or not to join than discussions with anyone else. By contrast, only 14% reported that they had discussed membership with HR staff and only 7%

18

ALEX BRYSON AND RICHARD B. FREEMAN

Table 4.

Importance of Discussions with Others in Membership Decision.

Fellow workers Yes, important Yes, not important No Supervisor Yes, important Yes, not important No HR manager/department Yes, important Yes, not important No Family/friends outside the company Yes, important Yes, not important No Financial/legal adviser outside the company Yes, important Yes, not important No

Non-Member

Member

All

23 26 51

38 28 34

32 27 42

13 12 76

19 14 68

16 13 71

7 7 85

7 6 87

7 7 86

17 30 70

29 19 52

24 16 60

6 5 89

10 7 83

8 6 86

Notes: (1) Employees were asked: ‘‘Have you/did you ever talk to any of the following people about membership of the Plan? If yes were they important in the decision you made?’’ (2) Table presents column percentages.

viewed discussions with HR staff as important. The influence at the workplace that leads some locations to join the Plan more than others thus appears to rest with coworkers rather than with management. As a further test of the potential influence of coworkers on decisions, we introduced a set of dummy variables for whether an employee had talked to a particular group about Plan membership into the Table 2 model of individual decision regarding the Plan. Table 5 gives the estimated coefficients on these variables from the new estimated model. The estimates show that talking to fellow employees was associated with an 8 percentage point increase in joining the Plan while talking to supervisors was associated with a 6 percentage point increase. But it also shows that talking to family and friends increased the probability of Plan membership by 13 percentage points relative to not talking to them. Since proportionately fewer workers said they had talked with family members than said they had talked with fellow employees, the larger coefficient on talking with family members does

19

Factors Influencing Employee Share Plan Membership

Table 5.

Estimates of the Effect of Talking to Other Persons on Whether the Employee Joined the Plan.

Who Talked to About Plan Membership Fellow workers Supervisor HR manager/department Family or friends Financial or legal adviser

Probit Coefficients 0.208 (3.30) 0.152 (2.20) 0.127 (1.41) 0.332 (5.30) 0.042 (0.46)

Marginal Effects 0.082 0.059 0.050 0.129 0.016

Notes: (1) t-statistics in parentheses with asterisk signifying significance where 0.10, 0.05 0.01. (2) Controls are as per Table 2. (3) Predicted membership mean under the model is 0.572.

not mean that the family was more important than the workplace. The two routes of impact add roughly similar explanatory power in the augmented regression model.13 Finally, we asked employees another question that casts light on potential peer influences in decisions to join the Plan. This question related to workers perceptions’ of whether other workers are joining the Plan: ‘‘What percentage of workers in your business unit do you think are members of the Plan?’’ If workers are following some perceived norm at their workplace we would expect those who believe many others are members would also join. In fact, the probability of an individual being a Plan member rises steeply with the perception of the Plan membership rate among coworkers.14 The correlation coefficient for the employees’ perception of the Plan participation rate in the business unit and the actual Plan membership rate in business units as derived in our data was 0.23, which shows that the measure of perception does not simply reflect the actual rate of membership, which makes it hard to interpret in any causal manner (Manski, 1993). Taken together the evidence on the concentration of membership by office, employees reporting that coworkers were important sources of information, and on their perceptions of the participation of other workers on their joining the Plan directs attention at peer influences on joining decisions above and beyond those that influence individual decisions in isolation.

20

ALEX BRYSON AND RICHARD B. FREEMAN

4. CONCLUSION Many firms encourage employees to own company stock through share plans that subsidize the price at favorable rates, which should make the decision to participate in the Plan a ‘‘no brainer.’’ Even so, many employees do not buy shares. Our analysis of a survey of employees in a multinational with a share ownership Plan finds considerable variation in joining for observationally equivalent workers within the firm. Workers’ probability of joining the share Plan is higher the greater the potential payoff, pointing to an important role for rational economic calculations. But some nonmembers say they intend to join in the future, which forgoes the benefits of immediate membership. And the behavior of coworkers influences the purchase of shares while company HR information does not affect the decision. The evidence thus indicates that participation reflects a mixture of economic responses to incentives and behavioral economics responses to what others do.

NOTES 1. The 3,360 employees who responded to the survey included 2,707 with no missing data. 2. Because the Plan differs modestly across countries, we used variants of the survey instrument in each country but the differences were so slight that we pooled the country responses into a single firm data set. 3. Under the UK SIP scheme, employees can contribute a minimum of d10 each month up to a maximum amount of d125 or 10% of their monthly pre-tax earnings, whichever is the lower amount. This sum is tax-exempt. ShareCo matches each share purchased up to a value of d125 per month. All shares acquired by the employee are exempt from tax if held for 5 years. 4. The South African deferred Plan allows permanent employees to contribute at least Rand 1,800 per annum up to 50% of their after-tax salary to acquire shares under the Plan. The company matches each Rand contributed by the employee up to a maximum of Rand 24,000 per annum to purchase matched shares. The shares purchased by the employee vest after the first year and the matching shares from the company vest after the second year. Monthly contributions are deducted directly from the employee’s after-tax monthly salary and thus are not tax privileged as the shares are acquired with after-tax money at their full market value. Matching shares are subject to taxation, which the employer pays upfront, so that the employee is not liable for income tax. However, when employees sell their shares they are subject to capital gains tax. A second share Plan awarded employees 50 shares free of charge in 2005. 5. The US Plan was open to all full-time employees who work at least 20 hours per week and more than 5 months in a calendar year. Employees may contribute between $10 and $800 per month of their gross salary through a payroll deduction. 6. There are notable differences in the association between Plan membership and demographic characteristics across the four countries surveyed. For example, the

Factors Influencing Employee Share Plan Membership

21

possession of professional qualifications is associated with a lower probability of Plan membership in the United Kingdom and South Africa but not in the United States or Australia. 7. Taken together job characteristics accounted for 13% of the variance in Plan membership in ShareCo and 10% of the variance in contributions among members when entered into the model alone. 8. These are frequently cited as the reasons why employees often choose not to claim in-work benefits and tax credits (see, e.g., Bingley & Walker, 2001). 9. Financial constraints and opportunities dominated other aspects of Plan investment too: the need for money was the chief reason given for selling shares and the availability of more money was given as the primary reason for increasing monthly contributions. The need for money was also the chief reason for leaving the Plan, though very few employees had actually left (4% of all employees and 6% of those who had ever been a member). 10. This F-test for the homogeneity of variances is performed using STATA’s sdtest. F2.83, df 117, FWf 0.0000. 11. Thus, among all 118 office–business units the standard deviation for membership was 0.32 compared to 0.19 for the predicted rates across the same location/business units. 12. A chi-square test of variance confirms this difference is statistically significant. 13. The only communication channel associated with a significant increase in monthly investments was talking to family and friends, perhaps reflecting discussions regarding the ability of the family to find the money to invest in the Plan. We also found that the HR effect differed in country-level regressions. It was large and positive for South Africa but negative effect in the United States. 14. Adding this perception measure to the Table 2 regression model, the effect of perceived membership among coworkers being around half (40–59%) raised the probability of an individual’s membership by 35 percentage points relative to a case in which the employee believed no coworkers were members.

ACKNOWLEDGMENT We thank ShareCo (a pseudonym) for their collaboration with this survey and the British Academy for financial support (grant number BR100020).

REFERENCES Bingley, P., & Walker, I. (2001). Housing subsidies and work incentives in Great Britain. The Economic Journal, 111(471), C86–C103. Bryson, A., & Freeman, R. (2010). How does shared capitalism affect economic performance in the UK? In: D. Kruse, R. Freeman & J. Blasi (Eds), Shared capitalism at work: Employee ownership, profit and gain sharing, and broad-based stock options (Chapter 6, pp. 201–224). Chicago: University of Chicago Press.

22

ALEX BRYSON AND RICHARD B. FREEMAN

Dohmen, T., Falk, A., Huffman, D., Sunde, U., Schupp, J., & Wagner, G. G. (2005). Individual risk attitudes: New evidence from a large, representative, experimentally-validated survey. IZA Discussion Paper no. 1730. Bonn, Germany. Engelhardt, G. V., & Madrian, B. C. (2004). Employee stock purchase plans. NBER Working Paper no. 10421. Cambridge, MA, USA. European Federation of Employee Share Ownership. (2009). A political roadmap for employee ownership in Europe. Available at http://www.efesonline.org/ROADMAP/ A%20political%20roadmap%20for%20employee%20ownership%20in%20Europe.pdf Glaeser, E. E., Sacerdote, B., & Scheinkman, J. A. (1996). Crime and social interactions. Quarterly Journal of Economics, 111(2), 507–548. Kruse, D. L., Blasi, J. R., & Park, R. (2010). Shared capitalism in the US economy: Prevalence, characteristics and employee views of financial participation in enterprises. In: D. Kruse, R. Freeman & J. Blasi (Eds), Shared capitalism at work: Employee ownership, profit and gain sharing, and broad-based stock options (Chapter 1, pp. 41–76). Chicago: University of Chicago Press. Kruse, D. L., Freeman, R. B., & Blasi, J. R. (2010). Shared capitalism at work: Employee ownership, profit and gain sharing and broad-based stock options (http://www.nber.org/ books/krus08-1). Chicago: University of Chicago Press. Madrian, B. C., & Shea, D. F. (2000). The power of suggestion: Inertia in 401(k) participation and savings behaviour. NBER Working Paper no. 7682. Cambridge, MA, USA. Manski, C. (1993). Identification of exogenous social effects: The reflection problem. Review of Economic Studies, 60, 531–542. Pendleton, A., Poutsma, E., van Ommeren, J., & Brewster, C. (2005). Employee share ownership and profit sharing in the European union. Dublin: European Foundation for the Improvement of Living and Working Conditions. Pendleton, A., Whitfield, K., & Bryson, A. (2009). The changing use of contingent pay at the modern British workplace. In: W. Brown, A. Bryson, J. Forth & K. Whitfield (Eds), The evolution of the modern workplace (Chapter 11). Cambridge: Cambridge University Press. Rabin, M. (1998). Psychology and economics. Journal of Economic Literature, 36(1), 11–46.

DO INNOVATIVE WORKPLACE PRACTICES FOSTER MUTUAL GAINS? EVIDENCE FROM CROATIA Derek C. Jones and Srecko Goic 1. INTRODUCTION There is abundant evidence that innovative work practices (IWPs) of various kinds, such as teams, quality control circles, no-layoff policies, job rotation, and employee ownership, have spread rapidly in developed market economies during the past thirty years or so. Between 1983 and 1993, Freeman, Kleiner, and Cheri (2000) report survey evidence that the number of nonmonetary incentive programs offered by firms increased by 500% in the United States. Similar trends appear to be at work in other countries including the United Kingdom, Japan (e.g., Kato, 2000), Denmark (e.g., Datta Gupta, & Eriksson, 2004), and Finland (e.g., Kalmi & Kauhanen, 2008). Whereas the corresponding evidence for transition economies is much slimmer, the available evidence is also suggestive that such practices are limited though spreading.1 Unsurprisingly, both theoretical and an empirical literature have appeared to examine the impact of IWPs on both business performance and employee outcomes. As different scholars from diverse fields in the broad area of industrial relations have applied varying approaches to explore several research questions, those literatures have grown rapidly. At the same time, while it is clear that analytical work of the kind is becoming commonplace for advanced economies, work that focuses Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 23–68 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011006

23

24

DEREK C. JONES AND SRECKO GOIC

on developing and transition economies is very slim. As it is important to determine whether findings for firms in advanced market economies carry over to other economies, the first contribution of this chapter is to extend the geographical coverage of the empirical literature. This we do by assembling and analyzing new survey data set for a large Croatian manufacturing firm with our chapter perhaps representing one of the first such investigations for a former communist economy. Another reason why case studies of firms in transition economies is perhaps of special interest is that some have argued that the legacies of communism will help to shape and constrain contemporary labor-management relations in transition economies (e.g., Blanchflower & Freeman, 1997). At the same time, Croatia was once part of the Socialist Federative Republic of Yugoslavia (SFRY), in which a system of self-management existed within a communist state. Such a legacy means that examination of the potential of IWPs is perhaps of special interest. Finally, in many former communist countries, often privatization has been accompanied by significant employee and managerial share ownership in the firm at which they work. In our case in Croatia, the process of insider privatization was facilitated by the creation of an Employee Stock Ownership Plan (ESOP), and thus, it is interesting to try to uncover the ways in which the ESOP and the legacies of selfmanagement and communism may help to shape both the design and the effects of IWPs on the case. However, the main contribution of the chapter reflects key characteristics of the case and the nature of our data. Whereas most studies tend to examine cases in which attention is focused on one or two IWPs, our case is one in which several IWPs are present, notably offline teams, online teams, incentive pay, and employee ownership. In addition, there is much variation in participation in these practices among employees and also in membership in labor unions. Moreover, our survey has been designed so as to enable us to investigate the impact of IWPs on both firm and worker outcomes and to do so for a broader range of such outcomes than has typically been the case in much work in this area.2 Furthermore, while IWPs are designed to bolster firm performance, the empirical evidence on their actual economic effects is often quite mixed.3 Also, for the most part, there appear to be few studies that have endeavored to use such data to directly test hypotheses derived from economic theories such as the impact of IWPs on the nature and extent of peer monitoring and, more generally, on what the key mechanisms are by which different IWPs are expected to produce improvements in bottomline performance.4 Hence, in addition to providing evidence on worker outcomes, our survey has been designed to capture information for several

Do Innovative Workplace Practices Foster Mutual Gains?

25

other areas of interest that figure prominently in theoretical work, mainly by economists. These areas include the provision of discretionary effort by employees, the extent of horizontal monitoring, the degree of cohesion or cooperation within the organization, the extent to which employees are interested in product quality, and the degree to which employees have knowledge of and interest in their job and its relationship to company goals. In this econometric case study,5 we use these unusual measures that are derived from data we have collected using face-to-face surveys to investigate the impact of IWPs, such as employee ownership and participation in offline teams, on employee behaviors that are expected to affect firm outcomes.6 In turn, we are able to investigate whether IWPs deliver benefits and, if so, whether these accrue to employees or employers and to determine whether there are mutual gains (Appelbaum et al., 2000). Furthermore, given the importance many have attached to packages of IWPs (e.g., Ben-Ner & Jones, 1995), our data enable us to begin to see if there are varying effects to different combinations of IWPs and to see if conclusions reached using data for other cases (with more restricted combinations of IWPs) carry over to this case in which a rich set of IWPs is present. The structure of the chapter is as follows. In Section 2, we briefly review theory and empirical evidence. Then in Section 3, we overview the case study and its environment and also describe our most unusual data. In Sections 4 and 5, we examine the impacts of these practices on outcomes for workers and the firm. In Section 6, we summarize our findings and offer some concluding comments.

2. CONCEPTUAL FRAMEWORK AND EMPIRICAL LITERATURE We continue by reviewing theoretical and empirical literature as it mainly concerns relationships between IWPs and firm outcomes. This is interwoven with a brief review of the corresponding literature that concentrates on the impact of IWPs on worker outcomes. In both areas, we note that often the picture is muddled with theoretical work in particular tending to yield ambiguous predictions as to the impact of IWPs. As ours is not a theoretical contribution, here we merely highlight some key themes and focus on those theoretical issues that are most pertinent for our case. In particular, we discuss ways in which theory suggests that the key IWPs at our case, namely different kinds of teams, employee ownership, and incentive pay, might have

26

DEREK C. JONES AND SRECKO GOIC

potential impacts on important aspects of individual behavior. Although we concentrate on theoretical issues relating to the expected impact of individual IWPs such as teams, we also examine theory concerning the potential payoffs to combinations of IWPs. So far as the literature on firm outcomes is concerned, much literature recognizes that the organization of the firm matters for employee behavior (and thus ultimately on firm performance) and thus is necessary to go inside the ‘‘black box’’ that is usually ignored in simple micro-economic theory. Broadly speaking, we may distinguish mechanisms that focus on the effect of IWPs that emphasize financial participation, such as employee ownership, from those that stress employee involvement and skills, such as teams. Labor contracts are necessarily incomplete, providing opportunities for opportunistic behavior by employees and employers. As the interests of employers and employees may not be aligned, the degree of effort expended by employees is a variable, rather than something that is fully specified in a labor contract. In turn, a literature has emerged that shows how the degree of effort may be affected by particular IWPs such as teams and how these relationships may be affected by other aspects of the firm’s institutional set up with a substantial body of theoretical work concerning ‘‘team production,’’ and also a related literature on ‘‘teams.’’7 This literature on ‘‘team production’’ is concerned with the provision of incentives in situations where team output can be observed but individual productivity is unobservable. Dividing the output equally between team members leads to well-known free rider problems. The early literature concluded that the presence of an outside party is necessary for team production to function, either in performing a monitoring function (Alchian & Demsetz, 1972) or acting as a ‘‘budget breaker’’ (Holmstrom, 1982). Among subsequent works, the most relevant for our purposes is the literature stressing the role of horizontal monitoring.8 It has been argued that monitoring functions can be performed more efficiently by employees who observe each other’s work effort on a continuous basis. Similarly, it has been argued that teams can achieve first best outcomes once they develop norms against shirking. By ‘‘teams,’’ we refer to mainly work practices in online teams whereby employees work in groups rather than individually, and they also have some discretion over their working methods. According to this definition, to some degree, teams have internalized the monitoring function and thus may be less subject to the free rider problem associated with team production. Incentive problems are not the only problems faced by teams. Teamwork, both in online and in offline teams, usually involves regular meetings between team members that are not used in productive activities.

Do Innovative Workplace Practices Foster Mutual Gains?

27

Teams may also increase employee expectations about increased discretion. If these expectations are not fulfilled, it may create frustration and lower work morale (Heller, Pusic, Strauss, & Wilpert, 1998).9 Although there are distinct costs associated with working in teams, several authors have suggested that in many situations, the advantages outweigh the costs and many have stressed that teams are expected to be associated with improved supply of effort. Thus, the important question to understand is whether employees develop norms that protect from shirking and other undesirable side effects of team work. Disciplining of coworkers is likely to inflict psychological costs on employees, and therefore, it is not likely that they would engage in it without additional inducements (Freeman, Kruse, & Blasi, 2004). Increasing employee discretion without providing incentives for effort may produce detrimental effects: for instance, employees within autonomous teams that have wide discretion on production methods and the pace of work may use this discretion to work at a more leisurely speed. If production technology can be characterized mainly as team production, then pay structures tied to individual performance cannot be used. Two relatively widely used alternatives are company-based profit-sharing and employee share ownership schemes. Workers who receive more flexible pay are expected to be more committed to their firm, to work harder and smarter, and to be more likely to engage in the accumulation of firm-specific human capital. And there exists fairly consistent evidence that these schemes have modest but positive performance impacts.10 However, if employees are members of offline (rather than online) teams, then pay structures tied to individual performance can be used. It is also apparent that the willingness to engage in horizontal monitoring may vary between union and non-union environments. And the degree of cohesiveness or extent of cooperation between labor and management may be expected to vary between union and nonunion forms (although the direction of this effect may vary depending on one’s view of unions – contrast Freeman’s (not so) new view of unions with the traditional view). As briefly previously discussed, many stress the need for complementary initiatives. For example, Ben-Ner and Jones (1995) point out that employee involvement alone may not lead to enhanced business performance, especially in the absence of increasing return rights. The coupling of return rights with teamwork may provide the right incentives to engage in peer monitoring and also to withhold from opportunistic use of increased discretion. Several other authors, including Ichniowski et al. (1997) and MacDuffie (1995), suggest that combinations of various IWPs may be more

28

DEREK C. JONES AND SRECKO GOIC

effective than individual practices. However, this is not a universal position with, for example, Godard arguing against this conclusion. Godard (2004) also stresses the role that institutional conditions, notably union attitudes, might play for workplace innovations to be effective. And even among those who argue for benefits flowing from combinations of IWPs, disagreements exist on matters such as what exactly constitutes the best set of practices. Turning to the literature on worker outcomes, because there are very good recent reviews on this matter (e.g., Handel & Levine, 2004), our review will be quite brief. As with outcomes that are apt to be of main interest to employers, again there is no consistent picture in the literature as to whether IWPs are expected to deliver benefits to employees. One camp is often optimistic that workers (as well as employers) may benefit from IWPs. Thus, Appelbaum et al. (2000) argue that there are complementarities between IWPs that provide incentives, opportunities for participation and for skill formation. In such circumstances, they predict that there will be mutual gains as both employees and employers benefit. This view is shared by Freeman, Kruse, and Blasi (2006). By contrast, there are many who are much more pessimistic and who expect that workers may suffer in such high-performance workplace environments. For example, Godard (2004) and Ramsay, Scholarios, and Harley (2000) take this view. They expect that employees will not benefit materially from IWPs and that job stress will be apt to increase. In addition, as many have noted, IWPs may disappoint, and ultimately backfire, because of difficulties related to implementation.11

3. THE CROATIAN CONTEXT, THE CASE, AND THE DATA Although generalizing from case studies is always potentially a risky undertaking, in the case of a Croatian firm, one must be especially aware of the limitations that might be imposed arising from the heritage of the Yugoslav system of self-management as well as the transition process itself. Equally, because the research was undertaken more than 15 years after the end of self-management, it is clear that many workers at the case would have no direct experience of the self-management model. Although attitudes toward some IWPs (such as the ESOP) might be affected by experience of that earlier system, it is difficult to believe that such influences will be powerful after such a long passage of time. So far as the transition process

Do Innovative Workplace Practices Foster Mutual Gains?

29

itself is concerned, we note that ESOPs are not an uncommon institution in Croatia. In the only attempt to systematically survey the incidence of ESOPs in Croatia, Tipuric´ (2004) note that ESOP and ESOP-like models existed in 9.4% of enterprises. In their discussion of privatization Goic´, Zavrsˇ ak, and Brnabic´ (2006) also note the use of the ESOP model during privatization in Croatia. Hence, insofar as there were many other similar ESOP cases in Croatia during the transition period, the case firm, ADP Plastics (ADP), is far from unique in Croatia. Equally, it is also clear that the typical pattern of enterprise transition in Croatia did not involve an ESOP.12 Several kinds of data were collected from ADP. Most important are face-to-face data from surveys of workers. In addition, several interviews were conducted with diverse personnel, including ADP managers and union representatives. Finally, additional information was drawn from other sources such as annual reports and other internal documents. The firm that is listed on the Croatian stock market is quite important in the local economy. It is headquartered in Split and one major plant has been located there for more than 50 years. The firm has multiple plants and has established a solid market niche in the broad area of manufacturing plastic products, with a focus on high-quality plastic parts for the automobile industry. During the 1980s, when Croatia was part of the SFRY, what is now ADP was part of a larger group of companies that employed as many as 13,000 employees. ADP emerged as an independent company during transition and the disintegration of the SFRY when the original company was split into several parts. ADP is one of the few large industrial companies in Croatia that has managed to successfully navigate the problems posed by transition. An ESOP was established in 2001 as part of the privatization process; the aim was to transfer the bulk of ownership to employees and management and to avoid a takeover of the firm by a foreign company. Currently, about 53% of the firm is owned by the ESOP, and individual employees and managers still control a majority of votes (in fact, around 60%), with the majority of the balance owned by another corporate entity that is a long-time strategic partner of ADP. ADP is part of a larger group, namely the AD Plastik Group that includes ADP and several smaller companies some of which are located in Slovenia, Romania, and Russia. Recently, overall employment at ADP has averaged about 1300, whereas employment in the group has fallen from 2073 in 2003 to 1974 in 2005. More than 90% of ADP’s output is exported mainly to customers in Western Europe. The company is doing well and sales have tripled during

30

DEREK C. JONES AND SRECKO GOIC

the past three years, and data for several plants show that plants have recorded sustained growth over even longer periods. Investment is at high levels and is reported to average in recent years between 12% and 30% of sales. At the same time, the firm faces an environment that is increasingly challenging. Before 1990, most of its plants tended to face mainly domestic competition (within the SFRY) and had comfortable profit margins. More recently, managers perceive that these margins have become quite thin and that competitive pressures have grown, usually from overseas competitors.13 Many, including Appelbaum and Batt (1994), argue that globalization and regulatory changes that have increased competition have compelled firms to consider means for improving productivity by the application of advanced IWPs such as total quality management or autonomous teams. Such pressures appear to have played a part in the introduction of IWPs at our case. Indeed in interviews with managers, many mentioned that they faced growing competitive pressures in the 1990s in their product markets and tougher standards for product quality from their customers, including requirements for IOS certification. Accordingly, the firm has been required to make strategic responses to a fast-changing situation. Such pressures to change were felt especially strongly after the disintegration of the SFRY. They have been sustained in more recent times as Croatia prepares its candidacy for entry into the EU. Our interviews also indicated that the case uses a number of IWPs including offline teams, online teams, incentive pay, as well as the ESOP.14 During one site visit, we observed online teams in one department with each containing from 10 to 12 and mainly female workers. Offline teams are reported to have existed for many years. It appears that team practices at ADP, especially offline teams, were generally well regarded by management. So far as offline team membership is concerned, it appears that line management and top management usually select members. These teams serve mainly as ‘‘task forces’’ with tasks including developing new products, implementing new technologies, and general issues surrounding interplant and interdepartmental cooperation. ADP has other practices that provide for extensive information sharing, including quarterly meetings and a monthly newsletter. During quarterly meetings, the labor force learns confidential corporate information concerning new products, new strategies, and financial statements. In addition, employees receive a monthly newsletter in which they are informed about developments at the firm. Turning to financial matters, it appears that ADP provides higher starting wages than other comparable firms. In addition, ADP has a long history of

Do Innovative Workplace Practices Foster Mutual Gains?

31

financial participation, and the system of incentive pay has existed for many years, while the ESOP has been in place since 2001. At the same time, it was apparent that the firm did not have accurate information on the extent to which employees participated in many of these IWPs. Similarly, while a strong union is also present in the firm, estimates varied as to union density at the firm. To provide more accurate information on both the incidence of these IWPs and their effects on employee attitudes toward and behaviors resulting from these IWPs, we administered a survey in face-to-face interviews with individual employees. By surveying employees at two plants, we were fortunate to collect more than 470 surveys, which gave us an impressive response rate of more than 80%.15 All variables are defined in the Appendix. From Table 1, we see that the average worker at ADP is 39 years old and has worked at the firm for more than 11 years. About half of the workforce is female and about two in three are currently married. There is a wide spread in the highest level of educational attainment. Whereas 12.6% of workers have completed a four-year degree course, almost 10% of employees did not even complete grade school. Table 1.

Descriptive Statistics: Means (Standard Deviations).

Age Experience Tenure Female (%) Married (%) Single (%) Commuting time (minutes) Hours worked (months) Overtime hours (months) Vacation days (paid) Less than 8 years education (%) Completed grade school (%) Vocational level (%) High school (%) Some college (%) Completed four-year college (%) In ESOP (%) Incentive pay (%) In offline team (%) In online team (%) Of which in self-managed teams (%) In trade union (%)

39.0 11.8 11.5 49.1 68.9 26.1 32.7 181.8 12.7 24.4 1.7 7.8 23.7 45.9 8.2 12.6 55 45.6 67.2 42.8 58.1 50

9.91 9.61 9.58 0.5

23.1 14.48 16.9 6.34

32

DEREK C. JONES AND SRECKO GOIC

The descriptive statistics reported in Table 1 also indicate that there is wide dispersion among the labor force in their participation in IWPs. On average, between 42.8% and 67% of employees participated in one of the four key IWPs that we have identified. The highest rate of participation is in offline teams (67%), whereas fewer than 43% report that they are in an online team. However, for those who are in an online team, more than 58% of respondents report that they belong to teams that are self-managed. About half of the respondents were union members. The last row of Table 3 provides additional information on the incidence of participation rates in combinations of IWPs within ADP. Again, we observe much dispersion in participation rates in these practices. We see that 65 employees (about 14% of respondents) were in both offline teams and the ESOP; however, 137 (about 30%) were in neither plan. By comparison, 99 employees (about 21% of respondents) were in both the online teams and the ESOP, whereas only 80 employees (about 17%) did not participate in either plan. The last two columns provide information on the extent to which there was participation in all four practices by employees at ADP. Interestingly, 28 employees (about 7% of respondents) report that they were in both online and offline teams, as well as the ESOP, and also received incentive pay. By contrast, 53 workers (11%) report that they were in no practice. Or, in other words, almost 90% of employees participated in at least one IWP.

4. THE EFFECTS OF IWPS ON FIRM AND WORKER OUTCOMES: SIMPLE HYPOTHESIS TESTS To provide evidence on the impact of IWPs on firm and worker outcomes, we begin by comparing a wide range of outcomes for participants and nonparticipants and conducting simple hypothesis tests. As some of our interests are in what is a relatively new line of inquiry, the available literature with which to guide our research is rather limited. Hence, often, we employ alternative measures for key ideas – for example, to measure discretionary effort, we use measures of both absolute and relative effort. We proceed in two steps. First, in Table 2, the method is to focus on only one IWP at a time. We compare outcomes for those who are (are not) members of (i) offline teams; (ii) online teams; and (iii) the ESOP. Finally, for the subset of workers who are in an online team, we examine whether it matters to be in a self-directed team. Next in Table 3, and reflecting out interest in the effects of combinations of IWPs, we identify five interesting sets of IWPs within the case.16

Communication: ‘‘Management is usually open about sharing company information with employees at this company’’ ‘‘How often do you personally communicate about work issues with managers or supervisors in your work group or work team?’’ Proportion of employees replying ‘‘at least weekly’’ ‘‘How often do you communicate about work issues with managers or supervisors outside of your work group or work team within the firm?’’ Proportion replying ‘‘at least weekly’’ ‘‘How often do you personally communicate about work issues with workers outside of your work group or work team within the firm?’’ Proportion replying ‘‘at least weekly’’ ‘‘How often do you communicate about work issues with technical experts outside your work group/work team, e.g., engineers and technicians, in the firm?’’ Proportion replying ‘‘at least weekly’’

Empowerment: I have a lot to say about what happens on my job ‘‘My job allows me to take part in making decisions that affect my work’’

31.03

39.57

20.85

57.25

57.25

41.22

1.18

1.5

69.23

1.18

1.57

71.21

1.37

1.77

38.92

57.23

50.00

74.60

1.28

1.42

1.62

17.51

37.52

32.04

66.31

1.14

1.19

1.38

NonESOP member

29.58

45.75

26.62

40.13

32.41

61.39

73.23

43.6

1.15

1.19

1.44

1.26

1.4

1.52

45.45

64.77

63.22

76.67

1.43

1.63

1.86

18.70

32.79

29.51

70.24

1.15

1.23

1.28

Online Non-online Self-directed Non-selfteam team member online team directed team member member member

Mean

Outcomes: Based on One IWP.

Offline Non-offline ESOP team team member member member

Table 2.

Do Innovative Workplace Practices Foster Mutual Gains? 33

Teamwork/Peer monitoring: ‘‘I help my co2.19 workers when they need it’’ ‘‘To what extent have other employees at this 2.13 company taught you job skills, short cuts, problem solving, or other ways to improve your work?’’ 1 ¼ To a great extent, 4 ¼ Not at all ‘‘My effort at work is affected by the effort of 1.88 my co-workers’’ ‘‘The work of my co-workers affects my pay’’ 1.59 ‘‘If I saw a co-worker slacking off, I would say 1.99 something to that worker’’ Proportion of workers who have ever said 25.00 anything to a co-worker when they saw that worker slack off (%)

Effort: ‘‘How much effort do you put into your 3.67 work beyond what your job requires?’’ 1 ¼ None, 4 ¼ A lot Relative effort ¼ (effort put into a typical hour 5.40 or work – Effort put into a typical hour of watching TV) 0 ¼ Hardly any at all, 10 ¼ All your energy Days missed in the last year 2.28 Hours worked per week 46.95 ‘‘My effort at work affects my pay’’ 2.45

1.76 27.89

1.34 1.63 36.58

36.84

1.33 1.72

1.58

1.76

1.56 2.43

2.25

2.32

2.34

2.12

2.18

2.03 45.43 2,76

2.09

1.58 45.67 2,68

4.00

5.44

4.23

1.56 44.72 2,86

3.50

3.63

NonESOP member

3.52

Offline Non-offline ESOP team team member member member

Table 2. (Continued )

30.33

1.42 1.8

1.72

2.26

2.13

1.48 45.93 2,71

4.84

3.6

37.34

1.47 1.64

1.64

2.28

2.09

1.75 44.99 2,75

4.58

3.48

32.96

1.84

1.59

1.96

28.70

1.31 1.77

1.53

2.27

2.05

2.23 2.23

1.23 45.25 2,82

4.62

3.50

1.63 46.94 2.55

5.20

3.69

Online Non-online Self-directed Non-selfteam team member online team directed team member member member

Mean

34 DEREK C. JONES AND SRECKO GOIC

0.88 1.49 2.70

3.21

2.53

1.68 0.79

1.04 1.67 2.33 2.78

3.9

1.87 1.09 1.96 285

1.29

1.28

3.04 139

1.7

2.01

2.02 201

0.92

1.84

2.78

2.98

1.69 2.50

1.01

1.24

1.94

1.97 214

0.86

1.67

2.52

3.14

1.44 2.60

0.91

1.31

1.72

1.97 229

0.94

1.85

2.69

3.04

1.56 2.56

0.97

1.22

1.88

1.94 172

0.85

1.65

2.57

3.15

1.52 2.63

0.88

1.28

1.7

2.07 95

1.11

1.94

2.88

2.72

1.64 2.34

1.04

1.28

1.97

1.90 128

0.81

1.77

2.58

3.25

1.52 2.48

0.91

1.27

1.81

Notes: Unless otherwise indicated, each respondent is given four choices: 1 ¼ strongly agree; 2 ¼ agree; 3 ¼ disagree; and 4 ¼ strongly disagree. Mean differences significant at 1% level. Mean differences significant at 5% level. Mean differences significant at 10% level.

Job stress: ‘‘My job is stressful’’ Number of respondents

Intrinsic rewards: ‘‘My job makes good use of my knowledge and skills’’ ‘‘What I do at work is more important to me than the money I earn’’

Job satisfaction: ‘‘All in all, how satisfied would you say you are with your job?’’ 1 ¼ Very good, 5 ¼ Very bad

Trust: ‘‘I am treated fairly by the firm’’ ‘‘To what extent do you trust the management at the firm?’’ 1 ¼ To a great extent, 4 ¼ Not at all ‘‘In general how would you describe relations in your workplace between management and employees?’’ 1 ¼ Very good, 5 ¼ Very bad

Commitment: ‘‘I am willing to work harder than I have to in order to help this company succeed’’ ‘‘I would take almost any job to keep working for this company’’ ‘‘I would turn down another job for more pay in order to stay with this company’’

Do Innovative Workplace Practices Foster Mutual Gains? 35

2.93

2.3

68.72

24.12

32.37

54.55

55.36

23.28

30.25

2.97

2.85

2.72

67.86

2.45

2.27

2.14

68.33

2.99

2.77

2.11

2.34 Communication: ‘‘Management is usually open about sharing company information with employees at this company’’ 79.69 ‘‘How often do you personally communicate about work issues with managers or supervisors in your work group or work team?’’ Proportion of employees who replied ‘‘at least weekly’’ 64.06 ‘‘How often do you personally communicate about work issues with managers or supervisors outside of your work group or work team within the firm?’’ Proportion replying ‘‘at least weekly’’ 65.63 ‘‘How often do you personally communicate about work issues with workers outside of your work group or work team within the firm?’’ Proportion replying ‘‘at least weekly’’

Empowerment: I have a lot to say about what happens on my job ‘‘My job allows me to take part in making decisions that affect my work’’

Mean

Combinations of IWPs and Outcomes.

5.8.51

35.62

29.17

58.10

76.60

55.32

2.92

2.99

2.68

2.59

2.43

2.27

54.29

52.17

67.61

2.59

2.36

2.21

33.03

24.30

55.17

2.92

2.85

2.63

55.56

59.25

74.07

2.34

2.11

1.96

In both In offline Not offline In offline Not offline In online Not online In online No online teams and and no team and and not team and and no team and and not team and incentive incentive ESOP and ESOP ESOP incentive incentive ESOP ESOP incentive pay pay member member pay pay member member pay

Table 3.

30.61

19.15

59.18

3.08

3.02

2.79

In neither team, not in ESOP and no incentive pay

36 DEREK C. JONES AND SRECKO GOIC

Commitment: ‘‘I am willing to work harder than I have to in order to help this company succeed’’ ‘‘I would take almost any job to keep working for this company’’

35.56

20.00

2.38

2.74

2.78

2.63

2.73

2.33

2.35

1.98

2.32

1.93

2.68

2.27

2.75

38.66

2.43

1.71 44,41 2.95

2.09

2.20 48.34 2.27

2.2 44,89 2.94

4.09

17.24

2.42

5.72

32.72

3.99

13.79

1.77

Teamwork and Monitoring ‘‘To what extent have other employees 2.37 at this company taught you job skills, short cuts, problem solving, or other ways to improve your work?’’ 1 ¼ To a great extent, 4 ¼ Not at all ‘‘The work of my co-workers affects 2.18 my pay’’ ‘‘If I saw a co-worker slacking off, I 2.02 would say something to that worker’’ Proportion of workers who have ever 17.19 said anything to a co-worker when they saw that worker slack off (%)

6.25 Effort: Relative ¼ (effort put into a typical hour or work – Effort put into a typical hour of watching TV) 0 ¼ Hardly any at all, 10 ¼ All your energy Days missed in the last year 3.05 Hours worked per week 47.17 ‘‘My effort at work affects my pay’’ 2.44

55.56 ‘‘How often do you communicate y with technical experts outside of your work group or work team, such as engineers, technicians, within the firm?’’ Proportion replying ‘‘at least weekly’’

2.85

1.98

25.51

2.13

2.40

2.43

1.89 46.38 2.73

5.51

42.55

2.73

2.7

41.33

2.35

2.69

2.41

2.65 45,08 2.91

4.35

17.39

2.74

1.99

20.29

2.06

2.40

2.11

1.85 47.92 2.46

5.85

31.88

2.74

2.37

34.78

2.34

2.59

2.34

2.18 44,84 2.85

4.15

22.55

2.65

1.77

14.29

1.89

2.11

2.44

2.48 48.46 2.33

5.88

42.31

2.79

2.38

39.58

2.37

2.81

2.52

3.98 44,95 3.17

4.31

12.77

Do Innovative Workplace Practices Foster Mutual Gains? 37

2.75

3.26

2.61

2.39 3.27

2.22

2.66

1.97

2.03

2.86

2.02 137

2.62

2.19

1.85 65

3.17

2.91

Mean differences significant at 1% level. Mean differences significant at 5% level. Mean differences significant at 10% level.

Job stress: ‘‘My job is stressful’’ Number of respondents

Intrinsic rewards: ‘‘My job makes good use of my knowledge and skills’’ ‘‘What I do at work is more important to me than the money I earn’’

Job satisfaction: ‘‘How satisfied y say you are with your job?’’ 1 ¼ Very good, 5 ¼ Very bad

Trust: ‘‘I am treated fairly by the company’’ ‘‘To what extent do you trust the management at this company?’’ 1 ¼ To a great extent, 4 ¼ Not at all ‘‘In general how would you describe relations in your workplace between management and employees?’’ 1 ¼ Very good, 5 ¼ Very bad

‘‘I would turn down another job for more pay in order to stay with this company’’

Mean

1.91 58

2.95

2.16

2.04

2.75

2.21

2.20

2.98

2.07 204

3.24

2.36

2.57

3.35

2.76

2.58

3.09

1.98 99

3.07

1.97

2.15

2.91

2.40

2.27

2.99

1.99 80

3.31

2.39

2.56

3.15

2.59

2.56

3.19

1.81 72

3.03

2.15

2.04

2.82

2.30

2.33

3.06

2.06 125

3.16

2.42

2.50

3.25

2.65

2.59

3.09

1.88 28

2.85

2.04

1.81

2.54

2.07

2.19

2.81

In both In offline Not offline In offline Not offline In online Not online In online No online teams and and no team and and not team and and no team and and not team and incentive incentive ESOP and ESOP ESOP incentive incentive ESOP ESOP incentive pay pay member member pay pay member member pay

Table 3. (Continued )

2.02 53

3.38

2.40

2.77

3.36

2.68

2.70

3.31

In neither team, not in ESOP and no incentive pay

38 DEREK C. JONES AND SRECKO GOIC

Do Innovative Workplace Practices Foster Mutual Gains?

39

For example, the first combination is for those who participate both in an offline team and the ESOP; outcomes for that group are compared with those who participate in neither offline teams nor the ESOP. In the final set of such comparisons, we compare outcomes for those who participate in all core practices – in online teams, in offline teams, in the ESOP, and in the incentive pay scheme – with those employees who do not participate in any of these practices. In all exercises, findings are reported under several categories of outcomes. Whereas many of these dimensions are reasonably standard in the literature, namely empowerment, communication, commitment, trust, job satisfaction, intrinsic rewards and job stress, others are more novel, particularly effort and teamwork/peer monitoring.17 In all of these exercises, we use t-tests on means to determine if there are statistically significant differences in outcomes for employees who do/do not participate in the IWP or set of IWPs. We continue by first discussing the findings reported in Table 2. From the first two columns of Table 2, it is clear that membership in an offline team by itself tends to be associated with both enhanced worker and firm outcomes. For most categories, statistically significant differences exist between members and nonmembers for the bulk of questions within each block. Thus, both questions concerning empowerment indicate that offline team members believe that they are more empowered than those who are not in such teams. A similar picture prevails for four of the five questions concerning different ways of capturing communication. By using two measures of effort as well as data on hours worked, members in offline teams report that they work harder and longer than do their peers who are not in such teams. The data on offline teams in Table 2 also show that team members are more committed to the company than those who are not in such teams, are more trusting of the firm, have higher levels of job satisfaction, and they experience higher levels of intrinsic rewards. Furthermore, members of offline teams are much more likely than those who do not belong to offline teams to engage in peer monitoring – to say something to a worker who is slacking off. The only area in which offline team membership is not associated with a statistically significant enhanced worker of firm outcome is job stress where no differences are apparent. On the basis of the data reported in the three remaining three sets of columns in Table 2, we make two observations. First, in some cases, we observe that it is another practice that appears to be having a bigger effect on an outcome than does participation in an offline team. For example, compared to membership in an offline team, membership in a self-directed online team is observed to have larger effects on several outcomes

40

DEREK C. JONES AND SRECKO GOIC

concerning communications. Second, in the main, the general pattern observed for offline teams carries over to the other practices for which we report evidence. However, the evidence is not quite as compelling. Thus, membership in the ESOP does not reveal marked differences concerning commitment, and online team membership is not linked with many significant differences concerning teamwork/peer monitoring. Also while belonging to a self-directed team usually is associated with more teamwork and more peer monitoring, this is not found for all questions. At the same time in the bulk of instances, the findings reported in Table 2 indicate favorable worker and firm outcomes for participants. Consistent with those who hypothesize that IWPs will deliver benefits to both workers and firms, we also find that participants in all of the IWPs communicate more often than do nonparticipants with managers and supervisors outside of their work groups or teams and also communicate more often with workers outside of their work groups or teams. Participants in all practices are also found to put more effort into their work and are more satisfied with their work. Again, there is no evidence that stress levels differ for participants and nonparticipants in the other IWPs. In Table 3, we turn to combinations of IWPs. The evidence is broadly supportive of predictions that combinations of IWPs will be expected to be associated with better worker and firm outcomes (compared to situations when there is no participation at all). In the first eight columns of the table, we offer pairwise comparisons for outcomes for employees who are in a pair of IWPs alongside those who are in neither practice. Thus, from the first two columns, where we contrast outcomes for those who are (are not) in an offline team and an ESOP, we see consistent evidence that this pair of IWPs is associated with favorable outcomes for workers and for the firm. For most categories, statistically significant differences exist between members and nonmembers for the bulk of questions within each block. Workers who are in offline teams and in the ESOP report that they are more empowered, engage in more frequent communications, are more committed, have more job satisfaction, and have higher intrinsic rewards. They also work harder and undertake more peer monitoring. When we consider the impact of membership in offline teams together with incentive pay, the results are virtually identical to those found for participation in an offline team and an ESOP. By comparison with findings for the ESOP-offline combination, findings for the offline-incentive pay combination are slightly weaker only in the area of commitment. But again in most categories, statistically significant differences exist between members and nonmembers for the bulk of questions within each block. Workers who

Do Innovative Workplace Practices Foster Mutual Gains?

41

are in offline teams and who also receive incentive pay report that they are more empowered, engage in more frequent communications, have more job satisfaction, and have higher intrinsic rewards. They also work harder and undertake more peer monitoring. Very similar patterns of effects prevail concerning outcomes for the remaining pairs of IWPs. When we consider the impact of membership in online teams together with either participation in the ESOP or receipt of incentive pay, the results are virtually identical to those found for combinations involving participation in an offline team, especially the pair including participation in the ESOP. Moreover, the findings are virtually unchanged when we contrast outcomes for those employees who are in all practices with those who belong to none (findings are reported in the last two columns of Table 3). In sum, the evidence presented in Tables 2 and 3 provides good support for the general proposition that IWPs are associated with better worker and employer outcomes – IWPs can deliver mutual gains. The analysis indicates that when workers participate in IWPs, they develop a stronger sense of empowerment and achieve more intrinsic rewards from their jobs as well as higher levels of job satisfaction. In turn, these empowered and more satisfied workers tend to trust management more and develop stronger commitment to the firm. These attitudinal changes are accompanied by behavioral changes. When workers participate in IWPs, they tend to have more open and more frequent communication with management (as well as with their coworkers), exert more effort (shirk less), and engage in more peer monitoring (or horizontal monitoring). Finally, IWPs are not associated with increased stress. As such, IWPs appear to offer a strong point of hope, even in firms that face a difficult environment, such as those in transition countries.

5. THE EFFECTS OF IWPS ON FIRM AND WORKER OUTCOMES: MULTIVARIATE ANALYSIS To see if the conclusions yielded in the exercises reported in Section 4 carry over once we introduce additional controls, we estimate a variety of ordered probit models concerning diverse employer and employee outcomes. In essence, we estimate 3 sets of models for each of 13 outcomes when outcomes are grouped into 6 sets, including discretionary effort, monitoring, employee involvement, job satisfaction, and intrinsic rewards. In the baseline

42

DEREK C. JONES AND SRECKO GOIC

models, as well as controls for personal characteristics (tenure, age, and gender), we include only one IWP.18 Besides our four core IWPs (offline teams, online teams, ESOP, and incentive pay), for the subset of workers who are in offline teams, we also consider whether it matters to be in a self-directed team. We are also able to investigate the impact of union membership in a similar fashion. The remaining two sets of probits reflect our interest in the impact on worker and firm outcomes of combinations of IWPs. In the second set of exercises, we investigate the influence of several pairs of IWPs that we have previously examined using simpler methods. Reflecting our earlier theoretical discussion, these are constructed so as to combine membership in one type of team alongside some mechanism for financial participation. In the last sets of exercise, we include information on all four IWPs and then, in separate regressions, we also consider the effect of union membership as well as all four IWPs. In Tables 4–16 we report the coefficients that emerge from these exercises. The final task is to report selected marginal effects to see if the effects that we uncover are not only statistically but also economically significant.

5.1. Discretionary Effort The results reported in Tables 4 and 5 are for two alternative measures of discretionary effort – the measure used in Table 4 is a measure of absolute effort, whereas that in Table 5 attempts to get at relative effort.19 From the set of models reported in the first six columns of Table 4, we see that there is evidence that teams of both types enhance the provision of effort and that this is also the case for incentive pay. By contrast, neither union membership nor participation in an ESOP is found to have a statistically significant effect on the supply of discretionary effort. In the models reported in columns 7–10, we examine the impact of pairs of practices. We see that membership in an offline team enhances the supply of effort in both models (columns 9 and 10), although in these cases, there is no additional impact from financial participation. In the other two regressions, in one instance, online team membership is associated with more effort, other things equal, whereas in the other case, only incentive pay (and not online team membership) has a positive impact on the supply of effort. As in the most restricted models, neither union membership nor participation in an ESOP is found to matter for the supply of effort. The final set of models includes all IWPs. Reassuringly, we still find that offline team membership enhances the supply of effort, although now no other IWP or union membership has an effect.20

2

149.8 188

0.357 (0.192)

0.009 (0.012) 0.027 (0.188) 0.010 (0.040) 0.0001 (0.001)

3

317.0 378

0.277 (0.147)

0.016 (0.009) 0.045 (0.129) 0.034 (0.027) 0.001 (0.001)

4

304.5 367

0.117 (0.137)

0.015 (0.009) 0.059 (0.129) 0.018 (0.029) 0.0004 (0.001)

5

0.271 (0.150)

0.017 (0.009) 0.020 (0.139) 0.010 (0.030) 0.0001 (0.001) 0.254 (0.137)

7

0.043 (0.134) 321.9 278.3 385 328

0.017 (0.009) 0.020 (0.126) 0.031 (0.027) 0.001 (0.001)

6

0.183 (0.156)

11

0.214 (0.172)

0.016 (0.010) 0.134 (0.150) 0.018 (0.033) 0.001 (0.001) 0.078 (0.158) 0.499 (0.175)

12

0.216 (0.174) 0.157 (0.156) 0.076 0.011 0.058 (0.144) (0.159) (0.163) 285.3 260.5 247.2 345 311 297

0.019 (0.009) 0.119 (0.137) 0.005 (0.029) 0.0001 (0.001)

10

0.019 (0.010) 0.089 (0.147) 0.015 (0.032) 0.001 (0.001) 0.036 (0.154) 0.466 0.473 0.537 (0.153) (0.152) (0.171)

0.019 (0.009) 0.129 (0.140) 0.008 (0.029) 0.0001 (0.001)

9

0.070 (0.146) 282.9 281.8 335 339

0.018 (0.009) 0.028 (0.136) 0.008 (0.030 0.0002 (0.001) 0.256 (0.136)

8

Determinants of Discretionary Effort.

Notes: Discretionary effort is measured by responses to ‘‘How much effort do you put into your work beyond what your job requires?’’ (where 1 ¼ none and 4 ¼ a lot). Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

285.5 345

0.020 0.019 (0.009) (0.009) 0.032 0.116 (0.136) (0.137) 0.011 0.007 (0.030) (0.029) 0.0002 0.0001 (0.001) (0.001) 0.259 (0.136) 0.477 (0.152)

Log-like 283.0 No. of 335 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 4.

Do Innovative Workplace Practices Foster Mutual Gains? 43

403.6 175

0.051 (0.165) 0.110 (0.129) 0.003 (0.120)

0.022 (0.008) 0.220 (0.112) 0.002 (0.024) 0.0005 (0.001)

6

0.113 (0.136

0.025 (0.008) 0.219 (0.120) 0.009 (0.027) 0.0002 (0.001) 0.021 (0.120)

8

10

0.033 (0.134)

0.268 (0.129)

0.002 (0.144)

0.027 (0.009) 0.090 (0.129) 0.37 (0.028) 0.0005 (0.001) 0.270 (0.138) 0.391 (0.144)

11 0.025 (0.009) 0.110 (0.131) 0.043 (0.029) 0.001 (0.001) 0.261 (0.140) 0.345 (0.147)

12

0.057 (0.146) 0.001 (0.133) 0.347 0.326 0.344 (0.127) (0.137) (0.140) 758.8 684.3 657.7 315 288 272

0.247 (0.127)

0.028 0.021 (0.008) (0.008) 0.103 0.115 (0.123) (0.119) 0.013 0.016 (0.025) 0.0002 0.0002 (0.001) (0.001)

9

0.245 (0.128) 739.4 744.0 307 309

0.030 (0.008) 0.202 (0.110) 0.007 (0.027) 0.0003 (0.001) 0.022 (0.121)

7

0.250 (0.119) 827.4 722.5 344 301

0.024 (0.008) 0.224 (0.115) 0.009 (0.026) 0.0002 (0.001)

5

795.1 324

0.027 (0.008) 0.200 (0.116) 0.0003 (0.025) 0.001 (0.001)

4

810.6 338

0.035 (0.011) 0.198 (0.162) 0.022 (0.035) 0.0001 (0.001)

3

Determinants of Relative Effort.

Notes: Relative effort is measured as responses to how much effort and energy you expend on a typical hour of work less effort and energy expended on an hour watching TV, where each activity is measured on a 10-point scale and 0 ¼ hardly any at all and 10 ¼ all your energy. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

762.6 315

0.271 (0.127)

0.026 (0.008) 0.114 (0.119) 0.007 (0.025) 0.0003 (0.001)

0.029 (0.008) 0.227 (0.119) 0.001 (0.027) 0.0004 (0.001) 0.011 (0.120)

Log-like 741.3 No. of 307 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

2

1

Table 5.

44 DEREK C. JONES AND SRECKO GOIC

Do Innovative Workplace Practices Foster Mutual Gains?

45

In addition, and as in all other models reported in Table 4, we observe that, having controlled for tenure, it is older workers who report that they supply more discretionary effort. Turning to Table 5, the key finding is that, as with the results for absolute effort reported in Table 4, membership in an offline team is consistently found to enhance effort supply. This is found in all specifications, both those in which that IWP is included alone (column 2) or alongside other measures of financial participation (columns 9 and 10) and in the fully augmented models reported in the last two columns. In addition, and unlike findings reported in Table 4, in all models, we find that membership in an ESOP will enhance the provision of effort when effort is measured in this relative way.21 For this measure of effort, the fully augmented models provide evidence that some of the remaining IWPs, notably online team membership, also matter for effort supply, although this does not often show up in the more parsimonious specifications.22 As in Table 4, we find that older workers report that they work harder. In addition, the negative and significant coefficient on gender means that in many specifications, there is evidence that men believe that they are more likely than are women to work harder.23 5.2. Monitoring In Tables 6 and 7, we report evidence on the impact of IWPs concerning, respectively, whether respondents have engaged in horizontal monitoring and also their willingness to horizontally monitor. The results reported in Table 7 are particularly striking. There we find that membership in both online and offline teams is associated with a greater willingness to engage in horizontal monitoring. To some extent, this finding is mirrored for offline teams because, in some specifications reported in Table 6, we observe statistically significant and positive coefficients for that IWP.24 For the most part, no other IWP is found to affect monitoring. Also, the results for some controls reveal an interesting story. They indicate both that women are typically softer on monitoring (compared to men) and that older workers are apt to monitor more. 5.3. Employee Involvement In Tables 8 and 9, we turn to the first measure that relates to a worker outcome, namely empowerment (decision-making ability, reported in Table 8) or employee involvement (measured by my say in what happens

0.016 (0.010) 0.181 (0.147) 0.028 (0.033) 0.001 (0.001) 0.194 (0.148)

208.3 337

0.324 (0.161)

0.009 (0.010) 0.154 (0.148) 0.040 (0.031) 0.001 (0.001)

2

109.4 182

0.123 (0.208)

0.012 (0.014) 0.390 (0.206) 0.069 (0.046) 0.003 (0.002)

3

227.7 366

0.266 (0.159)

0.019 (0.010) 0.210 (0.141) 0.056 (0.030) 0.002 (0.001)

4

224.7 356

0.149 (0.149)

0.013 (0.010) 0.266 (0.140) 0.048 (0.031) 0.001 (0.001)

5

0.218 (0.173)

0.013 (0.010) 0.177 (0.152) 0.349 (0.033) 0.001 (0.001) 0.187 (0.151)

7

0.015 (0.144) 236.0 195.3 373 320

0.013 (0.009) 0.230 (0.140) 0.050 (0.030) 0.002 (0.001)

6

0.168 (0.167)

0.295 (0.163)

0.001 (0.001)

0.007 (0.010) 0.177 (0.152) 0.050

9

0.026 (0.158) 202.9 202.3 324 331

0.016 (0.010) 0.182 (0.148) 0.029 (0.033) 0.001 (0.001) 0.195 (0.149)

8

0.043 (0.155) 208.3 337

0.339 (0.161)

0.008 (0.010) 0.153 (0.148) 0.039 (0.031) 0.001 (0.001)

10

Determinants of Whether Have Engaged in Horizontal Monitoring.

0.11 (0.011) 0.226 (0.159) 0.048 (0.036) 0.001 (0.001) 0.164 (0.174) 0.200 (0.185)

12

0.222 (0.185) 0.113 (0.168) 0.036 0.622 (0.171) (0.175) 184.1 176.2 302 289

0.207 (0.183)

0.011 (0.011) 0.185 (0.159) 0.041 (0.035) 0.001 (0.001) 0.148 (0.170) 0.201 (0.182)

11

Notes: Whether one has engaged in horizontal monitoring is determined by responses to ‘‘Have you ever said anything to a co-worker when you saw that worker slack off?’’ where 1 ¼ yes and 0 ¼ no. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

Log-like 202.9 No. of 324 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 6.

46 DEREK C. JONES AND SRECKO GOIC

0.074 (0.167)

372.3 377

0.156 (0.124)

0.021 (0.008) 0.340 (0.115) 0.006 (0.025) 0.0005 (0.0008)

6

0.088 (0.144)

0.257 (0.008) 0.384 (0.0125) 0.003 (0.027) 0.000 (0.001) 0.345 (0.126)

8

0.045 (0.138)

11

0.067 (0.152)

0.025 (0.009) 0.354 (0.135) 0.008 (0.029) 0.00009 (0.0009) 0.240 (0.148) 0.342 (0.155)

12

0.100 (0.155) 0.093 (0.141) 0.105 0.058 0.073 (0.129) (0.143) (0.146) 351.8 310.2 294.3 360 324 307

0.21 (0.008) 0.317 (0.124) 0.000 (0.026) 0.000 (0.001)

10

0.025 (0.009) 0.354 (0.135) 0.008 (0.029) 0.00005 (0.0009) 0.255 (0.144) 0.455 0.477 0.333 (0.134) (0.133) (0.152)

0.019 (0.008) 0.311 (0.127) 0.003 (0.026) 0.000 (0.001)

9

0.067 (0.133) 335.8 347.8 346 354

0.024 (0.008) 0.359 (0.129) 0.005 (0.027) 0.000 (0.001) 0.337 (0.127)

7

0.123 (0.120) 395.4 330.4 398 339

0.024 (0.008) 0.345 (0.118) 0.003 (0.025) 0.0001 (0.0008)

5

Notes: Willingness to horizontally monitor is measured by responses to ‘‘If I saw a co-worker slacking off, I would say something to that worker,’’ where 1 ¼ strongly disagree and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

390.0 391

0.128 (0.132)

0.019 (0.008) 0.306 (0.119) 0.004 (0.025) 0.0003 (0.0008)

4

Determinants of Willingness to Horizontally Monitor.

0.026 (0.011) 0.457 (0.169) 0.017 (0.035) 0.0002 (0.001)

3

203.0 196

0.465 (0.132)

0.020 (0.008) 0.315 (0.124) 0.003 (0.026) 0.0002 (0.001)

2

352.1 360

0.025 (0.008) 0.380 (0.125) 0.005 (0.027) 0.0001 (0.0008) 0.341 (0.126)

Log-like 336.0 No. of 346 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 7.

Do Innovative Workplace Practices Foster Mutual Gains? 47

0.308 (0.130)

0.279 (0.141)

0.003 (0.008) 0.522 (0.123) 0.001 (0.027) 0.0002 (0.001) 0.329 (0.123)

8

0.249 (0.137)

11

0.130 (0.150)

0.001 (0.010 0.476 (0.135) 0.026 (0.029) 0.001 (0.001) 0.127 (0.145) 0.404 (0.153)

12

0.162 (0.152) 0.118 (0.139) 0.464 0.510 0.468 (0.128) (0.142) (0.145) 375.0 332.7 317.4 358 322 305

0.001 (0.008) 0.443 (0.122) 0.023 (0.025) 0.0005 (0.001)

10

0.0002 (0.008) 0.503 (0.133) 0.022 (0.028) 0.0005 (0.001) 0.152 (0.142) 0.459 0.461 0.369 (0.132) (0.131) (0.149)

0.005 (0.008) 0.419 (0.125) 0.003 (0.025) 0.003 (0.025)

9

0.524 (0.132) 363.2 371.6 346 352

0.001 (0.008) 0.521 (0.127) 0.014 (0.027) 0.0004 (0.001) 0.321 (0.124)

7

0.403 (0.119) 424.3 359.4 399 339

0.002 (0.008) 0.448 (0.113) 0.011 (0.024) 0.0002 (0.001)

6

Notes: Empowerment is measured by responses to ‘‘My job allows me to take part in making decisions that affect my work,’’ where 1 ¼ strongly disagree and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

405.2 378

0.183 (0.122)

0.004 (0.008) 0.441 (0.115) 0.008 (0.025) 0.0008 (0.001)

5

Determinants of Empowerment.

0.001 (0.008) 0.427 (0.117) 0.006 (0.024) 0.0001 (0.001)

4

417.3 392

0.416 (0.167)

0.009 (0.011) 0.456 (0.167) 0.001 (0.035) 0.0003 (0.001)

3

212.5 194

0.497 (0.130)

0.006 (0.008) 0.429 (0.122) 0.010 (0.025) 0.0002 (0.001)

2

381.5 358

0.003 (0.008) 0.524 (0.123) 0.008 (0.026) 0.0003 (0.001) 0.349 (0.122)

Log-like 212.5 No. of 346 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 8.

48 DEREK C. JONES AND SRECKO GOIC

0.103 (0.121)

0.011 (0.008) 0.715 (0.114) 0.027 (0.024) 0.001 (0.001)

6

0.361 (0.140)

0.012 (0.008) 0.722 (0.128) 0.013 (0.026) 0.0004 (0.001) 0.199 (0.113)

7

0.335 (0.136)

11

0.252 (0.149)

0.014 (0.009) 0.631 (0.136) 0.044 (0.029) 0.001 (0.001) 0.013 (0.143) 0.474 (0.152)

12

0.270 (0.151) 0.118 (0.137) 0.293 0.250 0.213 (0.126) (0.139) (0.142) 385.7 345.2 329.4 363 325 308

0.013 (0.008) 0.716 (0.124) 0.038 (0.025) 0.001 (0.001)

10

0.012 (0.008) 0.683 (0.134) 0.037 (0.028) 0.037 (0.028) 0.010 (0.140) 0.438 0.460 0.432 (0.131) (0.130) (0.149)

0.015 (0.008) 0.676 (0.126) 0.022 (0.025) 0.001 (0.001)

9

0.296 (0.129) 378.4 380.1 349 357

0.011 (0.008) 0.762 (0.125) 0.028 (0.027) 0.001 (0.001) 0.210 (0.121)

8

Notes: Employee involvement is measured by responses to ‘‘I have a lot to say about what happens on my job,’’ where 1 ¼ strongly disagree and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

0.251 (0.117) 437.3 371.6 404 342

0.016 (0.008) 0.682 (0.117) 0.031 (0.025) 0.001 (0.001)

5

415.8 383

0.391 (0.129)

0.012 (0.008) 0.660 (0.118) 0.013 (0.024) 0.0004 (0.001)

4

Determinants of Employee Involvement.

428.9 397

0.623 (0.169)

0.018 (0.011) 0.603 (0.168) 0.001 (0.035) 0.0005 (0.001)

3

214.2 196

0.490

0.016 (0.008) 0.709 (0.124) 0.030 (0.025) 0.001 (0.001)

2

388.4 363

0.014 (0.008) 0.769 (0.125) 0.017 (0.026) 0.0004 (0.001) 0.224 (0.121)

Log-like 381.0 No. of 349 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 9.

Do Innovative Workplace Practices Foster Mutual Gains? 49

50

DEREK C. JONES AND SRECKO GOIC

on my job, reported in Table 9). In several respects, the two measures yield similar results and findings. A clear picture emerges that all IWPs enhance employee participation, both alone and in combinations. This is perhaps most evident concerning the impact of offline teams. For both measures of employee participation, we find that in all specifications, offline teams lead to employees perceiving that they are more empowered. This pattern is also essentially repeated concerning ESOPs. Findings are also reasonably strong concerning the other two IWPs, the incentive system and online teams. In all specifications reported in Table 9, the incentive system is always found to be an IWP that enhances the average employees’ sense of empowerment; in Table 8, this is frequently the case as well. For online teams, the evidence of positive effects is a little spottier. The evidence reported in Tables 8 and 9 also indicates that union membership does not play a role in accounting for differences in perceived levels of participation. Finally, so far as employee characteristics are concerned, the coefficient on gender is consistently positive in both sets of findings, thus indicating that it is men, rather then women, who report that they have a greater say in their job, even after controlling for benefits for participation that flow from IWPs. Although cultural differences between men and women presumably play a role in accounting for this difference (and other gender differences in outcomes), an interesting line of inquiry would be to pursue the reasons for these differences more thoroughly.

5.4. Job Satisfaction In Table 10, we report findings on the relationships between IWPs and job satisfaction for our case, ADP. For membership in offline teams, participation in the ESOP, and receiving compensation in part through the incentive system, the evidence is very strong that each of these IWPs alone as well as in combinations is associated with enhanced job satisfaction. This is the finding for these three IWPs in all specifications. For online teams, the evidence is also strong, with coefficients on that variable not attaining customary levels of statistical significance only in the fully augmented models that are reported in columns 11 and 12.25 Although the pattern of the findings on job satisfaction thus mirrors some of the results reported for other outcomes for workers and the firm, there is an important difference in the factors that account for differences in job satisfaction. From Table 10, we see that membership in a labor union is also found to enhance job satisfaction – no similar effects of union membership

0.367 (0.176)

0.0001 (0.008) 0.571 (0.118) 0.712 (0.025) 0.001 (0.001)

6 0.002 (0.009) 0.601 (0.131) 0.086 (0.028) 0.002 (0.001) 0.229 (0.128)

8

11

Notes: Job satisfaction is measure by responses to ‘‘All in all, how satisfied would you say you are with your job?’’ where 1 ¼ very dissatisfied and 5 ¼ very satisfied. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

0.428 (0.167)

0.002 (0.009) 0.462 (0.147) 0.115 (0.031) 0.003 (0.001) 0.018 (0.153) 0.601 (0.168)

12

0.427 (0.169) 0.328 (0.148) 0.770 0.801 0.797 (0.138) (0.156) (0.159) 327.5 278.8 263.1 359 321 304

0.575 (0.149)

0.002 (0.008) 0.532 (0.130) 0.089 (0.027) 0.002 (0.001)

10

0.002 (0.009) 0.477 (0.144) 0.105 (0.030) 0.003 (0.001) 0.014 (0.150) 0.514 0.526 0.545 (0.142) (0.141) (0.165)

0.005 (0.008) 0.407 (0.132) 0.059 (0.027) 0.001 (0.001)

9

0.701 (0.140) 321.3 330.0 345 353

0.630 (0.154)

0.007 (0.009) 0.493 (0.134) 0.058 (0.028) 0.001 (0.001) 0.220 (0.129)

7

0.546 (0.124) 389.5 318.7 400 338

0.324 (0.125)

376.0 379

0.660 (0.140)

0.004 (0.008) 0.560 (0.120) 0.064 (0.025) 0.001 (0.001)

5

Determinants of Job Satisfaction.

0.004 (0.008) 0.455 (0.121) 0.492 (0.025) 0.001 (0.001)

4

380.9 393

0.010 (0.011) 0.628 (0.177) 0.062 (0.037) 0.001 (0.001)

3

188.6 193

0.576 (0.139)

0.006 (0.008) 0.492 (0.128) 0.064 (0.026) 0.001 (0.001)

2

343.4 359

0.010 (0.008) 0.594 (0.129) 0.057 (0.027) 0.001 (0.001) 0.255 (0.127)

Log-like 334.1 No. of 345 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 10.

Do Innovative Workplace Practices Foster Mutual Gains? 51

52

DEREK C. JONES AND SRECKO GOIC

were observed for other outcomes. There is also a novel finding concerning one of the controls, namely for tenure, for which the net positive coefficient implies a negative effect on job satisfaction for longer tenured workers. As with empowerment, men report more job satisfaction than do women, other things equal.

5.5. Communications, Commitment, and Teamwork In Table 11, we report findings for the impact of IWPs on communications. Again, there is evidence that IWPs play a role in accounting for variation in this employee outcome. This is especially the case for both types of teams for which, except in the fully augmented models, team membership is always found to matter at customary levels of statistical significance. There is also some weaker evidence that participation in the incentive plan is associated with greater frequency of communications between respondents and other groups. By contrast, participation in the ESOP (and union membership) plays no role. As for controls, the positive coefficient on gender means that women report that they are apt to communicate less frequently than men. The evidence reported in Table 12 offers quite strong and additional support for those who hypothesize that high-performance workplace practices such as teams and mechanisms for financial participation will deliver improved outcomes for employees. In all specifications reported in Table 12, it is the case that membership in offline teams and participation in the ESOP is associated with employees reporting higher levels of commitment to the firm. The evidence is also reasonably strong that membership in online teams leads to improved commitment, while, for incentive pay, there are also some indications that this too plays a role. By contrast, union membership has no effect one way or the other. As the net tenure coefficient is positive, our results also indicate that workers who have longer tenure have weaker commitment to the organization. Compared to some of the findings emerging from the other tables, the evidence on the impact of IWPs on the likelihood of employees engaging in team-like behavior (as they assist their peers in various kinds of on the job training), reported in Table 13, is relatively weak. As with many other outcomes, it is offline team membership that is found to have the clearest and most consistent effects, with membership in an offline team enhancing the likelihood of employees assisting other employees in job training. Although the evidence for other IWPs is patchier, nevertheless, such evidence is present and is especially evident in the preferred specification (column 11) when,

0.326 (0.136)

485.7 339

0.156 (0.127)

0.257 (0.143)

0.108 (0.135)

0.004 (0.008) 0.400 (0.129) 0.014 (0.028) 0.001 (0.001)

10

0.154 (0.142)

450.7 317

0.117 (0.135)

0.646 0.642 (0.135) (0.0134)

0.003 (0.008) 0.386 (0.132) 0.014 (0.028) 0.001 (0.001)

9

446.6 315

0.008 (0.008) 0.579 (0.128) 0.009 (0.028) 0.0004 (0.001) 0.379 (0.127)

8

449.5 315

0.007 (0.008) 0.549 (0.132) 0.008 (0.028) 0.0002 (0.001) 0.384 (0.128)

7

443.4 312

0.005 (0.008) 0.543 (0.122) 0.001 (0.027) 0.0001 (0.001)

6

422.7 298

0.035 (0.145)

0.126 (0.151)

0.003 (0.009) 0.435 (0.138) 0.006 (0.029) 0.0003 (0.001) 0.206 (0.145) 0.533 (0.150)

11

393.7 281

0.101 (0.154) 0.177 (0.143) 0.003 (0.148)

0.004 (0.009) 0.434 (0.140) 0.016 (0.030) 0.001 (0.001) 0.161 (0.149) 0.551 (0.154)

12

Notes: Communications are measured by responses to ‘‘How often do you personally communicate about work issues with managers or supervisors outside of your work group or work team within the firm,’’ where 1 ¼ never, 2 ¼ rarely, 3 ¼ monthly, 4 ¼ weekly, and 5 ¼ daily. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

453.4 320

0.153 (0.131)

0.007 (0.008) 0.541 (0.125) 0.006 (0.028) 0.0001 (0.001)

5

Determinants of Communications.

0.005 (0.008) 0.502 (0.125) 0.002 (0.027) 0.00003 (0.001)

4

479.1 336

0.542 (0.170)

0.003 (0.011) 0.387 (0.167) 0.006 (0.035) 0.0001 (0.001)

3

268.2 188

0.656 (0.133)

0.005 (0.008) 0.400 (0.129) 0.012 (0.028) 0.0005 (0.001)

2

451.1 317

0.010 (0.008) 0.581 (0.128) 0.012 (0.28) 0.0005 (0.001) 0.388 (0.127)

Constant Log-like 449.8 No. of 315 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 11.

Do Innovative Workplace Practices Foster Mutual Gains? 53

209.7 192

0.174 (0.172)

403.5 386

0.267 (0.132)

(0.008) 0.221 (0.118) 0.031 (0.025) 0.0004 (0.001)

0.019

4

391.8 372

0.046 (0.123)

(0.008) 0.259 (0.116) 0.046 (0.026) 0.001 (0.001)

0.020

5

8

0.285 (0.144)

10

0.164 (0.139)

0.096 (0.153)

0.014 (0.009) 0.154 (0.138) 0.097 (0.030) 0.002 (0.001) 0.127 (0.147) 0.502 (0.158)

12

0.124 (0.155) 0.004 (0.142) 0.395 0.495 0.481 (0.130) (0.145) (0.148) 363.7 315.6 299.7 354 317 300

(0.008) 0.199 (0.123) 0.069 (0.026) 0.001 (0.001)

0.015

11

(0.009) 0.186 (0.135) 0.094 (0.030) 0.002 (0.001) 0.109 (0.144) 0.499 0.485 0.512 (0.136) (0.135) (0.155)

(0.008) 0.174 (0.126) 0.052 (0.026) 0.001 (0.001)

0.021 0.018

9

0.433 (0.133) 347.7 360.9 341 348

(0.008) (0.008) 0.243 0.278 (0.128) (0.124) 0.054 0.075 (0.028) (0.028) 0.001 0.002 (0.001) (0.001) 0.313 0.303 (0.127) (0.126)

0.019 0.016

7

0.280 (0.120) 410.8 342.9 393 334

(0.008) 0.266 (0.114) 0.045 (0.025) 0.001 (0.001)

0.017

6

Determinants of Commitment.

Notes: Commitment is measured by responses to ‘‘I am willing to work harder than I have to in order to help this company to succeed,’’ where 1 ¼ strongly disagree and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

368.4 354

0.522 (0.134)

(0.011) 0.301 (0.169) 0.050 (0.037) 0.001 (0.001)

(0.008) 0.195 (0.123) 0.058 (0.026) 0.001 (0.001)

(0.008) 0.290 (0.124) 0.058 (0.028) 0.001 (0.001) 0.321 (0.123)

Log-like 353.0 No. of 341 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

3

0.022 0.023

2

0.021

1

Table 12.

54 DEREK C. JONES AND SRECKO GOIC

0.013

2

3 0.014

4

226.9 197

437.2 393

418.3 379

0.258 (0.140)

0.011 (0.008) 0.053 (0.121) 0.084 (0.027) 0.002 (0.001) 0.004 (0.121)

8

0.172 (0.136)

0.328 (0.130)

0.322 (0.129)

(0.008) 0.059 (0.120) 0.091 (0.025) 0.003 (0.001)

0.015

10

0.142 (0.148)

(0.008) 0.082 (0.130) 0.101 (0.028) 0.003 (0.001) 0.231 (0.0140) 0.440 (0.148)

0.014

11 0.016 (0.009) 0.163 (0.133) 0.114 (0.029) 0.003 (0.001) 0.199 (0.143) 0.424 (0.151)

12

0.151 (0.151) 0.175 (0.137) 0.220 0.242 0.153 (0.127) (0.140) (0.143) 398.6 354.2 333.0 361 325 308

(0.008) 0.008 (0.122) 0.083 (0.025) 0.002 (0.001)

0.013

9

0.185 (0.129) 384.6 393.7 349 355

0.010 (0.008) 0.015 (0.124) 0.078 (0.026) 0.002 (0.001) 0.014 (0.123)

7

0.148 (0.118) 445.1 377.0 400 342

(0.008) 0.147 (0.112) 0.071 (0.024) 0.002 (0.001)

(0.008) 0.069 (0.114) 0.080 (0.025) 0.002 (0.001)

0.113 (0.121)

0.015

6

0.014

5

Determinants of Teamwork.

Notes: Team work t is measured by responses to ‘‘To what extent have other employees at this company taught you job skills, short cuts, problem solving, or other ways to improve your work?’’, where 1 ¼ not at all and 4 ¼ to a great extent. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

400.1 361

0.005 0.009 (0.008) (0.008) (0.010) (0.008) 0.59 0.057 0.113 0.078 (0.120) (0.120) (0.163) (0.115)    0.77 0.085 0.085 0.080 (0.026) (0.025) (0.035) (0.025) 0.002 0.002 0.002 0.002 (0.001) (0.001) (0.001) (0.001) 0.006 (0.121) 0.345 (0.128) 0.120 (0.163) 0.257 (0.130)

Log-like 385.6 No. of 349 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 13.

Do Innovative Workplace Practices Foster Mutual Gains? 55

56

DEREK C. JONES AND SRECKO GOIC

beside offline team membership, teamwork is also facilitated by membership in the ESOP. However, and rather surprisingly, we also find that online team membership tends to undermine team-like behavior. The evidence also indicates that tenure and age play significant roles in accounting for differences in the propensity for employees to engage in team-like behaviors. Thus, the net positive coefficient implies that workers with longer tenure are apt to engage in less team-like behavior, as are older workers.

5.6. Intrinsic Rewards The findings reported in Table 14 and 15 offer evidence on the role of IWPs concerning intrinsic rewards. In Table 14 this is measured by how well employees regard their job as making use of their skills and knowledge, whereas in Table 15 we look at the importance of what workers do at work relative to what they earn. For both measures of intrinsic rewards, again, we find that the HR practice that appears to enhance this outcome in the eyes of employees is membership in most specifications in offline teams, although other IWPs are found to be statistically significant in the fully augmented specifications reported in columns 11–12. However, when intrinsic rewards are measured by knowledge and skill (Table 14), membership in online teams arguably plays an even more consistent role than does offline team membership. In both cases, IWPs that provide for financial participation typically do not appear to have much effect, although the ESOP variable is statistically significant in the specification reported in column 8. So far as controls are concerned, for this outcome, gender appears to be playing the biggest role.

5.7. Trust The last outcome we investigate is ‘‘trust’’ for which we report findings in Table 16. Perhaps, surprisingly, in view of our earlier findings, it is policies that promote financial participation that loom largest in the specifications reported in Table 16. The final task in the section is to compute and report marginal effects to see if the effects that we uncover are not only statistically but also economically significant. In order not to overwhelm the reader (and to keep the number of tables that we report to a manageable number), we do this on a selective basis. One set of illustrative exercises is to compute selective marginal effects concerning the willingness to horizontally monitor. When

220.6 195

0.163 (0.168)

0.009 (0.016) 0.127 (0.166) 0.045 (0.036) 0.002 (0.001)

3

445.5 393

0.114 (0.129)

0.010 (0.008) 0.227 (0.115) 0.015 (0.024) 0.0005 (0.001)

4

432.1 379

0.157 (0.121)

0.011 (0.008) 0.227 (0.114) 0.027 (0.025) 0.001 (0.001)

5

0.061 (0.140)

(0.008) 0.209 (0.125) 0.004 (0.026) 0.0001 (0.001) 0.344 (0.123)

0.014

7

0.141 (0.117) 451.5 376.5 400 340

0.011 (0.008) 0.238 (0.111) 0.024 (0.024) 0.001 (0.001) 0.366 (0.124)

6

0.040 (0.136)

0.315 (0.131)

(0.008) 0.150 (0.123) 0.033 (0.025) 0.001 (0.001)

0.015

9

0.226 (0.130) 381.6 401.1 347 354

(0.008) 0.203 (0.121) 0.018 (0.027) 0.0005 (0.001)

0.013

8

0.006 (0.149)

(0.008) 0.153 (0.131) 0.024 (0.028) 0.001 (0.001) 0.283 (0.140) 0.173 (0.148)

0.015

11 0.012 (0.009) 0.118 (0.134) 0.029 (0.028) 0.001 (0.001) 0.289 (0.143) 0.186 (0.151)

12

0.018 (0.151) 0.164 (0.137) 0.171 0.218 0.214 (0.126) (0.139) (0.142) 405.9 356.4 340.7 360 323 306

0.295 (0.130)

(0.008) 0.154 (0.120) 0.043 (0.025) 0.001 (0.001)

0.014

10

Determinants of Intrinsic Rewards (Knowledge and Skills).

Notes: Intrinsic rewards is measured by responses to ‘‘My job makes good use of my knowledge and skills,’’ where 1 ¼ strongly disagree and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

406.9 360

(0.008) (0.008) 0.212 0.154 (0.121) (0.120) 0.010 0.038 (0.026) (0.025) 0.0003 0.001 (0.001) (0.001) 0.355 (0.0122) 0.314 (0.129)

0.016

0.016

Log-like 383.2 No. of 347 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

2

1

Table 14.

Do Innovative Workplace Practices Foster Mutual Gains? 57

2

3

191.6 196

373.6 395

358.8 381

0.010 (0.126)

0.009 (0.008) 0.328 (0.119) 0.023 (0.026) 0.001 (0.001)

5

0.164 (0.123) 379.8 402

0.001 (0.008) 0.290 (0.116) 0.031 (0.025) 0.001 (0.001)

6

330.8 341

0.008 (0.144)

0.001 (0.008) 0.291 (0.129) 0.010 (0.027) 0.001 (0.001) 0.142 (0.127)

7

0.018 (0.139)

11

0.066 (0.152)

0.007 (0.009) 0.256 (0.138) 0.015 (0.029) 0.001 (0.001) 0.021 (0.149) 0.492 (0.157)

12

0.019 (0.155) 0.045 (0.142) 0.185 0.171 0.188 (0.131) (0.143) (0.147) 341.5 313.1 295.5 363 325 308

0.002 (0.008) 0.245 (0.125) 0.029 (0.026) 0.001 (0.001)

10

0.001 (0.009) 0.249 (0.135) 0.021 (0.028) 0.001 (0.001) 0.050 (0.145) 0.419 0.407 0.399 (0.135) (0.134) (0.153)

0.003 (0.008) 0.259 (0.128) 0.016 0.026 0.001 (0.001)

9

0.213 (0.134) 336.6 335.9 348 357

0.0004 (0.008) 0.266 (0.125) 0.027 (0.027) 0.001 (0.001) 0.132 (0.126)

8

Notes: Intrinsic rewards t is measured by responses to ‘‘What I do at work is more important to me than the money I earn,’’ where 1 ¼ strongly disagree y and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

342.5 363

0.046 (0.133)

0.002 (0.008) 0.301 (0.120) 0.018 (0.025) 0.001 (0.001)

4

Determinants of Importance of Work to Me (Relative to Money).

0.002 0.004 0.002 (0.008) (0.008) (0.011) 0.274 0.244 0.147 (0.125) (0.125) (0.170) 0.019 0.025 0.047 (0.027) (0.026) (0.036) 0.001 0.001 0.002 (0.001) (0.001) (0.001) 0.143 (0.125) 0.424 (0.133) 0.372 (0.171)

Log-like 337.9 No. of 348 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 15.

58 DEREK C. JONES AND SRECKO GOIC

0.014 (0.008) 0.274 (0.122) 0.034 (0.026) 0.0005 (0.001) 0.040 (0.121)

388.8 360

0.228 (0.129)

0.014 (0.008) 0.205 (0.121) 0.046 (0.026) 0.001 (0.001)

2

4

214.8 196

0.132 (0.166)

422.3 393

6

0.034 (0.122)

0.008 (0.008) 0.260 (0.122) 0.054 (0.027) 0.001 (0.001) 0.013 (0.122)

8

0.175 (0.130)

0.009 (0.008) 0.213 (0.122) 0.062 (0.026) 0.001 (0.001)

10

0.206 (0.151)

0.009 (0.008) 0.168 (0.132) 0.070 (0.028) 0.001 (0.001) 0.091 (0.141) 0.156 (0.149)

11 0.012 (0.009) 0.151 (0.134) 0.070 (0.029) 0.001 (0.001) 0.113 (0.144) 0.168 (0.152)

12

0.203 (0.153) 0.082 (0.138) 0.509 0.580 0.578 (0.130) (0.143) (0.145) 381.0 343.8 329.5 360 324 307

0.322 (0.140)

0.194 (0.131)

0.014 (0.008) 0.213 (0.122) 0.047 (0.026) 0.001 (0.001)

9

0.526 (0.132) 373.0 380.9 348 354

0.366 (0.142)

0.014 (0.008) 0.194 (0.125) 0.036 (0.026) 0.001 (0.001) 0.020 (0.123)

7

0.401 (0.120) 427.2 371.6 400 341

0.016 0.010 (0.008) (0.008) 0.262 0.263 (0.115) (0.113) 0.040 0.045 (0.025) (0.025) 0.001 0.001 (0.001) (0.001)

5

413.0 380

0.403 (0.133)

0.023 0.014 (0.011) (0.008) 0.297 0.183 (0.167) (0.116) 0.043 0.034 (0.035) (0.025) 0.001 0.001 (0.001) (0.001)

3

Determinants of Trust.

Notes: Trust is measured by responses to ‘‘I am treated fairly by the company,’’ where 1 ¼ strongly disagree and 4 ¼ strongly agree. Statistically significant at 1%. Statistically significant at 5%. Statistically significant at 10%.

Log-like 381.0 No. of 348 observations

ESOP

Union

Incent

Selfmanonline

Offline

Online

Tenure square

Tenure

Gender

Age

1

Table 16.

Do Innovative Workplace Practices Foster Mutual Gains? 59

60

DEREK C. JONES AND SRECKO GOIC

this is done, we find that the magnitudes of the effects typically are economically significant but not that large. Thus, if I am not a member of an offline team (rather than a member), then the probability of my strongly disagreeing with the statement concerning my willingness to horizontally monitor is calculated to be 1.9 percentage points higher (category 4), 10 percentage points higher (category 3), 6.5 percentage points lower (category 2), and 5.4 percentage points lower (category 1). As we have already noted, the evidence reported in Table 7 also suggests that women are less likely than men to be willing to monitor. Furthermore, we find that, compared to a man a woman has a 5.6 percentage points lower probability of strongly agreeing with the statement concerning willingness to horizontally monitor (category 1), and a 2.1% higher probability of strongly disagreeing with the statement concerning willingness to horizontally monitor (category 4). In sum, the evidence presented in Tables 4–16 reinforces findings from the simpler hypothesis-testing evidence reported in Tables 2 and 3. Often, there is pretty solid support for the broad hypothesis that IWPs are expected to be associated with better worker and employer outcomes. The analysis indicates that when workers participate in IWPs, they develop a stronger sense of empowerment and achieve more intrinsic rewards from their jobs as well as higher levels of job satisfaction. In turn, these empowered and more satisfied workers tend to trust management more and they develop stronger commitment to the firm. These attitudinal changes are accompanied by behavioral changes. When workers participate in IWPs, they tend to have more open and frequent communication with management (as well as with their coworkers). This leads them to exert more effort (to shirk less) and to engage in more peer monitoring (or horizontal monitoring). Finally, IWPs are not associated with increased stress. Although the evidence indicates that all IWPs have beneficial effects for both worker and firm outcomes, the most consistent findings that emerge from the multivariate analysis are for the favorable effects of offline teams. In addition to offline teams, we also find that other IWPs have favorable outcomes. Therefore, adoption of sets of IWPs that include offline teams and other practices such as financial incentives will yield benefits to both employees and firms.

6. CONCLUSIONS, FINDINGS, AND IMPLICATIONS In this chapter, we undertake various exercises including estimating various ordered probits models to investigate the impact of IWPs separately and in

Do Innovative Workplace Practices Foster Mutual Gains?

61

combination for a case located in Croatia. The most clear and compelling finding is that among the different IWPs, it is membership in offline teams that most often yields favorable worker outcomes for both workers and firms. As such, our findings are somewhat stronger than those contained in another recent study of offline teams in which several effects were found to be short lived (see Jones & Kato, forthcoming). But in addition to offline teams, we also find that other IWPs usually have favorable outcomes. As such, our findings are consistent with other econometric case studies of employee involvement including studies of online teams (e.g., Hamilton, Nickerson, & Owan, 2003; Jones, Kalmi, & Kauhanen, 2010). We also find that participation in sets of IWPs that include offline teams and other practices such as financial incentives yields benefits to both employees and firms. Membership in online teams (and to a lesser extent in offline teams, in an ESOP, and a flexible compensation system) is associated with workers who perceive that they are more empowered, satisfied, committed, trusting, and communicative. In turn, this results is their working harder and their being more apt to engage in peer monitoring. These conclusions consistently emerge from several exercises including our estimating various ordered probits models and when we investigate the impact of IWPs separately and when we investigate the impact of pairs and larger combinations of IWPs. In understanding why it is that among the several IWPs it is membership in offline teams that is associated most often with favorable outcomes, we conjecture that at ADP, this reflects the key role played by managers in selecting participants. Whereas for other IWPs, it is other factors, notably technology (e.g., in online teams) or tenure (being employed at ADP when the ESOP was introduced) that drives participation, for offline teams, it is line management and top management that decide who will be team members. Our findings strongly suggest that managers make very good use of their knowledge of their personnel. At the same time, in future work, we would plan to examine the broader question of what determines which workers participate in particular practices in more depth and why, for the case of offline teams in particular, participation in these IWPs is not even more widespread at ADP. One implication of our findings is to provide support for those who argue that IWPs will produce mutual gains and deliver benefits to both workers and firms. In this general sense, ours is not an original finding – while the overall body of evidence is ambiguous, some do argue that increasingly there is a body of evidence that suggests that, when properly introduced, IWPs may have beneficial effects.26 Thus our findings might be viewed as

62

DEREK C. JONES AND SRECKO GOIC

contributing to what might be viewed as an emerging consensus. But among that body of evidence, our findings may carry more weight than findings derived from some other studies. They are derived from a single econometric case, and thus, we avoid the problems of firm heterogeneity that plague attempts in many studies to disentangle relationships between IWPs and firm or worker outcomes. Moreover, in investigating relationships between IWPs and particular worker outcomes, notably absolute and relative work effort and peer monitoring, our approach arguably includes measures that improve over those used in most previous studies. As such, our findings may also be viewed as helping to identify some of those channels that might be playing the most important roles in accounting for the ultimate improvements in business performance that many associate with IWPs. In turn, sometimes our findings differ in important particulars from findings contained in other recent studies. We find that all four IWPs are often associated with favorable worker and firm outcomes. This differs, for example, from Freeman et al. (2006) who report that being a member of an ESOP was often associated with negative outcomes.27 In our case, the beneficial effects of ESOPs may reflect the particular historical circumstances attending the introduction of the ESOP. But perhaps, the legacy of self-management plays a role as might the coexistence of a strong labor union, as hypothesized by Godard (2004). Furthermore, we find that there are a number of interesting points concerning the role played by different controls. In particular, as others have found (e.g., Datta Gupta, & Eriksson, 2004), there are powerful effects of gender with women perceiving that they are less empowered and found to be less willing to engage in peer monitoring. Also different controls are found to play differing roles for different IWPs. In the main, these differences concerning controls do not appear to have figured prominently in previous work. It is also important to emphasize that the firm and workers that we investigate in this case tend to be different from those covered in most of the existing literature. Our study is one of the first to report findings on these diverse issues for workers in a transition economy. It is reassuring to find that support for optimistic hypotheses and evidence in the received literature concerning the impact of IWPs, which were largely based on firms and workers with other characteristics, also carry over to our case. Indeed, arguably, our findings provide some of the clearest evidence that IWPs can deliver benefits to all parties, even to firms that do not operate in the most advanced market economies and in economies where legacies might be expected to hinder the effectiveness of certain IWPs.

Do Innovative Workplace Practices Foster Mutual Gains?

63

NOTES 1. For example, Jones and Klinedinst (2006) report the diffusion of such practices in Bulgaria. 2. Most evidence that uses country surveys understandably has tended to concentrate on worker rather than firm outcomes. This is also true of most case studies that examine outcomes collected from employee surveys. By contrast, our survey is unusual insofar as it enables us to investigate possible links between various work practices and employee outcomes (including commitment, employee involvement, and job satisfaction) and outcomes of most interest to firms, such as peer monitoring and the supply of effort. There are other studies, including some of those produced by the shared capitalism project (e.g., Freeman et al., 2006), that take a similar tack although they tend not to go as deep within the black box as does this chapter. Also, there is an important and often neglected body of work that has used employee surveys to look at firm and worker outcomes in firms with very different organizational structures, notably worker-owned firms. Studies include work by Rhodes, Long, Greenberg, Ros for US firms and Bradley and Gelb for firms elsewhere. 3. Note that the bulk of evidence on economic outcomes has been derived from firm-level, cross-industry surveys. Such studies necessarily have shortcomings such as the inability to control for variation in IWPs within large multiplant firms. Equally most surveys of this evidence find that normally most innovative IWPs such as profit sharing and employee ownership are beneficial to firm performance. However, not all assessments are so optimistic – see, for example, Ben-Ner, Jones, and Han (1996). In part, assessments differ because of the varying reliability of findings. In turn, this reflects the different types of econometric methods used as well as the nature of the underlying data. 4. There is a neglected history of economists trying to measure some of these variables, including Stafford and Duncan (1980). Ros (2001, 2003) is a notable recent exception in attempting to provide evidence on economic hypotheses such as effort supply by using data derived from worker surveys although, unfortunately, his findings are restricted in part because of small sample sizes. More recent work of this kind includes papers emerging from the shared capitalism project (e.g., Freeman et al., 2004; Kruse et al., 2003). 5. For a discussion of this approach and related concepts including insider econometrics, see Bartel, Ichniowski, and Shaw (2004) and Jones, Kalmi, and Kauhanen (2006). 6. As such, we continue procedures that were developed in earlier work, beginning in 2000 for US firms. See, for example, Jones, Kato, and Weinberg, 2003. 7. Heywood, Jirjahn, and Tsertsvadze (2002) also make a similar distinction, although using slightly different terminology. 8. For other related literature, see the insightful discussion in Dow (2003, ch. 8.5). 9. Two recent econometric case studies of the impact of teams are Hamilton et al. (2003) who look at online teams, whereas Jones and Kato (forthcoming) investigate the impact of offline teams. Both find favorable firm outcomes. 10. For company-level evidence, see, for example, Jones and Kato (forthcoming).

64

DEREK C. JONES AND SRECKO GOIC

11. For example, Jones and Kato (forthcoming) document that initially there were positive productivity effects at the case they study but that these benefits were dissipated, arguably because of the failure to introduce a balanced set of IWPs. 12. For additional information on labor market institutions and industrial relations in Croatia, see Gregurek (2001). 13. Some plants are located in metropolitan areas (including one on Zagreb). Although we do not focus on those plants in this study, we note that they have faced a somewhat different situation from the plants we study. Two of them have aggressively entered emerging product markets such as wireless and broadband. Employment at these plants grew at an extraordinarily rapid clip between 1994 and 1999, and both had strong profit positions. However, some executives at these divisions expressed uncertainty about whether these new markets would be sustainable over the long haul – a view that unfortunately proved to be far-sighted. 14. Although studies of IWPs are scarce in Croatia, Goic et al. find that IWPs are less common in Croatia than in other Western countries. 15. The questionnaire builds on core questions that have been used in previous work with which one of the authors has been associated, for example, Jones et al. (2003). Thus, the survey is ‘‘customized’’ to reflect specific features of the HR set-up at a specific firm. Most questions use Likert-type scales and solicit employee responses in worker outcome areas such as job satisfaction and trust and on issues that are more likely to be regarded as relating to firm outcomes such as willingness to engage in peer monitoring and the provision of discretionary effort. 16. Of course, this is not an exhaustive list of combinations or categories. We could, for example, make comparisons with those in intermediate categories – such as participating in only one IWP. 17. Although many of these dimensions are frequently examined in the literature (e.g., Cappelli & Neumark, 2001; Batt, 2004; Freeman et al., 2000), unsurprisingly, the particular language used in questions varies. As already noted, the specific wording in our questionnaire builds on core questions that have been used in previous work with which one of the authors has been associated, for example, Jones et al. (2003). Also the questionnaires used in those earlier studies include some novel adaptations of questions in some areas, including the provision of effort. 18. In unreported regressions, we also include controls for several categories of education and wages (which is also measured as a categorical variable). By and large, findings are not sensitive to these modifications. 19. This variable is normalized with reference to effort spent watching TV. 20. On the basis of Wald tests, the preferred specifications are those reported in columns 11 and 12. 21. Again Wald tests lead us to select the preferred specifications as those reported in columns 11 and 12. 22. It is important to remember that for these and other results, because different numbers of observations are usually involved in the different specifications reported in the tables, that findings from different estimates are not usually directly comparable. 23. Given the variable definitions, an alternative interpretation provided by a colleague who is fan of the NY Yankees and the NY Giants is that men may work harder than women when watching TV.

Do Innovative Workplace Practices Foster Mutual Gains?

65

24. In Table 6, Wald tests lead us to select the specifications reported in columns 9 and 10 as preferred; in Table 7, the preferred specifications are those in columns 11 and 12. 25. These are the preferred specifications. 26. For the US studies include those by Appelbaum, Bailey, Berg, and Kalleberg (2000), Black and Lynch (1997), Freeman et al. (2000), Helper (1998), Ichniowski, Kathryn, and Prennushi (1997), Levine and Tyson (1990), MacDuffie (1995), and Neumark and Cappelli (1999). For the interesting case of Japan, see Jones and Kato (1995) and Kato (2000). This includes previous work by one of the authors where, together with Kato, some exercises similar to those reported in this chapter and for comparable outcomes are reported for a group of cases in central New York (see Jones et al., 2003). In addition, see again for Finland, Kalmi and Kauhanen (2008) and papers emerging from the Shared Capitalism project, such as Freeman et al., 2004 and 2006. 27. Similar findings concerning the ambiguous impact of ESOPs have been found in many other studies – see, for example, Logue and Yates (2001).

ACKNOWLEDGMENTS The research reported in this chapter was supported by a grant from WDI, University of Michigan. The authors wish to thank the many individuals at the case who gave generously of their time, comments from Jeffrey Pliskin and participants at a Seminar in Nijmegen, in 2006, notably Richard Long and Eric Poutsma, and in Split in 2008 as well as research assistance by Qi Ge.

REFERENCES Alchian, A., & Demsetz, H. (1972). Production, information costs, and economic organization. The American Economics Review, 62, 777–795. Appelbaum, E., Bailey, T., Berg, P., & Kalleberg, A. (2000). Manufacturing advantage: Why high performance work systems pay off. Ithaca, NY: Cornell University Press. Appelbaum, E., & Batt, R. (1994). The new American workplace: Transforming work systems in the United States. Ithaca, NY: ILR Press. Bartel, A., Ichniowski, C., & Shaw, K. (2004). Using ‘‘Insider Econometrics’’ to study productivity’’. American Economic Review, 94(2), 217–223. Batt, R. (2004). Who benefits from teams? Comparing workers, supervisors and managers. Industrial Relations, 43, 183–212. Ben-Ner, A., & Jones, D. C. (1995). Employee participation, ownership, productivity: A theoretical framework. Industrial Relations, 34(4), 532–554.

66

DEREK C. JONES AND SRECKO GOIC

Ben-Ner, A., Jones, D. C., & Han, T.-S. (1996). The productivity effects of participation in control and in economic returns: A review of the econometric studies. In: U. Pagano & R. Rowthorn (Eds), Democracy and efficiency in the economic enterprise (pp. 209–244). Routledge. Black, S. E., & Lynch, L. M. (1997). How to compete: The impact of workplace practices and information technology on productivity. NBER Working Paper no. 6120. Blanchflower, D., & Freeman, R. (1997). The attitudinal legacy of communist labor relations. Industrial and Labor Relations Review, 50(3), 438–459. Cappelli, P., & Neumark, D. (2001). Do ‘‘high-performance’’ work practices improve establishment-level outcomes? Industrial and Labor Relations Review, 54(4), 737–775. Datta Gupta, N., & Eriksson, T. (2004). New workplace practices and the gender wage gap. Working paper, 04-18. Aarhus Department of Economics, Denmark. Dow, G. (2003). Governing the firm: Workers control in theory and practice. Cambridge, UK: Cambridge University Press. Freeman, R., Kleiner, M., & Cheri, O. (2000). The anatomy of employee involvement and its effects on firms and workers. NBER Working Paper no. 8050. Freeman, R., Kruse, D., & Blasi, J. (2004). Monitoring colleagues at work: Profit sharing, employee ownership, broad-based stock options and workplace performance in the United States. Advances in the Economic Analysis of Participatory and Labor Managed Firms, 7, 101–127. Freeman, R., Kruse, D., & Blasi, J. (2006). Do workers gain by sharing? Presented at LERA, January, Boston. Godard, J. (2004). A critical assessment of the high performance paradigm. British Journal of Industrial Relations, 42(2), 349–378. Goic´, S., Zavrsˇ ak, D., & Brnabic´, R. (2006). Country Report-Croatia: Extended country Reports-Croatia. In: J. Lowitzsch (Ed.), The PEPPER III Report: Promotion of employee participation in profits and enterprise results in the new member and candidate countries of the European Union. Inter-University Centre at the Institute for Eastern European Studies, Free University of Berlin, The European Commission’s Directorate General Employment, Industrial Relations and Social Affairs (pp. 111–124). San Francisco CA, Berlin/Rome: Kelso Institute for the Study of Economic Systems. Gregurek, M. (2001). Transition in Croatia: Pace, level and effects of privatization. Fourth International Conference on ‘Enterprise in Transition.’ Proceedings, Faculty of Economics Split, Hvar, May 24–26, pp. 2003–2024. Hamilton, B. H., Nickerson, J. A., & Owan, H. (2003). Team incentives and worker heterogeneity: An empirical analysis of the impact of teams on productivity and participation. Journal of Political Economy, 111(3), 465–498. Handel, M., & Levine, D. (2004). Editors’ introduction: The effects of new work practices on workers. Industrial Relations, 43(1), 1–43. Heller, F., Pusic, E., Strauss, G., & Wilpert, B. (1998). Organizational participation: Myth or reality? Oxford: Oxford University Press. Helper, S. (1998). Complementarity and cost reduction: Evidence from the auto supply industry. NBER Working Paper no. 6033. Heywood, J. S., Jirjahn, U., & Tsertsvadze, G. (2002). Getting along with colleagues – Does profit sharing help or hurt? Kyklos, 58(4), 557–573. Holmstrom, B. (1982). Moral hazard in teams. Bell Journal of Economics, 13, 324–340.

Do Innovative Workplace Practices Foster Mutual Gains?

67

Ichniowski, C., Kathryn, S., & Prennushi, G. (1997). The effects of innovative work practices on productivity: A study of steel finishing lines. American Economic Review, 87(3), 291–313. Jones, D., Kato, T., & Weinberg, A. (2003). Managerial discretion, business strategy, and the quality of jobs: Evidence from medium-sized manufacturing establishments in central New York. In: E. Appelbaum, A. Bernhardt & R. Murnane (Eds), Low-wage America: How employers are reshaping opportunity in the workplace. New York: Russell Sage Foundation. Jones, D. C., Kalmi, P., & Kauhanen, A. (2006). Human resource management polices and productivity: New evidence from an econometric case study. Oxford Review of Economic Policy, 22, 526–538. Jones, D. C., Kalmi, P., & Kauhanen, A. (2010). Teams, performance-related pay, profitsharing and productive efficiency: Evidence from a food-processing plant. Industrial and Labor Relations Review, 63(4), 606–626. Jones, D. C., & Kato, T. (1995). The productivity effects of employee stock-ownership plans and bonuses: Evidence from Japanese panel data. American Economic Review, 85(3), 391–414. Jones, D. C., & Kato, T. (forthcoming). The effects of offline teams on firm performance: evidence from an econometric case study. IZA Working Paper and Forthcoming Industrial and Labor Relations Review. Jones, D. C., & Klinedinst, M. (2006). Advances in the economic analysis of participatory and labor managed firms, 9, 177–209. Kalmi, P., & Kauhanen, A. (2008). Workplace Innovations and employee outcomes: Evidence from Finland. Industrial Relations, 47(3), 430–459. Kato, T. (2000). The recent transformation of participatory employment practices in Japan. NBER Working Paper no. 7965. Kruse, D., Blasi, J., Freeman, R., Buchele, R., Scharef, A., Rogers, L., & Mackin, C. (2003). Motivating employees in ESOP firms: Human resource policies and firm performance. In: V. Perotin & A. Robinson (Eds) Advances in the economic analysis of participatory and labor managed firms (Vol. 8). Bingley, UK: Emerald Group Publishing Ltd. Levine, D. I., & Tyson, L. D. A. (1990). Participation, productivity and the firm’s environment. In: A. S. Blinder (Ed.), Paying for Productivity (pp. 183–236). Washington, DC: Brookings Institution. Logue, J., & Yates, J. (2001). The real world of employee ownership. Ithaca, NY: Cornell University Press. MacDuffie, J. P. (1995). Human resource bundles and manufacturing performance: Organizational logic and flexible production systems in the world auto industry. Industrial and Labor Relations Review, 48, 197–221. Neumark, D., & Cappelli, P. (1999). Do ‘high performance’ work practices improve establishment-level outcomes? NBER Working Paper W7374. Ramsay, H., Scholarios, D., & Harley, B. (2000). Employees and high-performance work systems: Testing inside the black box. British Journal of Industrial Relations, 34(4), 501–532. Ros, A. (2001). Profits for all? The cost and benefits of employee ownership. Huntington, NY: Nova Science Publishers. Ros, A. (2003). Do ESOPs motivate employees? Worker effort, monitoring and participation in employee-owned stock ownership plans. The determinants of the incidence and the effects of participatory organizations. Advances in the Economic Analysis of Participatory and labor managed Firms, 7, 83–103.

68

DEREK C. JONES AND SRECKO GOIC

Stafford, F., & Duncan, G. J. (1980). The use of time and technology by households in the United States. In: R. Ehrenburg (Ed.), Research in labor economics (Vol. 3, No. 1981, 41pp). Greenwich, CT: JAI. Tipuric´, D. (Ed.) (2004). ESOP i hrvatsko poduzec´e. Zagreb: Sinergija Nakladnisˇ tvo.

APPENDIX. VARIABLE DEFINITIONS HR Variables Offline team. ¼ 1 when a member of an offline team and ¼ 0 when not in such a team. Online team. ¼ 1 when a member of an online team and ¼ 0 when not in such a team. Selfmanonline. ¼ 1 when a member of a self-managed online team and ¼ 0 when not in such a team. Union. ¼ 1 when a member of a union and ¼ 0 when not a member. Incent. ¼ 1 when receiving incentive pay and ¼ 0 when not in such a plan. ESOP. ¼ 1 when a member of an ESOP and 0 when not in such a plan.

Controls Age. Measured in years Gender. 1 ¼ male; 0 ¼ female. Tenure. Years worked at the case firm. Exper. Years in the labor market.

Outcome Variables All outcome variables are defined in Tables 2–16. (a) Most are measured using a four-point scale. (b) Unless otherwise indicated, each respondent is given four choices: 1 ¼ strongly

disagree; 2 ¼ disagree; 3 ¼ agree; and 4 ¼ strongly agree. For example, this is the case for both of the empowerment questions. (c) But for some questions a different four-point scale is used. For example, discretionary effort is measured by responses to ‘‘How much effort do you put into your work beyond what your job requires?’’ where 1 ¼ none and 4 ¼ a lot. (d) Relative effort (e.g., Table 5) is measured using a 10-point scale. (e) Job satisfaction and one of the trust measures use a five-point scale.

THE PERFORMANCE IMPACT OF FINANCIAL PARTICIPATION: SUBJECTIVE AND OBJECTIVE MEASURES COMPARED Panu Kalmi and Christina Sweins 1. INTRODUCTION One of the major challenges for organizations is to increase productivity in order to compete in national and world markets. Another challenge managers encounter is to attract and retain key workers, and ensure the involvement and commitment of workers as part of the route to high performance (e.g., Guest, Michie, Conway, & Sheehan, 2003). One prominent way to reach these goals is the adoption of high performance workplace practices, including financial participation. The traditional rationale for financial participation schemes is that they align employee and employer objectives in maximizing firm value, by inducing them to work harder and smarter (e.g., Fakhfakh & Perotin, 2000; Bryson & Freeman, 2010). In many settings, employees possess private information on how to improve production processes. Some organizational practices, such as self-managed teams, rely heavily on employees utilizing their private information. However, they may not want to share or use this information unless given incentives to do so. One useful way of giving these incentives, and

Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 69–88 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011007

69

70

PANU KALMI AND CHRISTINA SWEINS

thus complementary with organizational practices such as teamwork, is financial participation (Levine & Tyson, 1990; Ben-Ner & Jones, 1995; Appelbaum, Bailey, Berg, & Kalleberg, 2000). However, these positive outcomes may not always materialize. In the literature, two reasons for skepticism in particular have been presented. Because the benefits of financial participation are shared collectively but the costs of effort are born privately, some researchers have argued that financial participation would generate free rider behavior on the employees’ part (Alchian & Demsetz, 1972). Another criticism, derived from expectancy theory (Vroom, 1964), is that employees may not sufficiently see the connection between the effort they exert and the materialized rewards. This ‘‘line of sight’’ problem (e.g., Boswell & Boudreau, 2001) is pronounced in financial participation programs where collective effort is rewarded and there are many intervening factors. Both free rider and line of sight problems rely on an individualistic perspective of decision making. Many authors have argued that taking social and other pressures regarding preferences into account changes these conclusions. The free riding problem may be limited when employees have a financial stake in the firm and employees are able to detect free riders via mutual monitoring and sanction them (e.g., Weitzman & Kruse, 1990; Ben-Ner & Jones, 1995; Freeman, Kruse, & Blasi, 2010). Moreover, as pointed out, e.g., by Mitchell, Lewin, and Lawler (1990), financial participation has a potentially large symbolic value for employees. Rousseau and Shperling (2003) argue that financial participation helps to create a mutual understanding between employer and employees where both sides recognize mutual commitments. In this case, free riding may no longer be a dominant response to financial participation schemes. Similarly, the line of sight problem may not be that pressing if the effects of financial participation do not depend on purely individualistic calculations. Mitchell et al. (1990) and Rousseau and Shperling (2003), among others, have stressed the role of financial participation in educating employees about the financial conditions of the employer, which may also enhance work motivation. Even though there have been very many studies related to the performance impact of financial participation, there are only a few studies that focus on a subjective assessment of financial participation on firm performance, and a complete lack of research comparing objective and subjective impact measures. Arguably, both objective and subjective measures can give valuable perspectives on workplace performance. Therefore, it is interesting to see to what extent these measures are

The Performance Impact of Financial Participation

71

congruent. Another issue that is tested in this chapter is whether it matters for the consistency of objective and subjective measures from whom the subjective measures are collected from. The single respondent bias identified by Gerhart, Wright, Mahan, and Snell (2000) suggests that it is generally better to collect subjective data from a larger number of respondents than just one respondent. However, if there are systematic differences in the quality of the data different respondents provide then it may no longer be optimal to aim for large pools of respondents. In our analysis, we use data from two different sources. In the analysis of objective performance data, we use standard financial data from 1999 to 2003. The analysis of objective measures of performance is done by using matched pair comparisons, with two comparison firms from the same sector for each firm that uses a financial participation scheme. The purpose of these comparisons is to obtain a measure for the firm-specific difference between the financial participation firms and their matched comparison firms, and use this as an objective measure of the impact of financial participation. The subjective data come from a questionnaire targeted for firms using the financial participation scheme. The subjective measure of the impact of the financial participation scheme is derived from the assessments of the respondents. In this chapter, we use data derived both from chairmen of the personnel funds (which are at least supposed to be more knowledgeable about the personnel fund system) and from employees in general. The aim is to compare if there are differences between these two respondent groups concerning the relation between subjective and objective impact measures. The contributions of this chapter are as follows: The subjective measures of the impact of financial participation, where the causal relation is embedded in the question, have not previously been compared to objective impact measures.1 Therefore, our chapter broadens the literature comparing subjective and objective performance measures. Moreover, we explicitly test whether it is better to rely on an informed single respondent (in our case, the personnel fund chairman) or on a large pool of less informed respondents. We start by reviewing the relevant literature measuring performance impact of financial participation. Next we describe the features of the unique Finnish financial participation scheme, namely personnel funds. These combine elements from deferred profit-sharing schemes and employee share ownership plans (ESOPs). Next we explain the methods adopted in our study before going to the core findings, summary, and conclusions.

72

PANU KALMI AND CHRISTINA SWEINS

2. CONCEPTUAL FRAMEWORK 2.1. Different Approaches to Impact Measurement We start by briefly describing studies using two different approaches. Economists usually measure impact with objective measures gathered from financial statements. Studies assessing the productivity impact of financial participation schemes usually use measures of output (typically value added or sales) and regress them on inputs (capital and labor) and augment the standard specification by including a measure for a financial participation scheme (usually a dummy variable).2 The coefficient of the dummy variable is then interpreted as the impact of the scheme. Other studies focus on differences in profitability, comparing firms with and without these schemes (e.g., Fitzroy & Kraft, 1986; Mitchell et al., 1990; Magnan & St-Onge, 2005). A smaller number of studies have essentially used the same identification strategy, but with subjective performance ratings. Several studies (e.g., Blanchflower & Oswald, 1988; McNabb & Whitfield, 1998; Bryson & Freeman, 2010) have used manager responses from various rounds of the British Workplace Employment Relations Survey (WERS), where the respondents are asked to rate the financial performance of their firm relative to the competitors. This is then used as the dependent variable and is regressed on the financial participation indicator. Other studies employ a rather different strategy of assessing the performance effects of financial participation, by asking the respondents directly about the performance effects of financial participation. The first of such studies in our knowledge was Poole and Jenkins (1990) who collected data from both managers and employees in a sample of UK firms using various financial participation schemes. Two studies collected subjective responses from managers – Long (2000) from Canadian firms using profitsharing and Kalmi, Pendleton, and Poutsma (2005) from the listed sector in four European countries – from firms using various financial participation schemes. Sweins, Kalmi, and Hulkko-Nyman (2009) report the results from data collected from employees of Finnish firms that have personnel funds.

2.2. The Advantages and Disadvantages of Different Types of Measures In the literature of performance measurement, there have been debates on the relative merits between the use of subjective and objective performance measurement. The advantages of subjective performance measures are that

The Performance Impact of Financial Participation

73

they are often less costly to collect and more easily available than subjective data (Forth & McNabb, 2008). The latter is because often objective financial information is not available at the level of establishment or information is commercially sensitive and therefore not reported. Subjective measures may also attract high response rates. Unfortunately, subjective measures also have distinct disadvantages. The respondents may overstate the performance. For instance, in WERS a large majority of the respondents claims to have above average performance while only a few firms maintain it is below average (e.g., Bryson & Freeman, 2010). When we consider about direct question of impact, the respondents may overstate the impact, if they have a stake in the continuance of practice. For instance, employees value a compensation scheme that generates them extra compensation, and thus they may be inclined to overstate its effects in a survey, at least if they perceive that the survey results are used in management decision making. Subjective performance variables suffer from the ‘‘common method variance’’ problem (Spector, 2006), which means that individuals may tend to give similar ratings to items asked by the same method (such as a survey questionnaire). Subjective measures are often limited because of their ordinal nature, and thus direct quantification of the impact is not possible (Forth & McNabb, 2008). Subjective performance measures are often gathered as cross-sections, and the lack of longitudinal data hinders the assessment of causality. A particular type of problem highlighted in the literature investigating the links between human resource management and performance is the single respondent bias (Gerhart et al., 2000). In essence, if there are idiosyncratic differences among the performance raters that are random, having more respondents for each observation reduces the measurement error. However, if the differences among the raters are not random but systematic, then the situation is different. If group A provides poorer quality ratings than group B, then even if including the responses from group A may reduce random error, the increase in systematic error may outweigh these benefits. In other words, if information is gathered from a broader group of employees the risk of having misinformed respondents is increased. Turning to objective measures, it is usually thought that their greatest advantage is that they are free from subjectivity, i.e., they are harder to manipulate and therefore ‘‘harder’’ data. For some, this indicates that they should be preferred whenever available.3 However, the absence of subjectivity may not be strictly true, although it is likely to be more limited than in the case of subjective data. For instance, when the statements are used for taxation purposes, firms try to minimize the tax burden, and this

74

PANU KALMI AND CHRISTINA SWEINS

might skew the income statements. Sometimes the bias may arise from differing accounting standards (Machin & Stewart, 1990). There is also a different kind of problem with objective measures that arises from the impact being assessed indirectly. In this case, any conclusions about causality depend on the model being correctly specified (i.e., not suffering from omitted variables bias). For instance, if firms using financial participation are systematically better than their comparison firms in some dimension (for instance, better managed), then the performance difference that is attributed to financial participation may in fact reflect other, unmeasured characteristics of management. These comments, of course, also apply to the WERS type subjective performance measures. However, the subjective measures where causality can be directly inferred from the questions do not in principle suffer from this problem.

2.3. Results from Earlier Research and Hypotheses The productivity effects of financial participation schemes have been a heavily researched area. Most of the studies tend to find positive effects of financial participation on productivity, although some studies cannot reject the null hypothesis of no productivity effects.4 There are fewer studies focusing on profitability, but those that do (e.g., Fitzroy & Kraft, 1986; Mitchell et al., 1990; Magnan & St-Onge, 2005) indicate that financial participation schemes are associated with higher profitability. Based on the previous literature that has found evidence of the positive effects of profit sharing on performance, our first hypothesis is: Hypothesis 1. Firms that have profit-sharing plans are likely to have higher measures of objective performance than those without profitsharing plans. As discussed above, there have been two alternative research designs involving subjective measures of performance. The results using WERS measures have been mixed. Blanchflower and Oswald (1988) found no significant association between profit sharing and subjective performance measures, whereas McNabb and Whitfield (1998) find that profit sharing and employee ownership have positive and statistically significant associations with the subjective performance measure when combined with participatory HRM practices. Using more recent data, Bryson and Freeman (2010) find that the subjective performance measure is correlated with employee share ownership but not with other shared compensation schemes.

The Performance Impact of Financial Participation

75

Probably more relevant for our purpose is the second class of studies, where the respondents were asked directly to evaluate the impact of financial participation schemes. These have often indicated relatively strong links between financial participation and productivity. Long (2000) studied Canadian firms with profit-sharing schemes. Ninety percent of his respondents, who were CEOs of these firms, indicated that the schemes had positive performance effects. However, other studies have found more moderate effects. In Poole and Jenkins (1990, p. 24), managers regarded financial participation schemes as much more effective in being tax-efficient means of compensation or sharing benefits equitably. For the item Incentive for greater productivity, managers gave an average rating of 3.3 for financial participation schemes on a scale from 1 to 5, where 5 means ‘‘very successful’’ and 3 ‘‘no impact.’’ When employees were asked about productivity changes, only 3% of employees said it had increased greatly, while only 1% said it had decreased (Poole & Jenkins, 1990, p. 60). The modest but positive effects of performance are also present in the study of Kalmi et al. (2005), who have senior HR managers from listed firms in four European countries as their respondents. The average rating for the statement Financial participation improves productivity is 2.27 on a scale from 1 to 4, where 1 is no impact. In our study, we use the second type of measure, i.e., direct questions on the effects of financial participation. Because studies using such a measure have usually yielded positive results, our second hypothesis is: Hypothesis 2. The subjective measures of the impact of profit-sharing are likely to indicate beneficial impacts from profit-sharing. There are a few recent studies comparing subjective and objective measures of overall workplace performance.5 The research is dominated by data gathered from the WERS where the subjective measures are the ordinal measures described above and the objective measures are different accounting measures. This literature tends to find a positive but small correlation between subjective and objective performance measures. The relevance of this literature to our case is limited, because we are comparing the measures of the impact of financial participation schemes, whereas the earlier studies have compared the measures as such and have not been concerned about whether the presence of financial participation schemes influences the distribution of the measure. However, the general lesson from these studies is, as Wall et al. (2004) and Forth and McNabb (2008) argue, that subjective and objective measures should give consistent results, although empirically this link has been a rather weak one. Even though

76

PANU KALMI AND CHRISTINA SWEINS

there is no prior research on our next research question, we take the analogy from the prior literature comparing subjective and objective measures in formulating our third hypothesis: Hypothesis 3. Subjective and objective impact measures are positively correlated. Our final hypothesis is related to the research question whether supposedly more informed respondents provide subjective assessments on performance impact that correlate more closely with objective measures than the subjective measures correlated from employees at large. In our case, the more informed respondents are fund chairmen. It is reasonable to expect that they are better informed about the effects of the funds than are employees at large, and therefore we expect that the correlation between their assessment and objective performance measures is stronger than the correlation between the general assessment of employees and objective performance measures. If this hypothesis will be supported by the data, it will generally support collecting information from more informed single respondents than on a larger pool of more uninformed respondents, and thus alleviate concerns regarding single respondent bias.6 Thus, our fourth hypothesis is: Hypothesis 4. Subjective impact measures derived from fund chairmen are more strongly correlated with objective impact measures than subjective impact measures collected from employees at large.

3. DESCRIPTION OF THE PERSONNEL FUND SCHEME Personnel fund scheme is a unique Finnish profit-sharing system that combines features of profit sharing and employee share ownership (see Sweins et al., 2009 for a fuller description). Personnel funds have been in existence since the early 1990s. Altogether 80 funds were established between 1990 and 2006, of which 28 have been closed mainly due to mergers and divestments. In more recent years, the popularity of the funds has increased, and in November 2009 there were 57 personnel funds with more than 136,000 members covering about 10% of the workforce. In 2000, when the subjective data were collected there were 36 funds. The capital paid to the fund is mainly accumulated based on firm profitability indicators from the income statement. The employer chooses

The Performance Impact of Financial Participation

77

the criteria of profit-related payments, but it must be fixed before the realization of income statements, typically a year in advance. Payments are made once a year. The funds invest their capital either in shares of their own firm or other firms, in investment funds, bonds, or in bank accounts. Through these investments, the financial gains of the employees extend beyond just profit-shares. The capital in the personnel fund is divided into individual accounts, and sometimes a collective part, which is used either for administrative costs or other costs of the fund. The shares are distributed to employees usually either in relation to compensation (base pay or total compensation) or in relation to hours worked. What makes the system a deferred profit-sharing scheme is that the individual shares of the members are locked in for the first 5 years of membership. After that, a member can withdraw at most 15% of the value of his or her accumulated fund share. The employee can withdraw the shares completely upon leaving the firm. The use of personnel funds has been encouraged by several tax advantages. For employees, 20% of the payouts from the fund are tax free. The fund pays no taxes on its earnings. Employers do not have to pay pension or social security contributions for the profit-shares paid to the fund.

4. DATA AND METHODOLOGY 4.1. Sample Characteristics and Methodology Our core sample was gathered during research on personnel funds conducted by researchers at Helsinki University of Technology in 2000. The sample consisted of 31 of the 36 firms using personnel funds at the time. Five firms did not want to participate in the questionnaire survey because they had ongoing large surveys in the firm, or they had decided (due to mergers) not to pay any profit into the fund. We then matched these firms to financial statements and balance sheet data for the years 1999–2003. We were able to match these data for 21 of the 31 firms available. Thus, our data include 58% of firms that had personnel funds in the year 2000. These 21 firms represent different sectors nationwide. Around 20% of the personnel fund firms and comparison firms are in the manufacturing sector, about one-fourth (24%) in media and 55% in trade and other services. Among the 10 firms that responded to the questionnaire but whose objective performance data were not matched, four were banks or insurance

78

PANU KALMI AND CHRISTINA SWEINS

firms for which the accounting data are not comparable. For others (e.g., national airline carrier), there were no convincing comparison firms. These firms were mostly large state firms that are the sole domestic firms in their industries (e.g., the national airline carrier). Thus, even though our sample covers most firms with personnel funds, the selection in the sample is not random, and this caveat should be taken into account when evaluating the results presented below. The idea in matching the firms with personnel funds with those without the funds is that if the firms are otherwise similar, then the difference in performance reflects the impact of having a personnel fund.7 This produces a firm-specific estimate for the impact of personnel fund, which is useful, because we can calculate the correlation between these measures and the subjective measures. The main challenge with this approach is that there might be other differences between the firms that influence the performance differences between the firms, such as the quality of the workforce, use of participatory workplace practices, use of comparable contingent pay schemes, or management quality. If these differences are correlated with the decision to adopt the personnel fund, the objective impact measure may be biased. Unfortunately, our data do not contain controls for these variables. An important limitation is that we do not observe adoptions of personnel funds during this period, and thus we cannot use before and after analysis.

4.2. Subjective Data The collection of the subjective data involved a questionnaire targeted at the personnel. The questionnaires were only distributed to some employees. The representatives of the firms were asked to distribute the questionnaires so that the employees who received them would represent all employee groups. The questionnaires were returned in sealed envelopes. The researchers did not monitor the distribution of the questionnaires, so it is possible that the employees selected for the sample are not representative of all employees. Among the 21 firms from which we use data in this chapter, the response rates in the firms (calculated as the number of respondents per questionnaires distributed) varied from 15% to 100%, the mean response rate being 52%. We received 651 individual responses out of 1,246 questionnaires, although the number of responses to individual items was often considerably smaller.8 The subjective performance effects of the profit-sharing system are measured with seven items. A factor analysis shows that all items are loading

The Performance Impact of Financial Participation

79

on one component. Thus we use all items as a sum variable called overall efficiency. This consists of the responses to the following questions: How has the personnel fund influenced economic efficiency, operational efficiency, quality, flexibility, delivery reliability, customer service, and forward planning? Because this question is quite broad, we also take out two more focused items, perceived impact on economic efficiency and on operational efficiency, and test the correlation of these with the objective variables. In addition to the questionnaires collected from all employees, we also have separate responses from the chairmen of personnel funds for 15 out of 21 firms in the final sample. We use this information to test the hypothesis that chairmen would actually be more knowledgeable about the impact of financial participation than ordinary employees.

4.3. Objective Data We use financial statements and balance sheet data for the years 1999–2003. The source of this information is the Voittoþ database, maintained by the Finnish firm Suomen Asiakastieto. This database covers financial statements from about 100,000 firms in Finland. We were able to match and find comparable data for 21 of the 31 firms available. We paired each personnel fund firm with two firms without the fund. The criterion for the matching process was narrow industry classification and similar size (measured by sales). We attempted to select for each comparison firm from the same industry cell, so that when firms are arranged in order of their sales in 1999, one comparison firm would be just smaller than the firm with the fund, and the other would be just larger than the firm with the fund. Most firms with personnel funds are relatively large, but there is also considerable variation among the firms. For instance, the mean turnover during the period 1999–2003 ranged from about 3.7 million euro to almost 4,700 million euro. The mean was 484 million euro. Employment ranged from 22 to 17,308 employees, the mean employment being 1,730. For comparison firms, the mean turnover was 217 million euro and mean employment 1,293. The discrepancy in the average size of the firms in these two groups is due to the fact that there are some fairly large firms for which it was not possible to find comparison firms of similar size. Among the smaller firms, there is always one firm in the same industry that is smaller and one that is larger. In the objective performance estimations, we measure the growth in productivity using two dependent variables: sales per employee and gross

80

PANU KALMI AND CHRISTINA SWEINS

proceeds9 per employee. Similarly, we use two measures of profitability: return on assets (ROA) and return on investment (ROI). We use growth equations because they help to control for unobserved heterogeneity (unobserved differences in firms we cannot measure).10 We use the four different outcome variables, and calculate the difference between each measure for a 5-year period 1999–2003, thus observing four differences for each variable per firm. Then we calculate the difference between the ‘‘treatment’’ observation (firm with the fund) and the control observations. The differences in growth of sales and gross proceeds are in logarithms and can be interpreted as percentage growth. The growth in ROI and ROA are in percentage units. There are some missing values that limit the number of observations (70 in gross proceeds growth and 78 in other regressions). Next we will look at our results.

5. RESULTS The first thing we do is to look at the different objective performance measures of both personnel fund firms and comparison firms (Table 1). We find that in terms of all profitability and productivity measures that firms with personnel funds are doing better, i.e., growing faster. This indicates that using objective measures, financial participation has a positive impact. However, there is a considerable variation in these measures, and therefore the differences are not statistically significant. Therefore Hypothesis 1 gets only weak support from our data. Next, we look at the subjective impact measures we have from the personnel fund firms. The means and standard deviations are all gathered Table 1. Objective Performance Measures for Firms with Personnel Funds and the Comparison Firms (Annual Growth Rates): Means and Standard Deviations (Standard Errors for the Difference).

Firms with personnel funds Control firms Difference

Sales Growth

Gross Proceeds, Growth

ROI, Growth

ROA, Growth

0.053 (0.236) 0.048 (0.125) 0.005 (0.028)

0.057 (0.357) 0.038 (0.172) 0.018 (0.041)

0.501 (11.411) 1.299 (11.324) 1.80 (1.71)

0.391 (7.722) 0.555 (7.223) 0.164 (0.964)

81

The Performance Impact of Financial Participation

Table 2.

Subjective Impact of Profit Sharing: Means and Standard Errors.

Subjective Measures

Overall efficiency Economic efficiency Operational efficiency

1. All Employees (N ¼ 21)

2. Fund Chairmen (N ¼ 15)

3. Difference (All Employees – Fund Chairmen)

3.44 (0.04) 3.56 (0.05) 3.53 (0.05)

3.59 (0.12) 3.86 (0.18) 3.93 (0.16)

0.15 (0.13) 0.30 (0.18) 0.40 (0.16)

Notes: (1) In columns 1 and 2, the null hypothesis is mean ¼ 3; in column 3, the null is that the means of the two groups are equal. (2) Significance levels: 10%; 5%; 1%.

into Table 2, and they have been listed separately for all employees and for fund chairmen. For the chairmen sample, the number of observations unfortunately drops to 15. The scale of the impact is 1 ¼ large negative impact from the fund, 2 ¼ small negative impact, 3 ¼ no impact, 4 ¼ small positive impact, and 5 ¼ large positive impact. We present the full distribution of the employee and chairmen responses in the appendix. However, we perform the analysis using methods for continuous (interval) data because for most indicators of interest, there are one or even two levels of aggregations that break down the categorical data. First, in employee responses, we use the firm-level means of employee responses. Second, one of our measures, overall efficiency, is an aggregation of the responses to seven different items. From the full distributions presented in the appendix and from the firm-level sample means, we find that the respondents tend to perceive funds having positive rather than negative impacts. As evident from the appendix, employees indicate extremely rarely that funds would have negative effects, and chairmen do not indicate that at all in their responses. Table 2 indicates that both overall scores and the two selected individual scores are always larger than 3 for both employees and for chairmen and in all cases significantly so. This means that our second hypothesis is supported. In most cases, the means tend to concentrate around 3.4–3.5, so the impact is not perceived to be very large. Perhaps not surprisingly, the fund chairmen believe the funds to have larger effects than employees in average, though the difference is significant only in regard to operational efficiency.

82

PANU KALMI AND CHRISTINA SWEINS

The next thing we do is to compare the subjective and the objective measures with each other. Table 3 presents the correlation coefficients and significance levels from comparing the subjective and objective impact measures of year-firm observations, where for subjective measures the same value from the questionnaire has been assigned for each year. Because the subjective measures do not vary over time, the levels of statistical significance have been obtained from regressions where the year effects have been controlled for and standard errors are robust for firm clustering. The surprising finding from Panel A of Table 3 is that all correlations are negative, although only 3 of 12 coefficients are statistically significant. The Hypothesis 3 said that these correlations were expected to be positive. However, the size of these correlations is small. In Panel B, where subjective Table 3.

Correlations Between Subjective and Objective Performance Measures. Perceived Overall Impact of PF

Perceived Impact of PF to Economic Efficiency

Panel A: Subjective measures derived from all employees Difference in growth in 0.03 0.03 sales (N ¼ 78) Difference in growth of 0.04 0.04 gross proceeds (N ¼ 70) Difference in growth of 0.06 0.12 ROI (N ¼ 78) Difference in growth of 0.04 0.12 ROA (N ¼ 78) Panel B: Subjective measures derived from fund chairmen Difference in growth in 0.04 0.10 sales (N ¼ 55) Difference in growth of 0.19 0.24 gross proceeds (N ¼ 51) Difference in growth of 0.35 0.37 ROI (N ¼ 55) Difference in growth of 0.26 0.22 ROA (N ¼ 55)

Perceived Impact of PF to Operational Efficiency

0.06 0.10 0.10 0.10

0.10 0.24 0.39 0.22

Notes: (1) The correlation coefficients have been obtained from pairwise correlations of the subjective and objective performance measures. The levels of significance are from regressions where year effects have been added and standard errors have been corrected for clustering. (2) Significance levels: 10%; 5%; 1%.

The Performance Impact of Financial Participation

83

measures are derived from fund chairmen, the situation is quite different. All correlations are positive, and 9 out of 12 coefficients are significant. Moreover, some of the correlations are of reasonable size (over 0.3). The correlations with profitability and subjective measures are clearly higher than those with productivity, but no clear pattern across the various subjective measures emerges. Hypothesis 3 is thus only supported when using chairmen’s estimates. The differences between the results of chairmen versus all employees are not due to the differences of sample composition. If we restrict the sample so that we use only those firm observations for which chairman evaluations are available and repeat the analysis using subjective assessments collected from employees broadly, roughly the same picture emerges than in Panel A: correlations are mostly negative, although there are now two positive correlations. However, in all cases (both positive and negative) the coefficients are not statistically significant. Hypothesis 4 therefore is supported: plausibly because of informational advantages, chairmen of the funds appear to be able to make judgments that are more in line with observed objective impact measures, and therefore likely to be more reliable.

6. CONCLUSIONS The main task of this chapter is to compare the objective and subjective measures of the impact of financial participation. In our case, the subjective impact measures were collected from an employee survey. The objective impact measures were gathered from financial accountancy data for a 5-year period and including two comparison firms for each firm with a personnel fund. The novelty of the research design is that this was the first study comparing explicitly subjective and objective impact measures in the context of financial participation. We also collected data from respondents from different levels, namely from employees at large and from fund chairmen, and we are able to compare the responses from these two groups. Our analysis indicates that when we compare the objective performance measures in personnel fund firms with the comparison firms, financial participation has a positive, but a nonsignificant impact. There can be several reasons for this. One reason could be that we do not have enough information about the comparison firms. We do not know for instance what kind of incentive systems they have in use. They might be using a cash-based profit-sharing plan, which is very common in Finland, even in international

84

PANU KALMI AND CHRISTINA SWEINS

comparisons (Jones, Kalmi, Kato, & Ma¨kinen, 2010b). If the control firms have such a scheme, they may not be that different from the ‘‘treatment’’ group and this biases the results. Our approach also relies on quite strong assumptions that the firms are comparable in other regards than the presence of the profit-sharing plan. If this is not the case, this discrepancy may bias the results to an unknown direction. Finally, the small sample size limits the ability to obtain statistically significant results. When we look at the subjective impact measures our results reveal that the employees perceive the funds to have a positive impact overall. This holds also separately for measures of economic efficiency as well as of operational efficiency. The most interesting results are related to the comparisons between objective and subjective measures. We find that with subjective measures collected from employees broadly, the correlations between objective measures are negative, though often insignificant. This is clearly against the hypothesis that the correlations would be positive. However, when we use subjective measures derived from the chairmen of the funds, the results are quite different. The correlations are always positive, and in 9 cases out of 12, they are significant. This brings us to the main recommendation of this study: despite the socalled single respondent bias, it is better to use an informed single respondent than a number of less informed respondents. Our results indicate that the bias arising from gathering data from less informed respondents is more severe than the bias of using only a single respondent. This would also mean that many surveys on compensation could also be done more economically, when there would be less need to gather a large number of respondents from individual firms. The only problem is to find the well-informed respondents, when making a survey.11 However, our findings depend on the assumption that the objective measures of impact correspond with the real impact. As discussed above, there may be various reasons why this may not be correct. Unfortunately, we can only test whether the results are similar or not when using two types of measures, but not whether one (or both) set of results is correct. There is clearly more room for research on these issues, preferably overcoming some obstacles we faced. These include small sample size and nonrepresentative sample, lack of changes in financial participation status (precluding fixed effects estimations), lack of information of many variables that may influence performance (such as the existence of complementary workplace practices), and the absence of longitudinal data on subjective assessments of performance.

The Performance Impact of Financial Participation

85

NOTES 1. Similar subjective impact measures have been used previously, e.g., by Poole and Jenkins (1990) and Kalmi et al. (2005). 2. Some key studies include Fitzroy and Kraft (1987), Cable and Wilson (1989), Wadwhani and Wall (1990), Kruse (1993), Jones and Kato (1995), Fakhfakh and Perotin (2000), Kato and Morishima (2002), Conyon and Freeman (2004), Robinson and Wilson (2006), and Bryson and Freeman (2010). There are of course some variations in the research design around the common theme. Oxera (2007) is a recent large-scale study done using a large UK data set and is among the most comprehensive studies. 3. For instance, in Oxera (2007) it is stated that subjective measures overstate the impact of financial participation schemes and therefore objective measures should be preferred. 4. See the surveys of Jones, Kato, and Pliskin (1997), Kruse and Blasi (1997), Sesil, Kruse, and Blasi (2003) and Perotin and Robinson (2003) and the papers cited in footnote 2. One recent study that fails to find positive effects using a range of alternative estimators is Jones, Kalmi, and Ma¨kinen (2010a) on Finnish stock option schemes. 5. These studies include Guest et al. (2003), Wall et al. (2004), Haskel (2005), Kersley et al. (2006), and Forth and McNabb (2008). 6. However, note that this rests on the assumption that the objective performance measures are, broadly speaking, correct. This assumption unfortunately cannot be tested. 7. Similar estimation strategy has been used, e.g., by Winther and Marens (1997). 8. In the total sample of 31 firms, the size of the targeted population was 1,956 of which 1,038 questionnaires were received (response rate 59%). 9. ‘‘Gross proceeds’’ is the sum of turnover added or deducted with following items: ‘‘other business revenues,’’ ‘‘production for own use,’’ ‘‘change in stock,’’ ‘‘purchases during accounting period,’’ and ‘‘external services.’’ It is roughly similar to value added. 10. As mentioned, we cannot deal with unobserved heterogeneity by fixed effects estimations, because we do not observe changes in personnel fund status of firms. 11. We thank Professor Stephen Perkins for pointing this out at the second Reward Management Conference in Brussels November 2009, where the second author presented an earlier version of this paper.

ACKNOWLEDGMENTS The authors acknowledge financial support from the doctoral program in Industrial Engineering and Management, Academy of Finland project no. 120234, and Silta Oy. The authors thank an anonymous referee and the editor Tor Eriksson, Andrew Robinson, and Stephen Perkins for a series of most helpful and constructive comments on earlier versions of this chapter.

86

PANU KALMI AND CHRISTINA SWEINS

REFERENCES Alchian, A. A., & Demsetz, H. (1972). Production, costs, and economic organization. American Economic Review, 62(5), 777–795. Appelbaum, E., Bailey, T., Berg, P., & Kalleberg, A. L. (2000). Manufacturing advantage. Why high-performance work systems pay off. Ithaca, NY: ILR Press. Ben-Ner, A., & Jones, D. C. (1995). Employee participation, ownership, and productivity: A theoretical framework. Industrial Relations, 34(4), 532–554. Blanchflower, D. G., & Oswald, A. J. (1988). Profit-related pay: Prose discovered? The Economic Journal, 98, 720–730. Boswell, W. R., & Boudreau, J. W. (2001). How leading companies create, measure and achieve strategic results trough ‘‘Line of Sight’’. Management Decision, 39(10), 851–859. Bryson, A., & Freeman, R. (2010). How does shared capitalism affect performance in the UK? In: D. Kruse, R. B. Freeman & J. R. Blasi (Eds), Shared capitalism at work: Employee ownership, profit and gain sharing, and broad-based stock options (pp. 201–224). Chicago, IL: University of Chicago Press. Cable, J., & Wilson, N. (1989). Profit-sharing and productivity: An analysis of UK engineering firms. The Economic Journal, 99, 366–375. Conyon, M. J., & Freeman, R. B. (2004). Shared modes of compensation and firm performance: UK evidence. In: R. W. Blundell, D. Card & R. B. Freeman (Eds), Seeking a premier league economy. Chicago: University of Chicago Press. Fakhfakh, F., & Perotin, V. (2000). The effects of profit-sharing schemes on enterprise performance in France. Economic Analysis, 3(2), 93–111. Fitzroy, F. R., & Kraft, K. (1986). Profitability and profit sharing. The Journal of Industrial Economics, 35(2), 113–130. Fitzroy, F. R., & Kraft, K. (1987). Cooperation, productivity, and profit sharing. The Quarterly Journal of Economics, 102(1), 23–35. Forth, J., & McNabb, R. (2008). Workplace performance: A comparison of subjective and objective measures in the 2004 workplace employment relations survey. Industrial Relations Journal, 39(2), 104–123. Freeman, R., Kruse, D., & Blasi, J. (2010). Worker responses to shirking under shared capitalism. In: D. Kruse, R. B. Freeman & J. R. Blasi (Eds), Shared capitalism at work: Employee ownership, profit and gain sharing, and broad-based stock options (pp. 77–103). Chicago, IL: University of Chicago Press. Gerhart, B., Wright, P. M., Mahan, G. C., & Snell, S. A. (2000). Measurement error in research on human resources and firm performance: How much error is there and how does it influence effect size estimates? Personnel Psychology, 53, 803–834. Guest, D. E., Michie, J., Conway, N., & Sheehan, M. (2003). Human resource management and corporate performance in the UK. British Journal of Industrial Relations, 41(2), 291–314. Haskel, J. (2005). Unions and productivity again: New evidence from matched WERS and business census data. Preliminary paper presented at the 2005 WPEG conference. Jones, D., & Kato, T. (1995). The productivity effects of employee stock-ownership plans and bonuses: Evidence from Japanese panel data. American Economic Review, 85, 391–414. Jones, D., Kato, T., & Pliskin, J. (1997). Profit sharing and gainsharing: A review of theory, incidence and effects. In: D. Lewin, J. B. Mitchell & M. A. Zaidi (Eds), The human resource management handbook, Part I. London: JAI Press.

The Performance Impact of Financial Participation

87

Jones, D. C., Kalmi, P., Kato T., & Ma¨kinen, M. (2010b). Financial participation in Finland: Incidence and determinants. Manuscript, Aalto University School of Economics. Jones, D. C., Kalmi, P., & Ma¨kinen, M. (2010a). Productivity effects of selective and broadbased stock option schemes: Evidence from Finland. Journal of Productivity Analysis, 33(1), 67–80. Kalmi, P., Pendleton, A., & Poutsma, E. (2005). Financial participation and performance in Europe. Human Resource Management Journal, 15(4), 54–67. Kato, T., & Morishima, M. (2002). The productivity effects of participatory employment practices: Evidence from new Japanese panel data. Industrial Relations, 41(4), 487–520. Kersley, B., Alpin, C. J., Forth, J. A., Bryson, A. H., Bewley, H. G., Dix, G., & Oxenbridge, S. (2006). Inside the workplace: Findings from the 2004 workplace employment relations survey. London: Routledge. Kruse, D. L. (1993). Profit sharing: Does it make a difference? Kalamazoo, MI: Upjohn Institute. Kruse, D. L., & Blasi, J. R. (1997). Employee ownership, employee attitudes, and firm performance: A review of the evidence. In: D. Lewin, D. J. B. Mitchell & M. A. Zaidi (Eds), The human resource management handbook, Part I (pp. 113–151). Greenwich, CT: JAI Press Inc. Levine, D. I., & Tyson, L. A. (1990). Participation, productivity, and the firm’s environment. In: A. S. Blinder (Ed.), Paying for productivity: A look at the evidence. Washington, DC: The Brookings Institution. Long, R. J. (2000). Employee profit sharing consequences and moderators. Relations Industrielles, 55, 477–504. Machin, S. J., & Stewart, M. B. (1990). Unions and the financial performance of British private sector establishments. Journal of Applied Econometrics, 5, 327–350. Magnan, M., & St-Onge, S. (2005). The impact of profit sharing on the performance of financial services firms. Journal of Management Studies, 42(4), 761–791. McNabb, R., & Whitfield, K. (1998). The impact of financial participation and employee involvement on financial performance. Scottish Journal of Political Economy, 45(2), 171–187. Mitchell, D. J. B., Lewin, D., & Lawler, E. E., III. (1990). Alternative pay systems, firm performance, and productivity. In: A. S. Blinder (Ed.), Paying for productivity: A look at the evidence. Washington, DC: The Brookings Institution. Oxera. (2007). Tax advantaged employee share schemes: Analysis of productivity effects. HM Revenue & Customs Research Report no. 33. Perotin, V., & Robinson, A. (2003). Employee participation in profit and ownership: A review of the issues and evidence. Paper prepared for the European Parliament. Poole, M., & Jenkins, G. (1990). The impact of economic democracy: Profit-sharing and employee-shareholding schemes. London: Routledge. Robinson, A., & Wilson, N. (2006). Financial participation and productivity: Insights from stochastic frontier estimation. Economic and Industrial Democracy, 27(4), 609–635. Rousseau, D. M., & Shperling, Z. (2003). Pieces of the action: Ownership and the changing employment relationship. Academy of Management Review, 28, 553–570. Sesil, J. C., Kruse, D. L., & Blasi, J. R. (2003). Sharing ownership via employee stock ownership. In: L. Sun (Ed.), Ownership and governance of enterprises: Recent innovative developments. Basingstoke: Palgrave/Macmillan and St. Martin. Spector, P. E. (2006). Method variance in organizational research: Truth or urban legend? Organizational Research Methods, 9, 221–232.

88

PANU KALMI AND CHRISTINA SWEINS

Sweins, C., Kalmi, P., & Hulkko-Nyman, K. (2009). Personnel knowledge of the pay system, pay satisfaction and pay effectiveness: Evidence from Finnish personnel funds. The International Journal of Human Resource Management, 20(2), 457–477. Vroom, V. H. (1964). Work and motivation. New York: Wiley. Wadwhani, S., & Wall, M. (1990). The effects of profit-sharing on employment, wages, stock returns and productivity: Evidence from UK micro-data. The Economic Journal, 100, 1–17. Wall, T. D., Michie, J., Patterson, M., Wood, S. M. J., Sheehan, M., Glegg, C. W., & West, M. (2004). On the validity of subjective measures of company performance. Personnel Psychology, 57, 95–118. Weitzman, M. L., & Kruse, D. L. (1990). Profit sharing and productivity. In: A. Blinder (Ed.), Paying for productivity (pp. 94–141). Washington: The Brookings Institution. Winther, G., & Marens, R. (1997). Employee ownership may go a long way: Comparative growth performance of employee ownership firms in New York and Washington States. Economic and Industrial Democracy, 18, 393–422.

APPENDIX. THE DISTRIBUTION OF RESPONSES ON THE IMPACT OF PERSONNEL FUNDS (IN PERCENTAGES) Impact of personnel funds on y Employees (N ¼ 521–531) Score 1 Quality 0 Economic efficiency 0 Flexibility 0 Operational efficiency 0 Delivery reliability 0 Customer service 0 Forward planning 1

2 1 1 2 0 1 0 1

3 63 44 62 47 65 60 57

4 32 48 31 47 29 30 36

5 4 7 5 6 6 10 5

Fund chairmen (N ¼ 13–15) Score Quality Economic efficiency Flexibility Operational efficiency Delivery reliability Customer service Forward planning

2 0 0 0 0 0 0 0

3 54 29 50 20 80 53 67

4 38 57 50 67 13 33 20

5 8 14 0 13 7 13 13

1 0 0 0 0 0 0 0

Notes: Scale of responses: 1 ¼ large negative impact from the fund, 2 ¼ small negative impact, 3 ¼ no impact, 4 ¼ small positive impact, and 5 ¼ large positive impact.

EMPLOYEE DISCRETION AND THE LABOR-MARKET ENVIRONMENT Jaime Ortega 1. INTRODUCTION This study provides an empirical analysis of the relationship between job design and the labor-market environment in which firms operate. In particular, I focus on one aspect of job design: the extent to which employees have discretion (autonomy) to organize their work. There has been considerable emphasis in the last 20 years on the importance of ‘‘highinvolvement’’ work practices, which seek to give employees more decision rights at work. This literature has been concerned with the introduction of work practices such as team work, job rotation, or quality circles, and with the use of performance pay contracts. Within this literature, there are also some studies that focus more particularly on the extent to which employees have job discretion or autonomy. Discretion is an important characteristic of jobs because much of the redesign effort that has been conducted in the last years has aimed at giving employees more power to make decisions at work, and performance gains are largely attributed to these changes. The diffusion of these practices is usually attributed to an increase in competition in the product markets, which has put pressure on companies to increase quality. For example, the widespread use of quality circles and total quality management among car manufacturers is largely credited to Japanese manufacturers’ success, which has lead to imitation by European and U.S. manufacturers. Human resource management (HRM) innovations Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 89–110 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011008

89

90

JAIME ORTEGA

in other industries (e.g., steel or apparel) appear to be driven by similar forces. However, in the last 15 years there has been not only an increase in product market competition, but also a significant reduction in unemployment rates in developed countries. Is it possible that the introduction of greater worker autonomy has been partly caused by an increase in workers’ bargaining power? There is actually little empirical evidence on the effect of workers’ bargaining power on job design, largely because most studies take the view that, if a human resource innovation increases value, then owners, managers, and workers should agree to introduce it, no matter how bargaining power is divided. The aim of this research is to study whether employees’ bargaining power influences their employers’ willingness to give discretion. For this purpose, I use individual data from the third European Working Conditions Survey (EWCS) matched with regional unemployment data and with information on national labor regulations. The main advantage of using European data is that labor markets are more regulated than in the United States, and we can therefore use national laws to obtain comparable measures of employee bargaining power across countries. The EWCS provides measures of job discretion by employee, and I use cross-country variation in labor laws and regional variation in unemployment rates within countries to estimate the effect of employees’ bargaining power on discretion. Since discretion also depends on many other factors – particularly individual and job characteristics – which are likely to vary across countries and regions, I use the EWCS data to construct individual-level controls and measures of other work practices. Since the data are cross-sectional, I cannot study whether the diffusion of new forms of work organization has been caused by changes in the labor market, but I can study the extent to which interregional and international differences in labor markets influence firms’ HRM choices concerning discretion. To guide the empirical analysis, I discuss two theoretical approaches, which yield different predictions. The first is a bargaining approach in which the firm negotiates the level of discretion with its employees. I hypothesize that if negotiation is costly the firm is more likely to give discretion when the party that has a preference for discretion has more bargaining power. Thus, if employee discretion is costly to employees and beneficial to the firm, the level of discretion will be higher the lower the employees’ bargaining power. Conversely, if discretion is beneficial to employees and costly to the firm, it will be higher the higher the employees’ bargaining power. If both the firm and the employees obtain a positive utility from an increase in employee discretion, I hypothesize that the division of bargaining power should have

Employee Discretion and the Labor-Market Environment

91

no influence on the level of discretion. The second approach is based on efficiency wage theory and posits that when unemployment is smaller employees have less incentive to work hard because the cost of being unemployed is smaller. In such cases, firms will monitor more intensively and employees will enjoy less discretion. More generally, I hypothesize that when employee protection is stronger, for example, due to higher firing costs, stronger unions or higher unemployment benefits, employees will also have less incentive to work hard and firms will react by increasing monitoring and reducing discretion. One potential criticism to this approach is that firm–union bargaining usually focuses on wages and employment and is not really concerned about job discretion or work organization in general. However, the fact that collective agreements do not usually include explicit clauses on employees’ decision rights does not imply that changes in work organization are not negotiated on an informal basis. In fact, there is evidence that a strong presence of unions is positively correlated with more employee discretion. For example, Doellgast, Holtgrewe, and Deery (2009) study work practices in call centers in Europe and North America and find that strong collective bargaining institutions are associated with more opportunities for employee participation and discretion. Clark, Clark, Day, and Shea (2001) study the effects of market-based healthcare reform on registered nurses in the United States. They find that nurses who experienced job restructuring had less voice in the hospital. Moreover, nurses who reported less voice were more likely to view unions as a potential solution and to vote for unionization. This study is most closely related to work done by Arai (1994), who estimates wage regressions using a measure of employee discretion in the righthand side. If discretion has a positive value to employees, a compensating wage differentials approach predicts a negative effect of discretion on wages. On the other hand, from an efficiency wage perspective, discretion should have a positive coefficient because when firms pay higher salaries, they can afford to monitor less. Arai (1994) finds a positive effect for private sector employees and a negative effect for public sector employees. The main difference with respect to this study is that I do not focus on the relationship between autonomy and wages; instead, I am interested in the effect of employees’ bargaining power on job discretion, controlling for wages. For the same reason, this chapter also differs from the efficiency wage literature interested in the relationship between supervision and wages (Leonard, 1987; Groshen & Krueger, 1990; Neal, 1993; Rebitzer, 1995; Ewing & Payne, 1999), which provides mixed evidence.1 The effect of bargaining power is interesting on its own because it has hardly been studied, and it can improve our

92

JAIME ORTEGA

understanding of employers’ choices and of the effects of policy changes on HRM practices. I also contribute to the literature on discretion (autonomy). Research on discretion has analyzed its effects on employee satisfaction (Karasek, 1979; Parker & Wall, 1998; Singh, 2000; Parker, 2003) and some of the literature on ‘‘high-performance’’ work practices and productivity includes measures of discretion (Appelbaum, Bailey, Berg, & Kalleberg, 2000; Ichniowski, Shaw, & Prennushi, 1997; Ichniowski & Shaw, 1999). The studies on employee satisfaction typically find that discretion reduces work stress, particularly when job demands are high, and the literature on productivity finds that certain work practices that imply higher levels of discretion contribute to increase performance. Other studies have analyzed the relationship with performance pay (Osterman, 1994; MacLeod & Parent, 1999; Nagar, 2002; Foss & Laursen, 2005; Ortega, 2009b), usually finding a positive relationship. Discretion regarding work schedules has also been analyzed in the context of the ‘‘work–family balance’’ problem (Goodstein, 1994; Osterman, 1995; Golden, 2001; Wood, de Menezes, & Lasaosa, 2003; Ortega, 2009a).2 However, none of these studies is interested in the effect of unemployment or labor-market institutions on employers’ decisions about discretion.

2. THEORY Suppose a firm has to decide whether to introduce some new HRM practices, which imply more employee involvement in decision making, that is, more job autonomy or discretion. We are interested in the issue of how the relative bargaining power of employers and employees will influence the decision. I consider two theoretical views on the problem: a bargaining approach and an efficiency wage approach.

2.1. A Bargaining Approach The first approach is one in which firm and employees negotiate the introduction of more discretion. The firm seeks to maximize profits, employees care about both their salaries and job content, and there is no asymmetry of information between employees and employers. The main problem is to make sure that the gains from the increase in discretion are divided in such a way that both employees and employers are willing to

Employee Discretion and the Labor-Market Environment

93

agree. As explained by Coase (1960) in a different context, there will be two distinctly different scenarios: with and without negotiation costs. If there are no negotiation costs, the firm will adopt the new HRM practices as long as total benefits outweigh total costs, considering total benefits and costs as the sum across all individuals who would be affected by the changes, that is, the Coase theorem will apply. In particular, the decision will not be affected by the relative bargaining power of employees and employers: for a given distribution of bargaining power each party will obtain a share of total surplus, and the larger the total surplus the higher the gains for both employers and employees. The division of bargaining power will determine how the total gains are divided among the parties but will not influence the decision to give more discretion. A more realistic scenario is one in which negotiation costs are important. First, changes in HRM practices typically affect a large number of employees at a time, which considerably raises negotiation costs, particularly when there are asymmetries in how different employees are affected by the changes. Second and more importantly, labor markets are usually regulated. For example, in most countries there are important restrictions to the types of contracts that can be written and the negotiation procedures are often subject to regulation (e.g., collective agreements). By contrast, the value maximization principle according to which HRM choices would be independent of bargaining power requires that the parties have enough contracting freedom to distribute the gains – thus, the contract can make sure that the parties that are negatively affected by the changes can be ‘‘compensated’’ by the parties that benefit the most (see Milgrom & Roberts, 1992). As an example, suppose the firm wants to introduce high-involvement work practices for a group of employees and suppose an increase in discretion is costly to those employees, for example, because their work will be more stressful. An agreement between the two parties will require some compensation for the group of employees who are involved in the new HRM practices. However, certain labor-market institutions can make it difficult to reach this type of solution: thus, if there is a collective agreement covering all employees in the firm and not all employees are affected by the HRM changes, the firm may have to offer the same increase in compensation to all employees covered by the same collective agreement or otherwise might have to negotiate a different treatment for the employees who are going to be involved in the new HRM practices and those who are not. In any case, it will be costly to reach an agreement. Similar problems can arise if employees place a positive value on discretion but the firm is not interested in giving more discretion; in that case, employees would have to

94

JAIME ORTEGA

accept a reduction in their salaries or would have to make other, nonpecuniary concessions to the firm, but such particular conditions for specific groups of employees can also be difficult to negotiate. In those cases, I hypothesize that it will be easier to reach an agreement when the party that places a positive value on discretion has more bargaining power. Thus, if the firm is interested in introducing more discretion but employees are reluctant, discretion is more likely to be increased if the firm has more bargaining power than the employees. Conversely, if employees would like to have more discretion but the firm considers that it will be too costly, discretion is more likely to be introduced if employees have more bargaining power.

2.2. The Efficiency Wage Model In the efficiency wage model (Shapiro & Stiglitz 1984), the firm seeks to maximize profits and the employee’s utility depends upon his salary, the effort level exerted at the job, and the probability and cost of being dismissed. The employee’s effort is not observable, and the main problem is to provide the right incentives to exert effort. Unlike in the bargaining setup, the two parties can agree at zero cost on a contract that divides any gains between them. In Shapiro and Stiglitz (1984), the firm has two instruments to motivate the worker: the level of supervision and the salary. If the firm raises the level of supervision, the employee’s cost from shirking will be larger because the probability of being ‘‘caught’’ increases. On the other hand, if the firm increases the salary the cost from shirking will also be larger because the employee will suffer a larger cost if he is dismissed and becomes unemployed. Thus, both an increase in the level of supervision and an increase in the salary can be used to elicit a higher level of effort. In equilibrium, the firm will pay a salary above the competitive level and will fire the employee if he is caught ‘‘shirking.’’ In the model, an exogenous reduction in the unemployment rate reduces the employee’s disutility from being dismissed because in case he is fired he expects to be unemployed for a shorter period of time. Since this reduces his incentives to exert effort, the firm will have to increase the level of supervision or the salary. Both options will raise the cost of shirking because in the former case there will be a higher probability of being ‘‘caught,’’ and in the latter case the utility loss from being fired will be higher. Therefore, controlling for the salary, we expect an exogenous reduction in the

95

Employee Discretion and the Labor-Market Environment

unemployment rate to raise the level of supervision and therefore reduce the employee’s discretion. For similar reasons, we can also hypothesize that labor regulations affecting the probability of being dismissed or the expected cost of being unemployed will also influence the levels of supervision. For example, an increase in unemployment benefits or in firing costs (both monetary and nonmonetary) should reduce the employee’s expected disutility from shirking and should lead to a reduction in employee discretion. Similarly, labor regulations that raise the power of unions or strengthen the employees’ bargaining power in collective disputes will lead to more supervision and less discretion if those regulations make it more difficult for firms to fire employees. Therefore, controlling for the salaries, a more protective labor-market environment should lead to more supervision and less employee discretion.3

2.3. Predictions Table 1 summarizes the predicted effects of the unemployment rate and employee protection on discretion. As just pointed out, efficiency wage predictions are conditional on salary levels, whereas the bargaining approach predictions are not, that is, they are valid as long as salaries are allowed to adjust. Since the bargaining outcomes depend upon employees’ and employers’ preferences for discretion, Table 1 distinguishes among three possible scenarios: one in which both sides prefer to introduce more discretion and two others in which only one of the parties has a preference for discretion. As shown in the table, the predicted effects are different in each case. On the other hand, the efficiency wage approach suggests a positive effect of unemployment and a negative effect of employee protection. Table 1.

Theoretical Predictions. Predicted Effect on Discretion

Bargaining approach

Efficiency wages

Discretion is beneficial to both parties Discretion is costly to employees and beneficial to the firm Discretion is beneficial to employees and costly to the firm

Unemployment rate

Employee protection

0 þ

0 



þ

þ



96

JAIME ORTEGA

3. DATA DESCRIPTION For the purposes of this study, I link data from three different sources: the third EWCS, the dataset of labor regulations constructed by Botero, Dankjov, Laporta, Lo´pez de Silanes, and Shleifer (2004), and Eurostat’s Labor Force Survey (LFS). The EWCS data are at the individual (employee) level, whereas the regulation and LFS data are at the national and regional level, respectively. The third EWCS was conducted in 2000 and 2001 by the European Foundation for the Improvement of Living and Working Conditions, a European Union (EU) government agency, and provides a cross section of employed and self-employed individuals in 27 EU member states.4 The survey includes information about discretion, education and pay, among other variables, and covers approximately 26,000 employees in total. Individuals were interviewed in their homes, and the overall response rate was 48 percent. The dataset from Botero et al. (2004) refers to 1997 and consists of a summary of national labor laws in 85 countries, and the LFS is the official source of unemployment statistics in the EU. The variables taken from these data sources can be classified into four groups: employee discretion variables, labor-market characteristics, controls for job characteristics, and other controls. All these variables come from the EWCS except those on labor-market characteristics, which are drawn from the two other data sources.

3.1. Employee Discretion The EWCS contains six questions about employee discretion, which concern the extent to which employees can choose the order in which they perform tasks, the methods of work, the speed of work, the number of work hours, the times at which they take breaks, and the days in which they take their holidays (see Tables 2 and 4). Answers to all six questions are dichotomous (yes/no). Following the job design literature, and particularly the work of Hackman and Oldham (1975) on job autonomy, I distinguish two aspects of discretion: discretion to choose the methods of work and discretion to choose the timing of work.5 Furthermore, factor analyses conducted in Ortega (2009b) with the same data suggest two summary measures of discretion, DiMeth and DiSched. DiMeth (discretion to choose work methods) is the average of the standardized variables on order, methods, and speed, and DiSched (discretion to choose work schedules) is the average of the standardized variables on work hours, breaks, and holidays.

Employee Discretion and the Labor-Market Environment

Table 2. Variables Order Methods Speed DiMeth Breaks Hours Holidays DiSched Complex Monotonous Repetitive Learning new things Interruptions Interruptions due to nature of work Work pace dependence

97

Description of EWCS Variables. Definitions/EWCS Questions

Are you able to choose or change your order of tasks? The methods of work? Your speed or rate of work? Average of standardized variables order, methods, and speed Can you take your break when you wish? Are you able to influence your work hours? Are you free to decide when to take holidays and days off? Average of standardized variables breaks, hours, and holidays Generally, does your main paid job involve complex tasks? Monotonous tasks? Short repetitive tasks of less than 10 minutes? Learning new things? How often do you have to interrupt a task you are doing in order to take on a unforeseen task? Are those interruptions mainly due to the nature of your work?

On the whole, is your pace of work dependent on the work done by colleagues? Team Does your job involve doing all or part of your work in a team? Job rotation Does your job involve rotating tasks between yourself and your colleagues? Vertical communication Do you exchange views with your superiors about your working conditions in general or about the organization of your work when changes take place? Improvement Do these exchanges of views lead to improvements at your own personal workplace/in your office or factory/in the organization as a whole? Piece rates Does your remuneration include piece rate or productivity payments? Group performance pay Payments based on the overall performance of a group? Profit sharing Payments based on the overall performance of the company (profit sharing) where you work? Stock ownership Income from shares in the company you work for? Performance pay Employee receives at least one form of performance pay (piece rate, group performance pay, profit sharing, or stock ownership) Number of subordinates How many people work under your supervision, for whom pay increases, bonuses, or promotion depend directly on you? Establishment size How many people in total work in the local unit of the establishment where you work? Firm tenure How many years have you been in your company or organization? (If less than one year) How many months? Income Net monthly income quantile at country level, based on Eurostat’s European Earnings Structure Survey 2002, adjusted for inflation (12 intervals) Occupation One-digit ISCO-88 occupation Industry One-digit NACE industry

98

JAIME ORTEGA

3.2. Labor-Market Characteristics I use six labor-market variables, which are defined in Table 3 and summarized in Table 4. The unemployment rate is taken from the LFS and refers to 2000 or 2001 (depending on whether the corresponding country was surveyed for the EWCS in 2000 or 2001, respectively). Because labor mobility is very low in Europe, I use the regional rates of unemployment (for two-digit NUTS6) instead of the national rates. The other variables are taken from Botero et al. (2004) and measure five different aspects of national labor regulations: firing costs, administrative constraints on dismissal, unemployment benefits, labor union power, and employee protection during collective disputes (see Table 3 for details). The main advantage of these variables is that they are consistently defined across countries. The firing costs index measures the monetary cost of dismissing 20 percent of a 250-employee firm’s workforce (10 percent for redundancy and 10 percent without a cause). It is computed as the pay equivalent of the notice period plus the severance payment and any other monetary penalties linked to the dismissal decision. The dismissal procedures index measures the extent to which there are administrative constraints on dismissal. Specifically, it measures whether employers have to notify or need approval from a third party to dismiss employees with or without a cause; whether employers must provide training or relocation alternatives for redundant workers; and whether there are compulsory priority rules applicable to dismissals, layoffs or reemployment. Unemployment benefits is an index including the number of months required to qualify for unemployment benefits, the percentage of monthly salary deducted to cover unemployment benefits, the waiting period and the percentage of salary covered by the benefits in the case of a one-year unemployment spell. Finally, labor union power measures union rights and collective bargaining rules, and collective disputes measure the level of employee protection during collective disputes (see Table 3 for details). All indices are defined in such a way that a higher value of the variable means that employees have more bargaining power.

3.3. Controls for Job Characteristics Since the aim of this study is to estimate the effect of labor-market institutions on employee discretion but data are cross-sectional and encompass a

99

Employee Discretion and the Labor-Market Environment

Table 3. Variable

Description of Labor Market Variables. Definition

Source

Unemployment rate

Unemployment rate in NUTS-2 regions

Eurostat

Firing costs

Index measuring the cost of firing 20 percent of the firm’s workers (10 percent for redundancy and 10 percent without cause). The cost of firing a worker is calculated as the sum of the notice period, severance pay, and any mandatory penalties established by law or mandatory collective agreements for a worker with three years of tenure with the firm. If dismissal is illegal, the cost of firing is set equal to the annual wage. The new wage bill incorporates the normal wage of the remaining workers and the cost of firing workers. The cost of firing workers is computed as the ratio of the new wage bill to the old one.

Botero et al. (2004)

Dismissal procedures

Index measuring protection against dismissal granted by law or mandatory collective agreement. It is the average of the following seven dummy variables which equal one: (1) if the employer must notify a third party before dismissing more than one worker; (2) if the employer needs the approval of a third party prior to dismissing more than one worker; (3) if the employer must notify a third party before dismissing one redundant worker; (4) if the employer needs the approval of a third party to dismiss one redundant worker; (5) if the employer must provide relocation or retraining alternatives for redundant employees prior to dismissal; (6) if there are priority rules applying to dismissal or lay-offs; and (7) if there are priority rules applying to reemployment.

Botero et al. (2004)

Unemployment benefits

Measures the level of unemployment benefits as the average of the following four normalized variables: (1) the number of months of contributions or employment required to qualify for unemployment benefits by law; (2) the percentage of the worker’s monthly salary deducted by law to cover unemployment benefits; (3) the waiting period for unemployment benefits; and (4) the percentage of the net salary covered by the net unemployment benefits in case of a oneyear unemployment spell.

Botero et al. (2004)

100

JAIME ORTEGA

Table 3. (Continued ) Variable

Definition

Source

Labor union power

Measures the statutory protection and power of unions as the average of the following seven dummy variables that equal one: (1) if employees have the right to unionize; (2) if employees have the right to collective bargaining; (3) if employees have the legal duty to bargain with unions; (4) if collective contracts are extended to third parties by law; (5) if the law allows closed shops; (6) if workers, or unions, or both have a right to appoint members to the Boards of Directors; and (7) if workers’ councils are mandated by law.

Botero et al. (2004)

Collective disputes

Measures the protection of workers during collective disputes as the average of the following eight variables: (1) if wildcat, political, and sympathy/ solidarity/secondary strikes are legal (legal strikes); (2) if employer lockouts are illegal; (3) if workers have the right to industrial action; (4) if there is no mandatory waiting period or notification requirement before strikes can occur; (5) if striking is legal even if there is a collective agreement in force; (6) if laws do not mandate conciliation procedures before a strike; (7) if third-party arbitration during a labor dispute is mandated by law; and (8) if it is illegal to fire or replace striking workers.

Botero et al. (2004)

Table 4.

Summary Statistics.

Variable

Mean

SD

Min

Max

N

DiMeth DiSched Unemployment rate Firing costs Dismissal procedures Unemployment benefits Labor union power Collective disputes

.074 .123 9.083 0.48 0.47 0.80 0.52 0.47

.848 .717 5.389 0.14 0.24 0.07 0.17 0.15

1.472 1.058 2.1 0.16 0.14 0.66 0.00 0.13

.682 .966 26.0 0.69 0.86 0.96 0.71 0.83

26,375 26,095 227 26 26 26 26 26

Employee Discretion and the Labor-Market Environment

101

great variety of employees, it is necessary to control for job characteristics that can potentially influence the choice of discretion. I classify these control variables into two groups. The first group includes job characteristics that are not tied to specific work practices, and the second group includes job characteristics that are implied by the use of certain work practices. Within the first group, I consider one measure of task dependence, three measures of job complexity, one measure of learning requirements on the job, and two measures of work interruptions (see Table 2 for definitions). The second group of variables includes measures of teamwork and job rotation, communication between employees and superiors, and performance pay. As far as the first group is concerned, task dependence is an important control because there are cases in which the work of an employee is so tied to the work of colleagues that there is simply little room for discretion. The measures of job complexity are based on three separate questions on whether the job is complex, monotonous, and repetitive. Controlling for job complexity is necessary because according to the literature on delegation firms give more discretion when employees have more specific knowledge, and this is more likely to happen when jobs are more complex (Baker, 1992; Prendergast, 2002; Raith, 2008). Moreover, using EWCS data Ortega (2009b) has found a positive relationship between discretion and complexity. The reason for including a measure of learning requirements is similar because jobs with stronger learning requirements are typically more complex. Finally, the two work interruptions variables capture the uncertainty that the employee is subject to. The aforementioned literature on delegation develops the idea that in jobs subject to more uncertainty, the potential benefits from delegation are higher because employees tend to have a greater informational advantage over their superiors. This suggests using some measure of work uncertainty as a control variable. The two measures of uncertainty that have been chosen for this purpose are interesting because they are directly related to the job (‘‘How often do you have to interrupt a task you are doing to take on an unforeseen task?’’ and ‘‘Are those interruptions mainly due to the nature of your work?’’). The second group of job-related control variables measures to what extent the employee is involved in ‘‘innovative’’ forms of work organization. Team and job rotation are dichotomous variables equal to one when the employee works in a team or rotates (respectively). I also include two measures for whether employees communicate with superiors (‘‘vertical communication’’) and whether such communication leads to work improvements (‘‘improvement’’), and a dichotomous variable equal to one when the employee

102

JAIME ORTEGA

receives some form of performance-related pay (‘‘performance pay’’). These are important controls because one of the objectives of these practices is to promote greater employee involvement.

3.4. Other Control Variables The remaining control variables are also taken from the EWCS and therefore vary across individuals. Specifically, I use the number of subordinates, establishment size, firm tenure, job tenure, firm tenure  job tenure, age, age squared, sex, net income, industry, and occupation as control variables. The number of subordinates is defined as the number of individuals whose pay or promotion are decided by the respondent (categorized in four intervals), and establishment size is measured by the number of employees (categorized in eight intervals). Industries and occupations are measured at one-digit level according to the Classification of Economic Activities in the European Community (NACE) and the International Standard Classification of Occupations (ISCO), respectively. As far as income is concerned, for each participating country the questionnaire provides 12 quantiles based on Eurostat’s European Earnings Structure Survey and asks respondents to choose the interval that corresponds to their net monthly income. Based on these responses, I create one dichotomous variable for each quantile. This has two important implications: first, cross-country differences in income levels are not captured by the data and second, differences in income among employees are reported with error.

4. RESULTS Correlations among the main variables of interest (employee discretion, unemployment, and labor regulations) are shown in Table 5. The unemployment rate is negatively correlated with the two measures of discretion and four (out of five) measures of labor regulations are negatively correlated with discretion over work schedules, DiSched. Moreover, the only labor law index that is not negatively correlated with DiSched does not have a significantly positive correlation. While there seems to be a clear negative correlation between labor regulations and DiSched, the correlations between labor regulations and the other measure of discretion, DiMeth, do not seem to follow a clear pattern: two indices (firing costs and dismissal procedures) have a positive correlation with DiMeth, two other

103

Employee Discretion and the Labor-Market Environment

Table 5.

1. 2. 3. 4. 5. 6. 7. 8.

DiMeth DiSched Unemployment rate Firing costs Dismissal procedures Unemployment benefits Labor union power Collective disputes

Correlations.

(1)

(2)

(3)

(4)

(5)

(6)

(7)

1.000 .364 .077 .022 .032 .056 .004 .091

1.000 .137 .021 .003 .058 .029 .050

1.000 .098 .059 .079 .039 .116

1.000 .342 .086 .053 .084

1.000 .303 .471 .248

1.000 .370 .243

1.000 .004

Note: Levels of significance: 10 percent; 5 percent; 1 percent.

indices (unemployment benefits and collectives disputes) have a negative correlation with DiMeth, and for the fifth index (labor union power) the correlation is positive but insignificant. Tables 6–7 report the results from different sets of seemingly unrelated regression equations with the two types of employee discretion as lefthand side variables. Table 6 reports results of minimalist regressions in which the only right-hand side variables are the unemployment rate, the labor law indices, and all control variables except the job characteristics. The unemployment rate, unemployment benefits, and collective disputes have a negative relationship with both types of discretion, and labor union power is insignificant in most cases. The two firing cost variables (firing costs and dismissal procedures) have a positive relationship with discretion over methods, but their effect on discretion over schedules is different – firing costs has a negative effect and dismissal procedures has a positive effect. Finally, the signs and significance of the variables of interest coefficients do not change with the introduction of control variables for income. Table 7 shows the results obtained in regressions with all the right-hand side variables used for Table 6 and all the job characteristics controls. The results are very similar to those in Table 6. In fact, except for some changes in the levels of coefficients, the only worth-noting difference is that labor union power is positively associated with discretion over work methods. Table 7 also shows that most job-related controls have the expected effect on discretion: the job complexity proxies, learning new things, interruptions, vertical communication, improvement and performance pay have positive coefficients and work pace dependence has a negative coefficient. The only two controls that do not have the expected sign are teamwork and job rotation, but those are only significant in half of the cases. .

104

JAIME ORTEGA

Table 6.

Regression Results, All Employees. (1)

Unemployment rate Firing costs Dismissal procedures Unemployment benefits Labor union power Collective disputes N R2

(2)

DiMeth

DiSched

DiMeth

DiSched

.010 (.001) .135 (.054) .270 (.030) .525 (.089) .006 (.037) .441 (.036) 22,157 .136

.015 (.001) .133 (.039) .187 (.025) .314 (.076) .109 (.032) .193 (.031) 22,157 .128

.009 (.001) .086 (.051) .298 (.032) .347 (.099) .026 (.043) .502 (.039) 17,810 .150

.013 (.001) .170 (.043) .217 (.027) .172 (.084) .046 (.037) .238 (.033) 17,810 .140

Note: Seemingly unrelated regressions for DiMeth and DiSched. In addition to the variables in the table, I use the following explanatory variables: Column (1): three dichotomous variables for the number of subordinates, seven dichotomous variables for establishment size, firm tenure, job tenure, firm tenure  job tenure, age, age squared, one dichotomous variable for female, two dichotomous variables for permanent contract and fixed-term contract (respectively), and 14 and 9 dichotomous variables for industry and occupation respectively. Column (2): same explanatory variables and 11 dichotomous variables for income quantiles. Standard deviations are shown in parentheses. Levels of significance: 10%; 5%; 1%.

I have also estimated other sets of seemingly unrelated regressions but results are generally similar to Tables 6–7, except for changes in the levels of significance (tables are omitted to save space). I have estimated separate sets of regressions for employees with a permanent contract and those with a fixed-term contract, but the signs of the coefficients do not change. There are no significant changes either when I take public administration and state-owned firms’ employees out of the sample. I have also restricted the sample to employees who are involved in work practices that are usually associated with more employee participation in decision making. Specifically, I estimate the seemingly unrelated regressions of Table 7 for employees who satisfy three conditions: (i) they are involved in teamwork or in job rotation, (ii) they receive at least one type of performance-related pay, and (iii) the level of communication with their superior is above the full sample average. I find results similar to the ones in Table 7. Finally, I have also added to the Table 7 regressions three interaction terms between tenure

105

Employee Discretion and the Labor-Market Environment

Table 7.

Regression Results with Controls for Job Characteristics, All Employees. (1)

Unemployment rate Firing costs Dismissal procedures Unemployment benefits Labor union power Collective disputes Complex Monotonous Repetitive Learning new things Interruptions Interruptions due to nature of work Work pace dependence Team Job rotation Vertical communication Improvement Performance pay N R2

(2)

DiMeth

DiSched

DiMeth

DiSched

.004 (.001) .323 (.059) .064 (.038) .398 (.112) .126 (.047) .219 (.046) .140 (.016) .096 (.015) .074 (.014) .184 (.019) .145 (.025) .000 (.015) .117 (.014) .032 (.016) .014 (.015) .098 (.024) .079 (.009) .058 (.019) 11,245 .151

.012 (.001) .115 (.054) .074 (.036) .162 (.103) .013 (.043) .166 (.042) .030 (.014) .056 (.014) .100 (.013) .039 (.017) .062 (.023) .020 (.013) .081 (.013) .006 (.015) .035 (.014) .083 (.022) .085 (.008) .093 (.017) 11,245 .157

.003 (.001) .294 (.066) .075 (.042) .325 (.123) .127 (.052) .245 (.050) .135 (.017) .090 (.016) .059 (.016) .185 (.020) .134 (.027) .012 (.016) .119 (.016) .039 (.018) .023 (.016) .094 (.026) .077 (.010) .042 (.020) 9,234 .171

.011 (.001) .217 (.061) .138 (.038) .013 (.113) .017 (.048) .254 (.046) .027 (.016) .065 (.015) .079 (.014) .029 (.019) .051 (.025) .011 (.015) .081 (.014) .015 (.017) .038 (.015) .076 (.024) .082 (.009) .086 (.019) 9,234 .161

Note: Seemingly unrelated equations with the same controls as in Table 6. Column (1) does not include income dummies and column (2) does. Standard deviations are shown in parentheses. Levels of significance: 10 percent; 5 percent; 1 percent.

106

JAIME ORTEGA

in the firm on the one hand and firing costs, dismissal procedures, and unemployment benefits on the other hand, to take into account that the costs of dismissing an employee and the level of unemployment insurance usually vary with seniority. However, the results are also very similar to those in Table 7.7

5. DISCUSSION AND CONCLUSIONS The data analysis shows a negative relationship between the unemployment rate and both types of discretion, which is significant through all specifications. This result contradicts the efficiency wage argument that firms tend to monitor less when the employees’ personal cost of becoming unemployed is larger. However, it is consistent with a setup in which (i) employees like to have discretion; (ii) firms are reluctant to give discretion; and (iii) employees obtain more discretion the larger their bargaining power. According to this approach, the reason why employees in high unemployment regions tend to have less discretion is that in those regions firms have more bargaining power over workers than in low unemployment regions. As far as labor-market institutions are concerned, I find differences among three areas of regulation: firing costs (both monetary and nonmonetary), unemployment benefits, and collective rights (power of unions and worker protection during collective disputes). Let us begin with firing costs. The analysis shows that while nonmonetary costs have a positive relationship with both aspects of discretion, monetary costs have a positive relationship with discretion over work methods and a negative relationship with discretion over schedules. Thus, the effect of firing costs on work methods runs clearly counter to the efficiency wage theory (and is consistent with the bargaining approach), but the effects on schedules are not fully consistent with any of the two theoretical frameworks. All in all, we could therefore say that the estimated coefficients for firing costs lend more support to the bargaining approach than to the efficiency wage approach, but this evidence is not unambiguous. Another important area of labor regulation is unemployment insurance, and here the results are clearly supportive of the efficiency wage story, that is, countries that offer higher levels of unemployment benefits tend to have lower levels of discretion. In an efficiency wage model, this result would be interpreted as evidence that when workers have higher unemployment insurance, the cost of being fired is smaller and their motivation goes down. As a result, the firm needs to monitor more. This negative effect of

Employee Discretion and the Labor-Market Environment

107

unemployment benefits on discretion would also be consistent with a bargaining setup in which discretion is costly to employees and beneficial to employers, but that setup would be inconsistent with the negative effect of the unemployment rate and positive effect of firing costs on discretion. The third type of labor regulations have to do with collective rights – the protection of workers during collective disputes and the power of labor unions. Stronger worker protection is associated with lower levels of discretion, which is consistent with efficiency wages, but labor union power is associated with more discretion over work methods (and has generally no significant relationship with discretion over schedules). Therefore the relationship between collective rights and discretion does not lend clear support to any of the two theoretical approaches. This study has some clear limitations. First, while I benefit from great variability in unemployment across regions, the fact that data on labor institutions only vary from country to country imposes an important restriction. This may actually explain why evidence on labor regulations is sometimes inconclusive. Second, all the variation is cross-sectional and therefore I cannot control for individual or firm-fixed effects, and I cannot claim causality for the empirical relationships that I estimate. Note, however, that a natural experiment study would have to face the problem that it takes time for firms to redesign jobs in such a way as to increase (or reduce) discretion. Third, data on employees’ net incomes are not available in monetary units and the results obtained do not change very much when I introduce the net income controls, which may be due to measurement error. Despite these limitations, I think this study makes an interesting point, namely that the extent to which firms adopt a more participative approach to work is influenced by the labor market in which they operate. Although the idea that labor regulations have great influence on firms’ employment decisions is widely spread, less attention is usually given to the fact that certain labor regulations can also influence the way work is organized within firms.

NOTES 1. Groshen and Krueger (1990), Rebitzer (1995), and Ewing and Payne (1999) find evidence of a negative correlation between supervision and wages, whereas Leonard (1987) and Neal (1993) do not find any evidence. 2. See also Ortega (2009a), which compares the view that employers give discretion to facilitate work–family balance with the view that discretion is mostly given to improve performance. Results are more consistent with the latter view.

108

JAIME ORTEGA

3. One caveat is that, when bargaining power changes, the firm may choose to elicit a different level of effort, in which case the above-mentioned effects would be ambiguous. For example, consider an increase in unemployment. On the one hand, it must cause an increase in discretion because there will be more market discipline, but on the other hand the firm may want to reduce the level of effort elicited from the agent, in which case there will be a reduction in discretion. This caveat, however, applies to all the empirical literature on efficiency wages, which tests for a negative relationship between supervision and wages and does not take into account these possible ambiguities (see Allgulin & Ellingsen, 2002). 4. Austria, Belgium, Bulgaria, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain, Sweden, and United Kingdom. 5. Hackman and Oldham (1975) define autonomy as ‘‘the degree to which the job provides substantial freedom, independence, and discretion to the employee in scheduling the work and in determining the procedures to be used in carrying it out.’’ 6. The NUTS system is based on both administrative and demographic criteria. In particular, NUTS-2 units are political or administrative units with a population comprised between 0.8 and 3 million inhabitants. As there are countries where the organization of regions does not include units in that population range, NUTS-2 units sometimes have a population larger than 3 million. 7. The interactions of dismissal procedures and unemployment benefits with firm tenure are always insignificant and all the variables of interest except firing costs (unemployment rate, dismissal procedures, unemployment benefits, labor union power, and collective disputes) have the same effects as in Table 7. The effect of firing costs on discretion over methods is also the same as in Table 7. Its effect on discretion over schedules is insignificant, but the interaction term (firing costs  firm tenure) has a negative coefficient. (Note that in Table 7 the effect of firing costs on discretion over schedules is negative.)

ACKNOWLEDGMENTS This research has been partially funded by the Spanish Ministry of Science and Innovation (research grant ECO2009-08278). I thank the European Foundation for the Improvement of Living and Working Conditions for kindly providing access to the third European Working Conditions Survey and two anonymous reviewers for helpful suggestions.

REFERENCES Allgulin, M., & Ellingsen, T. (2002). Monitoring and pay. Journal of Labor Economics, 20(2), 201–216.

Employee Discretion and the Labor-Market Environment

109

Appelbaum, E., Bailey, T., Berg, P., & Kalleberg, A. (2000). Manufacturing advantage. Why high performance work systems pay off. Ithaca, NY: ILR Press. Arai, M. (1994). Compensating wage differentials versus efficiency wages: An empirical study of job autonomy and wages. Industrial Relations, 33(2), 249–262. Baker, G. (1992). Incentive contracts and performance measurement. Journal of Political Economy, 100(3), 598–614. Botero, J., Dankjov, S., Laporta, R., Lo´pez de Silanes, F., & Shleifer, A. (2004). The regulation of labor. Quarterly Journal of Economics, 119(4), 1339–1382. Clark, P., Clark, D., Day, D., & Shea, D. (2001). Healthcare reform and the workplace experience of nurses: Implications for patient care and union organizing. Industrial and Labor Relations Review, 55(1), 133–148. Coase, R. (1960). The problem of social cost. Journal of Law and Economics, 3, 1–44. Doellgast, V., Holtgrewe, U., & Deery, S. (2009). The effects of national institutions and collective bargaining arrangements on job quality in front-line service workplaces. Industrial and Labor Relations Review, 62(4), 489–509. Ewing, B., & Payne, J. (1999). The trade-off between supervision and wages: Evidence of efficiency wages from the NLSY. Southern Economic Journal, 66(2), 424–432. Foss, N., & Laursen, K. (2005). Performance pay, delegation and multitasking under uncertainty and innovativeness: An empirical investigation. Journal of Economic Behavior and Organization, 58(2), 246–276. Golden, L. (2001). Flexible work schedules: Which workers get them? American Behavioral Scientist, 44(7), 1157–1178. Goodstein, J. (1994). Institutional pressures and strategic responsiveness: Employer involvement in work-family issues. Academy of Management Journal, 37(2), 350–382. Groshen, E., & Krueger, A. (1990). The structure of supervision and pay in hospitals. Industrial and Labor Relations Review, 43(3), 134S–146S. Hackman, R., & Oldham, G. (1975). Development of the job diagnostic survey. Journal of Applied Psychology, 60(2), 159–170. Ichniowski, C., & Shaw, K. (1999). The effects of human resource management systems on economic performance: An international comparison of U.S. and Japanese plants. Management Science, 45(5), 704–722. Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects of human resource management practices on productivity. American Economic Review, 87(3), 291–313. Karasek, R. (1979). Job demands, job decision latitude, and mental strain: Implications for job redesign. Administrative Science Quarterly, 24(2), 285–308. Leonard, J. (1987). Carrots and sticks: Pay, supervision and turnover. Journal of Labor Economics, 5(4), 136–152. MacLeod, B., & Parent, D. (1999). Job characteristics and the form of compensation. Research in Labor Economics, 18, 177–242. Milgrom, P., & Roberts, J. (1992). Economics, organization and management. Englewood Cliffs, NJ: Prentice-Hall. Nagar, V. (2002). Delegation and incentive compensation. The Accounting Review, 77(2), 379–395. Neal, D. (1993). Supervision and wages across industries. Review of Economics and Statistics, 75(3), 409–417. Ortega, J. (2009a). Why do employers give discretion? Family versus performance concerns. Industrial Relations, 48(1), 1–26.

110

JAIME ORTEGA

Ortega, J. (2009b). Employee discretion and performance pay. The Accounting Review, 84(2), 589–612. Osterman, P. (1994). Supervision, discretion and work organization. American Economic Review, 84(2), 380–384. Osterman, P. (1995). Work/family programs and the employment relationship. Administrative Science Quarterly, 40(4), 681–700. Parker, S. (2003). Longitudinal effects of lean production on employee outcomes and the mediating role of work characteristics. Journal of Applied Psychology, 88(4), 620–634. Parker, S., & Wall, T. (1998). Job design and work design: Organizing work to promote well-being and effectiveness. Thousand Oaks, CA: Sage. Prendergast, C. (2002). The tenuous trade-off between risk and incentives. Journal of Political Economy, 110(5), 1071–1102. Raith, M. (2008). Specific knowledge and performance measurement. Rand Journal of Economics, 39(4), 1059–1079. Rebitzer, J. (1995). Is there a trade-off between supervision and wages? An empirical test of efficiency wage theory. Journal of Economic Behavior and Organization, 28(1), 107–129. Shapiro, C., & Stiglitz, J. (1984). Involuntary unemployment as a worker discipline device. American Economic Review, 74(3), 433–444. Singh, J. (2000). Performance productivity and quality of frontline employees in service organizations. Journal of Marketing, 64(2), 15–34. Wood, S., de Menezes, L., & Lasaosa, A. (2003). Family-friendly management in great Britain: Testing various perspectives. Industrial Relations, 42(2), 221–250.

THE PRODUCTIVITY EFFECTS OF PROFIT SHARING, EMPLOYEE OWNERSHIP, STOCK OPTION AND TEAM INCENTIVE PLANS: EVIDENCE FROM KOREAN PANEL DATA Takao Kato, Ju Ho Lee and Jang-Soo Ryu 1. INTRODUCTION Compensation systems have been shifting away rapidly from a fixed-wage contractual payment basis in many nations around the world (Ben-Ner & Jones, 1995). Particularly prominent is the explosion in the use and interest in employee financial participation schemes, such as profit sharing, employee stock ownership, stock option, and team incentive (or gainsharing) plans. With the rising use and interest in such employee financial participation schemes, many studies have examined their effects on enterprise performance in industrialized countries.1 Most prior studies consider either profit sharing plans (PSPs) in which at least part of the compensation for no executive employees is dependent on firm performance (typically profit)2 or employee stock ownership plans (ESOPs) through which the firm forms an ESOP trust consisting of its nonexecutive employees and promotes Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 111–135 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011009

111

112

TAKAO KATO ET AL.

ownership of its own shares by the trust.3 Moreover, an increasing number of firms (in particular ‘‘new economy’’ firms) are extending the use of stock option plans (SOPs) to include nonexecutive employees in recent years.4 Finally, with the rising popularity of ‘‘high-performance workplace practices (notably self-directed teams),’’ more firms are introducing team incentive plans (TIPs), which makes at least part of the compensation for employees dependent on performance of the team or work group to which they belong.5 Outside of major industrialized countries, in particular newly industrialized countries (NICs), however, research on the performance effects of such schemes is extremely limited.6 The paucity of systematic studies on employee financial participation schemes in NICs (in particular South Korea) is surprising considering the growing importance of such economies (e.g., South Korea is currently among 15 largest economies in the world). This chapter provides much-needed rigorous evidence on the incidence, diffusion, scope, and effects of various forms of financial participation schemes in South Korea. We do so by using a new comprehensive survey of publicly traded firms in South Korea, the Survey on Human Resource Management (SHRM). The survey was conducted by the Korea Labor Institute in August–September in 2000. The sample universe of the SHRM was all firms listed in Korea Stock Exchange (KSE) at the time of the survey (there were about 600 of them). The response rate was over 60%, which compares very favorably to similar firm-level surveys conducted in the past in other industrialized countries.7 The unusually high response rate of the SHRM makes the Korean data particularly appealing. In the next section, we begin with documenting the incidence, diffusion, and scope of Korean financial participation schemes using our new survey data. We then combine the survey data with the corporate proxy statement database assembled by the Korea Listed Companies Association (KLCA) to create for the first time a panel data set for over 200 Korean manufacturing firms that provides data on the use of a variety of financial participation schemes as well as value-added, labor input and capital input over 1990– 2000. The data are then used to estimate production functions with fixed effects, augmented by variables to capture the effects of these financial participation schemes. We find consistently that the introduction of PSPs and TIPs will lead to a significant increase in productivity (a little over 10%), whereas we find no such evidence for stock-based schemes, such as ESOPs and SOPs. We also find that the productivity payoff appears to be more long-lasting for PSPs than for TIPs. Finally, our fixed-effect estimates suggest that PSPs and

The Productivity Effects of Financial Participation in Korea

113

TIPs tend to be substitutes rather than complements in their productivity effects. The chapter is organized as follows. In Section 2, we document the incidence, diffusion, and scope of employee financial participation schemes in South Korea. In Section 3, we offer general theoretical arguments for the productivity effects of such financial participation and argue that we expect strong productivity gains from PSPs and TIPs in Korea and somewhat weaker results for stock-based schemes such as ESOPs and SOPs. We then present evidence from our field research at two manufacturing firms in Korea, which demonstrates vividly the strong use of PSPs and TIPs and the weak use or absence of stock-based schemes. In Section 4, we begin by describing the basic empirical strategy and the data, and then present our main empirical results. A concluding section will follow.

2. THE INCIDENCE, DIFFUSION, AND SCOPE OF KOREAN FINANCIAL PARTICIPATION SCHEMES As discussed earlier, employee financial participation comes in various forms. Most prior studies investigate either PSPs or ESOPs, and lately researchers are beginning to study SOPs and TIPs. Fortunately, our Korean data provide information on all four schemes (PSPs, ESOPs, SOPs, and TIPs) and thus grant us an unusual opportunity to examine all four schemes together. In particular, the data enable us to address systematically one of the most often discussed issues in the literature or whether some of these schemes are complements or substitutes.8 According to the SHRM, as shown in Table 1 and Fig. 1, over 35% of publicly traded firms in South Korea had a PSP in 2000. The proportion of firms with a PSP was only 3% in early 1980s and grew steadily to 10% by early 1990s. A significant diffusion occurred during 1990s (in particular late 1990s) in large part due to governmental promotion of PSPs in the 1990s. Table 1 also indicates that there is no discernable difference in the incidence of PSPs between manufacturing and nonmanufacturing firms. As shown in Fig. 1, by mid-1980s, the proportion of firms that had an ESOP was less than 10%. In early 1990s, Korean government advocated ESOPs as a way to help firms obtain much needed capital for investment. The government, using informal channels as well as a tax incentive, encouraged firms to set up new ESOPs to accommodate employee investments in their stock. Perhaps, partly due to this government initiative,

3.18 3.18 3.18 3.18 3.18 3.18 3.47 3.76 4.34 4.62 7.80 8.38 9.25 10.12 11.56 13.87 15.03 16.47 19.65 30.92 35.84

3.52 3.52 3.52 3.52 3.52 3.52 3.96 3.96 4.85 5.29 9.25 10.13 11.01 11.45 12.78 15.86 17.62 18.06 21.15 28.19 32.16

5.92 5.92 6.54 6.85 6.85 8.41 8.72 12.15 18.07 19.00 24.92 27.10 28.35 29.60 33.33 37.69 42.06 47.04 47.66 52.02 52.34

4.57 4.57 5.02 5.02 5.02 7.31 7.76 11.42 17.81 19.18 25.57 27.40 28.77 29.68 33.79 36.99 40.64 45.21 45.66 50.68 51.14

Manufacturing

All

All

Manufacturing

Firms with ESOPs (%)b

Firms with PSPs (%)a

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.83 2.22 7.76 14.96

All 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.83 2.50 6.67 12.50

Manufacturing

Firms with SOPs (%)c

0.00 0.00 0.00 0.28 0.28 0.28 0.28 0.28 0.28 0.28 0.28 0.28 1.40 1.96 2.52 4.48 5.32 6.72 12.61 19.33 23.25

All 0.00 0.00 0.00 0.42 0.42 0.42 0.42 0.42 0.42 0.42 0.42 0.42 1.69 2.12 2.12 3.81 4.66 5.93 11.86 18.64 19.92

Manufacturing

Firms with TIPs (%)d

8.94 8.94 9.60 9.60 9.60 10.93 11.26 13.91 20.53 21.52 29.80 31.79 33.77 35.76 39.74 45.70 50.99 55.96 59.27 69.87 71.52

All

7.80 7.80 8.29 8.29 8.29 10.24 10.73 13.66 20.98 22.44 31.71 34.15 36.10 36.59 40.98 45.37 49.76 54.15 57.56 68.29 69.76

Manufacturing

Firms with At Least One Financial Participation Scheme (%)e

Proportion of Publicly Traded Firms with Employee Financial Participation Schemes in Korea from 1980 to 2000.

Source: Survey on Human Resources Management (SHRM). a Based on 346 firms (227 firms in manufacturing) who provided information on their use of PSPs. b Based on 321 firms (219 firms in manufacturing) who provided information on their use of ESOPs. c Based on 361 firms (240 firms in manufacturing) who provided information on their use of SOPs. d Based on 357 firms (236 firms in manufacturing) who provided information on their use of TIPs. e Based on 302 firms (205 firms in manufacturing) who provided information on their use of PSPs, ESOPs, SOPs and TIPs.

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Year

Table 1. 114 TAKAO KATO ET AL.

%firms with each scheme

Fig. 1.

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

1990

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1989

1988

1987

1986

1985

1984

1983

1982

1981

1980

Diffusion of Financial Participation Schemes among Publicly Traded Firms in Korea: 1980–2000.

Year

%firms with at least one financial participation scheme

%firms with TIPs

%firms witth SOPs

%firms with ESOPs

%firms with PSPs

The Productivity Effects of Financial Participation in Korea 115

116

TAKAO KATO ET AL.

the next decade was characterized by a rapid pace of diffusion of the institution, and over one half of firms came to have ESOPs by 2000. SOPs became legal in 1997 in South Korea. Since then, SOPs have been spreading among Korean firms rapidly, and by 2000, 15% of firms reported to use such a plan. Finally, according to the survey, as shown in Table 1 and Fig. 1, only less than 1% of publicly traded firms used TIPs before the 1990s. In early 1990s, the institution diffused steadily. The rapid diffusion took place after the Asian crisis. In 2000, almost one in four firms used TIPs. Again, the rapid diffusion of TIPs was in part due to public policy initiatives to respond to the Asian Crisis. Table 2 summarizes the scope of each scheme (or how widely each scheme applies to the firm’s labor force). The scope of Korean PSPs compares favorably to such schemes elsewhere. Thus, for Korean firms with PSPs, typically over 90% of employees are covered by a PSP, whereas Kruse (1993) reports that nearly 80% of employees are covered by a PSP in the United States. Furthermore, unlike the United States, Korean PSPs rarely have separate plans for different divisions and occupations or for union and nonunion members. This way, Korean PSPs are more similar to Japanese PSPs (Kato & Morishima, 2003). Nearly all Korean PSPs are cash plans, which is in sharp contrast to the United States where deferred plans are more popular (Kato & Morishima, 2003). As indicated in Table 2, about 30% of the labor force in firms with ESOPs actually participated in the plan. The figure is hardly trivial although still lower than in Japan and the United States (Jones & Kato, 1995). Furthermore, there is much attrition in U.S. ESOPs. This contrasts sharply with the situation in Korea (and Japan) where the rate of termination is negligible. However, in Korea, recently employees were allowed to withdraw their shares only after one year of vesting (it used to be seven years), and hence, there are currently a significant number of ESOPs in Korea with no employee participation. Our data on ESOPs exclude those defunct ESOPs.9 Table 2.

PSPs ESOPs SOPs TIPs

Percentage Employees Covered by Each Plan. N

Mean (%)

SD

133 138 31 65

90.46 29.41 25.88 57.26

24.19 38.19 34.00 34.23

Source: Survey on Human Resources Management (SHRM).

The Productivity Effects of Financial Participation in Korea

117

As given in Table 2, Korean SOPs are fairly broad-based with 26% of all employees in firms with SOPs covered by the plan. Finally, as Table 2 reports, over 50% of employees in firms with TIPs are covered by such pay systems. The penetration of TIPs once introduced is quite high in Korea.

3. PRODUCTIVITY EFFECTS OF FINANCIAL PARTICIPATION SCHEMES: HYPOTHESES Generally, formal economic theory is ambiguous as to the expected impact of employee financial participation schemes on productivity.10 Focusing on individual motivation and performance, however, several hypotheses predict positive productivity effects, of which the goal alignment effects are perhaps most important. In addition, there are a few hypotheses concerning the complementarities and substitutabilities among employee financial participation schemes. PSPs help align the interest of the firm with the interest of its workers by making pay for workers sensitive to firm performance (e.g., profitability). Similarly, the positive productivity effects of ESOPs and SOPs stem from firm success being reflected in a higher price of its stock, and thus increased wealth for workers who own stock in the ESOP. As such, the interest of the firm is more aligned with the interest of its workers. TIPs also help the goal alignment between employees and their teams/work groups and foster team cooperation. As shown in Section 2, considerably higher proportions of the labor force are covered by PSPs and TIPs than ESOPs and SOPs. Thus, we expect greater effects on enterprise-level productivity of PSPs and TIPs as compared to ESOPs and SOPs. The relative importance of cash-based group incentive pay, such as PSPs and TIPs, in Korean firms can be also demonstrated vividly by field research we conducted at two large manufacturing firms, firm A and firm B.11 Firm A is a large manufacturing firm with sales of over 20 trillion won and employment of over 50,000 workers worldwide (close to 30,000 in Korea) in 2003. Firm B is also a manufacturer with sales of over 7 billion won and employment of about 1,600 workers in 2003. Although considerably smaller than firm A, firm B is a leading manufacturer in terms of its market share in their products. Firm A is a publicly traded firm, listed on KSE, whereas firm B is privately owned. Both firms are unionized.12

118

TAKAO KATO ET AL.

Like many other Korean firms, each employee at firm B receives a regular bonus, which amounts to eight months (or 800%) of his/her monthly base pay. As in the case of the bonus payment system in Japan, most Korean employees consider such regular bonuses part of their regular pay. Separate from regular bonus, both firm A and firm B provide all of its employees with PSP bonuses. Firm B has a longer history of PSPs than firm A with firm B’s PSP in existence for at least 10 years. It is broad-based with all employees covered by the same plan with one exception that a specific formula is used to calculate the amount of PSP bonus for top management, whereas no such formulae used for other employees. Thus, except for top management, the amount of PSP bonus is decided through collective bargaining between labor union and management at the end of each year. Both union and management understands that the amount of PSP bonus reflects firm performance although there is neither specific measure used to gauge firm performance nor specific formulae used to calculate the amount of PSP bonus. When firm performance is weak, the amount of PSP bonus can be set to zero. In fact, at the Asian Crisis (1997 and 1998), employees at firm B received no PSP bonus. Except for those two years, the amount of PSP bonus has been positive and on average amount to about two months of monthly base pay. Firm A introduced a new form of PSP in 2000, which applies to all employees. Unlike firm B, firm A uses a specific firm performance measure. As such, once a year, based on firm performance measured by economic value added (EVA),13 the amount of PSP bonus which each employee receives is determined as a percentage of his/her monthly base salary (e.g., the last year of PSP bonus was 200% of base salary). Both firms also use TIPs. For example, at a division of firm A, which we visited, has two departments and each department consists of five production lines. Every month based on line performance (consisting of productivity and quality as compared to targets set by the supervisor and approved by the plant manager), one line in each department will win and receive 300,000 won. How to spend the prize monies is up to each winning line and the winning line often organizes dinner parties for all line members or save some prize monies to fund group trips to popular resorts. The team incentive pay has been in existence for at least 16 years. The supervisor we interviewed strongly believes that their TIP has been a significant motivating factor. Firm B also has a similar TIP with two distinct differences. First, instead of awarding a prize every month, firm B selects the winning team once a year, based on team performance. Team performance is assessed

The Productivity Effects of Financial Participation in Korea

119

based on two criteria: (i) how high the team sets its target in productivity, quality, and cost reduction; and (ii) whether the team actually achieves the target. Second, the prize is a three-day special training trip to a most popular and expensive resort place. For example, the firm paid for a three-day group trip to Je Ju Do (the Korean counterpart to Hawaii) for the winning team. Like in the case of firm A, the winner is decided by the plant managers. Members of the winning team are all excited about winning and proud of the team. Again, our informant at firm B (HR director) believes that their TIP is a great motivator for individual workers and boosts the team spirit. In spite of their reputation as highly participatory firms in Korea, stockbased schemes are much less developed in both firms. Firm A is a publicly traded firm and does have an ESOP for all employees. However, our informant at firm A (HR director) did not hesitate to reveal that the firm has not issued any new stock since 1993 and that the ESOP has been inactive. Firm A has no SOP currently while planning to introduce it to executives only next year. Firm B is not publicly traded and thus has no stock-based incentive plans. Lastly, several scholars have suggested that different forms of alternative compensation schemes may be complementary. Weitzman and Kruse (1990) argue, for example, that PSPs work only when the free rider problem is effectively controlled. When work is done in teams and the size of teams is sufficiently small that each team member can easily observe each other’s effort, TIPs can alleviate the free rider problem. As such, PSPs and TIPs can be complements. However, considering the asymmetric penetration of PSPs and TIPs (90% for PSPs and 60% for TIPs) in Korea, the complementarity effect of TIPs on PSPs may be limited. Our data will allow us to examine whether different forms of financial participation are complements.

4. PRODUCTIVITY EFFECTS OF FINANCIAL PARTICIPATION SCHEMES: FIXED-EFFECT ESTIMATES We follow a methodology used by Jones and Kato (1995) who provided the first systematic evidence on the productivity effects of Japanese financial participation schemes (in particular ESOPs and the Japanese bonus system). That is, we estimate Cobb–Douglas production functions with fixed effects, augmented by a dummy variable capturing the productivity effect of each

120

TAKAO KATO ET AL.

employee financial participation scheme: lnQit ¼ bK lnK it þ bL lnLit þ b1 PSPitj þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð1Þ

lnQit ¼ bK lnK it þ bL lnLit þ b1 ESOPitj þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð2Þ

lnQit ¼ bK lnK it þ bL lnLit þ b1 SOPitj þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð3Þ

lnQit ¼ bK lnK it þ bL lnLit þ b1 TIPitj þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð4Þ

where Qit is output of firm i in year t; Kit is the capital stock; Lit is labor; PSPitj is a dummy variable that takes the value of 1 if PSPs have been in existence for j and more years, and the value of zero otherwise; ESOPitj is a dummy variable that takes the value of 1 if ESOPs have been in existence for j and more years, and the value of zero otherwise; SOPitj is a dummy variable that takes the value of 1 if SOPs have been in existence for j and more years, and the value of zero otherwise; TIPitj is a dummy variable that takes the value of 1 if TIPs have been in existence for j and more years, and the value of zero otherwise; and bs are slope coefficients. For the disturbance term, uit, we assume uitBNID(0, s2). As a robustness check, we also consider translog production functions and find that our key results (the estimated productivity effects of financial participation schemes) do not change when we consider translog production functions.14 The coefficient on PSPj (for the sake of exposition, from now on, we omit subscripts i and t) can be interpreted as indicating the productivity difference between firms that have had PSPs for at least j years and firms with no PSPs. Analogous interpretations can be given to the coefficients on ESOPj, SOPj, and TIPj. We allow for lags (up to three years, or j ¼ 1, 2, and 3) in the productivity effects of employee financial participation schemes. First, it is highly

The Productivity Effects of Financial Participation in Korea

121

unlikely that instituting a financial participation scheme will instantly create significant interest alignment between labor and management. It takes time for the introduced scheme to generate an industrial relations climate conducive to the aforementioned positive productivity effects.15 Second, a newly established financial participation scheme may go through a significant amount of learning by doing in its early developmental years.16 Lastly, in investigating the productivity effects of employee financial participation schemes, ideally we want to use the presence of such schemes (a de jure measure) as well as the extent of employee financial participation schemes (a de facto measure), such as the penetration of such schemes to the labor force. Unfortunately, no panel data on such de facto measures are available for Korean firms. We can, however, measure the extent of employee financial participation schemes indirectly by allowing for lags in the productivity effects of employee financial participation schemes insofar as the extent of employee financial participation schemes is positively correlated with the length of time employee financial participation schemes have been in place (see Freeman & Kleiner, 2000; Kato, 2006 for such positive correlations). In sum, we assume that it will take at least j years ( j ¼ 1, 2, and 3) for a newly introduced employee financial participation scheme to mature and hence realize its full potential for creating significant interest alignment between labor and management. We include firm-specific fixed effects to capture the time-invariant unobserved heterogeneity of our firms. In particular, firm-specific fixed effects will attempt to control for differences among firms in managerial abilities and worker quality. As Wadhwani and Wall (1990) argue, PSPs might be adopted in firms that are better managed. If so, the coefficients on PSPj might indicate the effects of better managers and superior management in general as well as the actual effects of PSPs. To the extent that managerial differences across firms are time-invariant, firm-specific fixed effects will help separate the two effects. Moreover, as Conte and Svejnar (1990) argue, firms with ESOPs might have more productive and more qualified workers than do conventional firms. To the extent that they are time-invariant, firm-specific fixed effects will also capture these quality differences.17 In addition, we include year effects to capture technological change and other shocks that are common to all firms. Industry time trends are also included to capture industry-specific time trends that are common to all firms in each of 15 industries. To test if having had at least one employee financial participation scheme (regardless of which scheme) results in any significant productivity gains,

122

TAKAO KATO ET AL.

we create an additional variable, FPARTitj (a dummy variable that takes the value of 1 if at least one employee financial participation scheme has been in existence for j and more years, and the value of zero otherwise), and estimate: lnQit ¼ bK lnK it þ bL lnLit þ b1 FPARTitj þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð5Þ

Table 3 provides summary statistics. Output is measured by value added deflated by the Producer Price Index (PPI) for manufacturing products at the two-digit industry level (published by the Bank of Korea) for each year.18 The capital stock is proxied by the fixed assets of the firm deflated by the PPI for capital goods (published by the Bank of Korea). Labor is measured by the number of employees. We show data for the whole sample and also disaggregated by whether or not the firm has had each employee financial participation scheme for at least one full year. The table indicates that, compared to firms without a PSP, in PSP firms on average: (i) value added is 140% larger; (ii) employment is 9% smaller; (iii) capital stock is 267% larger; and (iv) capital/labor ratio is 182% higher. Likewise, compared to firms without an ESOP, in ESOP firms on average: (i) value added is 255% larger; (ii) employment is 97% larger; (iii) capital stock is over 400% larger; and (iv) capital/labor ratio is 165% higher. Furthermore, contrasted to firms without an SOP, in firms with SOPs on average: (i) value added is 6% larger; (ii) employment is 59% smaller; (iii) capital stock is 231% larger; and (iv) capital/labor ratio is 266% higher. Finally, compared to firms without a TIP, in firms with TIPs on average: (i) value added is 207% larger; (ii) employment is 9% smaller; (iii) capital stock is 351% larger; and (iv) capital/labor ratio is 253% higher. Table 4 reports the ordinary least square (OLS) estimates of Eqs. (1)–(5) with j ¼ 1.19 The estimated coefficients on the PSP dummy and TIP dummy are positive and significant at the 5% level. By using a simple formula, {exp(bE)1}  100, the size of the productivity gains from having had a PSP for at least one full year is estimated to be 10.6%, and likewise, the size of the productivity gains from having had a TIP for at least one full year is estimated to be 12.2%. On the contrary, the estimated coefficients on the ESOP dummy and SOP dummy are not statistically significant at the 10% level, suggesting no significant productivity gains from having had an ESOP or a SOP for at least one full year. It appears to matter which financial participation scheme is introduced. To confirm this, the estimated coefficient on the FPART

The Productivity Effects of Financial Participation in Korea

Table 3.

All observations N ¼ 1,974 PSP1 ¼ 1 N ¼ 283 PSP1 ¼ 0 N ¼ 1,511 ESOP1 ¼ 1 N ¼ 731 ESOP1 ¼ 0 N ¼ 1,063 SOP1 ¼ 1 N ¼ 18 SOP1 ¼ 0 N ¼ 1,776 TIP1 ¼ 1 N ¼ 85 TIP1 ¼ 0 N ¼ 1,709 FPART1 ¼ 1 N ¼ 870 FPART1 ¼ 0 N ¼ 924

Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD Mean SD

123

Summary Statistics.

Value Added in Billions of Year 2000 Won (Q  109)

Employment (L)

Capital Stock in Billions of Year 2000 Won (K  109)

Capital/ Labor Ratio (K  109/L)

107.43 364.47 211.13 665.30 88.01 269.58 187.02 555.77 52.70 67.77 114.23 352.71 107.36 364.68 300.00 596.73 97.85 346.40 164.68 512.81 53.53 66.72

1605.84 4109.28 1487.43 3257.42 1628.01 4250.43 2267.06 6143.38 1151.13 1434.87 660.39 1010.73 1615.42 4127.77 1462.76 1730.19 1612.95 4192.64 2010.95 5668.36 1224.40 1500.58

3437.14 12596.98 7261.10 23127.88 2720.93 9234.46 6194.15 18954.77 1541.20 3480.66 7841.21 27936.48 3392.50 12353.94 10792.54 24161.94 3071.30 11620.81 5416.03 17482.62 1573.89 3650.87

1.73 2.25 2.78 3.74 1.530 1.777 2.26 2.91 1.36 1.55 4.51 7.10 1.70 2.13 4.08 5.39 1.61 1.90 2.21 2.77 1.28 1.48

Source: Survey on Human Resources Management (SHRM) and corporate proxy statement data from KLCA (Korea Listed Companies Association). Notes: (1) PSP1 is a dummy variable that takes the value of 1 if PSPs have been in existence for at least one year, and the value of zero otherwise. Likewise, ESOP1 is a dummy variable that takes the value of 1 if ESOPs have been in existence for at least one year, and the value of zero otherwise; SOP1 is a dummy variable that takes the value of 1 if SOPs have been in existence for at least one year, and the value of zero otherwise; TIP1 is a dummy variable that takes the value of 1 if TIPs have been in existence for at least one year, and the value of zero otherwise; and FPART1 is a dummy variable that takes the value of 1 if at least one of the above four financial participation schemes has been in existence for at least one year, and the value of zero otherwise; (2) For all regressions, there were 1,794 observations on 205 firms over the 1990–2000 time period. All models include firm-specific fixed effects, year effects, and industry time trends.

dummy is also not statistically significant at the 10% level, suggesting no significant productivity gain from having had at least one financial participation scheme for one or more years with no regard to the types of financial participation schemes.

0.928

3.509 14.152 2.204

0.112 0.569 0.106

0.928

0.054

0.118 0.570

Parameter Estimates

1.287

3.699 14.154

Value of t-Statistics

0.928

0.036

0.116 0.568

Parameter Estimates

0.347

3.650 14.101

Value of t-Statistics

0.928

0.122

0.110 0.566

Parameter Estimates

2.199

3.460 14.066

Value of t-Statistics

0.029 0.928

0.117 0.567

Parameter Estimates

0.776

3.670 14.083

Value of t-Statistics

Source: Survey on Human Resources Management (SHRM) and corporate proxy statement data from KLCA (Korea Listed Companies Association). Notes: (1) PSP1 is a dummy variable that takes the value of 1 if PSPs have been in existence for at least one year, and the value of zero otherwise. Likewise, ESOP1 is a dummy variable that takes the value of 1 if ESOPs have been in existence for at least one year, and the value of zero otherwise; SOP1 is a dummy variable that takes the value of 1 if SOPs have been in existence for at least one year, and the value of zero otherwise; TIP1 is a dummy variable that takes the value of 1 if TIPs have been in existence for at least one year, and the value of zero otherwise; and FPART1 is a dummy variable that takes the value of 1 if at least one of the above four financial participation schemes has been in existence for at least one year, and the value of zero otherwise; (2) For all regressions, there were 1,794 observations on 205 firms over the 1990–2000 time period. All models include firm-specific fixed effects, year effects, and industry time trends. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level.

lnK lnL PSP1 ESOP1 SOP1 TIP1 FPART1 Adjusted R2

Value of t-Statistics

Fixed-Effect Estimates of Cobb–Douglas Production Functions, Augmented by Financial Participation Schemes with j ¼ 1 (Dependent Variable ¼ lnQ).

Parameter Estimates

Table 4.

124 TAKAO KATO ET AL.

The Productivity Effects of Financial Participation in Korea

125

As discussed earlier, the absence of significant productivity gains from ESOPs and SOPs may be due to longer lag in the productivity effects of ESOPs and SOPs. Tables 5 and 6 report the OLS estimates of Eqs. (1)–(5) with two-year lag and three-year lag ( j ¼ 2 and 3). The tables show that the estimated coefficients on the ESOP and SOP dummy variables are still statistically insignificant at the 10% level, confirming that neither ESOPs nor SOPs have significant impact on enterprise efficiency even after allowing for considerable gestation periods. Tables 5 and 6 also reveal an interesting difference in the time profiles of the productivity effects between PSPs and TIPs. The estimated coefficients on the PSP dummy continue to be statistically significant with longer lag whereas the estimated coefficient on the TIP dummy becomes substantially smaller and insignificant with three-year lag. The productivity effect of TIPs may start to dissipate to some degree after three years. We have two preliminary interpretations of the different time profile of the productivity effects between PSPs and TIPs although such interpretations will probably need to be tempered somewhat until a future analysis including more years of data and thus more observations with TIP ¼ 1 will confirm them20 First, as demonstrated in our field research at firms A and B, the size of the PSP bonus is considerably greater than that of the TIP bonus. For example, at both firms A and B, the average PSP bonus amounts to 200% of monthly base pay per person while a typical TIP bonus is an overnight group trip to a resort, which is definitely a nontrivial prize, yet rarely amounts to two months of base pay. As Kruse (1993) shows, the size of the prize matters for the productivity effects of financial participation schemes. Second, as we demonstrated earlier in our case reports, PSP bonuses are almost always given to each individual employee in the form of cash, whereas TIP bonuses are often provided to a group of employees in the form of funding for group activities, such as group trips. It is plausible that the marginal utility of group activities diminishes more rapidly than that of cash. We have found statistically significant productivity effects for PSPs and TIPs separately. To discern the relative importance of PSPs and TIPs in the goal alignment process, we further estimate a nested model: lnQit ¼ bK lnK it þ bL lnLit þ b1 PSPitj þ b2 TIPitj þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð6Þ

0.928

3.559 14.146 1.723

0.113 0.569 0.093

0.928

0.008

0.117 0.568

Parameter Estimates

0.205

3.680 14.097

Value of t-Statistics

0.928

0.053

0.117 0.568

Parameter Estimates

0.338

3.679 14.103

Value of t-Statistics

0.928

0.120

0.112 0.567

Parameter Estimates

1.680

3.521 14.076

Value of t-Statistics

0.045 0.928

0.117 0.568

Parameter Estimates

1.215

3.675 14.096

Value of t-Statistics

Source: Survey on Human Resources Management (SHRM) and corporate proxy statement data from KLCA (Korea Listed Companies Association). Notes: (1) PSP2 is a dummy variable that takes the value of 1 if PSPs have been in existence for at least two years, and the value of zero otherwise. Likewise, ESOP2 is a dummy variable that takes the value of 1 if ESOPs have been in existence for at least two years, and the value of zero otherwise; SOP2 is a dummy variable that takes the value of 1 if SOPs have been in existence for at least two years, and the value of zero otherwise; TIP2 is a dummy variable that takes the value of 1 if TIPs have been in existence for at least two years, and the value of zero otherwise; and FPART2 is a dummy variable that takes the value of 1 if at least one of the above four financial participation schemes has been in existence for at least two years, and the value of zero otherwise; (2) For all regressions, there were 1,794 observations on 205 firms over the 1990–2000 time period. All models include firm-specific fixed effects, year effects, and industry time trends. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level.

lnK lnL PSP2 ESOP2 SOP2 TIP2 FPART2 Adjusted R2

Value of t-Statistics

Fixed-Effect Estimates of Cobb–Douglas Production Functions, Augmented by Financial Participation Schemes with j ¼ 2 (Dependent Variable ¼ lnQ).

Parameter Estimates

Table 5.

126 TAKAO KATO ET AL.

0.928

3.531 14.179 2.399

0.112 0.570 0.135

0.928

0.018

0.117 0.569

Parameter Estimates

0.489

3.691 14.110

Value of t-Statistics

0.928

0.047

0.117 0.568

Parameter Estimates

0.175

3.679 14.101

Value of t-Statistics

0.928

0.042

0.115 0.568

Parameter Estimates

0.438

3.601 14.105

Value of t-Statistics

0.019 0.928

0.116 0.568

Parameter Estimates

0.541

3.654 14.107

Value of t-Statistics

Source: Survey on Human Resources Management (SHRM) and corporate proxy statement data from KLCA (Korea Listed Companies Association). Notes: (1) PSP3 is a dummy variable that takes the value of 1 if PSPs have been in existence for at least two years, and the value of zero otherwise. Likewise, ESOP3 is a dummy variable that takes the value of 1 if ESOPs have been in existence for at least two years, and the value of zero otherwise; SOP3 is a dummy variable that takes the value of 1 if SOPs have been in existence for at least two years, and the value of zero otherwise; TIP3 is a dummy variable that takes the value of 1 if TIPs have been in existence for at least two years, and the value of zero otherwise; and FPART3 is a dummy variable that takes the value of 1 if at least one of the above four financial participation schemes has been in existence for at least two years, and the value of zero otherwise; (2) For all regressions, there were 1,794 observations on 205 firms over the 1990–2000 time period. All models include firm-specific fixed effects, year effects, and industry time trends. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level.

lnK lnL PSP3 ESOP3 SOP3 TIP3 FPART3 Adjusted R2

Value of t-Statistics

Fixed-Effect Estimates of Cobb–Douglas Production Functions, Augmented by Financial Participation Schemes with j ¼ 3 (Dependent Variable ¼ lnQ).

Parameter Estimates

Table 6.

The Productivity Effects of Financial Participation in Korea 127

128

TAKAO KATO ET AL.

Furthermore, to test whether PSPs and TIPs are complements, we add the interaction term involving both PSP and TIP: lnQit ¼ bK lnK it þ bL lnLit þ b1 PSPitj þ b2 TIPitj þ b2 ðPSPitj  TIPitj Þ þ ðfirm-specific fixed effectsÞ þ ðyear effectsÞ þ ðindustry time trendsÞ þ uit

ð7Þ

Table 7 shows the OLS estimates of Eqs. (6) and (7). The estimated coefficients on the PSP and TIP dummy variables are statistically significant at the 10% level when we consider both variables simultaneously, pointing to the robustness of the significant productivity effects of PSPs and TIPs. When the interaction term is added, the estimated coefficients on the PSP and TIP dummy variables are statistically significant at the 5% level and the estimated coefficient on the interaction term is negative, yet not quite Table 7. Fixed-Effect Estimates of Cobb–Douglas Production Functions (Considering both PSPs and TIPs simultaneously; Dependent Variable ¼ lnQ).

lnK lnL PSP1 TIP1 PSP1  TIP1 Adjusted R2

Parameter Estimates

Value of t-Statistics

Parameter Estimates

Value of t-Statistics

0.107 0.567 0.092 0.106

3.340 14.110 1.904 1.898

0.109 0.564 0.121 0.160 0.153 0.928

3.409 14.015 2.326 2.421 1.527

0.928

Source: Survey on Human Resources Management (SHRM) and corporate proxy statement data from KLCA (Korea Listed Companies Association). Notes: (1) PSP1 is a dummy variable that takes the value of 1 if PSPs have been in existence for at least one year, and the value of zero otherwise. Likewise, ESOP1 is a dummy variable that takes the value of 1 if ESOPs have been in existence for at least one year, and the value of zero otherwise; SOP-1 is a dummy variable that takes the value of 1 if SOPs have been in existence for at least one year, and the value of zero otherwise; TIP1 is a dummy variable that takes the value of 1 if TIPs have been in existence for at least one year, and the value of zero otherwise; and FPART1 is a dummy variable that takes the value of 1 if at least one of the above four financial participation schemes has been in existence for at least one year, and the value of zero otherwise; (2) For all regressions, there were 1,794 observations on 205 firms over the 1990–2000 time period. All models include firm-specific fixed effects, year effects, and industry time trends. Significant at the 10% level. Significant at the 5% level. Significant at the 1% level.

The Productivity Effects of Financial Participation in Korea

129

statistically significant at the 10% level although fairly close to statistically significant. As such, there is no statistically significant evidence on the complementarity between PSPs and TIPs, suggesting that the complementary role of TIPs in easing the free rider problem of PSPs may be limited due to the fact that TIPs cover less than 60% of the labor force, whereas PSPs cover over 90%. PSPs and TIPs are more likely to be substitutes than complements.

5. CONCLUSIONS We have reported the first results for Korean firms on the effects of various employee financial participation schemes including PSPs, ESOPs, SOPs, and TIPs by estimating production functions using a unique panel data set containing firms with and without such schemes. We have found consistently that the introduction of a PSP or a TIP will lead to a significant increase in productivity (about 10%), whereas we have found no such evidence for ESOPs or SOPs. Further evidence has been found that the productivity payoff appears to be more long-lasting for PSPs than for TIPs. Finally, we have failed to find any statistically significant evidence on the complementarity between PSPs and TIPs, and evidence appears to suggest that they are more likely to be substitutes. Our findings are relevant for two broad sets of issues. First, our findings are relevant to the observation that pay systems are rapidly changing away from a fixed-wage contractual payment basis in many parts of the world (Ben-Ner and Jones, 1995). Foremost among these changes is the explosion in the extent of various forms of alternative pay methods, especially employee financial participation schemes in western economies in recent years. However, because of the limited nature of the available data, the econometric evidence on the economic effects of changing pay systems for major corporations in developing countries and NICs is quite thin. Consequently, by using panel data with information on one of the most comprehensive lists of employee financial participation schemes, our findings provide some of the first, more reliable evidence on the important issue of the effects on productive efficiency. Furthermore, our findings of positive productivity effects of PSPs and TIPs in Korea suggest that we are likely to see further evolution of pay systems away from fixed-wage contractual forms toward employee financial participation not only in mature industrialized countries but also in NICs and transition economies (notably China).

130

TAKAO KATO ET AL.

The second concerns the current public policy debate over economic reforms in Korea. Since the Asian Crisis, many policy makers and practitioners have been questioning the validity of their existing management practices and seeking new ways to improve performance of Korean firms.21 By providing rigorous evidence on what parts of the existing management practices have worked and what parts have not worked and whether complementarity exists among various practices, our chapter will help policy makers and practitioners make informed decision on their current reform efforts.

NOTES 1. For a survey of the literature on financial participation schemes, see for instance Kruse and Blasi (1997) on employee stock ownership and Jones, Kato, and Pliskin (1997) on profit sharing, gainsharing/team incentives. For a meta-analysis of the literature, see Doucouliagos (1995). For a more theoretical survey of the literature, see Gibbons (1997) and Prendergast (1999). For more recent works, see the shared capitalism literature (see, e.g., Kruse, Blasi, & Park, 2008; Bryson & Freeman, 2008). 2. For detailed discussion on the definition of PSPs, see Kruse (1993) and Jones et al. (1997). 3. See, for instance, Jones and Kato (1995), Blasi, Conte, and Kruse (1996), and Kruse and Blasi (1997). 4. See, for instance, Conyon and Freeman (2004) and Sesil, Kroumova, Blasi, and Kruse (2002). 5. See, for example, Hamilton, Nickerson, and Owan (2003), Jones and Kato (forthcoming), and Jones, Kalmi, and Kauhanen (forthcoming) for teams and TIPs. 6. In the English-language literature, we are aware of only one study that investigates the productivity effect of financial participation in NICs, that is, Cin and Smith (2002), which provide cross-sectional estimates on the productivity effects of employee stock ownership plans in South Korea. Our study extends Cin and Smith (2002) in two important ways. First, we are able to assemble long panel data (10 years), which enable us to estimate fixed-effect models and hence account for potentially serious omitted variable bias due to unobserved firm heterogeneity, such as managerial ability, which may be correlated with the adoption of financial participation schemes. Second, we are able to consider not only ESOPs but also other financial participation schemes, such as profit sharing, team incentives, and stock option, and test whether there is any synergic relationship among schemes. 7. Freeman, Kleiner, and Ostroff (2000), one of better-known examples of work using firm-level survey data, used the Human Resource Practice Survey of member firms of the Society of Human Resource Management in the United States and the survey’s response rate was 11%. The Human Resource Management Survey of Japanese Firms conducted by Kato and Morishima (2002) received about 20% of all

The Productivity Effects of Financial Participation in Korea

131

Japanese firms listed in Japan’s three major stock exchanges. In addition, in June of 1991, the Rengo Sogo Seikatsu Kaihatsu Kenkyu Jo (Rengo Research Institute of General Life Development) in Japan mailed their questionnaire asking questions on labor conditions and employee participation/involvement to 6,800 firms (including both public and private firms in Japan) and received usable responses from 689 firms (a response rate of 10%). In June of 1989, the Japan Productivity Center mailed their questionnaire asking questions on Human Resource Management Practices to 1,030 firms in Japan and received usable responses from 203 firms (a response rate of 19.7%). 8. See, for instance, Fitzroy and Kraft (1987), Weitzman and Kruse (1990), Levine and Tyson (1990), Jones and Pliskin (1991), Ben-Ner and Jones (1995), Kandel and Lazear (1992), Kruse (1993), Holmstrom and Milgrom (1994), Baker, Gibbons, and Murphy (1994), Milgrom and Roberts (1995), Ichniowski, Shaw, and Prennushi (1997), Helper (1998), Black and Lynch (2001, 2004), Kato and Morishima (2002), Eriksson (2003), Bayo-Moriones, Galilea-Salvatierra, & Merino-Dı´ az de Cerio (2003), Zwick (2004), DeVaro (2006), Jones, Kalmi, and Kauhanen (2006), and Heywood, Jirjahn, and Wei (2008). 9. This explains the difference in the estimated incidence of ESOPs between our study and Cin and Smith (2002). 10. For reviews, see the essays in Blinder (1990) and Gibbons (1997). 11. Our confidentiality agreements with both firms prohibit us from identifying the specific products they produce. 12. Firm A is known for successful representative participation and labormanagement cooperation through work councils and firm B is famous for the use of self-directed teams. As such, they may not be representative of the population of Korean firms in terms of employee involvement and participation. The purpose of presenting field evidence here is to demonstrate the stronger use of cash-based incentive pay as compared to stock-based incentive schemes in Korean firms with successful employee involvement and labor-management cooperation. See Kato, Lee, Lee, and Ryu (2005) for a more detailed description of the case studies. 13. EVA is essentially after-tax operating profit minus the total cost of capital. 14. We report the Cobb–Douglas results because the translog estimates yield an implausible (negative) distribution parameter for capital. These, and other, unreported regression results are available from the corresponding author at [email protected] upon request. 15. For similar arguments, see Pil and MacDuffie (1996) and Ichniowski & Shaw (1995). 16. See, for instance, Kato and Morishima (2002). 17. A similar argument is made for the case of profit sharing by Nakamura & Nakamura (1991) and Ehrenberg (1990). 18. The data on value added as a proxy for Q were created by adding labor costs and depreciation to operational profit. A similar procedure was used to calculate value added for Japan with similar accounting practices by the Oriental Economist (Toyokeizai Shinpo Sha) whose value-added data were widely used by both scholars and practitioners (e.g., Jones & Kato, 1995; Ohkusa & Ohtake, 1997; Kato & Morishima, 2002). Most prior studies use sales to proxy output. We believe that our use of the (theoretically preferable) value added instead of sales is a strength of the chapter.

132

TAKAO KATO ET AL.

19. The sums of estimated bK and bL for the five specifications are around 0.7, suggesting decreasing returns to scale. As explained in Hsiao (1986, pp. 26–28), the use of panel data in production function estimation leads to a smaller sum of estimated output elasticities. The sums of estimated output elasticities reported earlier are similar to those cited in Hsiao (1986, p. 27) as well as those obtained by Jones and Kato (1995) and Kato and Morishima (2002) for Japanese manufacturing firms. 20. Jones and Kato (forthcoming) find similar evidence on the dissipating performance effects of employee involvement. 21. ‘‘New ways’’ often mean ‘‘Anglo-American ways,’’ such as replacing allegedly rigid internal labor markets (characterized by long-term employment, senioritybased wage, and various human resource management practices) with flexible external labor markets. See, for instance, OECD (2000).

ACKNOWLEDGMENTS We are grateful for support from the Korea Labor Institute. An earlier version of the chapter was written while Kato was Distinguished Visiting Professor at KDI School of Public Policy and Management. Kato is thankful for their hospitality. The chapter has benefited from comments and suggestions from Dong-Bae Kim, Sam-Su Kim, and Yong-Jin Ro. JongKyu Lee provided excellent research assistance. Finally, we owe our great debt of gratitude to the HR managers, line supervisors, and union officials of Korean companies who granted us the opportunities to interview them.

REFERENCES Baker, G., Gibbons, R., & Murphy, K. J. (1994). Subjective performance measures in optimal incentive contracts. Quarterly Journal of Economics, 109(4), 1125–1156. Bayo-Moriones, J. A., Galilea-Salvatierra, P. J., & Merino-Dı´ az de Cerio, J. (2003). Participation, cooperatives and performance: An analysis of Spanish manufacturing firms. In: T. Kato & J. Pliskin (Eds), Determinants of the incidence and the effects of participatory organizations: Advances in the economic analysis of participatory and labormanaged firms (pp. 31–56). Amsterdam: Elsevier/JAI. Ben-Ner, A., & Jones, D. C. (1995). Employee participation, ownership, and productivity: A theoretical framework. Industrial Relations, 34(4), 532–554. Black, S. E., & Lynch, L. M. (2001). How to compete: The impact of workplace practices and information technology on productivity. Review of Economics and Statistics, 83(3), 434–445. Black, S. E., & Lynch, L. M. (2004). What’s driving the new economy? The benefits of workplace innovation. Economic Journal, 114(493), F97–F116. Blasi, J., Conte, M., & Kruse, D. (1996). Employee stock ownership and corporate performance among public companies. Industrial and Labor Relations Review, 50(1), 60–79.

The Productivity Effects of Financial Participation in Korea

133

Blinder, A. S. (Ed.) (1990). Paying for productivity: A look at the evidence. Washington, DC: Brookings Institution. Bryson, A., & Freeman, R. (2008). How does shared capitalism affect economic performance in the UK? NBER Working Papers: 14235. National Bureau of Economic Research, Inc., Cambridge, MA. Cin, B.-C., & Smith, S. C. (2002). Employee stock ownership and participation in South Korea: Incidence, productivity effects, and prospects. Review of Development Economics, 6(2), 263–283. Conte, M. A., & Svejnar, J. (1990). The performance effects of employee ownership plans. In: A. S. Blinder (Ed.), Paying for productivity: A look at the evidence (pp. 143–172). Washington, DC: Brookings Institution. Conyon, M. J., & Freeman, R. B. (2004). Shared modes of compensation and firm performance: U.K. evidence. In: D. Card, R. Blundell & R. B. Freeman (Eds), Seeking a premier economy: The economic effects of British economic reforms, 1980–2000 (pp. 109–146). Chicago and London: University of Chicago Press. DeVaro, J. (2006). Teams, autonomy, and the financial performance of firms. Industrial Relations, 45(2), 217–269. Doucouliagos, C. (1995). Worker participation and productivity in labor-managed and participatory capitalist firms: A meta-analysis. Industrial and Labor Relations Review, 49(1), 58–77. Ehrenberg, R. G. (1990). Alternative pay systems, firm performance, and productivity: Comment. In: A. S. Blinder (Ed.), Paying for productivity: A look at the evidence (pp. 88–94). Washington, DC: Brookings Institution. Eriksson, T. (2003). The effects of new work practices-evidence from employer-employee data. In: T. Kato & J. Pliskin (Eds), Determinants of the incidence and the effects of participatory organizations: Advances in the economic analysis of participatory and labormanaged firms (pp. 3–30). Amsterdam: Elsevier/JAI. FitzRoy, F. R., & Kraft, K. (1987). Cooperation, productivity, and profit sharing. Quarterly Journal of Economics, 102(1), 23–35. Freeman, R., Kleiner, M., & Ostroff, C. (2000). The anatomy of employee involvement and its effects on firms and workers. NBER Working Paper No. 8050. National Bureau of Economic Research, Inc., Cambridge, MA. Freeman, R. B., & Kleiner, M. M. (2000). Who benefits most from employee involvement: Firms or workers? American Economic Review, 90(2), 219–223. Gibbons, R. (1997). Incentives and careers in organizations. In: Advances in economics and econometrics: Theory and applications: Seventh world congress (Vol. 2, pp. 1–37). Cambridge; New York; and Melbourne: Cambridge University Press. Hamilton, B. H., Nickerson, J. A., & Owan, H. (2003). Team incentives and worker heterogeneity: An empirical analysis of the impact of teams on productivity and participation. Journal of Political Economy, 111(3), 465–498. Helper, S. (1998). Complementarity and cost reduction: Evidence from the auto supply industry. Working Paper No. 6033 (revised). National Bureau of Economic Research, Cambridge, MA. Heywood, J. S., Jirjahn, U., & Wei, X. (2008). Teamwork, monitoring and absence. Journal of Economic Behavior & Organization, 68(3/4), 676–690. Holmstrom, B., & Milgrom, P. (1994). The firm as an incentive system. American Economic Review, 84(4), 972–991.

134

TAKAO KATO ET AL.

Hsiao, C. (1986). Analysis of panel data. Cambridge; New York; and Sydney: Cambridge University Press. Ichniowski, C., & Shaw, K. (1995). Old dogs and new tricks: Determinants of the adoption of productivity-enhancing work practices. Brookings Papers on Economic Activity, 0(0), 1–55. Ichniowski, C., Shaw, K., & Prennushi, G. (1997). The effects of human resource management practices on productivity: A study of steel finishing lines. American Economic Review, 87(3), 291–313. Jones, D. C., Kalmi, P., & Kauhanen, A. (2006). Human resource management policies and productivity: New evidence from an econometric case study. Oxford Review of Economic Policy, 22(4), 526–538. Jones, D. C., Kalmi, P., & Kauhanen, A. (forthcoming). Teams, performance-related pay, profit-sharing and productive efficiency: Evidence from a food-processing plant. Industrial and Labor Relations Review. Jones, D. C., & Kato, T. (1995). The productivity effects of employee stock-ownership plans and bonuses: Evidence from Japanese Panel Data. American Economic Review, 85(3), 391–414. Jones, D. C., & Kato, T. (forthcoming). The impact of teams on output, quality and downtime: An empirical analysis using individual panel data. Industrial and Labor Relations Review. Jones, D. C., Kato, T., & Pliskin, J. (1997). Profit sharing and gainsharing: A review of theory, incidence and effects. In: D. Lewis, D. Mitchell & M. Zaidi (Eds), The human resource management handbook, part I (pp. 153–174). Greenwich, CT: JAI Press. Jones, D. C., & Pliskin, J. (1991). The effects of worker participation, employee ownership and profit sharing on economic performance: A partial review. In: R. Raymond & R. Veljko (Eds), Ownership and participation: International handbook of participation in organizations (Vol. 2, pp. 43–63). Oxford: Oxford University Press. Kandel, E., & Lazear, E. P. (1992). Peer pressure and partnerships. Journal of Political Economy, 100(4), 801–817. Kato, T. (2006). Determinants of the extent of participatory employment practices: Evidence from Japan. Industrial Relations, 45(4), 579–605. Kato, T., Lee, J. H., Lee, K.-S., & Ryu, J.-S. (2005). Employee participation and involvement in Korea: Evidence from a new survey and field research. International Economic Journal, 19(2), 251–281. Kato, T., & Morishima, M. (2002). The productivity effects of participatory employment practices: Evidence from new Japanese panel data. Industrial Relations, 41(4), 487–520. Kato, T., & Morishima, M. (2003). The nature, scope and effects of profit sharing in Japan: Evidence from new survey data. International Journal of Human Resource Management, 14(6), 942–955. Kruse, D., & Blasi, J. (1997). Employee ownership, employee attitudes, and firm performance. In: D. Lewis, D. Mitchell & M. Zaidi (Eds), The human resource management handbook, part I (pp. 113–152). Greenwich, CT: JAI Press. Kruse, D. L. (1993). Profit sharing: Does it make a difference? Kalamazoo, Michigan: W.E. Upjohn Institute for Employment Research. Kruse, D. L., Blasi, J. R., & Park, R. (2008). Shared capitalism in the U.S. economy? Prevalence, characteristics, and employee views of financial participation in enterprises. NBER Working Papers: 14225. National Bureau of Economic Research, Inc., Cambridge, MA. Levine, D. I., & Tyson, L. D. A. (1990). Participation, productivity and the firm’s environment. In: A. S. Blinder (Ed.), Paying for productivity. Washington, DC: Brookings Institution.

The Productivity Effects of Financial Participation in Korea

135

Milgrom, P., & Roberts, J. (1995). Complementarities and fit: Strategy, structure, and organizational change in manufacturing. Journal of Accounting and Economics, 19(2–3), 179–208. Nakamura, M., & Nakamura, A. (1991). Risk behavior and determinants of bonus versus regular pay in Japan. Journal of the Japanese and International Economies, 5(2), 140–159. OECD. (2000). OECD Economic surveys: Korea. Paris, France: OECD. Ohkusa, Y., & Ohtake, F. (1997). The productivity effects of information sharing, profit sharing, and ESOPs. Journal of the Japanese and International Economies, 11(3), 385–402. Pil, F. K., & MacDuffie, J. P. (1996). The adoption of high-involvement work practices. Industrial Relations, 35(3), 423–455. Prendergast, C. (1999). The provision of incentives in firms. Journal of Economic Literature, 37(1), 7–63. Sesil, J. C., Kroumova, M. K., Blasi, J. R., & Kruse, D. L. (2002). Broad-based employee stock options in US ‘new economy’ firms. British Journal of Industrial Relations, 40(2), 273–294. Wadhwani, S., & Wall, M. (1990). The effects of profit-sharing on employment, wages, stock returns and productivity: Evidence from UK Micro-data. Economic Journal, 100(399), 1–17. Weitzman, M. L., & Kruse, D. L. (1990). Profit sharing and productivity. In: A. S. Blinder (Ed.), Paying for productivity. Washington, DC: Brookings Institution. Zwick, T. (2004). Employee participation and productivity. Labour Economics, 11(6), 715–740.

STOCK OPTION SCHEMES AND FIRM TECHNICAL INEFFICIENCY: EVIDENCE FROM FINLAND Mikko Ma¨kinen 1. INTRODUCTION Many firms in many countries started to issue stock option schemes to their employees in the 1990s (Murphy, 1999).1 In the course of time, the mushrooming of schemes has generated a heated public debate on the pros and cons of this compensation method. In one camp are those who argue that stock options are nothing more but a compensation mechanism by which managers transfer excessive fortunes to themselves without a real enhancement in firm performance. On the other hand, proponents underline that options provide managers and employees financial incentives to make better decisions, work harder, and share valuable information in a way that enhance firm performance. Thus, they see options – more or less– as a major innovation in managerial and personnel compensation (or more generally in human resource management). However, at the moment there is no theoretical or empirical consensus how stock options and managerial equity ownership affect firm performance in economic literature (Core, Guay, & Larcker, 2003). In this chapter, we try to shed more light on this issue by using stochastic production frontier estimators. These estimators provide a statistical tool to study efficiency differences between economic decision-making units such as Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 137–160 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011010

137

138

MIKKO MA¨KINEN

firms. Thus, our research questions that arise are the following: (i) do stock option firms have a lower technical inefficiency compared to nonoption firms; (ii) does the impact on firm technical inefficiency depend on whether options are selectively allocated to particular employees (selective schemes) or whether all employees are eligible to participate (broad-based schemes). To do so, we use a novel panel data set on the publicly listed Finnish firms in the manufacturing and information and communication technology (ICT) sectors over the period 1992–2002. The data enable us to distinguish broad-based option schemes from those selectively allocated to particular employees. In the empirical models, we estimate simultaneously stochastic production frontier parameters, inefficiency scores, and marginal effects. To the best of our knowledge, we provide the first empirical evidence in literature on the link between employee stock option schemes and firm technical inefficiency, when an inefficiency term is modeled as a function of option schemes.2 Our main findings can be summarized as follows. First, we find that the shape of the inefficiency distribution itself differs notably between the sectors. For example, in the ICT sector mean inefficiency estimates are substantially higher compared to the manufacturing sector, although naturally efficient and inefficient firms exist in both sectors. Further, in the ICT sector the mean inefficiency estimates indicate no difference in inefficiency between option and nonoption firms, whereas in the manufacturing sector our findings suggest that broad-based firms have higher mean inefficiency than selective scheme and nonoption firms. Second, when assessing quantitatively the statistical significance of marginal effects of option schemes on inefficiency, we find that in the manufacturing sector the marginal effects of broad-based schemes on the mean and the variance of the inefficiency term uit are significantly positive. Other things equal, these findings suggest that broad-based schemes increase firm technical inefficiency and production uncertainty in the sector. For selective option schemes, we find no evidence of a statistically significant link with technical inefficiency. As such, our findings do not provide empirical support for the view that by using stock option schemes firms can reduce technical inefficiency and therefore be able to enhance performance. Furthermore, our finding that option schemes may not enhance firm performance is similar to Jones, Kalmi, and Ma¨kinen (2010), who, by using a different empirical approach, consistently find insignificant association between stock option schemes (both selective and broad-based) and firm productivity. Based on the existing theoretical literature, the potential reasons for the insignificant link of option schemes with firm performance might be (i) poor option scheme

Stock Option Schemes and Firm Technical Inefficiency

139

implementation; (ii) free riding; (iii) accounting myopia; (iv) line-of-sight; and (v) managerial rent-seeking. However, to some extent, differences in institutional arrangements might limit generalizability of our results to other contexts. For example, because of the collective wage bargaining agreements in Finland, nominal base wages are generally protected from downward falls and therefore stock options do not substitute for nominal base wages. The potential risks of stock options are further downgraded by the facts that (i) options are usually given for free or for symbolic compensation without real costs to employees, and (ii) options (without a need to purchase shares) are often traded in secondary markets once the vesting period begins (this is not possible in the United States). Furthermore, around 40% of stock option schemes in Finland include a dividend protection clause (the strike price for shares is adjusted downwards by the amount of dividends paid), whereas in the United States, this is the case in only 1% of cases (see Liljeblom & Pasternack, 2006). But even with these institutional differences, one might wonder why employee performance incentives in general are contingent on a firm’s stock market performance (a stochastic process) if the factors outside employee efforts largely dominate a firm’s share performance (e.g., random shocks). In such a case, employees are participating in a compensation lottery, which may pay wealthy in the realization of a positive shock, but in the case of negative shock yields probably nothing (both naturally conditional on the specific terms of a scheme). Furthermore, partially related to this, although the efficient market hypothesis postulates that share prices always reflect accurately firms’ value given all publicly available information, the current financial and economic crises have raised doubts how right stock markets truly get stock prices with respect to some economic fundamentals like firm earnings (firm performance). Note that already Keynes (1936) expressed contempt for financial markets, which he viewed as being dominated by short-term speculation with little regard for economic fundamentals. Hence, although stock market prices are easily observed and are defined outside a firm’s control, one may ask: ‘‘If and when the capital development of a country and employee compensation becomes a by-product of the activities of a casino, under what circumstances the job is well-done?’’ (see also Krugman, 2009). This chapter is organized as follows. Section 2 provides background discussion. Section 3 describes the pattern of stock option schemes’ adoption in Finland. Section 4 outlines the data and empirical strategy. Section 5 reports the estimation results. Section 6 concludes.

140

MIKKO MA¨KINEN

2. BACKGROUND DISCUSSION To provide a brief background, we next review the related theoretical and empirical work. We focus on two issues, but it is clear that many of the points we discuss are relevant to some degrees to both. The first concerns the impact of options in general on firm performance, the second relates to whether or not a scheme being broad-based or selective is expected to have some implications for firm performance.3 The theoretical literature concerning the expected impact of options in general on firm performance is ambiguous. In one camp are those who expect firm performance to be enhanced, because options can align the interest of employees and shareholders – for example, stock options may motivate employees to exert more effort and to take actions that are mutually beneficial to both owners and employees. Stock options have deemed also crucial in recruiting and retaining key employees, particularly in labor markets where employees are potentially highly mobile (Rousseau & Shperling, 2003). In addition, stock options may help to retain key employees, because they adjust pay according to market conditions (Oyer, 2004) and because they are a deferred form of compensation. However, there has also been criticism of stock option schemes. For one thing, options dilute shareholders wealth when exercised. Besides this, the potential performance’ impacts of options are questioned. For one thing, the free rider argument applies to all compensation mechanisms that are based on group performance (Alchian & Demsetz, 1972). Another argument is the line-of-sight; rewards based on performance can only be motivating if, by their actions, employees can influence the measures on which performance pay is based (Vroom, 1995). This is typically not the case with stock option plans, where employees can hardly perceive any direct link between their actions and the share price performance (with possible exception of top executives). Finally, especially executive stock option compensation may be motivated by rent-seeking activities (Bebchuk & Fried, 2003). Given that the theoretical literature concerning how stock option schemes affect firm performance is ambiguous, the question seems to be empirical one. Most of the previous empirical work has concentrated on the performance consequences of executive stock options. For instance, Hanlon, Rajgopal, and Shevlin (2003) find that changes in Black–Scholes values of option grants are positively associated with future operating income of the firm. But using similar data, Larcker (2003) find that results depend essentially on estimation strategy. Habib and Ljungqvist (2005) show that the fewer stock options the CEO holds, the short fall from the

Stock Option Schemes and Firm Technical Inefficiency

141

firm value-maximization benchmark is smaller. Core et al. (2003) conclude in their survey that the research consensus is that there is no clear consensus how executive equity compensation affects firm performance. Our second hypothesis concerns the expected impact of the different type of option plans (selective vs. broad-based). As with the first hypothesis, the free rider and the line-of-sight arguments are again evident, especially to broad-based schemes. On the contrary, it is argued that when employees have a financial interest in the company, they are much more likely to act against shirkers (Freeman, Kruse, & Blasi, 2008). Also, when rewards are based on group performance, it is in the interest of employees to develop a group norm to monitor the performance of their peers and prevent freeriding behavior (Kandel & Lazear, 1992). These arguments are applicable to broad-based schemes in particular and are expected to produce anti-shirking behavior (and finally to result in enhanced firm performance). Whether or not employee effort is observable may have implications on the impacts of option schemes. Jackson and Lazear (1991) argue that deferred wage compensation is a better motivator compared to stock and stock options when a worker’s performance is perfectly observable. Thus, for rank-andfile workers, the benefits of deferred wage compensation are likely to greater, but for CEOs and managers, stock and stock options may do a better job (selective schemes in our case), because managers’ output can be difficult to observe directly. As before, the issue of the impact of broad-based versus selective option schemes on firm performance is an empirical question. We are familiar with two studies that examine the effects of broad-based options with samples of listed firms from various industries, Conyon and Freeman (2004) for the United Kingdom and Sesil, Kroumova, Blasi, and Kruse (2000) for the United States. Both of these studies find a significant positive association between the presence of broad-based option plans and firm productivity. Two other studies analyze broad-based options in the US new economy firms. Sesil, Kroumova, Blasi, and Kruse (2002) find evidence that productivity is higher in firms with broad-based stock option plans, and Ittner, Lambert, and Larcker (2003) find that lower than expected option grants and existing option holdings are associated with lower accounting and stock price performance in subsequent years. The most related study to our chapter is Jones et al. (2010), who find consistently insignificant association between selective and broad-based stock option schemes and firm productivity. We utilize the same data set, but our estimation strategy is different; we specify stochastic frontier models to estimate technical inefficiency, whereas they focus on how stock options are associated with

142

MIKKO MA¨KINEN

firm productivity in a production function context. In addition, they pool the data over all sectors, but we instead estimate separate models for the manufacturing and ICT sectors.

3. THE PATTERN OF OPTION SCHEMES’ ADOPTION IN FINLAND In this section, we briefly review the pattern of option schemes’ adoption in Finland.4 Table 1 describes the evolution of option plans in the publicly traded firms at the Helsinki Stock Exchange (HEX) between 1987 (when the first employee stock option scheme was launched in Finland) and 2002. We have information on the presence of option schemes on the main list throughout the period and on the minor lists, NM-list (New Market) and I-list (Investor), since 1997. Column (1) gives the number of firms at the HEX main list. Column (2) shows the total number of listed firms, including the two minor lists (from 1997). It appears that the total number of listed firms fluctuates a lot with the business cycle. The first period of growth was the economic boom years 1987–1989, when the number of firms increased from 52 to 82. From 1989 onwards, the number of firms fell, reaching a low point of 60 firms in 1993. The main reason for this was the Finnish Depression in 1990–1993, when many Finnish firms had financial problems.5 After 1993, the number of listed firms started to rise, and the 1989 level was reached again in 1997. The increase continued until 2000, but thereafter, the number fell again. From 1997 onwards, we also include firms on the two minor lists. In some cases, firms switched from the minor to the major list. At the same time, however, there are new firms entering the minor lists, especially in 2000 when relatively many small ICT firms entered the NM-list. Column (3) indicates how many firms have adopted their first option scheme in a given year. Altogether, 127 firms have adopted a stock option plan. Whereas seven pioneering firms implemented their option plans as early as the 1980s, few plans were launched during the economic depression years of 1990–1993. The renewed interest in option plans began in 1994, when 20 firms (almost 40% of listed firms) adopted option schemes. Relatively few firms adopted schemes during 1995–1996 (possibly because the taxation of option gains changed from a moderate capital tax into a substantially higher marginal income tax), but since 1997, options have became widely popular. The rise of option schemes during 1999–2000 was fueled by new listings. When new listings stopped after 2000, so did the

143

Stock Option Schemes and Firm Technical Inefficiency

Table 1. Year

The Development of Stock Option Schemes in Finland.

(1)

(2)

No. of firms on the main list

No. of firms in total

1987

52

–

1

1

1988

70

–

2

2

1989

82

–

4

6

1990

77

–

2

3

1991

66

–

3

4

1992

65

–

1

1

1993

60



4

6

68



20

21

1994

74



5

7

1996

73



3

9

1997

82

115

12

22

1998

92

119

24

47

1999

102

137

21

41

2000

107

150

20

59

2001

103

145

4

32

2002

99

137

1

29

127

290

Total

(3)

(4)

(5)

No. of first No. of new No. of main option option list firms plan in this plans in having year this year option plans 1 (1.9%) 3 (4.3%) 6 (8.5%) 7 (9.1%) 9 (13.6%) 8 (12.3%) 12 (20.0%) 27 (39.7%) 34 (45.9%) 34 (46.6%) 40 (48.8%) 60 (65.2%) 77 (75.5%) 88 (82.2%) 87 (84.5%) 82 (82.8%)

(6)

(7)

No. of firms having option plan

HEX portfolio index, yearly changesa

1



3



7



8

0.380

10

0.113

11

0.077

15

0.657

34

0.164

38

0.062

36

0.322

46 (40.0%) 69 (58.0%) 91 (66.4%) 113 (75.3%) 112 (77.2%) 101 (73.7%)

0.273 0.138 0.541 0.242 0.191 0.150

Notes: For years 19901995, we have used the general index, because the portfolio index is calculated only since 1996. Changes are in logarithmic scale. a The portfolio index in trade-weighted average share returns, where a maximum weight assigned to one company is 10%.

144

MIKKO MA¨KINEN

introduction of new option schemes. Firms often launch new schemes once the previous schemes are close to expiring, or they may operate many schemes simultaneously: 84 of the 127 firms (66%) that have ever adopted a scheme have implemented more than one scheme (three firms have reached seven successive schemes).6 Column (4) shows the number of firms that adopted new option schemes in a given year. The total number of option adoptions we are aware of is 290. The early peak year was 1994 (21 firms adopted). From 1997 (22 firms adopted), the adoption increased further, but after 59 plans in 2000, the adoptions started to decline, with 32 new schemes in 2001 and 28 in 2002. Hence, the most active period of adoptions coincided with the height of the stock market boom in the late 1990s. But, as market prices started to fall after May 2000, so did the rate of stock option adoption (see also Jones, Kalmi, & Ma¨kinen, 2006). In column (5), we use the information on timing and launching of a scheme. A firm is treated as having a scheme in year t, if it has at least one scheme that has started in year t or earlier and if the final date for exercising options in this scheme is in year tþ1 or later. Column (5) indicates that the proportion of firms with an option scheme increased until 1993, by which time 20% of the main list firms had an option scheme. This proportion jumped to around 40% in 1994, after which it increased slowly for three years, until it jumped again to 65% in 1998. The temporary maximum was reached in 2001, when almost 85% of the main list firms had a stock option scheme. Column (6) shows the development for all firms, also for those outside the main list. The proportion of firms with stock option schemes is somewhat lower for all firms, due to many nonoption firms at the I-list. In sum, the extensive growth of stock option schemes reflects a major change in the Finnish corporate governance system in listed firms. In the end of the 1980s, the corporate governance system was very much bank-centered and resembled the German system (see, e.g., Hyytinen, Kuosa, & Takalo, 2003). After the stock market started its recovery from the depression in 1993, the importance of the equity market in financial intermediation grew throughout the 1990s. Both the turnover and the market value of firms listed on the stock exchange increased dramatically throughout the decade, with Nokia leading this development. In addition, now Finnish stock markets are much deeper, more transparent, and arguably provide more reliable information than in the past. At the same time, both monitoring of insider trading and legal punishments have become stricter. As such, we can say that during the last 10–15 years, Finland has shifted from a bank-based financial intermediation towards a market-based system.

Stock Option Schemes and Firm Technical Inefficiency

145

4. DATA AND EMPIRICAL STRATEGY 4.1. Data To examine the association of option schemes with firm technical inefficiency, we use new panel data on the publicly listed Finnish firms in the manufacturing and ICT sectors over the period 1992–2002. The option data set is combined from four sources: firms’ annual statements and general meeting reports; firms’ public press releases on the terms of schemes at time of an adoption; the data gathered by a professor at Helsinki School of Economics; and the data provided an investment bank that designs option schemes in Finland. We then carefully cross-checked the terms of option programs, and in a few cases when we found divergence between the data sets, we have trusted on firms’ own public announcements on the adoptions. One benefit of our option data are that we are able to make a distinction between selective and broad-based schemes. Thus, this allows us to investigate the inefficiency differences between the different types of option plans. Finally, we matched option data with firm-level accounting data, provided by a firm specialized in accounting information. Our data include all listed Finnish companies for a minimum of four consecutive years in the sectors. It is an unbalanced panel data set, because we miss some firm-year observations. Apparently, some of the yearly variation in the number of firms is due to the entry and exit (attrition) of listed firms at the HEX. Also, a few firms have merged in the time period, and in these cases, we included only new merged firms after the merger. In addition, concerning mainly recently listed firms, in a few cases we have added a firm’s financial statement information before the listing, if that information was available in the accounting data.7 To control for potential bias of very small and very large firms in the stochastic frontier estimations, we have excluded potential outlier observations, that is, an observation if employment was less than 50 persons, if fixed capital was less than h1,000,000, and if employment was more than 50,000 persons. We also deflated all nominal monetary variables by an industry based gross-output deflator at constant 2000 Euros, obtained from Statistics Finland. The final data set contains 571 firm-year observations regarding 62 firms in the manufacturing sector and 243 firmyear observations covering 32 firms in the ICT sector. The number of observations of a firm i, that is, Ti, is 4rTir11.8 Tables 2 and 3 present summary statistics for our key variables.

MIKKO MA¨KINEN

146

Table 2.

ln(va) ln(l) ln(k) dilua

dilussa dilubba opt ssopt bbsopt

Summary Statistics for the Manufacturing Sector. Name

Firm-year Observations

Mean Standard Minimum Maximum Deviation

Natural logarithm of value added Natural logarithm of labor Natural logarithm of fixed capital Potential dilution in the range of (0,1); a proxy of option program size Potential dilution for selective stock option programs Potential dilution for broad-based stock option programs Option program dummy Selective option program dummy Broad-based option program dummy

571 571 571 298

18.27 7.38 18.37 0.048

1.79 1.67 2.16 0.046

14.95 4.32 13.94 0.003

22.37 10.74 23.58 0.307

228

0.032

0.023

0.003

0.111

70

0.100

0.062

0.034

0.307

571 571 571

0.522 0.399 0.123

0.500 0.490 0.328

0 0 0

1 1 1

Notes: All value measures are deflated using an industry-specific gross output deflator at constant 2000 Euros obtained from Statistics Finland. The data contain 571 firm year observations regarding 62 firms. a Summary statistics for dilu, diluss, and dilubb only for those firms that have a stock option program.

Table 3.

ln(va) ln(l) ln(k) dilua

dilussa dilubba opt ssopt bbsopt

Summary Statistics for the ICT Sector.

Name

Firm-year Observations

Mean Standard Minimum Maximum Deviation

Natural logarithm of value added Natural logarithm of labor Natural logarithm of fixed capital Potential dilution in the range of (0,1); a proxy of option program size Potential dilution for selective stock option programs Potential dilution for broad-based stock option programs Option program dummy Selective option program dummy Broad-based option program dummy

243 243 243 139

17.13 6.35 16.62 0.070

1.50 1.38 1.78 0.045

13.56 3.95 13.87 0.002

22.16 10.62 21.17 0.214

64

0.058

0.038

0.002

0.187

75

0.081

0.049

0.018

0.214

243 243 243

0.572 0.263 0.309

0.496 0.441 0.463

0 0 0

1 1 1

Notes: All value measures are deflated using an industry-specific gross output deflator at constant 2000 Euros obtained from Statistics Finland. The data contain 243 firm-year observations regarding 32 firms. a Summary statistics for dilu, diluss, and dilubb only for those firms that have a stock option program.

Stock Option Schemes and Firm Technical Inefficiency

147

In the analysis that follows, we distinguish between broad-based and selective option schemes. The classification on broad-based and selective schemes is based on firms’ public stock exchange reports,9 because the Finnish Law on Joint Stock Companies requires listed firms to report all relevant terms of schemes to shareholders before an adoption of a scheme. The selective schemes are mostly managerial schemes, although they can also include other key personnel such as R&D employees. To qualify as a broad-based scheme, all employees (or at least a great majority) should be eligible to participate. Although a high rate of eligibility does not automatically guarantee a high participation rate, there are good reasons to believe that these are closely connected. For one thing, employees usually face only small costs when they subscribe to options – for example, by providing a zero-interest loan to the company, with the company repaying the loan at face value after a certain period, usually in 1–3 years. Thus, while employees face a cost in terms of foregone interest and liquidity, typically this cost is far below the real value of the options. Moreover, not all companies use this procedure, as they essentially give options for free to their employees.10 In the empirical models, we measure the presence of an option scheme, the size of the scheme, and whether the scheme is selective or broad-based. Two of the variables are binary indicators and one is a continuous variable. The first binary indicator is opt measuring the presence of a scheme in a firm i in a given year t. The dummy variable equals one for option firms and zero otherwise. Our second binary indicator measures also the presence of a plan, but it distinguishes between selective (ssopt) and broad-based (bbsopt) plans. As said before, by a selective plan, we mean a scheme that is targeted to a selected group of employees including managerial programs, but also schemes that are targeted to key personnel. Broad-based plans are all encompassing, including also managers, but they are not egalitarian in the sense of all participants have the same number of options. The third option scheme variable is potential dilution (dilu), which measures the (potential) size of effective schemes in a firm in a given year t.11 This is a continuous variable, that is, the ratio of the number of shares that may be awarded through effective plans in a given year t divided by the sum of the total number of shares and the number of new shares that may be awarded through options at the end of a year t. If a program ends in the middle of a year t, then year t1 is the last year in assessing a potential dilution. We also use distinct variables for selective (diluss) and broadbased (dilubb) schemes to examine whether there is a difference between the schemes.

MIKKO MA¨KINEN

148

4.2. Empirical Strategy In their pioneering work, Aigner, Lovell, and Schmidt (1977) and Meeusen and van den Broeck (1977) proposed independently stochastic production frontier models. Since then, the literature has proposed several specifications and estimation techniques.12 We base our empirical analysis on the basic firm-level Cobb–Douglas13 stochastic production frontier model14 lnðvait Þ ¼ bk lnðkit Þ þ bl lnðl it Þ þ bx xt þ vit  uit

(1)

where subscripts i and t index firm and time, respectively, ln(vait) is firm value added,15 firm fixed capital ln(kit) is the sum of a firm’s tangible and intangible assets at the end of the year, labor input ln(lit) is the (mean) number of employees16 in a given year, xt is a time trend to account for technological change, and a composed error term it ¼ vit  uit is the difference between noise component vit and inefficiency component uit.17 In the empirical analysis that follows, we estimate separate models for the ICT and manufacturing sectors, because production frontiers may differ between the sectors.18 To study whether firm-level technical inefficiency is higher in nonoption than option firms19, we use a pooled stochastic production frontier model over the period 1992–2002.20 To obtain inefficiency estimates, we apply the conditional estimator E(uit|eit) proposed by Jondrow, Lovell, Materov, and Schmidt (1982).21 By doing this, we are able to compare inefficiency differences between stock option and nonoption firms. However, we deviate a little from a basic setup in Eq. (1) in that we also account for heteroskedasticity in the inefficiency component uit (e.g., Caudill and Ford, 1993) and in the noise component vit (e.g., Hadri, 1999). The reason is that, according to Bottasso and Sembenelli (2004), unaccounted heteroskedasticity in the uit leads to biased estimates of the production frontier parameters and technical efficiency, whereas unaccounted heteroskedasticity in the vit leads to biased estimates of technical efficiency. We parameterize the variance of inefficiency term uit as an exponential function of firm size (measured by ln(lit)) and stock option variables as follows:22 s2u ¼ s2uit ¼ expðd0 zit Þ ¼ expða þ dL lnðl it Þ þ dopt optit Þ

(2)

s2u ¼ s2uit ¼ expðd0 zit Þ ¼ expða þ dL lnðl it Þ þ dssopt ssoptit þ dbbsopt bbsoptit Þ (3)

Stock Option Schemes and Firm Technical Inefficiency

149

Besides the variance of the inefficiency component, the symmetric noise component can be heteroskedastic with respect to the size of firms. Thus, we model vit as an exponential function of firm size as follows: s2v ¼ s2vit ¼ expðg0 zit Þ ¼ expða þ gL lnðl it ÞÞ

(4)

To address how stock option schemes influences on the mean (e.g., Kumbhakar, Ghosh, & McGuckin, 1991) and the variance of the inefficiency term uit (e.g., Caudill & Ford, 1993), we utilize a novel estimator suggested by Wang (2002).23 As before, we also model heteroskedasticity but now the mean and the variance of inefficiency term are measured by the unconditional statistics of E½uit  and Var½uit .24 Finally, by using the estimator suggested by Wang (2002), we can accommodate unconditional marginal effects of exogenous variables on the mean and the variance of inefficiency term uit and then examine the statistical significance of marginal effects by bootstrapping.

5. ESTIMATION RESULTS Table 4 presents the stochastic production frontier estimation results. For one thing, the production technology parameter estimates are in line with our prior expectations; capital input elasticities are higher and labor input elasticities lower in the manufacturing compared to the ICT sector. In the manufacturing sector, estimated elasticities for capital are 0.29 (0.16 in the ICT sector) and for labor 0.71 (0.85 in the ICT sector), indicating that a production process in the manufacturing sector is more capital and less labor-intensive compared to the ICT sector. As such, our choice to estimate separate frontier models for the sectors seems reasonable. The constant rate of technical change is significant and about 2.5% per year in the manufacturing sector, but the estimate is insignificant in the ICT sector. The estimates of l are statistically significant and higher than one, indicating the existence of inefficiency in the sectors. When comparing the estimates of su between the sectors, the estimates appear to be clearly higher in the ICT sector (0.68) than in the manufacturing sector (0.28). In addition, in the manufacturing sector, the presence of broad-based schemes seems to be associated with the variance of su. The choice of parameterizing the error term vit as a function of firm size is supported statistically in the manufacturing sector, but the parameter estimate of firm size is insignificant in the ICT sector.

MIKKO MA¨KINEN

150

Table 4.

Stochastic Production Frontier Estimates. Manufacturing

Constant ln(labor) ln(capital) Year The variance of v Constant ln(labor) The variance of u Constant ln(labor) opt ssopt bbsopt Year sv su s l ¼ su/sv Log likelihood function Finite sample corrected Akaike’s information criterion (AIC) Wald test for constant returns to scale (p-value) Ho: bkþbl ¼ 1

ICT

(1)

(2)

7.705 (0.156) 0.709 (0.019) 0.293 (0.012) 0.026 (0.006)

7.707 (0.157) 0.710 (0.019) 0.293 (0.011) 0.025 (0.006)

2.185 (0.519) 0.182 (0.074)

2.238 (0.505) 0.174 (0.072)

2.687 (1.920) 0.167 (0.324)

2.699 (2.095) 0.170 (0.354)

3.466 (0.997) 0.125 (0.116) 0.344 (0.183) 

3.889 (1.104) 0.197 (0.131) 

0.590 (0.654) 0.018 (0.108) 0.622 (0.174) 

0.964 (0.843) 0.051 (0.144) 

 0.037 (0.045) 0.175 0.279 0.330 1.59 6.585 7.223 0.88

0.039 (0.193) 0.791 (0.326) 0.052 (0.048) 0.176 0.276 0.327 1.57 10.211 2.050 0.85

(3) 9.502 (0.408) 0.853 (0.060) 0.161 (0.042) 0.007 (0.010)

 0.064 (0.064) 0.155 0.677 0.695 4.37 124.114 269.176 0.96

(4) 9.496 (0.410) 0.856 (0.061) 0.160 (0.043) 0.007 (0.010)

0.397 (0.425) 0.773 (0.169) 0.073 (0.063) 0.156 0.677 0.695 4.34 123.307 269.757 0.69

Notes: The dependent variable is ln(value added). Standard errors in parentheses. We have 62 firms/571 observations in the manufacturing sector and 32 firms/243 observations in the ICT sector. ssopt is a dummy variable for selective and bbsopt is a dummy for broad-based option schemes, respectively. As a control group, we use nonoption firms. Statistically significant at 1% level. Statistically significant at 5% level. Statistically significant at 10% level.

151

Stock Option Schemes and Firm Technical Inefficiency

Table 5 reports conditional inefficiency estimates. As can be seen from Table 5, the mean inefficiency is substantially higher in the ICT sector compared to the manufacturing sector. For example, the estimated mean inefficiency is 0.21 or 21% with a standard deviation of 0.13 in the manufacturing sector, whereas in the ICT sector, it is 0.45 or 45% with a standard deviation of 0.26. It is, however, important to notice that efficient and inefficient firms exist in both sectors. Table 6 summarizes the mean (conditional) inefficiencies by the type of option scheme. In the ICT sector, mean inefficiencies are in the range of 0.44–0.47, indicating little difference in mean inefficiency between option and nonoption firms. On the contrary, in the manufacturing sector our findings suggest that broad-based scheme firms have somewhat higher mean inefficiency than selective and nonoption scheme firms (0.28 compared to 0.22 and 0.21). Table 7 presents stochastic production function estimation results by using the estimator proposed by Wang (2002). Standard errors are adjusted for intragroup correlation, and the mean and the variance of the inefficiency terms are modeled as a function of option plans and firm size. Contrary to Table 4, where our option program variables measure the presence of a plan, now our program variable is potential dilution (dilu), measuring the size of Table 5.

Conditional Inefficiencies.

Estimated Inefficiencies u^it

Mean Standard deviation Minimum Maximum

Table 6.

Manufacturing ICT Total

Manufacturing

ICT

(1)

(2)

(3)

(4)

0.217 0.125 0.040 0.762

0.213 0.126 0.041 0.780

0.451 0.258 0.050 0.959

0.449 0.257 0.051 0.954

Conditional Inefficiencies by the Type of Option Scheme. (1)

(2)

(3)

(4)

Selective option scheme firms

Broad-based option scheme firms

Nonoption firms

Total

0.218 (236 obs.) 0.445 (64 obs.) 0.266 (300 obs.)

0.280 (70 obs.) 0.467 (75 obs.) 0.377 (145 obs.)

0.190 (265 obs.) 0.438 (104 obs.) 0.260 (369 obs.)

0.213 (571 obs.) 0.449 (243 obs.) 0.283 (814 obs.)

MIKKO MA¨KINEN

152

Table 7. Pooled Stochastic Production Frontier Maximum Likelihood (ML) Estimates When Parameterizing the Variance and the Mean of u. Manufacturing Constant ln(labor) ln(capital) Year The mean of u Constant Year diluss dilubb ln(labor)

7.791 0.707 0.286 0.023 17.634 1.302 12.697 3.324 0.194

ICT

(0.042) (0.042) (0.034) (0.007) (110.227) (8.040) (68.423) (33.001) (1.006)

9.570 0.863 0.150 0.003 1.541 0.423 26.733 10.269 0.249

(0.440) (0.052) (0.043) (0.014) (2.786) (0.463) (25.529) (9.962) (0.288)

The variance of u Constant Year diluss dilubb ln(labor)

0.873 0.221 0.573 5.055 0.065

The variance of v Constant

3.280 (0.194) 3.494 (0.339)

(6.222) 0.573 (1.654) 0.124 (0.077) (0.062) (11.964) 5.838 (4.787) 3.829 (4.233) (1.722) (0.091) 0.180 (0.190)

0.64 Wald test for constant returns to scale (p-value) Ho: bkþbl ¼ 1 Wald test for joint significance of variables in the mean 0.99 of u (without constant, p-value) Ho: year, diluss, dilubb, ln(labor) ¼ 0 Wald test for joint significance of variables in the 0.00 variance of u (without constant, p-value) Ho: year, diluss, dilubb, ln(labor) ¼ 0 Wald test for joint significance of diluss and dilubb 0.03 in the mean and variance of u (without constant, p-value) Ho: diluss, dilubb ¼ 0 Log pseudolikelihood 13.595

0.45 0.45 0.00 0.05

112.599

Notes: The dependent variable is ln(value-added). diluss is a selective and dilubb is a broadbased option scheme proxy variable, respectively. Standard errors in parentheses are adjusted for intragroup correlation. We have 62 firms/571 observations in the manufacturing sector and 32 firms/243 observations in the ICT sector. Statistically significant at 1% level. Statistically significant at 5% level. Statistically significant at 10% level.

Stock Option Schemes and Firm Technical Inefficiency

153

an effective scheme in firm i in year t.25 The following key findings emerge from Table 7. First, stochastic production frontier technology parameter estimates are in line with those presented in Table 4. Second, in both sectors, the assumption that all parameters (constant excluded) are jointly zero is rejected at 1% level in the variance of the inefficiency term but not in the mean. On the contrary, the Wald test for the hypothesis that selective (diluss) and broad-based (dilubb) option scheme parameters are jointly zero both in the mean and in the variance of the inefficiency term is rejected at 5% level in both sectors,26 supporting for the parameterization of the inefficiency term as a function of option schemes. Although informative, Table 7 does not provide an estimate of the magnitude of the effects of selective (diluss) and broad-based (dilubb) schemes on the mean and the variance of the inefficiency term uit. To provide a quantitative assessment of the marginal effects, Table 8 reports the marginal effects on E(uit) and Var(uit). The standard errors are bootstrapped Table 8. Marginal Effects on Inefficiency.

Marginal effects on E(uit) Year diluss dilubb ln(labor) Marginal effects on Var(uit) Year diluss dilubb ln(labor)

Manufacturing

ICT

0.003 (0.285) 0.191 (0.715) 0.616 (0.298) 0.013 (0.010)

0.012 (0.018) 1.473 (1.738) 0.026 (1.478) 0.029 (0.038)

0.001 (0.002) 0.050 (0.194) 0.178 (0.089) 0.004 (0.003)

0.000 (0.014) 0.454 (1.67) 0.197 (0.799) 0.025 (0.027)

Notes: Sample means of marginal effects. Standard errors of marginal effects in parentheses are bootstrapped results of 1,000 replications; statistical significant levels are based on biascorrected and accelerated confidence intervals. Statistically significant at 1% level. Statistically significant at 5% level. Statistically significant at 10% level.

MIKKO MA¨KINEN

154

results of 1,000 replications and significance levels are based on biascorrected and accelerated intervals. As can be seen from Table 8, in the manufacturing sector the marginal effects of broad-based schemes are significant on the mean and the variance of the inefficiency term uit. As such, the results on E(uit) indicate that an increase in the potential dilution of broad-based schemes (program size) in the manufacturing sector is expected to increase production inefficiency: the average marginal effect is estimated to be 0.62, that is, a one percentage point increase in the potential dilution of broad-based schemes increases firm technical inefficiency by 0.62%. As qEðlnðvaÞÞ=qdilubb ¼ qEðuÞ=qdilubb, the marginal effect on firm productivity would be about 0.62%. Also, the average marginal effect of the potential dilution of broad-based schemes on Var(uit) is positive, implying an increase in production uncertainty. Taken together, these results suggest that as time goes by broad-based scheme firms in the manufacturing sector achieve lower and more uncertain productivity growth (other things equal). For selective schemes, we find no evidence of significant marginal effects in the sectors. In sum, our empirical findings do not provide statistical support for the view that by using employee stock option schemes, firms can reduce technical inefficiency and therefore be able to enhance their performance.

6. CONCLUSIONS In this study, we have applied stochastic production frontier models to assess whether employee stock option schemes can be associated with firm technical inefficiency. Especially, we examined (i) do stock option firms have a lower technical inefficiency compared to nonoption firms and (ii) does the impact of options on firm technical inefficiency depend on the type of scheme (broad-based vs. selective). The data we have used is a novel panel data set on publicly listed Finnish firms in the manufacturing and ICT sector over the period 1992–2002. The data enabled us to distinguish broad-based schemes from those selectively allocated to particular employees. Our main findings can be summarized as follows. First, although efficient and inefficient firms exist in both sectors, the mean inefficiency estimates in the ICT sector are clearly higher compared to the manufacturing sector. Second, when firms are categorized by the type of option schemes, we find that broad-based scheme firms in the manufacturing sector have higher mean inefficiency than selective and nonoption firms. On the contrary, in the ICT sector the mean inefficiency estimates do not indicate substantial differences between option and nonoption firms. Third, the quantitative

Stock Option Schemes and Firm Technical Inefficiency

155

assessment of the marginal effects in the manufacturing sector indicates that broad-based schemes have a significant positive contribution to the mean and the variance of the inefficiency term uit; the average marginal effect on the mean of the inefficiency E(uit) shows that an increase in the size of broad-based schemes increases production inefficiency, and, respectively, the marginal effect on the variance of the inefficiency term Var(uit) implies an increase in production uncertainty. These findings suggest that, other things equal, broad-based scheme firms in the manufacturing sector achieve lower and more uncertain productivity growth compared to other firms in the sector. However, in the ICT sector the marginal effects are insignificant. As a whole, our empirical findings do not provide statistically significant support for the view that by using employee stock option schemes firms can reduce technical inefficiency and therefore be able to enhance their performance. The potential theoretical reasons for this might poor option scheme implementation, free riding, accounting myopia, line-of-sight, and managerial rent-seeking. Finally, our key empirical finding that option schemes may not enhance firm performance (by reducing technical inefficiency) is similar to Jones et al. (2010), who, by using a different empirical approach, consistently find insignificant association between stock option schemes (both selective and broad-based) and firm productivity. However, our findings differ noticeably from those earlier empirical studies that have found the evidence of a positive productivity effect of broad-based option schemes. The possible reasons for the divergence might be the issues like (i) more reliable and extensive data that encompass a period spanning both stock market boom and bust; (ii) our data are less likely to suffer from selection biases compared to previous work; (iii) differences in institutional arrangements, and (iv) differences in empirical approaches.

NOTES 1. In the beginning, stock options were allocated to executives. But the association of options with managerial compensation mechanism changed after companies started to issue options more broadly to their employees (e.g., Blasi, Kruse, & Bernstein, 2003; Jones et al., 2006). 2. See Kumbhakar and Lovell (2000), Wang (2002), and Bottasso and Sembenelli (2004) for stochastic frontier models. 3. See Jones et al. (2010) for more detailed discussion. 4. Ma¨kinen (2001) describes the evolution of stock option programs in Finland. Jones et al. (2006) study the determinants of option schemes adoption in Finland.

156

MIKKO MA¨KINEN

In addition, they summarize in more detail the evolution of options and the institutional background in Finland. See also Ma¨kinen (2007, 2008). 5. See Kiander and Vartia (1996) and Honkapohja and Koskela (1999) for more detailed discussion about the Great Finnish Depression during 1990–1993. 6. Firms may adopt schemes for different reasons. For example, the shareholders of a firm may prefer to broaden schemes to a larger set of employees, or there is a need to change the terms of a scheme for some reason. 7. This is done to increase the number of observations in the data. 8. We are aware that few unlisted Finnish firms, mainly in the ICT sector, have also adopted option schemes, especially during the bull market in the end of 1990s. Unfortunately, no information on these firms and option schemes was unavailable. We believe, however, that the number of these unlisted option firms is small, because option schemes probably work only in situations where the value of shares can be assessed in the stock market. Also, to study the impacts of stock option programs with publicly available data, our data seem to be a reasonable choice. 9. Our classification is different from Kroumova and Sesil (2006), who use a 50% threshold as a criterion for broad-based schemes. Our data do not include this information, but they have the important advantage of being derived from publicly reported sources that must be externally verifiable, rather than from confidential surveys. 10. We also interviewed a partner at an investment bank, who has been personally involved in setting up dozens of option schemes. He confirmed that there are dramatic differences in the participation rates for option schemes, depending on eligibility. 11. Unfortunately, we do not have information on exercise prices to calculate Black–Scholes values. 12. Early specifications focused on estimating technical inefficiency with crosssectional data. However, an access to panel data allowed a richer modeling approach in the form of the fixed effects (e.g., Schmidt & Sickles, 1984) and the random effects (e.g., Pitt & Lee, 1981) estimators enabling to relax some relatively strong distributional assumptions needed in cross-sectional models. Excellent literature surveys are Bauer (1990), Greene (1993, 1997), and Kumbhakar and Lovell (2000). To account for simultaneously heteroskedasticity and inefficiency, Greene (2001, 2002a, 2005) proposes a new ‘‘true’’ fixed-effects framework that more explicitly follows stochastic frontier modeling foundations applied frequently in cross-section frontier models. See also Prentice and Gloeckler (1978) and Sueyoshi (1993) for a formal derivation of the estimators. 13. Although the Cobb–Douglas functional form is more restrictive than other functional forms, such as the translog, we prefer the Cobb–Douglas production function, because the number of the translog production functions did not converge. 14. We recognize there is an issue of endogeneity bias in the model, because our covariates may not be exogenous. As a consequence, this can lead to (i) biased parameter estimates and (ii) potentially biased estimates of inefficiency. Unfortunately, because we cannot address endogeneity bias, this is a drawback of our study. 15. On theoretical grounds, firm value added is a preferable measure to sales, because value added does not include intermediate inputs that are purchased from other firms. 16. Finnish firms report only mean number of employees per year in financial statements. All value measures are deflated using an industry-specific gross output deflator at 2000 constant Euros, obtained from Statistics Finland.

Stock Option Schemes and Firm Technical Inefficiency

157

17. The noise component vit captures measurement errors and production function misspecifications, whereas uit is related to technical inefficiency. See also Aigner et al. (1977) and Meeusen and van den Broeck (1977). 18. For one thing, manufacturing firms probably use more capital in their production process compared to ICT firms. Respectively, labor may be a more important production factor in the ICT than in the manufacturing sector. Besides these divergences, also, other differences between the sectors can be imaged easily. As such, production frontiers likely differ between the sectors, supporting the estimation of separate models for the sectors. 19. See Kumbhakar and Lovell (2000) for detailed discussion on how to account for exogenous influences in the one- and two-step approaches. Also, according to the Monte Carlo studies conducted by Schmidt and Wang (2002), the one-step modeling approach is more favorable than the two-step approach, where inefficiencies and exogenous effects are estimated sequentially. 20. A priori, we conducted several fixed effects estimations by LIMDEP where we modeled the mean and the variance of the inefficiency term. Unfortunately, all models behaved extremely poorly, even when we tried the stratification method. As noted in LIMDEP’s manual, pp. E24–E27 (Greene, 2002b), fixed-effects formulations, especially based on the Newton’s method, can be ‘‘extremely problematic in all but the most favorable of cases.’’ Therefore, we use the pooled ML model. 21. It is also possible to obtain confidence intervals for the point estimates of technical inefficiency, but we do not examine this issue here. For more details, see Horrace and Schmidt (1996) and Bera and Sharma (1999). 22. This modeling approach has a lot of common with Bottasso and Sembenelli (2004), who provide evidence on the relation between identity of ultimate owners and technical inefficiency by estimating stochastic production frontiers on Italian manufacturing firms. 23. Note that here we use a different estimator. The reason is that the estimator, kindly provided by Wang, utilizes STATA’s maximum likelihood routines and assumes that the same z affects both the mean and the variance of uit. According to Bera and Sharma (1999), the mean of uit measures the expected value of technical inefficiency, whereas the variance measures production uncertainty. 24. Wang (2002) underlines that the marginal effects on the conditional mean and variance of uit are almost intractable, particularly when the variances of uit and vit are modeled. 25. The reason is that we had major problems in convergence of the ML estimator with option program dummy indicators. 26. We also specified models (not reported here) where the variance of vit was modeled as a function of firm size. All the estimated models and the performed Wald tests indicated that vit is not heteroskedastic with respect to firm size.

ACKNOWLEDGMENTS We are grateful for comments and suggestions to participants at the Annual Meeting of the Finnish Economic Association in Helsinki, ESEM in Milan,

MIKKO MA¨KINEN

158

and EALE in Amsterdam. We also thank Derek Jones, Panu Kalmi, and an anonymous referee for their helpful comments. In addition, we are grateful to Academy of Finland (Grant 120234) and Finnish Work Environment Fund (Grant 108708). The article is a revised version of Chapter 3 of the author’s PhD dissertation at the Helsinki School of Economics.

REFERENCES Aigner, D. J., Lovell, C. A. K., & Schmidt, P. (1977). Formulation and estimation of stochastic frontier production function models. Journal of Econometrics, 6(1), 21–37. Alchian, A., & Demsetz, H. (1972). Production, information costs, and economic organisation. American Economic Review, 62, 777–795. Bauer, B. W. (1990). Recent developments in the econometric estimation of stochastic frontiers. Journal of Econometrics, 46, 39–56. Bebchuk, L. A., & Fried, J. M. (2003). Executive compensation as an agency problem. Journal of Economic Perspectives, 17, 71–92. Bera, A. K., & Sharma, S. C. (1999). Estimating production uncertainty in stochastic frontier production function models. Journal of Productivity Analysis, 12, 187–210. Blasi, J., Kruse, D., & Bernstein, A. (2003). In the company of owners: The truth about stock options and why every employee should have them. New York: Basic Books. Bottasso, A., & Sembenelli, A. (2004). Does ownership affect firms’ efficiency? Panel data evidence on Italy. Empirical Economics, 29, 769–786. Caudill, S. B., & Ford, J. M. (1993). Biases in frontier estimation due to heteroscedasticity. Economic Letters, 41, 17–20. Conyon, M. J., & Freeman, R. B. (2004). Shared modes of compensation and firm performance: UK evidence. In: R. W. Blundell, D. Card & R. B. Freeman (Eds), Seeking a premier league economy. Chicago: University of Chicago Press. Core, J. E., Guay, W. R., & Larcker, D. F. (2003). Executive equity compensation and incentives: A survey. FRNBY Economic Policy Review (April), 27–50. Freeman, R., Kruse, D., & Blasi, J. (2008). Worker responses to shirking under shared capitalism. NBER Working Paper 14227. National Bureau of Economic Research, Cambridge, MA, USA. Greene, W. (1993). The econometric approach to efficiency analysis. In: H. O. Fride, C. A. K. Know & S. S. Schmidt (Eds), The measurement of productive efficiency. New York: Oxford University. Greene, W. (1997). Frontier production functions. In: M. H. Pesaran & P. Schmidt (Eds), Handbook of applied econometrics, Vol. II: Microeconometrics. Oxford: Blackwell Publishers. Greene, W. (2001). Fixed and random effects in nonlinear models. Working Paper no. 01-01. Department of Economics, Stern School of Business, New York University. Available at http://www.stern.nyu.edu/Bwgreene/panel.pdf Greene, W. (2002a). The behavior of the fixed effects estimator in nonlinear models. Working Paper no. 02-05. Department of Economics, Stern School of Business, New York University. Available at http://www.stern.nyu.edu/Bwgreene/nonlinearfixedeffects.pdf Greene, W. (2002b). The LIMDEP Reference Guide. New York: Econometric Software, Inc.

Stock Option Schemes and Firm Technical Inefficiency

159

Greene, W. (2005). Reconsidering heterogeneity in panel data estimators of the stochastic frontier models. Journal of Econometrics, 126, 269–303. Habib, M., & Ljungqvist, A. (2005). Firm value and managerial incentives: A stochastic frontier approach. Journal of Business, 78(6), 2053–2094. Hadri, K. (1999). Estimation of a doubly heteroscedasticity stochastic frontier cost function. Journal of Business and Economic Statistics, 17, 359–363. Hanlon, M., Rajgopal, S., & Shevlin, T. (2003). Are executive stock options associated with future earnings? Journal of Accounting and Economics, 36, 3–43. Honkapohja, S., & Koskela, E. (1999). The economic crisis of the 1990s in Finland. Economic Policy, 14, 400–436. Horrace, W. C., & Schmidt, P. (1996). Confidence statements for efficiency estimates from stochastic frontier models. Journal of Productivity Analysis, 7, 257–282. Hyytinen, A., Kuosa, I., & Takalo, T. (2003). Law or finance: Evidence from Finland. European Journal of Law and Economics, 16, 59–89. Ittner, C. D., Lambert, R. A., & Larcker, D. F. (2003). The structure and performance of equity grants to employees of new economy firms. Journal of Accounting and Economics, 34, 89–127. Jackson, M., & Lazear, E. (1991). Stock, options, and deferred compensation. In: R. Ehrenberg (Ed.), Research in labor economics (Vol. 12, pp. 41–62). Greenwich, CT: JAI Press. Jondrow, J., Lovell, C. A. K., Materov, I., & Schmidt, P. (1982). On the estimation of technical inefficiency in the stochastic frontier production function model. Journal of Econometrics, 19, 233–238. Jones, D. C., Kalmi, P., & Ma¨kinen, M. (2006). The determinants of stock option compensation: Evidence from Finland. Industrial Relations, 45(3), 437–468. Jones, D. C., Kalmi, P., & Ma¨kinen, M. (2010). The productivity effects of stock options schemes: Evidence from Finnish Panel Data. Journal of Productivity Analysis, 33(1), 67–80. Kandel, E., & Lazear, E. (1992). Peer pressure in partnerships. Journal of Political Economy, 100, 801–817. Keynes, J. (1936). The general theory of employment, interest and money. London: MacMillan. Kiander, J., & Vartia, P. (1996). The great depression of the 1990s in Finland. Finnish Economic Papers, 9(1), 72–88. Kroumova, M. K., & Sesil, J. C. (2006). Intellectual capital, monitoring and risk: What predicts the adoption of broad-based employee stock options. Industrial Relations, 45(4), 734–752. Krugman, P. (2009). How did economists get it so wrong? New York Times, September 6, p. MM36. Available at http://www.nytimes.com/2009/09/06/magazine/06Economic-t. html?ref=magazine&pagewanted=all Kumbhakar, S. C., Ghosh, S., & McGuckin, J. T. (1991). A generalized production frontier approach for estimating determinants of inefficiency in U.S. dairy farms. Journal of Business and Economic Statistics, 9, 279–286. Kumbhakar, S. C., & Lovell, C. A. K. (2000). Stochastic frontier analysis. Cambridge: Cambridge University Press. Larcker, D. F. (2003). Discussion of are executive stock options associated with future earnings? Journal of Accounting and Economics, 36, 91–103. Liljeblom, E., & Pasternack, D. (2006). Share repurchases, dividends and executive options: The effect of dividend protection. European Financial Management, 12, 7–28.

160

MIKKO MA¨KINEN

Ma¨kinen M. (2001). Optiot – Suomalaisjohtajien uusi kannustin (Title in English: Stock options  The new incentive of Finnish executives). In Finnish with English Summary. The Research Institute of the Finnish Economy (ETLA), Series B182, Helsinki. Ma¨kinen, M. (2007). Essays on stock option schemes and CEO compensation. HSE Acta Universitatis Oeconomicae Helsingiensis, A-291, Helsinki. Ma¨kinen, M. (2008). CEO compensation, firm size and firm performance: Evidence from Finland. In: L. Oxelheim & C. Wihlborg (Eds), Markets and compensation for executives in Europe. Bingley, UK: Emerald Group Publishing. Meeusen, W., & van den Broeck, J. (1977). Efficiency estimation from Cobb–Douglas production functions with composed error. International Economic Review, 18(2), 435–444. Murphy, K. J. (1999). Executive compensation. In: O. C. Ashenfelter & D. Card (Eds), Handbook of labor economics (Vol. 3B). Amsterdam: Elsevier Science B.V. Oyer, P. (2004). Why do firms use incentives that have no incentive effects? Journal of Finance, 59, 1619–1641. Pitt, M., & Lee, L. (1981). The measurement and sources of technical inefficiency in Indonesian weaving industry. Journal of Development Economics, 9, 43–64. Prentice, R., & Gloeckler, L. (1978). Regression analysis of grouped survival data with application to breasts cancer data. Biometrics, 34, 57–67. Rousseau, D. M., & Shperling, Z. (2003). Pieces of the action: Ownership and the changing employment relationship. Academy of Management Review, 28, 553–570. Schmidt, P., & Sickles, R. (1984). Production frontiers with panel data. Journal of Business and Economic Statistics, 2(4), 367–374. Schmidt, P., & Wang, H.-J. (2002). One-step and two-step estimation of the effects of exogenous variables on technical efficiency levels. Journal of Productivity Analysis, 18(2), 129–144. Sesil, J. C., Kroumova, M. A., Blasi, J. R., & Kruse, D. L. (2000). Broad-based employee options in the US: Do they impact company performance? Academy of Management Proceedings, HR, pp. G1–G6. Sesil, J. C., Kroumova, M. A., Blasi, J. R., & Kruse, D. L. (2002). Broad-based employee options in ‘‘New Economy’’ firms: Company performance effects. British Journal of Industrial Relations, 40, 273–295. Sueyoshi, G. (1993). Techniques for the estimation of maximum likelihood models with large numbers of group effects. Manuscript. San Diego: Department of Economics, University of California. Vroom, V. H. (1995). Work and motivation (Rev. ed.). San Francisco: Jossey-Bass. Wang, H.-J. (2002). Heteroscedasticity and non-monotonic efficiency effects of a stochastic frontier model. Journal of Productivity Analysis, 18, 241–253.

THE UNION WAGE PREMIUM, VOICE, AND NONUNION WORKERS’ ATTITUDES: BEFORE AND AFTER JAPAN’S LOST DECADE Tsuyoshi Tsuru 1. INTRODUCTION In their now classic book What Do Unions Do? Freeman and Medoff (1984) open their discussion as follows (p. 3): ‘‘Trade unions are the principal institution of workers in modern capitalistic societies. For over 200 years, since the days of Adam Smith, economists and other social scientists, labor unionists, and businessmen and women have debated the social effects of unionism. Despite the long debate, however, no agreed-upon answer has emerged to the question: What do unions do?’’ In the remainder of the book, the authors provide a coherent answer to this question and, as a result, What Do Unions Do? has become firmly established as a cornerstone for the economic analysis of labor unions. However, since the first publication of the book, developments have emerged that are inconsistent with Freeman and Medoff’s predictions regarding the effectiveness of unions. One such development, for example, is the precipitous decline in union density in many countries around the world; another is Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 161–204 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011011

161

162

TSUYOSHI TSURU

the growing importance of, and interest in, alternative forms of employee representation. In Japan, too, the work of Freeman and Medoff sparked an interest in the role of labor unions and stimulated a host of empirical studies. In this context, interest in Japan has focused on the effect of unions on wages. However, the findings of these studies remained largely inconclusive. Studies conducted at the beginning of the 1990s arrived at contradictory results regarding union wage effects, and it is only in the new millennium that clear supporting evidence for a positive effect of unions on wages was found. Rephrasing the original question posed by Freeman and Medoff, one could therefore ask: ‘‘What did unions do in the 1990s, and what do unions do today?’’ Against this background, and focusing on Japan, this chapter pursues the following three objectives. First, it seeks to discover why the conflicting conclusions with regard to union wage effects in the 1990s and 2000s emerged and whether the different results reflect changes in union behavior during this period. The second objective is to investigate changes in union voice effects, one of the major issues addressed by Freeman and Medoff (1984). Finally, the third objective is to examine changes in attitudes among nonunion workers with regard to unionization at their workplaces. To analyze these three issues, two unique datasets are used: the results of a survey of individual workers conducted by the Japan Institute of Labor (now the Japan Institute for Labor Policy and Training) in 1992, and the results of a repeat survey using the same questionnaire conducted by the Cabinet Office in 2007. The 2007 questionnaire survey exactly duplicated the 1992 survey questions in order to investigate and highlight key industrial relations issues such as union wage premiums, union voice effects, and support for unionization among nonunion workers. In other words, the data were gathered to analyze changes of worker attitude toward unions and industrial relations, matters on which conventional government surveys had shed little light. The remainder of the chapter is organized as follows. Section 2 provides an overview of preceding research on the economic effects of labor unions and outlines the research agenda. Section 3 presents a description of the data used for analysis. Section 4 begins the empirical analysis, focusing on union wage effects, while Section 5 concentrates on union voice effects. Section 6 then looks at nonunion workers’ attitudes toward unionization. A discussion of the result obtained is provided in Section 7, and Section 8 concludes the chapter.

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

163

2. THE ECONOMIC EFFECTS OF UNIONS: PREVIOUS STUDIES AND RESEARCH AGENDA 2.1. Previous Studies 2.1.1. Union Wage Effects1 The purpose here is to provide a brief look at principal studies relevant to the analysis in the remainder of the chapter. Table 1 presents an overview of major studies on union wage effects and their results. Three patterns stand out. First, studies using data for the 1980s find a significant negative wage effect. Second, studies using data compiled at the beginning of the 1990s find virtually no evidence of a union wage premium. And third, studies based on survey data for the 2000s indicate a significant positive wage premium. Two possible interpretations can be drawn from these results. One is that, as Blanchflower and Bryson (2007) found for the United States, the wage premium moves countercyclically, that is, it is large during economic downturns and small during economic upswings. However, it is difficult to determine whether this is also the case for Japan because there is hardly any academic research on Japanese wage premiums for the latter half of the 1990s. The other possible interpretation is that the prolonged recession of the 1990s (the ‘‘Lost Decade’’) brought about changes in the union wage effect in Japan that are quite unrelated to the question of whether changes in wage premiums are countercyclical or not. Using data for before and after the Lost Decade, the analysis below suggests that this is indeed the case. 2.1.2. Union Voice Effects2 The effects of unions are of course not limited to wages. Since Freeman and Medoff (1984), a large number of studies have provided evidence that unions can change the channels of communication between labor and management. Above all, unions can raise the bargaining power of workers with regard to a variety of workplace-related issues. In the United States, this union voice effect has often been used to explain low job separation rates. By helping to resolve workplace-related problems and thus providing an alternative to quitting (‘‘exit’’), unions decrease the likelihood that a worker will leave a company. One way to examine whether workers value the bargaining power provided by unions is to quantitatively examine the extent to which union membership affects job separation. If, in addition to resolving workplacerelated problems, union voice provides an effective alternative to quitting,

Survey Period

Data Source and Sample Size

1987

Questionnaire survey of 391 survey monitor registrants in greater Tokyo, Kansai region, and 12 major cities 2000–2003 Sample of 2,415 persons from the Japanese General Social Surveys (JGSS)

2003

Comment

Men: þ11.3% (statistically significant) Women: 5.3% (not statistically significant) Men and women overall: þ17% After controlling for industry and (statistically significant) occupation: þ7% (statistically significant); also provides factor decomposition following DiNardo, Fortin and Lemieux (1996) Men and women overall: þ12.6% Working hours are not controlled for (statistically significant) Men: þ5.1% (statistically significant) Women: þ24.1% (statistically significant)

979 union and nonunion manufacturing firms from the Yearbook of Japanese Unlisted Companies Questionnaire survey of 689 firms Not statistically significant nationwide Questionnaire survey of 1,104 Not statistically significant workers in the greater Tokyo area

Nitta and Shinozaki 2000–2003 Sample of 1,432 persons working at (2008) firms with 30 or more employees from the Japanese General Social Surveys (JGSS)

Hara and Kawaguchi (2008)

After the lost decade Noda (2005)

Tachibanaki and 1991 Noda (1993) Tsuru and Rebitzer 1992 (1995)

Brunello (1992)

Men: 13.2% (statistically significant) Women: 15.4% (statistically significant) 2.8% (statistically significant)

Union Wage Effect

Major Preceding Studies on Union Wage Effects.

Before the lost decade Kalleberg and 1981–1983 Questionnaire survey of 3,735 Lincoln (1988) employees at 46 plants in Kanagawa prefecture

Author(s) and Year of Publication

Table 1.

164 TSUYOSHI TSURU

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

165

one should observe that, controlling for other factors, the propensity for job separation among unionized workers is lower than that among nonunion workers. However, the propensity for job separation is only an indirect indicator of the value that workers place on their current job. A more direct method is to examine the effect that union membership has on workers’ degree of job satisfaction. Studies examining the effect of labor unions on voluntary job separation or the intention to change jobs in Japan include Muramatsu (1984), Tomita (1993), and Tsuru and Rebitzer (1995). Whereas Muramatsu (1984) and Tomita (1993) found that union membership had the effect of reducing voluntary job separations; Tsuru and Rebitzer’s (1995) result suggested that unions do not significantly reduce exit propensities. In this context, a recent study of great interest is the one by Todate (2007). Using data by industry from the Survey of Employment Trends for the years 1984 to 2002, Todate examined the impact of unionization rates on voluntary and firm-initiated separations. He found that for men a higher unionization rate significantly lowered the rate of voluntary separation and that although the estimated coefficient for women had the same sign, it was not statistically significant. Todate provided two reasons to explain the different results for men and women. The first is that, with the exception of grievance procedures, union policies tend to center on men. The second is that women tend to be less integrated into firms than men and their job tenures tend to be shorter, so that women cannot capture the long-term benefits of union membership. The advantages of union membership and voice are therefore small for women. 2.1.3. Nonunion Workers’ Support for Unionization3 In contrast with growing body of literature in the United States, such as Farber (1990) and Freeman and Rogers (1999), there is very little research on nonunion workers’ attitudes toward unionization in Japan. However, one notable study in this field has been conducted by Noda and Tachibanaki (1993). Using data from a survey they conducted among firm employees, they conducted a probit analysis in which the dependent variable was whether workers thought a union was necessary in their firm. They found that workers who thought that a union was necessary tended, for example, to expect that this would bring an improvement in wages and better access to information regarding the firm’s management and that such workers tended to be dissatisfied with personnel evaluation. However, Noda and Tachibanaki also found that certain groups of workers whose shares in

166

TSUYOSHI TSURU

overall employment had increased, such as women and white-collar workers, attached a relatively low value to labor unions. Finally, they found that, with more companies utilizing merit-based pay systems, a growing number of workers were seeking to improve their lives by raising their performance at work rather than trying to improve working conditions through joining a union, while the number of workers wanting to organize a union at nonunion firms decreased. This last point provides the main reason for the decline in unionization rates in Japan. This interpretation parallels Freeman and Rebick’s (1989) analysis showing that the declining rate of new union organization accounts for a larger share of the decline in union density than such structural factors as increases in part-time, female, and service jobs. The study by Noda and Tachibanaki focuses on nonunion workers’ attitudes as an important factor explaining unionization rates. In particular, the quantitative analysis, following the example of Farber (1990) for the United States, of how attitudes toward unionization among nonunion workers are related to dissatisfaction and worker attributes represents an important step, and their argument that the encroachment of the merit system has reduced the number of workers willing to take charge of organizing a union provides a novel explanation. The most important recent research on attitudes toward unionization among nonunion workers is two studies by Hara and Sato (2004, 2005). Using data from a questionnaire survey on private firm employees residing in the Tokyo area, the Kyoto and Osaka regions, and the 12 major cities of Japan, they search for the sources of union support. It is well known that dissatisfaction with working conditions is a major factor underlying support for unionization (Tsuru & Rebitzer, 1995). In addition, Hara and Sato found the following. First, support for unionization is affected by the extent of workers’ understanding of workers’ rights (such as basic workers’ rights and various worker protection laws). Second, support for unionization is positively correlated with the perception of labor unions’ effectiveness. And third, support for unionization is lower among workers in nonunion firms than among workers in union firms. While Hara and Sato found that support for unionization at nonunion firms is low, why this is the case remains an issue that has not yet been examined in detail. A possible reason is that in many nonunion firms in Japan, other, nonunion arrangements – such as labor-management consultations and employee associations – are in place (Tsuru, 2002; Tsuru & Morishima, 1999). Therefore, an intriguing research issue is what effect the existence of such alternative arrangements has on the support for unionization among nonunion workers.

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

167

2.2. Research Agenda Given the above considerations, this chapter analyzes the economic effects of unions from the following perspectives. 1. The union wage effect: In addition to calculating the union effect on wage levels by estimating wage functions, the effect of unions on wage distribution is examined by employing counterfactual kernel density estimates following the methodology of DiNardo et al. (1996) and Hara and Kawaguchi (2008). 2. The union voice effect: Using a probit model, the union effect on the propensity for voluntary quitting and the degree of job dissatisfaction is estimated. 3. The determinants of support for unionization: Again using a probit model, this part of the analysis looks at the effect of worker attributes such as educational attainment and of alternative voice mechanisms such as labor-management consultation arrangements and employee associations on support for unionization (i.e., the formation of unions) by nonunion workers.

3. THE DATA The analysis in this chapter uses cross-section data for two different years derived from the same questionnaire. The first set of cross-section data is from a survey conducted by the Japan Institute of Labor (now the Japan Institute for Labor Policy and Training) in 1992. The sample frame consists of 2,800 men and women aged 18 to 59, residing within a radius of 30 km from Tokyo Station (i.e., in Tokyo, Kanagawa, Chiba, and Saitama) and selected through stratified two-stage random sampling from the Basic Resident Register of the local government. Among the 2,800 men and women, 1,736 were in employment, and these were chosen as the survey subjects. A total of 1,104 replied to the survey, for a response rate of 63.6 percent. For the survey, respondents were interviewed at their residences, and the entire survey was conducted between February 7 and 24, 1992. The second set of cross-section data is from a survey conducted by the Cabinet Office in 2007. This is a repeat survey using the same questionnaire as in 1992. The sample frame consists of 5,000 men and women residing in the Tokyo metropolitan area (Tokyo, Kanagawa, Chiba, and Saitama) and randomly chosen from among the mail survey monitor registrants of Intage, Inc., one of the largest market research companies in Japan. Among these,

168

TSUYOSHI TSURU

4,371 responded to the survey for a response rate of 87.4 percent. The entire survey was conducted by mail between February 16 and 26, 2007. The Intage data boast high reliability. First of all, most of the survey monitor registrants were selected through random sampling from the Basic Resident Register of the local government, although some were selected through alternative methods including random digit dialing, a method for selecting people for participation in telephone surveys. Intage does not advertise on internet for the monitor registrants in order to avoid selection bias associated with computer access. A second important characteristic is that the sample frame accurately reflects the population: 5,000 of the 240,000 persons registered with the company were selected for this study in accordance with the male–female ratios specified for all adult 10-year age groups (e.g., 25–34-year-olds) in Tokyo and other metropolitan prefectures by the National Population Census 2005 of Japan, and the same ratio of regularto-nonregular employment reported in the government-conducted Labour Force Survey 2005 applied to each cohort. Third, the survey is conducted by mail, and e-mail submissions are not accepted. Fourth, an incentive payment of 500 yen (about 5 US dollars) to each respondent is paid by the company, yielding a high rate of response. In sum, the 2007 Intage-supplied data are closely comparable to the 1992 data that were randomly sampled from the Basic Resident Register. Table 2 provides a comparison of the population and the sample in each of the two years. As can be seen, the survey respondents represent a crosssection that quite accurately reflects the makeup of workers in the Tokyo metropolitan area. The ratio of labor union members in the samples is 32.1 percent in 1992 and 29.2 percent in 2007.4 These values are higher than the estimated national unionization rates of 24.4 percent (1992) and 18.1 percent (for 2007), but the estimated unionization rate for the Tokyo metropolitan area is several percentage points higher than the national average.5 The survey is confined to workers in the private sector. Summary statistics for each of the variables after outliers were removed from the sample are shown in Table 3. One difference is that the 1992 survey did not ask whether labor-management consultation mechanisms existed, but the 2007 survey did.

4. UNION WAGE EFFECTS This section begins the empirical analysis by measuring the effect of union membership on wage levels (total annual salaries including bonuses). The results are displayed in Table 4 and indicate that in the 1992 data, no wage

169

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

Table 2.

Comparison of Samples and Population.

(a) Sample Structure by Sex and Union Membership Number of Persons

Sample Share (%)

1992 Men Women Union members

654 450 354

59.2 40.8 32.1

2007 Men Women Union members

2544 1827 1146

58.2 41.8 29.2

(b) Age Distribution of Sample and Population (%) Age in Years 1992 Sample Men Women Total Population Men Women Total 2007 Sample Men Women Total Population Men Women Total

18–19

20–29

30–39

40–49

50–59

Total

3.2 2.7 3.0

27.7 29.8 28.5

32.3 22.0 28.1

21.9 29.1 24.8

15.0 16.4 15.6

100 100 100

5.9 5.9 5.9

28.7 26.4 27.6

21.7 21.3 21.5

24.7 26.2 25.4

19.0 20.3 19.7

100 100 100

0.7 0.3 0.5

20.5 22.6 21.4

34.4 29.5 32.2

23.1 22.4 22.8

21.2 25.3 23.0

100 100 100

1.7 2.2 1.9

21.6 24.8 23.0

28.9 25.2 27.2

23.4 24.1 23.7

24.3 23.8 24.1

100 100 100

Note: 1992: The population consists of 18- to 59-year-olds within a 30 km radius of Tokyo Station. The population data are taken from Management and Coordination Agency (1990), Population Census of Japan. 2007: The population consists of 18- to 59-year-olds residing in the Tokyo metropolitan area (Tokyo, Kanagawa, Chiba, and Saitama) among the 240,000 mail survey monitor registrants of Intage, Inc.

Union member dummy Voice-oriented employee association dummy Recreation-oriented employee association dummy Joint consultation system dummy Annual cash earnings (natural log) Male dummy Firm size (natural log) Educational attainment Specialized school dummy Vocational/junior college dummy University dummy Potential labor market experience Full-time regular worker dummy Straight time working hours Overtime working hours Tenure years Tenure^2/100 Voluntary job quitting dummy Degree of job dissatisfaction Degree of dissatisfaction with regard to wages Degree of dissatisfaction with regard to welfare benefits Degree of dissatisfaction with regard to working hours

0.300 0.055 0.126

14.952 0.588 5.502 0.082 0.104 0.305 8.644 0.763 37.231 2.826 8.330 1.438 0.213 2.688 3.054 2.958 2.782

1023

876 1023 997

1021 1021 1021 1003 1023 1016 1023 1020 1020 986 1023 1023

935

1022

Mean

1023 1023

No. of Obs.

0.738 0.492 2.518

0.332

0.459 0.228

SD

1992

1.110

1.030

1

1

0 0 0 0 0 0.800 0 0.083 0.000 0 1 1

13.618 0 0

0

0 0

Min.

5

5

1 1 1 44 1 98 38.3 45.25 20.476 1 5 5

16.670 1 9.031

1

1 1

Max.

Summary Statistics.

0.275 0.305 0.460 10.520 0.425 10.064 5.284 8.629 2.732 0.410 0.972 1.061

Table 3.

3348

3126

2.951

3.247

0.130 0.124 0.436 10.025 0.665 42.549 6.071 8.500 1.506 0.327 3.104 3.410

0.024 14.972 0.570 6.003

2331 3270 3382 3275 3367 3367 3367 3189 3382 3266 3274 3317 3317 3187 3375 3368

0.114

0.281 0.036

Mean

3382

3329 3382

No. of Obs.

1.214

1.105

0.336 0.329 0.496 10.637 0.472 15.746 8.793 8.855 2.796 0.469 1.098 1.165

0.152 0.714 0.495 2.455

0.318

0.450 0.186

SD

2007

1

1

0 0 0 0 0 0 0 0 0 0 1 1

0 13.651 0 0

0

0 0

Min.

5

5

1 1 1 41 1 140 95 41.75 17.431 1 5 5

1 16.649 1 9.252

1

1 1

Max.

170 TSUYOSHI TSURU

Potential labor market experience: The number of years since graduation from school minus current job tenure. If workers complete their education prior to accepting the current job, this variable measures the amount of potential labor market experience between school-leaving and beginning work for the current employer. Full-time regular worker dummy: A dummy variable equal to 1 if respondent is a full-time, regular employee. Straight time working hours: The number of hours worked in the previous week. (Hours do not include time spent on breaks.) Overtime working hours: The number of paid and unpaid overtime hours in the previous week. Tenure: Years at current employer. Voluntary job quitting dummy: A dummy variable set to 1 if respondent answered yes to the question, ‘‘Do you hope to change jobs in the next year or two?’’ Degree of dissatisfaction: The variables indicate respondent’s level of dissatisfaction with overall aspects of job, wages, welfare benefits, or working hours. In all four dissatisfaction variable, responses are coded as follows: 1 ¼ very satisfied; 2 ¼ satisfied; 3 ¼ neither satisfied nor dissatisfied; 4 ¼ dissatisfied; and 5 ¼ very dissatisfied.

University dummy: A dummy variable equal to 1 if graduated from a university or graduate school.

Vocational/junior college dummy: A dummy variable equal to 1 if graduated from a vocational school or junior college. Junior colleges offer 2-year programs after high school. These are often geared toward women who enter low-level white-collar jobs. Vocational schools offer 5-year training programs that begin after junior high school. These schools offer specialized training for blue-collar jobs.

Specialized school dummy: A dummy variable equal to 1 if graduated from a specialized school. These are private training institutes offering short (usuallyo2 years) training programs in such subjects as computer technology or cosmetology.

Notes: Union member dummy: A dummy variable equal to 1 if respondent was a member of a union. Voice-oriented employee association dummy: A dummy variable equal to 1 if respondent was a member of a nonunion employee association, which discusses working conditions with management. Recreation-oriented employee association dummy: A dummy variable equal to 1 if respondent was a member of a nonunion employee association, which is oriented toward recreational activities. Joint consultation system dummy: A dummy variable equal to 1 if respondent was covered by joint consultation system, which involves formal committee meetings of labor and management representatives in nonunion firm. Annual cash earnings: The log of earnings. Earnings data were collected in categories because Japanese respondents are reluctant to state their income levels. Annual earnings were collected in 16 categories. Each individual was assigned the log of the midpoint of the category. Midpoints for the top and bottom categories were assigned by fitting the empirical distribution of earnings to a log-normal distribution. Male dummy: A dummy variable equal to 1 if respondent was male. Firm size: The log of the number of employees at the respondent’s firm. Size data were collected in 10 categories and respondent was assigned the log of the midpoint of the category. Midpoint for the top category was assigned by fitting the empirical distribution of firm size to the lognormal distribution. Educational attainment (the omitted category is high school or less)

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes 171

Full-time regular worker dummy

Potential labor market experience

University dummy

0.120 (0.060) 0.134 (0.073) 0.354 (0.036) 0.017 (0.002) 0.422 (0.110)

0.035 (0.039) 0.014 (0.008)

0.031 (0.067) 0.021 (0.010)

0.058 (0.071) 0.018 (0.011)

Women

0.098 0.044 0.051 (0.059) (0.079) (0.082) 0.094 0.082 0.074 (0.074) (0.064) (0.067) 0.332 0.368 0.323 (0.039) (0.076) (0.084) 0.017 0.002 0.002 (0.002) (0.002) (0.002) 0.384 0.654 0.643 (0.108) (0.063) (0.065)

0.026 (0.039) 0.008 (0.008)

Men

1992

0.052 (0.030) 0.122 (0.046) 0.202 (0.022) 0.011 (0.001) 0.465 (0.028)

0.091 (0.021) 0.046 (0.004)

Men

0.033 (0.030) 0.091 (0.046) 0.172 (0.022) 0.012 (0.001) 0.448 (0.028)

0.089 (0.021) 0.045 (0.004)

0.011 (0.034) 0.030 (0.005)

0.007 (0.035) 0.032 (0.005)

Women

0.008 0.001 (0.037) (0.037) 0.083 0.057 (0.032) (0.032) 0.159 0.136 (0.034) (0.034) 0.004 0.003 (0.001) (0.001) 0.606 0.567 (0.032) (0.032)

2007

Private Sector Union Wage Premium (Annual Cash Earnings)a,b.

Vocational/junior college dummy

Educational attainment Specialized school dummy

Firm size (natural log)

Union member dummy

Table 4.

172 TSUYOSHI TSURU

0.003 0.002 0.015 0.016 0.002 0.002 0.017 0.017 (0.002) (0.002) (0.003) (0.003) (0.001) (0.001) (0.001) (0.001) 0.006 0.026 0.023 0.005 0.004 0.005 0.006 0.008 (0.003) (0.003) (0.010) (0.010) (0.001) (0.001) (0.003) (0.003) 0.041 0.044 0.041 0.040 0.041 0.016 0.017 0.040 (0.005) (0.005) (0.013) (0.013) (0.003) (0.003) (0.005) (0.005) 0.042 0.082 0.069 0.055 0.058 0.000 0.001 0.031 (0.016) (0.016) (0.060) (0.062) (0.009) (0.008) (0.018) (0.018) No Yes No Yes No Yes No Yes No Yes No Yes No Yes No Yes 14.348 14.576 13.202 12.949 13.990 13.921 13.432 13.371 (0.135) (0.159) (0.115) (0.325) (0.057) (0.066) (0.063) (0.081) 475 475 278 278 1525 1525 1032 1032 42.42 22.51 47.81 22.37 165.15 82.66 200.58 95.36 0.4901 0.5214 0.6502 0.6493 0.5423 0.5626 0.6804 0.6872

a

Notes: , , and  indicate statistical significance at the 1%, 5%, and 10% level, respectively. Estimated using ordinary least squares. The dependent variable is the log of annual cash earnings. Standard errors are given in parentheses. b Because the dependent variable is the natural log of annual cash earnings in the preceding year, tenure is estimated only for those with one or more years of tenure.

No. of obs. F-statistic Adj. R-squared

Occupation dummy Industry dummy Constant

Tenure^2/100

Tenure years

Overtime working hours

Straight time working hours

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes 173

174

TSUYOSHI TSURU

premium was observed either for men or for women. In contrast, in the data for 2007, again no wage premium was observed for women, but a positive wage premium of around 9 percent was observed for men.6 This result remains almost unchanged if the sample is limited to full-time regular employees, that is, excluding part-time workers (not shown in the table).7 Possible critiques to this result would be an endogeneity problem and sample selection of union status. Put simply, the endogeneity problem occurs when the union member dummy correlates to the error term. The solution is to find an instrumental variable that is highly correlated with the union dummy but that is not correlated with the error term. However, it is not easy to find a valid instrument in general, and it is difficult to find a valid instrument for union status in particular (Bound, Jaeger, & Baker 1995). Although we have not resolved the endogeneity problem, there is a good reason to believe that it is not very serious in the context of Japanese industrial relations since a majority of Japanese private sector unions have ‘‘union shop’’ clauses that require union membership as a condition of employment. Therefore, employees have virtually no choice whether they join unions or not (Tsuru & Rebitzer, 1995). This naturally suggests that workers are unlikely to self-select into union/nonunion jobs in Japan. Sample selection bias, on the other hand, is a potentially serious problem. Sample selection bias occurs when the union dummy variable is observed only for a restricted, nonrandom sample. Heckman’s selection correction model can correct the problem by using a two-step estimation: in the first stage, a probit model is used to predict the probability of union status, and in the second stage, the inverse Mill’s ratio is included in the wage equation. As is shown in the appendix (Tables A1 and A2), the values of the parameters are close to OLS estimates that treat union membership as an exogenous variable. In order to explore the underlying reasons for the differences between 1992 and 2007, I conduct graphical comparisons using counterfactual kernel density estimates based on the approaches of DiNardo et al. (1996) and Hara and Kawaguchi (2008). We utilize a counterfactual kernel density estimate in order to disaggregate key variables. In order to examine whether a particular attribute (such as union membership) accounts for differences in the distribution of wages of those with and those without that attribute in two different sectors or periods, a counterfactual kernel density estimate compares the actual wage distribution for those with the attribute of interest (e.g., union membership) at one sector or time and the counterfactual wage distribution for those without the attribute of interest after controlling for other

175

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

attributes (educational attainment, years of employment, etc.) at another sector or time. This means that, taking into account that workers’ wages are determined by the wage structure of the sector in which they work as well as their individual attributes, a counterfactual distribution is estimated in which only workers’ individual attributes are controlled for.8 The first comparison takes workers who were union members in 1992 and 2007 and compares the actual wage distribution of those who were union members and the counterfactual wage distribution of those who were not union members. The second comparison focuses on workers’ status in 2007 and compares the actual wage distribution in 2007 with the counterfactual wage distribution in 1992. Table 5 presents worker attributes, which are used for the counterfactual kernel density estimates, by year, by sex, and by union status. The mean Table 5.

The Mean Values of Worker Attributes used for Counterfactual Kernel Density Estimates. 1992 Men

2007 Women

Men

Women

Union

Nonunion

Union

Nonunion

Union

Nonunion

Union

Nonunion

Firm size (natural 7.391 log) Educational attainment Specialized school 0.068 dummy Vocational/junior 0.049 college dummy University 0.481 dummy Potential labor 2.649 market experience Full-time regular 0.995 worker dummy Straight time 39.398 working hours Overtime working 4.464 hours Tenure years 12.915 Tenure^2/100 2.590

4.703

7.629

4.554

7.878

5.361

7.861

5.121

0.078

0.056

0.112

0.088

0.129

0.102

0.172

0.040

0.278

0.149

0.028

0.044

0.286

0.233

0.412

0.222

0.093

0.619

0.534

0.286

0.243

6.904

7.814

15.734

3.742

8.829

9.236

14.526

0.960

0.889

0.354

0.981

0.794

0.720

0.385

41.723

35.992

31.100

49.451

49.189

37.958

33.313

4.231

1.224

0.806

9.597

7.851

3.874

2.607

11.238 2.164

7.016 0.801

5.189 0.465

15.200 3.340

9.444 1.674

11.200 2.030

6.367 0.735

Note: This table omits occupation dummies and industry dummies.

176

TSUYOSHI TSURU

Fig. 1. Counterfactual Kernel Density, Male Regular Workers. Notes: Broken line: Actual wage distribution of union workers. Solid line: Counterfactual wage distribution if union workers had worked as nonunion workers. The vertical axis represents the probability density function and its integral with respect to annual cash earnings (natural log) equals 1.

values differ by status. The counterfactual kernel density reported below is estimated after worker attributes indicated in Table 5 are controlled for. Beginning with Fig. 1, it can be seen that the counterfactual wage distribution of male union members in 1992 had they worked in the

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

177

nonunion group is not to the left of the actual wage distribution; that is, union members’ wages would not have been lower had they not been union members. In fact, the ‘‘bump’’ to the right suggests that for a certain group, wages would have been higher than they actually were. However, for 2007, the counterfactual wage distribution for male union members had they worked in the nonunion group lies to the left of the actual distribution. Put differently, the wage distribution of the union group is at a relatively higher position than that of the nonunion group. Fig. 2 presents the results for women. In contrast with the men’s case, the counterfactual wage distribution for female union members had they belonged to the nonunion group leans to the left of the actual wage distribution, indicating a positive union wage effect. However, the counterfactual wage distribution in 2007 for female union members had they belonged to the nonunion group is more concentrated than the actual distribution, and, compared with 1992, there is a much greater overlap. Next, as shown in Figs. 3 and 4, focusing on workers’ union membership status in 2007, the actual wage distribution for 2007, and the counterfactual wage distribution for workers with the same attributes had they worked in 1992 are examined. Beginning with the group of male unionized workers, it can be seen that the actual distribution for 2007 lies to the right of the counterfactual distribution for 1992. On the other hand, for the group of nonunion workers, the two distributions essentially overlap. For women, the actual distribution in 2007 lies to the right of the counterfactual distribution for 1992 irrespective of union membership status.9 Returning now to the significant positive wage premium for men observed in 2007, the above analysis suggests that this is the result of (a) the fact that in 2007 the wage distribution for union members was to the right of that (i.e., wages were higher than) for nonunion members and (b) while the wage distribution for nonunion workers in 2007 was almost unchanged from that in 1992, it has shifted to the right (implying higher wages) for union workers. On the other hand, for women, the wage distribution shifted toward the right (implying higher wages) irrespective of whether they were union members or not. Overall, the results for men provide concrete evidence that during the Lost Decade a positive union wage effect emerged in Japan.10

5. UNION VOICE EFFECTS The second important aspect of the Freeman–Medoff model is worker voice. In order to examine union voice effects, as a first step, the extent to

178

TSUYOSHI TSURU

Fig. 2. Counterfactual Kernel Density, Female Regular Workers. Notes: Broken line: Actual wage distribution of union workers. Solid line: Counterfactual wage distribution if union workers had worked as nonunion workers. The vertical axis represents the probability density function and its integral with respect to annual cash earnings (natural log) equals 1.

which labor unions provide voice is examined. Table 6 shows that in the data for 1992, the share of workers replying that ‘‘Even if there are problems at my workplace, I don’t tell anybody about my grievances’’ is lower among union members than nonmembers, thus indicating that labor unions do

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

179

Fig. 3. Counterfactual Kernel Density, Male Regular Workers. Notes: Broken line: Actual wage distribution in 2007. Solid line: Counterfactual wage distribution if workers in 2007 had worked in 1992. The vertical axis represents the probability density function and its integral with respect to annual cash earnings (natural log) equals 1.

offer a voice channel. This pattern, where the share of those saying they do not tell anyone about their grievances is lower among union than nonunion workers, remains basically unchanged in 2007, implying that unions continue to represent a voice channel for workers.11

180

TSUYOSHI TSURU

Fig. 4. Counterfactual Kernel Density, Female Regular Workers. Notes: Broken line: Actual wage distribution in 2007. Solid line: Counterfactual wage distribution if workers in 2007 had worked in 1992. The vertical axis represents the probability density function and its integral with respect to annual cash earnings (natural log) equals 1.

As a second step, the effect of union membership on the propensity for voluntary quitting and the degree of workplace dissatisfaction in 1992 and 2007 is examined.12 Table 7 shows the results, which indicate the following. First, in 1992, both for men and women, union membership did not have any measurable effect on either voluntary quits or the degree of job

181

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

Table 6.

Percentage of Employees Who Would Not Tell Anybody About Problems Arising at Work.

Dissatisfaction with Regard to

Men Union workers

Wage level Welfare benefits Working hours, holidays, and vacation leave Job security Performance ratings Transfer policies Harassment

Women Nonunion workers

Union workers

Nonunion workers

1992

2007

1992

2007

1992

2007

1992

2007

21.5 25.1 22.9

30.0 26.6 22.9

32.7 40.4 30.3

42.9 45.9 36.5

19.1 17.9 17.9

39.1 33.7 26.5

38.0 45.1 33.8

47.5 50.1 33.4

24.2 24.7 20.2 32.7

24.7 28.1 24.3 19.5

38.8 35.9 35.9 43.0

43.5 43.0 40.0 32.3

16.7 22.6 22.6 15.5

32.0 35.4 27.3 11.7

43.9 45.1 49.0 37.1

44.6 46.9 40.2 21.8

Note: The figures in the table show the percentages of employees responding ‘‘I don’t talk to anybody’’ to the following question: ‘‘If you have grievances at the workplace with regard to any of the following items, what would you do?’’

dissatisfaction. Second, in 2007, union membership did significantly lower the propensity for voluntary job quitting for men (and although the result is not significant at the 10 percent level, the coefficient with regard to the degree of job dissatisfaction now has the correct sign and the p-value is 0.111). However, for women still no such union effect can be observed in 2007. In sum, whereas no union voice effect was found in 1992, for men such an effect occurred in 2007. This pattern is in line with the results above, that is, that a union wage premium was observed only for men in 2007.

6. ATTITUDES TOWARD UNIONIZATION AMONG NONUNION WORKERS The preceding sections suggest that it is indeed possible to observe economic effects of labor unions. This being so, one would expect that this provides at least a partial incentive for nonunion workers to form labor unions at their workplaces. But what are nonunion workers’ attitudes to unionization in practice? As can be seen in Table 8, in 1992, the percentage of nonunion workers supporting unionization was less than half among both men (43.3 percent)

1992

426 162.699

1413 713.561

214 87.047

851 447.715

0.010 (0.047) 0.011 (0.045) 0.087 (0.016) 0.002 (0.008) Yes

2007

1525 2007.813

0.032 (0.015) Yes

0.064 (0.028) Yes 475 601.356

0.115 (0.072) 0.534 (0.088)

2007

0.183 (0.131) 0.560 (0.159)

1992

Men

0.025 (0.016) Yes

0.012 (0.100) 0.437 (0.091)

2007

278 1031 298.567 1358.218

0.031 (0.033) Yes

0.130 (0.214) 0.070 (0.189)

1992

Women

Degree of Job Dissatisfactionc

Notes:  and  indicate statistical significance at the 5% and 1% level, respectively. a The determinants of the propensity for voluntary job quitting are estimated using a probit (marginal effects) model, while the determinants of the degree of job dissatisfaction are estimated using an ordered probit model. Standard errors are in parentheses. b The propensity for voluntary job quitting is estimated using as the dependent variable a dummy that takes a value of 1 if survey respondents answered with yes to the question whether they were thinking of changing jobs within the next 1–2 years. c The degree of job dissatisfaction is measured based on respondents’ answer to the question: ‘‘Overall, how satisfied are you with your current working conditions?’’ Respondents were given the following options for their reply: 1 ¼ extremely satisfied; 2 ¼ quite satisfied; 3 ¼ neither satisfied nor dissatisfied; 4 ¼ quite dissatisfied; 5 ¼ extremely dissatisfied. d Other independent variables included in the estimations are the explanatory variables used in the estimation of the wage function in Table 4 as well as occupation and industry dummies.

No. of obs. Log likelihood

Other independent variablesd

Annual cash earnings (natural log) Degree of dissatisfaction with regard to welfare benefits Firm size (natural log)

0.093

2007

Women

0.016 0.064 (0.048) (0.031) (0.076) 0.006 0.064 0.045 (0.055) (0.040) (0.077) 0.106 0.154 0.068 (0.018) (0.012) (0.028) 0.012 0.0003 0.004 (0.010) (0.007) (0.013) Yes Yes Yes

1992

Men

Propensity for Voluntary Job Quittingb

Determinants of the Propensity for Voluntary Job Quitting and the Degree of Job Dissatisfactiona.

Union member dummy

Table 7.

182 TSUYOSHI TSURU

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

Table 8.

Attitudes Toward Unionization.

Support Unionization 1992 Men Women Total 2007 Men Women Total

183

Do Not Support Unionization

43.3 37.8 40.7

56.7 62.2 59.3

55.7 43.1 50.1

44.3 56.9 49.9

Note:  indicates that there is a statistically significant difference in the means for men and women at the 1% level.

and women (37.8 percent). However, in 2007, a majority of men (55.7 percent) supported unionization. On the other hand, among women the rate was still only 43.1 percent and thus clearly lower than for men (the difference between men and women is statistically significant at the 1 percent level). Given this background, the purpose here is to explore the sources of support for unionization among nonunion workers. This is done by regressing the ‘‘support for unionization’’ variable on factors that are likely to influence unionization support. Included as explanatory variables here are not only workers’ attributes and their dissatisfaction with working conditions, but also variables relating to voice mechanisms such as the presence of joint labor-management consultation mechanisms and employee associations. The results are presented in Table 9. The results, beginning with those for 1992, indicate that among men, dissatisfaction with wages had a positive effect on support for unionization; on the other hand, women’s support for unions strengthened as both firm size increased and dissatisfaction with welfare benefits increased. Looking now at the results of the same estimation for 2007, one prominent difference is that for men, the dummy for recreation-oriented employee associations is positive and significant. When such an organization exists, it increases support for unionization by more than 10 percentage points. On the other hand, the dummies for the presence of a joint consultation system and for a voice-oriented employee association are not significant. However, in the estimation for women, the coefficients on the recreation-oriented employee association dummy and voice-oriented employee association dummy are significantly negative. Specifically, the existence of a voiceoriented employee association decreases the support rate for unionization by

184

TSUYOSHI TSURU

Table 9. Support for Unionization Among Nonunion Workers in the Private Sector (Marginal Effect). 1992 Men

Women

2007 Men

Women

Joint consultation system 0.038 0.305 dummy (0.123) (0.150) Voice-oriented employee 0.015 0.144 0.029 0.003 0.152 0.399 association dummy (0.127) (0.196) (0.090) (0.094) (0.127) (0.093) Recreation-oriented employee 0.079 0.184 0.147 0.117 0.127 0.127 association dummy (0.089) (0.127) (0.050) (0.055) (0.070) (0.088) Degree of dissatisfaction with 0.110 0.033 0.070 0.056 0.079 0.105 regard to wages (0.039) (0.052) (0.020) (0.022) (0.023) (0.028) Degree of dissatisfaction with 0.042 0.092 0.085 0.083 0.056 0.043 regard to welfare benefits (0.042) (0.051) (0.021) (0.022) (0.026) (0.034) Degree of dissatisfaction with 0.018 0.023 0.003 0.006 0.012 0.020 regard to working hours (0.038) (0.051) (0.020) (0.021) (0.024) (0.030) Firm size (natural log) 0.027 0.051 0.048 0.053 0.051 0.068 (0.019) (0.024) (0.010) (0.012) (0.012) (0.016) Educational attainment Specialized school dummy 0.140 0.116 0.049 0.105 0.039 0.036 (0.123) (0.158) (0.069) (0.078) (0.080) (0.105) Vocational/junior college 0.070 0.065 0.028 0.071 0.080 0.077 dummy (0.178) (0.124) (0.104) (0.115) (0.070) (0.087) University dummy 0.134 0.120 0.029 0.011 0.154 0.225 (0.085) (0.215) (0.053) (0.059) (0.072) (0.085) 0.007 Potential labor market 0.005 0.001 0.001 0.002 0.007 (0.004) (0.005) (0.002) (0.003) (0.003) (0.004) experience Full-time regular worker Dropped 0.223 0.094 0.061 0.104 0.003 dummy (0.118) (0.064) (0.077) (0.066) (0.089) Straight time working hours 0.004 0.006 0.001 0.000 0.004 0.002 (0.004) (0.006) (0.002) (0.003) (0.003) (0.004) Overtime working hours 0.006 0.009 0.001 0.0001 0.017 0.014 (0.006) (0.019) (0.003) (0.003) (0.007) (0.009) Tenure years 0.007 0.024 0.002 0.0005 0.006 0.007 (0.012) (0.028) (0.007) (0.008) (0.013) (0.015) Tenure^2/100 0.029 0.042 0.017 0.015 0.036 0.045 (0.033) (0.132) (0.022) (0.023) (0.052) (0.059) Occupation dummy Yes Yes Yes Yes Yes Yes Industry dummy Yes Yes Yes Yes Yes Yes No. of obs. 255 159 721 622 483 343 Log likelihood 152.324 86.002 430.186 368.748 287.396 187.977 Adj. R-squared 0.1312 0.2004 0.1219 0.1233 0.1375 0.2090

Note: Estimated using probit estimation. The dependent variable is a dummy for supporting unionization (support ¼ 1). Standard errors are given in parentheses. , , and  indicate statistical significance at the 1%, 5%, and 10% level, respectively.

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

185

40 percentage points. These results suggest that for men, recreation-oriented employee organizations play a complementary role to unionization, while for women, both recreation- and voice-oriented employee organizations tend to serve as a substitute for unionization. The phenomenon of worker organizations influencing attitudes toward unionization was not observed in 1992 (for more direct evidence on a statistical difference between 1992 and 2007, see Table A3 that reports coefficients of interactions with 2007 dummies derived from estimates using the pooled 1992 and 2007 data).

7. DISCUSSION The main findings can be summarized as follows. First, although no union wage effects were observed in 1992, a wage premium for male workers had emerged in 2007. However, no such premium could be observed for female workers. Second, and similarly, no union voice effect was found for 1992, but in 2007, union membership reduced the propensity for job quitting (and, possibly, the degree of job dissatisfaction) among men. However, for women, no such voice effect could be confirmed. And third, perhaps as a result of these trends, the support for unionization among male nonunion workers seems to have increased. On the other hand, the increase in support for unionization among nonunion women was small when compared with men, and if there exists a voice-oriented employee association at the workplace, this largely substitutes for support for unionization among women. The following subsections provide an interpretation of the results.

7.1. Union Wage Effects To start with, the fact that a union wage effect was found for 2007 is likely to be related to the behavior of both labor unions and firms. Two different interpretations are possible. The first interpretation is to regard the union wage effect as a demonstration of the bargaining power of labor unions. Amid prolonged economic stagnation from 1992 to 2007, there was almost no increase in wages at nonunion firms and the wage distribution for men at such firms remained unchanged. Any wage increases won by labor unions in wage negotiations during the traditional ‘‘spring offensive’’ did not spill over to nonunion firms.13 As a result, wages in union firms increased relative to those in nonunion firms.

186

TSUYOSHI TSURU

The second interpretation is that the union wage premium was brought about unintentionally through the introduction of measures by firms to deal with the prolonged economic stagnation in Japan. Hara (2003), for example, using a firm survey conducted in 2003, found that although more union firms than nonunion firms had implemented restructuring measures, including personnel reductions, few of them cut wages. The second interpretation is also confirmed by the results of our analysis using microdata of the Ministry of Health, Labour and Welfare’s Survey on Wage Increases (for 1991 to 2007). As is shown in Fig. 5, when asked which measures firms considered the most pressing with regard to keeping personnel expenses in check, the percentage of firms citing ‘‘personnel retrenchment and natural attrition’’ as the most important measure was higher among union than among nonunion firms. At the same time, as reported in Table 10, the percentage of firms that implemented regular pay raises was also higher among unionized firms. Taken together, the pattern that these findings suggest is that union firms tended to keep their labor costs in check by curbing new hires, which then made it possible to maintain or even raise wage levels. On the other hand, nonunion firms, even if they 20

Union firms Nonunion firms

18 16 14

(%)

12 10 8 6 4 2 0 1991 92

93

94

95

96

97

98

99 2000 01

02

03

04

05

06

07

Fig. 5. Percentage of Firms Seeing ‘‘Personnel Retrenchment and Natural Attrition’’ as the Most Important Measures to Reduce Personnel Costs. Source: Tsuru, Yoshinaka, Enoki, and Tokuda (2009). Notes: (a) Based on an analysis of microdata of the Ministry of Health, Labour and Welfare’s Survey on Wage Increases (1991–2007); (b) based on firms with 100 or more employees. Figures were also aggregated by firm size, but the pattern was generally the same.

29.9 27.4 29.6 25.6 26.4

70.1 72.6 70.4 74.4 73.6

62.7 64.5 66.3 69.8 68.6

6.9 7.8 4.1 4.6 4.9

0.5 0.4 0.0 0.1 0.1

Postponed

32.9 36.2 33.2 28.3 26.4

67.1 63.8 66.8 71.7 73.6

48.5 54.1 53.8 61.7 65.4

18.4 9.3 11.9 9.6 8.1

0.2 0.4 1.2 0.4 0.1

No system System of Implemented/ Not Postof regular regular pay implementing implemented/ poned pay raises raises implementing

Source: Tsuru et al. (2009). Notes: (a) Based on an analysis of microdata of the Ministry of Health, Labour and Welfare’s Survey on Wage Increases (2003–2007). (b) Based on firms with 100 or more employees. Figures were also aggregated by firm size, but the pattern was generally the same.

2003 2004 2005 2006 2007

No system System of Implemented/ Not of regular regular pay implementing implemented/ pay raises raises implementing

Nonunion Firms

Regular Pay Raises, by Type of Firm (Share in %).

Union Firms

Table 10.

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes 187

188

TSUYOSHI TSURU

were already restraining labor costs through personnel adjustments, also tended to adjust wages downward.14,15 The above discussion could be applied both to male and female workers. For women, however, there have been different factors at work in terms of labor laws and the labor markets. First, a series of legal and regulatory measures was introduced to improve the situation of female workers. Followed by the Equal Employment Opportunity Law in April 1986, the Childcare Leave Law in April 1992, the ban on discrimination in the Revised Equal Employment Opportunity Law in April 1999, and the Revised Child and Family Care Leave Law in April 2002 were implemented. Thanks to those legal revisions, women’s willingness to work has reportedly been enhanced and their tenure has lengthened (see, for example, Cabinet Office (2005) White Paper on Gender Equality, Part 1, Chapter 2, Section 2). Due at least in part to these developments, wages for regular female workers relative to regular male workers have been increasing. As can be seen in Fig. 6, when focusing on typical regular workers (called ‘‘standard workers,’’ i.e., workers who directly entered a firm after graduating from school and then worked for the firm continuously, and excluding nonregular workers such as temporary workers), the wage differential between men and women appears to have narrowed during this period. Given this background, the working conditions for regular female workers in both union and nonunion firms have improved in a similar fashion, making it possible for nonunion women to ‘‘catch up’’ with union women. In contrast, the bottom of nonunion male wages has fallen out without such legal and union protections. Second, demand for female (presumably nonunion) jobs has increased in the Japan’s service industry. A conspicuous example would be rapid expansion of demand for medical and social welfare services, including long-term nursing care. Other salient examples would include the wholesale and retail trade industry and accommodations, eating and drinking services industry where more nonregular women have been working than regular female employees. On the other hand, as Kambayashi and Kato (2009) have pointed out, the job stability of regular employees eventually declined during Japan’s prolonged economic stagnation, and the burden of downsizing fell disproportionally on female regular employees in large (presumably unionized) companies. Put differently, labor demand has shifted away from regular to nonregular female workers and from union to nonunion female employees. We believe that these trends in labor laws and labor markets could contribute to the absence of a union wage premium for female employees.

189

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes 1992 85%

All firms

85%

80%

80%

75%

75%

70%

70%

65%

65%

60%

60%

55%

55%

50%

Firms with 1,000 and more employees

50% High school Vocational/junior University graduates college graduates graduates

85%

2007

Firms with 100-999 employees

High school Vocational/junior University graduates college graduates graduates 85%

80%

80%

75%

75%

70%

70%

65%

65%

60%

60%

55%

55%

50%

Firms with 10-99 employees

50% High school Vocational/junior University graduates college graduates graduates

High school Vocational/junior University graduates college graduates graduates

Fig. 6. Wage Differentials Between Men and Women by Firm Size. Notes: (a) Based on data from the ‘‘Basic Survey on Wage Structure’’ (1992 and 2007) published by the Ministry of Health, Labour and Welfare. The survey for 2007 covers approximately 78,000 establishments and about 1.61 million workers; (b) the figure shows the ratio of female standard workers’ wages to male standard workers’ wages for all ages by firm size and academic achievement; (c) the wage data were obtained by adding annual bonuses and other special cash earnings to the annualized amount of scheduled cash earnings; (d) values are for industries in total; (e) the comparison here is for ‘‘standard workers,’’ defined as those who directly entered a firm after graduation and then worked for the same firm continuously.

190

TSUYOSHI TSURU

7.2. Union Voice Effect The next issue to address is why a union voice effect was observed for men in 2007 when none was found for 1992. A likely explanation, which is similar to the one for the union wage effect, is that unions, on the one hand, accepted restructuring measures such as workforce reductions, but on the other hand, strove to protect the working conditions, such as wages, of existing workers.16 But this leaves unanswered the question why no union voice effect was observed for women. To gain some traction on this issue, it is necessary to start by looking at changes in the way workers voiced grievances at the workplace and whether they did so through individual channels (‘‘I talk to management by myself’’ or ‘‘I talk to my superior’’) or through collective voice channels (‘‘I talk to the labor union’’). Overall, both in 1992 and 2007, the most common reply given as to which route is chosen to voice grievances is ‘‘I talk to my superior,’’ followed by ‘‘I talk to the labor union.’’ Although for brevity no table is provided here, when examining the effect of union membership on the choice of individual or collective voice channels, the following pattern emerges. Compared with 1992, there was a larger number of items for which the share of both men and women replying ‘‘I talk to the labor union’’ increased rather than decreased.17 On the other hand, while there was a large number of items for which the share of male union members replying ‘‘I communicate with management through my superior’’ increased rather than decreased, the opposite was the case for female union members. However, the above results do not yet provide conclusive evidence of union voice, since it is possible that the observed greater exercise of collective voice has come at the expense of the exercise of individual voice. To take this possibility into consideration, it is useful to look at whether being a union member has the effect of opening up channels for expressing dissatisfaction.18 The results indicate that for men, union membership provided channels for expressing dissatisfaction with regard to a much larger number of items in 2007 than in 1992. On the other hand, for women, union membership opened up channels for expressing dissatisfaction for only some of the items. This suggests that although initiatives by labor unions mean that women now have a stronger collective voice than in the past to address their working conditions, these initiatives have been insufficient to provide them with as many channels to express dissatisfaction as men.

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

191

7.3. Nonunion Workers’ Support for Unionization Finally, the analysis also highlighted differences between male and female workers in terms of the support for unionization among nonunion workers. The most likely explanation for this, as shown by the estimation results obtained by Hara and Sato (2004, 2005), is that nonunion workers clearly perceive the different union wage and voice effects for men and women, and this leads to differences between men and women in their attitudes toward unionization. However, it is possible that nonunion workers are not clearly aware of the benefits of unionization. An alternative, although less rigorous, interpretation may therefore be more appropriate. For example, against the background of deregulation and weakening employment protection since the early 1990s, it could be argued that men, in their roles as main or sole providers of household income, have become more sensitive to the issue of employment protection and, as a consequence, are more inclined than women to believe that ‘‘labor unions can protect employment.’’ It seems rather difficult to explain why recreation-oriented employee associations raise the support for unionization among men, while voiceoriented employee associations lower women’s support for unionization. According to the empirical analysis in Tsuru and Morishima (1999) and Tsuru (2002), both recreation and voice-oriented employee associations express employees’ concerns with regard to welfare benefits; however, they differ in that recreation-oriented associations express workers’ concerns with regard to working hours, holidays, and vacation leave, while voiceoriented associations express workers’ concerns with regard to annual management plans and production schedules. In other words, some of the latter may be similar to labor unions in that they provide voice on issues such as working conditions and management issues. Moreover, compared with recreation-oriented employee associations, voice-oriented employee associations tend to be more strongly institutionalized from the perspectives of both managers and employees. Overall, therefore, men may tend to see voice-oriented organizations as an imperfect substitute for labor unions, while women consider them as a sufficient substitute. In this regard, recreation-oriented employee associations are weak alternatives to labor unions. And this, in turn, may lead men, who feel greater pressure to support the household, to prefer an institutionalized union. In other words, it is possible that, as a result of declining job stability due to the loosening of employment protection laws, there are moves toward turning simple, recreation-oriented employees associations into unions. In fact, this interpretation is backed up by the fact that, according to a

192

TSUYOSHI TSURU

government survey conducted at the end of June 2007, there has been an increase in the number of labor union members for the first time in 13 years, that is, since 1994, while another survey carried out at the end of June 2008 showed that the unionization rate stopped falling for the first time in 27 years, that is, since 1981, and held steady.19 These results suggest that, along with the change in the employment environment, there are also signs of changes in attitudes toward labor unions.

8. CONCLUSION What do unions do in Japan? This is the question posed at the outset of this chapter, and the analysis presented here has provided the following answers. First, although no union wage effect was observed in 1992, a wage premium for men could be observed in 2007. However, for women, no such wage premium was found. Second, with regard to union voice effects, again no such effect was evident in 1992, but in 2007, a decline among men in the propensity for voluntary quitting (and, to some extent, the degree of job dissatisfaction) was found. In contrast, among women, no such union voice effects could be detected. Third, among male nonunion workers, support for unionization increased between 1992 and 2007. On the other hand, the increase in support for unionization among female nonunion workers was quite small in contrast to that for their male counterparts. Furthermore, if there exists a voice-oriented employee association at the workplace, this largely substitutes for female support for unionization. These results have various implications for labor relations in Japan. First, the fact that a union wage effect was observed indicates that labor unions exercised their bargaining power in wage negotiations and in particular were able to win regular annual pay rises. On the other hand, the phenomenon that these wage improvements did not spread to nonunion firms suggests that shunto, the ‘‘spring offensive,’’ is no longer fulfilling its previous role of disseminating wage increases from larger to smaller firms and has become an empty shell. Second, it is notable that the positive union wage effect has been accompanied by personnel reductions at unionized companies. In other words, unions have been able to maintain wage levels, but only in exchange for reductions in employment. Third, and most critical, it appears that unions do not sufficiently understand the grievances and needs of female workers. This suggests that unions have generally done little to resolve women’s workplace grievances, and therefore failed to reduce voluntary quits. The impression that emerges

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

193

from this analysis is that unions focus their resources on protecting the interests of male union members in the firm. The data strongly suggest that unions continue to emphasize the interests of male employees. Unions have demonstrated a significant economic impact by helping to maintain wage levels and maintain employment conditions for members, particularly full-time male employees. However, at the same time, this narrow focus also limits the social significance of labor unions in Japan, leading to weaker support among the female labor force.

NOTES 1. The term ‘‘wage effect’’ here refers to the wage differential between workers belonging to a labor union and workers not belonging to a labor union (including workers employed at firms where there is no labor union). The wage differential is referred to as the ‘‘union wage premium.’’ 2. The term ‘‘voice effect’’ here refers to differences between workers belonging to a labor union and workers not belonging to a labor union (including workers employed at firms where there is no labor union) in terms of job separation rates and job dissatisfaction. Job separation rates and the degree of job dissatisfaction are assumed to be linked to whether matters improve when workers voice dissatisfaction regarding working conditions or management issues. 3. ‘‘Nonunion workers’ support for unionization’’ refers to the view among workers at firms with no union that ‘‘it would be better to establish a union at the firm.’’ 4. Union density by gender in each year of our dataset was as follows. Union density was 37.0% for male and 20.0% for female workers in 1992. In 2007, union density was 34.9% for male and 19.2% for female. 5. Using the Ministry of Health, Labour and Welfare’s Basic Survey on Labor Unions as well as the Ministry of Internal Affairs and Communications’ Labor Force Survey and Establishment and Enterprise Census to calculate unionization rates for 2007 by region, the estimate for greater Tokyo (Tokyo, Kanagawa, Chiba and Saitama) is 20.9% and that for Tokyo 24.7%. 6. A further notable difference between the 1992 and 2007 results is that the coefficient for the years of employment for women decreased by about one-third. To explore the reason for this, the estimation was repeated dividing the observations for women into full-time regular workers and part-time workers. The results indicate that although the coefficient for full-time women remained basically unchanged between the 2 years and that for part-time women declined. In other words, the decrease in the coefficient for the years of tenure for women overall is the result of the increased share of part-time workers rather than a decrease in the years of employment of full-time women. With regard to men, the coefficient for the years of employment remained almost unchanged, but the coefficient for the years of employment squared declined. This can be interpreted as indicating that the role of seniority in determining wages is dominant but diminishing over the life cycle for male employees. 7. The reason for estimating separate wage functions for men and women is that the structure of wage determination differs for men and women in the overall labor

194

TSUYOSHI TSURU

market. Of course, within many firms, including nearly all large firms, the structure of wage determination is likely to be identical for both men and women. Moreover, labor unions are likely to conduct wage negotiations without distinguishing between men and women. However, most of the workers included in this sample are employed in small and medium-sized firms, where union representation rates are low and wage structures often vary by gender and other factors. 8. For a more detailed explanation, refer to Appendix 2 of Kawaguchi and Hara (2007). 9. Figs. 3 and 4 can be interpreted as implying that the composition of workers is held fixed while the wage schedule was changed. To check these results, the analysis was repeated holding the wage schedule fixed while allowing the composition of workers to change. Similar to the results shown in Figs. 3 and 4, the only large change observed was for unionized men. What these results suggest is that the composition of workers changed more in union firms than in nonunion firms. 10. To obtain further confirmation, using microdata from the Ministry of Health, Labour and Welfare’s Survey on Wage Increases (1991 to 2007), changes in average wages were regressed on union dummy, firm size, and industry. The coefficient on the union dummy was positive and significant at the 1% level for the years 1993–1999, 2001, and 2005, indicating that a positive union wage effect can also be found when focusing on wage revisions by firms, thus confirming that union members gained wage increases from the early 1990s to the 2000s, as shown in the analysis above. 11. A notable pattern is that, with the exception of harassment, the ratio of nonunion workers not voicing dissatisfaction is higher than that of union workers. Especially with regard to wage levels and welfare benefits, almost half of nonunion workers say that they do not voice any grievances. 12. Labor unions not only provide workers with a collective voice, but, through the part they play in maintaining job security, also have a restraining effect on job separation rates. In addition, job separation rates tend to be low at firms with a labor union because such firms have generally adhered to long-term employment practices. However, it should be noted that while unions’ collective voice helps to lower quits initiated by employees themselves since it helps to resolve workers’ grievances, this differs from employment protection and long-term employment practices, which reduce separations initiated by firms (i.e., restructuring). In this paper, the union voice effect is examined by focusing on voluntary job quitting, which is used as the dependent variable, while variables related to long-term employment, such as firm size or workers’ years of employment, are used as independent variables, so that long-term employment as a factor is controlled for. 13. Cabinet Office (2007) reports that the proportion of firms that cited ‘‘going wages’’ as an important factor when revising basic wages was substantially lower in the early 2000s than in the 1990s, irrespective of whether firms were unionized or not. In addition, the time lag in determining wage rises between union firms and nonunion firms (which have tended to follow the lead of union firms) narrowed in the 1990s. Based on this evidence, it is possible to argue that wage spillovers from union to nonunion firms weakened in the 1990s and the early 2000s. 14. It is plausible that less effective unions with smaller union wage effects have been losing their members more rapidly than more effective unions with stronger

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

195

union wage effects. If this is the case, we will observe a greater union wage effect on average when the period economic stagnation comes to an end. 15. Another important factor to be considered is the aging of the population between 1992 and 2007. The average age in 1992 was 36.5 and that in 2007 was 38.6. In the wage equations, however, we incorporated as explanatory variables tenure and potential labor market experience, which closely correlate with age. This suggests that a union wage premium is can be observed after controlling for the aging of the population. 16. In terms of the proportion of those wishing to change jobs, that among male union members increased only by 2.2 percentage points from 1992 to 2007, while that among male nonunion workers increased substantially by 19.0 percentage points. For women, the corresponding figures were increases of 9.0 percentage points for union members and 8.8 percentage points for nonunion workers. 17. As for items on which job dissatisfaction might focus, the surveys provided the following seven categories: (1) wages; (2) working hours, holidays, vacation leave; (3) welfare benefits; (4) employment stability; (5) assessment and setting of salaries and bonuses; (6) staff rotation (including transfer policies); and (7) harassment. As for how workers voiced any dissatisfaction with regard to these items, the surveys provided respondents with the following options: (1) I talk to management by myself; (2) I talk to management together with my coworkers; (3) I communicate with management through my superior; (4) I communicate with management through the employee association or recreational association; (5) I communicate with management through the joint consultation system; (6) I talk to the labor union; (7) I talk to administrative bodies such as the labor administration office; (8) I talk to external organizations, such as telephone consultation services for part-time workers; and (9) I don’t talk to anybody. Looking at the share of union workers saying ‘‘I talk to the labor union,’’ this increased from 1992 to 2007 among men for five items and among women for four items. On the other hand, the share of union workers saying ‘‘I communicate with management through my superior’’ declined both among men (for three items) and women (for five items). 18. This is examined by conducting a probit analysis using a dummy variable for ‘‘I don’t talk to anyone’’ (taking a value of 1 if applicable, 0 otherwise) as the dependent variable. Doing so, the coefficient on the dummy variable for union membership was significantly negative in the estimation for 1992 for men in the case of zero items and for women in the case of two items, while in the estimation for 2007, it was significantly negative for men in the case of seven items and for women in the case of four items. 19. Ministry of Health, Labour and Welfare, Basic Survey on Labor Unions (2007, 2008).

ACKNOWLEDGMENTS This chapter is the result of joint research with Takashi Yoshinaka (Cabinet Office), Hiroyuki Enoki (Ministry of Health, Labour and Welfare), and Hidenobu Tokuda (Cabinet Office). For the earlier version in Japanese, refer to Tsuru, Yoshinaka, Enoki, and Tokuda (2009). For our analysis,

196

TSUYOSHI TSURU

we used microdata from a survey conducted for the 2007 Annual Report on the Japanese Economy and Public Finance. Masayoshi Matsumoto provided assistance in the data compilation as well as useful advice on the program writing for our estimation. In addition, we received valuable comments from seminar participants at Hitotsubashi University, Aoyama Gakuin University, and the 2009 Japanese Economic Association meeting: Hiroyuki Chuma, Toshie Ikenaga, Daiji Kawaguchi, Ryo Kambayashi, Tomohiko Noda, Hiroyuki Odagiri, Takuji Saito, and Shingo Takahashi. We would like to express our gratitude to Jun Saito and Fumihira Nishizaki at the Cabinet Office for comments received during the preparation of the discussion chapter. Further, we would like to thank Charles Weathers for his comments and editing. Finally, the usual disclaimer applies: the analysis and opinions presented in this chapter are the personal view of the author and should in no way be ascribed to the Cabinet Office or the Ministry of Health, Labour and Welfare of the Japanese government.

REFERENCES Blanchflower, D. G., & Bryson, A. (2007). What effect do unions have on wages now and would Freeman and Medoff be surprised? In: J. T. Bennett & B. E. Kaufman (Eds), What do unions do? A twenty-year perspective (Chap. 4, pp. 79–113). New Brunswick, NJ: Transaction Publishers. Bound, J., Jaeger, D. A., & Baker, R. M. (1995). Problems with instrumental variables estimation when the correlation between the instruments and the endogenous explanatory variables is weak. Journal of American Statistical Association, 90(430), 443–450. Brunello, G. (1992). The effect of unions on firm performance in Japanese manufacturing. Industrial and Labor Relations Review, 45(3), 471–487. Cabinet Office. (2007). Rodo shijo no henka to kakei bumon he no eikyo [Labor market change and its impact on the household sector]. In: Heisei-19-nendo nenji keizai zaisei hokoku [2007 annual report on the Japanese economy and public finance] (Chap. 3, pp. 171–245). Tokyo: Jiji Gaho Sha. DiNardo, J., Fortin, N., & Lemieux, T. (1996). Labor market institutions and the distribution of wages, 1973–1992: A semiparametric approach. Econometrica, 64(5), 1001–1044. Farber, H. S. (1990). The decline of unionization in the United States: What can be learned from recent experience? Journal of Labor Economics, 8(1), S75–S105. Freeman, R., & Rebick, M. E. (1989). Crumbling pillar? Declining union density in Japan. Journal of the Japanese and International Economies, 3(4), 578–605. Freeman, R. B., & Medoff, J. L. (1984). What do unions do? New York: Basic Books, Inc. Freeman, R. B., & Rogers, J. (1999). What workers want. Ithaka, NY: Cornell University Press. Hara, H. (2003). Kumiai ha nanno tame ni? Fukyo taisaku to chingin wo megutte [What are unions for? – Issues surrounding anti-recession measures and wages]. In: Research

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

197

Institute for Advancement of Living Standards (Eds), Rodo kumiai ni kan suru ishiki chosa hokokusho [Report on the survey on attitudes toward labor unions] (Part 2, Chap. 3, pp. 118–134). Tokyo: Research Institute for Advancement of Living Standard. Hara, H., & Kawaguchi, D. (2008). The union wage effect in Japan. Industrial Relations, 47(4), 569–590. Hara, H., & Sato, H. (2004). Rodo kumiai shiji ni nani ga eikyo wo ataeru ka? – Rodosha no kenri ni kan suru rikai ni chakumoku shite [ What affects labor union support? Focusing on an understanding of workers’ rights]. Nippon Rodo Kenkyu Zasshi [Japanese Journal of Labor Studies] (532), 54–70. Hara, H., & Sato, H. (2005). Kumiai shiji to kenri rikai [Union support and the understanding of rights]. In: K. Nakamura & Research Institute for Advancement of Living Standards (Eds), Suitai ka saisei ka? – Rodo kumiai kasseika he no michi [Decline or revival? – The way forward to reinvigorate the labor union] (Chap. 2, pp. 71–84). Tokyo: Keiso Shobo. Kalleberg, A. L., & Lincoln, J. R. (1988). The structure of earnings inequality in the United States and Japan. American Journal of Sociology, 8(Supplement), S121–S153. Kambayashi, R., & Kato, T. (2009). The Japanese employment system after the bubble burst: New evidence. Revised version of the paper presented at the ESRI Conference on Japan’s Bubble, Deflation and Long-term Stagnation, Federal Reserve Board San Francisco, December 11–12, 2008 (also circulated as CJEB Working Paper No. 268, March 2009). Kawaguchi, D., & Hara, H. (2007). Nihon no rodo kumiai ha yaku ni tatte iru no ka? [Are Japan’s labor unions useful?]. JILPT Discussion Paper 07-2. The Japan Institute for Labor Policy and Training, Tokyo. Muramatsu, K. (1984). Rishoku kodo to rodo kumiai: Taishutsu, hatsugen apurochi [Job separation behavior and labor unions: An analysis based on a quits and voice approach]. In: K. Koike (Ed.), Gendai no Shitsugyo [Contemporary unemployment]. Tokyo: Dobunkan. Nitta, M., & Shinozaki, T. (2008). Rodo kumiai no chingin koka no kensho [An analysis of the wage effect of labor unions]. In: I. Tanioka, M. Nitta & N. Iwai (Eds), Nihonjin no ishiki to kodo – Nihonban sogoteki shakai chosa JGSS ni yoru bunseki [The attitudes and behavior of the Japanese: An analysis based on the Japanese general social surveys (JGSS)] (Chap. 6, pp. 121–133). Tokyo: Tokyo University Press. Noda, T. (2005). Rodo kumiai no koka – Chingin to koyo chosei ni tai suru koka no kento [The effectiveness of unions: An investigation of the effects on wages and employment adjustments]. In: K. Nakamura & Research Institute for Advancement of Living Standards (Eds), Suitai ka saisei ka – Rodo kumiai kasseika he no michi [Decline or revival: The way forward to reinvigorate labor unions] (Chap. 2, pp. 71–84). Tokyo: Keiso Shobo. Noda, T., & Tachibanaki, T. (1993). Misoshiki rodosha to noryoku shugi [Nonunion workers and the merit system]. In: T. Tachibanaki & Research Institute for Advancement of Living Standards (Eds), Rodo kumiai no keizaigaku – kitai to genjitsu [The economics of labor unions: Economic expectations and reality] (Chap. 4, pp. 69–88). Tokyo: Toyo Keizai Shinposha. Tachibanaki, T., & Noda, T. (1993). Chingin, rodo joken to kumiai [Wages, working conditions and labor unions]. In: T. Tachibanaki & Research Institute for Advancement of Living Standards (Eds), Rodo kumiai no keizaigaku – kitai to genjitsu [The economics of labor unions: Economic expectations and reality] (Chap. 10, pp. 195–216). Tokyo: Toyo Keizai Shinposha.

198

TSUYOSHI TSURU

Todate, M. (2007). Rodo kumiai to rishokuritsu [Labor unions and separation rates]. Nippon Rodo Kenkyu Zasshi [Japanese Journal of Labor Studies] (568), 51–62. Tomita, Y. (1993). Rishokuritsu to rodo kumiai no hatsugen koka [Job separation rates and union voice effects]. In: T. Tachibanaki & Research Institute for Advancement of Living Standards (Eds), Rodo kumiai no keizaigaku – kitai to genjitsu [The economics of labor unions: Economic expectations and reality] (Chap. 9, pp. 173–193). Tokyo: Toyo Keizai Shinposha. Tsuru, T. (2002). Roshi kankei no nonyunion-ka – Mikuroteki, seidoteki bunseki [The nonunionization of labor relations: A micro and institutional analysis]. Tokyo: Toyo Keizai Shinposha. Tsuru, T., & Morishima, M. (1999). Nonunion employee representation in Japan. Journal of Labor Research, 20(1), 93–110. Tsuru, T., & Rebitzer, J. (1995). The limits of enterprise unionism: Prospects for continuing union decline in Japan. British Journal of Industrial Relations, 33(3), 459–492. Tsuru, T., Yoshinaka, T., Enoki, H., & Tokuda, H. (2009). Rodo kumiai no chingin hatsugen koka to misoshiki rodosha no soshikika shiji – Ushinawareta 10 nen no zengo hikaku [Union wage and voice effects and nonunion workers’ attitude toward unionization: Before and after Japan’s lost decade]. Keizai Kenkyu (Economic Review), 60(2), 140–155.

APPENDIX: SAMPLE SELECTION BIAS The earnings equations estimated in Table 4 presume that the coefficients on all variables other than the constant term are the same for union and nonunion employees. A more general specification would be to estimate separate earnings equations for union and nonunion employees: lnW ui ¼ bu X ui þ ui ,

(A.1)

lnW ni ¼ bn X ni þ ni ,

(A.2)

where subscripts u and n represent union and nonunion status for individual i, lnW is the log of annual earnings, X is the vector of exogenous variables (including the constant term), and eui, eni are mean zero, normally distributed error terms. None of the earnings equations reported so far, however, consider the possibility that union membership may itself be determined by the size of the union–nonunion wage differential. A simple alternative approach that takes union endogeneity into account would be to estimate (A.1) and (A.2) together with I i ¼ d0 þ d1 ð1nW ui  1nW ni Þ þ d2 X i þ vi ,

(A.3)

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes

199

where v is a mean zero, normally distributed error term and I i is a latent variable such that if I i 40 the individual belongs to a union. Since the dependent variable in (A.3) is determined in part by eui and eni, applying OLS to (A.1) and (A.2) produces inconsistent estimates of bn and bu. Consistent estimates of bn and bu when union membership is endogenous can be obtained using a two-step procedure. First, a probit equation is estimated in which the probability of being a union member is a function of the vector of exogenous variables, X (including a constant term). The parameters of this reduced-form probit equation (g) are used to calculate predicted values of the latent variable, ci, where ci ¼ gXi. The second step is to use values of ci to estimate the following equations: fðci Þ þ Zu , Fðci Þ

(A.4)

fðci Þ þ Zn , 1  Fðci Þ

(A.5)

ln W ui ¼ bu X ui þ su  ln W ni ¼ bn X ni þ sn 

where f is the density function and F the distribution function of the standard normal; parameter s is the covariance between the error of the reduced-form probit equation (e) and the error term from the union (eu) or nonunion (en) wage equations; and Zu and Zn are normally distributed, mean zero error terms. Applying ordinary least squares to (A.4) and (A.5) produces consistent estimates of b and s. The results for men are presented in Table A1. The values of the parameters are close to OLS estimates that treat union membership as an exogenous variable. The estimate s is also statistically insignificant. Table A2 repeats the analysis of Table A1 for female workers. The results parallel those for men. The restrictions implicit in the earnings equations presented in Table 5 above cannot be rejected at conventional significance levels. Ordinary least squares estimates of (A.4) and (A.5) are quite close to OLS estimates that treat union status as exogenous, although, for the case of female union status in 2007, estimate s is statistically significant. In sum, the above estimates using a two-step procedure generally support the assumption of Table 4, implying that the union member dummy would be exogenous.

0.002 (0.010)

0.032 (0.016)

Straight time working hours

0.003 (0.004)

0.109 (0.073)

Union 0.029 (0.025)

Nonunion

Endogenous

0.050 (0.007)

Union

0.053 0.020 (0.040) (0.048) 0.122 0.095 (0.060) (0.058) 0.177 0.203 (0.031) (0.033) 0.011 0.016 (0.001) (0.003) 0.425 0.289 (0.034) (0.263) 0.001 0.004 (0.001) (0.002)

0.043 (0.006)

Union

0.054 (0.039) 0.147 (0.061) 0.186 (0.033) 0.012 (0.002) 0.384 (0.054) 0.004 (0.002)

0.026 (0.015)

Nonunion

Endogenous

0.006 (0.039)

2007

Nonunion

Exogenous

0.099 0.048 0.103 0.011 (0.073) (0.110) (0.089) (0.043) 0.020 0.172 0.046 0.185 (0.098) (0.129) (0.074) (0.068) 0.338 0.314 0.337 0.183 (0.050) (0.063) (0.049) (0.031) 0.018 0.004 0.020 0.014 (0.002) (0.011) (0.003) (0.002) 0.300 0.481 0.389 (0.111) (0.116) (0.085) 0.002 0.005 0.002 0.001 (0.002) (0.004) (0.002) (0.002)

Nonunion

Union

Exogenous

1992

Earnings Equations for Private Sector Men (Annual Cash Earnings)a,b.

0.072 (0.108) Vocational/junior college dummy 0.042 (0.119) University dummy 0.318 (0.068) Potential labor market experience 0.012 (0.005) Full-time regular worker dummy

Educational attainment Specialized school dummy

Firm size (natural log)

Table A1.

200 TSUYOSHI TSURU

185 9.17 0.4825

290 15.65 0.5383

185 9.18 0.5453

290 17.27 0.5784

578 25.19 0.4909

947 39.05 0.4805

578 24.29 0.5128

814 35.69 0.4947

0.010 0.006 0.004 0.008 0.005 0.004 0.004 0.003 (0.005) (0.003) (0.009) (0.003) (0.002) (0.002) (0.003) (0.002) 0.036 0.043 0.038 0.043 0.042 0.036 0.039 0.049 (0.009) (0.007) (0.010) (0.007) (0.004) (0.004) (0.008) (0.005) 0.057 0.031 0.060 0.065 0.045 0.059 0.064 0.025 (0.025) (0.022) (0.026) (0.021) (0.010) (0.013) (0.018) (0.013) 14.855 14.624 14.235 14.730 14.119 13.855 14.725 13.890 (0.241) (0.195) (0.646) (0.198) (0.138) (0.085) (0.616) (0.114) 0.347 0.204 (0.326) (0.191) 0.199 0.129 (0.132) (0.091)

a

Notes: , , and  indicate statistical significance at the 1%, 5%, and 10% level, respectively. Estimated using ordinary least squares. The dependent variable is the log of annual cash earnings. Standard errors are given in parentheses. b Because the dependent variable is the natural log of annual cash earnings in the preceding year, tenure is estimated only for those with one or more years of tenure.

No. of obs. F-statistic Adj. R-squared

j(ci)/(1F(ci))

j(ci)/F(ci)

Constant

Tenure^2/100

Tenure years

Overtime working hours

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes 201

0.034 (0.024)

Union 0.062 (0.071)

Union

0.042 (0.078) 0.069 (0.086) 0.432 (0.126) 0.002 (0.003) 0.672 (0.088) 0.017 (0.003)

0.030 (0.015)

Nonunion

Endogenous

0.030 0.340 (0.091) (0.165) 0.059 0.071 (0.083) (0.094) 0.411 0.085 (0.119) (0.116) 0.002 0.022 (0.003) (0.006) 0.633 0.447 (0.072) (0.528) 0.017 0.006 (0.003) (0.015)

0.021 (0.012)

Nonunion

Exogenous

1992

0.078 (0.086) 0.131 (0.065) 0.316 (0.072) 0.005 (0.003) 0.507 (0.079) 0.020 (0.003)

0.025 (0.015)

Union

0.008 (0.041) 0.051 (0.038) 0.094 (0.039) 0.003 (0.001) 0.572 (0.037) 0.017 (0.001)

0.033 (0.006)

Nonunion

Exogenous

0.202 (0.069)

Union

0.009 (0.042) 0.051 (0.036) 0.094 (0.040) 0.003 (0.001) 0.571 (0.048) 0.017 (0.002)

0.033 (0.011)

Nonunion

Endogenous

0.210 (0.105) 0.164 (0.067) 0.321 (0.073) 0.002 (0.003) 0.977 (0.202) 0.020 (0.004)

2007

Earnings Equations for Private Sector Women (Annual Cash Earnings)a,b.

0.430 (0.194) Vocational/junior college dummy 0.078 (0.112) University dummy 0.190 (0.110) Potential labor market experience 0.016 (0.006) Full-time regular worker dummy 0.885 (0.175) Straight time working hours 0.006 (0.010)

Educational attainment Specialized school dummy

Firm size (natural log)

Table A2.

202 TSUYOSHI TSURU

0.017 (0.012) 0.039 (0.014) 0.096 (0.057) 12.805 (0.318)

218 26.74 0.7317

0.009 (0.006) 0.036 (0.009) 0.040 (0.027) 13.372 (0.214)

814 64.35 0.6418

0.006 (0.004) 0.016 (0.007) 0.011 (0.028) 13.372 (0.091)

218 65.51 0.7692

0.010 (0.007) 0.060 (0.013) 0.056 (0.029) 10.521 (1.147) 0.799 (0.316)

814 76.90 0.652

0.005 (0.117)

0.006 (0.004) 0.016 (0.007) 0.012 (0.029) 13.374 (0.091)

Note: , , and  indicate statistical significance at the 1%, 5%, and 10% level, respectively. a Estimated using ordinary least squares. The dependent variable is the log of annual cash earnings. Standard errors are given in parentheses. b Because the dependent variable is the natural log of annual cash earnings in the preceding year, tenure is estimated only for those with one or more years of tenure.

214 19.90 0.6402

64 27.67 0.7551

0.033 (0.016) 0.018 (0.030) 0.024 (0.080) 14.650 (1.158) 0.475 (0.347)

No. of obs. F-statistic Adj. R-squared

214 14.62 0.5952

0.018 (0.013) 0.037 (0.016) 0.094 (0.080) 12.890 (0.359)

0.146 (0.140)

64 6.12 0.6193

0.025 (0.017) 0.050 (0.020) 0.045 (0.091) 13.322 (0.436)

j(ci)/(1F(ci))

j(ci)/F(ci)

Constant

Tenure^2/100

Tenure years

Overtime working hours

The Union Wage Premium, Voice, and Nonunion Workers’ Attitudes 203

204

Table A3.

TSUYOSHI TSURU

Support for Unionization Among Nonunion Workers in the Private Sector (Marginal Effect). Pooled 1992 and 2007 data Men

Women

Interactions with 2007 dummy Voice-oriented employee association dummy Recreation-oriented employee association dummy Degree of dissatisfaction with regard to wages Degree of dissatisfaction with regard to welfare benefits Degree of dissatisfaction with regard to working hours Firm size (natural log) Educational attainment Specialized school dummy Vocational/junior college dummy University dummy Potential labor market experience Full-time regular worker dummy Straight time working hours Overtime working hours Tenure years Tenure^2/100 Occupation dummy Industry dummy Time dummy (2007 ¼ 1) No. of obs. Log likelihood Adj. R-squared

Interactions with 2007 dummy

0.015 (0.129) 0.081 (0.093) 0.111 (0.040) 0.042 (0.043) 0.018 (0.038) 0.027 (0.019)

0.047 (0.156) 0.225 (0.093) 0.039 (0.045) 0.046 (0.048) 0.014 (0.043) 0.022 (0.022)

0.151 (0.211) 0.186 (0.126) 0.034 (0.053) 0.095 (0.052) 0.024 (0.053) 0.052 (0.025)

0.032 (0.285) 0.284 (0.104) 0.040 (0.058) 0.040 (0.058) 0.006 (0.058) 0.005 (0.028)

0.137 (0.116) 0.072 (0.187) 0.134 (0.085) 0.005 (0.004) 0.773 (0.036) 0.004 (0.004) 0.006 (0.006) 0.007 (0.012) 0.029 (0.033) Yes Yes Yes

0.174 (0.136) 0.025 (0.212) 0.121 (0.101) 0.004 (0.005) 0.976 (0.223) 0.005 (0.005) 0.006 (0.007) 0.009 (0.014) 0.046 (0.040) Yes Yes Yes

0.122 (0.170) 0.067 (0.129) 0.121 (0.214) 0.001 (0.005) 0.228 (0.120) 0.006 (0.006) 0.009 (0.020) 0.025 (0.029) 0.044 (0.136) Yes Yes Yes

0.177 (0.198) 0.138 (0.150) 0.033 (0.225) 0.007 (0.006) 0.126 (0.137) 0.002 (0.006) 0.024 (0.021) 0.017 (0.032) 0.006 (0.145) Yes Yes Yes

981 585.730 0.1337

644 377.954 0.1460

Note: Estimated using probit estimation. The dependent variable is a dummy for supporting unionization (support ¼ 1). Standard errors are given in parentheses. , , and  indicate statistical significance at the 1%, 5%, and 10% level, respectively. This table shows estimates used to determine whether changes from 1992 to 2007 in variables used in Table 9 are statistically significant. Due to the lack of a ‘‘joint consultation system dummy’’ variable in the 1992 data, however, it is not possible to compare changes in joint consultation systems.

WORKER AND COMMUNITY COOPERATIVES: A MULTI-CRITERION MODEL Roger A. McCain Criticism of corporations and of corporate capitalism leads naturally to a discussion of the possible alternatives to capitalist corporate organizations, and thus to some resurgence of interest in worker cooperatives. However, the critics may hesitate to endorse cooperation without qualification, since cooperatives are interest organizations and the parochial interests of a group of workers may conflict with the ‘‘public interest,’’ where, for example, externalities are concerned. Thus, for example, Magnuson (2008) writes ‘‘In a mindful economy, the purpose of a cooperative must be clearly stated and defined in its corporate charter. As a true alternative, its most important purpose is the serve the needs of the people y An exclusively producer cooperative would create an unbalanced market in which producers maintain ownership, control, and a position of advantage y Balancing a plurality of interests is essential for functional democracy’’ (pp. 361–362). The first two sentences of this passage suggest instead a nonprofit corporation, that is, a mission-driven enterprise (McCain, 2006a, 2010); but mission-driven enterprises are not permitted to distribute their net revenues. The second passage brings to mind the solidarity cooperatives developed in recent years in Quebec.1 Magnusson uses the term ‘‘community cooperatives.’’ ‘‘Balancing a plurality of interests’’ seems unavoidable where workerowned enterprises are concerned, even when we view such an enterprise in

Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 205–224 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011012

205

206

ROGER A. MCCAIN

the more traditional way as an organization to advance the mutual interest of a group of workers. Ward’s (1958) original model identified the interests of a worker cooperative with a single dimension – increasing the wage. It is plausible that even the simplest reasonable concept of a worker cooperative would recognize other dimensions of worker interest, such as working conditions, effort, pension benefits, and security from layoffs. Yet there was nothing inevitable about Ward’s representation even 50 years ago. By that time, Scitovszky’s (1943) utility-maximizing model of the proprietary enterprise was well known, and the Kuhn-Tucker (1952) analysis of multicriterion optimization was known. By the 1970s, these methods were widely applied in economics and management (Intriligator, 1971). Yet, to the best of my knowledge, no model of worker-owned enterprises in terms of multicriterion decision making has ever been proposed.2 Multi-criterion decision making also seems to be the appropriate tool for a formal model of Magnusson’s ‘‘community cooperatives.’’ The first two sections of this chapter first develop a multi-criterion model of a pure worker-owned firm (i.e. all of the criteria for a decision reflect the parochial interests of the workers). The model is distinguished from existing models in that (1) it models layoffs as distinct from dismissals and assumes that the cooperative’s decision-process is averse to layoffs of members and (2) benefits to worker-members include retirement pension funds as well as wages, and the trade-offs among wage levels, investment, and retirement benefits is explicitly allowed for. The third section modifies this model by assuming open membership, a condition expected of cooperatives in general and worker cooperatives in particular (International Cooperative Alliance, 1995). The fourth section of the chapter then extends the model to represent a hybrid ‘‘community cooperative’’ with decisions that reflect both the interests of the workers and those of a broader community with specific reference to a polluting effluent and to the membership of the firm. A comparison of the models tends to support Magnusson’s position. The last section of the chapter, however, considers in ordinary language terms the realism of an ideal economic system that relies strongly on missiondriven enterprises to guide economic activity toward the public good.

1. A WORKER-OWNED FIRM WITH INVESTMENT, LAYOFF, AND POLLUTION DECISIONS: VARIABLES AND CONSTRAINTS We envision an enterprise owned by an association of its employees. The enterprise may employ nonmember as well as member labor. All are paid the

Worker and Community Cooperatives: A Multi-Criterion Model

207

same wage, but member-employees also accrue credit toward a pension as a benefit of their employment. Nonmember employees do not receive this benefit. Nonmember employees may be allowed to ‘‘buy in’’ as members for a share value that does not dilute the retirement fund, but only at the discretion of the existing members. As a consequence, the workers’ association is not strictly a cooperative, since membership is not ‘‘open.’’ One of the purposes of this inquiry, however, will be to determine under what circumstances the workers’ association will recruit new members. One objective of this chapter is to address the occurrence of layoffs. In most models in the ‘‘theory of the firm,’’ no distinction can be made between layoffs and dismissals: any separation of an employee from the firm is just that, and nothing else. In order to make the distinction, we suppose that decisions are made in two stages. At the first stage the ‘‘state of the world’’ is unknown, although probabilities can be assigned to different possible ‘‘states of the world.’’ In particular, the price at which output will be sold varies according to the ‘‘state of the world.’’ An employment decision is made at the first stage, on the basis of the expected values of the output price and other second-stage variables. At the second stage, once the state of the world is known, some further decisions can be made, including layoffs in cases in which the ‘‘state of the world’’ is unfavorable for a large labor force. In addition, investment decisions are made at the first stage. At the beginning of the period, the cooperative has a cash reserve fund derived from production and sale of output in previous periods. Some part of that fund can be spent on replacement or maintenance of capital goods. The balance will flow into the member-workers’ retirement fund. This decision is again based on expected values of the state-of-the-world variables. After the state of the world is known, output and the wage are determined, with both member-workers and nonmember-workers receiving the same wage. Any net revenue from the sale of output flows into the reserve fund, and so is available for investment or retirement benefits in the next period. The association also may choose techniques of production that create a negative externality, such as a polluting effluent, and the decision as to how much effluent to create is made at the second stage. Thus, the first stage decisions correspond roughly to the Marshallean long run, whereas second-stage decisions correspond roughly to the Marshallean short run. Accordingly, the decision to admit new members is also made at the first stage. This chapter makes use of the methods of formal maximization. It may be argued, and is probably correct, that real human decision-makers are characterized by bounded rationality, so that they do not literally make maximizing decisions. However, adaptive learning can result in convergence of boundedly rational decisions to the maxima. In this sense propositions of

208

ROGER A. MCCAIN

maximization theory should be considered as hypotheses subject to further consideration in a dynamic theory as well as empirical confirmation. Further, the hypothesis may have differing force for different kinds of organizations. These refinements must be beyond the scope of the chapter. Decisions at both stages are made by a representative body and are made in such a way as to maximize the value of a multi-objective criterion function subject to various constraints. The dimensions of the criterion function are dimensions of the work and productive process which affect the interests of the member employees, but the criterion function is not simply an aggregate of the utility functions of the members (supposing that such things exist). Rather, it reflects a more or less political balancing of those interests on the part of the board. Among the objectives to be decided are wage rates, contributions to the retirement fund, effort norms for the work process, and layoffs of members. A list of variables for the model follows. Variables subscripted by i are chosen at the second stage of the decision-process and so conditioned on the state of the world, whereas other variables are decided at the first stage. i denotes the state of the world Wi denotes the wage in state i, assumed to be the same for member and nonmember labor Ei denotes the effort norm, again the same for member and nonmember labor Gi denotes layoffs of members Hi denotes layoffs of nonmembers Li denotes the quantity of labor deployed, net of any layoffs Fi denotes the flow of the negative externality Ri denotes the remaining reserve (retirement) fund at the end of the period Qi denotes the output of the firm N denotes the number of members M denotes the number of nonmember employees recruited at the first stage K denotes the physical capital of the firm R0 denotes the initial level of the reserve fund N0 denotes the initial membership of the cooperative pi denotes the price realized in state i, assumed to be parametric qi denotes the probability of state i V i ¼ ð1  rÞðRi =NÞ, the discounted expected fund for a retirement pension per member.

Worker and Community Cooperatives: A Multi-Criterion Model

209

Among the constraints assumed, we have Li ¼ N  Gi þ M  H i

(1)

Note that this equality constraint is, in effect, equivalent to two inequality constraints. Essentially, what is assumed here is that labor can only be recruited in the first stage, before the state of the world is known. This is appropriate for a model with layoffs: otherwise, there is no meaningful difference between layoffs and discharges – a problem with the classic literature. Second, members are not expelled, so that initial membership net of retirements is the lower limit on membership for the second period. N0  N

(2)

The reserve fund is depleted by replacement of the capital stock, but augmented by the quasirent earned in the particular state of the world. The retirement fund is supposed to be of portable, defined-benefit type, so that employees who separate from the enterprise retain their share. For simplicity it is assumed that all physical capital is used up in every period. Ri  R0  K þ pi Qi  wi Li

(3)

Output is determined by a conventional production function augmented with the values of the negative externality (which is assumed to be correlated with higher levels of production), an effort norm, and is variable with the state of the world. Qi  f ðLi ; E i ; F i ; K; iÞ

(4)

The firm’s decisions are supposed to maximize a multi-criterion objective function with the wage and the discounted present value of the reserve (retirement) fund as positive criteria and effort norms and layoffs of members as negative criteria. U i ¼ gðW i ; E i ; Gi ; V i Þ

(5)

The partial derivatives with respect to Wi and Vi are positive. The partial derivative with respect to Ei is globally negative. The partial derivative with respect to Gi is negative for positive Gi and is zero for nonpositive Gi. It is assumed that the wage is the same for member and nonmember labor and that nonmember labor is available at that wage without any important supply constraint. Hi  M

(6)

210

ROGER A. MCCAIN

Layoffs of nonmember employees are constrained by the number of such employees. The same is true of layoffs of member labor, but this constraint will not be significant for the decision process as it corresponds to shutdown of the enterprise in some states, a contingency we will not consider.

2. A WORKER-OWNED FIRM WITH INVESTMENT, LAYOFF, AND POLLUTION DECISIONS: VARIABLES AND CONSTRAINTS: ANALYSIS We maximize the expected value of Eq. (5) subject to Eqs. (1)–(4) and (6). The Lagrangian function is   X n n X   Ki þ qi g W i ; E i ; Gi ; ð1  rÞ li f ðLi ; E i ; F i ; K; iÞ  Qi L¼ N i¼1 i¼1 þ

n n X   X mi R0  K þ pi Qi  wi Li  Ri þ ui ð N  G i þ M  H i  L i Þ i¼1

þ

n X

i¼1

fi ðM  H i Þ þ yðN  N 0 Þ

ð7Þ

i¼1

Among the necessary conditions are @L @g ¼ qi  mi Li  0 @W i @W i

(8)

@L 1  r @g  mi  0 ¼ qi @Ri N @V

(9)

Supposing that Wi and Ri will be positive, these inequalities will be satisfied as equalities and @g Li @g ¼ ð1  rÞ N @V @W i

(10)

The motive to increase the wage without limit is balanced by the resulting depletion of the discounted present value of the retirement fund, and if nonmember labor is used in the particular state, this partly offsets the discounting and the wage is decreased and reserve fund increased as a result. @L ¼ li þ mi pi  0 @Qi

(11)

Worker and Community Cooperatives: A Multi-Criterion Model

@L @f ¼ li  mi  ui  0 @Li @Li

211

(12)

Here again we may assume that Qi and Li are positive for a meaningful case, so that these will be equalities. Together these yield pi

@f ui ¼ Wi þ mi @Li

(13)

As we will see, ui will be zero so long as Hi is positive and less than M. Thus, so long as nonmember labor is available at the stipulated wage, the labor force will be adjusted so that the value of the marginal product is equal to the wage. Superficially, this resembles the case for a profit-maximizing firm, but it is important that the wage is differently determined. @L @g @f ¼ qi  li 0 @E i @E i @E i With Eqs. (8) and (11), this yields   @f ð@g=@E i Þ @g pi 40 since ¼ Li o0 @E i ð@g=@W i Þ @E i

(14)

(15)

That is, effort is set so that the value marginal product of effort is equal to the money value of a reduction in effort, where this money value is the money wage increase necessary to offset a unit increase in effort, times the number of employees who experience the increase in effort. @L @g ¼ qi  ui  0 @Gi @Gi

(16)

Thus u i  qi

@g @Gi

(17)

In case Gi is positive (members are laid off), this will be an equality. Note that ð@g=@Gi Þo0 in that case. Since ui is the multiplier for an equality constraint, it may be negative. In this case pi ð@f =@Li Þ  W i ð@g=@Gi Þ o0 ¼ Li ð@g=@W i Þ

(18)

That is, layoffs of members are adjusted so that the loss of income due to maintaining members at work is balanced against the money value of the

212

ROGER A. MCCAIN

negative weight on member layoffs in the criterion function. @L ¼ ui  fi  0 @H i

(19)

@g @Gi

(20)

Therefore fi  ui  qi

If HioM, by the complementary slackness conditions fi ¼ 0, therefore ui ¼ ð@g=@Gi Þ ¼ 0 which can only be true at Gi ¼ 0. That is, member labor will be laid off only after all nonmember labor has been laid off. Moreover, it follows also that, in states with HioM, the inequalities in lines (16–20) are equalities. @L @f ¼ li 0 @F i @F i

(21)

That is, the externality-producing input will be used so far as it is profitable to do so. The cooperative will generate negative externalities just as would a profit-maximizing firm. n n X @L X @f li m 0 ¼  @K @K i¼1 i i¼1

(22)

Once again, for a meaningful case we may assume that K is positive so this is an equality. Using Eqs. (8) and (9), this yields n X i¼1

qi

n X N @g @f @g pi ¼ ð1  rÞ qi Li @W i @K i @V i¼1

(23)

That is, investment is adjusted in such a way as to balance the discounted expected marginal benefit of the retirement reserve against the expected weighted value of the marginal product of capital. The value of the marginal product of capital is weighted by (1) the marginal benefit of increasing the wage and (2) the proportion of the labor force that receives that wage who are members. n @L X ui þ y  0 ¼ @N i¼1

(24)

Worker and Community Cooperatives: A Multi-Criterion Model

213

Since NZN0Z0, this must be an equality. Therefore y¼

n X

ui ¼ 

i¼1

n X

qi

i¼1

ð@g=@Gi Þ Li

(25)

The summation is the absolute value of the expected sacrifice of the criterion function due to layoffs of worker members, that is to say, the expected ‘‘marginal cost’’ of member layoffs. Notice however that by the complementary slackness conditions, yðN  N 0 Þ ¼ 0. Therefore, yW0.N ¼ N0. That is, new members will be recruited only when there is no possibility that members will be laid off in any state of the world, however unfavorable. n n X X @L ¼ ui þ fi  0 @M i¼1 i¼1

(26)

For MW0, this is an equality. Thus, if any nonmember employees are recruited n X i¼1

fi ¼ 

n X

ui

(27)

i¼1

  Returning to Eq. (12), ui ¼ pi ð@f =@Li Þ  W i mi . Thus, from Eq. (25), we have    n X ð@g=@W i Þ @f (28) y¼ qi pi  Wi Li @Li i¼1 The right-hand side of Eq. (28) is the expected value of losses due to retaining a labor force with pi ð@f =@Li ÞoW i , weighted by the value of the wages lost in the criterion function, per employee, in each state. Now, pi ð@f =@Li Þ  W i might be negative if further reductions of the labor force were restrained by aversion to layoffs, or it might be positive if further increases in the labor force were impossible because Li is constrained by MþN. However, we will see that the latter case does not occur. From condition Eq. (19), fi ¼ ui in every state in which Hi is positive. In the remaining states, if pi ð@f =@Li Þ  W i is positive, we have f  ui and since ui is positive, fP iZ0, therefore indeed fi 4  ui . If there are any such states, P n n f 4  i¼1 i i¼1 ui , which violates condition Eq. (27). It follows that there cannot be any states in which the labor force is limited because LirMþN; rather, M will be expanded until it is great enough to assure that pi ð@f =@Li Þ  W i in every possible state. It follows, however, that the value of ui will be nonnegative and the expected value in Eq. (25) will be positive so long as there is any state bad enough that pi ð@f =@Li ÞoW i with Li ¼ N, however small the probability of

214

ROGER A. MCCAIN

the state. Supposing that there will always be such a state, perhaps with very small probability, we may conclude that the worker-owned firm will never recruit new members. Thus, worker ownership per se is selfextinguishing.

3. A WORKER COOPERATIVE A worker cooperative will be a worker-owned firm in the sense used in the previous two sections, with one exception. For the cooperative, membership is ‘‘open,’’ which we take to mean that nonmember employees have an opportunity to buy in as members at an equitable charge. (This is consistent with the principles of the International Cooperative Alliance (1995).) As before, the association begins its decisions with a certain number of members, N0, who are holdovers from earlier periods. Members are not expelled. The association can then actively recruit new members or can seek to hire nonmember employees to supplement the member labor force. Since the membership is open, however, some of those recruited as nonmember employees may choose to become members. As an ‘‘equitable’’ charge to enter membership, we might suppose that the charge should be sufficient that the members’ retirement fund would not be diluted, that is that the charge might be R0/N0. Presumably not all of the newly hired nonmember employees will choose to become members. Some may elect not to do so because they have alternative investment opportunities that are more attractive. Realistically, many may choose not to join because they do not have sufficient capital to buy in and cannot borrow it at an opportunity cost interest rate. They are ‘‘liquidity constrained.’’ As a simple approximation, we might suppose that the proportion of new recruits who choose to be members would decline as R0/N0 increases according to some functional relationship v ¼ h(R0/N0). Since R0/N0 is a given constant for the period, however, we may treat v as a parameter, so that N 0 þ vJ  N

(29)

where J is the number recruited at stage 1. Constraint Eq. (29) replaces Eq. (2), and we add the constraint M ¼ ð1  vÞJ

(30)

Worker and Community Cooperatives: A Multi-Criterion Model

215

The Lagrangian function will then be   X n n X   Ki þ qi g W i ; E i ; Gi ; ð1  rÞ li f ðLi ; E i ; F i ; K; iÞ  Qi L¼ N i¼1 i¼1 þ

n n X   X mi R0  K þ pi Qi  wi Li  Ri þ ui ð N  G i þ M  H i  L i Þ i¼1

þ

n X

i¼1

fi ðM  H i Þ þ yðN  vJ  N 0 Þ þ cðM  ð1  vÞJ Þ

ð31Þ

i¼1

Eqs. (8) through (25) will again be applicable, and the conclusions from them are applicable also to the worker cooperative. In place of Eq. (26) we have n n X X @L ¼ ni þ fi þ c  0 @M i¼1 i¼1

(32)

from which c¼

n  X

u i þ fi



(33)

i¼1

There is one more necessary condition for this maximization. It is @L ¼ vy  cð1  vÞ  0 @J

(34)

and this will be an equality if JW0. From Eqs. (13), (25), (32) and (34) we have n X i¼1

ui ¼ ð1  vÞ

n X

fi

(35)

i¼1

We may now distinguish four families of states of the world. (a) States of no layoffs, with the quantity of labor actively constrained by first-stage hiring. In these states ui ¼ qi ð pi ð@f =@Li Þ  W i Þ40 may be observed; however, fi ¼ 0. (b) States involving some nonmember layoffs. Then ui ¼ fi ¼ 0. (c) States in which all nonmembers but no members are laid off. Then fi ¼ ui  qi ð@g=@Gi Þ ¼ 0. But moreover as fi  0 by the Kuhn-Tucker necessary conditions, we have fi ¼ ui ¼ 0.

216

ROGER A. MCCAIN

(d) States in which some members are laid off. Then qi ðpi ð@f =@Li Þ  W i Þ ¼ ð@g=@Gi Þo0. That is, some members are retained despite the excess of the wage over the value marginal product of labor, to the extent that these losses are balanced against the marginal aversion to layoffs of members. In these states fi ¼ ui 40 may be observed. We see that the summations in Eq. (35) sum terms that differ from zero only in states of types a and d, in the case of ui and of type d, in the case of fi. Therefore  n X X ð@g=@W i Þ  @f (36) fi ¼  qi pi  Wi Li @Li i¼1 d n X i¼1

 X ð@g=@W i Þ  @f ui ¼  qi pi  Wi Li @Li a ! X ð@g=@W i Þ  @f þ qi pi  Wi Li @Li d

ð37Þ

P P where a denotes summation over states of type a and d denotes summation over states of type d. Combining Eqs. (35)–(37),   X ð@g=@WÞ  @f X ð@g=@W i Þ  @f ¼ v qi pi  Wi qi pi  Wi Li @Li Li @Li a d (38) The right-hand side is the expected losses from retaining members when the value marginal product of labor is less than the wage, and the left-hand side is the expected loss from the constraint of the labor supply to first-stage recruitment, weighted by the proportion of new recruits who become members. This balance determines the rate of recruitment. As recruitment, J, increases, both member and nonmember labor forces increase, but the proportion of member labor declines from 1 (at J ¼ 0) toward vo1 in the limit as J approaches infinity. Suppose that the technology is characterized by constant returns to scale, and that the capital-labor ratio remains constant as the work force increases with increased J. Then the proportion of nonmember workers will nevertheless shift the profitability of production in a given state. As the proportion of nonmember employees increases, states of type a are fewer and those of type d more common, reducing the LHS and increasing the RHS. It might then be that, even with the proportion of nonmember employees at v, the left-hand side might remain greater than the

Worker and Community Cooperatives: A Multi-Criterion Model

217

right-hand side. In that case the optimal number recruited would be infinite. (This is a familiar paradox that when returns to scale are constant and the price parametric the optimal scale of the firm may be infinite.) In such a case, more realistically, pi ð@f =@Li Þ may decline because (a) of decreasing returns to scale, (b) the capital-labor ratio may decline because liquidity constraint prevents the cooperative from raising an unlimited quantity of capital, or (c) the firm enters a range of monopoly power and the price declines. Any of these will reduce the LHS and increase the RHS until the balance is struck. The efficiency implications of these conditions are unclear,3 apart from Eqs. (15) and (21). Condition Eq. (15) can be interpreted as characterizing an efficient effort norm. Condition Eq. (21) clearly implies an inefficient level of effluent. As for Eq. (23), there is a suggestion that the marginal productivity of capital is underweighted to the extent that N/Li o1 on the left-hand side, whereas a more global opportunity cost of capital on the right-hand side would define efficiency. But we have not allowed for the worker-owned or cooperative firm to raise outside capital or to invest the retirement fund in outside capital markets. In the ideal world of perfect capital markets, those opportunities would be sufficient to assure efficiency. In the real world of liquidity constraint, there are no consensus benchmarks to apply. At the least, Eq. (23) tells us that a worker owned or workercooperative firm as conceived here would consider the marginal productivity of capital and would have a motive to improve the allocation of capital in order to improve its retirement fund. As for Eq. (38), there also is no benchmark for efficient layoffs or for recruitment of labor that may be laid off, but again we can say that an increase in the value marginal product of labor, by increasing the likelihood of states of type a and reducing the likelihood of states of type d, will tend to increase hiring. Conversely, the cooperative is not self-extinguishing, in that it will recruit some new members in appropriate circumstances to the extent that employees wish to join.

4. COMMUNITY COOPERATIVE Now suppose instead that we have a hybrid cooperative with a criterion function that sets a negative value on the production of the negative externality and a positive value on membership in the cooperative. Letting   N (39) ni ¼ N þ Mi

218

ROGER A. MCCAIN

The criterion function becomes Ui ¼ g ðW i ; E i ; Gi ; V i ; F i ; ni Þ

(40)

We do not impose an openness requirement since the same normative standard is expressed by the inclusion of ni in the objective function. With a similarly constrained maximization of the expected value of U i , we have Lagrangean function L¼

n X

qi gðW i ; E i ;Gi ;V i ;F i ;ni Þ þ

i¼1

n X   li f ðLi ;E i ; F i ; K;iÞ  Qi i¼1

n n X   X þ mi R0  K þ pi Qi  wi Li  Ri þ ui ðN  Gi þ M  H i  Li Þ i¼1

þ

n X

i¼1

fi ðM  H i Þ þ yðN  N 0 Þ

ð41Þ

i¼1

Eqs. (8)–(20) are again unchanged. Eq. (21) becomes @L @f @g ¼ li þ 0 @F i @F i @F i

(42)

In a state in which a positive amount of effluent is produced, this will be an equality. Note that ð@g =@F i Þo0, so that equality is consistent with ð@f =@F i Þ40. Equivalently, pi ð@f =@F i Þ ð@g =@F i Þ   Li ð@g =@W i Þ

(43)

The left-hand side is the marginal contribution of increased effluent to the net revenues of the enterprise, per member of the work force. The right-hand side is the marginal trade-off, in terms of values of the criterion function, between reductions in the externality and increases in wage rates. Applying the usual assumption of diminishing marginal productivity, the upper limit on the marginal productivity of effluent is equivalent to a reduction in the effluent quantity. Thus, where the pure worker-owned firm increases the effluent so long as its marginal productivity is positive, the hybrid cooperative limits the effluent at some positive value of its marginal productivity. The exact limit would depend on the relative negative weight of the externality in the criterion function.

Worker and Community Cooperatives: A Multi-Criterion Model

219

Eq. (24) becomes   n @L X @g 1  n ¼ 0 ni þ y þ @N @ni N þ M i¼1

(44)

Thus, in place of Eq. (25), y¼

n X i¼1

  @g 1  n ni  0 @ni N þ M

(45)

Therefore, y will be zero, consistent with NWN0 and recruitment of new members, in case    n   X @f ð@g =@WÞ @g ¼ ðN þ M Þ qi pi  Wi ð 1  nÞ (46) @ni @Li Li i¼1 That is, recruitment of new members will be carried up to the point that the number of employees times the expected value of the sacrifice of revenues due to limitation of layoffs in some states, weighted by the criterion-function marginal weight on wage increases per worker, is equal to the criterion function marginal weight on increasing the membership times the deviation from 100% of the whole labor force before layoffs are made.

5. MISSION-DRIVEN ENTERPRISE AS AN ECONOMIC SYSTEM ‘‘Nonprofit corporations’’ and ‘‘Non-Governmental Organizations (NGOs)’’ generally refer to similar organizations. They are not prohibited from earning profits but are not permitted to distribute their profits. Instead, they are required to use profits and other resources at their command to advance their mission, which generally is stated in the charter. Thus, I prefer the term ‘‘mission-driven enterprises’’ (MDE). A Mission-Driven Enterprise is controlled by a board of directors or trustees. It is their responsibility to assure that the organization operates in accordance with its mission. If they fail to do so they may be guilty of failure of trusteeship, a criminal offense. Apart from criminal penalties, many MDE’s within our capitalist economic system depend on gifts, and that feedback can constrain the management of the MDE to enact the mission. Here, however, we will be concerned with mission-driven organizations in commodity production, and gifts will play no part in the analysis.

220

ROGER A. MCCAIN

Interest-driven enterprises will always have some tendency to act against the general interest when there are externalities, social dilemmas, or other ‘‘market failures.’’ With this in mind, one might argue (as Magnusson does) that organizations at the basis of a better society should not be interestdriven but mission-driven. We might think of an alternative economic system, certainly not capitalist but perhaps not socialist either, in which the local productive units are MDEs. A first question is this: how would the missions of these MDEs be determined? Historically, existing MDEs in countries like the United States have been established by the bequests of wealthy people, and the mission has been determined by the bequest. Thus, there are MDEs with the mission of promoting for-profit free market economics, and others with the mission of promoting the opposite. The ideal we seek here is a decentralized economy in which the local units act in accordance with the public interest, so there must be some mechanism to assure that the mission of each local MDE is consistent with the public interest. Presumably this would be a political process. The mission can influence, but not determine, the objective function of the enterprise. To take the example of the externality in Eq. (43), we may hope that @g =@F i , the marginal (negative) weight on the externality in the objective function, would correspond to the marginal social cost of the effluent. Even the best-intentioned trustees could direct their decisions in this way only if they have reliable information on the marginal social cost of the effluent. Similarly, in the context of Eq. (46), the appropriate weight to be attached to increasing the membership of the community cooperative should correspond to the marginal social benefit of membership, a complex question of fact to which the answer may simply not be known. In practice, the marginal weights in both cases are likely to be only rough approximations to socially optimal weights, determined as much by political compromise among the trustees as by any objective information on social costs and benefits. This leads to the second question: how are the trustees to be designated? If they are elected by the members or employees of the enterprise then they are likely to express the interests of those people. Whatever the charter of the organization may say about the importance of reducing effluent, it seems likely that trustees motivated primarily to increase the pay and pension contributions will often be able to do that, at the cost of increased effluent, without seriously risking prosecution for failure of trusteeship. If the trustees are elected by the local population, or designated by local government, then they may feel a stronger motivation to reduce effluent that

Worker and Community Cooperatives: A Multi-Criterion Model

221

has local bad effects, and to promote local employment. However, where externalities transcend regional boundaries, local representatives would have no more tendency to reduce them than would employee representatives. If the trustees were appointed by national government or international organizations, then they might have motives to take boundary-crossing externalities into account, but any political process for the appointment of trustees is likely to be influenced as much by the objective of keeping the dominant political party in office as by any economic concerns. It may be that the board of trustees would be a mixed body, with some members elected and others appointed locally or nationally, with consequent different objectives, so that the objective function would be determined by a process of negotiation among trustees with different concerns. To configure a board that would implement even an approximately optimal policy would require both a precise theory of bargaining and a fine balance of power that is hardly plausible. A self-coopting board of trustees is a group of trustees who recruit their own successors. A self-coopting board of trustees might maintain a consistent value position as against the pressures of member interests, local and national politics, by recruiting replacement members who share their values. Thus, if an appropriately motivated board can initially be formed, its appropriate value standards could be quite stable. Certain religious foundations provide plausible examples. However, such a process of selfcooption would also make it difficult if not impossible to replace a corrupt board or one that systematically perverts the mission of the enterprise, without actually committing provable failure of trusteeship. Examples of such boards are also easy to supply. All in all, it seems that the decision process envisioned by Section 4 could be reliably implemented only if (1) organizational missions were directed by some political authority, (2) trustees were initially appointed by some political authority and (3) were subject to routine review by some enforcement authority to assure that the board continues to make decisions in accordance with the mission. In the light of these points, the distinction between community cooperatives and government organizations seems impossible to make. That said, decentralization of state control via community cooperatives differs in important ways from the Soviet-type economic system and from the national state enterprises established in western European and other democracies as steps toward socialism, and the problems indicated here probably are soluble in that framework; so that community cooperatives might prove a more viable form of state organization than has been established in the past.

222

ROGER A. MCCAIN

Some similar issues might be raised with respect to the standard of open membership that defines the worker cooperative in Section 3. However, open membership is a regulatory standard that can be included in statutory law establishing the worker cooperative as a distinct form of (legal) corporation, as is the case in France (Kamshad, 1997), rather than a broad general goal that might be implemented more or less by a board of trustees. Moreover, while worker cooperatives operating strictly in the interests of their members or employees would not abate externalities efficiently, just as profit-seeking enterprises do not, here again a regulatory solution is reasonable. Pigovian taxes or cap-and-trade regulations both convey information along with incentives to make decisions in accordance with the marginal cost of the externality. (Both rely to some extent on centralized information: in the case of cap-and-trade systems to determine the targets or caps, and in the case of Pigovian taxes to estimate the marginal social costs of the externalities.) Moreover, the parallel between worker cooperatives and profit-seeking corporations is not necessarily direct. Unlike corporations, but like proprietorships, worker cooperatives choose their objectives according to the free will of the members. If the members of a cooperative are persuaded that they ought to abate externalities, then their objective function may include a negative weight on the production of effluent, contrary to the worst-case assumptions of Sections 2 and 3. By contrast, if a profit-seeking corporation were to do so, at a sacrifice of profits, shareholders would have a case for a lawsuit. Thus while we might not want to rely on persuasion to solve problems of externality, we might expect a considerable degree of voluntary compliance with environmental regulations from worker-cooperatives, by contrast with profit-seeking corporations. A system of worker cooperatives would rely on the parochial interests of employees in firms, together with both markets and prudent regulation, to promote efficient production. A system of ‘‘community cooperatives’’ would rely to a much greater extent on the good intentions of government authorities and their appointees. One’s estimate of the likely success of the two will be correlated with the confidence one has of the power of good intentions in politics.

6. CONCLUSION AND SUMMARY Many critics of profit-seeking enterprise argue that new forms of enterprise must balance multiple objectives or interests, and with this in view, oppose worker cooperation as too narrow in its objectives. This suggests a

Worker and Community Cooperatives: A Multi-Criterion Model

223

multi-criterion approach to decision making, an approach that is explored in this chapter. The models explored here allow for distinctions between member and nonmember labor and for layoffs. Although (on the one hand) worker-owned and worker cooperative enterprises can best be thought of as pursuing multiple goals, nevertheless (on the other hand) there are clear distinctions in the conditions that characterize maxima for worker-owned firms, worker-cooperatives, and ‘‘community cooperatives.’’ Worker-owned firms (without open membership) will not ordinarily recruit new members, while the other two forms will. This parallels the ‘‘self-extinction forces’’ in the traditional literature on labor-managed enterprises, in the case of the closed worker-owned firm. Neither worker-owned nor worker-cooperative enterprises have incentives to abate externalities, while the community cooperative, assumed to include aversion to externalities in its decision processes, would do so. There is, however, some question as to how the ‘‘community cooperative’’ decision model could really be implemented. Whereas the decision models for worker-owned and cooperative enterprises reflect the material interests of their members, and an adaptive learning process might plausibly tend toward the maxima of the criterion functions assumed, the ‘‘community cooperative’’ depends on good intentions expressed in a mission statement. Another advantage of the multi-criterion decision model is that it can clarify the motives to invest in labor-managed enterprises, which has been a subject of some controversy in the literature. For worker-owned and cooperative firms, including a pension fund as one of the benefits to members leads to the conclusion that investment will be determined by balancing the marginal productivity of capital against the discounted benefits of the pension, although the marginal productivity can be underweighted to the extent that nonmember labor is employed.

NOTES 1. See e.g. Girard (2003), Girard and Langlois (2005), Wall, Duguay, and Rohan (2004). These cooperatives admit both employees and customers or clients as members, and also allow ‘‘supporting membership’’ by individuals who disinterestedly support the purposes of the cooperative. Supporting members may be no more than one-third of the board of directors. Although this proposal relies on the mixed membership rather than on a mission statement to integrate the objectives of different groups, the objectives seem rather similar, and case studies suggest that a mission concept also plays an important role. Note also the remarks in the recent papal encyclical, Caritas in Veritate, paragraph 40.

224

ROGER A. MCCAIN

2. Some of my models, postulating individual utility functions of workers and simplifying by imposing a representative-agent interpretation, use the same mathematical formulations as multi-criterion models, but were not suggested by that approach, and strictly assume only one criterion, average member utility (see e.g. McCain, 2006b, 2007). 3. The efficiency implications would also be influenced by differences in factor neutral productivity between worker-owned (including cooperative) and profit-seeking enterprises. Thus, the discussion ignores the widespread evidence that labor productivity in worker cooperatives is no less, and often greater, than that of comparable profit-seeking firms. Coupled with the equally widespread observation that the cooperatives employ less capital per worker, this would imply systematically superior factor-neutral productivity, although the results are usually phrased as differences in labor productivity. See references in McCain (2007).

REFERENCES Girard, J.-P. (2003). Revolution within a revolution. Making Waves, 14(3), 13–16. Girard, J.-P., & Langlois, G. (2005). Tracking the social impact of solidarity co-ops. Making Waves, 16(1), 5–7. International Cooperative Alliance. (1995). Statement on the co-operative identity. Available at http://www.ica.coop/coop/principles.html. Retrieved on June 20, 2006, last modified on September 2, 2006. Intriligator, M. (1971). Mathematical optimization and economic theory. Englewood Cliffs, NJ: Prentice-Hall. Kamshad, K. (1997). A model of the free-entry producer cooperative. Annals of Public and Cooperative Economics, 68(2), 225–245. Kuhn, H. W., & Tucker, A. W. (1952). Nonlinear programming. Second Berkeley Symposium on Mathematical Statistics and Probability, University of North California Press, Berkeley, CA. Magnuson, J. (2008). Mindful economics. New York: Seven Stories Press. McCain, R. (2006a). Cooperative partnership of not-for-profit and conventional corporations: A proposal. Proceedings of the 13th Conference of the International Association for the Economics of Participation (IAFEP), Sanctuario de Arantzazu, Spain. McCain, R. (2006b). Worker cooperatives, effort and shirking: A semi-effective game model. Proceedings of the 6th Meeting on Game Theory and Practice, Zaragoza, Spain, July (pp. 10–12). McCain, R. (2007). Cooperation and effort, reciprocity and mutual supervision in worker cooperatives. In: S. Novkovic (Ed.), Advances in the economic analysis of participatory and labor-managed firms. Greenwich, CN: JAI Press. McCain, R. A. (2010). Game theory: A nontechnical introduction to the analysis of strategy (Revised Edition). Singapore: World Scientific Publishers. Scitovszky, T. (1943). A note on profit maximisation and its implications. Review of Economic Studies, 11(1), 57–60. Wall, J., Duguay, P., & Rohan, S. (2004). New synergies. Making Waves, 15(1), 32–36. Ward, B. (1958). The firm in Illyria: Market syndicalism. American Economic Review, 48(3), 566–589.

ANTECEDENTS AND CONSEQUENCES OF THE ADOPTION OF MARKET-BASED COMPENSATION BY ISRAELI KIBBUTZIM Raymond Russell, Robert Hanneman and Shlomo Getz 1. INTRODUCTION This chapter analyzes the antecedents and consequences of transformations that have recently been occurring among Israeli kibbutzim. After serving for nearly a century as some of the world’s best known examples of organizations that distribute resources ‘‘from each, according to ability, to each according to need,’’ most kibbutzim now pay their members differential salaries on the basis of the market value of their work. Both inside and outside the kibbutz movement, this innovation has been widely seen as signifying a fundamental change in the identity of the kibbutz. Since October 2005, Israeli law has acknowledged two major types of kibbutzim. A kibbutz that retains need-based compensation is a ‘‘collective kibbutz.’’ A kibbutz that pays market-based differential salaries is now known as a ‘‘renewed’’ kibbutz. An additional number of kibbutzim provide members with budgets that are based mostly on the needs of their Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 225–246 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011013

225

226

RAYMOND RUSSELL ET AL.

households and partly on the market value of their individual work. In the eyes of the law, these kibbutzim continue to qualify as collective kibbutzim, but academic and kibbutz movement publications have generally treated them as constituting a distinct ‘‘mixed’’ type. The present study is an empirical analysis of transformations of collective kibbutzim into the mixed and renewed types of kibbutzim in the years 1995–2005. It treats these transformations as being of interest both in and of themselves, and as manifestations of more general phenomena. For the literatures on labor-managed and communal organizations, they once again illustrate the ‘‘degeneration’’ of democratic workplaces (Mill, 1909, p. 790) and the ‘‘decommunalization’’ of utopian communities (Kanter, 1968; Pitzer, 1989). For the sociology of organizations, they provide new arenas in which to apply theories of ‘‘inertia’’ (Hannan & Freeman, 1984) and ‘‘deinstitutionalization’’ (Oliver, 1992) in organizations and of the diffusion of innovations through organizational fields (DiMaggio & Powell, 1983). This chapter asks what each of these theoretical perspectives can contribute to our understanding of these transformations.

1.1. Changes in the Kibbutzim, 1990–2009 In the latter half of the 1980s, the kibbutzim fell into an economic and demographic crisis from which they have still not emerged. Many kibbutzim suffered heavy losses of both money and members. Because the kibbutzim were collective guarantors of each other’s debts, the entire kibbutz movement descended into bankruptcy. In the face of such reverses, critics both outside and inside the kibbutzim soon began calling for the kibbutzim to introduce reforms that could increase their incomes and reduce their costs. If members could bring in more revenue by working outside the kibbutz economy than within it, it was now suggested that they should be encouraged to take those higher paying jobs. If kibbutz economic ventures could increase their incomes by hiring more nonmembers to work in them, they were now encouraged to do so, even though the rising use of hired labor would be widely seen as a sign of the transformation of the kibbutzim from cooperative into capitalist firms. If kibbutz members now ate their meals in their individual apartments rather than collective dining halls, then those dining hall should be closed. In 1990, the kibbutz federations formally committed their member kibbutzim to adopt measures of this sort as a condition of their emergence from bankruptcy, but they left each individual kibbutz free to decide which

Antecedents and Consequences of the Adoption of Market-Based Compensation 227

specific innovations it would or would not introduce. Many kibbutzim set up ‘‘innovation teams’’ to identify reforms that looked especially appropriate for their kibbutz (Near, 1997, p. 352). In making decisions about which changes to adopt, many kibbutzim have also sought advice from their federations or from professional consultants. Despite all of these influences encouraging them to change, many kibbutzim found themselves deeply divided between reformers and traditionalists, and uncertain about what to do next (Ben-Rafael, 1997; Palgi, 1994, 2004). Reforms tend to be backed by ‘‘technocrats,’’ who seek a freer hand in management, and by younger, kibbutz-born or Israeli-born members, who seek lifestyles more similar to those of Israelis who live outside the kibbutzim (Topel, 2005). Traditional kibbutz values and practices tend to be defended by older members and by members born outside of Israel (Ben-Rafael, 1997). For most of the 1990s, the kibbutzim embraced a number of mostly modest reforms that were ‘‘convergent’’ with their traditions (Tushman & Romanelli, 1985), while avoiding ‘‘radical’’ changes (Greenwood & Hinings, 1996) that threatened to alter the identity of the kibbutz. Most kibbutzim transferred authority for governing their economic ventures from the general assembly to boards of directors, but they refused to abolish the principle of rotating managers at the end of fixed terms. Most kibbutzim gave their members the right to take jobs outside the kibbutz, but the income that kibbutz members earned on such outside employment continued to be paid to the kibbutz, not the member. A growing portion of consumption expenditures, such as the costs of electricity and travel, became private rather than public, but most members’ budgets continued to be based not on work but on need. Although the reforms that were most widely adopted in the 1990s were relatively modest, the kibbutzim found decisions regarding them very difficult to make, and they introduced these reforms with great ambivalence. Many more changes were proposed and discussed than were accepted, and many changes that had supposedly been decided upon were never or only briefly put into practice. Relatively innocuous proposals like having members pay for their own energy use successfully passed through all of these stages, but more radical proposals either were never brought up for discussion or were dropped in early stages of consideration. Kibbutzim that adopted more radical reforms, such as paying differential salaries to their members, remained a small and daring minority until very late in this period (Russell, Hanneman, & Getz, 2006). In the early 1990s, Ein Zivan and Snir became the first kibbutzim to pay differential salaries to their members, but other kibbutzim remained

228

RAYMOND RUSSELL ET AL.

reluctant to take this step, because it was so explicitly contradictory to traditional kibbutz values. Beginning around 1995, kibbutzim such as Gesher Haziv and Naot Mordecai found a way to make payment of differential salaries more palatable to kibbutz members. They created the so-called ‘‘safety net’’ budget, in which members receive differential, marketbased salaries, but their incomes are taxed to support a minimal standard of living and level of social services for all kibbutz members. The safety net budget sounded more like Scandinavian socialism than like capitalism, making it a less radical departure from the traditional political culture of the kibbutzim. Politically, by using taxes collected from healthy working-aged members to meet the needs of older and weaker members, the safety net budgetary system had the potential to create alliances between the two. Participants in decisions about compensation were well aware that any innovation that deviates from the traditional principle of ‘‘from each according to ability, to each according to need’’ would signify a fundamental change in the identity of the kibbutz. Because these transformations involved changes in the founding documents of the kibbutz, regulations applicable to Israeli cooperatives required the consent of 75% of the members before such changes could be introduced (Gavron, 2000, p. 93). The distribution of nonreligious kibbutzim by type of compensation in each year from 1995 to 2009 is shown in Table 1. From 1996 to 1999, Table 1.

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009

Kibbutzim by Type, 1995–2009.

Collective

Mixed

Renewed

247 238 223 207 194 171 146 120 107 89 76 73 66 58 55

0 6 15 24 30 37 37 32 30 31 23 18 15 11 9

2 5 11 18 25 41 66 97 112 129 150 157 167 179 184

Note: In 2006, Kibbutz Shomria ceased to be a kibbutz after all members left. This reduced the total number of nonreligious kibbutzim from 249 to 248.

Antecedents and Consequences of the Adoption of Market-Based Compensation 229

the number of kibbutzim using the safety net budgetary system rose slowly from less than 2% to nearly 11%. During these years, kibbutzim with safety net budgets were outnumbered by kibbutzim choosing mixed forms of compensation. The proportion of kibbutzim with compensation systems that mix differential salaries with traditional budgets peaked in the year 2000, when it rose to 13.9%. That year, however, was also the first year in which safety net kibbutzim outnumbered those with mixed systems. By 2003, the percent of kibbutzim with safety net budgets had grown equal to the share that remained collective, at 44.1%. By 2009, the proportion of kibbutzim with safety net budgets exceeded 70%, and less than a quarter of all nonreligious kibbutzim continued to share incomes purely on the basis of need. As kibbutzim with safety net budgets increased in number, heated debates ensued over whether a kibbutz that pays differential market-based salaries to its members still deserves to be called a kibbutz. For the Israeli government, this issue took on practical significance, because it needed a ruling on how members of kibbutzim who pay differential salaries should be taxed. In a traditional kibbutz, income is earned collectively, and taxes are paid collectively. As growing numbers of kibbutzim began to pay differential salaries to their members, the Israeli government took the position that members of these kibbutzim should pay taxes individually, rather than collectively. In 2002, the Israeli government asked a Public Committee chaired by Professor Eliezar Ben-Rafael of Tel Aviv University to answer the question of whether a kibbutz that paid differential salaries to its members could still be considered a kibbutz. The Public Committee recommended in 2003 that the legal definition of a kibbutz should be amended to acknowledge that kibbutzim now came in more than one type (Manor, 2004). A kibbutz that adhered to the traditional model would henceforth be known as a ‘‘collective kibbutz.’’ A kibbutz that pays differential salaries to its members, introduces private ownership of housing, or distributes the assets of the kibbutz to individuals, would be called a ‘‘renewed kibbutz.’’ The Public Committee’s recommendations were later approved by the kibbutz movement and the Israeli government and became law in October 2005. ‘‘Renewed kibbutzim’’ had come to outnumber the ‘‘collective kibbutzim’’ even before these definitions became law. In addition to their systems of compensation, renewed kibbutzim differ from collective kibbutzim in a number of other ways. Table 2 lists reforms that had been adopted by 2001 in most kibbutzim of all types, while Table 3 identifies reforms that had been adopted by most renewed kibbutzim but not most collective kibbutzim. Table 2 indicates that the most widely adopted reforms are those that increase the rights and entitlements of members,

230

RAYMOND RUSSELL ET AL.

Table 2.

Innovations Adopted by Most Kibbutzim by 2001.

New rights and entitlements Secret ballot Members work outside the kibbutz Responsibility of member to select work Pension plan Health insurance Rental of housing to kibbutz-born young adults Differentiation and rationalization Board of directors in industry Shadow wages Human resource division Involvement of nonmembers Increase use of hired labor Hiring managers for businesses Agricultural cooperation with other kibbutzim Selling services to the outside Accepting outsiders in educational system Rental of housing to the outside Rental of housing to kibbutz-born salaried employees Privatization of services Comprehensive budget Pay for electricity Pay for meals Pay for travel OK to own or use private car Young study by special agreement Cancel evening meal

expand the role of nonmembers, and privatize consumption. Large numbers of changes in each of these three categories have been adopted by substantial majorities of collective, mixed, and renewed kibbutzim. New rights and entitlements gaining general acceptance include secret mail ballots, pension plans, health insurance, members working outside, and rental of housing to the kibbutz-born. By 2001, most kibbutzim in all three categories had also decided at some time or other to increase the use of outside labor and were ready to hire outside managers to run one or more of their businesses. Most were partnering with other kibbutzim in agricultural ventures, selling services to outsiders, and welcoming outsiders in their educational systems and rental apartments. In consumption, most kibbutzim in all three categories had adopted comprehensive budgeting, required members to pay for their own electricity, meals, and travel and were permitting members to own or use private cars. Most had also cancelled the evening meals in their dining halls on all evenings except Fridays.

Antecedents and Consequences of the Adoption of Market-Based Compensation 231

Table 3.

Innovations Adopted by Most Renewed and Mixed Kibbutzim by 2001.

Differentiation and rationalization Abolishing the rotation of managers Economic branches as profit centers Separating the economy from the community Involvement of nonmembers Replace members by outside labor Nonmember as chair of board of business Privatization of services Pay for recreation Increase size of house at own expense Privatization of health services Privatization of the laundry Parent’s budget includes enrichment Higher education in members’ budgets Privatization of education Canceling breakfast in dining hall Material rewards and incentives Pay for overtime or additional work Connection between days worked and budget Payments for additional work in services

Reforms that had been adopted by 2001 by most renewed and mixed kibbutzim, but not most collective kibbutzim, are listed in Table 3. By that year, the mixed and renewed kibbutzim had differentiated themselves from the collective kibbutzim in a number of significant ways. In these kibbutzim, privatization had been taken much farther than in collective kibbutzim. In addition to requiring members to pay for electricity, meals, and travel, these kibbutzim had also privatized laundry and health services, recreation, and education and had cancelled the morning meal in their dining halls. Among mixed and renewed kibbutzim, most also reported that they were in favor of separating the economy from the community, treating economic branches as profit centers and outsourcing services.

2. THEORIES OF ORGANIZATIONAL TRANSFORMATION The adoption of market-based compensation systems by Israeli kibbutzim can take on widely differing significance, depending on which theoretical perspective one views them from. One research tradition emphasizes the

232

RAYMOND RUSSELL ET AL.

uniqueness of the kibbutzim, and of more general classes of organizations with which they are associated. Whether we classify the kibbutzim as labormanaged workplaces or as communes, both of these literatures view their subject matters as inherently ephemeral phenomena. Insofar as empirical studies and theories differ within these literatures, it is over the question of which influences play the greatest roles in accelerating or retarding the spread of widely observed signs of degeneration and decommunalization. Another set of theories invite us to ignore completely the unique features of the kibbutzim, and instead to treat them as organizations like many others. From this more general point of view, any transformation from one type of kibbutz to another is an instance of change in organizations. Like most instances of change in organizations, these transformations require the organizations involved to overcome their internal tendencies toward ‘‘inertia’’ (Hannan & Freeman, 1984). Because the practices that these new compensation systems displace had previously been widely legitimated and institutionalized within the kibbutz movement, they emerge as ‘‘illegitimate changes’’ (Kraatz & Zajac, 1996) and their adoption depends on the ‘‘deinstitutionalization’’ (Oliver, 1992) of the preexisting organizational form. Insofar as these new organizational forms become widely adopted, they illustrate the diffusion of innovations within organizational fields (DiMaggio & Powell, 1983). This heterogeneous mix of relevant theories yields a diverse and occasionally contradictory set of predictions regarding the influences that make organizations more or less likely to undergo these transformations. We review these expected influences below.

3. THE KIBBUTZIM AS DEMOCRATIC AND COMMUNAL ORGANIZATIONS 3.1. Size and Age of Organizations Democratic (Weber, 1978) and communal (Pitzer, 1989) organizations have often been said to become less democratic and less communal, the older and larger they become. In the literature on producer cooperatives, this is the old idea that cooperatives have a tendency to ‘‘degenerate’’ over time (Mill, 1909). In a study of 49 kibbutzim in the years 1976–1979, Rosner and Tannenbaum (1987) found that both the age and the size of a kibbutz had significantly negative effects on the rotation of officers, while the age of the kibbutz also had a negative effect on participation in the assembly.

Antecedents and Consequences of the Adoption of Market-Based Compensation 233

Batstone (1983) argues that periods of degeneration in democratic workplaces are often followed by periods of ‘‘regeneration,’’ in which members reassert their democratic traditions. Degeneration followed by regeneration would produce a curvilinear effect for age on changes that take democratic workplaces away from their traditions, positive for low values of age, and negative at higher values. Two studies of the use of nonmember workers in labor-managed workplaces (Estrin & Jones, 1992; Simons & Ingram, 1997) have reported effects for age of this type. To be sensitive to the possibility of similar curvilinear effects of age in this study, our models include both the linear and the squared form of age.

3.2. Ideology Communes, cooperatives, and other labor-managed workplaces often originate in explicit decisions to reject conventional capitalist alternatives. This rejection is often rooted in ideology. Many perspectives therefore attribute the transformation of these organizations to the gradual loss of the ideological and moral motivations that initially sustained them. In research on the kibbutzim, differences between the two major kibbutz federations are often taken as convenient markers of ideological differences. By such measures as the use of hired labor, kibbutzim affiliated with the smaller and more ideologically coherent Artzi federation have often been found to be more strongly attached to kibbutz traditions and ideology than kibbutzim affiliated with the larger and more heterogeneous Takam (Rosner & Tannenbaum, 1987; Simons & Ingram, 1997). For these reasons, we expect Artzi kibbutzim to be slower than Takam kibbutzim to transform themselves into renewed or mixed types of kibbutzim.

3.3. Geographic Isolation In a study of utopian communities in the United States, Kanter (1968) reported that communities that were geographically isolated and that avoided contact with outsiders were more likely than others to retain their communal structures for long periods. Regarding the kibbutzim, Ben-Ner (1987) noted that kibbutzim that are near cities tend to become increasingly integrated into the urban market economy and adopt increasingly individualistic patterns of consumption in response. These works suggest

234

RAYMOND RUSSELL ET AL.

that kibbutzim located at greater distances from cities will be less likely than others to transform themselves into new or mixed types of kibbutzim.

4. KIBBUTZIM AS INSTITUTIONALIZED ORGANIZATIONS 4.1. Causes and Consequences of Organizational Inertia In many theories of organizations, the most important institutional influences on organizations are those that are felt throughout a society. In the so-called ‘‘old’’ institutional theories of Selznick (1949) and Stinchcombe (1965), on the other hand, the unique structures and practices of individual organizations also gradually acquire the status of institutions in the eyes of their own members. In 1984, Hannan and Freeman invoked this idea to support their claim that organizations live or die by their current structures, rather than being free to imitate the structures of more successful competitors. They labeled this tendency for some organizations to be more resistant to change than others as organizational ‘‘inertia.’’ The notion that individual organizations are characterized by more or less inertia is relevant to this study in two ways. First, although the literatures on democratic degeneration and decommunalization see age and size as making organizations more likely to abandon their unique practices, Hannan and Freeman’s theory of organizational inertia sees these influences as making organizations less likely to introduce changes of any kind. The effects of age and size on transformations among types of kibbutzim therefore constitute a head-to-head test of the relative relevance of these two bodies of theory. In addition to testing Hannan and Freeman’s predictions about the causes of organizational inertia, this study also provides opportunities to examine its consequences. Along with age and size, many additional characteristics are likely to make one organization more or less likely to introduce innovations than another. Kibbutzim also differ from one another in the average age of their members and in the occupations, educations, and gender of their members. These differences contribute in turn to large differences among kibbutzim in the degree of support for or opposition to reform among their members (Ben-Rafael, 1997; Palgi, 2004). To capture differences among kibbutzim in the attitudes of their members toward reforms, our analyses include an index of the proportion of 52 reforms that diffused earlier among the kibbutzim that were adopted by

Antecedents and Consequences of the Adoption of Market-Based Compensation 235

each individual kibbutz. This index is based on surveys of all kibbutzim conducted annually by the Institute for Research of the Kibbutz at the University of Haifa in the years 1990–2001 (Russell et al., 2006). For all years in which these survey data were missing or in which the survey was not conducted (2002–2005), we use the most recently observed index of prior innovations for that kibbutz as our best estimate of the current value.

4.2. Resource Scarcity and Deinstitutionalization According to Oliver’s (1992) theory of deinstitutionalization, organizations are most likely to abandon institutionalized practices when scarcities of resources make it increasingly difficult to afford their costs. On the basis of this theory, we would expect to find that kibbutzim that are in greatest economic difficulty would be most likely to introduce differential salaries. In the literatures on communal and democratic workplaces, the perceived ‘‘degeneration’’ of their practices is often viewed as a consequence not of hardship, but of prosperity. Democratic firms of many kinds have been accused of becoming more likely to abandon their democratic structures, the more capital they accumulate or the more profitable they become. In her history of utopian communities in the United States, similarly, Kanter (1968) identifies ‘‘sacrifice’’ and ascetic ‘‘renunciation’’ of worldly pleasures as characteristics that help to make some of these organizations unusually long-lived. The implication is that communal organizations like kibbutzim become more tempted to introduce changes, the more affluent they become. This thus constitutes another instance in which theories of organizations as institutions and theories of democratic and communal organizations make contradictory predictions. The problems that have shaken the kibbutzim in recent years have been not only economic but also demographic. As they have lost faith in the economic futures of their kibbutzim, many kibbutz members and adult children of members have left their kibbutzim. Calls for change in the kibbutzim have therefore aimed not only to make the kibbutzim more efficient and productive, but also to make them more attractive to present and future members. This anecdotal evidence is consistent with the expectations that can be derived from relevant organizational theories, such as that of Oliver (1992). From this perspective, the loss of members can be viewed as another form of resource scarcity that leads institutionalized organizations to abandon

236

RAYMOND RUSSELL ET AL.

costly traditions. For these reasons, we include a measure of the percentage change in membership from one year to the next in the analyses that follow.

4.3. Mimetic Isomorphism Whatever motivates certain organizations to overcome their inertia and pioneer the use of new forms of organization, a large body of research and theory indicates that other organizations become more likely to adopt an innovation, the higher the number and proportion of similar organizations that adopt it. DiMaggio and Powell (1983) called this phenomenon ‘‘mimetic isomorphism.’’ A number of subsequent studies have documented the relevance of this process to many instances of diffusion. For this reason, we include the proportion of all kibbutzim that had previously transformed themselves into either mixed or renewed kibbutzim in all models.

5. MEASUREMENT Testing the relevance of these diverse theories to recent transformations in the kibbutzim requires a wide range of measures. Means, standard deviations, and mutual correlations for all predictors included in these analyses are shown in Tables 4 and 5. The average kibbutz was 50 years old at the time of these observations and had 218 members. Age and size are highly correlated with each other (.60). Kibbutzim belonging to the Artzi federation constitute 33% of the total. The average kibbutz is located 76 km from the nearest large city (Tel Aviv or Haifa). Rural kibbutzim are on average younger (.44) and smaller (.31) than other kibbutzim. The proportion of other kibbutz reforms previously adopted by the kibbutzim, and the proportion of kibbutzim that have previously transformed themselves into renewed or mixed kibbutzim are strongly correlated both with each other (.68) and with type of kibbutz. Because kibbutzim are responsible for each other’s debts, their federations have learned to keep a close watch on their finances. Our measure of economic crisis is taken from evaluations of the economic condition of individual kibbutzim by kibbutz movement economists. It combines data on debt, profit per member, liquidity, and living expenses into a composite index. It ranges from 0 to 100, with scores of 0–30 indicating that the condition of a kibbutz is ‘‘good,’’ 31–50 that it is ‘‘OK,’’ 51–65 signifying that it ‘‘needs improvement,’’ 66–80 signs of ‘‘hidden crisis,’’ and 81–100 showing

Antecedents and Consequences of the Adoption of Market-Based Compensation 237

Table 4.

Descriptive Statistics.

Variable

Mean

Standard Deviation

Economic crisis Change in members Collective Mixed Renewed Years since change Age Age squared (100s) Number of members Federation Rural Prior innovations Percent renewed or mixed

59.89 1.64 .78 .06 .15 .83 50.39 28.04 218.43 .33 75.62 508.05 21.70

18.96 8.73

1.91 16.29 15.73 133.70 57.43 505.93 24.82

Notes: Economic crisis is measured on a scale from 0 (excellent condition) to 100 (severe crisis); change in members is the difference between the reported membership from the year of observation to the next; ‘‘collective,’’ ‘‘mixed,’’ and ‘‘renewed’’ are dummy variables with the value one if a kibbutz is of the given form at the year of observation, and zero otherwise; age is the number of years since the founding of the kibbutz until the year of observation; members is the membership of the kibbutz at the year of observation; federation is a dummy variable taking the value 1 if the kibbutz is a member of Artzi, and zero otherwise; rural is the distance, in kilometers, from the kibbutz to the nearest urban place. Prior innovations is an index created by calculating the percentage of all surveyed innovations that had been adopted by a kibbutz in all years prior to the current year of observation. Percent renewed or mixed is the percentage of all kibbutzim that had adopted either the mixed or renewed form at the year prior to the observation.

Table 5.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12.

Economic crisis Change in members Collective Mixed Renewed Age Age squared Number of members Artzi federation Rural Prior innovations Percent renewed/mixed

Correlations for Pooled Data.

1

2

3

4

1.00 .08 .11 .05 .09 .04 .04 .21 .05 .03 .12 .04

1.00 .05 .04 .03 .14 .16 .11 .03 .03 .08 .03

1.00 .50 .81 .13 .12 .16 .10 .08 .70 .60

1.00 .11 .06 .05 .07 .07 .02 .30 .20

5

6

7

8

9

10

11

1.00 .11 1.00 .11 .97 1.00 .15 .60 .58 1.00 .07 .05 .02 .09 1.00 .07 .44 .38 .31 .11 1.00 .60 .24 .25 .11 .19 .12 1.00 .55 .28 .30 .06 .00 .00 .68

238

RAYMOND RUSSELL ET AL.

‘‘active crisis’’ (Yoffe, 2005, p. 47). The incidence of economic crisis is negatively correlated with the size of a kibbutz (.21), indicating that larger kibbutzim are on average in better economic condition than smaller ones. Change in members averages 1.64, meaning that the average kibbutz was losing 1.6 members per year between 1995 and 2005. This variable is negatively correlated with both the age of the kibbutz (.14) and the number of members (.11), indicating that both older and larger kibbutzim are more likely to be losing members. It is also negatively correlated (0.08) with economic crisis. This means that kibbutzim that are losing money are the kibbutzim most likely to be losing members. Abramitzky (2008) notes that while members of kibbutzim in economic difficulties leave as they lose faith in the futures of their kibbutzim, members of wealthy kibbutzim are in contrast ‘‘locked in’’ by the rising and nontradable value of their memberships.

6. MODELING TRANSFORMATIONS In Table 6, we present a series of models that analyze the effects of these predictors on the probability that a collective or mixed type of kibbutz will transform itself into a mixed or renewed type of kibbutz between one year and the next. The observations are a panel of 249 kibbutzim observed across the years from 1995 to 2004. The models presented are multinomial logistic regressions, with fixed effects for year and kibbutz. Coefficients are additive effects on the log-odds of a transition during a year relative to the odds of no change. Estimates by maximum likelihood are obtained from SPSSx version 15.0. For each of the three observed transitions, two models are estimated. In each pair of models, the first model omits the number of innovations that had previously been adopted on each kibbutz, which is our measure of organizational inertia, and the second model includes it. In the first model in each pair, predictors like age, size, and economic condition are allowed to exert their full effects, regardless of how that effect is brought about. A given influence might be making an organization more or less likely to make this particular change, as in theories of degeneration, or it could be making organizations more or less likely to introduce changes of all types, as in theories of inertia and resource dependency. By adding the number of innovations previously adopted by each organization, the second model in each pair attempts to control for differences among organizations in their relative openness to change or inertia. Under these circumstances, the second

Antecedents and Consequences of the Adoption of Market-Based Compensation 239

Table 6.

Effects on the Probabilities of Transitions to the Mixed and Renewed Forms. From Collective to Mixed

Economic crisis Change in members Age Age squared Number of members Federation Rural Prior innovations Percentage renewed or mixed Intercept

From Collective to Renewed

From Mixed to Renewed

I

II

I

II

I

II

.015 .053 .072 .062 .003 .607 .004

.027 .002 .057 .062 .004 .588 .009 .068

.023 .013 .043 .072 .003 .056 .007 .003 .037

.023 .050 .101 .097 .001 .703 .002

.042

.010 .044 .070 .074 .002 .185 .002 .003 .010

.023 .050 .101 .096 .001 .746 .002 .000 .027

6.166

6.679

6.205

6.824

.028 1.968

2.048

Notes: po.05, one-tail. Multinomial logistic regression coefficients (estimated with SPSSx version 15.0), with fixed effects (not shown) of year and kibbutz. Coefficients are additive effects on the log-odds of a transition during a year relative to the odds of remaining in the origin status. Models I and II differ only by the inclusion of the prior innovations variable. Nagelkerke pseudo R2: Model I transitions from collective ¼ 0.174; Model II transitions from collective ¼ 0.263; Model I transitions from mixed ¼ 0.127; Model II transitions from mixed ¼ 0.127. For definitions of the variables and their metrics, see the text and notes to Table 4.

estimates of the effects of each predictor variable estimate the effects of those predictor variables not on the likelihood of change in general, but on the unique attractiveness to the organization of this particular transformation.

7. INFLUENCES ON TRANSFORMATIONS Estimates of the effects of each potential influence on transformations to the mixed and renewed types of kibbutzim are shown in Table 6. In models of transitions from collective kibbutzim to kibbutzim of the mixed type, the age of the organization shows the positive but curvilinear relationship predicted by Batstone’s theory, as indicated by a significantly positive effect for age in Model I and a significantly negative effect for age squared in Model II. In models of transitions from the collective kibbutz model to the

240

RAYMOND RUSSELL ET AL.

renewed types, coefficients for age and age squared are similar in sign and magnitude, but lack statistical significance. The effects of size in these models, on the other hand, are as in theories of organizational inertia, not theories of degeneration. The significantly negative effects of size on transitions from collective kibbutzim to both mixed and renewed types of kibbutz indicate that as Hannan and Freeman (1984) predict, kibbutzim become less likely to abandon their traditions, the larger they become. In the year 2000, the Artzi and Takam federations merged their formerly separate organizations into a single kibbutz movement. Table 6 indicates that as in Abramitzky (2008), Artzi kibbutzim continued to differ significantly from Takam kibbutzim in their readiness to embrace these transformations in the years covered by this analysis. As expected, kibbutzim affiliated with the Artzi federation are less likely than other kibbutzim to depart from traditional practices, whether we look at transitions from the collective kibbutz to the mixed type or to the renewed kibbutz. When the number of innovations previously adopted by the kibbutz is included, however, estimates of the effect of differences between federations are greatly reduced and lose statistical significance. This suggests that Artzi kibbutzim were less likely than Takam kibbutzim to make these transitions because they were less likely than Takam kibbutzim to embrace any or all of the recently proposed kibbutz reforms. When this generally greater fidelity to kibbutz traditions among Artzi kibbutzim is incorporated into our models (Model II), Artzi kibbutzim are revealed to show no greater resistance to any particular reform than they do to all the others. As Kanter and Ben-Ner led us to expect, distance from cities makes collective kibbutzim less likely to transform themselves into renewed kibbutzim. In this case, the effect remains significant, even when the negative relationship between rural locations and the adoption of other innovations has been controlled. Prior innovations, which we use as a negative indicator of organizational inertia, behave as expected. The more reforms a kibbutz has adopted in the past, the more likely it is to transform itself from a collective kibbutz into a kibbutz of the renewed or mixed types. Once a collective kibbutz has transformed itself into a mixed kibbutz, however, prior innovations have no further effects on whether or not a kibbutz takes the next step of becoming a renewed kibbutz. Economic crisis behaves as predicted by Oliver’s (1992) theory of deinsitutionalization. It stimulates transitions of all three types: from collective to mixed, collective to renewed, and mixed to renewed. Abramitzky

Antecedents and Consequences of the Adoption of Market-Based Compensation 241

(2008) reports a similar finding, in a cross-sectional study of the relationship between the ‘‘wealth’’ of a kibbutz and its ‘‘equality of compensation.’’ Demographic crisis stimulates transformations of collective kibbutzim into mixed kibbutzim, but has no effect on the other transitions. We think this may be due to unique features of the mixed type. The mixed form of compensation is essentially a compromise between traditionalists and reformers. It retains needs-based budgets as the traditionalists want, but also adds market-based differential payments as the reformers want. Which kibbutzim are most likely to choose this compromise? Table 5 suggests that it is the kibbutzim that are losing members, and cannot afford to lose any more of them, that are most likely to choose this form of compensation that tries to please everyone. Turning to the effects of transformations to the renewed or mixed form by other kibbutzim, the first model in each pair indicates that as expected, they stimulate all three transitions. When the number of innovations previously adopted by a kibbutz is controlled for, however, the estimated effect of transformations to the renewed or mixed forms is greatly reduced and in one case loses statistical significance. Given that most of the transitions incorporated in this measure are to the renewed kibbutz, not the mixed type, it is unsurprising that it stimulates transformations to the renewed kibbutz more strongly than it promotes transitions to the mixed type.

8. EFFECTS OF TRANSFORMATIONS ON PERFORMANCE Table 6 supports the popular impression that kibbutzim are abandoning their traditional structures in response to economic and demographic problems. In light of this evidence, it becomes appropriate to ask what effect these transformations are having on the economic and demographic challenges that led the kibbutzim to adopt them. Do they ameliorate the problems that gave rise to them, or do the kibbutzim that introduce these changes continue to decline? Of the theories of organizational transformation that were reviewed earlier in this chapter, the one that is most relevant to answering such questions is the theory of Hannan and Freeman (1984). While most prior theories of change in organizations had viewed change as rational and adaptive, Hannan and Freeman argued that because the structures and

242

RAYMOND RUSSELL ET AL.

Table 7. Effects on Economic Crisis and Number of Members, by Period. Change in Economic Crisis

Change in Members

1996–1999 2000–2004 1996–2004 1996–1999 2000–2004 1996–2004 Mixed Renewed Years since change Economic crisis Age Age squared Number of members Federation Rural Percent renewed/mixed Intercept

5.375 6.012 2.469 .370 .115 .032 .072 1.008 .071 .205 .005

1.060 1.022 .406 .263 .054 .029 .105 .224 .035 .073 .009

1.086 .397 .902 .316 .083 .006 .081 .596 .062 .023 .017

2.865 .660 1.374 .043 .104 .167 .069 .164 .040 .002 .035

2.603 3.394 .065 .038 .082 .190 .030 .184 .209 .054 .008

1.933 1.787 .311 .045 .083 .161 .057 .117 .081 .044 .018

Generalized estimating equations results with fixed effects for kibbutz and year (not shown). po.05, one tailed.

practices of organizations are institutionalized, efforts to change them are more likely to be disruptive and counterproductive than helpful. Models predicting the financial and demographic performance of kibbutzim are presented in Table 7. The dependent variables for these analyses are the change in the level of economic crisis or the number of members from one year to the next. As in Table 6, models are estimated on a panel of all kibbutzim from 1996 to 2004. The figures shown in Table 7 are maximum likelihood estimates based on a ‘‘generalized estimating equations’’ approach for panel data, using SPSS version 15. The generalized estimating equations approach is one method for adjusting for the nonindependence of observations in pooled crosssection and time-series (i.e., panel) data. It accomplishes this by, essentially, estimating the test model on the time series with each case (kibbutz), and pooling (population averaging) the results. The nonindependence of the observations in the time domain is managed by fitting ‘‘working correlation structure’’ to the residuals or prediction errors. The models include terms to show whether each kibbutz was, in the year of observation, a ‘‘collective,’’ ‘‘mixed,’’ or ‘‘renewed’’ kibbutz. When a kibbutz changes form, the ‘‘mixed’’ or ‘‘renewed’’ variable in Table 7 reflects the effect of that change on the dependent variable in the first year, net of other effects. The models also include a variable that counts the number

Antecedents and Consequences of the Adoption of Market-Based Compensation 243

of years since a transition occurred, and scored as zero for all observations of a kibbutz in which no change ever occurred. The coefficient of this variable reflects the effect on performance of each additional year since the change occurred, adjusting for whatever initial shock may have occurred in introducing the change, and all other modeled factors. Greve (1999) notes that if organizations that perform poorly by one or more criteria are the organizations most likely to introduce change, then change is likely to be followed by improvements in performance, simply due to the statistical phenomenon of ‘‘regression toward the mean’’ (Greve, 1999). In this study, we attempt to mitigate this problem by including the level of economic crisis and size of the organization as predictors in our models. As in Greve (1999), the coefficients relating the current year’s values of these variables to those for the following year are negative in both models. Preliminary analyses disclosed that the consequences of transformations differed significantly between the earlier (1996–1999) and later (2000–2004) periods of the diffusion of this innovation. To capture differences between early and late adopters, Table 7 presents results separately for each period, before reporting results for the entire period 1996–2004 in the third column in each group. The first column of Table 7 shows that in the period 1996–1999, transformations to the mixed or renewed types of kibbutz were followed by significant reductions in economic crisis in the year following the change. In the same period, the number of years since the change had an opposite effect, causing levels of economic crisis to rise. The net result of these two counteracting influences is that two or three years after the first year, the level of crisis had climbed back to where it was at the time of the change. In the years 2000–2004, this favorable short-term effect of transformations on economic performance was not repeated. Turning to the effects of transformations on the number of members, Table 7 indicates that the transition to a mixed kibbutz led to an average loss of 2 or 3 members in the first year after the change in both time periods. For kibbutzim that transformed themselves into renewed kibbutzim, the effects on demographic performance differ by period. Kibbutzim that became renewed kibbutzim in the years 1996–1999 did not experience significant losses of members in the year immediately following the change and gained an average of 1.374 members in each subsequent year. In 2000–2004, however, transitions to renewed kibbutzim were followed by the loss of an average of 3.394 members in the first year, and no offsetting gains in later years.

244

RAYMOND RUSSELL ET AL.

Regarding other variables included in these analyses, the age of a kibbutz had a curvilinear effect on change in size, indicating that kibbutzim of medium age were gaining members, while the oldest kibbutzim were losing members. Rural kibbutzim lost members at a faster rate than urban kibbutzim, especially in the more recent years. Prior introduction of other reforms helped reduce levels of economic crisis 1996–2004 (.017) but was also followed by additional losses of members (.018), especially in the earlier years (.035). Increasing percentages of kibbutzim having previously transformed themselves were associated with reductions in economic crisis in the period 1996–1999, but led to worsening of economic performance in the years 2000–2004. When one judges by the number of members gained or lost after transformations, previous transformations by other kibbutzim have a significantly positive effect on the number of members, but only in the later years, not 1996–1999.

9. CONCLUSIONS Earlier in this chapter, we noted that theories of democratic and communal organizations and more general organizational theories make contradictory predictions about the effects of three influences: organizational age, organizational size, and economic condition. In the case of size and economic condition, the effects observed are those predicted by theories of organizational inertia and deinstitutionalization. In the case of age, we report a positive but curvilinear effect. This finding is consistent with the predictions of Batstone and can therefore be taken as support for his theory. This curvilinear effect of age can with equal justice be interpreted as providing partial support for each of the competing theories of the relationship between organizational age and transformations, depending on which ranges of age one focuses on. At low values of age, older organizations are more likely to undergo transformations, as classic theories of democratic and communal organizations predict. At higher values of age, the oldest organizations are less likely to undergo transformations than those of intermediate age, as Hannan and Freeman’s (1984) theory of organizational inertia predicts. While head-to-head confrontations between these two bodies of theory produce mixed results, when one takes into the account the influence of all of the variables in Table 6, it is the general organizational and institutional processes whose effects are most prominent. If one asks whether these changes can best be characterized as the ‘‘degeneration’’ of the kibbutz,

Antecedents and Consequences of the Adoption of Market-Based Compensation 245

the ‘‘decommunalization’’ of the kibbutz, or the ‘‘deinstitutionalization’’ of the kibbutz, they are clearly all of these things, but it is processes of deinstitutionalization that predominate. The collective kibbutz had been institutionalized in the twentieth century, both by being embodied in the customs and traditions of individual organizations, and by being standardized and legitimated by the kibbutz movement as a whole. These traditions broke down first in individual organizations under the impact of resource shocks, and then later in larger numbers of kibbutzim, with the examples set by other kibbutzim as an increasingly important accelerant. The view of kibbutzim as institutionalized organizations with which we emerged from Table 6 is reinforced by the results shown in Table 7. Because the collective kibbutz had been institutionalized, efforts to transform the kibbutz are disruptive, costing an average of two members per kibbutz. And for similar reasons, the changes in compensation examined here have not succeeded in improving the economic performance of the kibbutzim that adopt them. Putting together both sets of results, it is clear that the kibbutzim have been changing, because they are losing both money and members, and because they see other kibbutzim adopting the same change. Once a given kibbutz begins to introduce changes, one change leads to another, with the result that most kibbutzim have now transformed themselves into kibbutzim of a completely new type. But these transformations have not succeeded in solving the problems that they were introduced in response to. Given that as of our last year of observation, the kibbutzim were continuing to lose both money and members, the series of changes that began in the 1990s can also be expected to continue.

REFERENCES Abramitzky, R. (2008). The limits of equality: Insights from the Israeli kibbutz. Quarterly Journal of Economics, 123(3), 1111–1159. Batstone, E. (1983). Organizations and orientation: A life cycle model of French co-operatives. Economic and Industrial Democracy, 4, 139–161. Ben-Ner, A. (1987). Preferences in a communal economic system. Economica, 54, 207–221. Ben-Rafael, E. (1997). Crisis and transformation: The kibbutz at century’s end. Albany: State University of New York Press. 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.

246

RAYMOND RUSSELL ET AL.

Estrin, S., & Jones, D. C. (1992). The viability of employee-owned firms: Evidence from France. Industrial and Labor Relations Review, 45, 323–338. Gavron, D. (2000). The kibbutz: Awakening from utopia. Lanham: Rowman & Littlefield. Greenwood, R., & Hinings, C. R. (1996). Understanding radical organizational change: Bringing together the old and the new institutionalism. Academy of Management Review, 21, 1022–1054. Greve, H. R. (1999). The effect of change on performance: Inertia and regression toward the mean. Administrative Science Quarterly, 44, 590–614. Hannan, M. T., & Freeman, J. (1984). Structural inertia and organizational change. American Sociological Review, 49, 149–164. Kanter, R. M. (1968). Commitment and social organization: A study of commitment mechanisms in utopian communities. American Sociological Review, 33, 499–517. Kraatz, M. S., & Zajac, E. J. (1996). Exploring the limits of the new institutionalism: The causes and consequences of illegitimate organizational change. American Sociological Review, 61, 812–836. Manor, R. (2004). The ‘‘renewed’’ kibbutz. Journal of Rural Cooperation, 32(1), 37–50. Mill, J. S. (1909). Principles of political economy. London: Longmans, Green, and Co. Near, H. (1997). The Kibbutz movement: A history. Volume 2, Crisis and achievement 1939–1995. London and Portland, OR: Vallentine Mitchell. Oliver, C. (1992). The antecedents of deinstitutionalization. Organization Studies, 13, 563–588. Palgi, M. (1994). Attitudes toward suggested changes in the kibbutz as predicted by perceived economic and ideological crises. Journal of Rural Cooperation, 22(1–2), 113–130. Palgi, M. (2004). Organizational change and ideology: The case of the kibbutz. International Review of Sociology, 12(3), 389–402. Pitzer, D. E. (1989). Developmental communalism: An alternative approach to communal studies. In: D. Hardy & L. Davison (Eds), Utopian thought and communal experience (pp. 68–76). London: Middlesex University. Rosner, M., & Tannenbaum, A. S. (1987). Organizational efficiency and egalitarian democracy in an intentional communal society: The kibbutz. British Journal of Sociology, 38, 521–545. Russell, R., Hanneman, R., & Getz, S. (2006). Demographic and environmental influences on the diffusion of changes among Israeli kibbutzim. Research in the Sociology of Work, 16, 263–291. Selznick, P. (1949). TVA and the grass roots. Berkeley: University of California. Simons, T., & Ingram, P. (1997). Organization and ideology: Kibbutzim and hired labor, 1951–1965. Administrative Science Quarterly, 42, 784–813. Stinchcombe, A. L. (1965). Social structure and organizations. In: J. G. March (Ed.), Handbook of organizations (pp. 142–193). Chicago: Rand McNally. Topel, M. (2005). The new managers: The kibbutz changes its way (Hebrew). Ramat Efal: Yad Tabenkin. Tushman, M. L., & Romanelli, E. (1985). Organizational evolution: A metamorphosis model of convergence and reorientation. Research in Organizational Behavior, 9, 171–222. Weber, M. (1978). Economy and society. Berkeley: University of California. Yoffe, D. (2005). Concepts in analyzing financial reports of a kibbutz (Hebrew). Tel Aviv: Department of Economics, The Kibbutz Movement.

FROM DESTRUCTIVE TO CREATIVE TRADE THROUGH ECONOMIC DEMOCRACY Jaroslav Vanek We live in a capitalist world where most people believe that globalization based on free trade and free markets in all areas is the optimal or more efficient state. In this chapter, I would like to accomplish the following: 1. Show the fallacy of the optimality conclusion 2. Demonstrate the validity of an alternative analysis of today’s processes of globalization – the theory of destructive trade 3. Propose a remedial procedure that would lead to a globalization that would be optimal 4. Indicate in both theory and practice solutions that already exist in today’s world and that approach optimality.

1. THE FALLACY OF GLOBAL OPTIMALITY During the great depression of the 1930s, facing drastic unemployment and declining incomes, many of the European countries affected engaged in protecting their national markets through import protection. This beggarthy-neighbor policy gave the economists’ profession an impetus to show Advances in the Economic Analysis of Participatory and Labor-Managed Firms, Volume 11, 247–253 Copyright r 2010 by Emerald Group Publishing Limited All rights of reproduction in any form reserved ISSN: 0885-3339/doi:10.1108/S0885-3339(2010)0000011014

247

248

JAROSLAV VANEK

what was happening and to conclude that restriction of free trade – the beggar-thy-neighbor policy based on repetitive retaliation – is detrimental to the economy, and thence the conclusion and recommendation of free trade. This notion remains to this day the ruling principle of most capitalist economies based on principles of modern liberalism. But we are about to show that there is a fundamental flaw in this strategy – a flaw perhaps also indicated by the fact that many countries and social groups are violently opposed to such a strategy. As all students of neoclassical economics know, the theoretical model of supremacy of free international trade is based on some necessary conditions or assumptions, most significantly (1) full employment of productive resources, including labor and (2) a tendency toward factor price equalization. In fact, the pure theory of trade leads to complete factor price equalization, and in some situations such equalization is even the essence of the proof of comparative advantage (e.g., see Vanek, ‘‘The Factor Proportions Theory: The N-Factor Case,’’ Kyklos, October 1968). In the real world, we live in the two central assumptions that are most dramatically violated – and yet the capitalist economists are using a theory based on them in calling for and supporting free trade globalization. Even the Scriptures – and Our Lord – were aware 2,000 years ago that the poor will always be with us. Indeed, in our world of the 21st century, wage rates and incomes can be some one to 1,000 percent divergent between the poorest members of the human family and the most affluent labor force in different parts of the world. The second postulate of full employment of resources – that is labor – is equally dramatically violated. Without elaborating on these obvious facts, let us see a little more closely what happens when a simple theory is constructed that recognizes these two conditions of the real world.

2. A REVISED THEORY: DESTRUCTIVE TRADE Some years ago, aware of these fundamental facts I wrote a study entitled ‘‘Destructive International Trade: From Justice for Labour to Global Strategy.’’ The study was published by the Jan Tinbergen Institute of Development Planning, Rohtak, India, as a special issue of its International Journal of Development Planning Literature (Vol. 13, No. 1, January–March, 1998) because the capitalist liberal economists would not recognize its validity here in our Western countries, even though little children would have no difficulty with its simple logic. While I use some basic geometrical

From Destructive to Creative Trade Through Economic Democracy

249

analysis in my small book, I will substantiate my argument in a simple verbal manner that everyone can understand, as follows: Suppose there is a capitalist enterprise somewhere in the United States, representing a 10-dollar per hour wage world, producing tons of gadgets, 1 ton per worker, and there are 1,000 workers in the firm. The owner or owners of the enterprise can lay off the 1,000 workers and produce the same 1,000 tons of gadgets in the following way: He retains 100 of the workers (the 900 remaining unemployed) and with these he produces 100 tons of food, and exports the food to a low-wage country where wage rates are only 1 dollar and where 10 workers can be supported by 1 ton of food. And thus the entrepreneur from the United States disposes of a labor force of 1,000 workers in the poor country. These workers can, using machinery shipped from the United States to the poor country, produce the same 1,000 tons of gadgets. And these can be exported to the United States, and the US entrepreneur has the same output as before, and in this simplified theoretical model, he has a profit equal to 900 tons of gadgets. Of course, this is an oversimplified model, but it shows or can be used in describing several conclusions: 1. There is an increase in unemployment in the 10-dollar countries. 2. There is increase of employment in the poor countries, but hardly an increase in wage rates as there is a virtual infinity of labor availability there. 3. Of course, there are inefficiencies in the whole situation, such as lower productivity of workers in the poor world, cost, and ecological deterioration through long distance transportation, etc. 4. This is very significant although not immediately recognizable; namely, the profit motive (the 900 tons of gadgets) remains, even if reduced, as long as the inefficiencies noted are enormous, down to the point where only 100 tons of gadgets are recuperated from the international transaction. Thus, having recognized the two major alterations of the classical theory corresponding to the situation in the real world (absence of full employment and wide differentials in wages represented by the 1,000 percent in our simple example), we come to the conclusion that trade will exist, generate unemployment in the advanced economies, produce all kinds of serious inefficiencies, but increase the relative gains and incomes of the capital owners and entrepreneurs. The neoliberal capitalist promoters using fraudulently the traditional theory of comparative advantage tell us that while some jobs will be lost in

250

JAROSLAV VANEK

the import-competing industries, other jobs will be gained in the export industries. But applying our theory of destructive trade as long as the two fundamental assumptions are altered, we must conclude that in equilibrium, or a final long-term solution, all that is transportable will be produced in the low-wage part of the world, and there will be a tendency to unemployment. The only remaining jobs in the rich part of the world will be those in services (workers in hamburger industries or construction cannot commute from Bangladesh, although many will try to from Mexico at the cost of their lives) and in armaments protected by law or in the defense sector. Thus wars and the military will be (or are) the salvation of the affluent countries and of the affluent capital owners there, eating hamburgers, and building and living in expensive homes. Some construction jobs will also be created in constructing more modest homes, many of which are to be foreclosed under the subprime mortgages. That these tendencies exist in capitalist markets of advanced economies such as the United States is confirmed by the reactions of labor unions to jobs leaving for China, the upheavals facing the WTO and others. The destructive trade tendencies were partly covered up by virtually zero real interest rates practiced for many years by Mr. Greenspan (as Chair of the United States Federal Reserve Bank) permitting the banking and credit institutions to introduce subprime instruments stimulating cancerous building expansions, and of late the foreclosure calamities in the United States, and banking bailouts around the world.

3. REMEDIAL ACTION Of course, the final equilibrium state of destructive trade is a dismal vision, which will never be realized fully under real world conditions. But it indicates the destructive forces in today’s globalization and confirms some of the tendencies actually observed. For a greater precision of analysis and especially for a more careful analysis of remedial policies, the reader may turn to my study noted above. But the purpose of this writing is to show the fundamental causes of the suboptimal state, and using such analysis, indicate how an optimal globalization could be attained, and more important, show that germs of such optimality already exist in our world. In my Unified Theory of Social Systems (available on Internet, http:// hdl.handle.net/1813/642), I attempted to define true optimality of any social system, economic, political, family, religious or other, in terms of forms of

From Destructive to Creative Trade Through Economic Democracy

251

participation and participative processes. The reader may want to refer to the study for more detail. Here let me just use a subset of the optimality conditions to indicate that the crisis outlined above in the first two sections is closely linked to the nonfulfillment of theoretical optimality. The first and most relevant condition is that in all social situations people, or members of society, ought to participate in decision or determination according to both intensity (such as one person one vote, or one share one vote in capitalist firms) of their involvement in a given social situation, but also according to the nature or quality of the involvement. And the various types of qualities or natures must be ranked in nonconvertible higher and lower rankings. For example, the mothers or wives of soldiers killed in Iraq are vitally and parentally and lovingly involved in their children, and the decision as to whether they should participate in a war of aggression should be dominated by the relatives and not by a misguided president. Similarly, those suffering from the subprime mortgage crisis in losing their homes and lives’ happiness are vitally involved in the strange if not criminal credit arrangements and should dominate the control and regulation processes exploiting their ignorance. Similarly the capitalist system is based on the principle of capital control of firms and employment. And yet the capitalists’ involvement typically is indirect and often atomized (in such portfolios), where the owners never even saw the firm they own. By contrast, the workers in the firms are directly and often vitally involved in their places of work often for long portions of their lives – and yet they have no say in decisions concerning their livelihood. Worse than that, the capitalist principle of profit maximization is tantamount to wage and income minimization as long as such minimization can be exercised. The case of destructive trade, the first principal subject of this study, is the extreme and most dramatic situation of such minimization: namely reduction to zero for those who lose the jobs in the 10-dollar country, and minimization to 1 dollar in the 1-dollar country of our theoretical example. The direct and often vital involvement of the workers in the firm should take precedence over the indirect and possibly atomized involvement of the owners of modern corporations. Yet these principles are violated most dramatically in the situation of destructive trade, which rules our modern world of globalization. This analysis and application of the unified theory of optimality of social systems indicates the problems of the world we live in and of capitalist globalization. It also indicates the way or ways out of the dilemma.

252

JAROSLAV VANEK

4. CONCRETE COOPERATIVE ACTION Now consider a firm in the advanced (10-dollar) economies where as called for by the unified theory of social systems for optimality, the firm is a worker cooperative or some other form of worker control or ownership. The worker managers of the firm will also be aware of the 10-to-1 differential in the world, but they will try to defend the interests of their workers to preserve their employment. It would be absurd to move the firm with its workers to the 1-dollar world! Rather they will engage in some action that would protect their jobs and livelihood. They could create a powerful lobby that would convince the government to impose protective tariffs on products that could come from abroad. Such policy is one of the options in the capitalist world facing destructive trade and that option is discussed as one possible strategy in my study on destructive trade. But this will not concern us here as we are speaking about a situation of optimal globalization under the precepts of the unified theory of social systems. The strategy is as follows: Our democratic firm obviously has a significant – if not absolute – advantage in having, knowing and advancing its technology, organization, user markets for its products, etc. This way it can move to any part of the lower-wage world and start a new firm using all the comparative advantages it has. It can now use its more labor-intensive machinery and equipment and ship it to the new offspring firm where labor presumably is abundantly available, and produce in the advanced economy with new machinery that is less labor demanding. It can cooperate with the offspring and engage in all kinds of division of labor. Perhaps more importantly, it will give employment to workers in the poor country and a possibility of somewhat higher than the ‘‘1-dollar’’ wages. Most likely, it will give them new technical training and education. Then there is the question of the efficient or optimal organizational and legal form. The parent firm coming with its project will have to incorporate somehow in the host country. Obviously it cannot come to a place in the host country and declare the new offspring firm a democratic cooperative. The new workers are happy to have well-paying jobs, and they hardly have any notion about a democratic firm. Consequently, the new firm will have to be incorporated under local law, presumably as a limited liability corporation of some sort. But the founder’s firm, the parent cooperative, wants to preserve its democratic structure, at least in the long run. Here again the unified theory of optimality gives us a helping hand: the parent firm becomes parentally

From Destructive to Creative Trade Through Economic Democracy

253

involved, similarly to the involvement of parents toward a new baby. The baby – or the new offspring firm – cannot manage itself; it is the parent or parents who must assume the task, under the new form of local law incorporation. But as with maturation in the domain of the human family, perhaps in 20 years or so, the infant or offspring must grow into a mature democratic or cooperative form, akin to the nature of the parent. How this will be accomplished, no one knows yet, because it will depend on the participation of both, the parent and the maturing offspring, through an educational and praxis-based process, defined but not given also in my Unified Theory of optimal and suboptimal states. Perhaps the most significant thing is that if the process of regeneration and reproduction can be sustained in a democratic context, from parent to child and from parent to child again, the whole economy and society can gradually be transformed from suboptimal to optimal, truly and fully democratic form, where even the transportation and low labor efficiency losses noted in Section 2 will disappear and so will the horrendous profits and maldistribution of income, represented in our theory of destructive trade by the 900 tons of gadgets.