Retrospectives on Malleliiom wlatelarer | EDITED BY LORRAINE EDEN Digitized by the Internet Archive in 2021 with funding from Duke University Libraries https://archive.org/details/retrospectiveson01unse Retrospectives on Public Finance Duke University Press Durham Retrospectives on Public Finance EDS TED BY L.OUR RA | NE ED EN © 1991 Duke University Press All rights reserved Printed in the United States of America on acid-free paper © Library of Congress Cataloging-in-Publication Data appear on the last printed page of this book. CONTENTS Preface | Introduction ie LORRAINE EDEN Retrospectives on Public Finance: An Introduction to the Issues Il The Shoup Tax Missions 2). 3. 4. oF CARL S. SHOUP Melding Architecture and Engineering: A Personal Retrospective on Designing Tax Systems MALCOLM GILLIS Legacies from the Shoup Tax Missions: Asia, Africa, and Latin America MINORU NAKAZATO The Impact of the Shoup Report on Japanese Economic Development CHARLES E. MCLURE, JR. Income Tax Reform in Venezuela: Thirty Years after the Shoup Mission 31 51 67 Contents JOHN BOSSONS Comment: Evaluating Tax Missions Ill The Tax Mix 6. RICHARD GOODE Changing Views of the Personal Income Tax 7. WILLIAM VICKREY The Corporate Income Tax and How to Get Rid of It 8. WAYNE THIRSK Intellectual Foundations of the VAT in North America and Japan 9. JOHN F. GRAHAM The Place of the Property Tax in the Fiscal System STANLEY WINER Comment: Choosing an Appropriate Time Horizon for Taxation W. IRWIN GILLESPIE Comment: Taxing Consumption and Wealth IV The Expenditure Mix 10. LORRAINE EDEN AND MELVILLE L. MCMILLAN Local Public Goods: Shoup Revisited 86 93 118 133 149 168 172 ie We 12. CLIFF WALSH Public Goods Provision with Price Exclusion: Market Behavior and Market Performance JOHN G. HEAD Merit Wants: Analysis and Taxonomy ENID SLACK Comment: Retrospectives on Local Public Goods EDWIN G. WEST Comment: Merit Wants and Public Goods Theory V_ Macro Public Finance 13: 14. 15. RICHARD M. BIRD Tax Structure and the Growth of Government PEGGY B. MUSGRAVE Fiscal Coordination and Competition in an International Setting DOUGLAS A. L. AULD Compensatory Fiscal Policy: Evolution or Revolution? JOHN B. BURBIDGE Social Security and Public Debt in Historical Perspective JACK M. MINTZ Comment: The Role of Economic Policy in the Tax Reform Process Contents 203 229 J8)3) 255 263 276 306 323 338 Vii vill Contents JOHN SARGENT Comment: On Macro Public Finance VI Conclusion 17. RICHARD A. MUSGRAVE Tableau Fiscale References Contributors Author Index Subject Index 344 363 393 397 402 PREFACE This volume contains papers and discussant comments presented at the conference “Retrospectives on Public Finance: A Conference in Honour of Carl Sumner Shoup,” held at Carleton University, Ottawa, Canada, 20—21 March 1989. The conference was jointly organized by the Norman Paterson School of International Affairs, the Department of Economics, and the School of Public Administration at Carleton Univer- sity. The conference was funded by grants from the Japan Foundation, the Social Sciences and Humanities Research Council of Canada, and Carle- ton University. The assistance of the Canadian embassies of Japan, Venezuela, and the United States is also gratefully acknowledged. My co- organizers, W. Irwin Gillespie and Harvey Lithwick, conference secre- tary, Mrs. Edie Landau, and conference assistant, Michele Kovanchak, provided invaluable advice and support. Judith van Walsum and Susan Olsen assisted with preparing the manuscript. I would also like to thank Richard Bird, Malcolm Gillis, Fraser Taylor, Stan Winer, and my family for their assistance. Lastly, like the other participants in this volume, I want to thank Carl Shoup for his help, encouragement, and friendship. Lorraine Eden Ottawa, Canada July 1990 | Introduction 1. LORRAINE EDEN Retrospectives on Public Finance: An Introduction to the Issues The purpose of this chapter is to introduce both the subject matter of this book and the papers within it. Retrospectives on Public Finance contains the edited proceedings of a conference held at Carleton University, Ottawa, Ontario, Canada, in May 1989, which brought together scholars and policy makers from Canada, Australia, Japan, and the United States for a retrospective on the discipline of public finance, in honor of one of its most famous practitioners, Carl Sumner Shoup, McVickar Professor of Political Economy Emeritus at Columbia Univer- sity. The 1989 conference marked the twentieth anniversary of Shoup’s 1969 major treatise, Public Finance (Chicago: Aldine Publishing). The date also marked the anniversaries of the Shoup Mission to Japan in 1949-50 and the Shoup Mission to Venezuela in 1958—59. From his early work in the U.S. Treasury Department, through the Shoup Tax Missions to Japan, Venezuela, and Liberia, from his ongoing interest in public expenditures and size of the government sector, through the value-added tax and taxation of multinationals, Carl Shoup has suc- cessfully intermingled theoretical advances in public finance with practi- cal advice to policy makers around the world. Although the conference was not formally in honor of Richard A. Musgrave, his great contributions to the discipline, the similarities in his and Shoup’s approaches to public finance, and the fact that 1989 is the thirtieth anniversary of his pathbreaking treatise, The Theory of Public Finance: A Study in Political Economy (New York: McGraw-Hill), are also marked by this book. These events, the publication of two important treatises and the conclusion of two tax missions, had a substantial impact on the theory and practice of public finance from the 1930s through the 1980s. Carl Shoup and Richard Musgrave, both leading world figures, represent a branch of public finance that we can call the “prescriptive” or “political 4 Lorraine Eden economy” school. This school is older than the optimal tax and the deadweight loss literature one associates with Harberger, Atkinson, Stiglitz, and others. The Shoup-Musgrave branch combines rigorous theory with practical policy advice. Public finance as practiced by these economists is a distinct approach, one that is prescriptive, hands-on, and firmly rooted in the political economy tradition that economics must be applicable to real world problems. These economists believe that fiscal institutions are important to our understanding of public finance and that theory and practice must interact for success in making democracy function. Carl Shoup and Richard Musgrave have trained and been colleagues with over two generations of public finance specialists, including Rich- ard Bird, Sijbren Cnossen, John Due, Lyle Fitch, Irwin Gillespie, John Graham, Richard Goode, Lowell Harriss, John Head, Peggy Musgrave, Charles McLure, Jr., Hirofumi Shibata, and William Vickrey, to name just a few. Their research spans the entire field of public finance, from the theory of public goods to the integration of the personal and corporate income taxes to macroeconomic policy to fiscal harmonization. These economists believe in a practical approach to public finance, one that uses theory to design realistic fiscal structures. The purpose of the conference was to take a retrospective look at this political economy branch of public finance in honor of Carl Shoup’s contribution to it. The participants were asked to review the literature of this school and to assess its relevance to current problems in public finance theory and policy making. Each participant was asked to focus on the long view: to address the issues of what was really important, what has stood the test of time, what this branch can say to modern public finance problems, what advice can it offer to developing countries and to tax reform movements. Each paper was to be both a backward and a forward look. Given the current round of worldwide fiscal reforms (e.g., the United States, Canada, Japan, Australia, the United Kingdom, the European Community), such a retrospective look seemed par- ticularly appropriate. Outline of the Book Retrospectives on Public Finance is drawn from the papers and discus- sant comments presented at the conference which were subsequently Introduction to the Issues revised and edited for inclusion in this book. It is intended to offer the economics profession surveys that are useful, not only to academics and their graduate students, but also to practitioners of public policy in both developed and developing countries. The goal of the book is to help improve the level of policy making and economic analysis applied to public policy issues through a careful review of the prescriptive school of public finance. Retrospectives can be a particularly effective research and learning tool. They are useful to academics, policy makers, and graduate students who wish to gain an understanding of a particular field of study. Through taking a long-run view of a field’s development, authors can isolate key contributions, areas where further work needs to be done, and relevant work that has been lost to current researchers. Second, assessments of previous fiscal reform missions can be helpful in designing successful new reform missions. Such assessments link the importance of “architecture” and “engineering,” to use Carl Shoup’s words, to successful fiscal reform. Lastly, bringing together the senior statesmen in the profession with younger academics can lead to fruitful exchanges, linking the practical experiences of the older generation with theoretical advances of the younger group. This volume is divided into six parts, consisting of seventeen chapters and seven comments. It is framed by two chapters, this one (Part I), which summarizes the papers and their themes, and the last chapter (Part VI) written by Richard Musgrave, which offers a personal view of the history of public finance from the 1930s through today. Part II opens with a personal retrospective by Carl Shoup on the relative importance of tax architecture, engineering, and administration in successful fiscal missions. Shoup’s paper is followed by three chapters assessing the success/failure of the Shoup missions and the lessons these missions offer policy makers. Part III looks at the personal and corporate income taxes, the value added tax, and the role of the property tax in the fiscal system. Part IV contains papers on local public goods, excludable public goods, and merit wants. Part V deals with impacts of the tax structure on expenditure growth, international public finance issues, compensatory fiscal policy, and social security and public debt. The range of topics covered in Retrospectives on Public Finance is very broad, as befits a book designed to assess the political economy school of public finance and Carl Shoup’s role within it. A brief outline of the key points of each of the chapters is provided below. 6 Lorraine Eden The Shoup Tax Missions Carl Shoup’s paper on “Melding Architecture and Engineering” asks a question he considers neglected but important: How should a. tax mis- sion allocate its resources between tax architecture, engineering, and administration? Shoup defines tax architecture as the choice of taxes to form a country’s revenue base and the outlining of the essential features of each tax. Tax engineers must decide each of the substantive issues within the tax. Tax administration is concerned with how to implement the provisions in the tax law formulated by the architects and engineers. He argues that, contrary to some critics, tax architecture is an essential task for tax missions; while tax engineering and administration are both of growing importance. Malcolm Gillis’s paper on “Legacies from the Shoup Missions” notes that both Richard Musgrave and Carl Shoup are synonymous with tax missions; both are economists who can be referred to as “tax doctors.” Gillis argues that the Shoup mission to Japan was a prototypical, path- breaking study since it was the first mission where distinguished public finance academics applied theory to the practical problem of reforming an entire tax system. All four of Shoup’s missions emphasized tax fair- ness, the relation between tax systems and inflation, and the centrality of good tax administration in public finance. Gillis concludes that the legacy of the Shoup missions was not only the actual measures enacted, but also the impact of the missions on defining the agenda for tax reform in other nations. “The Impact of the Shoup Report on Japanese Economic Develop- ment” by Minoru Nakazato notes that the report has influenced Japa- nese tax policy for nearly forty years. Nakazato concentrates on the tax reform aspects and asks whether the recommendations were appropriate at the time. He notes that fairness and the ideal of a comprehensive income tax were basic features of the report. Nakazato faults the mission for possibly encouraging overincorporation within Japan as a tax- minimizing technique. Charles McLure, in “Income Tax Reform in Venezuela: Thirty Years after the Shoup Mission,” compares the recent World Bank report on the income tax system in Venezuela with the report of the 1958—59 Shoup mission, which he considers a classic in the field. McLure focuses on the tax engineering aspects and argues that good tax advice has had little impact in Venezuela and that little has been learned from mistakes. He Introduction to the Issues attributes this to the absence of a tax research group, the lack of political will to introduce reforms, and excessive reliance on high petroleum rents as a source of tax revenue. McLure concludes that tax reform may only occur when governments are forced by revenue needs to make unpopu- lar choices. The Tax Mix Richard Goode in “Changing Views of the Personal Income Tax” de- scribes and comments on changing views of the proper form and role of the personal income tax in the economy. Goode deals with three crucial issues: the definition of the base of the tax, the appropriate rate struc- ture, and the choice between income and expenditure taxes. The paper uses Carl Shoup’s taxonomy of consensus and conflict criteria to identify three conflict and four consensus variables that influence discussion of income tax rates. The conflict variables all deal with progressivity of the rate structure: the exemption of incomes below poverty level, mild prog- ressivity over middle range, and steep graduation at high income levels. The consensus criteria are general fiscal goals: adequate revenue, avoid- ance of excessive interference with macro goals and macro stability, and practicable administration and compliance. Goode then compares the personal income tax to the expenditure tax in terms of the consensus cri- teria and argues that the income tax is still superior. He concludes that the academic view of the personal income tax has changed greatly be- cause of economic analysis and the economic environment. In “The Corporate Income Tax and How to Get Rid of It,” William Vickrey notes that the corporate income tax faces the widest gap be- tween political popularity and professional esteem. He attributes this gap to the public belief that the tax is borne by someone else, while economists disagree on the incidence of the tax. Vickrey argues that the tax in the United States was changed in 1936 without a clear underlying rationale and that its structure increases the cost of equity capital for firms, having a twofold depressing effect on business investment. In addition, he notes that the tax discriminates in favor of debt-financed and speculative investment, pressures firms to increase leverage ratios, and encourages leveraged buyouts and tax-driven mergers. Uncertainty as to its incidence makes it difficult to calculate benefits and costs, while the foreign tax credit mechanism creates an incentive for other nations to adopt similar taxes in order to take advantage of the credit. Vickrey’s 8 Lorraine Eden solution is a flat tax levied on a source basis and a progressive tax levied on a destination basis. He notes that there are problems with getting rid of an old, established tax due to windfalls, uncertainties, and incentives for tax avoidance. His conclusion is that we may have to live with the tax and that returning the tax to its pre-1936 structure may be.the only feasible solution. Wayne Thirsk’s “Intellectual Foundations of the var” identifies Carl Shoup as the intellectual father of the value-added tax (vat). The pur- pose of the chapter is to identify Shoup’s contributions to growth and application of value-added taxes, now used by more than fifty countries worldwide. Thirsk notes that Shoup does not provide a blanket endorse- ment of the vAT, on the grounds that it is not a “magic tax potion for solving all fiscal ailments”; the choice of sales tax depends on the struc- tural peculiarities of each country. Thirsk concludes that Shoup’s work on the vaT is an “impressive intellectual achievement” which developed the taxonomy used to describe value-added taxes, established the essen- tial equivalences, compared advantages of vAT to other commodity taxes, and linked tax administration capabilities to the appropriate VAT for developing countries. The last chapter, John Graham’s “The Place of the Property Tax in the Fiscal System,” is broader in scope than the other chapters in this section because it deals with both sides of the government ledger, revenues and expenditures. Graham develops an analytical framework for establish- ing an efficient and equitable system of local public finance, one that takes into account property taxes and the nature of municipal services and that can be operationalized by policy makers. He argues that the distinction between people and property services is erroneous since only people can benefit from public services. A more satisfactory distinction is between general and local services. Graham discusses the old and new views of incidence of the property tax; he sees the tax as a tax on the value of the service rendered by property to the owner/tenant over some time period. The paper develops a rational assignment of functions and revenues between upper- and lower-tier jurisdictions, one that is de- signed to achieve both vertical and horizontal fiscal balance. The Expenditure Mix “Local Public Goods: Shoup Revisited” by Lorraine Eden and Melville McMillan reviews Carl Shoup’s writings on local public goods in the light of developments in this area of public finance over the past twenty Introduction to the Issues years. The authors identify four key issues around which Shoup’s work has centered: the characteristics of public services, the definition and measurement of output, the determination of cost functions for level of service, population, and area of the jurisdiction, and the distribution and incidence of the benefits from local public goods. The paper concludes that Shoup’s major contributions lie in his work on the characteristics and distribution of local services. Cliff Walsh’s chapter on “Public Goods Provision with Price Exclu- sion” deals with market provision of price-excludable public goods. He notes that jointness is the key characteristic of such commodities and that jointness has important and not well understood implications for the nature of market equilibrium and public policy. Walsh investigates the implications of excludable public goods for pricing, market struc- ture, decision making, and market behavior. He concludes that market provision may not be superior to government provision for this class of commodities. The last chapter in this section, John Head’s “Merit Wants: Analysis and Taxonomy,” shifts the emphasis from social wants (public goods, ex- ternalities, joint goods) to merit goods. Head notes that social wants have dominated work in the public finance area because they embody a whole range of cases where market failure can justify government provision in order to give effect to individual preferences. Merit wants, on the other hand, appear to interfere with consumer sovereignty and thus have led a “shadowy existence on the periphery.” Head proposes a broader analyt- ical framework that extends the concept of individual sovereignty to in- clude both psychic externalities and self-paternalism. Head’s framework uses a hierarchy of preferences that includes behavioral aberrations such as impulse or weakness of will. He concludes that his framework makes the scope of merit wants fully comparable to the scope of social wants and that merit wants could perhaps be used to justify public provision in cases where congestion estimates are used to suggest privatization. Macro Public Finance The first paper in this section, Richard Bird’s “Tax Structure and the Growth of Government,” argues that tax structure and the size of gov- ernment are linked. For Bird fiscal changes are made through a political, not an economic, process, one that depends on the interplay of interests and actors. Five propositions link tax structure to expenditure growth: 10 Lorraine Eden (1) the tax structure is quasi-constitutional in nature and (2) an impor- tant political symbol; (3) tax structure changes are due primarily to exogenous shocks; and (4) tax structure is linked to expenditure struc- ture and (5) to the growth of government spending. Bird concludes that understanding tax reform and understanding government growth are different facets of the same problem. Peggy Musgrave’s chapter, “Fiscal Coordination and Competition in an International Setting,” examines the roles played by fiscal coordina- tion and competition when goods, labor, and capital can move across jurisdictions. She focuses on two approaches in the literature. The first approach, associated with Shoup and the European Community, recom- mends coordination of fiscal systems where individual countries imple- ment their own equitable tax structure that retains individual fiscal characteristics, while using harmonization to neutralize fiscal differen- tials between countries. The second approach argues that tax competi- tion, like competition between firms in the private marketplace, can promote efficiency and responsiveness of governments. Musgrave com- pares these two approaches and concludes that the analogy to competi- tion in the private sector does not hold. Even if domestic fiscal structures are efficient, fiscal competition cannot secure an efficient and equitable fiscal structure internationally. Where domestic policies are inefficient, she concludes that fiscal competition is a “clumsy and costly means” of correcting these problems. Douglas Auld in “Compensatory Fiscal Policy: Evolution or Revolu- tion?” assesses the changes in the economics profession’s views about fiscal policy since 1960. Auld first discusses the contributions of Carl Shoup’s and Richard Musgrave’s treatises to this area. He then identifies seven key theoretical developments which have affected the profession’s assessment: rational expectations, supply-side economics, the Laffer curve, tax-push inflation, Ricardian equivalence theory, crowding out, and the natural rate of unemployment. In each case Auld goes through the definitions and discusses how each development is interpreted in a fiscal policy context and its implications for compensatory fiscal policy. Turning to textbooks in public finance and macroeconomics, he looks to what conclusions have been reached on compensatory fiscal policy and what major issues still confront policy makers dealing with short-term stabilization policy. Auld notes that stabilization policy as a topic is no longer included in most public finance textbooks. He concludes that compensatory fiscal policy is unlikely to play a major role in counter- cyclical policy today due to high budget deficit and debt levels. Introduction to the Issues John Burbidge’s chapter, “Social Security and Public Debt,” finishes this section. Burbidge notes that the reigning view of social security and debt is based on the Samuelson-Diamond life-cycle growth model. Two critical assumptions underpin this model: (1) relative savings rates differ by age bracket, and (2) taxpayers believe they will receive benefits while debt holders believe the debt will not be repudiated in the future. Bur- bidge then compares this approach with Shoup’s views in his 1969 treatise. Burbidge argues that the older approach forces the reader to think about the uncertain environment in which real world governments must try to function, whereas the current models have only a “nodding acquaintance with reality.” He concludes that knowledge is not cumula- tive, that important ideas have been lost, and that each generation of economists takes important assumptions for granted that previous gen- erations questioned. Burbidge recommends that research be done “in the eclectic middle ground where model selection is more closely tied to empirical facts than to the appeal of its policy conclusions.” Conclusion: Tableau Fiscale The last paper in this volume is by Richard Musgrave. Musgrave notes that Carl Shoup has contributed to the public finance community for sixty years. The purpose of Musgrave’s chapter is to look at changes in the field of public finance over the same time period. Musgrave provides a table organized into three rough time periods, 1930-50, 1950-70, and 1970-90, and eight broad topics: general public sector theory, taxation, expenditures, public enterprise, public debt, macro theory, utility and social welfare, and ideology. By going through each of the cells in this matrix, he provides a broad, sweeping view of the growth and dynamics of public finance thought. Lastly, Musgrave stresses the importance of ideology and values. He argues that many of the advances in public finance have been achieved due to new analytical tools; how- ever, the questions that each generation poses and the projects it under- takes are influenced by the time period and its value sets. Main Themes of the Book From the papers in this book emerge six themes which run, either implicitly or explicitly, through all the papers. The first theme is the importance of the AEA functions—architecture, engineering, and admin- Lorraine Eden istration—in successfully applying theoretical models of public finance to practical policy problems. Carl Shoup’s chapter defines these func- tions and applies them to tax missions; however, the concepts are broader than that. All policy makers must take account of the AEA functions in designing successful fiscal reform. Several papers in the book link theory to practice. While Shoup focuses on the links between the three functions, Malcolm Gillis argues that good tax administration is central to public finance. Charles McLure, on the other hand, stresses the engineering aspects. Wayne Thirsk argues that Shoup’s solid work on the taxonomy and administrative feasibility of value added taxes is an example of “pragmatism in its pristine form.” John Graham, in his chapter on municipal public finance, stresses practicality in his design for an efficient and equitable fiscal structure. The second theme that runs through this book is the political economy of fiscal reform. This is most clearly argued by Richard Bird in his chapter on the links between taxes and expenditure, where he argues that fiscal reform is a political not an economic process that depends on the inter- play of interests and actors. Jack Mintz draws this distinction more sharply. He argues there are two groups of economists, those who argue for practical policy advice on the grounds that analysis is irrelevant be- cause it ignores the behavior and attitudes of governments and those who believe their role is to recommend policies using normative criteria for tax reform, regardless of the political process. Richard Goode, using Shoup’s taxonomy, stresses the importance of conflict and consensus cri- teria in determining the degree of progressivity of the personal income tax. The importance of political economy is illustrated in McLure’s chap- ter which argues that the failure of income tax reform in Venezuela was due to the lack of political will to introduce reforms. It is also seen in William Vickrey’s assessment of the longevity of the corporate income tax despite its well known inefficiencies and inequities. Edwin West also uses the political economy approach to explain public provision of ser- vices usually considered private like education and nonprofit hospitals. A third theme is the changing agenda within the public finance profes- sion. Several of the authors compare the Shoup-Musgrave prescriptive school with the more recent optimal tax tradition. The broadest sweep is provided by Richard Musgrave’s tableau fiscale. He makes the link be- tween ideology and values: the questions each generation of economists poses are influenced by the time in which they write and the values of that time. Burbidge is perhaps most worried by this changing agenda. In his comparisons of new and old models of social security and debt he notes Introduction to the Issues that knowledge is not cumulative and that some of the practical insights of the older literature have been lost. In terms of specific agenda changes Richard Goode discusses the controversy over the relative merits of in- come and expenditure taxes and argues that both economic analysis and the economic environment affect theoretical debates. Graham discusses the new and old views of the property tax. Eden and McMillan compare Shoup’s work on local public goods to developments in this area over the past twenty years. John Head builds a new, broadened view of individual sovereignty and applies it to the merit wants concept. Peggy Musgrave compares the old view of fiscal harmonization with the new view of tax competition. The paper on compensatory fiscal policy by Douglas Auld identifies seven key developments over the past thirty years. Another theme is the importance of the international dimension in public finance. Peggy Musgrave’s chapter contrasts fiscal harmonization with fiscal competition among states and concludes that harmonization is preferable. Shoup insists that tax architects need a broad view of international fiscal policy changes in order to design effective tax policy. Vickrey stresses that domestic taxes like the corporate income tax have international economic and political effects. McLure notes that changes in the international environment, by constraining tax revenues, can induce meaningful fiscal reform. Tax structure change results primarily from exogenous shocks, according to Bird. Mintz gives examples of difficult policy recommendations in the international sphere. Several of the authors address unresolved issues within their area of public finance. Goode discusses the debate between income and con- sumption taxes as the appropriate tax base. Thirsk identifies the subtrac- tion method of calculating value added taxes as an unresolved issue. Eden and McMillan argue that work on defining and measuring both the output and the incidence of public services is needed. Cliff Walsh stresses that jointness in the context of public versus private provision is little understood. Full integration of a broadened concept of the role of merit wants into the theory of public finance remains to be done, according to Head. Richard Bird calls for more understanding of the linkages between tax structure and government growth. Stan Winer notes that the appro- priate time horizon for tax purposes is an unsettled issue. The last theme that threads through Retrospectives on Public Finance is the importance of Carl Shoup’s work. Each of the chapters in this book identifies and assesses his contribution to the theory and practice of public finance. Malcolm Gillis argues that the Shoup missions defined the agenda for tax reform in developing countries. Nakazato emphasizes 13 14 Lorraine Eden the impacts of the Shoup report on economic development in Japan. Thirsk identifies Shoup as the “intellectual father of the var” (although Shoup would argue that he is not “at bat for the vat”). Eden and McMillan conclude that he made major contributions to the theory of local public goods. Peggy Musgrave credits Shoup with much of the underlying work on international fiscal harmonization. From these as- sessments it is not hard to draw the conclusion, which Malcolm Gillis states in his chapter, that Carl Shoup is one of the great public finance figures of this century. Using This Book in the Classroom As a comprehensive survey of the prescriptive school of public finance, the contents of Retrospectives on Public Finance should be useful to graduate students and practitioners of public finance. Possible courses where this book could be used include senior undergraduate and gradu- ate courses in public expenditure, public sector economics, taxation theory and policy, public finance in developing countries, international taxation, public administration and policy analysis, public choice the- ory, and intergovernmental relations. The chapters and comments are readable by senior undergraduates and graduate students in public finance courses, so that the book can be used to supplement standard public finance textbooks, in addition to Musgrave’s and Shoup’s own treatises. A half course in public expendi- ture should focus on Parts IV, V, and VI; one on taxation, on Parts II, III, V, and VI. A full course on public finance would use all of the readings. Courses emphasizing public finance in developing countries might con- centrate on Parts II, II], and V; while courses on public choice and welfare economics would benefit from Parts IV, V, and VI. The table below is included for those professors who wish to use the book as a supplementary book of readings to accompany one of the standard textbooks in public finance. The table cross-references each chapter in Retrospectives on Public Finance to the following textbooks: (A) Atkinson, A. B., and J. E. Stiglitz. 1980. Lectures on Public Economics. London: McGraw-Hill. (B) Boadway, Robin, and David Wildasin. 1984. Public Sector Economics. 2nd ed. Boston and Toronto: Little, Brown. (C) Brown, C. V., and P. M. Jackson. 1986. Public Sector Economics. 3rd ed. Oxford: Basil Blackwell. Introduction to the Issues 15 (D) Due, John F., and Ann F. Friedlaender. 1981. Government Finance: Economics of the Public Sector. 7th ed. Homewood, IIl.: Irwin. (E) Musgrave, Richard. 1959. The Theory of Public Finance: A Study in Political Economy. New York, Toronto, and London: McGraw-Hill. (F) Musgrave, Richard and Peggy Musgrave. 1984. Public Finance in The- ory and Practice. 4th ed. New York: McGraw-Hill Ryerson. (G) Prest, A. R., and N. A. Barr. 1985. Public Finance in Theory and Prac- tice. London: Weidenfeld and Nicholson. (H) Rosen, Harvey. 1988. Public Finance. 2nd ed. Homewood, Ill.: Irwin. (I) Shoup, Carl. 1969. Public Finance. Chicago: Aldine. (J) Stiglitz, Joseph. 1986. Economics of the Public Sector. New York and London: W. W. Norton. (K) Tresch, Richard. 1981. Public Finance: A Normative Theory. George- town, Ont.: Irwin-Dorsey. The chapters in Retrospectives on Public Finance that correspond to the chapters in the textbooks listed above are outlined in the matrix below. This matrix lists chapters vertically and textbooks horizontally. Each cell in the matrix gives the chapter in the textbook which corresponds to the chapter in this volume. R mE P Public Finance Textbooks Ip A By 6 D E F G H I J) K 2 18 3 2 — 1S 137 7 DD} 17 16,25 1,14 3 18 —_- — 19 — 3 Ti 15 7 16,25 — 4 18 —- — — _ 37/ vi 15 17 16,25 — 5 18 —- — —- — 37 7 15 17 16,25 — 6 D3 3 4h NO aN ALG A 7/ 13 16,17 11 20,25 13 7 5) 13} lg 12 14 eS) Nei 7/ 19 12 Pi 72) 17 8 1 Wil al 15 13 20 19 20 8,9 17 a 9 3 WA iG 16 14 21 15 21 aS) 22 13 10 7 15 9 — 1 23; 3 22 5 26,27 29 11 16 4 3 2,3} 1 3,4 3 5 4,5 4,5 5 12 a 6 4 — 1 4 — 4 4 3,4,6 31 i Ot! Seer zs 6,7 8 6 20 Deca ses 14 N7/ 15 — 18 8 36 18 723} BS) 26 == 15 a= —- — — 18,21 29,30 6 18 22,24 _ -- 16 8 14 eels) 6 23,24 33,34 14 10,11 6 13 13 U7 tii) 3 2 — 1 1 2 2 ez 4 1,14 = tp A eh te nal | awed byl juts he Bi Fa iis 7 HDi “ wha o ‘ av ‘whe “4 2 ve = ae | infra ' I! The Shoup Tax Missions 2: GUAURL EL Si..coaht OUP Melding Architecture and Engineering: A Personal Retrospective on Designing Tax Systems Tax Architecture, Tax Engineering, Tax Administration The purpose of this paper is to look at the tax missions which I have directed or helped direct, and ask a question that has been some- what neglected but is important: how should a tax mission allocate its resources among tax architecture, tax engineering, and tax administra- tion in conducting its research and in its report? Tax architecture is the choice of taxes that will form the revenue system for the country in question and the outlining, broadly, of the essential features of each tax. Tax engineering makes the decisions on each of the numerous substantive issues within each tax. Tax administra- tion is distinct from both of these; it is concerned with how to implement the provisions in the tax law formulated by the architects and engineers. The three tasks must be accomplished almost simultaneously: the tentative decisions reached in any one of the areas will influence the decisions to be made in the other two. This melding sometimes requires a continuous interchange of suggestions and ideas between two or more members of the mission; sometimes the melding takes place in the mind of one person, since most tax mission members have interests and skills in all three fields. (I rather like the word “meld” here; my dictionary tells me that it signifies a blend of “melt” and “weld.” A large part of a tax mission’s time is taken up with melting and welding). The tax mission, I assume, is made up of foreign experts. This fact, we shall see, tends to influence the allocation of time and effort between the three fields of architecture, engineering, and administration. The distinction between these three can be illustrated by the decision Carl S. Shoup whether to allow a carryover of a business loss, under an income tax, to other taxable years. That there is to be an income tax and that it is to allow some kind of carryover of loss are architectural decisions. The engineering task is to decide, on substantive grounds, such questions as: Shall the carryover be a carryforward or a carryback or both? If there is to be a carryback of the loss, how many years shall it be carried back? Shall it be absorbed first in the earliest year to which it is carried back, or to the latest year, working back from there? Is the carriedback loss to be computed by retaining all the tax preferences or only after adding them back in? These are issues to be settled on grounds of equity and economic efficiency, with administrative considerations playing a minor role, if any, except perhaps on the tax preferences issue. But administrative expertise must be depended on to answer such a question as, if there is an unabsorbed loss after all the carryback, shall it give rise to an immediate tax refund, or must the taxpayer wait to apply it in carryforwards? A somewhat similar problem exists under the value-added tax, with respect to freeing certain goods or services from the tax. There is a choice between full freeing through zero rating and exemption only of the one stage of value added by exempting the sale but disallowing a credit for earlier-stage VAT on the components of the favored good. This is an engineering, substantive issue. But the zero-rating technique may require a tax refund to be immediately and completely effective. In some coun- tries this raises an important administrative issue: is it practicable? If it is not, the engineering answer may have to be modified. Several more illustrations of the differences between these three con- cepts may be drawn from the report of the Japan tax mission. Indexing for inflation is an engineering task. In Japan the income tax on business profits was foundering in 1949, partly because the basis for computing depreciation had not been adjusted upward to keep pace with inflation. Many firms were forced to evade tax in order to survive. The mission report recommended such an adjustment, which was carried out. Similarly, personal exemptions and allowances for dependents had not been moved upward quickly enough, with corresponding demoral- ization of taxpayers and tax officials as many of the former strove to evade taxation that cut deeply into the necessities of life. Another example of an engineering change was the inclusion of all fixed assets, including machinery and equipment, in the property tax base (local tax), but this comes close to an architectural change. In contrast taxing recipients rather than donors, with consequent On Designing Tax Systems change in the basis of progression of the tax rates, under the transfer taxes (death and gift tax) was an architectural alteration. Substituting a value-added tax for the income tax at the prefectural level (we thought the three tiers of income tax—national, prefectural, and municipal—put too much pressure on one tax) was another archi- tectural recommendation. The engineering and administrative issues were not, however, spelled out adequately, and, had the tax been en- acted, they would have posed important problems, especially for sales across prefectural boundaries. In retrospect we did not give enough attention to these two aspects of the VAT. An entire volume of the Japan report was devoted to administration of the income tax. It covered four groups of topics: (1) voluntary com- pliance and self-assessment; (2) procedural steps in compliance, returns handling, collection, and litigation; (3) office structure, procedures, and personnel within the administration; (4) collateral matters. Most of these details would have been applicable whatever the decisions on the role of the income tax (its degree of importance) and its engineering provisions. In drafting a personal income tax there is a choice between a tax credit and a deduction from income to allow for spouse and dependents. If the difference in tax between two families of different size but the same income should increase as income increases, the deduction from income is the preferred method. I consider this an engineering rather than an architectural issue. It seems scarcely an administrative issue. In contrast the details of withholding income tax from wages and salaries involve almost entirely decisions that are tax administration not engineering related. Shall the wage or salary earner supply information to the employer, not only about family status, but also relative to items of income or outgo not connected with the employment income, for exam- ple, interest paid, dividends, or interest received? The issue here is not the amount and timing of the tax, but practicality and administrative feasi- bility. The distinction between tax architecture and tax engineering is there- fore chiefly one of general outline versus detail. Consider a country where the outlines of the tax system are generally appropriate. Let it be a highly industrialized one, with a retail trade considerably concentrated in large firms. The country depends chiefly on an income tax and a retail sales tax (abstracting from local taxes). Assume the details of the tax law may, for both these taxes, be inconsistent, incomplete, and vague. By 72 22 ~=Carl S. Shoup analogy to a building, the structure has been correctly designed, but the hot water faucets run cold, the elevators are programmed to stop only at the top and bottom floors, the fire doors open inward, and so on. The elevator operators are well trained, the doorman is courteous and help- ful, the fire drills are well conducted. Administration of the building is thus admirable, and the building well proportioned, but the engineering has been faulty or at least inconsiderate. At the other extreme is the tax system for which the taxes have been meticulously and appropriately drafted and are administered efficiently and honestly, but where these virtues are largely wasted because the important taxes are unsuited to the country’s economic and cultural background. The country may be, for example, almost entirely a nation of small farmers and fishermen and exporters of raw materials, with a retail trade fragmented among small shopkeepers, yet it may have no export tax but instead a refined value-added tax and a complex income tax. Veterans of tax missions will correctly see these sketches as somewhat exaggerated, yet indicative of certain broad truths. The choice of rates for a certain tax is better viewed as an architec- tural rather than an engineering problem. It is closely associated with the choice of taxes. Repeal of taxes that are clearly unsuitable is commonly counterbalanced not entirely by the introduction of new taxes (if any), but also by increases in the rates of new taxes. Again, these decisions must be taken with the engineering problems in mind. In passing I note that repeal of a certain major tax, in accordance with a tax mission’s recommendation, has commonly been ignored in later appraisals of the degree of success achieved by the mission. In the many critiques of the Japan mission report that I have read (all of them in English, to be sure), I do not recall any that hailed the 1950 repeal of the excess profits tax as a major victory for the mission. Repeal of a turnover tax has been almost equally ignored in appraisals of the Japanese report and, later, the Venezuelan report. For these kind of architectural change, gone is indeed forgotten. The Role of a Tax Mission How should a tax mission allocate its time and effort among architec- ture, engineering, and administration? One view is that the mission will On Designing Tax Systems be most helpful if it concentrates on engineering and administration. Two interesting examples of this view may be cited, one with respect to the Japan tax mission, the other with respect to the Venezuelan mission. Hayato Ikeda, minister of finance in Japan in 1949 and 1950 when the tax mission carried out its work, later wrote an appraisal of the public finances of Japan which included the following comments on the tax mission’s assignment: “The suggestion that we should get Shoup to come [to Japan] originated from about the time when in February of that year [1949] I finally failed in my negotiations with Dodge [the United States adviser to SCAP on how to stop inflation] for tax reductions. I had no particular intentions that Shoup should give me theoretical recom- mendations for the Japanese tax system. The tax system as such was based on a fairly careful study of various European [sic] systems, and its theoretical underpinnings were well developed, so there was no particu- lar need to seek the guidance of foreigners on the system itself. But there was at the time quite a lot of confusion in the operation of the tax laws, that is, in tax administration.” (Ikeda 1952, in James 1987:19). As director of the tax mission, I did not take this view and indeed was not even aware that it was held by Ikeda. Our group did not hesitate to recommend important architectural changes: repeal of the excess profits tax, repeal of the general sales (turnover) tax, introduction of a wealth tax for the national government and a value-added tax for the prefec- tures, as well as structural changes in the income tax and death tax. As to what are here termed engineering changes, | infer that at least some of them were in Ikeda’s mind when he wrote of administration. Still more directly to the point at issue here was the opinion voiced by Enrique F. Gittes in 1968 on the report by the tax mission to Venezuela. Gittes, a research associate at the Harvard Law School’s International Tax Program, asserted that “More often than not, it is in the working out of technical details that expert help is needed; today foreign technicians will not be the first persons in the host countries to suggest broad policy changes such as the elimination of schedular [income] taxes [in favor of a global income tax] or the taxation of dividends.” (Gittes 1968:172). I strongly disagree with the view expressed by Gittes. It is the foreign experts who are likely to have the broad view of what is going on in the world at large, in taxation, and what styles of tax architecture have proved successful for what types of country. In the other Venezuelan tax mission, that to the Federal District of Venezuela in 1959, information was gathered by the mission on the revenue systems of the District of 23 Carl S. Shoup Columbia, New York City, London, Paris, Buenos Aires, and Monte- video and was published in the mission’s report. Moreover, the reasoning supporting one or another kind of tax archi- tecture for a particular country involves sophisticated analyses of social and political influences, and here too the mission expert who has had occasion to study several tax systems has an advantage. Finally, to support the architectural recommendations, the tax archi- tect must review not only the new taxes that are recommended, but also other major possible choices that are deemed unacceptable, with the reasons why. Thus the Venezuelan Federal District report explained why various types of general sales tax would not be appropriate for the district. This broad viewing of all major possible alternatives is likely to be more feasible for those who are well acquainted with a number of tax systems. In contrast to the Ikeda-Gittes view is that of Bronfenbrenner and Kogiku in their 1957 appraisal of the Japan tax mission report. By their conclusion that some of the architectural recommendations should have been different, they imply that this is a proper area for a mission report to cover. Bronfenbrenner and Kogiku pose the conflict between “equity,” viewed as “leaving the distribution of income and wealth more equal after taxes than before, “ and “capital formation” by encouragement of saving and investment (“economic growth”) regardless of equity consid- erations. They appraise the mission report as a “pragmatic compromise between equity and capital formation” (354). This compromise would of course involve both architectural and engineering features of the tax system. My own recollection is that we were much more concerned with both equity, especially horizontal equity, and reduction of excess burdens (economic efficiency) and were willing to let capital formation reach whatever level the market would then call for. In any event this choice of ultimate objectives clearly involved both the general structure of the tax system and dozens of important engineering provisions within each tax. In the main Bronfenbrenner and Kogiku fault the mission for mistakes in architecture. They charge that the report ran counter to certain “vis- ceral attitudes” in Japan: (1) traditionally, they assert, direct taxation was, and indirect taxation was not, associated with “inquisitorial methods of enforcement and with tyrannical government generally” (336), and the mission’s emphasis on the key role of the income tax On Designing Tax Systems seemed to ignore this tradition, or reflect an unawareness of it; (2) in Japan “periods of local independence have been periods of lawlessness and civil war,” and the increase in local fiscal power recommended by the mission seemed to take no account of this historical lesson; (3) the large “undivided family” in Japan was socially beneficial in several ways and “the accession principle in succession taxation flies in the face of this feeling” (337). A criticism implicitly directed at the mission’s engineering is the Bronfenbrenner-Kogiku observation that the mission, or at least a few of its members, should have stayed in Japan for some time after submission of the report to correct “technical defects” (not specified) that were discovered as the reform measures were examined. The Gittes view might seem to have been adopted by the tax mission to Liberia, in its 1969 report, which was devoted largely to engineering recommendations, with few architectural changes; no new taxes were proposed, and those to be repealed were not very important. Suggested changes in tax rates were limited. This result did not, however, reflect any initial desire by the mission or by Liberian officials that the mission’s scope not include tax architecture. Rather, it reflected Liberia’s comfortable budget condition and the gen- eral acceptability of the existing tax structure. Total tax revenue did not need to be increased at once, though the system was thought not to be responsive enough to growth of the Liberian economy. The engineering changes recommended were concentrated, not in the general income tax or in a general sales tax (there was none), but in two areas common to many less developed countries. One was the conces- sion agreements with foreign-owned companies that were extracting large amounts of iron ore from Liberia’s rich mines; these agreements contained many income tax anomalies that reduced the tax bill for the concessionaires. The second area was the network of taxes on imports and exports. Here, indeed, some rate changes were recommended, but there were also analyses of ad valorem versus specific rates, undue differences in rates on closely related products, and great discrimination among firms in the granting of tariff exemptions or reductions. A long list of recommended tariff reductions, although architectural in princi- ple, was tax engineering in its high degree of specificity. Tax administration in Liberia needed reform but was given restricted treatment in the mission report, owing to an expectation that technical assistance from other sources would soon be available. 25 26 Carl S. Shoup Recent Trends In recent years tax missions have been larger and have operated over a longer period of time than those of the earlier days of 1930-60. The number of mission members on hand in the host country at any one time may not be greater than before, but any one person’s time is spread over a longer period, perhaps in installments, and totals more, or the number of persons engaged is greater. Some of the research and writing is done at the individual’s home base. The work of the tax missions that I directed was in contrast completed in a few months, and almost all the members, typically four to seven in number, were present in the host country at the same time. Total man- hours devoted to a mission have probably increased considerably in recent years. How have these changes in mission pattern affected the relative amounts of time spent and report space occupied by tax architecture, tax engineering, and tax administration? My impression is that both engi- neering and administration, but especially engineering, have increased their shares. At least this would seem to be expected. Tax engineering is a very time-consuming process, dealing as it does with a myriad of inter- locking details and often presenting new problems that have to be thought out with little help from the tax literature (even from reports of other missions). Perhaps the time has come for a revival of tax engi- neering handbooks and articles of the kind published by the Harvard Law School’s International Tax Program. Have the larger tax missions been more effective than the smaller ones were? Barring exceptional cases (for example, Japan), the answer is probably yes. But here we have a cause-effect conundrum. Have the larger missions been more effective because, being larger, they have produced more analysis, description, and more detailed recommenda- tions? Or have they been larger because the host country has been more eager to get their help, more receptive to their advice—an attitude that assures a greater chance of success for the mission while simultaneously causing it to be larger? In any event the increased size of tax missions, measured in total man- hours, has eased the task of deciding how to allocate the mission’s resources over the three areas (architecture, engineering and administra- tion). Crucial research in one area does not have to be sacrificed for still On Designing Tax Systems more crucial research in another. Here we come close to the engaging questions: what is the optimum size of a tax mission and spread over what period of time? to which I shall certainly not attempt answers here, except to conjecture that there will be no more one-month tax missions (Liberia, 1969), but no five-year tax missions either. The larger missions are better able to cope with the growing problem of the underground economy (illegal sector, informal economy) in cer- tain developing countries. This problem was not addressed in detail in the earlier mission reports. It calls not only for special measures in tax administration, but also for special provisions in tax engineering, that is, in the tax law, to specify the details of a tax base that is suitable for the informal economy. Involved, for example, are the levels of exemption from an income tax or a sales tax and the minimum-tax provisions (taxable income to be at least x percent of sales, or taxable sales to be at least y percent of inventory). Some such tax engineering seems essential if the underground economy is anywhere near as large as it is said to be in many developing countries. Future Missions and Implications for Tax Engineering What topics are likely to occupy more and more of tax mission resources in the years ahead, and what are the implications for the role of tax engineering? Indexing a tax system for inflation is likely to attract increasing attention in the coming years. As prices rise, bases for depreciation become inadequate, inventory profits may be illusory, a small-firm ex- emption set in terms of sales will apply to so few taxpayers in a short time that the tax administration will become overwhelmed, property tax assessments will commonly lag behind market values, and so on. In developing countries serious inflation seems to be spreading, not retreat- ing, and tax missions may encounter this set of problems more often than in the past. The task of fitting a tax system to cope with inflation is primarily an engineering one. LIFO accounting, tax brackets linked to price-level indexes, and present-value depreciation all require detailed provisions in the tax law. It is also an architectural issue in that the degree of difficulty of coping with inflation differs markedly among taxes; in such an en- vironment a sales tax may be preferred to an income tax. 27 Carl S. Shoup All this will absorb much of a tax mission’s time and energy, diverting resources from more fundamental tax issues—another reason for larger missions than in the past. The missions that I directed were fortunate in not having to construct an indexed tax system (except, to a limited degree, in Brazil in 1964): Japan, 1949; Venezuela, 1958—60; Liberia, 1969. The basic accomplishments of these missions would have been far less if much of their time had had to be directed to the formulation of inflation-coping measures. We need a handbook on how to adapt a tax system to inflation. A web of informal societal obligations that are met by contributions of money from households and business firms is found in some develop- ing countries. This drain on one’s resources contributes to an unwilling- ness or even inability to meet tax obligations. These informal obligations exist in all countries, but, in proportion to the firm’s or household’s means, seem particularly heavy in some developing countries. In Liberia “the tax system must compete with an informal system of more or less voluntary contributions to family, friends, political party, local political chiefs, charitable organizations, and the like. . . . The extended family and, to some degree, various secret societies obligate those who gain to share with others.” These obligations “dampen incen- tive to enter the monetary economy [from the subsistence sector] and to build an economic base upon which there can function a tax system. ... Another example is the strong pressure on Government employees to make political contributions that have on occasion amounted to as much as one month’s salary. Equally serious are the informal and arbitrary levies said to be made by county officials, chiefs, and others, without statutory authority, to finance amenities for traveling officials, to defray fines incurred by the county officials, chiefs, etc. . .. Under these circum- stances it is understandable that the formal tax obligations are relegated to second place (if not lower); the social sanction for not paying taxes may often be less severe than for not participating in the informal contributions.” If “the load of informal contributions . . . continues in its present intensity and extent, meaningful reform of the formal tax system will probably not be possible.” (Shoup et al. 1970:5—7). One is reminded of the extralegal, quasi taxes in the Japan of 1949, “the so-called voluntary contributions which the hard-pressed localities have exacted from their citizens” (Shoup et al. 1949:2, 210-11). The prevalence of socioeconomic payments, or voluntary contribu- tions, affects tax architecture and especially tax administration more On Designing Tax Systems than it does tax engineering. There are few, if any, provisions that can be inserted in the substantive part of the tax law for a particular tax that will lessen the drain on the system caused by this environment. It is primarily an administrative problem, one which is linked to an architec- tural policy that depends heavily on the kinds of tax that the taxpayer does not view as being in close competition with the informal require- ments: the sales tax, for example, compared with the income tax. If tax engineering is to become a larger part of a mission’s task, the composition of the group may change somewhat. The disciplines from which the membership of missions is drawn have been, in the past, economics foremost, and tax law. The directors have almost all been economists. Tax engineering makes more intensive use of more disci- plines than do either tax architecture or tax administration. The tax lawyer and the tax accountant can often suggest technical ideas and technical language that might not occur to the economist. Sociologists or social psychologists might contribute to fit the tax engineering as pre- cisely as possible to social patterns in the particular country’s household and business environments. So far as I know they have not been called on to serve on tax missions, nor have the political scientists, who could have a special understanding of the political limitations surrounding any tax proposal. Tax lawyers helped guide three of the earlier missions (Cuba, Japan, Venezuela), influencing decisions in all three mission aspects, but espe- cially in tax administration. Economists have devoted the most effort to formal analyses of equity, horizontal and vertical, to economic efficiency (absence of excess burden), and to particularized economic goals such as economic growth and stimulus of this or that sector. It seems likely that economists will continue to be the directors of tax missions, though it might be tax lawyers instead, if tax engineering problems demand more and more attention. These conjectures on the assignments and composition of future tax missions assume that such missions will continue to be in demand. From the developed countries’ viewpoint the need for tax missions to develop- ing countries seems not to have abated, given the heavy foreign debt payments that the latter have contracted to make and given the present degree and spread of inflationary finance. Nevertheless, the future of the tax mission is in doubt for two reasons. First, the funds to finance them may not be forthcoming. Those countries that are most in need of a tax mission are the least able to pay 29 30 Carl S. Shoup for it. The financing will have to come from developed countries or from international organizations. There is precedent for this. The Japan mis- sion was financed by GHQ, scap in Tokyo (General Headquarters, Su- preme Commander for the Allied Powers), and the one-man mission to Brazil was financed by usarp. But the two Venezuelan studies and the Liberian were financed by the host countries, as were those, if I recall correctly, to Cuba. The question now is have the potential donor countries, the Interna- tional Monetary Fund, and the World Bank been sufficiently impressed by the record of tax missions to keep their purse strings open? Perhaps. In any case they have not a great deal to lose, since the financing of even a large mission calls for a sum that is rather modest compared with other requests for aid. Second, some of the inflating countries may really consider inflation superior to taxation over the long term. There are important arguments for this view, notably inflation’s absence of specification of who, what social or income classes, shall bear the financing burden. In taxation this pattern, at least as to tax impact if not tax incidence, must be spelled out in detail in the tax law. Agreement on such a pattern is hard to get in a country where social classes are economically and emotionally almost warring with each other. If, indeed, substantial inflation is accepted (though perhaps not explicitly) as a major financing instrument, the tax system, and its faults, are seen to be of minor importance. Why ask for aid to reconstruct and strengthen a means of finance that no one wants to use heavily anyway? Public finance analysts have for too long written as if everyone was against inflation, but of course inflation is a consciously adopted policy and must have a certain appeal. An inflation mission would be more appropriate for some countries—to advise the authori- ties how the inflation might best be structured, to distribute the burden in a certain way, and to achieve certain economic objectives. The archi- tectural, engineering, and administrative aspects of inflation, however, lie well beyond the scope of the present paper. On balance it would seem that tax missions will not become extinct, but may be employed more and more by those developing countries where tax systems are already strong enough to prevent inflation and therefore bear heavily enough on the country’s households and business firms to motivate requests for technical help, especially tax engineering. 3. MALCOLM GILLIS Legacies from the Shoup Tax Missions: Asia, Africa, and Latin America In the early decades of this century the economist Edwin Kem- merer gained worldwide renown, or notoriety, as an international “money doctor” as a result of his financial reform missions abroad, particularly in Latin America (see, for example, Ascher 1985). As one consequence the name Kemmerer is forever linked with the word mis- sion. In the second half of the century the names of two other American economists have become almost as tightly associated with the phrase “tax mission:” Carl Shoup and Richard Musgrave. Perhaps because of their stronger records of academic distinction in economics generally and public finance in particular no one has made so bold as to apply the sobriquet “tax doctor” to either Shoup or Musgrave. I do not propose to do that here, even though their imprints upon tax systems worldwide have been at least as enduring as those left abroad by Kemmerer in banking and finance legislation. The Shoup and Musgrave legacies may be found not only in tax reform laws actually enacted, most notably in Japan (Shoup) and Colombia (Musgrave), but also in the residue of theoretical and applied insights where the proposals of their missions were stillborn, as in Liberia (Shoup) and Bolivia (Musgrave), in the impact of their missions in developing greater tax sophistication among the public in almost every instance, and finally, in the influence of all the Shoup and Musgrave missions upon tax reform programs initi- ated on all continents in the past four decades. The Shoup mission to Japan in 1948 was, however, the prototypical study, and the focus of that mission foreshadowed his approach to tax reform not only in later missions to Venezuela (1958), Brazil (1964), and Liberia, led by Shoup, but in missions organized by others in other countries, including Indonesia in 1981—84. The Shoup mission to Japan was the first example wherein a group of distinguished American ana- Malcolm Gillis lysts had the opportunity to apply their formal skills to the problem of reforming an entire tax system. The very path-breaking nature of this undertaking has led some observers, Japanese and American alike, to offer, ex post, some criticisms of the Shoup-Japan mission not widely shared among the American public finance fraternity. According to Nakazato (chap. 4, this volume) the mission members were “idealistic new deal scholars (who) wanted to test their theory, and (who) imple- mented in Japan what they could not do in the United States.” Similarly, Bronfenbrenner and Kogiku (1957) contended that the “Shoup system included several novel fiscal experiments, which the mission suggested for Japan without benefit of large-scale experience with other countries.” This implies that many Japanese opponents of the Shoup proposals were correct in claiming that the mission was using the Japanese people as “guinea pigs” (Bronfenbrenner 1986), the only admonition ever voiced to Shoup in instructions from the commander of the occupying army, General MacArthur (Shoup 1989b). The Shoup mission report for Japan was also notable for containing the first concrete, detailed proposal for value-added taxation (VAT) in the English language. Indeed, it now seems clear that Shoup and his colleagues developed the vat proposal by stages, without much reference to previous tax literature or tax practice (Shoup 1989b). Later Shoup missions to Venezuela, Liberia, and Brazil bore many of the earmarks of his approach in Japan. In all four missions Shoup’s strong emphasis upon tax fairness and, in particular, horizontal equity was readily evident. Indeed, a few later writers who apparently see tax systems as an appropriate means of promoting capital accumulation have viewed this emphasis as a weakness, rather than a strength, of Shoup’s handiwork (Bronfenbrenner 1989:8; Nakazato, chap. 4, this volume). In all four missions (except that for Venezuela) from 1949 through 1964 Shoup displayed acute awareness of the issues arising from the interaction of tax systems and inflation, an awareness not common among American public finance specialists until almost the middle of the seventies, when the problem became critical in the United States. This was particularly the case in Japan, where the Shoup model for revalua- tion of fixed assets (including taxes on the revalued assets themselves) became the standard for International Monetary Fund (IMF) and other missions to inflationary economies for the next twenty years. One of the hallmarks of Carl Shoup’s long career has been his insist- Legacies from the Shoup Missions ence upon the centrality of good tax administration in public sector finance. To him, tax policy in developing nations has always been the backbone of development policy, but tax administration has always been the key to tax policy. All of the larger Shoup missions (Japan, Venezuela, and Liberia) devoted a large share of their efforts to formula- tion of proposals for administrative changes consistent with the struc- tural reforms recommended. Time pressures and resource limitations precluded Shoup from extensive work on administration in Brazil. The Venezuelan mission, like the Japanese undertaking, utilized lawyers not only in what Shoup (this volume) calls the architecture and the engineer- ing of tax reform proposals, but in administrative dimension of reform. All four Shoup missions operated under extremely tight time constraints, relative to subsequent large-scale missions in Colombia (Gillis and McLure 1978) and recent missions in Indonesia (Gillis 1989a) and Jamaica (Bahl 1989). To illustrate, the work of the seven-man team in the 1949 mission began in May and ended with a set of full-blown proposals in August (Shoup 1989d:180). The Liberia study was com- pleted within three months of its inception in 1969 (Shoup et al. 1970), the Brazil study within a few weeks in 1964. Moreover, Shoup insisted, successfully, on full and immediate publication of all four reports pro- duced by his missions. The Japanese report was published, with twenty thousand copies, within two months after it was completed (Shoup 1989d), the Liberian, Brazilian, and Venezuelan studies within a year (Shoup 1989c). But befitting Shoup’s lifelong stress on learning and education, the later missions also featured innovations not found in the mission to Japan. The Shoup study of Venezuelan taxation was perhaps the first to provide a modern, systematic treatment of natural resource issues in the context of tax reform. The Shoup study for Liberia brought to the tax reform literature among the first applications of innovations in eco- nomic analysis outside of public finance, including topics in effective protection and national income accounting. Subsequent sections examine the origins, approaches, proposals, and results of each of the four Shoup tax missions of the past four decades. Substantially more space is devoted to the Japan mission than the others for two reasons. First, considerably more information is available on the Japanese experience, which has also been critiqued most often. By con- trast, critiques of the Shoup mission to the other countries have been few (Gittes 1968) and relatively mild (Peacock and Andic 1966). Second, 68 34 ~=Malcolm Gillis Professor Shoup already had, in the author’s view, most things right in 1949, with the possible exception of his continued insistence on taxation of the increase in value of revalued assets. Rationales offered for particu- lar reform measures in Japan were also offered, appropriately modified for local conditions, in Venezuela and Brazil. Professor Shoup saw no reason for changing his mind on these questions after Japan; I see no reason for repeating them in the shorter section dealing with these later missions. The Shoup Mission to Japan Background The first Shoup mission to Japan! in 1949 is best viewed against the backdrop of the occupation’s overall economic plan as guided by the Supreme Commander of the Allied Powers (scar), General Douglas MacArthur. Sometimes called the Dodge Line (after the name of Chicago banker Joseph Dodge, invited by MacArthur to end the great Japanese postwar inflation), this plan was, according to Bronfenbrenner (1986), intended “to end the burden which occupied Japan then represented for the U.S. taxpayer.” Draconian stabilization measures dictated by the Dodge Line resulted in a significant success in reducing budgetary defi- cits fueling inflation. This was achieved in large part by a series of occupation tax adjustments involving increases in sales and excise taxes and higher and more progressive rates of income taxes (Bronfenbrenner and Kogiku 1957). Revenues in this bloated tax system were maintained only by tax collection quotas which were revised upward frequently and arbitrarily, in a manner later described by Shoup as “forceful to the point of brutality” (Shoup 1989b), bringing hatred of tax officials that, accord- ing to Bronfenbrenner and Kogiku (1957), rivaled that inspired by the prewar “thought police.” Instances of reassessments in excess of total income were widely reported; the marginal tax rate applicable to the yen equivalent of a $1,000 annual family income was 85 percent (Bron- fenbrenner 1989); corporate taxes and an excess profits tax combined to preclude the replacement, let alone the expansion, of capital within firms (Bronfenbrenner and Kogiku 1957). The case for fundamental tax re- form was therefore clear-cut. In response to a request from General MacArthur the seven-person Legacies from the Shoup Missions Shoup mission began work on tax reform in May 1949. By that time the postwar inflation had been largely tamed. Largely for this reason the Shoup team was able to focus upon the fashioning of a reform that was intended to be essentially revenue-neutral, at least over the first two fiscal years (Shoup 1989b; Gillis 1989b). Reform Proposals The mission recommended four basic types of changes in the Japanese tax system: structural reforms in existing taxes, repeal of unworkable taxes, adoption of new taxes, and measures to improve tax administra- tion and compliance. These are presented in encapsulated form, as details are readily available in recent, accessible publications by Shoup (1989b, 1989d). Structural Reforms: Existing Taxes on Income, Property, and Inheritances Three themes permeate the Shoup report on Japan. The first was the mission’s emphasis upon fairness in taxation, particularly the pursuit of horizontal equity. The second was an aversion to tax preferences of all kinds, a view driven largely by considerations of tax fairness. These themes were most evident in the mission’s income tax recommendations. The third was the mission’s concern that the tax system not interfere with future economic growth in Japan (Shoup 1989b). Tax measures deliberately designed to encourage growth were rejected; reform in- tended to remove tax barriers to growth was seen as the best use of taxation in facilitating growth. Principal changes in income, death, and property taxes included: (a) A reduction in bracket rates for personal income taxes, with the top rate reduced from 85 percent to 55 percent. (b) Substantial increases in personal exemptions and dependent allowances, coupled with a decrease in earned-income credits favoring wage earners relative to farmers and small businessmen. (c) Full taxation of capital gains, to replace a previous exclusion of 50 percent, coupled with averaging for capital gains, as well as full loss offsets. (d) Partial integration of personal and company income taxes, through provision of a dividends-received credit of 25 percent (of dividends). 36 Malcolm Gillis (e) Full taxation of interest income, coupled with denial of interest deduc- tions for consumer credit. (f) Major changes in taxation of business profits of all enterprises, corporate or not, including (i) Revaluation of all depreciable assets, as well as land, to compensate for 1940—49 price level increases of nearly 200-fold, combined with imposition of a 6 percent tax on increases in asset values. (ii) Adoption of unlimited loss carryforward and two-year loss carry- back. (g) Conversion of preexisting estate and gift taxes to a single, cumulative accessions tax on the recipient of gifts inter vivos and property passing at death. (h) Expansion and updating of the base of the real estate tax, to be reverted to the municipalities and entirely administered by them. Proposals for Repeal of Taxes New The implications of the Shoup mission’s success in securing repeal of several types of taxes have been widely overlooked by both supporters and detractors of the mission’s work. The two most significant examples of repeal were a complex, troublesome excess profits tax and an equally troublesome and unworkable multiple-stage turnover, or cascade, tax. Both of these little-lamented taxes had been introduced under the pres- sure of revenue duress in 1947 and 1948. Other notable examples of repeal included a 20 percent withholding tax on dividends paid, as well as assorted indirect taxes on textiles, sugar, and local enterprises. Taxes Three major new tax instruments were recommended by the mission: a net wealth tax, a tax on retained earnings of corporations, and a value- added tax for the prefectures. Both the net wealth tax and the tax on retained earnings were to be imposed at very low rates (one percent) except that much higher rates were proposed for earnings retained after July 1949 by family corporations. Both measures were intended as means of maintaining higher direct taxes on the wealthy with less harm- ful incentive effects than a high income tax rate. The vaT was developed by the mission in lieu of an income tax because of their concern over the cumulative effect of three levels of income tax—national, prefectural, and municipal—and because mission members believed that at the pre- Legacies from the Shoup Missions fectural level a lower rate tax on a broader base might increase business activity (Shoup 1989b). Administration and Compliance The mission concluded early on that, even without reform, total Japa- nese tax revenues might increase anywhere from 25 to roo percent, if all laws had been fully enforced (Shoup et al. 1949:15). And in spite of the fact that the mission viewed details of administrative reform as outside the scope of their work (Shoup et al. 1949), several of the most notable innovations of the 1949 Shoup mission were administrative in character, the most widely known of which was the blue-return system. Other mission proposals for improvement in administration and compliance were also far-reaching. They included the abolition of anonymous bank accounts and bearer shares which presented a major problem in tax enforcement (Shoup, chap. 2, this volume), establishment of provisions enabling taxpayers to litigate administrative claims, a recasting of an outdated system of fines and penalties, the abandonment of involvement by trade associations in tax collections (wherein the associations would bargain for an overall tax payment figure and then allocate liabilities among members), a new system of estimated tax payments, and a shift away from presumptive standards in assessments on farm income. The blue-return system was, however, the centerpiece of this set of recommendations. It was developed by the mission for two reasons: to provide some taxpayer freedom from excessive administrative forceful- ness and to encourage small firms to set up simple books (Shoup 1989b). Any income taxpayer agreeing to keep books and records as specified by the tax administration would be distinguished by filing a return on a form of distinct color (Shoup, 1989d). The color blue was selected following a somewhat whimsical conversation between Shoup and his frequent collaborator from Columbia University, Professor William Vickery (Shoup 1989b). Taxpayers filing blue returns were offered both carrots and sticks to do so. The carrot was that blue-return filers could not be subject to administrative reassessment until after an actual field investigation, and if reassessments were made, specific reasons for it would have to be provided. The stick was that deductions for deprecia- tion, loss carryforwards, and other outlays would be denied to taxpayers failing to maintain adequate books, and therefore ineligible for blue- return treatment. ay 38 Malcolm Gillis Closely related to the mission’s proposals for strengthening tax ad- ministration were a set of recommendations for improving the “tax atmosphere,” and therefore tax compliance. One reason for the poor tax atmosphere in 1949, according to Shoup, was the lack of transparency in the formulation of tax policy and the implementation of tax administra- tion: “when we came, Japanese tax policy was a closed little secret between the Finance Ministry and some members of the Diet. . . . [W]e broke that shell of secrecy” (Shoup 1989b). The mission sought to accomplish this goal through a variety of recommendations: establish- ment of new university courses in tax law and new tax professorships, encouragement of new technical journals related to taxation, promotion of fora on taxation, involving bar and accounting associations, and finally, the formation of a Japanese National Tax Association. Results Among tax analysts, Japanese and foreign alike, there is even today little consensus regarding the long-term impact of the Shoup mission upon the Japanese tax system and upon Japanese economic development in gen- eral. A greater degree of consensus, however, has been present regarding the short-term consequences of the mission, that is to say the first three years following submission of the mission report in August 1949. Exam- ination of the immediate aftermath of the Shoup proposals yields in any case lessons in attempts to plumb the longer-run results. Short-term Results As a set of proposals endorsed by the occupying powers, there was little chance that the government would reject the mission report (Nakazato, chap. 4, this volume; James 1987), even though then Minister of Finance Ikeda made a futile trip to Washington in April 1950 in order to mobilize pressure from Washington against several scap financial proposals, in- cluding aspects of the Shoup report (Bronfenbrenner and Kogiku 1957). Nevertheless, implementation of the recommendations of the mission proceeded in piecemeal rather than immediate fashion, even under the watchful eye of scap (James 1987; Bronfenbrenner and Kogiku 1957). The piecemeal approach was itself inconsistent with the principles of the report. Indeed, the report itself strongly counseled against such an ap- Legacies from the Shoup Missions proach, arguing that most of the recommendations were interconnected: “if any of the major recommendations are eliminated, some of the others will become of less value, or even harmful” (Shoup et al. 1949). Virtually all of the proposals for structural reform of existing taxes were adopted in the year following submission of the report. All of the taxes Shoup marked for repeal were in tact repealed except a tax on sugar (James 1987). The new taxes proposed in the report were also enacted, as well as the proposed tax on net wealth. However, the ap- plication of the vAT was suspended shortly after the tax was enacted (Shoup, chap. 2, this volume). The mission recommendations concerning improvement of tax ad- ministration, tax compliance, and tax atmosphere fared almost as well as the structural proposals in the short run. The blue-return system continues to this day (see below). Moreover, participation of trade associations in tax matters was abolished, as was the goal system for tax collectors. Moreover, income tax returns were made confidential. How- ever, the proposal to abolish anonymously held bank deposits was re- jected. It is noteworthy that this provision was opposed by the business section of scap (Shoup, 1989d). Also, the mission proposals to deny deductions for depreciation for non-blue-return filers was put aside, as well as the mission’s recommendations for liberalizing procedures for securing tax refunds and special record-keeping requirements for wealthy individuals (James 1987). The short-term results of the mission were therefore very substantial, even in the face of quite lukewarm support for the report by the generally pro-American leadership of the Japanese government, the higher bureaucracy, and business interests (Bronfenbrenner and Kogiku 1957). Long-term Results The short-term success of the mission’s work began to unravel with the end of Allied occupation in April 1952. In short order the prefectural VAT was, after having been twice postponed, permanently laid to rest as a result of strong opposition from the Japanese business community, labor groups, and the news media, particularly individual journalists who objected to the bother of keeping receipts (necessary under the type of addition method vaT proposed by the mission). In the view of one long- term observer of the Japanese scene, the media may have played the most important role in derailing the vaT (Bronfenbrenner 1986). 39 40 Malcolm Gillis The capital gains tax on the sale of securities was abolished in 1953, following strenuous objections by business groups (Nakazato, chap. 4, this volume); capital gains on the sale of houses and land continued to be taxable. The tax on capital gains from securities was repealed on grounds that this aspect of the Shoup reforms allegedly hampered the development of the securities market (Shoup, 1989d). Capital gains from sale of securities remained exempt through 1988, although the govern- ment did include in its (ultimately unsuccessful) 1987 reform bill some strengthening of the taxation of such gains (Noguchi 1988). The net wealth tax was used for three years and then discarded, largely on grounds of its very limited revenue productivity (Bronfenbren- ner and Kogiku 1957) and ease of evasion by property owners (Shoup, 1989d). The fate of this levy was perhaps preordained by the rejection of the mission’s proposals on anonymous bank accounts, bearer certifi- cates, and special reporting requirements for the very wealthy (Shoup 1989b; James 1987). The tax on undistributed profits of corporations met a similar fate. It was used for two years but was also repealed, except for family corporations. In retrospect Shoup himself viewed this tax as perhaps “not worth the fuss” (Shoup 1989b). Several other Shoup mission proposals survived essentially intact until the mid to late fifties. The inheritance (accessions) tax designed partly to encourage the dispersion of large estates (Shoup et al. 1949) remained in force until 1958, although its cumulative feature was re- pealed in 1953 (James 1987), and other substantial modifications were made in 1957 (Bronfenbrenner and Kogiku 1957). The report’s counsel against tax preferences (Nakazato, chap. 4, this volume; Bronfenbrenner and Kogiku 1957) of all kinds went largely unheeded after the end of the occupation. The number of income tax preferences including accelerated depreciation and tax-free reserves, grew steadily through 1986, when the preferences became a target of proponents of tax reform in 1987 and 1988 (Shoup, 1989d). Nevertheless a number of important structural and administrative features of the mission’s proposals have survived forty years of legislative tinkering with the tax system. The provision for asset revaluation re- mains in the law in roughly the same form as in 1950 and is considered by even some of the report’s detractors as one of the major achievements of the mission. Nakazato (chap. 4, this volume), for example, considers this provision as perhaps the only positive incentive to capital accumula- tion in the entire report. But Bronfenbrenner (1989) while citing this as Legacies from the Shoup Missions one of six accomplishments of the mission, maintains that it had evolved so far from its original form that it may be inappropriate to call it part of the “Shoup system.” Also, the basic outline of dividends-received credit proposed by the mission has become a permanent feature of the tax system, now supple- mented by a dividends-paid deduction at the corporate level (Homma et al. 1986). Similarly, the Shoup recommendations on loss carryovers remain deeply imbedded in Japanese income taxation. The mission’s proposals for improving administration and com- pliance that did survive the end of the occupation have proven especially durable, particularly the blue-return system. The system proved so pop- ular that it spawned an active Blue Return Association, which even by 1980 involved 40 percent of blue-return filers (Shoup, 1989d). By 1984 over half of individuals with business income as their main income and 90 percent of corporations were using the blue return (Shoup 1989b). Moreover, in 1986 the government went part of the way toward revival of the mission’s original proposal to abolish anonymous bank accounts, when verification of identity was required of all deposit holders with interest bearing accounts (Shoup 1989d). Problems in collecting tax from small businesses, the self-employed, and farmers were addressed in some detail in the mission report. But neither the Shoup mission nor forty subsequent years of tax legislation have apparently done a great deal to reduce evasion and avoidance from these groups. The problem, often regarded as the most serious in the present tax system (Noguchi 1988), now has a name: the 9-6-4 problem, or alternatively the 10-5-3 problem. The ratios refer to inequalities in assessments of salaried income, business income, and agricultural in- come: reportedly, taxation is 90 to 100 percent of salary income, 50 to 60 percent for business income, and only 30 to 40 percent for agricul- tural income (Noguchi 1988). These percentages have been generally verified by a number of researchers including Ishi (1984). Moreover, Kaneko (1985) notes continuing extreme difficulties inherent in taxing farm income, particularly since many farmers grow much more diversi- fied crops today than in 1950. Finally, the Shoup mission proposals for improving “tax atmosphere” have been seen as clearly successful both by the report’s critics as well as its advocates (Nakazato, chap. 4, this volume; Bronfenbrenner and Kugiku 1957; Noguchi 1990; Bronfenbrenner 1989; James 1987). Early on the report helped to mollify the strong antitax sentiment in the 41 42 Malcolm Gillis Japanese public just after the war (Bronfenbrenner 1989). The report’s stress on curtailing what was euphemistically called the “forcefulness” of tax officials and its fostering transparency in tax policy, dissemination of tax information, and the confidentiality of returns contributed to the abatement of such sentiment. Moreover, dozens of tax professorships in universities, the Japan Tax Association, the Blue Return Association, and several technical tax journals are all directly traceable to the Shoup mission proposals. In the words of one of the most consistent critics of various aspects of the mission’s works, the Shoup mission “raised the standards of tax administration in the government, and of tax-return preparation by taxpayers themselves and the professionals they hire” (Bronfenbrenner 1989). Missions to Developing Countries Carl Shoup organized and directed two major tax reform missions and one tax reform study in the fifteen years following his mission to Japan. The first mission was to Venezuela in 1958—59, where Shoup again assembled a team of economists and lawyers to provide options for reform, in this case reform of central government and municipal taxes as well as important aspects of the expenditure system. However, the mis- sion report, The Fiscal System of Venezuela (1959), was much more than a public finance document; it was also a wide-ranging book on critical aspects of Venezuela’s socioeconomic development, covering problems of income distribution, education, trade and natural resource policy, foreign investment, issues in factor choice, and public health. Five years later Shoup journeyed to Brazil, this time alone, at the invitation of the Getulio Vargas Foundation to conduct what was in- tended to be a survey of the tax system of Brazil. But the report, The Tax System of Brazil (1965), went weil beyond a mere survey, to set forth options for new taxes and recommendations for repeal of others. Un- characteristically for a Shoup undertaking, the report involved very little focus on tax administration. The principal thrust of the report turned out to be the identification of tax measures to check, or at least cope with, inflation. The final Shoup-led effort was a five-man mission consisting entirely of economists, a notable departure from the Shoup approach in Japan and Venezuela. The mission’s report, The Tax System of Liberia (1969, published as Shoup et al. 1970) was in fact focused largely on taxes, Legacies from the Shoup Missions duties, and nontax revenues, but there were echoes of the Venezuela study in that almost one fourth of the volume was devoted to a cogent and perceptive analysis of Liberia’s overall development prospects. Principal Similarities and Differences in the Three Missions The differences in population, land area, living standards, literacy, cul- ture, and tax systems between the Venezuela of 1958—59, the Brazil of 1964, and the Liberia of 1970 were at least as striking as any similarities that one might have observed two and three decades ago. Indeed the differences between the Brazil of 1964 and the Brazil of 1988, with average annual real growth of 4.3 percent in the intervening years (World Bank 1988), are at least as notable as those between the other two countries. Change has also come to Venezuela and Liberia over the last quarter century: Venezuela remains an oil-rich nation with relatively high income per capita, not much more diversified now than in 1958 and with sizable international debt problems, while per capita income in Liberia today is appreciably lower than when Shoup journeyed there in 1969 (World Bank 1987). Notwithstanding these differences, there were a number of interesting commonalities in the three tax reform episodes discussed in this section. First, in all three cases, just as in Japan, revenue crisis was not a factor in the request for expatriate assistance in tax reform (Gillis 1989b). This is not to say that the partisans of tax reform who requested Shoup’s assistance in Venezuela, Brazil, and Liberia did not seek revenue en- hancements from reform. Clearly in Venezuela there were four types of revenue motives for tax reform. In 1958, in the days just before the creation of OPEC, when oil prices were low and when reserves of Venezuelan heavy crude were uncertain, it seemed important to have in place a tax system that could be relied upon to supply large increases in non-oil taxes if and when petroleum revenues dwindled. This also was the initial motivation for the Indonesian tax reforms of the 1980s: install a tax system capable of revenue enhancement in a future that may hold lower oil prices (Gillis 1985). The old Venezuelan tax system, consisting primarily of internal indirect taxes—particularly sumptuary excises—could clearly not be counted upon to replace oil revenues should they eventually decline (Shoup et al. 1959). A second revenue-related motive for reform in Venezuela was to 43 44 Malcolm Gillis finance a major expansion in public expenditures for human capital: in education primarily but also in public health. An expanded educational program was seen as particularly vital given that only 50 percent of the school-age population was in school (Shoup et al. 1959). Also, while there was no immediate revenue crisis, there were impor- tant medium-term revenue needs to be addressed, since the national government expected a sizable deficit in 1958—59. Finally, unlike many reform programs of the 1980s, an explicit goal of the Venezuelan effort was to increase progressivity, at least in some limited degree. Goals sought in the author’s Brazil report of the mid-sixties were not so explicitly expressed, partly because there was no clear governmental commitment to tax reform at the time. But here again, there was no short-term fiscal crisis to confront. Indeed, 1964, a year of moderate (for Brazil) inflation, may have been the only year in recent Brazilian history when the government budget deficit was not viewed as a problem requir- ing urgent attention. Rather, Professor Shoup was requested to identify problems in taxation and resource allocation, taxation and income dis- tribution, multilevel government finance, and tax structure issues in general. Liberia also faced no immediate threat of fiscal crisis in 1969; in- creases in tax rates for existing taxes were not an available option in any case, for reasons discussed at length in the report (Shoup et al. 1970). Instead, the tax mission focused heavily upon restructuring the tax system to achieve equity, resource allocation, and economic growth goals. A second similarity among the three cases was the manner in which the mission reports were utilized early on. In none of the three cases did the work of the mission have any significant immediate impact on tax policy, but in two instances (Venezuela and Brazil) there were important longer-term results traceable to Shoup’s work. In Venezuela virtually none of the mission’s recommendations were adopted over the next seven years (Shoup 1989c) with the one exception noted below. But, as in the case of Colombia in 1968, delayed reform was not lost reform (Gillis and McLure 1978). The Shoup 1958-59 mission proposals provided the foundations for a fundamental reform of the Venezuelan income tax in 1966. Since 1966 a number of the mis- sion’s proposals have been enacted but on a piecemeal basis. These are noted below. Still, one may view the results of the 1958—59 Venezuelan mission with mixed feelings. Even today, thirty years after the Shoup Legacies from the Shoup Missions mission, Venezuela remains ripe for fundamental tax reform (McLure, chap. 5, this volume). The Brazil proposals experienced a broadly similar fate. Except for the state-level VAT, only small bits and pieces of the Shoup program were adopted over the ensuing twenty-three years. But if the Brazilians made only very selective use of the Shoup proposals, they have made even less use of other tax reform recommendations over the past quarter century. With one very small exception, no new taxes other than those recom- mended in the 1964 Shoup report are in force today. And Brazil has not repealed any major taxes other than those Shoup recommended for repeal in 1964 (Shoup 1989c). Finally, the Liberian reforms were stillborn in almost every sense. There the failure of the tax mission to influence policy seems almost total. Shoup (1989c) has recently offered several reasons why that might have been so. One final similarity between all three programs was that in each case Professor Shoup agreed to undertake the projects only after receiving assurances that all tax reform studies worth publishing would in fact be published in timely fashion. As in Japan this was done to facilitate open and wide-ranging discussion of tax reform issues in each of the countries. There were also significant differences between each tax reform un- dertaking. First, the Venezuelan reform proposals were the product of a team of well-known economic and legal specialists in taxation and government spending. The composition of the Liberian team has already been noted. By contrast the Shoup report on Brazil was a one-man show from start to finish. The differences in resources made available for each of the three reform efforts is reflected in the coverage of the reform proposals. The work of the Venezuelan and Liberian missions extended into several areas not heretofore covered by tax reform programs: examples include analysis of implicit taxes arising from the Venezuelan system of exchange controls, the analysis of rates of return to investments in education, and, in both Venezuela and Liberia, an assessment of the role of government accounts in compiling national income accounts. Moreover, the Vene- zuelan effort, but not the Brazilian and Liberian studies, involved a strong focus on reform tax administration, while in both Venezuela and Liberia, but not Brazil, substantial attention was paid to reform of customs duties. Also, the Brazilian report involved much heavier stress upon struc- 45 46 Malcolm Gillis tural issues of income taxation than was the case for either Venezuela or Liberia. Reasons for this relative emphasis likely had more to do with the time and resources available to the author than with his preferences. Finally, two very distinct features of the Brazilian study were virtually absent from the reports on Liberia and Venezuela. First, the Brazil study included, for its day, a rather extensive discussion of issues involved in inflation-proofing income taxation. Second, only in Brazil did Professor Shoup delve deeply into consideration of a value-added tax as a major tax reform option. The remainder of this section provides sketches, rather than detailed discussion, of the principal results of the Shoup missions to these three developing nations. The Venezuelan tax system is in any case much more fully analyzed elsewhere in this volume (see McLure, chap. 5, this vol- ume), while Shoup himself has recently provided a detailed accounting of the substance of each of these reports (Shoup 1989c). Venezuela The single most important structural change proposed in the mission’s report was, by Shoup’s own testimony, the transformation of the income tax structure from a schedular system to a unified tax on individuals and a separate income tax on corporations (Shoup 1989c). This reform was enacted seven years after the Shoup report and still remains in force. Other important features of the Shoup income tax proposals that eventually found their way into Venezuelan tax law in 1966 and in later years included the mission’s proposals on capital gains taxes, on the taxation of foreign source income, on loss carryforward, and a few administrative reforms, including self-assessment of income taxes. The general thrust of virtually all of these proposals bore a good bit of resemblance to those recommended for Japan nine years earlier. The only immediate measure undertaken in response to the mission’s report was the abolition of the general sales tax, a grossly defective form of turnover tax with the usual cascade features. This tax was repealed in 1958. No changes in response to the mission’s report were made in the structure or the administration of customs duties, nor were any of the recommended changes ever adopted for integration of income taxes, par- tial exclusion of interest and rent income, limits on deduction of personal expenses, field audits, or “tax atmosphere” (tax associations, promotion of technical tax studies, provision of tax information to the public). Legacies from the Shoup Missions Brazil The most important elements of the Brazil study, by Professor Shoup’s Own account, were proposals for repeal of four important taxes, the implementation of a value-added tax for the states, several changes in income taxation, particularly to help check inflation, and certain changes in state and local taxing powers. While neither the present author nor Professor Shoup (Shoup 1989c) claim any cause-effect linkages between the more important Shoup pro- posals and the reforms actually undertaken, some of the subsequent Brazilian legislation seems geared closely enough to the Shoup proposals to justify discussion here. The Shoup proposals for repeal of an excess profits tax, a system of compulsory loans, schedular income taxes, and a legal-entity tax on sole proprietorships were accepted in part with varying lags. Both the excess profits tax and the compulsory loan system were abolished between 1964 and 1975. A significant vestige of the schedular income survives today; although gross income of individuals is distributed among eight schedules, the same tax rate schedule applies to all categories of income (Shoup 1989c). Moreover, the legal-entity tax not only remains in force, but the applicable rates have been increased at least twice. Perhaps the most significant result of the Shoup study was the repeal, in 1967, of the state-level cascade turnover tax and its replacement by a value-added tax. While Shoup himself has been at pains to avoid attribu- tion of this result to his 1964 efforts, other observers, including the present author, view the Shoup study as the clear inspiration for the adoption of the world’s first system of value-added taxation at the subnational level. In any case the state-level vats in Brazil were the first operating examples anywhere of value-added taxes extending through the retail level (Gillis, Shoup, and Sicat 1989), and Professor Shoup clearly had a hand in this occurrence. Liberia The Shoup team in Liberia focused its energies in 1969 upon a variety of issues ranging from tax revenues from foreign investment in the natural resources sector, real estate taxes, import taxation, export taxes, and excise taxes. The commission recommended no major changes in per- 48 Malcolm Gillis sonal income taxes and generally discouraged initiatives for installing ambitious broad-based indirect taxes, including value-added taxes (Shoup et al. 1970). Virtually none of these recommendations were accepted then or in subsequent years, although no broad-based indirect taxes. have been enacted. The value of the mission, however, is not to be judged by the degree to which its proposals were implemented but by the calibre of analysis in the mission’s report. The report provided valuable insights into approaches both to complex tax issues such as the tax treatment of multinational minerals enterprises as well as deceptively simple issues in the design and collection of per capita (“hut”) taxes. Indeed, in the opinion of the present author, sensible discussion of the major issues in taxation of foreign concessionary enterprises date from the Shoup report of 1969. Conclusion The legacy of Carl Shoup’s long career, now approaching seven decades, to economics in general and public finance in particular has not been the focus of this paper. If it had been, the author would have been amply justified in describing him as one of the four central figures in public economics of this century, still publishing at a rate that would do proud any assistant professor scheduled for a tenure decision the following year. The legacy of the Shoup tax missions may be less easily limned than the full body of Shoup’s voluminous work. The question may be ap- proached in several ways. Are the accomplishments of these missions best assessed by toting up the number of important proposals that were ultimately enacted into law or those that remain in force decades later? If that is to be an acceptable measure, then no applied social scientist can lay claim to greater effectiveness than Professor Shoup. This is clearly so even by standards used by critics, usually friendly, of the various Shoup missions, whether in Japan (Bronfenbrenner and Kogiku 1957; Bron- fenbrenner 1986 and 1989; Nakazato, chap. 4, this volume) or in Vene- zuela (Gittes 1968; Peacock and Andic 1966). Critics of the Shoup missions to Brazil and Liberia, if any, have been unusually silent. Are we to believe Noguchi (1990), who views the basic structure of the Japanese tax system as essentially “unchanged since the Shoup re- Legacies from the Shoup Missions form of 1949,” or Nakazato (chap. 4, this volume), who contends that even by the mid-1950s the “Japanese tax system bore very little resem- blance to the system the Report had recommended”? Or perhaps Bron- fenbrenner and Kogiku were correct in 1957 in stating that through 1956 “the Japanese national tax system retained the essential features of what one newspaper called the ‘patched and tattered’ Shoup plan, while the local tax system had departed further from the Shoup framework.” The Bronfenbrenner of 1989, however, describes the present Japanese tax system as a mixture of pre-Shoup, Shoup, and post-Shoup elements, while describing as an exaggeration the label “Shoup system” when applied to the existing laws. Nakazato, while noting that the Shoup report is “the orthodoxy for some, the dream for others and for still others a myth,” nevertheless concedes that the report “has directly or indirectly influenced Japanese tax policy for years, that even if the tax system based on the report was short-lived, the Report’s intellectual impact was tremendous.” Ishi (1987) argues that the present tax system has retained major features of the Shoup framework and would be considered by many as “the most successful” tax reform program in the world. Pechman and Kaizuka (1976) conclude that while the Shoup system was quickly discarded, major legacies nevertheless arose from the mission’s work, in making income taxation acceptable as the basic source of revenue in Japan, in raising the level of tax sophistication, and in improving tax administration. These observations, plus a little reflection, provide ample support for the following conclusion: what remains or does not remain of the par- ticulars of the original mission proposals is essentially immaterial. The essence of the tax principles and the administrative and tax atmosphere recommendations espoused in the Shoup report are deeply imbedded in the present tax system. And at least one of the innovations, or what Bronfenbrenner and Kogiku (1957) disapprovingly called “novel fiscal experiments,” proposed by the mission have resurfaced time and again. The specter of the value-added tax continues to haunt Japanese politics, having been raised and toppled on two separate occasions, in 1977 and 1988 (Noguchi 1990). More recently a new low-rate (3 percent) value- added tax adopted in April 1989 contributed to the downfall of a premier and heavy losses for the governing Liberal Democratic party in elections to the upper house of the Diet in July 1989. The legacy of the Shoup missions can be assessed, however, in yet another way: the influence of the four Shoup reports upon later tax 49 50. Malcolm Gillis reform programs in several LDCs, including Colombia, Bolivia, Indo- nesia, and Jamaica. The successive Shoup missions clearly provided important lessons not only in organizing for reform, but in the attempted scope of reform efforts, timing of reform, professional orientation, and composition of mission members (Gillis 1989b). Perhaps more impor- tantly, after Shoup’s early efforts in Japan and Venezuela no self- respecting mission chief could afford to ignore the critical administrative aspects essential for the success of any tax reform program. That these lessons were taken seriously by later organizers of tax reform missions is a tribute to Professor Shoup’s talent for distilling the essence of careful, first-class scholarship into innovative and workable tax policy options. Moreover, he has always been at pains to insure that future architects of reform learned as much from his successes as his putative failures (Shoup, 1989c, 1989d, and chap. 2, this volume). The true significance of the four Shoup missions, then, lies not so much in the actual measures enacted or not enacted in each country, but in the effects of this body of work in defining the agenda for tax reform in dozens of other nations. Notes 1. There were in fact two Shoup missions to Japan, in 1949 and 1950. The second mission was much smaller in size and scope than the first. It is discussed in some detail in Shoup (1989d), but is examined only peripherally here. 4. MINORU NAKAZATO The Impact of the Shoup Report on Japanese Economic Development For some The Report on Japanese Taxation by the Shoup Mission (hereinafter referred to as the Shoup report or the report) is the ortho- doxy. For others it is a dream for the future. And for still others it is a myth. The Shoup report has directly or indirectly influenced Japanese tax policy for nearly forty years. Much has been said about it in Japan. Whenever tax reform is discussed, commentators compare the tax sys- tem of the time with the “idealistic” one in the Shoup report. Little has been written, however, about how the Shoup report actually influenced Japanese economic development. This article thus focuses on the influ- ence of the Shoup report on the postwar Japanese economy. Most commentators evaluate the Shoup report from the point of view of whether it was truly accepted in Japan, whether it had a long life, or whether it ever truly took effect (see Shoup 1989d:207—18). From this perspective they conclude that the Shoup report was “only a limited and partial success” (Bronfenbrenner and Kogiku 1957:237). But is it ade- quate to evaluate the Shoup report by its degree of implementation? My impression is that these commentators are only observing the surface and only from a short-run view. Consider, in this respect, Professor Martin Feldstein’s distinction be- tween “tax reform” and “tax design”: [T]he optimal reform of an existing tax system differs from the optimal design of a new tax structure. Even if a uniform Haig-Simons tax were the appropri- ate standard for de novo tax design, some piecemeal changes in that direction would reduce efficiency. (Feldstein 1976a:123—24) Tax reform is a change from the existing tax structure. In practice, tax reform is piecemeal and dynamic in contrast to the once-and-for-always character of tax design. . . . [T]he optimal tax reform depends on the starting situation. Equivalently, the optimal tax laws for next year are not the same as 52 Minoru Nakazato they would be if taxes were being introduced for the first time. Optimal taxation depends on the historical context. (Feldstein 1976b:90—91) We could say that most commentators, however unintentionally, use this distinction between “tax design” and “tax reform.” They think that the Shoup report was an idealistic tax design but had only limited suc- cess as tax reform as it was accepted by the Japanese legislature for a very short time. There are several problems with this approach. First, those who believe in the superiority of the consumption tax question the tax design of the Shoup report, with its inclination toward the Haig-Simons defini- tion of income. Second, as tax reform, we should evaluate the Shoup report not by the fact that it was short lived, but by the impact it had on the economic conditions that followed. Whether it lasted long or not is not necessarily important. Instead, we should examine whether the publication of the Shoup report at that time and under those particular circumstances was appropriate. In this paper I focus on the tax reform aspects of the Shoup report and examine whether its recommendations were appropriate at that time and the impact that the Shoup report had on Japanese economic de- velopment. I discuss the tax design aspect of the Shoup report only secondarily. Tax Reform Before the Shoup Report The individual income tax was introduced in Japan in 1887. As a schedular tax, it applied only to the income types which were listed in the law (for example, capital gains were not taxed). Between the tax’s introduction and 1945 several reforms were made. The most important occurred in 1940 when the individual income tax was divided into two categories: a schedular tax in which the flat rate differed according to income category and a progressive tax on general income.! The income tax became an important revenue source after this reform (Hayashi 1974:207). The Japanese government made several attempts to reform the tax system between 1945 and the Shoup report (see Kaneko 1983:1—15).2 After World War II a wide range of political, social, and economic reforms were undertaken by the occupation to democratize Japan. For example, there was land reform (Eyre 1956:113—20) and the dissolution of zaibatsu conglomerates (Bisson 1954; Hadley 1970). The economy The Shoup Report and Japanese Development was in a shambles (see Japan 1949:330—33) and inflation was out of control (Kato 1974:3—26).? The occupation forces began to change the basic structure of the Japanese tax system (see Bronfenbrenner and Kogiku 1957:237—38; Shavell 1948). In 1947 a progressive, global income tax was established in Japan for the first time, capital gains were also taxed, and a self- assessment system was instituted. The tax reform was a landmark change from an European-style schedular tax to a global, progressive income tax (Kaneko 1983:11).4 This tax reform was a bridge from the prewar tax system to the tax system under the Shoup report and con- stitutes an early experiment with some of the tax provisions the report would later recommend (Kaneko 1983:16; Hayashi 1974:210). Corporate income began to be taxed in 1899. At that time dividends were not taxed to individual or corporate shareholders. Since 1920, however, dividends became taxable without any attempt at integration. In the tax reform of 1940 the corporate income tax became entirely independent from the individual income tax. The biggest change made after World War II and before the Shoup report was the introduction of the dividends-received credit of 15 percent in 1948 (Hayashi 1974:211, 240-42). In short, by the time of the Shoup report the reforms of 1940 and 1947 had already transformed the Japanese income tax system into quite a modern one. Most particularly, the individual income tax reflected the postwar American tax theory reasonably closely. The Japanese tax system before the Shoup report had some serious defects. One study, for example, cites unfair administration, not enough incentives for capital accumulation, and the excessive centralization of tax revenue (Bronfenbrenner and Kogiku 1957:237—38). Another study points to administrative difficulties, overly high tax rates, and a shortage of tax specialists (Hayashi 1974:216—21). The tax reforms after World War II and before the Shoup report, especially those in 1947, amounted to tax increases under inflation (Kaneko 1983:8; Bronfenbrenner and Kogiku 1957:237). It was in this situation that the Shoup report was issued. Recommendations in the Shoup Report The Shoup report reflected the most advanced American tax theory of the time.> As the report covered the entire tax system, it is impossible to 53 54 Minoru Nakazato describe it completely here (see Shoup 1989d:182—95; Kaneko and Gomi 1988:148—50; Hayashi 1974:224). Instead, this paper concen- trates on the features which became important in later discussions of the Japanese tax system. The Individual Income Tax Fairness is basic to the Shoup report. One of the report’s principal aims was establishing a fair tax system (Kaneko 1988:60). Because it is difficult to consider a taxpayer’s ability to pay with indirect taxes (Ka- neko 1988:60; Hayashi 1974:227, 230), the Shoup report tried to estab- lish direct taxes, especially the individual and corporate income taxes, as the main revenue sources in Japan. The report thus strongly recom- mended the use of a comprehensive income tax base (Kaneko, 1983, pp. 19-22). We can point out several factors in this respect. First, the report emphasized the idea that a dollar is a dollar (Shoup et al. 1949, 1:12—13).® At the time the idea that a dollar is a dollar was unfamiliar to Japanese tax specialists. Today many commentators still argue that capital income should be taxed more heavily than labor income, even though this argument contradicts the idea that income tax is a personal tax imposed according to a taxpayer’s ability to pay. Second, capital gains taxation in the report was very strict. Its basic idea was full taxation of realized capital gains and full deduction of capital losses (Shoup et al. 1949, 1:91). A sophisticated method of averaging, using a progressive tax rate and the realization principle, was proposed (Shoup et al. 1949, 1:90, 3:B1—B19). Gifts and bequests were considered realization events (Shoup et al. 1949, 1:92, 3:B1z). Little consideration was given to the effect of this capital gains taxation on capital accumulation, however, except for the revaluation of fixed assets. Third, the Shoup report attacked the Japanese tradition of taxing interest income separately with a flat rate. The report also denounced attempts to promote savings by giving preferential treatment to interest income. Instead, the report “recommended that interest of whatever kind be included in individual incomes in full for tax purposes” (Shoup et al. 1949, I:101). Fourth, strangely enough, nothing was said about an interest deduc- tion, although it is theoretically necessary to deduct all interest payments under a comprehensive income tax (see White 1959:366; White and The Shoup Report and Japanese Development White 1977; Bradford 1986:41—42). Thus, in Japan only interest paid on business or investment debt has been deductible, and the interest paid on personal debt (debt incurred for present consumption or buying consumer durables or houses) has never been deductible. Fifth, the Shoup report recommended that individuals rather than households be the taxable unit: “It is recommended that the aggregation of the income of the ‘co-living family’ be abandoned, and that each taxpayer be permitted to file a return and pay a tax on his income separately” (Shoup et al. 1949, 1:73—74; see also Kaneko 1983:40). This recommendation on the taxable unit was to cause serious diffi- culties later. The Corporate Income Tax The Shoup report denied the independent status of corporations as taxable entities and recommended the integration of individual and corporate income taxes by allowing shareholders a credit for dividends received. It also recommended the exclusion from corporate income of intercorporate dividends. Both measures were necessary for avoiding the double taxation of dividends. In short the report tried to restore tax neutrality between corporate and other forms of doing business. The report’s position on the nature of corporations caused disputes among several Japanese scholars who found it difficult to understand why corporations should not be independent taxable entities (see Hay- ashi 1974:238—44; Hatakeyama 1975:212—15). This may have been largely because of the Marxist tradition among prewar Japanese econo- mists. Capital Accumulation The report contains some measures intended to promote, or at least measures that did promote, capital accumulation (Hayashi 1974:252). The reduction in the income tax rate and the introduction of the divi- dends-received credit were such measures. Perhaps the most important measure was the revaluation of fixed assets (see Kaneko 1983:50-53; Bronfenbrenner and Kogiku 1957:239—40). After the war Japan experienced rampant inflation. Be- cause firms could base their depreciation allowances only on the book 56 Minoru Nakazato value of their assets, their taxable income did not reflect real economic profit. Instead, taxable income reflected the nominal profits unadjusted for the real capital costs of producing the goods sold. A similar problem existed in the context of capital gains. Firms had to pay tax on nominal income caused by inflation when they sold their assets. The Shoup report recommended revaluation of assets, with a 6 per- cent tax levied on the revaluation write-up (Shoup et al. 1949, 2:123- 25, 1:92). This “was designed to allow accumulation of adequate de- preciation reserves and to prevent the draining away of these reserves in income tax” (Bronfenbrenner and Kogiku 1957:239). Generally, however, the Shoup report’s attitude toward capital ac- cumulation was less positive. This attitude was shown in the denuncia- tion of tax incentives on the ground of tax equity (Kaneko 1988:63). Moreover, even the tax rate reduction and the dividends-received credit were measures that the report largely advocated not so much to preserve the neutrality between consumption and savings as to promote equity. The Politics of the Shoup Report The Shoup report was forced on the Japanese government by the General Head Quarters of the Allied Powers. The only practical choice the Japanese government had was to accept it. From the point of view of the Japanese government the Shoup report “came from the sky.” It was a given factor. Thus, in the 1950 tax reform the government adopted almost all the recommendations (Shoup 1989d:195—96). Idealistic New Deal scholars wanted to test their theory, and they implemented in Japan what they could not implement in the United States. Since the report was not a product of power struggles among Japanese interest groups,’ not surprisingly, soon after the report was imple- mented, various interest groups began to try to change the system (Shoup 1989d:196).8 Business circles particularly tried to lower the effective tax on capital income. The following changes were later made: (1) the exemption for individuals of capital gains from the sale of securities; (2) the exemption or the separate taxation at a lower rate of interest income for individuals; and (3) various preferential measures in business income taxation, such as provisions for accelerated depreciation and tax-free re- serves (Hayashi 1974:253—58; Bronfenbrenner and Kogiku 1957:241- 45; Kaneko and Gomi 1988:150). The Shoup Report and Japanese Development Through the first two changes the government at least partly intro- duced a consumption-type income tax (of the labor income tax type), which promoted capital formation. Through the third it promoted the investment of the capital accumulated in productive activities.? The Economic Effects of the Shoup Report Every tax system affects the national economy; no doubt, so did the 1950 tax system implementing the Shoup report. It would be quite difficult, however, to fully evaluate the effects on the Japanese economy because the measures were in effect for such a short period. By the mid-1950s the Japanese tax system bore little resemblance to the report’s recommendations. It is possible, however, to study the effect on the Japanese economy of some aspects of the Shoup report. Not all the changes were short lived, and those that survived must surely have influenced the economy (even if in different ways than the drafters envisioned). For example, the influence of the revaluation of fixed assets on Japa- nese capital accumulation can be easily seen. It would be hard to deny that revaluation helped businesses in the severe inflationary period in the early 1950s,!° and it seems to have done so without significantly distort- ing investment decisions. Yet this revaluation of fixed assets may be virtually the only positive incentive toward capital accumulation in the entire Shoup report. Saving and Investment Incentives Income Taxes and Consumption Taxes Modern scholars increasingly support a consumption tax rather than a comprehensive income tax on the grounds that the former does not affect a taxpayer’s choice between consumption and savings. A labor income tax which exempts capital income has essentially the same effect as a consumption tax (see Andrews 1975:953—57). On this Professor Kai- zuka says: These divergences . . . [that is, the exemption of capital income which replaced the system recommended in the Shoup report] have converted the Japanese income tax virtually to a highly progressive tax on wages. In certain 58 Minoru Nakazato conditions, a wage tax is equivalent to a consumption tax; in effect, therefore, the Japanese personal income tax has diverged considerably from a com- prehensive income tax and instead has been approaching a consumption tax. (Kaizuka 1988:163—64; see also Shoven and Tachibanaki 1988:52) Thus, it might be possible to say that the broad-based income tax that the Shoup report recommended was less neutral than the Japanese in- come tax system later implemented in the wake of the report which taxed capital income more lightly. The Shoup report showed very little interest in positively stimulating capital accumulation or economic development. Its main goal was eq- uity (Hayashi 1974:244—45). It used the notion of horizontal equity to denounce the idea of taxing capital income more lightly than the other income. For example, during a visit to Japan in 1988, Professor Shoup noted in his interview with Professor Mark Ramseyer of UCLA School of Law: “(T]he importance of having a tax system acceptable to the public in terms of equity was in our minds all the time. We did not consciously try to shape the tax system to encourage economic growth of this kind or that kind. We felt that was best left to the marketplace to decide” (Shoup and Ramseyer 1989:2). The Shoup report’s ideal was a comprehensive income tax. The report obviously thought that an income tax was fairer than a consumption tax, but it did not give any adequate explanation why. Of this kind of attitude, Professor Richard Musgrave wrote: “[A] great deal of work has been done on refining the broad-based income concept and exploring how it should be applied in practice. . . . But income tax theorists have been less successful in explaining just why accretion (assuming that it can be implemented) should be the best index of equality. Instead, its superi- ority has been taken largely for granted” (Musgrave 1983:302; see also Stiglitz 1982:27).1! In fact, Professor Feldstein showed that a com- prehensive income tax can itself be a source of horizontal inequities (Feldstein 1976b:86—89). Whatever the merits of Professors Musgrave and Feldstein’s points on equity, however, income taxes of the sort advocated in the Shoup report are significantly biased against savings. As a result, most scholars agree that a consumption tax is more neutral than an income tax. The Interest Deduction The Shoup report did not mention the deduction of interest expenses, although a comprehensive income tax base theoretically requires the The Shoup Report and Japanese Development deduction of all interest payments including interest payments on debts for consumption. Thus, Japan has not allowed the deduction of interest payments on debts incurred for present consumption or for the purchase of owner-occupied houses. There have also been some limitations on the deduction of interest payments on investment debts. For instance, no deduction for interest payments on debt incurred to produce “interest income” is permitted, and net losses from “dividend income” and “mis- cellaneous income” are not deductible from other categories of income (all terms as defined in the Income Tax Law). These limitations have had the following effects. First, the denial of interest payments on debts incurred for consump- tion does not offset the savings-reducing effect of a comprehensive in- come tax: It is true that such conformity of taxable income to Simons income implies a tax law that is not neutral in its treatment of consumption and savings but rather is biased against the latter. But to disallow deduction of interest cost on consumer debt would not remove this un-neutrality since it is inherent in the definition of income. Instead, the disallowance would introduce another un- neutrality, a tax law bias favouring asset finance over debt finance. (White and White 1977:4—5) However, this “non-neutrality with respect to the consumption-savings choice can be avoided by exempting positive earnings on assets and disallowing the deductibility of interest on debt” (White and White 1977:6). By exempting capital income from tax and limiting the interest deduction, the Japanese tax system, post—Shoup report, did just that. Second, the Japanese restrictions on the interest deduction have made “tax arbitrage” more difficult. With the preferential treatment for certain types of income (such as a capital income exemption), taxpayers may decide to exploit the treatment by borrowing money. Borrowing to invest in preferentially taxed projects while deducting the interest pay- ments from fully taxed income is called tax arbitrage (see Steuerle 1985). Through arbitrage a taxpayer can reduce taxable income by deducting interest payments and investing the borrowed funds in tax-preferred investments. By limiting the interest deduction the Japanese tax system minimizes this tax arbitrage. Third, no matter how much money people deposit in banks, the funds will not promote capital accumulation if that money is lent out to households for consumption. The denial of the deductibility of interest payments on debts incurred for present consumption or for the purchase Sy) 60 Minoru Nakazato of owner-occupied houses has had the effect of reducing such noninvest- ment debt. Thus, when people bought houses, they first saved money for large down payments and benefited from the interest-income exemption. Once house owners, they then tried to pay back the debt as soon as possible because the mortgage interest was not deductible. ° In short, the limitation on the interest deduction, although not ex- plicitly recommended in the Shoup report, has probably stimulated savings in Japan (Nakazato and Ramseyer 1987). Savings Patterns Under the Bretton Woods system after the war, the exchange rate was fixed at 360 yen to the dollar, and the yen was, at first, overvalued. Although this exchange rate impaired Japanese exports at that time, the Japanese government tried very hard to defend this rate by not borrow- ing from foreign countries and thus avoiding huge current account deficits. At the same time, in order to provide capital to the private sector, the government tried to avoid government deficits (Lincoln 1988:73—75), and stimulate households savings. During the 1950s and 1960s these policies enabled the government to supply favored indus- tries with a relatively cheap source of investment capital. Because of the exemption of interest income and the limitation on the interest deduction, the Japanese savings rate has stayed quite high compared to other industrialized countries (see Boskin and Roberts 1988:121—43). Of course, the tax system is hardly the only factor that influences a nation’s savings rate. There are other factors which might explain the high savings rate in Japan, such as “low levels of social security benefits, privately borne educational costs, a lack of consumer credit, the bonuses paid twice a year by corporations to their employees, and the high cost of housing (and the relatively high down payments required)” (Lincoln 1988:78). Still we should not deny the importance of the tax system (see Homma, Maeda, and Hashimoto 1987:416). And “we know that in the absence of international capital flows and govern- ment deficits a savings incentive would also be an investment incentive: since saving must equal investment, one cannot encourage one without encouraging the other” (Bradford 1986:209). Besides a high savings rate, Japanese household savings patterns have been characterized by the tendency for households to hold financial assets rather than real assets. It may be possible to explain this through The Shoup Report and Japanese Development the tax system. For several reasons the tax on capital gains from the sale of houses and lands has been heavier than the tax on other types of investment income such as capital gains from the sale of securities and interest income. First, the strict capital gains taxation regime recommended by the Shoup report has remained in place virtually only for gains from the sale of houses and lands, except that under the postreport regime such gains are taxed separately from other income. Second, in the computation of taxable capital gains from the sale of owner-occupied houses, the tax- payer’s basis is reduced by the amount of “depreciation,” even though this “depreciation” cannot be deducted from taxable income. In effect the system partially recaptures the nontaxable imputed income from owner- occupied houses (see Epstein 1971:457—59). As a result, fewer taxpayers have chosen to borrow additional funds and invest in houses in the way that Europeans and Americans have tended to do. Instead, more peple have invested their money in financial assets (Tajika 1986:29). In other words household savings have tended to flow into financial intermediaries and from there into the business sector for their invest- ment (Tajika 1986: 30). As a result, the strict capital gains taxation regime in the Shoup report, remaining as it does only in the case of houses and lands, has combined with the preferential treatment of other capital income to promote investment in the industrial sector. A third reason is that Japanese households “rely heavily on bank time deposits in their portfolios of financial assets” (Lincoln 1988:133). Thus “the exemption from tax of much household interest income provides corporations, which are able to deduct interest expense, with a consider- able advantage from highly leveraged, debt-financed investment, . . . [and] opens up the possibility that capital income may be taxed lightly in Japan” (Makin and Shoven 1987:307, 309). Note that there is no provi- sion similar to Section 395 of the U.S. Internal Revenue Code which limits deduction of interest payments. Other Investment Incentives The Shoup report also contained several other recommendations that have worked as incentives for investment. First, the dividends-received credit reduces the tax burden on capital income—even though the Shoup report introduced the credit for the principle of equity rather than on capital accumulation grounds.!2 Also, the complete exemption (now 61 62 Minoru Nakazato only a partial exemption) of intercorporate dividends from the corporate income tax worked in the same direction, though it was intended only “that all extra burdens on the holding of shares by corporations and on the payment of intercorporate dividends be removed as far as possible” (Shoup et al. 1949, 1:121—22). Second, the Shoup report recommended the carryforward and car- ryback of losses in both the individual and corporate income taxes (2:133-35). These generous treatments reduce the tax burden on in- come from investment capital. Third, the report recommended “that reserves for bad debts be al- lowed with respect to all types of businesses” (2:141). This recommenda- tion might seem only an improvement in accounting; however, it later proved to have important results. Due to the generous standards for the amount of the reserves, financial institutions were able to increase their deductions more than necessary (Kaneko 1988:257). Negative Effects The Shoup report also, of course, had negative effects on the Japanese economy. One particularly important phenomenon that may have been due to the report is tax avoidance caused by the greatly increased use of the corporate business form. According to the report, on March 31, 1949, there were 153,636 general corporations (“joint stock companies” in the Shoup report), 43,630 close corporations (“limited companies”), 82,644 limited part- nerships, and 25,475 general partnerships (“unlimited partnerships”). !3 By 1984 there were 868,955 general corporations, 704,099 close corpo- rations, and 6,471 commercial partnerships in Japan (Kokuzeicho 1986). No one is quite sure why the numbers have increased so much. Taxation may be one of the reasons (Kaneko 1988:149). Initially, one might suggest that the substantial use of the corporate form derives from the fact that the corporate income tax rate has been lower than the highest individual income tax rate. Unfortunately, this does not explain fully why there are so many corporations in Japan, because many of these corporations have deficits or only very small amounts of income. According to the National Tax Administration statistics, in 1984, among the 1,624,261 entities subject to the corporate income tax, 899,186 (55.4 percent) had deficits, 281,706 (17.3 percent) had taxable The Shoup Report and Japanese Development income of less than 1 million yen, 92,044 (5.7 percent) had taxable income of over 1 million yen and less than 2 million yen, and 119,140 (7.3 percent) had taxable income of over 2 million yen and less than 5 million yen. Only 14.3 percent of the firms had taxable income of over 5 million yen (Kokuzeich6 1986). Instead, the explanation probably lies in the following combination of factors. First, the corporate income tax applies broadly to any entity with a judicial personality (including close corporations and part- nerships in the Commercial Code). Note that the code gives a judicial personality to any general or limited partnership established under it. Second, the Commercial Code makes it rather easy to establish corpora- tions and partnerships. Third, the tax unit for the individual income tax after the Shoup report has been the individual. The U.S. system of joint returns has never been seriously discussed by the Japanese government. As a result, by establishing a family corporation and employing family members, a taxpayer can divide the income among the family members. This is more advantageous than using an unincorporated firm under the individual income tax because the Income Tax Law permits payments to family members from unincorporated firms to be deducted from the taxpayer’s income only within certain limits. Of course there are some limits under the corporate income tax, yet the limits under the individual income tax are significantly more stringent. Were the incomes of all family members aggregated, income division using a family corporation would become impossible. Fourth, it is possible to reduce the amount of the tax still further. The amounts the family members receive from the family corporation are treated as “employment income,” which is defined as one of the income categories in the Income Tax Law. However, under that law the recipient is entitled to a standard deduction for a specific percentage of the gross income as an employee business expense, even if the expense is not actually incurred. This standard deduction system was in use before the Shoup report. Although the report allowed the deduction to continue (however reluctantly), we should not blame the report for this system. More recently, the government has moved to limit the use of the stand- ard deduction by family members. As a result, many taxpayers, such as small business owners, actors, writers, accountants, and doctors, are able to obtain tax benefits by establishing family corporations. This phenomenon of using corpora- tions to reduce the individual income tax is nonequitable. It might also 63 64 Minoru Nakazato be nonneutral because it distorts the taxpayer’s choice of business form. Some people argue that there are too many small businesses in the Japanese distribution process, which makes things more expensive. Even this may be the result of the overincorporation which, at least in part, is caused by the tax system. Note that a taxpayer may be able to stay in business by incorporating because incorporation reduces the tax burden. Conclusion In this paper I have discussed a few limited aspects of the Shoup report. One might just as validly have chosen to discuss such matters as tax administration, the value-added tax, the net wealth tax, or the accessions tax. Nor ought one to ignore the Shoup Report’s educational impact. As Professor Shoup himself noted: “The mission thought of its work as largely an educational task, as well as a direct guide to policy; care was taken to explain the reasons for every recommendation, and the Shoup Report was widely distributed over the country” (Shoup 1989d:181). Even if the tax system based on the Shoup report was short lived, the Shoup report’s intellectual impact was tremendous. It remains today an inspiration to many Japanese tax scholars. Tax law professors especially owe much to the recommendation that: “The Law Department of the Universities should institute courses in Tax Law... . Attention should be directed toward the substantive and technical provisions of the tax laws and to the specialized aspects of tax administration” (Shoup et al. 1949, 4:D67). Recently Japan experienced two major tax reforms (see Kaneko 1986). In the tax reform of September 1987 the tax-exempt savings system was abolished (Nagano 1988:158—61). In December 1988 the tax exemption of capital gains from the sale of securities was abolished and a value-added tax was introduced. Once again the Japanese tax system has begun to resemble the one in the Shoup report. Much remains to be learned from the Shoup report. The current Japanese tax system retains many of the problems which the report identified forty years ago. In this sense it will remain a textbook for the future. As Professor Kaneko put it: The Shoup Report was a great experiment. It attempted to design a country’s entire tax system according to a consistent theory. American tax theory was at The Shoup Report and Japanese Development the time the world’s most advanced, and the Report systematized that theory. Academically, the Report was excellent. As the selection of the members of the Shoup Mission showed, the Report was a reformist document. These members knew what was wrong with the American tax system and had intelligent ideas about how to correct it. Ultimately, they tried hard to estab- lish a modern, equitable, optimal tax system. (Kaneko 1988:66—67) Notes 1. This system resembled that of France after the 1914-17 tax reforms, le “systeme Caillaux”, where the income tax consisted of les impots cédulaires et Vimpot général sur le revenu. See Gest and Tixier 1986:373-77. 2. For a brief history in English, see Aoki (1985):435. 3. Joseph Dodge, an American banker sent to Japan as a financial adviser, introduced a policy known as the “Dodge Line” designed to use taxation as a weapon against inflation. The Dodge Line “aimed at stabilizing the yen value by establishing a true balance in the consolidated budget and by eliminating the subsidies which had been the prime cause for the continuing growth of fiscal deficits.” It succeeded in halting inflation, but a depression followed (Ishi 1988:10 n.20). 4. The underlying idea was equity (Kaneko 1983:9). 5. Professor Shoup’s ideas about tax reform appear in Committee on Taxa- tion of the Twentieth Century Fund (1937). 6. The Shoup report proposed a lower standard deduction on salary income (see Kaneko 1983:35—38). 7. Of course, there are a few things the Japanese government resisted, such as a value-added tax. One could also examine the process of negotiation between the General Headquarters (GHQ) and the Japanese government, the influence of various Japanese interest groups on the policy formation, and the power strug- gles within the GHQ among the countries involved (each country was effectively an interest group). 8. The value-added tax, for example, was repealed before it took effect. g. Perhaps it was in part because of these measures that the Japanese econ- omy grew so rapidly; that is, the result of interest group pressures was an efficient tax system. At one level this conclusion seems somewhat contrary to Stigler (1971), though in line with the theorem in Becker (1983). 10. The report also changed the rate structure of the individual income tax. 11. Note, as Professor Stiglitz says, that “it is ex post horizontal equity in which we are interested” (Stiglitz 1982:1). See also Warren (1980:1102—05). 12. A reduced corporation tax rate on dividends paid was later introduced. 13. A “joint stock company” (kabushiki-gaisha) is comparable to a société 65 66 Minoru Nakazato anonyme in France and a Aktiengesellschaft in West Germany. A “limited com- pany” (yuigen-gaisha) is comparable to a société a responsabilité limitée in France and a Gesellschaft mit beschrankter Haftung in West Germany. A “limited partnership” (g6shi-gaisha) is comparable to a société en commandite simple in France and a Kommanditgesellschaft in West Germany. An “unlimited part- nership” (g6mei-gaisha) is comparable to a société en nom collectif in France and offene Handelsgesellschaft in West Germany. 5. GiRVA:R LE, SE. “Mie EU RE |) JR... Income Tax Reform in Venezuela: Thirty Years after the Shoup Mission In 1958 Professor Carl Shoup led a mission to examine the tax system of Venezuela. The names of the members of the mission read like a “Who’s Who” of tax reform: Carl Shoup, John Due, Lyle Fitch, Donald MacDougall, Oliver Oldman, and Stanley Surrey. Not surprisingly, this group produced a report (Shoup et al. 1959) that became a classic in the field; fortunately, Shoup had the good judgment to provide for the publication of the report in both English and Spanish. The report has had a lasting impact on tax policy in Venezuela, though probably not as much as its authors would have hoped. The early 1980s saw yet another tax reform commission in Venezuela, the Comision de Estudio y Reforma Fiscal (Commission for Fiscal Study and Reform) or CERF. This commission was headed by Dr. Tomas Enri- que Carillo Batalla, who as a young man just returning from graduate study at Columbia University had been instrumental in bringing the Shoup mission to Venezuela and had then served as the chairman of the 1958 commission (Shoup et al. 1959:v—vii; Shoup 1989c¢:254—55). Though this commission has as yet had relatively little direct impact on policy making, it has provided substantial valuable information about and analysis of the tax system of Venezuela, leaving as its legacy an agenda for tax reform. This paper compares the 1958 income tax of Venezuela and the Shoup proposals for its improvement with the 1988 income tax and its defects, as a precursor to a discussion of why the Shoup mission has had a relatively small long-run impact on the tax system of Venezuela. The 1958—88 comparison is, for the most part, a comparison of snapshots, rather than a movie. That is, it compares the present income tax system of Venezuela and its need for reform with what the Shoup mission found thirty years ago and the reforms the mission recommended. Little effort 68 Charles E. McLure, Jr. is made to trace the evolution of the income tax system (what might be called “episodes” of tax reform) between 1958 and 1988.! This focus is dictated by both lack of adequate familiarity with the detailed history of income tax reform efforts in Venezuela and the belief that the movie version of history would not be substantially more informative than the snapshot version for present purposes. But in certain instances addi- tional snapshots are shown, usually in footnotes, especially where they indicate either the extent to which the 1961 and 1966 reforms (effective in 1967) built upon the Shoup recommendations or led to a worsening of the situation described by the mission. Nor is there a systematic discus- sion of the CERF recommendations for tax reform, though the discussion of the needs for reform of the 1988 system is largely based on a study done for CERF in the early 1980s (Aguirre et al. 1986). Perhaps more important are the CERF comments on why Venezuela has not made more progress in tax reform. The expositional approach taken is first, to describe briefly (in Section II) salient features of the pre-Shoup income tax system of Venezuela and the Shoup proposals for its reform and then (in Section III) the income tax system of 1988 and its defects. The appraisal of the current income tax distinguishes between improvements recommended by Shoup, other improvements, defects noted by Shoup, and defects not discussed by Shoup; often the last category involves “tax deforms” introduced since the Shoup mission visited Venezuela. The final section comments on progress in Venezuelan income tax reform—or the lack thereof—over the past thirty years. Following the 1983 report of the CERF, it suggests that much of the explanation for the failure to adopt a better income tax can be traced to the country’s good fortune in being blessed with rich petroleum resources; in a word, Venezuela could afford not to follow sound tax policies. Now that Venezuela’s fiscal fortunes have deteriorated with the decline in world oil prices, it will be interesting to see whether it becomes more receptive to proposals for tax reform. Attention focuses almost entirely on what Shoup (this volume) might call engineering aspects of the income tax. The paper has little to say about “tax architecture” or tax administration. The type of general sales tax Venezuela should adopt, if and when it adopts one, is addressed only briefly. There is no discussion of customs duties and commercial policy, stamp taxes, excise taxes, gift and inheritance taxes, or taxes of subna- tional governments. Even within the income tax, coverage is not com- Income Tax Reform in Venezuela 69 prehensive; there is no discussion of the taxation of oil and gas, an especially important topic for Venezuela,” and little discussion of tax administration or international aspects of taxation.? What Shoup Found and Recommended4 The Income Tax in 1958 The Venezuelan income tax of 1958 was an extremely complex com- bination of nine schedular taxes, each levied at low flat rates, and a “complementary” tax levied on a relatively global basis at graduated rates. Separate rates, ranging from 1 percent to 3 percent, were levied on rents from real property (2.5 percent), interest and royalties (3 percent), business profits (2.5 percent), oil and mining profits (2.5 percent), agri- cultural profits (2 percent), income from noncommercial professions (2 percent), wages and salaries (1 percent), and gains of real estate (3 percent); in addition there was a ro percent tax on gambling winnings.° Losses or excess deductions in one schedule could not be used to offset income taxable under another schedule. Dividends and interest received from several sources (interest on saving accounts, certain agricultural and industrial development bonds, and certain government bonds) were exempt from both tax regimes. Both individuals and corporations were subject to the same schedular income tax rates and the same graduated rate schedules under the com- plementary tax (to be described further below). Partnerships and limited liability companies were subject only to the schedular taxes; partners included their share of partnership income in the base of the complemen- tary tax, whether distributed or not. There were separate schedular exemptions for income from wages and salaries (B 1,000 per month, available only if such income for the month did not exceed B 1,600) and for agricultural income (B 30,000, available only if agricultural income did not exceed B 50,000). A per- sonal exemption totaling B 12,000 was available to reduce income from other sources (other than gambling) subject to schedular taxation; this exemption could be applied against whichever of the taxpayer’s incomes would otherwise be subject to the highest rate. These exemptions were not allowed in calculating liability for complementary tax; instead there was a separate system of exemptions for the taxpayer (B 12,000), spouse 70 Charles E. McLure, Jr. (B 4,000), and dependents (B 3,000 for parents, grandparents, children, and grandchildren; B 900 for others supported by the taxpayer). The twenty-nine rates under the complementary tax (which applied to the total of schedular income other than gambling winnings) ranged from 1.5 percent to 26 percent, with rates at the bottom of the income scale differing from each other by as little as 0.25 percentage points. It is worth noting that only eleven of these twenty-nine rates, including the highest four, were integers (whole numbers).¢ The 1958 Venezuelan income tax was obviously quite complex. Com- plexity was far greater than even this description might suggest. De- preciation allowances were deductible in calculating business profits, but not in calculating rent from real estate. Gains from the sale of stocks and securities were taxable as business income. The tax administration (but not the taxpayer) could use ro percent of gross sales as a presumptive measure of income from agriculture. Income subject to four of the schedules could be reported only under cash accounting; that under the other four schedules could be reported only under the accrual method. Taxpayers could use fiscal years to report three kinds of income, but only calender years for the other five. Needless to say, the existence of nine different schedules, plus the complementary tax, made compliance and administration needlessly complicated. Because the filing requirements were not consistent with exemption levels, taxpayers with no liability might be required to file tax returns. There were also complicated rules determining whether various types of income could be split among spouses for income tax purposes. (Vene- zuela has community property.) Finally, the use of marginal tax rates specified as decimal fractions—not to mention the unnecessary pro- liferation of rates—contributed needlessly to difficulties of compliance and administration.” Since the exemption for wages and salaries was a monthly exemption, unused exemptions from one month could not be used to offset income earned in another month. There were no deductions for costs of earning wage and salary income; itemized deductions were only allowed for medical and dental expenses and for charitable contributions, provided the contributions exceeded a given amount. A taxpayer with two em- ployers was required to advise one of them of the total amounts received so that withholding could be adjusted. These features of the tax and the fact that the tax was levied at a flat rate helped assure that withholding would exactly discharge tax liability for a large number of taxpayers. Income Tax Reform in Venezuela Taxpayers in the top three rate brackets were eligible for reductions in tax rates applied to reinvested income. The normally applicable rates of 21 percent, 23 percent, and 26 percent could be reduced to 19 percent, 20 percent, and 22 percent on a limited amount of reinvested income. The system was generally characterized by “notches,” segments of income (or expenditure) over which tax liability would rise (or fall) dramatically when income or some other measure of economic activity changed only slightly. Notches resulted, inter alia, from the fact that certain exemptions (e.g., for labor and agricultural income) and deduc- tions were available only if income fell below a certain level, from the fact that charitable deductions were available only if contributions ex- ceeded a given amount, and from the way the reinvestment incentive was structured. Finally, and perhaps most important in the minds of Shoup and his colleagues, the complexity of the two-tiered system (the existence of a number of schedular taxes and a complementary tax) made it difficult to think clearly about the structure of the Venezuelan income tax and its effects (its impact on taxpayers and the burdens it involved). It was necessary to keep in mind several sets of exemptions, treatments of family status, and rate structures. In the words of the Shoup report, “It becomes very difficult to grasp the cumulative effect of these taxes on any taxpayer or class of taxpayers, in order to perceive readily and clearly just what their burdens may be.” For some taxpayers the low-rate sched- ular taxes were more important than the complementary tax with its higher graduated rates, due to differences in exemption levels; but for others the complementary tax was the more important tax. Under some circumstances agricultural workers would pay more tax than their em- ployers with equal income (Shoup et al. 1959:103—6; quotation is from 103). The Shoup Recommendations for Income Tax Reform The recommendations of the Shoup mission would, in most respects, be judged to be quite consistent with the conventional wisdom of the time—wisdom that for the most part has not changed. For that reason they are only described here, with little discussion. The mission recommended that the schedular taxes (including the schedular exemptions for particular types of income) should be elimi- Til iy? Charles E. McLure, Jr. nated or absorbed into the complementary tax to create a modern global income tax. Shoup (1989c:267) has noted that this was the most impor- tant recommendation of the report. The mission noted (Shoup et al. 1959:104—5) that the problems of definition and classification involved in the schedular taxes were so great that they were tolerable only because the tax rates—and the differences between them—were so small. It also recommended the elimination of “notches,” the closer conformity of declaration requirements with exemption levels, more uniform account- ing rules (dealing with the choice of tax years and cash versus accrual methods of accounting), the elimination of the preferential rates for reinvested income, the taxation of all interest income, and the elimina- tion of differences in the treatment of expenses of earning income in different activities. It also recommended continuation of the policy of not allowing deductions for personal and living expenses. The mission (Shoup et al. 1959:111—25) devoted considerable atten- tion to the taxation of corporate taxpayers and shareholders in an excellent discussion of that topic. First, it argued that corporations should not be subject to the same progressive rate schedule as individ- uals. Rather than proposing a flat-rate corporate tax, the mission fa- vored a system of two or three rates, say 10 percent, 20 percent, and whatever top rate was chosen. If fragmentation of corporate operations to take advantage of the lowest rates were thought to be a problem, a possibility that the mission thought unlikely, the mission would propose a requirement for the consolidation of the returns of commonly owned firms (Shoup et al. 1959:113—14). Second, the mission recommended that dividends should be taxed when received by individual shareholders and that withholding be ap- plied to dividends, especially those on bearer shares. It then examined various ways of integrating the two taxes or avoiding double taxation of dividends (partnership method, corporate deduction for dividends paid, and shareholder credit for dividends received, with and without “gross- up”). In the end it recommended providing partial relief from double taxation, preferably through the shareholder credit with “gross-up,” which it chose over the economically equivalent corporate deduction for dividends paid on administrative grounds. The mission also advocated establishment of a Tax Research Group “to advise the Minister of Finance on economic policy and legal aspects of taxation.” An important part of its task would be helping to determine what statistical data on the tax system should be compiled and made available (Shoup et al. 1959:236, 239). Income Tax Reform in Venezuela Income Tax Topics Not Addressed There are several topics in income tax reform about which the Shoup mission had relatively little to say, at least by today’s standards. For example, it did not really discuss the proper time pattern for depreciation allowances (or other timing issues) or the need for inflation adjustment of either the measurement of income from business and capital or amounts fixed in nominal terms. Today the standard tax reform mission would be likely to pay particular attention to such timing issues, and perhaps to the need for inflation adjustment in the measurement of income, using the recently developed analytical technique of marginal effective tax rates to quantify deviations from a neutral income tax system (see McLure et al. 1988). In addition the Shoup mission generally appears to have taken the tax system at face value, without stressing that “things ain’t what they seem” once the lawyers and accountants finish their work. Though the Shoup mission mentioned the difficulties of monitoring transfer prices used for transactions between affiliated taxpayers, it did not stress this problem in its discussion of either the income tax or the sales tax. This is a problem that has become increasingly important, for reasons to be noted below. Finally, the mission seems to have paid relatively little attention to opportunities for tax arbitrage. The General Sales Tax The Shoup mission recommended that Venezuela should eliminate its low-rate (/ of 1 percent) tax on gross receipts, a tax generally agreed to be highly defective (Shoup et al. 1959, chap. 10). It devoted only a short chapter (chap. 11) to the question of whether Venezuela should replace the gross receipts tax with a general sales tax, and if so what kind. It considered five types of taxes: a turnover tax, a retail sales tax, a whole- sale sales tax, a manufacturers’ sales tax, and a value-added tax (vat). It recommended against imposition of any form of general sales tax, pri- marily because the revenue potential of the income tax was far from being exhausted. In retrospect two characteristics of the sales tax chapter are striking: the recommendation that the wholesale tax would be the most appropri- ate form of sales tax for Venezuela and the fact that only a single 74 Charles E. McLure, Jr. paragraph is devoted to the value-added tax. This stands in sharp con- trast to the much longer chapter devoted to sales taxation in a study done for CERF in 1981—82, which focused almost entirely on the advantages and implementation of a vAT. This difference almost certainly reflects the nature of professional thinking about sales taxation at the time the two reports were prepared, plus differences in the economic conditions and mandates under which they were conducted. In 1959 thinking about the vat was still quite rudimentary, and there was little experience with it. Moreover, the Shoup mission thought that it would not be desirable to introduce a general sales tax until the income tax reached several times its then- current level. It is thus hardly surprising that sales taxation received little attention and the vAT was not proposed for adoption, or even subjected to careful study. The situation in 1981—82 was quite different. There had been a decade and a half of experience with the vaAT in Europe, and other countries in Latin America had already shown that the vat could be administered effectively by a nation at Venezuela’s level of economic development. While it still appeared in the early 1980s that expanding income tax revenues should be the first order of business if more revenues were needed, there was growing recognition that it might be wise to have a general sales tax “on the shelf” in case it were needed. Thus Aguirre et al. (1986:206) wrote, “It is clear, however, that additional revenues are going to be needed, possibly in the near future, and that sales taxation is one of the best potential sources” (see also CERF 1983:28). The Present Situation The present income tax system of Venezuela shows clear evidence of the Shoup recommendations. In some beneficial respects it even goes beyond the advice of the Shoup mission. And yet it suffers from many defects. Some of these result from failure to accept the advice of the mission. Others reflect “tax deform” in areas not really touched on in detail by the mission. “Shoupian” Improvements Venezuela no longer has a schedular income tax for either individuals or corporations. Rather, it now has a global income tax levied at rates Income Tax Reform in Venezuela ranging from 4.5 percent to 45 percent for individuals. This is the most important reform that can clearly be traced to the influence of the Shoup mission. Even so, it might be noted that Venezuela did not move in one step to a global income tax on individuals. The 1967 law contained separate progressive rates for commercial and business income and for wages, salaries, dividends, and other income of individuals (other than from oil and mining, which was subject to yet a third schedule) (see Gittes 1968:133—42). Of course, this system produced considerably more complexity than a simple global income tax. In addition, dividends are now subject to tax; indeed, dividends are subject to withholding tax, but only at a rate of 5 percent. The Shoup mission recommended such reforms. Consistent with the Shoup recommendations, since 1961 corpora- tions are no longer taxed under the same regime as individuals. There are now three corporate rates, 15, 35, and 50 percent. One prediction of the Shoup mission, that fragmentation of corporations to gain the benefits produced by the application of graduated rates to corporate income would not be a problem, did not prove to be accurate, perhaps because the corporate rates are higher and more differentiated than the mission recommended (see Shoup 1989c:267). It appears, however, that the middle 20 percent rate bracket—covering income from B 100,000 to B ro million—was so broad under the Shoup proposal that there would have been substantial incentives for fragmentation, unless the top rate were very close to the recommended middle rate of 20 percent. A requirement for the consolidation of the returns of closely affiliated corporations was enacted in 1986 to reduce the benefits of fragmenta- tion. Unfortunately, this was done in the context of a three-rate system that leaves intact substantial benefits of fragmentation. Moreover, the provision for consolidation appears to be largely unworkable for admin- istrative reasons and unfair under some circumstances. These problems could be avoided if the graduated rate structure for corporations were eliminated, or reduced at most to two rates, the lower of which is applied only to income below a fairly low limit, as proposed by CERF (1983:27). Other Improvements Withholding has been extended to twenty-eight different types of pay- ments, in addition to wages and salaries. (For administrative reasons withholding is required on many of these payments only when they are 7S 76 Charles E. McLure, Jr. made by companies, state entities, and tax-exempt organizations.) Many of these apply to payments to nonresidents and thus are not of great interest for purposes of the present discussion. Payments to residents covered by withholding include honoraria, commissions for the transfer of real estate, merchandise discounts of employees, interest, payments to contractors, rental of real estate and furniture, payments by credit card companies, and payments to insurance agents. While this is an important improvement, it should be noted that the information provided by withholding statements may be every bit as important as the small amount of revenue resulting from application of the relatively low rates of withholding. If the fiscal authorities do not actually make full use of the information contained in withholding statements, there will be little long-run effect on revenues, beyond the relatively minor direct effect through increased withholding.® Uncorrected Defects Noted by Shoup Venezuela has replaced personal exemptions with credits for the tax- payer (B 500), spouse (B 300), and dependents (B 300). But, as in the analogous situation in 1958, the combination of tax rates and credits is inconsistent with the filing requirement. As a result, there is a “notch,” since a slight amount of additional income can push the taxpayer across the threshold of the filing requirement and subject him or her to substan- tial tax liability.? Gittes (1968) indicates in various footnotes that many new notches were created by the 1966 legislation. In addition, continu- ing the unfortunate pattern found in 1958, only seven of the fifteen marginal tax rates for individuals are integers. Now that it taxes dividends, Venezuela employs the classical system of taxing both corporate income and the dividends received by share- holders; it has not adopted one of the conceptually correct approaches to integration discussed by the Shoup mission.!° Venezuela has not heeded the advice of the Shoup mission to avoid allowing deductions for personal and living expenses. Indeed, it has moved in exactly the opposite direction and now has perhaps as gen- erous and ill-conceived a set of itemized deductions as found in any country with a similar level of income. In addition to the deductions for charitable contributions and medical expenses that were in existence in 1958, there are now deductions for automobile insurance, education Income Tax Reform in Venezuela costs, payments to certain professionals, public utilities, life insurance premiums, mortgage interest, residential rent, and contributions to so- cial security and employee savings funds. With almost no exceptions there is no good justification for these provisions. Indeed, these provi- sions undermine vertical equity (e.g., automobile insurance and educa- tion) and encourage congestion and pollution (automobile insurance), as well as substantially reducing revenues and increasing problems of tax administration. These provisions are particularly troublesome, given the way they are treated in withholding and the inadequate level of auditing of taxpayer returns. The taxpayer reports estimated deductions to his or her em- ployer, who adjusts withholding accordingly. Since employers have little incentive or ability to police such claims for deductions and administra- tion of withholding is lax, there is little to prevent taxpayer abuse. The taxpayer is obliged to file a return at the end of the year in which amounts actually spent on deductible items are itemized, but monitoring of such claims is also lax. Venezuela has not effectively implemented the recommendation of the Shoup mission that a tax research group be established. Nor has data collection improved much. A 1982 report prepared for CERF described a situation that appears to have improved little since 1958 (Aguirre et al. 1986, appendix 4). The situation in 1988 was little better than in 1982. Other Defects In addition to the many defects noted by the Shoup mission, there are many that were not noted. Some predate the Shoup mission, but others are of more recent origin. Fiscal Incentives One of the most troubling developments since the Shoup mission visited Venezuela was the seemingly inexorable growth of exemptions and exonerations for income from a wide range of activities. In 1958 there were exemptions for interest on saving accounts, certain agricultural and industrial development bonds, and interest on certain government bonds. For the most part these did not extend to business or agricultural income. There was a rate reduction for reinvested income, but this was not targeted to particular sectors or forms of investment. 78 Charles E. McLure, Jr. During the intervening period, beginning in 1961, targeted exemp- tions and investment incentives have been granted for a wide range of activities.!! Incentives have been provided, inter alia, for agriculture, investments in pollution control, construction, fishing, mining, tourism, transportation, regional development, exports, and a variety of financial activities. From 20 to 100 percent of the income earned in certain activities is tax-free. In addition, the interest paid on loans to finance certain investments is tax-exempt, as are the dividends paid from income earned in agriculture and some construction. Investments in some sec- tors are eligible for credits against tax of 15 percent of the investment (25 percent in the case of producers of capital goods). It is difficult to know either the costs or the benefits of fiscal incentives. Of course, by being offered for so many activities, exonerations and other incentives lose much of their potential effectiveness—though not their adverse impact on revenues. Although the income tax return has a place for the calculation of “fiscal sacrifice” resulting from incentives, many taxpayers do not bother to complete this part of the return, and the government has no idea how much revenue is forgone. Besides directly reducing the tax base drastically, these tax benefits create opportunities for evasion (thereby reducing the tax base even further than would be apparent from a reading of the law), distort resource allocation, and reduce the progressivity of the tax system. There appears to be little realization that generous and pervasive incentives can create negative marginal effective tax rates and cause the waste of invest- ment capital, perhaps because there is not a competent tax research group. Particularly noteworthy from an administrative point of view is the fact that the tax return does not request information on transactions between affiliated firms; of course, such transactions can be manipulated to shift income from taxable to tax-exempt activities (see Aguirre et al. 1986:166—68). Until recently taxpayers could also easily use expenses incurred in earning tax-exempt income to offset taxable income. Inflation Adjustment As noted above, the Shoup mission did not consider whether it would be appropriate to provide inflation-proofing for the income tax. In chapter two of the mission report, “General Economic Considerations Relevant to Public Finance,” inflation was not included as an important determi- nant of the tax structure. Chapter thirteen did consider means to increase Income Tax Reform in Venezuela the sensitivity of tax collections to both inflation and unemployment, in order to increase the built-in stability of the system, but presumably the mission would not have proposed inflation adjustment, since that de- creases such sensitivity. To some extent this attitude reflects the conven- tional wisdom of the day; inflation adjustment simply was not seen to be important, or even desirable, especially if it reduced the automatic sta- bilizing properties of the income tax or resistance to policies that pro- duced inflation. !2 Recent developments suggest that it may be appropriate to reconsider whether Venezuela should adopt inflation adjustment in the measure- ment of income. First, inflation has been substantially greater in recent years than in the 1950s. Even since 1982 inflation has been more rapid than during the period before 1979. (The three-year period of especially high inflation, 1980—82, is omitted from this comparison as a possible aberration caused by poor management of the demand pressures created by the oil bonanza.) Whereas the inflation rate averaged 3 percent from 1969 to 1973 and 8 percent from 1974 to 1978, it averaged 10.3 percent during 1982—86 and exceeded 4o percent in 1987. Moreover, both economists and policy makers in many countries have come to appreci- ate better how resource allocation and equity are undermined by the interaction of even moderate amounts of inflation and a tax system that makes no allowance for inflation.!3 Worldwide Taxation In a change generally consistent with the recommendations of the Shoup mission, Venezuela has introduced worldwide taxation of the income of its residents. Certain foreign-source gross income (dividends, interest, rental income, royalties, life pensions, and income from trusts) is now subject to tax under a provision introduced in the 1986 tax reform. For administrative simplicity such income is taxed at a flat rate of 14 percent, rather than at the rate applicable to other income. The merits of this change are questionable, since the administrative difficulties of implementing such a scheme might easily outweigh the conceptual advantages of worldwide taxation. It simply is not realistic for a country such as Venezuela to attempt to tax the foreign-source income of its residents. The tax on foreign-source income may under- mine taxpayer morale, since it is likely to be paid by honest taxpayers and evaded by the dishonest. !4 79 80 Charles E. McLure, Jr. Multiple Taxes on Labor Income The national social security institute and the national institute for coop- erative education each levy payroll taxes to support their activities. Together these yield more revenues than is collected from the individual income tax. It is highly anomalous to have two payroll taxes levied by different agencies of the central government, in addition to the income tax levied by the government itself. There is unnecessary complexity, since the three taxes are not levied on the same definition of labor income, and there is considerable duplication of administrative effort. Moreover, the three taxes produce a pattern of marginal rates that can only be de- scribed as bizarre, at least if one ignores the benefits of social security. Comments on Venezuelan Tax Reform Views on just how influential the Shoup mission was for the future of Venezuelan tax reform are mixed. Oliver Oldman, a member of the mission, wrote the following in his “Foreword” to Gittes (1967): The Shoup Report is still considered a principal reference work on the tax problems of less developed countries. . . . It is difficult to measure the impact of the Report, but it is certain that Professor Shoup’s insistence that the Report be published and that it be published in Spanish as well as English was crucial to the Report’s having substantial impact. The Report has been and is probably still being used in Venezuelan universities. As Mr. Gittes’ article shows, a number of the Report’s suggestions were ultimately followed; and a number were not. The role of the Report in stimulating interest in tax structure reform and in identifying problem areas ought not be underesti- mated. (126)1!5 Twenty years later, Carillo Batalla provided a similar assessment: There is no doubt whatsoever that the “report on the Venezuelan 1958 fiscal system” constitutes the most valuable technical analysis ever made of the Venezuelan tax system. The recommendations contained in the same have shown the way to the national fiscal reforms since 1958 up to 1986... the principal merit of the Shoup Report has been its permanent applicability in the last thirty years as fundamental technical orientor of the Venezuelan fiscal system. (Quoted in Shoup 1989c:275) These views are reflected in the following quotation from the 1983 CERF report: Income Tax Reform in Venezuela The history of Venezuela’s tax system, written by a group of experts for the Commission on Fiscal Study and Reform, shows that the guidelines and point of reference for reforms implemented in the 1960s and early 1970s were based on the recommendations contained in the work of the 1958—60 Com- mission and those made by Professor Shoup and his collegues. (Quoted in Shoup 1989c¢:275) Gittes, on the other hand, offered this less glowing appraisal in his 1967 article: It is clear that the Report was not the initiating force behind the enactment of the new law; rather, the decision to enact a new law was a political decision reached independently of the Report. Those responsible for the formulation of the new law no doubt studied their copies of the Report and found that its observations and recommendations had not lost relevance. One or more of those responsible for the new law were also members of the enthusiastic group of Venezuelans who took an active part in exchanging views with the members of the Shoup Commission at the time the Report was drafted. (Gittes 1968:172) In reviewing this experience my own judgment is that it is remarkable how little impact good tax advice has had in Venezuela, how little appears to have been learned from mistakes of the past, and how much bad policy there has been. It is true that the most egregious problem— the use of a complicated schedular system—has been overcome with the switch to a global income tax and that the turnover tax has been re- pealed. But there are still many problems with the Venezuelan income tax. The filing limit is inconsistent with the combination of tax rates and personal credits; the proliferation of both tax incentives and undesirable itemized deductions has eroded revenues, equity, and administrative simplicity; graduated corporate rates continue to be used, despite their incentives for fragmentation. The system remains defective even in minor and easily remedied respects: many more marginal rates are provided than are needed to provide smooth progression in average rates; many of these rates continue to be decimal fractions, complicating compliance and administration; and withholding is required for de min- imis amounts. In summary the income tax of Venezuela has important structural defects, and it is, by all accounts, poorly administered. More- over, in many respects the tax system of Venezuela has become worse, not better, over the past thirty years. All these problems—and others not named—reflect the absence of the competent tax research group recommended by Shoup, as well as the 81 82 Charles E. McLure, Jr. lack of political will needed to introduce reforms. Moreover, it is possi- ble that, as Gittes (1968:172) has suggested, the tax policy of Venezuela over the last thirty years would have fared better if foreign advisers had played a continuing role in the formulation of policy. There is good reason to believe that in some instances the country attempted to move in the right direction but simply lacked the technical expertise to know how to do so. (Examples include the techniques used for integration of the corporate and personal income taxes, the belief that many marginal rates are needed to achieve progressivity in average rates, and the con- tinued proliferation of notches.) Experience in Jamaica and Indonesia suggests clearly that continuing involvement with foreign tax advisers can be crucial to the tax-reform process (see Gillis 1989a; Bahl 1989). The historic abundance of revenues from petroleum helps to explain Venezuela’s substantial disregard for sound advice on income tax policy. In 1958 it appeared that falling oil prices and dwindling petroleum reserves would create a need for increased non-oil revenues in the me- dium term (see Shoup 1989b:252). In fact the actions of opEc and the continued availability of reserves have meant that Venezuela could con- tinue until recently to rely almost entirely on revenues from the pe- troleum sector. Since it has not been important for the country to have a well-functioning income tax, none has been created.!¢ A previous report on tax administration undertaken for CERF drew the following conclu- sion, which seems as applicable for structural tax policy as for tax administration: Twenty years ago, taxation in Venezuela was the subject of one of the classic studies of tax systems, the Shoup Commission Report. . . . At the time of the Shoup Report, and ever since, one of the main characteristics of the Venezue- lan system has been the public sector’s great dependence on petroleum reve- nue. At the central government level, non-petroleum revenue averaged 36 percent between 1977 and 198o. It is perhaps the consequence of this low contribution of non-petroleum taxes to the country’s treasury that there has been no sense of urgency as regards reform of the administrative system. After all, any inefficiency that may have existed in the Tax Directorate affected only the collection of one third of revenue. A change of a few dollars in the international price of petroleum would have a much more marked effect on government revenue than a difficult and costly improvement in administrative techniques for internal taxes and customs. !7 The abundance of petroleum revenues may also help to explain the proliferation of fiscal incentives found in Venezuela. Because of its enor- Income Tax Reform in Venezuela mous earnings of petro-dollars, Venezuela has long suffered from what has variously been called “Dutch disease” or “the Kuwait effect,” the inability to export anything besides petroleum because of the strength of the bolivar.!8 Not surprisingly, there was local interest in attempting to swim upstream against current competitive advantage by providing in- centives for investment in various nonpetroleum sectors. Of course, this explanation is easily over done, because many countries that have not been cursed with petro-dollars have nonetheless adapted extensive sys- tems of fiscal incentives. Looking back at Carl Shoup’s influence in Venezuela there may be a tendency for disappointment that so little has been done correctly and so much has been done badly. It seems however that his experience simply suggests a converse to the maxim proclaimed by Malcolm Gillis (1989b) that reform often occurs only when governments are forced by revenue needs to make unpopular choices.!? When they are under no such pressure, governments are not likely to embrace tax reform simply because it is the right thing to do. Notes The author wishes to thank Carl Shoup for his useful comments on earlier drafts of this paper, As always, Carl’s wisdom is equaled only by his generosity. Of course, any views expressed here should be attributed only to the author. 1. By comparison, McLure and Zodrow (1989) trace the evolution of tax reform in Colombia over roughly the same period. 2. It should be noted that the Shoup report covers the taxation of the petroleum industry. Taxation of petroleum is omitted from the present discus- sion, as it is in many tax reform studies, because it is highly complex and a separate issue from the taxation of income in the nonpetroleum sector. Since Venezuela nationalized the oil industry in 1976, after the Shoup mission, it would be extremely difficult to include an adequate discussion of changes in its taxation in an overview of policy changes such as this. 3. On tax administration, see, however, Shoup et al. 1959, chaps. 5 and 6; Shoup 1989c:263—66 and 271—74; and Gittes 1967:158—71. Griffith, Escobar, and Pavesi (1986) deals exclusively with tax administration. 4. This discussion is based primarily on Shoup et al. 1959, chap. 3, especially 87—96; see also Shoup 1989¢:257—-59. 5. In 1961 the rates applicable to wages and salaries and to gambling 83 84 Charles E. McLure, Jr. winnings were tripled; the remaining rates, except for the rate on income from oil and mining (left unchanged), were doubled (see Gittes 1967:130). 6. In December 1958 the rate schedule for the complementary tax was modified so that there were thirty rates, ranging from 2 percent to 45 percent; again only eleven of these were integers. See Shoup et al. 1959, appendix C. 7. One is reminded of the view of a former U.S. commissioner of internal revenue that American taxpayers can add and subtract fairly well, but they have difficulty multiplying and dividing. Since one would not generally expect more of Venezuelan taxpayers, the gratuitous use of decimal fractions seems quite inap- propriate. 8. Some of the provisions for withholding are almost certain to be subject to considerable uncertainty and controversy. For example, the requirement that the buyer of real estate take account of the circumstances of the seller in withholding on real estate transactions appears to be unworkable. Moreover, withholding is required on de minimis amounts in some cases (e.g., for amounts of tax as small as B 30 or B 50). Such unrealistic requirements can only breed contempt for the law and (if, as is unlikely, compliance is high) create excessive costs of admin- istration and compliance. g. A taxpayer with an income below the filing limit of B 48,000 pays no tax. But application of the existing marginal rates to an income of B 48,001 produces a tax liability before personal credits of B 3,335, well above the B 500 that would be offset by the personal credit for the taxpayer. Only if the taxpayer had ten dependents, itemized deductions of almost B 47,000, or some combination thereof, would the taxpayer be as well off as immediately below the tax thresh- old. 10. From 1966 to 1978 Venezuela did employ a quite defective system of dividend relief under which recipients of dividends were allowed a credit equal to 40 percent of dividends received, without “gross-up” (see Gittes 1967:150). For a complete discussion of this provision and of integration in general, see Aguirre et al. (1986:120-30). 11. Gittes (1967:144—47) describes the situation following the 1966 changes in the law. 12. On the influence of this view in Colombia and its gradual demise, see McLure (1990). Shoup (1989c:263) has noted that nontax revenues were so small relative to national income that the stabilizing effect could not be great, in any event. 13. See McLure et al. (1988) for a detailed discussion of inflation adjustment. 14. See McLure (1989) for full development of this point. 15. Oldman also notes the impact the Shoup report has had in other develop- ing countries, a point also noted, inter alia, by Gillis (chap. 3, this volume). 16. Shoup (1989c:286) draws a similar conclusion, though for a more limited period: “In Venezuela in 1958—59 there was no urgency for improving the tax system.” Income Tax Reform in Venezuela 17. This is a translation of Griffith, Escobar, and Pavesi (1986:13—14). This view has been repeated in the final report of CERF (1983:24). 18. Gelb (1988:88—89) notes in writing about the effects of the oil shocks of 1970s that “Venezuela had long since adapted to the economic situation of an oil monoexporter.” Gelb also notes (82) that the bolivar actually depreciated slightly during the first oil shock. Nonetheless, this does not overcome the fact that during the last thirty years Venezuela has had difficulty exporting anything besides petroleum. 19. Harberger (1988) has made a similar point on the expenditure side of the budget; budget-constrained public officials can resist request for uneconomical expenditures, but those blessed with ample revenues have difficulty in doing so. 85 Comment JOHN BOSSONS Evaluating Tax Missions In discussing papers that appraise tax missions it is useful to start off by discussing the criteria by which tax missions are evaluated. The criteria used by critics of a tax mission may reflect objectives which differ radically from those which guided the mission. The organizer of a tax mission (or, for that matter, any tax consultant) must have a pragmatic and therefore limited definition of what con- stitutes “success.” Knowing that the mission’s report is merely an input into a country’s political process, the leader of a tax mission must define objectives which are realistic goals for that country, so that the mission’s report cannot be dismissed as irrelevant. At the same time, if the mission is to be of lasting value, the objectives must be ambitious enough to define that country’s agenda for tax reform. Balancing ambition against realism, a tax mission should be given high marks for achievement if only half of its recommendations are adopted. On this criterion of achievement the Shoup 1949 mission to Japan has to be regarded as outstanding. The extent to which a tax mission’s recommendations are imple- mented is of course only one criterion by which a mission’s success should be evaluated. A second criterion is whether the recommendations which were adopted survive the tests of administration, time, and vested interests. A third criterion is whether the report’s recommendations and analysis succeeded in changing the nature of arguments over tax policy in the country, redefining the agenda for tax reform. The truly successful missions—like the Shoup mission to Japan—are those which have re- defined the tax policy agenda not only in the country for which they are done but in other countries as well. There is of course a fourth criterion implicit in any critic’s evaluation of a tax mission report, which is whether the report’s recommendations are consistent with the critic’s own preferences and reform agenda. Tax mission reports, like modern music, often fail this critical test. In terms of all criteria other than the fourth, the Shoup mission to Japan was unusually successful. Certainly not all of the recommenda- Evaluating Tax Missions tions were implemented, and not all of those that were have survived the test of viability over a long period of time. The recommendations on interest, for example, ran into substantial opposition from domestic vested interests and conflicted with other policy directions within the Allied Supreme Command. The value-added tax never did get off the ground. But nevertheless the list of accomplishments is substantial: the reduction in tax rates, the partial integration of personal and corporate income taxes, the important changes in corporate taxes, the introduction of the revaluation for inflation. These are important changes which stood the test of time. One might argue from a critical viewpoint that it is “unfair” to regard these accomplishments as evidence of unusual success. After all, some critics may argue, the Shoup mission’s report was advice to an occupying power which was able to impose its will on the Japanese nation. But this argument merely changes the test of what constitutes “success.” I regard the true test as what survived the end of the Allied occupation in 19§2. It is a measure of the success of the mission that so much did survive. To find that so many of the recommendations which were implemented have remained in place is a substantial testimonial to their wisdom and practicability. Turning to the consistency of the recommendations with a critic’s own views, the 1949 mission to Japan had a very clear set of objectives. These objectives were focused on attempting to implement a national income tax which was comprehensive, fair, and administratively practicable. Even though it recommended a mix of taxes, including a value-added tax at the prefectural level, the Shoup mission was clearly motivated by its concept of a comprehensive income tax as the best way of achieving fairness in income taxation. A comprehensive income tax is, by design and objective, not meant to be neutral in savings decisions. The effect of income taxation on savings was obviously a consideration which informed the mission’s analysis and recommendations, but increasing aggregate Japanese savings was not the mission’s primary objective. To suggest, as does Nakazato, that this is a defect in the report is simply to say the Shoup mission clearly put the goal of fairness ahead of neutrality. In most countries the political consensus continues to do so, whether or not economists concur. Percep- tions of fairness largely determine the degree of public acceptance of a tax system. The debate over the Shoup mission’s goals reflects the continuing 87 88 John Bossons debate among public finance specialists over the validity of the com- prehensive income tax base as a tax reform objective. Over the years since 1949 the public finance profession has moved in the direction of giving much higher weight to the objective of tax neutrality, whether achieved through tax rate reductions or through increasing the impor- tance of consumption taxes. But it would be unfair to suggest that the Shoup mission was unsuccessful because its objective differed. The mis- sion’s objectives were different. Its primary goal was to implement a more comprehensive income tax that was practicable and that would be accepted by the Japanese people as an improvement over the tax system previously in place. And in this goal it succeeded. Turning to the ingredients of success, Carl Shoup’s paper sets out an interesting and useful taxonomy. Using Shoup’s terminology, all success- ful missions can be characterized as visionary engineering missions. No recommendations for tax changes will be successful unless successful in engineering terms. But at the same time the long-term impact of the mission will depend on the clarity with which it articulates a compelling vision of its objectives. What are the criteria which define good engineering? A good engineer is one who has both a vision of what is potentially feasible and a clear, concrete understanding of how to implement that vision. Carl Shoup’s division of a mission’s task into the three components of architecture, engineering, and administration reflects an emphasis on deriving a solu- tion which is practicable, as does the strong weight Shoup has consis- tently given to administration. Malcolm Gillis aptly summarizes the importance of this in his statement that Shoup’s fine sense of judgment as to what is feasible is the key determinant of his success as an adviser. To illustrate the usefulness of the taxonomy let me use adjustment of the income tax base for inflation as an example. As Shoup states in his paper, it is extremely difficult to devise a means of making such an adjustment which is both well designed and easily implemented but it is very important that this be done. The vision that one has as an architect of an income tax is (following Shoup) that an income tax should be based on a comprehensive definition of real income. Having defined this architectural goal, the engineering problem is to devise the best feasible way of approaching this goal. The problem is one of finding a “second best” solution within the constraints imposed by other policies, admin- istrative problems, and political perception. The difficulty of dealing with the inflation adjustment issue is greater Evaluating Tax Missions now than it was in 1949 when the Shoup mission considered this prob- lem in Japan. At that time the impact of the Japanese tax system could realistically be analyzed as a “closed-economy” problem, making the analysis much simpler. Today, it is impossible to look at the problem of adjusting the income tax system for inflation (whether in Europe, North America, or Japan) without taking into account the extremely high mobility of capital across international boundaries, the tremendous growth in tax arbitrage skills and techniques over the past several dec- ades, and the consequent importance of harmonizing a national tax on income from capital with the tax provisions in other countries. These considerations make it much more difficult to implement comprehensive inflation adjustments in a single country. The solution to this engineering problem is much more constrained (and therefore even more a problem of finding the second best) than it was four decades ago. At the same time the importance of the nonneutralities introduced by the absence of inflation adjustments has also become greater. Carl Shoup is to be congratulated for focusing on the need for some form of solution (albeit second-best) and for recommending an assets revaluation scheme in his 1949 report. A practicable solution has to focus on what is most important to achieve, what can most easily be achieved, and on the effect of in- complete achievement on other parts of the tax system. For a country experiencing significant inflation, such as had been the case in Japan, restating depreciation and capital gains (asset revaluation) are the most important inflation adjustments because of their impact on investment decisions. Asset revaluation is also the easiest component to implement of a comprehensive set of inflation adjustments. The openness of capital markets makes it very difficult for a single country to move beyond the revaluation of physical assets or equities to restate interest income and expense on a real basis. It is easy for multina- tional companies to move their borrowing from one country to another, finding techniques to keep the advantages of nominal interest deduct- ibility through international tax arbitrage. It is difficult to introduce any restriction on interest deductibility (whether through inflation adjust- ments or other limitations) that does not disadvantage locally financed business relative to multinationals. The potential importance of tax arbitrage in the current environment of global capital markets complicates the design problem for any reform of capital income taxation. Indeed, the growth in international tax 89 90 John Bossons arbitrage makes it increasingly difficult for a small open economy to vary even the statutory rates of corporate income tax—let alone the definition of the tax base—from the standard set by the largest countries with open capital markets (Bossons 1988). These difficulties impose constraints which make the: engineering component of tax system design very much a problem of finding the second best. In this context ad hoc adjustments which take account of interactions with other elements of the tax system may be preferable to “purer” solutions which are better only when considered in isolation. The problem for the tax mission is to find engineering solutions which do not excessively compromise the architectural vision. The definition of a good architect of a tax system is not dissimilar to the definition of a good tax engineer. The architect must develop a structure of compelling vision and consistency, while at the same time ensuring that the associated engineering design problems are soluble. The architect must be able to demonstrate that the vision is practicable, not just administratively but also in terms of the more difficult engineer- ing problem of taking into account the realities of political and policy constraints. The tax architect, like the tax engineer, must understand the constraints in order to find the engineering solution that is most consis- tent with the architectural design concept while at the same time deliv- ering substantial improvements over the existing tax system. And the tax architect must be able to articulate the architectural vision in compelling terms. The great tax architects are both good engineers and visionaries. They must set out a compelling design that redefines the policy agenda, deal effectively with policy constraints, relax constraints where possible through improvements in administrative infrastructure, and reflect the realities of that infrastructure through developing reform proposals that are feasible. In all these respects Carl Shoup’s 1949 mission to Japan set the standard for tax missions. All currently practicing tax architects owe him a substantial debt. Ill The Tax Mix 6. RICHARD GOODE Changing Views of the Personal Income Tax In the preface of his classic work, Personal Income Taxation, dated December 1937, Henry Simons wrote, “the reader may wisely make some allowance for the fact that several of these chapters were prepared originally with regard for prevailing opinions which, in both popular and academic circles, have since changed markedly” (1938:v). This paper describes and comments on changing views of the proper form and role of the personal income tax, as revealed by the writings of economists and public finance scholars. During the twentieth century the personal income tax has been trans- formed from a minor revenue source in a few countries to a leading one in the industrialized market economies, from a tax directly impacting only a tiny minority to a mass tax. The increase in tax rates, the rise in the fraction of the population covered, the use of the tax to promote nonfis- cal purposes, and the interaction of taxation with inflation attracted attention to the shortcomings of existing taxes and provoked discontent. Also changes in political attitudes and economic thought affected views of the personal income tax. After a brief section on the appropriate definition of the base of the personal income tax, this paper will review discussions of tax rates, particularly views on progressivity, and look at questions of tax justice, incentives, macroeconomic implications of progressivity, and the politics of elastic revenue yield. Next, the intellectual challenge to the income tax presented by consumption tax advocates will be examined. Since these topics would offer scope for one or more substantial books, my treat- ment must be summary in nature. No doubt it will omit significant issues and deal too briefly with subjects that merit detailed examination. The Tax Base Legislation and administration usually define the income tax base by enumerating items to be included, deducted, or excluded. However, both 94 Richard Goode the courts and scholarly writers have had to recognize that the meaning of “income” is by no means obvious and that some guiding principle is needed to help decide specific questions. Simplifying, one may say that two broad concepts have been proposed to anchor tax practice (Goode 1977). The first is the idea that income is a flow from a continuing source—in a favorite metaphor, the fruit from a tree. The second defini- tion, which is associated with the names of Schanz, Haig, and Simons, is the accretion concept—the sum of consumption and change in net worth. The two definitions agree on most items but differ in the treat- ment of capital gains and other nonrecurrent items. The Schanz-Haig- Simons (S-H-S) definition includes income capital gains, gambling win- nings, gifts and bequests, and all kinds of casual receipts. The other definition excludes these items. The United States has always included the nonrecurrent items, except gifts and bequests, in taxable income, whereas in the past the United Kingdom and members of the Common- wealth excluded them. Capital gains (with important exceptions) are now taxed in Australia, Canada, and the United Kingdom but in the latter two countries at lower effective rates than apply to other income. This is not the place for an examination of the rival concepts. There appears to have been a convergence of expert opinion on the S-H-S concept (Carter 1966; Meade 1978, Ireland 1982; McDaniel and Surrey 1985). In practice there are many departures from the S-H-S concept. Capital gains are not taxed in many countries and when taxed are subject to lower rates than other income in all industrial countries except Australia and the United States; imputed rent from owner-occupied dwellings is not taxed in about half the industrial countries and usually is under- assessed in the other countries; the service value of consumer durables is not taxed; and many fringe benefits of employment are omitted from the tax base. However, recognition of the significance of the omissions has been growing, and some moves to curtail them have been made. Statistical studies have revealed wide differences between nominal personal income tax rates and effective rates in relation to a comprehen- sive income definition. These differences cause misunderstanding of the burden of the personal income tax and also result in inequities and inefficiencies. The innovative work of Pechman, extending over three decades, has been especially influential in bringing the facts to light (for example, 1957, 1985, 1988; Pechman and Okner 1974). Recent reforms and reform proposals have included broadening the income tax base and reducing nominal rates. The Personal Income Tax Tax Rates Most academic discussions of income tax rates have proceeded on the implicit assumption that the base is properly defined. In this section I shall follow that tradition and, except when otherwise noted, shall assume that taxable income is approximately equal to S-H-S income. The subject to be discussed is the degree of progressivity of the personal income tax and the tax system as a whole rather than the average tax rate in the sense of the ratio of tax revenue to aggregate income. The ap- proach of the writers whose work is mentioned and of my comments is normative: what rates should be rather than an explanation of how actual rates were chosen. Questions of justice, income distribution, economic incentives, and macroeconomic policy will be encountered. At the outset it is useful to recall Shoup’s distinction between con- sensus and conflict criteria for taxation. Consensus criteria relate to horizontal equity and economic efficiency, which everyone will approve in the abstract though there are disputes over their application. Progres- sivity or vertical equity, in contrast, is clearly a conflict criterion. Progres- sion, or its absence, Shoup writes, is the result of a compromise “that reflects other features of the environment.” It cannot be decided by theoretical arguments, “But the compromise is likely to be more durable, the more the discussion has been informed by abstract analysis” (Shoup 1969:35). Neoclassical Sacrifice Theory Passing over a large volume of rather tedious material, I select Edge- worth (1897) and Pigou ({[1928] 1947) as authoritative expositors of what may be called the neoclassical sacrifice theory justifying progres- sive taxation. Assuming that people are equally sensitive to pleasures and pains and that income is subject to diminishing marginal utility, both Edgeworth and Pigou reasoned that minimum aggregate sacrifice would result from draconian progressivity. From the top down incomes would be subject to a 100 percent marginal rate until the needed revenue was obtained, and indeed the process might be carried further until incomes were equalized by a combination of taxes and transfer payments. Both writers emphasized that the conclusion was subject to an important qualification: Adverse effects on capital accumulation and work effort 96 Richard Goode had to be taken into account and somehow balanced against the gain from using progressive rather than proportional or regressive taxes. Within the utilitarian framework the minimum-aggregate-sacrifice theory was criticized on the grounds that people in fact are not equally capable of feeling satisfactions. Two answers were given.to this argu- ment. One was that equal enjoyment capacity is the only ethically or politically tolerable assumption. Any apparent differences were due to habits conditioned by the existing income distribution. The other answer was that enjoyment capacities, though unequal, are randomly dis- tributed and not correlated with income. Hence, the probable value of total utility will be maximized by equal distribution, since errors made by allowing too little to the sensitive low-income persons will outweigh the equally probable errors of allowing too little to the sensitive high- income persons (Lerner 1947:29—32).! A second criticism was that income, unlike particular goods, cannot be assumed to be subject to diminishing marginal utility. If, however, people use their income so as to maximize their satisfactions, the things they buy with a given income must offer greater satisfaction than any other bundle obtainable with that income. It follows that things bought with additional income would yield less satisfaction than those bought with the lower income. Some allowance must be made for complemen- tarity and lumpiness (Lerner 1947:26—28).2 The neoclassical sacrifice theory went into decline and virtual extinc- tion as a result of an attack on utilitarianism launched by Robbins (1935). The dominant opinion became that interpersonal comparisons and cardinal measurement of utility are impossible. It was held that economists qua economists can recommend only Pareto improvements, that is, changes that would make at least one person better off and no one worse off. This drastically narrowed the scope of economists’ advice, since virtually all tax revisions benefit some and harm others. Kaldor’s compensation principle (1939) won considerable support as a way of allowing economists to evaluate a broader range of actions. This princi- ple holds that an action is advantageous if the gainers could compensate the losers and still be better off. But the test was not accepted by those who insisted that potential compensation is insufficient, while pointing out that actual compensation is rare. Until recently, moreover, the Kal- dor test was hardly applied to the question of tax progressivity, perhaps because there was no agreement on how potential compensation could be calculated. The Personal Income Tax Sociopolitical Theories There were always economists who denied that the justification for progressive taxation should be sought by attempting to estimate its impact on individual taxpayers’ satisfactions. For example, S. J. Chap- man asserted in 1913 that the basis of progressive taxation is “in part at any rate, somewhat as follows... : that the wants satisfied by the earlier increments to income are usually of more importance socially than the wants satisfied by later increments to income, whether the satisfaction of the former causes more utility or not. In speaking of the equity of taxation we are obviously talking ethics, and therefore the wants pri- marily dealt with must be adjudged not according to the value of their satisfaction in fact (positive value), but according to the value of their satisfaction in a moral scheme of consumption (normative value)” ([1913] 1959:12). Chapman’s statement is wiser, in my opinion, than that offered by exponents of the neoclassical sacrifice theory. Phrased in more contem- porary language it won some later adherents (e.g., Fagan 1938; Groves 1956). But the idea lacked the appearance of rigor and technicality that appeals to economic theorists and hence did not form an important chapter in the changing views of economists on the appropriate degree of tax progressivity. Simple Egalitarianism Henry Simons caustically rejected the neoclassical minimum-aggregate- sacrifice theory on the grounds that sacrifice is unknown and unknow- able and in any case is an ethically unattractive standard. Instead, he advocated progressivity simply as a means of reducing inequality. In his words “the case for equality (for less inequality) is enormously stronger than any utility foundation on which it can be rested” (1938:14). The case for drastic progressivity, he wrote, “must be rested on the case against inequality—on the ethical or aesthetic judgment that the prevail- ing distribution of wealth and income reveals a degree (and/or kind) of inequality which is distinctly evil or unlovely” (18—19). Simons scorned the proposition that income tax rates should reflect some view of “rea- sonableness” as vague and a surrender to “fickle sentiments” (31). He 98 Richard Goode conceded that, as regards mere distributional effects, he could suggest no limit to progressivity short of “substantial equality among those taxed” (17—18; emphasis supplied). This formulation presumably was chosen because Simons did not advocate a negative income tax to lift people below the personal exemption level up to the income level of those subject to the income tax. He recognized that in order to contribute significantly to equality the income tax would have to take a much larger role in the revenue system than it had in the United States in the 1930s. Like Edgeworth and Pigou, Simons conceded that progressive taxa- tion could have an economic cost. His generalization was “‘it is reason- able to expect that every gain, through taxation, in better distribution will be accompanied by some loss in production” (19). But he tended to minimize the loss. The effect on highly paid personal services and enter- prise was likely to be “negligible” because the activities were themselves attractive and because relative rewards were much more influential than the absolute amount of after-tax income. The effect on saving was potentially more significant, but the reduction of personal saving could be offset by retirement of the public debt and later by a pay-as-you-go policy for public works and government investment in public utilities and other industries (27). Although Simons’s rejection of sacrifice theories was consistent with the growing agnosticism concerning utility measurement, his policy pre- scription went far beyond the modest role that the new welfare eco- nomics allowed economists. His justification of progressivity gained considerable acceptance among tax specialists. Pechman, writing for the International Encyclopedia of Social Sciences, concluded, “The basic justification for the progressive personal income tax is now probably the socio-economic objective of reducing great disparities of welfare, oppor- tunity and economic power arising from the unequal distribution of income” (1968, emphasis supplied). | doubt, however, whether simple egalitarianism ever received as wide support among academic writers as sacrifice theories earlier held or gained wide popular acceptance. While Simons’s attack on other theories was powerful, he failed to advance detailed arguments for his own position, merely stating what seems to be a subjective preference. Propositions in ethics and esthetics cannot be settled by unaided intuition; they are subject to analysis and debate. However, because Simons proposed to treat gifts and bequests as income to the recipients, he is less vulnerable than other supporters of steeply progressive income taxation to the critcism that it will actually The Personal Income Tax increase concentration of economic power by preventing the accumula- tion of new fortunes to rival the established ones. Stagnation For a mature economy threatened by secular stagnation, Alvin Hansen favored a personal income tax at fixed, steeply progressive rates. Such a tax would raise the propensity to consume, which would make for greater stability, he held. In particular progressive taxation would raise the level of national income at which consumption would equal total income and thus would limit the fall in income and activity due to a failure of investment demand (1941, chap. 13). Hansen also maintained that, whereas an increase in government expenditures financed by a proportional income tax would raise national income by an equal amount, expenditures financed by a progressive tax would raise income considerably more because a larger proportion of the revenue obtained from the progressive tax would come from potential private saving rather than consumption (195 1:200—205). In Hansen’s view progressive taxation had the advantage of enhancing the power of countercyclical government spending and lessening the need for budget deficits. The stagnation thesis attracted attention and a good deal of support in the late 1930s and early 1940s. In the decade or so after World War II, it still figured in policy discussions. Gradually, however, concern and policy analysis shifted to the problem of fluctuations around an average that was assumed to represent a tolerable level of employment/un- employment, if not necessarily an acceptable growth rate. Accordingly, the view that a progressive income tax was especially suitable for a stagnation-prone economy became obsolete. A policy of encouraging private saving came to be generally accepted. Built-in Flexibility By the 1940s the traditional canon holding stability of yield to be a desirable characteristic of a tax had been reversed. Samuelson, in a 1942 article, pointed out that a high marginal propensity to tax (later called a high built-in flexibility of yield) acts as a stabilizer against both upward and downward movements of the economy and judged it a desirable feature of the personal income tax (1942:582). 100 Richard Goode The progressivity of the personal income tax made its yield elastic with respect to national income; the high rates established during World War II meant that its built-in flexibility was large. The stabilizing prop- erty of built-in flexibility of the personal income tax was attributed mainly to its impact on consumption, operating through its moderation of changes in disposable income. The emphasis on built-in flexibility, and the analysis of its operation, was part of the Keynesian revolution. Already in 1938, however, Si- mons—certainly not a Keynesian—wrote that wide fluctuations in an- nual revenues would be “one of the greatest merits” of his scheme for making the personal income tax the centerpiece of the revenue system (1938:221). He attributed the merit, not to a direct impact on disposable income and consumption, as Keynesians did, but to the injection and withdrawal of money and effective money substitutes in connection with the financing of budget deficits and disposal of surpluses (221-22). A decade later Milton Friedman favored primary reliance on a progressive personal income tax and proposed that budget deficits be financed solely by money creation. In his scheme built-in flexibility of the income tax would contribute to stability by its impact on disposable income and on the quantity of money (1948:248—52). A great tactical advantage attributed to built-in flexibility was that both the difficulty of forecasting and lags due to delays in recognizing changing conditions and in legislative action—necessary for discretion- ary tax and expenditure changes for stabilization purposes—were avoided. Stein pointed out that reliance on variable tax yields was more palatable to business groups than the earlier emphasis on expenditure variations. A respected business group, the Committee for Economic Development, published an influential statement, Taxes and the Budget, in 1947 proposing that taxes and expenditures be set to produce a budget surplus at “an agreed high level of employment.” Built-in flex- ibility was expected to exert sufficient stabilizing force unless there was “some major change in. . . condition of national life” (Stein 1969:177— 90, 220-32). While the majority opinion among economists still appears to be that built-in flexibility is an advantage of the personal income tax, much less emphasis is placed on it than at times in the past. In addition to other reasons, the permanent income (Friedman 1957) and life cycle hypoth- eses (Modigliani and Brumberg 1954; Ando and Modigliani 1963) are particularly significant. These hypotheses indicate that temporary The Personal Income Tax changes in disposable income—whether due to built-in flexibility or discretionary tax changes—have little impact on consumption expendi- tures. Blinder and Solow in a 1974 survey concluded that “‘a reasonable guess” was that a temporary tax change would have about half as much effect on consumption as would a permanent change (109). Other inves- tigators placed the effect of temporary tax changes at much smaller fractions (Hubbard and Judd 1986). Some economists have adopted a strict interpretation of the perma- nent income-life cycle hypothesis, which attributes no effect to a tempo- rary tax change. Hall presents evidence that he interprets as showing that policy affects consumption only as it affects permanent income and that only new information on taxes and other policy instruments can affect permanent income, thus denying any influence to changes in the yield of existing taxes (1978). Other analysts have reported econometric results that they interpret as being inconsistent with a strict life cycle hypothesis. One suggested explanation is that people simply have horizons shorter than a lifetime. A second explanation (in principle more amenable to econometric inves- tigation) is that at any time a significant proportion of consumers are subject to a liquidity constraint that compels them to conform their spending to their disposable income regardless of their long-term prefer- ences. This constraint arises from inability to borrow against future earnings and from lack of information, high transaction costs, and high- risk premiums on the personal loans that are available (Hayashi 1982; Hubbard and Judd 1986). These considerations, which | think have merit, support the belief that built-in flexibility of the personal income tax has some stabilizing power, though less than formerly supposed. Reliance on the liquidity-constraint explanation, however, has the un- pleasant implication that young people and those with temporarily or chronically low income and little wealth bear the burden of adjusting to tax-caused variations in disposable income. Even on a strict life cycle interpretation of consumer behavior, built- in flexibility could exercise stabilizing influence via the impact on money and credit associated with the financing of a budget deficit, as monetarists have argued. That possibility is denied by the new classical economics, which postulates rational expectations, continuous market clearing, and an equivalence between taxes and government debt, and thus leaves no constructive role for stabilizing fiscal or monetary policy. 101 102 Richard Goode Optimal Tax Theory In the 1970s a number of theorists concerned with taxation overcame inhibitions on cardinal measurement and interpersonal comparisons of utility. They undertook to identify which taxes are optimal in terms of individual utility and an implicit or explicit social welfare function. For personal income tax rates they returned to the Edgeworth-Pigou frame- work, with the important difference that they incorporated effects on work effort, and sometimes saving also, in their formal models. Al- though Edgeworth and Pigou had recognized that those effects should be taken into account, their failure to offer a unified treatment had contrib- uted to the neglect of the neoclassical sacrifice theory of progressive taxation. The literature on optimal taxation generally is technical and complex and is growing rapidly. | cannot attempt an extended or rigor- ous description and evaluation of its treatment of personal income tax rates. Since, however, the literature conveys yet another significant change of views on the income tax, I will mention some of the main ideas and comment briefly on them.4 The optimal tax theorists assume identical utility functions and di- minishing marginal utility of income, as did Edgeworth, Pigou, and other writers of the neoclassical sacrifice school. Mirrlees (1971) initi- ated the papers on the optimal income tax rate schedule. On the basis of an assumed utility function and assumed distribution of earning capac- ity, he reached three conclusions that have greatly influenced economists’ thinking: (1) Graduated tax rates are nonoptimal; the optimal rate schedule is approximately linear, that is, a constant marginal rate with a negative tax below the exemption level; (2) the optimal marginal rate is rather low—lower than Mirrlees had expected; and (3) the income tax is much less effective for reducing inequalities than had been thought in the past. Other writers, employing similar methodology but somewhat dif- ferent assumptions concerning the utility function and labor supply, reached similar conclusions (Atkinson 1973; Feldstein 1973; Seade 1977). One conclusion that attracted attention was that the marginal tax rate on the person with the highest income in the community should be zero. If the marginal rate is reduced to zero and the highest-income person works more, he/she will be better off, the government will lose no revenue, and other taxpayers will suffer no loss of utility—a Pareto improvement. If he/she fails to react, nothing will be lost. The Personal Income Tax It seemed that optimal tax theory had turned the neoclassical sacrifice theory on its head and had provided a sophisticated indictment of graduated income tax rates that would be welcomed in some conserva- tive or reactionary circles but that would lend itself to populist carica- ture. The conclusions on tax rates, course, represented only the numerical implications of combining particular functions for utility, labor supply, and social welfare. They did not, and could not, report observed facts. Mirrlees adopted as the social welfare function a simple unweighted sum of individual utilities (as indicated by his assumed Cobb-Douglas utility function in income and leisure). This implies that society is indifferent to the distribution of income and values equally one unit of utility for a rich person and a poor one. Different assumptions about social welfare and individual utility functions will yield different numbers for optimal tax rates. Stern, for example, using different assumptions found optimal linear marginal tax rates of 50 percent to 60 percent or more (1976, 1987:41—43). He reasonably concluded that optimal income tax rates increase with the aversion to inequality and the amount of revenue to be raised and decrease with the (compensated) elasticity of labor supply. The suggestion that the marginal tax rate should be zero at the very top of the income distribution has no practical significance. Tax rates are set not for individuals but for income classes, with an open-end class at the top, which will include a number of persons. If rates could be set separately for each individual, the reasoning that suggests a zero rate for the highest-income person would apply equally to any other taxpayer. Contrary to the conventional Pareto principle, it is doubtful whether people would regard a zero marginal rate as fair even if it applied to only one person and cost no revenue. Strict versions of the permanent income or life cycle hypothesis are a significant feature of optimal tax theory. Hubbard and Judd (1986) point out that analyses of the welfare implications of tax policy are “especially sensitive to assumptions made regarding individuals’ ability to use capital markets to transfer income across time.” They argue that, if account is taken of liquidity constraints on consumers, calculations of welfare costs of taxation are altered and sometimes reversed in sign. For example, a graduated income tax, which optimal tax theorists maintain is more burdensome than a linear tax, may be advantageous for liquid- ity-constrained consumers whose earnings rise with age. Unless the exemption and negative tax component of a linear tax are very large 103 104 Richard Goode (which will require a high tax rate), it will collect more from young people and others subject to liquidity constraints and will exact a welfare cost from them that should be balanced against the benefits that others obtain from a constant marginal rate. The authors report simulations that support their reasoning. The most favorable assessment of the optimal tax literature on in- come tax rates that can be supported is that it offers a useful framework for discussion and debate. That is the position of Atkinson and Stiglitz, who write: “it is a misunderstanding of the purpose of the literature to suppose that it can yield definite policy recommendations. ... [T]he aim is rather to explore the ‘grammar of arguments’ ” (1980:456). Rosen, in an especially clear and succinct critique, states, “the accurate thing to say about the optimal tax literature is that it allows us systematically to draw out the implications of alternative ethical and behavioral assumptions, thus allowing a coherent discussion of tax policy” (1988). But in my judgment these comments may take too favorable a view of the literature. For many economists it seems to be seen more as a set of policy conclusions than as the description of a methodology. These conclusions favor low income tax rates with no graduation and rank other taxes above the income tax. The underlying assumptions of opti- mal tax theory have received too little critical evaluation. The cardinal utility assumption is a matter of convenience rather than demonstrated validity. Market clearing and the absence of liquidity constraints and of externalities in production and consumption are assumed, though usu- ally not emphasized. The distribution of benefits from government ex- penditures on goods and services and from the legal system is not explored. The discussion frequently proceeds as if the personal income tax were the only tax. The social welfare functions employed are simplis- tic interpretations of complex social and political attitudes toward eco- nomic inequality and the good society. In the circumstances great cau- tion on the part of writers and skepticism on the part of readers are appropriate. Applied General-Equilibrium Models Applied general-equilibrium tax modeling is related to optimal tax the- ory, but the use of large amounts of data and computations in the models developed in the 1970s and 1980s and the form in which results are The Personal Income Tax presented distinguish this work from the optimal tax theorizing dis- cussed in the preceding section. My comments relate particularly to the U.S. model developed by Ballard et al. (1985a; see also Shoven 1983; Ballard et al. 1985b). Like the optimal tax theorists, general-equilibrium modelers often assume the interpersonal comparability and cardinal measurement of utility and adopt utility functions on the basis of their simplicity, plausibility, and mathematical tractability rather than any empirical evidence. Other behavioral relations are selected from reports of previous research and by calibration to ensure that the benchmark case yields realistic aggregates (Whalley 1988:19). However, sensitivity tests have been applied to allow readers to appreciate the importance of the parameters of the particular standard functions chosen. Externalities in production and consumption and also risk ordinarily are ignored. Numerical results usually are related to discounted values of income, consumption, revenue, and other variables in an economy that moves from an initial competitive equilibrium to another steady state after a tax change.> In the U.S. model an additive social welfare function is calculated by applying the Kaldor compensation test, that is, the sum of the compen- sating variations that would be needed to ensure that the gainers from the tax changes considered could compensate the losers (Shoven 1983). Since, however, it is not assumed that the losers are in fact compensated, an estimated welfare gain due to a tax change does not indicate an actual Pareto improvement. Conflict criteria are not so easily transmuted into consensus Criteria. For present purposes the most relevant feature of the U.S. model is the estimates the authors derive for the marginal welfare loss (or marginal excess burden) for revenue from increases in various taxes. The estimates presented by Ballard and his coauthors in their 1985 paper (1985b) apply to a small additional amount of revenue that is used to pay for additional government expenditures on goods and services. No allow- ance is made for the utility of the goods and services provided by the government since the purpose is to furnish only the tax side of cost- benefit calculations for expenditures financed by alternative taxes. With the model builders’ standard assumptions concerning supply functions (uncompensated elasticities of saving and labor supply of 0.4 and 0.15, respectively), the marginal excess burden or welfare loss at- tributed to the U.S. personal income tax is 31 cents per dollar of mar- ginal revenue. This may be compared with figures of 28 cents for output 105 106 Richard Goode taxes, 23 cents for labor taxes at industry level, and 46 cents for capital taxes at industry level (Ballard et al. 1985b:136). The personal income tax is not a pure S-H-S tax but the U.S. federal tax at 1973 rates (ranging from 14 percent to 70 percent) with the assumption that 30 percent of savings is currently deductible (to allow for the exclusion from the base of the imputed rental value of owner-occupied dwellings and consumer durables and other provisions). The estimates are sensitive to the various elasticities in the model, about which econometric studies differ. On the assumption that the elasticities of saving and labor supply are both zero, the estimated marginal excess burden for the income tax is roughly half that for the standard assumptions, as are the estimates for output and labor taxes; the reduction of the estimate for capital taxes is greater. The results are affected to an unknown extent by other assumptions: perfect competi- tion with complete and costless information for all agents, full employ- ment, equilibrium in the base year (1973) on a steady growth path, constant returns to scale, unchanged relative prices in the absence of policy actions, a balanced government budget, immortal consumers, a constant 4 percent discount rate. A highly important assumption, em- phasized in the full description of the model (Ballard et al. 1985a) but not mentioned in the paper on marginal excess burden, is that interna- tional capital movements are unaffected by tax changes. On this assump- tion changes in saving result in equal changes in domestic investment, but that is not so in an open economy. An assumption of high interna- tional capital mobility could change the estimate of the effect of a tax- induced increase in saving from a gain to a loss in the taxing country (Ballard et al. 1985a:237). The most striking feature of the estimates is their large numerical values. Consistent with received theory, the estimated marginal excess burden of taxation increases with tax rates, the elasticities of the factors affected, and differences in tax rates applied to a factor of production in different uses. But the usefulness of numerical estimates of the marginal welfare cost of alternative taxes is questionable. The estimates derived from the U.S. model are stated as aggregates or averages without regard to differences in distribution and without allowance for social or individual prefer- ences regarding the distribution of income and wealth. Potential com- pensation is not the same as actual compensation of losers. Even if actual compensation were paid that would not, and should not, settle the The Personal Income Tax ethical and political questions of whether welfare has been improved or harmed in a meaningful sense and justice done or impaired when a change in tax rates and government expenditures occurs. Welfare—even if modified by the word “economic”—is not to be understood in a narrow, technical sense. It is clear, nevertheless, that some kind of general-equilibrium analysis is needed to evaluate a big change in taxation or other policy. The general-equilibrium modelers are to be commended for their work and for pressing ahead even though the data base is unsatisfactory in many respects and critical functions are problematical. In this rapidly develop- ing field model builders are attempting to incorporate elements not adequately treated in early general purpose models, including risk, finan- cial decisions, international trade and factor flows, and market struc- ture. Specialized models are being developed to deal with particular subjects. In all the general-equilibrium models dealing with taxation, as in partial-equilibrium simulations, a difficult issue relates to the treatment of the distant future. The usual practice is to apply a uniform discount rate. The choice of the rate is important since a low rate will favor a policy promising remote gains and a high rate will favor one yielding quick results. Discounting cannot satisfactorily allow for uncertainty (unique events for which a probability distribution cannot be devised) or balance the interests of the present generation against those of future generations. While I agree with Shoven’s statement that general-equilibrium anal- ysis is now “a tool that can be used for policy purposes” (1983:417), the function of the tool needs explanation. In my opinion far too much emphasis has been placed on numerical estimates of the aggregate welfare consequences of tax rates and tax changes. I do not think such numbers are useful for policy making. Working with general-equilibrium models, analysts can make the hum- bler, though essential, contribution of calling attention to effects ne- glected in popular thinking and in partial-equilibrium analysis and con- veying to legislators and other policy makers a sense of economic magnitudes that they can use in forming their own value judgments. In particular they can try to refine estimates of the distribution of taxes by income classes and socioeconomic categories such as income source, wealth, and age (Bovenberg 1987:37—38). They can try to identify and quantify the effects of graduated income tax rates on labor supply, 107 108 Richard Goode private saving, investment, national income, and the balance of pay- ments. Policy makers are demonstrably interested in such analyses if not in estimates of the present value of “welfare” over a long time period. Modelers can try to make their models more realistic and the estimates more reliable, but in the effort the models will become more complex and may increasingly take on the character of a “black box” (Whalley 1988:50). So far, numerical general-equilibrium analysis appears to have influ- enced economists’ views less than optimal tax policy has, probably because of the models’ structural complexity, lack of tested specifica- tions, and use of imperfect and unreliable data (Whalley 1988:15). In my judgment, however, the general-equilibrium approach, if directed as suggested in the preceding paragraph, holds more long-run promise than does largely deductive optimal tax theory. Meanwhile policy makers and their advisers should be aware that the predictive value of point esti- mates may be very limited (Bovenberg 1987:36) and that alternative general-equilibrium models can assess income tax rate schedules quite differently. Like forecasting models, general-equilibrium models may come to be used by lobbyists to attack or support legislative proposals. Ability to Pay and Power: Consensus and Conflict In democratic countries the choice of the income tax rate schedule does not flow from any great overarching theory but from a compromise or series of compromises. All the normative analyses reviewed so far are incomplete and suffer to a greater or lesser degree from oversimplifica- tion, despite the technical virtuosity and ambitious scope of some of them. Here I will continue in the normative mode but will comment very briefly on the criteria that figure in public debates on tax rates and in the formulation of recommendations of civil servants and other specialists in applied tax research. While I believe the normative criteria influence legislation, any attempt to evaluate the degree of influence would go far beyond the limits of this paper.® Referring again to Carl Shoup’s insights, I identify three conflict and four consensus criteria that appear to influence discourse on income tax rates. The conflict criteria are (1) exemption of incomes below a subsis- tence or poverty level; (2) mild progressivity over a broad middle range obtained by personal exemptions and some rate graduation; and (3) The Personal Income Tax steep graduation at high income levels. (Other conflict criteria deal with the choice of the tax base and tax mix and relate only indirectly to income tax rates.) The first criterion is the least controversial and, indeed, might be classified as a consensus criterion. It is founded on humane sentiments and on the recognition that in contemporary conditions in rich countries income taxes collected from the very poor are likely to be offset by increased welfare benefits. The second criterion reflects a fairly vague sacrifice notion—how much one is hurt by paying taxes. It can be sharpened by two observations. One is that the poverty level should not be rigidly defined but may be seen as a zone over which humane senti- ments gradually give way to revenue demands. Also, it is plausible, as previously mentioned, to suppose that people use their income first to satisfy their most urgent needs and use successive increments of income for purposes that they and the community value less highly. Income tax progressivity can help offset regressive indirect taxes and payroll taxes. The third criterion is less firmly based than the first two and currently receives the least overt support. It reflects concern over the economic and political power of persons with high income and great wealth. It also serves the purpose of helping to reconcile people with smaller incomes to their own tax bills and to distasteful measures such as wage controls in times of emergency. The three criteria are matters of informed opinion, which cannot be captured by a numerical formula or wide public opin- ion polling but which can be refined by discussion and expert assistance in determining the poverty level and in assessing the consequences of taxation above that level. The main consensus criteria bearing on income tax rates are (1) adequate revenue; (2) avoidance of excessive interference with employ- ment, production, and growth; (3) macroeconomic stability; and (4) practicable administration and compliance. Of these stability, in the sense of automatic or discretionary variations in income tax liability to limit fluctuations in aggregate demand, now receives less emphasis than in the past but seems not to have been discarded by the majority of expert advisers of policy makers. Administration and compliance, in contrast, are increasingly stressed owing to concern over tax avoidance and eva- sion, which have been facilitated by weaknesses in tax legislation and procedures, by deregulation and innovations in financial markets, and by increased international mobility of capital and labor and which are believed to be stimulated by high tax rates and perceived inequities. 109 110 Richard Goode By the political process the conflict and consensus criteria have to be balanced to arrive at a rate schedule. As I see it, policy makers do not seek—and should not accept—from economists and other experts a definitive finding on what rate schedule is best. Economists and other experts can contribute by advising on the impact of income taxes on the points covered by the criteria. It is also appropriate for them to express opinions on the validity of the criteria and the weight to be attached to each of them and on the trade-offs between the conflict and consensus criteria. Ideally, analysis of the consensus items would not be influenced by opinions on the conflict criteria, but experts, being fallible human beings, find it nearly impossible to be totally objective. Participants in the debates on taxation often allege consensus-criteria arguments to support positions that actually are based on conflict criteria.” The Expenditure Tax Challenge Until recently most writers on public finance took it for granted that the personal income tax is the best tax. That view has been vigorously challenged during the past two decades, and many writers have held that a direct tax on personal consumption (here called an expenditure tax) would be preferable. Indeed, the latter position may now be that of a majority of academics. The idea that, abstractly considered, an expenditure tax is better than an income tax is not new. John Stuart Mill, Marshall, Pigou, and Einaudi espoused it but thought that a direct consumption tax was impractica- ble. Irving Fisher advocated such a tax, which he called an income tax, and argued that it was practical but made few converts. The recent wide acceptance of the expenditure tax idea appears to be due to a restatement and extension of the theoretical arguments, a different assessment of the administrative problem, and certain broad economic and political de- velopments.? The early statements of the case held that an income tax results in unfair double taxation of savings because the tax applies to income from which savings are made and also to the return on the savings. This could be remedied either by deducting savings from the tax base or by exclud- ing the return on savings. The early discussions concentrated on the former approach, presumably partly because of the difficulty of identify- The Personal Income Tax ing the returns that should be exempt and partly because saving seems more virtuous and praiseworthy than receiving interest or other prop- erty income. Lately attention has been given to the alternative of exempt- ing investment returns rather than deducting savings. The exemption approach was brought to economists’ notice by the U.S. Treasury De- partment report, Blueprints for Basic Tax Reform (1977), and the Meade Committee report (Meade 1978). The alternatives and other problems of administration will be taken up again later. The double-taxation-of-savings phraseology is now seldom encoun- tered. The equity point is made by asserting that the income tax discrimi- nates against persons who save early in life and consume later, whereas an expenditure tax is neutral. This emphasis, of course, is congenial to analysts who adopt the life cycle hypothesis for consumption. While the underlying substance remains the same, the equity argu- ment looks weaker when we note that the income tax does not in fact impose a tax on the act of saving but a tax on any return obtained if the savings are successfully invested. Persons who choose to postpone con- sumption are not penalized; they simply are prevented from increasing their consumption power (or opportunities) as much as with an expen- diture tax. In this sense an income tax treats recipients of labor income and interest recipients similarly, whereas an expenditure tax favors the latter. This is contrary to a long tradition holding that labor income deserves preferential treatment because personal effort is required to earn it and because it disappears with the disability or death of the earner. Ironically, Kaldor, whose influential book, An Expenditure Tax (1955), introduced a new generation of economists to the subject, as- serted that the case for an expenditure tax does not rest on any need to correct a discrimination against saving under an income tax but arises from the failure of the income tax to apply to “the spending power that is exercised through ‘dissavings’ (or spending out of capital).” He added, “taxation according to ‘income’ introduces a bias in favour of property owners whose taxable capacity is underrated relatively to those who derive their income from work” (14). His specific proposal was to add an expenditure tax to the income tax. The rates of an expenditure tax would have to be higher than those of an income tax to obtain the same current revenue yield and more steeply graduated to maintain the same degree of progressivity with respect to income. The significance of these observations, however, can be ques- 111 112 Richard Goode tioned by maintaining that government revenue should be measured in present value terms and that progressivity should be judged on a life cycle basis. If all postponed consumption is taxed, and if gifts and bequests are classified as taxable consumption of the donor, it can be argued that in present value terms the yield and burden of the expendi- ture tax will be the same as that of an equal-rate income tax. (Actually, the present value of the expenditure tax will be greater to the extent that it applies to the use of preexisting wealth for consumption, gifts, and bequests.) This interpretation led Aaron and Galper to write that a tax on lifetime consumption, gifts, and bequests would be equivalent to an S-H- S income tax in present value terms (1985:29). This conclusion depends on the strong assumptions that tax rates and the discount rate are constant and the capital market allows everyone to lend and borrow at the discount rate. As explained in the discussion of built-in flexibility, imperfections of the capital market can significantly affect the distribu- tion of the burden of taxation; failure to allow for these effects biases the analysis in favor of a consumption tax (Hubbard and Judd 1986). Aaron and Galper also disregard the fact that tax revenue and the budget balance are not perceived by either policy makers or the public in present value terms over long periods of time. It is hard to believe that gifts and bequests would be treated as consumption of the donors, since this is contrary to common under- standing and would run counter to the attitudes that have allowed the transfer taxes to decline in the United States and to disappear at the central government level in Australia and Canada. Without effective taxation of gifts and bequests, an expenditure tax regime would end the mild restraint that present-day income taxes impose on the growth of dynastic fortunes. Even with taxation of the transfers, the unchecked growth of wealth during a person’s lifetime could be seen as unfair and could raise political and social questions. From the standpoint of economic policy the original case for the expenditure tax was that it would stimulate private saving, which was regarded as unquestionably desirable. That view of saving is again gener- ally accepted, after a period in which it was challenged. A simple partial- equilibrium analysis indicates that the increase in the net return on invested savings made possible by substituting an expenditure tax for the income tax would tend to stimulate saving via the substitution effect; the income effect, which pulls in the opposite direction when an exogenous The Personal Income Tax change in the rate of return occurs, would be largely neutralized in the case of equal-yield taxes. But the extent of the increase in saving is uncertain. Econometric studies of interest elasticity are less numerous and less conclusive than might be expected in view of the importance of that parameter. Boskin’s 1978 paper—which is still influential, despite severe criticisms—reported a much higher positive elasticity than had been indicated by earlier research, a figure of about 0.4 with respect to the real after-tax rate of return on capital (see also Howrey and Hymans 1980; Pechman 1980; Bosworth 1984:67—96; and Starrett 1988). According to received theory, economic distortions caused by taxa- tion increase with the supply elasticities of the taxed factors. It can be argued, moreover, that even a zero interest elasticity does not imply an absence of distortion of consumer preferences and hence of inefficiency (Feldstein 1978; Mieszkowski 1980). This argument, however, is some- what esoteric and is likely to carry less weight with policy makers than results derived more simply from estimates of interest elasticity. At present there is no consensus on the relative economic efficiency costs of income and expenditure taxes. The usual theoretical analysis indicates that an income tax distorts both labor/leisure and saving/ consumption decisions and that an expenditure tax avoids the latter distortion but, because its rates must be higher, may distort the la- bor/leisure choice more severely. An additional point is that a new expenditure tax by striking consumption financed by preexisting wealth would to that extent resemble a nondistorting lump-sum tax (Auerbach and Kotlikoff 1987b:79—81).? But persons who hold little wealth and are liquidity constrained cannot take full advantage of the presumed opportunity that an expenditure tax offers to optimize saving/consump- tion decisions. Hence it is uncertain which tax is less inefficient. My opinion is that this analysis is correct so far as it goes but that it reflects too narrow an interpretation of efficiency and welfare. The recent endorsement of an expenditure tax by many writers is based not only on theoretical considerations but on a reversal of the previous belief that the tax is administratively impractical. That judg- ment was based on the supposition that an expenditure tax would be assessed by first determining income and then subtracting savings. Dis- honest taxpayers, it was pointed out, could understate their consump- tion by reporting savings but concealing dissavings. Reviving Fisher’s proposals for assessing the expenditure tax on the basis of cash flows supplemented by partial balance sheet information, als) 114. Richard Goode Kaldor concluded that an expenditure tax was practicable though more complicated than the existing British type of income tax (but no more complicated than an S-H-S income tax) (1955:222). | argued that com- plete balance sheets would be needed to verify reported cash receipts and disbursements for investment purposes (including changes in cash bal- ances) and thus to arrive at a residual equal to consumption, and that it would be infeasible to obtain balance sheets from everyone. I concluded that an expenditure tax would be more difficult to administer than a tax on realized income but less difficult than an income tax including ac- crued capital gains, that is, an S-H-S income tax (1964:33). Andrews, in a paper that was especially influential with American lawyers, claimed that an expenditure tax (which he called a cash-flow income tax) could be successfully administered solely on the basis of cash flows (1974) but did not dispel my skepticism. Critics of the income tax now often maintain that an expenditure tax would be easier to administer. They call attention particularly to avoid- ance of the difficulties of assessing accrued capital gains, measuring economic depreciation, and correcting interest income and expense for inflation. They also hold that the so-called prepayment feature of an expenditure tax would neatly tax the imputed return on consumer dur- ables and owner-occupied dwellings. The reasoning is that, in a cash- flow expenditure tax, if outlays for durables and houses are not treated as deductible investments, the imputed return is implicitly taxed. On certain assumptions the “prepaid” tax equals in present value the annual expenditure taxes that would apply over the useful life of the item under an imputation system. Some writers have proposed that taxpayers be allowed to choose either prepayment or annual taxation. The reasoning and practical sug- gestion also have been extended to financial and business investment. Taxpayers could choose either (1) to claim a deduction for an investment and include the future returns in receipts or (2) to claim no deduction when the investment is made and exclude future returns from receipts. Owing to uncertainty and differences in rates of return, however, the two methods are not equivalent. The so-called prepayment system would favor those who realize above-average returns and would omit from the tax base part of their consumption expenditures. The option would allow manipulation and gaming of the tax system (Graetz 1980; and report of discussion in Pechman 1980). My summary judgment is that recent analysis has correctly shown The Personal Income Tax 115 that an expenditure tax would have some administrative merits, espe- cially in inflationary conditions, but has not convincingly supported the contention that it would be administratively preferable to an income tax that assesses capital gains only when realized. I remain convinced that the personal income tax is superior to an expenditure tax. The familiarity of the income tax and its compatibility with customary accounting systems, adaptability to collection at the source, use by many countries, and incorporation in tax treaties are advantages that should not be overlooked. Continuation with the in- come tax avoids the practical problems and equity issues that would arise in transition to an expenditure tax. Income is a broader index of ability to pay; consumption expenditure omits the part of ability to pay associated with individual decisions to save. Even an income tax that fails to assess capital gains as they accrue reaches better the personal benefits of wealth accumulation. An expenditure tax that failed to treat gifts and bequests as taxable expenditures would allow the perpetuation of family fortunes unless accompanied by heavy and effective gift and death taxes. If an increase in national saving is a policy objective, a smaller budget deficit is a more certain and more powerful action than replacing the income tax with an expenditure tax. Concluding Remarks This account—summary and selective and indeed cursory in many re- spects—has shown that views of the personal income tax have changed greatly during the twentieth century. As Simons cautioned, readers of the literature should make some allowance for its relationship to prevailing opinions. The changing style and substance of economic analysis clearly influenced the views of scholars whose work has been noticed here. Less clear, but profoundly important I feel sure, is the influence of the eco- nomic environment; political trends; the preconceptions, values, and self interest of the scholars and the policy makers whom they sought to advise. And the impact of the proffered advice is doubtful. Legislators did not adopt the S-H-S definition of taxable income, did not enact draconian progressivity in an effort to minimize aggregate sacrifice or lessen in- equality, were unconvinced by the stagnation hypothesis, have lost inter- est in the stabilizing value of discretionary tax rate changes and built-in 116 Richard Goode flexibility, and have been oblivious of refined theoretical criticisms of the inherent equity and efficiency of the income tax. Possibly economists who were influenced by optimal tax theory reinforced policy makers’ growing doubts about the wisdom of highly progressive tax rates. Legis- lators appear to have responded to a modest degree to normative and empirical work on the relationship between the tax base and tax rates— but also to the fantasies of supply siders. Probably some of the other ideas, good and bad, are percolating through the seedbeds in which future legislation will sprout. I will not try to predict future views of the personal income tax. But I am sure that they will be diverse enough and changeable enough to offer ample scope for scholars with different values and different comparative advantages. Notes I acknowledge with thanks access to an early version of Alan A. Tait’s paper on general equilibrium tax modeling and optimal taxation (1989) and helpful comments by A. Lans Bovenberg and Joseph A. Pechman on drafts of this paper. 1. Friedman (1947:409—11) criticized Lerner’s “appeal to equal ignorance” but did not dispute the conclusion. 2. For detailed critiques, see Fagan (1938); Blum and Kalven (1953, 1963). 3. Elasticity equals (AT/T) (AY/Y), while built-in flexibility equals AT/AY, where T is tax liability and Y is national income. 4. For general surveys and comments on the literature, see Atkinson and Stiglitz (1980) and Stern (1987). 5. For a general survey by a leading modeler, see Whalley (1988). 6. Two recent contributions to positive analysis of income tax rates, the first by a political scientist and the second by two economists, are Witte (1985) and Hettich and Winer (1988). Another paper by Hettich and Winer (1989) presents an informative review of formal models combining economic and political elements that are intended for positive analysis of decisions on income taxation. My approach is compatible with Witte’s but quite different from the formal models discussed by Hettich and Winer. 7. An illuminating treatment of conflict and harmony in taxation and the roles of pressure groups, experts, legislators, and administrators can be found in Blough (1952). 8. From the extensive literature three items that offer convenient reviews of the issues may be cited: Pechman (1980); United States, Department of the The Personal Income Tax Treasury (1977); Meade (1978). The first book includes references to many other publications. 9. According to Auerbach and Kotlikoff, however, if compensation were paid to the losers (largely the elderly), it would absorb most of the long-run gains in efficiency that they attribute to substituting an expenditure tax for an income tax. IU7/ VAs WELL LAsMer¥ bCik RoEPY The Corporate Income Tax and How to Get Rid of It Why Do We Have a Corporate Income Tax? Popularity versus Rationality Of all the major U.S. federal taxes the corporation income tax in its present form is the one with the widest gap between its political popularity among the public at large and its esteem by economists. The naive public seems to think of the tax as in some way borne by the corporation, while even the more sophisticated may think of the tax as being borne primarily by stockholders or management or by investors generally. Many among management, however, believe firmly that the tax is treated as a business cost and is passed on to consumers. Thus a large part of the public believes that the burden of the tax is not borne by them but by somebody else. Economists are by no means unanimous in their opinion as to where the burden falls. Some have claimed that the tax is passed on to con- sumers by more than 100 percent. Others see a burden distributed over all types of investors. The difficulty is that where a tax differs markedly from other reference revenue sources or expenditures in its macroeco- nomic impact, a great deal depends on the assumptions made, implicitly or otherwise, as to how the macroeconomic impact is handled. Indeed under some not unreasonable assumptions it appears that the burden will be felt largely by future generations rather than by present tax- payers. The Lack of Underlying Rationale In the United States, especially, the tax has developed largely without any clear economic rationale, other than possibly Colbert’s cynical principle The Corporate Income Tax 119 of getting the most feathers from the goose with the least squawking, or the peculiar legal logic that insists that since a corporation is defined to be a “juridical person” a “personal” income tax must somehow be applied to corporations. Indeed, in some jurisdictions the corporate tax has been applied at the same graduated rates as the individual tax. While in most cases some concession has been made to avoid or mitigate “double taxation,” the tax has not been treated as merely a collection at source of a part of the individual income tax, unlike the former British practice. Problems with the Corporate Income Tax Taxes Above and Below the Capital Market The crucial fact to be taken into account is that since 1936 the U.S. corporation tax has been a separate tax operating to a large extent as an addition to the market rate of return as a component of the cost of equity capital that must be expected to be covered if a new investment is to become attractive. Up to 1936 dividends were exempt from the normal tax so that at least part of the tax could be considered a collection at source of a part of the individual income tax, and thus to that extent not an addition to the cost of capital relative to market rates in general. While the corporation tax rate was generally higher than the normal tax, there was an additional effect through the removal of the corporation tax from the surtax base, so that in extreme cases the tax burden on a stockholder of a corporation distributing all of its income would be lower than that on a member of a partnership with a comparable amount of pretax income. The net effect varied from year to year and from taxpayer to taxpayer, but the overall effect of the tax as an addition to the cost of capital was relatively slight. At the time the change was made, to be sure, subjecting dividends to an additional 4 percent individual income tax seemed a relatively mild measure that appeared to increase the progressivity of the tax. But the pattern thus established came to have serious consequences as during the decades after the war investors came to expect that combined corporate rates above the market at combined state and federal rates on the order of 50 percent would continue indefinitely. The matter is somewhat complicated by the fact that where part of the profits remain undistributed there is some reduction in the burden of the 120 William Vickrey individual tax burden. Even if the profits are eventually distributed as dividends, or realized as a fully taxable capital gain, the deferment of the realization of taxable income results in a burden reduction equivalent to an interest-free loan of the amount of tax deferred. In addition the income may become totally exempt if the stock is held to the death of the taxpayer or used as a vehicle for a charitable contribution. Until recently the gains were favored by lowered rates in addition to the deferment factor, even if the stock were sold outright. There is thus some leeway for considering that a part of the current corporate tax is absorbed below the market as an offset to the favored treatment of corporate earnings under the individual tax. But the differ- ence remains: the pre-1936 tax was primarily a tax levied after the market that caused slight if any differentials between the market rate of return and the returns required for investment of equity capital, while the post-1936 tax became largely an additional requirement to be cov- ered by the prospective return on equity investment. Dependence of Tax Impact on Monetary and Fiscal Policy The special macro impact of the corporate income tax can be seen in the scenario of what might happen if the individual income tax is reduced by $1 billion and the corporate tax increased by a billion so as to leave the budget balance unchanged, while market rates of interest remain un- changed, either as a result of monetary policy in a closed economy or by reason of the openness of the international capital markets. The result is a curtailment of equity-financed investment with little or no offsetting increase in consumption expenditure, assuming the income tax reduc- tion to be so graduated that the marginal propensity to consume from the tax reduction would be roughly comparable to that out of corporate earnings. The net result will be a reduction in aggregate demand and a reduction in employment. Such a reduction in employment is not what one normally thinks of as part of the incidence of the tax. In this scenario the corporate income tax in effect has a double depressing effect on GNP: one effect through the subtraction of the revenue from the flow of purchasing power, offset by an equal reduction in the individual income tax, and another effect through the reduction in investment, not so offset. To arrive at a version of incidence in which total Gnp will be main- tained, some adjustment will be necessary. One way would be to increase The Corporate Income Tax the budgetary deficit sufficiently to return to circulation the savings that are no longer being recirculated through private investment. In a simple but possibly rather extreme case, suppose that an increase of $1 billion in current government outlays is financed by a like increase in the corpora- tion income tax revenue. Assume further, as one possibility, that this produces, in the absence of any change in the level of market interest rates, a reduction of $1 billion in private domestic investment, so that to maintain overall GNP some way of assuring the employment of the resources thus released must be adopted. One way of doing this would be to reduce the individual income tax by $1 billion, bringing disposable income back to its former level. The savings out of this income that were formerly used to finance private investment would now be used to finance the corresponding deficit. Resources formerly devoted to private investment will be diverted to providing the new government services. If individuals consider govern- ment bonds to be as good as the private investment they replace, disre- garding the collective future liability to service the bonds, the govern- ment expenditure will have been financed without any current aggregate burden being felt by present taxpayers, although some may gain and others lose to the extent of a distributional mismatch between the im- pacts of the corporate tax and of the change in the individual tax. The main eventual burden will be felt by those who are deprived of the benefits from the private investment, that is, future wage earners whose marginal productivity is impaired by the lack of capital with which to work. If the impact of the corporate tax on private investment were more than the amount of the tax, say $1.2 billion, the reduction in individual income tax required to maintain full employment would be $1.2 billion so that the net result is to make individuals feel $.2 billion better off. This and the government expenditures being financed by the corporate tax would then in effect be obtained at the expense of $1.2 billion of burden deferred to the future. Conversely, if the reduction in private investment were only $.6 billion, the net effect would be that $.4 billion of the corporation income tax revenues would be felt as a net aggregate burden on present taxpayers and $.6 billion of the burden would be passed on to the future. These figures would have to be further modified to the extent that the government expenditures being financed consist of capital out- lays not competitive with private investment, or outlays on current account that elicit complementary private investment outlays. i2il 122 William Vickrey In these scenarios in which the imposition of a corporate income tax is coupled with a monetary policy that maintains market rates of interest and a fiscal policy that maintains full employment, the corporate income tax becomes in effect the active ingredient in a Dr. Feelgood cocktail which provides current gratification at the expense of a hidden burden passed on to the future. Conversely, abolition of the corporation income tax (in sharp contradistinction from the reduction of taxes on capital gains) could be an essential ingredient in a “supply-side” austerity pro- gram that would permit full employment to be maintained in the face of budget deficit reduction. An alternative scenario in a closed economy is for the monetary authority to lower interest rates sufficiently to maintain aggregate invest- ment unchanged, though probably altered in composition (assuming that interest rates are sufficiently high to permit this to be done without encountering a liquidity trap). In this case the long-run effect is to cause the burden to be felt in the form of lower rates of return on capital, though short-run transitory effects following a change would involve offsetting windfall gains and losses. A full offsetting of the impact of the corporate income tax on invest- ment in this fashion seems unlikely, however, even if conditions are such that the liquidity trap is not approached. Unless the economy is a closed one, or interjurisdictional capital movements are tightly controlled, capi- tal movements from abroad are likely to frustrate any attempt to move domestic interest rates far from levels prevailing in the major interna- tional capital markets. Even under closed economy conditions, the inher- ent interest of the financial community, and of the monetary authorities that operate in a symbiotic relationship with it, is strongly in favor of higher interest rates with the associated higher profit margins in financial transactions and the higher incomes derived by investors and owners of capital. In addition, to the extent that monetarist dogmas prevail, suffi- cient expansion of the money supply to offset the impact of the corporate income tax on investment is likely to be resisted for fear of generating inflation. This will be true in general even in the case where there are no multinational corporations and all international trade takes place be- tween independent agencies operating only within single jurisdictions, and international investment takes place in terms of trade in the se- curities of such entities. The complexities of any analysis of situations in which multinational corporations are of importance with all of the The Corporate Income Tax 123 intricacies involved in the taxation of such entities boggle the mind and will not be tackled here. On balance there is a strong likelihood that the corporation income tax is among the most powerful influences tending to reduce investment and thereby real savings and economic growth. In contrast it should be noted that attempts to promote savings and growth through encourage- ments to individual savings with 1RA and similar provisions may have the opposite of the intended effect. If I save by forgoing a haircut for a month, there is five dollars more in my bank account and five dollars less in the barber’s, and no increase in the funds available to finance invest- ment. On the contrary the likely consequent decrease in the barber’s consumption would be a reduction in demand that would if anything discourage investment and with it aggregate savings, even without a “multiplier effect.” Those who are concerned to promote economic growth would do better to work to eliminate the corporation tax than to tinker with methods of promoting individual savings. The savings pro- motion provisions, including the concessions to capital gains on the pretext that this would promote savings, may at a deeper level be moti- vated by a subtle desire to weaken the income tax generally in the interests of those with a substantial financial stake. Effects of the Tax on Economic Efficiency Whatever the conclusions may be as to the incidence and macroeco- nomic impacts of the tax, the indictment against the corporation income tax is a long one. Even if the impact of the tax on the rate of economic growth and the choice between current consumption and investment for the future were countered by other macroeconomic measures, its effects on the organization and efficiency of the economic system would be serious. One such impact is to discriminate heavily in favor of types of investment that can readily be hypothecated and financed with debt, such as buildings and structures adaptable to a wide variety of uses or vehicles that can be easily shifted to new locations and uses, as con- trasted with specialized equipment, or speculative investment in mineral exploration or research and development that is more likely to require financing through equity instruments. This is the more serious in that the types of investments discriminated against tend to be the ones most likely to generate positive externalities, for example in the form of knowledge, 124 William Vickrey positive or negative, concerning the potentials for progress in different lines of development. A related inefficiency concerns the pressure provided by the tax to trade on equity and push leverage beyond natural equilibrium levels. This increases the frequency of bankruptcy and other similar procedures with their attendant disruptions and costs. It is perhaps ironic that while one of the pretexts often advanced for a tax on corporations is the externality of the losses imposed on creditors through the limited lia- bility feature, the effect of the tax is to increase the likelihood of such losses. A tax that included interest and rents paid in its base would be more to the point. Still another baneful effect of the tax is the lubrication it provides for the spate of leveraged buyouts that involve the conversion of taxable net income into interest on junk bonds. Indeed in many cases most if not the entire gain to the perpetrators of the takeover or merger comes from reductions in taxes. If there is any residual increase in managerial effi- ciency such as is touted as the ostensible justification for the operations, it is often more than offset by costs of the brokers and lawyers fees and other unproductive use of resources in the process that does not contrib- ute to the national welfare, however much it may be counted in the nominal GNP. It has also been alleged that the large number of highly leveraged situations might in the event of a business downturn lead to a spate of defaults that could seriously aggravate the slump. And the “heads I win, tails you lose” aspect of the tax as administered in practice is another element in the discriminatory impact of the tax against risky ventures. A related defect that is not inherent but is the result of the failure of the tax in practice to provide adequate allowance for losses is the incen- tive provided for mergers and other transactions carried out for the purpose of obtaining a fuller tax allowance for various losses and deduc- tions. Aside from the direct costs involved in such transactions, there is a tendency for the resulting agglomerations to develop a degree of monop- olistic market power that is economically inefficient, however profitable it may be. A more subtle defect of the tax rests in the very uncertainty concern- ing its incidence, which makes it difficult to make rational choices as to the balance between the benefits of government outlays and the taxes needed to finance them. Privatizers who suspect all government outlays of being inefficient or wasteful should be especially antagonistic to a tax that hides its burden so effectively. Even though some advocates of The Corporate Income Tax 125 higher levels of government activity might welcome financing from a tax that would encounter less political opposition, real democracy would call for their defending their proposals before the court of public opinion in terms that reflect a realistic appraisal of the alternatives. Actually our political institutions are poorly designed to promote reasoned balancing of the benefits of an expenditure and the burdens of the tax needed to finance it. Authorizations and appropriations are made by and large without reference to the taxes that may be needed to finance them, and taxes are levied by and large without reference to the outlays that they are to finance. Ideally, at some point a reasoned decision should be possible for or against an added outlay together with specifications as to the mode of its financing, whether by contemporaneous taxes or by borrowing coupled with specified future taxes. But how to achieve this would take us too far afield for the present occasion. International Tax Effects Not only does the tax have a baneful effect on the domestic economy, but through the foreign tax credit it creates an incentive for foreign jurisdic- tions to adopt similar taxes not only as a matter of naive emulation but as a means of taking advantage of the tax credit. In effect the levying of an eligible tax by a foreign government on an American corporation transfers revenues from the U.S. treasury to the foreign treasury without imposing a net additional burden on the corporation. The converse of this is that an attempt by a foreign country to encourage investment in particular enterprises by granting tax conces- sions may be frustrated by the resulting reduction in the foreign tax credit, so that instead of being an effective encouragement the net effect is merely to transfer funds from the foreign government to the U.S. treasury. This has led to proposals for “tax sparing” provisions in which a credit would be granted for a tax that would normally have been paid even though a special exemption is provided. But the notion of giving a credit for a hypothetical tax which is not actually paid and which is assessed by a tax administration that may not have the highest standards of integrity is one which understandably is not greeted with enthusiasm by U.S. tax authorities. The foreign tax credit also presents a temptation to engage in various charades of levying taxes with one hand and dis- pensing subsidies, either in cash or in kind, with the other. The foreign tax credit is indeed a fundamentally flawed method of 126 William Vickrey dealing with the problem of possible double taxation of income pro- duced in one jurisdiction and received in another. The basic problem arises from including in the base for a single tax both income produced and income received. Even with the tax credit device, the net effect is to subject income to whichever of the tax rates of the origin jurisdiction or the destination jurisdiction is the greater. Whenever these two rates differ, there will be an effective discrimination against international income payments. While this problem is mainly one involving the corpo- ration tax, it would have to be faced for individual income tax purposes even if the corporate tax is eliminated. How to Get Rid of the Corporate Income Tax Given the baneful impact of the corporation income tax, there remains a serious problem of how to get rid of it. One cannot simply abolish the tax out of hand: there would remain the problem of dealing with the ineq- uities involved in the postponement of the realization of individual income through the retention of undistributed earnings in the corpora- tion. There is also in the short run the problem of arranging the transi- tion with a minimum of disruptive transitional impacts and inequitable windfalls. Cumulative Assessment In a closed economy the sovereign long-term remedy would be to assess the individual income tax on a cumulative basis. This would not only solve the problem of deferment of income realization through retention of corporate earnings, but would constitute a master stroke of tax simplification as well, eliminating all questions as to the timing of in- come, as in determining of depreciation rates, timing of assets sales, and the like. This is achieved by first computing a global tax on the income thus far reported since some starting date, according to tables and schedules appropriate to the period covered, and then treating all pre- vious tax payments with respect to this income as deposits in an interest- bearing account and crediting the amount of this account against the global tax. In this way any deferment of the reporting of income results merely in borrowing the amount of the deferred tax at a suitable rate of interest. The Corporate Income Tax 127 One difficulty with this method arises where taxpayers enter and leave the jurisdiction, especially if such movements are repeated. For such taxpayers special provisions may be required that might seriously detract from the overall simplicity of the tax structure as a whole. Another difficulty is that to be fully effective in achieving simplification many of the special privileges and discriminations embedded in the present tax might have to be abrogated or modified, which might be opposed by those who fear that the substitution of more explicit subventions for the indirect “tax expenditures” would expose them to criticism. Converting the Tax to a Collection at Source Politically subtle opposition to such a solution is likely to arise from those with a vested interest in the complexities of the present law. Major simplification might deprive politicians of bargaining chips to play with, and tax practitioners might find that much of their expertise is of little value and that they are in need of a more productive form of employ- ment. Also, there is likely to remain a political demand for something that will at least give the appearance of a tax on the corporation. More significantly, a corporation tax may be needed as a means of collecting tax on a source basis from the operations of foreign-based corporations. This leads naturally to a return to the pre-1936 practice of using the corporation tax as a collection at source of a “normal” tax, distinguished from the surtax which would comprise the upper reaches of the progres- sive scale. Preferably one would use the British practice of using the same rate for the corporation tax as for the normal tax on individuals from which dividends would be exempt. As an alternative to the “grossing up” of net dividends for purposes of computing the base for the surtax, however, one might consider allowing the normal tax on other income to be deducted in computing the surtax base. To some naive observers this will tend to make the top rates appear to be more progressive than they actually are: for example, a top 50 percent rate would combine with a normal tax rate of 20 percent to produce an overall top rate of 60 percent rather than 70 percent, which may appear to be a slight advantage to those who feel that populist sentiment tends to carry the progression of rates too far. The deduction procedure may also be more acceptable to some taxpayers than the seemingly more adverse practice of “grossing up.” It would also fit more naturally with the treatment of dividends from foreign corporations as noted above. 128 William Vickrey Dealing with Undistributed Profits In the absence of cumulative assessment there would remain the advan- tage gained by the deferment of income realization through the retention of earnings in the corporation. The amount of this advantage is equiv- alent to an interest-free loan of the amount of the tax deferred. This naturally leads to the consideration of some form of undistributed prof- its tax. The types of tax considered and experimented with in the thirties proved unsatisfactory, however, as they were oriented at least in part toward compelling the full distribution of earnings for the sake of dis- tributing additional purchasing power as a counterdepression measure, as well as providing investment funds for new companies or companies growing more rapidly than can be financed internally, subject to the test of the market rather than biased by the empire-building proclivities of managements. No consideration was given to the duration of time over which the distribution might be postponed, with the result that very sharp discriminations often occurred dependent on the exact time of distribution. Elaborate provision for taxable stock dividends or “con- sent dividends” was felt to be needed. A form of undistributed profits tax that would result in a more nearly neutral overall impact would consist of an annual tax on the accumu- lated undistributed surplus of the corporation at a rate that would represent an interest payment on the amount of surtax deferred. Thus if one were to assume a modal marginal surtax rate of 40 percent and a rate of interest of to percent this would imply that an annual tax rate of 4 percent applied to the accumulated undistributed surplus would be required to offset the advantage to the individual taxpayer. The neutrality would still be only approximate, however. Stockhold- ers subject to a marginal surtax rate higher than the assumed modal rate would derive an advantage from the deferment of profit distribution, while those with lower surtax rates or subject only to normal tax would be disadvantaged. More important, if realization takes place prior to the distribution of the earnings in dividends, as when the stock is sold and the accumulated earnings become reflected in the capital gain, further taxation of the previously accumulated undistributed profits would con- stitute an excessive burden. This could be roughly compensated for by permitting an annual discount or amortization of the accumulated un- distributed profits for this purpose. This would be an approximate The Corporate Income Tax 129 adjustment to reflect some modal rate of the turnover in shareholders. While in principle one might adjust this amortization rate to the rate of turnover in shares on the books of the company, the possibilities for manipulation that this would open up seem to make this undesirable. The method is an approximate one at best, but it is far superior to the proposals advanced in the thirties. A possible alternative is to compel the distribution of all corporate earnings to stockholders in some taxable form. But doing this through penalty taxes, such as the notorious section 102 tax on “improper accu- mulations of surplus” or the taxes on “personal holding companies,” is messy and often capricious. Requiring the reporting of income on a partnership basis may be administratively unfeasible where there are several layers of corporations, though one might be reconciled to dealing somewhat severely with such situations on the ground that such layering is seldom necessary to the accomplishment of a socially desirable result. In some cases a partnership treament may result in a substantial tax liability for a taxpayer without liquid funds with which to pay the tax, as indeed sometimes happens with actual inharmonious partnerships. Spe- cial arrangements for deferred payment of tax such as are available with the estate tax might be able to deal with this problem but would be messy. Adjustments in the event of a reassessment of a corporation tax as a result of audit would be a headache. It seems that in the absence of a system of cumulative assessment for individuals, a substantial degree of nonneutrality in the assessment of taxes on corporations accumulating undistributed surplus would be inevitable. Allocation of Domestic Income by Source Another problem is that as long as one is going to levy a tax in part ona source basis, some method of allocation of domestic income among jurisdictions is necessary. Here attempts to allocate income by account- ing procedures run into all kinds of difficulties. The matter is bad enough when it involves merely the question of the transfer prices to be set on goods imported or exported within the company or among related companies. When it comes to allocating income from intangibles such as patents, research and development, or design activities or from the distribution of books, movies, computer programs, or other products with a heavy component of fixed or head-end costs, allocation on the 130 William Vickrey basis of the usual accounting procedures is almost impossible to keep in line with economic reality, especially when the allocation is under pres- sure from tax-rate differentials across jurisdictions. While this problem arises mainly with respect to corporation tax, it can also occur with respect to the individual tax. The solution that has been generally adopted in the allocation of income among the states of the United States is in terms of formulas based on the relative amounts of various activity measures deemed to be correlated with the production of net income and which can readily be allocated among the various jurisdictions. However, the form of most of these formulas actually in use leaves much to be desired, appearing to have been concocted, like the camel, by a committee, in this case appar- ently made up primarily of lawyers and accountants with little or no input from economists or econometricians. One of the most popular, the so-called Massachussetts formula, uses a simple average of three frac- tions of the amounts of payrolls, capital assets, and sales within each given jurisdiction as compared to the total for the firm. Thus if B; is the tax base for the ith jurisdiction, and L;, $;, and K; the respective payrolls, sales, and assets located within the ith jurisdiction, with t indicating the totals for the firm, we would have ; = B, X [L,/L, + S;/S, + K,/K,]/3. The results produced by this formula can be quite capricious, how- ever, with possible undesirable distortionary incentives. For example, if a consulting firm operating from rented office space in Manhattan and having very little in the way of fixed capital assets were to buy property in Vermont, possibly on the pretext of providing recreational facilities for its employees, it would thereby shift one-third of its corporate tax base to Vermont, with no corresponding shift in the real sources of its income. A superior formula using the same data would be BS Be, Et Gx Si Gray IG xl, + Cc, xs, + Gx Ki. (1) The coefficients C), C,, and C,, could be uniform over all firms or could be varied across classes of firms. One way of determining these coefficients would be by running a multiple regression of the B,’s for the various firms in a given class on the corresponding L,, S,, and K,’s. In effect one would be predicting the tax base for the ith jurisdiction on the basis of The Corporate Income Tax 131 this regression and then adjusting the results proportionally to agree with the actual tax base for the firm as a whole. Handling Foreign Source Income The solution is to separate the tax that is to be levied on a source basis from that levied on a destination basis. Since progression is appropri- ately based on the total income received by a family or other unit the appropriate combination would be a flat rate “normal” tax on a source basis and a separate tax at progressive rates on the aggregate income received. Since income received from a foreign jurisdiction is most con- veniently reported on a net after-tax basis (indeed it would be difficult to determine the amount by which a net distribution should be “grossed- up” to include the source-based tax), it would be appropriate in the case of domestic income to allow the normal source-based tax to be deducted in computing the base for the progressive “surtax.” Conclusion: Getting Rid of the Tax Is Difficult Regardless of the form which the reform or abolition of the corporation tax takes, the transition will present difficulties. Even at best, there is a good deal of wisdom in the saw that “any ‘old’ tax is a good tax.” Any tax change inherently involves not only the upsetting of expectations that may have achieved a certain degree of legitimacy, but also involves additional complications in dealing with the transition from the old regime to the new. Any proposal for tax reform carries a considerable burden of overcoming the costs of the transition. This is particularly true with basic reform or abolition of the corporation income tax. Sudden transition from one regime to the other would involve sub- stantial windfall gains to shareholders, as well as possibly some losses. This could, in principle, be avoided by making the transition sufficiently gradual so that financial arrangements and investment levels could be adjusted, investment being increased at a rate that would keep the marginal productivity of investment in line with the pretax return re- quired. For this to come off smoothly, however, would require that investors have confidence that the gradual reduction in the tax would actually take 132 William Vickrey place. Congress would in effect have to make a commitment to this effect, would have to adhere to this commitment over a considerable period of time in the face of strong political pressures, and would have to convince investors that the commitment would in fact be met. Prospects for successfully carrying out such a program are at best dim. And the uncertainties created in the period during which the enactment of the change is being proposed and discussed would themselves have undesir- able effects. Even if the program could be carried out, the pre-announced changes in tax rates would constitute a powerful incentive for tax avoidance by deferring the reporting of income by one means or another: precisely the type of manipulation that cumulative averaging is designed to deal with. Applying cumulative average, however, would bring back the windfalls. On the whole the best procedure might be to enact the change in one fell swoop with as little advance discussion and warning as possible. It is an additional item on the bill of indictment against the tax that getting rid of it is so difficult. 8. WAYNE THIRSK Intellectual Foundations of the VAT in North America and Japan Carl S. Shoup should be recognized as the intellectual father of the value-added tax (vaT).! He has given us a great deal of the vocabu- lary that is currently used in describing the operation of the vaT and has endowed us with the cognitive framework utilized by most economists and policy makers who try to understand how the vat works. If nothing else, Shoup has made us acutely aware that there is no single vAT, only a large variety of vaT options that he himself has meticulously outlined and distinguished in his own writings on the subject. While it is never an easy task to assign, or even apportion, credit for the origination, let alone evolution, of a particular method of taxation, I will in what follows attempt to identify Carl Shoup’s numerous contri- butions to the growth and application of value-added taxes in the world. The paper divides into four sections. It starts with an appraisal of Shoup’s pioneering efforts to enlighten us on how a vaAT differs from other kinds of broad-based commodity taxes and the choices confront- ing policy makers in their task of choosing one form of vaT from the family of vaT possibilities. Certain equivalence propositions, that be- tween consumption and wage taxation as demonstrated by Shoup and that between the origin and destination principles as disputed by Shoup, are examined in the second section. The third section focuses on Shoup’s careful consideration of the relative merits of a VAT versus a retail sales tax. A fourth and final section looks at the issues posed by the exempt- ing, or zero rating, of certain products, firms, and sectors and the way in which these issues are affected by the method of computing the varT. The VAT—What Is It? As a tax on the incremental value that a firm adds to a product, over and above the value of its purchases from other firms, the vAT sounds like a 134 Wayne Thirsk disarmingly simple method of taxation. This apparent simplicity disap- pears, however, in the face of the wide variety of forms that the vaT may assume. As Shoup has emphasized in a recent paper (1989a), political and administrative pressures combine to produce a complex choice set that has the following dimensions: . (1) the nature of the vat base—consumption, net income, or gross income; (2) the treatment of international trade flows—taxing consumption, regard- less of source of supply, the destination principle, or alternatively, taxing production by exempting imports and making exports taxable, the origin principle; (3) the computation of tax by either the addition, subtraction, or credit- invoice methods: (4) the designation of products, firms and sectors, if any, to be free of tax; (5) the method of realizing tax-free status—exemption or zero rating; (6) the application of special regimes for hard-to-tax sectors such as agricul- ture; (7) the adoption of a single rate or a multiple rate structure; (8) the imposition of tax on prices which either exclude or include the tax itself. From the large number of categories in the above taxonomy, it is evident that a considerable number of vaT possibilities exist.4 However, not all of these combinations make good economic sense as Shoup (1989a) has been careful to point out. For example, the income type of VAT appears to fit well only with the origin basis of trade treatment and the use of the addition method of calculating tax liability.5 Despite the striking array of potential vats, Shoup (1989a) has shown that in practice there is considerable conformity among countries that rely on the vaT. Most countries using the vAT are observed to employ the consumption/destination vaT along with the credit-invoice method of computation, at least two rates and an amount of tax quoted separately from price (tax-exclusive basis). Countries differ mainly in the scope of the VAT as it is applied to different firms and sectors, their method of achieving tax-free status, and in the design of special tax regimes. Many features of the vaT that are conventionally familiar to us now must have been unfamiliar to most economists thirty-five years ago. Shoup was the first to draw clearly the basic distinction between a VAT of the consumption type (CvAT), which allows a full and immediate deduc- tion for investment, and the income type (IvAT) which allows a deduc- tion only for depreciation. Although this distinction appears formally in The Intellectual Foundations of the VAT 135 his 1955 article, it is important to realize that, reading between the lines of history, the recognition of the essential difference between a CvAT and an IVAT must have occurred prior to this date. In the Report on Japanese Taxation by the Shoup mission in 1949 one finds the first practical proposal for the implementation of a vAT. Viewed as a replacement for an enterprise tax and a national transactions tax, the Japanese VAT was intended to be applied at the prefectural (state) level using the subtraction method of tax computation.® Motivation for this proposal came from several sources. Having written extensively on the turnover sales taxes in France (1930) and having witnessed the emergence of state retail sales taxes in the United States during the early 1930s, Shoup was cognizant of the vat’s ability to solve many of the problems that plague those other forms of sales tax. The VAT promised to avoid the cumulative element of taxation and the bias in favor of large enterprise which are inherent in turnover taxes and unavoidable in practice with a manufacturers’ sales tax. On administrative grounds the vAT seemed superior, in the context of the early postwar Japanese economy, to its nearest competitor, the retail sales tax. With numerous small retailers, most of whom kept poor records, if any at all, the opportunities for tax evasion were much greater with a retail sales tax. Moreover, the joint operation of households and small businesses on the same premises meant that it would have been extremely difficult to apply a suspension technique for excluding resales from tax without encouraging tax free division of goods to household consumption. Moreover, with a sales tax many producers’ goods would be subjected to either double taxation (on purchase and subsequent resale if not exempted) or no tax at all (if exempted). Because any VAT paid on purchase would have been deductible in calculating taxable value-added resulting from resale, the vAT avoids the need to distinguish business use from final or household use. And to the extent retailers either did not collect the tax or turn in their collections rather than merely underre- porting their sales, at least a sizable fraction of the tax would have already been collected at the preretail level. To moderate any regressive impact a partial exemption was to be provided for food purchases, while for ease of administration agriculture and forestry were to be exempted. Although Shoup had argued for a uniform rate, standard rates of 4 and 3 percent were to apply to business and professions, respectively. 136 Wayne Thirsk Under the subtraction approach interest, rent, and dividends received by ordinary business were considered as value-added of the payer, in- stead of the payee, in an attempt to curb evasion and apply the tax at source.” Thus capital income received was allowed as a deduction from the vaT base. In the case of financial enterprises, however, interest and dividends received were to be included in the vat base, while interest paid out was to be ignored.8 Despite the care with which the vaT proposal was crafted, it was not adopted in Japan, although it did receive legislative approval at one point. A number of factors—its novelty at the time, the perception of a significant shift in tax burdens from business to consumers, and the traditional Japanese concern with maintaining the centralization of fiscal powers—conspired to doom this pioneering proposal for a VAT. There was also considerable confusion over where the burden of the vaT would ultimately rest. Capital income was apparently caught in the tax net, yet businesses were to be allowed to deduct their purchases of physical capital, including inventory. Established businesses complained that this feature of the tax favored new over older businesses. In the end businesses were given the option of deducting depreciation on their existing assets as a transition device. The second Shoup report, however, was unambiguous in claiming that the proposed vat was, by allowing an immediate deduction for investment, a broad-based consumption tax that would stimulate savings and investment in accordance with the needs of postwar reconstruction.? A prominent characteristic of Shoup’s work is the intertwining of theory and practice, such that experience in grappling with real world tax questions has frequently led to refinements in the economics of taxation that have enlightened generations of public finance economists. Shoup’s explanation of the difference between an IvaT and a cvaT in his 1955 paper is a good example of the useful complementarity that exists between tax theory and empirical work on tax policy. In describing the cvaT Shoup (1955:9) suggested that “the first type . . . be called the consumption type of value-added concept because the value added that is represented by the capital equipment does not appear in the accounts until later years, when the equipment is being consumed in the process of production” (of consumer goods). In contrast, the income type of VAT (IVAT) “is based on the accounting concepts familiar in computing net income. No subtraction is made for the cost of equipment in the year of purchase but depreciation is deducted in years of use.” The Intellectual Foundations of the VAT 137 An instructive numerical example is then presented by Shoup (1955: 10—11) to demonstrate that “the consumption type of value added concept gives the same total as retail sales, year by year,” while the “income type of value-added concept gives totals identical with those for income, year by year.” Thus, for the same rate of tax, “the value-added tax (consumption type) produces the same yield as the retail sales tax, while the value-added tax (income type) produces the same amount as the income type.” Another numerical example is deployed in the same paper as a vehicle for explaining the interest-inclusion variant of the cvaT. A simple two- period paradigm is invoked to show that “It comes to the same thing whether deduction of the purchase price in the first year is allowed, or whether it is disallowed and, in exchange, the taxpayer is allowed to deduct in later years both depreciation and the return earned on the investment” (1955:12). This statement is a remarkable forerunner of one of the major expenditure tax principles, that expensing of invest- ment under either the value-added or corporation income tax effectively exempts from tax the normal return to capital.!° It also foreshadows one of the major equivalence propositions in public finance that is attributa- ble to Shoup!! and which is discussed next. Equivalences: Good, Bad, and Approximate As a natural outgrowth of his earlier work on the interest exclusion variant of the vAT, Shoup explored the relationship between a tax on consumption and an equal-rate tax levied on wage income. The fruits of this exploration appeared in an article (1969a) and a book (1969b). Shoup approached the issue initially by examining the following set of national income identities which define net national income, Y,: am) Y= Cia 0 where C is total consumption, I is gross investment spending, and D is depreciation for the (closed) economy as a whole; (2) Y,=W+P where W represents labor income and P denotes property income. By combining these two identities it is easy to show that: (a) (Ca NG Ae Bw = 1h 138 Wayne Thirsk From this result Shoup recognized that in any given year consumption and wages would be equal if, and only if, | = P + D. While on a dis- counted basis this equation linking I, P, and D might hold for a single in- vestment, Shoup concluded that for a particular year observed amounts of property income and depreciation derive from a mixture of invest- ments, old and new, and that in the absence of discounting there is no reason why the aggregate value of investment should match the sum of property income and depreciation. Shoup progressed from this conclusion to establish the present value equivalence of taxes on wages and taxes on consumption. Using the subscripts p, f, and d to denote, respectively, present value, future value, and discounted value, Shoup invited us to consider a two-factor econ- omy in which a representative firm employs only labor to produce some consumption goods and a capital good. With these simplifications the firm’s income and sales statement could be written as: (4) Ww, = C, +P ie = G. ar Pra ar Dea Accordingly, the present value of a CvAT base is (OWS Sa es rena Dis where the first two terms measure the value of current consumption and the other terms capture the present value of future consumption. Clearly this tax base will equal current wage payments W,, if I, = Peg + Dea. This latter condition is nothing more than the normal equilibrium condi- tion for capital markets so that as long as it holds, as long as investment equals the present value of future consumption flows, a tax on wages and an equal-rate tax on consumption will yield the same present value of tax collections. What are we to make of this result twenty years later? In one sense, apart from the fact that a tax on wages is inherently origin-based in an open economy, the result withstands the glare of subsequent scrutiny de- spite some minor qualifications. This result in many ways foreshadows economists’ later preoccupation with the taxpayer’s lifetime budget con- straint. Over the taxpayer’s entire lifetime the present value of income, including inheritances, is matched by the present value of outlays, in- cluding gifts and bequests. If savings are made from after-tax wage income and are used to fund retirement period consumption, capital income basically vanishes from the lifetime budget constraint. Thus it is common to assert now that a tax on wages and inheritances would have The Intellectual Foundations of the VAT the same present value of revenue as an equal-rate tax imposed on household expenditures and bequests. !2 However, recent research utilizing dynamic general equilibrium mod- els to investigate the consequences of alternative tax regimes strongly suggests that wage and consumption taxation are not close equivalents. In these models savings behavior is geared to the age structure of the population, and differences in the timing of taxes over the taxpayer’s lifetime can have significant effects on savings levels. Intuitively, a tax on consumption would raise tax burdens more toward the end of a tax- payer’s lifetime, curbing the dis-saving of older households and inducing a higher level of saving than would a wage tax yielding the same present value of revenues. In his simulations for the United States, Summers (1981) found that a shift from a comprehensive income tax to a wage tax would eventually increase the economy’s capital-labor ratio by 58 percent compared to an increase of 74 percent that would be expected to occur if a shift to a consumption tax were made instead.!3 Recent work on the same subject by the Economic Council of Canada (1987) confirms this result when the issue is examined in the context of a small open economy. Under a wage tax vis-a-vis an income tax, the aggregate capital-output ratio rises by 22 percent. A much higher gain in capital deepening of 47 percent occurs when an equal-yield consumption tax replaces the income tax. Another equivalence relationship that has attracted Shoup’s attention is the claim that in a flexible exchange world there is no substantive difference between the destination and origin principles and that coun- tries have nothing to either gain or to lose in switching from one trade regime to another. As a specific application of the purchasing power parity (ppp) the- orem, Shoup first noted in 1953 that if a country adopted a comprehen- sive and uniform rate of sales tax, its choice of trade regime would produce no real consequences apart from the distribution of tax revenue among the taxing countries. A country that replaced the origin principle with the destination principle would experience a trade neutral apprecia- tion of its currency, while a country making the opposite switch would be insulated from any trade repercussions by a depreciation of its cur- rency. Shoup was quick to point out that this exchange rate protection would be valid only to the extent that the conditions required for the ppp theorem to hold were satisfied, namely, no international flows of capital 139 140 Wayne Thirsk and no transfers between countries. In other words, if a country’s ex- change rate were continuously influenced by capital account elements, the required long-run adjustment of exchange rates would be prevented from happening. Thirty-five years later Shoup (1988c) repeated his skepticism over the utility of this equivalence theorem as a practical guide for policy makers. On reentering this debate, however, he concedes that a “tax structure change will surely not be without some influence on the exchange rate,” and that “a move from the origin principle to the destination principle will probably increase exports and decrease imports somewhat, but not as much as the tax rate alone might indicate” (Shoup 1988c:368). Drawing on some of his work with Cnossen, Shoup also makes the point that the border controls, often thought necessary for applying the destination principle, could be eliminated by making the required border tax adjustments from books of account (Cnossen and Shoup, 1986). If this were to happen, as is likely in the European Economic Community, the attraction of the destination principle would reign unchallenged, especially if it were perceived as being able to confer a trade advantage to a country. Choosing between the VAT and a Retail Sales Tax Shoup has had an abiding interest in the merits as well as the defects of real world sales taxes and has made explicit comparisons of the vat and a retail sales tax (RST) on at least two occasions (1969a and 1973b). First, in evaluating European experience with the vAT (1969a) Shoup concluded that the vAT was more adept than the RsT in removing taxes from business inputs and incorporating final services into the tax base. Therefore, the vaT is better than the RsT in avoiding undesirable tax cascading and in allowing accurate border tax adjustments to be made to both exports and imports. As is well known, thanks to Shoup and others, the RsT faces an intractable designation dilemma. Producers’ goods under an RST must be classified either as a final good or service, and thus taxable, or as a business related input, and therefore exempt from tax. The dilemma is particularly acute when a commodity can be used for either purpose. Will shovels be used in someone’s backyard to plant flowers or on a construction site? The vaT does not have to agonize over this question because tax is always charged on every sale and if the The Intellectual Foundations of the VAT purchaser is a business, it can avoid the tax on its inputs by either claiming a credit for it (under the credit-invoice method) or by claiming a deduction for the tax inclusive amount (under the subtraction method). In a balanced assessment of the vaT, Shoup (1973b) has pointed to the following advantages that the vaT has over the RST: (1) With a vat failure to remit tax at the retail level will not produce a total loss of revenue since a large portion of the tax will have already been collected at the preretail levels of business operation. Yet, Shoup recog- nizes that if evasion takes the form of underreporting sales, both the vaT and the rst will suffer from the same amount of revenue leakage. As mentioned above, the vAT is superior to the RST in removing tax from producer goods. In embracing the vat both Sweden and Denmark were motivated by this factor.!'4 Shoup conceded, however, that producers could divert some producer goods to their personal use by filing a falsified invoice but claims that this activity is apt to be less of a problem than the inclusion of some producer goods in the base of the RsT. If for administrative reasons small retailers are exempted from sales tax, the shrinkage in the tax base will be smaller for a vat than for the RsT, and for a given amount of revenue the vAT tax rate will be lower for that reason alone. It is easier with a VAT to reach services consumed by households and prevent tax from being applied to business purchases of services. Once again the problem arises from the fact that many firms provide services to both households and other businesses and the RsT is incapable of distin- guishing among these transactions. A service must be either taxable, thereby taxing some business inputs, or it must be exempted from tax, thereby shrinking the size of the tax base. On the other side of the coin, Shoup also takes note of the vaT’s relative disadvantages: (1) (2) A much larger number of firms, at all levels of operation besides retailing, must be registered for the vAT, a requirement which raises both the administrative and compliance costs of sales taxation. For any particular firm there is more paperwork and higher compliance costs for the firm with a credit-invoice vaT than there would be with an rST.!5 This burden under the vaT stems from the need to account for each purchase to obtain an offsetting credit and the need to arrange for refunds if total credits exceed tax liabilities as could easily occur in export activities. On a number of other dimensions, for example, the difficulty of including housing and financial services in the tax base and the ability to 141 142 Wayne Thirsk implement a multiple rate structure for distinguishing between luxuries and necessities, Shoup rates the vaT and the rsT as being about even. Overall, however, at least for the United States and probably for many other countries as well, Shoup ranks the vat ahead of the RsT. Shoup’s work on the vaT is an example of what may be called prag- matism in its pristine form. Shoup has been loath to give a blanket endorsement of the vat as is evident from the conclusions of one of his recent papers (1988b) in which he argues “no generalization seems justified on the suitability of the value-added tax for developing coun- tries as a group.” In the end the choice of sales tax must conform to the structural peculiarities of each country. If a country is possessed of weak administration skills, a small trade and a large agricultural sector (with numerous small farmers), a highly fragmented retail sector, little scope for vertical integration, and poor record keeping, it might be advised to adopt either a turnover tax or a single-stage manufacturing level sales tax. It is not hard, however, to envision circumstances in which either of these taxes would be dominated by a preretail vat, one that applies to all firms except retailers. What Shoup urges us to do, and what he has done for himself, is to judge events by their effects and not be charmed into believing that we have discovered a magic tax potion for solving all of our fiscal ailments. Unfinished Business—Exemptions and the Method of Collecting Tax It seems singularly appropriate in a paper honoring Shoup’s contribution to our understanding of the vAT to raise some unresolved issues that bring us to one of the frontiers of current knowledge. There is a touch of irony in the fact that Shoup recommended the use of the subtraction method of vaT collection in 1949 in Japan only to see the rest of the world subsequently adopt the credit-invoice technique for collecting tax. Now, forty years later, Japan promises to provide us an experimental observation in the use of the subtraction method, and Canada may soon follow suit. Will the subtraction approach be successful? Is it a reasonable alter- native to the European style credit-invoice approach to implementing the vaT? Because it relies on books of account and moves the tax collection apparatus from a firm’s sales force to its accounting depart- ment, the subtraction method is attractive, particularly to smaller firms, The Intellectual Foundations of the VAT because of the savings in compliance costs that it promises. It is moot, however, whether these benefits outweigh all of the disadvantages that are alleged to be inherent in the use of the subtraction method. The case against the subtraction method has been persuasively pre- sented by Charles McLure (1987). McLure concludes that unless the vAT is applied on a comprehensive basis at a single rate it will be impossible for the subtraction method to render accurate border tax adjustments. Numerous exemptions and multiple rates, it is argued, would spell disaster for a subtraction style vAT.1© Moreover, because preretail ex- emptions, unlike what happens under the credit-invoice method, create permanent holes in the vat tax base, there is concern that a subtraction style VAT will invite the assault of preference seeking lobbyists.!7 As well, firms may be encouraged to employ transfer pricing practices which channel an artificially large amount of value-added through tax-exempt stages of production. McLure (1987) also demonstrates that many of these defects of a subtraction-based approach can be overcome using a sophisticated sub- traction technique that permits a deduction only for purchases on which tax has already been paid. With a less than fully comprehensive base and no break in the chain of deductions McLure has indicated how the sophisticated subtraction method is capable of producing accurate bor- der tax adjustments. With multiple rates, however, not even this compli- cated procedure is up to the task of avoiding tax-induced trade distor- tions. Nonetheless, even where the sophisticated subtraction method is expected to work reasonably well, it requires that firms keep track of their tax-paid and tax-free purchase invoices and thus reimposes the compliance costs on firms that the subtraction approach was intended to circumvent. While an early advocate of the subtraction method, Shoup’s later writings reflect a steadily growing admiration of what the credit invoice approach has to offer. In several places (1969b, 1973b), for example, Shoup emphasizes the efficacy with which the credit-invoice technique frees exports from vaT and solves the problem of underevaluation of imports. !8 Still, Shoup is cognizant of one of the major blemishes on the face of the credit-type VAT, the overtaxation that results whenever ex- emptions give rise to a disruption in the chain of tax credits. In the volume Public Finance (1969b) and subsequently in Shoup (1973a) he draws on several numerical illustrations to indicate how exemptions create double taxation of the input purchases made by exempt entities. For administrative and compliance reasons vAT levying countries 143 144 Wayne Thirsk typically exempt small firms in all sectors, some hard-to-tax sectors such as agriculture, housing, and financial services, other services such as health and education, and often some professional services as well. The presence of these, and possibly other, vat exemptions serves to introduce unintended origin elements into an otherwise destination-based tax. Exports, for instance, will be inadvertently and indirectly taxed through their purchases of exempt banking, insurance, and other services. On the import side of the trade ledger, Shoup (1969b:263) has pointed out that exemption of some pre-import domestic stage for import competing goods also imposes higher rates on them than on imports. Given the inevitability of exemptions under a vAT, an important policy question is whether the subtraction type of vAT is better able to accommodate their effects than the credit type of vaT.19 McLure (1987) offers one perspective on this issue. Table 1 reproduces the simple numer- ical model on which McLure bases his critique of the naive subtraction approach. As is shown in the top part of that table, if there were no exemptions a ro percent vAT would collect $30, $40, and $30, respec- tively, from producers at stages A, B, and C. In the middle part of that table producers at stage B are assumed to be exempt and as a result their value-added disappears from the tax base. Revenue falls by the product of the tax rate and the decline in the tax base. Finally, the lower part of the table indicates that if exporters at stage C can claim a rebate for input taxes equal to the amount of their purchases times the tax rate, they will be overcompensated and receive instead an effective export subsidy. As is well known, a credit type of vAT would produce the opposite set of distortions, overtaxation of the product and its export by $30, the amount of vAT paid by the vendors to the exempt stage at zero. These results, however, seem to rest on the manner in which the model has been closed. McLure and Shoup before him have chosen a linear closure rule in their numerical models in which initial stages of production never interact as purchasers with later stages. An alternative way of closing the model is to allow for the possibility of a purchasing loop that permits sales by retailers to primary or initial stages of production. With the addition of this loop producers at stage A would buy some imports from stage C and would exhibit nonzero purchases on which some vaT would have been paid, rather than the zero amount shown in table r. Making that modification in assumption changes the way in which an exempt-laden vaT affects prices and costs in the economy. Under a credit-type of vAT it magnifies the degree of overtaxation due to the existence of exempt entities. With a subtraction-type VAT there are no The Intellectual Foundations of the VAT 145 Table 1: The Impact of VAT Exemptions under the Subtraction Method Basic Information A B @ Total Sales 300 700 1000 2000 Purchases 0 300 700 1000 Value-Added 300 400 300 1000 Stage B Exempt from Tax (10% rate) Taxable Sales 300 = 1000 1300 Purchases 0 == 700 700 Taxable Value-Added 300 0 300 600 VAT 30 0 30 60 Rebate of Tax on Exports with Stage B Exempt Taxable Sales 300 = 0 300 Purchases 0 — 700 700 Taxable Value-Added 300 0 —700 —400 VAT 30 0 = TAY) —40 Source: McLure (1987) Note: All values in the table are expressed on a tax-exclusive basis. further tax implications since whatever gets added to cost at stage A is subsequently subtracted at later stages and the existence of an exempt stage at some point would not affect the value of the deductions that could be claimed. However, by itself the vAT included in early stage costs would exert a direct effect on final prices and make tradable goods and services more expensive than otherwise. Seen in this light, payment of taxes on inputs purchased by exempt entities may serve to offset the undertaxation that occurs with exemp- tions under a subtraction-style vAT.2° Conversely, the same phenomenon may work to exacerbate the exemption-induced price distortions arising under a credit-type of vaT. In any event it seems to be an issue on which we have little empirical knowledge. It also warns us of the dangers of comparing real world vats with theoretically ideal vats, a warning which is frequently sounded in Shoup’s work. Conclusions Forty years ago the vAT existed only in the minds of a few people like Carl Shoup. Since then, thanks largely to the influence of Shoup in his 146 Wayne Thirsk dual role of adviser and academician, the vat has flourished to become an important revenue component in the tax systems of nearly fifty countries. Shoup’s particular contributions to the growth and develop- ment of the vAT are wide ranging. He perhaps has done more than anyone else to provide us with a taxonomy of the vat and make us aware of the enormous number of options that exist in applying the var. Beyond that he has been instrumental in establishing some of the essen- tial equivalences between consumption taxes and wage taxes. Finally, he has thoroughly investigated the advantages that a VAT may claim over other forms of commodity taxation, and in the context of developing countries he has outlined how tax administration capabilities will deter- mine the appropriate form of vAT to adopt. Overall, it is an impressive intellectual achievement. Many econo- mists hope that their work will improve conditions in the real world. Carl Shoup has the satisfaction of knowing that his work has. Notes Irwin Gillespie, Charles E. McLure, and Carl Shoup provided useful comments on this paper. Only the author, however, is responsible for any errors or omis- sions that remain. 1. Shoup led the commission to Japan that recommended a var for that country. His 1955 article was a major landmark in the literature on value-added taxation, as was Clara Sullivan’s 1965 dissertation, which he directed. Shoup played a key role in the acceptance by the Neumark commission of the vaT as the standard form of sales tax in the European Community (see Pohmer 1983). In the dedication of his most recent work on the vAT, McLure (1987) explicitly acknowledges his intellectual debt to Shoup. 2. Starting in the early 1950s when a crude form of vaT existed in France, the vat has been adopted by some forty-eight countries in both the developed and the developing world and is soon to include, it seems, Canada. Many other countries, particularly in Africa, employ the vaT principle at a single stage, usually manufacturing, in order to avoid the cascading difficulties associated with other forms of commodity taxation. Tait (1988) provides a recent survey. 3. A fourth possibility, the personal exemption vaT discussed in the U.S. Treasury’s tax proposals for reform in 1984, is not considered here because it represents a radical departure from impersonal commodity taxation and raises the question of whether to tax consumption directly or indirectly. 4. If the choices in the fourth and sixth dimensions are restricted to two in The Intellectual Foundations of the VAT each case, Shoup (1989a) indicates that there may be as many as 768 VAT options. 5. Shoup (1989 a) has dealt thoroughly with which combination of features may be compatible with each other. 6. To illustrate the dictum that “what goes around, comes around,” Japan has recently adopted a subtraction method var at the national level. 7. To check vAT evasion arising from the padding of expense accounts, all expenses of this nature were to be treated as nondeductible wages. 8. With the benefit of hindsight it is understood now that this is not the right method of taxing financial institutions under a vaT. However, no country has as yet devised a satisfactory method of doing so. Those that have tried have resorted to a variety of special regimes such as the temporary use of the addition method in Israel. g. In the first report no deduction for either depreciation or investment was suggested. In addition to permitting expensing of new investment, the second report allowed firms to alternatively select depreciation on all assets, old and new. 10. How would capital goods built by the firm for its own use be treated? Shoup explains that, under the deduction variant of the CvAT, activity of this type would be ignored for tax purposes. By so doing the tax authorities would recognize an implicit deduction allowance for the investment equal in value to the implicit sale made by the firm to itself. 11. Shoup (1969b:301—2) looks at the matter from the angle of the corpora- tion tax by noting that “Completely accelerated depreciation is thus equivalent to tax exemption, and transforms the corporate income tax into a value-added tax, consumption type, so far as corporate income is concerned.” 12. Another minor qualification to the Shoupean equivalence is the need to differentiate between the normal return to investment and the earning of su- pranormal returns, or rents. For instance, if P in equation (3) contains a mixture of normal returns N and rent R, then it can be rewritten as C = W + (R+N+D — I). Since I is the present value of N + D ina capital market equilibrium, it is more accurate to claim that the present value of a tax on consumption would be equal to the present value of an equal-rate tax imposed on wages and rents rather than a tax levied only on wages. 13. Although the terminology changes over time the issues facing tax analysts tend to remain the same. In the language of today’s expenditure tax literature, the wage tax (investment nondeductible, capital income not included for tax purposes) is referred to as the tax-prepaid method of taxing expenditure while the consumption tax (savings deductible, capital income included in the tax base) flies under the banner of the cash-flow tax. 14. In the case of Denmark the vat replaced a wholesale level tax rather than a retail sales tax. 147 148 Wayne Thirsk 15. The promise of a substantial cut in these compliance costs if the subtrac- tion method were instead applied to books of account is one of the issues examined in the next section. 16. Both exemptions and multiple rates would make it impossible for the tax authorities to calculate how much var is embedded in the input costs of exports and import substitutes and therefore how much tax to rebate on exports and to impose on imports. 17. Political optimists might contend that adoption of the subtraction method would represent a preemptive strike for tax purity making it unthinkable for governments to resort to multiple rates. 18. Undervaluation of imports produces a diminished size of the input tax credit for purchasers so that undertaxation at one stage is exactly offset by overtaxation later on. By the same token Shoup (1969b:262) observes that if an importer has understated value upon importation, he will have larger value- added to pay tax on under the subtraction method. Shoup (198g9a) has also pointed out that both the subtraction and credit-invoice methods generate about the same amount of “paper trail” for tax administrators and both engender a similar incentive for firms to overstate their purchases. 19. Although he did not address this matter at length, Shoup (1969b:260) nonetheless concludes that the subtraction exemption is “less serious”—in terms of revenue loss—“‘than the injustices that can occur under the tax credit system.” 20. The undertaxation associated with a subtraction style vAT may also counteract any overtaxation of tradables arising from applying national excise taxes or subnational sales taxes. 9. JOHN F. GRAHAM The Place of the Property Tax in the Fiscal System This paper draws upon the author’s extensive involvement with two Royal Commissions in Nova Scotia and New Brunswick (the first as chairman, the second as consultant) in grappling with the issues of municipal public finance and related matters. Both of these commissions considered comprehensively the questions of allocation of responsibility for public services between provincial and municipal governments, the tax bases to be assigned to the respective levels of government and the provision for fiscal transfers to the municipalities to ensure that vertical and horizontal fiscal balance were established within the province. Both commissions also addressed all three of the elements of architecture, engineering, and administration in accordance with Carl Shoup’s dis- tinctions in his article in this volume, both on the tax side of the budget and, simultaneously, on the expenditure—public services—side. The paper also draws upon the author’s lesser involvement with similar commissions in Newfoundland, Ontario, and British Columbia. In all cases the question of the place of the property tax in the fiscal system was of central importance. Where provinces are referred to in the exposition, in most cases the arguments and propositions can be applied to states in other federal countries or to central governments in unitary states, although possibly with some modification. The argument in the paper is much informed by Carl Shoup’s work in this area (esp. Shoup 1969b, chaps. 14 and 15), and it is the intention to address the subject in the spirit of the political economy approach that epitomizes his writing, an approach the author wholeheartedly shares. Given the focus in this volume on retrospectives on public finance, what has been attempted here, rather than provide a summary of de- velopments with respect to property taxation,! is to bring together some 150 John F. Graham of the elements of public goods theory and of the principles of property taxation that have been developed over the years into what is seen as the culmination and synthesis of the two sides of local government finance. The aim is to suggest an analytical framework for establishing an effi- cient and equitable system of public finance that takes cognizance of the attributes of property taxation and of the nature of municipal services and that can serve as an operational guide to policy makers. “Property tax” is interpreted here to mean the real estate tax on land and improve- ments. Goods and Services: The Nature of Municipal Services The planning and provision of municipal services determine much of the character of the environment in which citizens live. The intent should be to foster a community within which citizens can collectively determine the form and character of their municipality that will enable them most effectively to pursue their common and individual interests. The munici- pality should not be seen simply as a machine for governing, but as a polity that touches almost the whole of life. In pursuing the goal of economic efficiency in the narrow sense, it is important to avoid engag- ing in a technical exercise that ignores efficiency in the broader sense. This paper is guided by such a view. On the issue of the principle on which to assign the costs of providing public services, that is, pricing, user-pay, some other benefit basis, or ability to pay, it is necessary to have full regard for the nature of the services in question. In considering the nature of municipal public ser- vices, it is common practice to use a distinction between people-related services, such as education, health services, social assistance, and the administration of justice, and property-related services, such as water, sewerage, streets and protection, or even to speak of the distinction between services to people and services to property (e.g., Economic Council of Canada 1987:99, 103; Thirsk 1982:389). This is an unfortu- nate and, indeed, an erroneous distinction. Only people can experience benefits from public services. Some services may well affect the value of property, and this connection may be a legitimate basis for assigning their cost to property owners, but even local improvements such as lateral streets and water and sewerage facilities are rendering services to the people in the properties and not to the properties as such. Even in the case of nonresidential (industrial and commercial) property, where many Property Tax in the Fiscal System of the municipal services are essentially intermediate goods, the ultimate beneficiaries are the people working on those properties and the con- sumers of the goods produced. A much more satisfactory distinction is that between general and local services, a distinction found in some of the literature (Graham 1963:56—-58; New Brunswick Royal Commission 1963:3—4; Nova Scotia Royal Commission 1974, 1:1). General services are services the benefits from which extend beyond the local communities in which they are provided and affect the whole province. They should be provided at the same level to citizens throughout the province and not at levels that differ because of differences in the property tax bases of the munici- palities or because of differences in the importance attached to them by particular municipal councils. In this category those general services that municipalities help to provide in most provinces are elementary and secondary education, health services, social services and public housing, and the administration of justice. Logically, these services should be provided by the provincial government, for then the externalities are internalized (an efficiency consideration), the levels of services will be uniform throughout the province, and provincial taxes to finance them will also be at uniform rates throughout the province; that is, the princi- ple of fiscal equity (equal treatment of equals or horizontal equity) will be met (a combined equity and efficiency consideration). This was the solution recommended by the Royal Commission for Nova Scotia, and the one actually adopted in New Brunswick, and later by Prince Edward Island, on identical reasoning. An alternative is for the province to require the municipalities to provide prescribed levels of general services and to give conditional deficiency grants (for vertical fiscal balance) and also equalization grants (for horizontal fiscal balance) that will enable the municipalities to provide their share of the cost at uniform tax rates on their property tax base, under uniform assessment, that is, where the total assessments of all municipalities are measured at the same percentage of market value. Variations on this practice are found in most of the other Canadian provinces. The objection to this solution is that the municipalities are then in the position of providing provincially determined, that is, man- datory, services and of levying taxes that are de facto provincial taxes on their property tax base. Municipalities are not then free to use their only substantial revenue base, property taxation, in accordance with their own priorities. It should be noted that there need not be a loss of local autonomy, as is 151 152. John F. Graham sometimes alleged, when the province takes over the administrative and financial responsibility for these general services. Indeed, it is entirely possible for these services to be administered by the province in a highly decentralized way, to take local circumstances into account, even though levels of funding would be uniform. In the case of education a great deal of autonomy can rest even with the individual school; so local auton- omy, which need not be municipal autonomy, can even be enhanced under provincial administration. Local services are, by definition, those services whose benefits accrue entirely to the residents of the municipal region or local area in which they are provided. Local services include general local government (legis- lation and administration); fire and police protection and other protec- tive services; some transportation services and transportation-related services, such as traffic control, public transit, street, sidewalks, and snow removal; other public works; water supply and sanitary and storm sewerage; garbage and waste collection and disposal; other pollution control; recreational and cultural services (including parks); beautifica- tion; business, port, and tourist promotion; physical planning, including zoning and subdivision regulations; and financial and social planning. These services are clearly of enormous importance to the community’s well being. Moreover, practically all of these services are regional in character in that they affect not only citizens residing in existing munici- pal units that provide them, but also citizens in certain neighboring municipal units, all of which together constitute a single regional com- munity. The externalities in question are of considerable importance, given the prevalence of fragmented municipal structures, although some provinces have moved by various means and in varying degrees toward regional municipal government. Such externalities exist for most local services, and particularly for police and fire protection, transportation facilities, water supply, sanitary sewerage, other pollution control, recre- ational and cultural services, industrial development, and, above all, for the physical planning of residential, commercial, and recreational land use. These local services are the raison d’étre of municipal government and should be provided by that level of government in accordance with local choices and priorities. But, in recognition of their essential regional nature, it is desirable that they be provided by single municipal govern- ments, each embracing a whole regional community, so that the same municipal government would have the responsibility for both planning Property Tax in the Fiscal System 153 and providing those services. With such municipal consolidation, the externalities associated with these services would be internalized. Larger administrative units will often also make possible substantial gains in administrative capability. At the same time the establishment of a rela- tively small number of large regional municipalities needs to be balanced by considerable provision in the political structure for very local, even neighborhood, interests to be taken into account (see Nova Scotia Royal Commission 1974, vol. 2, chap. 6, sec. 5). The need for a regional municipal structure and the suitable form of it will of course depend upon demography and physical geography. What is appropriate for Nova Scotia or Southern Ontario may not be appropriate for Northern Ontario or Saskatchewan. The important issue of the extent to which local services should be financed by user charges, property tax, or some other tax will be ad- dressed below. Revenue Sources: Property Tax Much of the recent writing on the property tax has been preoccupied with the so-called New View of the incidence of that tax and, paradox- ically, with an emphasis on the financing of local services on a user-pay or pure-benefits basis. It is the contention here that much of this discus- sion, interesting though it is, is at worst misleading and at best is not very helpful in guiding the policy maker to establishing a rational structure for financing public services at the municipal level. It is important in considering any part of our fiscal system to be concerned with the incidence of taxes and of public services, for only when this is known is it possible for legislators to assign benefits and burdens in the desired way. According to the Old View of incidence, the burden of the tax on land is borne by the owner and of the tax on improvements by the occupier, whether tenant or owner, in the case of residential property. The degree of short-run shifting to the tenant of the tax on improvements will in fact depend on the state of the real estate market in a given municipality. In the case of nonresidential (commercial and industrial) property the tax on land is borne by the owner and the tax on improvements is largely borne by the consumers of the goods produced. The taxes are generally considered to be quite strongly regres- sive, particularly at the lowest income levels, since a larger share of the 154 John F. Graham income of the poor than of the rich is spent on housing or on consumer goods. The regressivity is reduced if lifetime income, rather than annual income, is used as the base. According to the New View of incidence, the tax, wherever levied, reduces the overall rate of return on all capital throughout the country (and beyond in an open economy), through a transfer of capital from taxed to untaxed assets, and so ultimately rests uniformly on all property of all types. Since property is largely owned by citizens in high-income groups, the tax is judged to be progressive. Account needs to be taken of excise effects where the rate of the tax differs in given jurisdictions from the average rate; that is, there will be higher prices to consumers in the higher taxed areas. But, as the Eco- nomic Council study points out: “The effect . . . will either cancel out across jurisdictions, if local residents are not mobile, or be capitalized into land rents if they are mobile. Hence, whether the property tax varies between jurisdictions or not, its real incidence is still on owners of capital” (Economic Council of Canada 1987:101). There is much more that can be said about these two views.” While it would be useful to include a full statement and evaluation of the two positions, to attempt to do so would take up more space than is avail- able. Given the stated particular purposes of this paper, this brief de- scription should suffice. The New View of incidence is one of many instances in our discipline of general equilibrium theory derived from highly unrealistic assump- tions—fixed supplies of labor and capital, perfectly competitive factor and product markets, the payment for factors being equal to the value of their marginal products, and perfect mobility of capital—being taken as describing the real world and as providing a guide to policy. There is also a tendency to ignore that, even within its assumptions, it is a very long- run view. It is not the intention to disparage this kind of theoretical exploration. If the assumptions can be relaxed sufficiently to make the analysis accord with the real world, we should become clearer about the important question of incidence, but in its present form there is a danger of becom- ing preoccupied with the theory as an intellectual toy and of losing sight of whether it is a valid guide to policy. In any case the New View may be less relevant in the context of a particular local jurisdiction, or in the context of the municipalities in a province, which is the focus here, than when considering the national or global context. Moreover, try to tell Property Tax in the Fiscal System the hard-pressed pensioners striving to meet their property tax bills that they are not bearing the tax, but that it is borne by and is progressively distributed among all capital owners throughout the land in proportion to the value of their capital! For the above reasons the discussion that follows rests largely on the Old View, with recognition that if the validity of the New View were established under actual, real world, conditions, the policy conclusions might have to be substantially altered. Full accept- ance of the New View, as it stands, would leave little basis for a rational relationship of the property tax and benefits from local public services. To determine the proper place of the property tax in the fiscal system it is necessary to understand its nature. The tax is variously perceived as a tax on capital income; as a wealth tax; as a pure-benefit, or user-pay, tax; and as an excise tax (see Hobson 1988). In all cases it is a real property tax levied on assessment based on market value that is being considered. There are some respects in which the property tax can be legitimately viewed as a tax on capital income, real and imputed, and no doubt it does have some effect on the supply and allocation of capital. But it is questionable that this is the most useful or appropriate way to view it. It is certainly not a wealth tax, although it is often so regarded. The owner may not have full equity in the property and indeed may have only a small proportion of equity, and yet he or she is liable for the tax on the full value of the property. The tax is most correctly and usefully seen as a tax on the value of the service rendered by the property to the owner or tenant over the tax period, where the assessment at market value is the index of the value of the service rendered by the property. The connection is an integral and direct one, for market value is simply the capital value of the imputed or actual income yielded by the property. In the pursuit of equity, uniform market value assessment is therefore clearly called for. Seen in this way the property tax translates into an excise or expenditure tax that, ex- pressed as a percentage of net rent (the value of the service rendered by the property), has the relatively high rate of around 15—20 percent in most municipalities, compared with the provincial retail sales taxes, the highest of which is 12 percent. However, when the federal manufac- turers’ sales tax is added on and, a fortiori, if it is replaced, as the federal government intends, by the proposed broad-based goods and services tax (a value-added tax), the excise rate of the property tax seems not excessively high. Moreover, a higher rate on the value of service provided 155 156 John F. Graham by residential property may be justified in view of the exclusion of the imputed income and of capital gains from such owner-occupied property in the income tax base (Thirsk 1982:394). Although there are some elements of benefit in relation to it, par- ticularly in the broad sense of benefit, as is explained below, the property tax is by no means a benefit tax in the narrow pure-benefits sense of being payment for public services in relation to the direct private benefits received by individuals. There is no quid pro quo relationship between the taxpayer and the taxing municipality. Moreover, it is inconsistent to argue, according to the New View, that the incidence of the real property tax is borne by all property owners of all property, real and otherwise, throughout the country and to regard that tax as being paid for benefits received by particular property owners and tenants; for if it is to be seen as a pure-benefits tax, the incidence must be shown to be on the particular property owners and tenants who benefit. But the notion of the property tax as a pure benefits tax is also questionable on other grounds. It is important to distinguish between two meanings of benefit, the narrow and the broad meanings, in relation to taxation. The narrow meaning has to do with individuals paying taxes according to the benefits they receive. Such taxes are then virtually tax prices. There are then no distributional aspects, except in the negative sense that only those who can pay for a public service will receive it. It is in this sense that the term “pure-benefit” principle is used. In the broader meaning of benefit taxation the argument is that, while citizens in a community should collectively pay for the services from which they benefit, the payment need not be on a quid pro quo basis. Rather, where vertical equity is a consideration, as in the cases where the goods have public goods characteristics, and where there are distribu- tional considerations, payment should be on an ability-to-pay basis. That is, the concept of ability to pay does not exclude the broad concept of benefit. Indeed, the two are complementary. And, assuming that, by and large, residential property owners and tenants living in higher- valued, and therefore higher-taxed, properties have higher incomes, they tend to have greater ability to pay. The general residential property tax is therefore appropriate for financing local public goods. There is, of course, also the element of capitalization to be taken into account, where capitalization of the property tax to some degree likely offsets the cap- italization of the benefits from some local public services. In addition the Property Tax in the Fiscal System issues of possible regressivity and of heavy burdens on the poor have to be considered and addressed. While the first, narrow, concept of benefit applies to some local services (those that are essentially private goods), it cannot be addressed by the general property tax. However, there are important pure-public- good, mixed-good and merit-good aspects, as well as outright distribu- tional considerations, with respect to many local services, that make the application of pure-benefit taxation inappropriate and inequitable with respect to those services and that require attention to ability to pay. In the case of these services the general property tax is suitable. These are the reasons for the earlier statement that, on the issue of the principle on which to assign the costs of providing public services (that is, pricing, user-pay, some other benefit basis, or ability to pay), it is necessary to have regard for the services in question. Municipal sales and income taxes are frequently suggested as alterna- tives to the use of the property tax at the municipal level. It is question- able that municipal use of these tax bases is desirable. Not only are there serious boundary problems (Thirsk 1982:399, 401) and the usual prob- lems from adding another layer to the jurisdictions using these bases, but, if fiscal equity is a concern, as is contended here, there would then have to be equalization grants to compensate for the municipal differ- ences in fiscal capacity with respect to those tax bases. It is preferable for the provinces to share these and other revenues with the municipalities through grants made on an equitable basis, as all provinces now do in one way or another. The sharing of set percentages of the income tax, as is done in Manitoba, for example, makes the sharing more visible, but it is not superior in principle to formulas that provide for municipal grants that increase at the same rate as total provincial revenue, as is the case in Nova Scotia. It is generally agreed that the integrity of the real property tax, particularly the maintenance of equity, depends largely on maintaining up-to-date accurate assessments at market value for all classes of prop- erty. Failure to do this can ultimately lead to the breakdown of the property tax when the revision of assessments would lead to politically unacceptable shifts in the tax burden both within and between classes of property. Some parts of Ontario, particularly Toronto, are at present in this position. This situation would seem to call for a phasing in of assessment revision to ease the pain of those who would bear increased burdens, even though it would delay the full restoration of equity for 157 158 John F. Graham those whose burdens have been and, for a time, would continue to be excessive. Since even this gradual reform is politically difficult, an alternative of abandoning the tax based on assessed value and moving to a tax based on frontage, with geographical zoning of properties according to types and levels of services provided, is being considered (Slack 1988; Bossons 1981; Hobson 1988:20—22). This would be virtually a poll tax on houses that would eliminate any connection between tax and ability to pay and would regard the taxes simply as crude fees for services. This is the direction in which England has recently moved. This sytem would ignore the public goods characteristics of many local goods and would abandon any notion of value of property reflecting ability to pay. It would also lead to substantial shifts in burdens, which, although the redistribution effects would be different, could be as difficult politically as assessment revision. To substitute a poll tax on property for taxation on market value would be a decidedly retrograde step. Its advocacy reflects a naive, or at least desperate, view of the issues. It is best to leave the poll tax in the oblivion to which most of the public finance literature has rightly con- demned it. Assignment Problems This section shows how the principles of assignment and correspond- ence in the public finance literature would be largely met by the proposed arrangements, both with respect to public services and to location (see also Oates 1972, chaps. 2 and 4; Breton and Scott 1978). Goods and Services Assigning responsibility for general services to the province would inter- nalize the externalities (spillovers) that would otherwise ensue if those services were provided by the municipalities, would therefore help to ensure that the full benefits from these services were taken into account in determining what levels of services to provide, and would be the most effective way of ensuring that all citizens were equitably treated in both the provision of the services and in the imposing of the tax burden to finance them. Property Tax in the Fiscal System 159 Since most local services are regional in character, the assignment of local services to regional municipalities would internalize the exter- nalities that otherwise exist with their provision by the existing munici- palities and would therefore permit the full benefits and costs to be taken into account in determining their level. Moreover, there would be, as is mentioned earlier, some administrative economies of scale and gains in administrative capacity with larger municipal units. If political obstacles to establishing the proposed regional munici- palities proved to be insuperable, as is quite likely, the financial structure proposed is still valid for the existing municipal units, for all of the reasons given. The municipal structure would be less efficient than it could be but may be the best that could be achieved within the realm of practical politics. Some efforts can still be made to deal with externalities by such devices as two-tier metropolitan government or by intermunici- pal cooperation, but, judging from experience with them, such efforts are almost bound to produce rather poor second-best results. Revenue Sources On the financial side, since the province would be assuming respon- sibility for providing the general services, some transfer of property tax revenue to the province would likely be necessary. At the same time it is desirable that the municipalities have a major tax base that is exclusively theirs to use in accordance with their own priorities in the planning and provision of local services. Moreover, it is equally desirable, for effi- ciency reasons, that municipalities be directly financially responsible and accountable to their citizens. Analysis of the nature of the residential and nonresidential elements of the property tax base indicates that it is appropriate to assign the residential portion to the municipalities, on the reasoning that the incidence and effects of the residential property tax are principally confined to the municipality, in accordance with the Old View, as are the benefits from local services. There are a number of reasons why the nonresidential property tax should be assigned to the provincial government. The incidence is borne widely through the increased prices paid for goods sold outside as well as inside the municipality by industries and commercial establishments. A further consideration is that, on grounds of neutrality, it is desirable that the tax on business be uniform throughout the province, so that indus- trial location will be determined by basic economic considerations of 160 John F. Graham efficiency, rather than be influenced by deliberate or inadvertent tax differentials. A still further consideration is the relatively greater stability of the tax rate under provincial jurisdiction, which would provide a more favorable business climate. While the rates of municipal property taxes are generally changed annually once the required amount of reve- nue is determined in the municipal budget, the rates of provincial taxes are changed much less frequently, since there is a greater range of taxes and of borrowing possibilities to work with. (The experience of New Brunswick supports this expectation of stability.) Overall, provincial levying of this tax would internalize the externalities that are associated with it, at least to the provincial level—an important efficiency consider- ation (see also Thirsk 1982:396—407). The revenues required by the province would come only partly, of course, from the nonresidential property tax. They would come mainly from the province’s income and expenditure taxes and from its transfers from the federal government. It is important to remember that these recommendations rest on the earlier argument favoring the Old View over the New View of incidence. While the ground for confidence in this argument is believed to be very strong, it must be acknowledged that if a variant of the New View were found to hold up under real-world assumptions, the assignment solution could be quite different and considerably more complex than the one postulated here. Based on the above argument, the financial arrangements for the municipalities would appropriately consist of: (a) a municipality-wide tax rate on residential real property for munici- pality-wide services having public goods characteristics; (b) area tax rates on residential real property for services, also having public goods characteristics, provided by the municipality, where such services are confined to an area within the municipality; (c) local improvement charges based upon frontage for capital facilities of primary benefit to particular property owners, such as local access roads, sidewalks, curbs, gutters, storm sewers, and sanitary sewer and water main laterals, having principally private goods characteristics; and (d) user charges for most of the municipal share of water and sanitary sewerage capital and operating costs other than laterals, on residential, industrial, commercial, and institutional users, and for parking, similar to the way in which prices are paid for the services of other public utilities, such as those of electricity and telephone companies. Property Tax in the Fiscal System It should be noted that all properties, both residential and nonresiden- tial, and as well as the property of churches and of other charitable organizations which may be otherwise tax exempt, would be required to pay local improvement and user charges to the municipalities. This means that, although the province would receive the revenues from the nonresidential property tax, the municipalities would levy and receive local improvement and user charges with respect to such property in direct payment for services listed above in (c) and (d). There can, and it is contended here should, also be provision for comprehensive unconditional equalization grants to enable all munici- palities to provide a standard level of local services at the same tax rate (see Graham 1963:239—43; New Brunswick Royal Commission 1963: 270-87; Nova Scotia Royal Commission 1974, 2:57—81). Such a provi- sion would be necessary to ensure horizontal equity in accordance with the fiscal equity principle and would recognize the province’s ultimate overall responsibility for provision of local as well as general services in an equitable and efficient manner. In summary the principle of assignment on the financing (revenue) side calls for local improvement and user charges, where appropriate, and, otherwise, for taxes at uniform rates within the jurisdiction or area in which the public services are provided. The general services would be financed by various provincial taxes levied at uniform provincial rates throughout the province. Certain local services, with private goods char- acteristics, would be financed by local-improvement and user charges within the narrow meaning of the benefit principle. Local services with public goods characteristics would be financed by property taxes levied at uniform regional municipal rates for region-wide services and at uniform area rates for services limited to particular areas within a munic- ipality. Furthermore, the recommended provincial equalization grants to the municipalities would ensure that all municipalities could provide standard levels of local services with comparable tax rates; that is, the important principle of horizontal equity, equal treatment of equals, would be met at the municipal level as well as at the provincial level. (That principle is met, in large part, at the provincial level through unconditional federal equalization grants.) In the broader meaning of benefit used here, those citizens benefiting from province-wide general services would pay for them with province- wide taxes and those citizens benefiting from local services with public goods characteristics would pay for them with local region-wide, or area, 161 162 John F. Graham taxes. This does not mean, however, that these taxes would be paid by in- dividuals according to benefits received from public services—the other, quite different, narrow, “pure-benefits” meaning of benefit taxation. At the municipal level the tax burden for public goods would be borne in relation to the value of residential property owned and, to the extent that property taxes are shifted to tenants, on the value of property occupied. On the assumption that owners and occupiers of higher valued properties are generally in higher-income groups, the burden of property taxes would generally be distributed according to income, that is, roughly according to ability to pay. There is quite substantial provision for local improvement and user charges in these proposals, but there is also recognition of the very considerable pure-public-good, mixed-good, merit-good, and distribu- tional elements in most local services that call for an ability-to-pay approach, rather than a pure-benefit approach. If we consider the nature of local services, we find considerable such public good aspects in them, and particularly in physical, social, and financial planning; in control of development; in environmental protec- tion and control; in protective services; in transportation-related ser- vices; in beautification; in recreational and cultural services; in tourist and industrial promotion; in general government; and even among the services most closely related to property. What is puzzling in much of the current advocacy of the pure-benefit principle is not the recognition that there is a substantial and perhaps expanded place for the user-pay principle, which is strongly supported here, but that there seems to be a bias toward assuming, without ade- quately examining their nature, that municipal services are prepon- derantly in the nature of private goods, while accepting with little question that the services provided by higher levels of government are principally public goods evincing market failure.4 The applicability of the user-pay principle, while considerable, is decidedly limited. Certainly, the greater use of user and local improvement charges beyond those proposed here may be explored (see Bird 1976b), for example, for some elements of recreational and protective services, but it should not be forgotten that local improvement and user charges, in most cases, con- tain some element of tax. The more extensively such charges are used for essential local services, the greater will be the burden on low-income members of the community in relation to that on the high-income mem- bers. A major weakness of much of the current discussion of municipal Property Tax in the Fiscal System 163 finance is the obsession with a pure benefits approach, where it is quite inappropriate for many local services. Of course, if general services (education, social services, etc., all of which fall into one or more of the market failure categories for which the user-pay principle is not appro- priate) remain in part the responsibility of municipalities, then a general property tax must be continued on that account as well. With respect to the use of the property tax to finance general services such as elementary and secondary education and social services, it is not so much a matter of the inappropriateness of the property tax for this purpose that is at issue; rather, it is that the residential property tax is a particularly appropriate tax base for financing local services and that it is highly desirable that the municipalities have a tax base that they can use exclusively in relation to their own priorities. Such an arrangement is also very much in accordance with the important principles of direct financial responsibility and accountability. The Property Tax: Other Issues With respect to regressivity of the residential property tax, the question is not only the rate of the tax with respect to income, but also the weight of the tax on low-income families. There is often an excessive preoccupa- tion with regressivity as such. A tax can be highly progressive and yet the burden on low-income groups still be great. Assuming that there is, in general, a positive relationship between residential property value and income, if regressivity is considered to be significant, that problem, and the problem of the burden on the poor, can be addressed in a number of ways. The one favored here is an assessment exemption that would decrease the property tax for homeowners and tenants in lower-valued properties and for larger families with greater than average housing needs. For any given total tax levy, this adjustment would transfer some of the burden systematically from the owners of lower-valued properties to the owners of higher-valued properties, for the exemption would be a declining proportion of assessment as assessment increases. This adjust- ment would be self-financing within each municipality, and so would not mean any revenue loss to the province, unlike the property tax credits applied against the personal income tax now provided in some proy- inces. This is a considerable advantage in low-income provinces that have less room for financial maneuver than have wealthier provinces (see 164 John F. Graham Nova Scotia Royal Commission 1974, 2:191—201). A possible formula for the exemption, proposed by the Nova Scotia Royal Commission (2:191—201), Is: 2 + Actual number of children per dwelling x 1 Exemption = — § —————————— oo iit 4 2+ Average number of children per dwelling x Average assessment per dwelling. There is a particular problem for some retirees who have low incomes and therefore may have difficulty in paying property taxes and in con- tinuing to live in their homes, homes which they may have spent a lifetime acquiring. At the same time retirees may have significant wealth in the form of equity in their homes (often 100 percent). This matter can readily be dealt with at no ultimate public expense by devices for advanc- ing these citizens the funds, which would become charges against their equity in their property, to meet their property taxes and, indeed, other costs of remaining in their homes (2:201—5). There is much discussion and questioning in the early and recent literature about the appropriateness of the general property tax on nonresidential property (in particular, Thirsk 1982:396—407). The ob- jection is not to local improvement and user charges of the sort proposed here, but to a general tax based on market value and particularly to the tax being imposed at a higher rate than that on residential property, especially when supplemented by a business tax that is also usually based on market value. The argument generally emphasizes the relatively low level of benefits from municipal services accruing to owners and oc- cupiers of nonresidential property compared to owners and occupiers of residential property. It has been argued here that a nonresidential property tax is more appropriately levied at the provincial level, especially if the provinces assume full financial responsibility for the general services as has been done in New Brunswick and Prince Edward Island. But it is not clear that the tax, even when levied at a higher rate than the rate on residential property, is quite the villain it is made out to be, even when levied by municipalities. Again, the argument seems clouded by an obsession with the property tax as a pure-benefits tax. There are good economic grounds for the nonresidential rate being higher than the residential rate. Carl Shoup, in his treatise, Public Finance, has so argued. He points out that the residential property tax is concentrated on one portion of a family’s Property Tax in the Fiscal System spending—housing—which is a necessity and is a large part of the total spending by low-income households, while the business property tax, to the extent it is passed on to consumers, is spread across all items of expenditure, including luxuries, and hence is not so heavy a burden on the poor. In Professor Shoup’s words, in referring to the residential property tax: No other tax strikes so severely, relative to other expenditures, an outlay that is so large a proportion of the total family budget of a low-income house- hold. .. . There is a good case for imposing the housing part of the real estate tax at a much lower rate than the business-property part, since the latter does not concentrate its pressure so heavily on one aspect of the consumer budget, an aspect that is relatively more important in the low-income budget. (1969b: 391) Moreover, to the extent the tax on nonresidential property is shifted to consumers, the after-tax return is not affected by the tax. (If this is the case, it means that, along with the already doubtful validity of the new, general equilibrium view of incidence, that view is irrelevant in the case of the nonresidential property tax, at least to the portion on improve- ments.) There is the further consideration that taxes on nonresidential property are deductible in arriving at taxable business income, a priv- ilege not accorded residential property owners with respect to personal income taxation in Canada. It is significant that in practice the nonresi- dential effective tax rate is almost universally higher than the residential rate (Economic Council of Canada 1987:101; Thirsk 1982:388), al- though it should be acknowledged that governments may simply be taxing what the “traffic will bear,” rather than following any well thought out principle of taxation. As elsewhere in public finance, politi- cal decisions are often not based solely, or even primarily, on economic principles. Throughout this paper it has been important to bear in mind the distinctions among absolute and differential tax and expenditure inci- dence and budget incidence and their relation to the analysis. The central discussion in this paper, which emphasizes the importance of consider- ing the property tax in relation to the nature of the public services provided, is implicitly couched in terms of budget incidence in that it takes account of the combined budgetary effects, both allocative and distributive, of expenditure and revenue decisions. When discussing questions of regressivity and weight of the property tax and the differen- 165 166 John F. Graham tiation between residential and nonresidential property taxes, the focus is implicitly on absolute tax incidence. There would, however, seem to be no conflict here when it is borne in mind that these questions are ex- plored after establishing the desired budget incidence. Conclusions As stated at the beginning, the aim in this paper has been to suggest an analytical framework for establishing an efficient and equitable system of municipal finance that takes cognizance of the attributes of property taxation and of the nature of municipal services and that can serve as an operational guide to policy makers. The framework and rationale of the paper can be summarized briefly. Using local improvement and user chargers for local private-type goods and the general property tax for local public goods incorporates both the narrow and broad concepts of benefit, as they apply, with provision for the element of ability to pay in the latter case. Assigning general services to the province and local services to regional municipalities internalizes the externalities for both types of services. Assigning the residential property tax to the municipalities, where the incidence is confined within the municipality, and the nonresidential property tax to the province, where the incidence is at least province-wide, provides the correspond- ing fit on the revenue side. The provision of services is thus brought into harmony with the revenue sources. The use of unconditional grants in accordance with the principle of fiscal equity rounds out the picture. The objectives of efficiency and equity are then largely met within a frame- work of rational assignment of functions and revenues. Vertical and horizontal fiscal balance can thus largely be achieved, with sufficient flexibility for year-to-year adjustments that will always need to be made. Above all, the municipalities can function as real communities in which citizens can effectively pursue their common interests in the plan- ning, provision, and enjoyment of public services. Notes The author is much indebted to Irwin Gillespie, Malcolm Gillis, and to an anonymous reviewer for their most valuable criticisms and suggestions regarding the organization and content of this paper. Property Tax in the Fiscal System 1. For a useful summary and evaluation, see Economic Council of Canada (1987, chap. 9). While some of the argument in that chapter may be questioned, it does nicely lay out the main elements and issues concerning the tax. Another very good article covering similar ground is Thirsk (1982). Also useful through- out is Kitchen (1984). 2. See Bird (1976a) for a good summary and evaluation. See also Economic Council of Canada (1987:101) for a succinct statement. For fuller treatment, see Aaron (1975), Miezkowski (1972), and Miezkowski and Zodrow (1984). For a good recent examination of the old and new views in relation to the benefits view, see Hobson (1987). For a thorough earlier theoretical and empirical examination of the Old View, see the widely acknowledged monograph by Netzer (1966). 3. This point is made in Economic Council of Canada (1987:100) and in Bird (1976a:331—32). This suggestion may be “comforting,” but the reasoning is not very convincing. If the theory is valid, it would seem to apply whatever the level of jurisdiction. As Hobson (1988) points out, “... even a tax change in a single jurisdiction will result in a reduction in the net return on capital as a whole in the taxing jurisdiction.” 4. Carl Shoup’s distinction between collective goods and government- provided goods is instructive regarding this point and the preceding discussion (1969b:68—74). 167 Comment STANLEY WINER Choosing an Appropriate Time Horizon for Taxation The papers by Richard Goode and William Vickrey provide broadly ranging, lively discussions of two of the pillars of contemporary revenue structures, the personal and the corporate income taxes. Richard Goode surveys the enormous body of normative theory on the personal income tax from the perspective of one who is firmly committed to a progressive tax on comprehensively defined annual personal income. William Vick- rey’s paper is more narrowly focused on the structure and impact of the modern corporation income tax. This is a tax which Vickrey would like to see abolished, though much of his paper is concerned with the many practical problems that make such a reform unlikely. I would like to draw attention to an important issue in tax theory underlying both these papers, namely the choice of an appropriate time horizon for taxation purposes. It is probably an understatement to say that the authors have discussed the choice of time horizon before. I hope they do not mind if I raise the issue once again, for it is an important one about which a consensus remains elusive. Richard Goode’s rich discussion of the tapestry of tax theory is marked by a strong emphasis on the short run. Equity, he thinks, should be assessed and reassessed periodically. Incentive effects over long time horizons, in distinction to such effects in the shorter run, are not consid- ered by him to be of great relevance for policy purposes. Part of this emphasis on short horizons comes, no doubt, from his considered belief that there are substantial uncertainties in the economic environment over long horizons coupled with his view that liquidity constraints have serious consequences for a substantial number of people. He points out that liquidity constraints and other imperfections in capital markets have wide-ranging implications for tax theory and practice. Such con- straints imply that a tax on lifetime consumption, bequests, and gifts is not equal in present value terms to the yield and burden of an equal rate Shanz-Haig-Simons income tax. Moreover, people who are liquidity Choosing a Tax Time Horizon constrained cannot take advantage of the presumed opportunity that consumption taxes offer to optimize consumption-savings decisions. The conclusions about equivalency in present value terms of taxes also require constancy of tax rates and discount rates and assumptions about the long run which are not likely to hold and which bias analyses of the choice of tax base in favor of a consumption tax. Thus, he notes, argu- ments for utilizing short time horizons complement other reasons for the choice of annual income as the ideal tax base. To summarize Richard Goode’s view governments should not make judgments concerning allocation, equity, or the choice of tax bases on the basis of ex ante present value calculations over long time horizons. Tax designs should acknowledge ex post changes in an individual’s needs and capacities and in the economic environment over long periods of time. William Vickrey’s ideas about the appropriate time horizon are rather different. He has proposed, in Goode’s words from a paper that begins by thanking without implicating Vickrey, a highly sophisticated pro- posal to make the income tax more or less neutral with respect to the timing of income receipts and tax payments during a well specified period of time (Goode 1980). This would involve “considering all pay- ments of income tax, with respect to income reported since some starting date, as interest-bearing deposits in a taxpayer’s account with the trea- sury. The accumulated balance in this account would then be available as a credit against whatever tax is found to be due for the entire period to date, on the basis of the total income thus far reported for the period, ac- cording to the tax schedule appropriate to the period covered” (Vickrey 1980:117—18). In the revenue system without a corporate tax Vickrey would like to implement, cumulative averaging applied to the personal income tax would neatly eliminate the problem of the deferment of income realiza- tion through the retention of corporate earnings, as well as eliminating all problems stemming from the timing of corporate income. If the rate of interest applicable to the calculation of credits in the tax account is competitive, the incentive to retain earnings in the corporation is much reduced. Professor Vickrey is quick to raise the problem for cumulative averag- ing posed by taxpayers who leave the tax jurisdiction, as well as the necessity of collecting tax on a source basis from the operations of foreign-based corporations. In the absence of cumulative assessment, 169 170 Stanley Winer there would remain the incentive to gain advantage through the reten- tion of earnings in the corporation once the corporation income tax was abolished. In this case he suggests that it would be desirable to levy an annual tax on the accumulated undistributed surplus of the corporation at a rate that would represent an interest payment on the amount of tax deferred. While individuals may voice displeasure at cumulative averaging of personal income taxes for reasons suggested in Richard Goode’s paper, in response to this suggestion for a tax on undistributed profits corpora- tions might similarly complain about the uncertainties of novel invest- ment projects and argue for the necessity of retained earnings as a source of financing these projects. Could there ever be a reconciliation of the views of Goode and Vickrey concerning the choice of time horizon? Richard Goode’s argu- ments in favor of a short horizon are all qualitative, and I wonder if, say, three years might also be considered the short run by him. Perhaps it would be practical to incorporate into personal or corporate taxation a limited form of cumulative averaging, and I wonder if this might be useful from William Vickrey’s point of view. Would efficiency be suffi- ciently enhanced by this modest lengthening of the horizon over which personal taxation is assessed to compensate for the extra cost of admin- istration? Toward the end of his paper Professor Vickrey briefly raises another fundamental problem for the implementation of long horizons, one not discussed by Richard Goode. Vickrey recognizes that governments al- ways have an incentive to revise their programs after individuals have committed themselves to economic actions, in order to gain further political advantage. In the present context this inconsistency in political behavior over time may lead governments to change tax systems sub- stantially in midstream, making the actual implementation of a stable system of cumulative averaging or lifetime income taxation extremely unlikely in representative democracies, as well as making the transition to a world without a corporation income tax very difficult. It seems to me that we do not want to dispense with the electoral system we have evolved, and this fact, if I may call it that, suggests the need to reformulate the way one thinks about the choice of time horizon for tax purposes. Perhaps the very long time horizon is not a feasible alternative in democratic regimes without fundamental changes in the nature of our political institutions. Some time ago Richard Bird sug- Choosing a Tax Time Horizon gested the establishment of a quasi-independent tax commission to re- move some aspects of tax policy from the vagaries of day-to-day politics. Perhaps such a commission would make longer time horizons politically stable, though how such an institutional reform would come about remains as puzzling to me as the engineering of the end of the corpora- tion income tax is to William Vickrey. 171 Comment WR WoUN Gi DES Pale Taxing Consumption and Wealth The two papers by Thirsk on the value-added tax and by Graham on the property tax provide an interesting contrast. The value-added tax (vaT) is a comparatively new addition to the family of taxes, whereas the property tax has been around for a very long time. Thirk’s paper traces the seminal contributions of Carl Shoup to our understanding of this new tax, whereas Graham’s paper develops the role of the property tax as a major component in the efficient and equitable provision of local public goods. Wayne Thirsk eloquently develops the contributions of Carl Shoup as the “intellectual father” of vAT taxes. Thirsk identifies one common source of all these contributions as Shoup’s “intertwining of theory and practice.” From the taxonomy of vaT choices, through the equivalence theorems, through comparative evaluations of vat with retail sales taxes and turnover taxes, to the methods of collecting vat, Shoup’s applica- tion of “pristine pragmatism” is urging us to “judge events by their effects and not be charmed into believing that we have discovered a magic tax potion for solving all of our fiscal ailments.” This is excellent advice to all who wish to develop useful recommen- dations for tax systems within a normative framework that encompasses objectives such as efficiency and equity. There is another closely related common source of Carl Shoup’s contributions to the growth of vat, which is alluded to in Thirsk’s paper—the political economy of vaT taxation, in the broadest sense. Value-added taxes—and all tax reforms, for that matter—are proposed by ministers, legislated by parliaments or congresses, administered by bureaucrats, and legitimized by voter-taxpayers. No idea, regardless of how cogently and eloquently developed, will be transformed into action unless it convinces at least a core group of each of these politicians, bureaucrats, and voters—that it is in their own self interest to do so. Carl Shoup’s analytical approach and policy recommendations pro- vided a framework that could accommodate governments choosing to implement the kind of vats observed today. The careful analysis of Taxing Consumption and Wealth replacing an existing tax (say, a manufacturer’s sales tax) with a vaAT tax, the detailed comparisons of a VAT tax with a retail sales tax, and the extensive discussion of structural, institutional, and political features of the country were integrated by Shoup and his fellow commissioners to provide a detailed blueprint for action. These detailed country-specific blueprints provided, in addition to sound economic policy advice, some information on the potential winners and losers from such action. This could form the basis for political coalitions that would see the recom- mended reforms through to implementation. Thirsk correctly identifies the importance of the administrative costs of collection, compliance, and enforcement in determining the type and structure of VAT tax chosen by very different countries. However, these costs are just one source of the political pressures that influence govern- ments in their choice of tax reform measures. The high administrative costs of including the agricultural sector in a developing country may account for its exemption from the vaT base; such administrative costs could not explain the exemption of basic foods from the proposed Canadian VAT tax. John Graham, like Carl Shoup, develops the case for a tax—the property tax. And like Shoup, his analytical approach and policy advice are based on an intertwining of theory and practice with careful consid- eration of the institutional, structural, and political context. Graham’s paper develops an analytical framework that encompasses both sides of the local-provincial public household (the spending side and the financ- ing side), assigns expenditure functions and revenue sources to local, regional, and provincial levels, and highlights the special attributes of the property tax as a major instrument of municipal finance. The strength of this paper is the development of a comprehensive conceptual framework that stresses the crucial link between the spend- ing and financing of the local-provincial level. The nature of the goods and services provided is important—and Graham carefully discusses just how important—in devising revenue sources that accommodate their provision. The approach, in terminology advanced by Richard Mus- grave and Carl Shoup, is one of “budget incidence.” This conceptual “budget incidence” framework provides the crucial conclusions on the assignment issues. Local improvement and user charges are assigned to municipalities for private type municipal goods and services, the residential property tax is assigned to municipalities for the financing of local type municipal goods and services (that can be 173 174 W. Irwin Gillespie coordinated by regional municipalities), and the nonresidential property tax is assigned to the provinces for the financing of general, province- wide services. Graham develops the case that this assignment is an efficient and equitable system of municipal finance. IV = The Expenditure Mix 10.9 LORRAINE EDEN AND WEEVIIPLEE Li. MCMIELAN Local Public Goods: Shoup Revisited Since the 1954 publication of Samuelson’s “The Pure Theory of Public Expenditure,” a considerable literature has emerged which fo- cuses on the concept and characteristics of public goods. The purpose of our paper is to review the literature on local public goods (LPGs), a subset of public goods characterized by benefits that taper off spatially. The intent is to situate this literature in relation to Carl Sumner Shoup’s work on particular government services, notably fire prevention and police protection services. Our concern is with the issues identified by Shoup, as early as 1964, and their subsequent treatment (or nontreatment) in the literature. Shoup’s writings focus on the characteristics of public services, the definition and measurement of output of individual services, the slopes and shapes of the cost functions as service level, population, and area vary, and the distribution and incidence of the benefits from such ser- vices. Much of his work deals with fire protection and crime prevention services. Some of these areas have been actively pursued by subsequent researchers; others have not. Alternatively, some of the current areas of interest in local public finance have not been addressed by Shoup.! While his taxonomy of public services has not supplanted the more popular Musgrave-Samuelson approach, Shoup’s concern with the particular- ities of individual government-provided services has generated a large amount of subsequent research. Our paper looks at developments in those areas of local public finance which we believe link most closely to Shoup’s contributions. The first section of the paper focuses on the characteristics of LPGs, and is fol- lowed by sections on defining and measuring output, cost functions, and distributional issues. The fifth section summarizes these results and addresses the question: have the issues raised by Shoup been successfully answered, or are there areas that remain to be addressed? We conclude 178 Lorraine Eden and Melville L. McMillan that empirical and theoretical studies of local public goods have just begun to handle the four issues Shoup addressed twenty years earlier and that much work remains to be done. The Characteristics of Local Public Goods The Musgrave-Samuelson Approach Two of the key early writers on the characteristics of public goods were Paul Samuelson (1954, 1955, 1958a) and Richard Musgrave (1959, 1969). The widely quoted definition of a public good as one where “each individual’s consumption of such a good leads to no subtraction from other individuals’ consumption of that good” (Samuelson 1954:179) defines the first characteristic, nonrivalness in consumption. In his 1959 book Richard Musgrave refers to social wants as “those wants satisfied by services that must be consumed in equal amounts by all” (8). He argues that nonexcludability, that is, the inability of private providers to exclude potential consumers through the price mechanism (and thus be unable to require consumers to reveal their preferences), implies that social wants must be met through public provision. This thinking is formalized in R. Musgrave (1969), which identifies the two characteris- tics of social goods as nonrivalness in consumption and nonexcludability from consumption. Nonexcludability means that the market is likely to be an inefficient means of providing such commodities; however, whether public provision is necessary or superior to the market depends on the circumstances (see Head 1972). Pure public goods have both characteristics. Figure 1(A), based on Stiglitz (1986:103) and Musgrave, Musgrave, and Bird (1987:45), shows a box with combinations of two characteristics, rivalness and excludability.2 Point a represents the polar case of a pure public good; point c represents the polar case of a private good. Where local public goods like fire, police, and recreational services fit within the box depends partly on the author and the circumstances. In practice fire and police services, and often recreational services, are publicly produced and provided free of direct charge at the municipal level to residents of the jurisdiction. Stiglitz (1986:100) situates fire protection toward the lower right-hand corner, that is, the marginal cost of protecting an additional household in a community is low, but it is Local Public Goods: Shoup Revisited 179 (A) The Musgrave-Samuelson Box vel by private good pure public good nonrival a not feasible costless Excludability (B) The Shoup Circles Group-Consumption Goods Collective-Consumption Goods (C) Integrating the Musgrave-Samuelson and Shoup Approaches A rival B | 344 Quasi-private | © | Goods Quasi- GCGs GCGs | Quasi-CCGs nonrival a | | +2 d v nonexcludable excludable eC CGsE = = Figure 1: The Musgrave-Samuelson and Shoup Approaches to Public Goods 180 Lorraine Eden and Melville L. McMillan feasible to exclude the household from the service. Police protection, however, may fall toward the lower left. Although the cost of protecting another household is low, it is not feasible to exclude it (fully) from the benefits which police services afford the community (even if direct ser- vices were not provided to the specific unit). The Shoup Approach to Public Goods Shoup uses a different taxonomy for public goods than the Musgrave- Samuelson one outlined above.? Commodities are distributed free of charge by the government for four general reasons, according to Shoup: preservation of the nation-state against aggression, group-consumption goods, redistribution in kind, and miscellaneous reasons such as diffi- culty in measuring output and uninsurable costs (Shoup 1969b:65). In general only services are freely distributed because goods can be resold whereas few services are resalable. Of these four reasons Shoup concentrates on the second: the group- consumption good (GCG) (see also Shoup 1974). A Gcca is defined as a good or service that can be supplied more efficiently through a non- marketing method than by a rationing (price or nonprice) method. The nonmarketing or group method means that the Gcc is supplied simulta- neously to all members of the group, and exclusion is not feasible. The key criterion for choosing between the two methods is efficiency, defined as production and distribution at lowest cost including the cost of resources used in operating the method. Efficiency must be defined in incremental terms as the cost of an additional unit of the service in per capita terms. It is possible for a particular good to switch from being a GCG to anon-Gca, depending on the relative costs of the two methods at different levels of supply. Within the group-consumption good framework Shoup addresses the specific characteristics of individual GccGs. When the service is govern- ment provided, it is available within a given area from one or more points of input (e.g., a fire station). If the benefits taper off as the recipient is located further away from the point of input, different households receive differing amounts of the service depending on their location within the jurisdiction. This type of GCG is now identified as a local public good. Shoup notes that the point of input can be mobile (e.g., patrol cars) and there may be several input points (e.g., recreation centers). Spillovers of benefits between units can also occur. Local Public Goods: Shoup Revisited 181 While nonexcludability of a GcG means that discrimination between individual households or firms is impossible (or at least less costly than exclusion), discrimination at the level of the group may be possible. Such intragroup discrimination may be involuntary or deliberate on the part of the local government. For example, involuntary discrimination be- tween rich and poor neighborhoods can happen even though equal numbers of police are provided to both areas because the inputs (number of police) do not produce equal benefits (the same risk of becoming a victim of crime) in rich and poor areas. Deliberate discrimination be- tween groups can occur if distinct subunits exist within a jurisdiction and the local government discriminates in the level of service provided to different areas (e.g., by deliberately providing higher service levels in wealthy areas). Separate from the concept of a GCG is a collective-consumption good (ccG), which Shoup defines as one that, if supplied to one person, can be supplied to additional people at zero incremental cost. Thus the total cost of supplying a given level of service of a CcG to each household remains unchanged as the size of the consuming group expands. A CcG may be excludable. Shoup illustrates the range of excludability with theaters and mosquito abatement. A ccc thus has the characteristic of nonrivalness of consumption but may be either excludable or nonex- cludable, whereas a Gcc has the nonexcludability characteristic but may be either rival or nonrival. The distinction between GcGs and ccGs is illustrated in figure 1(B) which is based on Shoup (1969b:73, and chap. 5). The left-hand circle, composed of segments 1, 2, 3, and 4, represents GCGs (nonexcludable public services); the right-hand circle, composed of areas 1, 2, and 5, is ccGs (nonrival public services). The Gcc circle contains a smaller circle, areas 2 plus 3, where intragroup discrimination is feasible; outside this circle, in area 1 plus 4, intragroup discrimination is not possible. Outside both Gcc-cce circles are goods that are excludable and rival, that is, private goods. The individual segments can be explained as follows. Where the two circles overlap (categories 1 and 2), the services are nonrival and nonex- cludable; this is the polar case of a public good. In category 1, pure public goods where intragroup discrimination is not feasible, Shoup includes military expenditures, public health, space research, contract enforcement, and externalities. Shoup speculates that category 2, pure public goods where intragroup discrimination is feasible, is nearly empty. Category 3, GCGs with feasible intragroup discrimination, in- 182 Lorraine Eden and Melville L. McMillan cludes fire and police protection, street maintenance, and flood control. Shoup conjectures that category 4, GCGs with nonfeasible intragroup discrimination, is also empty. He hypothesizes that category 5, exclud- able ccGs, includes education and medicine, waste removal, and recre- ational services.* Shoup (1969b, 1974) clearly situates both fire and police protection as publicly provided, rival services with feasible intragroup discrimina- tion (area 3). He argues that the marginal cost of extending fire and police services to an additional consumer is nonzero so the services are rival. Once the service is provided by the municipality, it is not feasible to exclude individual households from consumption. Intragroup discrimi- nation is possible, but not at the level of the individual household or firm. Shoup positions recreational services, however, in category 5 as nonrival goods that are most efficiently produced and distributed by private firms.’ Walsh in this volume, on the other hand, argues that the case for private as opposed to public provision is not clear and that work remains to be done in this area. The Musgrave-Samuelson box in figure 1(A) has an advantage over Shoup’s circles in that movements within the box can be interpreted as corresponding to more or less rivalness and/or excludability; that is, rivalness increases in a northward movement, excludability in a west- ward movement. Such directions cannot be read from figure 1(B). Figure 1(A), however, does not allow for the intragroup discrimination/nondis- crimination split that exists in Shoup’s circles. Shoup’s approach also encourages thinking about the relative mix of GcGs and ccGs through changing the sizes of three circles and their degrees of overlapping. For example, a small right circle implies fewer joint goods and a smaller public sector; a large overlap between the two main circles increases the range of pure public goods. The Musgrave-Samuelson box, on the other hand, encourages linear thinking along the diagonal of the box. We compare Shoup’s GcG-ccG taxonomy with the Musgrave-Sam- uelson public goods taxonomy in figure 1(C). The comparison between the Shoup circles and the standard box treatment of public goods does not appear to have been made before. The comparison is a bit forced since exclusion in the Shoup taxonomy refers to all nonmarketing modes including queuing and rationing, whereas in the Musgrave-Samuelson approach exclusion normally refers to sale through the private market. Categories 1 plus 2 in figure 1(B) correspond to point a (pure public goods), categories 3 plus 4 (rival, nonexcludable services) to point b, and category 5 (excludable, nonrival services) to point d.° Private goods, the Local Public Goods: Shoup Revisited area outside the circles in figure 1(B), are represented as the single point c in figure 1(C). Thus the polar case of a group-consumption good can be translated as the vertical line ab in figure 1(C); the collective-consump- tion good, as the horizontal line ad. The Shoup focus on the polar GCG-CCG cases (the ab-ad lines) is therefore quite different from the Musgrave-Samuelson focus on the polar case of a pure public good, the single point a. In the Musgrave-Samuelson approach everything in the box other than points a and cis a grey area of quasi-public goods of varying degrees of rivalness and excludability. In the Shoup taxonomy, since GCGs are based on the criterion that nonexclusionary supply is less costly than rationing, we can interpret quasi-GcGs as the left half of the box in 1(C). Similarly, ccGs can be extended to include public services with positive, but low marginal costs per user. Quasi-ccGs are thus the bottom half of the box. Since these areas overlap, the lower left-hand quadrant can be seen as quasi-GCG/CccG cases. This leaves the last quadrant as quasi- private goods with c as the polar case. Shoup’s definitions of group-consumption goods and collective- consumption goods have not supplanted the more popular Musgrave- Samuelson rivalness-nonexcludability approach. It is interesting to spec- ulate as to why the Shoup taxonomy has not been more widely used. Perhaps the terminology is somewhat confusing (i.e., the difference be- tween the terms “group” and “collective” is not obvious). In addition Shoup’s definition of a group-consumption good depends on both the costs of production and distribution of the service; whereas subsequent literature has separated the issue of production (government versus private sector) from the question of provision (financing). The defini- tions of nonexcludability also differ; Shoup’s includes both price and nonprice forms of rationing (e.g., he defines public education as exclud- able because it is rationed). Most authors (see Head 1972) use exclud- ability to mean exclusion via the price mechanism only. In the following section we review the recent literature on the characteristics of local public goods that deals with the issues Shoup raised. Recent Developments: The Characteristics of LPGs Shoup’s definition of group-consumption goods depends on the relative efficiency in the production and distribution of local public goods under the marketing versus nonmarketing (i.e., nonrationed) modes. In order 183 184 Lorraine Eden and Melville L. McMillan to determine which method is most efficient, costs of private production, public production, private sale, public distribution free of charge, public distribution with user fees, and public distribution with rationing must be computed and compared. The most efficient method is the least cost combination of production and distribution methods for a given level of service. Given the data requirements, it is not surprising that little work has been done on nonexcludability as defined by Shoup.” Most work in local public finance has focused on rivalness, that is, on estimating points in a vertical direction in the Musgrave-Samuelson box. The methodol- ogy of Bergstrom and Goodman (BG, 1973), Borcherding and Deacon (BD, 1972), and other papers using their approach is the basis for much of our review. The Basic Rivalness Model and Results It was only with the pathbreaking BG-BD papers that estimates of the rivalness of public services became available. Both papers introduce into their models of the demand for public output a term to allow for the potential rivalness or congestion of the publicly produced good. They assume that the amount of the public output to the individual (q) is determined by the amount of output produced (Q) and the number (N) sharing that output as follows: QriQine (1) If « = o, q = Qand the public output is a pure public good. If a = 1,q = Q/N and the public output has private good characteristics in that each person benefits from only a per capita share of the total output. Inter- mediate values imply quasi publicness/privateness. The rivalness factor a is referred to variously as the crowding, capturability, congestion, and/or publicness parameter. It is the congestion elasticity and measures the percentage change in public output Q needed to maintain constant at q the amount of the public service benefiting each beneficiary for a given percentage change in N. The congestion parameter is then introduced into a utility maximizing demand model. Continuing to follow the Bergstrom and Goodman specification, an individual’s utility (u,;) is a function of private goods consumed (X;) and q; that is, u; (X;, q). The person’s budget constraint is: Xe Oar (2) Local Public Goods: Shoup Revisited 185 where the price of X; is one, r; is the tax share of the person, and p is the unit cost of Q. Recognizing that Q = N« q, the budget constraint is: cy DENG Gi XG; (2.1) where r; p Nis the price per q-unit to the individual. Assuming demand is Cobb-Douglas with constant price and income elasticities, the demand for q is: Gis (Gp N2)® 2 (3) and the demand for Q is: Or (cip) RANGED Ye 1" (4) In application the form of the estimated equation is: InE=oInN+6lnr+6lnp+olnY +28; X; (5) where E is municipal expenditures (pQ) on a particular function, equals a (6 + 1), and X, denote other socioeconomic characteristics influencing expenditures and conditioning utility. This approach as- sumes that public decisions reflect the median voter’s preferences so that rand Y represent the median voter’s tax share and income. The value of the publicness parameter a can be derived from the estimated values of and 6. The results from the BG-BD demand studies yield an « value close to unity. Since the BD-BG papers, many investigators have incorporated the publicness specification into their own models. The prevailing empirical result is that a is close to one, indicating that publicly provided services are subject to congestion. The rivalness parameter values derived for fire and police services specifically tend to be somewhat more diverse (see below). Specification of the Congestion Term The BD-BG specification of the congestion term is a simple one, q = Q/N«. An implication of this specification is that for a > o congestion is decreasing at the margin. That is, dqg/dN < 0 and d2q/dN2 > o. Further- more, the congestion elasticity, a = (dq/dN)/(q/N), is a constant. Such a form may be unduly restrictive. Indeed, one can readily think of public services such as roads and swimming pools for which congestion could 186 Lorraine Eden and Melville L. McMillan be increasing at the margin as population or the number of users ex- pands. Concern for the restrictiveness of the publicness function N« has led various investigators to consider alternative forms. Edwards (1986) considers several models allowing for variable congestion elasticities and for increasing as well as decreasing marginal congestion. Where permit- ted, decreasing marginal congestion performed better than other specifi- cations. Hayes and Slottje (1987) test two alternate congestion relation- ships—(N« +#!2.N), which reduces to the original N« when ¢ = o, and e® N, which allows congestion elasticities to vary with population— against the simple N¢ specification. The former is found superior to the exponential e® N form, but the BG-BD specification dominates both as statistical tests cannot reject the hypothesis that @ = o. These results suggest that the original BD-BG congestion parameter is quite robust. Instead of the usual single-equation approach to public expenditures, Hayes (1985, 1986) and Hayes and Slottje (1987) use a simultaneous system of equations to study the demand for fire protection, police services, and other local public goods. Hayes (1986) allows a to be influenced by demographic variables—metropolitan (versus nonmetro- politan) and growing (versus nongrowing) cities. These demographic variables do not significantly influence the congestion effects. Output and the Publicness Parameter Several researchers have suggested that the observed nonpublicness of some government services may be partly a measurement anomaly. Most of the forementioned studies circumvent (albeit neatly) the problem of actually measuring the output of the public service itself. As Bradford, Malt, and Oates (1969) and Shoup have noted, estimates of rivalness are likely to be inaccurate when the underlying output of LPGs is incorrectly measured. In this section we look at the impacts of output definition, size and number of facilities, and varying distribution of benefit levels on congestion measures. Brueckner (1981) examines congestion in local fire protection services by estimating the reduction in expected fire losses resulting from expen- ditures on fire suppression capability. The expected reduction in losses is based on fire insurance premiums which vary among communities de- pending upon the Insurance Services Office rating of the quality of fire protection in each. He finds congestion elasticities that range from Local Public Goods: Shoup Revisited —.1143 to —.2379. These estimates indicate a high degree of publicness for fire protection. Craig (1987) models police protection as a process by which police labor inputs produce a clearance rate and clearances deter crime and produce safety. The rivalness specification Craig selects is (u — N)¢ which permits increasing rather than decreasing marginal congestion as population increases. If N = », congestion stalls production and nothing is produced; for smaller N, output is realized and increases as N falls. The congestion elasticity, [—o N/(j — N)], varies with population, and, for o < o < 1, increasing marginal congestion costs occur and conges- tion elasticity becomes, absolutely, larger. Using data for Baltimore police beats, Craig finds that the total elasticity of the final output (safety) to increases in N is small (—o.016) when N is low; however, as population increases, the congestion elasticity reaches —1.0, implying privateness. Craig also estimates his model with the conventional BD-BG specification. The estimated elasticity (—0.047) implies a high degree of publicness, but the implications of the two models for larger beats differ somewhat. No test of the alternative specifications is provided. McGreer (1989) compares the performance of the BD-BG with Craig’s and Edwards’s exponential specification of the congestion term using Australian local government data for recreation and culture, roads, and total municipal expenditures. A test of alternative specifications reveals that the BD-BG specification dominated the Craig version in each case. The BD-BG model also appears superior to the exponential form for the first and last categories while neither dominate for road maintenance. Other economists have been concerned that the data studied and the approach of the analysis tended to mask the underlying publicness of LpGs. Many services, for example, fire and police stations and cultural and recreational facilities, have limited service areas and/or popula- tions. In larger communities a single facility may be inadequate to meet local demands. If within a single community multiple facilities are re- quired, replication of units will make numbers and cost correspond closely to population (unless there are substantial economies in over- head) even though the service at the individual facility level has public- ness characteristics. To test this idea McMillan, Wilson, and Arthur (1981) estimate congestion parameters, using the BG model, for small and large municipalities (less and more than 10,000 population). They find that the congestion elasticities for general services, recreation and culture, and fire protection in communities under 10,000 persons are 187 188 Lorraine Eden and Melville L. McMillan typically about one-half as large as those for the over 10,000 group which (with the exception of that for recreation and culture) ap- proaches one. Communities with larger populations may be able to provide a greater range of services than smaller jurisdictions. This is what Oates (1988a) refers to as the “zoo effect,” since only larger municipalities can usually afford, for example, zoos. If expenditures for services expand with population, because the range of services expands, estimates of the congestion parameter not accounting for this change will be upward biased. This situation may partly explain the observed nonpublicness of local public goods. Some evidence of Oates’s contended zoo effect is provided by McMillan (1989) in a study of fire protection in Ontario municipalities. The quality of fire protection service (Oates’s range of service) is measured by the fire protection grade assigned by the Fire Underwriters’ Survey. The grade of service tends to improve with popu- lation size. McMillan finds that inclusion of fire protection grade in a standard demand equation reduces the congestion parameter substan- tially. If the personal level of service can vary among residents, the distribu- tion of q can affect the publicness parameter. Gramlich and Rubinfeld (1982) introduce this possibility into their demand model by assuming that the level of q residents realize is conditioned by a relative income term. The distribution of q favors the rich or the poor as the income distribution parameter is greater or less than zero. Since values for this parameter estimated from household data exceed zero, this indicates a (perceived) pro-rich distribution of local schooling. The associated esti- mates of the publicness parameter from jurisdictional and individual household data yield values of about one.® Summary of Evidence on the Congestion Parameter Columns 5 and 8 in table 1 report the publicness parameter values for police and fire expenditures, respectively, for fifteen different studies published during the 1972—87 period. The statistically significant esti- mates range from 1.02 to 1.63 for police services and from 0.39 to 4.27 for fire services. Shoup’s placement of fire and police services as GCGs is neither confirmed nor denied by this evidence; however, the estimates do suggest that the marginal cost of provision is nonzero. Local Public Goods: Shoup Revisited 189 Table 1: Estimates of Price, Income, and Congestion Elasticities? Price Police Protection Fire Protection Term ; ; 5 5 Study price income a price income a Bergstrom/Goodman (1973) tax (O25) ae Oal 1.07 NA NA NA Borcherding/Deacon (1972) wage (0.97) 0.81 1.02 — 0.88 1.01 Brueckner (1981) wage NA NA NA 0.03 — (0.24)f Deacon (1978) index (0.76) — — (0.99) a — Ehrenberg (1973) low wage (0.01) NA NA (0.23) NA NA high (0.35) (0.31) Gonzalez/Mehay (1985) none NA 0.23 NA NA 0.69 NA Gramlich/Rubinfeld (1982) wage (0.06) NA NA (0.06)> 3 — NA Hayes (1985) wage (0.71) — 1.50 — — 0.95 Hayes/Slottje (1987) wage (1.23) 0.40 — (1.03) ae 4.27 McMillan/Wilson/Arthur tax NA NA NA (0162)\0 = 0.39 (1981) to 0.69¢ Pack/Pack (1978) tax (0.19) 0.53 1.63 (0.12)4 0.474 1.324 (0.21)© (0.82)¢ 1.58¢ Perkins (1977) wage (0.73) 0.21 NA (0.75) 0.14 NA Pommerehne/Frey (1976) tax NA NA NA (0.33) NA 0.56 Santerre (1985) wage (0.45) 0.49 11335) (0.48) a 1.66 Vehorn (1979) tax NA NA NA (0.15) — 1.0 to (0.88) to 2.0 Range of values? low (0.01) 0.21 1.02 (0.06) (0.82) 0.39 high (25) ee OnSIL 1.63 (1.03) 0.88 4.27 a. NA indicates that no elasticity was estimated; a dash (—) that the elasticity did not differ significantly from zero at the 10 percent level or better. Estimated for fire plus police; the t-value is 1.8. For municipalities with less than 10,000 population. For towns of 5,000 to 50,000 population. Elasticity of congestion of fire protection. b. CG d. For towns of 1,000 to 5,000 population. ce; i g. When significantly different from zero. Defining and Measuring the Output of Local Public Goods The Shoup Approach to Output Measurement Shoup (1969b:78) establishes a clear distinction between units of input and units of output of a GcG. He argues that physical units should be measured in three dimensions: time, number of consumers served, and 190 Lorraine Eden and Melville L. McMillan level or intensity of the service. A correct measure of output is in house- hold-weeks of units of service. A creative service, a GCG agreeable to household consumption and/or which creates intermediate or final products for firms, can occasionally be measured in this manner (e.g., area size of a park). Measuring output of a GCG is difficult at the best of times; when the service is a preventive one, the difficulties are worse. A preventive service is “one that is not in itself agreeable to consume or receive, but is valued nonetheless because it prevents, or is deemed to prevent, something more disagreeable from happening” (1969b:78). Shoup argues that most GCGs are preventive services, for example, crime reduction, limiting fire dam- age, and restricting the spread of disease. By its very nature quantity of a preventive service cannot be measured by the amount of input; the only way to measure the level of a preventive service is indirectly by measur- ing what it fails to prevent. The analysis must be in decrements of service from an unknown total. Shoup suggests that inputs can be distinguished from outputs as follows: if the service is nonmarketable because discrim- ination at the level of the household is not feasible, anything that is marketable is an input, not an output. Fire protection services consist basically of limiting damage once a fire has started together with education in and inspection for fire prevention. Shoup argues that the output of fire protection, as a preventive service, must be measured indirectly by the mean and variance of the expected damage from fire over a stated period in a stated area. Output can be measured incrementally by changing the level of fire protection inputs and observing the subsequent fall in fire losses and/or insurance pre- miums. The physical output of police protection, another preventive service, is best measured by the number of undesirable (weighted) events that are prevented by police inputs (see Shoup 1964, 1969b, 1974, 1988a). Since the end product is the reduction of exposure of the house- hold to crime, output should be measured as the reduction in the proba- bility of loss of a certain dollar amount of property from crime over a certain period due to a rise in police inputs. Recent Developments: Output Measurement Shoup clearly identifies a major issue affecting all theoretical and empiri- cal work on local public goods, that is, how to define a physical unit of a publicly provided service. In the absence of a quantifiable measure of Local Public Goods: Shoup Revisited 191 output, demand and cost functions for the LPG, benefit distribution and incidence measures, and elasticities of demand and supply cannot be accurately calculated. Much of the early work on local public goods suffers from this deficiency. (See the review in Beaton 1983.) There are, however, a few papers which distinguish output from expenditure. Shoup’s approach is echoed in Bradford, Malt, and Oates’s (1969) distinction between C- and D-output, which uses a two-stage production function. In the first stage inputs of labor, capital, and materials are used to produce a vector of directly produced output (D-output), in the form D = f (I), where I is inputs and D is direct output. In the second stage D-output, in conjunction with environmental variables, E, determines the output that affects the consumers’ utility functions, C-output. That is, C = g (D, E) = g [f(I), E]. For example, in the case of police services, I could include police officers and cars, D the number of blocks under surveillance, E the proportions of residential to business and urban to rural properties, and C the probability that a resident would not be subject to crime. A few recent papers have been based on the Shoup—Bradford, Malt, and Oates (BMO) approach. Schwab and Zampelli (1987) use this ap- proach to argue that zoning regulations and fiscal migration between communities can affect the production of LpGs through changing popu- lation characteristics. They note that the tax price variable common to most median voter analysis is the price of D-output, not C-output, so that all empirical estimates of demand determinants are biased. To test their hypothesis they include income in the production function to act as a proxy for community characteristics. They find that the usual income elasticities underestimate the true values once income is allowed to affect both the demand and supply of LPGs. Five recent papers have been based on the Shoup—BMo approach to preventive services, four on police and one on fire expenditures. Scic- luna, Foot, and Bird (1982) break the BMo distinction between D- and C-output into an intermediate category, D’-output. They argue that C-output can differ from D-output for two reasons, environmental and productivity/efficiency factors. The intermediate category, D’-output, is defined as operational or D-output adjusted for quality differences, but excluding environmental factors; D’-output is thus a measure of output proper. C-output is then D’-output adjusted for environmental factors. The paper uses these distinctions to measure police productivity as distinct from environmental influences. Craig (1987), cited above, models police protection as a two-equation 192 Lorraine Eden and Melville L. McMillan process by which police labor inputs produce a clearance rate (D-output) and clearances produce safety (C-output) as reflected in a lower crime rate. Both production processes are subject to congestion. Gyimah- Brempong (1987) estimates a multiproduct translog cost function where arrests for each type of crime are the primary output measures, inputs consist of two types of labor (police and civilian labor) and a measure of capital, and cost shares are taken as the value of property stolen relative to total cost. The third paper by Levitt and Joyce (1987) regards arrests as an intermediate output, where final output is crime deterred. The sole fire paper, Brueckner (1981) cited earlier, estimates the reduction in expected fire losses in a community resulting from expenditures on fire suppression capability. Cost Functions for Local Public Goods The Shoup Approach to LPG Cost Functions Table 1 in Shoup (1969b:143) summarizes his hypothesized effects of level of service, population, and area size on the cost functions of pro- ducing and distributing group-consumption goods. Category 1 and 2 services (pure public goods) face the standard rising total cost and U-shaped marginal and average cost curves as the level of service in- creases, holding population and area constant. Increases in population, for a given level of output and area, leave output unchanged, while marginal cost falls because the marginal cost of provision to an addi- tional consumer is zero for a pure public good. Increases in area size, holding population and level of output fixed, vary by function, but generally leave the elasticities unaffected. However, as population increases, rivalness may increase as public services become congested. The cost of maintaining service quality may increase and/or service quality may diminish; for example, response time increases as new homes are located further from the fire station, or costs rise as increases in density require more sophisticated and more expensive manpower and equipment. Shoup argues that police and streets face rising total, average, and marginal costs over all three vari- ables (level, population, area). Fire protection faces rising costs as the level of service increases, but U-shaped costs for increases in population and area size. In general in category 3, an increase in population, with a Local Public Goods: Shoup Revisited 193 fixed amount of input and fixed area, causes a decline in the level of service. Migration between jurisdictions can therefore cause fiscal exter- nalities by affecting per capita cost. An increase in area size, however, with population and service level fixed, causes costs to increase. The links between population, rivalness and growth of government share are explored in Shoup (1976, 1984). He concludes that little work has been done in this area. Eden (1984) extends Shoup’s analysis to examine the impacts of publicness on government share measured in current and constant dollar terms and in an open economy. See Bird (chap. 13, this volume) for additional extensions linking tax reform and government growth. Recent Development: LPG Cost Functions Cost of public service studies would often have benefited from Shoup’s insights. Many of the analyses examine scale effects in the sense of per capita cost rather than as expenditure per unit of output (E/Q). It is the latter which is of interest, but that approach is complicated by problems of public output measurement. Constant returns to scale is often as- sumed in recent studies because it is necessary in the BD-BG framework in order to identify the structural parameters (Inman 1979:295—96). With- out constant returns only the product, s a, of the returns to scale param- eter (s) and the publicness parameter (a) is available from the estimates. Evidence of returns to scale in local public services is mixed. Fox (1980) reviews studies of size economies in fire and police protec- tion plus other services. Output measures used in those studies, if not simply population, usually include population with some effort to con- trol for service quality. With respect to fire protection Fox concludes that there are some economies of scale for communities up to 10,000 persons but that they are very limited beyond that size. For police protection he finds that unit cost savings seem to exist with size but that the cost savings are offset by the addition of more services. Rider (1979), how- ever, in a study of fire protection in New York that regards scale as a multidimensional factor involving population, area, and workload, ob- tains a U-shaped average cost curve. Brueckner (1981) finds some evi- dence of increasing returns to scale in his fire data for one hundred cities. The results of police protection studies are mixed. Walzer (1972) concludes that when an index of police service is used, a U-shaped cost 194 Lorraine Eden and Melville L. McMillan function is observed, but when population is used to reflect scale, no economies are revealed. Scicluna, Foot, and Bird (1982:269), in their study of police productivity using the concept of D’-output, find evi- dence of substantial amalgamating of police forces into regional groups. Gyimah-Brempong (1987), using a multiproduct translog cost function where arrests are the primary output measures, rejects the Cobb-Doug- las production function specification. For the average police department decreasing returns are observed; the scale of decreasing returns rises as population increases. Also, he finds no evidence of cost advantages to providing a variety of services; that is, there appear to be no economies of scope in public production. Levitt and Joyce (1987) and Carr-Hill and Stern (1973) also observe diseconomies of scale in public services. Eden and Millar (1990) find that smaller municipalities in north Ontario have, ceteris paribus, higher exogenous costs (that is, higher fiscal needs) and lower fiscal capacities than their larger, southern Ontario cousins, pro- viding some evidence of size economies. We conclude that, while progress is being made in the study of econo- mies of scale in local public services, there is as yet no clear evidence as to whether or not constant returns prevail. One consequence is that, with- out this evidence, caution is advised when interpreting estimates of publicness. Distribution and Incidence of the Benefits from LPGs The Shoup Approach to Benefit Incidence and Distribution Shoup’s work on the benefits from LPGs focuses on three areas: measure- ment of the benefits, intragroup discrimination, and the shifting and incidence of LpGs. Group-consumption goods provide valuable benefits to their consumers. Shoup argues that their value can be measured by the sum of the amount by which households and firms reduce purchases of private competing services, plus savings in insurance, plus the increment households and firms are willing to pay in higher building and land prices to enjoy the Gcc (Shoup 1969b:116).? If the service is a preventive one that is not prized for its own sake (e.g., fire protection), Shoup suggests that its benefit can be measured by the saving of market ex- penses and the reduced disutility from fear and inconvenience. The benefits from GcGs are not necessarily uniform to all consumers Local Public Goods: Shoup Revisited within a jurisdiction. Shoup (1969b) argues that equality of inputs across the jurisdiction is no guarantee of horizontal equity, equal treat- ment of equals, because benefits from LPGs depend not only on inputs, but also on the environment in which these inputs are used. Horizontal inequity in the distribution of benefits is usually due to different costs necessary to attain a certain level of service across various subgroups (e.g., urban/rural, residential/business, rich/poor). Such intragroup dis- crimination may be involuntary or unrecognized since it is due to so- cioeconomic variables that affect the technology of public service provi- sion. Shoup (1988a) hypothesizes that economists have ignored the hori- zontal equity issue for several reasons: preoccupation with the pure public good concept; the difficulty of measuring output of GCGs; preoc- cupation with vertical, as opposed to horizontal, equity; lack of avail- ability of micro data; lack of a connection between public services and money valuation; little demand for such studies by recipients; preference of social scientists for measuring public inputs, compared to outputs; and stress on equity among input suppliers rather than output con- sumers. He calls for more work on the horizontal equity impacts of local public goods, arguing that political scientists and lawyers have contrib- uted more to this issue than economists and that economists can still provide a valuable perspective. Governments distributing services free of charge to households and firms may deliberately discriminate in the level of service provided to subgroups within the jurisdiction. If such intragroup discrimination is feasible, the jurisdiction must consider how to optimally discriminate among the subunits. Different goals will lead to different methods of intragroup differentiation. For example, the government may choose from several goals for fire protection: maximum reduction in the number of fires, maximum protection from becoming a victim of fire, equal protection per dollar’s worth of property. Similarly, there are several goals for crime prevention: equal protection per person, maximum pre- vention of crime, equal work load per police officer (see Shoup, 1964, 1969b, 1974, 1988a). A third issue related to the benefits from GcGs is the shifting and incidence of such benefits. A free government good exerts a positive income effect on its recipient; this is the impact incidence of the GcG (Shoup 1969b:86). If there are locational differences in benefit levels, market forces will cause the final incidence to differ from its original 195 196 Lorraine Eden and Melville L. McMillan impact (e.g., higher rents for households living near city amenities). Shoup uses the term relinquishment rather than shifting, to stress the loss of GcG benefit as opposed to the removal of a tax burden (1969b:88). Such relinquishment is most likely to occur when government services are locational; that is, enjoyment of the GcG depends on geographic location of the household or firm relative to the service input. The shifting of benefits is manifested by a rise in the price of marketable goods bought by the Gcc consumer and/or a fall in the recipient’s factor rewards. Some examples are given in Shoup (1969b:92—93; 1988a:9). Shoup (1988a) notes that the concept of relinquishing benefits is un- familiar to most public finance scholars and is not discussed in most textbooks, whereas expenditure incidence has been extensively ana- lyzed. Recent Developments: Benefit Shifting and Incidence Measuring the Benefits from Local Public Goods The conventional approach to measuring the benefits from the local public sector is to focus on the vertical distribution of income (Dodge 1975; Gillespie 1976; Dahlby 1985). Using reasonable assumptions, these studies divide public outputs into those which are specific to identifiable groups and those which are indivisible. Allocation of the benefits of indivisible services is done according to income, property, per capita, and other bases. Concerns about this methodology stimulated Aaron and McGuire (1970) to propose an alternative, distinguishing between the benefits of public output imputed from the marginal value of the public good and the impact of redistributive transfers resulting from its finance. Though not without fault, this approach has been employed in many studies. For example, Dean (1980) integrates the Aaron and McGuire method with the BD-BG approach to estimate the benefit incidence of Canadian municipal services. See also studies by Martinez-Vazquez (1982), Weicher (1971), Behrman and Craig (1987), and Hewitt (1987).1° Another way to estimate public output benefits is to identify house- holds’ willingness to pay from differential property values attributable to variations in local public services. Although a common approach for estimating the benefits and costs of local public goods and bads, this Local Public Goods: Shoup Revisited 197 technique has rarely been used to study the distributional implications of the local public sector. We are aware of one application: Chaudry-Shah (forthcoming) uses this approach to measure the capitalization effects of public service and residential property tax variations across twenty- seven subcommunities in Edmonton. He then determines annual resi- dential property tax burdens and the annual value of local public sector benefits and relates these to community income levels. He finds that the amount of the annual tax burden increases with income, whereas the annual value of local public service benefits first diminishes and then increases with income. Intragroup Discrimination Twenty-five years ago, Shoup (1964) addressed the problem of how a free government service should be distributed. He refers to services such as fire protection and roadways and deals explicitly with police protec- tion. The question is how the supplier with a specific budget should allocate service across service units that differ in terms of incomes, property values, race, age, etc. In the context of his police protection example Shoup discusses three possible criteria: equal crime rates in each area, equal marginal cost of crime prevention, and tangency of the production and social welfare trade-offs among service areas, with the latter as the preferred alternative. He subsequently notes equal work load per patrolman as the criterion actually utilized in New York City. Until recently, economists have devoted relatively little attention to intragroup discrimination.!1 However, there are a few papers applied to fire and police services. Rider (1979) examines the distribution of fire companies in New York City. He considers as potential objectives: equal travel times to fires, minimum average city-wide travel time, and equal work load plus fire hazard (as proxied by average assessed value per building). Within residential areas Rider finds that work load per unit area and average assessed value are good predictors. Kennett (1982) looks at distribution of police benefits in New York City. He predicts the number of crimes which would result from applying each of three opera- tional allocation criteria: (i) output equalitarian (equal crime rates per precinct), (ii) input equalitarian (equal input per capita), and (iii) max- imum output/minimum offenses (equal marginal crime per patrolman). The actual distribution of police resources does not appear to conform to that implied by any of these standards. 198 Lorraine Eden and Melville L. McMillan Understanding of the distribution of police services has advanced recently as a result of the work of Behrman and Craig (1987) and Craig (1987) whose models incorporate the interdependence of the public good production and allocation decisions. This interdependence is im- portant because the social value of the product the inputs generate depends upon where they are assigned and whom they benefit. As alter- native objectives, Behrman and Craig consider service maximization and equal service as two extreme criteria. They also are able to assess whether the supplier’s degree of concern for public outputs varies with neighborhood characteristics; for example, whether poor neighbor- hoods are favored over the rich. They find evidence of a significant degree of inequality aversion. Police are not allocated to maximize safety (crime reduction), but neither are they allocated to achieve equal services in each area—that is, there is an equity-efficiency tradeoff. In addition the provider’s concern appears to vary with neighborhood characteris- tics. Price and Income Elasticities of Local Public Goods Although this paper is motivated by other issues, it would be an in- complete commentary on demand for local public good studies without some mention of estimated price and income elasticities. Appendix I gives a brief discussion of the elasticities of police and fire protection, a summary of which is provided in columns 3, 4, 6, and 7 of table r. Conclusions Carl Shoup’s writings on local public goods are based on the taxonomy of group-consumption and collective-consumption goods, a framework quite different from the better-known Musgrave-Samuelson approach. While this methodology has not been more widely used, Shoup’s work on public goods that are spatially limited is well known. He emphasizes four aspects of LpGs: the definition and measurement of output, their rivalness and nonexcludability characteristics, the cost functions associ- ated with service level, population, and service area, and the distribution and incidence of benefits. In this paper we have outlined Shoup’s views on each of these issues Local Public Goods: Shoup Revisited 199 and then surveyed the subsequent literature. In each of the four areas his work has been at the lead in identifying issues and contributing to the understanding of these questions and, as such, has been at the forefront of the debate. In the case of output definition and measurement it is only recently that empirical studies have begun to take account of his insights. The literature on rivalness, while enormous, still suffers from a reliance on total and per capita expenditures to estimate publicness parameters. As Shoup warned in 1969, such reliance can lead to faulty estimates (1969b). Moreover, Shoup’s recognition in his concept of GcGs that nonmarket provision and rivalness are not inconsistent avoids some of the confusion and puzzlement found in some interpretations of conges- tion parameters implying rivalness. Shoup’s recognition that rival goods can be legitimately provided by government (as part of GcG) is itself a contribution. Little work has been done on Shoup’s definition of nonex- cludability, partly because of the difficulty of making the necessary cost comparisons. In terms of cost functions there is a long history of cost function estimation, but the issues Shoup identified are still being wres- tled with today. Lastly, the horizontal distribution and incidence of the benefits of LpGs have been little studied, but are becoming increasingly recognized. Also, the study of LpGs has gone in directions Shoup did not pursue (cf. note 1). We conclude that Shoup offers economists a broad menu of research topics in local public goods, some of which have hardly been sampled and other areas which, despite active investigation, still have major problems to solve. For those of us working in the area of LPGs, this news is both heartening and disheartening: disheartening because it shows how little we have progressed in particular areas; heartening because there is a good deal of scope for independent contributions to an exciting area of research. Appendix |: Price and Income Elasticities for Local Public Goods This appendix provides a brief review of the literature on the price and income elasticities of local public goods. Calculation of price elasticities for public outputs is complicated by the appropriate definition of the price term. With the exception of Deacon (1978) who calculates a price index based on all inputs, the studies noted here are all macro models which use either a median voter tax share or a public sector wage level as the price variable (see column 2z, table 1).!? 200 Lorraine Eden and Melville L. McMillan The price elasticities reported in table 1 are all negative, and though typically inelastic, the range is broad. For police services the estimated price elasticities range from —o.01 to —1.23, with only one value exceeding unity. For fire protection the range of the elasticities is only slightly narrower and the distribu- tion is also rather uniform. The estimated income elasticities for police services are less than one and fit into a somewhat narrower band, about o.2 to 0.8. For fire protection most of the estimated income elasticities are not significantly different from zero. To this point there has been no consideration of developments in the literature suggesting that elasticity estimates from the standard median voter model using cross-sectional data may be biased. However, the literature has recently focused on Tiebout bias, a potential bias that can occur because people select their jurisdiction of residence according to their public good preferences in the spirit of the Tiebout (1956) model. This problem was first identified by Goldstein and Pauly (1981). Their model predicts that demand estimated on observed data with fiscal migration overestimates the true income elasticity when account is not taken of other factors affecting public good preferences. Tiebout bias is a poten- tially serious problem if households’ locations are influenced by local public output. Gramlich and Rubinfeld (1982) provide strong evidence of such sorting. Recent papers by Holtz-Eakin (1986), Rubinfeld, Shapiro, and Roberts (1987), and Bergstrom, Roberts, Rubinfeld, and Shapiro (1988) allow for Tiebout bias in their models. While lower elasticities sometimes result, neither income nor price elasticities are uniformly reduced. While Tiebout bias implies that the conventionally estimated elasticities are too large, Schwab and Zampelli (1987) argue that, because income may affect both supply and demand, the usual income elasticities underestimate true values. The reasoning for this result is that income is related to household or community characteristics that favorably affect the price of certain services; schooling and crime protection are good examples. In this situation the response of expenditure to income depends not only on the income elasticity of demand, but also on the response to the price change due to the income change. Schwab and Zampelli find that the estimated income elasticity of demand increases from 0.59 to 1.20 when income is allowed to be a determinant of price. Wildasin (forthcoming) also argues that a second bias, in addition to Tiebout bias, exists when public goods are financed with distortionary taxation. He finds that distortionary taxes raise the effective tax price of local public goods to the median voter. The distortion biases the price elasticity toward unity and the income elasticity away from unity when log-linear demand functions for public goods are estimated, with the error in the price elasticity being much larger than in the income elasticity. Given these potential sources of bias, estimates of price and income elas- ticities from public sector demand models should be treated carefully. Local Public Goods: Shoup Revisited 201 Notes We would like to thank Malcolm Gillis, Paul Hobson, Wade Locke, Carl Shoup, and Enid Slack for comments. 1. For example, there is little mention in Shoup’s work on local public goods of the optimal size of the government jurisdiction, the distinction between production and provision, tax exporting, or interjurisdictional migration in response to fiscal differentials, all topics of current research. 2. Carl Shoup has pointed out to us that neither Samuelson nor Musgrave explicitly used the box diagram to illustrate the characteristics of social goods, although Musgrave’s table clearly leads to the diagram. Both of us have used the box for years in class lectures and would appreciate knowing who was the originator of this technique. 3. Our discussion is based on Shoup (1969b, Chaps. 4, 5, and 21; 1964; 1974; 1988a). 4. There may be some disagreement as to the public services Shoup includes in category 5, since some of them do not appear to meet the zero marginal cost criterion. The Walsh paper in this volume is an analysis of nonrival but exclud- able services also. Walsh includes services such as cable television, the output of composers, toll bridges, and entertainment. Since our paper emphasizes GCGs, the Walsh paper can be considered a companion piece. 5. The rivalness/nonrivalness of LPGs is a subject of ongoing controversy. Empirical estimates of publicness can provide insight into the interaction be- tween the rivalness and excludability characteristics. Sometimes, however, the empirical estimates conflict with intuitive positioning (see section entitled “Re- cent Developments” below). 6. We would argue that category 2 services are less than pure public goods due to the discrimination possibility which, ina sense, is a form of exclusion; that is, those services in category 1 are at point a and those in 2 are a bit to the right of point a. 7. Some work has been done on the contracting out of public services to private producers. The tendency toward privatization of LPGs is a recent phe- nomenon partly encouraged by rising costs of urban public services. 8. Blecha (1987), however, criticizes the Gramlich and Rubinfeld conclu- sions about publicness. Once the distributive share term is introduced, she notes that a must vary with population size. Thus the effects of the publicness and distributive share parameters cannot be separated. She argues that this bias is large enough to nullify their results. 9. We note that there could be double-counting here as land prices might reflect lower insurance or private service costs; that is, these have been cap- italized into the property. 202 Lorraine Eden and Melville L. McMillan 10. One should be cautious about imputing individual benefits from informa- tion of income elasticities derived from interjurisdictional data. Results in Gram- lich and Rubinfeld (1982) suggest that elasticities of residents within a commu- nity may be quite low relative to the interjurisdictional estimate. They also find that the wealthy within a community perceive themselves as getting greater benefits from public outputs. 11. For reflections from other disciplines, see Benson and Lund (1969), Bloch (1974), Lineberry (1977, 1978), and Rich (1982). 12. For discussions of the price term, see Bahl, Johnson, and Wasylenko (1980), Inman (1979), Rubinfeld (1987), and Bergstrom, Rubinfeld, and Shapiro (1982). ea Ga RF UWA S.A Public Goods Provision with Price Exclusion: Market Behavior and Market Performance Television and radio broadcasts accessible only through cable or coin-in-the-slot devices; meteorological information, or information on credit ratings, available only on payment of a fee; the outputs of re- searchers, inventors, composers, and writers, usable only on payment of a royalty as a result of enforcement of patent or copyright laws; perform- ances of plays, operas, or symphony concerts for which admission is restricted to (single-performance or season) ticket holders; restricted access roads and bridges usable only on payment of a toll; attendance at football matches, visits to fairgrounds or parks, and trips by aircraft, ships, trains, and buses and many other entertainment, recreation, infor- mation, and transport services are all examples of a distinctive class of commodities. They exhibit (in varying degrees) both the “Samuelsonian jointness” characteristic associated with pure public goods, and the “price-excludability” characteristic associated with pure private goods: accordingly, they can be termed price-excludable public goods.! That these goods (or services) constitute an important class of com- modities is almost self-evident: their consumption bulks increasingly large in developed economies and few have escaped intervention, in some form, by the public sector. Yet analysis of market provision of excludable public goods—a prerequisite for the development of a case for public intervention and an important (positive) exercise in its own right—is in its infancy. Indeed, where such goods have been examined with a view to de- veloping public policies toward them, attention has frequently focused not on what is arguably their central analytic feature—their jointness characteristic—but rather on what may be more peripheral features of 204 Cliff Walsh particular cases, for example, the “cultural externalities” associated with the performing arts or the “natural monopoly” elements that also arise in many of the contexts cited. Such features may not be unimportant; they may be policy relevant; and in those cases where the degree of Samuelsonian jointness is sufficiently small that the good or service is likely to be highly congested, treatment of it as if it were nonjoint may not always lead to egregious error. Nonetheless, the jointness charac- teristic is relevant; in some cases it is the dominant characteristic, and it has important and not well understood implications for the nature of market equilibrium and for public policy. While the published literature on market provision of excludable public goods is as yet sparse, the results have proved to be profoundly different from those of analogous models of market provision of private goods. In particular: (i) Earl Thompson (1968), in the first formal analysis of excludable public goods provision, considered a perfectly competitive market but with the extreme assumption that all producers (and consumers) are perfectly informed about all consumers’ preferences and purchases. He concluded that, despite the competitive nature of the market, consumers could and would be faced with interpersonally discriminating per-unit prices, so that market output would be fully consumed by all individuals. Output, however, would exceed that which is efficient, and competitive outcomes sometimes could be as inefficient as having nothing at all produced. (ii) William Oakland (1974) showed that, with the more conventional as- sumption that consumers and producers have no more information than the market process would normally generate for them, competitive equi- librium is associated with consumers facing an (identical) step function of prices, with prices inversely related to the number consuming each particular output unit. As a result, some consumers will be inefficiently excluded from consumption of some units of output, and (in contrast to Thompson’s result) the market output will, in general, be inefficiently low. (iii) Thompson, in the course of his competitive analysis, demonstrated that perfectly discriminating monopoly provision of excludable public goods would result in fully optimal output (and consumption). A more recent series of papers on monopoly provision when information necessary to permit interpersonally discriminatory pricing is not available to pro- ducers (Brennan and Walsh (1979, 1981), Brito and Oakland (1980), and Burns and Walsh (1981)) have shown that monopoly provision will generally involve some (high-demand) individuals being rationed by output rather than by price and that over some output ranges a positive price-output relationship may prevail (i.e., an upward shift in MC (mar- Public Goods with Price Exclusion 205 ginal costs) results in a lower profit-maximizing price, as well as lower output). Not surprisingly, monopoly performs worse than competition in a welfare sense.” A formal survey of the models and methods employed in generating these results is not intended or attempted in this paper. Rather, its central purpose is to explain and interpret the unfamiliar aspects of market behavior that emerge with price excludable public goods, as a contribu- tion to a better understanding of their nature and significance. Section I attempts to provide a clear characterization of the nature of excludable public goods. In Section II some implications of the jointness characteris- tic of excludable public goods for pricing strategies and for market structure are examined, while in Section III the implications of jointness for decision making and market behavior are explained and compared to the more familiar world of private goods. Section IV turns to consider the question of market performance in the provision of price excludable public goods, with and without exclusion. Section V offers brief conclud- ing comments. The Nature of Excludable Public Goods John Head’s influential surveys (1962, 1968) of the early literature on public goods provide a valuable characterization of pure public goods as a (double) polar case: they exhibit both Samuelsonian jointness in supply (or nonrivalness in consumption) and impossibility of exclusion. The ‘Sointness” characteristic (following Samuelson 1954, 1955) implies that a unit produced for any one individual could be consumed fully and equally by all others at no extra resource cost. The “impossibility of exclusion” characteristic implies that a unit produced for any one indi- vidual will be accessible for consumption by all others, whether or not they offer payment. Standard examples of goods which most nearly meet these requirements include: defense; law and order; charitable giving motivated by altruism; preventative health measures (such as vaccina- tion against communicable disease); uncongested, unlimited access roads; and environmental protection measures. Pure private goods also represent a double polar case but with totally opposite characteristics: they exhibit zero jointness (output units are strictly rival in consumption) and are subject to complete (“costless”) excludability. Basic consumption goods—food and drink in particu- lar—provide the purest examples. 206 Cliff Walsh The characteristic features of public goods and private goods can arise in varying combinations. For example, there exist goods which are “private,” but for which impossibility of exclusion prevails: common property resources (as in the fisheries case) where private property rights cannot be defined, or are not enforced or enforceable, provide a case of such nonexcludable private goods. By the same token some goods which are “public” nonetheless can be made subject to price-exclusion in some form, as clearly exemplified in the case of cable TV or weather informa- tion accessible only on payment of a fee. It is this latter context which is the focus of the literature on exclud- able public goods, and it is clear that these goods can be usefully viewed as another polar case, exhibiting the complete jointness characteristic of a pure public good but also the complete (and in the limit, costless) excludability characteristic of a pure private good. Although pure cases may be rare, the examples cited earlier indicate that the list of com- modities which are appropriately analyzed as public (though excludable) goods, rather than as private goods, is significant. Figure 1 serves to place goods with varying (polar) combinations of “Jointness” and “excludability” into their relative positions and gives us some clear initial bearings. Since the central focus of this paper is on markets supplying goods for which exclusion is possible and econom- ical, and more specifically is concerned with the peculiarities which emerge when those markets are in public as compared to private goods, we are concerned exclusively with cases that lie along the right-hand vertical in that figure. This perspective of the excludable public-private goods spectrum suggests an important link with the literature on what, following Bu- chanan (1965), are termed “club goods,” that is, goods for which some “consumption sharing” may be efficient and can in fact be secured through the formation of private (i.e., excludable) consumption clubs. Thus, pure private goods are those for which optimal club size is neces- sarily unity, while pure excludable public goods are those for which optimal club size is infinite (all-inclusive).3 However, while some contributions to the club goods literature have raised the question of how well “firms” might perform in this context, its central concern has been with the notion of optimal club size and structure, and little attention has been given to the development of positive models of market provision. Conventional microeconomic mar- ket models of private goods provision provide one special case: analo- gous models of excludable public goods evidently provide another. Public Goods with Price Exclusion 207 Jointness Pure Public Good Excludable Public Good (complete) 1 x x Non-Excludable Private Good 4 Pure Private Good x > 1 7 (complete) Excludability Figure 1 Two further points might, however, be made. First, though “special” cases, neither extreme need be considered an entirely empty set. Second, in an important sense excludable public goods may be the more interest- ing case: by allowing for varying degrees of crowding or congestion of excludable public goods, the entire spectrum of commodities may be covered, with private goods emerging as the case of a “completely congested” excludable public good. We would expect, indeed, many of the distinctive features of market behavior in the provision of excludable public goods to persist until the pure private good pole is reached, and it is to an examination of those features that we now turn. Some Implications of Jointness The key feature of excludable public goods—that which distinguishes them analytically from conventional private goods—is evidently the Samuelsonian jointness which is exhibited completely by the former but not at all by the latter. The implications of this distinction for the decision making of exclud- 208 Cliff Walsh able public good producers and for market behavior derive essentially from one simple, but profoundly important, consequence of jointness itself: namely, that (unlike for private goods) output units do not coin- cide with consumption units. For an excludable public good each output unit gives rise to many (in the limit, an infinite number of) consumption units: each separate TV program transmitted by cable can be viewed by as many as are prepared to pay the access fee and monthly charges; each theatrical performance generates as many potential consumption units as there are seats available, and so on. This is, of course, a feature of public goods whether excludable or not. But when price exclusion is possible, and is practiced, it imparts a num- ber of analytical complexities that are of little relevance with nonexclud- able public goods and do not arise with private goods. For example, there is, now, a need for a clear distinction between the price that individuals pay for access to a particular unit of output and the price producers re- ceive (the latter being a summation of the former). If we take the (natural) meaning of price to be what individual consumers pay, it follows that conventional price-cost relationships need no longer apply: in particular, per-unit prices less than marginal production cost may (and perhaps commonly will) apply to the provision of excludable public goods. And since producers are concerned with the total price (or revenue) from each unit produced and need never produce more units than are demanded by any single consumer at any particular unit price, the pattern of demand will clearly play an important role in excludable public good markets in circumstances where it does not for private goods. Before examining the direct consequences of these implications of jointness in the context of some specific market models, we deal briefly with two related, but logically prior, matters: first, the possibility that jointness may permit price discrimination across consumers to be prac- ticed in circumstances where it would not be possible for private goods and, second, the possibility that jointness may lead to a “natural” mo- nopolization of markets in excludable public goods. Jointness and Price Discrimination One implication of the output/consumption distinction drawn above is that, unlike for private goods, individuals are not in direct competition for access to particular output units.4 As Earl Thompson (1968) pointed out, this would seem to imply that even perfectly competitive suppliers of Public Goods with Price Exclusion an excludable public good could impose and sustain a set of interperson- ally discriminating prices, if they were fully informed about the prefer- ences and consumption levels of all individuals (and retrading is no problem). In that case all firms could ensure that their entire output is consumed by all individuals by facing each with a per unit price equal to their average valuation for the total quantity of the good they consume. Thus every firm sets the same price for any given individual, but different prices for different individuals; above normal profits are driven to zero by expansion of industry output; and in equilibrium the sum of prices (equal to average valuations) equals marginal cost. Thus, Samuelsonian jointness does appear to make interpersonal price discrimination possible in circumstances where it would not be for private goods. It is, nonetheless, difficult to see how producers (even monopoly producers) could fully exploit this fact in the absence of greater information than that which would be generated through re- peated market transactions. However, together with a rather natural lumpiness in units of exclud- able public goods (an entire opera or TV program being treated as a “unit”), jointness in consumption may facilitate the practice of “bun- dling” of units in various ways. The producer has some incentive to induce everyone to consume the entire output, though it may not prove possible, or profitable, to always do so. In practice a rich variety of alternative pricing schemes seem to be applied. Per-unit prices (e.g., per program for pay TV or per visit for swimming pools), lump sum charges (e.g., monthly charges for cable TV, flat rate fares for buses, or “season ticket only” offers for plays) and multipart pricing (e.g., entrance fees plus per-ride charges at fairgrounds) are all observed; and alternative pricing strategies may exist side by side (e.g., season ticket and single ticket offers for buses, plays, etc.). As with private goods, however, alternative pricing arrangements can be seen as being (rationally) chosen by profit-seeking entrepreneurs (subject to market constraints): they are not in any sense predetermined by the nature of excludable public goods.5 Jointness and Decreasing Costs A second implication of Samuelsonian jointness is that it gives rise to an extreme form of decreasing cost: the average cost per consumer of a 209 210 Cliff Walsh particular output unit decreases as the number of consumers of that unit increases (and the marginal resource cost of extra sales is, in this sense, zero when pure jointness prevails). This is, of course, to be clearly distinguished from the (conventional) decreasing cost with respect to output that can arise with private goods where significant “common costs” exist. Although in the provision of excludable public goods both forms of decreasing cost may exist side by side, each has quite different implications for market structure. Consider the case of production and transmission of TV programs, with exclusion through coin-in-the-slot devices or cable charges, where both forms of decreasing cost do occur. On the one hand, there are significant common cost elements associated with the provision of facili- ties for production and transmission of programs, so that average costs decrease as output (the number of programs produced and transmitted) increases. Evidently, the (quite familiar) tendency toward monopoliza- tion of such facilities is likely (and, perhaps with regulation, may be socially desirable). On the other hand, each separate unit of output (each program broadcast) is joint in Samuelson’s sense, so that average cost falls, and marginal cost per viewer is more or less zero, as the number of viewers of any particular program increases. What this suggests is that the market for particular homogeneous units (e.g., a particular type of program ina particular time slot) will tend to be monopolized (and should be), but it clearly does not suggest that all separate output units need be, or will be, produced by a single firm. Competition in the production of different units (different John Wayne movies, for example) is perfectly consistent with decreasing con- sumption cost for any one of them. And such competition may exist though there may be a tendency toward monopolization of the facilities within which production takes place.® Jointness and Market Behavior We now turn to a more direct examination of the impact of jointness on producers as decision makers and of the implications which follow for market behavior. Specifically, we examine: why competitive equilibrium is characterized by consumers facing a “step function” of prices; why market equilibrium for both competition and monopoly is typically Public Goods with Price Exclusion associated with the rationing of some consumers by output rather than by price; and why this may give rise to positive price-output relation- ships (i.e., why price may fall when a rise in MC causes output to fall and vice versa). Excludable Public Goods and Revenue-Cost Comparisons For a private good the “exclusive” consumption of each and every unit implies that, in equilibrium, aggregate production and aggregate con- sumption are equal, and hence comparisons of costs and revenues are simple and direct. For an excludable public good, however, the level of production need never exceed the number of units consumed by the highest demand individual but, depending on the price charged, some of those units provided will be consumed by many individuals. Even for the simplest of market models, comparisons of cost (which depend on production) and revenues (which depend on consumption) are thus more complex for public good producers than for their private good counterparts: aspects of the distribution of demands are as relevant as aggregate demand. This is illustrated, with the aid of figure 2, for a simple uniform per- unit pricing monopolist, selling an excludable public good, G, to two consumers. D, and D, are the individuals’ demand curves; Dy, is the market demand curve derived in the usual manner as a horizontal sum- mation; MR,, is the monopolist’s market marginal revenue curve; and MC indicates marginal production cost. If G had been a private good, G would be the profit-maximizing output and total consumption, with MR,, and MC equated. However, since G is an excludable public good, at the price p, while total consump- tion is G (and MR, correctly identifies marginal revenue with respect to consumption), output need not exceed G,, the highest demand at that price. Hence, marginal production cost is that relevant to G5, not G. The nature of the complexities introduced by this separation of the consumption and production dimensions can be explained as follows. If price were lowered, the slope of D,, would identify the extra total consumption, and the relevant area under MR,, would measure the extra revenue that would be generated. Since the slope of Dy, is the sum of the slopes of the individual demand curves (i.e., 2} dqi/dp, henceforth denoted &S,), the extra revenue is approximated by MRyy X 2S,. How- 741 212 Cliff Walsh vi MC ) G, G G ‘ Units of G Figure 2 ever, the slope of the demand curve of the highest demand individual (S, from D, in figure 2, but S,, more generally) determines the extra output required and hence the area under MC in this vicinity would measure the extra production cost involved, approximated by MC x §S,. From this it follows that profit is maximized where MRy, X 2S; = MC x S, or, rearranging: Sh MRy = MC x 3S, (1) The left-hand side of (1) is simply market marginal revenue: the revenue gained from an extra consumption unit. The right-hand side represents the production cost of an extra consumption unit, which in the exclud- able public good case is less than the cost of an extra production unit (MC). Thus, while the conventional rule “MC = MR” remains concep- tually valid, its interpretation is much changed by the separation of output and consumption dimensions.” The information requirements of the public good producer are quite Public Goods with Price Exclusion 213 different from those it would face in the private good case. Even under simple uniform pricing, the producer is as concerned about the pattern of demand as about aggregate demand for the product, since aggregate demand no longer provides a direct revenue-output and hence revenue- cost relationship. In fact this two-person case oversimplifies the pro- ducer’s problem (as we shall later see), but it serves as adequate warning that conventional (private good) modes of analysis are inadequate, and conventional expectations about the characteristics of equilibrium are unlikely to be fulfilled. A Step Function of Prices under Competitive Provision For perfectly competitive producers of an excludable public good, a similar “complication” is introduced, although the nature of market conditions reduces its impact on individual producers. However, it does give rise to what is perhaps the most immediately surprising result of the difference between private and excludable goods—namely (as Oakland demonstrated) that competitive equilibrium (with normal information assumptions) is associated with consumers facing a “step function” of prices. In long-run competitive equilibrium, no unit in production may earn above-normal profit: average (and marginal) revenue for all production units must equal average (and marginal) production cost. However, if the same per unit price in consumption applied to all units produced (as it does for private goods), given differences in demands, the jointness characteristic would lead some units to have more consumers than others and hence producers of “more intensively consumed” units would earn greater revenue from their production. The process of competition must eliminate such differences, and to do sO output units which are more intensively consumed must have lower unit prices in consumption: in long-run equilibrium the price to all consumers of any unit in consumption must be MC/m (where m is the number of consumers of that particular unit) so that the revenue earned from its production is m X p,, (=MC). For example, in a community of n individuals, all may jointly con- sume some units at a price p,, = MC/n and the revenue of the producer of any such unit would just cover costs since n p, = MC (=AC under com- petitive equilibrium conditions). The lowest demand individual would, 214 Cliff Walsh P,= MC = MC/, Py ic/2 P3= MC/3 0 G, & G Units of Public Good Figure 3 at this price, have their demand fully met and would drop out. However, the (n—1) remaining individuals may be willing to (jointly) purchase additional units at a price of P,, , = MC/(n—1) and again the producers would just cover costs; and so on until, if MC is relatively low, the highest demand individual may, finally, purchase some units alone at P, = MC. Such an equilibrium is illustrated in figure 3 for a simple three person case: all three individuals consume OG, units at p; = MC/3, an addi- tional (G, — G,) units are purchased jointly by individuals 2 and 3 at P, = MC/2, and individual 3 alone purchases another (G3 — G,) units at P, = MC.8 In a long-run equilibrium each and every production unit earns the same revenue, equal to marginal (and average) cost: producers are thus indifferent about which “segment” of the market they are serving. Con- sumers will have arranged their purchases so as to obtain the units with the lowest consumption price first, buying additional units from such progressively higher-priced categories as are available so long as the Public Goods with Price Exclusion 215 marginal price they must pay is no higher than their marginal valuation across all units purchased. Thus, because of the jointness property of excludable public goods, competitive equilibrium is characterized by a form of “reverse discrimi- nation” across units of output—but not across consumers, all of whom face the same step function of prices (i.e., face the same price for any given unit). This pricing pattern need not occur under monopoly, however, since any profit earned on intensively consumed inframarginal units is not eliminated by competition. It is, nonetheless, possible that a monopoly producer might choose to establish an Oakland-like pattern of prices (but such that m X p,, > MC for all units) in effect as a means of “exploiting” different elasticities of demand among groups of consumers and inducing greater consumption of more intensively consumed units (see Brennan and Walsh 1979). Such an outcome could not, however, be said to be characteristic of monopoly, and its profitability compared with simple uniform pricing would depend crucially on the pattern of demand. Rationing by Output With private goods in profit-maximizing equilibrium all consumers will be rationed by price alone: since each unit is “exclusively” consumed, if it is profitable to produce at all (p = MC), it cannot pay to restrict consumption of any individuals below the number of units they would wish to buy at that price.? In contrast, for excludable public goods a consumption price below MC can be profitable, but then it will not always be profitable to allow all individuals to consume all that they would wish at that price. Consider, initially, a simple example. TV broadcasts are provided by a monopolist who utilizes coin-in-the-slot devices to set uniform per-unit consumption prices for output; the firm faces constant costs (MC = AC); but MC exceeds the marginal valuation of all consumers for any unit of output (i.e., MC(q) > p;(q) for all i and q). If TV programs were private goods, production could not be profit- able. However, the jointness property of programs implies that if there exists some per-unit consumption price, p’ < MC, such that the number of individuals, m, who would jointly tune in is sufficient to cover costs 216 Cliff Walsh (i.e, m X p’ = MC), it is profitable to produce at least one program. Indeed, several may be profitable at p’, even though the number of individuals who will consume additional programs decreases as output is increased (i.e., some individuals will be rationed out by price). How- ever, once the number of consumers of additional programs has fallen to a level, g, such that g x p’ = MC, it is no longer profitable to produce additional units, even though some individuals would be willing to purchase them at that price: those individuals are clearly rationed by output.) The assumption that MC exceeds all individual marginal valuations may characterize many examples of excludable public goods. It is not, however, a necessary condition for the emergence of rationing by output, as can be seen by further consideration of the implications of equation (1) above. At the output selected by reference to (1), the profit-maximizing consumption price may be less than marginal production cost for a wide range of MC levels, and if some of the units produced at this price would be consumed by the highest demand individual alone, it cannot be profit- able to produce them. Figure 4 illustrates the point. D,, is the highest individual demand curve and by construction has a slope one-sixth of that of Dy, the market demand curve (i.e., $,,/2S; = 1/6). Given MC as illustrated, (1) is satisfied with the price p* and a total consumption of G*. At p* individ- ual n wishes to consume (and production need not exceed) Gn units. However, the next highest demand individual would (jointly) consume only G,,_, of them, so (G,, — G,,_,) units would be exclusively consumed at p* < MC. Profit would be increased by reducing output at least to G, since two individuals, (n—1) and n, would then be consuming all units produced (and others will also be consuming some of them), profit will be made on all G,,_, units. Of course, this output and price may not be the most profitable arrangement when n is “rationed” in this way: a new profit-maximiza- tion calculation is now required similar to that in (1), but in which n is treated “as if” it had the demand curve of (n—1). And as the logic of the argument makes clear, that must involve the persistence of output ra- tioning of at least the highest demand individual.!! It also makes clear how much more complex is the producers’ problem when jointness prevails. Rationing by output can also arise with competitive provision as can be seen by reference back to figure 3. Had D, intersected MC to the left Public Goods with Price Exclusion 217 Figure 4 of G,, individual 3 would not have consumed additional units alone: it would simply consume G, units (some at P, and the rest at P,) and hence would have excess demand at the marginal price it faces.!2 In fact, in a similar sense, all but the lowest demand individual in the competitive case invariably face quantity rationing on “low” price units, as well as the possibility of such rationing with respect to their marginal consump- tion units. Evidently “rationing by output” is a somewhat characteristic feature of market outcomes for excludable public goods. A Positive Price-Output Relationship The most striking result to emerge from analysis of excludable public goods provision is the possibility that an upward shift in the MC sched- ule may result in a reduction in profit-maximizing price (e.g., profit- maximizing airfares may be lowered when oil prices rise). Such a result seems impossible in the private good case and, in fact, emerges in the public good context as a result of the possibility of profit-maximizing outcomes with rationing by output. 218 Cliff Walsh Consider the case of a monopolist facing a constant MC schedule who maximizes profit at the price/output combination (pp, Go) with some individuals rationed by output. The revenue received from the last out- put unit is k X po (where k is the number consuming that unit), and a necessary condition for Gy to be a profit maximum is that k X pp = MC. It must also be true that a change in price, Gg given, would decrease revenue: thus, if Py were reduced, the revenue gain from having the “price-rationed” individuals increase their consumption must be more than offset by the reduced revenue on all Gp units consumed by the “output rationed” individuals (and vice versa for a price increase). Now suppose MC shifts up to MC’, so that at (pp, Gg) some marginal production units are no longer profitable (k * Py < MC’). Profit would be increased by reducing output to G, such that the number of con- sumers of the marginal unit is sufficient to cover its cost (i.e., k’ X pp = MC’). It is, of course, unlikely that Py) remains the profit-maximizing price (or that G, is the finally appropriate output); but, unlike in the private good case, it is by no means clear that raising price is the unambiguously correct response. At (po, Gy) more individuals are rationed by output and fewer ra- tioned by price than at (pp, Go). Hence if price is lowered at output G,, the revenue gained from price-rationed consumers will be positive, but less than at Gp. But, while there are also more output-rationed con- sumers so that as price is lowered the loss of revenue from them on each individual output unit is greater at G, than at Go, that loss is sustained over fewer output units. If the reduction in output needed to restore profitability on the marginal output unit is “large” relative to the number of additional individuals who become rationed by output, it is entirely possible that lowering price is profitable at (pg, G,), even though it was not at (pp, Go)!3: increased revenue from price-rationed individuals exceeds reduced revenue from those rationed by output. Thus jointness can give rise to an altogether unusual price-output relationship over some output ranges (Brennan and Walsh 1981). The competitive model generates an analogous, but not identical, result. While the characteristic price function, P,,, = MC/m must gener- ate higher prices in each “step” when MC shifts up, it is possible (as inspection of figure 3 would reveal) that the top marginal price bracket would now be vacated. Thus, while the price in each step would be higher, the highest price step in which nonzero consumption occurs could be lower. This must be associated, of course, with high demand individuals facing (greater) output rationing. Moreover, the degree of Public Goods with Price Exclusion 219 quantity rationing high-demand consumers face in low price brackets must also, in general, be increased. Market Provision of Public Goods With, and Without, Exclusion: Normative Issues Market Equilibrium and Efficiency We turn now to begin the explicit consideration of normative issues associated with the provision of excludable public goods with a simple question—what is the relationship between market equilibrium out- comes for excludable public goods and the outcomes required for op- timality in their provision?!4 In the context of private goods provision it is well known that, in the absence of decreasing costs, externalities, etc., both perfect competition and perfect (interpersonally discriminating) monopoly result in full effi- ciency. However, in the context of the provision of excludable public goods only perfect monopoly guarantees an efficient outcome. The ear- lier preliminary discussion of the requirements for efficient provision of excludable public goods suggests why this should be so. If producers are unable to engage in perfect price discrimination (i.e., if consumers must face identical prices for all units they consume, or at least identical price- consumption schedules) firms cannot appropriate the benefits of admit- ting all consumers to all units for which they have positive marginal valuations: hence they will tend to produce inefficiently low output levels. In the case illustrated in figure 3 of an Oakland competitive outcome, total output is established where the marginal valuation of the highest demand individual is equal to marginal cost: the fact that at least one other individual in that case has a positive marginal valuation for the final output unit cannot be converted into revenue to the firm in the absence of price discrimination, and so a socially desirable output ex- pansion is not privately profitable.!5 The inefficiency of the outcome is further compounded by the fact that lower demand individuals are excluded from some units that are produced. Much the same can be said of the uniform pricing monopoly case also considered earlier: indeed, in general, output will be less than under perfect competition and con- sumption will also be more severely rationed for all consumers, whether rationing is by output or price (see Brennan and Walsh 1981). Perhaps the only really surprising result in this context is that of 220 Cliff Walsh Thompson’s model of a perfectly informed competitive market, where he argues overexpansion of output (and consumption) will arise. In fact, this is also simply explained once one appreciates that, with perfect information, competitive producers of excludable public goods can en- gage in pricing which is, in effect, perfectly discriminating. That is, each firm sells access to all units it produces to all individuals, but each different individual pays a per-unit price equal to the average valuation over all units consumed. Thus, the revenue to all firms at any industry output level is given by the sum of all individuals’ average valuations. When the socially efficient output level is reached (where the sum of marginal valuations is equal to marginal cost), industry profit is still positive since the sum of average valuations exceeds the sum of marginal valuations. In fact, industry profit is here maximized, and a perfectly discriminating monopolist, or a cartel, would set just this output level. However, with free entry profit will be driven to zero by a further expansion of output, until the sum of average valuations is equated with marginal industry cost and an inefficiently high output (and consump- tion by all) is produced. The Thompson result is interesting, and especially so since it implies that cooperative (cartel) behavior by firms would be socially desirable in this context. But it is based on very special assumptions about informa- tion, the relaxation of which leads to totally opposite conclusions (i.e., Oakland’s underexpansion result). In general, excludable public good producers will produce too little output and will also inefficiently ex- clude some individuals from consumption of some units of output actu- ally produced. In markets for excludable public goods, not only our notions of the nature of market inefficiency, but also the appropriate recipe for corrective action, must undergo substantial revision. The Separate Influence of Non-Excludability and Jointness on Market Failure Those who have analyzed the problems of market failure with respect to pure public goods have sought (often implicitly) to attribute the market failure problem primarily to one or other of their two fundamental characteristics. John Head (1977b) points out that the two founding fathers of modern public goods analysis adopt opposite conclusions. Samuelson, focusing on the fact that jointness alone changes the condi- Public Goods with Price Exclusion tions for optimal provision, and recognizing that the ability to price- exclude cannot be expected to result in full optimality, emphasizes the role of the jointness characteristic of public goods. Musgrave, perhaps because in the early development of his work he was inclined to argue that market failure would be complete for pure public goods (i.e., noth- ing would be produced), seems to have emphasized the significance of nonexcludability. If the separate roles of jointness and nonexcludability in explaining market failure for public goods could be disentangled, we might be able to go some way toward generalizing the theory of market failure. Refer- ring back to figure 1, we might be able to subdivide the whole set of goods which exhibit varying degrees of jointness and nonexcludability into subsets, grouped according to the degree of market failure they are likely to exhibit, and ultimately to form some picture of the likely shape of “iso-market-failure lines” depicting different combinations of degrees of jointness and nonexcludability which result in the same level of market failure. However, using a comparison between excludable and pure public goods as the basis for attempting such a task, while super- ficially attractive, does not seem an appropriate approach. As with other contexts in which two or more sources of inefficiency are simultaneously present, the separate effects of each are compounded when they arise in combination. Thus, examining public goods with and without exclusion cannot establish whether nonexcludability or jointness is the more vir- ulent source of market failure when they occur together. !¢ The comparison of market provision of public goods with and with- out exclusion is not, however, a pointless exercise and may in fact be of some policy relevance. For example, in some countries TV broadcasts funded primarily through voluntary public subscription exist side by side with others funded by cable charges. Though the continued exist- ence of both may suggest both are “profitable,” and/or socially desir- able, neither is likely to be fully efficiently produced. The appropriate extent and nature of public intervention requires a comparative positive and normative analysis of their provision. In other contexts govern- ments may be called on to define, or make enforceable, property rights that would convert more or less pure public goods into excludable public goods. The provision of a system of patent and/or copyright laws in the market for ideas and information provides a case in point, one in which the relevant issues are, by common agreement, often difficult to resolve. Ju) 222 Cliff Walsh Provision With and Without Exclusion: The Problem Stated To facilitate such a comparative analysis of market provision of pure public goods and excludable public goods, we must establish what the nature of market equilibrium is likely to be in each case. For a variety of reasons the Oakland competitive model is an appealing benchmark for the excludable case. For one thing its characteristics are easily depicted as in figure 3 earlier (this figure, but with additional information to be used later, is reproduced as figure 5). More fundamentally, for market models with reasonable information assumptions, competitive provision performs best in a welfare sense: if we have doubts about whether competitive provision of public goods with exclusion dominates their provision without exclusion in this case, those doubts would be strength- ened in the context of noncompetitive models. In many respects it is the relevant model for market provision of nonexcludable (pure) public goods that is most difficult to specify. If it were true, the early suggestion of Musgrave that market failure will be complete for pure public goods would be convenient: market provision with exclusion could not do worse! However, the Musgrave position seems unsupportable, in general. Subscription TV and radio survive; altruistic donations are given to private charities; and ideas and informa- tion are produced even when exclusion cannot be practiced. The point is that if there are private net benefits for some individuals or firms from providing units of a public good, even if provision would be entirely at their own expense, then zero output cannot be a reasonable long-run equilibrium prediction. !7 Even without attempting to further specify what output will be for a pure public good, the fact that it may be nonzero makes the comparison of outcomes with and without exclusion more complex than might have been expected. Intuition initially suggests that the “power to exclude” would give excludable public good production an unbeatable compara- tive (efficiency) advantage. All consumers are obliged to reveal some- thing about their demand curves in order to gain access to units when exclusion is practiced, whereas when exclusion is not possible we are likely to get limited, and probably distorted, information on preferences embodied in market outcomes. This intuitive response, while valid, is incomplete. For while the power to exclude does constitute an important source of information, its use also constitutes a potential source of Public Goods with Price Exclusion 223 G Units of Public Good Figure 5 inefficiency. Its use may result in output being “rationed” in different quantities across consumers, even when all have positive marginal valua- tions for all units produced, and the fact that nonexcludable public goods will not be rationed among consumers in this way gives markets in pure public goods an offsetting advantage. Except where the output of a public good provided without exclusion would be less than the con- sumption of every individual if it were provided with exclusion, the relative market failure cannot be assessed without detailed knowledge of the pattern of individual demands as well as knowledge of the relative output levels. A familiar example can be offered. It has been commonly argued that a patent/copyright system is likely to increase the output of new ideas by allowing producers of ideas to appropriate more fully the benefits to others from their successful efforts. It is also recognized, however, that the ability to exclude implied by patent or copyright protection has a 224 Cliff Walsh cost: the social marginal cost of dissemination of ideas may be low and, with price-exclusion, dissemination is likely to be unduly restricted from a social viewpoint. The justification for the patent system must, then, rest on the view that the value of the increased output of ideas exceeds the value of the lost (or restricted) consumption of those ideas that are produced. Whether this is, in fact, the case cannot be established from a priori reasoning alone, since there cannot be a general presumption that everyone’s consumption of ideas will be larger from the larger output produced when patents or copyright apply. A Suggestive Analysis Olson and Zeckhauser (1966) have suggested that the equilibrium out- put of a pure public good might be predicted to be where the highest individual’s demand curve intersects MC, since this is where the net private benefit is maximized, with the most to lose from failure to produce at least this output. Accepting this as at least a suggestive model for public goods production in the absence of exclusion, it can be readily seen how both the pattern of demands and the level of costs are critical to the relative efficiency of public goods provision with, and without, exclusion. Consider, initially, the case illustrated in figure 5, when marginal cost is MC and the individual demands are D,, D,, and D3. The pure public good output and consumption level of all individuals is G;. This is also the output in the Oakland competitive solution, with exclusion, but in the Oakland case the consumption of both 1 and 2 is less than G3. Here, then, provision without exclusion unambiguously dominates provision with exclusion. The effect of a different pattern of demand can be seen by considering a shift inward of D; to D5. The Olson and Zeckhauser output is now G3, consumed by all. The Oakland equilibrium, however, would now in- volve an output of G,(> G3) since individual 3 would not buy units alone at p3. The combined gain to individuals 2 and 3 from exclusion must be compared to the loss to individual 1, and the balance would clearly change as the precise position of D, is changed (and could be further altered by adding in more individuals with different intermediate demand curves). Finally, consider the effect of assuming MC’, rather than MC, as the Public Goods with Price Exclusion 225 marginal cost schedule. Now the pure public good equilibrium in the Olson and Zeckhauser analysis would presumably be zero, in which case provision with exclusion cannot result in a worse outcome. In fact, as illustrated in figure 5, Oakland output would be Gj fully consumed by all three consumers at a price MC’/3 (additional units at MC’/2 and MC’ would not be taken up by 2 and 3). Of course, as MC increases yet further, the output in the Oakland model will progressively change toward zero too. Evidently, we cannot establish a priori whether the market is gener- ally likely to perform better in the provision of public goods with exclu- sion than without it. This observation, which initially appears strongly contrary to intuition, is much less surprising once one rejects the rather firmly imbedded notion that market failure will be complete for pure public goods: it is then clear that provision of public goods without exclusion has some advantages, given the jointness characteristic of these goods, and comparative assessments of market performance are obliged to proceed on a case-by-case basis.!8 Interestingly, despite the relatively recent origin of models of excludable public goods provision, it is the degree of uncertainty about the likely market outcomes without exclusion that renders the analysis presented here suggestive rather than definitive. And this, in turn, may be the unfortunate by-product of an age in which a rather optimistic view of government’s ability to correct market failure has prevailed. Concluding Remarks Excludable public goods represent a distinctive class of commodities. Their essential feature is that they combine the jointness characteristic of a pure public good with the excludability characteristic of a pure private good. This combination gives models of their market provision an en- tirely unfamiliar structure, deriving primarily from the fact that units in production and units in consumption must be distinguished for exclud- able public goods in a way which is unnecessary for private goods (and, in general, in a way which is also unnecessary for pure public goods). Because of this, market outcomes are significantly different from those predicted for analogous markets in private goods and, equally impor- tantly, different from those expected for pure (nonexcludable) public goods. Moreover, it is far from clear that the “ability to exclude” im- 226 Cliff Walsh proves the market’s performance in the provision of public goods: the jointness characteristic of public goods, whether excludable or not, not only generates unusual market behavior, it also imposes unique and demanding requirements if the market is to be even tolerably efficient. Notes 1. Other terms have been applied to them, such as: privately producible public goods, marketable public goods, collective goods, or joint goods. How- ever, price-excludable (or simply excludable) public goods seems the most accu- rately descriptive name. 2. Demsetz (1970) and Auster (1977) produce alternative competitive mod- els, but in both cases their analysis is flawed in critical respects (see, for example, Head 19774). In a more recent paper Brennan, Lee, and Walsh (1980) extend Lee’s (1977) earlier incomplete analysis of monopoly provision with all-or-none exclusion. In this case it appears that monopoly provision may generate overex- panded output and consumption in efficiency terms. 3. In fact only narrowly defined basic consumption goods may be inherently private. Many goods for which consumption sharing in some form is possible may nonetheless be treated as private goods because individuals (rationally) choose to consume them “exclusively” (e.g., houses, cars, even clothing), even though they could (reasonably efficiently) form “consumption clubs” for them. For further analysis, see the survey by Sandler and Tschirhart (1980) and the references given there. 4. It was failure to fully appreciate the significance of this fact which led Demsetz (1970) into error in his attempt to show that competition would actually generate information on preferences sufficient to allow discriminatory per unit prices to be set for excludable public goods. Head (1977a) provides the relevant critique of this and other common errors. 5. Auster (1977) and Lee (1977) argue that all or nothing pricing is a characteristic feature of excludable public goods—in the former case, even where imperfectly informed competitive markets exist. This seems entirely er- roneous. Brito and Oakland (1980) and Burns and Walsh (1981) explicitly consider alternative choices of pricing strategy. 6. We are, of course, warned by the nature of the discussion that the homo- geneity assumption may become somewhat strained. Nonetheless, at the concep- tual level we are not compelled to think of each unit as a unique product nor to think of monopoly as natural. Thus basic meteorological data (collected by a monopoly firm) may be purchased by a number of competing firms which interpret and sell weather information to final consumers. More competition Public Goods with Price Exclusion 227 (including the provision of production facilities) will exist as congestion becomes more important. 7. The simple MC = MR requirement for private goods follows directly from the fact that an extra production unit is needed for every extra unit in consumption. For any price change extra revenue is MR,, = S,. The public good producer is actually likely to work in terms of revenue (and cost) per production unit. Rearranging (1) and noting that MR,, = = MR, S,/>S,, we obtain > MR; S,/S,, = MC as the more useful formulation. For expository purposes, however, we retain (1). 8. It is assumed that income effects are negligible, so that demand curves are unaffected by inframarginal prices. Different positions of the demand curves, or of the MC curve, would influence the number of price categories that actually emerge, as well as the quantities supplied in each. 9. This is not to say that “rationing by output” is never observed in the world of private goods. Nonprice rationing is widely practiced, often as a result of government intervention. There are, moreover, some cases, such as that of the producers of “prestige” cars, where (presumably in part for nonpecuniary rea- sons) availability is rationed rather than price raised to eliminate excess demand. 10. To give an obvious example, you cannot catch a bus after midnight, even though you are willing to pay the regular fare, because there are insufficient individuals like you to cover the cost. 11. Brennan and Walsh (1981) and Burns and Walsh (1981) offer further analysis of the rationing outcome. Note that for MC sufficiently low, rationing by output will not occur. 12. In fact, it is possible to show in this context, for some positions of the relevant demand curve, that both 2 and 3 could be rationed by output, or even that 2 but not 3 might be rationed in this sense. Since, in general, not all potential price categories will exist, many people may be rationed by output on their marginal consumption units. 13. It might be noted that if price is lowered, output will have to be raised above G,. The logic of the earlier reasoning makes clear, however, that output cannot rise as far as Gp. A “concrete” example may help to capture the flavor of the outcome described in the text. A commuter airline currently provides seven return flights per week between two cities and does so profitably even though each flight is less than fully booked. If the price of fuel increases, the airline might find it now more profitable to reduce the flights to three per week and to reduce the airfare. Those who previously made the return trip each day will have their consumption cut (and will be rationed by output, not price), but each flight will now have a larger number of passengers. 14. It should be remembered that an efficient ouput can be associated with inefficient consumption rationing (and vice versa) when public goods provision is at issue. 228 Cliff Walsh 15. The outcome illustrated in figure 3 (and others later discussed in figure 5) may be taken to be “typical” of the Oakland outcomes but may not always apply. For example, if in figure 3 all other MV’s had fallen to zero at G,, output (but not consumption) would have been fully efficient. The Oakland outcome, more generally, can range from complete inefficiency (zero output and consump- tion) to full efficiency (optimal output fully consumed by all), depending upon the pattern of demand and the level of MC. See “A Suggestive Analysis” below. 16. My colleague, Geoffrey Brennan, forcefully impressed upon me the signif- icance of the observations contained in this paragraph. 17. Public goods provided as complements to private goods (e.g., information which will increase sales of a private good) provide a clear example. In some cases the public good may, in fact, be inherently joint in a Marshallian sense with a private good—as with the spillovers from education or vaccination against a communicable disease. Even a nuclear deterrent conceivably might be provided by a large firm in the absence of government provision. 18. Whether public provision with or without exclusion would be better than market provision is an interesting question but beyond the scope of this paper. 12, JOHN G. HEAD Merit Wants: Analysis and Taxonomy In The Theory of Public Finance (1959) Richard Musgrave puts forward two principal concepts, social wants and merit wants, which provide the foundation for public expenditure analysis in the allocation branch. The social wants concept serves to epitomize the range of cases in which, due to market failure, public provision may be justified in order to give effect to individual preferences. Policies for the provision of social wants thus fall squarely within the consumer sovereignty tradi- tion. Reflecting, however, some of the doubts and reservations of econo- mists regarding the ultimate normative authority of the consumer sov- ereignty principle, Musgrave is at the same time unwilling to limit the allocation branch function to provision for generalized social wants, mixed goods, or externality problems. The concept of merit wants is accordingly put forward to epitomize a further range of cases in which, absent social wants problems, individual preferences expressed in a market setting may nevertheless require correction. He therefore draws a clear distinction between social wants “where corrective policy is re- quired in order to secure an allocation of resources that is in line with consumer preferences” and merit wants “where the reason . . . for budgetary action is to correct individual choices” (1959:9). From the outset, however, Musgrave is also concerned to justify the place of the merit wants concept “in a normative theory of public economy based upon the premise of individual preference in a demo- cratic society” (13). He argues, for example, that “interference with consumer sovereignty, narrowly defined, may derive from the role of leadership in a democratic society,” and that circumstances may there- fore arise “where an informed group is justified in imposing its decision upon others” (14). Public intervention in areas such as health, education, sale of drugs, protection of minors and antidiscrimination policy can, he suggests, be explained at least in part under this general heading. Sim- ilarly, due to misleading advertising, “There may rise a distortion in the 230 John G. Head preference structure that needs to be corrected. The ideal of consumer sovereignty and the reality of consumer choice in high pressure markets may be quite different things” (14). A partial reconciliation of merit wants with a somewhat broadened consumer sovereignty concept is therefore explicitly envisaged. At the same time, however, Musgrave is alive to the dangers of a preference correction concept indiscriminately applied, and he warns against authoritarian aspects which cannot be reconciled with the consumer sovereignty tradition. Musgrave’s original discussion of the merit wants concept, although suggestive and useful, is very brief and invites more extensive analysis and further clarification. In an early commentary on Musgrave (1959), Head (1966) argues that there are indeed important public policy issues which lie beyond the scope of the traditional consumer sovereignty principle. These include problems of ignorance and irrationality which could serve to justify the central merit wants function of correcting individual choices in the allocation branch. They also include issues of interpersonal and intergenerational equity and related problems of inter- temporal preference distortions which are handled in the context of Musgrave’s multibranch framework by assignment to separate distribu- tion and stabilization branches where, by implication, individual prefer- ences may not be decisive. ! Head argues that at least some of these functions may be handled within the framework of a richer and more sophisticated consumer sovereignty concept. What might otherwise appear to be elitist and authoritarian interference with the preferences of others may, for exam- ple, be justified in terms of a generalized externality concept extended to encompass psychic externalities. Similarly, in the case of interpersonal and intergenerational redistribution, utility interdependence appears to offer a basis for reconciliation of further controversial and apparently paternalistic functions with an individual sovereignty principle. In this analysis it appears, nevertheless, that there remain important and legiti- mate merit wants policies which lie beyond the reach of even a broad consumer sovereignty principle. To the extent that merit wants functions can be reconciled with consumer sovereignty by way of a generalized social wants or exter- nalities concept it may, however, be doubtful whether a truly distinctive merit wants concept has in fact been achieved. Indeed, the generalization of the social wants concept and its extension in the work of Baumol (1952), Buchanan (1968), Hochman and Rodgers (1969), and others to a much broader range of public functions do appear to have reduced | Merit Wants: Analysis and Taxonomy the need for an independent and more controversial merit wants notion. Partly as a result, there has been remarkably little real progress in the discussion and development of the merit wants concept in the subse- quent literature in spite of a steady trickle of articles.2 Most public finance specialists, including Shoup (1969b:97—98), appear to have concluded with McLure (1968) that the merit wants concept is either completely unnecessary or irretrievably authoritarian in character and therefore totally irreconcilable with a normative framework based on individual preferences. Of Musgrave’s twin concepts of social wants and merit wants, which were to provide the foundation for public expendi- ture analysis, it is accordingly the social wants concept which has held center stage, while the merit wants notion, though not entirely forgotten, has led a somewhat shadowy existence on the periphery. Although there has been little progress in the analysis of merit wants in the public finance literature of the past twenty years, there have, however, been important related developments in some of the more directly relevant disciplines and subdisciplines. Within welfare econom- ics, including the neighboring areas of social choice and public choice, serious attempts have been made to broaden the analytical framework in order to address, at least in a preliminary way, some of the more chal- lenging but previously neglected issues, and these efforts have been usefully complemented by contributions from neighboring disciplines such as social philosophy. In this review we propose a considerably broadened analytical frame- work which draws heavily on these interdisciplinary developments. In the application of this framework we attempt to clarify both the distinc- tions and the common features in some of the more important examples of merit wants policies traditionally cited by Musgrave and others. We argue that the “thin theory” typically applied by economists in the analysis of complex public services such as health and education, as in their occasional forays into related areas of social policy and into the analysis of political processes, cannot possibly do justice to the richness and complexity of these important social phenomena. The Analytical Framework When the concept of merit wants was first developed in the late 1950s, the normative framework prevailing in the New Welfare literature of the period gave little attention to the distinctions fundamental to merit 231 232 John G. Head wants analysis. The social welfare function was usually presented in the following form: WW (UF, U2... 5 Us... «5 9) (1) where the individual utility indicators U! were generally taken to repre- sent the individual’s revealed preferences (e.g., Little 1950; Graaff 1957). The central feature of this literature was accordingly the elabora- tion of a simple, and by implication unique, consumer sovereignty ideal in the form of a Pareto optimality concept based on overt or revealed preferences. Controversial questions regarding the normative authority of revealed preferences, like the issue of income redistribution and the complications of utility interdependence and preference endogeneity, were largely ignored. In the development of a broader and more relevant welfare frame- work for purposes of merit wants analysis, an essential starting point is the recognition, as for example in the work of Sen (1977), of the exist- ence of a hierarchy of individual preference sets. The notion of such a hierarchy of preferences was already familiar in the earlier utilitarian welfare economics of Pigou; and the Pigovian distinctions between de- sires, satisfactions, and welfare offer a useful starting point in the de- velopment of an appropriate classification of policy-relevant levels of preference.3 In the Pigovian taxonomy the concept of “desires” (or “ex ante” preferences) clearly corresponds to the standard notion of overt or revealed preferences, whereas the concept of “satisfactions” leaves room for possible divergencies between actual choices based on revealed pref- erences and the individual’s true preferences or satisfactions (ex post). Satisfactions or true preferences are in turn distinguished from the con- cept of individual “welfare” or “real interests.” As shown below, diver- gencies between desires and satisfactions may be explained in terms of impulsiveness or lack of information, whereas a divergence between satisfactions and welfare may reflect some more fundamental impair- ment in the individual’s capacity to choose. Merit wants policies may accordingly be required to assist the individual in giving effect to his or her own “higher order” preferences. It is clear from Pigou’s own analysis and illustrations that the domain of the Pigovian preference hierarchy is strictly limited to the individual’s own economic interests and concerns, and no extension to the wider realm of social concerns, ethical preferences, or moral values is en- visaged. Within these limitations we show that Pigovian distinctions and Merit Wants: Analysis and Taxonomy 233 levels of preference remain relevant and important for merit wants analysis. Application of the Pigovian distinctions must presuppose some possibility of value criticism though, in principle at least, of a relatively straightforward and uncontroversial character which could readily be performed or acknowledged by the individual alone. Pigovian merit wants policies should accordingly be theoretically quite soundly based, though perhaps mundane and unexciting. More exciting possibilities emerge when the domain of the preference hierarchy is extended to embrace the individual’s ethical preferences or moral values. That individuals have, and to a greater or lesser degree choose to live in accordance with, certain ethical principles or moral values is a matter of common observation. So also is the existence of tension and conflict between such moral values and the individual’s “lower-order” desires, inclinations, satisfactions, or even real interests (in the standard narrow sense) from the Pigovian preference categories. Unquestionably the most important development at this level for merit wants analysis has been the emergence from an interdisciplinary literature of a broad but fundamentally individualistic framework of ethical principles within which some of the more challenging merit wants issues can at least be systematically addressed if not definitively resolved. With roots in the contractarian tradition of political philoso- phy and in Kantian ethics, and with seminal contributions from Har- sanyi (1955) and Buchanan and Tullock (1962), among others, this development culminates in Rawls (1971). Whatever its deficiencies in matters of detail and specifics, the Rawlsian edifice potentially provides what was manifestly lacking in the New Welfare Economics of the 1950s: a broad framework for the development of “a normative theory of public economy based upon the premise of individual preference in a democratic society” (Musgrave 1959:13). Ethical preferences in the Rawlsian framework are those which would find expression in an appro- priately specified institutional context: in the “original position” under the “veil of ignorance.” They are accordingly “disinterested prefer- ences,” purged of any possible taint from the individual’s vested interests in the status quo. Although of profound importance for the development of a com- prehensive normative framework, the analysis of ethical preferences is, however, inherently controversial. Individual value systems differ widely, and a Rawlsian approach, though logically persuasive, is also highly abstract and philosophical and may not be very widely accept- 234 John G. Head able. Merit wants policies to give effect to ethical preferences, whether Rawlsian or other, can accordingly be distinguished from more narrowly Pigovian merit wants policies as arguably more important and exciting but also considerably more controversial. Combining the Pigovian distinctions within the general category of subjective preferences with the extension to the domain of ethical prefer- ences as proposed by Harsanyi and Rawls, we come up with a four-tiered hierarchy of individual preference orderings which can conveniently be presented as a set of four variations on the simple consumer sovereignty ideal from the New Welfare Economics of the 1940s and 1950s as represented by the social welfare function of equation (1), viz.: WW (DI 2) oe crs De is or OD) (1a) We Wo (ShS47 Sie oan) (xb) W = W (Rt Re ca Ri ee ghR3) (1c) and W = W (E!, E,..., E,..., E) (1d) In this much richer analysis the D' represent the standard (ex ante) utility indicators of revealed preference (Pigovian “desires”); the S' represent the corresponding (ex post) indicators of “true” preference (Pigovian “satisfactions”); the R' represent the individual’s “real interests” (Pigo- vian “welfare”); and the E' represent ethical preferences (Rawlsian or other). This broader conceptual framework need not, of course, rule out the possibility that for some individuals D', S', R', and E' may coincide; nor does it exclude the unlikely possibility that (1b), (1c), and (1d) may collapse completely into (1a) as in the standard New Welfare analysis of the 1950s. More generally perhaps it might be expected that D', S', R', and E' would coincide for most individuals across a broad range of choice situations. The proposed formulation should in any case help to ensure that important policy issues of a merit wants character requiring “value criticism” are not completely overlooked or brushed aside. In- stead of a unique and simple concept of consumer sovereignty based on (1a), three further concepts accordingly emerge, based on (1b), (1c), and (1d). A second dimension in which some broadening of the conventional New Welfare framework would seem to be essential is in the recognition of the realities of utility interdependence. As Musgrave (1959:11) em- phasizes, there is no need to assume that people are “selfish monsters.” In this, as in other respects, both New Welfare theory and the Pigovian Merit Wants: Analysis and Taxonomy 235 preference theory appear to be too narrowly drawn. Just as Pigovian externalities or Musgravian social wants may serve to justify public policies which give effect to individual preferences, might not the gener- alized externality phenomena of utility interdependence serve as the basis for a further range of “merit wants” policies within the framework of a somewhat broadened consumer sovereignty ideal? Under each of our four consumer sovereignty concepts the issues raised by utility inter- dependence must therefore be explicitly addressed, and a choice must be made between a notion of consumer sovereignty based on consumer preferences in the standard narrow sense and a broader concept of individual sovereignty which allows the well-being or consumption of others to enter as arguments in individual utility functions. In the simple consumer sovereignty framework individual preferences were assumed to be not merely independent but also exogenous or autonomous. The latter assumption is of course manifestly absurd and unrealistic. A third dimension of generalization accordingly requires us to recognize the obvious if somewhat disturbing fact that individual preferences are endogenously determined in the process of social interac- tion. Traditionally it has been thought that preference endogeneity must render any normatively authoritative concept of consumer sovereignty simply impossible. The development of a broad and realistic normative framework clearly requires, however, that possible merit wants or pref- erence correction issues raised by important real-world phenomena of preference endogeneity be squarely faced. The following subsections consider the possibilities and problems of merit wants analysis which emerge in the context of this three-dimensional broadening of the nar- row consumer sovereignty or traditional New Welfare framework. Merit Wants and the Preference Hierarchy As we have already seen, the recognition of a hierarchy of individual preference orderings suggests a possible rationale for merit wants pol- icies to assist individuals in giving effect to their own higher-order prefer- ences. Level 1: Desires and Satisfactions Within the narrowly “Pigovian” domain of subjective preferences, diver- gencies between desires and satisfactions can perhaps best be explained 236 John G. Head in terms of two major influences: lack of information and impulsiveness or weakness of will. Information problems. Since individual choices must frequently be based on incomplete or inadequate information, an obvious source of divergence between ex ante desires and ex post satisfactions would appear to be a lack of information.? It is, of course, true that information is costly to acquire, and a considerable degree of ignorance could there- fore be expected under rational behavior. It is also possible, however, that the supply of information has significant externality-generating characteristics, and public intervention may therefore be justified in order to promote informed choice. The existence of such informational externalities appears indeed to be an important fact of economic life in spite of legislative efforts to establish and extend the concept of property rights in information. Merit wants policies may therefore be justified in order to promote informed choice and to curtail the flow of false or misleading information in advertising and elsewhere. There has been some lively debate in the merit wants literature as to whether the form of public intervention need be confined to the supply or subsidy of informa- tion as such. However this may be, the category of merit wants policies based on informational externalities is conceptually quite uncontrover- sial and is precisely analogous to the generalized social wants policies required to give effect to the individual’s overt or revealed preferences within the framework of the narrow consumer sovereignty principle represented by equation (1a). Even after all Pareto-relevant policies to internalize informational externalities have been implemented, a considerable degree of ignorance and uncertainty must generally remain to complicate individual decision making. The difficulties involved in defining rational behavior under risk and uncertainty have been analyzed in great detail by economists and others in the literature on decision theory and cannot be reviewed here (see Machina 1987; Elster 1984, chap. 3). Suffice it to say, however, that the problems of individual decision making under conditions of igno- rance and uncertainty do raise issues of some relevance in the merit wants context. There is, for example, strong evidence in the work of Kunreuther (1978) that individuals have an irrational tendency to ignore completely the low probabilities attaching to possible disastrous events. Further evidence of systematic bias in individual choices under condi- tions of risk is also to be found in the important analysis and experimen- tal work of Tversky and Kahneman (1974) on the incomplete “heuris- Merit Wants: Analysis and Taxonomy 237 tics” or rules of the thumb commonly adopted by individuals in everyday life in order to simplify complex decisions. Bias due to irreversibility and cost asymmetry in decision making under uncertainty has been explored by Arrow and Fisher (1974), building on the early insights of Weisbrod (1964) on “the value of the option.” There is also evidence from the individual and social psychology literature of strong tendencies to irra- tional behavior under conditions of great uncertainty. Across a broad and important range of decisions involving such complications, some delegation of decision making to government may have considerable appeal as a device for simplifying complex choices and reducing risk and uncertainty. Part of the strong and widespread support enjoyed by compulsory national health insurance and pension schemes can no doubt be explained in this way (Diamond 1977). A second subcategory of merit wants policies can therefore be distinguished under the general heading of information problems which can be construed as self-paternalistic strategies which might be adopted in the face of risk and uncertainty. This case clearly lies beyond the scope of the generalized social wants or externality concept. Some specific examples and impor- tant special cases will be developed in more detail below. Problems of impulsivenss or weakness of will. Even if we abstract from information problems as conventionally understood, individual choices may require correction due to common “character deficiencies” or be- havioral aberrations such as impulsiveness or weakness of will. Diver- gencies between actual choices and true preferences may thus simply reflect the difference between an ill-considered or impulsive choice as compared with a carefully considered or reflective choice. Consumer protection legislation against misleading advertising and “cooling-off periods” in the case of door-to-door sales provide examples of merit wants policies which fall in this general category. Here also government may serve as a useful device to achieve the benefits of self-paternalism and thus give effect to the individual’s true preferences. More generally perhaps such choices reflect problems under the head- ing of weakness of will. The one important example of a significant divergence between desires and satisfactions cited by Pigou arises in the intertemporal context in the choice between present and future con- sumption and reflects problems of myopia or a failure to recognize the multiperiod character and future implications of certain types of choices. Elster (1984, 1985) provides a penetrating interdisciplinary analysis of such problems in terms of the concept of weakness of will. In such cases 238 John G. Head individuals may rationally choose to bind or constrain lower-order de- sires or overt preferences through what might otherwise appear to be coercive (including government-imposed) measures of a merit wants character. The case of Ulysses and the Sirens provides, of course, the classical example. Although this standard Pigovian merit wants example has long been the subject of much dispute, in particular for its apparently authoritarian or paternalistic character, it now seems clear that it can be completely reconciled with a somewhat expanded notion of consumer sovereignty as a voluntary self-paternalistic delegation of decision making to govern- ment. This argument may well carry significant weight in matters of considerable policy importance, notably in relation to future-oriented decisions in major government spending areas of health, education, and social security. Here again the issue is obviously conceptually quite distinct from externality problems which, however, also figure promi- nently in the case for public provision in these same areas of government spending. Another obvious application of the concept of weakness of will is in the analysis of the important social policy issue of drug addiction. Here again major problems arise from the failure to recognize the multiperiod character and implications of an initial decision to consume. In contrast to ordinary “habits,” such as Pigovian improvidence or overeating, the consumption of addictive drugs directly weakens the will. An impulsive or ill-considered decision to consume accordingly has long-term implica- tions which may be both extremely damaging and virtually irreversible. Even at the most mundane level of divergence between desires and satisfactions there has clearly been much controversy and the issues at stake for merit wants analysis are far from trivial or unimportant. Level 2: Satisfactions and Real Interests At the upper level of the Pigovian preference hierarchy is the distinction between satisfactions and welfare. Whereas divergencies between desires and satisfactions can be explained in terms of lack of information and a range of cognitive and volitional problems which commonly affect the behavior even of mature adults who are generally capable of rational decision making, divergencies between satisfactions and welfare tend to reflect some more fundamental impairment in the individual’s capacity to choose. Thus, for example, in the case of the psychologically dis- turbed, the mentally retarded, or the immature the supply of adequate Merit Wants: Analysis and Taxonomy 239 information, at least in the conventional sense, is clearly insufficient to guarantee that actual choices will serve the individual’s real interests. Here the concept of an informed or considered choice is almost a contra- diction in terms as the individual lacks the judgment required for ra- tional decision making over a range of important issues. Merit wants policies in the broad and important area of child protection, involving public intervention in such matters as health and education, clearly find their justification under this general heading. Beyond these relatively straightforward examples in which the prob- lem of irrationality attaches to a particular category of immature or psychologically disturbed consumers, divergencies between satisfactions and welfare are also commonly inferred from the irreversible, extremely damaging, and/or drastically life-shortening implications of certain types of decisions. Even where information is complete and accurate, a decision to sell oneself into slavery, commit suicide, consume heroin, or otherwise to expose oneself to unreasonable risk of extreme damage may, at least prima facie, be presumed to be irrational and contrary to the individual’s real interests. In most such cases, to be sure, public intervention becomes compel- ling only because of associated problems of impulsiveness, lack of infor- mation, immaturity, etc., which remain even under the most comprehen- sive (feasible) policies of information provision and custodial choice. It may accordingly very seldom be necessary at the practical policy level to confront in their pure form the more controversial issues raised at the conceptual level by the possible rejection of a fully informed and consid- ered decision to consume. However this may be, it is nevertheless clear that merit wants policies at this second level are considerably more difficult to reconcile with the consumer sovereignty notion since a truly voluntary self-paternalistic delegation of decision making authority is hardly conceivable. At least in the more routine cases of custodial choice, however, this merit wants function could hardly be described as unduly controversial or beyond the scope of a broad concept of individual sovereignty reasonably inter- preted. Level 3: Merit Wants and Ethical Preferences As already suggested, the extension of the domain of the narrowly Pigovian categories of subjective preferences or “tastes” in the standard economist’s parlance to the realm of ethical preferences, though chal- 240 John G. Head lenging and more controversial, is crucial for a comprehensive analysis and taxonomy of merit wants. Reserving for later discussion the applica- tion of the Rawlsian framework in the development of some specific examples, we shall focus here on the more general phenomenon of conflict between the individual’s tastes and moral values. Thus, for example, it is a matter of common observation that, for those of a particular religious, philosophical, or ideological persuasion, certain types of market choices may pose genuine moral problems. Familiar examples in the area of medical procedures and health care would include abortion, blood transfusions, in vitro fertilization, and euthanasia. Prostitution, pornography, and racial or sexual discrimina- tion provide additional instances. In all these cases it could clearly be argued that the individual’s lower-order market choices may require correction in the light of his or her own higher-order preferences or moral values. A further top-level category of merit wants policies can therefore be distinguished which could help give effect to the individual’s ethical preferences. Here again the basis for public intervention is not authoritarian but rather self-paternalistic and rests firmly on commonly observed behavioral aberrations and cognitive problems analyzed pre- viously under the general heading of impulse or weakness of will. Difficulties arise, as in lower-level examples, because some types of government measures of a merit wants character, such as legal prohibi- tion, must typically be universal or nondiscriminating. The practical issue may be further complicated by the fact that those of the relevant religious, philosophical, or ideological persuasion may be no less con- cerned to impose their own ethical preferences on the rest of the commu- nity than they are to achieve the benefits of self-paternalism. At the implementation level authoritarian elements may thus intrude and may be difficult to disentangle from legitimately self-paternalistic policies. The same problem, however, arises under imperfect real-world political processes in the provision for generalized social wants. Utility Interdependence To an economist much the most obvious way to broaden the traditional consumer sovereignty framework to embrace the individual’s social con- cerns or ethical preferences is to extend the standard Pigovian externality concept from spillovers of ordinary goods and services to the realm of Merit Wants: Analysis and Taxonomy psychic externalities or utility interdependence (Arrow 1951; Buchanan 1968). In this setting it becomes possible to accommodate even some of the most elitist and authoritarian merit wants notions. Thus, for exam- ple, the desires of some individuals and groups to impose their own moral values upon society can be explained and justified in terms of their altruistic concern for the welfare or moral well-being of others or in terms of the disutility they suffer as a result of others’ immoral behavior. Just as markets fail in the presence of “tangible” externalities, they may fail also due to the existence of psychic externalities. In this wide individ- ualistic framework it is tempting to suggest that many of the most challenging, value-laden issues of social policy, in areas such as abortion, prostitution, racial and sexual discrimination, surrogate motherhood, etc., could be analyzed and resolved by means of this generalized exter- nality concept. A further important category of merit wants policies can therefore be distinguished based on psychic externalities.® Although tempting, at least to an economist, this extension of exter- nality analysis remains, however, in some respects extremely problem- atic. Many would no doubt reject an analysis in which ethical preferences are simply equated with tastes as a source of possible Pareto-relevant (psychic) externalities. What people might be able or willing to pay in order to reduce immoral behavior in others may well be regarded as an inadequate criterion for policy where moral values are involved. At the other extreme it might be questioned whether any policy weight whatsoever should be attached, even where willingness to pay exists, to psychic externalities reflecting one individual’s views of an- other person’s preferences. It has, for example, been pointed out by Sen (1970) that interventionist policies based on the existence of such “nosy” or “meddlesome” preferences may conflict with liberal values. In a liberal democracy there may be general agreement that nosy preferences, however well intentioned, should simply be ruled out, and market choices of the type in question should remain a matter of individual conscience. In many cases of such nosy preferences it is moreover possi- ble to question very seriously the motives involved, which may range widely from the pure altruism of a truly moral preference to much less wholesome concerns such as envy, malice, and sadism (Sen 1976, 1979). Even those who would have no hesitation in acknowledging the policy relevance of psychic externalities where the motives are generally well intentioned may wish to draw the line in cases such as envy, malice, and schadenfreude. Even if utility interdependence is admitted into the 241 242 John G. Head framework of the analysis, a considerable measure of value criticism or “preference laundering” may be required (Goodin 1986). It is at this point that the Rawlsian analysis comes into its own in providing an appropriate basis for the evaluation and correction of preferences, including ethical preferences, in a liberal democracy. Ex- treme aberrations such as sadism are, for example, rather easily ruled out by the presuppositions of liberal democracy such as the requirement of equal respect, and the central Rawlsian notion of a “disinterested prefer- ence” as conceived in the “original position” under the “veil of igno- rance” provides a useful standard which can be applied in some of the more difficult and controversial cases. Distributional Issues The possible uses and limitations of the utility interdependence concept and the potential contribution of the Rawlsian framework are nicely illustrated by developments in the analysis of other functions of a merit wants character, notably the redistribution function which had been handled by Musgrave (1959) through assignment to a separate “branch” in which, by implication, individual preferences might need to be over- ridden. Cash transfers. Hochman and Rodgers (1969) showed that income redistribution can be justified within an individualistic framework if the rich are altruistically concerned about the incomes or welfare of the poor (see also Brennan 1975; Head 1966). Where, for example, large num- bers of middle and upper-income people are concerned with the welfare of those below some basic poverty line, redistribution becomes a public good and private charity can provide no adequate solution. Provided the appropriate conditions are met, such “general” utility interdependence can be used to explain and justify public redistributive programs of a guaranteed minimum income character and can help to account for the widespread acceptance of far-reaching welfare state measures over the postwar years. The nature and extent of the general utility interdependence phenom- enon is, however, an empirical question about which little is yet known. In a given society it may be that little if any redistribution could be justi- fied in this way. A normatively adequate treatment of the distribution branch function might therefore seem to require much stronger value judgments of an elitist or authoritarian character. Musgrave (1970) Merit Wants: Analysis and Taxonomy points out that “primary” as well as “secondary” (or Pareto-optimal) redistribution may be required. In relation to the basic preference correc- tion characteristic of merit wants, value criticism is implied of the rela- tively small weight attached to the incomes of the poor in the preference functions of the better off whose preferences may be distorted as a result of prejudice and self-interest. In the face of such obvious limitations of the utility interdependence concept, the Rawlsian approach offers in effect a standard for value criticism and an alternative individualistic basis for the redistributive merit wants function. In Rawls at the constitutional level a “primary distribution” is established based on individual preferences corrected for such distorting influences as prejudice and self-interest by a veil-of- ignorance construct. This construct requires individuals to behave as if they had no knowledge of any special factors which might determine their incomes under a market system in the postconstitutional setting. Emphasis is placed on the observation that, if individuals are risk averse, a high floor under the income distribution might be established by unanimous agreement.” In-kind redistribution. For Musgrave the redistributive in-kind sub- sidy represents in many ways the archetypal example of a merit wants policy overlapping the distributional and allocation branch functions. Here again the psychic externalities concept in the specific form of “particular commodity interdependence,” stemming from original con- tributions by Buchanan (1968) and Olsen (1969), appears potentially very useful in offering an individualistic foundation for what might otherwise seem to be a coercive or paternalistic function. Under particu- lar commodity interdependence, utility of the better off depends not on the incomes or utility of the poor but rather on their consumption of commodities of special distributional significance, such as health, educa- tion, or housing. In this context a standard Pigovian in-kind subsidy can serve not only the distribution branch objective of vertical redistribu- tion, but also the allocation branch objective of efficiency through inter- nalizing the psychic externality involved. It would appear therefore that Musgrave’s merit wants example of the redistributive in-kind subsidy can easily be reconciled with the individualistic or consumer sovereignty framework, a point also emphasized by Musgrave (1969:143-44; 1987). Such policies are hardly more “paternalistic” than the in-kind subsidies or free provision strategies routinely advocated for standard externalities or social wants. 243 244 John G. Head More recent analysis by Brennan (1975:248—50) and Brennan and Walsh (1977) clearly shows, however, that redistribution of income, whether in cash or in kind, cannot generally be justified in this way, since the psychic externality involved can perfectly well be internalized with- out the need for any redistributive benefit to the low-income direct consumer. Indeed, in a simple two-commodity setting it can easily be seen that the allocation branch objective can equally well be achieved by taxing low-income consumption of the “other commodity” such as liquor as by subsidizing consumption of merit wants such as education or housing. At most we could say that where both forms of utility interdependence, general and particular commodity, are simultaneously operative, the redistributive in-kind subsidy might be employed as a package measure combining the distinct and separate distribution and allocation branch functions involved, any redistributive component being justified, however, solely in terms of general utility interdepend- ence. In the face of these further limitations of the utility interdependence approach, it is interesting to observe that the Rawlsian framework offers once again a more viable approach. It is clearly true that cash transfers provide the preferred instrument for the establishment of the “primary distribution.” However, a significant measure of in-kind redistribution can also be justified under the Rawlsian conception of “primary goods.” This concept extends well beyond endowments of income and wealth to a variety of specific rights and opportunities, with important implica- tions in such areas as health and education.® Particular emphasis on the distribution of some of the major Musgravian merit goods may therefore be appropriate on grounds of equality of opportunity, viewed from a constitutional perspective under the veil of ignorance. Recently, Mus- grave has placed increasing emphasis on this categorical equity concept which appears much less vulnerable to objections of the sort raised above.? Preference Endogeneity Although the existence of utility interdependence need not imply prefer- ence endogeneity and is therefore consistent in principle with the con- ventional welfare framework, it is nevertheless obvious that people are preeminently social animals and individual preferences are heavily con- Merit Wants: Analysis and Taxonomy ditioned by the social environment. Since the 1930s economists such as Knight (1935), Galbraith (1958), and Scitovsky (1964) have attempted to highlight the phenomenon and importance of preference endogeneity. By and large, however, these observations have had little impact on mainstream economic analysis.!° It has, however, been generally recog- nized that, if individual preferences are to a significant degree influenced by the process of social interaction, the normative authority of the consumer sovereignty ideal is seriously affected and may need substan- tial modification. Thus, for example, at the lowest “D-level” of desires or inclinations, individual preferences are clearly influenced by the behavior and opin- ions of others. Indeed the influence of fashion is so great that decisions are in a very real sense delegated to “leaders” and “opinion makers” in the relevant areas of choice. This influence is used by economists such as Galbraith and Scitovsky to criticize the naive consumer sovereignty principle. In the limit, as Galbraith (1958) points out, people simply buy what they are persuaded to desire and any autonomous concept of individual preferences appears completely impossible. Merit wants pol- icies designed to protect the individual from some of the more blatant forms of intentional preference manipulation, notably in the area of advertising, could clearly be justified in this way and are easily reconcil- able with the broadened consumer sovereignty notion. Such problems must, however, be sharply distinguished from the more general phenom- enon of preference endogeneity where individual preferences may be affected unintentionally in the process of social interaction. Such endo- geneity does not raise difficulties for a normative framework based on a broad interpretation of consumer sovereignty. The phenomenon of preference endogeneity is clearly no less impor- tant in the analysis of the individual’s ethical preferences or moral values. Views on major issues of social or economic concern evolve in the historical process of social interaction and are transmitted or reinforced within the family and through more formal educational processes in the schools and in a variety of other community organizations. Musgrave (1987) gives particular prominence to such “community values” as the basis for a further specific category of merit wants policies for the preservation of historic sites, the observance of national holidays, pro- tection of the environment, and the promotion of respect for learning and the arts.!! Musgrave stresses the conservative and conservationist character of such policies, rooted as they are in the history of society. He 245 246 John G. Head observes that these policies may be accepted without question even by those who would not embrace the particular values they reflect. It is indeed a fundamental observation in recent interdisciplinary analysis by Sugden (1986) and others that “community values” or “ethi- cal preferences,” including concepts of justice or fairness, evolve in the historical process largely as a rational community response to a variety of challenging coordination and free rider problems. Major features of the “common culture,” including the basic institutions and associated value structures embodied or reflected in a political and legal system, common language or common religious heritage, can accordingly be quite prop- erly characterized as public capital which serves as an important input in the provision by society for the satisfaction of social wants. !2 Policies to “entrench” or preserve important aspects of the common culture against the possibility of impulsive, ill-considered, or myopic behavior, including majority decisions, by members of the current generation may therefore be supported virtually unanimously in the appropriate constitutional or quasi-constitutional perspective without resort to any “organic” concept of community. A major category of merit wants policies for the preserva- tion or conservation of public capital in the form of the common culture can accordingly be distinguished which extends far beyond the relatively mundane examples cited by Musgrave under the general heading of his community values concept. As in the more narrowly “Pigovian” exam- ples of preceding sections, the merit wants problem arises from the obvious danger, due to lack of reflection or weakness of will, that large numbers of people may fail to recognize the multiperiod character and future implications of certain major types of social choices. Preference endogeneity is, however, a slippery concept. Some care is therfore necessary if we are to be sure that the specific institutions and associated value structures which emerge ostensibly as a rational com- munity response to various coordination and free rider problems do not at the same time reflect significant elements of arbitrariness, inequity, and discrimination. Here again some of the more significant insights for purposes of merit wants analysis come from the broader interdiscipli- nary literature. Thus, for example, it has been pointed out that our basic institutions and value structures may have evolved under historic power relationships which are totally arbitrary and inequitable. The concept of “adaptive preferences,” developed in the work of Elster (1983) and others, serves to remind us that existing preferences may have come about through social and educational processes which help condition the re Merit Wants: Analysis and Taxonomy individual to accept certain types of deprivation or discrimination. The well-known preference endogeneity phenomenon of “sour grapes,” highlighted by Elster, clearly provides a potentially very powerful expla- nation of acceptance by the victim of traditional forms of sexual, racial, or religious discrimination. The rationalization and acceptance of such distinctions can serve to reduce cognitive dissonance as individuals rec- oncile themselves to unpleasant situations which they may be powerless to alter. Similarly, prevailing “community values,” as articulated by the beneficiaries of discrimination under the evolutionary status quo, may well reflect a manifestly self-serving “preference for discrimination.” Institutionalized discrimination may thus be “generally accepted” (in the sense of Musgrave), and there may seem little case on standard individ- ual sovereignty criteria for affirmative action or related policies to achieve equal rights. A proper appreciation of the preference endogeneity phenomenon of adaptive preferences clearly serves, however, to undermine the norma- tive authority of the simple consumer sovereignty notion in such cases. The “values” and “preferences” which support discrimination are re- vealed as morally suspect and, in the case of the beneficiaries, painfully self-interested. An important category of merit wants policies can ac- cordingly be distinguished based on value criticism of endogenous pref- erences which are manifestly “adaptive” and reflect the results of social conditioning under existing and totally arbitrary power relationships. It certainly appears reasonable to argue that antidiscrimination policies find an important part of their justification under this general heading. Since, however, all preferences including “ethical preferences” are to a significant degree endogenous, this approach would appear to pose serious conceptual and practical problems. Less ambiguous standards for value criticism are clearly required, and the Rawlsian framework based on disinterested choice behind the veil of ignorance would seem here again to provide a much firmer theoretical foundation for a merit wants category of antidiscrimination policies. Political Provision for Merit Wants The preceding section has reviewed three possible dimensions in which the standard consumer sovereignty or New Welfare framework of the 1950s can be broadened in the light of both earlier and more recent work 247 248 John G. Head across a range of disciplines. This analysis strongly suggests that a much wider range of policies of an apparently paternalistic character can be justified within a broad individualistic normative framework by refer- ence to the concept of psychic externalities or, more importantly, in terms of the existence of a hierarchy of individual preference orderings in combination with impulsiveness or weakness of will. It is, of course, a matter of common observation that individual preferences appear fre- quently to be overridden in political processes. It might accordingly be inferred that, where markets fail in the provision for merit wants, politi- cal processes should lend themselves rather well to the merit wants function of preference correction. Modeling by economists of democratic political decision making pro- cesses began in earnest during the 1950s, not long before Musgrave’s original formulation of the merit wants notion. A characteristic feature of these efforts has been the assumption that individuals will vote instru- mentally in political processes in order to achieve satisfaction of their market preferences (Downs 1957). A major concern of this and the subsequent public choice literature has been to avoid any romantic presumption that political processes lend themselves uniquely to the expression of higher values or ethical preferences. !3 In the traditional public finance literature the argument that the pref- erence orderings which find expression in voting behavior differ system- atically from those which motivate market behavior is perhaps most familiar from the work of Gerhard Colm (1955, 1956, 1960, 1965). Colm and the so-called public interest theories of politics maintain that individuals take broader and longer views in democratic political deci- sion making processes and that expressed preferences accordingly reflect the individual’s conception of the good society. The argument that mar- ket and political processes are simply alternative mechanisms which are available to the individual for the satisfaction of the same overt or ex ante preference ordering is completely rejected; for example, Colm (1965:213) notes that “[a]s there is no strict comparability between the preference scales of two individuals, there is also no cogent relation- ship between the preference scales of the same individual as homo oeconomicus and homo politicus.” Analyses by Brennan and Buchanan (1984) and Brennan and Loma- sky (1983) of the dominant median voter model, which underlies the Downs analysis, in fact lend considerable support to Colm’s position. Since the likelihood that an individual’s vote could affect electoral out- Merit Wants: Analysis and Taxonomy 249 comes becomes very small in large number electorates, they argue that individuals will vote expressively rather than instrumentally in demo- cratic political processes. This peculiarity of the voting process serves in effect to reduce dramatically the opportunity cost of expressing higher- order preferences reflecting broader or longer views and community val- ues. Self-paternalistic merit wants policies reflecting higher-order prefer- ences may therefore find expression through democratic political pro- cesses. Such outcomes, although not desired by individual voters, result unintentionally from expressive voting under conditions in which the incentive to vote instrumentally is reduced to negligible proportions. It would, however, be misleading to infer from this analysis that majority voting in a democratic political setting provides an ideally efficient mechanism for implementing merit wants policies. Apparently coercive or paternalistic policies may simply reflect familiar aberrations of political processes such as majority exploitation, bureaucratic distor- tions, rent seeking or Leviathan tendencies. It is also obvious that expres- sive voting may equally well reflect preferences which are racist, xeno- phobic, impulsive, or shortsighted. Although there may be considerable benefits from democratic discussion and deliberation, it would clearly be absurd to equate the hurly-burly of real world democracy with some Rawlsian ideal of reflective and disinterested choice in an “original position” under a “veil of ignorance.” What these and related insights from the literature on political deci- sion making suggest is that some provision for merit wants is likely through democratic political processes, accompanied however by a vari- ety of unpleasant aberrations, outright coercion, and inefficiency. The possibilities for achieving more efficient outcomes through the political process depend, as Buchanan (1975) has emphasized, on the ability of politicians and voters to distinguish, in the hurly-burly of postconstitu- tional political decision making, those issues which are of a multiperiod or quasi-constitutional character where an appropriately long-term per- spective and a properly reflective attitude are required. Concluding Reflections Central to the analysis of the merit wants concept, as we have seen, are issues of paternalism which sit uneasily both with the traditional notion of consumer sovereignty and with the conventional normative frame- 250 John G. Head work of welfare economics. This paper has tried to show how a range of apparently coercive policies can be reconciled with a broad concept of individual sovereignty which encompasses both psychic externalities and, in particular, the notion of self-paternalism developed within the framework of a hierarchy of preferences and based on behavioral aber- rations such as impulse or lack of reflection and weakness of will. The illustrations, although developed only briefly, are drawn from most of the major areas of government spending. They strongly suggest a poten- tially very broad area of application for the Musgravian merit wants concept based on the premise of individual choice in a democratic society. Indeed it would therefore appear that the scope and importance of such a merit wants concept are fully comparable to that of its more prominent rival, the externality or generalized social wants concept. The major public policies and public services most commonly cited as examples of merit want issues are, however, complex and multidimen- sional. It is seldom a simple matter—indeed it is frequently almost impossible—to disentangle the uniquely merit wants aspects from asso- ciated social wants aspects (Walsh 1988). Health and education services, for example, raise issues of custodial choice, Pigovian shortsightedness and self-paternalism, externalities (both tangible and psychic), commu- nity values, paternalistic giving, and categorical equity. Much the same is true in other public policy areas. Given the economist’s natural hesita- tion to complicate the analysis with notions which smack of paternalism or elitism, these further dimensions have been very largely ignored. Where an explanation in terms of conventional social wants or Pigo- vian externalities appears possible, why bother with the problematic complications of merit wants analysis? The obvious answer is that the case for public intervention is frequently very much stronger than might be inferred from the “thin theory” developed on the basis of the Pigovian externalities concept. The standard economic analysis of major public services and regulatory policies does not therefore do justice to the richness and complexity of the issues involved. The same is clearly true of the economist’s occasional forays into the analysis of controversial social policy issues such as abortion, prostitution, pornography, and discrimination, not to mention the whole question of income distribu- tion. It has accordingly been a major purpose of this paper to show how developments in related disciplines and subdisciplines can be used to enrich greatly the analysis of important public policy issues with signifi- Merit Wants: Analysis and Taxonomy 251 cant merit wants dimensions. The concept of merit wants simply cannot be developed comprehensively and systematically without the broader perspective offered by a multidisciplinary approach. The analysis pre- sented here is of course suggestive rather than definitive, but may serve to draw the attention of economists to an array of important but tradi- tionally neglected policy issues. One of the great disappointments of modern public expenditure anal- ysis has been the dearth of relevant empirical research. Very little has, for example, been achieved in estimating the externalities or publicness of services such as health or education. In the conservative political and social environment of the 1980s it has become common for economists to argue, in the absence of relevant empirical evidence, that a wide range of existing public services are essentially private in character and should be returned to the market. Such arguments are no doubt facile, even on their own terms. The extra dimensions of the merit wants concept should serve, however, to remind us that there is more, and frequently much more, to the case for public intervention than can properly be encom- passed by the social wants or Pigovian externality concept. Notes 1. Musgrave (1959:49) particularly emphasizes the redistributive in-kind subsidy as an example of a merit wants policy overlapping the allocation and distribution branches. 2. See McLure (1968), Andel (1969), Head (1969), Schmidt (1970), Culyer (1971), Auld and Bing (1972), Pulsipher (1971-72), Pazner (1972), Ballentine (1972-73), Braulke (1972-73), Buttler (1973), Folkers (1974), Roskamp (1975), Basu (1975-76), Charles and Westaway (1981), Wenzel and Wiegard (1981), Brennan and Lomasky (1983), and Andel (1984). 3. See Pigou (1928, chap. 8; 1932, chap. 2). The possible role of these distinctions in developing a broader and more relevant welfare economics has been emphasized by Bergson (1966). 4. Lack of information as a possible basis for merit wants policies was particularly emphasized in Head (1966, 1969). 5. See, for example, Head (1966:5), McLure (1968:481), Head (1969:219— 20), Auld and Bing (1971-72), Ballentine (1972-73), and Charles and Westa- way (1981). 6. The possible role of psychic externalities in the analysis of the merit wants concept was first emphasized in Tiebout and Houston (1962). See also Head (1966:15—16). 252 John G. Head 7. Abstracting from possible disincentive effects and assuming that all indi- viduals are risk averse, unanimous agreement could be achieved on the desir- ability of a completely equal distribution of income. This observation goes back to Samuelson (1963—64). More recent debate has focused on the Rawlsian maximin principle as the appropriate criterion for trade-off between equity and incentive aspects. See Rawls (1971). 8. On the definition of primary goods, see Rawls (1971:62). On implica- tions in the specific area of education, see Rawls (1971:101, 107, 275). 9. See, for example, Musgrave and Musgrave (1984:79, 99-100) and R. Musgrave (1987). See also Tobin (1970) and, in the context of the early merit wants discussion, Head (1966:7-8). 10. Significant exceptions include Duesenberry (1949) on the consumption function and, within welfare economics, more recent work based on Carl von Weiszacker (1971). 11. In the evolution of Musgrave’s views on merit wants, this particular category is first discussed under the heading of “communal wants” and with a much stronger “organic” flavor in Musgrave and Musgrave (1980:83—87). 12. For an analysis of the law as public capital, see Buchanan (1975, Chapter 7). 13. For example, Head (1966) applies the Downs model and concludes, counterintuitively, that provision for merit wants through the democratic politi- cal process is problematic and unlikely. Comment ENID SLACK Retrospectives on Local Public Goods The Eden-McMillan paper on local public goods is truly a retrospective in that it begins with Shoup’s views on local public goods (in particular, fire and police services) and then reviews the subsequent theoretical and empirical literature on this topic. The authors conclude that Shoup’s findings remain central to the discussion on local public goods and, further, that the subsequent literature has not progressed very far. The four issues highlighted by Shoup (1969b) are reviewed in the paper: the definition and measurement of output, the characteristics of public goods, cost functions, and the distribution and incidence of benefits. The definition and measurement of output is the critical first step in any empirical work on local public goods. Early research on output used data on inputs as a proxy, that is, expenditures on police and fire services were used to measure output. Not surprisingly, this measure produced unsatisfactory results and was replaced by other proxies for output: in the case of police, for example, crime reduction was used. In more recent studies the output variable has been disaggregated into its component parts, recognizing, for example, that the output of crime prevention is different than the output of traffic control or community and social services. Eden and McMillan recognize that more work needs to be done on the definition and measurement of output to include the quality of service as well as relevant institutional knowledge of the public goods being measured. Most of the literature on local public goods has focused on the characteristics of public goods: nonrivalness in consumption and nonex- cludability. Carl Shoup simply assumed that fire and police are rival in consumption without empirical evidence. Subsequent empirical testing to determine the presence of these characteristics has resulted in largely inconclusive evidence: the characteristics of public goods depend on the definition and measure of output used. The third issue concerns the estimation of cost functions. In the empirical studies the costs of fire and police services are estimated as functions of population, area size, level of service, and other variables. 254 Enid Slack As with the characteristics of local public goods, the evidence on econo- mies of scale in the provision of these services is inconclusive and de- pends crucially on the output measure used. The literature addresses, to a lesser extent, the distribution and inci- dence of benefits of local public goods. The incidence of benefits by income class, property, population, and other variables has been esti- mated by several authors. Again, the results are inconclusive in terms of whether the distribution of local public goods is progressive or regres- sive. Another aspect of distribution of particular interest to Shoup con- cerns how to allocate funds available for police services across service units on the basis of various criteria such as equal crime rates in each area, equal marginal costs of crime prevention, and the tangency of the production and the social welfare trade-offs among service areas. Again, research on this aspect of local public goods is still relatively under- developed. In addition to the four topics that Shoup addresses and that Eden and McMillan review, at least three other policy issues merit attention. First, there is the issue of the optimal size of governmental unit to provide services such as police and fire. Economies of scale, discussed by Eden and McMillan, is only one criterion for determining the optimal size of local jurisdiction. Others include accountability, demand considera- tions, and externalities. A second issue, and one in which the literature is sparse, is the efficiency of service delivery. How can the provision of local public goods be more efficient? Some literature exists on privatization and on the use of volunteers, but again, to do a proper study requires information on expenditures per unit of output. A third issue concerns not only the distribution of the benefits but also how local public goods, such as fire and police services, will be financed. There is a more substantial literature on alternative methods of finance including income taxes, property taxes, lot levies, and other revenue sources. In summary Eden and McMillan note that not much advancement has been made in the literature on local public goods, largely because of data and measurement problems. The evidence that does exist is gener- ally inconclusive and, in most cases, depends on the researcher’s choice of output measure. The authors conclude the paper on a positive note: the lack of progress to date on research into local public goods provides interesting research opportunities for economists. Comment EDWIN G. WEST Merit Wants and Public Goods Theory Of all aspects of John Head’s interesting paper, to me the most arresting is his discussion of issues of political implementation of a merit goods policy. He warns us that the political process is a suspect means to such an end. He observes that the voting process can serve as the vehicle for the expression of preferences which are “racist, xenophobic, bigoted, short-sighted, and extremely ill-considered.” If we are to respond to his request for further research and detailed analysis of interdisciplinary character in order to determine whether or not the merit wants concept is an empty box, we must surely keep in mind this strong possibility of government failure. There has always been a danger that, when examining the real world of government intervention, public finance specialists will attempt to use received theories to rationalize every aspect of government intervention to the neglect of institutional structure. The large parts of the govern- ment budget that are devoted to defense are most readily and plausibly explained by application of the Samuelson public goods theory. Addi- tional forms of intervention that exist are then justified in terms of externalities and much of this too seems convincing. But at the end of the day there remains evidence of forms of government activity that are not so easily explained by public goods and externality theory. One reason, indeed, is because we encounter the public provision of private goods. If a benevolent government is assumed, then the public finance theorist will naturally seek a benign rationalization for these “residual activities.” For instance government may allegedly be interested in improving the qual- ity of certain private goods and services, and its entry into production might be explained on these grounds. It may be that the idea of a merit wants policy originally emerged from this sequence of thought. As is now well known, however, the newer economics of politics, or public choice, challenges the benevolent government assumption at every opportunity. We now have to consider what John Head calls “the dark side” of public choice theory which is represented by supply-driven models based on the role of bureaucracies and the importance of rent- 256 Edwin G. West seeking or revenue maximization motives. Had our original public fi- nance specialists possessed the advantage of the later contributions of public choice analysis, it would no doubt have more readily occurred to them that there were hypotheses in addition to the merit wants idea that might be even more plausible as an explanation of what I have just called the “residual government activities.” John Head mentions the health sector as one in which a merit wants intervention seems to him to be fairly obvious. One aspect of this inter- vention is to encourage state run or nonprofit hospitals. This policy is usually argued to be necessary in order to protect consumers from making impaired or uninformed choices. Recent research suggests, how- ever, that it is the supply interests that have the most influence in encour- aging governments to undertake such policies. Doctors, for instance, prefer nonprofit hospitals because they can better influence their direc- tion and share their net revenues (Clark 1980). It is interesting that the typical fee-for-service-payment system in such hospitals allows physi- cians to earn much more per hour when they work in nonprofit or state hospitals than when they work in their offices (Sloan 1988). Empirical evidence by Herzlinger and Skrasker (1987) also supports the proposi- tion that doctors are the main financial gainers from nonprofit hospitals. What I am arguing, therefore, is that the newer theories of institu- tional organizations are suggesting hypotheses that compete with that of merit wants. So if we are to take seriously John Head’s invitation to undertake more research, each stated example of a merit wants policy should be tested empirically alongside an alternative hypotheses from other branches of economics. Toward the beginning of his paper John Head notes that strong tendencies to irrational behavior frequently emerge, especially in circum- stances of great uncertainty. In several of these cases, he suggests, some delegation of decision making to government may be appropriate. The public choice theorists’ reaction to such argument is once more to focus on the institution called government that is expected to correct things. They will ask the direct question: If individuals facing great uncertainty tend to make impaired choices, will not this same impediment reappear when they come to vote for politicians who will eventually form the government that is expected to correct things? John Head’s reasoning about impaired decision making envisages particularly those situations in which individuals have to make choices that are complex and long- term. A prime example of such choices, however, is that of selection of a Merit Wants and Public Goods Theory political representative for, say, the next five years. This being so, the argument for surrendering decision making to government is in danger of becoming circular. With regard to John Head’s concept of self-paternalism, the proposi- tion is that individuals voluntarily hand over future-oriented decisions to government in order to constrain their lower-order desires, for these are based on “weakness of the will.” His favorite example, and one that is used also by Richard Musgrave, is Pigou’s claim that, because of myopia, future consumption tends to be undervalued relative to present con- sumption. This argument, which is described by Pigou in terms of what he called “the deficiency of the telescopic faculty,” originated in Bohm- Bawerk’s second reason for the existence of a rate of interest. To accept this reasoning, however, simply because of Pigou’s personal opinion that the market rate of discount is too high is surely unwarranted. Moreover there are a number of arguments that have since been advanced against the alleged myopia (Blaug 1985, p. 503). In addition, if, in fact, governments are found to be employing a social rate of discount that is lower than the market rate, public choice theo- rists might explain this event in terms of the newer economics of bu- reaucracy: If, as is likely, the imposed social rate of discount has the effect of increasing the bureau’s budget, then the fact that this serves the purpose of furthering the bureaucrat’s own self-interest provides us with a potential alternative explanation of events. The fact remains, mean- while, that if people are myopic in their long-term decisions, there seems nothing to guarantee that this deficiency will not also show itself at the ballot box. In Richard Musgrave’s article on merit goods in the Palgrave Diction- ary (1987) there appears one example that, in his view, “goes to the heart of merit concept.” This is the situation where individuals, as members of the community, “accept certain community values and preferences, even though their personal preferences might differ.” What is called for here is an empirical study to test whether the personal preferences of individuals do in fact differ significantly from those in the community. Musgrave mentions examples such as concern for maintenance of an historical sites, regard for the environment, or for learning and the arts. In some recent research that I conducted for Ontario (West 1985, chap. 9) I found that many taxpayers were willing, within limits, to have themselves taxed to support the performing arts even though they did not them- selves attend the theater. On confronting the same individuals with the 257 258 Edwin G. West apparent paradox in their position, I elicited further explanations. Most of these, however, were in terms of the external benefits that the individ- uals expected from the taxes they were agreeing to. It is true, however, that this was only weak evidence and that some other nonusers declared there was no benefit that they could perceive. Could it be argued that the attitude of this second group might be consistent with the merit good reasoning that Richard Musgrave is referring to? Additional research based on more specifically designed questionnaires is obviously needed. Further reflection by Musgrave (1987) on some of his original exam- ples of merit goods now persuades him that they are not fully appropri- ate because they are ultimately consistent with consumer sovereignty. This judgment is made with respect, for instance, to cases where rational choice requires correct information. The case of education is one exam- ple, and I would like to dwell on it for a while since I have done some work in this area. When teaching public finance in England in the 1960s, and using Richard Musgrave’s great new treatise, first published in 1959, I was struck by the particular words he used on the subject of education: “The advantages of education are more evident to the informed than to the uninformed, thus justifying compulsion in the allocation of resources to education” (Musgrave 1959:14). This sentiment appeared to me to echo that of another important writer of the previous century, John Stuart Mill. Written in 1848, Mill’s words were as follows: “The uncultivated cannot be competent judges of cultivation. Those who most need to be made wiser and better, usually desire it least” (Mill 1969:953). This connection with utilitarianism is interesting. In Musgrave (1987), second thoughts led Musgrave to conclude that intervention in education is consistent with the dominance of individual preference at the normative level because the beneficiary’s own prefer- ence is ultimately improved “by initial delegation of choice to others whose prior information is superior.” This again was the same position held by John Stuart Mill and, more explicitly, by his colleague Nassau Senior (West and McKee 1983:1114). The implication is that interven- tion should be self-liquidating. Choice is to be imposed temporarily; however, as preferences are improved, the imposition will be withdrawn because it will have become superfluous. If this is intended to be a scientific proposition, we should be informed about the approximate time frame that is relevant to the problem. If we are given a prediction which is not falsifiable within a specified period of time, we can never Merit Wants and Public Goods Theory 259 falsify the theory because at every future moment we would be urged to be more patient. We have already had about a century of universal free and compulsory education, and there seem no signs that the authorities are going to terminate their intervention. If the theory of temporary intervention has failed in this example, we obviously need an alternative hypothesis to help explain the continuance of public education. Perhaps, after all, education reveals a merit good policy in Musgrave’s strongest sense where imposed preference rules the day. Once more, however, the public choice theory also provides an explanation. There are strong reasons to believe that the perpetuation of education intervention in its present form is due to the political strength of organized teachers and administrators. It may be contended that free and compulsory public education started out as an initial delegation of choice to others while the interest group obtrusion occurred later in the form of an entrenched monopoly that prevents the policy from being self-liquidating. The historical evi- dence, however, shows that the predominant campaigners for free and compulsory education were the teachers’ associations (West 1967). The rent-seekers, in other words, were active right from the beginning of public education. To come now to the paper by Cliff Walsh. This piece of course is not involved with John Head’s multiple preference theory. That is to say, it is not concerned with the higher preference orderings but confines itself to the conventional ex ante utility indicators of revealed preference. To this extent the analysis is relatively easy. But the uncovering of hitherto unexpected differences between public goods with and without exclu- sion brings with it new levels of intellectual interest and sophistication. The special feature of his important paper is that it concentrates entirely on the market provision of public goods. And, as he observes, this is a prerequisite for the development of a case for public interven- tion. And since Walsh does not pretend to have produced the final definitive analysis, the full and comprehensive case for public interven- tion presumably also awaits further research. At the end of Walsh’s paper his readers, having been equipped with a variety of interesting market provision models, will be desirous to know where to go next with them. It would certainly be helpful if Cliff Walsh could first help us to see which model fits which example out of the intriguing list in his opening paragraph. Take from this list, for instance, the examples of performances of plays, operas, or symphony concerts 260 Edwin G. West for which admission is restricted to ticket holders. Readers will be interested to know which of these situations fit the monopoly model of public goods with exclusion and which situation fits the competitive model. One would have thought, indeed, that monopolistic competition was a relevant market structure here. Although monopolistic competi- tion is a model that Walsh’s analysis at present does not include, it would seem worthy of attention in further development of his argument. With regard to the last section of the paper which compares the efficiency of provision without exclusion to provision with, the analysis is fascinating as an intellectual exercise. It seems intrinsically, however, to be unoperational. Starting with a scenario where market provision is without exclusion, we are in a world where, by definition, the pattern of demand curves is unknown. It is impossible, therefore, simultaneously to compare this model with one where exclusion is present because this assumes that some knowledge of demand curves is known. These two worlds cannot coexist. One has to wait until Walsh’s last footnote to learn that he is not going to jump into the controversial matter of the comparison of public provi- sion compared with market provision of public goods. Yet some of his reasoning is already pertinent to this issue. His argument, for instance, that zero output of public goods is not a reasonable long-run equilibrium prediction of market provision, implies that market failure is not as severe as was once thought. But when we couple this with the new public choice emphasis on arguments about government failure, Walsh’s point becomes a new and relevant contribution to an area of ongoing lively debate. It is to be hoped that he will soon participate in it more explicitly. VV Macro Public Finance 13. RICHARD M. BIRD Tax Structure and the Growth of Government It has long been evident that in some sense the growth of govern- ment is limited by the possibility of raising taxes. The pioneering analysis of government expenditure growth by Peacock and Wiseman (1967), for example, stressed the constraints imposed on expenditure by the fact that it was hard to raise tax rates significantly in the absence of some great socioeconomic disturbance such as a war or a revolution. Gupta (1967) later added a third horseman, major economic depression, to this apocalyptic team, noting both that Keynesian-influenced governments might spend more, not less, in a depression and that with a falling national income even a constant expenditure level would yield a higher ratio of expenditure to national income. That there is considerable merit in this approach is suggested by experiences such as that in Nicaragua where, after the Sandinista revolution, the share of government in the GDP rose rapidly to 40 percent in the course of five years—aided, it must be said, by a major economic depression which reduced Gpp substan- tially (Bird 1985). As Peacock and Wiseman (1967) also noted, even in the absence of such major crises, the tax ratio could keep growing, as it were, invisibly so long as the economy grew and the tax structure was elastic. The introduction of a progressive mass income tax in the 1940s undoubtedly provided revenue fuel for the rapid expansion of public spending that took place after the war in most Western countries. The importance of this factor may be illustrated also by Canadian experience after 1974, when the indexation of the personal income tax substantially reduced the elasticity of the tax structure by removing much of the revenue response to the marked price level changes of the period. Not entirely coincidentally, the growth of the expenditure ratio also slowed down and even reversed. The apparent link between expenditures and revenues was of course clear long before the arrival of the income tax. As Gillespie (1988) notes, 264 Richard M. Bird for example, in Canada at the turn of the last century, when rapidly growing foreign trade boosted tariff revenues substantially the increased revenues were soon matched by increased expenditures. Similar budget- ary sensitivity to trade flows is evident in the many developing countries which remain heavily dependent on trade taxes for revenues. Tax structure, like the factors that determine it such as the openness of the economy (Cameron 1978), thus appear to be related to government growth. Changes in tax technology such as the introduction of withhold- ing during World War II or the indexation just mentioned may have an important effect on the growth of tax revenues and hence, over time, on the growth of government expenditures. In an important sense, however, attributing observed trends in gov- ernment growth to changes in tax technology or tax structure can at best be a half-truth. In democratic states people must in the end get the (size of) government they want, unless one is prepared to believe in some sort of perpetual fiscal illusion. Is it really believable that taxpayers will con- sistently underestimate the true tax prices facing them and hence con- tinue indefinitely to support an excessively large and growing govern- ment? As Oates (1988a) has argued, most “illusion” arguments seem to be both theoretically weak and empirically unsupported. What may per- haps be called the “Barnum” theory of public sector growth—“there’s a sucker born every minute”—seems less plausible than the contrary “Lin- coln” theory that “You can fool all of the people some of the time, and some of the people all of the time, but you cannot fool all of the people all of the time.” Such, in any case, is the basic argument for democracy as a survival strategy for modern states. The theme of this volume is “retrospectives on public finance.” Twenty years ago, when | first wrote on the subject of public expenditure growth (Bird 1970), although recognizing the importance of revenue elasticity in fueling incremental expenditure growth, I nonetheless argued that it was legitimate to study expenditures alone without reference to tax con- straints because, in the end, organized states with modern tax admin- istrations could—at least within the ranges then experienced—finance whatever level of expenditure they wanted. In short, I ended up agreeing with Adolf Wagner (1958:8), the nineteenth-century pioneer of studies of government growth, that “in the long run the desire for development of a progressive people will always overcome these financial difficulties.” Looking back, both Wagner and I were perhaps precipitate in our judg- ments: revenues probably do constrain expenditures, just as the man in Tax Structure and Government Growth 265 the street has always thought. In addition, however, I shall argue below that expenditures also constrain revenues. This conference is also to honor Carl Shoup, so it is only fitting to note that he of course has long been aware of the link between taxes and expenditures. Indeed, Shoup (1981) says that the proposition that ex- penditure and tax levels and structures are interdependently determined is so obvious as to be not worth stating! So it may be. Nevertheless, even some recent studies on the growth of government have not taken this fact adequately into account and have therefore missed what appears to be a crucial connection between recent changes in taxes and expenditures in OECD countries. As one recent review (Diamond and Tait 1988:26—27) concluded: “It is a sad commentary on progress in this field that almost a century after Wagner’s initial speculations, and after almost three dec- ades of extensive empirical research, we are only at the stage of guessing at the mechanisms that affect the transformation inside the [black] box [of public expenditure growth].” The limited purpose of the present paper is to sketch what may perhaps be a few of the connections between taxation and expenditure lurking inside that box. Tax Reform in the OECD Tax reform is in the air. Not only Canadians and Americans, but also Britons, Danes, New Zealanders, Japanese, even Frenchmen are doing it—or at least talking about it. Indeed, according to recent press reports on the possible revival in the Soviet Union of progressive income taxes— once thought to have been safely “buried” by Nikita Khruschev—even Russians may be poring through their dictionaries trying to decipher the meaning of the strange phrase “tax reform.” Tax reform—or, to avoid prejudging the issue, major tax change— has thus been the subject of active public discussion in many countries in recent years, with special emphasis on changes in the personal income tax. Of eleven major country reforms discussed in a recent OECD (1987) study, for example, ten proposed reductions in personal income tax rates. The 1986 American tax reform has of course received the most publicity and in some ways has had most impact on others both by force of example and by the economic weight the United States still carries in the world. The U.S. pattern of personal income tax rate cuts and base broadening (including full taxation of capital gains) accompanied by 266 Richard M. Bird similar changes in the corporate tax, but with some net shift in tax burden from the personal to the corporate level, has been repeated to varying degrees in many countries. In 1987, for example, Canada followed the United States with respect to corporate tax reform and in 1988 to some extent with respect to personal income tax reform. One big difference between the two coun- tries, however, is that Canada plans to shift its federal sales tax to a value-added tax in 1991. The United States, of course, does not have a federal sales tax in the first place. Britain preceded rather than followed the United States down the tax reform road. Indeed, the beginning of the current wave of tax reform lies not in Washington but London, where Mrs. Thatcher’s government led the way in 1979 with personal tax cuts and a shift to a value-added tax, followed in 1984 by a corporate rate cut and base broadening very similar to the 1986 American pattern noted above. Australia too had a personal rate cut and a base broadening reform (including an innovative tax on fringe benefits, which remain sacred in most countries) in 1986. With respect to corporate taxes, however, Australia went in precisely the opposite direction to the United States, first narrowing the tax base greatly (by introducing a full dividend tax credit system) and raising the corporate tax rate, although the latter move was later reversed. Like Canada, Australia has a preretail stage federal sales tax which makes tax experts unhappy. However, proposal for a national retail sales tax died a quick death in 1985. In contrast, New Zealand did practically everything Australia did in the tax area and more besides as its socialist government dazzled citizens with a brilliant display of left-wing foreign policy and right-wing eco- nomic policy including, to a much greater extent than any other OECD country, a shift in the tax mix toward greater reliance on the value-added tax, corporate tax changes similar to those in Australia, a much lower personal income tax, and (again as in Australia) a fringe benefit tax. Even outside the English-speaking world, tax reform has been in the air almost everywhere. In Japan, for example, first the corporate tax rate was cut; then the personal tax base was broadened (in precisely the opposite direction to that urged by expenditure tax advocates, namely, by eliminating savings deductions) and rates cut. Moreover, after dec- ades of discussion and in the face of considerable political dissent a low- rate (3 percent) value-added tax was introduced in 1989. Nothing quite as dramatic has happened in most continental Euro- pean countries, although there have been many significant tax changes in Tax Structure and Government Growth recent years. Two such changes worth noting are the move in the Nether- lands to closer integration of personal income tax and payroll tax (in effect moving most personal taxpayers to a virtually flat wage tax) and the move in Denmark to a flat-rate schedular tax for capital income, with progressive rates only on labor income. Although the flat rate is a high 50 percent, the latter change is especially striking in the Scandinavian heart- land of welfare statism. Five points emerge from this rapid tour of the complex tax reform picture in developed countries (see also Pechman 1988). First, tax reform has focused on the personal income tax and espe- cially on specious simplification in the form of reducing the number of rate brackets and genuine lowering of the top rate, with the average top rate coming down from 60 to 50 percent in the OECD as a whole— although interestingly, less so in the European community where the rates were (and are) higher (OECD 1987). Outside of the English-speak- ing countries, however, there has been little emphasis on offsetting base- broadening changes. Secondly, corporate nominal rates have also been reduced from 46 to 40 percent on average (OECD 1987). Outside of the United States, Can- ada, and Britain, however, there have been few attempts to increase the base of corporate taxes. Indeed, Australia and New Zealand have actu- ally reduced the corporate tax base substantially. Thirdly, the value-added tax played a central role in tax reform in Britain, Japan, Canada, and especially New Zealand. In both Japan and Canada, however, public discontent with the resulting increase (prospec- tive in Canada’s case) in the visibility of national sales taxes continues to be manifest. The value-added tax has not been an issue recently in the European community, of course, since all member countries already have such taxes. In Australia, after arousing bitter opposition, a pro- posed retail sales tax was dropped in 1985. Only in the United States has national sales taxation so far remained beyond the pale. Fourthly, only in New Zealand are the recent reforms likely to result in much change in the tax mix. Not one country has yet moved in the direction of either direct personal expenditure taxation or corporate cash flow taxation, although proposals along these lines have dominated recent academic literature on tax reform. On the contrary, as already noted, tax reforms in several countries have moved more in the direction of the older normative standard for “fair” taxation—the comprehensive income tax. Finally, in almost every case the result of tax reform has been more a 267 268 Richard M. Bird change in the composition of tax revenues than an increase in tax revenue. Even where taxes have increased, the additional revenues have served more to reduce deficits than to increase expenditures. In Canada, for example, taxes rose from 31 to 33 percent of GpP in the 1982—87 period (and total revenues from 40 to 41 percent), while expenditures fell from 47 to 45 percent of Gpp (Canada, Department of Finance 1988). A similar slowdown and even reversal of government expendi- ture growth in the 1980s is visible in a number of other OECD countries also. From Tax Reform to Government Expenditure Growth Tax reform is a political, not an economic, process. It results from the interplay of interests and actors characteristic of the political process rather than the application of the “rational man” (or “benevolent dicta- tor”) approach that underlies the conventional economic analysis of tax reform. The same is of course true with respect to expenditure change (which, interestingly, is seldom labeled “reform’”). Both tax reform and expenditure change can thus best be understood as a process of change in what may be called “political equilibrium” (Hettich and Winer 1988). From this perspective the tax reforms of the 1980s must have reflected much the same forces as those that resulted in the marked slowing down of the pace of government expansion in the same period. The five propositions which constitute the balance of this chapter suggest some of the ways in which tax reform and expenditure growth may be linked. Proposition 1: The Tax Structure is “OQuasi-Constitutional” in Nature This straightforward extension of the notion of political equilibrium implies that the way in which the cost of government is shared among different groups in any society is such a central part of the political “balance” that major changes in it are unlikely to come about easily (Head and Bird 1983). As recent Canadian budget experience suggests, it would appear that much the same can be said of changes in expendi- tures. The wartime adoption of mass income taxes, the postwar growth of payroll-financed social security schemes, and the adoption of mass consumption taxes were the major tax changes of the last fifty years or so in most countries (Bird 1988a). As the Peacock-Wiseman argument Tax Structure and Government Growth suggests, in most countries these changes accompanied major political events—some endogenous and some exogenous. Although it is by no means clear that the tax reforms summarized above are of equal impor- tance, it seems likely (as developed further in Proposition 3 below) that the same can be said of them. Thinking of tax reform in these terms also suggests that federal states—or other states (such as Japan) requiring a high degree of con- sensus before action—may find it harder to introduce major tax changes and therefore to raise revenues (and expenditures) than unitary states. The relatively slow move to value-added taxation in Australia, Canada, the United States, and Japan suggests there may be something to this proposition. So does the generally smaller size of the public sector in these countries compared to other OECD countries. However, in one of the most recent econometric explorations of this theme, Saunders (1988) found the federalism variable to be insignificant—a result which Breton (1989) takes to support his theoretical argument that the influence of a federal structure on the growth of government should be indeterminate. This first proposition also emphasizes the need to see the tax system as a whole. The tax system seems less a villain in the political play than a victim: it does not so much determine the nature of social and political outcomes as it reflects them in the fiscal arena—an arena in which, from time to time, the crowd demands a human sacrifice, usually in the form of an unfortunate finance minister but sometimes a whole government. Governments therefore tend to walk very carefully through the tax reform minefield and to introduce changes only when “the time is ripe.” “Ripeness” may be deemed to occur, for example, when tax changes can be slipped by because people are distracted by crisis (Proposition 3), when enough people want some expenditures to be willing to pay for them (Proposition 4), or perhaps when politically important groups view the change largely in favorable ideological or symbolic terms (Proposi- tion 2). The first of these possibilities is exemplified by the wartime adoption of withholding, the second by social security finance, and the third, perhaps, by the latest round of tax reforms. Such at least is one conclusion that might tentatively be drawn from the striking U.S. evi- dence on the changing perceptions of the fairness of the personal income tax (ACIR 1988) and the emphasis in all countries on reducing income tax rates. In the eyes of some prominent experts the outcome of recent attempts to reform the complex, inequitable, ineffective, and distortionary income 269 270 Richard M. Bird tax appears to have been to leave us with a distortionary, ineffective, inequitable, and complex income tax. One of the original architects of the recent American reform, Charles McLure (1988), has argued that while the 1986 reform probably moved the system as close to a proper income tax as politically possible, its results are such that instead of proceeding further in that direction future reformers should turn to the consumption tax instead. A similarly pessimistic look at tax reform experience in Britain argues that twenty years of so-called reform have resulted in a tax system that is still inequitable, complicated, and distorting (Sandford, 1988). The whole experience is, according to Sandford, depressing; and so indeed it must be to anyone who views the outcome of reform as most academic commentators seem to do, in terms of how it accords with a precon- ceived normative standard of what an ideal tax system should look like. But of course this is not necessarily the most meaningful way in which to view tax reform. Henry Simons (1938:219) once referred, dispar- agingly, to the loophole-ridden progressive personal income tax as amounting to little more than “dipping deeply . . . with a sieve.” This theme has since been much developed and much lamented under the label of “tax expenditure” analysis (e.g., Surrey and McDaniel 1985). It may be, however, that both the dip (progressivity) and the sieve (loopholes) are essential ingredients of fiscal balance in a democracy. Joseph Schumpeter (1954:769), in many ways the father of the new political economy, once said that “nothing shows so clearly the character of a society and of a civilization as does the fiscal policy its political sector adopts.” As Proposition 2 suggests, he may well have been right. Proposition 2: The Tax Structure is an Important Political Symbol Economic analysis quite properly stresses the instrumental role of taxa- tion in achieving such goals as equity and efficiency. But tax structure may also in an important sense be an end in itself. Webber and Wildav- sky (1986:526), for instance, labeled the income tax “a mirror of democ- racy” in the sense that over the years this tax in its role of correcting the more egregious outcomes of the market has come to symbolize the strength of egalitarianism and commitment to social justice in a society. Different societies may, of course, attach different weights and inter- pretations to this symbol, and these weights may change over time. The slowing down of economic growth may, for example, reduce the price Tax Structure and Government Growth 271 people are willing to pay for more egalitarianism. The very success of the welfare state in abolishing the worst aspects of poverty in some devel- oped countries may reduce the perceived value of redistribution through the tax system. And if, as has been the case in recent years in some countries, both of these factors are at work, it is perhaps not so surpris- ing that income tax structures have tended to flatten out (and wealth taxes to wither away). That the same thing has happened in the United States—clearly a country which comes near the tail of the welfare state parade among developed countries—can perhaps be explained by the much stronger persistence there than in most countries of the belief that each man (or woman or child?) is ultimately the master of his own fate and responsible for himself. Despite recent trends, however, it seems unlikely that the character of modern democracies has yet changed to the extent that they will no longer levy income taxes, however increasingly imperfect a redistributive instrument such taxes may have become. The crisis of confidence in government that has precipitated the revival of market-oriented policies in recent years has probably not yet penetrated deeply enough to reverse the long-standing perception that taxation is primarily about fairness, not efficiency. The reinforcement of the income tax and the rejection of the expenditure tax in the recent reforms appears to offer some support for this position. Finally, a quite different sense in which tax reform may have as much symbolic as instrumental value is that the high public profile of the income tax in most countries makes it an obvious lightning rod for fiscal discontent, however caused. If income growth slows, people complain about the income tax. If they value government services less than before for any reason, they complain about the income tax. If both of these things happen simultaneously, as has been suggested was the case in Canada at the end of the 1970s (Bird 1982a), they complain even more. This point is further developed in the fourth and fifth propositions set out below. Proposition 3: Tax Structure Change Results Primarily From Exogenous Shocks To some extent the first two propositions fit awkwardly within the framework of the new economics of politics because they come uncom- fortably close to talking about changes in “tastes” or preferences, a 272 Richard M. Bird subject with which economists always have trouble. In contrast, this third proposition, taken directly from Hettich and Winer (1988), falls squarely within that framework. Taking tastes and political institutions as given, outcomes will change only when the system is shocked so that the marginal cost (political, economic, administrative) of raising rev- enues—or spending them—in one way rather than another changes. There have been at least two such shocks that may in this sense lie behind the recent wave of tax changes in most countries: the end of the era of rapid postwar growth and the growing internationalization of the world economy. The first of these was already mentioned in connection with Proposition 2. In any case growth may revive—and, indeed, already has to some extent. Short of catastrophe, however, world integration seems unlikely to be reversed soon. The present discussion therefore focuses on the “shock” to the political-economic foundations of tax and expendi- ture structure as a result of the increasing international mobility of capital. The openness of advanced economies to international capital flows and the sensitivity of such flows to tax factors greatly constrain the freedom of policy makers to decide on their own tax system. No country exists in splendid isolation; each is part of the world—and the world is beyond the control of any nation, even the largest. To get the design of taxes right in this international context is both conceptually and prac- tically difficult. At the conceptual level many conflicting factors need to be balanced, and in fact one cannot have an “optimal” policy for all countries simultaneously (Gersovitz 1987). There is therefore much room for intercountry bargaining. At the practical level most of the information needed to implement any tax policy with respect to interna- tional income flows must in the end come from taxpayers (since, by definition, their jurisdictional span exceeds that of the governments trying to tax them), and hence to a certain extent there is always bargain- ing between taxpayers and authority. In this game investors are so far outplaying tax authorities with the result that the effective taxation of international capital flows seems likely to spiral down close to a least common denominator of zero (Bird 1988b). As Bird and McLure (forthcoming) argue, the future of not only the corporate income tax but also the personal income tax on non-wage income seems bleak in the absence of an unlikely degree of coordination among nation states—coordination of the sort that, it must be noted, should in any case be viewed with distrust by believers in the state as Tax Structure and Government Growth 273 Leviathan (Brennan and Buchanan 1980). Consideration of the interna- tional dimension of tax reforms is not likely to make one sanguine about the future of progressive personal income taxes in an increasingly inte- grated world. As Lee and McKenzie (1989) put a similar argument, increased international capital mobility seems to be one important rea- son why marginal income tax rates have been coming down everywhere in recent years. Proposition 4: Tax Structure is Linked to Expenditure Structure This proposition may be developed in the form of three linked argu- ments. First, if, as argued above, tax structure is part of the political balancing equation, its precise nature is dependent, inter alia, on expen- ditures. For example, regionally based direct expenditures may sub- stitute for regional tax incentives or intergovernmental transfers and vice versa—and often do (Bird 1982b). What remains unchanged as a rule is the overall net regional flow of public sector resources. This phenome- non is perhaps clearest in the context of federal states, but it is by no means limited to such states. Secondly, and probably more controversially, tax reform may be linked to changing perceptions of what may be called “taxpayer sur- plus,” that is, the perceived surplus accruing to taxpayers in the form of benefits from state action in excess of perceived tax costs. Changes in this surplus, which may lead to tax revolts and pressure for tax reform, may clearly result from changes in the composition of expenditures, not taxes. A crude index of taxpayer surplus developed for Canada in Bird (1982a) suggests that the pressure for tax reform in the early 1980s, in part at least, may have resulted from the diminution in the real level of such voter-desired services as health and education partly owing to the growth in the 1970s of (less desired) social welfare transfers. The num- bers fit. Whether the argument is valid is of course another matter since, as usual with such broad issues, the problem is not that we cannot explain what happened but that there are too many plausible explana- tions that cannot be distinguished. A related proposition has been as- serted in a long-run historical context by Paul Kennedy (1988), who emphasizes the dampening effect of the growing military expenditures of hegemonic states on their economic and taxable capacity. Finally, a third clear linkage between tax and expenditure structure is that taxes earmarked to desired expenditures are more likely to increase 274 Richard M. Bird over time than general fund taxes. This proposition, originally put for- ward by James Buchanan (1963), has been tested a number of times (e.g., Deran 1965; Eklund 1969) and seems to hold up pretty well, although of course more empirical work is, as always, needed. This argument seems especially relevant in explaining the apparently inexorable rise in most developed countries of payroll taxes linked to social security finance. Proposition 5: Tax Structure is Linked to the Growth of Government The apparent robustness in the face of fiscal adversity of payroll tax— financed social security systems is one piece of evidence for this proposi- tion; a second, quite different, might be the well-known “value-added tax as a money machine” argument, which appears to have some merit in the sense that once a broad-based relatively efficient sales tax is in place not only does its revenue yield tend to be elastic but its rate tends over time to move upward (Tait 1988). One reason for this finding—again related to propositions put forward earlier by Buchanan (1967) and the Italian scholar Puviani (not to mention Machiavelli)—may be that gov- ernments find it easier to raise invisible than visible taxes—and once the initial struggle is over, value-added taxes seem in most countries to have become virtually invisible. The current furor in Canada over replacing a hidden federal sales tax—which has been incrementally raised by 50 percent (from a standard rate of 9 percent to one of 13.5 percent) over the last few years, with no perceptible political backlash—by a (presum- ably) more visible one suggests there may be something in this argument. All arguments, however, based on illusion are, as noted earlier, suspect both conceptually and empirically (Oates 1988a). Of course, Buchanan and his illustrious predecessors are also closely related to what has come to be known as the Leviathan view of govern- ment—the argument that government revenues, and hence expendi- tures, will expand relentlessly unless constrained by the adoption of political or fiscal institutions such as balanced budgets or narrow-based and inelastic taxes (Brennan and Buchanan 1980). The more drastic propositions sometimes associated with this school of thought seem overdone, however (Musgrave 1981). Bureaucrats are, for example, constrained by politicians, and politicians are constrained both by the need to retain support over time and by the design of such political institutions as the length of time between elections and the nature of Tax Structure and Government Growth political financing (Breton 1974). Such factors therefore also clearly influence the outcome of both tax and expenditure systems. In turn, and in much the same way, the tax structure adopted will at least to some extent govern the growth of government. Wagner (1958), like Bird (1970) and many others, in effect argued that causation flowed from desired expenditures to acceptable revenues. The public choice school at times has it almost the other way, with revenue acceptability governing expenditure levels (Mueller 1987). In fact, as Breton (1989) has emphasized, both the supply and the demand blades of the scissors seem likely to be needed to cut this Gordian knot. Much work remains to be done before all the mysteries of tax and expenditure change can be unraveled. What seems undeniable, however, is that, as Carl Shoup (1981) in effect told us some years ago, under- standing tax reform and understanding the growth of government ex- penditure are different facets of the same problem. Considering either tax or expenditure change in isolation is not likely to advance much our understanding of what makes the public sector tick. Considering their interaction, perhaps by developing further some of the preliminary lines of inquiry sketched above, may in the end prove a more productive approach to this important question. 275 4, PEGGY, B. MUSGRAVE Fiscal Coordination and Competition in an International Setting Throughout modern history problems have arisen from the taxa- tion of goods, labor, and capital which move across jurisdictions. These problems have involved such questions as the entitlement to tax by one jurisdiction of income, products or wealth which belong to residents of another, or which have been created in another; questions of fair treat- ment of taxpayers resident in one place but earning income or wealth in another have also been raised; and there has been concern with the effects of multijurisdictional taxation on trade, capital, and labor move- ments across regions, localities, and nations. Within federal systems such as the United States, Canada, Australia, the Federal Republic of Ger- many, and Switzerland, these issues have been largely resolved by consti- tutional or statutory rules designed to coordinate subnational fiscal systems under the overarching and unifying influence of central govern- ment finance. At the international level fiscal economists have tradi- tionally sought means by which these conflicts of interest and possible impediments to commerce and factor flows might be resolved by interna- tional agreement. For instance, after World War I, Professor Seligman (still teaching at Columbia in Carl Shoup’s early years there) along with Professor Einaudi of Italy and Sir Josiah Stamp of Great Britain played leading roles in formulating principles for tax entitlements and coordi- nation which were influential in shaping the format of international tax treaties sponsored by the League of Nations (League of Nations 1923). Following World War II the United Nations and the oEcD have partici- pated in promoting multilateral agreements on international tax treaty provisions. A central objective of the European Economic Community has been coordination of the differing fiscal systems of member countries to allow a neutral fiscal environment in which goods and factors could freely move within that Common Market. Professors Carl Shoup and Fiscal Coordination and Competition Fritz Neumark were two key contributors in laying the basis for move- ment to a coordinated system which many hope will reach fruition in 1992. Similar coordinating efforts have been made by groups of nations in other parts of the world (P. Musgrave 1986). The term “harmoniza- tion” of fiscal systems has been the theme of that work, rather than “equalization,” recognizing the desire of member countries to implement an equitable system of taxation of their residents and to retain as much of their individual fiscal characteristics as possible while neutralizing effects of those differentials on the free flow of trade, labor, capital, and people within the Common Market (Shoup 1967). This approach of fiscal harmonization or coordination through inter- national cooperation, with which Carl Shoup has been identified, rests on the premise that the decision process leading to tax and expenditure decisions within each country is more or less efficient or that fiscal competition is not the appropriate way to correct for domestic inefficien- cies as may exist. This premise has recently come under challenge, based on the following argument:! As commodity, labor, and capital markets open up one to another, superior public sector performance will be rewarded by attracting resources, residents, and trade, just as the supe- rior performance of firms in the market results in increased profits. This process of competition will thus promote efficiency and responsiveness of governments. According to this view, harmonization may be coun- terproductive, since it insulates government budgets from the disci- pline of international competition, and may be likened to collusion of firms in the private sector. Fiscal competition, rather than coopera- tion, should be the keyword. The following quote typifies this point of view: For those of us who believe that free markets guarantee the highest possible standard of living, the words cooperation and coordination ring like euphemisms for collusion against market outcomes and sound a threat to a proven source of lasting prosperity. . . . Similarly, the increased ease with which manufacturing and financial firms can move about the globe places a check on regulation and taxation. Simply stated, greater international inter- dependence increases the opportunities for investors and traders to protect their wealth from the misguided policies of individual countries. (Hoskins 1989) Seen from this perspective the application of normative fiscal rules as traditionally developed becomes invalid and is replaced by a very dif- ferent set of requirements. UT 278 Peggy B. Musgrave The purpose of this paper is to examine the roles to be played by fiscal coordination and competition. The problem has many dimensions which are difficult to sort out. In seeking a workable approach, Part I begins with the special case of benefit taxation in the context of local finance and the relevancy of the Tiebout model in the broader context of national finances in an international setting. We then turn to the role of competition and coordination in the latter setting. Part II considers major policy goals of and obstacles to an efficient fiscal process linking diverse national jurisdictions. Part III goes on to discuss the implications of fiscal competition in the absence of coordinating arrangements. It assumes first that each jurisdiction conducts its fiscal affairs in a more or less efficient fashion so that there is no need for using fiscal competition as a remedial device and then allows for imperfections in the domestic conduct of fiscal affairs and considers the potential uses of international fiscal relationships for remedial purposes. Part IV concludes with an examination of the requirements for and implications of fiscal coordina- tion. Fiscal Competition and the Special Case of Local Benefit Taxation The efficiency conditions for the provision of public goods as given in Samuelson’s basic paper (1954) apply to a single jurisdiction and assume the benefits of public services to be available therein. The question of how this outcome might be achieved through a voting system was first raised by Wicksell (1896), expanded upon by Lindahl’s system of benefit taxation (1919), and then explored further by Downs (1956) and the emerging theory of public choice. Tiebout’s theory of local public goods (1957) offered a first examination of how the argument may be applied to the case of multiple local jurisdictions. With the benefits of public goods limited to a particular local region, efficient provision calls for individuals with homogeneous public good preferences to congregate in particular regions. Moreover, this result is brought about by the ten- dency of individuals to move so as to seek a congenial preference en- vironment. Alternatively, one may think of residents of any one region as “competing” for equal-preference individuals. Mobility of residents and competition in this sense thus serves as an ordering device in securing an efficient outcome. Fiscal Coordination and Competition 279 Limitations of Tiebout Model While this vision of competition as an ordering device offered a challeng- ing new insight, it has since been recognized that it is subject to many limitations (Zodrow and Mieszkowski 1986). Pure benefit taxation ap- plies, and the movement of residents is assumed to be costless and governed by the search for fiscal advantage in the provision of public goods only. The model in its simplest form also assumes equal incomes. Once income inequalities are allowed for, but still retaining the assump- tion of benefit taxation, low-income individuals will find it to their advantage to seek coresidency with high-income individuals, especially if the latter share the same preferences for public goods. A more complex and potentially inefficient pattern of movement then results, and even more so if the assumption of benefit taxation is replaced by an income tax and other forms of general taxation, a problem given numerical illustration in the Appendix. Finally, and most important, the spirit of benefit taxation which is crucial to the Tiebout model largely rules out allowance for redistributional uses of tax and expenditure (transfer) policies. Only to a limited extent can such adjustments be dealt with as based on the spatial dimension of utility interdependence (Pauly 1972). Introduction of redistributional policies thus leads to the poor pursuing the rich, the rich trying to exclude the poor, resulting in unstable equi- libria which greatly complicate the simple pattern of the original Tiebout model (R. Musgrave 1959). Nevertheless, the basic idea of the Tiebout model retains considerable merit. Benefits from the provision of most public goods are spatially limited and preferences differ. Thus there is good reason for provision of public goods differing in spatial scope to be provided by a multiple set of governments. Local, regional, and national goods are to be financed in each case by tax contributions of their respective residents. At a formal level this argument can be extended to a set of national governments, capped finally by a worldwide unit designed to deal with services having a global benefit scope. In principle the valuable sights of the Tiebout model can thus be extended to the international level. At the same time there are serious limitations to such extension. The weight carried by fiscal gains of the Tiebout type is obviously much less when it comes to residence choice among nations than when it comes to the choice among suburban local governments grouped around a central place of employ- 280 Peggy B. Musgrave ment. In addition the underlying assumption of benefit taxation becomes wholly unrealistic. For the case of local governments it is not unreason- able to neglect redistributional concerns, especially if these are assumed to be taken care of at the national level. But national governments cannot do so. Taxation in line with ability to pay is called for and distributional considerations cannot in practice be separated from the choice of public services.? Finally, and of crucial importance for tax coordination once capital income is allowed for, it can no longer be assumed that an individual when moving carries his/her resources along. The international setting is thus concerned not only with the mobility of residents but also with the mobility of economic resources and capital in particular. The principle that residents of each location should pay for the public services available in that location—as it were, an interregional benefit rule—continues to make good sense in the national setting. But its implementation is now greatly complicated because the flow of re- sources may be affected. For these reasons it is evident that the role of fiscal competition for residents as an ordering device, while meaningful in the simple Tiebout model, largely disappears in a more realistic view of the international case. Policy Goals and Obstacles We begin with the assumption that each nation manages to expedite citizens’ preferences regarding public sector issues through its demo- cratic process, including the input of voters, legislators, and the civil service. The question then is how this supportive view of the public sector can be translated into the international setting so as to permit the participating nations to seek their respective fiscal objectives. Policy Goals Proceeding in this spirit, and reaching back to the Seligman-Shoup tradition, the task is to provide a setting which will permit each jurisdic- tion to reflect the preferences of its own electorate and to do so without interfering with other jurisdictions and efficient resource flows in the private sector. The problems that arise are multifold, and there are many objectives involved in the choice of particular jurisdictional relation- ships, some calling for conflicting policies. Fiscal Coordination and Competition 281 Efficiency of Public Services Efficiency in the provision of public services requires that each govern- ment provides a level and mix of public goods consistent with the preferences of its residents and that such provision be paid for by them. Tax burdens are not to be shifted abroad, and domestic-source income which accrues to foreign investors is to be burdened only to the extent that benefits are received through the provision of intermediate public goods or, as will be discussed below, in compliance with other agreed upon rules of internation equity. In the special case of public goods, the benefits of which spill across national borders, an international rule by which appropriate cost shares are assigned in line with benefit shares is required. Allocative Efficiency in the Private Sector With efficient resource use by the public sector thus provided for, a further policy goal is to avoid interference with efficient resource use by the private sector within and across boundaries. As noted before, this calls for a neutral fiscal environment in which goods and services can be exchanged and factors of production move across boundaries in ways which are not dictated by fiscal concerns. This requires the consumers of any unit to pay the same tax on the goods which they consume whether produced in their own jurisdiction or another, and it also requires that investors and workers should face the same effective tax rates on their returns whether they choose to invest or work at home or abroad.? In the absence of such fiscal neutrality the flow of goods and the location of production will be diverted from their efficient pattern. With capital the most mobile factor, the tax treatment of capital income and in particular the role of the corporation income tax are of special impor- tance in this context. Nor should fiscal differentials affect the location of residency and thus introduce a further interference with individual choice. Internation Equity Next, fiscal arrangements should be such as to provide for internation equity (Musgrave and Musgrave 1972). A clear understanding of this concept and how it differs from that of interpersonal equity is important in sorting out alternative tax arrangements. The question at issue is to 282 Peggy B. Musgrave determine to what extent a country is entitled to tax income of foreign residents or of foreign corporations operating within its boundaries. International agreements such as the GATT preclude the use of fiscal instruments to exploit other jurisdictions through turning the terms of trade or of factor prices by way of exporting tax burdens or gaining balance of payment advantages. At the same time it is reasonable to permit the country of income source or product origin to take a “fair” tax share in the income or product value arising within it, even though owned or consumed by foreign residents. The appropriate magnitude of that share depends on an international view of property rights and standards of internation equity and must be set as a matter of multiunit consensus.* Interresident Equity As distinct from internation equity which is subject to agreement among nations, any one jurisdiction will wish to implement its own concept of interpersonal equity. In the absence of benefit taxation this requires that people in equal positions (capacity to pay) should pay the same tax and for payments among unequals to differ in line with what each electorate may consider a fair distribution of shares. This pattern tends to be interfered with in an open economy setting since trade and factor move- ments among jurisdictions raise the possibility of either over- or under- taxation of the taxpayers of a particular unit. This means that residents of one unit who choose to work, invest, or consume in another may be subject to higher or lower tax burdens than those who choose to do so within their own country. As a result, the effect on the distribution of tax liabilities among persons becomes a major concern as do resulting effects on factor movements. It is generally agreed that residents of one country who choose to transact abroad should pay the same as their fellow residents who do not. The question then arises how this can be imple- mented, a matter to be discussed below. Differential Mobility Difficulties which arise in reconciling these goals with the coexistence of different fiscal systems are attributable in large part to the interjurisdic- tional mobility of goods, factors, and residents/citizens. This mobility can take many forms, each reflecting varying responses to differentials in fiscal variables. The impact of this mobility on the public sectors and Fiscal Coordination and Competition 283 resource bases of the jurisdictions concerned will differ accordingly. Mobility of factors of production or of consumers (whether through trade or direct purchase by individuals outside their jurisdictions of residence) may provide opportunities for tax avoidance and these oppor- tunities differ with differentials in mobility. Favorable tax treatment given to individuals who choose to earn their income abroad or to consume abroad thus serves to enhance those opportunities. Capital Financial capital is perhaps the most mobile of resources, followed by real, or direct, investment. Indeed, the growing mobility of such capital has transformed international economic relationships in recent decades. Technology is transferable: either embodied in the direct investment, in skilled management, or by other contractual means. Both forms of investment flows, but particularly that of direct investment, carry with them embodied forms of capital which arise from public sector benefits and social externalities conferred by the “home” country and enhance the productivity of that capital when invested in other jurisdictions. Capital is not only the most mobile factor of production, but it is separable from its ownership. Shareholders may reside in one jurisdic- tion while investing their wealth abroad where it is subject to the fiscal system in that source jurisdiction, or a corporation may be registered, controlled, and managed in one country and invest in another. This feature potentially allows individuals or corporations to avail themselves of public services in their country of residence (incorporation) but to avoid payment of taxes therein. A further characteristic of capital is that it is generally put to work by businesses which (as multinational corpo- rations) may operate in a number of jurisdictions simultaneously. This is particularly the case, of course, in a federal setting. For these various reasons the fiscal treatment of capital and capital income poses a (if not the) major problem in a multijurisdictional setting. Labor Fiscal effects on labor migration are most likely to arise from differentials in income and payroll taxes and in transfer payments such as social security and income support programs. But these effects are less severe than those on the movement of capital. As a factor of production, labor services differ from those of capital in that mobility of the service gener- 284 Peggy B. Musgrave ally requires an accompanying mobility of the persons performing the service. In border areas, such as that between the United States and Canada, people can fairly readily reside in one jurisdiction and work in the other, but more generally a change of work venue must be accom- panied by change of residency. The necessity for changing residency obviously serves as a barrier to changing place of work, a barrier involv- ing psychological as well as economic costs of moving. This is one reason why labor is much less mobile than is capital, which can move from one jurisdiction to another without the necessity for its owners changing residence. In the case of international labor mobility very large differen- tials in real wages are generally needed to induce mobility of labor, examples being the movement of Mexican labor into the United States or the “brain drain” of skilled professionals from the developing countries. Of course, barriers to labor mobility will be much stronger in the case of movement from one country to another than from one jurisdiction to another within the same national boundaries. As to other economic effects, the labor-importing country will experi- ence an increase in its supply of labor and (given full employment) will result in an increase in national income and tax base, but it may also be confronted with an increased need for public services and transfers. For countries with underemployed labor, the loss of relatively unskilled labor may have little effect on domestic product, indeed remittances out of higher wages earned abroad may more than compensate for that loss. The more skilled the workers, however, the greater is likely to be the loss. The loss of educated people is in fact a loss of human capital, often accumulated at public expense through publicly financed education. More than purely economic considerations are involved, including the question of on what basis the tax allegiance of individuals or their rights to transfers should be made to depend. Consumers Next consider effects of fiscal differentials on the location of consump- tion. Populations living in border locations may choose to purchase consumer goods or utilize public facilities outside their own jurisdiction. This is a problem not only for local jurisdictions but also internationally as in the Common Market context. Furthermore, and more broadly, a related problem arises with respect to product mobility, where consumer choice may be influenced by the tax treatment according to origin of product. Fiscal Coordination and Competition 285 Residents Finally, there is residence mobility wherein those who are not bound to their place of work have the choice of location. These would include retired persons and recipients of capital income. Fiscal differentials can provide strong incentives to residence mobility. As in the case of labor mobility, it raises questions of the basis for entitlement to tax, for example, whether it be residence and/or citizenship. This poses the question whether primary tax allegiance should be based on both resi- dence and citizenship, so that a taxpayer continues to bear a tax obliga- tion to the country of citizenship even though resident in another. These are philosophical questions which have for long presented problems in international taxation. Given these various forms of mobility, and especially the much greater mobility of capital, the question is how to secure an outcome which (1) secures an efficient provision of public services by foreclosing free riding, that is, preventing residents from enjoying the benefits of public services or to partake in distributional goals without contributing to their share of the cost; (2) ensures a fair distribution of shared tax bases among nations; (3) provides for a fair interpersonal distribution of the tax burden among residents of each jurisdiction; and (4) leaves the efficient operation of factor and commodity flows in the private sector unaffected. As we shall see, this solution is not resolved by forces of fiscal competition, but requires a complex pattern of coordination. Fiscal Competition Without Coordination Consider first a multinational setting without any fiscal coordination. Each jurisdiction sets its own level and composition of public services and determines its own tax rates and tax structure. No allowance is made for taxes paid to other jurisdictions or ownership by residents or nonresidents. As is readily seen, such a system will generate inefficiencies in both the public and the private sector. Distorting Effects of Fiscal Competition Suppose first that in the absence of tax competition each country oper- ates a reasonably efficient public sector, with resulting differences in tax 286 Peggy B. Musgrave and expenditure systems. Now fiscal competition is introduced.5 What will happen? Short-Run Responses We begin with short-run responses to fiscal differentials: rs Resources and capital in particular will flow to locations where taxes (or more precisely, net fiscal residues) are lower, thereby distorting the regional allocation of factor use and thereby impairing the efficiency of the private sector. Movement, in particular of capital, to low-tax locations permits the owner who resides in a high-tax location to act as a free rider enjoying a high level of public services without contributing to their cost. As a result, voting patterns will be distorted, burdens will be shifted, and an inefficient level of public provision will result. Each jurisdiction taxing on a source basis will tax income accruing to foreigners so as to maximize the advantages it can derive therefrom. Lower rates of tax will attract foreign capital and raise the base, while higher rates will increase revenue from a given level of foreign capital. The outcome will depend on the elasticities of capital inflow responses, but there is no reason to expect that they will match the domestic share called for by the rules of internation equity. Interpersonal equity will be interfered with as tax burdens are avoided by residents capable of shifting their income-earning assets abroad, given the fact that the jurisdiction of residency in the absence of cooperation has no way of reaching foreign income. Such will be the case especially where the residency country wishes to implement progressive taxation, thus interfering with or avoiding the possibility of redistributional pol- icles. Commodity flows and consumption choices are interfered with as tariffs and subsidies are imposed by individual countries. In the absence of coordination there will be an undersupply of public services, the benefits of which extend beyond borders since no cost contribution is received from external beneficiaries. Longer Run Responses If all taxes are in the nature of benefit taxes, differentials in tax rates across jurisdictions will have no distorting effects on resource flows, but such distortion will result if there are differentials in net fiscal burdens, that is in taxes other than strictly benefit taxes. Resources will then move to low-tax jurisdictions. Such will be the case in the short run, but tax Fiscal Coordination and Competition 287 differentials are not likely to endure if resources are sufficiently mobile. If country A finds that its resources, capital in particular, move to country B where net burdens are lower, A will then be inclined to lower its own tax rates to stem this loss. Otherwise, A will suffer not only a loss of revenue but also, and more important, a loss of national income as capital outflow reduces its productivity. If carried far enough, this pro- cess becomes a zero sum game for both countries. As nonbenefit taxes are reduced, inefficient resource flows will be reversed and as such taxes drop to zero, so will their distorting effects. The Tiebout model with its benefit taxes thus offers an efficient solution. This however is a quite unrealistic picture, especially so when it comes to the international setting. The very nature of public goods renders their financing via strict benefit taxation impracticable. Moreover, distribu- tional considerations must also be accounted for. They may perhaps be overlooked when it comes to local finance where they may be assumed to be taken care of at the national level. But this does not apply in the international context. Here distributional considerations in the domestic policies of various countries must be taken into account. Once income taxes, rather than strictly benefit taxes, are used, and even more so as redistributional programs are accounted for, the benign role of fiscal competition breaks down (see appendix). As taxes are bid down, the level of public services now becomes suboptimal, and the equity of the tax structure will deteriorate as governments seek to maintain their revenue positions by shifting the burden of taxation away from the mobile factor (usually capital) and toward the less mobile factors (labor and land), and away from wealthy residents and toward the less wealthy (and less mobile). Conclusion As even this brief survey of the problem shows, it is evident that “forces of competition” cannot secure an orderly—efficient and equitable— arrangement of public finances in the international setting. The analogy to competition in the private sector does not hold. What in fact is the role of competition in the private sector as Adam Smith viewed it, or as later and more sophisticated theorists saw it, in bringing about an efficient equilibrium in the Walrasian system? It is crucial to note that the argument begins with a setting of entitlements which delineate the economic resources at the disposal of the individual player. These resources may then be used in voluntary exchange so as to 288 Peggy B. Musgrave maximize gains therefrom. Firms will compete for resource inputs by offering the highest returns based on their ability to channel them into those markets which in the end will merit the consumer’s highest evalua- tion. With large numbers of firms engaged in bringing about the efficient solution, each player will seek to maximize the gains to be derived from his/her initial entitlement and thereby contribute to an efficient outcome for the system as a whole. While this image of a benevolent market has to be qualified by allowing for problems such as distribution and exter- nalities, it nevertheless offers a powerful guide to efficient resource use. How then can this Smithian model be translated into the realm of fiscal competition in the international setting? It cannot. Most basically, and in the absence of coordinating measures, there is no initial set of entitlements established within an international legal framework on the basis of which fair competition can proceed. Any one country can exploit any other with which it transacts and use whatever forms of discrimination are most profitable. The responses of the abused will not involve a mechanism by which an equitable outcome in line with some set of predefined entitlements is secured. Nor does such competition offer a way by which resources are drawn into efficient uses. Inefficient spatial allocation is induced instead. Much the same holds regarding any one country’s ability to provide the level of public services or that pattern of interpersonal equity which it would choose were it not for the need to fend off factor movements induced by the actions of competing jurisdictions. While the outcome may or may not lead to an equilibrium, such equilibrium as may be reached will bear little relation to the optimal one, except of course for the wholly unrealistic case where there is universal adherence to benefit taxation. Note also that the international fiscal game involves relatively small numbers only with minor players dominated by a few large players so that strategic behavior and game-theoretic considerations become of major importance, once again, departing from the premises of the effi- cient private model. Fiscal Competition as Remedial Policy Our assumption so far has been that domestic public sector operations are reasonably efficient and that international arrangements should not interfere therewith. Beginning in the 1960s there has developed an exten- sive literature which challenges this premise (Buchanan and Tullock Fiscal Coordination and Competition 289 1962). Instead, it postulates a built-in bias which produces inefficiency and, in particular, an overexpansion of the public sector (Niskanen 1971; Borcherding 1977). This finding was followed by a search for corrective arrangements in domestic policy determination (Brennan and Buchanan 1980) and beyond this by the use of interjurisdictional fiscal competition. Applied first to local governments, the remedial role of fiscal competition has now come to be seen as extending to the interna- tional level. Whether or not the Leviathan proposition of systemic overexpansion is valid will not be examined here. Suffice it to say that an analysis of budget determination does not yield an unequivocal support for the Leviathan hypothesis. Voting failure may lead to under- as well as over- expansion (R. Musgrave 1981). Power-building politicians may be drawn and find support (finance and otherwise) in either direction. Bureaucrats depending on legislative and executive bosses may find their advancement served by either expansion or retrenchment, depending on how the political wind blows. The very assumption that civil servants must necessarily act in their self interest as empire-building bureaucrats is itself questionable. They may also be guided by concern for the public welfare. However this may be, let us assume for purposes of this section that Leviathan does indeed generate excessive budgets. The question then is whether fiscal competition offers an effective remedy, what are its costs, and how does it compare with other and perhaps less costly approaches. Deterring Leviathan via Tax-Base Flight: Effectiveness As we have repeatedly noted, capital mobility once more plays the crucial role. Suppose first that countries A and B have a fully coordinated system with foreign taxes on capital income credited against the domes- tic tax of residence. In this case an increase in A’s taxes will not result in a loss of tax base and overexpansion of A’s budget can proceed without penalty. Next suppose that B does not credit taxes which its residents have paid to A. An increase in A’s source tax on income accruing to foreigners now causes B’s residents to withdraw capital from A. As a result, A’s tax base shrinks, revenue falls, and the productivity of A’s workers is depressed. Next suppose that A increases its income tax on its own residents. This may induce residents to migrate to B, once more with a resulting loss of tax base and revenue in A. Such loss will result even if departing residents leave their investment in A. Finally, an in- 290 Peggy B. Musgrave crease in A’s income tax on residents may invite capital flight to B, and do so without change of residence in situations where A’s tax administration cannot effectively trace the outside income of its residents. A tax increase will thus induce a loss of tax base, whether (1) via capital withdrawal by B residents in the absence of crediting by B, (2) via migration of A residents, or (3) via capital flight due to A’s failure to reach the foreign income of its residents. Working through these and other responses, a tax increase will be punished by base loss thereby raising the cost of public services to remaining residents. If the budget has been overexpanded, this will point to a more efficient budget level. The next question is how this loss of tax base will be translated into effective deterrence of budget expansion. A reduction of tax base, as noted before, means that the rate of taxation needed to yield a given revenue is increased and the cost of public services per taxpayer is driven up. Without further action this need not restrain bureaucrats from press- ing for program extensions. If their bias is introduced by misleading cost- benefit presentations—for example, by showing legislators that total benefits will still exceed total costs, rather than by comparing the mar- ginal costs and benefits of the expansion (Niskanen 1971)—such prac- tice will continue even though taxes have been raised. Bureaucrats them- selves will have to pay more but this will hardly be a sufficient deterrent, especially if the taxes are on capital income. For bureaucrats to be restrained action has to come from the executive level, that is, by linking advancement to success in restricting rather than expanding their bu- reaus. Executives and politicians in turn must look at the response of voters. They must weigh what is to be gained by not raising taxes and what is to be lost by not expanding programs. Since base loss will mean that the tax cost per taxpayer rises, it may be expected that this will induce politicians to tighten budgets and to restrain departmental pres- sures for expansion. The effectiveness of fiscal competition to restrain budget expansion in the last resort thus depends on voter responses to rising tax costs which result from base loss. The impact is not as direct as appears at first sight, but it remains apparent that fiscal competition, mainly via inducing capital flight, will exert a downward pressure on budgets. Deterring Leviathan via Tax-Base Flight: Costs Turning now to the costs of such budget restriction, previous mention has been made of resulting inefficiencies in the location of capital. Such Fiscal Coordination and Competition 291 inefficiencies obviously result if a worldwide view is taken, even though the capital-gaining jurisdiction may benefit in the process. This efficiency cost should not be taken lightly by advocates of competition while at the same time voicing concern over the efficiency cost of excess budgets. The efficiency cost of budgetary overexpansion which prevails with fiscal coordination may well be less than that of misallocation of capital which results from budget restriction by competition. Nor is there a simple way of setting the degree of fiscal competition (for example, the degree to which less than full crediting is allowed) at its correct level, so as to match precisely the degree of overexpansion in the budget which is to be corrected. As we have noted earlier, complete lack of coordination may well result in a downward spiral leaving budget levels severely deficient or at zero. While this process will also reduce or eliminate inefficiency in capital location, it will at the same time lead to mounting inefficiency in suboptimal budget levels. Restraining budget expansion by tax competition will also tend to carry a heavy cost in tax equity in both its horizontal and vertical dimensions. The point to be stressed once more is that capital is the most mobile factor. Since the share of income received from this source rises sharply when moving up the income scale, it follows that budget re- straint by inducing capital flight is in head-on conflict with progressive taxation. The use of fiscal competition thus restrains the ability of jurisdictions to implement its distributional objectives of tax equity if these call for progressive taxation. By the same token efficient use of fiscal competition as a remedial device not only requires that the budget level is excessive but also that domestic budget determination has re- sulted in an excessively progressive pattern of taxation, where “exces- sive” in both cases is defined in comparison with an outcome in line with “true” voter preferences. As a matter of political economy, it also follows that budget restraint via the threat of competition should be of special appeal to high income groups. The imperfections which thus arise are not limited to the finance of general public services but apply even more strongly in the case of tax- transfer programs. Here fiscal competition would take the form of pro- gram reduction, thereby stemming the influx of welfare clients and thus once more reducing budget size. The ability of any one country to provide redistributive programs for its residents is thus impaired by the failure of others to do the same. Just as we noted that the effectiveness of tax coordination breaks down with the residence mobility of high- income recipients, so does that mobility break down coordination in the 292 Peggy B. Musgrave context of redistributive programs. In either case effective coordination would require unacceptable restraints on residence mobility, or else some degree of continued fiscal allegiance (with regard to both tax obligations and transfer claims) based on continued citizenship in the country of emigration. Conclusion Fiscal competition, it appears, is a clumsy and costly means by which to remedy inefficiencies in domestic budget determination, but it is not the only remedy. Remedies may be applied also at the domestic level. These may involve inefficiencies of a new kind, in the extreme case by deterring voters from budget support by the use of taxes which maximize dead- weight loss, thus reversing the goal of optimal taxation. Or rigid limita- tions may be introduced which hold the rate of budget expansion to a fixed proportion of the increase in national income. Such measures again balance one set of inefficiencies against another, in the hope of reaching a net gain in the process. Preferably, however, remedies will be sought which increase the efficiency of domestic budgeting without adding new distortions. This might involve better methods of securing and present- ing program evaluations; budget voting may be rendered more respon- sive by building a bridge between ability-to-pay and benefit taxation; and a more effective agency staff may be secured by replacing bureaucrat baiting with respect for public service. Also ways may be found of reconciling the macro needs of fiscal policy with the discipline of balanc- ing the budget. These and similar measures would be designed to im- prove the process without departing from the premise, itself controver- sial, that the budget is overexpanded to begin with. The aim would be to correct inefficiencies, in whichever direction they may point. Obviously, these remedies are not simple and difficulties will remain. Nevertheless, they offer a more promising path toward the goals set forth in part II than that offered by interjurisdictional fiscal competition. Fiscal Coordination Fiscal coordination can take many forms, depending on the political setting, the objectives of participating jurisdictions, the mobility of fac- tors and economic agents, and the type of fiscal instrument. The purpose of fiscal coordination in its minimal form is to prevent one jurisdiction Fiscal Coordination and Competition 293 from engaging in discriminatory fiscal practices to the detriment of other jurisdictions. In its fullest form fiscal coordination seeks to meet the goals set forth in Part II above. This is to establish an interjurisdictional fiscal environment which is neutral with respect to flows of trade, fac- tors, and residents, and at the same time secures fair tax shares by each jurisdiction in gains accruing to nonresidents, while preserving standards of taxpayer equity prevailing in the residence jurisdiction or jurisdiction of primary tax allegiance (Seligman 1921:110—25). Full coordination ensures that multijurisdictional finance is compatible with the goal of an internationally neutral fiscal system which does not interfere with effi- cient factor use, meets standards of interjurisdictional and taxpayer equity, and permits each country to pursue its own public sector choices. Minimal Coordination Coordinating measures which are currently applied by countries and members of federations might be characterized as minimal coordination. With respect to commodity trade, these include the limitations or pro- hibitions on taxes or subsidies for traded commodities, including tariffs on imports, or subsidies to or taxes on exports for purposes of improving the terms of trade or the balance of trade. Another rule (common in international tax treaties and in federal constitutions) is that tax treat- ment of income earned by foreign investors or nonresident labor not be discriminatory: that the same treatment apply to it as to that of domestic investors and workers or that reciprocally equal withholding taxes apply to income remitted from one country to another. Another minimal coordinating rule might include a prohibition against the full double taxation of foreign investment income. None of these measures ensures interjurisdictional fiscal neutrality or equity. For instance, prohibitions against taxes on and subsidies for traded commodities leaves open the possibility of origin-type product taxes which apply to exports but not imports. Subsidies or tax incentives to investments may lower the cost of both domestic goods and exports. Reciprocity of withholding tax rates leaves open the possibility of nonreciprocal individual and corporate income tax rates (which the nondiscrimination rule virtually ensures). If interjurisdictional equity is defined as reciprocally equal revenue shares, this rule would not be consistent with this standard. Avoidance of double taxation is not enough to secure tax neutrality with respect to capital labor flows, for single taxation by the jurisdiction of source of income 294 Peggy B. Musgrave will expose factor flows to the tax differentials among those jurisdictions of source. Full Coordination: Internation Equity Full coordination as understood here, far from calling for centralization and uniformity, is designed to permit a maximum of fiscal diversity consistent with a setting of interjurisdictional neutrality and equity. As before, we assume that separate governments, whether national or sub- national, exist in order to implement the preferences of their voting residents. These preferences encompass individual choices with respect to the public goods and services provided by the public authority, as well as the social welfare function which determines the level and structure of redistributive expenditures, of the social rate of discount, and of applica- tions of standards of taxpayer equity to the tax structure. At the same time economic transactions and flows of resources and residents across borders give rise to changes in the levels and distribution of income for all participating regions. There are also fiscal impacts arising from the interregional shifts in the levels and distribution of tax bases and claims on government expenditures. Acceptable interregional fiscal relation- ships should be such as to avoid conflict regarding these fiscal claims and liabilities and to permit each unit to achieve its fiscal goals. Securing Internation Equity: Division of Tax Bases The term “internation equity” as noted before applies these consider- ations to a fair sharing of the mobile tax base among nations (P. Mus- grave 1987). This approach to interjurisdictional equity incorporates the widely recognized concept of “territorial entitlement,” whereby the ju- risdiction of source or location of economic activity giving rise to the income, consumption or property, is assigned the primary right to tax whether the income accrues to its own or foreign residents. The entitle- ment by the jurisdiction of source of income or origin of product to this tax is a widely and long recognized one both at the international and subnational level (Harding 1933).¢ But its implementation involves two problems: (1) how the tax base is to be divided among countries of source and (2) what rate of tax the source country should apply to the divided tax base. Division of the tax base poses a problem wherever economic activities Fiscal Coordination and Competition involve multiple locations. This is a particularly severe problem in the case of multijurisdictional corporations with operations integrated across political boundaries (McLure 1984). Consistent and uniform rules, whether separate accounting (if feasible) or formula apportion- ment, are needed to assign the tax base fairly and to prevent overlapping of the base (P. Musgrave 1984). In practice this has been found to be one of the more obdurate problems in interjurisdictional taxation. It requires that there is agreement on the definition of “source” and that in cases where the tax base nexus extends over more than one region of source, that it can be equitably divided, either by applying rules in current usage (such as, for instance, separate accounting) or by mutually agreed rules applied by formula. Securing Internation Equity: Tax Rates on Divided Base Tax rates then applied by each country of source to its share of income so determined must also be consistent with equity rules. Equitable tax shares by the countries of source might be interpreted in different ways, but perhaps the most reasonable is that of reciprocally equal tax rates applied to income flowing between each pair of countries. This would suggest either that corporate tax rates would be everywhere equal or that withholding tax rates in combination with differential corporate tax rates should yield the same combined rate on income paid to nonresi- dents. Thus jurisdictions which opt for lower than average corporate tax rates for domestic reasons would apply higher than average withholding tax rates to bring the total tax share up to the agreed total. If the corporate tax is integrated with the individual income tax, it should be done via a dividends-received credit to the dividend recipient rather than through a dividends-paid credit which would have the effect of relieving foreign recipients from the corporation tax and interfering with interna- tion equity. Taking the same approach concerning the value-added tax, tax rates on exports after rebates should be equalized and not necessarily brought to zero. The reciprocity rule is theoretically applicable to gov- ernment transfer payments and subsidies. Thus country A would under- take to apply to temporary residents from country B, a level of transfers and subsidies equal to that provided by region B to temporary residents from region A. A few further points might be added regarding the role of internation equity. First, it is evident that implementing internation equity by mutual agreement on the rate of source taxation may interfere with the principle 295 296 Peggy B. Musgrave of nondiscrimination. Though widely accepted, this principle is however applicable only within its appropriate context. On the one hand, it is quite reasonable to argue that a country in taxing its residents should not discriminate between various sources from which their incomes are derived; on the other hand, it is quite appropriate, and indeed required by the concept of internation equity, that the rate of tax which a country imposes on nonresident income derived within its borders should differ from the rate of tax which it applies to its own residents. The tax applied to nonresident income is an in rem tax justified by rules of internation equity while the tax on residents is a personal tax, and should reflect considerations of interpersonal equity. Finally, note that how the country of residence chooses to tax the income or economic activities of its residents which are generated out- side its borders has no bearing on interjurisdictional equity, since the tax imposed by the residence country merely determines the share of the “national” return which goes to the treasury. Securing Internation Equity: Benefit Spillovers On the expenditure side internation equity calls for joint provision for public goods which have spillover effects across borders, involving com- pensation to the providing jurisdiction by the beneficiary of spillover effects. While benefit spillovers are most frequent in the case of local finance, they may also assume major importance in the international context, for example, joint concerns with defense, waterways, acid rain, oil spills, and other environmental matters. Full Coordination: Interresident Equity Here a distinction must be drawn between problems which arise because taxpayers whose residence is fixed in one country derive income from abroad and others which arise because effects of taxation may induce taxpayers to switch their country of residence. Foreign Income Moving now to interpersonal equity, attention shifts from the country of source to that of residence. Here the basic principle is that all residents of any one country should pay the same total tax (including that paid at home and abroad) if their total income (foreign and domestic) is the Fiscal Coordination and Competition 297 same. In a system of income taxation full implementation of this rule would require the country of residence to include both domestic and foreign (before foreign tax) income in the base, impose its tax thereon, and then credit the foreign tax. This procedure not only meets the requirement of horizontal equity, but also assures neutrality with respect to the location of investment or work. Nor does it interfere with interna- tion equity. Note, however, that where the foreign tax is higher, full coordination would require the crediting rule to extend to refunds. As in the case of loss allowances in the domestic context, this may not be readily acceptable. The same approach may be applied to a setting where different countries not only differ in rates of income tax, but also in corporation tax rates. The country of residence may again tax on a grossed-up basis and then credit foreign tax. Since capital is a highly mobile factor, effective coordination is of special importance in this case, even though it may be difficult to accomplish. Special provisions are needed to deal with situations where either one or both countries integrate the corporate with the individual income tax. As previously noted, for internation equity reasons countries with integrated systems should integrate via crediting the corporation tax to the shareholder when dividends are received by shareholders, thereby denying the credit on dividends paid to foreigners. Regarding foreign dividends received by its residents, the rule should be to apply the domestic corporate tax to a base arrived at by grossing up dividends by the foreign corporate tax and then to credit the latter against the domestic corporate tax, after which the dividends would be treated in the same way as domestic-source dividends. The entire problem of coordinating the taxation of capital income and its complexities would, of course, be avoided if all countries were to dis- card income taxation and adopt an expenditure tax instead. While longed for by the Meade report (Meade 1978) as the ideal solution, this approach while indeed removing a difficult problem does so at the ex- pense of the freedom of individual jurisdictions to design their tax sys- tems as they wish. Moreover, the neatness of such a solution may depend on just how the consumption tax is designed (P. Musgrave 1989). Change in Residence Without expanding further on the technical details of coordination, it appears that a satisfactory solution can be worked out, a solution which maintains equity and neutrality regarding factor movements. The mutu- 298 Peggy B. Musgrave ally reinforcing system breaks down, however, when it comes to the movement of residency of individuals. As noted before, differences among jurisdictions in income tax rates and degree of progressivity, as well as in transfer programs, if sufficiently great, will induce taxpayers to shift their place of residence so as to avoid tax liabilities. High income citizens of the United States may for tax purposes become residents of the Caribbean, or U.K. citizens for tax purposes may become residents of Luxembourg. While the inducement for residents to invest abroad can be checked by coordination and credit arrangements, the inducement to migrate cannot be voided. We are thus returned to a situation noted ini- tially to prevail in the absence of coordination where wealthy individuals congregate and redistributional objectives of fiscal policy are voided. Such practice might be discouraged by conditioning both citizenship and residence on tax allegiance, but no country so far has moved in this direction. It should be added that the difficulties caused by residence mobility apply primarily to the residence choice of individuals. Regard- ing corporations, the switching of residence is accomplished more easily, but it also carries with it the possible disadvantage of foreign incorpora- tion. The consequences of tax avoidance by residency change assumes particular importance in the context of Lpcs. As their capable profes- sionals move to industrialized countries where earnings are higher and taxes lower, a significant loss to their home economy occurs. To make matters worse the home country may have made a substantial invest- ment in the migrants’ education, a form of human capital which is taken abroad by emigration. Suggestions have been made for an exit tax or other means by which the LDC retains some tax claim on its emigrants, especially so where the latter retain their home citizenship (Bhagwati 1976). International versus Subnational Coordination Our discussion so far has been directed mainly at coordination among nations, but the need for coordination also arises among constituent states within a federal system. The theory of fiscal decentralization would assign certain expenditure functions and the taxes to finance them to different levels of government—state, local, and central (McLure 1983, 1986; Oates and Schwab 1985). It is generally agreed that re- Fiscal Coordination and Competition distributive policies should be largely the responsibility of the central government, while the provision of goods and services, the benefits of which have limited spatial dimensions, should belong to lower-level governments. Provided that the latter use revenue sources which are closely related to the distribution of benefits of the public services pro- vided, one would not expect to see a strong need for coordination among the fiscal systems at that level of government. Corporation income and payroll taxes are still appropriate as source-entitlement taxes and should be of equal rate. This equality in rates is of particular importance at lower levels of government since capital and labor are highly mobile between local jurisdictions and the administrative feasibility of coordi- nating any rate differentials is quite limited. Provided the central govern- ment carries out an adequate degree of redistribution, personal income taxes at that level are less suitable than are property taxes, fees and charges, and possibly retail sales taxes. It is to be noted that the retail sales tax, being a destination tax, does not call for border-tax adjust- ments as would a value-added tax at this level, but it assigns a zero tax share to the country of origin. This theoretically optimal division of fiscal responsibilities, however, is not generally achieved in practice. As is the case in the United States, member states find it politically difficult to agree on common rates of corporation and payroll taxes, which in consequence are competed down to very low, but differential rates. Broad-based individual income taxes are widely used, indeed are often seen as necessary to compensate for the lack of progressivity in the federal system. In the real world one sees only a partial attainment of the normative model and fiscal residues remain. However, they pose much less of a problem at the subnational level than do differentials between nation-states. First, central govern- ment expenditures and revenues account for much of the weight in the overall system, so that lower-level rates and differentials are much smaller and therefore less of a problem even though factors are more mobile at that level. Secondly, the central government provides an over- arching fiscal structure which can help to correct for such differentials which exist at the subnational level. The deductibility of state-local income taxes in the United States thus serves to mitigate the effective differentials which exist, though at some cost to the equity and integrity of the federal tax. Common markets such as the European Economic Community represent political systems somewhere between the federa- tion and a collection of independent nation-states. With economic inte- 299 300 Peggy B. Musgrave gration, the increasing mobility of resources, and improvements in com- munications systems, a process of consolidation of public preferences is to be expected so that a central government emerges acting on behalf of the entire community in areas of common interest. Nevertheless, this process allows for a substantial degree of decentralization. Spatial fac- tors in the incidence of benefits and costs and production technologies may all call for separate and often overlapping jurisdictions whose job it is to provide different public services. Furthermore, nation-states are likely to exhibit more deeply entrenched differences and greater varia- tion in preferences than is observable between regions within a nation- state. It follows that the preservation of diversity in fiscal systems to reflect differences in underlying social preferences is a fundamental crite- rion in evaluating multijurisdictional fiscal relationships. Yet in a world of mobility of residents, consumers, labor, and capital, as well as com- modity trade, differences in fiscal systems and the differentials in net benefits and burdens which ensue can result in nonneutralities leading to the movement of resources from more- to less-productive uses or of residents from more- to less-preferred locations. Not only are such fiscally induced relocations inefficient, but they can also sever the link between tax burdens and expenditure benefits and sunder social con- tracts with regard to distribution policies. Differences in tax rates can also result in unequal, nonreciprocal tax shares in income accruing to residents of other countries. Appendix: Limitations of the Tiebout Model The Tiebout model of efficient resident mobility breaks down when public goods are financed through general rather than benefit taxation, and coordinating measures will be needed to protect diversity of preferences for social goods, while securing fiscal neutrality with respect to location of work, investment, residency, and consumption. This may be illustrated with reference to the individual in- come tax, the corporation income tax, and a general consumption tax. Individual Income Tax Under the highly restrictive conditions of the Tiebout model, with equal incomes but differing preferences, little is changed by moving from benefit taxation to Fiscal Coordination and Competition income taxation. However, there is a minor qualification to this conclusion in that low-savers will want to move to jurisdictions of high-savers, for under the income tax, the latter will pay a larger contribution to social goods. Once differing incomes are allowed for, however, income-tax finance can have distort- ing effects in the absence of coordination. Table A.1 gives numerical illustration to this conclusion. Consider jurisdictions, J, and J,, each with two residents. H, and H, are high-income taxpayers with wages of 150 and corporate-source income of 50, while L, and L, are low-income residents with wages of 100. To avoid the earlier discussed tendency in a Tiebout model to have low-income residents seeking to reside with high-income individuals because of the income effect on demand for social goods, we assume that the income-elasticity of demand for such goods is zero. We also assume that H, and L, have identical preference for social goods which is weaker than that of H, and L, who also have identical tastes for social goods. Further, it is assumed that the spatial limits of the benefits and costs of government operations coincide with the geographical boundaries of J, and J,. We begin with a system of pure benefit taxation (columns I through V), so that the Hs and Ls have sorted themselves out, with H, and L, residing in jurisdiction 1 and choosing to spend 100 on social goods with each deriving 50 of benefits and each contributing 50 in benefit taxes, since their preferences are identical and there is no income effect on such demand. Individuals H, and L, likewise choose to reside together in jurisdiction 2, spending 150 on social goods and each deriving 75 of benefits and paying 75 in benefit taxes. Thus total real income will be 200 for the Hs and too for the Ls in each location. The situation will be Tiebout-efficient, and there will be no fiscal incentive for any further rearrange- ment of residency. Furthermore, the fiscal environment will be neutral with respect to location of capital, labor, or consumption, provided that any benefits provided by one jurisdiction to those factors owned by residents of the other will be matched by corresponding benefit taxes, and similarly for consumption activities. The situation might also be said to meet standards of interjurisdic- tional equity, since each jurisdiction is permitted to charge for benefits provided to residents. We now substitute proportional income taxes at rates of 33 and 50 percent respectively in J, and J, for the benefit tax, to cover the cost of the social goods (columns VI and VII). Since the tax now paid by the Hs exceeds the benefits they get from the social goods, they experience a negative fiscal residue, and their real income falls. At the same time the Ls now find that the income tax they pay falls short of the benefits they derive from the social good, and they experience a positive net fiscal residue with their real incomes rising. While overall real income in each jurisdiction has remained as before, a redistribution has occurred from high to low. More important for the analysis in this section, the distribu- tional effects are stronger in J, with the larger public sector. Assuming zero 301 71 07 = €€ OOL = Oot SZ 'H 8ZI o¢ 0 L9 007 os OST SL ei %0T %0 YEE OST Z uonsipsiin{ €TI 0 = LT 001 = 001 os lay LAL 0 ST 8b 007 os OST os 'H ALI%O APIY%OS 91 %/T OOL [| uonsipsiin{ > O.4 >a) B,< IITIAX IIAX IAX AX AIX IIIxX If1>,< 9WIOSUT Xe Xe “DUT xXel d9UIODUT IUODUT Spoor) “D0§ [eo *suoy ‘dioy QUIODUT [BIOL jeudey sodey, sayauog Jo soy wiayshs paxil Sil 09 Sz os 001 SZ OOT == 001 GZ ey ssl 06 SZI O01 007 SL 007 os OST ¢L “YH %09 %OS OST Z uonsipsiin{ OLL OF LII €¢ OOT os O01 = 001 0s I 061 09 €8I £9 007 os 007 os OSI 0s NEI a1e1 % Or a1e1 %Ee 001 [ uonsipsiin{ 18,2 x XI ILIA IIA IA A AI III II I IWIOU] Xe] IUIOSU] Xe] IWIODUT Xe] QUIODUT ==dWIODUT Spoor *305 eps | *suoy [Poy IWIODUT [Poy qyousg [eI0L jeudey sadey, siyoucg jo 1sOy [302] Xe] Iaueg YUM UONeXe] [eIIUID JO uOsIIedWOD [eoOWINA :][°Y aqeL Fiscal Coordination and Competition 303 residence mobility, the situation will remain efficient with respect to the choice of social goods. However, with residence mobility assumed for the Hs, the higher income tax paid in J, may induce H, to move to J,, if the lower tax there more than compensates for the less preferred and lower levels of social goods. Plus, the addition of H, to the voters of J, may result in a level of social goods higher than would be the case if tastes were fully homogeneous. As H, moves to J,, L, will have to reduce the level of social goods since income is insufficient to cover the previous level when shared with H,. Similar inefficiencies result if mobility of residence for the Ls is allowed for. Now suppose that there is no residence mobility, but there is labor mobility. Further, we assume that a general system of source taxation prevails, that is, each jurisdiction applies its income tax to labor income arising within its boundaries but not that which is earned by its residents outside. If workers can reside in one jurisdiction and work in the other, inefficiencies of resource allocation will arise since it will pay both H, and L, to work in J, where they will pay lower taxes. This will necessarily result in lower levels of social goods in J, as tax revenue is diverted to J, where the income is earned. Thus, whether a shift in work location is or is not accompanied by a shift in residence location, inefficiencies in resource allocation and in the operation of the public sector will result. General Consumption Tax The case of a general consumption tax is shown in columns VIII and IX. Since this is equivalent to a tax on wages only, the required rate is higher than in the case of the proportional income tax, and the redistributive effects are less. Everything that was said regarding the income tax also applies here, though the incentive to move on the part of labor and residents will be weaker. But there is the additional problem of consumer mobility. If residents of J, can shift their consumption purchases into J,, they will gain a tax advantage and, furthermore, the public sector in J, will lose tax revenue and be compelled to reduce the level of public services. Corporation Income Tax Finally, let us consider the corporation income tax allowing for capital mobility. This is a case where owners of capital can remain resident in the jurisdiction of their choice while investing their capital outside. For this and other reasons capital mobility is much greater than is labor and resident mobility. Therefore differentials in corporate tax rates may be expected to result in capital movement 304 Peggy B. Musgrave and, as in the labor mobility case, result in misallocation of resources as well as a reduction in the otherwise efficient level of public services in the residence country. Alternatively, in response to the capital outflow and in an effort to retain its tax base, the high-tax jurisdiction may decide to lower its corporate rate and shift the tax burden to less mobile factors such as labor and land. But in the low- tax capital-importing jurisdiction the gain of resources and tax base may permit a further lowering of tax rates with a further round of capital outflow from the higher-tax jurisdiction. This will particularly be the case if the inflow of resources is unaccompanied by commensurate increase in demand for public services, as would be the case for financial capital flows. The existence of tax differentials on capital income, combined with the source principle of taxation, leaves capital open to strong nonneutralities and poses the most serious problem for the preservation of diversity (and thus efficiency) within the public sectors. Mixed General Taxation Depending on the relative mobility of the various factors which we have dis- cussed—labor, capital, residence, and consumption—and with uncoordinated, source-based taxation, governments may seek to protect their tax base by choos- ing a tax structure which has as little effect on resource and residence movements as possible. This means that diversity in the level and provision of social goods can only be protected in an uncoordinated tax system by the sacrifice of choice with regard to tax structures. The likely result is that jurisdictions choose to have low or zero taxes on capital income, owing to the high mobility of capital, compensated by higher taxes on consumption, since consumer mobility or labor mobility is less possible. Land taxation may also be favored, if residence mobility is considered to be less of a problem, while benefit taxation may also be empha- sized. This is illustrated in columns XII through XX, in which J, uses a personal and corporate income tax mix, while J, applies a personal income and consump- tion tax mix. A high degree of neutrality is achieved by the use in J, of a consumption (wage) tax rather than the corporation income tax, which offsets the disadvantage J, would suffer in the interjurisdictional competition for capital and labor owing to a larger public sector entirely income-tax financed. Notes Not surprisingly, I discussed this topic extensively with my husband while preparing the paper which benefits greatly from his insights. The errors, how- ever, are all mine. Fiscal Coordination and Competition 1. For a good discussion of this point of view applied in a federal setting, see Brennan and Buchanan (1980, ch. 9). 2. Ata highly theoretical level it might be possible to conduct this analysis by initially distinguishing between the allocation and distribution functions of the budget, viewing their respective implications in the international settings and then proceeding to a netting out of the two concerns. This, however, would involve too elaborate an analysis for present purposes. 3. Stated more carefully, reference throughout should not be to tax rates only, but to net fiscal residues allowing also for expenditure benefits received. 4. The case of source-country participation might also be based on benefit consideration, that is, the country of source should be compensated for inter- mediate services rendered. However, I do not think that this benefit aspect fully covers the issue of source-country entitlement. 5. Note that the analysis is limited to tax competition. Other forms of compe- tition, for example, through less stringent environmental requirements, may also be considered. 6. 1 am indebted to Vogel (1988) for these references. 305 5. DOUGLAS AccLeAU LD Compensatory Fiscal Policy: Evolution or Revolution? The Contribution of Carl Shoup A primary reason for this conference is to honor the breadth and depth of Carl Shoup’s contribution to economics, and especially to the field of public finance. It is only fitting, in my view, to anchor this paper on fiscal policy in the groundwork established by Carl Shoup, and there is no more appropriate foundation than chapters 18, 19, 22, and 24 of Shoup’s 1969 Aldine treatise, Public Finance. The timeliness of these chapters is almost uncanny, having been written at the zenith of fiscal policy’s popularity and the widespread devotion to macroeconomic pol- icy for achieving full employment and price stability while sustaining economic growth. It was widely held that macroeconomic fiscal policy or compensatory finance, as it was labeled by Richard Musgrave (1959), was responsible for the economic recovery of the world’s greatest eco- nomic power. The theories of Keynes, while having been tested in more limited experiments, were now “proven.” Free enterprise, industrialized countries could look forward to policy makers steering a steady course through unchartered economic water using, judiciously, a variety of fiscal instruments. That was the world of the late 1960s, the time of Shoup’s treatise. The theoretical principles of compensatory finance were not em- braced by all the experts, far from it; and Shoup’s chapters capture, mostly in nontechnical terms, the doubts and anxieties regarding the “new” economics. Chapter 19 sets the stage for most of the macroeconomics in Shoup’s treatise, drawing heavily on the work of Bent Hansen (1958) who was responsible for placing much of the macroeconomics of Keynes’ work in the context of a broader picture of public finance objectives. Shoup Compensatory Fiscal Policy reminds us, quite rightly, at the outset that “one cannot afford to become too engrossed with some one end, forgetting that the measures taken to achieve this end will quite likely affect appreciably the degree to which other ends can be achieved” (466). Another important issue, somewhat overlooked at the time, was the risk associated with the timing of fiscal policy. The policy maker, accord- ing to Shoup, ought to contemplate a series of outcomes stemming from the use of any instrument (or instruments). These outcomes have to be viewed in terms of not only the degree of success in achieving the economic goal but the speed of attaining the goal (see Shaw and Peacock 1971). In many popular treatments of compensatory fiscal policy, there was an overemphasis on the deterministic nature of models and lack of debate pertaining to the time element involved. Part of chapter 22 is devoted to a careful exposition of lags in the private sector’s planning process pertaining to consumption, investment, and the price of household and corporate assets. In a nontechnical overview Shoup illustrates how the results of the work by Allen (1967), Musgrave (1959), and Phillips (1954) underscore the absolute need to understand the dynamics of the economic systems before fiscal manage- ment can be considered as a method of “controlling” the economy’s direction. Much of the foundation for the discussion of policy means and ends of the compensatory fiscal policy of the 1960s was established by Han- sen (1958). Shoup provides a concise overview of this issue, defining clearly what is and is not a policy instrument. More important, perhaps, he is careful to draw out the monetary implications of a fiscal policy decision and define the difference between fiscal and monetary policy. The relationship of instruments to goals has particular applications to compensatory finance. It is also a subject which lends itself to complex modeling and mathematical formulation. In a remarkably short four pages of text and a two page appendix of simple algebra, Shoup is able to reduce much of the complexity down to the basic principles. In addition, Shoup’s emphasis on the practical aspects of fiscal policy is incorporated in a separate section on the constrained value of certain fiscal instru- ments and conflict of goals in a policy setting. This chapter recognizes that compensatory finance in a globally inte- grated economy may have very different impacts than in the closed economy model. While pioneering work by Meade (1951), Mundell (1963), Swan (1960), and Oates (1966) had been done on this question, 307 308 Douglas A. L. Auld it was frequently overlooked in many popular discussions of demand management policy. The chapter includes an appendix where a simple “open economy” model is presented to underscore the balance of pay- ments effects in an open economy. The external equilibrium issue is revisited in chapter 24 along with the relationship between fiscal policy and price stability. Shoup discusses how rising taxes can reduce demand and thus alleviate the pressure on prices due to an imbalance of aggregate demand and supply at constant (or slightly) rising price levels. But he points out “An increase in the personal income tax may check consumer spending very little in coun- tries where wage negotiations that are based, ostensibly, on gross pay are in fact influenced by the relationship between take home pay and the cost of living index” (601). This aspect of tax policy had, in 1969, received only limited attention (Brennan and Auld 1968) and underscored the importance of recognizing the underlying equations in macroeconomic models. Shoup might have added that the extent to which wage negotia- tions are heavily influenced by the cost of living index, an increase in consumption taxes will also lead to higher wages and, if validated by the monetary authorities, higher money demand. Chapter 18 examines government borrowing and inflationary finance. For this review of compensatory finance we deal with only the first four pages of this chapter. Nevertheless, in those four pages Shoup is able to provide a concise statement of the basic incidence issue related to debt financing, an issue which in the 1970s and early 1980s became an impor- tant focal point for those who argued against debt-financed fiscal policy. Shoup’s framework in this chapter is not entirely applicable to com- pensatory finance since one of the premises in the chapter is that the use of resources by government “requires that the private sector reduce its spending on investment goods or consumption goods” (442). In a full employment economy that is true but not necessary if there are under- utilized resources. The main point that Shoup underscores is the possibility that if people realize that the cost of servicing today’s debt implies future taxes, they may reduce current consumption. “If the stream of future additional taxes is fully discounted, the bond issue will cause the private sector to consider itself... poorer by an amount equal to the bond issue” (442). Shoup discusses a range of possibilities depending on uncertainty about future tax incidence, the use of the bond proceeds, and the type of taxes used by the government. In a compensatory finance framework, how- Compensatory Fiscal Policy 309 ever, a reduction in income taxes to stimulate demand may have little effect on the target if taxpayers discount the future tax payments to the present. Shoup also emphasizes the point that if bond finance is used in place of taxation to finance current expenditures, future generations could be worse off if there is a reduction in private fixed capital forma- tion. In this context it should be noted however that debt financed public capital expenditures in the current period may enhance wealth in the future by providing public sector infrastructure. In addition to demonstrating a broad and detailed understanding of current and past issues related to fiscal stabilization policy, these four chapters provide insight into the reasons for the decline and fall of stabilization policy as the “fiscal wonder drug” of the 1950s and 1960s. In broad summary form, then, these are 1. The uncertainties associated with the responses to changes in fiscal variables and the complex lags involved in discretionary policy. 2. The constraints imposed on domestic policy by trade and financial links with the rest of the world. The financial market implications of changes in fiscal instruments. 4. The concern that present generations would anticipate (discounted) costs of deficit finance and limit the usefulness of antirecessionary fiscal policy. 5. The growing belief that wage rates were being “set” in terms of “after tax” compensation, thereby limiting the effectiveness of anti-inflationary tax increases. >) In reviewing Shoup’s work one cannot help but notice the influence of Richard Musgrave’s classic, The Theory of Public Finance (1959). Shoup, in fact, refers to this work no less than ten times in his analysis of the macroeconomic aspects of public finance. This is not the place or time to provide a detailed review of Musgrave’s treatise and its contribu- tion to the development of fiscal policy analysis. Nevertheless, its impor- tance in setting the stage for a considerable literature on fiscal policy in the 1960s and 1970s leads us to conclude that a brief overview is warranted. Fiscal policy or the “theory of stabilization,” as Musgrave labeled the first chapter in this section of his book, is developed in a general mac- roeconomic model where the rate of interest depends on the transaction demand for money. The model is manipulated to allow for a variety of assumptions including wealth effects on consumption, money wage ri- gidity, and interest floors. The role of money and the price level are also 310 Douglas A. L. Auld incorporated in the model and examined in detail. The importance of the relationship between consumption expenditure and fiscal policy instru- ments is highlighted in a separate chapter, incorporating the role of money illusion, changes in the marginal propensity to consume, and price effects. The investment aspects of fiscal policy also receive a very detailed analysis. Many of the footnotes in both these chapters contain mathematical models which became standard components of articles and other textbooks in the 1960s and 1970s. Musgrave’s discussion of fiscal instruments and inflation concludes: “The ideal tax policy. . . would be a tax on wages that applies to wage gains in excess of produc- tivity gains and through a tax on profits that applies to profits from increases in prices not justified by increased costs, both taxes being applied at marginal rates of 100 percent” (471). It would be a decade before tax-based incomes policies were being hailed as the “new” eco- nomic policy. Musgrave’s treatment of compensatory finance also examines the problem of timing fiscal policy changes, the role of built-in stabilizers, and the dynamics associated with response lags. Again, while the models used to examine these issues would be considered basic foundations today, they were, in 1959, ahead of their time. They have also stood the test of time. The thoroughness of Musgrave’s work is further underscored in the final three chapters of his book where the liquidity aspects of fiscal policy, the theory of debt, and compensatory aspects of debt policy are examined in penetrating detail. Musgrave’s highly theoretical (at the time) treatment of fiscal policy and Shoup’s detailed analysis of many of the practical aspects of fiscal policy together provide an excellent over- view of compensatory finance in the late 1950s and 1960s. Both treatises raise a number of concerns relating to the effectiveness of fiscal policy while supporting the principle of countercyclical policy. How wide- spread were these concerns? What other factors led governments to slowly limit compensatory fiscal actions? Was fiscal policy overrated from the beginning? These are important questions which cannot be answered fully given the scope of this paper. Complicating the Basic Fiscal Policy Model Prior to the last half of the 1960s the context within which fiscal policy was supposed to work was relatively simple. With the exception of a few Compensatory Fiscal Policy minor irritations related to financing deficits, exchange rate inflexibil- ities, and the balance of payments, manipulation of real output (and hence prices and employment) could be achieved by manipulating the levers of fiscal (and sometimes monetary) policy. The context changed dramatically during the 1970s and into the 1980s. The following terms provide some idea of the ever-tightening loops fiscal policy has had to jump through during the past fifteen to twenty years. rational expectations supply-side economics the Laffer curve tax-push inflation Ricardian equivalence theory crowding out the natural rate of unemployment EON eat To those who studied or practiced fiscal policy in the nineteen sixties, the terms either did not exist or were consigned to some abstract theoriz- ing. There are those who would still argue that all these terms can be so consigned, but there would be little support for such a position. What do these terms mean? How are they interpreted in the fiscal policy context? Are they part of conventional wisdom? Do the implica- tions derived from these phenomena all but rule out compensatory fiscal policy in modern industrialized economies? Rational Expectations The notion of rational expectations can be traced back to Muth (1961), and its application to macroeconomics is found in Lucas (1976). Barro (1984) defines rational expectations in the following way: “This ap- proach says that if people do not observe something directly—such as the current price level—then they form the best possible estimate of this variable, given the information that they possess. In other words people make efficient use of their limited data, so as not to commit avoidable errors” (469). Notwithstanding other difficulties associated with the application of countercyclical fiscal policy, in a world of rational expec- tations people would respond to an anti-inflationary rise in taxation or reduction in expenditure by, for example, reducing inventories and postponing investment expenditures which, in the short run, may sustain inflation while creating more unemployment. The direct effects of the 311 S12 Douglas A. L. Auld expenditure reductions (occurring with a lag) add to the unemployment and precipitate a decline in the rate of inflation. Complicating this are wage contracts that prevent cost adjustments if they cover long periods of time. In short, by forming expectations as to the effect of policy changes, the actual fluctuations are greater than would have been the case were there no fiscal intervention. While this is an oversimplification of rational expectations as it applies to fiscal policy, the growth of a sophisticated literature in this area and its widespread adoption in eco- nomic textbooks have raised new questions regarding the effectiveness of countercyclical fiscal policy. Supply-Side Economics Supply-side economics is, to some, a doctrine which has evolved out of the disappointing economic performance of the mid-1970s and apparent lack of success of discretionary fiscal policy. “Supply-siders” argue that governments should abandon policies that affect aggregate demand and develop policies to shift the short run aggregate supply curve (to the right), thus expanding output and employment and reducing price infla- tion. Traditional anti-inflationary fiscal policies such as increases in personal and consumption taxes may actually threaten price stability (see below). Reductions in real purchases of federal services may be an appropriate policy but only as a temporary measure. The concept of supply-side economics is also inextricably tied to the Laffer curve. Laffer Curve Arthur Laffer postulated that as the government raised the tax rate, revenue would reach a maximum long before the tax rate reached 100 percent. This occurs because as the tax rate rises, the rate of increase in taxable income declines due to disincentives, mobility of high-income earners, etc. Thus Ri = ti-yi and Y' = f(t') where dYi/dti < o when ti > ti max where R' is revenue from the ith tax source, and t' is the average tax rate (at the margin) imposed on the ith tax base, Y. No one knows what the “magic” tax rate is although Stuart (1981) suggests that for Sweden it is Compensatory Fiscal Policy 313 approximately 70 percent. While the very existence of this curve is questioned, the concept has found its way into many basic textbooks (for example, Barro 1984), and to those who oppose interventionalist anti- inflationary fiscal policy, the Laffer curve is one more argument for such a position (see Fullerton 1982). Tax-Push Inflation It has always been recognized and accepted that an increase in any ad valorem or excise tax could well Jead to a rise in prices. However, in the context of compensatory finance this would be a once and for all in- crease which would shortly dampen demand and reduce the rate of price inflation (Musgrave 1959). In the late 1960s at the same time as Shoup’s treatise was published, Brennan and Auld (1968) demonstrated that both sales tax and personal income tax rate decreases could, in a dy- namic context, be anti-inflationary. Since then there has been a growing literature on the subject, both theoretical and empirical (see Auld 1977; Nowotny 1980). In a recent paper Auld and Wilton (1988) have shown that in Canada, three quarters of the increase in effective personal in- come tax rate increases have been pushed onto higher gross wages. To examine the link between income tax rate increases and inflation, the models used assume that workers bargain in terms of their net or after-tax wage rate. Increases in income tax rates (average or marginal) drive a wedge between net and gross wage rates reducing the ratio of take home to gross pay. To restore the original ratio workers bargain for a higher (inflationary) gross wage. These models must, of course, recognize that overall demand condi- tions, the growth in money supply, and labor market conditions will all influence the actual impact of higher taxes on gross wage rates. Nev- ertheless, the plethora of scientific and popular articles on the subject has raised doubts about the strictly Keynesian prescription for inflation using income taxes. Lowering indirect taxes to reduce inflation is also “un-Keynesian” but fits well into the arguments of supply-side propo- nents. ! Ricardian Equivalence Theory Keynesian fiscal policy calls for the government to sustain a deficit in periods of high unemployment and less than capacity resource utiliza- 314 Douglas A. L. Auld tion. Unless the deficit is financed out of a pure monetary expansion, the debt issued to cover the gap between current expenditures and revenue, along with interest on the debt, will have to be repaid. If the deficit is generated by a reduction in personal income taxes of T, households will have T dollars more to spend in the current period. However, if they have the foresight to look ahead, they will have to pay out T + R dollars in the future, where R is the interest on the debt issued. Now if house- holds use the extra T dollars to buy the government bonds, they will have T + Rextra funds in the next period to pay for the higher taxes needed to redeem the debt. The result is no aggregate change in wealth, work effort, or demand. This is referred to as the Ricardian Equivalence Theorem.2 Households may not save money in the current period to pay taxes in the future period but may merely spend all of the increase in disposable income. When the taxes are increased in the future (to redeem the debt), household consumption will then decline with the rise in taxes. “Overall, fiscal policy turns out to be an instrument that can influence the timing of real economic activity. But if we use this policy to get more output today, then we have to accept less output later on” (Barro 1984:387). The debate surrounding the issue centers on whether or not house- holds are sufficiently farsighted or concerned about the effect of deficit financing on future generations to adjust current spending to balance the future tax liability. Those who consider this to be the case believe that short-term fiscal policy is a useless exercise. If it is assumed that house- holds are shortsighted, there is still an argument against antirecessionary fiscal policy on the grounds that today’s gain is tomorrow’s loss. Crowding Out “Crowding out” is a term with more than one meaning, but it has been used to bolster arguments against countercyclical fiscal policy, partic- ularly policy-induced deficits to reduce unemployment. Put simply, the term “complete crowding out” means that any extra real expenditure precipitated by fiscal policy is offset by a reduction in private expendi- ture. This is, however, an oversimplified view of the world. “Crowding out” may refer to the long understood “transaction crowding out” associated with standard IS-LM models and recognized long ago by Compensatory Fiscal Policy 315 Hicks (1937). There is also the matter of “portfolio crowding out” emphasized by Friedman (1972) and analyzed by Blinder and Solow (1973) and Tobin and Buiter (1976). In these models the focus is on how bond issues affect private investment through asset substitution. All aspects of “crowding out” received considerable attention during the late 1960s and into the 1970s. The econometric models became larger and the theory much more complex, leading to a variety of conclusions, mostly couched in terms of “given the assumptions about.” Despite the lack of a clear consensus on whether or not deficit-induced increased spending crowds out private spending, the concept offers one more reason to be cautious, even suspicious, about fiscal policy. The Natural Rate of Unemployment/NAIRU/Price Expectations Augmented Phillips Curve In 1967 Milton Friedman coined the term “the natural rate of unemploy- ment”—the rate of unemployment at which the rate of change of the price level is neither accelerating nor decelerating (Non Accelerating Inflation Rate of Unemployment [NArRU]). The concept is based on the now familiar price expectations Phillips curve. The actual “natural rate of unemployment” depends on such factors as labor productivity, unemployment compensation, the pace of technical change, and skills development. It corresponds to an output level which is labeled full- employment national income and is associated with any money-vali- dated price expectations. If the natural rate were an absolute constant, increases in the rate of unemployment could well be interpreted as demand-deficient unemploy- ment and a candidate for correction by a stimulatory fiscal policy. If the measured change in unemployment is a shift in the natural rate, attempts to stimulate demand will only cause a rise in the rate of inflation. The uncertainty about the causes of changes in measured unemployment is therefore an important contributing factor in the argument against fiscal stabilization policy. What Has Survived? Notwithstanding the steady stream of criticism against discretionary fiscal policy, particularly deficit financing strategies to stimulate demand, 316 Douglas A. L. Auld the prescriptions of even traditional fiscal policies are not dead. The idea of government injecting a “quick fix” is, for the time being, out of favor with more emphasis on economic policy being directed toward the long- term health of the economy. The publication of Shoup’s treatise coincided with a heated debate over the usefulness of both fiscal and monetary policy which was high- lighted by the publication of a debate/dialogue between Milton Fried- man and Walter Heller (1969). Friedman’s underlying thesis was that while “The state of the government budget has a considerable effect on interest rates .. . the state of the budget by itself has no significant effect on the course of nominal income, or inflation, on deflation or on cyclical fluctuations” (50o—5 1). Much of Friedman’s analysis assails the effective- ness of fiscal policy because there is no evidence that it has worked. Heller concentrates his attack on monetary policy on the imprecise nature of the definition of money used in monetary analysis, the long lags of monetary policy, and monetary models which are set “in a frictionless Friedmanesque world without price, wage and exchange rigidities” (26). While the debate resolved nothing, it might well be viewed as the beginning of a very intense scrutiny of not only fiscal policy but the role of monetary policy to offset short-run shocks in the economy. Since the Friedman-Heller debate in 1969 there have been many studies of fiscal policy’s place in the modern economy with several major studies con- cluding that fiscal policy need not be neutral or perverse in its impact (Blinder and Solow 1973; Friedman 1972). Furthermore, new ap- proaches to measuring fiscal policy’s impact on the economy also ap- peared in the 1970s (Blinder and Goldfeld 1976). Without a commitment to write a full-length monograph on the subject, it would be impossible to review all of the scientific studies concerned with fiscal policy since the late 1960s. Many of the major points of conflict and consensus have, however, been captured in text- books dealing with public finance and macroeconomics. In this section of the paper we summarize briefly some of the conclusions reached by several authors on the efficiency of compensatory finance and the major issues that still confront policy makers regarding short-run stabilization policy. In Robert Barro’s (1984) macroeconomics textbook fiscal policy is accorded slightly more than one page in the chapter on the Keynesian theory of income fluctuations, but there are lengthy sections dealing with the effects of changes in government purchases on output and the price Compensatory Fiscal Policy level, as well as a lengthy chapter on government deficits and the public debt. Although it is difficult to capture, in a brief summary, all the important issues raised by Barro, the following appear to be the salient points. 1. When public spending rises, private expenditures decline by something less than the increase in government expenditures. Furthermore, while output rises due to increased government purchases, this higher output rises less rapidly, the larger the increases in public spending. The effect is an increase in aggregate supply which is less than the rise in aggregate demand; higher interest rates and a “crowding out” of investment de- mand follow. 2. If government purchases are considered in the real money demand func- tion, and given that there are fewer transactions associated with govern- ment purchases, there is less real money demanded (total output fixed). The result: “there is little to predict systematic effects of government purchases on the price level” (324). 3. With respect to government deficits, financed by borrowing, the Ricar- dian Equivalence Theory indicates that such action will have no effect on output, investment, or the real interest rate. In summary, crowding out and the Ricardian Equivalence Theory suggests a very limited role for fiscal policy in a demand-deficient econ- omy. Such a policy may alter the timing of aggregate demand, but today’s gain will be balanced by tomorrow’s loss. The emphasis on fiscal policy in Wilton and Prescott (1986) contrasts with Barro. The theoretical debate between Keynesians and monetarists and empirical evidence on the shape of the IS-LM schedules is carefully analyzed and the authors conclude that “many of the important differ- ences between Keynesians and monetarists reduce to empirical, not theoretical, propositions” (170). The open economy model with fiscal policy instruments is carefully scrutinized by the authors who contend that the effectiveness of fiscal policy will depend on such factors as balance of payments equilibrium and the nature of the exchange rate regime. The “basics” seem to have survived. As to the success of supply-side tax reductions to lower infla- tion and raise output, the authors conclude that it is only workable if price expectations are reduced permanently. Wilton and Prescott incorporate a price expectations Phillips curve in the IS-LM framework and examine fiscal and monetary policy in a comparative static framework. They conclude: Sil 7/ 318 Douglas A. L. Auld While the basic conclusions . . appear to be very damning for Keynesian economists, one should not lose sight of the fact that during the disequilib- rium adjustment period . . . the economy experiences income levels which differ from the equilibrium level . . . [and] if one focuses on the short run... change in government expenditure does affect output and inflation. (302) The authors conclude that in a temporal framework their model is consistent with both monetarists and Keynesians; it simply depends on one’s definition of the short and long runs. A tax- or expenditure- induced shock to the model will, in the long run, have no impact on output, income, or employment (ceteris paribus). In the short run it will have an impact.3 Finally, in summarizing the monetarist-Keynesian debate over the use of fiscal policy, the authors conclude: there are indeed serious economic and political limitations to the implementa- tion of a successful, consistent, Keynesian stabilization policy. . . . On the other hand, if a large demand shock is predictable, . . . then Keynesian aggregate demand offsets which are reversed when the shock ceases would stabilize output levels and avoid painful short run adjustments implicit in non interventionist policy. (470) Robert Gordon’s Macroeconomics (1984) lists fourteen parts of chapters and two full chapters on the subject of fiscal and stabilization issues. The conflicting goals of stabilization policy are introduced with references to the efficacy of changes in taxes or public spending to stabilize the economy when either unemployment or price inflation problems need corrections. The “crowding out” effects of fiscal policy are reviewed and the size of the effect related to the LM schedule. The debate over the elasticity of the LM schedule and the interest response in the demand for money is scrutinized, and the author concludes that the strict claim of some monetarists of a vertical LM curve should be dis- missed. The concept of crowding out is taken up again in the context of flexible prices. Here the author demonstrates that fiscal policy or any change in autonomous spending has no effect on real output when prices are completely flexible. This theme is again reviewed later in chapter 15. The nonmonetarist/monetarist debate is dealt with in considerable detail, and Gordon argues that the “old-fashioned interpretation of the monetarist debate” is centered on the potency of fiscal policy and shape of the LM schedule. “The real dispute between monetarists and non- monetarists has little to do with the relative potency of monetary versus Compensatory Fiscal Policy 319 fiscal policy.” At the heart of the debate is the source of instability in the economy. Gordon then provides an overview of the two sides of the debate in terms of spending instability, price flexibility, perverse effects of discretionary policy, and the volatility of the natural rate of unem- ployment. Each of these issues is dealt with in detail, concluding with a utopian “activists’ paradise.” As an introduction to the section on policy review, the author states “we will find there is reason to doubt the validity of several of the characteristics necessary for activist stabiliza- tion policy to achieve perfect control . . . the argument for activist government intervention must be that imperfect activist control is better than the monetarist approach, not that activist control is perfect” (404). Gordon sums up the activism versus noninterventionist debate in the following way: the opposition of monetarists to countercyclical activism is mainly based on the poor past performance of the government, the inability of economic forecasts to look far enough ahead to overcome long policy lags and, perhaps more fundamentally, a deep distrust of government. Most proponents of activism do not deny these accusations, except the last... lags . . . do not appear to be more than a year in duration, certainly a short enough delay to allow ample room for an activist policy to counter a three or four year investment slump, even if not enough room to counter a six-month economic pause.” (553) If we examine recent public finance textbooks, the subject of fiscal stabilization policy is excluded in most of them (Boadway and Wildasin 1984; Stiglitz 1986; Tresch 1981). The topic has survived in Richard and Peggy Musgrave (1984) and Auld and Miller (1982), and these conclude our review of recent thinking about fiscal policy. In Musgrave and Musgrave the authors have assigned over 125 pages, one complete section of the book, to compensatory finance. We cannot even begin to summarize the content of these pages. Instead, the focus will be on summaries or conclusions associated with the usefulness of fiscal policy to offset fluctuations in output, employment, and prices. The Musgrave text recognizes clearly the complicating factors which have rendered fiscal policy less attractive in the 1970s and 1980s. The authors recognize that the economy may exhibit rising prices at less than full employment due to a shift in the aggregate supply schedule, making traditional anti-inflationary policy unattractive since a rise in taxes would likely increase unemployment and perhaps do little for inflation. 320 Douglas A. L. Auld Furthermore, the model is complicated by introducing price expecta- tions and relationships between wage inflation, the growth in monetary stocks, and actual price inflation. Reviewing this model with all its interrelationships, the authors conclude: “policy makers take a severe risk if they rely on the equilibrating adjustments of the above models to run their course hoping that this will terminate inflation and re-establish full employment” (629). After reviewing the basic tenets of the rational expectations model and the conditions required to ensure ineffective discretionary policy, they conclude “While of interest in defining a model in which policy would be ineffective, the theorem is hardly a realistic appraisal of policy potentials. Given the real-world setting, stabilization policy though diffi- cult, can be effective” (630-31). The authors also conclude that the Ricardian Equivalence Theorem, which renders loan-financed fiscal pol- icy ineffective, also lacks validity because of the restrictive assumptions required to ensure neutral fiscal policy. Auld and Miller devote four chapters to fiscal policy theory, opera- tion, and evaluation. There is also a chapter on public debt and analysis of budget deficits. The theoretical chapter is based on the standard IS- LM framework, incorporating such topics as “crowding-out,” the dy- namics of fiscal policy, and the impact of tax increases on the rate of inflation. The theoretical chapter also includes a discussion of the role of fiscal policy (in an open economy) and its effect on the trade-off between inflation and output. There is a chapter that details the built-in aspects of fiscal policy and a further chapter that examines the role of specific fiscal instruments, for example, the personal income tax rate and the sales tax rate. The importance of time lags on fiscal policy efficiency are also reviewed. Notwithstanding the various limitations of fiscal policy, the authors, in the end, conclude: “Some instruments can be used selectively to alter demand in specific regions or affect specific products or income groups. ... [T]hese instruments depend upon certain conditions in the economy or particular patterns of behavior if they are to achieve their goal” (296). Summary and Conclusions To summarize the debate surrounding the effectiveness of fiscal policy is to also summarize the changes that have occurred in the profession’s Compensatory Fiscal Policy thinking about macroeconomic theory and monetary economics; it is the debate between Keynesians and monetarists over such issues as the causes of business fluctuations, the inherent stability (or instability) of the economy, the elasticity of the quantity of money demanded with respect to the interest rate, the elasticity of the supply of money, and the response to real expenditures to changes in the interest rate. As Blinder and Solow stated, “There is no theoretical controversy over this second level of crowding out. The only contested issues are empirical” (1973: 321). The empirical debate continues. For example, on the question of whether or not bond-financed tax cuts stimulate expenditure, Boskin (1987), Eisner (1986), Feldstein (1982), and Modigliani and Sterling (1986) have found evidence to support this relationship. However, Aschauer (1985), Barro (1978), Leiderman and Razin (1988), and Tan- ner (1979) have found no such evidence. These differences lead of course to very different assessments regard- ing the relative effectiveness of fiscal policy and monetary policy al- though the “gap” between the two sides narrowed throughout the 1970s. Still, certain fundamental differences remain, linked perhaps more closely to political views than those of a strictly economic nature. Today monetarists tend to support a noninterventionist role for gov- ernment while neo-Keynesians, while recognizing the limits of discre- tionary policy, would support fiscal and/or monetary policy to stabilize the economy. Monetarists tend to support a “rules” approach to eco- nomic control: let the money supply increase at a constant predeter- mined rate and, given flexible prices, the economy will stabilize itself without wild fluctuations. Keynesians argue that experience shows that prices are not downward flexible and waiting for the economy to natu- rally stabilize itself after a shock would be unacceptably costly. While not unanimous, the answer to Blinder and Solow’s question, “Does Fiscal Policy Matter?” (1973), would appear to be a cautious and qualified yes: cautious because there are still widespread differences of opinion about the degree of crowding out that occurs with bond-f- nanced tax reductions or expenditure increases and qualified in terms of whether or not the alternatives to fiscal policy are more efficient. In many countries today, however, the use of fiscal instruments to avert a major recession is constrained by the size of current deficits and accumulated public debt. Until these are reduced to more “manageable” levels, it is unlikely that compensatory finance will play a major role in counter- cyclical policy. 321 322 Douglas A. L. Auld Notes I would like to thank Lorraine Eden for her insightful comments on and sugges- tions for improving the first draft of this paper, and John Sargent for his com- mentary on the paper during the conference. 1. Reducing indirect taxes to reduce inflation is somewhat “un-Keynesian,” but does support the prescriptions of supply-side economics. 2. A detailed analysis is found in Barro (1981, 1984). The validity of the theorem is questioned by O’Driscoll (1977). 3. This is not significantly different from the position taken by Barro (1984) and others. But is this really new? The fiscal policy models of the 1950s were, by and large, static models that concentrated on the effects of fiscal policy after a single time period. What we tend to forget is that fiscal policy was seen as a way of “filling-in” when there was an autonomous change in investment or exports which was pushing the economy toward undesirable levels of inflation or unem- ployment. It was assumed, perhaps not always explicitly, that these autonomous components of aggregate demand would return to their long-run equilibrium levels permitting a withdrawal of temporary fiscal policies. Today, we have a somewhat better idea of what the adjustment processes involve and the costs associated with some of those adjustments. 16. JOHN B. BURBIDGE Social Security and Public Debt in Historical Perspective It is second nature for economists of my generation to think about social security and public debt in terms of the Samuelson (1958b) and Diamond (1965) life-cycle growth model. We know, or think we know, that a more generous social security program, or an increase in govern- ment debt to finance transfer payments to the current older generation, benefits the current old at the expense of younger generations whose welfare declines as a consequence of decreases in the community’s capi- tal stock.! To understand what Carl Shoup and the economists of his generation have to say about social security and public debt one must escape from the modern mind set. Perhaps others find this transition easy, but it is one that I have found to be very difficult. The present paper should be regarded as a progress report; others will have to judge to what extent I have been successful. Escape from “habitual modes of thought” requires a heightened awareness of important, often implicit, assumptions. In the next section, I argue that there are at least two critical assumptions underlying the Samuelson-Diamond results: (1) the relative saving rates of different groups in the economy and (2) the beliefs of those who pay the taxes or hold the debt about the likelihood of their receiving benefits or of the debt being repudiated. With regard to the first, Samuelson and Diamond adopted the as- sumptions of the simple life-cycle model—people are born with no assets, die leaving no bequests, and save when young to dissave in old age. These assumptions were not typically made by earlier writers; moreover, recent research has cast doubt on this model; in particular, it is far from obvious that the saving rate of the old is lower than the saving rate of the young. If it were the case, for example, that the young and old 324 John B. Burbidge had the same saving rate, then redistributions of income between them would have no impact on aggregate saving. Since changes in social security and public debt are, at least in part, intergenerational transfers, it seems clear that the first assumption is important. On the second assumption much has been written about the expecta- tions held by younger generations concerning their future social security benefits or the likelihood of certain governments repudiating debt. In an extreme case, if the young expect to receive no social security benefits or believe that government debt is nearly worthless, then expansion of social security taxes and benefits, or forcing government securities on the public, will have effects much like those of a series of temporary taxes on the young. Here steady-state equilibria are of little help, and the analyst must contemplate models out of long-run equilibrium. These two points are discussed further in the next section. Sections 3 and 4 take up Shoup’s work on social security and public debt. The last section presents a summary and some conclusions. Identifying Key Assumptions The purpose of this section is to persuade the reader that relative saving rates and expectations of future benefits are critical for the standard conclusions on the effects of social security. I proceed in three steps. First, I examine redistribution without public debt; the example discussed is tax reform and, in particular, switching between wage and consumption taxation. The modern result here is that a consumption tax induces a higher capital stock and a higher level of steady-state utility than a wage tax.? I adapt Diamond’s (1965) model to show that one can reverse this result if the old have a higher propensity to save than the young. Next, I look at the effects of social security when those who are being taxed do not expect to receive any future benefits, which means there is no debt. Finally, I introduce debt—those who are taxed when young “know” they will receive benefits when old. Redistribution—Tax Reform Consider a setting in which each generation consists of identical individ- uals who live for two periods. Each individual supplies one unit of labor Social Security and Public Debt in the first and 6 = 1 in the second. Let each person have a utility function that is additively separable across time and exhibits constant relative risk aversion in each period.? Let the production technology be cEs with constant returns to scale. It is an identity that private assets plus public assets less public debt equal the capital stock. Since the capital stock must be positive and I wish to admit the possibility that private assets are negative, net public assets must be positive. Unless the government uses some instrument such as a tax rate to control the evolution of public assets and debt, these will determine the dynamic behavior of the model. To make the role of public assets as neutral as possible, I assume that the government uses the proportional consumption tax as an instrument to fix the ratio of public assets to labor supplied. For the moment I set the level of public debt equal to zero. In this paper I shall employ only numerical simulations to illustrate how the models work and the critical assumptions upon which they rest. Let the model be in steady-state equilibrium in periods 1 and 2, with the wage-tax rate, ¢, set equal to zero. Suppose that at the beginning of period 3, ¢ is, unexpectedly and permanently, raised to 0.05. The cap- ital-labor ratio, k, is predetermined in period 3, and thus output, Y;3, the interest rate, r,, and the wage rate, w3, are set for period 3. Under adaptive expectations* the model can be solved for c 5, €12, k4, X43, and h,. Using this procedure recursively, one can generate this model’s dy- namic response to substituting a wage tax for a consumption tax. It is well known that in a partial-equilibrium life-cycle model propor- tional wage and consumption taxation are identical. If it so happened that, in the corresponding general-equilibrium model, young and old had identical saving rates, then the tax switch, which is simply a re- distribution of income, would have no impact on aggregate saving and therefore no impact on the capital stock or the level of utility. This possibility is illustrated in table 1. In a life-cycle model, where initial and terminal assets are zero, the borderline case in which the tax switch has no effects occurs when the young and the old have a zero propensity to save. The standard result that a consumption tax is more efficient than a wage tax depends on the usual life-cycle assumption that the young have a higher saving rate than the old. If the young had a lower propensity to save than the old, then the substitution of a wage tax for a consumption tax would raise aggregate saving and the welfare of each individual in steady-state equilibrium. S25 326 John B. Burbidge Table 1: A Simulation of the Effects of Switching from a Wage Tax to a Consumption Tax Period Che Gy u, y Ww, f 1 0.17027 0.12771 —7.25334 0.27027 0.05113 2.19138 2 0.17027 0.12771 = LISS 0.27027 0.05113 2.19138 3 0.17027 0.12771 = e2IOO4 0.27027 0.05113 2.19138 4 0.17027 0.12771 —7.25334 0.27027 0.05113 2.19138 5 0.17027 0.12771 —7.25334 0.27027 0.05113 2.19138 oo 0.17027 0.12771 —7.25334 0.27027 0.05113 2.19138 Period h, e, a, k, , A, 1 0 0 0.10000 0.10000 0 —0.69970 2 0 0 0.10000 0.10000 0 —0.69970 3 0 0 0.10000 0.10000 0.05000 —0.71471 4 0 0 0.10000 0.10000 0.05000 —0.71471 5 0 0 0.10000 0.10000 0.05000 —0.71471 oo 0 0 0.10000 0.10000 0.05000 —0.71471 Parameters: a = 2; B = 0.3; y = 2; 6 = 0.75; p = 4.67315; n = 1. The most recent literature is surprisingly obscure on why consump- tion taxation appears to be more efficient than wage taxation in growth models, and it may be worth digressing for a moment to put the results derived above, that is, the pivotal role played by relative saving rates, into historical perspective. Summers attributed the efficiency gains from switching to consump- tion taxation to the “postponement” effect—on the assumption that consumption is concentrated later in the life cycle than earnings, “con- sumption taxes postpone tax payments relative to wage taxes . . . [and thus] . . . consumption taxation . . . reduce[s] the steady-state present value of taxes paid by the representative individual” (1981:539). How- ever, Auerbach and Kotlikoff observed that the consumption tax, unlike the wage tax, acts as a lump-sum tax on initial wealth (1987a:52). While these “explanations” touch on the key factor—relative saving rates— Social Security and Public Debt 327 they are not nearly as clear as Musgrave who, in the context of a simple Keynesian model, draws out the implications of different propensities to consume for tax policy (1959:438—43). Likewise, Atkinson and Stiglitz contrast the effects of substituting capital taxation for wage taxation in “classical” and “life-cycle” savings models, and conclude that: with “pure classical savings .. . [where] . . . the capitalist class had the larger savings propensity .. . a transfer to wage earners reduced the rate of accumulation. [With the standard life-cycle savings model, in which]... the wage-earning generation are the only savers, . . . a transfer to this generation raises savings” (1980:248). Redistribution—Social Security, with No Debt The logic of the previous argument shows that any program that is purely redistributional (and therefore creates no debt) will have no effects on the capital stock so long as young and old have the same saving rate. As an example, consider a one-shot social security program in which a temporary wage tax on the young and old in period 3 is used to finance transfers to the old in period 3. Observe that there will be no debt because the young cannot anticipate any benefits from the program.> Table 2 reports simulation results for a one-period wage tax which is implemented at time 3. One can be quite explicit about the gains for the old of the time-2 generation and the losses for the young of the time-3 generation. Since the capital-labor ratio does not change, and therefore y does not change either, the change in c), equals (1 + n) times the change in c,3, which can be verified by inspection of table 2. That (1 + n)c;, + C>,_1 is constant also implies the consumption tax rate does not change. It may be helpful to contrast these results with the ones that derive from the Diamond model. Although a temporary wage tax would not cause any permanent changes in the Diamond model, it would induce a temporary decline in the capital stock because in the standard life-cycle growth model the young save and the old dissave. The temporary decline in the capital stock would mean lower levels of welfare for some genera- tions born after time 3. Redistribution—Social Security with Debt Even if young and old have the same propensity to save, the introduction of a permanent social security program will cause a reduction in the 328 John B. Burbidge Table 2: A Simulation of a One-Shot Social Security Program Period Cre recy. u, y, W, fi 1 0.17027 0.12771 —/a2008 0.27027 0.05113 2.19138 2 0.17027 0.14473 —7.09096 0.27027 0.05113 2.19138 3 0.16176 0.12771 —7.56245 0.27027 0.05113 2.19138 4 0.17027 0.12771 —7.25334 0.27027 0.05113 2.19138 5 0.17027 0.12771 —7.25334 0.27027 0.05113 2.19138 oo 0.17027 0.12771 aL OSA 0.27027 0.05113 2.19138 Period h, e Ey. k, , A 1 0 0 0.10000 0.10000 0.00000 —0.69970 2 0 0 0.10000 0.10000 0.00000 —0.69970 3 0 0 0.10000 0.10000 0.05000 —0.69970 4 0 0 0.10000 0.10000 0.00000 —0.69970 5 0 0 0.10000 0.10000 0.00000 —0.69970 oo 0 0 0.10000 0.10000 0.00000 —0.69970 Parameters: a = 2; 8B = 0.3; y = 2; 8 = 0.75; p = 4.67315; n = 1. capital stock because the young will consume more in anticipation of future social security benefits—this is the sense in which social security creates “debt.” The only difference between the simulation summarized in table 3 and that in table 2 is that the payroll tax rate increase is permanent in table 3. The presence of debt induces a permanent reduction in the capital stock because the young expect to receive social security benefits in their old age and therefore consume more. Looking back across the three cases it should now be apparent that the standard results on social security or government debt are heavily dependent on these assumptions: (1) the saving rate of the young is higher than the saving rate of the old, and (2) the young expect to receive benefits in accordance with the same formula that has been used to pay benefits to the current elderly, that is, the “debt” will not be repudiated. Social Security and Public Debt Table 3: A Simulation of a Permanent Social Security Program Period Ci Cor ur y, W, ff 1 0.17027 0.12771 —7.25334 0.27027 0.05113 AMS) 2 0.17027 0.14223 =e Z il 0.27027 0.05113 2.19138 3 0.16479 0.12696 —7.45667 0.27027 0.05113 2.19138 4 0.16955 0.12722 —7.28358 0.26885 0.05060 2.19677 5 0.16947 0.12721 —7.28649 0.26887 0.05060 2.19668 20 0.16947 0.12721 —7.28645 0.26887 0.05060 2.19668 Period h, er ae k, , A, 1 0.00000 0.00000 0.10000 0.10000 0.00000 —0.69970 2 0.00000 0.00000 0.10000 0.10000 0.00000 —0.69970 3 0.00000 0.00000 0.10000 0.10000 0.05000 —0.69441 4 0.00014 0.00079 0.10000 0.09935 0.05000 —0.70616 5 0.00015 0.00079 0.10000 0.09936 0.05000 —0.70597 00 0.00015 0.00079 0.10000 0.09936 0.05000 —0.70597 Parameters: a = 2; B = 0.3; y = 2; 6 = 0.75; p = 4.67315; n = 1. Against this background we can begin to understand Shoup’s research on social security and public debt. Shoup on Social Security It is startling to reread® the four pages of Public Finance devoted to the analysis of old-age and survivors’ benefits programs. Here pay-as-you-go social security is described, possible rationales for it are discussed, and its effects are considered. The pages are notable for their evenhandedness and common sense, and they stand in marked contrast to much of the literature that has emerged subsequently. One wonders where our knowledge of social security would stand today if these four pages had 529) 330 John B. Burbidge been used to establish the profession’s research program on social se- curity. Throughout his book Shoup forces the reader to think about the environment in which real-world governments must try to function. One theme emphasized is the replacement of the family by the state, which is clearly a (the?) dominant feature of most economic development. His discussion of social security fits naturally into this setting. Care of the elderly used to be the responsibility of the family. In the early decades of this century there arose substantial pressure, from the young as well as the old, for a system of publicly controlled transfers. Many research topics emerge from these pages. A central one is that researchers should attempt to construct models that simultaneously explain: (a) the demand for impersonal public transfers to the elderly and infirm; these may reduce family tensions caused by deciding who is going to contribute what to those relatives no longer able to care for themselves; public pensions may also reduce the horizontal inequities induced across families by differences in parental longevity and health; (b) the desire, by a large fraction of the population, to make personal contributions to particular charities. Somewhat surprisingly, economists have been very reluctant to pursue this topic;” surely it merits more attention in future research. By way of contrast, much of the more recent literature assumes smoothly functioning families and leads logically to a world where not even relative prices matter (see Bernheim and Bagwell 1988). The “Ricardian equivalence” or Barro “debt neutrality” (1974) result is stated only to be dismissed (correctly, in my view) as being unlikely: “At the extreme, the older generation would finance its own benefit pro- gram; the current taxes paid by the younger generation would be offset exactly by the reduction in intra-family transfers” (Shoup 1969b:166). With regard to the effects of social security on aggregate saving, Shoup states: “under a government-financed old-age system compared with no such system, saving may increase over a period of decades and hence, if total input is the same in either case, investment rises and consumption falls. But there are so many forces at work that no general- ization seems feasible” (167). He points out that the reduction in uncer- tainty under a public system might encourage or discourage saving by the young; “intense intra-family demands on the younger generation” in the absence of a social security system might depress the consumption of both young and old relative to what would occur with a public system; if Social Security and Public Debt 331 social security benefits are heavily income conditioned, and therefore are reduced significantly for each extra dollar of interest income in old age, the introduction of a social security program may decrease saving. In some respects section 2 above illustrates his points—the effects depend on the relative saving rates of young and old and the degree of uncertainty in the economic environment in which agents are responding to changes in the program. Looking back from our present vantage point at the immense literature that has grown up on the effects of social security, how much have we learned? | think “no generalization seems feasible” sums up the present state of the subject pretty well. It is clear that the effects of social security programs depend on the model used; is the standard framework used to study social security rich enough to encompass the key reactions people have exhibited to changes in social security programs? Here I think the answer is no; future research should concentrate on identifying the main reasons government intervention is desirable and then turn to understanding the effects of government actions in these environments. This part of Public Finance is a sensible place to start. Shoup on Public Debt The older literature on “the burden of the debt” has been obscured by the literature that has mushroomed from the Samuelson (1958b), Dia- mond (1965), and Barro (1974) articles on the effects of public debt. While this older literature had its fair share of confusions, contradic- tions, and controversy, and Shoup played a key role in sorting out the issues, this literature was not centered on models with extreme assump- tions as is the modern literature—for example, the assumptions of either the pure life-cycle model with no bequests, or Barro’s dynastic model, in which there is consistent family planning with an infinite horizon. Earlier authors appear to have been trying to understand the effects of debt versus tax financing of expenditures in the large middle ground left unexplored in the modern literature. For Shoup debt is “a device to substitute future taxation for current taxation” (1969b:441—42). He identified the “Ricardo-Pigou effect,” the effect of debt as opposed to tax financing on the stock of capital, as being the key issue in whether or not presently incurred debt burdens future generations: “only through a reduction in the size of the capital 332 John B. Burbidge stock inherited, can debt finance burden a future generation” (443 n.2). Once again, Ricardian or Barro debt neutrality is anticipated and then dismissed as being an unlikely outcome: “If the stream of future addi- tional taxes is fully discounted, the bond issue will cause the private sector to consider itself, on balance, poorer by an amount equal to the bond issue”—but since no one can be sure about who will pay the future taxes, this equivalence of present and future taxes is unlikely to hold (442). Like his ideas on social security, Shoup’s work on debt is character- ized by common sense and evenhandedness. These traits are particularly evident in his 1962 article in the Economic Journal. Here he sorted out the confusion in Bowen, Davis, and Kopf (1960) and in the work of other authors to the effect that debt could somehow burden future generations in some way other than by decreasing the capital stock. While one can construct strange examples in which this proposition fails to hold, the point is extremely important in understanding the effects of debt in a closed economy context.® Summary and Conclusions This paper has juxtaposed old and new approaches to social security and public debt. Certain contrasts seem to stand out. The older approach, while less mathematical and perhaps less precise as a consequence, was directed at understanding what governments do in environments charac- terized by a good deal of uncertainty. Its hallmark is an evenhanded, “chips-fall-where-they-may” style which contrasts with much of the more recent literature’s strident attack on social security and public debt. Here very strong policy conclusions are deduced from models that seem to have only a nodding acquaintance with reality. I would draw the following lessons from this evolutionary process. First, knowledge is not cumulative; important ideas that existed in the older debates on the effects of government debt or social security have been lost. For example, very few modern writers on social security have paid enough attention to the “intense intra-family demands” endured by young and old in a world without social security, as each family at- tempted to care for aging or disabled relatives. Indeed, parts of the modern literature hold up the model of the nineteenth-century family as an ideal to which society should return.? Second, each generation tends Social Security and Public Debt 333 to take critical assumptions for granted, assumptions which are strange or questionable to people in other times and places. The rule to state all important assumptions cannot be (re-)emphasized too much. Third, surely significant gains in knowledge could be achieved by doing re- search in the eclectic middle ground where model selection is tied more closely to empirical facts than to the appeal of its policy conclusions. Appendix Notation ¢;, consumption in period i by someone born at time t u, utility level of someone born at time t y, Output per unit of labor supplied at time t W, wage rate at time t iB interest rate at time t h, private wealth per unit of labor supplied at the beginning of period t e, public debt per unit of labor supplied at the beginning of period t a, public assets per unit of labor supplied at the beginning of period t k, capital stock per unit of labor supplied at the beginning of period t , proportional wage tax rate at time t h, proportional consumption tax rate at time t a the inverse of the elasticity of substitution between labor and capital B capital’s share in output if technology were Cobb-Douglas (i.e., if « = 1) y the inverse of the intertemporal elasticity of substitution in consumption h) fraction of second-period time endowment devoted to labor supply p rate at which each individual discounts future utility n the population growth rate The Equations of the Model Redistribution—Tax Reform Let someone born at time t have the utility function: 1— i 7 Cit v 1 Cor us u, a aaa - ! > 0: A. thon ilap@ th 9y u Cae where c;, is consumption in period i for generation t, the inverse of y is the intertemporal elasticity of substitution in consumption, and p is the rate at which 334 John B. Burbidge the individual discounts future utility. Assume the individual faces the budget constraint: (1 if Art) Cre (1 — O141)8We4 (1 +.A,)c,, + = = (1 — 6,)w, + 58 wlis2) etree ee ye where w,, f,, A,, and ¢, are the wage rate, the interest rate and the proportional consumption and wage tax rates, all at time t. Maximization of (A.1) subject to (A.2z) yields: (1 = 541) 8We4 (Sos) weet oe Ce ; (A.3) 4 ey hee seen ( ‘ 1 pee = where 1 (1 +A.) (1 + 2,4,) |” a | Ae et A. Me ream | ia Let h,, a,, €,, and k, denote the beginning-of-period stocks of private assets, public assets, public debt, and physical capital per unit of labor supplied. As in Gale (1973) or Burbidge (1983), the asset identity can be written as: he ae eke (A.5) To make the role of public assets as neutral as possible I assume the government uses the proportional consumption tax, A,, aS an instrument to set a, = a > 0, for all t, and, for the moment, I set e, equal to zero. Public assets will therefore evolve in accordance with (A.6): (Neva + ON aia = (N, + ONL.) (1 4 pay ida (No ONS) eee (irc N,—1€2¢—1)5 (A.6) where N, represents the number of people born at time t. If population grows at rate n, that is, N,,, = (1 + n)N,, (A.6) can be rewritten as (A.7). (r, —n)(1 +n + d8)a + (1 +n + 8)w, + AL (1 + ne, + CG, al SOA) Private wealth at the beginning of period t+ 1 is given by: (N.+1 + SNJh.+1 = [(1 — ow. — (1 + A)cIN, or (fn he, = (f—dwet ee (A.8) I assume technology is constant elasticity of substitution (CEs) with constant returns to scale and thus output per unit of labor, y,, is: Social Security and Public Debt 335 1 ye = [(1-8) + Bk,*-o] (A.9) where the inverse of a is the elasticity of substitution between labor and capital and 8 would be capital’s share in total output if technology were Cobb-Douglas (a equal to unity). The competitive factor-pricing equations are: r= Atal (A.10) and Wey nk.. (Atta) Redistribution—Social Security, with No Debt This experiment requires some changes in the formal model. Under this pay- as-you-go social security program, we must have for t =3: N,_1b, = (N, af ON, 1)O,W; or b, = (1+ n+ 8)b,w,, (A.12) where b, represents the benefits paid to each old person at time t. And whereas, in the tax-substitution setting described above, both tax rates entered the public asset identity, equation (A.7), now only the consumption tax rate enters to maintain the target of a fixed level of public assets, a. Finally, those who are old at time 3 will receive an unanticipated windfall from the program and thus their budget constraint implies that, for t =3: (ar r,)[( = When) ere Uae Ne) Gell te Wie O Owe t b, (1 + A,) Ka = (A.13) As I observed in the text of this paper, one can be quite explicit about the gains for the old of the time-2 generation and the losses for the young of the time-3 generation. The individual budgets constraints (A.2) and the asset identity (A.5) imply the gross national expenditure (GNE) identity, equation (A.14). il spin 1 = | == |e, + | ae (il sr = A. Me F fe yal 3 5) (; +n+ 5) gs) ee ee) Since the capital-labor ratio does not change, and therefore y does not change either, the change in c,, equals (1 + n) times the change in c,;, which can be verified by inspection of table 2. That (1 + n)c,, + cj, is constant also implies the consumption tax rate does not change. !° 336 John B. Burbidge Redistribution—Social Security with Debt Here, the N, people born at time t expect to receive (N,,, + 8N,)b,4 W441 in social security benefits at time t+ 1. This means that there must be a third term on the right-hand side of the individual’s budget constraint and thus: (1 ai .41)OWe4t he $.44(1 te 5)Wevt (1 ="o)w, + lar Se. Were co —— (A.15) (UO) eel Lt Ti+ In addition, the equation for the debt per unit of labor supplied at time t+1 will be: ase O.41Wr41 = ; A.16 a 1 a oe The numbers in table 3 are consistent with these equations. Notes I should like to thank John Sargent, Carl Shoup, Lorraine Eden, other conference participants, and Mike Veall for many helpful comments, and the Art Research Board of McMaster University and the Social Sciences and Humanities Research Council of Canada for financial support. 1.1 am making the usual assumption that the interest rate exceeds the population growth rate. 2. See, for example, Summers (1981), Auerbach, Kotlikoff, and Skinner (1983), Kotlikoff (1984), and Auerbach and Kotlikoff (1987a, 1987b). 3. Many of the details of the model and its equations are set out in the appendix. 4. Under rational expectations there are not enough assumptions to solve for all the unknowns in the usual way. Thus some form of the “multiple- shooting” algorithm must be employed to find the dynamic adjustment path. See Blanchard and Kahn (1980) and Lipton et al. (1982). 5. This experiment requires some changes in the formal model, which are described in the appendix. 6. Christopher Green, who was my first instructor in public economics, used Carl Shoup’s Public Finance as the main text. I know that the class was asked to read these pages, and I am sure that all of us did so, but I wish I had come to understand the significance of what he was saying sooner. Social Security and Public Debt 7. Veall (1986) is one exception. 8. An example can be constructed from the model in the section on social security with no debt outlined above. Suppose debt is issued to the young at time 3 to pay benefits to the old at time 3. Before the young can make their consump- tion decision, the government announces that it is not going to redeem the debt; that is, the young at time 3 are told that the debt certificates are worthless. In this setting the debt and the one-time tax policy should be equivalent. If the young and the old have the same propensity to save, this policy would result in a transfer from the young to the old, with corresponding changes in lifetime welfare, but no change, not even a temporary one, in the real capital stock. This way we could have the younger generation “burdened” even though the capital stock does not change. 9. See Gagan (1981) for a description of some familial systems of providing social security in nineteenth-century Upper Canada. His discussion of the par- ticipants’ sense of “domestic calamity” nicely illustrates the effects of “intense intrafamily demands” mentioned above. 10. This follows from the modified equation (A.7), from which the wage-tax term has been dropped. 337 Comment JACK M. MINTZ The Role of Economic Policy in the Tax Reform Process The papers in this volume reflect different views by economists of the limitations of their analysis in the tax reform process. Richard Bird’s paper stresses the importance of the political process for economic policy making. He argues that tax reform is often driven by exogenous events and that the size of the public sector depends on the tax structure and vice versa. The papers by Charles McLure, Jr., on “Income Tax Reform in Venezuela: Thirty Years after the Shoup Mission” and by Peggy B. Musgrave on “Fiscal Coordination and Competition in an International Setting” offer different perspectives of the role that economists have in the tax reform process. McLure’s paper examines the impact of the 1958 Shoup mission on the Venezuelan tax reform process over the past three decades. He reviews the recommendations made by the Shoup mission and suggests what policies were eventually adopted by the Venezuelan government. He concludes that “there may be tendency for disappoint- ment that so little has been done correctly and so much has been done badly. . . . [G]overnments are not likely to embrace tax reform simply because it is the right thing to do.” Peggy Musgrave’s paper examines the efficiency and internation equity arguments for tax coordination at the international level. Although Musgrave considers to some extent the behavior of governments, her paper concentrates on the type of advice that would be given to a government in designing international tax agreements. One view of economic analysis is that it is irrelevant because it does not properly account for the behavior and attitudes of governments in designing public policies. As a result, economists are “off the mark” in their recommendations for tax policy and often fail in pressing govern- ments to change their policies to meet normative criteria for tax reform. Instead, economists should make tax policy recommendations that take into account the behavior of governments so that their recommenda- tions are more practical. The alternative view is that the role of the economist is to recommend Economic Policy and the Tax Reform Process 339 policies using normative criteria for tax reform such as efficiency, equity, and simplicity. The economist provides guidance that would otherwise not be voiced in the tax policy process. Even if the economist is not always successful in having recommendations accepted, at least someone in the process is providing a viewpoint that is based on certain principles for tax design. The Shoup mission in Venezuela exemplifies this princi- ple. As pointed out by McLure, the mission made several solid recom- mendations that were consistent with conventional economic analysis at that time. The fact that many recommendations were adopted with limited success is indeed unfortunate. However, one might argue that the outcome of the policy process would have been far worse if the Shoup mission had not taken place. There is, nonetheless, an important limitation to this latter view. Even though economists have a role in normative tax policy advice, there remains considerable disagreement with respect to what may be viewed as “conventional” economic thinking. As a result, it is not clear whether the normative advice provided by one economist is in agreement with what would be recommended by other economists. To illustrate limitations to economic policy advice I discuss two exam- ples related to open economy considerations. The first is the problem of providing the correct normative policy advice. As McLure notes in his paper, a conventional recommendation for a good tax system is that corporate and personal taxes should be fully integrated. I wish to con- sider this policy in light of recent analysis that considers integration in an open economy. The second example is related to the relevance of an economist’s policy advice when the behavior of governments is impor- tant in determining the best policy for a country. This issue is well illustrated by recommendations for fiscal policy coordination, the pri- mary topic discussed by Musgrave. Integration What policy advice would an economist give with respect to the integra- tion of corporate and personal taxes? There are two sorts of problems that arise. The first is the type of personal tax base (consumption or income) which affects the design of the corporate income tax. | will not dwell on this issue since it is discussed in detail in other parts of the volume. The second is the problem of integrating corporate and personal taxes for both domestic and foreign shareholders. In an economy closed 340 Jack M. Mintz to international markets it is quite sensible to argue for the full integra- tion of corporate and personal taxes. But what about an open economy? As McLure notes, integration of corporate and personal taxes can be accomplished in two ways. One method is to allow distributed profits to be deducted from corporate taxable income (the “dividend deduction” method). A second method of integration requires that a tax credit be paid to shareholders equal to the amount of underlying corporate taxes on distributed profits (the “imputation” system). The shareholder would pay personal taxes on the “grossed-up” value of dividends (dividends plus corporate tax) net of the tax credit for corporate taxes paid by the firm. This way corporate and personal taxes on dividend income is fully integrated for the owner. Many economists have argued in favor of the imputation system for integrating corporate and domestic personal taxes. However, integra- tion cannot be successfully accomplished in this manner. For an open economy, such as Venezuela, integration of domestic and corporate taxes have a limited impact on the decisions of the firm. The reason is that the firm finances investments from both domestic and international capital markets. Integration of personal and corporate taxes for domes- tic owners does not achieve integration for foreign owners. If firms finance capital at the margin from international markets, a tax credit paid to domestic owners cannot undo the corporate tax paid by the firm and thus fails to achieve integration. Integration of corporate and personal taxes would fully undo the effects of the corporate tax if it is applied to both domestic and foreign shareholders. This could be accomplished by the “dividend deduction” method or the “imputation” method, with the latter requiring a tax credit for corporate taxes paid by both domestic and foreign share- holders. Would an economist wish to recommend integration of this sort? I doubt it in the absence of international fiscal cooperation. For a capital importing country, like Venezuela, integration for foreign share- holders would lead to considerable revenue leakage with, perhaps, little economic gain. The reason is that the foreign shareholder may be able to fully credit any corporate taxes paid to the capital importing country against the tax liabilities owed to the capital exporting country. If a capital importing country used the “dividend deduction” or the “im- putation method” for foreign shareholders, the effect would be a transfer of tax revenue from the importing to exporting country without any change in economic behavior. Economic Policy and the Tax Reform Process Taking into account international considerations, it is not clear what policy advice should be given to a government with respect to integra- tion. If integration is done for domestic shareholders only, it fails to accomplish desired objectives. However, if integration is provided for foreign and domestic shareholders, the capital importing country loses tax revenue which may be “lump sum” in nature. The conventional view in favor of integration of corporate and personal taxes loses some of its appeal in an open economy. Conventional analysis may thus provide wrong policy advice from a normative point of view if the characteristics of the economy are not taken into account. Fiscal Coordination As reflected in Peggy Musgrave’s paper, economic policy advice with respect to the coordination of international tax systems is very difficult to formulate. As Musgrave notes, many issues arise: (1) the appropriate treatment of foreign-source income in a tax system, (2) standards for taxation in fully coordinated systems, and (3) standards for taxation in “minimally” coordinated tax systems. What policy advice would an economist give to a government with respect to fiscal coordination? Is fiscal coordination desirable? If so, how should it be accomplished? An economist interested in providing norma- tive advice only would have some difficulties providing policy advice in this area. The reason is that fiscal coordination depends on the behavior of governments in determining their systems of taxation of inward and outward bound capital. Advice must depend on more than normative criteria—strategic interactions by governments are also important. An economist interested in only normative policy advice would sug- gest that international taxation of foreign- and domestic-source income should lead to worldwide efficiency and internation equity as defined by Musgrave. With residence-based taxes, such as the personal income tax, foreign- and domestic-source income could be taxed at the same rate. The main distortion arises in labor markets if individuals are able to change their residence. As Musgrave notes, in an international economy, mobility of households is restricted so that residence-based taxes can easily vary across countries without being harmonized so long as double taxation of intercountry flows of income is avoided. The problem arises with source taxes such as the corporate income 341 342 Jack M. Mintz tax. To achieve worldwide efficiency, countries would need to fully harmonize their tax systems so that all capital income is taxed at the same rate at the international level. Otherwise, capital flows among industries and countries would be impaired in that firms of either dif- ferent residence or location of capital would be taxed at different rates of corporate income tax. This would require tax systems to be uniform across countries and to avoid double taxation of income flowing from one jurisdiction to the other. Governments would lose considerable independence if they fully har- monized their corporate tax systems. As a result, they try to achieve “minimal” levels of coordination in which only certain conditions for cooperation are agreed upon. An economist interested in providing policy advice on fiscal coordination must consider not only normative criteria but also the behavior of governments. Two views of the behavior of governments have been developed. The first is the “political economy” approach used by Richard Bird that is premised on assumptions such as bureaucrats seeking to maximize tax revenues or governments seeking to maximize political support. Under this approach an economist might conclude that minimal tax coordina- tion may cause governments, that would otherwise compete for tax bases, to choose higher tax rates and greater public expenditure. Fiscal decentralization may be viewed as better for society since it encourages tax competition and less government expenditure. The second approach is to examine tax coordination as if the govern- ment maximizes the welfare of its citizens. In this framework economists examine noncooperative behavior of governments that only take into account national welfare, ignoring the welfare of citizens in other countries. These noncooperative policies can be compared to a coopera- tive policy in which governments fully coordinate their tax systems. Whether governments choose tax rates that are too high or low depends on the nature of fiscal externalities arising from national tax policies. Taxes are chosen too high from a worldwide viewpoint if the corporate tax is paid by foreigners whose welfare is ignored by the capital import- ing country. Tax rates are chosen too low if capital flight results, benefit- ing other countries that experience an expansion in their tax base. Using Musgrave’s terminology, does “minimal coordination” im- prove worldwide welfare? Here, we have little analysis. But one pos- sibility, with respect to corporate taxation, is the following. One of the virtues of corporate tax competition is that it limits the ability of coun- Economic Policy and the Tax Reform Process tries to “export” their taxes. As remarked earlier with respect to integra- tion, taxes on foreign capital withhold income from foreigners or with tax crediting arrangements withhold revenues from foreign treasuries. From the point of view of a single government, reliance on corporate taxation could be highly desirable even though worldwide efficiency considerations would require less corporate taxation. If governments agree to a double taxation agreement, two provisions are often adopted: (1) agreement not to discriminate between domestic and foreign capital and (2) agreement not to tax intercountry flows of income more than once. The first provision imposes a cost on capital importing countries that try to tax foreign capital. That cost is the lost domestic-owned investment that is deterred by corporate taxation. The desire to “export” corporate taxes is reduced by the nondiscrimination clause. The second provision may increase the rate of corporate tax. As discussed above with respect to integration, the crediting system could encourage too much corporate taxation since a capital importing coun- try does not wish to give up revenue that would otherwise accrue to foreign countries. Thus, a provision such as an agreement to credit taxes may make countries worse off. What form of policy advice with respect to international fiscal coordi- nation should be given by an economist? In principle one would recom- mend fiscal coordination. However, if the coordination is of a “minimal” type as discussed by Musgrave, it might lead to a worse situation than having no coordination at all. This conclusion depends on what predic- tions are made with respect to the behavior of governments which is important in determining policy advice in this context. Economists that simply stress normative criteria in this context fail to provide the infor- mation needed by policy makers who must take into account the strate- gic behavior of other governments. 343 Comment JOHN SARGENT On Macro Public Finance The papers by Douglas Auld and John Burbidge use very different approaches to interpret, assess, and extend some aspects of Carl Shoup’s writings in macro public finance from the 1960s. Douglas Auld reviews Shoup’s judicious assessment of what could be achieved and should be attempted through fiscal policy and considers how this assessment has stood the test of theoretical developments in the 1970s and 1980s. A survey of recent intermediate-level macroeconomic texts is used to ob- tain a reading on current views in the profession of the significance of a number of the developments which have gained prominence over the last two decades and which have potential implications for the macroeco- nomic impact of fiscal policy. John Burbidge provides a formal analysis, using an overlapping generations model of household savings in a closed economy with endogenous capital stock, of issues concerning the impact of social security and public debt on savings and on intergenerational burden transfer, which Shoup had earlier analyzed. The perspective which I bring to this discussion is that of someone who was involved in macroeconomic policy analysis in the Canadian Department of Finance from 1971 to 1983 and in the macroeconomic policy research of the Royal Commission on the Economic Union and Development Prospects for Canada in 1983 and 1984. I shall offer some reflections on Auld’s discussion of the analytical developments which have altered conventional views of the potential role of fiscal policy between the 1960s and now and shall point to elements in the experience of this period which also seem important in understanding the evolution of views toward fiscal policy and public debt. While these are personal views, I believe they are reasonably typical of those of many in the applied policy analysis communities in the developed market economies (see, for example, Higgins 1986). Before discussing Auld’s interpretation of recent analytical develop- ments, it is worth noting that notions of the potential role of fiscal policy among policy advisers in the latter part of the 1960s were generally closer to the balanced views of Carl Shoup than to the more utopian On Macro Public Finance visions referred to by Auld as being common at the time of the “zenith of fiscal policy’s popularity and the widespread devotion to macroeco- nomic policy for achieving full employment and price stability.” There was general recognition that the uncertainty to which macroeconomic forecasting is inevitably subject coupled with the lags with which fiscal policy actions have their impacts substantially limited the ability to fine tune the course of the economy. Quantitative macroeconomic models which were then coming into use also suggested that, even abstracting from uncertainty, any attempt to keep the economy within a narrow range of a target path would require continuing alterations in tax rates and expenditure levels, and in the deficit, of a size and frequency suff- cient to give pause to most policy makers and advisers. While the term may not always have been used, “instrument instability” was perceived as a real constraint. I know few if any “literal fine tuners” among those active as practical policy advisers in the 1960s. It is the case that analyt- ical and actual developments of the last two decades have deepened our understanding of these practical constraints on the role for fiscal policy, as well as having raised new issues concerning this role to which I now turn. Among the seven analytical developments which Auld views as hav- ing changed, and generally reduced, expectations as to the macroeco- nomic role of fiscal policy, the “natural rate of unemployment,” or NAIRU, view would seem to be the most important. This view, that there is no long-run trade-off between inflation and unemployment, has gained the widest acceptance and had the most profound impact on perceptions as to the objectives which should be pursued through both monetary and fiscal policy. It probably took five to ten years, after it was first put forward in formal form in 1967 and 1968, for this view to gain wide influence among policy advisers and to become a common feature of quantitative macroeconomic models used inside and outside govern- ment to analyze the impact of policy actions. The inflationary experience of the 1970s, particularly that which preceded the 1973 oil crisis, played a critical role in gaining acceptance of the view. In contrast, full acceptance of the rational expectations approach is much less general, as of course is also the case among academic mac- roeconomists. Certain of the more common sense elements of this ap- proach have gained ground, as reflected in an increased willingness to attribute importance to forward-looking expectations in policy analysis. It should be recognized, however, that policy makers and advisers, 345 346 John Sargent perhaps particularly central bankers, had always recognized the impor- tance of expectations. The natural rate of unemployment model is of course an elementary application of the view that expectations are ra- tional in the longer term. As with rational expectations, several of the themes of supply-side economics have influenced applied policy analysis without there having been a wholesale conversion to the more extreme positions, such as the view that the peak of the Laffer curve occurs at lower tax rates than those commonly in effect. The oil price shock and the productivity slowdown directed attention to the supply side of the economy, both as a source of macroeconomic disturbance, which added to the uncertainties of fore- casting, and as the fundamental factor underlying potential economic welfare. The possibility of trade-offs between short-term demand stabili- zation and growth of potential output—for example, through uncer- tainty resulting from variations in investment tax incentives—became of greater concern. The recent, internationally widespread lowering of marginal tax rates presumably reflected a judgment that supply-side gains dominated any (probably very minor) lessening in the automatic stabilization properties of the tax system. “Tax push” as a factor in inflation could be incorporated in supply- side approaches which call attention to elasticities of factor supplies, but it is also somewhat reminiscent of cost push, nonmonetary explanations of inflation which have generally lost favor. From time to time the tax push analysis did exert some influence on policy; an example was the Canadian decisions to provide fiscal stimulus in 1978 in the form of sales tax reductions. In sum, I agree that the factors listed by Doug Auld have influenced policy thinking; the above comments are intended to provide some further perspective on the way in which they have done so and on their relative importance. In considering the evolution of views or the role of fiscal policy, and macroeconomic policy generally, it is also important to take account of the role of actual economic experience. The inflation of the 1970s, the slowdown in productivity growth, and the growth of public debt have all had direct influence on policy thinking and have affected the receptiveness of policy makers and analysts to the messages of the theoretical developments discussed by Auld. As suggested, experi- ence with rising inflation greatly increased receptiveness to the natural rate view. The oil shocks and the productivity slowdown increased the importance attached to supply-side factors including efficiency and in- centives. On Macro Public Finance 347 The growth in the relative importance of the public debt, for example at the federal government level in Canada from below 20 percent of GDP in the mid-1970s to above 50 percent currently, has had an important influence on perceptions as to the appropriate objectives for fiscal policy which has been somewhat independent of the influence of the analytical developments discussed by Auld. In a world in which nominal interest rates are generally higher than nominal economic growth rates, simple arithmetic indicates that, assuming constancy in the ratios of revenues and noninterest expenditures to Gpp, the debt/cpp ratio will grow without limit unless the deficit is at a level at which revenues exceed noninterest expenditures by a margin that depends on interest rates, the growth rate, and the stock of debt. If the deficit is above that critical level, as it has been in Canada for much of the 1980s, some combination of increased taxes and reduced expenditures relative to GDP is an almost inevitable prospect. Such a situation implies that tax rates are below their equilibrium value for the current relative level of expenditures; the longer they remain below, the higher future tax levels will have to be and the larger will likely be the deadweight efficiency loss associated with future taxes. This would seem to constitute a way in which current deficits can impose burdens on the future additional to the classic routes of lowering physical capital accumulation and increasing foreign indebt- edness. These comments are offered primarily as an extension of Auld’s as- sessment of the factors that have influenced the evolution of views on the role of fiscal policy since the publication of Carl Shoup’s Public Finance in 1969. As Auld and Burbidge state, Shoup’s careful assessment of the range of factors which should be considered in arriving at judgments on fiscal and social security policy issues identified or foreshadowed, and offers many insights regarding, several of the issues that have been prominent over the past two decades. = a! i» sect cor md wa = + (hant) abet ioe -. 5 be > GO ers ) Jue es al nivad bioe