215 72 3MB
English Pages 240 [238] Year 2011
REPUTATION-BASED GOVERNANCE
R E P U TAT I O N - B A S E D GOVERNANCE
Lucio Picci
S TA N F O R D ECO N O M I C S A N D F I N A N CE An Imprint of Stanford University Press Stanford, California
Stanford University Press Stanford, California ©2011 by the Board of Trustees of the Leland Stanford Junior University. All rights reserved. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or in any information storage or retrieval system without the prior written permission of Stanford University Press. Printed in the United States of America on acid-free, archival-quality paper Library of Congress Cataloging-in-Publication Data Picci, Lucio, 1965– Reputation-based governance / Lucio Picci. p. cm. Includes bibliographical references and index. ISBN 978-0-8047-7329-4 (cloth : alk. paper) 1. Political planning. 2. Reputation—Political aspects. 3. Internet in public administration. 4. Transparency in government. 5. Political participation. I. Title. JF1525.P6P53 2011 352.3'4—dc22 2010035569 Typeset by Westchester Book Group in 10/13.5 Sabon Cover image: Palácio do Alvorada, Brasilia
a Marghe, Ale, e Cate
Contents
Preface 1 2 3 4 5 6 7 8 9 10 11
Introduction Reputation and Trust Reputation and Good Public Governance Entities, Roles, and Functions of Reputation-Based Governance Computing Measures of Reputation The Production of Statistical Information and the Analysis of Policies Managing Policies: Accountability, Rent-Seeking, and Corruption Applications of Reputation-Based Governance Interdependence Between the Choice and Execution of Policies Reputation-Based Democratic Participation Final Considerations
Notes Bibliography Index
ix 1 20 45 57 77 92 110 123 144 160 176 183 199 213
Preface
In this book, I discuss the role and the relevance of reputation in governance, and argue that an intelligent use of widely available Internet technologies would strengthen reputational mechanisms and significantly improve public governance. I will propose a governance model that is “reputation-based”—hence the title of the book. The study I present offers an analysis of public governance and public administration and is multidisciplinary. Overstepping the barriers that separate different disciplines presents risks, and my first challenge in writing this book was to guarantee its intelligibility. In this respect, my goal was to provide a discussion that is rigorous, but accessible to experts from different areas. As a result, the reader may occasionally find explanations of concepts that are well known. In those instances, I beg the reader for patience, and I grant permission to overlook the few lines or pages in question, as I fully endorse Daniel Pennac’s “Reader’s Bill of Rights,” Rule no. 2: The right to skip pages.1 The second basic challenge to writing a multidisciplinary book is, of course, for the writer, who should be at ease in each of the disciplines involved. Though I have several years of experience in the relevant fields, this is no reason for confidence. I have, however, been fortunate enough to benefit from the comments, advice, and criticisms of several colleagues and friends who have certainly helped me to correct at least some of my shortcomings.
x
Preface
The scope of the book explains the long list of intellectual debts that I accumulated over the years. Some of the ideas presented here were initially discussed at meetings that took place more than a lustrum ago at the Astrid think tank in Rome, benevolently presided over by Franco Bassanini. The issue of reputational incentives for public procurement allowed for lively discussions within the scientific committee of Consip (the Italian public procurement agency), of which I was a member between 2002 and 2005. Consip’s president in those years, Gustavo Piga, contributed to creating a lively intellectual environment where new ideas were always welcome. Many thoughtful colleagues made useful criticisms or brought to my attention work that contributed to improvements in the argument. They include Evelyn Brodkin, Marco Casari, Osvaldo Croci, Giampiero Giacomello, Nicholas Gruen, Raimondello Orsini, and Susan Rose-Ackerman. Others gave me their comments after reading a draft of the whole book. My gratitude goes to Brunetta Baldi, Marco Bertamini, Giliberto Capano, and Gennaro Zezza for putting up with a manuscript the bolts of which still needed tightening. A particular thank-you for their very attentive reading and for the sometimes crucially valuable suggestions that followed goes to Fred Thompson and an anonymous reader of a previous version of this book. Over the years, I have enjoyed conversations on different aspects of governance with Miriam Golden, with whom I coauthored more than one paper. One of these was an article that we published in 2005 in Economics and Politics, which is amply cited in Chapter 7 of this book. My gratitude to her oversteps the boundaries of this book, on which I received her thoughtful comments, because of the many lessons in political science that she has taught me over the years. In 2007, I was able to make real progress on this book after I temporarily abandoned the University of Bologna to move to Seville, Spain, where I joined the Institute for Prospective Technological Study (IPTS), part of the European Commission’s Joint Research Centre. IPTS, conveniently placed in a sort of meteorological heaven that reminded me of my years as a PhD student in Southern California, proved to be the perfect place to conclude, in a lively (intellectual) environment, a first draft of this book. IPTS, being part of the European Commission, is a privileged observatory for policy making and, for me, it was a salubrious submersion into the real world away from the abstractions of academia. I thank David
Preface
xi
Broster, head of the Information Society Unit at IPTS, for cheerfully accommodating the different souls working there—sometimes even the different souls within each one of us. The result was an environment characterized by multidimensional diversity: of approaches, sometimes of objectives, and of disciplines—in short, the ideal place to carry out a project that has many facets and implications. From IPTS, I would particularly like to thank Andrea De Panizza, Stefano Kluzer, and David Osimo. I single out Marc Bogdanowicz, who was my direct boss at IPTS, for his special role in this project and for his essential contribution in making my stay in Seville an extremely pleasant and fruitful one. This book has an evident normative slant, as it suggests that “reputation-based governance” would represent a step forward with respect to the models of governance in use today. To give flesh to this idea, a software demonstrator that implements a particular application of reputation-based governance has been developed. It serves two main purposes: first, it constitutes a test bed for the proposed governance model, and second, it gives interested people a firsthand account of how a particular model of reputation-based governance would work. The software, named Rebag-Ware, is accessible on the Web site www.rebag.it. Readers are encouraged to spend some time using it. However, those who are not interested in playing around with an Internet application should not worry: the book is fully self-contained. The software has been managed by some fine programmers at the Forlì Branch of the University of Bologna. My thanks go to Roberto Confalonieri, the main programmer; Cristiano Leoni, the project manager, Matteo Pompoli, the Web designer, and Alessandro Bravi, who helped in various ways. The book greatly benefited from the advice and experience of Stanford University Press Acquisitions Editor Margo Beth Crouppen. Margo proved that an impeccable professionalism can go hand in hand with a human touch that I very much appreciated. I would also like to thank Judith Hibbard, senior production editor at Stanford University Press, for guiding me through the production process; Barbara Goodhouse, production editor at Westchester Book Group, and Susan Burke, copy editor, for the meticulous editing of the manuscript. Last, I would like to thank Patricia Farrer, who with much care and good humor corrected the shortcomings of my prose, written in a language that is not mine. When, almost two hundred years ago, the Italian writer
xii
Preface
Alessandro Manzoni decided to put his new manuscript into impeccable language, he chose to travel from his city, Milan, to Florence—arguably the center of irradiation of the Italian language. There, as he put it, he “washed his clothes in the water of the Arno,” the river that flows through Florence. My prose needed similar treatment. It is telling of our sometimes confusing age that, on this occasion, an Italian’s linguistic clothes got their English-style washing almost literally on the banks of the river Guadalquivir, in the south of Spain. The world has changed, and changes to the world is what this book is about. Manzoni’s manuscript would become The Betrothed (compulsory reading for generations of Italian students)—a book that some consider a masterpiece, and others (guess who), the dullest tome ever written in the Italian language. In this respect, the world has not changed much: there are good books, and there are mediocre ones. I hope that the reader will like mine.
REPUTATION-BASED GOVERNANCE
1
Introduction
Today we work for a reputation. Tomorrow our reputation will work for us. Russian saying
EBay and Other Stories
EBay’s electronic market is one of the most successful Internet applications, generating 15 million new listings on the average day and allowing more than 90 million users worldwide to buy and sell all sorts of new and secondhand items (eBay 2009, 2010). On eBay, traders almost invariably do not know each other and usually will not have an encounter more than once. It would be easy to cheat on eBay—for example, you could fail to send the merchandise after having received the requested payment, or send a product of lower value than the one advertised, or change your mind on a purchase and not follow it through. If dishonesty were widespread, people would not trust each other and would not use the market. Beyond a certain level of mutual mistrust, eBay would not function. However, an essential characteristic of eBay prevents this from happening. At the end of a transaction, both buyer and seller may (and often do) write a “feedback” (as it is called) on each other, which can be positive, neutral, or negative. The sum of each type of feedback received by each participant during the previous six months forms what amounts to an “index of reputation” that is visible to all. The availability of such an index provides a strong incentive to be honest: a seller who receives negative feedbacks from previous buyers would find it difficult to continue
2
Introduction
trading, because prospective buyers would be wary of doing business with him. A buyer would also have a hard time doing his shopping, because sellers would not believe in his resolve. On eBay, fear of acquiring a bad reputation represents a strong incentive to be honest and efficient, and the observed outcome is that most people, in fact, honestly describe the merchandise that they plan to sell, ship it quickly, and pay their bills. The result is the success of eBay, one of the poster cases of the Internet age. Beyond what may be concluded from impressionistic observations, rigorous studies have assessed the role of reputational effects, using data on eBay transactions together with appropriate econometric techniques. Resnick et al. (2006, table 1) summarize fifteen such studies. The broad picture that emerges is one where such effects exist and are relevant. More recent research also points in the same direction. For example, Resnick et al. (2006) set up an experiment where an established dealer with a good reputation is compared with a new dealer, with no reputation at all. They both provide exactly the same good and quality of service. The authors find that the seller with an established reputation enjoys a significant price premium. Cabral and Hortaçsu (2010) find that when a seller receives a negative feedback for the first time, his sales drop significantly, and they conclude that overall “the eBay reputation system gives way to noticeable strategic responses from both buyers and sellers.”1 The case of eBay is a good starting point for our inquiry into the role of reputation in public governance. True, the outcome that we observe—a viable and thriving market—does not apply to public governance but to the private domain. It also turns out that there are more examples available of private governance where reputation plays an important role. One of the main theses of this book is that this divergence occurs precisely because the role of reputational considerations in public governance today is not as important as it could, and should, be. However, the eBay example does hint at two issues that we will encounter over and over again in our reasoning on public governance. First, reputational considerations may induce people to act in useful ways even without the presence of a formal institution threatening to punish them should they misbehave. What induces most users of eBay to be honest is not the fear of the police knocking at their door, should they cheat. For many, and possibly for most, an interiorized sense of honesty may certainly play a role in
Introduction
3
this respect. However, the presence of eBay’s widely visible feedbacks, and their consequent reputational effects, has a much more compelling role in guaranteeing the viability of eBay as a marketplace. Secondly, the success of eBay hinges upon the presence of a communication technology: Internet and the Web. In general, all solutions to governance problems need appropriate technologies in order to function. For example, the Roman Empire would not have existed as we know it without its network of roads and an efficient postal system, allowing the transmission of information and orders from the capital to the legions stationed in the provinces. Today’s aviation technology, which permits, for example, government representatives to meet frequently, is an integral part of the system of international relations. In the case of eBay in particular, the Internet is used to transmit information on the reputation of traders. Also, the eBay case illustrates that this reputational information typically has to be appropriately organized: though individual comments that people post on concluded transactions are accessible to all, it is the aggregation of information provided by a reputation index that is most useful. We should exercise care when considering the role of technology within a governance model. If the Internet, or something similar to it, were not available, eBay could not exist. However, while being able to record information and to communicate is indispensable for reputation to play a role, obviously reputation information can also spread using different and much less sophisticated technologies. The following description of a well-studied historical episode serves as a convincing example of the relevance of informal reputation considerations in very different technological contexts. In the early Middle Ages, a group of Jewish traders, originally based in Egypt, formed a coalition to better pursue their business (see Greif 1989 and 1993, on which this account is based). We know quite a lot about the dealings of the Maghribi traders, as they are known, because they are documented in an archive found in Fustat (Old Cairo, in Egypt), which contained letters and accounts of various types. The merchants lived scattered in several trade centers in the Muslim area of the western Mediterranean, and they helped one another in their work. Each trader could act either as a merchant or as a merchant’s agent, in this case
4
Introduction
supplying the merchant with the services needed to operate long-distance trade, such as loading and unloading the ship, paying customs and transportation fees, and so forth. Depending on the occasion, one member of the coalition acted as a merchant or as an agent. For a merchant, operating through agents had some important advantages over taking his own goods to market, because it kept costs down and reduced the overall risk, as many parallel undertakings could go on in different markets. Where long-distance trade was administered through agents, however, there was ample opportunity for cheating, and the anticipation of widespread dishonesty would have been enough to discourage any trade. Just as eBay would be unviable as a market if potential participants did not trust each other enough, so it would have been for the activities of the Maghribi traders. As Avner Greif (1993) wrote, “To gain from cooperation, there was a need for an institution capable of surmounting this commitment problem, an institution through which an agent could commit himself ex-ante, before receiving the merchant’s capital, to be honest ex post.” The legal system of the time was not capable of offering effective protection to merchants, and there was no formal institution that was capable of forcing agents to respect pacts. The solution the Maghribi traders came up with was to inflict a collective punishment in the form of a boycott on agents who cheated. Agents were also allowed to cheat those merchants who had been dishonest, without themselves being subject to collective punishment, thus reinforcing the threat to cheaters. The existence of this “reputational equilibrium,” supported by historical documents, allowed the Maghribi traders to prosper. The example of the Maghribi traders, to which we will return in the next chapter, illustrates that successful reputational mechanisms may use very different communication technologies. In fact, it has been suggested that the Internet “digitalizes word-of-mouth” (Dellarocas 2003), thus enhancing the possibilities of passing information (which traditionally was communicated informally) to others on the behavior of people or organizations. In the past, before choosing a restaurant or a hotel, we asked friends in the know for a tip, but now we access specialized Web sites providing ratings by former customers, such as www.zagat.com for restaurants or any of the many online hotel booking services available. In all
Introduction
5
cases, the presence of a communication channel, transmitting information about past performances, stimulates the provision of a good service. However, if we simply relegate the Internet to the category of instruments that allow us to digitize activities that humans have always carried out—for example, exchanging information on the performance of people and institutions—we run the risk of missing the novelty that it represents in this and many other respects. Today, it is not only the availability of Internet technologies that permits the working of information systems, but also the fact that they treat digitized reputational information in a coherent manner, compute useful measures of reputation, and disseminate the relevant information to all concerned parties, all at negligible cost. Also, one distinctive trait of these systems is that they allow the gathering and processing of reputation information to be specifically designed, unlike the spontaneous informal arrangements of the past. Today, reputational systems may be engineered far more than before, because their working is mapped into the architectural characteristic of a technological artifact, an Internet-based information system. The emerging possibility of crafting reputational systems to serve particular governance goals is an important theme of this book to which we will dedicate much thought. The example of the Maghribi traders is evidence of the fact that reputational considerations have always played an important role in governance and, more generally, in human affairs. We can distinguish between two main beneficial effects of reputation: one that we may call “static” and the other “dynamic.” Let us start with the static effect. A good reputation, for the owner, is the result of past behavior and is analogous to a capital good, which can be kept and increased by respecting one’s obligations, but also easily wasted when shortsightedness prevails. Rational individuals are constantly worried about their reputations, and such worries are reflected in the working of the governance institutions to which they contribute. We think twice before deciding to squander, in a moment of madness, a string of past investments in socially acceptable behavior. The ancient Romans understood this well when they crafted the adage semel in anno licet insanire, or “it is acceptable to be crazy once a year,” which does grant the right to insanity, but only within well-specified
6
Introduction
limits on the number of transgressions allowed. In this respect, self-control, which in part is certainly motivated by reputational considerations, translates into predictability of human behavior, which in itself is an important prerequisite for most social intercourse. Implicit in the private calculations people make on what course of action to take is a weighing up of the short-term gain that they could obtain from dishonest and predatory behavior, with the long-term damage that would follow from losing the trust of their acquaintances. The more important reputational considerations are, the more likely it is that the balance will tilt in favor of virtuous behavior. When this happens, the individual actors of governance are better off, and often society as a whole also benefits. We can define this as a static effect of reputation, in the sense that it applies to a given situation, that is, to the relevant actors as they are at a given time. Reputational incentives, however, also have a dynamic effect, in that over time they tend to cause an improvement in the characteristics of the relevant actors. They do so in two distinct ways. By rewarding good performances, they encourage actors to invest resources in improving their skills. If professionals know they will move up the career ladder if they do well in the job, they may decide to spend evenings taking classes to learn new skills. On the other hand, if they know that all they need are good connections, then they are quite likely to spend their evenings in networking activities. One effect of the presence of reputational incentives is to curb unproductive, rent-seeking activities, as economists call them. Another, more radical way in which reputational incentives have a dynamic effect on improving the quality of the actors of governance is by supporting a process of selection. Being able to observe past performances helps us to discriminate between high- and low-quality suppliers. For example, given a choice of movies with similar plots, we would probably choose one directed by Martin Scorsese rather than one directed by some unknown director. Similarly, we would rather buy a car manufactured by Audi than one produced by a less reputable firm, even though the appearance and price tags were similar. Such reputational effects set in motion selection forces that weed out the least fit from the competition. Film directors who are unable to secure an audience for their creations have to
Introduction
7
find new jobs, and uncompetitive firms lose importance and, eventually, go out of business. One consequence of these static and dynamic effects is that they also lead to more predictable human interactions—if I can read the strategic considerations that inform the choices of others, I can predict them better. An alternative route to achieving better predictability of behavior is through contractualization, with courts intervening when necessary to force people to respect their obligations. Indeed, large parts of our lives are influenced by contracts of various types, as the high—and possibly increasing—litigiousness of our societies demonstrates. However, courtrooms and hefty lawyers’ fees provide only a partial solution to the problem. Most relationships are not easily contractualized because they are fraught with unforeseeable contingencies and are simply too complex to be fully described in writing. Also, the transaction costs involved in a contract may be too high, and, last but not least, the courts may be unavailable, as we saw in the case of the Maghribi traders, or ineffective. The desire to maintain and improve one’s reputation, and to build or safeguard trusting relationships with others, in many instances constitutes an alternative to the contractualization of behavior, and in many others supplements the capacity of contracts and formal institutions to guarantee the necessary smoothness of human interactions.2 An important theme of this book is that there is a complex relation between reputational incentives, their nature, and some key characteristics of governance. To clarify this point, we look at the example of open-source software production. In this production method, highly relevant reputational considerations go hand in hand with a mode of governance that is rather horizontal and nonhierarchic. Open-source software3 is very different from proprietary software where the code cannot be modified, or even accessed. Often, it also reflects a specific mode of production, characterized by the presence of very little (apparent) structure, and of horizontal and nonhierarchical relationships among participants, within which collaborative effort and experimentation play an important role. The tight relationship between production and experimentation is shown by one of the slogans of the open-source community: “Deliver early, deliver often.” There is no clear distinction
8
Introduction
between the planning and the production of a product and, in general, open-source software developers do not pay much attention to the codified tenets of the software-engineering discipline, which establish how software projects should be conducted and determine, among other things, that the requirements of the software to be developed should be analyzed formally and at length (Crowston et al. 2005). Open-source software development, on the other hand, often sets out to solve a problem faced by the developers themselves, and does not include a proper analysis of requirements. This horizontal mode of production is in sharp contrast with the more traditional and vertical structure of proprietary software projects. A fortunate description of this contrast can be found in the title of the account of open-source software by Eric Raymond (2000), The Cathedral and the Bazaar. It is in fact rather amazing that this “bazaar” successfully deals with the degree of complexity of some of the open-source projects. A good example is the way the operating system Linux grew from the project of a college student, Linus Torwalds, to become a serious competitor to products by firms like Microsoft and Sun Microsystems.4 While some contributors to open-source software projects are employed by important corporations (IBM and Sun Microsystems being noteworthy cases), most of them participate in their spare time without receiving any monetary compensation. An important question, then, is what makes such highly skilled people work for free—or, to express it in the words of an economist, what is the incentive structure of the opensource mode of production? The presence of an altruistic motive seems to be contradicted by the fact that a truly altruistic person would probably prefer to spend time on finding a solution to one of the many great problems plaguing humanity, and the development of a software product (which may later be adopted by possibly greedy corporations [Lerner and Tirole 2002]) would not appear on this list. The personal satisfaction of being recognized by one’s peer group certainly plays a role. However, accounts written by insightful advocates (as an example, see Raymond 2000) and academic enquirers (as in Dalle et al. 2005) agree on the pivotal role of reputation in the governance of open-source projects. Having a good reputation, besides advancing one’s position within the open-source community, also signals to potential employers that one has the qualities of a good programmer (Lerner and Tirole 2002, 2005).
Introduction
9
Previously, we noted that there is a relation between technology and the presence of reputational considerations in governance. A given technology may support and allow the functioning of a (reputation-based) governance model, just as the Internet is necessary for eBay to work, and a network of roads was essential for the functioning of the Roman empire. However, the Internet does not determine eBay, nor did the technology that was available to the Romans two thousand years ago cause the characteristics of their empire. The analysis of the relation between the available technology and governance will have to be careful and nuanced.5 We would now like to discuss some examples of public governance where reputational incentives are relevant. Obviously, reputational incentives also matter in the public sphere in general—for example, citizens in democracies, when electing representatives, care about the past record of the available candidates. Today, we are witnesses to attempts to translate forms of public governance where reputational incentives play an important role, that are Internet-based, and, incidentally, that are rather horizontal in character. One example, besides being interesting in its own right, also received the important endorsement of being cited in the program of U.S. president Barack Obama (Waters 2008). The project Peer-to-Patent (Allen et al. 2009), running from June 2007 until June 2009, was developed by the New York Law School Institute for Information Law and Policy in cooperation with the U.S. Patent and Trademark Office (USPTO), and aimed to try out a collaborative solution to fix some of the many problems that affect the patent system, in the United States and elsewhere. Patents are legal documents that provide the assignee with limited monopoly over an invention. They do so in order to reward inventors for their successful endeavors and to encourage their efforts. The monopoly, however, is limited, because after the invention is done, it is in the best interests of society to allow people to use it freely. A patent is then a compromise between two opposing needs: to reward inventors, and to allow inventions to be adopted as widely as possible.6 According to many observers, the patent system is in crisis, and most acutely so in the United States. There, the number of patent applications has risen conspicuously of late, giving rise to a “patent inflation” that has put strain on the USPTO and has resulted in lengthening waiting times for patent applications to
10
Introduction
be examined. Moreover, there is a widespread perception that over the last few years, more than in the past, patents have been granted for inventions of very dubious utility, and that in general the technological and economic relevance of granted patents has decreased (Bessen and Meurer 2008). In order to assess the fix that the Peer-to-Patent project advocates, we should remember that a patent in the United States is required by law to have three main characteristics. First, it should describe an invention that is nonobvious. Secondly, the invention should be potentially useful. Last, it should be novel. Patent examiners, who are employees of the USPTO, have to examine each application to determine if all these criteria are met. In particular, to determine the novelty of the invention, patent examiners have to consider what is called the “prior art”—that is, all relevant existing knowledge that in principle was available to the inventor. If, in the prior art, there already is a description of the invention, the novelty condition is not satisfied, and the patent should not be granted. However, patent examiners typically have little time at their disposal to be so diligent, and the analysis of the previous art often turns out to be one of the trickiest aspects of their work. The Peer-to-Patent project addressed precisely this problem by attempting to get outside experts to participate in the search. Anyone, in fact, could participate in this experimental project by accessing its Web site (www.peertopatent.org). The Peerto-Patent Web site allowed people to discuss the claims made by a given patent application, within an available selection, and suggest instances of relevant prior art. At the end of the discussion, all the material, together with a selection of up to ten suggestions of prior art, was transmitted to the patent examiners, who were then free to use any, or none, of it. After the conclusion of the project, a formal assessment of its results indicated an encouraging degree of participation, and both the external participants and the examiners expressed favorable opinions (Allen et al. 2009). Outside experts contributed on a voluntary basis and did not receive any pecuniary compensation for their work. Certainly, many of them were employees of big companies like IBM, GE, Intel, and Sun Microsystems, which were also stakeholders in the project, and we may expect these people to have contributed in their working hours. However, others were independent reviewers who made their contributions in
Introduction
11
their free time, like the independent programmers who work on opensource software projects. Reputational considerations were important, at least in the view of the designers of the system: whenever the patent office examiners chose a prior art that was suggested by one of the outside experts, this expert was awarded a symbolic “prior artist award,” and his name appeared on the Peer-to-Patent Web site. The workings of a patent system, if not the need for it, may seem rather arcane to most. National security, however, is a sector of public action that certainly feels closer to our daily lives, particularly in the aftermath of the terrorist attacks of September 11, 2001. Many observers, and also the official commission that was established in the United States to investigate the circumstances surrounding the attacks, coincided in chastising the relative lack of collaboration among the many agencies that had responsibilities in the field of national security (National Commission on Terrorist Attacks upon the United States 2004, Section 13). Sweeping reforms to correct the problem followed, the most visible of which was the creation in 2002 of the U.S. Department of Homeland Security, which aims to coordinate the work of over forty federal agencies (Moynihan and Roberts 2002).7 One initiative that was then taken deserves our attention. In 2005, the National Intelligence Council launched a pilot project called Intellipedia, as a way of increasing the sharing of classified information among participants in the intelligence community and putting a check on the “silo mentality” that almost naturally arises along the borders of separate organizations. This online collaborative environment runs on a classified network but is based on the same software used by Wikipedia. It is divided into three parts corresponding to increasing levels of classification, and it allows members of the intelligence community to share information across agency boundaries and to develop intelligence on particular topics collaboratively (Thompson 2006; Mazzetti 2007). Intellipedia has been a success, at least judging by the number of users and the amount of information they circulate and generate over the network. Charles Fingar, at the time deputy director of National Intelligence for Analysis, made it very clear that reputational incentives had to have an important role in Intellipedia, that it was to be “the Wikipedia on a classified network,
12
Introduction
with one very important difference: it’s not anonymous. We want people to establish a reputation. If you’re really good, we want people to know you’re good. If you’re making contributions, we want that known. If you’re an idiot, we want that known too” (Fingar 2008).8 In the public sphere, the Peer-to-Patent project and Intellipedia reflect an increasing interest in forms of governance that are open, that are relatively horizontal, and where reputational considerations play a role. However, given the ubiquity of reputational incentives in governance in general, it is striking to note that in public governance, it is possible to find cases where reputational considerations are almost completely absent. We consider one such case—the buying of a product or service— that is very often treated in diametrically opposed ways in the private and public spheres. In particular, let us consider the purchase of a service that cannot be fully contractualized. The difficulty in doing so may be due to the complexity of the service, making it close to impossible to describe it in full detail, and because the specifics and timing of its delivery depend on outside factors that are also complex and not fully predictable, so that the simple enumeration beforehand of all possible contingencies would be a daunting prospect. In fact, it does not take anything very fancy to fall into the category of “incomplete contracts,” as economists call them, and it may be argued that all contracts are, to some extent, incomplete, precisely because they cannot contemplate all possible contingencies. In real-life situations, the more incomplete a contract is, the more we observe the presence of postcontractual agreements of various types. As a simple and mundane example, consider refurbishing a family kitchen. Assume that the maker of the tiles that were specified in the contract goes bankrupt. In such cases, the owner of the house would typically talk with the workers and agree on a substitute—a case of postcontractual agreement. There are often more compelling reasons it may be impossible to fully contractualize the quality of a product or service. Its quality frequently becomes known in detail only after its delivery has taken place or sometimes after prolonged use. All products are to some extent “experience goods”: We fully learn about our car after having used it for several years, and we need to watch a movie to the end in order to judge it.
Introduction
13
Let us return, then, to refurbishing the family kitchen and selecting the firm to do the job. When choosing from alternative quotes, price will obviously be an important element. However, the family will also be concerned about the expected quality of the work, and this cannot be fully described in the contract—just as the menu of a restaurant, detailed though it may be in the description of the dishes being offered, does not say everything about them. The reputation of the professional candidates for carrying out the job then provides valuable information. Typically, the family would be willing to pay more for the same contractualized service to a provider suggested by a friend than to a complete unknown, or to one with a bad reputation. In other words, a good reputation would command a higher price. Let us now contrast this private transaction with public procurement. The Italian Ministry of Economy and Finance, among its many tasks, is responsible for the upkeep and improvement of a number of digital information systems that are crucial for the working of the Italian state. Together, these systems allow the working of, among other things, the system of payments of the Italian central administration, whose overall transactions represent an important share of the Italian national product. These information systems are complex not only because of their scope, fault tolerance, and security requirements, but also because they are the result of many interventions that have been carried out over many years. Starting in the 1960s, the central government has been one of the largest purchasers of mainframe computers in Italy, and, in the language of computer specialists, the various interventions that have taken place over the last few decades present an important “legacy problem.” To maintain and upgrade its information systems, the Ministry of Economy and Finance outsources a multiplicity of tasks through contracts that may be on the order of tens of millions of euros, last a few years, and require interventions that range from the maintenance of existing systems to major upgrading of these systems. The choice of suppliers is made through public tenders. A very detailed itemized technical description of the services required is prepared, and, for each of the many items contemplated, a number of maximum points are assigned, according to the quality of the proposal. An ad hoc committee made up of independent experts evaluates the proposals that are received. Their work is
14
Introduction
carried out in distinct phases. First, each member of the committee assigns a number of points, not exceeding a given maximum, to each of the items specified in the tender. At the end of this examination, each proposal is assigned an overall quality “technical score,” resulting from the sum of the itemized scores, each one equal to the average of the scores assigned by each member of the committee. The committee carries out its evaluation without knowing the amount of the bid. This very relevant piece of information is, in fact, contained in a sealed envelope that is opened only at a public meeting, once the technical scores have been assigned and publicly declared. The bid and the technical score for each proposal are then inserted into a mathematical formula that is known to all participants and that represents a tradeoff for the administration between cost and quality. This formula is used to calculate the overall score. The enterprise receiving the highest overall score wins the tender. Public administration lawyers are present at all times during the committee sessions, and the relevant envelopes are opened in sessions attended by representatives of the participating firms, so they can verify that all operations are executed according to the rules. In general, the committee has to proceed very prudently (hence the presence of the lawyers to assist its members), so as not to make mistakes that could later be used by the losing firm to dispute the final decision in court. I have described the selection procedure in some detail in order to give an idea of its extreme complexity. This meticulousness is motivated by the desire to avoid decisions that may be arbitrary or, worse, motivated by private interests—read, corruption. The result, unfortunately, is a procedure that, besides being very costly, presents some fundamental shortcomings deriving from the incomplete nature of the contract that will eventually bind the winning bidder with the public administration. On the one hand, the complexity of the contract often allows the firm to renege even on the commitments that happen to be well contractualized, without having to fear that the public administration will start administrative proceedings against them, possibly leading to the imposition of sanctions. In fact, imposing sanctions on a firm is costly to the public administrations, so that once the contract has been signed, the public administration has an interest in maintaining good relations with the contractor, since the latter could retaliate by adopting an uncooperative attitude and by
Introduction
15
strategically reading the contract so as to take advantage of any loophole it may have, in order to minimize effort. In other words, any poorly specified parts in the contract end up weakening its well-contractualized aspects. Second, the quality of the services that will be delivered is not necessarily correlated with the quality of the proposal that was presented to the tender. Enterprises can decide to spend resources on learning to write impeccable proposals, so as to be granted many projects, at the expense of investing to improve the quality of their products. To participate in the tender, enterprises must satisfy some very stringent conditions to show they have the size and expertise needed for such a daunting job. The result is that the market of potential suppliers includes only a handful of (mostly multinational) firms. An important implication of the reduced size of the market is that the competing enterprises are well known to the public administration. Some of these firms may have been former suppliers of a ministry for similar services, maybe including some that are up for tender. Due to the thinness of the market (this certainly explains why the whole selection procedure is planned so carefully to minimise the possibility of corruption), members of the selection committee often have first-hand information on the past performance of a given participant, that is, its reputation. Explicitly considering such information is, however, illegal, and justifying the attribution of a low score by a past poor performance would provide the lawyers of a losing bidder with a compelling argument to get the committee’s decision reversed in court. To conclude, public procurement in Italy (and in the whole of the European Union, for that matter) rules out the use of reputational information.9 Public procurement represents an important area of public governance, so that what we have just discussed is not only an example indicating that not all (public) governance is reputation-based. More than that, it shows that in many countries an important area of public governance explicitly excludes the use of reputational considerations. The cases that we have considered, albeit anecdotal, have hopefully convinced the reader at least of the plausibility of the main theses that this book sets out to demonstrate: Reputational considerations have traditionally been important in governance, but less so in public governance. The new information technologies—in short, the Internet—allow for new and creative ways of organizing and disseminating information
16
Introduction
pertaining to the reputation of the actors of governance. They allow the engineering of appropriate Internet-based information systems so that in planning new forms of reputation-based governance, there is an explicit focus on the possibility of designing them in ways that are tailored to our needs and desires. Also, one of the Internet’s essential characteristics is the inexpensive access to the information it provides and the ease of interaction and collaboration it supports. These factors, if appropriately harnessed, may favor citizens’ involvement in policy. I will argue that reputation-based governance could, in fact, be conducive to instances of governance that are not only efficient but also characterized by very advanced forms of democratic participation. A Note on Methodology
The analysis that follows has both a positive and normative character. It is positive when it analyzes the role of reputation in governance and when it considers the foreseeable consequences of adopting governance modes where reputation occupies a central role. The book also has a normative slant, deriving from the conviction that the adoption of reputation-based governance would improve the quality of public governance. This is not, however, a how-to book, nor a manual for policy makers who desire to embark on the next wave of fashionable reforms. The goal is more modest and, at the same time, more ambitious: to inform and to spur the debate on the new opportunities that the use of reputation in governance affords. I adopt what political scientists define as a “rational choice” approach by assuming that individuals are self-interested and that the outcome of governance is the aggregate result of the behaviors of actors who are motivated by the incentives they are offered. Assuming instrumental rationality is certainly common in economics and increasingly so in the social sciences in general. It is not, however, an ideological choice, and it is not meant to imply that human behavior is uniquely driven by selfinterest. What I do assume is that self-interest is an important driver of human behavior, to the extent that its selection as the central behavioral assumption, possibly at the expense of other factors that I omit, allows us to develop analyses that have satisfactory explanatory power. Also, ratio-
Introduction
17
nal choice comes in different types, and the one that I adopt is rather flexible. I accept that the motivations informing our actions are varied and that among them are some that would be important for sociologists and psychologists.10 Governance, in this book, is about institutions of governance: that is, sets of individual incentives, appropriate organizations and procedures that provide constraints on actions and deliver behaviors, the aggregation of which determines societal outcomes. I subscribe to Douglas North’s (1991) well-known definition, which holds that institutions are “the humanly-devised constraints that structure political, economic and social interactions,” consisting of “both informal constraints (sanctions, taboos, customs, traditions, and codes of conduct), and formal rules (constitutions, laws, property rights).” North’s definition is centered on the concept of constraints on human behavior and is quite ample. It includes formal settings, where formal organizations fulfill a key role in constraining the behavior of their members, and cases where behavior is influenced by informal arrangements, as in the case of the Maghribi traders. When attention is on formal institutions—for example, public administrations—I will often refer to the concept of organization, which I define as a social construct having one or more explicit goals and a rather formalized structure. The distinction between institutions and organizations serves two purposes. On the one hand, it is coherent with the generality of North’s definition of what an institution is—a human artifact that, one way or another and not necessarily by means of a formal organization, succeeds in constraining human behavior. On the other, it hints at the organizational qualities of formal institutions (as opposed to informal ones), and of public administrations in particular, that have a set of collective goals that may be more or less explicitly stated, and are characterized by internal working procedures that often are codified, by a structure and by a boundary separating them from the outside environment. Of all the institutions of governance, both formal and informal, we will be particularly interested in public administrations, especially when we discuss how strong and appropriately engineered reputational incentives, together with Internet technologies, could improve public governance. Our focus on formal institutions will be so pronounced that the normative message of the book will suggest how public administrations could be
18
Introduction
reformed. In this respect, public governance will be about how public administrations implement policies, and the focus of our discussion will in fact shift at times from general public governance issues to the characteristics of government. This focus on public administrations is partly justified by a simple observation: the vast majority of resources that are dedicated to the administration of the res publica are controlled by public administrations, and for this simple reason they are the central actors of public governance. This does not mean that other actors are not important. Organizations that are an expression of civil society, for example, do play a role, and we are witnessing several attempts by nonprofit organizations to harness the power of reputational incentives to improve governance.11 Such experimentations at times are relevant in their own right, to the extent that they influence governance. Moreover, they contribute to showing the general public that today’s Internet technologies offer very interesting opportunities to enhance transparency, participation, and, eventually, the people’s control of government. However, thinking that public governance can be changed from below is only an idealist’s dream, because it forgets the most basic reality of public governance in our age: for the most part, it is the big bureaucracies that shape it, and we simply cannot change it without reforming public administrations. This explains why, in elaborating a proposal on how to improve public governance, I will in fact propose changes to public administrations. A second reason for my focus on public administrations is better seen in the light of the debate on ICT-enabled changes in public governance—e-government, in short. Over the past twenty years or so, many commentators have predicted a brilliant future for the Internet where it would, almost by itself, generate advanced forms of democracy and popular participation in government.12 For the simple reason that this book also navigates its way through a debate that, at times, has indulged in techno-utopian excesses, it may be appropriate to be rather clear right from the start on this issue with a statement whose veracity would otherwise be obvious. Democratic participation is about processes and, more to the point, about institutionalized processes guaranteeing specific forms of participation. In discussing reputation-based governance, and particularly its implications in terms of democratic participation, I will anchor my argument firmly to this (at least to me) self-evident truth.13
Introduction
19
This observation leads us to a further issue. Our focus on public administrations will not leave the general public out of the picture. To the contrary, the type of reputation-based governance that I have in mind attributes an essential and active role to the people as actors of governance. Advocating an active role for the people will be a continuous thread in what lies ahead, leading to the conclusion that reputation-based governance could support very advanced forms of democratic participation. I hope that, by the end of the book, all readers will be convinced that my choice of the word governance for the title of the book, instead of government, was indeed appropriate.14
2
Reputation and Trust
In an anonymous market, conduct in a given period has no effect upon the reward in subsequent periods. The agent thus has nothing to lose by cheating the merchant. Avner Greif
Toward the end of his Commentaries on the Gallic War, we learn how Julius Caesar confronted one of the last ordeals that barbarous Gaul inflicted on him, not long before he moved to what is now Northern Italy and to the Rubicon. The important defeat of Vercingetorex in Alesia was behind him, but not everyone accepted the Pax romana. A few local chieftains challenged Caesar and withdrew to a fortified camp named Uxellodunum.1 The Romans laid siege to Uxellodunum and diverted its only water, and the defendants, dying of thirst, eventually surrendered. Caesar then had to decide what to do with the prisoners. The author of the story makes it very clear that the Roman general was well known for his leniency before he offers an account of his decision. That is, he had a reputation, and that reputation influenced his decision. Julius Caesar responded to the call in an interesting way that goes a long way to showing the complexity of the concept of reputation. We will now define and discuss this concept, as we were rather informal about it in the introductory chapter. Once we are on firmer analytical ground, we will be ready to mull over Caesar’s decision. At the base of our discussion, we draw a distinction between reputation and the related concept of trust, and we begin our analysis precisely with the latter, by discussing how it can refer not only to interpersonal relations, but also to interactions that involve organizations. Addressing this issue is important because public governance is largely the result of
Reputation and Trust
21
the actions of public administrations, that is, of very big organizations. We draw on the large body of literature that analyzes these themes using the tools of game theory, describing interactions among individuals that are repeated in time, of the type that could exist between a person and the grocer down the street. In fact, it could even be that their relationship consists of only one meeting, so long as information on the outcome of the interaction is passed on to others—for example, when a friend who went to a bad restaurant tells us about it. We discuss two basic relationships of this type. The first one is known as the “folk theorem,” which is a simple and very important result in the theory of repeated games describing the emergence of trust. The second one implies a learning process, characterized by a continuous updating of one’s beliefs about the nature of one’s counterpart. This type of relation describes the presence of reputational effects. The two mechanisms, obviously, could be at work at the same time to describe situations where trust interacts with reputational effects. The concepts of reputation and, particularly, of trust have been approached by many disciplines, producing a vast body of literature that is noteworthy for its lack of consensus. In what follows, I refer to that debate, but not exhaustively. Our analytical framework, based on game theory, has two advantages over the alternatives that are available. First, it helps to keep our minds focused on well-defined concepts, whereas part of the literature at times seems to be mostly occupied in trying to solve terminological issues, following the fact that “trust” is a very semantically charged term, as demonstrated by the sixteen separate meanings of it in The Shorter Oxford Dictionary (Barbalet 2009). Also, our rational-choice, game-theoretic approach has the added advantage of connecting nicely with the analysis of economic incentives within public administrations—the “economics of career concerns”—that we will discuss in the next chapter. The second part of the chapter provides a different and broader historical perspective on the role of reputational incentives in governance, with a foray into the economic history of the Middle Ages. We consider again examples of forms of governance where trust and reputational considerations played an important role in guaranteeing longdistance trade in the Middle Ages, before the modern state, with its institutions and constitutions, could establish the needed guarantees of private property.
22
Reputation and Trust
Trust and Trustworthiness
In the simplest game-theoretic setting, the rise of trust can be seen as a result of the repetition of what social scientists call the prisoner’s dilemma game.2 The prisoner’s dilemma is probably the best known, and certainly one of the most relevant, characterizations of strategic relations among different actors. In its simplest form, it describes a situation where two agents would be better off if they could agree on a mutually advantageous deal, but individually they have an incentive to cheat each other. The worst that can happen to them is to decide to be “honest” when the other person chooses to cheat, so that, anticipating this outcome, both find it convenient to cheat. We may express this simple idea by introducing a payoff matrix, which is a simple scheme of what each of the two actors obtains, conditional on the decision of the other actor. Table 2.1 provides an example, each cell reporting what the two players obtain at the end of the game for given combinations of strategies. For example, the bottomleft cell says that if player 1 chooses “cheat” and player 2 chooses “be honest,” player 1 obtains $50, and player 2 gets nothing. If both players could somehow make a binding deal—for example, if there were an outside institution that could enforce a pact—they would both agree to be honest so that they would each gain $30. No other deal can be struck, because in all other cases one of the two players would get less than $30 and would not agree to the pact. However, in a situation where it is not possible to commit to playing “honest,” each player, if he believes that the other player will play “honest,” has a clear incentive to cheat to gain $50, which is better than $30. But if both agents think the same, then they both find it convenient to cheat, and the observed outcome is that they each gain $10. This, in fact, is the only (Nash) equilibrium of the game.3 We may think of two ways to avoid the prisoner’s dilemma, that is, to induce the players to play “honest” and to secure for themselves a higher outcome than the Nash equilibrium. The first one, as already noted, requires the existence of an arrangement of some kind that effectively constrains the actions of individuals so as to make credible any promise to play “honest.” Such a possibility is tantamount to modifying the payoffs of the game by imagining, for example, that dishonest players are invariably caught by the police and sentenced to pay, say, $50. Or we
Reputation and Trust
TABLE 2.1
23
The Payoff Matrix of a Prisoner’s Dilemma Player 2
Player 1
Be honest Cheat
Be honest
Cheat
$30, $30
$0, $50
$50, $0
$10, $10
can imagine that the two players may reach an agreement beforehand and that there is an organization—call it the mafia, for the sake of argument— that levies a fine on them if they do not respect the agreement. Last, we may assume that there is a social norm, possibly interiorized by individuals, compelling people to keep their word. However, of more interest to us is the possibility of escaping the dilemma without the need for any formal institution that, one way or another, is able to enforce agreements between the two players. This is where the folk theorem comes into the picture. It is a result now familiar to all students of social sciences, and here it suffices to provide a simple explanation. Imagine that the two players live forever and that they play the prisoner’s dilemma game an infinite number of times (I will argue later that this assumption is not as unrealistic as it may seem at first). Also, let us assume that they start by playing “honest,” that each player continues playing this way as long as the other player does the same, and that each player responds to his counterpart’s cheating by playing “cheat” forever afterward. Starting from the second round of the game, each individual has to decide whether to choose the same strategy as the previous round (playing “honest”) or to deviate from it and start playing “cheat.” Once a person starts doing so, in this simple characterization he will continue doing so forever. In such a context, where the game is repeated over and over again, in deciding which course of action to choose, each individual has to weigh up the benefits and costs of each alternative over an infinitely long time horizon. In particular, by opting to start cheating, he will obtain an immediate extra payoff (in Table 2.1, $20, the difference between $50 and $30), but he will then forever get $10, when both players start playing “cheat,” instead of $30 (the payoff when both play “honest”). Whenever
24
Reputation and Trust
the player values the future loss of revenues more than the present oneshot gain, he will opt for not cheating. These possibilities both depend on the amount of the payoffs and on the player’s time discount factor, expressing his subjective evaluation of future payoffs with respect to the payoffs received today (that is, in everyday speak, the size that an egg must be today to make it preferable to a chicken tomorrow). In general, if the future is not discounted too much in relation to the payoffs of the game, both players will decide to play honest forever. This is the essence of the “folk theorem,” its name reflecting that it is not clear who first had this (admittedly rather simple) intuition. A few caveats are in order. First, the folk theorem does not hold when the game is played for a known and finite number of times. In fact, in this case, the players know that in the last round they will have an incentive to play “cheat,” since there is no future loss in doing so. So the last round is described by the prisoner’s dilemma. But by anticipating this fact, they will also find it convenient to play “cheat” in the last round of the game but one, when they will also have nothing to lose in the future. So they will play “cheat” in the round before—and so on, backward. Each round is played as if it were a one-shot game, and the players play “cheat” forever. The solution of the repeated game, in this case, is just a repetition of the solution of the one-shot game: a repetition in time of the prisoner’s dilemma. Given that real-life games cannot logically be played with an infinite time horizon, since people eventually die, at first sight the folk theorem could seem to be just a curiosity. However, this is not the case. It is true that people do not live forever, but the institutions where they work can be thought of, at least as a first approximation, as infinitely living.4 Also, for adequate discount rates, the result of the folk theorem can be obtained for games that are played a number of times that is finite but unknown beforehand. Some simple algebra would show that by attaching to each date a probability that it is the last round of the game, the same reasoning that leads to the folk theorem would still apply. Finally, and of particular relevance to us, note that the results illustrated above, strictly speaking, do not depend on repeated interactions of the same players. Under certain conditions, players could even meet only once. What matters is that they observe the history of past interactions, so that if, say, player A cheats player B, then player C, whose turn it is to
Reputation and Trust
25
interact with A, is correctly informed of what happened. The desired incentive to be honest would still be in place, just as a restaurant owner would be afraid of losing not only a single client by serving him poorly, but also other potential customers to whom the disappointed client may speak. The presence of an effective communication channel to spread this type of information is then crucial for the folk theorem to apply when relations are impersonal. A further consideration regards the nature of the punishment that is meted out to cheaters. Above, I assumed that any cheating is punished with cheating forever. In most real-life situations, however, we may expect people not to react so harshly, and to eventually decide to give a second chance to a cheater. Different punishment strategies may give rise to multiple equilibria for the repeated game—a source of ambiguity, given that it is hard to determine a priori which punishment strategy the players will adopt. The folk theorem explains the emergence of trust,5 which is here firmly grounded on the self-interest of the actors involved, to the point that this characterization is often said to give rise to a concept of calculative trust (as opposed to noncalculative trust, which we will consider later on). Within the calculative approach, particularly in the political sciences, the emergence of trust is sometimes seen to follow from the presence of “encapsulated interests”: that is, you trust a person because you believe that this person has interests that are coherent with your own (see Hardin 2002 and Warren 2004). In the simplest of accounts, encapsulation may be seen precisely as the consequence of the repetition of a prisoner’s dilemma. In this case, “You value the continuation of our relationship, and you therefore have your own interests in taking my interests into account” (Hardin 2002, 1). Strictly speaking, such a precondition for the rise of trust may only happen within a thick dyadic relation between two people or organizations, as in the relation between me and the shopkeeper down the street: I go there for an indefinite time span, and he knows that if he cheats me, I will retaliate by going somewhere else. In that sense I trust him, because I am aware of the fact that my interests—not being cheated—are encapsulated in his—not losing a long-term client. Also note that the constituent game of this example is not a prisoner’s dilemma, and the situation is more one of a “one-way trust.” This is like the situation described in Dostoyevsky’s
26
Reputation and Trust
The Brothers Karamazov between a lieutenant colonel who regularly gives a large sum of public money to the merchant Trifonov, who in turn invests the money to later return it to the dishonest soldier, together with a personal gift. In such a relation, it is only the merchant who can cheat—and cheat he eventually does, after the lieutenant colonel loses his post and the prospect of the continuation of the relationship stops providing an incentive for “honesty” (see the account in Hardin 2002, Chapter 1). More prosaically than in Dostoyevsky’s immortal invention, the relation between a shopkeeper and his client is asymmetric. He can cheat, for example, by selling low-quality products. All that the client can do is to discontinue being a client, just as the lieutenant colonel could have stopped, possibly before it was too late, secretly giving large sums of public money to the merchant Trifonov to obtain an illicit profit. However, the encapsulated-interest story holds also for cases when the relation is not thick or dyadic. Interactions may be not repeated, taking place between individuals who possibly meet only once, but who benefit from the presence of a communication mechanism providing them with (possibly noisy) information on the past performances of their counterparts. This is the situation on eBay. Though actors usually interact only once, before deciding whether to trade they can read the “feedbacks” previously received by their potential partners. Such digitalization of word-of-mouth will play a central role in the forms of reputation-based governance that we will discuss. Trusting Organizations
The concept of trust does not translate automatically to situations where the interacting parties are not two individuals but two organizations or, as is typical in public governance, one individual and a big public administration. Can a public administration be trusted as we would trust an individual or, possibly, a small organization that is closely identified with its members? To discuss this important issue, we take as our starting point the work of Hardin (2002), who, while not completely dismissing the possibility that organizations may be trusted, raises some doubts, and in particular he makes a clear distinction between the presence of trust, arising from interests that are perceived to be encapsulated by the person who trusts, and what we may merely call an expectation
Reputation and Trust
27
that follows the presence of an empirical regularity. For example, we expect that tomorrow the sun will rise just as it did today, given our past experience and observations, but we cannot really affirm that we trust the sun to rise, given that it does not encapsulate our desire to be warm or the quest of mankind to continue living on this planet. For a more earthly example, Immanuel Kant was very methodical in his daily routine, to the point that his neighbors in Kaliningrad (or Königsberg, as it used to be called), the city where he lived, were said to set their watches when he appeared for his morning walk. However, they could have not said that they trusted him in the sense that he interiorized their desire to have a human watch of sorts to serve their interests. Quite likely, Immanuel Kant simply enjoyed having his morning walk always at the same time, regardless of the beneficial effects that his appearances may have had on others (Hardin 2002, 5). In Hardin’s account, statements of the type “I trust an institution” are difficult to support, because most often the vernacular use of the term simply denotes an expectation of future behavior that is based on a past regularity, not dissimilar from our trusting the sun to rise tomorrow. Determining the presence of encapsulated interests becomes increasingly complex when we refer not only to dyadic personal relations, but also to relations of individuals with organizations. It becomes somehow clumsy to affirm that I trust, say, some public administration because it encapsulates my interests. In this respect, two questions naturally arise. Who encapsulates my interests, the administration as such, or its employees? In the latter case, in what sense does the collective action of a great number of self-interested employees aggregate, so as to allow me to say that the overall interests of that organization are coherent with mine? It is even more problematic to assume that there is an ongoing repeated game between a public administration and a single citizen. Public administrations tend to be monopolies in their respective field of activities, so that the threat to quit the relation in favor of a competitor is not credible. Sometimes it is possible to obtain a public service from more than one organization—for example, when there are private firms that provide services similar to the ones supplied by a public administration (such as private hospitals in countries where such choice is possible). In these cases, halting a relation with a given administration is not necessarily a threat. When the resources received by an administration are to some
28
Reputation and Trust
extent fixed, as they often are, having fewer clients provides bureaucrats with more leisure time on the job, so that they may even have an incentive to encourage their clients to punish their organization by turning to a competitor. In most cases, a single citizen does not have much room for maneuver against an institution, and threatening to interrupt a long-term relationship is simply laughable. For these reasons, administrations can often be seen to take little notice when individuals, having voiced their dissent, decide to stop using their services. Obviously, at times the problem becomes so urgent that citizens take collective action and voice their dissent loudly enough for the powers that be to take notice and set about correcting it. Such a situation could be best characterized by one of those bureaucratic “vicious circles” described by Crozier (1963, 258). According to Crozier, changes are hard to obtain within bureaucracy, because “decisions, even when they regard small changes, are usually taken at the top.” When top managers are eventually made aware of outstanding problems, “they find it very difficult to take action, because of the weight imposed on them by the presence of impersonal rules that risk to be affected. A bureaucratic organisation . . . only gives in to change when it is put at risk by the presence of serious dysfunctions that have not been amenable to correction.” Adaptation to outside pressure within a big bureaucracy, then, happens in a discontinuous fashion, Crozier writes: “Change can’t come piecemeal. To cause a reform of the organization, what is needed is a dysfunction that is serious enough to put at risk the life of the organization itself.” Defending the U.S. Constitution in the Federalist Papers, Madison writes that “in framing a government which is to be administered by men over men, the great difficulty lies in this: you must first enable the government to control the governed; and in the next place oblige it to control itself.” His recommendation is that “ambition must be made to counteract ambition” (Hamilton et al. 2008, 289). How? “If I violate the norms determined by our bureaucratic mission, you and others are likely to find it in your interests to oppose me” (Hardin 1988, 526–527, cited in Hardin 2002, 20). The approach of encapsulated interests could possibly be applied to the realm of big organizations by taking James Madison’s analysis down to the level of individuals.
Reputation and Trust
29
In principle, Hardin (2002, 154) recognizes this possibility, but generally speaking he holds that our trust with respect to organizations may be such only in a vernacular sense, and it should really be seen as a mere expectation of their future behavior based on past regularities. I argue that the difference between trusting an organization and having an expectation about its future behavior that is based on past observed regularities is one of degree, not of kind. An example will clarify the issue. Assume that a person—let’s call her Ann (Audrey Hepburn’s name in Roman Holiday, the 1953 movie directed by William Wyler)—decides to visit Italy for the first time in her life and, being worried about possible abuse by the local police, wants to learn more about the situation that she is likely to face. Will she decide to trust the Italian police not to misbehave? If she is stopped by a police officer, can she expect him to behave professionally and respectfully? Let us consider three possible cases. In all cases, Ann is a person with good common sense—perhaps unlike the Ann of the movie—but her skills and eagerness to learn about Italy vary in each case. Case 1. Commonsense Tourist. Ann does not know anything about Italy, and has never studied any political science or any related discipline. She has not interiorized any formal model of big public organizations; however, she has an intuitive understanding, possibly based on her own personal experience and on her knowledge of history, of the fact that formal institutions tend to behave in ways that are largely constant in time and that major abrupt changes only happen rarely. She checks the Internet, and she quickly learns that Italy has shown a remarkable degree of institutional stability over the past few decades at least and that no revolution seems to be imminent. Then Ann decides to consult on the Internet the summary of the latest available Amnesty International report on Italy, because she believes in the impartiality of the organization producing it. This report indicates that, at least in comparison with many other countries, the Italian police force tends to be well behaved. Indeed, some worrying episodes are present. For example, some accounts relate discriminatory behavior against Roma, and of suspicious “renditions” within the so-called “war on terror” (Amnesty International 2009). As she is not a Roma, and is very unlikely to be suspected of terrorism, Ann estimates that she will not be harassed by the police during her stay in Italy. Can we conclude that she trusts the Italian police? Not quite, according to Hardin’s view of what qualifies as trust, because the conclusion
30
Reputation and Trust
is simply based on an observation of the past, and it should be classified as a case of “expectations about the future behavior of an institution,” not as a case of trust— not dissimilar to our expectation that the sun will also rise tomorrow. Case 2. Eager Political Scientist. Ann is now a researcher in political science embracing the theory of rational choice. Before coming to Italy, she decides to learn more about the country, and she quickly becomes fairly knowledgeable about its institutional framework. She thinks of a model to describe the behavior of the Italian police. If police officers started harassing peaceful individuals, they would be outraged and would draw the attention of the media. The minister of Internal Affairs would then be held accountable by Italian public opinion, and the governing coalition, fearing electoral defeat, would add to the pressure. If the abuses were very serious, the minister could even be forced to resign. This lingering menace would pressure him ex ante to avoid such trouble, by making sure that the police behave properly. At the end of her study, Ann would conclude that the police officials in the street would understand that they would be held accountable for any serious abuse and that her desire to be left alone while strolling the streets of Italy is encapsulated in their interests. The aggregation of individual incentives and behavior is, in this case, straightforward: if all police officers face the incentives described above, they all tend not to misbehave, and the aggregate observed behavior of the Italian police follows straightforwardly. To test the implication of her theory, Ann decides to read the last available report on Italy by Amnesty International. What it contains is largely consistent with the prediction of her model, crude as it may be. Can we conclude in this case that Ann trusts the Italian police? Yes, because her interests are now seen to be encapsulated in those of police officers. As it is for Madison, “ambition must be made to counteract ambition”: the ambition of the journalist acts against the possible greed of a police officer, who otherwise would have an incentive to corner hapless unarmed people in dark alleys so as to steal their belongings, or worse. Case 3. Lazy Political Scientist. Now Ann is also a sophisticated researcher, who has no desire to learn about Italian institutions. However, she wants to enjoy her trip as a carefree tourist, and she also firmly believes in the theory of rational choice. For her, it is almost a truism to conclude that, whatever the observed societal outcome in Italy or elsewhere, it must
Reputation and Trust
31
be the result of the interplay of a set of incentives faced by the relevant actors. To put it differently, if we observe some facts that are the results of collective action, there must be a true model that explains how the incentives affecting all the relevant actors shape what is observed. What is real is also rational, to cite Hegel and to mock Voltaire’s Pangloss. Moreover, Ann knows that institutions tend to be rather stable, and she easily finds out that in Italy there are no visible signs of major institutional changes. She would then read the summary of Amnesty International’s latest available reports on Italy. All of them would indicate that the Italian police behaved fairly consistently over the past years. Given such observed empirical regularity, she would conclude that the right incentives to behave in such a way are in place. In other words, once Ann observes a regularity in the institutional outcome, she does not feel that she needs to make an effort to come up with a theory of what these incentives are and how they play their effects. Can we conclude that Ann trusts the Italian police? Or is it again simply a case of formulating reasonable expectations based on an observed empirical regularity? The answer is not obvious. Ann has not formulated a clear theory of why her interests are encapsulated in those of the law enforcement officers. She believes, however, in what resembles a meta-model, possibly bordering on a simple rational-choice epistemic conviction, according to which some model should be there, explaining what is observed. The conclusion of this imaginary story follows naturally. In cases where formal institutions are involved, the difference between an empirical regularity and trust is one of degree and not of kind. Our Ann number 1 was not trained in the subtlety of political science, but she was endowed with common sense. She should not be, so to speak, discriminated against for her lack of a Ph.D. Ann number 2 had a clearer understanding of Italian politics compared to Ann number 3, and ended up formulating a model of sorts. Ann number 3 did not bother to develop a formal theory but reached the same prediction as Ann number 2. For the last of our Anns, an empirical regularity plays within a very broad model of what determines societal outcome, without the need to specify in any detail how that actually happens. Observing the empirical regularities documented by Amnesty International’s reports has different implications in the three cases. It is
32
Reputation and Trust
an empirical regularity without a theory in the first one. In the second case, it is an empirical regularity that provides an informal test of a theory. Finally, in the last case, it is an empirical regularity that simply shows a high degree of persistence of behaviors and, hence, of the underlying incentive systems. The picture that emerges is nuanced. We may talk of trust, rather than mere expectations, when we adopt a model that predicts the behavior of whom we trust and when this prediction indicates that our interests are encapsulated in the interests of the trustee. When trust is about organizations, not only must we have a theory that describes the incentives of the members of the organization, but also we need a theory describing their collective actions, that is, how the individual incentives and behaviors aggregate to determine the observed outcome. The case of firms is relatively easy to solve. If we assume their textbook characterization, adopting the single objective function of profit maximization, then the collective action problem of the firm nicely collapses into one of the wellknown “black boxes” that economists are so fond of (even this case, however, turns out to be not so straightforward whenever the owner of the firm is not its administrator, as so often happens). In general, however, unfolding the effects of the multifaceted incentives affecting a vast multitude of actors within an institution is a daunting problem. Uncertainty is pervasive in trust relationships, and the ensuing risk is one of its quintessential properties, as the lieutenant colonel in The Brothers Karamazov understood well, when the money that he had entrusted to the merchant Trifonov was not returned to him. However, again, the conceptual difference that exists between trusting a person and trusting a complex organization is one of degree, not one of kind. A second issue emerges from the previous analysis. The trust that our imaginary tourist may have in the Italian police does not result from a strategic repeated interaction as exemplified by the folk theorem. In particular, the strategic relationship described by the folk theorem is just one of the several that may give rise to an encapsulation of interests. There is no repeated interaction between Ann and the police officers she may meet. The model formulated by the second of Ann’s avatars, in particular, may more appropriately be seen as describing a situation where her interests are aligned with those of police officers and, in turn, of the police administration.
Reputation and Trust
33
We take the expression “encapsulation of interests” to be a synonym of the presence of “aligned incentives,” a formulation that is more common among economists. The desired alignment of incentives can be obtained in many ways. Indeed, organizations can be seen, at least to some extent, as artifacts that produce at least some degree of alignment of the incentives facing their members, in order to solve a collective action problem and allow for the pursuit of the organization’s common goals.6 The repetition of interactions is one important way of allowing individuals to establish trust relations, but it is certainly not the only one. Self- and Other-Regarding Types
When the folk theorem is at work, agents believe of each other that they will continue being honest because they find it advantageous. For all practical purposes, all agents are assumed to be identical. The whole idea that an individual may be intrinsically honest, or dishonest (or, as in BarIsaac and Tadelis [2008], hard worker or lazy), is ignored. Individuals are seen to be one or another solely according to what they find convenient in a given situation. As is typical in crude rational-choice accounts of human behaviors, humans are seen to be equally cynical in pursuing their selfinterest. What the folk theorem does is to show that the repetition of the game may solve what economists call a problem of “moral hazard” by making it in the players’ best interests not to cheat. As Cabral (2005) notes, “The folk theorem can be interpreted as a model of trust: if players are patient enough (that is, if the future matters a lot), then there exist equilibria of mutual trust.” We all agree that people do not always keep their word, and Dasgupta (1988, 53) rightly notes that “The problem of trust would of course not arise if we were all hopelessly moral, always doing what we said we would do in circumstances in which we said we would do it. . . . A minimal non-congruence between individual and moral values is necessary for the problem of trust to be a problem.” At the same time, if noncongruence is an obvious fact of life, individuals can hardly be characterized as being hopelessly immoral. There is a vast body of experimental literature that proves beyond any reasonable doubt that many people, instead of being entirely self-regarding, are at least to some extent “other-regarding.” They tend to trust and to cooperate more than predicted by game-theoretic models, and also they are often
34
Reputation and Trust
willing to punish purely self-regarding behavior of others, even if they incur a cost by doing so (see Olstrom and Walker 2003a, 2003b). In this respect, not only are there important individual variations in behavior, but also there are differences in mean behaviors across societies (Henrich et al. 2001). The reasons behind such differences in behaviors are not well understood. There could be a biological sense in which a will to cooperate is, in some sense, hardwired into many humans. Concepts such as group or kin selection have been invoked (Bateson 1988) by biologists, who have “struggled with the notion of altruism and cooperation, as these two behaviours appear to go against the principles of natural selection” (Olstrom and Walker 2003a). Such individual differences may influence observed aggregate outcome in various ways. In some cases, a minority of other-regarding individuals may push self-regarding ones to cooperate.7 In other situations, the opposite may happen, with the aggregate outcome closely resembling that of the individual cynical “homo economicus” (Camerer and Fehr 2006). Notwithstanding its practical limitations, a behavioral model that assumes a self-regarding individual is useful for at least two reasons. Trustworthiness is context-dependent: a person may be trustworthy in one respect and not in another and, more importantly, “[i]f the incentives are “right,” even a trustworthy person can be relied upon to be untrustworthy” (Dasgupta 1988, 49). For this reason, the consideration of incentives is essential in understanding trusting behaviors. Secondly, a self-regarding, calculative approach, provides us with a point of reference, akin to a Weberian ideal type, where only egoistic incentives matter. In this way, their role is isolated and made manifest. As economists often do, we obtain an insight (on a very complex problem) through simplification. Other Theories of Trust
It is rather common in the literature to define trust as a relation where the person who trusts believes that the trusted one will do something that is “beneficial or at least non detrimental” to him (Gambetta 1988, 217; see also Newton 2007, 343; Williamson 1993, 483; and many others offering similar definitions). Limiting trust relations to expectation of beneficial actions by one’s counterpart certainly conforms to a vernacular definition of trust, but within a calculative approach, it leads
Reputation and Trust
35
to terminological dire straits. Let us consider individuals A and B, who are linked by a long-term business relationship, namely, highway robbery. To fix ideas, we may think that A is the robber and that B takes care of selling whatever stuff A manages to expropriate. A and B reciprocally believe that the other will act beneficially, so that they are bound by a mutual (calculative) trust relation. A merchant runs into A while traveling. This merchant is very well aware of the incentives that A faces (let us assume that in his country, all highway robbers act in pairs just like A and B, so that having met one pair, you have met them all). It follows that the merchant expects that A will rob him. Whereas the merchant’s expectation regarding A’s behavior is identical to B’s, A’s action is beneficial to B but obviously prejudicial to the merchant. However, the robbing is very much the same, regardless of who the observer is, and it would be reasonable to describe such a phenomenon using a single concept. Affirming that “the merchant trusts that A will rob him” sounds clumsy. Looking for a negative of trust does not do, because the English language lacks such a term—distrust, according to the Oxford English Dictionary, means “lack of trust,” which does not mean an expectation that someone will hurt us.8 In fact, the concept of calculative trust hinges not so much on the expectation of something beneficial happening as on the legibility of incentives. The crux of the matter is that we (think) that we may predict the behaviors of other individuals (or organizations) because we are able to read the incentives that they face, and we succeed in mapping those incentives into behaviors.9 Calculative trust is about reading into others’ incentives and motivations, and the fact that the behaviors that follow are beneficial to one and possibly detrimental to others should be rather inconsequential as far as the definition of concepts is concerned. Many researchers would not agree with deriving the concept of trust from rational calculations. Trust is a “human passion,” according to Dunn (1988, 74), a relation that implies the “acceptance of dependence” in a situation where “information about the other’s reliability” is absent (Barbalet 2009), and where individuals, by and large, “act on some basis other than private interests” (Ruscio 1996), possibly more with reference to ideas of appropriateness and moral judgement, to the point that trust is prone to be “evidence resistant,” that is, “I have a tendency to read your behaviour in a way that tends to confirm my trust” (Weinstock 1999).
36
Reputation and Trust
The desire to believe is so strong that one characteristic of trust would be the absence of monitoring. The conclusion of this line of thought: calculated trust is a “contradiction in terms” (Williamson, 1993), an expression that should not be used for relations that are regulated by contracts or determined by the interplay of incentives. These views tend to confine the relevance of trust relations to personal relations, but in doing so, they also condemn the concept to a practical irrelevance. True, “there is a wide range of anecdotal, historical, and socio-psychological evidence to suggest that our capacity for self-delusion far exceeds rational optimistic expectations, and that we can indeed make ourselves and others ‘believe’ ” (Gambetta 1988, 232). However, there is also evidence of various types of selection forces, and among them the most brutal Darwinian ones, which take care of at least the most serious cases of gullibility. So it is questionable whether, even in the thickest personal relations, trust is completely based on acceptance of dependence and absence of monitoring. In important personal relations, precisely because they are so important, people tend not to be so naïve. Most importantly, these relations, taken individually, have little impact on societal and economic outcomes, and less so now than decades ago, considering the increasing degree of impersonality of our societies, which causes the “strength of weak ties [to become] more important as the scale and impersonality of society grows” (Granovetter 1973). So that, because “small and intense communities of the kind capable of teaching and sustaining thick trust, are increasingly rare in modern, large-scale societies . . . they have to rely on more diffuse and amorphous mechanisms to sustain trust” (Newton 2007, 345). The experimentation with Internet-based information systems that we commented upon in the introductory chapter has to be interpreted precisely in light of the historical juncture that is affecting our large-scale societies. Reputation-based governance, with its reliance on a technological infrastructure and institutions fostering the building of large-scale reputation and trust, is also a proposal to better manage the increasingly important “weak ties.” Learning About One’s Counterpart and Investing in Reputation
People may be intrinsically more or less energetic in what they do, managers are of varying levels of competence, and some firms are capable
Reputation and Trust
37
of producing products of better quality than others. As we discussed, people may be more or less intrinsically trustworthy. Such situations typically involve a degree of asymmetric information, where each agent knows about his own characteristics—or type—better than anyone else, while others can only try to infer such information from what they observe. Attentive observation is often not enough to unequivocally determine an agent’s type. For example, experiencing the quality of a product may not allow us to discriminate between a high- and low-quality producer, because the latter may succeed in producing high-quality goods at the cost of putting a lot of effort into the process. Also, high-quality producers may be mistaken for low-quality ones if, for some reason, they do not put enough effort into what they do. Moreover, exogenous shocks may influence the characteristics of what we observe, and observing a poor-quality product may leave us in doubt as to whether the producer is mediocre or has been the victim of bad luck. Such ambiguities often explain why agents engage in strategic behaviors that aim to send signals about their characteristics to their counterparts. For example, low-quality producers may decide to exert effort in order to convince buyers that they are of high quality and that, as such, they command a higher price for their products. Buyers, however, anticipate this possibility, so that when they observe a high-quality product, they may remain in doubt as to whether it has been produced by a high-quality producer, by a bad-quality producer who put a lot of effort into what he has done so as to be taken for a high-quality producer, or finally just by a lucky producer who benefited from a favorable shock. Such different considerations may lead to the presence of many possible outcomes of the interactions (of multiple equilibria, again in the parlance of game theorists), where producers strategically decide how much effort to put into their work to look better than they are (so as to pool with the high-quality producers) or not to be taken for low-quality producers (or to separate from them). Consumers observe the quality of products and continuously update their previous opinion on the characteristics of their counterparts. If a high-quality product is observed, for example, the consumers will think it more likely than before that the producer is of high quality.10 Different outcomes of the strategic interaction depend on many factors, whose discussion would lead us astray. What matters to us is that the amount of effort exerted can be seen as an investment in reputation. In most situations, a player who, through his previous efforts, convinced
38
Reputation and Trust
others of his high quality, has an interest in keeping up his virtuous behavior so as not to squander that investment.11 When actors are of different types, reputational considerations may effectively solve a problem that economists define as adverse selection, by allowing observers to screen types—that is, to determine with certainty the characteristics of their counterparts. When this happens, everyone knows the characteristics of others, and investment in reputation successfully serves as a signaling device for one’s characteristics. The reputational effects that we observe may be defined as static, in the sense that, when they are present, they induce given players to exert effort to deliver high-quality products. However, as we discussed in the introductory chapter, they also eventually have what we may loosely call a dynamic effect, by supporting a selection mechanism that tends to weed out low-quality actors. More mundanely, restaurants that persistently mistreat their clients, or firms that consistently produce lowquality products, eventually go out of business. It is this type of strategic interaction, where people learn about others’ characteristics by (imperfectly) observing the output of their actions, that will serve us in the next chapter to develop a characterization of how public administrations work, and of the present and prospective role of reputational incentives within them. Reputation Without the Internet: Lessons from the Middle Ages
The presence of repeated encounters is an inescapable characteristic of the human experience, and even in the darkest periods of human history some type of institution probably existed that extended the perceived time horizon of human interactions. Moreover, informal word-ofmouth has always been available to spread information on the behaviors of individuals and organizations, hence providing a way to escape prisoner’s dilemma–like situations to people who meet only infrequently, but who belong to the same community. Also, an updating of beliefs on the intrinsic quality of agents, based on past personal experience or, again, on word-of-mouth, has invariably provided some form of incentive to exert effort in developing a reputation. To summarize, the strategic relations
Reputation and Trust
39
that we have described in the sections above do indeed capture some important aspects of human interactions. In all the cases that we have considered, the interplay of these forces produces a system of incentives that effectively condition human behaviors, leading to the emergence of institutions—that is, of constraints that “structure political, economic and social interactions” (North 1991). These are institutions of the informal type. The distinction between informal and formal institutions, the ones provided by reputational considerations belonging to the former category, is also useful because it highlights the possibility that informal institutions in certain situations may develop to compensate for the absence of suitable formal ones. We have already discussed the case of the Maghribi traders’ coalition, where long-range trade was made possible by the presence of a form of collective punishment—a boycott—of the person that had been recognized as a cheat.12 This allowed the Maghribi traders to prosper, until they dispersed in the second half of the eleventh century, following the conquest of Maghrib by the Bedouins, and the decline of Muslim trade caused by the expansion of the Latin world (Greif 1989). Before that happened, the functioning of the Maghribi trade shows how an informal arrangement, based on the presence of strong reputational effects, effectively compensated for the lack of formal institutions. The tenth century—when the episode of the Maghribi traders took place—corresponds to the early portion of a very interesting period in history, heralding a revival of trade, after the contraction that accompanied and followed the demise of the Roman Empire, and extending to the rise of modern states toward the end of the fifteenth century.13 During this period, a fairly sustained level of trade, also of long-range trade, was possible even without the help of formal institutions that could safeguard it, in a situation where the jurisdictions of the then feudal states were much smaller than the reach of their economic activities. The situation changed with the development of the modern state, marking a process of consolidation of formal institutions and progressively creating a situation where the presence of effective courts, police, and some type of constitution constraining the sovereign provided a sense of security of property rights. Before that happened, the lack of formal institutions could have been fatal for the viability of long-range trade. Traders could not count on
40
Reputation and Trust
formal protection abroad, should a fellow trader cheat them. Also, merchants abroad always ran the risk that their goods would be expropriated by local rulers, whose best interest was to promise to respect prospective merchants, so as to convince them to set up shop in their land. Once they had done so, however, the local rulers would find it advantageous not to honor the promise and to expropriate the merchandise. However, merchants would anticipate the presence of such ex post incentives and, not believing the promises received, would not start trading in the first place. In the jargon of economists, the interaction between rulers and merchants was one characterized by a problem of time inconsistency: a situation where a shifting of interests (of the rulers) between a “before” and an “after” is perceived and anticipated (by the merchants), leading to a suboptimal outcome—in the case in point, the lack of trade. Greif, Milgrom, and Weingast (1994) (henceforth GMW) consider the role of medieval merchant guilds as institutions that effectively obliged a ruler to make credible promises to safeguard the property and lives of merchants who belonged to a guild within his jurisdiction. They note that “one of the central questions about the institutional foundations of markets concerns the power of the state. The simplest economic view of the state as an institution that enforces contracts and property rights and provides public goods poses a dilemma. A state with sufficient coercive powers to do these things also has the power to withhold protection or confiscate private wealth, undermining the foundations of the market economy.” While merchant guilds, to the extent that they could, certainly acted as cartels to benefit their members by inflating prices, it is not obvious why again in the words of GMW, “powerful rulers during the late medieval period cooperated with alien merchants to establish guilds.”14 The answer is that the presence of these guilds allowed a ruler to credibly commit to respecting the merchants and their properties, since it was clear to all that otherwise the guild could have retaliated by boycotting the ruler, interrupting trade, and stopping the flow of related benefits to the host country. For the ruler, the presence of the guild was a device that allowed him to make credible promises. The guild’s implicit (and, sometimes, explicit) threat to the prince was credible, because it was accompanied by an equally credible threat that the guild was able to make to its members, should they not respect a boycott: exclusion from the common benefits
Reputation and Trust
41
that derived from the price-setting power that the guild enjoyed. It was the power base at home that allowed members of a merchant guild to act as one when dealing with a ruler abroad, so that reputational considerations, based on the ruler’s past behaviors, could arise in what GMW define as a “bilateral reputation mechanism”—as opposed to the “multilateral reputation mechanism” in place within the coalition of Maghribi traders that we have considered above.15 Obviously, this mechanism was effective only as long as the ruler feared the threat of interruption of trade from a single merchant guild, a threat whose importance was directly proportional to the relative weight that the guild had in the trade occurring within the jurisdiction of a prince and that crucially depended on the lack of alternative sets of merchants ready to take the place of the boycotting ones. As an example, GMW consider the evolution of the German Hansa, a particular form of guild that aimed to coordinate and control the behaviors of merchants from different northern German towns so as to reach a critical weight that would make the announcement of a trade boycott a credible and effective threat. In 1358, after earlier fiascos, the Hansa eventually succeeded in staging a successful embargo against the city of Bruges. GMW comment that the “institution of the German Hansa was now crystallized. It was a nexus of contracts among merchants, their towns, and foreign cities that advanced exchange.” Suitable formal institutions not being available, this advancement in bargaining power was supported by informal, but effective, reputation mechanisms. The Hansa faced a problem of coordination that was not so relevant, for example, for the most important Italian merchant cities of the time, none of which was a marginal player in the trade game. Concluding Comments
The two distinct historical episodes that we have discussed show what we may define as examples of reputation-based governance ante litteram, or at least well before the appearance of the Internet allowed for the creation of ad hoc reputational systems. These examples are characterized by reputational considerations appropriately constraining individual behaviors, and they also illustrate the working of some of the incentives that we described while using a game-theoretic approach.
42
Reputation and Trust
Game theory gave us two different sets of models that are useful when thinking about reputation and trust. The first one shows how repeating a game may convince identical players to behave honestly, whereas if the same game is played only once (or a finite and known number of times), players would rationally not trust each other and contribute to a (Nash) equilibrium where social welfare is not maximized. Such a family of games is more aptly seen as describing trust (or lack thereof), not reputation. This characterization is useful in order to grasp the concept of “encapsulation of interests.” It should be kept in mind that the constituent game is not necessarily a prisoner’s dilemma—remember the example of the lieutenant colonel and of the merchant Trifonov in Dostoyevsky’s The Brothers Karamazov, which is based on a different type of strategic interaction. The concept of reputation is more intuitively seen within situations where agents are not all equal, but belong to different types. A series of questions naturally arises when one approaches a potential counterpart in a business relation. Is he an able producer or a mediocre one? Is he going to deliver on time, or is he known for his tardiness? Is he an able manager or public administrator or a poor organizer? In particular, making room for these considerations allows us to realistically assume that people are not all equally (im)moral. Certainly, one of the characteristics that we are very much interested in learning from our counterparts is a measure of what, as a rough approximation, we may call their innate qualities, one of which may be trustworthiness. The presence of repeated interactions (or of a communication channel substituting for them) is still central in this context, because it is the knowledge that one’s past performances are known to others that may convince players to exert effort today to acquire a better reputation that will be valuable tomorrow. One feature of most, if not all, models of this type is that people’s characteristics are thought to be predetermined. However, they are in part endogenously determined. People and organizations may decide to invest today’s resources not only to produce for today, but also to improve their capabilities of producing for tomorrow. For example, this is what individual investments in education are all about. Quite often, organizations also decide to spend considerable resources in reorganizing themselves, an investment that may be seen as a conscious effort to change their type. In fact, we can assume that a producer always
Reputation and Trust
43
has to decide between two forms of investment. First, producers, given their current characteristics, must decide how much effort to exert in order to produce today’s output. This is an investment in reputation. Second, producers must decide how much energy to put into improving their characteristics. Since both types of efforts would compete for the same scarce resources, it would be interesting to examine how these two strategic dimensions interact in different contexts. Unfortunately, this issue has not attracted much attention so far.16 To conclude, if there is a consensus in the literature, it is about the nuanced complexity of the concepts of reputation and trust. Intense research effort is under way to better understand, both theoretically and with recourse to laboratory experimentation, how real people behave when faced with interactions where trust and reputational considerations are relevant. Talking about real people, we left Julius Caesar with the problem of deciding what to do with the inhabitants of Uxellodunum, the village that surrendered to him after a stubborn fight. Caesar had been very aggravated by this episode of rebellion, which he saw as an act of inexcusable obstinacy. Also, he feared that others could follow the same example. On the other hand, Aulus Hirtius, the author of the story, reminds us that Caesar was known for his leniency.17 In the end, presumably, several considerations shaped his decision. We are informed that Caesar’s reputation was such that “being under no fears of being thought to act severely from a natural cruelty [he] resolved to deter others by inflicting an exemplary punishment on these. Accordingly he cut off the hands of those who had borne arms against him. Their lives he spared, that the punishment of their rebellion might be the more conspicuous” (Caesar 1869). We are left asking ourselves what explains Caesar’s tough decision. Possibly, as Hirtius suggests, he could get away with an act of incredible cruelty without damaging his reputation, because it was so well established. Hirtius’ explanation represents a sort of double somersault, at the end of which Caesar can be evil precisely because, in fact, he is good. Caesar knows that everybody knows that he is good, and that rationally he would not decide to squander his investment in reputation by turning bad, so that it is understood by all that a single episode of evilness means nothing. Or maybe, Caesar knew that with that act of cruelty, he would indeed damage his fame, but he thought that he had good
44
Reputation and Trust
reasons for spending part of his reputational capital. We know that he was in a hurry to end his campaign, because his term as a consul was drawing to a close and he needed free rein to prepare his political future. He calculated that achieving that objective was worth the price. Yet another possibility is that after observing the obstinacy of those who kept taking up arms against him, Caesar may have come to believe that they did so because they thought that, should they be defeated, thanks to his leniency they would get away with it. He may then have reconsidered the value of his present reputation, concluded that he needed a different one, and acted accordingly. Last, Caesar’s decision may have been influenced by the fact that he lived in a world where cruelty was more widespread than it is today. Maybe his overall behavior with respect to the barbarian tribes that he subjugated was what in game theoretic parlance we would define as a “mixed strategy” or, in the present case, a random alternation of forgiveness and cruel rage.18 The truth is that it is hard to tell: the concepts of trust and reputation are complex ones. Believing in others, who may then betray us; or having opinions of what others are, and then finding out how wrong we can be—this is the stuff of life and life’s tragedies. Our discussion of these concepts, which has favored internal consistency and simplicity, should not lead us to forget these broader facts.
3
Reputation and Good Public Governance
Public management is not an arena in which to find Big Answers; it is a world of settled institutions designed to allow imperfect people to use flawed procedures to cope with insoluble problems. James Wilson
We aim to go beyond intuitive arguments in favor of reputational incentives, to understand their role in shaping the observed behavior of public administrations and how their power may be harnessed to improve public governance. First, we need to clarify what we mean exactly by “good public governance.” Answering this question is not as easy as it may seem, fundamentally because we would want public governance to satisfy many characteristics at the same time—to be effective and efficient, to reflect the needs and aspirations of the people, to be responsive to prevailing values, et cetera. Unfortunately, vague concepts quite often are of no use. We develop two different definitions of good governance that, by being rather narrow, have the advantage of being applicable. Once it is clear what “good” we are aiming for, we will discuss a framework where reputational incentives play a central role and that accommodates some important aspects of public administrations. Here, the reputational incentives take the form of “career concerns”: the desire that bureaucrats have to step up the career ladder. I argue that career concerns play an important role in motivating public officials and in explaining the overall behavior of public administrations. Focusing on career concerns is not meant to exclude other types of individual motivations that are arguably important, and we will also discuss these. As a result of our efforts, I will provide a first description of how an appropriately engineered institution could support an advanced form
46
Reputation and Good Public Governance
of reputation-based governance—a form of public governance that is good, and where reputational considerations play a central role. The Quality of Public Governance
When can we say that public governance is good? Such an apparently simple question turns out to be rather problematic. We would like public governance in general, and public administrations in particular, to be many things at the same time. We expect public administrations to solve our problems efficiently, and we appreciate courteous public officials. We would also like public administrations to carry out their business on the cheap, because we do not enjoy paying high taxes. A particularly thorny issue regards the definition of the goals of public administrations, and of the public sector more generally. The problems that public administrations solve should reflect the needs of society at large, but eliciting such needs, defining the public good, is difficult. Moreover, in doing all these things, we generally admit that not only the identification of objectives is important, but also how they are pursued. Vibrant democracies are those where the citizens find venues for participation in a public discourse and may constructively condition government even between election days. Also, most people would like public administrations to take their social responsibilities seriously, by dedicating resources, for example, to fostering equal opportunities in the workplace, by hiring disadvantaged people, and so on. Such a wide array of goals is obviously at odds with the desire to formulate a viable definition of what good public governance may be, or, said otherwise, any simple definition that we may propose is bound to be highly selective. And selective we will be, by adopting first a rather narrow concept of quality of public governance: We say that public governance is good according to the efficiency criterion if public administrations pursue their goals (whatever they are) with a minimum use of resources. To borrow from the language of philosophy, this definition corresponds to a concept of instrumental rationality, expressing an exclusive focus on how a specific end is achieved, without any consideration on the worthiness of that end. Public administrations could be efficient in pursuing unworthy objectives, and there are even tragic examples in history of public ad-
Reputation and Good Public Governance
47
ministrations that efficiently pursued intentions that were indefensible from a moral point of view. The efficiency criterion alone is not enough for good public governance, and we supplement it with a further one, concentrating on the identification of the goals to be pursued. We define public governance to be good according to the democratic criterion if the goals pursued by public administrations are in accord with the desires of the people. Defining what the desires of the people are requires tackling the issue of aggregation of individual preferences, a very thorny issue from which we want to stay away. Again in the spirit of simplicity, we affirm that in order to be in accord with the desires of the people, active participation of the people in governance is a necessity. At this time, we want to avoid any discussion on what this exactly means, and we simply posit that public governance where citizens may participate more, all else being equal, is relatively more democratic.1 Just as an efficient public administration may pursue morally indefensible goals, so we may witness publicly stated goals that are very much in accord with the desires of the people and are morally impeccable but are pursued by very inefficient administrations. While the two criteria of quality of governance are distinct, they are not unrelated. For example, citizens tend not to identify with public administrations that are very corrupt and, as such, inefficient, so that their actions typically do not entice much democratic participation from below. Public governance, to be democratic, has to exhibit some degree of efficiency: an utterly inefficient administration, strictly speaking, cannot induce the type of identification of the people that is a precondition for the functioning of a democratic process. An overall good public governance is one that is good according to both the efficiency and the democratic criteria, where public administrations are efficient in pursuing their goals, and their goals are coherent with the desires of the people. Often, in what follows, we will be concerned only with the efficiency criterion of good governance, that is, we will discuss how well public administrations are pursuing their goals, whatever they are. In this chapter in particular, we describe the functioning of public administrations with the aim of clarifying how reputational incentives may make these administrations more efficient. Toward the end of the book, I will argue that reputational considerations can be usefully
48
Reputation and Good Public Governance
exploited not only to increase the efficiency of public administrations, but also to make public governance more responsive to the will of the people. It is at that point that we will appreciate the full message of this book: reputational incentives, if appropriately engineered, are conducive to public governance that is good according to both the efficiency and the democratic criteria. Incentives in Public Administrations
For public administrations to be efficient in pursuing their goals, the actions of many public officials ought to be appropriately coordinated so as to contribute to some common objectives. Within big administrations, a desired coherence of individual behaviors cannot simply be achieved in a command-and-control fashion. True, job descriptions may be lengthy and detailed but, just as all contracts are to some extent incomplete, so, too, are job descriptions. The observed behavior of a given public administration is the result of the aggregation of many individual actions, and it is by no means obvious that they smoothly produce the desired collective outcome. To the contrary, the possible contrast between individual incentives, the actions that result from them, and the collective purposes of an organization, creates a fundamental problem of alignment of incentives that we should consider. A critical assessment of how to achieve the necessary alignment of individual actions within a public administration, then, requires an analysis of the individual motivations. Dissatisfaction with red tape is widespread, and the rhetoric on the need to run public administrations as if they were private corporations is part of the political discourse almost everywhere. Researchers of public administrations tend not to share this somewhat stereotyped view, point out that the overall context of public service imposes the systematic differences that we observe with respect to private organizations, and stress that the perceived inefficiencies of public administrations should be considered in the light of such differences.2 James Wilson lists a series of reasons why public bureaucracies may not operate so efficiently. One of them has to do with the incentives faced by bureaucrats: “Public executives have weaker incentives than do private executives to find an efficient course of action. The former have no property rights in the agency; they are not, in the language of economists, ‘residual claimants’ who can
Reputation and Good Public Governance
49
put into their own pockets the savings achieved by greater efficiency” (Wilson 1989, 349). Bureaucrats are not the residual claimants of the successes of their organizations in the sense that, while their fortunes may depend on the outcomes of their actions, their compensation typically is not increased dramatically as a result of an important success. It is true, in principle, that bureaucrats’ salaries can be linked to their performance, but, as Perry et al. (2006) note, “merit pay and pay-for-performance systems in the public sector have generally been unsuccessful, have little positive impact on employee motivation and organizational performance, and fail to show a significant relationship between pay and performance.” Such practices hinge on the effectiveness of a difficult assessment, and when they have been introduced (for example, as part of New Public Management–styled reforms), they have often led to acrimony between the people who are awarded the bonuses and those who are not. Saying that bureaucrats are not residual claimants, however, does not imply that their incentives are weak. In fact, anyone who has some hands-on familiarity with the world of public administrations would conclude that, with exceptions, bureaucrats are anything but apathetic figures who try to do as little as possible. On the contrary, most of them, at least those who have not given up on the idea of obtaining a promotion, seem to be very busy participating in both formal meetings and informal corridor talks, in a work environment where office politics tends to reach a high level of sophistication. Bureaucrats do face strong incentives, and if the observed outcome is unsatisfactory, quite often it is not because individual incentives are weak but because they are poorly aligned with the common goal. In a world of self-interested employees, then, what makes a public official tick? Several studies have appraised the incentives of bureaucrats and assessed to what extent they may differ from those generally available in the private sector. Perry et al. (2006) summarize a vast body of literature and conclude that bureaucrats are affected by a wide range of motivations. An active participative role of the employee is conducive to a positive affective reaction to the organization, but not necessarily to better performance. Challenging and well-defined goals lead to good performance, as does in general the presence of a perception of meaningfulness of the assigned tasks. The strength of individual motivations is
50
Reputation and Good Public Governance
characterized by the presence of strong interaction effects, so that a factor may be significant in one organizational context, but not in another. Public sector and private sector employees also tend to differ significantly in their motivations. The former tend to have a stronger service ethic, they value the social impact of their work more, and they have a stronger sense of social justice.3 Regardless of salaries, employees who feel that their contribution to the common goals of the organization is valued, and who think that they have a say, if not in what the administration does, at least in how it does it, are often happy workers who put effort and enthusiasm into what they do. To the contrary, workers who judge their superiors to be unjust in their decisions and poor leaders overall often decide to play the “exit” strategy by making little effort and by not caring much about what they do.4 A good way to summarize a substantial part of bureaucrats’ motivations is to refer to the concept of “career concerns.” Strictly speaking, career concerns refer to a public official’s hope and expectation that if he does well today, his career may benefit tomorrow, either through promotion, or by landing of a new job elsewhere. However, we are advised to interpret the concept flexibly, in the sense that a promotion, or a better job elsewhere, is not a necessary prerequisite for this concept to apply. The good things that can happen following virtuous behavior today may be in the form of being chosen to lead or to participate in more exciting projects, the satisfaction of being respected by one’s co-workers, or of feeling valued by one’s superiors. It was Holmström (1982) who provided the starting point for the study of the “economics of career concerns,” describing a situation where self-interested employees face incentives that are not explicitly written in a contract, as happens, for example, with a salesman whose salary is contractually contingent upon his sales. The hope of a promotion is not a contractual obligation of the employer, under any contingencies, so the incentives that result from such a hope follow a contract only implicit in nature. The strategic situation described by Holmström, which we adopt as our general model to explain the behaviors of public officials, is like the learning games with reputational incentives that we discussed in the previous chapter. A bureaucrat produces an output, the result of a single task, that depends positively on his talent, his effort on the job, and a random shock that may be interpreted as the effect of “luck.” The easiest
Reputation and Good Public Governance
51
way to describe the implications of Holmström’s model is to focus on its two-period formulation, which we can think of as corresponding to the first and second part of a bureaucrat’s career. In the first period, the salary is fixed. In the second period, it depends positively on the output achieved during the first period. To fix ideas, imagine that if the output is high in the first period, then the employee obtains a promotion, but not otherwise. As is typical in the learning games already discussed, we assume that the ability of the bureaucrat is known only to himself, so that observing a high output at the end of the first period could be due to a combination of talent, effort, and good luck. Such output leads to an updating of prior beliefs about the agent’s ability: the bureaucrat’s boss will think it more likely that he is of the high-quality type. A public official can then try to produce a high output by putting a lot of effort into the job, thus signaling high ability, which in turn makes a promotion more likely in the second period. The formal solution of Holmström’s model indicates that in the first period the incentive to work hard is actually too strong, in the sense that bureaucrats would put in more than the level of effort desired by society as a whole. Intuitively, this is so because young bureaucrats are highly motivated to work in order to signal that they deserve a promotion. On the other hand, bureaucrats put no effort into the second period, since they cannot expect further promotions in the future. Overall, “young people will overinvest in labour supply because the returns from building a reputation are highest when the market information is most diffuse,” which “seems to accord nicely with causal empiricism (including introspection)” and with “some scientific evidence as well” (Holmström 1982). Another result concerns the role of the external shock that affects output. The solution to the model shows that effort in the first period is higher when the variance of that shock is smaller—or, to put it differently, when luck plays only a minor role. When the observed output is more informative of talent and effort, the bureaucrat has a stronger incentive to work hard in the first period. The role played by the dispersion of the error term in the output equation is particularly relevant. If it is high, then the observed output is a poor signal of the sum of talent and of effort on the job, the building of a reputation by the employee is
52
Reputation and Good Public Governance
problematic, and we observe a lower effort in the first period of Holmström’s two-period model. In an extreme case, if the environment is so noisy that the observed output does not carry any information about the talent of the employee, then there would be no incentive to exert effort in order to build a reputation. Also, a very noisy environment presents an opportunity to embezzle public resources allocated to the task, that is, to be corrupt, since such a loss could simply be attributed to a stroke of bad luck. The abatement of this noise, the provision of accurate information on the output of the actions carried out by the administration, is a key requirement of good governance and, as we will discuss, is a property of the type of reputation-based governance that this book favors. The description provided so far, centered on the presence of an implicit contract linking current performance with the future wage, succinctly represents a salient aspect of many employment relations even outside the public sector. For example, it describes the incentive problems of private managers, whose contracts are typically written in broad terms, with an implicit understanding that their future salary may change according to their performance today (Fama 1980). Implicit contracts are, however, particularly relevant within public administrations, where formal contracts providing explicit incentives to employees are notoriously scanty. Holmström’s simple model can be enriched to obtain further useful insights, following two companion contributions by Dewatripont, Jewitt, and Tirole (1999a, from now on DJT1; and 1999b, from now on DJT2). DJT1 extends Holmström’s model in several dimensions and, in particular, allows for the presence of a multiplicity of tasks that the bureaucrat may be asked to carry out, an aspect that mimics the lack of focus often observed in bureaucracies. In the words of Wilson (1989, 349), “The public officials must serve a variety of contextual goals as well as their main or active goal and they have little guidance as to what might constitute an acceptable trade-off among these goals.” The presence of multiple tasks goes hand in hand with the observation that bureaucracies must serve a variety of contextual objectives, and it increases the difficulty of monitoring performance. In DJT2, the extended model is adopted in order to study a set of stylized facts of public administrations, leading to the following conclusions. First, an increase in the number of tasks carried out by public officials weakens their incentives. In fact, having many tasks to do raises
Reputation and Good Public Governance
53
the overall level of noise, defined as the sum of the random shocks related to each task. This complicates the signaling of a high talent via a high level of effort on the job. As a consequence, the employer (or the general public, the “principal” of bureaucrats) has an interest in keeping public officials focused on a limited number of tasks, an intuition that again is not new within the community of researchers in public administrations.5 DJT2 also considers the possibility of having a “fuzzy mission,” meaning a situation where not only do bureaucrats carry out more tasks, but also employers do not know for sure how public officials allocate their total effort among these tasks. Under certain assumptions, it can be shown that such an added element of uncertainty also dampens incentives. A possible solution to this latter problem is for the employer to hire “narrow specialists,” that is, professionals who excel in carrying out a limited range of tasks. The employer would not have to worry about their engaging in activities they are not meant to do, because they would rather spend their time on tasks for which their specialization makes them more productive. The monitoring problem faced by the employer would then be less severe, and the employee in turn could be granted a greater degree of autonomy. Such insight is coherent with many existing narrative accounts on professional bureaucracies. Reputational Incentives and Good Public Governance
The career-concern model that we described allows us to see better in what sense the use of an appropriate Internet-based information system could lead to an improvement of public governance. We imagine a world where the people would be allowed to access an Internet-based information system to rate the effects of public policies. Measures of such ratings would be visible to all, and also the information system would make available a lot of relevant information on policies, such as their costs and the names of the people responsible for them. The effects of such an arrangement would be twofold. First, it would increase personal accountability, because a public evaluation of the outcome of a policy would immediately be linkable with the identity of the people in charge of it. In terms of the career concern model, this would facilitate monitoring the output of individual actions and, as we discussed, would strengthen
54
Reputation and Good Public Governance
reputational career-related incentives. The high visibility of a wealth of policy-related information, in turn, would allow for better monitoring from inside the organization, allowing for more timely and better informed readjustments. In the world of public administrations, the status quo has a strong degree of latency, and often the underperforming workers of an inefficient branch have an ex-ante expectation that they will not be disturbed—particularly if the career incentive is not a strong motivation in their behavior. As an example of these difficulties, consider the case of redundancy within an administration. Redundancy adds flexibility, as it allows an organization to better cope with new tasks and, by definition, to immediately find a substitute executer for an existing task, should the relevant department fail. I do not refer only to situations where there is a literal duplication of organizational functions. Within a big public administration there is always some overlap of capabilities between branches, so that, for example, the decision to allocate a new task to one or another may be influenced by consideration not only of their specializations, but also of their relative efficiency. Let us consider, however, a case of literal redundancy, where exactly the same service may be provided by more than one branch of an administration and where the clients of the service may choose among alternative suppliers. Demand would shift toward the branch that is more effective in getting things done. Assume that the resources allocated to each branch are either fixed or, as often happens, they display a high degree of persistence. As a consequence, the most efficient branch would eventually become overworked, while bureaucrats at the less efficient units would be able to put in less effort. These forces could eventually lead to an equilibrium where the perceived benefit of the services is the same across units, because an excessive burden on the best offices rations their services so that, at the margin, the positive differential in quality just equates the disutility of waiting for delivery. The result would be a perverse situation where the best offices are punished by a higher workload. In such a situation, where the rewards of exerting effort are dubious and highly uncertain, all branches have an incentive not to put much effort into what they do, to avoid a reorientation of demand that would increase workload. Crozier’s “vicious circles” again come to mind.
Reputation and Good Public Governance
55
The policy-related information that would be available under reputation-based governance would help address problems of this type. Inside the administration, its directors would be able to take notice of any miscarriage and to react appropriately. Should that not happen, the availability of the same information would empower administration’s clients to ask for change. More generally, the availability of very informative timely data on the output and outcomes of policies would put a brake on various types of rent-seeking behavior, an issue that we will discuss at length. Also instructive in this respect is the case of redundancy that occurs, at a wider level, within decentralized governance models, where many functions that the centralized structure carries out at the center are duplicated locally. In this case, organizational redundancy may allow for intergovernmental competition, at least to the extent that citizens may “vote with their feet” by moving to the locality that better suits their preferences (Tiebout 1956; Brennan and Buchanan 1980). Also, it is sometimes possible to have competition among different administrations even without citizens physically relocating, when they are allowed to apply to different territorial offices for a given service—say, the delivery of a passport.6 Decentralized governance has both merits and shortcomings. On the one hand, local knowledge allows public officials to make more-informed choices on local projects. However, leaving the power to local bureaucrats presents an important monitoring problem and opens the possibility of capture by local elites, who may hijack the local political process to bend local policies to their desires, something that they would find difficult to accomplish if decisions were made in the faraway national capital. Of course, national governments, too, can be captured by national elites. However, the problem of capture is arguably different at the national level, where there are more institutional checks and balances compared to the local level, because, as Bardhan (2002) notes, “collusion may be easier to organize and to enforce in small proximate groups involving officials, politicians, contractors and interest groups,” while at the local level “risks of being caught and reported are easier to manage, and the multiplex interlocking social and economic relationships among local influential people may act as formidable barriers to entry into these cozy rental havens.”7 This problem is now widely recognized in policy circles, in the context of development policies and an organization such as the World
56
Reputation and Good Public Governance
Bank, complaining that “too often, services fail poor people,” proposes “[putting] people at the centre of service provision: by enabling them to monitor and discipline service providers, by amplifying their voice in policymaking, and by strengthening the incentives to serve the poor” (cited in Bardhan and Mookherjee 2006). In this respect, reputation-based governance would have two distinct positive consequences. On the one hand, the availability of much policy-related information would facilitate monitoring, both by controllers at the national level and by the citizens, the media, and nongovernmental organizations, nationally and locally. The local “cozy rental havens” would find it more difficult to thrive. Second, consider that the possibility of elite capture follows from a series of political advantages that elites enjoy. First, compared with ordinary people, elites have more political resources generally. Also, they typically enjoy lower political participation costs. Finally, they face a collective action problem that is intrinsically easier to solve than the one faced by ordinary people, who are more numerous. Note that the Internet, per se, does not make it more difficult for elites to capture the political process, and in some instances it may actually help them, as Rethemeyer (2006, 2007) convincingly argues. However, the way the Internet is used within reputation-based governance would tend to level the playing field between elites and ordinary people, whose political participation would be favored by a legible policy space (an issue which we will discuss at length) and by the very low costs of political participation. How all this could happen precisely depends on the characteristics of the information system supporting reputation-based governance. So far, these characteristics have only been hinted at by means of analogy with already existing examples of Internet-based reputational systems. In the next chapter, we discuss the main characteristic of an information system capable of supporting a reputation-based advanced model of public governance.
4
Entities, Roles, and Functions of Reputation-Based Governance
Legibility is a condition of manipulation. James C. Scott
Internet-based reputational systems compute measures of reputation that are based on the assessments made by individuals who have a stake in the object of a transaction or decision. Who these individuals are and what they assess, in each of the cases that we considered in the introductory chapter, was easy to determine. If the purpose is the trading of goods, such as on eBay, then what is assessed is the quality of the experience of trading, and the people who do the assessment are the traders. If the system is for finding good restaurants, then it is restaurants that are assessed, and the people who dine at them do the assessments. If the purpose is to adopt an Internet-based reputational system that aims to improve public governance, identifying who should assess what is not as straightforward. Public governance is the result of an uncountable number of acts that affect many people, so that it is difficult to determine what its constituent parts are and who exactly are the stakeholders of each single component (however defined) of public governance. To settle on the use of a suitable reputational system for the purpose at hand, we must be very precise about the concepts on which it should rest, and this is one of the goals of the present chapter. The main gist of the solution I propose is to assume that public governance can be seen as the sum of conceptually uniform building blocks, or policies. In particular, when looking at a single public administration, it is useful to think about its general action as a bundle of such
58
Entities, Roles, and Functions
Policies’ space
Selected policy
Policy selection
Policy execution
Completed policy
Public administration
Firms’ space
Contractor selection
Contractor
Steps of Governance = actors; rectangles = processes; thin arrows = an input or an output; thick arrows = actions by actors. FIGURE 4.1
NOTE: Hexagons
policies. It is then their outcome that receives an assessment. Regarding the identity of who does the rating, for the time being we simply assume that it is possible to identify the people who are affected by the policies and who are in the best position to assess them. The assumption that public governance is defined by a sum of policies, far from being innocuous, not only influences the characteristics of the information system supporting reputation-based governance, it also affects in a subtle way some important characteristics of governance—an aspect that we will discuss later in the book. For the purpose of fixing ideas, we begin by presenting a strippeddown characterization of public governance, where we focus only on the role of the actors that are called to materially carry out a policy. The realization of a policy is the result of three subsequent stages, or processes, depicted as rectangles in Figure 4.1: policy selection, contractor selection, and policy execution. A public administration carries out a given policy with the help of one firm (or contractor). Almost invariably, public policies use products and services that are produced outside of the public sector, so explicitly introducing a procurement problem adds a touch of realism to a characterization that otherwise is an extreme simplification of a very complex reality. The input of governance is represented by the policy space, that is, the set of countless possibilities for action that the administration faces, and the firm’s space, that is, the set of potential contractors. We will
Entities, Roles, and Functions
59
gradually add texture to this very simple model. In this chapter, in particular, we introduce the stakeholders who, by assessing the outputs of policies, are instrumental in the computation of measures of reputation. We discuss the execution proper of the chosen policy in Chapter 7. After the policy-selection phase, the administration chooses one or more firms to help out in the execution of the selected policy, a public procurement problem that we consider in Chapter 8. For the time being, we focus only on how a given policy is executed—that is, on the efficiency criterion of good governance—and we do not consider how policies are chosen. We consider this issue in Chapters 9 and 10, where we discuss methods of policy selection that are also reputation-based and that are democratic in the sense that they hinge on a high level of public participation. Organizing the Policy Landscape
We distinguish between two broad types of policies: programs and projects. A project results in deliverables that are left for future use, such as a new bridge, a school, or a clearly specified outcome of a scientific research project. A program is more akin to a service to be provided over a period of time. The upkeep of a road in a given year, a job training program, and the functioning of a call center providing counseling services to disabled people are all examples of programs. Building a road, establishing a training center that is needed to run a training program, and establishing a call center, on the other hand, are cases of projects. In order to run, programs typically need some sort of infrastructure to be available, which is the result of previously executed projects. We further assume that policies can be segmented into parts, or deliverables, each one leading to a well-identified outcome. For example, the project to build a road connecting town A to town B could be divided into several portions of the road, possibly to be assigned to different contractors. Project deliverables could correspond to “work packages” within a “work breakdown structure”; both familiar terms in the project management literature (see Project Management Institute 2005, 113– 116). In what follows, we use the terms deliverables and work packages interchangeably. Policies usually do not exist in isolation from one another. We identify the following possible relations. A connecting relation applies
60
Entities, Roles, and Functions
when two policies are related but are not hierarchically ordered. For example, a road may give access to another road, and in general a (section of a) network to another (section of the same) network. A job training program in a region has a connecting relation with an analogous job training program in another region. Two policies that are connected could alternatively be described as two work packages of a single policy. In slicing up the whole of governance into policies, and policies into work packages, the issue of granularity naturally arises, the determination of which is inevitably affected by a degree of arbitrariness. A consequential relation applies when a policy follows another, both logically and in time. For example, the paving of a road follows its construction. A research project, once finished, may be followed by a further project that builds on the results of the first. A consequential relation between two projects is analogous to the relation between two work packages of a project where one needs the completion of the other in order to be started. Programs may be related to projects by the presence of a consequential relation: the maintenance of a road (a program) is consequentially related to its construction (a project). Policies may be usefully aggregated into bigger logical units. A nesting relation applies when a policy is part of a bigger one, which we call the parent policy. For example, a bridge may be part of a freeway, and a program aimed at training minorities may be part of a more encompassing training policy. A nesting relation implies the presence of hierarchies between the nested and the parent project, just as a work package is lower in hierarchy with respect to the project to which it belongs. Programs may be nested within projects, and vice versa. For example, the building of a training center may be necessary for the implementation of a training program. In this case, a project is nested within a program. Also, a training program may be needed within the construction of a public work, a case where a program is nested within a project. The relations present among policies define a rich web of relations that could also be represented graphically, in ways that are similar to the sociograms that are used in the social psychology literature. A sociogram, a tool for the analysis of networks first used by the Romanian psychiatrist Jacob Levi Moreno, along with Helen Jennings, is a visual representation of members of networks, indicating with arrows the presence of a direct relationship between a couple (Wasserman and Faust 1994, 11–12).
Entities, Roles, and Functions
Road A-B
Training program in A
Road B-C
61
Road C-D
Training program in D
Connecting relation
Training center in D
Nesting relation
Consequential relation FIGURE 4.2
An Example of a “Policygram” indicate projects, and pentagons indicate programs.
NOTE: Rectangles
Members that are most central in a network are highly connected with other members. The relations among policies also could be represented graphically in a similar way, through what we may call a policygram.1 Figure 4.2 provides an example. In the illustrations, the project to build a road from city A to city D is divided into three sections. The road is also meant to facilitate the construction in city D (which we assume to be rather isolated otherwise) of a building to be used to carry out a training program, part of a wider training policy that affects both city D and city A. Figure 4.2 illustrates the mutual relation among all these interventions. Policies may be aggregated into logical units, characterized by the presence of nesting relations. Two such broad encompassing policies are indicated in the figure: the construction of the road from A to D, and the overall training program in city A and D. Policygrams convey the idea that a set of policies form a policy landscape, in the sense that the mutual relationships among the building blocks of public governance may be represented using a spatial metaphor. They suggest that one of the effects of the atomization of public governance into policies is precisely to transform a generic problem of public governance into a landscape formed by a set of constituent
62
Entities, Roles, and Functions
blocks—the policies—mutually connected by a rich web of relations. Also note that policygrams could be made selective to represent a policy landscape as it applies to particular categories of subjects—for example, policies aimed at students, or policies relevant for a certain territory. A policy landscape organized in such a way is highly “legible.” I borrow this term from Scott (1998), who argues that we may interpret many initiatives of (early) modern states as attempts to make society more legible. For example, the imposition of a common national language allows the immediate intelligibility of all written records produced without the need to interpret; the introduction of permanent surnames eases the identification of citizens, and thus facilitates important state activities such as the collection of personal income taxes and army conscription. Even scientific forestry belongs to this category. Introduced in Prussia and Saxony at the end of the eighteenth century, through schematization and tree monoculture, it facilitated not only the maximization (at least in the short term) of land yields, but also their computation. Such state activities may be welcome or unwelcome. Compulsory vaccination, which needs a legible society to be carried out effectively, is an example of the former. On the other hand, many ambitious plans to modify reality were manifestations of human hubris. Some of the cases of what Scott calls “high modernism” have been hardly commendable, or have even turned into tragedies: collectivization in Soviet Russia successfully imposed a rational uniformity to a vast countryside, but at the price of millions of lives. These are cases of the state trying to improve the legibility of society, motivated by the fact that a “society that is relatively opaque to the state is . . . insulated from some forms of finely-tuned state interventions, both welcome (universal vaccinations) and resented (personal income taxes)” (Scott 1995, 77–78). Here, we are concerned with the opposite—that is, with the extent to which the state is legible to society. I surmise that a scarcely legible state is insulated from the control and the interventions of its citizens. In Manzoni’s The Betrothed, a lawyer aptly named Azzeccagarbugli (“picker of quarrels”) speaks to Renzo, an honest but uneducated worker, in “latinorum,” an obscure technical language that, being unintelligible to the worker, vaguely sounds like Latin (Manzoni 1909, 28). Just as society has always looked for ways to escape from finely tuned state interventions, so the state has an interest in being unintelligible to society
Entities, Roles, and Functions
63
in order to avoid control and escape accountability. Max Weber affirmed that all administrative organizations are preoccupied with secrecy as a way of insulating themselves from outside control (see the discussion in Bendix 1977, 452). Weber’s observation refers to an attitude that has effects within a given institutional context. Here, the focus of interest is not the strategies that take place within a fixed setting, but the institutional context itself. I argue that the state would prefer institutional set-ups more conducive to secrecy, where the strategies highlighted by Weber are more easily carried out.2 The state, having limited material and cognitive resources to read a complex reality, desires to modify society so as to make it more legible. Similarly, the people have limited time and bounded rationality to make sense of the state,3 and, in principle, they prefer it to be structured in such a way that it is easier to understand. Policies legislating transparency, which in many countries have arguably made a contribution to legibility, may be interpreted in this light, and it is not by chance that such policies are often decided after some traumatic event focuses public opinion and creates a policy window.4 Proposals for administrative reform, then, should also be judged on their implications in terms of the cognitive demands that they impose on the people. Unfortunately, too often these preoccupations have not been heeded. Consider some of the excesses of New Public Management, which suggested the disaggregation of large administrations into smaller units, in the same way as large private corporations, during the past few decades, had moved away from the traditional structure that dominated in the first part of the twentieth century (Reschenthaler and Thompson 1996). Particularly telling was the outcome of “New Zealand’s pioneering [New Public Management] structural changes [that] have left a country of 3.5 million people with over three hundred separate central agencies and forty tiny ministries, in addition to local and health service authorities” (Dunleavy et al. 2006). Such a “luxuriant administrative fragmentation,” again in the words of Dunleavy et al., was hardly easily legible to those citizens whom it was meant to serve.5 The Pitfalls of Simplification
A coherent structuring of the policy space is a precondition for reputation-based governance, because personal accountability is needed
64
Entities, Roles, and Functions
in order to compute measures of reputation for the actors of governance. Accountability, in turn, needs the clear assignment of responsibilities that only a well-ordered policy space can guarantee. Coherent structures, unfortunately, always come at the cost of simplification. In particular, for reputation-based governance to be applicable across the board, whatever public administrations do should be definable as a policy, and for any policy, there should be a group of identifiable beneficiaries who are allowed to assess it. Quite clearly, this cannot be literally true, because within the intricacy and complexity of the public sphere, there will be cases where either one condition or the other, or both, are not easily satisfied. For example, it would indeed be difficult to specify the beneficiaries allowed to assess the policy for the management of bilateral international relations between two countries. Moreover, many public activities that could reasonably be defined as a policy would in fact produce intermediate services to be used as inputs by the administration itself. Simplification has its cost, and the previous reference to the case of New Zealand allows us to look at the issue from a wider angle. Commenting on the reforms that took place there, Robert Gregory (1995) writes that “the main thrust of state-sector reforms has been to encourage all public agencies in New Zealand to treat all their tasks as if they were or could be made into production ones.”6 Such an approach was coherent with New Public Management–inspired reforms, which have attributed particular emphasis to the computation of performance measures in the public sector, even if the idea of measuring public administration performance is certainly not new.7 The computation of performance measures goes hand in hand with the idea that the activities carried out within an administration can be precisely defined and described, and that they are, in other words, a product. However, administrations are of different types, and some of them are not easily amenable to the computation of performance measures.8 This, in turn, often led (and leads) to trouble when applying forms of performance management, the most important of which is represented by behaviors of a “teaching-to-the-test” type—that is, by the substitution of the goal of an administration with the maximization of the index that is meant to measure its performance.9 The problem, in fact, results from a general law. Whenever we apply a conceptual straightjacket to a complex reality, reality always attempts to escape out of the straitjacket’s inevitable cracks and holes.
Entities, Roles, and Functions
65
So, before that happens, let me clearly suggest that, when applying reputation-based governance, the straitjacket should be worn rather loose. Every attempt at legibility has its limits, and, when applied thoughtlessly, it leads to unintended (and sometimes unpleasant) consequences.10 The imposition of legibility on the policy space upon which reputationbased governance rests is no exception. Certainly, we can make the point that, in fact, most state activities could fit the mold. For example, those policies leading to intermediate services could be rated by those public officials who use those services as an input. Also, we may decide to believe that, coherently with the paradigm shift toward more horizontal and “netcentric” organizations (Jones and Thompson 2007), the public administrations’ production processes will become progressively less fragmented. Thus, well identifiable teams, working with job-oriented structures, will increasingly execute policies from beginning to end (Reschenthaler and Thompson 1996). Last, to refer again to the example of bilateral relations between country A and B, these would indeed be difficult to assess. However, the groups of beneficiaries of the services of country A’s consular network in country B, for example, could easily rate the services that they receive. Still, we should exert caution and adopt a flexible mindset to avoid the pitfalls that always follow when we identify a model with a recalcitrant reality that it is meant to represent. We will reinforce this message in Chapter 8, where we will discuss some concrete cases of possible applications of reputation-based governance, and where the need for flexibility will emerge again. The Actors of Governance
The policy landscape is juxtaposed to an equally rich web of relations among many individuals and organizations that, in various ways, play a role in shaping both the choice of policies and the ways in which they are executed. We single out those actors who are essential within reputation-based governance. The individuals involved are the citizens who are affected by the policies and public administrators. Individuals may play more than one role: the administrator running a policy may also be the citizen who is affected by other (and possibly even the same) policies.
66
Entities, Roles, and Functions
The organizations involved are, first of all, public administrations executing policies and the firms that they may contract.11 Organizations are created, live their lives, and eventually cease to exist. Sometimes they merge with other organizations, or they split to form two or more entities. It is important to keep track of these changes, so as to allow the past record of an organization to matter even when, for example, it decides to change its name or to merge with another organization.12 Public administrations are typically divided into parts (departments, branches, etc.), each one of which may be responsible for a policy. Also, policies may be the result of collaboration between several administrations, or of parts of them. When this is the case, this information should be recorded appropriately, together with a weight that expresses the degree of involvement in the policy. Such weights are important in computing measures of reputations, because they determine how the assessments of policies propagate to people and to organizations. Suppose that a disastrous policy has been carried out jointly by two administrations, with the first one receiving 95 percent of the resources and the second one a meager 5 percent. It would be reasonable for the stigma of failure mainly to fall on the leading administration, and that it would only slightly affect the administration that contributed to the execution of the policy as a very secondary partner. Administrators work for an administration and assess the performance of firms that collaborate with them in the execution of policies. Administrations execute policies with the help of firms. A contractor is a firm in charge of the execution of (part of) a policy. When more firms, either as partners or in subcontracting relations, partake in the execution of a policy, the information system associates them with weights depending on their share in the work, so as to guarantee appropriate computation of their reputations. We could imagine a situation where the assessments of policies allow us to compute measures of reputation that propagate not only to the public administrations, their administrator, and the firms contributing to their execution, but also to the political parties that are politically responsible for the policies. In this case, political parties should be recorded appropriately within the information system. Political appointees—those people who have a political responsibility within part of an administration— could also be allowed to build up a reputation. They are often members of
Entities, Roles, and Functions
67
a legislature, but this need not always be the case. Often political appointees are either formally or informally affiliated to a political party. Sometimes they are “independent,” and sometimes, particularly within coalition governments, they are considered as being affiliated with a political area rather than a party. This is inevitably an area where details are hard to pinpoint and are highly context-dependent. For example, sharply distinguishing between administrators and politicians makes more sense in some contexts than in others. In the United States, the distinction tends to be blurred, and, according to a widespread view, the federal bureaucracy has two masters: the president, on the one hand, and Congress, which is endowed with sweeping powers, on the other.13 What matters is that reputational information, at least in principle, could also propagate to the political parties and to the political appointees that are involved in the execution of policies. Administrators assess contracting firms, and citizens assess the outcome of a policy. We say that the citizens who may assess a policy belong to the constituency of that policy, which we define as the set of actors affected by a given policy. For example, if the policy is a project to build a school, its constituency could be defined as the group of people who live in the area where the school is located, the personnel working there, and the students. For network-type public works, such as a freeway, the constituency may also include people who live far away from it. In the case of a job placement service, its constituency could be the set of unemployed people. Each group may be allowed to assess policies in different ways. For instance, teachers may be interested in characteristics of the school that are different from those parents care about—say, they may prize the availability of comfortable office spaces to use in their nonteaching working hours, while playing down the importance of having a good gym. To reflect such differences, members of each group would typically be allowed to assess a policy on different dimensions. In principle, constituencies may include not only individuals but also organized entities, such as whistle-blower associations, consumer and other civil society organizations, that could also be allowed to assess policies. People could then voice their opinions as both individuals and members of a constituency-representative organization. Such a possibility, however, raises a series of problematic implementation issues. One of them is the double counting of opinions, since the organization’s participants
68
Entities, Roles, and Functions
would also be able to assess policies individually. How to appropriately count the assessments of organized entities is not obvious. Weighting collective assessments according to the size of the organization’s membership would establish an incentive to artificially inflate them. Also, organizations would be more prone than individuals to engage in collusion. Political parties, for example, could encourage the formation of ad hoc organizations for this purpose, possibly leveraging on fake membership by party activists. Enterprises, also, could encourage the formation of similar “astroturfing” organizations to support their projects and to discredit their competitors.14 Certainly, a set of precautionary measures could be taken to limit scheming and collusion. For example, citizens’ organizations could be required to post their budget sheets and to declare the origin of any contributions they receive. Their members could be required to register within the system and to declare any link that they may have with firms or administrations. However, in most if not all democratic contexts, such measures would rightly be frowned upon and would be considered a serious infringement of personal liberties. It would then seem more reasonable to allow only individual participation in the system. Reputation-based governance reduces transaction costs of various kinds, and to the extent that organizations are instruments to counter transaction costs, they would be less needed than before. However, the costs of carrying out watchdog activities would also be lower, establishing an incentive for the formation of citizens’ organizations. Later in the book, I will argue that a rich ecosystem of nongovernmental organizations could take advantage of the availability of a vast array of policy-related information that would be routinely available. The Functions and the Units of Governance
The emerging governance model, which already significantly enriches the simple description of Figure 4.1, combines two distinct elements. On the one hand, we have a series of actors, mutually linked, who are stakeholders and participants in the public governance process. On the other hand, we have a characterization of what public governance does as a sum of policies that are also linked by a rich web of relations. We now characterize how the relevant actors may carry out the chosen
Entities, Roles, and Functions
69
policies, so as to describe the most relevant choices that confront the application of the governance model. In our case, most importantly, choices have to be made on the criteria for being permitted to express one’s assessment of a policy outcome, on the modality for expressing such assessments, on how individual assessments are aggregated, and on how the aggregated assessments of policy outcomes are used to compute reputations. We call the several tasks that have to be accomplished in the process of governance functions. I employ the term function here in a loose mathematical sense, as a rule that takes an input and produces an output. For example, one function regards the criteria determining whether a person is allowed to rate the output of a given policy. In this case, the input is the set of all people, and the output is a presumably smaller group formed by all the people that are indeed allowed to express their assessment. Such function is the rule describing how to decide whether, for a given policy, a certain person belongs to that group or not. Some functions are related, and we group them into logical units. For example, all the functions that describe what has to be done in order to compute reputation measures form the reputation unit. Grouping functions into units, which allows the families of procedures that accompany reputation-based governance to be identified, may be seen in two complementary ways. First, as part of the information system: the reputation unit, for example, is that part of the information system that carries out all the computations necessary to transform the various assessments of policies into measures of reputation. Also, a unit may be seen as an institution, both in an informal sense—a set of rules that constrain behaviors—and in a formal one, because the functions of governance need appropriate organizations of support. We define six units, shown in Figure 4.3 together with their reciprocal relations, each one comprising a varying number of functions. The first five are: the policy unit, allowing for the description of policies; the constituency unit, determining who has the right to assess a given policy; the voice unit, describing how such an assessment may take place; the assessment unit, describing how individual assessments on the outcome of a policy may be aggregated, and the reputation unit, permitting us to compute measures of reputation, using as an input the aggregate assessments of policy outcomes. The main characteristic of these units is that,
70
Entities, Roles, and Functions
Policy Unit
Constituency Unit
Voice Unit
Assessment Unit
Functions:
Functions:
Functions:
Functions:
1 2 3
4 5
6 7
8 9 10
Publisher Unit
FIGURE 4.3 NOTE: Arrows
Reputation Unit Functions: 11 12 (13) (14)
Functions: 15, 16
The Governance Units and Their Functions represent flows of information.
with one possible exception that we will discuss in the next chapter and that needs not bother us now, they work sequentially, in the order in which they are listed, so that in the workflow of reputation-based governance the assessment unit, for example, is upstream with respect to the reputation unit and downstream with respect to the voice unit. For this reason, each one needs all the preceding units to be in place in order to function. The sixth element is the publishing unit, determining what information has to be transmitted from one unit to the other, and what has to be made public. It is horizontal in its function and is needed by all units. Coding policies: The policy unit. The policy unit contains general information on policies, including their work breakdown structure or division into work packages. It includes all relevant accompanying information on policies, such as their expected and actual costs, Web links to any parliamentary discussions that may be of relevance, cost-benefit analyses, etc. The policy unit includes the following functions: F1: Determine whether a chosen policy is to be included into the policy unit. If reputation-based governance is applied to all policy domains, then the rule simply states that all policies to be executed are included in the system. If not, only policies within the relevant policy domain would be included.
Entities, Roles, and Functions
71
F2: Document the work-breakdown structure of the policy and, in particular, its granularity. This information is important because different responsibilities may correspond to different work packages. F3: Determine what accompanying information has to be included for each policy. It is the working of the policy unit, in particular, that imposes the possibility of identifying the actions of public governance as a wellordered policy landscape. In principle, the policy unit could be present in isolation and still serve an important purpose: allowing citizens to access information on policies and their cost.15 Such a possibility is increasingly seen as essential to empowering citizens as active players in the governance process. For example, a United States law requires the “existence and operation of a single searchable website accessible by the public at no cost that includes for each federal award of federal financial assistance and expenditures (excluding individual transactions below $25,000 and credit card transactions before October 1, 2008): (1) the amount; (2) information including transaction type, funding agency, the North American Industry Classification System code or Catalog of Federal Domestic Assistance number, program source, and an award title descriptive of the purpose of each funding action; (3) the name and location of the recipient and the primary location of performance; and (4) a unique identifier of the recipient and any parent entity” (U.S. Public Law 2006). According to our definition of the policy unit, to implement the requirement of this law, F1 would state: “Include all policies with federal funding above $25,000”; F2 would not be relevant; and F3 would require the information specified in entry 1-4 above. Being allowed to assess a policy outcome: The constituency unit. It determines the constituency of a policy, that is, the set of the stakeholders that should be allowed to rate its outcome. A constituency is typically formed by a number of homogeneous groups (let us call them subconstituencies). The functions of this unit are: F4: Identify the subconstituencies, which together form the constituency of the policy. F5: Determine membership of each subconstituency.
72
Entities, Roles, and Functions
One simple way to establish membership of subconstituencies would be to determine a maximum number of policies that each person can assess in a given year. In this way, citizens would self-select the policies that are most important to them. The shortcoming of such an approach is that a person could be barred from expressing opinions on the outcomes of policies that affect him. If the maximum number of assessments is kept high enough, however, this problem should not be too severe. This solution would not work for very contentious policies, on which minorities of citizens hold strong ideological opinions and could decide to assess them even without having direct experience of their outcomes. So, for example, a public health center performing abortions could hardly be subject to this regime. It can be argued that policies of this type, while being highly visible precisely because they are contentious, are only a small minority of all policies and could be treated ad hoc for the purpose of establishing their constituencies. An advantage of leaving citizens to decide for themselves which policies to rate, up to a maximum number of policies allowed, is that it would limit disparities between citizens who have different propensities for participation, possibly arising from a different valuation of the time that it takes to do so. Considering that a different valuation of time is typically related to social and economic characteristics, this property would be particularly desirable. An important issue in the working of the constituency unit regards the information on citizens’ identity that it requires.16 At one extreme, the constituency unit would be fed data from a national civil registry, which would be allowed to require citizens to identify themselves in order to access the system. Such a solution would present two important advantages. First, it would minimize the cost of access. For example, in contexts where citizens own smartcard-type identity cards, they could very easily access the information system supporting reputation-based governance from any terminal equipped with an identity card reader. Minimizing the cost of access in turn amplifies participation and typically reduces social class bias. Second, a system that identifies its users impedes highly interested individuals from casting multiple votes on specific projects. However, using a national civil registry would also be a reason for concern, particularly in countries such as the United States (as opposed, for example, to European countries), where for historical reasons there is resistance to forms of citizen identification.17 An alternative would require
Entities, Roles, and Functions
73
citizens to voluntarily register in the system in order to participate. This extra participation cost would quite probably result in less participation and a higher degree of class bias.18 Voice activities: The voice unit. We include the functions that regulate the assessments of outcomes of policies in what, as a tribute to Hirschman (1970), we call the voice unit. It implements the procedures connected with such activities. For example, the opinions expressed by administrators on a public procurement relation could be in draft for a certain period, so as to allow the contracting firms involved to express any counter-argument that they may have. Any such procedure should be appropriately codified and implemented. The voice unit also determines the characteristics of the assessments, and different subconstituencies may be allowed to carry out assessments on different dimensions. For example, the doctors working in a hospital may be allowed to rate the quality of the common spaces that are for their private use, but the patients would not be. The various individual assessments are the inputs for the computation of the policy’s aggregate assessments by the assessment unit, to be discussed next. To summarize, the voice unit carries out two functions: F6: Determine the type and dimensions of voice activities for each subconstituency and for each type of policy. F7: Determine the process according to which the assessments of policies are finalised, such as the rules for posting rebuttals, etc. It is obviously debatable whether people would be willing to spend time on assessing public policy outcomes. These assessments are a public good, and, as such, we should expect them to be underprovided, since people would rather not spend their time on an activity that mostly benefits others. However, current experience with online reputation systems indicates that people tend to participate much more than purely egoistic considerations might lead us to expect.19 There may be an analogy, here, with the paradox of voting. The paradox lies in the fact that people tend to vote much more often than purely selfish motivations would dictate. This has been widely studied, and there is ample evidence that “voter turnout is inversely related to the costs of voting,” where the “factors that may impact the costs to vote and turnout include the
74
Entities, Roles, and Functions
weather, registration requirements, time required to think about the voting decision, distance to the polling place, and so on” (Feddersen 2004). Reputation-based governance would make assessing the outcome of policies as cheap as it can be, as it would provide a legible policy space and an appropriate technological infrastructure accessible to all. Inexpensive voice activities would maximize citizens’ participation. An important issue concerning the working of the voice unit is the privacy of participating citizens. If the system requires some strong form of individual identification, then an obvious concern would be the publicity that individual assessments could receive. In many instances, we may assume that citizens do not have anything against letting everyone know that, for example, they strongly disliked the outcome of a given policy. We could go as far as envisaging a situation where individual assessments are part of an enlightened public discourse, as in an idealized town hall in New England or a Greek agora in classical times. However, publication of one’s assessments would raise two broad issues. The policies assessed could deal with sensitive areas, such as medical treatments, where privacy rights must be managed with care. Also, the secrecy of one’s assessments, just as the secrecy of one’s vote, is an important guarantee against forms of patronage, an obstacle to elite capture, and, more plainly, a safeguard against the control of the citizen by the state. The voice unit should certainly guarantee such privacy rights, by appropriately making assessments anonymous before passing them along to the next unit. Citizens should be confident that the voice unit does, in fact, make their assessments on policies anonymous without letting anyone use them unlawfully. This is a technological issue only in part and, more importantly, has to do with the strength of the underlying institutional texture. It is not unlike what happens in traditional electoral processes, which can also be tampered with in various ways in principle, and in reality they often are.20 The problem of safeguarding the secrecy of digitally recorded opinions, however, is not confined to reputationbased governance, but to any model of governance that posits forms of computer-mediated political participation. Computing aggregate assessments: The assessment unit. The finalized individual voice activities that are produced by the voice unit feed into the assessment unit, whose role it is to aggregate them. On the one
Entities, Roles, and Functions
75
hand, we have the assessments of policy outcomes. Also, administrators may assess the work carried out by contractors within a procurement relation. Finally, contractors may assess the administrators that they interact with. Since more than one firm may be involved in the execution of a policy, such assessments would be carried out for as many firms as there are. The relevant functions are: F8. Aggregate individual assessments to produce subconstituency assessments, and then aggregate the overall assessments of subconstituencies into a summary assessment of the policy outcome. F9. Aggregate the assessments of the role of administration by the different contracting firms. F10. Aggregate the assessments received by firms by administrations. The output of the assessment unit then feeds into the reputation unit, which we will discuss next. Computing reputation: The reputation unit. The assessments of past policy outcomes, together with the assessments of the performance of contractors in helping public administrations to carry out policies, are used to compute measures of reputation for public administrators, for firms, and eventually for other actors of governance. Individual reputations in turn may propagate to the organization that the individual belongs to—for example, the reputation of a public administration may be computed as a function of the reputations of its administrators. The necessary computations are taken care of by the following two functions: F11. Establish how policy outcome assessments determine reputations of individual actors (administrators and, if desired, political appointees) and of firms. F12. Establish how the reputations of administrators (and possibly of political appointees) determine the reputation of administrations (and, if the possibility is contemplated, of political parties). There are obviously different methods of computing a reputational score, an issue that we discuss in the next chapter, where we will also consider the possibility that the reputation unit carries out two more functions, which are indicated as F13 and F14 in Figure 4.3. Publishing information: The publisher unit. The publisher unit determines which information produced by the other units must be visible
76
Entities, Roles, and Functions
and to whom. The problem of producing information may be divided into two parts: F15: Determine which information from each unit must be transmitted to the unit downstream. F16: Determine which information produced by the other five units should be made public, and in what form. The last of these two functions includes the production of summary views of the data and of statistics of various kinds. It turns out that the overall organization of information on policies that we discussed would have interesting implications for the production of policy-related statistical information, a topic that we discuss in Chapter 6.
5
Computing Measures of Reputation
A good reputation is better than money. Publilius Syrus
Individuals assess policies, and these assessments aggregate to form a summary measure of reputation. There are, however, many ways of computing this measure, and we can predict right away that a given set of policy-output assessments could give rise to many alternative types of reputation indicators, each one characterized by properties that would probably have an impact on the characteristics of governance. In this chapter, we present two alternative measures of reputation that, with their differing characteristics and properties, will give us an insight into the multiplicity of reputation indicators that could be computed in principle. We consider first the computation of reputation indexes, which are the most immediate measures of reputation, since they depend directly on the aggregate evaluations of policies. The details of how the computation of such indexes is carried out have important implications that deserve discussion. Then we present an alternative measure of reputation, the reputational budget. While the reputational budget is also computed from the assessments of policies, and it can be expressed as a function of the reputation index, it is different in other respects. We present both measures using a simple public procurement example, anticipating a theme that we will discuss at length in Chapter 8. Last, we consider the concept of “beauty contest reputation,” a type of reputation that depends on the mutual relationships among the assessments of the policy outcomes.
78
Computing Measures of Reputation
Though this concept of reputation is in some ways ancillary in our reflections on reputation-based governance, it has properties and implications that are worth discussing.
The Reputation Index
Imagine a public administration executing a number of public projects, each one with the help of a single contracting firm. Once the projects are finished, citizens are allowed to express their assessment of their quality. These assessments serve as the basis for the computation of measures of reputation. Table 5.1 shows an example. For simplicity, assume that the works have been executed in only two years, 2011 (“this year”) and 2010 (“last year”). Assume further that there is a total of nine firms, identified in column 1 of Table 5.1, and that each one of them has been contracted for one project in each year. Also, only a single characteristic of the project (say, “overall quality”) is assessed by the public. Votes range from 0 (very bad) to 5 (excellent).1 Column 2 of Table 5.1 reports the overall assessment received by each project, computed as the average (or possibly the median) of all the assessments cast by citizens. Column 3 indicates the size of the project, expressed in thousands of euros. We assume that project outcomes lose relevance as time goes by, with a time-discount factor equal to 0.8. Column 4 shows the time-discounted values of projects that, one year back in time, are simply equal to the face value of the projects multiplied by the time-discount factor. The reputation index of each firm is a weighted average of the assessments received by its projects. Doing well on bigger projects, or in more recent ones, influences overall reputation more than doing well in small or old projects. Column 5 shows the reputation indexes of the nine firms, computed as: Ri , 2011 = asi , 2010 ⋅
pvi , 2010 pvi , 2010 + pvi , 2011
+ asi , 2011 ⋅
pvi , 2011 pvi , 2010 + pvi , 2011 EQUATION 1
where Ri,2011 is the reputation index of Firm i, computed at the end of 2011; asi,year is the assessment received by the project carried out by Firm
Computing Measures of Reputation
TABLE 5.1
The Computation of the Reputation Index
Columns 1 Year
1 2 3 4 5 6 7 8 9
Firm i 1 2 3 4 5 6 7 8 9
2
3
4
5
asi,year
Cost of project
pvi,year
4 3 2 0 2 1 4 4 5
150 800 350 50 1,100 945 300 130 75
120 640 280 40 880 756 240 104 60
Avg = 2.78
∑ = 3,900
∑ = 3,120
asi,year
Cost of project
pvi,year
Ri,year
3 1 4 2 4 4 3 4 2
300 700 550 40 950 1,300 600 100 60
300 700 550 40 950 1,300 600 100 60
3.286 1.955 3.325 1.000 3.038 2.897 3.286 4.000 3.500
Avg = 3
∑ = 4,600
∑ = 4,600
2010
Firm i
Year
79
2011
notes: asi, year : Assessment received by the project executed by Firm i in a given year. pvi, year : Present value of the project executed by Firm i in a given year. Ri, year : Reputation index of Firm i at the end of a given year. The time discount factor equals 0.8.
i in a given year; and pvi,year indicates the present value of the project. For Firm 1, the reputation index is obtained as: R1, 2011 = 3.286 = 4 ⋅
120 300 + 3⋅ 120 + 300 120 + 300 EQUATION 1′
80
Computing Measures of Reputation
The firm with the best reputation is Firm 8, which received a high mark (equal to 4) in both years. The example clearly shows in what sense reputation indexes derive straightforwardly from the aggregate assessments received by the projects carried out by each firm, being a weighted average of these assessments, where the weights depend on the relative sizes of the projects and on the time-discount factor.
The Reputational Budget
The reputational index is computed regardless of the scale of activity of the firm. If two firms carried out the same number of projects each year, with one firm doing projects twice as big as the other, and both received exactly the same assessments, they would have the same value for the reputation indexes. An example is provided by the outcomes of Firm 1 and Firm 7 in Table 5.1. The two firms received the same assessments each year, and they have the same reputation (equal to 3.286), even though the projects of one of these firms were twice the size of the other’s. We now discuss an alternative measure of reputation, which we call the reputational budget. While the reputation index is computed in a fairly straightforward manner, I ask the reader to be patient in following the construction of the reputational budget. To facilitate a comparison between the two, we will use the same example illustrated in Table 5.1. The reputational budget depends positively on the size of the activities of the subject for which reputation has to be measured. Obviously, there are many ways to make such a concept operational. To narrow down the field of possible candidates, we establish that the reputational budget should have the following characteristics. a. Every year, all firms are allocated a certain number of reputational points, that is, the units that form the reputational budget. Some are positive, some negative, and the (algebraic) sum of the reputational points that are allocated to all firms in a given year equals zero (that is, if there were only two firms, and one of them received five reputational points in a given year, then the other would necessarily receive minus five points). b. The reputational points that are allocated depend linearly on the size of the projects, so that, all else being equal, if a project
Computing Measures of Reputation
81
is twice as big as another project, it commands twice as many reputational points (be they positive or negative). c. The reputational budget of a firm is equal to the sum of the time-discounted reputational points it has received over time. d. At any time, the sum of the reputational budgets of all firms is zero. Consider Table 5.2. Columns 1–5 are the same as in the previous table. Columns 6, 7, and 8 show the computation of reputational points that satisfy the desired properties of the reputational budget. Column 6 computes a set of weights, wi,year, expressing the assessments in given year weighted by the size of the respective projects. For each firm and year, these are equal to the overall assessment, asi,year, times the relative share of the present value of that firm’s project, within the overall present value of all projects carried out by all firms: wi , year = asi , year ⋅
k
pvi , year
∑ pv
i , year
i =1
EQUATION 2
In the summation symbol at the denominator of the fraction, k is the number of firms (and of projects executed in a given year), 9 in our example. For example, for Firm 1 in year 2010, this weighted assessment equals: w1, 2010 = 0.154 = 4 ⋅
120 3120 EQUATION 2′
Column 7 shows the demeaned assessments. They are equal to the assessments minus the average weighted assessments, which is simply the sum of the weights of column 6: k
asdemi , year = asi , year −
∑w
i
i =1
EQUATION 3
1 2 3 4 5 6 7 8 9
Firm i
Year
150 800 350 50 1,100 945 300 130 75
∑ = 3,900
Avg = 2.78
Cost of project
3
4 3 2 0 2 1 4 4 5
asi,year
2
∑ = 3,120
120 640 280 40 880 756 240 104 60
pvi,year
4
The Computation of the Reputational Budget
2010
Columns 1
TABLE 5.2
5
∑ = 2.292
0.154 0.615 0.179 0.000 0.564 0.242 0.308 0.133 0.096
wi,year
6
1.708 0.708 −0.292 −2.292 −0.292 −1.292 1.708 1.708 2.708
asdemi,year
7
∑=0
204.923 452.923 −81.846 −91.692 −257.231 −976.985 409.846 177.600 162.462
RPi,year
8
9
300 700 550 40 950 1,300 600 100 60
∑ = 4,600
Avg = 3
Cost of project
3 1 4 2 4 4 3 4 2
asi,year
∑ = 4,600
300 700 550 40 950 1,300 600 100 60
pvi,year 3.286 1.955 3.325 1.000 3.038 2.897 3.286 4.000 3.500
Ri
∑ = 3.304
0.196 0.152 0.478 0.017 0.826 1.130 0.391 0.087 0.026
wi,year −91.304 −1,613.043 382.609 −52.174 660.870 904.348 −182.609 69.565 −78.261
−0.304 −2.304 0.696 −1.304 0.696 0.696 −0.304 0.696 −1.304 ∑=0
RPi,year
asdemi,year
notes: wi,year : Assessments weighted by size of project. asdemi,year : Demeaned weighted assessments. RPi,year : Reputational points at the end of a given year. RBi,year : Reputational budget at the end of a given year. The time discount factor equals 0.8. See also Table 5.1.
1 2 3 4 5 6 7 8 9
Firm i
Year 2011
∑=0
113.619 −1,160.120 300.763 −143.866 403.639 −72.637 227.237 247.165 84.201
RBi
84
Computing Measures of Reputation
For Firm 1 in 2010, the value is: asdem1,2010 = 1.708 = 4 − 2.292 EQUATION 3′
Column 8 computes the reputational points at the end of a given year, equal to the demeaned assessments of column 7 times the present value of the project carried out in that year: RPi,year = asdemi,year · pvi,year EQUATION 4
For Firm 1 in 2010, the value obtained is: RP1,2010 = 204.923 = 1.708 · 120 EQUATION 4′
Column 9 shows the reputational budget, for each firm, at the end of 2011, simply obtained as the sum over the two years of the time reputational points: RBi,2011 = RPi,2010 + RPi,2011 EQUATION 5
For Firm 1, the reputational budget is equal to: 113.62 = 204.923 − 91.304
EQUATION 5′
The sum, as required by condition (c) above, time-discounts past reputation points, since these are computed using the present value of projects—see equation (4). Also, the sum of the reputational budgets of
Computing Measures of Reputation
85
all firms equals zero, and the reputational points in a given year sum to zero (see the bottom of column 9 and, for each year, the bottom of column 8), as required by property (a) above. Moreover, note that such computations satisfy property (b), as again it is made clear by a comparison of Firms 1 and 7. The two firms receive the same assessment in each year, and the latter carried out projects that are exactly twice as big as the former. As already noted, their reputational index is identical, but the reputational budget of Firm 7 is twice that of Firm 1 (227,237 reputational points vs. 113,619). At the end of each year, a number of firms will have a positive number of reputational points, and others will have a negative reputational budget. In principle, these reputational points could be traded, with firms that have positive reputational budgets selling points to other firms. Trading of this kind could balance the reputational budget of all firms, for the simple reason that, by construction, the reputational points given out every year sum to zero according to property (a) above, so that the stock of reputational point is also equal to zero. Uses of the Reputational Budget
Consider a situation where a service provider of some kind has to be chosen, and where, in principle, reputation may play a role. To fix ideas, let us imagine again a public procurement problem, where a public administration has to select a firm to help it in carrying out some policy. We will discuss in more detail reputation-based forms of public procurement in Chapter 8, so here it suffices to anticipate the key elements that allow us to appreciate how adopting one measure of reputation or another may have differing implications. Using reputation considerations in public procurement is not new. In the United States, a reform in public procurement that took place in the early 1990s (part of the “reinventing government” effort) effectively forced public officials to consider the bidders’ reputations in source selection, together with the proposals’ price and perceived quality (Kelman 2002). There, the firms’ reputations are assessed by a public official, mainly after consulting a database where public administrators routinely record the performance assessments of the firms they have interacted with. The type of reputation-based governance we are discussing is different in many respects, and, most importantly, the
86
Computing Measures of Reputation
assessment of firms, instead of being the precinct of bureaucrats, would be carried out with the fundamental contribution of the citizenry. Reputational measures could enter into the picture in at least two ways. The administration could be willing to pay more in order to hire a firm that, all else being equal, has a higher measure of reputation. Alternatively, the administration could allow bids only from those firms whose measure of reputation is above a certain threshold. Adopting one measure of reputation or another has differing implications that we now discuss. As we noted above, the reputation index does not depend on the relative importance of the firm, while the reputational budget does, and it conveys information on how good, or bad, a given firm was, and on the size of the projects that it carried out in the past. Defining a cutoff point to be admitted into the bidding process that equals, say, the first quartile of the reputational budgets of all firms, would select firms that typically are of above-average size—or that are of medium size but performed relatively well in the past. Such a rule would make sense when the size or the complexity of projects suggests that the job should not be contracted to a small firm. In these cases, the administration may wish to deal with a big firm that has done at least reasonably well in the past, but at the same time the administration may be willing to give smaller firms a chance, provided that they have shown they are exceptional performers. Using the reputational budget to define a cutoff point, in other words, would incorporate a trade-off between reputation and size into the considerations of the public administration. On the other hand, if the project is small, the public administration may wish to define a cutoff rule for presenting a bid that is a function of the reputation index, rather than the reputational budget, expressing in this way an indifference about the size of the bidding firms. The cases where we admit the possibility that firms can trade reputational points are of particular interest. To discuss the implications of such a possibility, let us refer again to the economics of career concerns. Holmström (1982) concludes that career concerns may be too strong in the early stages of a career and too weak as retirement age approaches. As in all finite-horizon moral-hazard repeated games,2 as the end date of the game gets closer, the future expected stream of payoffs tends to vanish, and cheating today becomes more advantageous compared to the early stages of the game (see Kreps et al. 1982). In this context, trading reputation may
Computing Measures of Reputation
87
be of help in setting the incentives right. Fama (1980) already noted that managers may be disciplined by the fact that their reputation today positively influences the expected pay they may secure by moving to another firm—a situation where the trading of reputation is in fact taken care of by the presence of a market for managerial labor. Trading reputation could also be achieved by transacting brand names. To analyze how such a possibility affects life-cycle incentives (of firms’ owners) to exert effort—again, in the spirit of Holmström—Tadelis (2002) considers a model for both moral hazard and adverse selection, concluding that brand-name trading provides the sought-for incentives to build a good reputation—that is, past good behaviors are rewarded— and that it alleviates the moral-hazard problem even with short-lived agents. However, Tadelis’s results, as he himself states, are conditional on the assumption that clients do not observe this reputation trading (brand names). Since the model is also one of adverse selection, it follows that clients who realize that a firm is buying another firm’s reputation would take it as a sign of bad quality.3 Tadelis concludes that, for example, secrecy “is not reasonable in all industries (e.g., medical practices),” which, for the particular case cited, comes as sobering news for the reader. According to Tadelis, the “model of the paper . . . seems to fit small owneroperated firms with transient clients, such as restaurants and small service businesses, but is harder to link to larger firms.” Note also that the market for reputation illustrated by Tadelis is in fact “only” a market for brand names. Sales of brand names are perforce rare, and the ensuing thinness of the market does not bode well for its efficient functioning. In this respect, the computation and use of a reputational budget would be innovative in several ways. First, reputational points could be traded at any time, allowing for the presence of a market that functions well and is not limited to rare purchases of someone else’s brand name. Second, a market for reputational points could more easily guarantee the secrecy that is necessary in Tadelis’s treatment, given that the very visible brand names not only would not change hands, but also would possibly become irrelevant over time, to the extent that the reputational budget would contain all relevant information on the value of reputation. Different institutional solutions may be devised in order for reputation points to be exchanged and for reputational budgets to be balanced. For example, firms could be required to balance their reputational budget by
88
Computing Measures of Reputation
the end of each year, so as to carry over to the next year zero reputational points. For this purpose, a market could function, say, for a period of time at the end of each year, where firms could trade reputational points. Just as market forces impose the balancing of a budget, so an appropriate institutional setup could impose on firms the obligation to start a new year with non-negative reputational points (which is just the same as asking all firms to equalize their reputation indexes by trading reputational points). In public procurement, such a result could be obtained by adopting a rule that public administrations can only admit tender proposals from firms whose reputation is not below average. In this way, firms that have negative reputational budgets and are interested in public contracts would have a compelling incentive to buy reputational points, and firms with positive reputational budgets would increase their profits by selling them. A reputation market like this would also present firms with interesting opportunities for arbitrage. Assume that firms differ in technology in the following way. Some firms have a comparative advantage in producing at low cost but relatively low quality. Other firms are the opposite. Trading reputational points would then allow firms to exploit their comparative advantages. Firms of the low-cost/low-quality type would find it convenient to focus on what they do well—produce cheaply— expecting they will obtain below-average assessments but knowing they will be able to buy the needed reputational points on the market. Firms with a comparative advantage in quality would also focus on what they do well, thus acquiring reputational points that they would later sell for a profit to the low-cost firms.4 Finally, the presence of a reputational points market leading to a balancing of all reputational budgets would, in fact, allow the public administration not to directly consider reputation in source selection, and thus would require fewer changes in government when making the transition to a reputation-based governance of public procurement. At the closing of the market, all firms would have the same reputation (and zero reputational points). Good performances would be prized because they command reputational points that can later be sold. The role of reputation in source selection would be present, but it would be mediated by the presence of a market for reputation. The concept of the reputational budget, like the framework of reputation-based governance into which it is embedded, would lend itself
Computing Measures of Reputation
89
to applications in domains other than public procurement. For example, bureaucrats who are forced to balance their reputational budgets at the end of each year could trade reputational points using part of their salaries. True, in such a context, an unconstrained market for reputational points could be seen as running the risk of reverting to a pre-Weberian and patrimonial model of bureaucracy. In order to recoup the moneys spent on the reputational points needed to remain viable as a public official, some bureaucrats would have strong incentives for corruption. A more modest solution could be to make a fixed part of the bureaucrat’s salary conditional on the reputational points acquired every year. “Beauty Contest” Reputation
In the characterization of reputation-based governance that was offered in the previous chapter, reputation measures of one type or another are computed for public administrators, public administrations, firms that are involved in public procurement, and, possibly, political appointees and political parties. On the other hand, the citizens who assess policies have been presented as subjects who, with their evaluations of policies, allow the computation of measures of other actors’ reputations but are not themselves objects of possible evaluation.5 However, citizens may be allowed to build a different type of reputation, which for reasons that will soon become clear we call “beauty contest reputation,”6 which is a function of the actors’ assessments of policy outcomes. While the concept of beauty context reputation is not central within reputation-based governance, it represents an interesting and somehow intriguing supplementary type of reputation that is worth discussing, since it would give members of constituencies a reputation and, with it, incentives to do what they are entitled to do well: express assessments on policy outcomes. Beauty-contest reputation expresses the ability of a person to “objectively” assess policies, where the objective assessment is taken to correspond to the overall assessment of the policy output—typically, as discussed in Chapter 4, the average (or the median) of the individual assessments. It derives from the history of the overall coherence of a person’s assessments with respect to the assessments of others. We do not intend to spend time on the definition of appropriate measures of this type of reputation, and simply provide an example to clarify the concept.
90
Computing Measures of Reputation
Suppose that a person systematically assesses the outcomes of policies positively even when most people would think that they were disastrous, or imagine that, vice versa, someone always denigrates whatever the public administration does. Being systematically out of touch with what, according to prevailing opinion, qualifies as reality would result in a poor beauty contest reputation. The necessary computations to produce measures of beauty contest reputation would also be carried out by the reputation unit. We should then supplement the list of functions that such a unit may perform to include the following: F13. Determine the rules according to which a comparison of each individual assessment with the overall assessment by the constituency allows us to compute that individual’s beauty contest reputation. Beauty contest reputation could eventually be used to weight personal assessments of policy outcomes. The better one’s beauty contest reputation is—that is, the record at assessing past policy outcomes—the higher the weight that one’s future assessments receive. People whose past assessments were often outliers would see their future assessments of policy outcomes weighted down. The reputation unit would also take care of the management of beauty contest reputation for the purpose of discounting the assessments of the people who, in the past, produced poor assessments, as described by the next function: F14. Establish the weighting of individual assessments (that are the output of the voice unit) according to their beauty contest reputation, as they feed into the assessment unit.7 Needless to say, deflating the opinions on policies according to a measure of the agents’ beauty contest reputations would be controversial, to say the least. The whole idea of weighting down the opinions of oddball citizens dangerously contradicts the one head–one vote principle of democracy. Also, determining the formula needed to compute such a reputation would require a series of judgment calls that would almost inevitably become contentious. Moreover, beauty contest reputation could introduce an incentive to express conformist assessments, just as in Keynes’s story: the name chosen for this type of reputation serves primarily as a reminder of that risk. At one extreme, such incentives could be seen as representing a type of censorship or, more to the point, a form of “weighted censorship.”8
Computing Measures of Reputation
91
Regardless of these issues, the concept of beauty contest reputation has the merit of clarifying further the fact that several metrics of reputation could be usefully developed. These also include measures that depend on the assessments of policy outcomes in subtle ways, as in the example that we just discussed. From our discussion on measures of reputation, we can therefore draw the following general conclusion. Within reputation-based governance, there is ample room for devising different ways to compute reputation measures for actors in governance.9 As we are about to see in the next chapter, this conclusion has a broader reach, because the architecture of the information technology, that enables our governance model, allows the production of quantitative information on a vast array of aspects of governance.
6
The Production of Statistical Information and the Analysis of Policies
There can’t be measurement without theory. Tjalling C. Koopmans
The set of procedures and units of governance that we described in Chapter 4 provides a coherent framework for the processing of a host of policy-related information. As we noted, there is an underlying correspondence between the institutions of governance and the parts of the information system that we have described. It is useful to consider a further possible identity: that which exists between the organizational dimension of a particular governance model—its formal institutions—and the flow of information taking place within and among them. In this spirit, we described a possible incarnation of a model of reputationbased governance in terms of information flows taking place among different components of an information system (see Figure 4.3). Identifying organizational arrangements with the flows and processing of information is certainly not new and, in fact, amounts to the adoption of the metaphor of an “organisation as a brain,” to use the terminology of Gary Morgan (2006). Apart from the general insights that the adoption of evocative metaphors for the study of formal organizations may provide, what is of interest to us here is more limited in scope. Our departure point is a simple observation: an appropriate information system that coherently organizes data in digital form permits the production of inexpensive and effective summaries of the information contained therein. This fact is apparently obvious, but it represents an important departure with respect
Statistical Information and Analysis
93
to current methods of obtaining quantitative information on governance. In what follows, we first discuss some relevant characteristics of the functions carried out by the publishing unit, so as to understand their implication for the production of policy-related statistics. In the last two sections, we discuss how the data-rich context of reputation-based governance could lend itself to innovative ways of analyzing public policies. More on the Publisher Unit
Within the Internet-based information system supporting reputation-based governance, a great deal of data would be available, containing information on policies, their costs, and the identity of those responsible for them. Also, obviously there would be a wealth of assessments on the outputs observed. While in principle all of these data could be made publicly available, there are reasons for recommending discretion in this respect. First, there may be privacy issues. A policy dealing with the treatment of patients with serious diseases should certainly guarantee their privacy. Second, humans are not particularly good at processing vast amounts of raw data, so the information available needs to be filtered and summarized. Fung et al. (2007), in their comprehensive analysis of disclosure policies in the United States, make the point that even if information cost nothing to acquire and were equally available to all parties, individuals would still be only boundedly rational: they cannot process vast amounts of information efficiently, and they “are prone to a host of cognitive distortions that may lead them to make decisions different from those predicted in a world of perfect rationality” (33). In fact, one of the preconditions for the success of disclosure policies is the effectiveness and comprehensibility of the information provided (Fung et al. 2007, chap. 4). The same reasoning applies to our case. I upheld the merits of a legible policy space, and the importance of the easy readability of information that pertains to policies cannot be overstressed. In the proposed governance model, the task of filtering and summarizing information to make it usable is facilitated by the structure of the policy space, where policies are carefully defined and distinguished from one another. Also, assessments are typically in quantitative form, and both aggregate assessment and reputation measures routinely provide aggregate summaries of
94
Statistical Information and Analysis
important characteristics of governance. In this respect, the possibility of creating effective summaries of vast amount of data is already built into the system. In the description of the functions of the information system provided in Chapter 4, the task of making available policy-related information to the public is the duty of the publisher unit. The publisher unit has two broad responsibilities related to the processing of information. One is to feed information from upstream to the next unit downstream, after appropriate filtering. For example, it receives assessments on policy outcomes from the voice unit and eventually passes them along after some processing (for example, to make them anonymous) to the assessment unit (see Figure 4.3). The other task of the publisher unit is to produce summary information for the benefit of the general public. A relevant issue is what information, besides and beyond what has to be kept confidential because of privacy considerations, we may desire not to make available to the public at large, or to subsets of the public. Within reputation-based governance there is a general emphasis on the availability of much policyrelevant data, and the whole idea of making governance hinge on the reputations of the actors of governance goes hand in hand with the principle that, in order for reputation to form, information on actors must be structured, managed, and published. However, there could be reasons for omitting some of the information contained within the information system. For example, eBay reports feedback not older than six months, and federal public procurement in the United States, as we discuss in Chapter 8, uses reputational information on firms, but only from their performance on projects not older than three years. The idea underlying this arrangement is probably not to be too rash with firms, telling them that they should behave well, but that if they misbehave, they still have an incentive to mend their ways and build a positive record again. In this respect, the fine-tuning of the system would require experimentation and some flexibility.1 The overall success of the publisher unit in making policy-related information available to the public can be measured by how well it provides views of the data so as to effectively summarize important characteristics of governance. To this end, we may imagine a multitiered structure for presenting the information. The most relevant and important information on actors would be published so as to be easily
Statistical Information and Analysis
95
accessible. People wanting a quick overview of a given policy domain would be content with a high-level summary of this kind. Anyone would also be able to browse the information available in greater detail, in order to peruse the past record of an administrator or a firm. The reader could seamlessly browse from one type of information to another and at different levels of detail, taking advantage of the hypertextual characteristics of the World Wide Web. For each policy, the publisher unit provides summaries of the assessments of the policy outputs, like the average, the median, and a measure of dispersion, such as the inter-quintile range (the higher the dispersion of the individual assessments, the more controversial the policy outcome is seen to be), together with the overall ranking of each policy within classes of comparable policies, expressed in terms of percentiles. Different views of the data would be provided to different actors. Some personal information would only be private, while summaries of various types would be accessible to all. In general, great attention should be dedicated to the accessibility and usability of the information, in a context where the taxonomy of policies, and the rigorous definition of the entities and roles of governance, provide structure and simplify the overall task of making a vast amount of policy-relevant information instantly available. Ad Hoc Versus Integrated Statistics and Their Degree of Institutionalization
Among the many types of summary information that the publisher unit would produce, some have a particular public relevance and coincide with what we currently call statistics. For example, the sum of expenditure on policies of a certain category, or the length of new roads of a certain type that have been constructed in a given year, all qualify as useful statistics. They would be produced relatively easily as simple views of the underlying quantitative information on policies that are available. An example of how statistics can be produced in this way is provided by the data on accesses to a Web site. These are produced on a regular basis by dedicated software running on a computer without the need for any outside intervention, apart from an initial effort to set up the service. This way of producing statistical information represents an important innovation with respect to the way we usually obtain and represent
96
Statistical Information and Analysis
quantitative information on policy-related variables. As a consequence of the increasing impact of digital information systems in different government domains, these changes are already under way. The use of “administrative data,” as statisticians call them, for the purpose of producing official statistics is on the rise and is generating increasing interest within national statistical offices and the profession as a whole (for a discussion, see Wallgren and Wallgren 2007). Adopting reputation-based governance would imply an important step forward in this direction. Since the relevant data would be recorded in a digital form, it would not be necessary, as it still is today, to first gather the necessary raw data and then process it to finally produce official statistics. Within reputation-based governance and, for that matter, within any model of governance that rests on an adequate digital information system, the data are already available, and statistics may simply be produced as views of the data managed by the information system. To consider the implications of this innovation, we need to consider the role of official statistics and the institutional framework that allows their production. Today, the gathering and provision of statistics is mainly a responsibility of national governments through a set of dedicated agencies, especially the National Statistics Offices (NSOs), such as the Census Bureau in the United States. In democratic societies, NSOs are granted a degree of statutory autonomy as a way of minimizing the risk of political manipulation of the data that they produce. Trustworthy official statistics play a key role in democratic governance, since citizens need accurate information to form an opinion on policies and on politicians. Official statistics are important in public discourse and provide an essential basis for the formation of the public consensus on which democratic rules depend. Statistical information is often gathered by means of surveys that are carried out on samples drawn from the population of interest. However, surveys have shortcomings, and errors of various types can arise. Moreover, the time needed to administer a survey and to process the data introduces an important time lag between the reference time of the statistical information and the date of its availability. This has obvious negative consequences, particularly in decision-making contexts, where the most recent information, almost invariably, is also the most valuable. The way information is organized and treated within reputationbased governance would cause a shift away from the current practices of
Statistical Information and Analysis
97
statistics production for a potentially vast domain of statistics on public policies. All, or most, relevant information related to the public policies to which reputation-based governance applies would be digitally recorded, and statistics would simply be available as a view of the available data.2 We see an analogy with what is happening in the private sector, where business intelligence and data-mining practices are based on the possibility of extracting information from digitalized repositories to produce valuable summaries. Such information may or may not be public within the organization. Some pieces of information may be deemed to be relevant for all—such as bad news on labor productivity, to incite everyone to exert extra effort—or they may be considered by the management as strategic and confidential. In a public governance context where policy-related information is in digitalized form, public official statistics could be produced similarly. To better frame the problem, I introduce a simple and useful classification of official statistics. We define as “ad hoc” those statistics that, to be produced, need statistical information to be gathered and processed. Most statistics that we are familiar with today are of this type. On the other hand, those statistics that are produced as a view of the digital information within an existing information system, we call “integrated.” The contention is that, while at present most statistics are of the ad hoc type, we will witness a progressive shift toward integrated ones. Another important dimension in which statistics may vary is their degree of institutionalization, by which we mean the prospect for their continuing provision. Ad hoc statistics, which represent the core activities of National Statistical Offices, are highly institutionalized. For example, we may expect national accounts to be produced for an indefinite time in the future. On the other hand, ad hoc statistics that are produced for the first time, or one-shot surveys for which future funding and political support are not secure, are weakly institutionalized because they are vulnerable to a weakening of the political will that initiated them. While the degree of institutionalization of ad hoc statistics may vary greatly, all integrated statistics tend to be highly institutionalized. Once they have been produced for the first time, their future provision depends only on the continuing existence of the underlying information system and on the maintenance of the software producing the necessary views of the data. Strong opposition is needed for these statistics to be
98
Statistical Information and Analysis
discontinued, not just lack of support. To put it differently, ad hoc statistics require adequate funding and need an explicit decision to be continued. On the other hand, integrated statistics, once the continued functioning of the underlying information system is ensured, need an explicit decision to be interrupted. The two types of statistics can also be seen from the point of view of their cost structure. Ad hoc statistics generally have sizable fixed and marginal costs. On the other hand, integrated statistics have negligible marginal costs, since to produce them, once the continuing existence of the information system is guaranteed, all that is needed is the programming of the first release of the statistics. In this respect, integrated statistics share a well-known property of many information goods in the age of the Internet. With respect to other types of information goods (such as music, programs, or written text), integrated statistics have very low marginal costs with respect not only to further copies of the same information, but, more importantly, also to further releases of the same type of statistic in the future. Production of Statistics Under Reputation-Based Governance
The adoption of reputation-based governance would cause a shift toward integrated and highly institutionalized statistics and would make policy-related quantitative information, which today can only be produced at a high marginal cost, easily available. Let us take as an example public works, which contribute to the available stock of public infrastructure. Each piece of public infrastructure would be recorded in the information system, together with the flow of payments to the contractor(s). As the project is completed, its final cost becomes known, allowing for a comparison between realized and predicted costs. A simple view of the data would provide aggregate information on these variables, divided by type of infrastructure and by geographic or administrative area. The system would also record information on the physical characteristics of each good (such as span and width of a bridge or the total floor space of a school). Standardized costs of public works could be computed according to given measurement units, such as cost per mile for roads of comparable width or, for buildings, cost by square foot. The comparison between physical and monetary measures of infrastructure would provide
Statistical Information and Analysis
99
very useful information about the presence of rent-seeking activities, an issue that we take up in the next chapter. All policy domains adopting reputation-based governance would allow for the automated production of statistics, some of which are available today in ad hoc mode and many others that would be new. In the case of public works, a wealth of data would be available on the assessments received by each piece of infrastructure, which would form a very valuable body of information to evaluate the soundness of the overall policy. Moreover, government policies would generate a host of data on the geographic location of the recipients of the benefits, an interesting issue not only for political scientists and economists studying the topic of distributive politics, but also for the policy maker and for the general public. Insofar as most policies regard the provision of services of various types, their treatment within a reputation-based governance framework would usually contribute to creating statistical information on the (public) service sector. This new way of producing policy-related quantitative information contributes to the debate on performance measures, the importance of which have been stressed in New Public Management reforms. The definition and adoption of such measures has been fraught with problems, the main one probably being that “the implications of maximizing performance on a partially valid indicator raises the spectre of sub-optimization . . . or even systematic cheating” (Bohte and Meier 2000). In Chapter 8, we will discuss how this serious problem could be alleviated under reputationbased governance. A more venial sin of “performance management” is its cost, in terms of the resources required to collect the necessary raw data and then to process them to compute the desired performance measures. One implication of reputation-based governance is that these statistics, too, would come to be of the integrated type: their variety, sophistication, and timeliness would increase, their marginal cost would decrease, and, as a consequence, they would be more widely used. Taking a broader view, the ready availability of integrated and highly institutionalized statistics within an appropriate governance framework would contribute to meet today’s needs of measuring the service economy. This represents a priority for National Statistical Offices, whose original task, during the first part of the twentieth century, was framed by the need to measure the industrial economy of yore (see Abraham 2005).
100
Statistical Information and Analysis
Official Statistics and Their Institutional Setting
The diffusion of integrated and highly institutionalized statistics raises a series of questions on the role of statistical information in society. Official statistics have always been much more than mere numbers. They speak of the achievements—and failures—of the authority in power, and the language of the numbers to some extent is the language of power. Just as power has a ritualistic dimension, so have the official numbers that it produces. The Roman emperor Augustus, in his autobiography, recollected with pride his achievements in running the census every five years (Augustus 1991, 7). The name of the ceremony that took place at the end of every census, lustrum in Latin, is a word that has made its way into the English language—“a period of five years,” according to MerriamWebster’s Collegiate Dictionary, but also “a purification of the whole Roman people made in ancient times after the census every five years.” Statistics, with their ritualistic dimension, talk about policies, and as such are part of a discourse on the consensus that power concocts. The need of every political regime for a degree of consensus provides a reason to attempt to manipulate official statistics to depict a rosy reality. Dictatorships manipulate statistics to serve their interests, and when they cannot, they limit their publication. Today’s democracies have found ways to protect themselves against self-serving manipulations of official statistics, mostly by developing professionalized institutions, for example, the National Statistical Offices, which enjoy a degree of autonomy while being subject to some form of public scrutiny. Professionalization and institutionalization contribute in shielding National Statistical Offices from outside intervention to avoid situations where official statistics are manipulated or interrupted on a political whim. These considerations apply not only to existing statistics, but also to the initiation of new ones, an act requiring adequate political will, which could originate within the National Statistical Offices in a technocratic fashion and is also obviously open to pressures from outside.3 The presence of a dedicated institution, with a set of well-specified statutory obligations, provides a natural focus for any lobbying activity in this respect, shaping in turn a public discourse on statistics that democracies should allow and guarantee. These considerations contrast with the situation that would prevail within reputation-based governance. In particular, the sweeping changes
Statistical Information and Analysis
101
that we discussed in the condition and cost structure of official statistics provision raise the question of what organizational traits, within reputation-based governance, the publisher unit should have to adequately fulfill its responsibility for producing policy-related statistics, and on whether it could be accommodated within the structure of the National Statistical Offices. To provide some insight on this question, let us consider that the problem of guaranteeing the provision of statistical information has two aspects. The first one is that the effort of producing the desired data persists. The second is that the way the data are produced is stable, so as to guarantee their comparability in time. Both these requirements are currently met, thanks to dedicated institutions producing official statistics and whose stable internal culture, where professionalization plays an important role, contributes to the development of codified protocols for the production of statistics and to their relative stability over time. Whatever institutional form the publisher unit takes within reputation-based governance, it should guarantee the same two requirements—the continuous production of data and their comparability over time. How the publisher unit exactly carries out its tasks will depend on the general architecture of the institutional setup for the management of reputation-based governance, an issue of implementation that we do not discuss. Broadly, we can distinguish between two polar cases: a centralized and a decentralized architecture. If a centralized architecture prevails, then all the data will be under the jurisdiction of a single entity, within which the computation of statistics would occur. At the opposite end of the spectrum, within a decentralized information system, each administration running part of the overall information system would legally own its data and have corresponding responsibilities regarding any data privacy issue that may arise. The sharing of these data among different institutions would constitute a critical issue that would have to be addressed appropriately by means of data-sharing protocols. Here, the challenges are only partly technical and related more to preventing agencies from manipulating the data to serve their own interests (or their principal’s interests) and guaranteeing the privacy of the actors of the system. Quite clearly, traditional National Statistical Offices have not been designed to serve these ends. One of their current roles, as we discussed,
102
Statistical Information and Analysis
is to provide a place where the provision of (ad hoc) statistics can be usefully “parked,” because the solid institutional framework guarantees their institutionalization. Such a role would not be needed, at least not in the same form, in a context where integrated statistics, by being intrinsically institutionalized, would not need the strong protection provided today by the National Statistical Offices. What would be critical in the new context is the guarantee that the data will be continuously provided from the different units of governance to the publisher unit, so that the latter may produce the desired views of the data. It is the continuous functioning of the overall information system, and the sharing of the data among its parts, that becomes the paramount objective of governance, and not, as it still is today, the institutionalization of ad hoc statistics. Under reputation-based governance, then, the set of institutions guaranteeing the provision of statistical information, among which National Statistical Offices play a central role, would look different from what they are today. If the assessments on policies have to be used to compute measures of reputation, and if these are adopted, say, in the decision process to determine promotions within the public administration, then certainly the overall information system should have standardized characteristics, and the administrative center should be able to perform the desired computation on the raw policy-related data. But contemporary off-the-shelf Internet technologies would also allow much more horizontal access to such raw data, to allow anyone to use and manipulate them as desired. This option is strongly advocated, for example, by David Robinson et al. (2008), who invite the U.S. government “to require that federal websites . . . use the same open systems for accessing the underlying data as they make available to the public at large.” A shift of focus in government, from presentation of information to “creating a simple, reliable and publicly accessible infrastructure that ‘exposes’ the underlying data” (D. Robinson et al. 2008) would allow citizens to collaborate on formulating their own metrics of policy effectiveness, thus fueling a public discourse on official statistics that literally for the first time in history would not be biased toward the top of the power pyramid, but could be truly collaborative and democratic. The center, as represented by future incarnations of today’s national statistical offices, would retain its prerogatives, but their official statistics, now mostly produced as views of the underlying raw data, would compete in a dia-
Statistical Information and Analysis
103
lectical discourse with other initiatives and with other views of the same data created by other societal actors.4 New Methods for the Analysis of Public Policies
Even though we live in the computer era and in the age of highpowered statistical methods, for the purpose of assessing policies we employ quantitative methods of analysis only rarely. Certainly, policies that have contributed to the definition of their epochs—Roosevelt’s New Deal comes to mind—have become the object of many research studies, at least some of them employing sophisticated quantitative techniques. But these are exceptions to the rule, and, for the most part, policy assessment is an activity carried out as part of a political discourse, where quantitative information is often used purely in a rhetorical manner (see Picci 2006b). There are obviously some important exceptions to this state of affairs, the main ones represented by economic policies. These have frequently been analyzed using quantitative methods, and the pursuit of this objective was one of the reasons for the birth in the 1930s of the econometric discipline. In particular, the availability of econometric techniques for the estimation of “treatment effects,”5 together with a wider availability of the necessary micro-data, over the last decades has to some extent popularized studies on the effects of particular policies. Overall, however, the cases mentioned above are little more than a drop in the ocean of policies that are carried out daily, and the average policy is not subject to a serious quantitative analysis of its effects. The contrast between the ubiquity of instruments for dealing with quantitative analysis—computers, statistical and econometric methods, and software—and the infrequency of their use for the analysis of public policies is striking. Two broad reasons explain this state of affairs. On the one hand, the professionals who traditionally analyze public policies are relatively lacking in the necessary technical expertise. They are typically better versed in the use of qualitative methods of analysis, and they have shown, perhaps understandably, little desire to venture outside their turf. Most importantly, quantitative studies have been precluded by a lack of data. We very rarely have at our disposal enough quantitative information on policies and their effects outside the domain of strictly defined economic policies. Lack of data is both a cause and a consequence of a
104
Statistical Information and Analysis
paucity of methods for the quantitative analysis of policies. If the necessary data were available, there would be a greater demand for the development of the appropriate techniques to use them, while it is also true that a greater availability of the latter would increase the demand for the former. This reciprocal relation between the availability of data and the methods that make use of them is well illustrated by the development of the econometrics discipline, and considering this historical antecedent provides some useful perspectives on today’s prospects for progress in the quantitative analysis of public policies. It is no accident that econometrics developed together with the making of those national accounts that were needed to provide a quantitative translation of Keynes’s General Theory (1973, originally published in 1936), within a remarkable series of intertwined parallel developments. There was the economic crisis that followed the 1929 financial crash, with the questions that it raised, and then the emergent solution: Keynes’s theory—or, better, the “neoclassical synthesis” first exposed by Hicks (1937). For several decades, this provided the shared framework for thinking about macroeconomic phenomena—what in economic textbooks is explained under the heading of the “IS-LM model.” These theories, already announced by experiments that were carried out in the prewar years by both liberal and dictatorial regimes, were adopted on a grand scale after the war. The development of national accounts, with the consolidation of the National Statistical Offices, provided the necessary data to the first econometricians to estimate and solve their models, translating Keynes’s theory into systems of equations whose parameters were estimated using methods that also were progressively developed from the early 1930s. In fact, the flagship academic journal of the profession, Econometrica, was founded in 1933.6 While the lack of appropriate computing devices did not preclude the first applications of the newly created methods of analysis, the postwar availability of the first computers provided the means to propose and implement techniques of increased sophistication. The heyday of this tradition corresponds to the end of the 1960s and the beginning of the 1970s. The availability of sufficiently powerful computers and the necessary data, and three decades of progress in the development of the required techniques for estimating, testing, and solving macroeconometric models, allowed economists to provide mathematical
Statistical Information and Analysis
105
descriptions of national economies and their international interactions. One of the aims of these studies was to estimate the best course of action for economic policy instruments, corresponding to the maximization of an appropriate measure of collective welfare. Interestingly, the mathematical instruments that were used to this end—the tools of optimal control theory—were the same as those adopted by rocket scientists at that time to compute the optimal course for their missiles. Space engineers at NASA optimizing the route of the first men to the moon, econometricians in academia, central banks and finance ministries around the world attempting to fine-tune the economy along a path of full employment, all shared various elements in their respective tool boxes. While these functions were applied to a set of apparently disparate ends, they had in common both analytical instruments and technological artifacts—the computers, the use of certain programming languages such as Fortran, and an overall, rather optimistic view of how the world could be transformed to improve—actually, to maximize—human well-being. In the words of Fernández-Villaverde (2008), “there is no purer example of the high modernism impulse in economics than the vision behind the design and estimation of macroeconometric models.” The 1970s brought the first cracks in the previously shared culture. An economic crisis characterized by a mix of economic stagnation and inflation was at odds both with the accepted wisdom and with the forecasts of the econometric models. At the same time, new voices pointed out the presence of inconsistencies in the dominating economic paradigm and started setting the stage for a new direction in theorizing that, cumulatively, would later amount to a sharp shift away from the Keynesian tradition. The whole discipline of econometrics reoriented itself to partly different tasks and tools. However, the progress in terms of data availability, econometric techniques, and, obviously, computing power continued undisturbed. The development of the previous decades had produced effects the relevance of which no longer depended on the original research project that had set them in motion. This sketch, in its simplicity, allows us to appreciate the intrinsic social character of the production of statistics and their use.7 To advance to today’s wealth of economic-related statistical information, an important series of processes, intrinsically social in their nature, had to mature. They took place in the context of particular historical times and of the
106
Statistical Information and Analysis
ideas of those times. Just as the choice of policies is never carried out in a vacuum, so it is for the analyses of those policies. Both are part of a complex discourse taking place within a concrete societal and institutional landscape, using information of various types, among which is quantitative information produced by institutions that are also the children of their time, and contribute to shaping it. This brief historical account, by showing a particular case of the interplay between societal demands, the development of theories to interpret social and economic events, and the growing supply of quantitative information, provides the backdrop that allows us to appreciate the possible implications that reputation-based governance would have for the analysis of policy effects. Today, the ongoing transformations that follow the widespread use of information and communication technologies are generating the changes in the supply condition of statistical information that we have already observed and that are likely to gain momentum in the future. The wealth of integrated and highly institutionalized statistics that would allow citizens to reach an unprecedented degree of awareness on public governance would also encourage the development of new techniques of analysis of policies and of their effects. In this respect, we can identify two main directions of development. On the one hand, reputation-based governance routinely produces aggregate assessments of policy outcomes—the output of the assessment unit. These same assessments, besides allowing the computation of measures of reputation, would also be available to gauge the effects of policies. They could be supplemented by other information—for example, the number of feedbacks recorded in the system could somehow proxy the audience of a given policy. Second, the wealth of statistics on variables related to policy would allow us to adopt forms of policy evaluation that rest on the adoption of various types of models. To provide an example, we summarize a model for the evaluation of the impact of two e-government policies, one carried out by a central administration (the “government”) and one by a local administration (the “region”). Model-Driven Policy Evaluation: The Case of E-Government
Policies do not exist in isolation and are ingrained in a complex web of relations. They are related to each other and establish relations
Statistical Information and Analysis
107
connecting many societal actors of various kinds. Rarely, if ever, is it possible to draw a one-to-one correspondence between policies and their effects. Almost invariably, a single policy has more than one goal, and a single observed effect is the result of the interplay of more policies. As a consequence, it is not straightforward to discern the effects of public interventions, particularly so in the absence of knowledge on their mutual interplay. In other words, to paraphrase the quote from Tjalling Koopmans at the beginning of this chapter, and to adapt it to our purposes, it is difficult, if not outright impossible, to measure the effects of policies in the absence of a theory. While it is clear, at least intuitively, how a theory may succeed in connecting policy instruments to objectives that are economic in nature and, as such, quantifiable, there could be doubts on how to translate such an idea into practice. For this purpose, I present a brief overview of a model, fully explained in Picci (2006a), describing the effects of e-government policies in a multilevel governance context, where both a central government and a regional government are active. The purpose of the exercise is to show how a theory-based approach allows us to model the interactions of policies with convergent goals, and to indicate how the availability of data implied by reputation-based governance could be put to use. Figure 6.1 proposes a graphical representation of the model. There are two public administrations—a central government and a regional government—each one carrying out its own e-government policy. For each administration, the investment in ICT technologies and in projects that are enabled by the resulting ICT infrastructure together define that administration’s e-government policy (in the figure, the ovals labeled “e-gov central” and “e-gov regional”). Their joint action determines the overall e-government policy (the oval labeled “e-government”). The relation between distinct policies carried out by different institutions is typical of multilevel governance, and it is very often a source of confusion when it comes to attributing responsibilities for successes or failures. Suppose that a failure is observed. If the policies of the two administrations were substitutes, in the sense that one well-meaning administration would be enough to produce good results even when the other administration is inactive or ineffective, then we would have to conclude that both administrations share the responsibility for failure. However, if a success is observed, the possibility that one of the two administrations
108
Statistical Information and Analysis
Regional Administration
E-Gov Projects
Co-financing
Central Administration E-gov Projects
Savings Services Persons
Invest in ICT
Services Firms
Public Capital
Employees Private Sector
E-Gov Regional
Invest in ICT
E-Gov Central
E-Government
Technology Output Private Sector
Private Capital
A Structural Model of E-Government with square corners indicate administrations, and ovals indicate policies. Rectangles with rounded corners are inputs to the production function, and the main variable of interest is the regional output of the private sector. Arrows indicate causality relationships. The figure is from Picci (2006a). FIGURE 6.1
NOTE: Rectangles
took a free ride on the dedication of the other could not be dismissed. More often, however, policies carried out at different levels of government are complements, not substitutes. When a failure is observed, all that can be concluded is that at least one of the two administrations is responsible, but the possibility that one of the two acted effectively but alone cannot be dismissed. Following a success, on the other hand, the conclusion should be that both administrations share the merit. The model further posits that aggregate e-government policy may influence the prevailing technology used by the private sector positively, which in turn will affect output. Both administrations, however, as indicated by the left part of the figure, have at their disposal other possible productive uses for their scarce resources, such as building traditional
Statistical Information and Analysis
109
infrastructure or providing services to people or firms. They should then optimally allocate their resources to maximize their overall positive effect, which in the model is measured in terms of the aggregate output of the private sector. In Picci (2006a), this simple structural model is formalized as a set of equations to be solved in order to explore the impact of different scenarios of interest. In particular, the interplay of the relevant variables is analyzed for different degrees of substitutability between central and regional policy and for different profiles of the delays with which e-government policies influence productivity. What this model does not offer at present is a fully fledged analysis not just of scenarios, depending on several assumptions that cannot be backed by data, but of the effects of policies as they manifest themselves in a concrete context. However, at least in principle, many of the data that are needed but not available today would be so in the information-rich context of reputation-based governance. Such a structural modeling approach, then, would provide a route to accommodate newly available integrated statistics into a framework allowing for quantitative analysis of policy effects. One final note regards the social dimension of these modeling exercises. The use of these types of models would not necessarily be for specialists only but could also be integrated within the overall information system supporting reputation-based governance to make them accessible to all. They could be accessed through the Internet and would feed on real-time raw data. Anyone, then, could carry out analyses corresponding to different hypotheses and scenarios and compute forecasts of the outcomes of policies. Societal actors who are sufficiently sophisticated, moreover, could run their own models, simply by accessing the necessary raw data from the governance information infrastructure. The overall result would contribute to a democratization of policy analysis.8
7
Managing Policies Accountability, Rent-Seeking, and Corruption
Little by little, it is being understood that abiding to the laws also brings economic advantages. It is not so horrible to be honest. If people understand it, then, little by little, they come to believe it. You just have to get the habit. Giuseppe Vitale
The management of policies is the centerpiece of governance, and we now discuss how it would be under reputation-based governance. More precisely, we focus on two issues that allow us to adopt two distinct points of view on the matter. The first regards the implications of adopting a strong project-oriented approach that, as we argued, goes hand in hand with the organization of a policy landscape, where the set of actions by a public administration is partitioned into distinct projects and programs. The resulting policy landscape is scattered with a high number of policy objects. Work packages (in the language of the project management discipline) may be seen as the modules of their constituent projects, and each policy and program as the modules of the overall action of a public administration. In this sense, reputation-based governance naturally leads to a modularization of the policy landscape. This has important consequences that we discuss in the first part of this chapter. In the second part of the chapter, we consider the impact that a high degree of personal accountability—one of the implications of reputation-based governance—has on rent-seeking behavior and particularly on its most nefarious manifestation, corruption. The data-rich environment brought about by reputation-based governance would be particularly effective in limiting this phenomenon.
Managing Policies
111
The Modularization of the Policy Space
Within reputation-based governance, policies may be carried out according to the tenets of the project management discipline (as codified, for example, in Project Management Institute 2005). In fact, this would not be an inevitable outcome, in the sense that the adoption of reputationbased governance has no direct implication in terms of the procedural aspects of policy execution. It does imply, however, a broad adoption of what we may loosely call a project management approach, indicating a situation where the following holds: (a) the overall set of actions carried out by a public administration is partitioned into a number of projects, and each one of these is, in turn, divided into subcomponents, or work packages, and (b) there is a precise attribution of responsibilities for each project and for each work package within a project.1 The latter property is indeed a strong precondition of reputationbased governance, because personal accountability is needed to compute measures of reputation for the actors of governance. Accountability needs a clear assignment of responsibilities so that the assessments received by a policy can be used to compute reputations. The individual accountability of administrators is reflected in the accountability of public administrations, and to this end, policies should be attributed to those administrations, and departments within them, that are effectively responsible for their execution. When the responsibility for a policy is shared by more departments, or even by more administrations, individual work packages within a policy should be assigned to only one department within an administration, so as not to blur responsibilities. To be effective, the partition of the policy space should be communicated appropriately so the citizens can easily identify the particular policies they wish to observe. We may envisage a situation where each policy is assigned a unique code, which is given adequate publicity, for example by being reported in all official documentation pertaining to that policy. Just to provide an example of the possibilities that contemporary technologies offer, we may imagine that citizens could feed the barcode identifying a given program, taken from leaflets publicizing it, into a barcode reader in kiosks available in public spaces, or into their mobile phone cameras, and quickly access all the information available on that policy from the information system supporting reputation-based governance.
112
Managing Policies
If the policy landscape has to be composed by many mutually related but clearly distinguishable policy objects, how can the overall action of a public administration be broken down into these pieces? There are two aspects to this question. First, there is the “how.’ ” Policies should be defined in such a way as to be, as far as possible, self-contained. Reality may be more complex, and in the previous chapter we considered the example of two concurrent e-government policies carried out at different levels of governance. In accordance with some literature on modular designs, we call the identification of autonomous policies the decoupling problem, and in the next section we consider some of its implications. Given a decoupling strategy, presumably aimed at identifying as far as possible autonomous units of action, there is the related issue of the dimension, or granularity, of the policy objects and of their constituent parts. In fact, defining policies and their components presents two issues that are also common in the project management literature. The first one refers to the most appropriate size for policy work packages. Given the strong emphasis that we place on the principle of accountability, as a general rule work packages should reflect the attribution of responsibilities. For example, if the firms that will build two adjacent sections of road are chosen with two separate tenders, then each section should correspond to a different work package. On the other hand, work packages that are too small add an unnecessary element of complexity to the project. The second granularity issue regards the size of policies per se. Policies that are defined narrowly will tend to have few work packages, or possibly just one, while policies that are broadly defined will have a complex work-breakdown structure comprising many work packages. More complexity within the boundaries of the policy will result in the presence of fewer relations among policies, and vice versa: policies narrowly defined will be connected by many relations. The granularity of policies and the complexity of their workbreakdown structure is a well-known cause for concern in the project management discipline. Common sense should be used in this decision, and “the project team needs to seek a balance between too little and too much in the level of work breakdown structure planning detail” (Project Management Institute 2005, 114). Policies that are broad will better decouple the policy space, in the sense that they will be more independent from other policies—just as work packages that are broad will better
Managing Policies
113
decouple the space within a policy, in the sense that they will be more independent from other work packages. However, broader policies, or bigger work packages within them, will identify responsibilities with less precision and will be accompanied by a lower degree of accountability.2 The Implications of Policy Modularity
The partitioning and decoupling of the actions of public administrations into policies creates a landscape defined by the modularization of public governance. We use the term modularity in the same sense in which it is used in the literature on architecture designs. In a modular architecture, the overall design is divided into parts, also called modules or subsystems, interacting with one another according to rules that in principle are designed ex-ante. These rules should remain stable during the development of the object, so that the development of each module may proceed independently. As a result, each module can be developed and tested in parallel with the others, resulting in greater speed of project execution. The effectiveness of a general architectural philosophy like this is such that most complex artifacts today are produced according to modular designs, from software, as codified in the discipline of software engineering, to computers (Baldwin and Clark 2000), to airplane jet engines (Prencipe 1997). Overall, modular architecture design manages complexity quite effectively. We define complexity, as is somehow customary, as the presence of a multitude of relations among a great number of parts of a whole. The decoupling resulting from effective modularization is such that most relations happen within each module (akin, in fact, to a black box hiding its complexity inside) so that the chief engineer (or project manager) can focus on the inter-module relationships while delegating to collaborators the management of intra-module complexity. However, these advantages come at a cost. A modular design depends on an overall design that is decided beforehand, freezing part of the characteristics of a product at an initial stage. The possibility of making changes during the development phase is conditional on the overall predetermined architecture. If this architecture has fundamental flaws, such as ineffective decoupling of the overall task, then serious problems are bound to arise. An interesting anecdotal example is provided by the
114
Managing Policies
design of the Itanium integrated circuit by Intel (Ethiraj and Levinthal 2004). During the development process, it became clear that the modularization decisions had not been effective in slicing up the problem into smaller manageable parts, in the sense that changes in one module often had a knock-on effect on other modules. Thus, one of the key benefits of modularity—the possibility of parallel development and testing of the modules—was in fact lost. In a context where a team is trying to develop a highly innovative product (as was the case with Intel), a better alternative to modular design could be its opposite, an integral approach. Though it does not benefit from the pluses of modularization, this more traditional design philosophy guarantees a more holistic view of the whole product as it is developed and does not commit designers to an early decision that, in a context characterized by uncertainty, might not be appropriate. The cost incurred by adopting an integral approach instead of a modular one would obviously be a high degree of complexity of execution, in a context where a modification to one component may require modifications to many other parts of the project. This may have consequences that are hard to predict beforehand because of the potential systemic effects. This line of reasoning leads to an intuitive conclusion. A modular design is better suited to relatively stable situations, where architectural decisions taken ex-ante have a good chance of producing an adequate decoupling of the overall problem. On the other hand, in contexts characterized by considerable degrees of complexity—for example, in highly innovative projects—the advantage of an integral approach that allows for changes in architecture during the development phase could outweigh the benefits of modularization. Some recent studies, also using formal models, have provided arguments in favor of this intuition. In particular, modular architecture has been shown to be more efficient than an integral one in achieving an optimum solution within a problemsolving process. However, this optimum solution may be only local and trapped within an inferior design. A holistic point of view, on the other hand, seems better suited to not losing sight of the global optimum.3 The findings of the literature on modular architectures have implications for the decoupling and modularization of public policies within reputation-based governance. Just as in product architecture the interdependency among modules should be minimized to guarantee the
Managing Policies
115
stability of the interfaces through which they communicate with each other, the definition of work packages within a policy should be such that they make their execution as independent as possible. At a higher level, the division of an administration’s overall activity into separate policies also hinges on an adequate solution to the decoupling problem.4 The systematic modularization of the actions of a public administration arising from reputation-based governance would therefore be particularly desirable in relatively stable contexts. However, in unstable and uncertain situations, and when bold policies are sought, freezing a policy’s architecture at an early stage may reduce the freedom of governance actors excessively. This consideration should influence the choice of the degree of granularity of policies and of their work packages. Granularity should be coarser in uncertain situations to allow for a more holistic view of the issues to be addressed, even at the cost of a lesser degree of accountability of the governance actors. Rent-Seeking and Corruption
Rent-seeking behaviors often plague public governance by subtracting valuable resources from productive ends and by tilting private choices toward the pursuit of unproductive rents. Rent-seeking sometimes takes the form of corruption. The damage caused by corruption is widely documented, deriving in part from an immediate drain on public resources and, more importantly, from the distortions in both private and public choices that it provokes.5 Reliance on rules within public administrations is meant to put a check on corruption (Kassel 2008), and governments tend to overregulate administrative activities in a knee-jerk reaction to the fear of corruption, resulting in high administrative costs, distortions of various types, and strategies that aim to circumvent the rules. In the debate on corruption, there is at least some agreement on what measures could contribute to its cure. These involve increased accountability, to be obtained through a higher degree of transparency, regular monitoring of the activities that could lead to corruption, and a proper set of incentives, both in the form of punishments (such as a well-perceived chance of being caught following dishonest behaviors) and of rewards (for example, a civil servant’s expectation of promotion following a sustained good performance).
116
Managing Policies
There is a growing appreciation of the fact that corruption cannot be considered in isolation, since it is part of a complex nexus of relations within a given governance model. It follows that anticorruption policies that focus narrowly on the corruption issue will miss the complexity of the relations and are therefore likely to fail. More appropriate, then, are policies that aim to suitably reform prevailing governance systems. In this respect, consider the relations between corruption and the efficiency and effectiveness of a policy. A public official who receives an illegal sum of money for placing someone at the head of a queue for a service has an interest in keeping the line long enough. Hence, corruption may cause inefficiency and resistance to change. It is also true that inefficiency may cause corruption, which, at least in part, is a crime of opportunity. For example, the opportunity is there for an official to exploit when there is a long line of people waiting, and not otherwise. Moreover, lack of efficiency may lead to corruption, because the higher the complexity and length of a bureaucratic procedure, the more numerous the occasions to circumvent its rules that dictate honesty. There is also a possible relation between corruption and the ability of an administration to choose projects that respond well to public needs. A corrupt official or politician may choose a project because it allows him to extract unlawful rents more easily. Corrupt regimes dedicate more resources to military spending (Gupta et al. 2001), they consume more cement (Rose-Ackermann 2004), and, in extreme cases, they indulge in “white elephant” projects so lopsided that they constitute a credible form of patronage, because it is clear that they would be discontinued were someone else to get into power (J. Robinson and Torvik 2005). Corruption may cause ineffectiveness, but it is also true that the incapacity of a public administration to choose its projects well may cause corruption. For example, if a project is unsound and is widely understood to be a waste of money no matter how well it is executed, then moral restraints may loosen up and people in a position to embezzle funds may do so without being assailed by their consciences. Corruption, efficiency, and effectiveness, then, are strictly linked phenomena. This understanding tells us that it does not make much sense to propose narrowly designed anticorruption policies. On the contrary, both analysis and policy recommendations ought to be forged at a higher level, that is, in terms of general governance. In this respect, reputation-
Managing Policies
117
based governance represents precisely the appropriate level of analysis, because it rests on the availability of a set of integrated and highly institutionalized statistics that allow a series of incentives aimed at curbing rentseeking activities. In the case of corruption in particular, incentives may be based on the availability of measures that somehow proxy the extent of corrupt activities. We discuss next the measures of corruption currently available and how the adoption of reputation-based governance would innovate in this respect. Assessing Corruption
There are three prevailing types of assessments of corruption: judicial assessments, perception- or experience-based measures, and objective measures. A judicial assessment of corruption is the outcome of official judicial statistics stating, for example, how many people have been convicted in a given time period for corruption-related crimes. Judicial data provide a very dubious indication of the level of corruption. A high number of convictions could signal the presence of a lot of corruption or the efficiency of the judiciary in cracking a relatively low quantity of crimes. Similarly, an increasing number of convictions could indicate that crime is on the rise or the imposition of a successful crackdown, leading to a lower level of corruption. The extent to which corruption crimes are successfully prosecuted by the judiciary depends on many factors, explaining why much caution must be exerted when using judicial data as a measure of the phenomenon. Regardless of how effective the judiciary is in uncovering corruption cases, public opinion develops an informal assessment, a societal assessment of sort, of the severity and characteristics of the corruption phenomenon. Such an evaluation may follow from personal experience (particularly for petty corruption cases), hearsay, or the observation of the indirect effects of corruption. For example, seeing a police officer eating in a restaurant, in a country where it is known that his wage does not allow such luxuries, may be considered as a sign of petty corruption, while a bureaucrat driving a Ferrari suggests his involvement in corruption on a grander scale. A societal assessment of corruption is facilitated by the presence of a free press, which processes and gives visibility to information that individual citizens would otherwise obtain with difficulty,
118
Managing Policies
particularly regarding important corruption cases that are typically less visible to the general public. Corruption indexes are measures that summarize one or more aspects of corruption-related phenomena. These indexes are usually based on perceptions of the phenomena. This is the case with the Transparency International Corruption Perception Index, the result of the aggregation of several other indexes,6 and also the case with the Global Corruption Barometer, which Transparency International commissioned Gallup to produce and which focuses on petty corruption.7 Depending on the questions asked and on the identities of the intended recipients of the survey, the perceptions will be more closely related to one or another variety of corruption, where the main distinctions regard the size of the bribes (important vs. petty corruption), the identity of the persons committing a crime (politicians, bureaucrats, entrepreneurs, etc.), and the sector of economic activity involved. What is measured may be some incarnation of the corruption phenomenon, or may represent occurrences that are understood to be related to it, such as public integrity, or various measures of the quality of governance. Other indexes are based on descriptions of broad aspects of governance and, as such, are more tenuously linked with the corruption phenomenon (see Kaufmann et al. 1999). Some indexes, moreover, try to assess direct experiences of corruption, using a careful wording of questions so as to induce the respondents to answer honestly questions on whether they personally have been involved in corruption deals.8 These types of assessments may be linked by relations of reciprocal causation. Intense exposure of the public to corruption scandals may cause an artificial increase in perception-based indexes. In the 1995 Transparency International Corruption Perception Index, Italy’s ranking implied that it had more corruption than Mexico and Colombia and marginally less than Thailand, India, and the Philippines (Transparency International 1995). Such an unrealistic ranking was presumably explained in part by the spate of attention that serious corruption episodes had received in Italy in the early 1990s, following to the so-called Tangentopoli judicial inquiries.9 Moreover, the visibility in the media of corruption indexes could also influence the very same perceptions on which they are based. Judicial activity may also be influenced by perception of corruption, either of the informal type or as expressed by an index, to the extent that the judiciary has a say in deciding where to allocate its efforts.
Managing Policies
119
While most corruption indexes are based on perceptions, it is also possible to develop objective measures, that is, measures that derive from an observation of (a consequence) of corruption proper. Golden and Picci (2005) compute a corruption measure for the Italian regions based on the contrast between two alternative indicators of the public capital stock in different parts of the country. The first measure derives from the total amount of money spent over the years to endow Italy with infrastructure and is computed using the permanent inventory technique.10 The second is an appropriately aggregated physical inventory of the infrastructure that is actually present—length of roads and railroads, number and sizes of public buildings, and so on. The two measures are strikingly different in southern Italy where, over the years, a disproportionally high amount of financing appears to have been received in relation to the amount of work actually completed. This contrast forms the basis for the computation of a “corruption index.” Obviously, geographic differences in the effectiveness of public investments to generate finished infrastructure could be explained by factors other than corruption, such as disparities in the efficiency of the construction industry, in costs, or in the efficiency of local public administrations doing their part to ensure that infrastructure is built—all issues that are explicitly considered in Golden and Picci (2005).11 With reference to the taxonomy of statistics that we discussed in Chapter 6, all corruption indexes currently available are of the ad hoc type, and their degree of institutionalization varies. The Corruption Perception Index, computed under the sponsorship of Transparency International since 1995, presents a rather high degree of institutionalization. Indexes that depend on the effort of single researchers, such as the one in Golden and Picci (2005), have a one-shot character and are not institutionalized. While objective integrated statistics on corruption are not currently available, reputation-based governance would give rise to objective and integrated corruption-related measures, which would be routinely computed as views of the digitalized data contained in the information system supporting governance. In particular, within reputation-based governance objective measures such as the ones proposed in Golden and Picci (2005) could be routinely produced by the publisher unit as simple views of the available data.
120
Managing Policies
Concluding Comments
We started this chapter with a curious quotation from Giuseppe Vitale, at that time the CEO of Sacal, the corporation managing the southern Italian airport of Lamezia Terme in the Calabria region, as delivered to a journalist of the Italian newspaper La Repubblica (Caporale 2007). Giuseppe Vitale is a former judge, and Lamezia Terme is in an area where the ‘Ndrangheta, the local incarnation of the Mafia, is thriving. According to our source, the former judge obtained very good results as the airport administrator, underlined by a 16 percent increase in air traffic in his first year of tenure, compared to an 11 percent decrease the year before. This success followed an intelligent strategy that, on the one hand, improved the quality of the service provided and, on the other, succeeded in attracting low-cost operators such as Ryanair. In November 2006, after he had been in the job just one year, the majority of the Sacal board members asked for his resignation because they did not like his “style.” Later, a judge decided that the decision to fire him was not legitimate and reinstated him in the job. About one-third of Sacal is privately owned, and the rest is owned by different local public administrations. The appearance of such a positive article in a newspaper such as La Repubblica, which has the secondhighest circulation in Italy, probably helped Giuseppe Vitale in fending off further attacks.12 Not all good administrators in dire political straits have the good fortune to be contacted by the national press. Reputationbased governance would help in this respect. As we discussed in the previous chapter, the abundance of well-organized information pertaining to policies, and its accessibility to all, would allow many societal actors to participate creatively in making sense of it. As a result, the ecosystem supporting democracy would change. There is a debate worldwide on what can be done to save the great tradition of professional journalism, which is currently suffering from the double pinch of decreasing readership and diminishing advertising revenues. What arguably is suffering more is investigative journalism, which is the most expensive type of journalism and also the most useful to democracy.13 Traditional media plays, and will continue to play, an essential role in this respect, precisely for the reason that investigative journalism is expensive, and only a solid organization can sustain it for any length of time. However, a change in
Managing Policies
121
the cost structure of the information-gathering phase of news production would probably have an impact on the overall system. Part of the journalistic work would change its nature, since it could use the newly available policy-related data creatively. Investigative journalism could be directed toward those public administrations or officials whose activities are relatively more expensive, possibly indicating the presence of corruption, or that systematically receive low assessments by the public. Besides helping the traditional media to carry out investigative journalism, these tools—as they are available to all—could support investigative activities by other societal actors, such as whistle-blower organizations and even prominent bloggers. In conclusion, the reputation-based governance information infrastructure could help the news-generating ecosystem to be richer and more varied than it is today. A better informed citizenry would be more likely to act when faced with poorly executed policies and to support good administrators, who could boast good reputation ratings when conducting their political battles for survival in adverse environments. A higher degree of transparency and stronger career-concern incentives would make it correspondingly easier for administrators to resist the call of illicit money and would help to convince them that, after all, as Mr. Vitale put it, “It is not so horrible to be honest”—once the right incentives are in place. Reputation-based governance, with its emphasis on data collection and the organization of automatic data-processing procedures, should be seen within an ongoing process that is gradually changing our appreciation of the quantitative aspects of public administrations and, more generally, of governance. Good examples of this issue can be found in the study of corruption. In the 1990s, as we discussed, several organizations made available perception-based corruption and governance indexes. Golden and Picci (2005) provide a measure of corruption that is based not on perceptions but on hard data. The move from measuring perceptions to constructing measures that are based on objective data deserves encouragement. However, such a process is difficult to realize, because hard data are scanty and very time-consuming to organize. The methodology proposed in Golden and Picci (2005) could certainly be implemented in countries other than Italy, and has been. Such an endeavour is intrinsically expensive, and it could hardly be sustained on a regular basis. The availability of an information system of the type
122
Managing Policies
proposed here, however, would allow for the inexpensive production of integrated statistics, which would help to address the more general lack of suitable quantitative information that besets current analyses of public policies. These considerations allow us to view the proposed model of governance in a new light. The stated purpose is to provide a better model of governance. However, it is also about the creation of a conceptual and technological model for the systematic organization and collection of quantitative information on the working of a governance system. The two issues, in fact, should be seen as two sides of the same coin.
8
Applications of Reputation-Based Governance
It’s time to turn off the pork barrel spigot and deliver for our children’s future. Bill Clinton The defense budget is more than a piggy bank for people who want to get busy beating swords into pork. George H. W. Bush We need earmark reform, and when I’m president, I will go line by line to make sure that we are not spending money unwisely. Barack Obama
Reputation-based government depends on there being a policy landscape that is defined by the union of many policies. As we discussed, the parcelization of the policy space into a series of well-ordered policy building blocks is an abstraction at least in part, because the imposed conceptual uniformity hides a vast variety of actions that are carried out by governments. We would like to know more about the domains in which reputation-based governance is to be applied. Would reputation-based governance be meaningfully applicable to all policy domains? Probably not, even if the most obvious transparency-enhancing elements of the overall information system—say, the policy unit and the publisher unit—could indeed be introduced across the policy space. To discuss this theme, instead of supplying a tedious list of policy domains with a series of pertinent considerations related to our question for each one, I adopt a more informal approach by discussing a few distinct ways of applying forms of reputationbased governance. These examples will be relevant enough to convince us of the wide applicability of the governance model here proposed, but also different enough to suggest that practical applications would require some flexible thinking on how to adapt them to a particular policy domain.
124
Applications
Public officials are the most obvious public governance actors, but they are not the only ones. In particular, private firms assist public administrations in carrying out their tasks. While the participation by private firms in public governance is certainly not new,1 over time administrations have come to rely more and more on the private sector to carry out their tasks. There are two main motives for this progressive change. First, the complexity and technological content of public policies has grown over time, making it increasingly necessary for public administrations to tap into professional and technological resources from the private sector. Second, over the last few decades, public administrations have cut back to their core business and gotten rid of many in-house units of production. Under New Public Management–styled reform, outsourcing has become fashionable and, sometimes, faddish. As a result, private firms are today essential public governance actors, so that the selection of contractors from the firms’ space—again referring to Figure 4.1—is an issue that must be dealt with. Just as we needed a theory for bureaucrats’ incentives, we have now to discuss the quality of the work that private firms do as collaborators with public administrations. The textbook “objective function” of a firm is the maximization of profits and, unlike the often fuzzy tasks of public administrations, the outcome is easy to quantify and to assess. However, contractors’ incentives are often at odds with public administrations’ objectives. A profit maximizer has an obvious interest in minimizing the cost of what it delivers, typically at the expense of quality. Sometimes, as in the case of a corrupt relationship, a firm’s goals may coincide with those of a public official, but both are contrary to the public good. In this chapter, we describe a concrete example of reputationbased governance of public procurement of public works. The purpose of public works is to add to the stock of infrastructure available to an economy, or to provide for its maintenance.2 The public procurement case will lead us to discuss the combined use of traditional performance measures with user-generated assessments of reputation-based governance, a possibility that would help solve some of the known shortcomings of “performance management.” The second policy domain that we discuss is the provision of professional services, such as those provided by doctors working in public health systems. We will first refer to known examples that are based on
Applications
125
“scorecards,” that is, public disclosure of measures of service quality. Typically, they convey objective performance measures (for example, mortality rates following some types of medical treatment) or information that is provided by official inspections (of restaurant hygiene, for example). They have been extensively studied, particularly in their application to health care, thanks to the availability of suitable data. Then we will analyze a form of professional services market that would be based on an assessment of expected outcomes provided by the service suppliers themselves. Possibly the most important message to emerge from these very different cases speaks of the need for a flexible approach when applying forms of reputation-based governance. Public Procurement and Incomplete Contracts
Some anecdotal evidence on public works provides the starting point for our discussion. Between 1995 and 2000, 43.3 percent of the construction projects carried out by the private sector in Northern California were awarded after private negotiations, and the remainder after some form of bidding. In the same period, public administrations in Northern California used private negotiations in only 1 percent of the cases. Even when the private sector used bidding, a remarkable difference occurred: public administrations almost always chose open bidding, while the private sector more often opted to invite selected participants to bid.3 Such a striking difference is easily explained. While the private sector is free to choose the selection system, public procurement is typically constrained by a panoply of rules, the main purpose of which is to avoid abuse by public officials or, more to the point, corruption.4 The public sector, with few exceptions, is forced to choose (open) bidding. In the United States, for example, the Federal Acquisition Rules strongly limit the use of awarding methods not characterized by “full and open competition” (U.S. Government 2005). Full and open competition has much to commend it, because it places the buyer in the best position to exploit the competitive forces of the market in order to obtain the best deal. Limiting competition, by using invited bidding, where the buyer identifies the participating firms, or private negotiations, would seem to be illogical, because it excludes from the selection process firms that are potentially more efficient than the chosen ones.
126
Applications
However, the construction data example here provided, by showing that the private sector only rarely uses auctions to solve its procurement needs, suggests that open bidding has its downsides. The benefits of competition may be less important when the object of procurement is so complex that it is impossible to fully specify within a contract all the possible contingencies. In these cases, contractual flexibility allows both the buyer and the seller to better deal with unforeseen occurrences. Moreover, and most importantly, when the good to be delivered is not a commodity, the buyer benefits from the seller’s good reputation. It signals his ability to deliver high-quality goods and commits him to doing so. Current legislation on public procurement, though the motives behind it are commendable, constrains bureaucrats and distorts their choices of procurement methods. This distortion has a series of negative consequences, the seriousness of which is directly related to the degree of complexity of the procurement needs. First, contractors, if faced with a completely predetermined contract for work where the outcome is uncertain, may ask a higher price as a way of insuring themselves against the risk of cost overruns. Second, the products delivered may be worse. A fixed contract may force the contracting firm to take actions that could have been optimal ex-ante, before production started but that, ideally, would have been revised once the production process revealed information previously unknown. True, changes within an open bid setup are possible, as with the change orders in the U.S. construction industry. However, changes require a renegotiation of the original stipulations, are costly to obtain, and often are the source of acrimony between the parties. Moreover, in most private negotiations, reputation considerations play a prominent role. The greater the difference is between the object to be procured and the ideal-typical commodity with well-defined characteristics, the more important these considerations are. In open bidding, the reputation of the participating firms matters at best indirectly (for example, through their ability to find the necessary financial guarantees, whenever these are a precondition for participating in an open bid). Excluding bidders from a selection procedure because they have a bad performance record is, however, infrequent, even when possible. Such a lack of concern with respect to reputation considerations has a series of negative effects, as we already discussed in the introductory chapter. It pro-
Applications
127
vides the firm with weaker incentives to behave virtuously, because opportunistic behavior is less likely to be punished by the administration. Moreover, it weakens market discipline, because contracts may be awarded to firms that produce low-quality products simply because they have successfully specialized in the writing of impeccable tender proposals. On the administration’s side, besides provoking an overregulation of procurement, it puts an emphasis on the drawing up of overly detailed “technical specifications”—see the “cookie specs” example offered in Kelman (2002), referring to how the U.S. military had gone to great lengths to specify the cookies that it deemed fit for its troops. The emphasis on the minute description of the good or services to be procured has organizational implications for the administration, as it may dedicate resources to the writing of specs at the expense of those needed for managing contracts following their adjudication. In the 1990s, the United States embarked on a sweeping reform of federal public procurement, as part of then vice president Al Gore’s “reinventing government” initiative. The reform introduced reputation considerations into the public procurement process and aimed, among other things, to achieve more-effective outcomes by simplifying procurement regulations and increasing the responsibility of administrations and administrators. To this experience we now turn. Contractor Performance in the United States
The most relevant changes occurred under the guidance of Steven Kelman, professor at the Kennedy School of Government, who served as the administrator of the Office of Federal Procurement Policy (OFPP) from 1993 to 1997. This reform mostly affected what Kelman refers to as “the first two legs of contracting management,” namely, business strategy management and source selection, and “largely ignored the ‘third leg,’ ” or contract administration (Kelman 2002). While the collection and consideration of past-performance information had been around for a long time in the United States, the use of this information was strongly increased by a series of decisions started by OFPP Policy Letter No. 92-5 (December 30, 1992). This letter required all agencies to prepare past-performance evaluations for new contracts and include past-performance information as an evaluation factor in
128
Applications
awarding contracts. The whole process of reform was marked by two pieces of legislation, the Federal Acquisition Streamlining Act (FASA) in 1994 and the Federal Acquisition Reform Act in 1995. In particular, the changes that occurred during the 1990s brought about a situation where reputation considerations entered the computation of the relative merits of an offer, and now contribute to the final decision, together with quality and price, for all contracts exceeding $100,000. FASA, dated October 13, 1994, codified the requirements to consider past performance in making awards. Federal Acquisition Circular 90-26 (dated April 4, 1995) implemented OFPP Policy Letter No. 92-5 and FASA requirements into the Federal Acquisition Rules (Office for Government-wide Policy 1997). Currently, the Federal Acquisition Rules deal with reputation issues in Part 9.104, Part 15.608, and Part 42.1501. In particular, the 1998 rewrite of Part 15, dealing with source selection procedures for large buys, requires past performance to be one of two mandatory evaluation factors, with cost/price the other, for all competitively negotiated acquisitions exceeding established thresholds.5 The extent and quality of a bidder’s past performance is assessed by the source selection authority, using reviews of past projects carried out by a given firm, eventually together with other indicators of past performance that may also originate from the private sector. Of paramount importance for the formation of firms’ reputations are the opinions expressed by public administrators on the quality of work carried out by their contractors. These must be kept on file for three years. Considering the legal implications that a negative assessment may have, the procedure that dictates how a public administration will judge a contractor is of key relevance. The assessment has four separate dimensions: •
Quality of performance—as defined in the contract standards.
•
Cost performance—how close does this come to estimates?
•
Schedule performance—timeliness of completion by interim and final milestones.
•
Business relations—history of professional behavior and overall businesslike concern for the interests of the customer, including timely completion of all the administrative requirements and customer satisfaction. (Office of Federal Procurement Policy 2000).
Applications
129
Evaluations of these four dimensions are expressed in scores from 0 to 5. Each competing firm has ample room to voice its disagreement with scores assigned by the administration within a codified process. The relevant official documentation places much emphasis on the firm’s right to know beforehand how its performance is to be evaluated and on continuous communication between the administration and the firm. The shortening of the traditional arm’s-length distance between the administration and the contractor is, in fact, one of the goals of the reform, which aimed to transform a relation that was often acrimonious, into one among “partners.” It was hoped to achieve this result partly by emphasizing the responsibilities of the contracting official and correspondingly by allowing more freedom of movement, as is evident, for example, from the concomitant introduction of credit cards for all purchases below $2,500 (Kelman 2002). The Federal Torts Claims Act plays a relevant role in ensuring that evaluations are not a mere formality. It protects the evaluating official “from personal liability for common law torts. In those instances, if an agency official were sued, upon certification by the Attorney General, the official would be dismissed from the lawsuit and the United States would be substituted as the defendant” (Office of Federal Procurement Policy 2000). Another important characteristic of the evaluations is that they are not made public, being accessible only to those public officials who, for the purposes of source selection, need to assess a firm’s reputation. The relatively long experience with the system shows that firms take their reputation scores seriously, as they know that a bad reputation would damage their future prospects of being awarded new contracts. This constitutes basic evidence that the system works, in the sense that reputation considerations represent relevant incentives,6 even if past experience has also raised concerns about the possibility of strategic manipulation of the system.7 We now consider how reputation considerations could be applied to the procurement problem under the framework of reputation-based governance. Reputation-Based Governance of Public Works
Behind the proposed governance model is a taxonomy of possible projects. For example, and limiting our attention to the construction of
130
Applications
new pieces of infrastructure, public works can be roads, bridges, buildings, and the like. Within each type of good are further subclasses: buildings may be schools, police headquarters, and so forth. Upon approval, the public administration inputs each project into a database that can be accessed (via a Web application) by public administrators, firms, and the public. Preliminary information about a project would include its general description according to the codified taxonomy (type of good, location, expected cost, etc.), technical drawings, and a set of data pertaining to any documentation that may accompany the early stages of the project (such as an environmental-impact and cost-benefit analysis). As the project evolves, more information is provided, eventually to include details on costs and information on the final outcome, such as pictures of the completed works. Each project is associated with the administration and with the contracting firms responsible for its execution. In particular, the responsibilities of the contracting firms are recorded, and the information system keeps track of all the projects that each administration and each firm have done. Summary views of the past records of individual administrations and firms should be easily accessible. To summarize, reputation-based governance of public works rests on the availability of an information system that allows a careful definition of the types of goods and of their characteristics and that, through a set of suitable procedures, permits the storing and the publication of data on projects, administrations, and firms. Project outcomes are assessed by a set of relevant actors, who are allowed to express their opinions on given characteristics of the finished works. Public administrators in charge of a project express their opinion on the quality of the work carried out by the contracting firms. Firms assess how the public administration managed a given project, and the public judge public works that affect them. For example, a school is assessed by the local community for which it was built, by the personnel working there, and possibly by its students. A local road is judged by the people who live nearby, while a freeway is assessed by residents of a larger area. Such ratings are highly structured and refer to a small set of well-defined characteristics. For example, the public could judge aesthetic qualities, usefulness, and accessibility of public works.8 This digitalized information allows the provision of a series of automatically generated statistics of the types that we discussed in Chapter 6.
Applications
131
The statistics produced in this fashion include a comprehensive set of summary information: the general characteristics of public works, the unit costs of projects of the same type (for example, of a mile of road of a given category), and their completion time. Moreover, they express the overall assessment by the public. Also, the availability of information on many projects allows the computation of rankings of individual projects on several dimensions. Summary statistics could also be available for various subsets of the data, providing views by geographic unit, type of administration, and contracting firm. The assessments on the quality of completed public works would feed back to administrations, administrators, and firms. The reputation of all these actors is a function of the assessments received by their past projects. The system supplies information at two levels. First of all, it provides highly usable and easy-to-read summaries, using only very simple descriptive statistical concepts. A further level includes more detailed information and could use more sophisticated tools of analysis. The interested person, depending on needs and time available, could choose a quick tour to examine a single project or opt for a more indepth analysis. Reputation-based governance of public works would establish a number of incentives and disincentives. The general public could access various rankings of completed public works by the administrations that carried them out, the administrators involved, and the contracting firms. This opportunity alone puts a premium on honest and efficient behavior, inasmuch as such information influences electoral choices. Also, public administrators would be pressured to perform well in order not to be looked down on by their peers. To the extent that reputation matters in determining the outcome of public procurement, firms, too, are obviously interested in having good ratings. Moreover, if a firm is active in both the public and private sectors, a good reputation in the former would be useful in the latter. Such effects could then be made more cogent by deliberate publicity activities, such as publishing in newspapers the “ten best (worst) projects” of the year. The media would also play a role in providing incentives, since journalists could access a very powerful tool to obtain information on public works and expose both worst and best practices. Most importantly, reputation considerations could be employed within source-selection procedures. The experience accumulated in the United States would provide useful guidance in this respect. In Chapter 5,
132
Applications
we discussed the concept of the reputational budget, suggesting a marketoriented route that would allow the consideration of reputation in public procurement, without the need to explicitly include reputation among the criteria for source selection. In general, the adoption of reputationbased governance is accompanied by the presence of strong incentives for the actors of governance. This, in turn, provides them with more responsibility and gives bureaucrats room to maneuver. It makes an allocation mechanism based on off-the-shelf purchases more attractive than one based on competitive bidding, an evolution of procurement systems already hinted at in Rose-Ackerman (2004). There are obvious similarities between the U.S. experience and the present proposed model of reputation-based governance of public works. Both provide an institutional framework within which the reputation of the contracting firms can be built up and used as part of the sourceselection process. Also, the U.S. experience makes ample use of the Internet. Several administrations have developed their Internet-based information systems to record firms’ assessments, and the most popular of them, the Contractor Performance System run by the National Institutes of Health (http://cps.od.nih.gov), has been adopted by several administrations. This system and the ones that are run by the National Aeronautics and Space Administration and the Department of Defense have teamed up within the Past Performance Information Retrieval System (PPIRS) program to exchange reputation information. The resulting integrated system has been available since 2002. An emerging integrated infrastructure of this kind represents a portion of the information system that would serve the proposed model of reputation-based governance. However, the differences between the U.S. experience and fully fledged reputation-based governance of public works are as important as their similarities. First, in the United States only the opinions of the administration with respect to the contractor’s job are recorded. Firms are not allowed to produce their assessments on the administrations that they interact with, and, most importantly, the people who are affected by the projects are not allowed to voice their opinions.9 Also, the opinions expressed by the administration on the contracting firms are kept private. In this sense, public procurement in the United States, after the reform of the 1990s, represents an interesting experiment that only hints at some of the characteristics of a fully developed reputation-based governance model.
Applications
133
Reputation and Contract Administration
Kelman (2002) underlined the fact that the reform of federal procurement in the United States mainly addressed the issues of business strategy management and source selection and generally ignored the “third leg,” or contract administration, that “was largely a stepchild of procurement reform” in the United States. We consider this issue here in light of the changes that a fully fledged model of reputation-based governance could bring about. In also laying down a route for reform of contract administration, Kelman (2002) underlines that “contract management is mainly about management. The vast majority of what a good contracting manager needs to be good at are the same things any good manager needs to be good at.” Good managers, however, are in relatively short supply, and there is an issue of allocation of human resources among tasks where some do things and others check other people’s work. The former jobs are more rewarding within public administrations, and talented managers would rather be doers than checkers. Streamlining of procedures in the United States was also meant to contribute to making procurement jobs “more employee-friendly” within a “larger effort to make government a more attractive place to work.” This general consideration takes on particular force within public administrations, which are often at a disadvantage in hiring talented people. Kelman (2002) goes on to discuss how such goals could be achieved, and identifies “three things that the government should do if contract administration leadership is to become a core organizational competency: (1) properly positioning and training for the job, (2) splitting off lower-level tasks from executive-type tasks, (3) making an investment in performance measurement as a discipline.” The second item has to do with an obviously useful organizational change, away from a situation where contracting officials are overloaded with many administrative chores, such as taking minutes of meetings or updating spreadsheets. If such tasks were carried out by lowerlevel or junior staff, the contracting official could focus on more important issues, his work would be more rewarding, and in due time, contract administrators’ positions would attract more talented people. Regarding the “proper positioning” for the job, Kelman notes that senior contract administration jobs “should be positioned as management jobs with exciting challenges and stimulation similar to that of senior
134
Applications
executive positions.” Stimulations would consist, for example, of career opportunities based on a fair evaluation of the contract administrator’s performance, in accordance with the career-concern model that we discussed in Chapter 2. Reputation-based governance would provide an innovative framework to achieve this important goal: each contract administrator would daily build his reputation based on the appreciation of the policies he helps to carry out. Measures of this reputation would be public and visible to all and could serve for deciding promotions and other rewards. Since the reputation of the contract administrator would depend on the outcomes of the policies he is associated with, they would partly depend on the contractor’s measured performance. Reputationbased governance would deliver the desired alignment of incentives by putting contractor and contracting administrator in the same boat, both of them with a high stake in the ultimate success of the policy that they are working for. Regarding performance measurement, Kelman argues that just as the Defense Department, following its needs in the management of a complex weapons program in the 1960s, contributed to the development of the project management discipline, so today there should be an analogous “initiative to develop the discipline of non-financial performance measurement” (Kelman 2002). We discussed how reputation-based governance would help to obtain performance measures and also to put them to use for the purpose of analyzing the outcomes of policies. The mention of the project management discipline is also relevant to a discussion, from yet another angle, of the role played here by information technologies. Project management can be carried out, and initially was carried out, without the help of computers and dedicated software. However, extensive use of project management techniques, for example within matrix organizations that are strongly project-oriented, depends on the availability of an adequate information infrastructure, which permits the necessary horizontal communication, sharing of documents, project management and control, and time sheeting by employees to allow project managers to monitor costs. The computer age, by facilitating the development and adoption of project management techniques, has allowed them to fully play their systemwide role within organizational contexts that have shifted toward more horizontal configurations. The adoption of reputation-based governance would constitute a further step within this process.
Applications
135
Open Issues
The consideration of a concrete application of reputation-based governance allows us to critically appraise some problematic aspects that would require a careful examination in the implementation of reputationbased governance, not just in public procurement but in other domains as well. I have already mentioned the possibility of collusion between public administrators and firms. Unfortunately, these are not the only actors who could have incentives in reaching agreements to the detriment of the common good. The public could also have an interest in colluding with both administrations and firms. Public works are localized goods, which are sometimes supplied as part of a patronage relation between politicians and their clients, or, more crudely, they are manifestations of “pork barrel.” Up to a point, people benefit from local public works regardless of their quality. For example, an expensive but poorly built school may be a failure from the point of view of society as a whole, but it may still be preferable, for both parents and homeowners who benefit from any increase in real estate valuations, to a good school located far away. Administrations, firms, and local constituencies could team up to support policies that provide them with a common benefit, regardless of its social cost. The possibility of collusion between inefficient and possibly dishonest administrations, firms, and local constituencies that, in exchange for political support to their patrons, cynically appreciate any “pork” they may get should be an obvious cause for concern. However, it would only be in the interests of ordinary people to collude in this way if they expected that an unwarranted positive assessment of today’s policy would procure a favorable allocation of resources in the future. Local elites would be able to capture the political process only if this expectation were in place: for example, it would enable them to effectively convince the local population to turn a blind eye to their corruption, in exchange for crumbs to be distributed in the future. This perverse incentive should and could be avoided by implementing a policy selection process (an issue that we will consider in the next two chapters) that does not link the quantity of locally allocated resources to the assessments that local projects have received in the past. If this perverse channel is not present, then the people who do not belong to local elites will not have an incentive to assess projects they dislike positively.
136
Applications
The argument that tomorrow’s localized allocations should not depend on today’s assessments is relevant for public works, which cannot be taken away after they have been completed, and for projects in general, but not for those programs whose continuation has to be periodically decided—for example, via a budget law. Quite simply, if it is understood that the implicit threat to discontinue a program would become more cogent should such program receive negative assessments, then any beneficiaries would have an incentive to publicly pretend that they are happy with it, even when they are not. Some measures could be adopted to address these concerns. On the one hand, efforts could be made to limit these types of occurrences, which would probably be a desirable goal in itself. A degree of certainty about the identification of the needs that warrant public intervention of some type would seem to be a desirable prerequisite for providing a stable service. True, such a prerequisite is often purely academic. For example, street-level bureaucracies are often disturbingly characterized by the tendency of legislators “to delegate down the more contentious specifics of social policy” (Brodkin 2006). Also, other characteristics of the proposed model of governance would in fact favor a monitoring of local policies, regardless of whether they are programs or projects, by both the center and other communities that may fear they will be disadvantaged in the allocation of resources. First, the information systems would routinely produce statistics on standardized costs. These would have to be computed and read with care, because the costs of comparable tracts of roads, for example, are influenced by, among other factors, the local cost of labor and the nature of the terrain. However, the more information gathered by the system, the greater the sophistication that can be achieved in computing standardized costs of given types of public goods. The availability of standardized costs could allow for smart targeting of auditing activities by a dedicated agency. The probability of an audit for an administration could be determined by a rule, which would depend on both the cost of the project, relative to what standardized costs would dictate, and its quality assessment. A “probabilistic auditing rule” of this kind would temper the incentives that local actors could have to team up to grab disproportionate resources for themselves. Other characteristics of reputation-based governance would work favorably in
Applications
137
this respect. The information system allows the computation of very detailed geographic statistics showing the effects of distributive decisions. In such a highly transparent context, it would be relatively difficult to sustain pork-barrel expenditures, as these are to some extent shrouded in secrecy—about both the outcomes, whenever detailed and timely geographic information on the allocation of resources is scarce, and the political process, given that resources are often distributed by parliamentary committees, where logrolling practices favor the creation of vast majorities.10 Last, local projects could be audited from the outside. Besides formal auditing by a central authority, possibly following a probabilistic auditing rule, the public could also be involved in the assessment of public works located in different communities. If a public work is not highly localized, as in the case of an important freeway, such a possibility would automatically be present. To sum up, a combination of different types of measures could be used to assess policies. Considering this possibility within the wider debate on performance management will permit us to observe the issue from yet another angle. Combining Indicators
Evaluating public agency performance has typically been based on objective measurements, and clients’ perceptions of agency performance have not figured nearly as highly in reform blueprints. On the other hand, we are very much concerned with the assessment of policy outcomes by the public and on how these may feed into measures of reputation of governance actors. In fact, as we discussed, the information system supporting our governance model would also routinely produce a vast amount of objective information on policies, allowing the computation of a host of objective output measures of the type that form the basis for what we usually call “performance management.” Objective and subjective measures could, and should, be used in conjunction, as a way to correct the shortcomings that each of them may have if used in isolation. The pitfalls of objective performance measures are known. Boyne et al. (2005) note that “the implications of maximizing performance on a partially valid indicator raises the spectre of sub-optimization (Blau 1956) or even systematic cheating.” The latter was discussed in Bohte
138
Applications
and Meier (2000), who further add that “when agency performance is evaluated in terms of numerical outputs, bureaucrats have an incentive to maximize outputs, regardless of whether maximizing outputs is the preferred strategy for achieving desired social outcomes (a form of goal displacement).” Kelly and Swindell (2002) remind us that “as early as 1973, Elinor Ostrom pointed out that reliance on a single indicator of service quality could restructure internal incentives and lead to ‘pathologies.’ She recommended developing multiple indicators of service quality (including citizen perceptions) and studying the relationships among indicators.” The pathologies are, primarily, of the “goal displacement” type.11 Brodkin (2008), considering the management of street-level bureaucracies, affirms that “common accountability measures are too crude to capture the complex realities of informal practice,” and that “quantitative measures of inputs and outputs tend to reconstruct policy in its own terms rather than illuminate the qualitative content of what bureaucrats do and how they ‘make’ policy on their own terms. Even advanced efforts to improve accountability by applying New Public Management (NPM) techniques of performance measurement and ‘pay for performance’ contracting, at times, may do more to provide the appearance of accountability than accountability-in-fact.”12 I advocate the combined use of objective performance measures with subjective client perceptions. This possibility, also suggested by others, has so far been applied in traditional settings, where user-generated assessments are obtained through expensive surveys, focus groups, etc., and not at zero (marginal) cost, as they would be were there an appropriate information system and a clearly legible policy space. For example, Shingler et al. (2008) note that “agency managers must often negotiate a trade-off in efficiencies when determining which factors are important in the evaluation of their agency.” This would be less true in a context where subjective assessments on policy outcomes are routinely produced, summarized, and made available to all. Shingler et al. (2008) also note that “obtaining subjective client perceptions has considerable value beyond just providing data for performance evaluation” because it helps “give visible evidence to the public that agencies are genuinely interested in what their clients think and will use their responses in performance evaluation and policy making. This provides tangible evidence that government agencies can be responsive to
Applications
139
the public they serve.” However, the most important reason for suggesting a combination of objective performance measures and subjective assessments by the public—“hard” and “soft” measures, as Brudney and England (1982) define them—would be precisely to limit the possibility that administrators game the system by displacing the goals of the policy they work for with the maximization of the performance measures that their supervisors look at. In particular, as we discussed in the public procurement example, the presence of contrasts between objective performance indicators and user-generated assessments would be informative. The appearance of a poor assessment for a policy that otherwise presents impeccable performance measures, or vice versa, would call for an auditing activity, and would draw the attention of the ecosystem formed by the media and the varied array of whistle-blowing organizations that we described in the previous chapter. The Provision of Professional Services
The use of report cards on services of various types has become fairly widespread, particularly in the United States. Report cards are public disclosures of the quality of a certain service that aims to influence consumer choice. When this happens, we expect the usual positive effects of reputational incentives to be at work: low-quality providers would see the demand for their services decrease, so that eventually they would go out of business. Moreover, all providers in the market would have an incentive to exert effort in what they do. Report cards have been used mostly in the context of services that are at the intersection between the private and public sphere or that, while being clearly of a private nature, are publicly regulated. An example of the latter type is the hygiene grade cards that, since 1998, Los Angeles County restaurants are obliged to show in their windows. Jin and Leslie (2003) show that the new ruling caused restaurant health inspection scores to increase, with a subsequent decrease in food-borne illness hospitalizations. Also, they demonstrate that the positive effect was partly the result of a reorientation of customer demand toward those restaurants with better hygiene grades, and partly of hygiene quality improvements by restaurants. Scorecards also have been used for a rather long time in the health-care field, giving rise, thanks to the availability of suitable data, to
140
Applications
a host of research studies estimating their effects. For example, in the State of New York, the first publicly disseminated information on mortality rates in hospitals for a wide range of conditions was released in 1984. Then in December 1990, “New York State published the first comprehensive hospital cardiovascular surgery report cards” (Dranove and Sfekas 2008). One problem with report cards on health care that use the mortality rate as a proxy for quality is how to take into consideration the fact that mortality for a given illness is obviously influenced by a series of factors linked to the risk condition of the patient. An indispensable “risk adjustment” is performed using the patient’s clinical data, typically using regression techniques, so as to factor out the characteristics of the patient. However, not getting this adjustment right could have serious consequences, such as an attempt by the hospital to game the system by refusing to accept the most serious cases. This perverse effect was detected by Dranove et al. (2003), who analyzed the use of cardiac surgery report cards in New York and Pennsylvania. The results of the use of report cards in medical care are mixed (see Dafny and Dranove 2008 and Dranove and Zhe Jin 2010), with more recent studies pointing to this system’s positive effects. For example, Dafny and Dranove (2008) study “the release of HMO [Health Management Organization] report cards in 1999 and 2000 to 40 million Medicare enrollees, on the subsequent health-plan choices of enrollees” to conclude that “report cards are an effective means of disseminating quality information,” producing “substantial swings in market shares among Medicare HMOs.” Chernew et al. (2008) estimate the impact and value of information “using data from a large employer, which started distributing health plan ratings to its employees in 1997” and find that “the release of information had a statistically significant effect on health plan choices.” Dranove and Zhe Jin (2010) summarize the literature by affirming that “empirical studies confirmed the theoretical arguments that quality disclosure has strength and pitfalls. On the positive side, there is fairly strong evidence from health care and finance that disclosure enables consumers to identify superior sellers” and that “disclosure motivates sellers to improve quality.” On the minus side, “sellers have attempted to game the system at the expense of consumers, especially if the measured quality does not cover all dimensions of quality or does not adjust for characteristics of consumers that can affect the rankings.”
Applications
141
Most of the quality measures reported by scorecards are either objective, although partial, data (such as mortality rates in health care) or information that is collected and made easily readable by some third party (such as restaurant hygiene that, in the Los Angeles County case, is measured by hygiene inspectors). We now discuss the possibility of doing without any such information by appropriately eliciting the opinions of the service providers themselves. We consider the provision of professional services, of which one or more outcomes can be objectively observed. A typical example would be the services of a doctor for treatments whose outcome may be specified clearly. Assume that the more people he treats, the more he earns. Then, if we ask the doctor about the ex-ante probability of the treatment’s success, he will have an interest in overestimating it so as to get more clients. Similarly, if we want to sell a house, a real estate agent has an interest in telling us that he will be able to sell it for a high price, as this will encourage us to hire him. To deal with the choice of providers for services of this type, Gruen (2002, 2008) proposes a type of tender that, thanks to appropriately engineered reputational incentives, would induce professionals selling their services to offer an honest estimate of the outcome of their actions. This mechanism would work as follows. First, the potential service provider would offer a forecast of the outcome of his action. For example, a doctor could predict a given patient’s chances of successful obstetric delivery, and a real estate agent could predict the selling price of a house. These forecasts would be logged into an information system, together with the observed ex-post outcomes (when the service was effectively chosen and performed). The comparison of the ex-ante prognoses with the ex-post results allows for the computation of an “optimism factor”: “for instance, if the service provider has on average been 10% more optimistic than his results would justify, the ‘optimism factor’ would be − 0%.” Then, once enough data are available, which compare ex-ante prognoses with ex-post results, any new prognoses by a given professional would be weighted according to his past accuracy in forecasting the outcome of his actions. So, for example, “if a real estate agent’s optimism factor was −10%, and his raw prognosis for selling your house was $400,000, the optimism factor would see the raw prognosis reduced by 10% in the moderated prognosis to $360,000 ($400,000 − 10% of $400,000)” (both citations from Gruen 2008).
142
Applications
A client shopping for a service of a particular type—say, a person who can choose among several public hospitals for a particular treatment—could then simply compare the prognoses of the outputs corrected for the “optimism factor.” If Doctor A predicts a 90 percent probability of success, and Doctor B a 95 percent probability, but Doctor A in the past was on average right in his prognoses, while Doctor B on average overestimated by a factor of 20 percent in his, then it would be wise to hire Doctor A. An arrangement like this would obviously encourage service providers to provide honest ex-ante forecasts of the outcomes of their actions. In turn, the system could set up a series of incentives to reward high-quality work, taking advantage of the fact that, as Gruen (2008) observes, “this system produces a lot of information that will aid in the discovery of good practice by both consumers (reacting to the highest quality tender ‘bids’ for work) or by health administrators seeking to allocate large numbers of clinical jobs to the highest quality providers.” Good practices, once they have been detected, may then be appropriately rewarded—through promotions or by means of monetary bonuses. The tendering mechanism proposed by Gruen is based on the past records of the potential providers, and in this sense it also represents a case of a reputation-based solution to an allocation problem. However, it evidently differs with respect to the reputation-based governance model that we have discussed so far. The evaluation of the quality of the policy (say, a medical service) is made in the former by the “policy maker” himself—the doctor or the real estate agent—and not, as in the latter, by the people who are affected by their actions. The incentive to provide biased prognoses is countered through the comparisons with ex-post outcomes: the type of tender that Gruen advocates represents a form of reputation-based governance, but with an interesting twist with respect to what we have discussed so far.13 Conclusions
We have discussed two quite different cases of reputation-based governance applications—first, public procurement of public works, and second, the provision of certain types of professional services. The two examples tackle a similar general problem—how to best choose a product or a service—but in quite different ways. Appreciating the differences
Applications
143
between the two cases showed that there are many ways to harness the power of the Internet and of reputational incentives to improve governance. Different reputation concepts may be adopted, having differing properties—remember our discussion on “beauty contest reputation”— and different ways can be established to obtain unbiased assessments of the quality of a policy outcome, as the tender model discussed above demonstrates. Also, different measures of policy outcomes could and should be used, and the subjective assessments that are routinely generated within reputation-based governance would be usefully combined with the objective performance measures that have received so much emphasis in New Public Management reforms. The model of reputation-based governance, with the characterization of the supporting information system that was discussed in Chapter 4, should be seen as a concrete proposal, but not as an inevitable solution. The discussion on Gruen’s tenders, which constitute a reputation-based governance different in spirit from what we have discussed so far, also helps to clarify this point. The contrast between the cases that we have discussed shows, quite simply, that the implementation of reputation-based solutions to problems of governance admits and encourages creativity.
9
Interdependence Between the Choice and Execution of Policies
I left Paris, and France itself, because I realized that the Eiffel Tower was bothering me way too much. Not only do you see it from everywhere, but you find it everywhere. It’s made of all known substances, it’s exposed to all eyes, it’s an unavoidable and torturing nightmare. Guy de Maupassant
As the end of this book approaches, we get closer to completing a description of the different parts of the governance process. We will now discuss the selection of the policies to be executed from all the possible alternatives. This issue is strictly connected to what we may call the “democratic characteristics” of a given model of governance, by which we mean the ways in which the will of the people is allowed to influence this selection. In this and the next chapter, I argue that reputation-based governance may indeed lead to a choice of policies that are better attuned to citizens’ preferences, and that it opens up opportunities for advanced forms of popular participation in government. We proceed in two steps. In this chapter, we discuss how the characteristics of governance may influence the choice of policies. We have seen that the ex-post checks on policy makers’ actions implied by reputationbased governance, which makes administrators and politicians accountable to an unprecedented degree, causes strong ex-ante incentives to carry out policies efficiently. However, the modification of ex-ante incentives may also lead to a different choice of policies, because politicians, who are typically concerned about their own reelection, are likely to choose policies that are better attuned to the people’s wishes. Determining how the characteristics of governance influence the type of policies that are chosen is not straightforward. We focus on a single important policy characteristic— boldness—and, with the help of a simple model, we determine under what
Choice and Execution of Policies
145
conditions the stronger incentives associated with reputation-based governance would lead to the choice of more or less ambitious policies. In the next chapter, we consider a more direct route toward greater democracy within reputation-based governance, namely, via the direct participation of citizens in the choice of policies. Initially, we will pretend that the selection of a policy can be reduced to the choice of one finalized policy blueprint from an infinite policy space containing all possible policy options. This is only an abstraction, which, however, is useful because the selection of policies is typically quite complex and highly structured. A closer approximation to reality would be to assume a multitiered process in which, at the beginning of the selection phase, the policy space is of a finite dimension and comprises broadly defined families of available options. Once a policy type has been selected, the administration progressively perfects its choice, considering several subvariants within a multistep process that eventually converges to a finalized plan. The formal authorization of the project is also a multistep process in which several administrations, and possibly legislatures, in turn green-light a proposal that at one level may need to be expressed only in very broad terms but eventually has to be specified in detail. The authorization is typically conditional on changes to the original proposal within a complex bargaining process that takes place between different levels of governance, different administrations, and even different offices within the same administration. To make matters even more complex, after formal authorization the policy is carried out by officials who will at least interpret their mandate, almost invariably bending the policy to suit their interests. In many cases, the policy-selection process does not end with a formal adoption of a detailed policy blueprint that authorizes the execution phase, because the selection and execution of a policy are, to some extent, meshed together. Sometimes policies are activated in a tentative fashion, with many subsequent adjustments in the course of their lives. Moreover, they may start hurriedly, before a firm consensus is secured, as a device to impose a commitment on recalcitrant actors. Initiating only partially financed public works may be an instrument to force the administration to eventually find new resources, so as not to leave the work unfinished. More generally, whatever is spent on the early stage of a policy typically constitutes, in the language of economists, a “sunk investment.”
146
Choice and Execution of Policies
This investment does not enter subsequent economic evaluations on whether to discontinue the policy, thus making its continuation cheaper and tilting the balance in favor of its completion. There are other reasons that policies, once they have been started, often build their own momentum that tends to encourage their continuation. If part of an administration is able to impose the beginning of a policy on the administration as a whole and is then able to publicize it well enough to form a public expectation of its future outcome, then the administration as a whole, and not just the policy’s immediate sponsors, may become committed to its completion in order not to lose face. To summarize, in today’s multilevel governance systems, where many administrations partake in the design and the authorization process of a policy, the selection process tends to be very complex and is often characterized by an uncertain attribution of responsibilities, resulting in conflicts among the interested parties and also in delays that may arise in all phases.1 Therefore, thinking of the policy space as the set of all possible policy blueprints, from which a particular one is eventually selected for execution, is only a useful simplification of a very complex reality. The Interdependence Between Policies and Governance
Strong ex-post checks on the outcome of policies, when they are expected, become ex-ante incentives that modify how the actors of governance act. For example, if the introduction of ex-post controls makes it more difficult for a politician to get away with corrupt behavior, he is quite likely to notice that before he decides what course of action to take, and consequently his valuation of his personal interests changes. If an ex-post incentive links the chance of promotion of a bureaucrat more tightly to his past performance, he may ex-ante decide to put more effort into the job than he would have done otherwise. However, the presence of strong ex-post incentives would also influence the selection of policies. While considering the case of corruption, we have mentioned how countries where there is more corruption spend their resources in a systematically different way than countries where administrators are more honest. This is so because in situations where corruption is widespread, policies are also chosen in order to provide the possibility to
Choice and Execution of Policies
147
be corrupt. If strong ex-post checks are introduced, effectively impeding corrupt behavior, the choice of policies would also be modified, because the possibilities that they offer for diverting public resources to private pockets would stop being one of the criteria for their choice. Ascertaining the consequences of a given set of ex-post incentives for the choice of policies is certainly not straightforward. In the last chapter, we discussed how the modularization of policies may be related to wider modifications in the structure of organizations adopting reputationbased governance. Here we look at a different aspect of the relations between the model of governance and the characteristics of policies, and we ask ourselves if the stronger incentives that characterize reputation-based governance would have an impact on the types of polices that are chosen, and in particular on their boldness, a characteristic on which we focus our attention. Some policies are very courageous and aim high. These are the types of policies that stand a chance of making the pages of history books, for better or worse. Other policies are just bland and, at times, plain boring, and they are usually forgotten. March and Olsen (1995, 194) argue that democratic processes tend to be associated with incremental policies, never to be overambitious, and to discourage major planned experiments. Similarly, Scott (1998, 89) notes that very ambitious plans of the type that he defines as cases of “high modernism” need “a weakened or prostrate civil society that lacks the capacity to resist [them].” Here we consider an analogous problem, only more narrowly defined, and ask ourselves how the strength of ex-post incentives influences the degree of courageousness of chosen policies. To provide some preliminary intuition on the issue, let us consider an amusing historical episode: the construction of the Eiffel Tower in Paris between 1887 and 1889 for the Exposition Universelle, which celebrated the hundredth anniversary of the French Revolution. What would eventually become one of the most famous landmarks in the world and a symbol of the positivistic age was at first received disdainfully, at least by many of the local intelligentsia. The writer Guy de Maupassant, for example, wittily declared that his preferred restaurant was the one on the Eiffel Tower, because it was the only place in Paris where he was sure he would not have to see it (cited in Barthes 1989). And at one point in his life, Maupassant even decided to leave the
148
Choice and Execution of Policies
country—also because of the tower, a carcasse metallique, as he defined it (Maupassant 1890). A noisy petition was signed by the likes of Maupassant, and countless trenchant criticisms were leveled against the tower as construction proceeded. However, Gustave Eiffel, who had been contracted to design it, was allowed to complete his project. We could argue that, had the stronger incentives of reputation-based governance been adopted, matters could have turned out differently. A collective negative assessment on the work would have tarnished the reputations of both the contractor and the public administrations that gave him the job. Possibly, had they anticipated this, the administrations would have preferred a less courageous blueprint, and Gustave Eiffel would have confined the use of his metallic structures to more conventional domains, such as bridges and railroad stations, where he was also an expert.2 The result could have been a bland project that today few would notice. More generally, by not leaving the architect and city planners alone, and by putting constant pressure on them to deliver artifacts and cities that look appealing right away, strong ex-post incentives could deliver conservative and bland public works and spaces. This intuitive and anecdotal conclusion rests on the idea that strong ex-post incentives lead to the ex-ante choice of more prudent policies. However, the analysis of a simple model shows that this is not necessarily true. Bold Policies and Ex-Post Incentives: A Simple Model
The model that we present uses some simple math. The reader may skip it and only consider an intuitive interpretation of the results. In particular, the main result that we obtain in this section indicates that the preferences of public officials with respect to risk will depend on the incentives they face. If public officials receive a sizable reward for success and only a modest punishment for failure, then they may prefer to execute risky but ambitious policies, and vice versa. This simple intuition applies to different contexts. For example, an investment manager who receives a large bonus if his client’s portfolio does well, and incurs no significant costs even if there are serious capital losses, may tend to choose assets that are characterized by a combination of high risk and high expected return, regardless of his client’s attitude to risk. However, this out-
Choice and Execution of Policies
149
come, while possible, is not inevitable, as our model will show. The topic we consider in the next section also represents a case where the use of some math helps to clarify an issue. There, too, I will provide an intuitive explanation of the results. To simplify our analysis, we pretend that there is an identity between the selector and the executor of a policy. Such an assumption runs counter to the idea that, in fact, these two roles are often separated in governance. In particular, it could be objected that administrators simply implement policies, the choice of which, in a democratic context, is the prerogative of legislative assemblies or of elected officials. However, this is only partially true, because administrators typically are handed down general indications of what they have to accomplish, and choosing how to is tantamount to the selection of a course of action. In what follows, we will simply indicate the person selecting and executing the policy as the “policy maker,” which simplifies a collective decision problem. Inasmuch as the conclusions that we will draw from the model apply to all the actors involved in the choice and the execution of policies, we may simply imagine that their collective choice will converge to the individual intentions. The model focuses on the relationship between the strength of the incentives faced by the selector of the policy and its ambitiousness. Policies may have two outcomes: they either succeed or fail. The expected outcome for society of a policy, X, is as follows: X = pis Si + (1 − pis ) Fi ≥ 0
where pis is the probability of success of a policy of type i, Si is the payoff to society in the case of success (net of the total costs of the policy), and Fi (a negative number) is the payoff to society in the case of failure. Fi is the negative of the total cost of the policy, and a failure is defined as the policy having no positive effects. Policies may be of two types: ambitious policies, for which i = A, and prudent ones, i = P. We will see shortly how they differ. The expected outcome for society of a policy is assumed to be always greater than zero; policies with a negative expected value would simply not be chosen from the policy space.
150
Choice and Execution of Policies
We assume that the existing mode of governance may differ in the extent to which administrators’ performances are rewarded. A lack of reward or punishment may have different explanations. First of all, it could be that the monitoring is defective, so the organization simply is unable to put in place strong rewards or punishments. Or it could be that, while the final output is monitored efficiently, it is the result of the work of several actors whose contributions are not efficiently monitored. Last, it may be that, while the administration is effective in assessing the relative merits of its members in determining the outcome of a policy, it does not want to establish strong incentives, possibly because it prefers to allocate rewards according to different principles, such as nepotism. Here we do not need to distinguish between these cases, and we can simply focus on the strength of incentives, regardless of its determinants. The policy maker receives an expected reward, E(R), that is commensurate with the payoffs of the policy: E( R ) = ρ S pis Si + ρ F (1 − pis ) Fi − efforti ,
0 ≤ ρS, ρF ≤ 1
where ρ S and ρ F are the proportion of the policy payoff accruing to him in case of, respectively, success and failure. We may interpret this personal reward as a career advancement following a job well done, or even as direct monetary incentives. Generally, ρ S and ρ F will be fairly close in value to zero, to reflect the typical situation in public administrations, where officials are far from being residual claimants of their initiatives. Note that distinguishing between these two parameters allows us to consider different incentive schemes. In particular, ρ F could be particularly small or even equal to zero, again to reflect typical working conditions in public administrations, where policy failures are rarely seriously punished. The effort variable is the cost that the policy maker has to put in to carry out a given policy. We assume that: efforti = 0
if i = P, and efforti >
0 if i = A.
The effort variable, in fact, represents the extra work that it takes to accomplish an ambitious rather than prudent policy, the underlying idea
Choice and Execution of Policies
151
being that ambitious policies are inherently more challenging. Ambitious policies require more effort than prudent ones, but they typically provide bigger payoffs—be it in positive numbers in cases of success, or negative ones in cases of failure: SP = SA · λ , FP = FA · γ ,
λ, γ ≤ 1
The payoffs of prudent policies are just a fraction of the payoff of the ambitious ones, and the multiplying parameters do not have to be the same. For example, a value of γ close to 1 and a smaller value for λ would express a situation where, in the case of failure, policies have similar costs, but in the case of success, the ambitious policy is much more rewarding. Finally, ambitious policies are typically less likely to succeed than prudent ones: pAS = pPS − α ,
0 ≤ α ≤ pPS
In the expression above, pPS , the probability of success of prudent policies represents a base-level probability of success. The parameter α is non-negative, so that we assume that ambitious policies are not more likely to succeed (and in general, less so) than prudent policies. The bigger α is, the greater is the risk associated with the ambitious policy, as compared with the prudent one. We are interested in understanding under what conditions, within the trade-off between risky but rewarding ambitious policies and not so risky but less rewarding prudent ones, the public official will choose the former instead of the latter. Assume that public officials are risk-neutral— that is, they take into consideration only the expected payoffs. They will choose an ambitious policy whenever its expected payoff exceeds that of the prudent one: If, and only if: E(R)⎪POL = A − E(R)⎪POL = P > 0, then choose the ambitious policy (Pol = A), otherwise choose the prudent policy (Pol = P).
152
Choice and Execution of Policies
Using the previous definition, we obtain: E(R)⎪POL=A − E(R)⎪POL=P > 0 if
Φ = ρ s ⋅ ( pPS − α ) ⋅ SA + ρ F ⋅ (1 − pPS + α ) ⋅ FA − effort A − ρ s ⋅ pPS ⋅ λ ⋅ SA − ρ F ⋅ (1 − pPS ) ⋅ γ ⋅ FA > 0
Rearranging, we obtain: Φ = −α ⋅ ( ρ S SA − ρ F FA ) − effort A + ρ S pPS SA (1 − λ ) + ρ F (1 − pPS ) FA (1 − γ ) > 0 EQUATION 9.1
or: ρ S pPS SA (1 − λ ) + ρ F (1 − pPS ) FA (1 − γ ) > α ⋅ ( ρ S SA − ρ F FA ) + effort A EQUATION 9.1′
If Equation 9.1′ is satisfied, then it is convenient to choose the ambitious policy. The left-hand term of Equation 9.1′ represents the expected benefit of the ambitious policy with respect to the prudent one. It is equal to the sum of the advantages accruing to the policy maker of the ambitious policy for both success and failure, weighted by their respective probabilities. The right-hand side represents the extra cost that has to be incurred in choosing the ambitious policy. The first term represents the expected loss that follows from the fact that the ambitious policy has a lower probability of success. The second one is the extra effort requirement for the ambitious policy. Consider first the special case λ = γ = 1. In this case, SP = SA and FP = FA, that is, the ambitious and the prudent policy are not distinguishable in terms of payoffs. Equation 9.1′ reduces to: −α · (ρ SSA − ρ FFA) − effortA < 0,
which is always smaller than zero, since α, effortA, ρ S, ρ F, and SA are positive, and FA is negative. In this case, there is no reason the ambitious
Choice and Execution of Policies
153
policy would be chosen, since it only has disadvantages with respect to the prudent alternative. Consider instead the case λ = γ = 0, which corresponds to a situation where the prudent policy is indeed very prudent, and its outcome is always equal to zero, so that, in fact, choosing the prudent policy corresponds to not choosing any policy. Then, it will be convenient to choose the ambitious policy whenever: PPS ⋅ ρ S SA + (1 − pPS ) ⋅ ρ F FA > α ⋅ ( ρ S SA − ρ F FA ) + effort A
The left-hand side is now simply equal to the expected return of the ambitious policy, since the expected return of the prudent one is equal to zero regardless of the outcome. The right-hand side expresses the two extra costs that accompany the choice of an ambitious policy. In this case, the ambitious policy is chosen whenever its expected value is positive. To gather further insight into the problem, we compute the first derivatives of Φ, the expression of Equation 9.1, with respect to some relevant variables. The interpretation of the results will be as follows. If the first derivative of Φ with respect to a certain variable or parameter is positive, the implication is that, as this parameter or variable increases, it becomes relatively more convenient to choose the ambitious policy rather than the prudent one, and vice versa if the first derivative is negative. We start with computing the first derivative of Φ with respect to λ and γ. We obtain: ∂Φ ∂Φ = − ρ S ⋅ pPS ⋅ SA < 0 and = − ρ F ⋅ (1 − pPS ) ⋅ FA > 0 ∂λ ∂γ
Increasing λ always makes it less convenient to choose the ambitious policy, because it implies that, in the case of success, the advantage of the ambitious policy with respect to the prudent one is smaller. The opposite happens for increases of γ, making the disadvantages of the ambitious policy with respect to the prudent one smaller in cases of failure. A more interesting case emerges when we consider the variations of Φ for variations of parameters representing the strength of incentives ρ F and ρ S. The following holds:
154
Choice and Execution of Policies
∂Φ = α ⋅ FA + (1 − pPS ) ⋅ FA (1 − γ ) < 0 ∂ρ F EQUATION 9.2
and: ∂Φ = −α ⋅ SA + pPS ⋅ SA (1 − λ ) > 0 ∂ρ S
if, and only if: pPS ⋅ (1 − λ ) > α EQUATION 9.3
Any increase in ρ F has the unambiguous effect of making the ambitious policy less advantageous. Equation 9.2 shows that the variation is equal to the sum of two effects, both of which are negative. First, an increase of ρ F causes an expected loss that is commensurate with the difference in the probability of failure between the two types of policies, α, times the loss following the failure of the ambitious policy, FA. Secondly, an increase in ρ F hits the ambitious policy harder, to an extent that is proportional to its relative higher cost in the case of failure (the (1 − γ ) term), weighted by the probability of failure. An increase of the reward for success parameter ρ S has two opposite effects. On the one hand, it makes the choice of the ambitious policy less advantageous with respect to the prudent one, because it has a lower chance of success—this is represented by the first term of Equation 9.3. However, the more ambitious policy provides a higher payoff, and the advantage becomes more important as ρ S increases. Whichever of the two effects prevails determines whether an increase in ρ S makes it less or more advantageous to choose the ambitious policy. The overall effect is positive (favoring the ambitious policy) if pPS ⋅ (1 − λ ) > α . If the ambitious policy has a very high outcome in the case of success, relative to the prudent one (if 1 − λ is close to one), or if the ambitious policy is not much less likely to succeed with respect to the prudent one (α is close to zero), then a bigger prize for success will favor bold choices. This is a key result of our model. In discussing the case of the Eiffel Tower, we advanced the hypothesis that the presence of strong incentives would lead to the choice of bland, prudent policies. Our model shows instead that the answer de-
Choice and Execution of Policies
155
pends on a series of factors. Increasing the harshness of punishment indeed favors the choice of prudent policies. The higher the personal cost of failure to the administrator, the less interesting is the extra reward that successful bold policies carry with them. However, an increase in the incentive that applies in the case of success has an ambiguous effect, depending on the relative strength of the two factors that we have singled out. If ambitious policies are not much less likely to succeed than prudent ones, and/or if their outcome in the case of success is sufficiently higher than the corresponding output of prudent policies, then this type of stronger incentive may, in fact, make ambitious policies more appealing. Incentives and Organizational Capability
The particular form of governance also influences the choice of policies through their impact on the capability of the administration to carry out activities of a given complexity. Some organizations are better suited to carrying out complex projects, and complexity just does not come in one size, so that some institutions may be good at dealing with one type, but not with another. For example, the US Federal administration proved remarkably capable in winning the military confrontation with Saddam Hussein’s army during the second Gulf War, but not nearly as capable in supporting the difficult institutionbuilding process that should have followed. Both tasks may be judged, at first glance, to be equally complex, and the US administration arguably proved to be much more effective in dealing with one type of complexity than with the other. Some governance contexts allow the execution of complex policies; others do not. In particular, within a framework where checks on past performances are weak, policies will tend to be poorly executed, because administrators are not chosen according to meritocratic criteria and because existing administrators do not have strong incentives to exert much effort in the job. In such a situation, an innovative politician, with some good and bold ideas in mind, would be wary of proposing them as policies, even when he could manage to build the necessary consensus to get them approved, because he would anticipate that a poorly working administration would not be up to executing them effectively. Anticipating that intelligent but bold ideas may turn into a
156
Choice and Execution of Policies
nightmare for their sponsors, he would rationally choose more modest projects, the ones that could be realistically carried out by a modest administration. The choice of policies would adapt to this unfavorable environment. We include these considerations into our model by allowing the strengths of the incentives to influence the capability of the organization. In particular, we assume that the probability of success of policies now also depends on the strength of incentives. pAS = pPS − α + ϖρ S +ηρ F,
0 ≤ α ≤ pPS , ϖ , η ≥ 0, ϖρ S + ηρ F ≤ α
The above expression makes the probability of success of ambitious policies positively depend on the strength of the incentives—to capture the gist of the argument presented above. Our formulation assumes that strong incentives only improve the probability of success of ambitious policy, the underlying idea being that prudent policies are easier to execute, so that the probability of success is the same regardless of the organization that carries them out. Note that treating the two parameters ϖ and η separately allows us to admit an asymmetric impact of the two types of incentives on an organization. For example, we may imagine that a very high value of ρ F, corresponding to very severe punishments in cases of failure, would in fact have a negative effect on morale and eventually on the capability of the organization—a regime of terror within an administration, moral considerations aside, is hardly conducive to effective governance. The algebra of the model can be easily recomputed to take into consideration the effect of the strength of incentives on the probability of success. In particular, Equations 9.2 and 9.3 now become: ∂Φ = α ⋅ FA + (1 − pPS ) ⋅ FA (1 − γ ) − ρ F F A (1 − γ )η > 0 ∂ρ F if, and only if: ρ F F A (1 − γ )η < α ⋅ FA + (1 − pPS ) ⋅ FA (1 − γ ) EQUATION 9.2′
Choice and Execution of Policies
157
and: ∂Φ = −α ⋅ SA + pPS ⋅ SA (1 − λ ) + ρ S SA (1 − λ )ϖ > 0 ∂ρ S S S if, and only if: pP ⋅ (1 − λ ) + ρ SA (1 − λ )ϖ > −α .
EQUATION 9.3′
We now see that an increase in ρ may even have the effect of favoring the choice of ambitious policy if the impact of the change on the capability of the organization is strong enough. Changes in ρ S, as before, have an ambiguous effect, but the extra term (with respect to Equation 9.3), depending on the parameter ω, makes it more likely that stronger incentives work in favor of ambitious policies. F
Concluding Remarks
The adoption of reputation-based governance would lead to better monitoring of the outcomes of policies and more precise attribution of responsibilities. This, in turn, implies an increased degree of personal and organizational accountability and stronger incentives. Better monitoring is a necessary condition for the adoption of stronger incentives, but obviously not a sufficient one: The boss of an organization may be perfectly aware of who performed well among his staff and still prefer to reward his friends, relatives, lover, etc. An organization that achieves higher quality in monitoring, possibly through the adoption of reputation-based governance, has to consider next how to model its incentive structure. In this respect, we found an important asymmetry between punishments in cases of policy failures and rewards following success. In our simpler model, stronger punishments always lead to more prudence in the choice of policies. On the other hand, the effect of bigger prizes in case of success is ambiguous. From a normative point of view, organizations that do not want to discourage their members from taking bold actions will keep punishments low. The presence of positive feedback from incentives to organizational capability may modify this outcome, because a good organization, in turn, makes it more likely for an ambitious policy to succeed. Such an
158
Choice and Execution of Policies
effect would probably take time to manifest itself, since it would follow, at least in part, the presence of selection forces within the organization that, through career promotions, recruitment, and possibly dismissals, would lead to populating the organization with more capable personnel. It is quite likely that a modification of incentives would not provoke any such changes in the short term; therefore the conclusions of the first simple version of the model would be more relevant. In our model, we did not take the policy maker’s personal attitude toward risk into consideration, so as to focus on the other factors that we have discussed. More generally, in a governance context characterized by a high degree of personal accountability, risk-averse agents would be wary of accepting responsibility for risky projects, unless their expected results were worth it. Also, policy makers would self-select into administrations that suit their degree of risk aversion the best: there are not many risk haters to be found in the casinos of Las Vegas. A further consideration regards what we may loosely define as a psychological effect: the public’s assessment of some artifacts may change over time. As with the Eiffel Tower, a bold design may need time to be accepted, and a bland design may be well received at first, with its lack of originality manifesting itself only after time. Is this a shortcoming of reputation-based governance and its strong incentives? Hardly so: criticizing a model of governance on the grounds that it could lead to people’s conservative tastes’ obtaining too much influence is tantamount to affirming that the public is not able to judge for itself and that, for this reason, the specialists should be given free rein. This point of view, if not outright undemocratic, is at least quite snobbish. The presence of a high degree of accountability, with all its consequences, would put pressure on the public to become more educated and conscious of the implications of their assessments on the policies that will eventually be adopted, and it is a general property of democratic practices to require and to encourage the education of the people. Looking at the matter from a greater distance, the higher level of influence by the public on the actions of the actors of governance takes us again to the issue of how difficult it is, in democratic contexts, to accommodate ambitious long-term policies. To consider an extreme case, it took Stalin and his forced collectivization of the countryside in 1930 and 1931, at the expense of much suffering, to set in motion the irre-
Choice and Execution of Policies
159
versible industrialization of the Soviet Union. While this in no way justified the horrors of those years, most would agree that one of its presumed consequences, the industrialization of the Soviet Union, was in itself a positive occurrence. For a more palatable example, it took a special occasion, the Paris Universal Exposition of 1989, and a very gifted engineer and entrepreneur, Gustave Eiffel, to raise his tower. Daily life in today’s democracies is quite often rather unexciting or even plain dull, a fact that we often overlook because we tend to focus on the few exceptions that stand the test of time, and we forget that, in our regimes, the matching of a gifted leader with an effective policy window is a rare event.
10
Reputation-Based Democratic Participation
Freedom is not staying on a tree, and it also isn’t holding an opinion. Freedom is not a free space. Freedom is participation. Giorgio Gaber
We have discussed how strong ex-post incentives should lead to a choice of policies that is more responsive to citizens’ preferences. Reputation-based governance provides a more direct route for citizens to influence the choice of policies by directly ascribing to them a role in their selection, and here we consider how forms of participative democracy could be established. We focus on the procedural aspects of democratic practices: within a governance process, every chance of democratic participation rests on the presence of procedures of some type. These may be formally coded, as in the case of a referendum to abrogate a law, or they may be quite informal, as in a public demonstration attended by citizens who hope to put pressure on their government in favor or against some policy. Notwithstanding the obvious differences between these two cases, both can usefully be thought of as procedures. We will also illustrate these issues by referring to a well-known case, namely the “participative budget” of the Brazilian city Porto Alegre. Even though this example was cast in a traditional mold and the Internet played only a marginal role, it lends itself to interesting reflections. Reputation-based governance provides a framework that easily accommodates procedures that allow citizens to interact with policy options. These procedures may include, if desired, the possibility that citizens directly influence the choice of policy to be executed. These
Reputation-Based Democratic Participation
161
procedures, moreover, may themselves be reputation-based, in ways that we will discuss. Participation as a Process
When we discuss reputation-based governance, we assume that the general framework is a representative democracy, where policies are chosen by elected representatives. Their decisions, as long as they respect the rule of law, are perfectly legitimate regardless of whether or not they listen to their constituency’s opinions, as their legitimacy derives from the democratic process that led to their election. Obviously, whether elected representatives listen to these opinions could have negative consequences, certainly for their prospects of reelection, but also in some cases of remaining in power. Carrying out highly unpopular policies may result in discontent and even civil unrest, and citizens typically have ways to boycott policies that they strongly dislike. All regimes, not just democratic ones, need some level of consensus on what they do in order to be viable, and the historical examples of leaders who managed to rule through terror and by applying severe reprisals are exceptions to the rule. Some Nazi occupations during World War II come to mind. Tacitus’ (1877, 29) account of Galgacus’ view of Roman conquest, epitomized by his sentence “They make a desert, and they call it peace,” indicates that terror certainly played a role in the Roman empire.1 Terror-based rule, however, does not usually last very long. It either ends in defeat, as it did for the Nazis, or it transforms itself into something more palatable for the oppressed, as it did in the case of the territories conquered by the Romans, who were absolute masters in the assimilation of territories and integration of cultures. Still, all regimes, including dictatorships, to some extent consider the point of view of their subjects and make an effort not to overstep the boundary that could lead to open opposition. For example, according to a stylized model of Stalin’s “consensus problem” in the Soviet Union (P. Gregory 2004, chap. 4), even this quintessential tyrant was very careful in trying to avoid open unrest among the workers and was willing to deliver to them the necessary resources to prevent discontent from threatening industrial production. Stalin was a dictator who had few apparent checks from
162
Reputation-Based Democratic Participation
below and from the people around him, and, strictly speaking, there was no formal obligation for him to collect information on the living conditions of factory workers or to act on them. However, he probably felt some broad political obligation in this respect, which pushed him to collect information, as we know he did, through provincial party cadres and the secret police to make sure that any discontent remained under control. Even these forms of “listening,” as practiced by the likes of Stalin, may be considered forms of citizen participation in the choice and design of policies, although of a negative type. We may assume that Stalin and his acolytes would not choose policies that were so demanding in terms of material and human cost that they ran the risk of serious unrest. By judging the very high cost of at least some of the policies that were indeed carried out—such as the collectivization in agriculture that started in 1929––we may also add that the threshold of what Soviet citizens were prepared to take was quite high, because they feared harsh punishment and also had no feasible alternatives. Nonetheless, they had their limits, and these dictated the presence of some form of popular participation in the design of policies, peculiar though it was. Within democratic governance, elected officials also regularly use informal procedures to learn about citizens’ opinions, for example, through the media and specialized opinion polls. Moreover, they typically act on this information to improve their chances of staying in power and being reelected. In democracies, also, citizens obviously have more options to make their opinions on policies heard, apart from and beyond deciding whom to vote for in elections. The available options may include the possibility of referenda and initiatives: by 2005, more than 70 percent of the U.S. population, for example, lived in a state or a city where these were available, and in Europe ten countries allowed initiatives, as do six postSoviet states (Matsusaka 2005). Initiatives and referenda are examples of highly procedurized processes that permit citizens to take part in the determination of policies. Where they are allowed, some clearly specified conditions have to be met in order for a ballot proposition to go to the voters. The consequences of the vote are also clearly specified: the result of a positive pronouncement by the voters may be either a new law or the abrogation of an existing one. The examples of participation that we have considered differ greatly, but they also have important traits in common. They can all be
Reputation-Based Democratic Participation
163
described as procedures, the purpose of which is to allow the policy maker to learn people’s opinions on past, present, and prospective policies, and eventually to act on them. These procedures may be more or less formalized. A democratic leader learning what the people think by reading the newspapers in the morning while having breakfast is an example of a very informal procedure. Addressing the same need by consulting the monthly report of a contracted pollster represents a more codified routine, and reading the results of popular initiatives that have been allowed by the Constitution is an even more codified option. The procedure allowing citizens to influence the selection of policies may be divided usefully into two separate phases. In the first, citizens express their preferences as regards the policy options available, and these revealed preferences are aggregated and made known to the government. In the second phase, the government decides how to act on these preferences. We will now discuss these two phases in more detail. The Two Phases of Participation
Within the same overall participative process, its two subphases— elicitation of the people’s opinions and ensuing action by the rulers— may have varying degrees of codification. For example, according to the Italian constitution, at least half a million electors have to ask for a referendum for a law to be abrogated.2 In this case, both the elicitation of the citizens’ preference and the consequences to be drawn are highly codified: whenever the objective criteria are met, a referendum is organized according to the electoral law. The impact of a referendum is just as highly codified: if the majority of valid votes is in favor, and if the turnout is above 50 percent, then the relevant law is abrogated. Consider instead the case of the consultative referendum that was carried out in the Netherlands on June 1, 2005, on whether the proposed constitution of the European Union should be ratified. In this case, the referendum was not binding for the government, which had, however, declared that it would follow its indications, provided the turnout exceeded 30 percent. In fact, it was 63 percent, with 61 percent of the votes against. The Dutch government kept its word and did not seek to ratify the proposed constitution. In this case, we see very different levels of formalization of the two phases. On the one hand, the referendum
164
Reputation-Based Democratic Participation
was carried out according to a very codified procedure, just like a general election, while the impact of the outcome of the vote on the choice of policy was much less codified. The Dutch government, in principle, could have decided to ratify the proposed text anyway, albeit at a high political cost, but in full accordance with the legality principle. The obligations of policy makers when it comes to acting on what they know about the desires of the people may also be variously defined. As we have seen, initiatives and referenda may be accompanied by a formal obligation to take consequent action according to the popular will. An opposite case would be the extreme example of Stalin’s Soviet Union: once the dictator learned that the workers’ conditions were very bad, he did not have any formal obligation to take notice. However, it may well have been the case that he still felt obliged not to overstep some limit. Similarly, a democratically appointed prime minister may have no obligation to act on requests to drop an announced new policy by a multitude participating in a major demonstration on the streets of the nation’s capital. However, a sensible leader would think again about the proposed policy after observing such clear signs of popular discontent. Both cases, as different as they may seem, are examples of informal procedures dictating how a leader should take into account citizens’ opinions. Note that informal obligations may be as cogent as formal ones. In a healthy democracy, for example, the desire to be reelected may oblige officials to listen very carefully to what the people have to tell them, even in the absence of any formal obligation to do so. The two stages of the participation process—elicitation of opinions and action upon them—interact with each other. In particular, the obligations that politicians have with respect to citizens’ opinions on prospective policies influence the character of the elicitation of such opinions. If policy makers had few obligations with respect to the opinions expressed, then the citizens would have little incentive to participate in the deliberation process, apart from the solace derived from venting their dissatisfaction with real or prospective policies. On the other hand, if the citizens perceive that the results of their deliberation contribute to the choice and design of policies, then we could expect more participation. The strengths of the effects of citizens’ deliberations influence not only their level of participation but also their allocation of time between
Reputation-Based Democratic Participation
165
alternative forms of collective action. If politicians have few obligations with respect to popular deliberation, citizens could decide that other forms of action are better suited to provoking change, such as angry public demonstrations, public service strikes, or maybe a revolutionary insurrection. The causal link also runs from the first to the second phase, that is, from the character of the activities that elicit the people’s opinions on policies to the presence of procedures to translate those opinions into policy changes. Intense participation from below in the formulation of policy proposals, possibly as a consequence of a strong civil society, would create the demand for procedures to translate proposals into real policies. The debate on Porto Alegre’s Orçamento partecipativo, or “participatory budget,” which we discuss below, reveals the importance of these effects of reciprocal causation. The characteristics of the two phases of the process, as we have seen, can imply widely different degrees of democratic participation, from a dire dictatorship on one extreme to, say, a futuristic Internet-based utopian democracy on the other. A crucial issue on the quantity and quality of participation that is allowed has to do with the definition of the prerogative of agenda setting. Citizens could be allowed to express their opinions only on prospective policies chosen by governments. Again, a good example of an extreme case is provided by Stalin’s Soviet Union. We may argue there that Stalin’s agenda-setting power was absolute; the citizens could therefore only express their lack of satisfaction—say, by writing anonymous graffiti on factory walls, which the secret police would read and eventually report to the dictator—on the policies adopted and on their announced continuation.3 In a more democratic setting, the people could be allowed to discuss only a limited set of options for the same policy, all of them chosen by the policy maker. A more participative posture would also allow citizens to propose new options for policies, but only as different varieties of a more broadly defined policy proposal. Even more power would be placed with the citizens if they could fully control the political agenda by proposing policy alternatives as they deemed fit. Once again, note that what really matters is the joint outcome of the two phases of the participatory process. A highly participatory debate on prospective policy does not translate into effective participation if the elected officials ignore or manipulate the results of the popular
166
Reputation-Based Democratic Participation
deliberations. Even when there are apparently stringent obligations for the government to act on deliberations from below—such as when referenda and initiatives are admitted—the resulting law will not necessarily be enough to procure the desired change. For this to happen, political change very often needs not just laws, but also a proactive administration. An administration that has certain policies imposed on it against its will may have several ways of either procrastinating with their execution or executing them partially or ineffectively so as to only pay lip service to the popular will. The opposite is also true, that is, policies often require some degree of general consensus to be successfully carried out. Just as the government has many ways of not carrying out changes that it dislikes, so have the people, at least in democratic contexts. Participatory Budgeting
Deciding what policies to execute is tantamount to choosing how to allocate a public budget. Procedures that aim to allow citizens direct participation in the choice and design of policies, then, also address the issue of the “participatory budget.” This concept has been the object of debate worldwide, inspired by the experiences of many administrations and, most notably, of the Brazilian city of Porto Alegre, which has had forms of popular direct participation in the allocation of the city budget since 1989. These practices, at least in their most successful incarnations, show a high degree of popular involvement determining the allocation of the city public finances, within a process characterized by the interaction of elements of direct democracy with the presence of representative democratic institutions, resulting in what we may define as a mixed direct-representative democracy. The Porto Alegre experience is a good starting point for our discussion on how reputation-based governance could be amenable to participatory budget practices and on the innovation that it would bring about in this respect. In particular, the Porto Alegre participatory budget, where the Internet does not play any essential role, allows us to consider some key issues that arise when a traditional representative democracy is made to interact with procedures of direct participation. I start by providing a brief and selective account of the history and characteristics of this experience, which I have based on the work of Baiocchi (2005, 7ff.)
Reputation-Based Democratic Participation
167
whence all quotations below are drawn (see also Baiocchi 2001 and Santos 1998). Porto Alegre is a city of 1.5 million inhabitants in the southernmost Brazilian province of Rio Grande do Sul, of which it is the administrative and economic center. It compares relatively well with most other Brazilian big cities, in terms of both economic and social development. At the same time, the city has its share of typical Brazilian problems: “almost a quarter of Porto Alegre’s population today lives in ‘irregular housing,’ such as slums or invaded areas”; also, the city “lost 30% of net formal jobs between 1986 and 1995, mostly in industry,” years during which, like “much of Brazil, the city also experienced an exacerbation of urban poverty and a relative decline in revenues.” Porto Alegre has a long tradition of civic participation. From the end of World War II until the military coup of 1964, the populist Partido Trabalhista Brasilero (PTB) and the more conservative Partido Democratico Social (PTS) ruled the city alternately. The former had the support of a wide variety of organized working-class communities. The relationship between government and civic society had a populist trait that is typical of Brazilian politics, one characteristic of which was the co-optation by the rulers of civic associations within a patronage relation. This relationship did not completely disappear even in the years that followed the military coup of 1964, starting up again more strongly from the late 1970s, when the regime relaxed its pressure on civil society and renewed efforts of co-optation. The changed environment of the early 1980s led to a renaissance of civil activities and to the formation of new umbrella organizations that aimed to coordinate civil society efforts and broaden the participation base. The year 1985 marked the formal end of the dictatorship, and in 1986 Alceu Collares, of the PDT (the heir of the old populist Partido Trabalhista Brasilero), was elected mayor. There were pledges that civil society organizations would be involved in decision making. However, in the two years that Collares remained in office, whatever participation was obtained was perceived to follow the old populist lines of co-optation of friendly associations in exchange for favors and to lack the necessary determination to execute the projects desired by the population. It is against this backdrop that, in 1988, the opposition Frente Popular won the municipal elections on a platform that also included a commitment
168
Reputation-Based Democratic Participation
to increased popular participation. After a rocky start, the administration succeeded in introducing a set of procedures that gave rise to the Orçamento participativo, or participatory budget. Its introduction began with a first round of participatory meetings on the budget that took place around the city in August 1989. These led to a progressive consolidation and routinization of procedures that, by the end of the 1990s, had shaped a welldefined set of widely attended participative practices. Porto Alegre’s participatory budgeting takes places over an annual cycle (Baiocchi 2005, 75). The overall purpose of the process is to allow the citizens, through an elaborate series of meetings, to propose and build consensus on projects that later the administration will be called on to carry out. The cycle begins with a “first regional plenary” where previous years’ projects are discussed and delegates are elected. These delegates attend a series of intermediary meetings where they learn technical criteria and start discussing priorities in each region of the city. In a “secondary plenary,” a number of councilors are elected and regional priorities are decided upon. The newly elected councilors then start discussing the criteria for the allocation of resources, and later discuss and approve the new budget, which is sent to the legislature for approval. Last, the councilors debate sectoral investment and rules for next year’s round of meetings. Several characteristics of the process are worth noting. The first has to do with the role of the city’s administration in both setting up and running the deliberative process. Porto Alegre’s Orçamento participativo is a result of a deliberate act of institution building on the part of the new ruling coalition that was put in place after the 1988 municipal elections. It was certainly encouraged from below, and it responded to popular demands that had developed over the previous decades. However, it was the municipal administration that made the innovations possible. In the daily working of the participatory budget process, the administration also plays an important role, since it provides facilitators who participate in meetings to teach the rules of the game and help discussions stay on track. Moreover, while the deliberative process is ongoing, the administration can ensure that a citywide outlook is present, rather than an approach where each local group lobbies for its local needs, regardless of the broader picture.
Reputation-Based Democratic Participation
169
A crucial aspect relates to the intersection between the participative process, which leads to a set of deliberations forming the proposed budget, and the deliberations of the local legislature. There is no formal rule that requires the local legislature to accept the budget-allocation proposal it receives. However, so far, the legislature has always approved the proposals made by the elected councilors. It is useful to map Porto Alegre’s participatory budget into the two-phase characterization of participative democracy that we have adopted. The first phase—eliciting citizens’ preferences and publishing them—is defined by the several steps in the annual cycle of the participatory budget. The deliberations are made public as the budget proposal that is sent annually to the legislature. The second phase—namely, how the government reacts to the budget proposal—corresponds to the legislature’s decision on whether to adopt the budget proposal itself. The two phases of the participation process in Porto Alegre are roughly subsequent. Phase two, the deliberation by the legislature, takes place after a budget proposal has been prepared through lengthy discussions in a string of meetings. However, there is a degree of contemporaneity between the two phases, inasmuch as during the participatory budget cycle there is an ongoing public debate, where demands and opinions from below interact with the official and unofficial positions expressed by the administration. The two phases interact with each other. In particular, an interesting question is the extent to which the city government succeeds in influencing the participatory budgeting debate, thereby avoiding the risk that the budget proposal eventually presented to the legislature will be too controversial. The elected representatives have a chance to orient the debate through the administration’s representatives who take part in participative budget meetings as facilitators. In principle, the facilitators sent by city hall to the various meetings may also let the people know of various political constraints that may affect the chances of the budget proposal being accepted by the legislature, and there are certainly ways for city hall to make its point of view known while the debates are ongoing. Also, the fact that the legislature, so far, has always accepted the budget proposal does not mean that it has surrendered its prerogatives. Participants in the various meetings of the Orçamento participativo probably
170
Reputation-Based Democratic Participation
perceive not accepting the budget proposal as a credible threat, providing an incentive to discuss proposals that are judged to be compatible with the orientations of the legislature.4 Also note that the role of information technology is secondary in the working of Porto Alegre’s participatory budget. Certainly, basic Internet technologies are used extensively—for communication purposes, to exchange documents, etc.—and Porto Alegre’s participatory budget website5 provides a wealth of information on both the working of the process and the deliberations engaged in each year. However, the process itself essentially works through participation in traditional face-to-face meetings, and the presence of Internet technologies is not essential. Some of the implications of the Porto Alegre experience are worth discussing. One of its first effects was to make administrators more accountable by creating a procedure where past projects are discussed and evaluated (during the first plenary meeting of the participative budget) before the discussion of the new budget starts. During this first meeting, administrators have to account for the progress of projects included in the previous year’s participative budget and, throughout the annual cycle of the participative budget, citizens can question administrators on their ongoing policies (Baiocchi 2005, 104ff.). In this way, the participative budget contributes to strengthening the possibility of tracking policies, an effect that within reputation-based governance would be magnified. Second, the participative budget limited clientelist practices by providing a forum that is not biased in favor of the wishes of “friendly” organizations. As Baiocchi notes, “Indirect evidence of clientelism, such as requests by city councillors for executive projects and the concentration of support of individual politicians in specific areas, is also lower in Porto Alegre than other cities in Brazil. Because of the transparent nature of [Orçamento participativo] projects, it is practically impossible for city councillors to offer specific urban improvements in exchange for support” (2005, 45–46). The key here is the transparent nature of the process, a democratic element in itself since it levels the playing field for the societal actors who interact with the administration. The presence of clientelist relationships is accompanied by rent-seeking activities by those actors who find it convenient to cultivate relations in order to place themselves in the inner circle of those receiving favors. On the other hand, transparent rules of engagement with the administration, like those offered by the participatory budget process,
Reputation-Based Democratic Participation
171
provide a channel through which all the efforts of civil society organizations must pass to have a chance of success. In a clientelist context, to parts of society who strive to be co-opted there often correspond others who have no chance of co-optation and decide to engage in vociferous protests to pursue their goals. The participative budget also puts a brake on these protests by offering a venue where societal demand can be heard and where consensus on a proposal may be built. As Baiocchi notes, “Instead of combativeness or personal mediation, new practices in civil society now involve proposing and evaluating justification of public policies through participatory channels,” and overall, “The [Orçamento participativo] became the central axis for interfacing with civil society and a means of mediating different demands” (Baiocchi 2005, 46–47). Last, the participative budget, by providing new opportunities for social action, encouraged the formation of new civil society associations. The nature of the deliberative process, and the need to consider complex issues in detail, provided an important role for civil society organizations, as these are required to coordinate individual participation in the participative budget. Participatory Design / Budget and Reputation-Based Governance
To the extent that reputation-based governance gives voice to the people who are affected by the policies that elected officials choose, and that it provides a framework to increase the accountability of governance actors, it can be seen as a set of procedures that, broadly speaking, make governance more democratic. An informed citizenry is certainly an important prerequisite for a vibrant democracy. Moreover, as we discussed in the previous chapter, ex-post checks on administrators and politicians turn into ex-ante incentive to better cater to the interests of electors. However, the availability of a system that organizes information on a vast spectrum of public policies would lend itself to uses beyond those described so far. Not only could citizens assess the system to express their valuation of completed policies, but, in principle, appropriate procedures could allow citizens to also express their views on prospective policies, and reputation-based governance would provide a platform on which to implant procedures of participative democracy.
172
Reputation-Based Democratic Participation
Participatory budgeting practices could be introduced within a reputation-based governance framework by taking advantage of the coherent taxonomy of policies and the existing information system. In particular, the availability of a suitable information system containing all the necessary data could support a set of procedures that would allow citizens to participate in the design of policies and the allocation of a budget. The same citizens who use the information system to assess the outcomes of policies could be allowed to express their opinions on prospective alternative policy options. Here, too, it is useful to distinguish between the two phases of participation—preference elicitation and ensuing actions by the elected officials. During the first phase, the citizens could express their preferences over policy alternatives. The definition of the available options could follow different patterns. In the most conservative situation, only a limited number of policy alternatives defined by the government would be contemplated. At the opposite end of the spectrum, citizens would be allowed to evaluate whichever policies they fancy and to take a shot at building enough consensus to support them. In those situations where citizens are allowed to put forward their own proposals, the issue of how these are crafted arises. In particular, the governance framework could support collaborative forms of policy design, allowing citizens to interact appropriately, possibly by means of online collaboration tools. The other side of a decision on policies is the allocation of a public budget, and the publication of the aggregate forecasted cost of all policies for a year would add up to that year’s part of the budget that has been allocated to the participatory exercise. As the preparation of the budget law by the relevant legislature gets under way, the information system would give visibility to all the proposed policies, their costs, and the geographic distribution of their intended beneficiaries. The information system would compute in real time a set of statistics on where and to whom the benefits of a given budget proposal would accrue, to be used by citizens, their organizations, and the press. Publication of all acts and the presence of an even playing field would limit clientelist practices. The insights that we have drawn from Porto Alegre’s experience are useful in trying to predict the likely properties of participative budgeting under reputation-based governance. First, the transparency of the
Reputation-Based Democratic Participation
173
process would limit occurrences of co-optation of societal actors. In parallel, this very same transparency would take the steam out of those actors on the fringe of the political process by giving them the option to join a constructive discourse with the administration. An interesting question is whether, as happened in Porto Alegre, the new participative practices within reputation-based governance would also encourage the formation of new civil society organizations and, more broadly, a denser societal texture. Within reputation-based governance, two forces are pulling in opposite directions. On the one hand, as in Porto Alegre, the new participation possibilities would encourage the aggregation of individuals to coordinate and allow a division of labor in learning the many details arising from each of the items discussed. However, the clear organization of information would diminish the fixed costs of participating in the deliberative process. It is unclear which of the two effects would dominate. The relation between the two phases of participation deserves further reflection. Parts of them could happen at the same time. For example, there could be contemporaneous debates on some issues by both citizens and their representatives within a deliberative assembly. Proposals drafted by citizens could inform the debate of their representatives, the results of which could in turn feed back into the deliberations of citizens. Also, the first phase—the elicitation of citizens’ opinions—could be made contingent on the elected officials’ preferences. Assume a situation where citizens’ opinions count but where the government can take responsibility for deciding as it wishes when it strongly disagrees with the opinion expressed by citizens. An easy way to obtain such a result would be to allocate to the representatives the power of veto on a limited number of occasions. In such a situation, citizens could be allowed more participation in those cases where their proposals do not stray too far from the intentions of the elected officials. If, on the other hand, a policy option stands no chance of being adopted, elected representatives would be able to transparently cut the discussion short in order to prevent citizens from wasting their time on developing the proposal. The general point here is that the particulars of the interaction between the two phases of the participative process may be quite complex. Finally, when participatory design of policy is allowed, constituency members also could develop a form of policy-derived reputation.
174
Reputation-Based Democratic Participation
Their opinion regarding a policy proposal could be compared later on with the prevailing assessment of the outcome of that policy. People who favor policies that turn out to be failures would develop the reputation for being poor selectors of policy. This form of reputation could eventually be used to weight the opinions expressed on alternative policy plans during the participatory design phase. Needless to say, such a possibility sets in motion incentives that could have undesirable effects. At the participatory design stage, conformist behavior would be favored—of the same “beauty contest” type described in Chapter 5. At the policy evaluation phase, citizens would have an interest in assessing positively the policies that they had contributed to choosing. Reputation-based governance hinges on the principle that assessments have to be provided by third parties. When this principle is partially abandoned to allow for a more complex interplay between assessments and the formation of reputations, then various forms of incentive incompatibility in declaring one’s assessments are likely to arise. As a result, attributing a form of policy-derived reputation to citizens is probably not a good idea, but mentioning this possibility serves to reinforce one general message that we already discussed in Chapter 5, when we considered different types and measures of reputation. Within reputation-based governance, there is ample space to invent creative metrics that depend on the assessments of the outcome of policies, well beyond the computation of the simplest reputation indexes. What I have just described is a possibility—it certainly is not an inevitable outcome of reputation-based governance, however determined the pursuit of this ambitious plan might be. Many attempts at forms of cyber-democracy have failed, and we know that citizens are often wary of spending their free time on community affairs. Hibbing and TheissMorse, researching the attitudes of American citizens toward their government, reach the disheartening conclusion that people “do not want to make political decisions themselves; they do not want to provide much input to those who are assigned to make these decisions; and they would rather not know all the details of the decision-making process.” In sum, “the last thing that people want is to be more involved in political decision making” (Hibbing and Theiss-Moss 2002, 1).6 However, this thesis can be turned around to argue that, precisely because the people would rather not bother, it is of paramount impor-
Reputation-Based Democratic Participation
175
tance to facilitate their participation by lowering the costs—that is, if we agree that people’s participation in policy is desirable (as Hibbing and Theiss-Morse themselves do—see chapter 9 of their book). Participation, once again, has to do with the procedures and institutions that allow and guarantee it. The thesis here is that reputation-based governance provides an appropriate framework to think about these processes and institutions, and certainly not that it would magically transform apathetic citizens into dedicated politicos.
11
Final Considerations
And when you get there, there is no “there” there. Gertrude Stein
In this book we have reflected on a world where a continuous distributed process of assessment of policy outcomes, enabled by an appropriate information system together with appropriately engineered incentives, would contribute to a governance model characterized by effectiveness, efficiency, and a minimum of rent-seeking activity. Moreover, if candidate policies were also recorded in the information system before they were selected, then room would be created for advanced forms of popular participation in public governance. In this context, the public could be entitled not only to assess the outcomes of policies, but also to contribute to their design and choice. Reputation-based governance would then support far-reaching electronic democracy practices, where the word “democracy” would be taken to mean both participation and accountability. I have avoided raising an apparently important question: What can be done to implement reputation-based governance? The simplest reason for this omission is that this is not meant to be a “how-to” book. However, a more compelling explanation may be offered on this score. To an increasing degree, policies are managed with the help of information systems where they are codified, together with a vast amount of accompanying information. We may deem the continuation of this process to be almost inevitable. The real question, then, is what concrete shape will the emerging governance practices take? Will citizens be empowered, or not?
Final Considerations
177
Will the new practices support a horizontal, network-shaped, organizational landscape, or, to the contrary, will they tighten the bolts of traditional vertical organizations? Will the new ways of governance consider explicitly governance actors’ reputations, or will their reputations depend only on traditional word-of-mouth? Reputation-based governance, then, is a proposal for the shape that forms of Internet-enabled governance should take, a direction for changes that are already occurring and that, to some extent, regardless of what we think and do, will lead toward the full adoption of information technologies and an all-digital management of policies. In light of this, the question on how to transform the present models of governance into reputation-based ones changes its nature. The task is not the difficult one of setting in motion the forces that will introduce, from scratch, reputationbased governance. The real issue is how to steer a process that is already ongoing, to provide a coherent shape to it, and to develop the broadness of vision that this requires. In this sense, then, “When you get there, there is no ‘there’ there.” This book has not been about giving instructions or about making sharp predictions for the future. It has attempted to reveal interesting possibilities. In the social sciences, there is always a tension between positive and normative analyses, between understanding how things are and shaping opinions on how they should be. This distinction is not as sharp as we sometimes pretend, because explaining how things are leads to better ideas on how to improve them, and proposing how they should be requires a good positive understanding of how they are. The normative aspects of our discourse on reputation-based governance assume a particularly cogent positive character, if we agree with the view that many of the changes that would lead us to the proposed model of governance are happening right now, under our noses. So, our recipes have been shaped, so to speak, by a “conditional positive” analysis: we tried not to describe how governance should be, but how it is bound to become, if the transformation in governance that we are witnessing today takes on the suggested reputational character. Every book is defined both by what it contains and what it leaves out. One issue that we discussed only cursorily is privacy protection. While I hope I made it clear that the privacy of the citizens who are called to assess policy outcomes should and could be guaranteed, one reason for my
178
Final Considerations
relative lack of attention to this important theme is my conviction that the whole concept of privacy is in flux following the Internet revolution. Developing a sufficiently articulate discourse on this topic would have justified a separate project. If citizens deserve privacy protection, it is not immediately clear how to consider any claim that public officials may make in this respect. Consider that reputation-based governance (and, for the matter, any governance model that prizes personal accountability) hinges upon the possibility of tracking the organizations and people responsible for projects. In principle, at least to some extent, forms of tracking are obviously possible even today. If a citizen receives a public service of poor quality and wants to complain, in most democratic contexts he can find out which part of a given administration is responsible for that service, and he has a nominal right to know who is in charge of that office. The Web has simplified this type of search, since many public administrations nowadays publish their organigrams. However, it is one thing to embark on a lengthy search to get even with a shirking public official, and it is another to, for example, feed the barcode for a particular public policy into one’s mobile phone camera and in no time learn who is in charge, how other citizens rated the same service, its cost to the taxpayer, and comparisons with the costs of similar services provided elsewhere. Faced with such a prospect, some governance actors would feel threatened and resist the changes. Quite probably, the defensive line they would take would have a mix of two arguments. First, they would question the correctness and the legitimacy of the judgments expressed by the people who are affected by policies. For example, this argument has been used in the debate on whether to allow students to rate their teachers in many educational institutions around the world. Many universities routinely run surveys and publish aggregate results for each individual course and teacher. Often, these services are student-run.1 Online private services, such as Rate My Professor (www.ratemyprofessor.com), also provide summary information at course level. Other publicly run services, such as the UK’s National Student Survey (www.unistats.com), on the other hand, make public the results only at an aggregate level, say, by subject areas. Teachers faced with requests by their institutions to be rated by their students may resist by arguing that students are not in the best position to assess the quality of the courses that they take.2
Final Considerations
179
Second, they would invoke their right to privacy. What is the right granularity of the information to be made public? For example, if a public official is given a credit card, is it reasonable to allow the public to read on the Web the detailed list of expenses that he incurred, down to the name of the restaurants he chose for his business lunches? In 1958, the Palácio do Alvorada, a creation of architect Oscar Niemeyer, was inaugurated in Brasilia. This is the official residence of the president of Brazil, and it is built on the edge of a small lake. It is a striking structure, a beautiful masterpiece of modern architecture reflected in the water, and it reminds one of a glass box. A house of glass for a president is an elegant metaphor that hints at two parallel themes we have discussed at length. Transparency of what is inside from the outside represents the legibility and the accountability of power. Transparency also allows those on the inside to fully appreciate what is outside, because the availability of a host of policy-related information would obviously be useful for the conduct of policies. A metaphor, unfortunately, is not enough to shape reality. A few years after the inauguration of the Palácio do Alvorada, a military coup d’état in Brazil began a dictatorship that would last more than two decades. However, that metaphor eloquently represents a powerful longing for forms of governance where the people matter. Transforming thick walls into glass windows inevitably puts a strain on some preconceived ideas that we may have, for example, on our rights to privacy of information. If the house indeed has to be made of glass, quite simply, policy makers as such (as opposed to citizens), do not own many privacy rights. People who say the opposite either miss the point or, more probably, do not want the house of power to be made of glass. There are two relevant themes that I find intriguing and that, as the cliché obliges me to declare at this point, I leave for future research. Over the past decades, capitalist economies have shifted away from the public provision of a host of services, mostly public utilities, toward their private provision. This transition accompanied a paradigm shift in economic theory, and it followed a series of technological developments that “dramatically transformed information costs, the relative efficacy of various institutional arrangements and, thereby, the boundaries between the government, the market, and organizations” (Reschenthaler and
180
Final Considerations
Thompson 1996). The changes have been beneficial insofar as market competition provides incentives that are typically absent within a (public) regulated monopoly. However, this has come at the cost of the duplication of essential facilities, considerable regulatory overheads, significant transaction costs, and search costs for consumers who have to choose between competing and often complicated tariff plans. The choice of provision mode for a given service should depend, among other factors, on a balancing of the benefits and costs of the available alternatives. The introduction of reputation-based governance could modify present perceptions of where this balance lies. On the one hand, by making public administrations more effective, reputation-based governance could cause a swing back from regulated markets. On the other hand, the introduction of elements of reputation-based governance within the regulated (private) sector could have the opposite effect, because the wider range of information (on private regulated firms), automatically generated by the information system, would reduce the informational asymmetry between the market regulator and the regulated firms, which is one of the main obstacles to effective regulation. To express this point differently, we have focused on public governance. However, reputational considerations are also obviously relevant in the private sphere, and it would be interesting to understand the impact of reputational-based governance on the particular interface between public and private governance occupied by regulated industries and public utilities. I have argued that reputation-based governance could bring about forms of computer-mediated participation in policy design. In this respect, there is an interesting convergence between two issues that deserves our attention. The first issue is “openness” (as in “open”-source software), and the second is “reputation.” In open-source software the traditional distinctions between inventor, producer, and user are blurred, because quite often a software project is started to solve a problem that programmers themselves face, so the inventor and producer is also the first user of the software. A weak distinction between such roles is also observed in the production of many innovative physical goods, such as some scientific instruments and the gear used for recreational activities like windsurfing, skateboarding, and
Final Considerations
181
snowboarding (Shah 2005). Forms of production of goods that overcome the traditional distinction of inventor, producer, and consumer are on the rise. They are encouraged by three convergent factors: the availability of cheap digitalized design tools; increasingly modularized designs that allow complex problems to be sliced up into manageable units; and the availability of the Internet as an inexpensive coordination instrument that enables communities to collaborate on the solution of complex but decomposable problems (von Hippel 2005; von Hippel and de Jong 2010). The RepRap desktop 3D printer is an open-source initiative that, for a few hundred euros, allows tech-savvy users to put together a machine capable of creating plastic objects according to a design. Incidentally, the printer is also able to print (a good share of) its own parts, so that it can be seen as a (partly) “self-generating machine.”3 This project allows us to envisage a world where “people who have these machines in their homes would no longer need factories to make the goods that they want” (Bowyer 2009). We do not need to embrace this futuristic vision, and factories, with their economies of scale, are probably here to stay. We should recognize, however, that a collaborative mode of production and innovation characterized by a tenuous distinction between innovators, producers, and consumers is not confined to the realm of software and other information goods, such as Wikipedia. It is a mode of production that is widely applicable and where, as we already discussed in the introductory chapter of this book, reputational considerations play an important role.4 In this book we have argued in favor of a clearly legible policy space. Within reputation-based governance, policies are the building blocks of such a space. Often we may interpret a policy as a “mechanism,” in the sense that economists attribute to this term: an institution that can be designed to attain a given goal (Maskin 2008). More generally, a policy is an artifact that people construct to serve a purpose. It is at least legitimate to imagine that the production of policies may share characteristics that we observe elsewhere, such as a lack of distinction between producers (governments and parliaments) and consumers (citizens), and a rather horizontal production process, maybe not unlike what we described in the previous chapter, where reputational considerations play an important role. A sort of subterranean river, then, would be responsible for the set of apparently different phenomena that we observe. Here, the river emerges as advanced forms of reputation-based governance, where
182
Final Considerations
policies are collaboratively crafted; there, it comes up in the form of open-source software. Elsewhere, it surfaces in the guise of groups of lead users who, connected through the Internet, innovate outside the perimeters of firms. In yet another place, the water gushes out in the curious shape of an affordable 3D printer that, among other things, is able to replicate itself by generating its own parts. Are we really looking at the same water, which is possibly about to surface as a majestic river? This is precisely the question that we would like to understand better. To a limited extent, we can answer affirmatively. What is certainly common among these instances is their dependence on a set of enabling information and communication technologies that have come of age and are making all this possible. The disruptive power of the Internet has obliged us to walk a thin line in this book. On the one hand, its impact cannot be downplayed. On the other, technologies, strictly speaking, never determine anything. For example, depending on the context, a better communication technology may support both more intense horizontal or more vertical relations within an organizational landscape. The Roman Empire of the first century had many characteristics that resembled what today we would call a network organization. This situation derived, in part, from the difficulties associated with the communication technologies of the time; for example, navigation across the Mediterranean Sea was impossible for most of the year. A century or so later, pretty much the same communication technology was supporting a much more centralized political and military structure (Luttwak 1976). For a contemporary example, the emergence of the Internet today is enabling not only horizontal hackerdom, but also quite vertical Walmart. The presence of complex and nuanced relationships between forms of governance and the technologies available to support them has been a constant preoccupation in our reflections. Warning once more against forms of technological determinism cannot be anything but an appropriate way to conclude this book.
Notes
Preface 1. Pennac (1992, 145). All translations into English in this book are mine, unless otherwise indicated. I also hope that the reader will be satisfied with this waiver and will not decide to also invoke Rule no. 1 and Rule no. 3 of Pennac’s list, granting, respectively, the right not to read and, possibly worse, the right not to finish the book. Chapter 1 1. Also see Bajari and Hortaçsu (2004) for a review of the available empirical evidence on other aspects of the workings of Internet auctions. Dellarocas (2003) and Jøsang et al. (2007) provide overviews of Internet-based reputational systems, which also highlight their shortcomings. We will discuss these in due course, as they may apply to reputation-based governance. 2. See MacLeod (2007) on formal and informal mechanisms for the enforcement of contracts, the latter involving reputational incentives. 3. Open-source software is software for which the original code is made available under a copyright agreement, specifying rights of modification and of redistribution. According to the prevailing definition, its copyright should satisfy the general conditions laid out by the “Open Source Definition” (Open Source Initiative official Web site; www.opensource.org/DOCS/OSD. All the Web resources cited in the book were last checked on October 20, 2010). 4. The debate on open-source software has benefited from an enlargement of perspective, thanks to the consideration of historical antecedents, such as the rise
184
Notes to Chapter 1
of open science in the sixteenth and seventeenth centuries (David 2004; Willinsky 2005). 5. We may say that the emerging forms of governance are subject to a form of “technological coloration,” to borrow the expression that Dunleavy et al. (2006) apply to the processes leading to what they call “digital era governance.” This coloration is meant to mark the difference, in their words, to “any simple technological determination” of those processes. 6. An idea that results with notable clarity in a letter written in 1813 by Thomas Jefferson (1855, 181): “Considering the exclusive right to invention as given not of natural right, but for the benefit of society, I know well the difficulty of drawing a line between the things which are worth to the public the embarrassment of an exclusive patent, and those which are not.” 7. This reform represented a U-turn with respect to the practices of (what used to be called) “New Public Management,” which had led in many countries to a proliferation of relatively small goal-oriented agencies. Moynihan and Roberts (2002), arguing that the new context represented a “complete turnabout in reform priorities,” situate the reform in the debate. 8. Jackson (2009) refers to a “midlife crisis” of Intellipedia, linking its early success to the enthusiasm of a vast number of early adopters, while lamenting the fact that the instrument has not yet entered the main routines of many agencies. Meyer and Rowan (1977) reflect on the relation between the institutionalization of organizational artifacts and their being perceived as “appropriate.” Every reformer knows that it is difficult to institutionalize an innovation, even though it may be crucial for the success of the organization concerned. 9. According to Italian law, there is only one case when, at least in principle, an administration can exclude a firm from a tender; i.e., if the firm had already performed very poorly in a work carried out for the same administration (Decreto Legislativo 2006, Art. 38e). The word “reputation” never appears in Italian legislation on public procurement, nor does it appear in the relevant Directives of the European Union (Directive 2004/17/EC and Directive 2004/18/EC). In U.S. federal procurement, however, reputational information is allowed to play a role, as we discuss in Chapter 8. 10. Addressing any shortcoming that the rational choice approach may have in explaining a social phenomenon (for example, and most importantly, collective action; see Whiteley [1995]) by inserting the needed motivational elements in the utility function runs the risk of delivering theories that explain anything we like and that, as such, cannot be refuted. Also, each motivation in the utility function immediately begs an explanation of the mechanisms (for example, evolutionary) that brought it there. These are serious questions in the social sciences, and I will not answer them in this book.
Notes to Chapters 1 & 2
185
11. See the examples and the discussion in Osimo (2008). 12. See the discussion in Dunleavy et al. (2006). Margolis (2005, 766) suggests that “cyber-democrats saw the internet as the means to break away from the cycle of soaring promises and failed fulfilment that each new mass medium had engendered since the Industrial Revolution.” 13. In the words of Rethemeyer (2007), “Reconnecting politics with society is still primarily the work of organizational and institutional reformers, not hardware and software engineers.” 14. See also the discussion in Boyte (2005).
Chapter 2 1. Possibly corresponding to today’s town of Vayrac, in the region of Haut Quercy-Dordogne in France. The episode is narrated in the eighth (and last) book of the De Bello Gallico, which, unlike the previous ones written by Caesar, is the work of Aulus Hirtius, who had been one of his legates. I thank Massimo Portolani for drawing my attention to this episode of the Gallic Wars. My source of information is Caesar (1996). In particular, see p. 621, n. 1, and p. 632, n. 32-2. 2. For a more formal and complete treatment of the themes treated in this section and in the next, see Cabral (2005) and Bar-Isaac and Tadelis (2008). For a similar “calculative” approach in the sociological literature, see Coleman (1994, 91–115). 3. Technically, the Nash equilibrium in this case is also an equilibrium in dominant strategies (both players are better off playing “cheat” for any choice of strategy of their counterpart), and there is an outcome of the game that is Paretosuperior to it—both players would be better off if they could agree to play “honest.” Any introductory book on game theory will describe in more detail both the prisoner’s dilemma and what happens when such a game is repeated (an issue that we discuss next). See, with a focus on reputation issues, Kreps (1990, 503ff.). In a situation where two criminals have been arrested and each one is interrogated in a separate room, with each one getting a lighter sentence by betraying his companion, the two strategies, “be honest” and “cheat,” could also be seen as “say nothing” and “turn your buddy in.” Hence the original name of this archetypal game, or “prisoner’s dilemma.” In a market context, the two strategies could correspond to “exert effort” (be honest) and “do not exert effort” (cheat)—for example, in the production of a good to be sold, as in Bar-Isaac and Tadelis (2008). 4. Cremer (1986) explicitly models this idea (using an overlapping generation model), which also allows him to draw indications on the differing incentives of an individual to exert effort at different ages, in the spirit of Holmström (1982), on which see the next chapter.
186
Notes to Chapter 2
5. Even when using the game-theoretic approach, not everyone calls “trust” what may emerge from the repetition of a prisoner’s dilemma game. Bar-Isaac and Tadelis (2008), for example, call it “reputation”—see their “pure hidden action” model in their Section 4. 6. Alignment of incentives is just the other side of the coin of a human artifact being an organization, defined as a social structure with one or more common goals. For these goals to be pursued, there must be some degree of coherence among the incentives faced by the members. 7. Kreps et al. (1982) show that “incomplete information about one or both player’s options, motivation or behaviour can explain . . . cooperation.” In their model, the presence of a belief that one or more players are other-regarding, in the sense that they cooperate when self-regarding players would not, may induce self-regarding players to cooperate more than they would do otherwise. 8. Castelfranchi et al. (2009) provide a formal ontology of the concept of trust where a distinction is made between distrust and mistrust (of agent A vs. agent B). In their characterization, distrust applies to B’s not being able, or willing, to do something that A likes. Mistrust refers to B’s being able, and willing, to do something that A dislikes. 9. Incentives may be such, moreover, as to cause not action, but inaction. 10. Since such updating is done using Bayes’s theorem, this family of games is known as “Bayesian.” In a nutshell, Bayes’s theorem is a simple formula that allows us to compute an update of the probability of something happening, after new relevant information arrives. 11. The possibility that the intrinsic type of a producer changes (in Cabral 2005, according to a Markov process), possibly following exogenous technology shifts, provides a further incentive to exert a high level of effort to those players for which effort influences product’s quality. In some cases, efforts today are not necessarily positively correlated with existing reputation (in Cabral 2005, chapter 3, the two results are obtained respectively for a “pooling” and for a “separating” equilibrium). 12. In game theory terms, the strategic situation facing the Maghribi traders may be described by a moral-hazard model (of trust) with imperfect monitoring. It is a moral hazard model in the sense that all agents are equally intrinsically honest and that there are no “types.” Each one may have an incentive to cheat following only his self-interest, and not any intrinsic qualities of his. The imperfect monitoring aspect derives from the fact that ascertaining whether an agent had cheated, or had in fact been victim of an adverse shock, was not straightforward. The possibility of monitoring rested on the presence of more than one coalition member in each trading post who could provide additional information. See Greif (1989 and 1993) for further details.
Notes to Chapters 2 & 3
187
13. See Cameron (1993) on the interruption of long-distance trade in the early Middle Ages, and Lopez (1976) on the revival of trade of the tenth century. 14. GMW note that their interpretation of the role of merchant guilds complements “the view more common among economic historians that merchant guilds emerged to reduce negotiation costs, to administer trade and taxation, to extract privileges from foreign cities, and to shift rent in their own cities.” They also underline that their analysis is relevant for merchant guilds, but not for craft guilds. 15. In fact, according to our characterization above, it would be more correct to talk about “trust” than about “reputation.” 16. A reputation model where producers invest in quality is in Rob and Fishman (2005), a moral hazard model that links investments in quality with the customer base, obtaining that the two are positively related. 17. Hirtius was writing after Caesar’s death, and we may doubt this and other particulars of his account. However, we have good reason to believe him in this instance. First, while we know that Caesar was perfectly able to commit atrocities, his reputation for leniency was recognized by (Roman) friends and adversaries alike (see Caesar 1996, n. 44-1). Secondly, the episodes recounted in De Bello Gallico had been witnessed by literally thousands of Roman soldiers and scores of patrician members of Caesar’s staff, many of whom were still alive when Hirtius was writing. Blatant lies would have been recognized and would have put the writer in a negative light. True, we may simply decide to ignore any historiographical debate about the veracity of De Bello Gallico and simply sit back and enjoy the multifaceted nuances of this account, as is. 18. The fact that a developed philosophical discourse arose in classical antiquity on the necessity to restrain rage (to cite the title of Harris’s 2001 book) could indicate that, precisely because the issue was seen to be so problematic, decisions motivated by anger were more common then.
Chapter 3 1. Popular participation in public affairs also increases the legitimacy of democracy, which in itself is a reason to prefer participative models of governance. Also, popular participation may lead to more trust in government and to higher citizen satisfaction. For example, Olken (2010) randomly assigns a group of forty-nine Indonesian villages to one of two alternative ways, characterized by differing degrees of popular participation, to choose public investment projects. He finds that, even when controlling for the type of chosen project, the more participative method leads to a higher average degree of citizen satisfaction with the selected project.
188
Notes to Chapters 3 & 4
2. Any discourse on red tape, moreover, should consider the democratizing role of bureaucratic rules, which “also contribute to democratic equality because they are relatively blind to the wealth and other resources of the citizens they serve.” (Olsen 2008; see also the discussion in Olsen 2005). 3. See the discussion in Buelens and Van den Broeck (2007) and in Wright (2007). On “public value” as a motivator for public officials, see Moore (1995). 4. Another strong incentive for bureaucrats may come from corruption. Receiving bribes without being caught usually requires a special talent and, often, hard work. In Chapter 7 we discuss the theme of corruption and of rent-seeking activities in general. 5. See Wilson (1989), who provides examples. Han Chun and Rainey (2005) offer empirical evidence on the relation between measures of goal ambiguity and organizational performance, using data from the 2000 National Partnership for Reinventing Government Survey of (U.S.) federal employees. Their measures of goal ambiguity are negatively related to managerial effectiveness, and overall they “support the desirability of trying to clarify objectives for government agencies.” 6. Such a possibility has long been recognized as a solution to certain types of corruption, because it limits the rent extraction possibilities of each single office (Rose-Ackerman 1978, chaps. 7 and 8). 7. Whether the advantages exceed the disadvantages is a complicated question whose answer depends on several factors, one of which that figures highly is the way that local projects are financed. In this respect, Bardhan and Mookherjee (2006) contemplate three broad alternatives: taxes raised autonomously by the local government, user fees, and fiscal grants from the center. Their model shows that when a local elite succeed in capturing the political process, the possibility of fiscal autonomy is of particular advantage to them, at the expense of the ordinary people who end up carrying the burden of regressive taxation.
Chapter 4 1. Also, such relations allow for the computation of measures of proximity among policies. For examples, two policies that are connected to each other are “closer” than two policies that are not directly related but which are both connected to a third policy. Appropriate metrics could be defined to indicate the distances among policies. 2. A sharp dichotomy between state and society is a severe oversimplification of reality, for which the only justification is the need for concision. With respect to the problem of legibility of society to the state, James Scott’s reading
Notes to Chapter 4
189
is much more nuanced than my brief summary here. Where the issue of the legibility of the state to society is concerned, I am not assuming the presence of a purposeful strategy of obfuscation by the powers that be, and even less am I suggesting the existence of a sort of conspiracy by the state against the people. I do argue, however, that many benefits that the elites may enjoy are partly based on the reduced legibility of their actions. This leads to the elites’ lack of interest in lifting the veil of opacity that, if not checked by countervailing forces and institutions, may in turn lead to institutional outcomes that are in fact observationally equivalent to the presence of a successful and purposeful strategy of obfuscation. 3. Aside from their human limitations, the people often have a limited desire to devote much time to this endeavour, as Hibbing and Theiss-Morse (2002) argue. 4. On such a “struggle toward openness” in the United States, see Fung et al. (2007, 24–27). The march toward the right-to-know was slow, and it suffered setbacks, such as a general retrenchment under President George W. Bush. 5. Every reform has its counter-reform, and the excesses of New Public Management are in the process of being corrected by Post–New Public Management, one of whose tenets is reaggregation of functions into the governmental sphere. Labels to summarize ideas that have been wearisomely discussed are but a simplifying tool to facilitate discourse. There is also a debate on whether the label “New Public Management” should be used at all. See Hughes, who defines New Public Management as “a spectre, a chimera or some other kind of ghostly apparition” (2008). 6. See also Alford (1996). 7. See Light (2006) on the U.S. historical perspective. 8. Gregory (1995) adopts Wilson’s (1989, chap. 2) differentiation of public agencies according to (a) the observability of an agency’s outputs and (b) the observability of its outcomes. This gives rise to four types of administration: (i) production (both outputs and outcomes are measurable); (ii) procedural (outputs are measurable, but outcomes are not); (iii) craft (outputs are not measurable, but outcomes are); and (iv) coping (neither outputs nor outcomes are measurable). He concludes that “especially in craft and coping situations, . . . because work cannot be made visible—unless the nature of the task is transformed in the process— accountability must be broadened to become responsibility.” 9. See, among many others, Bohte and Meyer (2000), Heinrich (2002), Boyne et al. (2005), Brodkin (2006), and Kassel (2008). On the concept of goal displacement, see Blau (1956). We will consider the issue further in Chapter 8, where I will argue that the joint use of performance measures with those user-generated assessments that reputation-based governance generates and organizes automatically
190
Notes to Chapter 4
would alleviate the problem of goal displacement that derives from the use of performance measures alone. 10. Hood and Peters (2004) survey the sociological literature on unintended consequences and apply it to New Public Management reforms. 11. We adopt a simplified view where firms are monolithic entities, so that their assessments are expressed by a generic representative, say, their administrator. Firms for us are simple “black boxes,” because we focus on public governance so that we do not describe or consider explicitly the governance of private entities; hence we do not model their internal structure. 12. Alternatively, organizations could be allowed to restart the computation of reputation whenever a change in name or merger occurs. However, this possibility (analogous to changing one’s identity at no cost on systems such as eBay) establishes an incentive to change one’s identity when one’s reputation is poor. See Jøsang et al. (2007) for a discussion in the context of Internet-based reputation systems. 13. For a historical analysis of the politics-administration dichotomy in the United States, see Rosenbloom 2008. 14. The backdrop for these considerations is the changing role of nongovernmental organizations and policy networks in public governance. See Rethemeyer (2006, 2007) for a concise explanation of the increasing role of policy networks in the U.S. historical context, and for a reflection on the impact of the Internet on their functioning and openness. 15. More precisely, for this to be true, the publishing unit should also be in place, guaranteeing the necessary publicity, on the Web, of the relevant information. 16. The issue of identification vs. anonymity in the Internet age, and of technologies that, often against our wishes, have the ability to identify us (think of the implications of face-recognizing software), is simply too complex a theme for us to consider. See Marx (2001) for a concise list of reasons rationalizing, respectively, anonymity and identifiability (“to judge reputation” is one of the reasons that appears in the latter category). 17. In the United States, following the terrorist attacks of September 11, 2001, there has been a greater willingness to force individuals to carry and disclose an identity card (for example, when boarding a domestic flight). Identification for the purposes of accessing the information system supporting reputation-based governance could possibly use the existing identification system, which is largely based on government-issued driving licenses. 18. The United States, where citizens have to voluntarily register in order to vote, is also a country where voting, the least expensive form of political participation, displays a relatively high degree of class bias. See Lijphart (1997). 19. See the discussion in Jøsang et al. (2007).
Notes to Chapters 4 & 5
191
20. In this respect, the distinction between what is traditional and what is new has become blurred. For instance, a mobile phone camera is an effective and accessible tool for providing proof of one’s vote within a patronage relation.
Chapter 5 1. In Chapter 8 we describe a more articulated setting, where public administrations and firms rate each other and the overall rating of a firm is a weighted average of the valuations that its projects received from both the public and the public administrations. Also, as we discussed in the previous chapter, assessments of a completed project will typically be on more dimensions. 2. Remember from Chapter 2 that the prisoner dilemma (eventually repeated, so as to raise the possibility of trust to emerge) was a case of moral-hazard game: Agents are all equal and, depending on the incentives they face, decide whether to cheat (shirk) or be honest (exert effort). Learning games, where players are of different types, lead to the emergence of adverse selection issues. 3. We witness here one of those cases where the availability of more information leads to worse overall outcomes, as in Hirshleifer (1971) and as discussed more generally in Bassan et al. (2003). 4. If, as in Tadelis (2002), trading in reputation remains secret, public administrations would not separate the type of firm that they face—low cost or high quality. Such a lack of separation has in itself a cost, because there are projects for which one type of firm could be preferred to the other. For example, in a case where quality correlates positively with the ability to carry out complex projects, it would be beneficial to the administration running such a project to choose a high-quality firm. 5. Obviously, a person may at the same time be a citizen who is affected by a set of public policies and an administrator in charge of one or more public policies. Only in his latter role will he own a reputation in the sense that we have discussed so far. 6. With reference to John Maynard Keynes’s “beauty contest,” mentioned in chapter 12 of his General Theory as a way to illustrate some mechanisms governing the formation of expectations. He writes of “those newspaper competitions in which the competitors have to pick out the six prettiest faces from a hundred photographs, the prize being awarded to the competitor whose choice most nearly corresponds to the average preferences of the competitors as a whole; so that each competitor has to pick, not those faces which he himself finds prettiest, but those which he thinks likeliest to catch the fancy of the others competitors, all of whom are looking at the problem from the same point of view” (Keynes 1973, 156).
192
Notes to Chapters 5 & 6
7. These weights would be produced by the reputation unit and would feed back into the assessment unit, so that these two units would no longer be defined by the upstream–downstream relation depicted in Figure 4.3. 8. However, note that the effects of outliers on beauty contest reputation could be made to depend on a measure of dispersion of the assessments. If a policy is considered to be controversial and it receives diverging assessments, then the beauty contest reputation of the people who cast outliers would not be (seriously) affected. On the other hand, if most people are concordant in their assessments to a policy, outlying assessments would translate into a worse beauty contest reputation for the people who cast them. 9. Miller et al. (2005) propose a scoring system able to induce honest reporting that, as in our case, focuses on a comparison of reports. Their system is based on the fact that “it is generally possible to use budget-balancing transfer payments to extract agents’ private information,” an idea that is well established in economic literature. The crafting of scoring systems and of metrics of reputation highlights the role of the economist as a “mechanism designer,” or as an “engineer,” to quote Hal Varian (2002). Chapter 6 1. Formal models also could help in clarifying those strategic reasons that would suggest the hiding of information. Dewatripont et al. (1999b), for example, consider a few cases of “information garbling” having opposite effects within different career-concern models. 2. This shift, in fact, is already happening in several contexts where information is generated and stored electronically. For example, regarding the studies of the open-source mode of software production, Dalle et al. (2005) note: “[There is a] growing abundance and accessibility of quantitative material concerning the internal workings of ‘open epistemic communities.’ The kinds of data that are available for extraction and analysis by automated techniques from repositories of open source code and newsgroup archives, and from the email subscriber lists . . . also offer a rapidly widening window for research.” 3. As an example, consider Abraham’s (2005) account of the role of the National Association for Business Economics (NABE) in lobbying for new statistics that better reflect the postindustrial character of the U.S. economy. 4. The tendency that I commented upon is already visible in initiatives such as DATA.gov, run by the U.S. government, whose purpose is to “to increase public access to high value, machine readable datasets generated by the Executive Branch of the Federal Government” (www.data.gov/about). However, initiatives such as this one are cast in a traditional setting where, for the most, the raw data
Notes to Chapters 6 & 7
193
are of the ad hoc type. See also the discussion in Ho (2007) on “community indicators,” giving “power to the public by empowering and supporting non profit organizations in a community to measure the quality of life and policy outcomes through self-organized efforts and collaborative partnerships between government, nonprofit, and business organizations.” In the context of performance measurement, Yang and Holzer (2006) note that “performance measurement is a social-learning process involving the evaluators and the evaluated. Participation, interaction, and communication are essential characteristics of such a process.” 5. See Wooldridge (2001, chap. 18) for a textbook explanation, and Imbens and Wooldridge (2009) for a review of the literature. 6. It is precisely in the light of these concurrent efforts that the opening quote to this chapter from Tjalling Koopmans (1947) should be seen. Koopmans was awarded the Nobel Prize in Economics in 1975 for his contributions in the development of the econometric discipline. For a historical account of these developments, see Mary Morgan (1990). 7. See also the discussion in Van Dooren (2009). 8. An example of an econometric forecasting model accessible through the Web is the model of Ray Fair (see http://fairmodel.econ.yale.edu/main2.htm). The statistics feeding that particular model are not integrated, but of the ad hoc type— national accounts, price and monetary variables, and others. Chapter 7 1. In principle, the information system supporting reputation-based governance could be integrated with project management software, the routine output of which would automatically communicate the project’s progress and costs. The adoption of such an extended information system could even be a requirement that the administration imposes on the contracting firm(s), particularly as a device to monitor costs when a cost-plus contract is in place, i.e., a contract where the risk of cost overruns is shared between the contracting firm and the administration. In other words, the adoption of reputation-based governance could also require not only public administrations, but also contracting firms to manage projects in a particular way. 2. Another important implication of adopting broad policies is that their project managers will be more powerful and visible inside their organization. Particularly in a transition toward a more horizontal and project-oriented public administration, defining broad (horizontal) policies would be a way to empower a group of project managers and to create a useful conflict, where project managers would fight against the control of the heads of department in order to obtain the resources and political clout they need to honor their mandate. Such a move
194
Notes to Chapter 7
would help to attribute power to the horizontal teams within the typical power struggle taking place between the horizontal and the vertical elements of a matrix organization. 3. See Dosi et al. (2002), Brusoni et al. (2004), Ethiraj and Levinthal (2004), and Marengo and Dosi (2005), who offer more formal arguments supporting this intuition. The third paper cited provides some considerations on the relationships between modularization and organizational structure, and also proposes a review of the antecedents to the literature on modularization. In particular, it refers to the work of Herbert Simon on the concept of decoupling of complex problems. 4. We could proceed further in our line of reasoning and consider the concept of modularity as very general and applying equally to the architecture of products, of actions such as policies and programs, and also of those particular artifacts that we call organizations. The general framework of the formal institutions (and information systems) supporting reputation-based governance, as illustrated in Figure 4.3, can be taken as an example of modular organization. 5. See Krueger (1974) and Shleifer and Vishny (1993). A vast body of literature on the economic damages of corruption has developed over the last years, an often cited precursor being Mauro (1995). Svensson (2005) offers a survey on the debate on corruption from an economist’s view point, as does Lambsdorff (2006), who focuses on the causes and consequences of corruption. 6. See www.transparency.org/policy_research/surveys_indices/cpi/ for the most recent data and Lambsdorff (2003) on methodology. 7. See www.transparency.org/policy_research/surveys_indices/gcb. 8. For a general assessment of these issues, see Treisman (2007). 9. See Della Porta and Vannucci (1992). Italy moved up significantly in the TI-CPI ranking (implying less corruption) in the years that followed. For a qualitative assessment of corruption (in public works) in Italy, see Golden and Picci (2006). 10. The permanent inventory technique allows computing an estimate of the stock of capital by adding current and past flows of investments. Investments on given types of goods that are older than their relevant “service life” are not included in the computation. Past investments may be counted (at constant prices) as they are, or they may be deflated in proportion to their age in order to take care of depreciation. In the former case, the estimate is of the gross capital stock. In the latter, of net (of depreciation) capital stock. See Golden and Picci (2005) for further details. 11. Other authors have measured corruption and/or waste using objective data. Di Tella and Schargrodsky (2003) consider the prices of some basic inputs bought by hospitals in Buenos Aires before and after a crackdown on corruption. Olken (2007) measures corruption in a group of Indonesian village road projects by discrepancies between official project costs and an independent engineer’s
Notes to Chapters 7 & 8
195
estimate of costs. Bandera et al. (2009) look at prices of standardized goods purchased by Italian public administrations through the national public procurement agency (Consip). They are able to distinguish between forms of corruption (that they call “active” waste) and inefficiencies (or “passive” waste). 12. Not for long. Vitale finally resigned from his post in June 2007. 13. On professional journalism in the United States from a historical perspective, see McChesney (2008, chap. 1). For a pessimistic outlook on the current crisis in professional journalism, see Nichols and McChesney (2009). For a more optimistic point of view on the overall working of the media system, see Gentzkow and Shapiro (2009).
Chapter 8 The first five sections of this chapter draw from Picci (2006c). 1. In republican Rome, there already was an articulate jurisprudence dealing with the procurement of public works. See Trisciuoglio (1998). 2. At the moment, infrastructure is usually public, but we are witnessing a shift toward its private provision, the English Channel Tunnel being a representative example of such a trend. While this shift is in itself an interesting (and wellstudied) research topic, we focus on the public provision of infrastructure. 3. See Bajari et al. (2009). The source of the data they consider, consisting of about 25,600 projects, is the Construction Market Data Group. 4. See Kelman (2002, 2005); Rose-Ackerman (1999, 59–68). 5. “Except as set forth in paragraph (c)(3)(iii) of this section, past performance shall be evaluated in all source selections for negotiated competitive acquisitions expected to exceed the simplified acquisition threshold.” Federal Acquisition Rules, 15.304.3.iii, available at www.acquisition.gov/far/current/html/ Subpart%2015_3.html#wp1088864. 6. A system providing strong incentives only rarely punishes misbehavior, because the incentives are enough to convince actors to behave properly and also because increased accountability of firms eventually weeds out poor performers. Statistical evidence shows that this is indeed the case in the U.S. Department of Health (more on this below), where the vast majority of projects are rated to be good, excellent, or outstanding—possibly a case of “grade inflation” (Contractor Performance System–National Institutes of Health Web site, http://cps.od.nih.gov/, official presentation dated January 28, 2005). 7. According to an observer, “The rules governing the evaluation of past performance offer broad discretion and a bonanza of opportunities to steer the evaluation in a desired direction” (Petrillo 2005, from which the following citations are
196
Notes to Chapters 8 & 9
also taken). One source of problems is that “when past performance is part of the evaluation, most solicitations require that participating companies identify some of their ongoing and completed contracts.” As a consequence, solicitors have scope for strategic behavior in choosing their references, and the possibility of excluding scores arbitrarily adds to the uncertainty of the process: “Varied scores make manipulation possible. The evaluator can lift a competitor’s evaluation by excluding low scores and trash it by doing the opposite. There is no rule that the evaluator needs to include every contract mentioned in a proposal.” The relevance of a given past job is also an issue, because “current law doesn’t require more relevant contracts to have greater weight than less relevant ones.” 8. In the incarnation of the system developed at the University of Bologna, citizens are allowed to select up to ten projects to assess, the rationale being that such a quota (or any similar one) would force citizens to choose the projects that they feel affect them most (see www.rebag.it). A list of usable user identifications allows anyone interested to log into the system as a citizen, public administrator, or firm to experiment with the range of actions available to each of them. 9. In the official documentation on Federal Public Procurement, lip service is paid to the need to listen to the opinions of final users (see, among others, Office of Federal Procurement Policy 2000). However, final users are taken to be the users within the administration, and not the public at large. 10. This is the case in the United States, where pork barrel is often wrapped in omnibus legislation that attracts support from both sides of the aisle. See Golden and Picci (2008) for more details and for references to the literature. 11. With reference to Blau (1956). See also Nicholson-Crotty et al. (2006). 12. See also the discussion in Van Dooren (2009) and Taylor (2009). 13. However, the treatment of information that is needed for the tender to work would be accommodated easily within the information system described in Chapter 4. For example, service providers would be the constituencies of their own services, so as to be allowed to register their ex-ante prognosis for the outcome of a service. Chapter 9 1. More than fifty years ago, Charles Lindblom, to describe this inherently messy situation, wrote of policy making as “muddling through” (Lindblom 1959). 2. Or statues. Eiffel also built the frame of the Statue of Liberty on Ellis Island, New York. Apparently that other bold piece of architecture did not raise much opposition, either because of a difference in tastes between the demanding French and a more progressive American public or because it would not have been polite to complain about gifts received or, more likely, because the carcasse of the Statue of Liberty was not visible.
Notes to Chapters 10 & 11
197
Chapter 10 1. See also Luttwak (2007). 2. The demand for a referendum can also originate from five Regional Councils; see Article 75 of the Italian constitution, www.quirinale.it/costituzione/costi tuzione.htm. 3. Strictly speaking, Stalin’s agenda-setting prerogatives could not have been absolute, his set of admissible policies being constrained by the shared ideology of the Soviet state. Had he proposed the privatization of the means of production, it is quite probable he would have met with fierce resistance from the Soviet nomenklatura and society. 4. This is only speculation, based on a reading of the institutional setup of Porto Alegre’s participative budget, since I was not able to find any explicit analysis on the issue in the literature available. 5. www2.portoalegre.rs.gov.br/op/. 6. The wider problem, clearly, is that of the “paradox of participation—the proposition introduced by Mancur Olson that rational actors will not get involved in collective action in order to achieve common goals” (Whiteley 1995, referring to Olson 1965). Chapter 11 1. As an example, this is the case of CAPE at the University of California at San Diego, where teachers may decline to have their courses evaluated. See www .cape.ucsd.edu. 2. For example, there is evidence that easier courses tend to be rated more favorably by students, creating a corresponding perverse incentive for professors. However, it is possible to take this attitude into consideration so as to provide measures of course quality that are not affected by this factor. See the discussion in Gruen (2009). 3. See www.reprap.org. 4. The whole of Internet governance—indicating with this loose term the way that the Internet has been ruled—in fact is to a great extent “open” and “reputation-based.” It is based on open standards, a lot of open-source software, and a series of institutions (such as the Internet Engineering Task Force and the World Wide Web Consortium) that are rather horizontal in the way they build consensus on technical standards, and where individual reputation plays a prominent role in determining who has a bigger say on a given problem.
Bibliography
Abraham, Katharine. 2005. What We Don’t Know Could Hurt Us: Some Reflections on the Measurement of Economic Activity. Journal of Economic Perspectives 19 (3): 3–18. Alford, John. 1996. Who Said Production Was Simple? Delineating the “Public Production Process.” Australian Journal of Public Administration 55 (4): 157–163. Allen, Naomi, Andrea Casillas, Jason Deveau-Rosen, Jason Kreps, Thomas Lemmo, Joseph Merante, Michael Murphy, Kaydi Osowsk, Christopher Wong, and Mark Webbink. 2009. Peer to Patent: Second Anniversary Report. Center for Patents Innovations at New York Law School. http://dotank.nyls .edu/communitypatent/CPI_P2P_YearTwo_hi.pdf . Amnesty International. 2009. Amnesty International Report 2009, Italy: Summary. www.amnesty.org/en/region/italy/report-2009. Augustus, Caesar Octavianus. 1991. Res Gestae Divi Augustus. Pordenone, Italy: Edizioni Studio Tesi. Baiocchi, Gianpaolo. 2001. Participation, Activism, and Politics. The Porto Alegre Experiment and Deliberative Democratic Theory. Politics and Society 29 (1): 43–72. ———. 2005. Militants and Citizens: The Politics of Participatory Democracy in Porto Alegre. Stanford, CA: Stanford University Press. Bajari, Patrick, and Ali Hortaçsu. 2004. Economic Insights from Internet Auctions. Journal of Economic Literature 42 (June): 457–486. Bajari, Patrick L., Robert S. McMillan, and Steve Tadelis. 2009. Auctions Versus Negotiations in Procurement: An Empirical Analysis. Journal of Law, Economics and Organizations 25 (2): 372–399.
200
Bibliography
Baldwin, Carliss Y., and Kim B. Clark. 2000. Design Rules. Vol. 1: The Power of Modularity. Cambridge, MA: MIT Press. Bandera, Oriana, Andrea Prat, and Tommaso Valletti. 2009. Active and Passive Waste in Government Spending: Evidence from a Policy Experiment. American Economic Review 99 (4): 1278–1308. Barbalet, Jack. 2009. A Characterization of Trust, and Its Consequences. Theory and Society 38 (4): 367–382. Bardhan, Pranab. 2002. Decentralization of Governance and Development. Journal of Economic Perspectives 16 (4): 185–205. Bardhan, Pranab, and Dilip Mookherjee. 2006. Decentralisation and Accountability in Infrastructure Delivery in Developing Countries. Economic Journal 116 (January): 101–127. Bar-Isaac, Hesi, and Steven Tadelis. 2008. Seller Reputation. Foundations and Trends in Microeconomics 4 (4): 273–351. Barthes, Roland. 1989. La Tour Eiffel. Paris: Editions du Seuil. Bassan, Bruno, Olivier Gossner, Marco Scarsini, and Shmuel Zamir. 2003. Positive Value of Information in Games. International Journal of Game Theory 32 (1): 17–31. Bateson, Patrick. 1988. The Biological Evolution of Cooperation and Trust. In Trust: Making and Breaking Cooperative Relations, edited by Diego Gambetta, 14–30. New York: Basil Blackwell. Bendix, Reinhard. 1977. Max Weber: An Intellectual Portrait. Berkeley: University of California Press. Bessen, Jim, and Michael J. Meurer. 2008. Patent Failure. Princeton, NJ: Princeton University Press. Blau, Peter M. 1956. Bureaucracy and Modern Society. New York: Random House. Bohte, John, and Kenneth J. Meier. 2000. Goal Displacement: Assessing the Motivation for Organizational Cheating. Public Administration Review 60 (2): 173–182. Bowyer, Adrian 2009. Video. www.youtube.com/watch?v=iMhG4fWQnlE. Boyne, George A., Kenneth J. Meier, Laurence J. O’Toole Jr., and Richard M. Walker. 2005. Where Next? Research Directions on Performance in Public Organizations. Journal of Public Administration Research and Theory 15: 633–639. Boyte, Harry C. 2005. Reframing Democracy: Governance, Civic Agency, and Politics. Public Administration Review 65 (5): 536–546. Brennan, Geoffrey, and James M. Buchanan. 1980. The Power to Tax: Analytical Foundations of a Fiscal Constitution. Cambridge, UK: Cambridge University Press.
Bibliography
201
Brodkin, Evelyn Z. 2006. Bureaucracy Redux: Management Reformism and the Welfare State. Journal of Public Administration Research and Theory 17: 1–17. ———. 2008. Accountability in Street-Level Organizations. International Journal of Public Administration 31 (3): 317–336 Brudney, Jeffrey L., and Robert E. England. 1982. Urban Policy Making and Subjective Service Evaluations: Are They Compatible? Public Administration Review 42 (2): 127–135. Brusoni, Stefano, Luigi Marengo, Andrea Prencipe, and Marco Valente. 2004. The Value and Costs of Modularity: A Cognitive Perspective. SPRU Working Paper 123. University of Sussex. Buelens, Marc, and Herman Van den Broeck. 2007. An Analysis of Differences in Work Motivation Between Public and Private Sector Organizations. Public Administration Review (January/February): 65–74 Cabral, Luís. 2005. The Economics of Reputation and Trust. Mimeo. New York University, Stern School of Business. Cabral, Luís, and Ali Hortaçsu. 2010. The Dynamics of Seller Reputation: Evidence from eBay. Journal of Industrial Economics (forthcoming). Caesar, Julius C. 1869. Caesar’s Gallic War. Translated by William A. McDevitte and William S. Bohn. New York: Harper & Brothers. ———. 1996. La guerra gallica. Edited and translated by Adriano Pennaccini. Includes notes by Albino Garzetti. Turin, Italy: Einaudi. Camerer, Colin F., and Ernst Fehr. 2006. When Does “Economic Man” Dominate Social Behaviour? Science 311 (5757): 47–52. Cameron, Averil. 1993. The Mediterranean World in Late Antiquity, AD 395– 600. London: Routledge. Caporale, Antonello. 2007. Lamezia, giudice manager all’aeroporto: Grandi profitti, ma lo vogliono cacciare. La Repubblica (March 6). www.repubblica .it/2006/a/rubriche/piccolaitalia/aeroporto-lamezia/aeroporto-lamezia.html. Castelfranchi, Cristiano, Rino Falcone, and Emiliano Lorini. 2009. A Nonreductionist Approach to Trust. In Computing with Social Trust, edited by Jennifer Golbeck, 45–72. London: Springer. Chernew, Michael, Gautam Gowrisankaran, and Dennis P. Scanlon. 2008. Learning and the Value of Information: Evidence from Health Plan Report Cards. Journal of Econometrics 144 (156): 174 Coleman, James S. 1994. Foundations of Social Theory. Cambridge, MA: Harvard University Press. Cremer, Jacques. 1986. Cooperation in Ongoing Organizations. Quarterly Journal of Economics 101 (1): 33–50.
202
Bibliography
Crowston, Kevin, Kangning Wei, Qing Li, Yeliz U. Eseryel, and James Howison. 2005. Coordination of Free/Libre Open Source Software Development. In Proceedings of the Twenty-Sixth International Conference on Information Systems. Atlanta: Association for Information Systems. Crozier, Michel. 1963. Le phénomène bureaucratique. Paris: Editions du Seuil. Dafny, Leemore, and David Dranove. 2008. Do Report Cards Tell Consumers Anything They Don’t Already Know? The Case of Medicare HMOs. RAND Journal of Economics 39 (3): 790–821. Dalle, Jean-Michel, Paul A. David, Rishab A. Ghosh, and W. E. Steinmuller. 2005. Advancing Economic Research on the Free and Open Source Software Mode of Production. In How Open Is the Future?, edited by Marleen Wynants and Jan Jan Cornelis, 395–428. Brussels: Brussels University Press. Dasgupta, Partha. 1988. Trust as a Commodity. In Trust: Making and Breaking Cooperative Relations, edited by Diego Gambetta, 49–72. New York: Basil Blackwell. David, Paul A. 2004. From Keeping “Nature’s Secrets” to the Institutionalization of “Open Science.” In CODE: Collaborative Ownership and the Digital Economy, edited by Rishab Aiyer Ghosh, 85–108. Cambridge, MA: MIT Press. Decreto Legislativo. 2006 (April 12). N. 163, “Codice dei contratti pubblici relativi a lavori, servizi e forniture in attuazione delle direttive 2004/17/CE e 2004/18/CE” (Law of public contracts on works, services and supplies, following the European directives 2004/17/CE and 2004/18/CE), Gazzetta Ufficiale 100, May 2, 2006, Supplemento Ordinario, n. 107. www.camera.it/ parlam/leggi/deleghe/testi/06163dl.htm. Della Porta, Donatella, and Alberto Vannucci. 1992. Un paese anormale. Come la classe politica ha perso l’occasione di Mani Pulite. Rome and Bari: Laterza. Dellarocas, Chrysanthos. 2003. The Digitalization of Word of Mouth: Promise and Challenge of Online Feedback Mechanisms. Management Science 49 (10): 1407–1424. Dewatripont, Mathias, Ian Jewitt, and Jean Tirole. 1999a. The Economics of Career Concerns. Part I: Comparing Information Structures. Review of Economic Studies 66 (1): 183–198. ———. 1999b. The Economics of Career Concerns. Part II: Application to Missions and Accountability of Government Agencies. Review of Economic Studies 66 (1): 199–217. Di Tella, Rafael, and Ernesto Schargrodsky. 2003. The Role of Wages and Auditing During a Crackdown on Corruption in the City of Buenos Aires. Journal of Law and Economics 46 (1): 269–292.
Bibliography
203
Dosi, Giovanni, Daniel Levinthal, and Luigi Marengo. 2002. The Uneasy Organizational Matching Between Distribution of Knowledge, Division of Labor and Incentive Governance. LEM Working Paper 2002/26. Scuola Superiore Sant’Anna, Pisa. Dranove, David, Daniel Kessler, Mark McClellan, and Mark Satterthwaite. 2003. Is More Information Better? The Effects of “Report Cards” on Health Care Providers. Journal of Political Economy 111 (3): 555–588. Dranove, David, and Ginger Zhe Jin. 2010. Quality Disclosure and Certification: Theory and Practice. Journal of Economic Literature (forthcoming). Dranove, David, and Andrew Sfekas. 2008. Start Spreading the News: A Structural Estimate of the Effects of New York Hospital Report Cards. Journal of Health Economics 27 (5): 1201–1207. Dunleavy, Patrick, Helen Margetts, Simon Bastow, and Jane Tinkler. 2006. New Public Management Is Dead—Long Live Digital-Era Governance. Journal of Public Administration Research and Theory 16: 467–494. Dunn, John. 1988. Trust and Political Agency. In Trust: Making and Breaking Cooperative Relations, edited by Diego Gambetta. New York: Basil Blackwell. eBay. 2009. eBay Marketplaces Fast Facts at-a-Glance (as of September 30, 2009). http://ebayinkblog.com/wp-content/uploads/2009/10/eBay-Marketplaces-FastFacts-Q309.pdf. ———. 2010 (January 20). Fourth Quarter and Full Year 2009 Results. http://files .shareholder.com/downloads/ebay/859964021x0x345224/b455630d-4bb94ba5-adb1-40dcf29e82ce/eBay_Q409EarningsRelease.pdf. Ethiraj, Sendil K., and Daniel Levinthal. 2004. Modularity and Innovation in Complex Systems. Management Science 50 (2): 159–173. Fama, Eugene F. 1980. Agency Problems and the Theory of the Firm. Journal of Political Economy 88 (2): 288–307. Feddersen, Timothy J. 2004. Rational Choice Theory and the Paradox of Not Voting. Journal of Economic Perspectives 18 (1): 99–112. Fernández-Villaverde, Jesús. 2008. Horizons of Understanding: A Review of Ray Fair’s “Estimating How the Macroeconomy Works.” Journal of Economic Literature 46 (3): 685–703. Fingar, Thomas. 2008. Intelligence Reform: Rush Transcript. Council of Foreign Relations. www.cfr.org/publication/15754/intelligence_reform_rush_tran script_federal_news_service.html. Fung, Archon, Mary Graham, and David Weil. 2007. Full Disclosure: The Perils and Promise of Transparency. Cambridge, UK: Cambridge University. Gambetta, Diego. 1988. Can We Trust Trust? In Trust: Making and Breaking Cooperative Relations, edited by Diego Gambetta. New York: Basil Blackwell.
204
Bibliography
Gentzkow, Matthew, and Jesse M. Shapiro. 2009. Competition and Truth in the Market for News. Journal of Economic Perspectives 22 (2): 133–154. Golden, Miriam, and Lucio Picci. 2005. Proposal for a New Measure of corruption Illustrated Using Italian Data. Economics and Politics 17 (1): 37–75. ———. 2006. Corruption and the Management of Public Works in Italy. In Handbook of Economic Corruption, edited by Susan Rose-Ackerman. Cheltenham, UK: Edward Elgar. ———. 2008. Pork Barrel Politics in Postwar Italy, 1953–1994. American Political Science Review 52 (2): 268–289. Granovetter, Mark. 1973. The Strength of Weak Ties. American Journal of Sociology 78 (6): 1360–1380. Gregory, Paul R. 2004. The Political Economy of Stalinism. Cambridge, UK: Cambridge University Press. Gregory, Robert. 1995. Accountability, Responsibility, and Corruption: Managing the Public Production Process. In The State Under Contract, edited by Jonathan Boston. Wellington, NZ: Bridget Williams Books. Greif, Avner. 1989. Reputation and Coalitions in Medieval Trade: Evidence on the Maghribi Traders. Journal of Economic History 49 (4): 857–882. ———. 1993. Contract Enforceability and Economic Institutions in Early Trade: The Maghribi Traders’ Coalition. American Economic Review 83 (3): 525–548. Greif, Avner, Paul Milgrom, and Barry R. Weingast. 1994. Coordination, Commitment, and Enforcement: The Case of the Merchant Guild. Journal of Political Economy 102 (4): 745–776. Gruen, Nicholas. 2002. Renovating the Agenda: With an Example from the Market for Information. Australian Journal of Public Administrations 61 (2): 90–105. ———. 2008. Outline of Gruen Tenders. Mimeo. Lateral Economics, Melbourne, Australia. ———. 2009. Improving Information Flows for Users of Post-Secondary Education. Mimeo. Lateral Economics, Melbourne, Australia. Gupta, Sanjeev, Luiz R. de Mello, and Raju Sharan. 2001. Corruption and Military Spending. European Journal of Political Economy 17 (4): 749–777. Hamilton, Alexander, John Jay, and James Madison. 2008. The Federalist Papers. Charleston, SC: Forgotten Books. Han Chun, Young, and Hal G. Rainey. 2005. Goal Ambiguity and Organizational Performance in U.S. Federal Agencies. Journal of Public Administration Research and Theory 15: 529–557. Hardin, Russell. 1988. Morality Within the Limits of Reason. Chicago: University of Chicago Press.
Bibliography
205
———. 2002. Trust and Trustworthiness. New York: Russell Sage Foundation. Harris, William V. 2001. Restraining Rage: The Ideology of Anger Control in Classical Antiquity. Cambridge, MA: Harvard University Press. Heinrich, Carolyn J. 2002. Outcomes-Based Performance Management in the Public Sector: Implications for Government Accountability and Effectiveness. Public Administration Review 62 (6): 712–724. Henrich, Joseph, Robert Boyd, Samuel Bowles, Colin Camerer, Ernst Fehr, Herbert Gintis, and Richard McElreath. 2001. In Search of Homo Economicus: Behavioral Experiments in 15 Small-Scale Societies. American Economic Review 91 (2): 73–78. Hibbing, John R., and Elizabeth Theiss-Morse. 2002. Stealth Democracy: American’s Beliefs About How Government Should Work. Cambridge, UK: Cambridge University Press. Hicks, John R. 1937. Mr. Keynes and the Classics: A Suggested Simplification. Econometrica 5 (2): 147–159. Hirschman, Albert O. 1970. Exit, Voice, Loyalty. Cambridge, MA: Harvard University Press. Hirshleifer, Jack. 1971. The Private and Social Value of Information and the Reward to Inventive Activity. American Economic Review 61 (4): 561–574. Ho, Alfred T. 2007. Engaging Citizens in Measuring and Reporting Community Conditions: A Manager’s Guide. Report of the IBM Center for the Business of Government. Washington, DC. Holmström, Bengt. 1982. Managerial Incentive Problems: A Dynamic Perspective. In Essays in Economics and Management in Honor of Lars Wahlbeck, 177–208. Helsinki: Swedish School of Economics. Hood, Christopher, and Guy Peters. 2004. The Middle Aging of New Public Management: Into the Age of Paradox? Journal of Public Administration Research and Theory 14 (3): 267–282. Hughes, Owen. 2008. What Is, or Was, New Public Management? Paper presented at the International Research Society for Public Management Conference (IRSPM12), Brisbane, Australia, March 26–28. Imbens, Guido W., and Jeffrey M. Wooldridge. 2009. Recent Developments in the Econometrics of Program Evaluation. Journal of Economic Literature 47 (1): 5–86. Jackson, Joab. 2009. Intellipedia Suffers Midlife Crisis. Government Computer News 18 (February). http://gcn.com/articles/2009/02/18/intellipedia.aspx. Jefferson, Thomas. 1855. The Writings of Thomas Jefferson: Being His Autobiography, Correspondence, Reports, Messages, Addresses, and Other Writings, Official and Private. New York: Riker, Thorme & Co.
206
Bibliography
Jin, Ginger Zhin, and Phillip Leslie. 2003. The Effect of Information on Product Quality: Evidence from Restaurant Hygiene Grade Cards. Quarterly Journal of Economics 118 (2): 409–451. Jones, Lawrence R., and Fred Thompson. 2007. From Bureaucracy to Hyperarchy in Netcentric and Quick Learning Organizations. Charlotte, NC: Information Age Publishing. Jøsang, Audun, Roslan Ismail, and Colin Boyd. 2007. A Survey of Trust and Reputation Systems for Online Service Provision. Decision Support Systems 43 (2): 618–644. Kassel, David S. 2008. Performance, Accountability, and the Debate Over Rules. Public Administration Review (March/April): 241–252. Kaufmann, Daniel, Aart Kraay, and Pablo Zoido-Lobatón. 1999. Aggregating Governance Indicators. World Bank Policy Research Working Paper 2195. Washington, DC: The World Bank. Kelly, Janet M., and David Swindell. 2002. A Multiple-Indicator Approach to Municipal Service Evaluation: Correlating Performance Measurement and Citizen Satisfaction Across Jurisdictions. Public Administration Review 62 (5): 610–620. Kelman, Steven. 2002. Remaking Federal Procurement. Public Contracts Law Journal (Summer). ———. 2005. Unleashing Change: A Study of Organizational Renewal in Government. Washington, DC: Brookings Institution Press. Keynes, John Maynard. 1973. The General Theory of Employment, Interest and Money In The Collected Writings of John Maynard Keynes, Vol. 8. Cambridge, UK: Macmillan and Cambridge University Press. Koopmans, Tjalling C. 1947. Measurement Without Theory. Review of Economics and Statistics 29 (3): 161–172. Kreps, David M. 1990. A Course in Microeconomic Theory. Princeton, NJ: Princeton University Press. Kreps, David M., Paul Milgrom, John Roberts, and Robert Wilson. 1982. Rational Cooperation in the Finitely Repeated Prisoner’s Dilemma. Journal of Economic Theory 27 (2): 245–252. Krueger, Anne O. 1974. The Political Economy of the Rent-Seeking Society. American Economic Review 64 (3): 291–303. Lambsdorff, Johann Graff. 2003. Framework document 2003. Background paper to the 2003 Corruption Perceptions Index. Mimeo. www.transparency .org/cpi/2003/dnld/framework.pdf. ———. 2006. Causes and Consequences of Corruption: What Do We Know from a Cross-Section of Countries? In International Handbook of Economic Corruption, edited by Susan Rose-Ackerman. Cheltenham, UK: Edward Elgar.
Bibliography
207
Lerner, Josh, and Jean Tirole. 2002. Some Simple Economics of Open Source. Journal of Industrial Economics 50 (2): 197–234. ———. 2005. The Economics of Technology Sharing: Open Source and Beyond. Journal of Economic Perspectives 19 (2): 99–120. Light, Paul C. 2006. The Tides of Reform Revisited: Patterns in Making Government Work, 1945–2002. Public Administration Review January/February: 6–19. Lijphart, Arend. 1997. Unequal Participation: Democracy’s Unresolved Dilemma. American Political Science Review 91 (1): 1–14. Lindblom, Charles E. 1959. Muddling Through. Public Administration Review 19 (2): 79–88. Lopez, Robert Sabatino. 1976. The Commercial Revolution of the Middle Ages, 950–1350. Cambridge, UK: Cambridge University Press. Luttwak, Edward N. 1976. The Grand Strategy of the Roman Empire. Baltimore: Johns Hopkins University Press. ———. 2007. Dead End: Counterinsurgency Warfare as Military Malpractice. Harper’s, February. www.harpers.org/DeadEnd.html. MacLeod, Bentley W. 2007. Reputation, Relationships, and Contract Enforcement. Journal of Economic Literature 45 (3): 595–628. Manzoni, Alessandro. 1909. I promessi sposi (The Betrothed). New York: Collier & Son. March, James G., and Johan P. Olsen. 1989. Rediscovering Institutions: The Organizational Basis of Politics. New York: Free Press. Marengo, Luigi, and Giovanni Dosi. 2005. Division of Labor, Organizational Coordination and Market Mechanisms in Collective Problem Solving. Journal of Economic Behavior and Organization 58 (2): 303–326. Margolis, Michael. 2005. E-Government and Democracy. In The Oxford Handbook of Political Behaviour, edited by Russell J. Dalton and Hans-Dieter Klingemann. Oxford, UK: Oxford University Press. Marx, Gary T. 2001. Identity and Anonymity: Some Conceptual Distinctions and Issues for Research. In Documenting Individual Identity, edited by Jane Caplan and John C. Torpey. Princeton, NJ: Princeton University Press. Maskin, Eric. S. 2008. Mechanism Design: How to Implement Social Goals. American Economic Review 98 (3): 567–76. Matsusaka, John G. 2005. Direct Democracy Works. Journal of Economic Perspectives 19 (2): 185–206. Maupassant, Guy de. 1890. La vie errante. Project Gutenberg. www.gutenberg. org/etext/14793. Mauro, Paolo. 1995. Corruption and Growth. Quarterly Journal of Economics 110 (3): 681–712.
208
Bibliography
Mazzetti, Mark. 2007. Intelligence Chief Announces Renewed Plan for Overhaul. New York Times, April 12, www.nytimes.com/2007/04/12/washington/12intel .html. McChesney, Robert W. 2008. The Political Economy of Media. New York: Monthly Review Press. Meyer, John W., and Brian Rowan. 1977. Institutionalized Organizations: Formal Structure as Myth and Ceremony. American Journal of Sociology 83 (2): 340–363. Miller, Nolan, Paul Resnick, and Richard Zeckhauser. 2005. Eliciting Informative Feedback: The Peer-Prediction Method. Management Science 51 (9): 1359–1373. Moore, Mark H. 1995. Creating Public Value: Strategic Management in Government. Cambridge, MA: Harvard University Press. Morgan, Gary. 2006. Images of Organizations. 2d ed. Newbury Park, CA: Sage. Morgan, Mary. 1990. The History of Econometric Ideas. Cambridge, UK: Cambridge University Press. Moynihan, Donald P., and Alasdair S. Roberts. 2002. Public Service Reform and the New Security Agenda. In Governance and Public Security, edited by Donald P. Moynihan and Alasdair S. Roberts. Syracuse, NY: Campbell Public Affairs Institute, Syracuse University. National Commission on Terrorist Attacks upon the United States. 2004. The 9/11 Commission Report: Final report of the National Commission on Terrorist Attacks Upon the United States. New York: W.W. Norton. Newton, Kenneth. 2007. Social and political trust. In The Oxford Handbook of Political Behaviour, edited by Dalton J. Russell and Hans-Dieter Klingemann. London: Oxford University Press. Nichols, John, and Robert W. McChesney. 2009. The Death and Life of Great American Newspapers. The Nation, March 18. www.thenation.com/doc/ 20090406/nichols_mcchesney. Nicholson-Crotty, Sean, Nick A. Theobald, and Jill Nicholson-Crotty. 2006. Disparate Measures: Public Managers and Performance-Measurement Strategies. Public Administration Review (January/February): 101–113. North, Douglas. 1991. Institutions. Journal of Economic Perspectives 5 (1): 97–112. Office for Government-wide Policy. 1997. White Paper: Past Performance. www .acquisition.gov/sevensteps/library/ASIwp-past-perform.pdf. Office of Federal Procurement Policy. 2000. Best Practices for Collecting and Using Current and Past Performance Information. www.whitehouse.gov/ omb/best_practice_re_past_perf.
Bibliography
209
Olken, Benjamin A. 2007. Monitoring Corruption: Evidence from a Field Experiment in Indonesia. Journal of Political Economy 115 (2): 200–249. ———. 2010. Direct Democracy and Local Public Goods: Evidence from a Field Experiment in Indonesia. American Political Science Review (forthcoming). Olsen, Johan P. 2005. Maybe It Is Time to Rediscover Bureaucracy. Journal of Public Administration Research and Theory 16: 1–24. ———. 2008. The Ups and Downs of Bureaucratic Organizations. Annual Review of Political Science 11: 13–37. Olson, Mancur. 1965. The Logic of Collective Action. Cambridge, MA: Harvard University Press. Olstrom, Elinor, and James Walker. 2003a. Introduction. In Trust and Reciprocity: Interdisciplinary Lessons from Experimental Research, edited by Elinor Olstrom and James Walker. New York: Russell Sage Foundation. ———. 2003b. Toward a Behavioral Theory Linking Trust, Reciprocity, and Reputation. In Trust and Reciprocity: Interdisciplinary Lessons from Experimental Research, edited by Elinor Olstrom and James Walker. New York: Russell Sage Foundation. Osimo, David. 2008. Web 2.0 in Government: Why and How? Luxembourg: Office for Official Publications of the European Communities. ftp://ftp.jrc.es/ pub/EURdoc/JRC45269.pdf. Pennac, Daniel. 1992. Comme un Roman. Paris: Gallimard. Perry, James L., Debra Mesch, and Laurie Paarlberg. 2006. Motivating Employees in a New Governance Era: The Performance Paradigm Revisited. Public Administration Review (July/August): 505–514 Petrillo, Joseph J. 2005. Using Past Performance to Steer Contract Awards. Government Computing News 24. http://gcn.com/articles/2005/06/29/anotherview-using-past-performance-to-steer-contract-awards.aspx?sc_lang=en. Picci, Lucio. 2006a. The Quantitative Evaluation of the Economic Impact of e-Government: A Structural Modelling Approach. Information Economics and Policy 18: 107–123. ———. 2006b. Misurare l’e-government: Il caso italiano. Rassegna Astrid 2. www.astrid -online .it/ E -governme/ Contributi1/ L _Picci -Misurare -l -egov _il caso-it-3.pdf. ———. 2006c. Reputation-Based Governance of Public Works. Rivista di Politica Economica (January-February): 161–183. Prencipe, Andrea. 1997. Technological Competencies and Product’s Evolutionary Dynamics: A Case Study from the Aero-engine Industry. Research Policy 25 (8): 1261–1276.
210
Bibliography
Project Management Institute. 2005. A Guide to the Project Management Body of Knowledge. 3d ed. Newtown Square, PA: Project Management Institute. Raymond, Eric S. 2000. The Cathedral and the Bazaar: Musing on Linux and Open Source by an Accidental Anarchist. Sebastopol, CA: O’Reilly Media. Reschenthaler, Gil, and Fred Thompson. 1996. The Information Revolution and the New Public Management. Journal of Public Administration Research and Theory 6: 125–143. Resnick, Paul, Richard Zeckhauser and John Swanson. 2006. The Value of Reputation on eBay: A Controlled Experiment. Experimental Economics 9 (2): 79–101. Rethemeyer, Karl R. 2006. Policymaking in the Age of Internet: Is the Internet Tending to Make Policy Networks More or Less Inclusive? Journal of Public Administration Research and Theory 17: 259–284. ———. 2007. The Empires Strikes Back: Is the Internet Corporatizing Rather than democratizing policy processes? Public Administration Review March/ April: 199–215. Rob, Rafael, and Arthur Fishman. 2005. Is Bigger Better? Customer Base Expansion Through Word-of-Mouth Reputation. Journal of Political Economy 113 (5): 1146–1162. Robinson, David, Harlan Yu, William Zeller, and Edward W. Felten. 2008. Government Data and the Invisible Hand. Yale Journal of Law and Technology 11: 160–175. Robinson, James A., and Ragnar Torvik. 2005. White Elephants. Journal of Public Economics 89 (2–3): 197–210. Rose-Ackerman, Susan. 1978. Corruption: A Study in Political Economy. New York: Academic Press. ———. 1999. Corruption and Government: Causes, Consequences, and Reform. Cambridge, UK: Cambridge University Press ———. 2004. The Challenge of Poor Governance and Corruption. Copenhagen Consensus Challenge Paper. www.copenhagenconsensus.com/Files/Filer/CC/ Papers/Governance_and_Corruption_300404_%286.4MB_version%29.pdf. Rosenbloom, David. 2008. The Politics-Administration Dichotomy in U.S. Historical Context. Public Administration Review (January/February): 57–60 Ruscio, Kenneth P. 1996. Trust, Democracy, and Public Management: A Theoretical Argument. Journal of Public Administration Research and Theory 6 (3): 461–477. Santos, Boaventura de Sousa. 1998. Participatory Budgeting in Porto Alegre: Toward a Re-distributive Democracy. Politics and Society 26 (4): 461–510. Scott, James C. 1998. Seeing Like a State. New Haven, CT: Yale University Press.
Bibliography
211
Shah, Somali K. 2005. Open Beyond Software. In Open Source 2, edited by Danese Cooper, Chris Di Bona, and Mark Stone. Sebastopol, CA: O’Reilly Media. Shingler, John, Mollie E. Van Loon, Theodore R. Alter, and Jeffrey C. Bridger. 2008. The Importance of Subjective Data for Public Agency Performance Evaluation. Public Administration Review (November/December): 1101–1111. Shleifer, Andrei, and Robert W. Vishny. 1993. Corruption. Quarterly Journal of Economics 108 (3): 599–617. Svensson, Jacob. 2005. Eight Questions About Corruption. Journal of Economic Perspectives 19 (3): 19–42. Tacitus. 1877. The Agricola and Germany of Tacitus: And the Dialogue on Oratory. Translated by Alfred J. Church and William J. Brodribb. New York: Macmillan. Tadelis, Steven. 2002. The Market for Reputation as an Incentive Mechanism. Journal of Political Economy 110 (4): 854–882. Taylor, Jeannette. 2009. Strengthening the Link Between Performance Measurement and Decision Making. Public Administration 87 (4): 853–871. Thompson, Clive. 2006. Open-Source Spying. New York Times, December 3, www .nytimes.com/2006/12/03/magazine/03intelligence.html. Tiebout, Charles M. 1956. A Pure Theory Of Local Expenditures. Journal of Political Economy 64 (5): 416–424. Transparency International. 1995. Press release, July 15, 1995. www.transpar ency.org/content/download/2915/18031/file/cpi1995.pdf. Treisman, Daniel. 2007. What Have We Learned About the Causes of Corruption from Ten Years of Cross-National Empirical Research? Annual Review of Political Science 10: 211–244. Trisciuoglio, Andrea. 1998. Sarta tecta, ultrotributa, opus publicum faciendum locare : sugli appalti relativi alle opere pubbliche nell’età repubblicana e augustea. Naples: Jovene. U.S. Congress. 1994. Federal Acquisition Streamlining Act. Washington, DC. U.S. Government. 2005. Federal Acquisition Rules. Washington, DC. http://thomas .loc.gov/cgi-bin/query/z?c103:S1587.ENR. U.S. Public Law No. 109-282. 2006. http://thomas.loc.gov/cgi-bin/bdquery/z?d109: SN02590:@@@D&summ2=m&. Van Dooren, Wouter. 2009. A Politico-Administrative Agenda for Progress in Social Measurement: Reforming the Calculation of Government’s Contribution to GDP. Journal of Comparative Policy Analysis 11 (3): 309–323. Varian, Hal. 2002. Avoiding the Pitfalls When Economics Shifts from Science to Engineering. New York Times, August 29. www.nytimes.com/2002/08/29/business/ economic-scene-avoiding-pitfalls-when-economics-shifts-science-engineering
212
Bibliography
.html ?scp = 1 & sq = Avoiding %20the %20pitfalls %20when %20economics %20shifts%20from%20science%20to%20engineering&st=cse. Von Hippel, Eric. 2005. Democratizing Innovation. Cambridge, MA: MIT Press. Von Hippel, Eric, and Joereng P. J. de Jong. 2010. Open, Distributed and UserCentered: Towards a Paradigm Shift in Innovation Policy. Scales Research Reports 2010. Zotermeer, the Netherlands. www.ondernemerschap.nl/pdf-ez/ H201009.pdf. Wallgren, Anders, and Britt Wallgren. 2007. Register-Based Statistics: Administrative Data for Statistical Purposes. Hoboken, NJ: Wiley. Warren, Mark E. 2004. What Does Corruption Mean in a Democracy? American Journal of Political Science 48 (2): 328–343. Wasserman, Stanley, and Katherine Faust. 1994. Social Network Analysis: Methods and Applications. Cambridge, UK: Cambridge University Press. Waters, Richard. 2008. Obama Weighs Putting the Wisdom of Crowds to Work. Financial Times, December 8. http://us.ft.com/ftgateway/superpage.ft?news_ id=fto120820081548186785&page=2. Weinstock, Daniel. 1999. Building Trust in Divided Societies. Journal of Political Philosophy 7 (3): 287–307. Whiteley, Paul F. 1995. Rational Choice and Political Participation: Evaluating the Debate. Political Research Quarterly 48 (1): 211–233. Williamson, Oliver E. 1993. Calculativeness, Trust, and Economic Organization. Journal of Law and Economics 36 (April): 453–486. Willinsky, John. 2005. The Unacknowledged Convergence of Open Source, Open Access, and Open Science. First Monday 10 (8) (August 1). http://firstmonday .org/htbin/cgiwrap/bin/ojs/index.php/fm/article/view/1265/1185. Wilson, James Q. 1989. Bureaucracy: What Government Agencies Do and Why They Do It. New York: Basic Books. Wooldridge, Jeffrey. 2001. Econometric Analysis of Cross Section and Panel Data. Cambridge, MA: MIT Press. Wright, Bradley E. 2007. Public Service and Motivation: Does Mission Matter? Public Administration Review (January/February): 54–64. Yang, Kaifeng, and Marc Holzer. 2006. The Performance-Trust Link: Implications for Performance Measurement. Public Administration Review (January/ February): 114–126.
Index
Accessibility, of modeling tools, 109 Accountability: democratic participation and, 171; Internet-based information systems and, 53–54; participatory budgeting and, 170; preventing corruption and, 115; public administrations and, 63–64; public policy management and, 110, 111 Actors, in public governance, 58, 65–68, 68–69; applications of reputation-based governance and, 124; incentives for, 132; privacy and, 178. See also Democratic participation; Public administrations Ad hoc statistics, 95–98, 119 Administrative data, reputational assessment systems and, 95–96 Adverse selection, 38, 87 Age, of reputational data, 94 Agenda setting, 165, 197ch10n3 Aggregate assessments, 74–75, 77, 89–91, 106 Agreements, enforcing. See Contracts
Aligned incentives. See Encapsulated interests Altruism, 8, 33–34 Ambitious policies, 144–145, 147–148, 149–155 Analytical methods, statistical information and, 103–106 Anticorruption policies, 115–116 Applicability, of reputation-based governance systems, 123–124 Arbitrage, 88 Architecture, modular design and, 113–114 Assessments: beauty contest reputation and, 89–91; collaborative working environments and, 66–67; econometric techniques and, 103; of public administrations, 57–58; public works functions and, 129–132; simplification and, 64, 65. See also Reputational assessment systems Assessment unit, 69, 70, 74–75 Asymmetric information, trust and, 37 Audits, 136–137, 139
214
Index
Automated statistical systems, 95–98 Availability, of data, 103–104 Bayes’s theorem, 186n9 Beauty contest reputation, 77–78, 89–91, 174, 191n6, 192n8 Beneficial actions, trust and, 34–35 Beneficiaries, of public policies, 99 Betrothed, The (Manzoni), 62 Bidding, procurement functions and, 125 Bilateral reputation mechanisms, 41 Boldness, public policies and, 144–145, 147–148, 149–155 Boycotts, 40–41 Brand names, trading, 87 Brazil, participatory budgeting and, 166–171 Brothers Karamazov, The (Dostoyevsky), 25–26, 42 Budgeting, participatory, 166–171, 172–174 Bureaucracies, 28, 48–49, 188n2. See also Public administrations Business intelligence, 97 Business reputations, 1–5 Calculative trust, 25, 34–36 Capital goods, reputation as, 5–6 Career concerns: information garbling and, 192n1; Internet-based information systems and, 53–54; public administrations and, 45, 134; public officials and, 49–53; reputational budgets and, 86; reputation-based governance and, 121 Cathedral and the Bazaar, The (Raymond), 8 Censorship, 90 Cheating, prisoner’s dilemma game and, 22–25
Civil registries, 72–73, 74, 190n17, 190n18 Civil society organizations, 167–168, 171, 173 Civil unrest, risk of, 161, 162, 171 Clientelism, 170–171 Coding systems, reputational assessment systems and, 111 Collaborative working environments: democratic participation and, 172, 173; national security and, 11–12; open-source design and, 7–9, 181; Peer-to-Patent program and, 9–11; public administrations and, 66 Collares, Alceu, 167–168 Collective action, 32, 48, 164–165 Collusion, 68, 135 Combined indicators, of performance, 137–139 Common objectives, contributions to, 50 Commonsense interpretations, trusting organizations and, 29–30 Communication, effective, 25, 26 Communications technology, 3, 182 Compensation, 8, 10–11, 49, 89, 138 Competition, reputation and, 6–7 Computers, econometric techniques and, 104, 105 Connecting relationships, 59–60. See also Relationships, among public policies Consequential relationships, 60 Constituency unit, 69, 70, 71–73 Constituencies, of public policies, 67–68 Continuous functioning, reputational assessment systems and, 101–102 Contract administration, 133–134 Contractor Performance System, 132 Contractors: applications of reputation-based governance and, 124, 127–129, 132; public
Index
administrations and, 66; public officials and, 128–129; reputational assessment systems and, 75 Contractor selection processes, 58 Contracts: business relationships and, 7; career concerns and, 52; contract administration and, 133–134; “incomplete contracts” and, 12–15; prisoner’s dilemma game and, 22, 23; public procurement functions and, 126 Controversial policies, reputational assessment systems and, 72 “Cookie specs,” 127 Cooperative behaviors, 33, 34 Corporate leadership, 50 Corruption: career concerns and, 52; economic damages and, 194n5; measurement of, 194–195n11; public governance and, 47; public officials and, 188n4; public policy management and, 110, 115–119, 121–122; public policy selection and, 146–147; public procurement functions and, 125; reputational points and, 89 Corruption indexes, 118–119 Cost-benefit analysis, 149–155, 156–157 Costs: assessing corruption and, 121–122; public policy selection and, 145–146; public services and, 98–99, 136, 179–180; statistical information and, 98, 99 Counterparts, learning about, 21, 36–38, 42 Criminal convictions, for corruption, 117 Cyberdemocracy, 185n12 Data analysis, 103–106, 121 DATA.gov, 192–193n4
215
Data-mining, statistical information and, 97 Decentralized governance, 55 Decoupling problems, 112, 113–114, 115, 194n3 Democratic criteria, public governance and, 47, 48 Democratic participation: beauty contest reputation and, 90; bold public policies and, 158; participatory budgeting and, 166–171; process of, 161–166; public administrations and, 46, 47; public policies and, 187n1; reputation-based governance and, 16, 120, 121, 144, 160–161, 171–175, 176; statistics and, 100; technology and, 18–19 Design processes, 113–114, 180–181 Dictatorships, 161–162 Digital information systems, 95–99, 130–131. See also Internet-based information systems Direct democracy, 166 Disclosure policies, 93–94 Discount rates, folk theorem and, 24 Dostoyevsky, Fyodor, 25–26, 42 Dynamic effects, of reputation, 5, 6–7, 38 EBay, 1–3, 26, 94 Econometric techniques, 103–104, 193n6, 193n8 Economic incentives, 21 Efficiency, 46–47, 48–49, 54, 116–117 E-government, 18–19, 106–109 Eiffel Tower, 147–148, 154–155 Elected officials, 155–156, 165–166, 169–170. See also Public officials Elections, democratic, 9, 168 Elites, governance and, 55, 56, 135 Empirical regularity, trust and, 27, 29, 31–32, 42
216
Index
Employees, of organizations, 27, 49–53 Employment, in public administrations, 133–134 Encapsulated interests, 25–26, 26–28, 30, 31, 42, 186n6 Expectations, of behavior, 26–27, 28–31, 32, 35 Experience-based measures, of corruption, 117 Experience goods, purchasing decisions and, 12
Governments: local governments, 55, 135–136; non-democratic, 161–162; transparency and, 63 Granularity, of policy objects, 112–113, 115
Failure of policies, possibility of, 149–155 Federal Acquisition Reform Act (1995), 128 Federal Acquisition Streamlining Act (1994), 128 Federal government, public procurement functions and, 127–129 Federal Tort Claims Act (1948), 129 Feedback: eBay and, 1–2; positive feedback loops, 157–158, 173 Financing, types of, 188n7 Fingar, Charles, 11–12 Folk theorem, 21, 23–25, 32–33 Functions and units, reputational assessment systems and, 68–76 Fuzzy missions, 53
Identification, of assessors, 72–73 Implementation issues, reputationbased governance and, 135–137, 176–177 Incentives: applications of reputationbased governance and, 135; assessing public works and, 131; encapsulated interests and, 25–26, 26–28, 30, 31, 42, 186n6; monitoring performance and, 157; organizational capability and, 155–157; perverse incentives, 137–138, 140, 141, 174, 190n12, 195–196n7; in public administrations, 48–53, 132; public policies and, 144–145, 148–155; punishments and, 195n4; trust and, 34, 35; word-of-mouth, 38–39 “Incomplete contracts,” 12–15 “Index of reputation,” 1–2 Individual assessments, beauty contest reputation and, 89–91 Inefficiency and ineffectiveness, corruption and, 116–117 Informal procedures, democratic participation and, 164 Information, availability of: assessing corruption and, 121–122; asymmetric information and, 37; career concerns and, 51–52, 192n1;
Game theory, 21, 22–26, 42, 185n3, 186n5, 186n8, 186n12 General Theory (Keynes), 104 Global Corruption Barometer, 118 Goal ambiguity, 188n5 Goal displacement, 138, 189n9 Governance. See Public governance Government actions, democratic participation and, 163–166, 172, 173
Hansa, the, 41 Health care, report cards and scorecards, 139–140 HMO report cards, 140 Honesty, prisoner’s dilemma game and, 22–23
Index
open-source design and, 192n2; participatory budgeting and, 170; public administrations and, 55, 56; reputational incentives and, 41; reputation-based governance and, 176, 177; reputation unit and, 69; word-of-mouth, 38–39. See also Internet-based information systems Information systems: assessing public works and, 136–137; Internet and, 5, 15–16; project management and, 193n1; public policies and, 171; reputation and, 3; statistical information and, 92–93; trust and, 36 Infrastructure: applications of reputation-based governance and, 136; assessing corruption and, 119; e-government and, 107, 108; private corporations and, 195n2; public procurement functions and, 129–132 Institutionalization, statistical information and, 97–98, 117, 119 Institutional setting, official statistics and, 100–103 Institutions of governance, 17–18, 39–41, 163 Integral approaches, to design, 114 Integrated statistics, 95–98, 98–99, 117 Intellipedia, 11–12, 184n8 Internet-based information systems: analytical methods, 103–106; e-government and, 106–109; information organization and, 15–16; participatory budgeting and, 170; privacy and, 190n16; production of, 98–99; public procurement functions and, 132; publisher unit, 93–95; reputational incentives and, 18, 53–54, 55; reputation-based governance and,
217
56, 57, 176, 177, 182; statistical information and, 95–98, 100–103; trust and, 36; word-of-mouth and, 4–5 Investigative journalism, 120–121 Investing in reputation, 36–38, 43 Investments, 194n10 IS-LM model, 104 Italian Ministry of Economy and Finance, 13–15 Italy: assessing corruption and, 118, 119, 120; public procurement functions and, 184n9; referenda and initiatives, 163; trusting organizations and, 29–31 Itanium integrated circuits (Intel), 114 Judicial assessments, corruption and, 117 Julius Caesar, 20, 43–44, 185n1, 187n17 Kant, Immanuel, 27 Kelman, Steven, 127, 133–134 Keynes, John Maynard, 104, 191n6 Koopmans, Tjalling, 193n6 Legal remedies, 2, 4, 7, 14–15, 39 Legibility, of policy landscapes, 62–63, 65 Local governments, 55, 135–136 Long-term gains, 23–24, 25–26 Madison, James, 28 Maghribi traders, 3–5, 39, 186n12 Management tasks, contract administration and, 133 Manzoni, Alessandro, 62 Market competition, public services and, 179–180 Maupassant, Guy de, 147–148 Merchant guilds, 40–41, 187n14 Merit pay, 49, 138
218
Index
Methodology, 16–19 Middle Ages, 3–5, 21, 38–41 Mixed direct-representative democracy, 166 Mixed strategies, reputation and, 44 Modularization, of public policy management, 111–115, 194n4 Monitoring performance, 36, 157 Monopolies, 27–28 Moral hazards, 33, 87 Morality, 33–34, 46–47 Moreno, Jacob Levi, 60 Nash equilibrium, 22–23, 185n3 National Association for Business Economics (NABE), 192n3 National governments, 55 National Institutes of Health, Contractor Performance System, 132 National Intelligence Council, 11–12 National security, 11–12 National Statistics Offices (NSOs), 96, 99, 100, 101–102 Negative actions, trust and, 35 Negative feedback, eBay and, 2 Nesting relationships, 60, 61 Netherlands, democratic participation and, 163–164 New Public Management, 63, 184n7; administrative reforms and, 64; applications of reputation-based governance and, 143; assessing performance and, 138; integrated statistics and, 99; private corporations and, 124 News media, 117–118, 120–121, 131, 195n13 New Zealand, administrative reforms and, 63, 64 Nongovernmental organizations, 67–68, 190n14 Normative analysis, 16, 177
North, Douglas, 17 Novelty, patents and, 10 Number of assessments, limits on, 72 Obama, Barack, 9 Objective measures: of corruption, 117, 119; of performance, 137–138, 141, 143 Office of Federal Procurement Policy (OFPP), 127–129 Official statistics, 96, 100–103 Open access, to statistical information, 102–103 Open bidding, 125, 126 Open-source design, 180–182, 192n2, 197ch11n4 Open-source software, 7–9, 180, 183n3, 183–184n4 Optimism factors, 141–142 Organizational capability, 155–157, 157–158 Organizations: actors in, 66; governance and, 17; reputational assessment systems and, 67–68; reputational incentives and, 18; trust and, 20–21, 26–33, 42–43 Ostrom, Elinor, 138 Outcomes: applications of reputationbased governance and, 143; cost-benefit analysis and, 149–155; democratic participation and, 173–174; participatory budgeting and, 169; project management and, 59; public works functions and, 130; reputational assessment systems and, 75 Outsourcing, 124 Over-regulation, corruption and, 115 Palácio do Alvorada, Brazil, 179 Paradox of voting, 73–74 Parent policies, 60
Index
Participation costs, 73–74, 173, 174–175 Participation rates, reputational assessment systems and, 73–74 Participatory budgeting, 165, 166–171, 172 Partido Democratico Social (PTS), Brazil, 167 Partido Trabalhista Brasilero (PTB), Brazil, 167 Past performance: assessing, 195–196n7; career concerns and, 51; folk theorem and, 24–25; organizational capability and, 155–156; public procurement functions and, 126–127, 127–128, 184n9; purchasing decisions and, 15; reputational budgets and, 85–86; service providers and, 142; trust and, 26, 42 Past Performance Information Retrieval System (PPIRS), 132 Patent inflation, 9–10 Patent system, 9–11, 184n6 Patronage, 116, 135 Payoff matrix, prisoner’s dilemma game and, 22, 23 Peer-to-Patent program, 9–11 Performance measures: applications of reputation-based governance and, 124, 125, 137–139; career concerns and, 49–50, 51–52; integrated statistics and, 99; public administrations and, 64, 133, 134; public procurement functions and, 128–129; reputational assessment systems and, 75 Personal accountability, 110, 111, 115 Perverse incentives, 137–138, 140, 141, 174, 190n12, 195–196n7 Policy assessments, 67–68, 106 Policy design, reputation-based governance and, 180–182
219
Policy execution processes, 58, 59 Policygrams, 61–62 Policy landscapes, 59–63 Policy objects, 110, 112–113, 181–182 Policy selection processes, 58, 145 Policy unit, 69, 70–71, 123 Political appointees, 66–67 Political parties, 167 Pork barrel politics, 123, 135, 137, 196 Porto Alegre, Brazil, 160, 165, 166–171 Positive analysis, 16, 177 Positive feedback loops, 157–158 Postcontractual agreements, 12 Post-New Public Management, 189n5 Preference elicitation process, 163–166, 172, 173 Price premiums, 2, 13 “Prior art,” patents and, 10 Prisoner’s dilemma game, 22–24, 42, 191n2 Privacy: Internet-based information systems and, 190n16; of performance appraisals, 129; publisher unit and, 94; reputational assessment systems and, 72–73, 74–75, 93; reputation-based governance and, 177–178 Private corporations, 190n11; applications of reputation-based governance and, 124; career concerns and, 52; efficiency and, 48–49; employee incentives and, 50; procurement functions and, 125–127; public administrations and, 191n1; reputation indexes and, 80. See also Contractors Private negotiations, procurement functions and, 125, 126
220
Index
Procedures, for reputation-based governance: democratic participation and, 160–166, 171–175; participatory budgeting and, 166–171 Procurement functions, 125. See also Public procurement functions Production, modes of, 181 Production, of statistical information, 95–96, 98–99, 101 Production policies, public administrations and, 64, 65 Professionalization, of government statistics, 100 Professional services, 124–125, 141–143 Profit motives, private firms and, 124 Programs, public governance and, 59–61, 136 Project completion, 145–146 Project management: contract administration and, 134; information systems and, 193n1; modularization, of public policy management and, 111–113; outcomes and, 59; public policies and, 193–194n2; public policy management and, 110 Projects, public governance and, 59–61 Property rights, 39, 48–49 Proprietary software development, 7, 8 Public administrations, 189n8, 191n1; accountability and, 63–64; actors in, 66; assessments and, 57–58, 85; bold public policies and, 149; democratic participation and, 166; employment in, 133–134; incentives and, 17–18, 19, 45, 48–53; investigative journalism and, 120–121; organizational capability and, 155–157; participatory
budgeting and, 168–169, 170; privacy and, 178; privatization of public services and, 179–180; public policy selection and, 145; secrecy and, 62–63; trust and, 21, 26–28. See also Public officials Public administrators, 66, 67, 75, 120, 132 Public availability, of reputational assessments, 93 Public demonstrations, democratic participation and, 160, 165 Public governance: actors in, 65–68; encapsulated interests and, 28; functions and units of, 68–76; incentives and, 9–11, 48–53, 53–56; public policies and, 59–63, 146–148; quality of, 46–48; reputation and, 2, 45–46; simplification and, 63–65; structure of, 57–59 Public health, 139 Public input, 130, 132 Public officials: acting on public opinion and, 163–164; applications of reputation-based governance and, 124; assessing public works and, 132; career concerns and, 49–53; contractors and, 66, 67, 128–129; corruption and, 188n4; participatory budgeting and, 169; privacy and, 178–179; reputational assessment systems and, 75, 120 Public opinion: assessing performance and, 137–138; corruption indexes and, 118–119; democratic participation and, 162–163, 164, 174 Public policies: analytical methods and, 103–106; assessments and, 58; boldness of, 158–159; democratic participation and, 165–166; econometric techniques and, 103–106; incentives and, 148–155;
Index
management of, 110–122, 193–194n2; organizational capability and, 155–157; policy landscapes and, 68–69; public governance and, 59–63, 146–148; selection processes and, 144–146, 157–159; separating overlapping effects of, 107–109, 112 Public policy management, 110, 111–115, 115–119, 120–122, 193–194n2 Public procurement functions: applications of reputation-based governance and, 124, 125–127; public governance and, 58, 59, 184n9; reforms and, 127–129; reputational budgets and, 85; reputational incentives and, 13–15 Public services, market competition and, 179–180 Public works functions, 98–99, 124, 125, 129–132, 142–143, 195n2 Publisher unit, 70, 75–76, 93–95, 101, 123 Punishments: for cheating, 25, 34; for policy failures, 149, 150, 157, 195n4 Purchasing decisions, 12–15, 85. See also Public procurement functions Quality: of goods and services, 13, 15, 37, 42, 139–142, 187n16; of public governance, 46–48 Quantitative analysis, 103–106 Random shocks (luck), 50–51, 51–52, 53 Rankings. See Report cards and scorecards Rational choice approach, 16–17, 30–31, 33, 35–36, 184n10 Rationing, of assessment opportunities, 196n9
221
Raymond, Eric, 8 Rebuttals, of assessments, 73, 129 Recruitment, public administrations and, 133 Red tape, 188n2 Redundancy, public administrations and, 54, 55 Referenda and initiatives, 160, 162, 163–164, 197ch10n2 Regulation, 180 Re-inventing government reforms, 127 Relationships, among public policies, 59–63, 106–109, 111–113, 188n1 Rent-seeking activities, 6, 55–56, 110, 115–117, 170–171 Report cards and scorecards, 125, 139–142, 178; assessing public works and, 131 RepRap 3D printer, 181 Representative democracy, 161, 166 Reputation: information organization and, 3, 15–16; public governance and, 45–46; strategic behaviors, to demonstrate quality and, 37–38; trust and, 20–26, 41–44 Reputational assessment systems: actors in, 65–68; beauty contest reputation, 89–91; combined indicators of performance and, 138–139; democratic participation and, 173–174; functions and units of, 68–76; investigative journalism and, 120–121; local governments and, 135–136; perverse incentives and, 195–196n7; reputational budgets, 77, 80–89, 132; reputation indexes, 78–80 Reputational budgets, 77, 80–89, 132 Reputational equilibrium, 4 Reputational incentives, 5–7; availability of information and, 41; bold public policies and, 148; contract administration and, 134;
222
Index
Reputational incentives (continued) open-source software and, 7–9; procurement functions and, 12–15, 126–127; public administrations and, 45; public governance and, 9–11, 47–48, 53–56; self-interest and, 16–17; service providers and, 141–142; trading of reputational points and, 87 Reputational points, 80–81, 84, 85, 86–88 Reputational points markets, 85, 86–88, 191n4 Reputational scores, computing, 75, 77–91, 192n9 Reputation and contract administration, 133–134 Reputation-based governance: applications of, 123–125, 142–143; contractors and, 127–129; democratic participation and, 144, 160–166, 171–175; implementation issues and, 135–137, 176–177; performance measures and, 137–139; policy design and, 180–182; privatization of public services and, 180; professional services and, 139–142; public procurement functions and, 125–127; public works functions and, 129–132; reputation and contract administration, 133–134 Reputation indexes, 77, 78–80 Reputation unit, 69, 70, 75, 90, 192n7 Residual claimants, 48–49 Resource allocations, 108–109, 135–136 Responsiveness, of government, 138–139 Restaurant health inspection scores, 139
Risks, assuming, 32, 148–149, 149–155, 156–157, 158 Roman Empire, 100, 182 Scorecards. See Report cards and scorecards Scott, James C., 62, 188–189n2 Secrecy: public administrations and, 62–63; trading of reputational points and, 87–88 Selection processes, 6–7, 37–38, 144–146, 146–147, 157–159 Self-interest: career concerns and, 45; of organizational employees, 27; public administrations and, 146; public officials and, 49–53; reputational incentives and, 16–17; trust and, 25, 33–34 Service economy, measurements of, 99 Service ethics, 50 Service providers, report cards and scorecards, 141 Short-term gains, 23–24 Simplification, public governance and, 63–65 Size, of firms, 80–81, 86 Social context, 105–106, 117–118 Socioeconomic development, 167 Sociograms, 60–61 Stability: project design processes and, 114–115; of statistical information, 101 Stakeholders, 57, 59 Stalin, Joseph, 158–159, 161–162, 197ch10n3 Standardization, of statistical information, 102 State powers, 39, 40–41, 62–63, 188–189n2 Static effects, of reputation, 5–6, 38 Statistical information: ad hoc versus integrated statistics, 95–98; analytical methods, 103–106,
Index
223
193n8; assessing corruption and, 117–119; assessing public works and, 136–137; e-government and, 106–109; official statistics, 100–103; production of, 98–99; public works functions and, 130–131; publisher unit and, 93–95; reputational assessment and, 92–93 Strategic behaviors, to demonstrate quality, 37–38 Structural models: of e-government, 107–109; of public governance, 57–59, 63–65 Subconstituencies, reputational assessment systems and, 71–72 Subjective measures, 137, 138–139, 143 Summaries, of data, 93–94, 94–95, 131 Sunk investments, 145–146 Supply and demand, public administrations and, 54 Surveys, statistical data and, 96
Time inconsistency, trust and, 40 Time lags, statistical data and, 96 Tourism example, trusting organizations and, 29–31 Trade, 39–41 Transaction costs, 7, 68 Transparency, 63, 71, 115, 121, 179 Transparency International Corruption Perception Index, 118 Treatment effects, econometric techniques and, 103 Trust: eBay and, 1–3; game theory and, 186n5, 186n8; investing in reputation and, 36–38; the Middle Ages and, 38–41; reputation and, 20–26, 41–44; self-interest and, 33–34; theories of, 34–36; trusting organizations, 26–33 Two-period model, career concerns and, 50–52
Tadelis, Steven, 87 Taxation, 46 Taxonomies, 129, 130 Teacher ratings, 178, 197ch11n2 “Teaching-to-the-test,” 64 Technical specifications, public procurement functions and, 127 Technological coloration, 184n5 Tenders, public. See Public procurement functions Terror-based rule, 161 Theory-based approach, reputational assessment systems and, 107–109 Time-discount factors, 78, 80, 81, 84
Vitale, Giuseppe, 110, 120, 195n12 Voice unit, 69, 70, 74–75 Voting, 73–74, 190n18
U.S. Patent and Trademark Office (USPTO), 9–10
Watchdog groups, 67–68 Weber, Max, 63 Weighted assessments, 80–81, 89–91, 174, 192n7 Wilson, James, 48–49, 52 Word-of-mouth, 4–5, 38–39 Workloads, public administrations and, 54 Work packages, 110, 111, 112–113 World Bank, 55–56