Atkinson, A. B. (Anthony Barnes), 1944– Lectures on public economics / Anthony B. Atkinson, Joseph E. Stiglitz ; introduction by Anthony B. Atkinson ; introduction by Joseph E. Stiglitz. pages cm “Originally published in 1980 by McGraw-Hill”—Title page verso. Includes bibliographical references and index. ISBN 978-0-691-16641-4 (hardback) 1. Finance, Public. 2. Welfare economics. I. Stiglitz, Joseph E. II. Title. HJ141.A74 2015 336—dc23 2014044336 British Library Cataloging-in-Publication Data is available This book has been composed in Minion and Myriad Printed on acid-free paper. ∞ Printed in the United States of America 10 9 8 7 6 5 4 3 2 1
For Siobhan, Michael, Richard, Sarah, Charles, and Edward
Introductionxi Prefacexxvii Introductory Note to the 1980 Edition
P A R T O N E T H E A N A LY S I S O F P O L I C Y Lecture 1 Introduction: Public Economics 1–1 Introduction 1–2 Role of the Government 1–3 Guide to the Lectures Note: The Public Sector—Statistical Background
3 3 4
Lecture 2 Household Decisions, Income Taxation, and Labour Supply 19 2–1 Introduction 19 2–2 Income Taxation and Labour Supply 26 2–3 Broader Models of Labour Supply 36 2–4 Empirical Evidence on Labour Supply 40 2–5 Concluding Comments 47 Note on the Expenditure Function 48 Reading50 Lecture 3 Taxation, Savings, and Decisions over Time 51 3–1 Intertemporal Decisions and Taxation 51 3–2 The Basic Intertemporal Model 56 3–3 Developments of the Model and Alternative Views 66 3–4 Empirical Evidence—Taxation and the Interest Elasticity of Savings 74 3–5 Concluding Comments 77 Reading78 Lecture 4 Taxation and Risk-Taking 79 4–1 Risk-Taking and Portfolio Allocation 79 4–2 Effects of Taxation 85 4–3 Special Provisions of the Tax System 91 4–4 Generalization of Results 96 4–5 Concluding Comments 102 Note on Risk Aversion 103 Reading104 Lecture 5 Taxation and the Firm 5–1 Taxes and the Firm 5–2 Corporation Tax and the Cost of Capital 5–3 Taxation and Investment
105 105 108 117
C ontents 5–4 A Wider View of Investment 123 5–5 Empirical Investigation of Taxation and Investment 127 5–6 Concluding Comments 130 Reading131 Lecture 6 Tax Incidence: Simple Competitive Equilibrium Model 132 6–1 Introduction: Tax Incidence 132 6–2 Static Two-Sector Model 136 6–3 Incidence of Corporation Tax 142 6–4 General Tax Incidence 147 6–5 Incidence in a Two-Class Economy 155 6–6 Numerical Applications of the Model 159 6–7 Concluding Comments 163 Note on the Cost Function 164 Reading165 Lecture 7 Tax Incidence: Departures from the Standard Model 166 7–1 Introduction 166 7–2 Market Imperfections 167 7–3 Monopolistic Competition 172 7–4 Structure of Production 180 7–5 Non-Market-Clearing 184 7–6 Concluding Comments 188 Reading188 Lecture 8 Taxation and Debt in a Growing Economy 189 8–1 Introduction 189 8–2 An Aggregate Model of Equilibrium Growth 191 8–3 Growth and Taxation 197 8–4 Taxation in a Life-Cycle Model 202 8–5 Burden of the National Debt 208 8–6 Concluding Comments 214 Reading215 Lecture 9 Distributional Effect of Taxation and Public Expenditure 216 9–1 Taxation, Spending, and Redistribution 216 222 9–2 Modelling the Distribution of Income 9–3 Distributional Incidence 231 9–4 Empirical Studies of the Redistributive Impact of the Government Budget235 9–5 Concluding Comments 244 Reading245 Lecture 10 Theories of the State and Public Economics 246 10–1 Introduction 246 10–2 Voting and Decisions 250 10–3 Administration and Bureaucracies 259 10–4 Power, Interest Groups, and Marxist Theories 263 10–5 Empirical Studies of Public Expenditure 268 Reading274
P art T w o T h e D es i g n o f P o l i c y Lecture 11 Introduction to Part Two: Normative Analysis 277 11–1 Introduction 277 11–2 Normative Theories of the State 279 11–3 Pareto Efficiency and Welfare Economics 285 11–4 Standard Public Finance Objectives 291 11–5 Range of Government Instruments 296 Note on the Measurement of Income Inequality 302 Reading303 Lecture 12 The Structure of Indirect Taxation 304 12–1 Introduction 304 12–2 The Ramsey Tax Problem 308 313 12–3 Application of the Ramsey Results 12–4 Partial Welfare Improvements and Tax Reform 318 12–5 Optimal Taxation in a Many-Person Economy 322 12–6 Concluding Comments 327 Reading327 Lecture 13 The Structure of Income Taxation 328 13–1 Introduction 328 13–2 A Simple Model 331 13–3 Linear Income Tax 338 13–4 General Income Tax 343 13–5 Concluding Comments 352 Reading353 Lecture 14 A More General Treatment of the Optimal Tax Problem 354 14–1 Introduction 354 14–2 Indirect Taxes and Linear Direct Taxation 357 14–3 Nonlinear Tax Schedules and Tax Exemptions 364 14–4 Taxation of Savings 370 14–5 Externalities in Consumption and Corrective Taxes 378 14–6 Concluding Comments 381 Reading382 383 Lecture 15 Public Sector Pricing and Production 15–1 Introduction 383 15–2 Departures from Marginal Cost Pricing 386 15–3 Choice of Technique and Production Efficiency 394 15–4 Cost–Benefit Analysis and Social Rate of Discount 398 15–5 Concluding Comments 402 Reading403 Lecture 16 Public Goods and Publicly Provided Private Goods 404 16–1 Introduction 404 16–2 Optimum Provision of Pure Public Goods—Efficiency 408 16–3 Optimum Provision of Pure Public Goods—Distribution 415 16–4 Publicly Provided Private Goods 417 16–5 Equilibrium Levels of Public Expenditure 424 16–6 Revelation of Preferences 430 Reading434
C ontents Lecture 17 Local Public Goods 435 17–1 Introduction 435 17–2 Optimum Provision of Local Public Goods 437 17–3 Market Equilibria and Optimality: Identical Individuals 446 17–4 Market Equilibria and Optimality: Heterogeneous Individuals 452 17–5 Fiscal Federalism 461 Reading465 Lecture 18 Public Economics: Theory and Policy 18–1 On the Sources of Disagreement in Policy Analysis 18–2 Thinking about Policy: Taxation 18–3 Thinking about Policy: Expenditures 18–4 Policy Reform and Political Economy
466 466 470 476 479
Bibliography481 Author Index
Public economics is an exciting topic. It deals with key issues facing governments all round the world. The subject matter—taxes, government spending, and the national debt—is very much in the minds of citizens and their elected representatives. What is more, it draws on the whole range of economics. To understand public policy, it is necessary to consider how households and firms make decisions, how they interact in a market economy, and how the economy develops over time. Excitement about the subject led us to write Lectures on Public Economics in the 1970s (the book was first published in 1980). Our aim was to bring together the recent developments in the subject, which had given it a central place in the modern microeconomics of general equilibrium and in the theory of second-best welfare economics. We believed this repositioning to be important. Those who had early contributed to such developments, such as Paul Samuelson, Leif Johansen, Peter Diamond, and James Mirrlees, were economists with wide interests, as were we ourselves. Public economics was very much part of the economics mainstream. Our book was directed therefore at graduate students in economics as a whole, and not only as a specialist treatise in public finance. It was—and remains—our view that all professional economists should understand the “grammar of arguments about policy”. Would we write the book differently today? Yes. But it would mostly involve addition, not subtraction. The subject has made significant progress. In this Introduction, we identify some of the areas where in recent decades there have been major advances. There are equally areas where the field of public economics, and economics more generally, is in need of new thinking. In highlighting these areas, we are of course being selective, providing hints of some of the topics that we might have covered had we ever managed to write a second volume of our Lectures. There are many other aspects that should be discussed. It is certainly not possible for us to sur vey the entire rich literature that has appeared in the third of a century since the book was written. We refer the reader to the Handbook of Public Economics, of which to date five volumes have appeared since 1985, edited by Alan Auerbach and Martin Feldstein (now joined by Raj Chetty and Emmanuel Saez). In the field of taxation, extensive coverage is provided by the background papers for the Mirrlees Review of taxation in the United Kingdom (Mirrlees, 2010) and by Robin Boadway (2012).
International and Global Public Economics Reading Lectures on Public Economics in the twenty-first century, one is immediately struck by the absence (noted on page 4) of any discussion of the international dimension of public finance. If we were writing the book today, this would be different. Indeed, one of us (JES) has written extensively on globalization, and the other (ABA) is writing a book on global public economics. Fiscal policy cannot be analysed simply in terms of a nation-state in a closed economy. This is true of tax policy where the threat of tax competition limits what national governments can achieve. The firm
Introduction analysed in Lecture 5 has to consider choices not just about investment and its financing but also about the location of production and indeed of its headquarters. National governments in turn have to allow for these reactions when designing tax policy. Per sonal taxpayers make choices regarding location and fiscal domicile. Migration decisions are influenced by taxation and by government spending policy. National government conduct of fiscal policy now has to be set in a context of international competition and international co-operation. What is more, it is evident that there are global problems that require a global response, such as the trading im balances, the funding of development, and the challenge of climate change. This means that we have to consider the public finances of global policy ambitions—such as the Sustainable Development Goals—and the design of policy by the responsible international organizations. Where there are global public goods to be financed, there has to be agreement on burden-sharing. We should emphasize, however, that the intellectual frameworks developed in our Lectures still provide the basis of much of the analyses of these questions. The theory of local public goods (Lecture 17), for instance, described the limiting case of a world in which factors of production were freely mobile, though the benefits of public expenditures were limited to those living in the community. It was a straightforward matter to go from national public goods (à la Samuelson) to local public goods, and from there to global public goods—goods (such as knowledge, or those that affect the global environment, global security, or global health) that affect the well-being of everyone in the world (Stiglitz, 1995, 1999a). Similar issues are raised in the ongoing discussions concerning fiscal federalism and decentralization, which have become the center of key political debates. There are important political economy questions (see the discussion below): are lower levels of government actually more responsive to their citizens? To what extent are the consequences of the decisions made by one political entity borne by others? With the extension to global public economics come new challenges, particularly to the normative branch of public economics. On what basis should we evaluate policy? Presumably we would not simply carry over a purely national evaluation of the costs and benefits. But do we move to a fully fledged worldview—global cosmopolitanism? This would mean giving equal weight to all citizens of the world. For many people that would be a step too far, not least because of the limited sphere of control of national governments. The challenge is illustrated by the UK Treasury document setting out the basis on which public investments should be assessed: “all impacts (including costs and benefits, both direct and indirect) on non-UK residents and firms should be identified and quantified separately”. But it goes on to say that “generally, proposals should not proceed if, despite a net benefit overall, there is a net cost to the UK” (HM Treasury, 2011, 21n). This appears to be adopting a global perspective but rejecting a global evaluation. In reality, national governments attach some weight to the well-being of those outside their borders but stop a long way short of giving equal weight. The implications of such an intermediate position need to be considered as part of global welfare economics (Atkinson, 2014).
Behavioural Public Finance If we were rewriting the book, we would certainly pay more attention to the issues raised in the recent literature on behavioural public economics. Behavioural economics provides a broader view of economic decision-making than the standard micro-
economics that largely underlay Lectures on Public Economics. Integrating insights from psychology, sociology, and neuroscience, behavioural models provide a richer account of the responses of households and firms to public policy. We now recognize that there are many arenas in which an individual’s behaviour is not well-described as if they maximized a well-defined utility function. This is especially true when it comes to making decisions under uncertainty. The probability judgements may not correspond to any reasonable estimate of the likelihood of the occurrence of the relevant events. Decisions can be affected by how the choices are posed: framing matters a great deal. One individual’s preferences may be affected by those of another. Again, this poses challenges for descriptive economics, and even more for normative economics. Even if an individual’s behaviour seems irrational (from the perspective of the standard utility maximizing model), it can still be predictably irrational (to use the term of Ariely, 2008). That is the whole thrust of behavioural economics. And because there are systematic aspects to these behavioural responses, governments need to take them into account in formulating public policy. Thus while behavioural economics provides a strong cautionary note against using the old models to analyse the consequences of certain policy changes, it also provides new instruments for government intervention. Standard theory, for instance, suggests that it should make little difference to consumption if, when the government lowers taxes, it simultaneously lowers the withholding rate. Individuals facing the same budget constraint behave the same way. But behavioural economists have suggested that if individuals see more money in their bank account or on their take-home pay slips, they will spend more. People would be “nudged” to spend (Sunstein and Thaler, 2008). (These ideas actually influenced the design of the US tax cut in 2009.) The importance of these issues came to the fore well before behavioural economics gained the prominence that it has today. In the United States, individuals could put up to $2,000 a year into tax-exempt “IRA” accounts (where interest was exempt while it was being earned). For higher-income individuals, saving more than $2,000 a year, there was no marginal incentive, and standard theory thus predicted there would be only an income effect. Since the individual lifetime budget constraint had shifted out, consumption should increase—just the opposite of the intended effect. But the evidence was that the IRA accounts “worked”—people did save more. The explanation was simple: banks could “sell” savings. Thus, not only may the impact of a public measure be markedly different from that predicted by the standard model, but behavioural economics suggests new ways of attaining public purposes—for instance, increasing savings—by changing the “frames” in which these issues are viewed and by nudging. In some cases, behavioural economics represents only a mild change; it does not even necessarily imply the abandonment of maximization. As put by Matthew Rabin, “people have reasonably focused goals, and maximize these goals reasonably well” (2013, 538); the problem is that they make “mistakes” in the maximand and in the identification of the choice set faced. Rabin describes this as “quasi-maximization”, listing four common mistakes: narrow bracketing, where people maximize their true utility function but focus on only one aspect of the budget set; present bias, as with hyperbolic discounting; projection bias, where people mispredict the utility from future situations (for example, acting myopically); and incorrect expectations regard ing future events. But in other cases, the effects can be more profound, as the examples given earlier illustrate. This is also illustrated by the equivalence results for taxation that we
Introduction stressed—for example, in the discussion of taxation in general equilibrium in Section 6–4. If we simply look at the household budget constraint, then a proportional income tax paid by the worker may appear equivalent to a proportional payroll tax paid by the employer, but these may be perceived differently. As expressed by Peter Diamond, “in standard modelling, we assume consistent behaviour across economic environments, captured in preferences that are defined only in terms of commodities acquired (absent externalities). One of the key messages of behavioural economics is that context (also referred to as situation) matters in ways that are not recognized in standard modelling” (2008, 1859). For example, standard analysis assumes that (somehow) individuals are aware of the expected net burden of social security taxes— that is, the difference between the expected receipts and expected contributions—and it is only this difference that affects behaviour, with both income and substitution effects. However, there is little evidence that this is the case; the wider context matters. As another example, “framing” the requirement that individuals purchase health insurance as a mandate, subject to a fine if the individual does not purchase insurance, is equivalent to a tax imposed on individuals who do not purchase health insurance. But there can be behavioural consequences to framing this requirement as a mandate rather than as a tax. While early work in behavioural economics focused on bringing insights from psychology into the analysis (not only framing, but confirmatory and preconfirmatory bias), more recent work has sought to integrate insights from other social sciences. Behaviour is affected by norms, which can be culturally determined, even if they may be affected, in part, by economics. Norms help explain markedly different behaviour in societies with seemingly similar economic circumstances; differences in work effort between Americans and Europeans and among European countries have as much or more to do with norms and culture as with income and substitution effects.1 Again, this means both that conventional policies may have effects that are different from those predicted by the standard model and that there is a whole new set of mechanisms for changing individual behaviour and societal equilibrium. Affirmative action may succeed in changing norms and expectations in a far more powerful way than standard price interventions. Arguably, changes in views about the environment have contributed more to changes in recycling practice than have environmental taxes.2 From a normative standpoint, once individuals act as quasi-maximizers, or depart more radically from the standard paradigm, we have to re-examine the welfare criterion. There is a potential conflict between basing welfare judgements on the “true” utility function and basing them on the mistaken utility function employed by individuals. As it was put by Arthur Robson and Larry Samuelson, “experienced utilities are of no interest to a fiercely neoclassical economist—decision utilities suffice to describe behaviour. However, if we are to consider welfare questions the difference may be important. If experienced utilities do not match decision utilities, should we persevere with the standard economists’ presumption that decision utilities are an ap propriate guide to well-being? Alternatively, should we use [experienced utilities] to over-ride their decision utilities?” (2011, 312). Difficulties in answering this question have led to the issue being ducked. Raj Chetty and Amy Finkelstein note in their report on the NBER’s Program on Public Economics, “partly because welfare analysis This is a subject that has generated a huge literature. See, e.g., Stiglitz (2008). The 2015 World Development Report (World Bank, 2015) focuses not just on the importance of behavioural economics for developmental policy but on the ways in which public policy can alter norms. Some of these insights should be applicable to the economics of the public sector more generally. 1 2
in behavioral models is complex, much of the growth in the behavioral public economics literature has been in positive empirical work” (2012, 4). But the nettle has to be seized.3 In particular, there are disconcerting consequences to focusing on “decision utilities”, or, in the case of decision-making under uncertainty, on the basis of subjective probabilities that might consistently differ significantly from any meaningful “true” probabilities. For instance, reforms that open up the possibility of more extensive gambling between parties with different probability assessments will lead to an increase in ex ante utility for all parties. Indeed, government can create information asymmetries that generate huge bets and seemingly huge gains in expected utility. But such gambles are just zero sum; they lead to consumption volatility—in the extreme, they can lead ex post to individuals living in a state of deprivation (Guzman and Stiglitz, forthcoming). Looking back on their lives, individuals may be filled with regret. Surely these ex post states of mind have as much or more reality as the ex ante perceptions (see Brunnermeier and Parker, 2005).
Empirical Public Finance Part One of Lectures on Public Economics contains discussion of empirical evidence on taxation and labour supply, taxation and savings, and taxation and investment, and on the empirical implementation of computable general equilibrium models. This is an area where the reader needs to supplement our discussion, since there has since been an explosion of research. Three elements generated this explosion. The first is the widespread availability today of micro-data from household surveys and administrative records. It is hard now to imagine how limited were the data available to researchers in the past. In the United States, the 1967 Survey of Economic Inequality was much analysed because it was one of the few sources of individual data—relating to a small sample (3,203 male heads of families). It is not just that there is more data. Administrative records, for instance, enable us to have data about the top 1 per cent—information that it would be almost impossible to glean from standard surveys. The second element, induced by the availability of data, was the development of econometric techniques for the analysis of such data. Here there have been fruitful exchanges between those working on public economics and econometricians, covering such aspects as the treatment of complex budget constraints. The third element has been the renewal of interest by economists in experimental evidence, including both field and laboratory experiments. We say “renewal”, since we did devote 3½ pages of Lecture 2 to the negative income tax experiments launched in New Jersey and elsewhere in the United States in the 1960s. This new empirical work has not only provided new evidence concerning household behaviour, but also (and perhaps most controversially) about the behaviour of firms—how, for instance, investment, employment, and production decisions might be affected by public policies. A great deal has been learned. At the same time, there is much still to be done. This is well illustrated by the debate concerning the optimal top rate of income tax, which depends on the estimated elasticity of taxable income. As is observed in the report of the Mirrlees Review in the UK, “there is no escaping the uncertainty around the estimate” (Mirrlees, 2011, 109). The implied 95 per cent confidence interval meant that the optimal tax rate on the top range could be as For a recent attempt to do so, see Brunnermeier, Simsek, and Xiong (2014).
Introduction low as 24 per cent or as high as 62 per cent. The views of most politicians could be encompassed.4
Political Economy This brings us to the way in which policy is actually decided. Lecture 10 is concerned with positive theories of government, “seeking to ‘close’ the system, by incorporating explicitly the behaviour of the state” (page 246). We review voting theories, models of bureaucratic behaviour, and interest group models including class theories of the state. Our discussion drew on the public choice school, associated particularly with the work of James Buchanan and Gordon Tullock. In each of these areas there have been important advances, but it is also clear that there is much more to be done. At the time we wrote the Lectures several conundrums motivated our analysis. It was evident even then that the median voter model (with “rational” voters) provides a poor description of the political equilibrium.5 It was also evident that there were many Pareto-inferior policies. Why couldn’t political processes lead to reforms that would make everyone better off? How could we explain such seeming collective irrationality?6 But why the model performs so badly is not always so clear. In some cases, there is ample evidence of disparities between voters’ perceptions and “reality”: they underestimate, for instance, the extent of inequality in our societies. Behavioural economics will undoubtedly come to play an important role in our understanding of political economy. As we noted, the standard economic model has individuals with well-defined utility functions (preferences). Advertisers do more than just provide in formation: they attempt to shape preferences and beliefs. So too do political parties and politicians. Marketing research and behavioural economics have given us some insights into how and the extent to which this can be done. But these insights have yet to be widely integrated into political economy analysis. One topic that has come to play a central role in political economy in recent years is corruption. Imperfections of information (see below) are, of course, essential to understanding corruption: if corruption could be costlessly observed, presumably it would not occur. Corruption can, in part, be understood as a breakdown in wellfunctioning principal agent relationships, where one party is supposed to act on behalf of another. Public corruption is associated with a public official acting to serve his or her own self-interest, rather than the public that he or she is supposed to serve. Such failures to fulfill fiduciary duties also arise, of course, within the private sector. Some policies, it is argued, are better than others on the basis that they are more corruption resistant.7 Closely related is the literature on rent-seeking, surveyed recently by Ignacio Del Rosal (2011). We emphasize throughout the Lectures the distributive consequences of policies. The fact that different policies affect different groups differently is a core part 4 On the other hand, Diamond and Saez (2011) use the standard model to argue strongly that the top marginal tax rate should be 80 per cent (in the case of the United States, taking into account other taxes, the top federal income tax rate should be between 48 per cent and 76 per cent). 5 For a discussion of some of the recent advances in social choice theory, see Maskin and Sen (2014). 6 Since then, one of us (JES) has spent considerable time working in the US government and for the World Bank. This has only deepened these concerns. Stiglitz (1997, 1998a) provides some alternative interpretations of these experiences, focusing on how different policies affect the political dynamics through information flows and helping create bargaining coalitions. 7 There has been a particular focus on corruption-resistant tax structures in developing countries. See Stiglitz (2010).
of the “political economy”. Political leaders attempt to put forward proposals that can muster sufficient support, a winning coalition of beneficiaries of the policies. But political debates often also center around differences in perceptions concerning the effects of these policies. Evidently our theoretical and empirical understandings of the economy are sufficiently uncertain that economists disagree among themselves; even when there is widespread agreement among economists, politicians can attempt to persuade voters of an alternative reality. We noted earlier the important role of “culture”. Subgroups within a population can come to beliefs that are not based on evidence, exemplified in the United States by attitudes about climate change. The fact that government policies can affect the distribution of income has further consequences for the political equilibrium, if economic inequality translates into political inequality (as it typically does in practice). There can be an economic-cum-political equilibrium with a low level of inequality, with the political process leading to rules of the game (including tax laws) that support this low level of inequality; and there can be another equilibrium with a high level of inequality.8 An important strand in modern political economy focuses on commitment—on the realization that with democratic processes policies can change. While traditional discussions of tax policy contrasted temporary vs. permanent changes, there is in fact no such thing as a permanent policy change, and especially so in democracies, where elections are typically about changing policies. In a world with incomplete contracting (and imperfect information), the ability of governments to commit themselves to unchanging policies has important consequences. When firms make long-term investments, what matters is not just the tax rate today, but tax rates in the future. While the current government in a democracy cannot control what future governments might do, it can change the transactions costs and alter, in other ways, the consequences of a changed policy. In so doing, it can affect the likelihood that future governments will change their actions and alter the behaviour of market participants.9 Institutional arrangements can also affect perceptions about the government’s commitment.10 The dynamic interactions among parties may sometimes lead to surprising results. Consider a two-party model, with one party being more pro-business than the other. If the first party lowers the tax on dividends, market participants will know, should the other party come into office, that they will raise the tax. But that means that firms will distribute profits during the low-tax regime, and the lower level of retained earnings will result in less investment—firms may not have the requisite finance when a good project comes along. On the other hand, when the other party comes into office, dividend taxes are raised, and firms, knowing that taxes are likely to be subsequently reduced, are induced to retain earnings, so that when a good project comes along, they can make the investment. Thus, the level of investment will be higher in the high-tax regime than in the low-tax regime, and the pro-business party curtails the extent to which it cuts taxes, given its awareness of these effects (see Korinek and Stiglitz, 2009). There are important normative questions in the design of the optimal commitment policies on the part of government. In the macroeconomic literature, for 8 The multiplicity of political-cum-economic equilibria has been noted in other contexts. In the absence of a rule of law, those who take advantage of such situations thrive, and they support a political equilibrium with a weak rule of law. But there can be another equilibrium with a strong rule of law. See Hoff and Stiglitz (2004a, 2004b, 2008). 9 The general theory of equilibrium in contestable democracies is set forth in Korinek and Stiglitz (2008). 10 Thus, in monetary theory, some argue that independent central banks, dominated by those from financial markets, can serve as an important commitment device that the central bank will take an antiinflationary stance.
Introduction instance, policies with strong government commitments are often advocated within models that assume that the structure of the economy is known and not changing. But policies that work in one environment may be inappropriate in another. If a government’s commitments are too strong, it will not be able to adapt to the changing circumstances. Within the public sector, the design of the structure of decision-making affects the decisions being made and what the public sector does. There are two aspects of this. One, the study of bureaucracy, has traditionally been the provenance of public administration. And yet the constraints imposed on bureaucracy (e.g., concerning pro curement, hiring and dismissal of public employees, and so on) are among the reasons that government is often accused of being less efficient than the private sector. In assessing the relative merits of public and private production, account has to be taken of the reasons that such constraints are imposed on the public sector, and whether there are alternative rules that might provide the desired protections at less cost. The other relates to fiscal federalism, the level of government at which different decisions are made (see Oates, 1999, and Faguet and Pöschl, forthcoming). There has been a widespread movement to put government closer to citizens, in the hope that this will make government more accountable. But such efforts at decentralization have not been totally successful, and often for political economy reasons. In some cases, local government seems more controlled by local elites. The theory of local public goods (Lecture 17), moreover, explains how competition among communities may limit the degree of redistribution that can occur. These issues may be playing out in important ways in the European Union.
Leaving the World of Competitive General Equilibrium The analysis of public policy can be no better than the understanding of the economy on which the analysis is based, as we warned the reader in the Introduction to Lectures on Public Economics. We went on to emphasize that the model underlying much of the Lectures—and much of public economics—was the Arrow-Debreu model of competitive general equilibrium. Looking back a third of a century later, we are struck that little seems to have changed in this respect. Even then, we were stressing the breakdown of the assumption of perfect competition, the absence of markets, and the role of imperfect information (see Lecture 11). In an earnest of taking these seriously, we set out in Lecture 7 a model of general equilibrium tax incidence with monopolistic competition à la Dixit-Stiglitz. With some notable exceptions, this attempt at bridging the public economics and industrial organization literatures has not been followed—in contrast to international trade theory, where “the theory of monopolistic competition has had a huge impact on modern trade theory” (Neary, 2001, 2). More troubling, since Lectures was published there have been significant advances in the development of the basic theory of the economics of imperfect information, and even considerable empirical research; but many strands of this work have not been adequately incorporated into modern public economics. To mention but two examples: in the presence of information asymmetries and incomplete markets, there are pervasive pecuniary externalities, providing scope for corrective taxation.11 And markets where information asymmetries are important may not “clear”: there can See Greenwald and Stiglitz (1986), and Arnott and Stiglitz (1986).
be unemployment and credit rationing (Stiglitz and Weiss, 1981, and Shapiro and Stiglitz, 1984). If that is the case, among the first-order effects of any policy change (such as taxes) are those on the unemployment rate or the availability of capital. Macro economics has belatedly taken these ideas on board, with a major thrust focusing on macroeconomic externalities, including the role that corrective taxes can play in dealing with such externalities. For instance, Olivier Jeanne and Anton Korinek (2010) explain that when firms borrow in foreign exchange, they do not take into account the general equilibrium consequences that might arise in the case of an adverse shock, so there is excessive foreign exchange denominated indebtedness. They propose a corrective tax to deal with the externality. These and other models incorporating macroeconomic externalities, information imperfections, and other rigidities can help explain the equilibrium level of unemployment (or, more generally, the probability distribution of unemployment and the shortfall between actual and potential output), and tax policies can have significant effects on the equilibrium. If that is the case, then tax policies may not only have the microeconomic effects upon which public finance has focused, but also even more important macroeconomic impacts.12 Imperfect information helps explain why corporate governance is so important— and why the standard model of the firm (in which the firm maximizes its profits or stock market value) almost surely provides an inadequate theory of firm behaviour. An important part of traditional public finance analyses firm responses to taxes imposed on it (on the labor it hires, on its profits, and on the goods it sells), and if firms do not behave according to the standard competitive theory, our predictions of firms’ responses may be incorrect, and so accordingly will our analysis of the consequences of taxes.13 And this is even more the case if firms are credit or equity constrained.14 These are not just matters of theoretical niceties: they have important policy implications. If, for instance, firms are credit constrained, then average tax rates (which determine the amount of cash a firm has available to invest) matter as much as marginal tax rates. The standard theory cannot explain why firms pay dividends rather than buy back shares. But were firms to replace dividends with share buybacks on a more significant scale than is already the case, government revenue would be substantially reduced. Finally, imperfections of information play a central role in determining the set of feasible taxes: one can only impose taxes on things that can be observed; and as we pointed out in the Lectures, the set of feasible taxes are critical for understanding the structure of optimal taxes—even for determining whether it is desirable to have differential indirect taxation. 12 The latter may be an order of magnitude larger than the former—typically Harberger triangles are small relative to the losses associated with the economy operating for extended periods below full employment. 13 Before our book was published, there was a large literature on managerial capitalism (e.g., Marris, 1964). But this work was largely dismissed by the mainstream. With the advent of the economics of in formation, the extent of managerial discretion has increasingly become recognized, and there is a bur geoning literature attempting to take this into account in the analysis of firm behaviour. But the implications for public finance have yet to be fully explored. 14 Equity-constrained firms cannot raise additional equity, or can do so only at high costs, and they will act in a risk-averse manner (Greenwald and Stiglitz, 1990). While an earlier literature (Domar and Musgrave, 1944) attempted to take this into account—suggesting that a profits tax would lead to more risk-taking, since the government is a “silent partner” in the firm—most of the literature over the past 50 years has paid little attention, focusing more on risk-neutral firms in the context of well-functioning financial markets.
Inequality and Redistribution Central to Lectures on Public Economics is the distribution and redistribution of income. In the years following its publication this may have seemed strange to some readers, as macroeconomics took a turn towards the assumption of representative agents with identical endowments and tastes. Indeed, there were influential mainstream economists who claimed that distributional concerns had no place in “sound economics”. However, public economics restricted to efficiency considerations in a world of identical agents would fail to address the key questions of public policy. Distributional concerns are central to public economics. The disputed issues in public finance—those on which economists are asked to contribute to the public debate— arise because people have different capacities to pay and different preferences. These differences among people are of first-order importance. We cannot put it better than Robert Solow: “Heterogeneity is the essence of a modern economy. In real life we worry about the relations between managers and shareowners, between banks and their borrowers, between workers and employers, between venture capitalists and entrepreneurs, you name it. . . . We know for a fact that heterogeneous agents have different and sometimes conflicting goals, different information, different capacities to process it, different expectations, different beliefs about how the economy works. Representative-agent models exclude all this landscape” (2003, 1). The assumption that everyone is identical is a powerful simplification, and were it true, public finance might be a simple matter: just impose lump-sum taxes on everyone. In fact, the analysis of public policy within models where people differ is a challenge, as is illustrated by the discussion in Lecture 13 of the way in which James Mirrlees, and before him William Vickrey, tackled the design of the optimal income tax schedule where people differ in their earning power. The individual utility maximization problem has to be embedded within the social welfare maximization by the government. Since then, there has been an important generalization of the basic optimal tax model in allowing people to differ in more than one dimension. Here we may distinguish between multi-dimensional treatments of the joint distribution of endowments and tastes, on the one hand, and those concerned with the joint distribution of two different dimensions of endowments on the other. The former raises issues about the way in which taste differences enter the social welfare function.15 The latter is par ticularly relevant to recent developments in the explanation of inequality. For much of the past 30 years, attention has focused on the determinants of earned income, privileging such factors as skill-biased technological change. That research has highlighted that relative wages can be affected by government policy (both taxation and expenditures). Most of the earlier work (discussed in Lectures) assumed that the before-tax relative income of, say, high- and low-productivity workers is fixed. But there have been large changes in the relative wages of skilled and unskilled workers, and government policy can affect those wages. If the before-tax distribution is less unequal, less of a burden is imposed on distortionary redistributive policies. Tax policy needs to take this into account.16 However, we have both had a long-standing interest in the distribution of capital income, and this has now come to the forefront of public discussion, with wider recognition of the role played by differences in inherited wealth. This in turn raises interesting issues regarding how the design of taxation can affect inherited wealth. Much of the work in this area was anticipated by Mirrlees (1976). For an attempt to do so, see Stiglitz (1998b).
Thomas Piketty and Emmanuel Saez (2013), for example, have investigated the optimal taxation of inheritance in a model where people differ in earning capacity and in bequest preferences that translate into unequal endowments in subsequent generations. The bi-variate composition of income more generally introduces the question as to how the optimal design of taxation depends on the degree of correlation of different income sources. If we are moving from a situation where different people are at the top of the distributions of earnings and the distribution of capital income (a “class model”—see page 72) to a situation where it is the same people at the top of both distributions, does this mean that there is a stronger case for higher taxes at the top? Further analyses of the design of optimal taxation will require further developments in the theory of the distribution of income and wealth. For instance, even if the stochastic process describing the evolution of wages is the same for all families, and even if all families have the same preferences for bequests, there will be rich and poor families, depending on particular experiences (luck). Even if government only cared about the utility of dynastic families, it would want to engage in redistributive taxation (including through the imposition of inheritance taxes), and even more so if social welfare depends on the well-being of individuals, not just dynastic families.17 Moreover, individuals differ in their ability to obtain returns on capital. One might treat the superior returns to capital as really a return to labor—to the individual’s productive abilities in managing capital—but our tax laws do not, and for good reason: it is impossible to identify such returns, and to distinguish them from high returns that are, for instance, a result of risk-taking. One of the striking—and disquieting—results of Atkinson and Stiglitz (1976) using a standard model (with utility functions that were separable between leisure and goods) was that there should be no taxation of capital. But once one recognizes these differences in the ability to obtain returns on capital, this result no longer holds.18 (There are other reasons, related to the fact that the competitive model does not provide a good description of the economy, discussed in earlier sections of this Introduction. For instance, if some of what is recorded as a return to capital is in fact a monopoly rent, or even the return to land or some other inelastically supplied factor, it should be taxed—and indeed taxed at a very high rate.) We noted earlier that the structure of optimal taxes depends on the set of feasible taxes. So too the burden on taxes for redistribution depends on the extent to which other instruments—in particular, expenditure policy—can be used. Often expenditure policy is essential—not just the provision of education and health, but also the provision of public transportation, enabling poor individuals to have better access to jobs.19 This analysis is related to another strand of advances: putting the theory 17 The analysis of the equilibrium wealth and income distribution even in simple specifications is rather complicated. See, e.g., Stiglitz (1969); Bevan and Stiglitz (1979); Becker and Tomes (1986); Kotlikoff (1988). Note that in models with strong dynastic utility functions, those who are lucky in having high incomes share their wealth with future generations. The result is that high inheritance taxes (in the absence of offsetting progressive income taxes) may actually result in greater inequality of consumption/individual utilities (Stiglitz, 1976). But even in these circumstances, the societal consequences of inherited in equalities may make inheritance taxation desirable, suggesting limitations in the usual formulation of (even inequality-averse) social welfare functions. Other models of inheritance attribute differences in inheritance to the lack of efficient annuity problems (partially associated with problems of asymmetric information), the consequence of which is that some individuals inherit more wealth because their parents died earlier. See Stiglitz (1978b). 18 We were aware of these limitations on our theorem even at the time we published our original paper; we should perhaps have given them more attention. See, e.g., Stiglitz (1985). 19 For a discussion of some aspects of this interplay between redistributive taxes and government ex penditures, see Stiglitz (1998b).
introduction into a general equilibrium context. There are four ways of going about changing the final distribution of income and well-being: changing endowments, changing before tax-and-transfer prices/wages, engaging in progressive taxation and transfers, and engaging in distribution-sensitive public expenditures. Optimal tax-and-expenditure policy entails coordinating all the relevant instruments. The optimal design of taxation and spending may also change if the social welfare function ceases to be exclusively determined by individual utilities or if these utilities are replaced by other indicators of individual well-being. Such a movement has a long history in public economics. Richard Musgrave (1959) defined as “merit wants” those where the state decides that the consumption of certain goods is to be encouraged (e.g., education). Or it could be that the government would strike out from the individual utility functions the elements corresponding to demerit goods that should be discouraged (e.g., tobacco). On the positive side, governments may attach greater weight than individuals to expanding employment (see, for example, Kanbur, Keen, and Tuomala, 1994) or to global warming or to the promotion of social trust or altruism.20 Governments may care about income and wealth inequality, and not just because of the impact on an inequality-averse social welfare function. Democratic processes may not work well when a society is too divided.
Finally: The Over-Specialization of Economics As we emphasized at the outset, we believe that public economics should be situated in the wider context of economics as a whole. In our view there is a risk that economics is becoming too specialized. The depth of specialization today has brought impressive results, and we fully agree that people should establish their reputations as x-economists. But at the same time they should retain an appreciation of what is being achieved in other fields. In this respect, we belong to an older generation. One of us (JES) defies classification: he could be a microeconomist or a macroeconomist, or a labour economist, or an IO specialist, or a development economist. He has lectured in all these fields. The other (ABA) has, over a teaching career from 1967 to 2013, taught—apart from public economics—microeconomics, applied economics (including macro), economic statistics, comparative economic systems, health economics, labour economics, and the economics of inequality. We hope that the republication of our textbook will contribute to a wider knowledge of public economics. A B Atkinson Nuffield College, Oxford, INET at the Oxford Martin School, and London School of Economics J E Stiglitz Columbia University
20 There are externalities associated, say, with global warming, and the analysis of these externalities has been a standard part of the analysis of public economics. But what if many individuals simply do not believe that climate change is real, in spite of the scientific evidence? Should we maximize their ex ante utility function, using their (misguided) beliefs? The questions raised here are analogous to those posed earlier in this Introduction.
References Ariely, D. (2008), Predictably Irrational: The Hidden Forces That Shape Our Decisions, Harper Collins, New York. Arnott, R. and J. E. Stiglitz (1986), “Moral hazard and optimal commodity taxation”, Journal of Public Economics, 29 (1): 1–24. Atkinson, A. B. (2014), Public Economics in an Age of Austerity, Routledge, Abingdon. Atkinson, A. B. and J. E. Stiglitz (1976), “The design of tax structure: Direct versus indirect taxation”, Journal of Public Economics, 6 (1/2): 55–75. Subsequently published in Modern Public Finance, 2, International Library of Critical Writings in Economics, No. 15, A. B. Atkinson (ed.), 82–102, Elgar, Aldershot, UK, 1991. Auerbach, A. J., R. Chetty, M. S. Feldstein, and E. Saez (1985–), Handbook of Public Economics, 5 vols., Elsevier, Amsterdam. Becker, G. S. and N. T. Tomes (1986), “Human capital and the rise and fall of families”, Journal of Labor Economics, 4 (3): S1–39. Bevan, D. and J. E. Stiglitz (1979), “Intergenerational transfers and inequality”, Greek Economic Review, 1 (1): 8–26. Blitzer, C. R., P. Dasgupta, and J. E. Stiglitz (1981), “Project appraisal and foreign exchange constraints”, Economic Journal, 91 (361): 58–74. Boadway, R. (2012), From Optimal Tax Theory to Tax Policy: Retrospective and Prospective Views, MIT Press, Cambridge. Boone, J. and L. Bovenberg (2002), “Optimal labour taxation and search”, Journal of Public Economics, 85 (1): 53–97. Brunnermeier, M. K. and J. A. Parker (2005), “Optimal expectations”, American Economic Review, 95 (4), 1092–1118. Brunnermeier, M., A. Simsek, and W. Xiong (2014), “A welfare criterion for models with distorted beliefs”, Working Paper. Chetty, R. and A. Finkelstein (2012), “The changing focus of public economics research, 1980–2010”, NBER Reporter, 1, 1–6. Dasgupta, P. and J. E. Stiglitz (1974), “Benefit-cost analysis and trade policies”, Journal of Political Economy, 82 (1): 1–33. ——— (1988), “Learning-by-doing, market structure and industrial and trade policies”, Oxford Economic Papers, 40 (2): 246–268. Del Rosal, I. (2011), “The empirical measurement of rent-seeking costs”, Journal of Economic Surveys, 25 (2), 298–325. Diamond, P. A. (2008), “Behavioral economics”, Journal of Public Economics, 92, 1858–1862. Diamond, P. and E. Saez (2011), “The case for a progressive tax: From basic research to policy recommendations”, Journal of Economic Perspectives, 25 (4): 165–190. Domar, E. D. and R. A. Musgrave (1944), “Proportional income taxation and risk-taking”, Quarterly Journal of Economics, 58 (3): 388–422. Emran, M. S. and J. E. Stiglitz (2005), “On selective indirect tax reform in developing countries”, Journal of Public Economics, 89 (4): 599–623. Faguet, J.-P. and C. Pöschl (forthcoming), Is Decentralization Good for Development? Perspectives from Academics and Policy Makers, Oxford University Press, Oxford. Gordon, R., ed. (2010), Taxation in Developing Countries: Six Case Studies and Policy Implications, Columbia University Press, New York. Gordon, R. and W. Li (2009), “Tax structures in developing countries: Many puzzles and a possible explanation”, Journal of Public Economics, 93 (7–8): 855–866. Greenwald, B. C. and J. E. Stiglitz (1986), “Externalities in economies with imperfect information and incomplete markets”, Quarterly Journal of Economics, 101 (2): 229–264. ——— (1990), “Asymmetric information and the new theory of the firm: Financial constraints and risk behavior”, American Economic Review, 80 (2): 160–165. Guzman, M. and J. E. Stiglitz (forthcoming), “Pseudo-wealth and consumption fluctuations”, paper presented to the World Congress of the International Economic Association and to be published in the IEA Proceedings.
introduction HM Treasury (2011), The Green Book: Appraisal and Evaluation in Central Government, HM Treasury, London. Hoff, K. and J. E. Stiglitz (2004a),“After the Big Bang? Obstacles to the emergence of the rule of law in post-communist societies”, American Economic Review, 94 (3): 753–763. ——— (2004b), “The transition process in post-communist societies: Towards a political economy of property rights” in Toward Pro-Poor Policies: Aid, Institutions and Globalization, B. Tungodden, N. Stern, and I. Kolstad (eds.), 231–245, World Bank/Oxford University Press, Washington, DC. ——— (2008), “Exiting a lawless state”, Economic Journal, 118 (531): 1474–1497. Jeanne, O. and A. Korinek (2010), “Excessive volatility in capital flows: A Pigouvian taxation approach”, American Economic Review, 100 (2): 403–407. Kanbur, R., M. Keen, and M. Tuomala (1994), “Optimal non-linear income taxation for the alleviation of income-poverty”, European Economic Review, 38 (8): 1613–1632. Korinek, A. and J. E. Stiglitz (2008), “Political economy in a contestable democracy: The case of dividend taxation”, 2008 Meeting Papers, Society for Economic Dynamics. ——— (2009), “Dividend taxation and intertemporal tax arbitrage”, Journal of Public Economics, 93 (1/2): 142–159. Kotlikoff, L. J. (1988), “Intergenerational transfers and savings”, Journal of Economic Perspectives, 2 (2): 41–58. Marris, R. (1964), The Economic Theory of Managerial Capitalism, Free Press of Glencoe, New York. Maskin, E. and A. Sen (2014), The Arrow Impossibility Theorem (based on the Seventh Kenneth J. Arrow Lecture at Columbia University), Columbia University Press, New York. Mirrlees, J. A. (1976), “Optimal tax theory: A synthesis”, Journal of Public Economics, 6 (4): 327–358. ——— (chair) (2010), Dimensions of Tax Design, Oxford University Press, Oxford. ——— (chair) (2011), Tax by Design, Oxford University Press, Oxford. Musgrave, R. A. (1959), The Theory of Public Finance, McGraw-Hill, New York. Neary, J. P. (2001), “Monopolistic competition and international trade theory” in The Monopolistic Competition Revolution in Retrospect, S. Brakman and B. J. Heijdra (eds.), Cambridge University Press, Cambridge. Oates, W. E. (1999), “An essay on fiscal federalism”, Journal of Economic Literature, 37 (3): 1120–1149. Piketty, T. and E. Saez (2013), “A theory of optimal inheritance taxation”, Econometrica, 81: 1851–1886. Rabin, M. (2013), “Incorporating limited rationality into economics”, Journal of Economic Literature, 51: 528–543. Robson, A. and L. Samuelson (2011), “The evolution of decision and experienced utilities”, Theoretical Economics, 6: 311–339. Sah, R. S. and J. E. Stiglitz (1992), Peasants versus City-Dwellers: Taxation and the Burden of Economic Development, Oxford University Press, Oxford and New York. Shapiro, C. and J. E. Stiglitz (1984), “Equilibrium unemployment as a worker discipline device”, American Economic Review, 74 (3): 433–444. Solow, R. M. (2003), “Dumb and dumber in macroeconomics”, available online at https:// www0.gsb.columbia.edu/faculty/jstiglitz/festschrift/Papers/Stig-Solow.pdf. Stiglitz, J. E. (1969), “Distribution of income and wealth among individuals”, Econometrica, 37 (3): 382–397. ——— (1976), “Estate taxes, growth and redistribution” in Essays in Honor of W. Vickrey, R. Grieson (ed.), 225–232, Lexington Publishing Company, Lexington. ——— (1978a), “Equality, taxation and inheritance” in Personal Income Distribution, W. Krelle and A. F. Shorrocks (eds.), 271–303, North-Holland Publishing Company, Amsterdam. ——— (1978b), “Notes on estate taxes, redistribution, and the concept of balanced growth path incidence”, Journal of Political Economy, 86 (2): 137–150.