Taxing the financial sector , directly and indirectly, has long been relatively easy even for governments with limited taxing capacity. But grow- ing awareness of the strategic importance of the financial sector in catalyzing economic growth has led governments to reconsider the financial sector’s potential as a budgetary cash-cow. Indeed, the pendulum may have swung to the other extreme, with special interests sometimes arguing for exaggerated fiscal concessions in the name of improved financial sector performance. This book examines the possibilities and pitfalls of successful financial sector tax reform, taxation of financial intermediation whether based on simplifications such as VAT or transactions taxes, or on subtle attempts to use taxes as corrective instruments. Highlighted is the need to make the financial tax system as arbitrage-proof and inflation-proof as possible. Contributors to this volume—all respected academics or practitioners—address distinct areas of taxation.Theoretical chapters model the impact of taxes on intermediaries, the design of optimal tax schemes, the role of imperfect information, and links between taxation and saving. Current practice in the industrial world and case studies of distorted Taxation of national systems provide an empirical underpinning. And discussion of practical issues financial intermediation honohan considers inflation, the income tax treatment of intermediary loan-loss reserves, deposit insurance, the VAT, and financial transactions taxes. Theory and practice for emerging economies This book will prove useful to finance and policy professionals, development specialists, scholars, and students of financial sector policy. edited by Patrick Honohan ™xHSKIMBy354346zv":&:/:$:/ ISBN 0-8213-5434-5 26511 Taxation of Financial Intermediation Taxation of Financial Intermediation Theory and Practice for Emerging Economies Edited by Patrick Honohan A copublication of the World Bank and Oxford University Press © 2003 The International Bank for Reconstruction and Development / The World Bank 1818 H Street, NW Washington, DC 20433 Telephone 202-473-1000 Internet www.worldbank.org E-mail feedback@worldbank.org All rights reserved. First printing June 2003 1 2 3 4 06 05 04 03 A copublication of the World Bank and Oxford University Press. Oxford University Press 198 Madison Avenue New York, NY 10016 The findings, interpretations, and conclusions expressed herein are those of the au- thors and do not necessarily reflect the views of the Board of Executive Directors of the World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the World Bank con- cerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The World Bank encourages dissemination of its work and will normally grant permission promptly. For permission to photocopy or reprint any part of this work, please send a re- quest with complete information to the Copyright Clearance Center, Inc., 222 Rose- wood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750- 4470, www.copyright.com. All other queries on rights and licenses, including subsidiary rights, should be ad- dressed to the Office of the Publisher, World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. The views expressed in chapters 2, 9, 10, and 11 are those of the authors and should not be attributed to the International Monetary Fund. Chapter 11 is printed by permission of IMF Staff Papers, where an earlier version appeared. ISBN 0-8213-5434-5 Library of Congress Cataloging-in-Publication Data Taxation of financial intermediation : theory and practice for emerging economies / edited by Patrick Honohan. p. cm. “A co-publication of the World Bank and Oxford University Press.” Includes bibliographical references. ISBN 0-8213-5434-5 1. Financial services industry—Taxation—Developing countries. I. Honohan, Patrick. HG195.T39 2003 336.2v783321v091724—dc21 2003050061 Contents Foreword xi Preface xiii Contributors xv 1. Avoiding the Pitfalls in Taxing Financial Intermediation 1 Patrick Honohan PART I: THEORETICAL AND SYSTEM-WIDE ISSUES 2. Theoretical Perspectives on the Taxation of Capital Income and Financial Services 31 Robin Boadway and Michael Keen 3. Taxation of Banks: Modeling the Impact 81 Ramon Caminal 4. Tax Incentives for Household Saving and Borrowing 127 Tullio Jappelli and Luigi Pistaferri 5. Corrective Taxes and Quasi-Taxes for Financial Institutions and Their Interaction with Deposit Insurance 169 Philip L. Brock PART II: PRACTICAL EXPERIENCE 6. Taxation of Financial Intermediation in Industrial Countries 197 Mattias Levin and Peer Ritter 7. Seigniorage, Reserve Requirements, and Bank Spreads in Brazil 241 Eliana Cardoso 8. Taxation of Financial Intermediaries as a Source of Budget Revenue: Russia in the 1990s 269 Brigitte Granville v vi CONTENTS PART III: PARTICULAR TAXES 9. Corporate Income Tax Treatment of Loan-Loss Reserves 291 Emil M. Sunley 10. Bank Debit Taxes: Yield Versus Disintermediation 313 Andrei Kirilenko and Victoria Summers 11. Securities Transaction Taxes and Financial Markets 325 Karl Habermeier and Andrei Kirilenko 12. Consumption Taxes: The Role of the Value-Added Tax 345 Satya Poddar 13. The Accidental Tax: Inflation and the Financial Sector 381 Patrick Honohan Index 421 CONTENTS vii TABLES AND FIGURES Table 4.1 Saving Incentives in Voluntary Retirement Funds in Major Industrial Nations 132 Table 4.2 Tax Treatment of Mandatory Retirement Funds in Latin America and East Asia 139 Table 4.3 National Saving Regressions 143 Table 4.4 Mandatory Housing Funds 146 Table 4.5 The Interaction between Mandatory Pension Funds and Housing Finance 149 Table 4.6 Tax Incentives for Education, Health, and Life Protection 150 Table 4.7 Tax Treatment of Borrowing, Selected OECD Countries 153 Table 6.1 Cutting Tax Rates 201 Table 6.2 Examples of Base-Broadening Measures in the 1980s 202 Table 6.3 Dual Income Tax Systems 204 Table 6.4 Tax Exemption of Interest Expenses Depending on the Purpose of the Loan 209 Table 6.5 Marginal Effective Tax Wedges in Manufacturing, 1999 210 Table 6.6 Taxation of Interest Income in Some Industrial Countries 212 Table 6.7 Tax Treatment of Dividends and Capital Gains on Shares, 1998, Resident Taxpayers 213 Table 6.8 Ways of Dealing with Double Taxation 215 Table 6.9 Taxation of General Insurance Companies 229 Table 6.10 Stamp Duties, 2001 231 Table 6.11 Reserve Requirements in Selected Industrial Countries 232 Table 7.1 Taxes and Contributions at the End of the 1990s, Brazil 244 Table 7.2 Loans as Percent of GDP, Brazil, 1989–2000 247 Table 7.3 Real Interest Rates, Brazil, 1970–2001 248 viii CONTENTS Table 7.4 Spreads between Active and Passive Annual Interest Rates, Brazil, 1970–2001 248 Table 7.5 Spreads between Active and Passive Monthly Interest Rates and Net Margins of Commercial Banks, Brazil, 1995–2000 251 Table 7.6 Rates of Required Reserves before the Real Plan, Brazil, 1969–93 253 Table 7.7 Rates of Required Reserves after the Real Plan, Brazil, 1994–2001 255 Table 7.8 Ratio of Bank Seigniorage to Loans, Inflation Rate, and Required Reserves, Brazil, 1970–2000 257 Table 7.9 Determinants of the Spread between Active and Passive Rates after the Real Plan 259 Table 7.10 Determinants of the Commercial Bank Net Margin after the Real Plan 260 Table 7.11 Determinants of the Spread between Interest Rates on Working Capital Loans and Interest Rates on Time Deposits 261 Table 8.1 Central Bank of Russia Credit Allocations, 1992–93 273 Table 8.2 Explicit Subsidies to State Enterprises, Russia, 1992 275 Table 8.3 Seigniorage on Currency, Russia, 1993–2001 279 Table 8.4 Seigniorage on Bank Reserves, Russia, 1993–2001 280 Table 8.5 Minimum Reserve Requirements, Russia, 1995–98 281 Table 8.6 Nominal Interest Rates on Loans and Deposits, Russia, 1994–2001 284 Table 8.7 Deposits in the Banking System, Russia, 1993–2000 285 Table 9.1 Income and Cash Flow from Loan Portfolio 299 Table 9.2 Income and Cash Flow from Loan Portfolio 300 Table 10.1 Bank Debit Taxes, Selected Latin American Countries, 1989–2002 317 CONTENTS ix Table 12.1 Five Main Categories of Financial Services 346 Table 13.1 Backing of the Money Base by Foreign Exchange or Claims on Government 388 Table 13.2 Inflation and Bank Profitability, 1988–95 394 Table 13.3 Inflation and Bank Profitability, 1995–99 397 Table 13.4 Inflation and Net Interest Margins, 1995–99 410 FIGURES Figure 4.1 Pension Fund Assets in Developing Countries 140 Figure 4.2 National Saving and Pension Fund Assets in Developing Countries 141 Figure 5.1 Debt Contracts with Alternative Liability Rules 171 Figure 6.1 Average Taxes on Interest Income on Residents in the EU, 1983–2000 210 Figure 6.2 Average Tax on Financial Wealth in the EU, 1983–2000 220 Figure 7.1 Loans from the Private Financial Sector as a Percent of GDP, Brazil, 1989–2000 243 Figure 7.2 Loans from the Public Banking Sector as a Percent of GDP, Brazil, 1989–2000 243 Figure 7.3 Passive Real Interest Rates before the Real Plan, Brazil, January 1970–June 1994 249 Figure 7.4 Real Interest Rate, Brazil, 1992–2001 251 Figure 7.5 Monthly Spread between Loan and Deposit Rates and Net Margin of Banks, Brazil, 1995–2001 252 Figure 7.6 Seigniorage Collected by Commercial Banks as Share of Loans, Brazil, 1971–2001 256 Figure 7.7 Risk Spread and Average Intermediation Spread, Brazil, 1995–2000 263 Figure 7.8 Operational Costs Relative to Credit, Brazil, 1995–2001 263 x CONTENTS Figure 10.1 Deadweight Loss from Bank Debit Taxes, Selected Latin American Countries, 1999–2001 320 Figure 10.2 Disintermediation Caused by Bank Debit Taxes, Selected Latin American Countries, 1999–2001 321 Figure 13.1 Alternative Sources of the Monetary Base and Their Correlation with Monetary Depth and Inflation 390 Figure 13.2 Bank Profitability and Inflation, 1988–95 393 Figure 13.3 Bank Profitability and Inflation, 1995–99 396 Figure 13.4 Bank Value-Added and Inflation, 1995–99 399 Figure 13.5 How Bank-to Market Ratios Change with Inflation 401 Figure 13.6 Tax Rates at Different Rates of Inflation 405 Figure 13.7 Elasticity of Tax Rates at Different Rates of Inflation 405 Figure 13.8 Net Interest Rates and Inflation, 1995–99 406 Foreword An impressive body of research now supports the proposition that improving the efficiency and effectiveness of domestic financial sys- tems has an important role in accelerating long-term growth. Fur- thermore, avoiding financial crashes through well-designed and im- plemented regulatory policy has a demonstrable and substantial effect in preventing short-term surges in poverty. These two results justify the focus which the World Bank has placed on financial sec- tor research in recent years, and which has greatly enlarged the em- pirical knowledge base on these matters and enhanced the analysis and formation of policy. The research continues, as we deepen our understanding of how this complex and adaptable sector functions and how public policy can best be crafted to ensure that it increases its contribution to the economies of our client countries. Taxation is high on the list of the relevant public policy dimensions. Deciding just how much the domestic financial sector should be taxed, and in what way, is a complex problem for policymakers. On the one hand, governments need revenue and the financial sector is an administratively convenient source. On the other hand, given the central role of finance achieving sustained economic growth, poli- cymakers should not repress the sector’s development by an onerous tax burden. This is clearly a very live topic at present. Quite apart from the Tobin tax—which is outside the scope of this volume, with its focus on domestic tax issues—the financial transactions taxes that have been adopted in Latin America in the last few years have been highly controversial. They have generated a substantial flow of much-needed revenue: revenue that is vital for expenditures that can offer opportunities and support for poor people. But has the cost in terms of distortions been too high? Striking a balance requires a good understanding of how finance works and how the system is likely to adapt to the taxes that are im- posed. This volume provides a valuable toolbox for this purpose. Moving beyond the simplistic mantra that distortions must be avoided, the authors recognize that the financial system should bear its share of taxation, and seek to define criteria for ensuring that the distortions are limited. xi xii FOREWORD A firm theoretical foundation, with several chapters devoted to modeling the behavioral impact of taxation and its potential use as a corrective device, is supplemented by descriptive material on cur- rent practice in advanced economies and some case studies illus- trating how problems can arise. Perhaps most useful to policymak- ers, each of the main types of financial sector tax has a chapter of its own, highlighting the options and pitfalls. In normal times, finance typically makes a sizable contribution to tax revenue, though failing financial systems often involve costly fis- cal outlays. Improved arrangements for explicit and implicit taxation of the sector offer the double prospect of a more stable net contri- bution to the exchequer combined with a pro-growth strengthening of finance. This is an impressive group of contributors, from a diversity of research and policy institutions. I am particularly glad to welcome several contributions from the International Monetary Fund. Re- markably close and productive Bank–Fund cooperation in the Fi- nancial Sector Assessment Program (FSAP) over the past four years has been a hallmark of financial sector work, and I am happy to see further evidence of that cooperation in several dimensions of our re- search activity. Nicholas Stern Senior Vice President and Chief Economist The World Bank March 2003 Preface Taxation of financial intermediation receives surprisingly little ana- lytical attention, despite its practical importance both for national budgets and for the efficient functioning of the financial system. This volume is an attempt to provide a coherent overview of the policy issues involved. The volume opens with a general survey by Patrick Honohan (chapter 1) of the major issues in financial sector tax reform, char- acterizing the main styles of reform that have been advocated. Drawing freely on the remainder of the volume, the first chapter proposes some broad recommendations that should guide policy. The remainder of the volume is organized in three parts. The first part presents the main issues at a theoretical and system-wide level. It leads off with a discussion by Robin Boadway and Michael Keen (chapter 2) of the theory of optimal taxation as it applies to taxation of capital income and financial services. The chapter pays special at- tention to the implications for tax policy of recent innovations in fi- nancial intermediation. Ramon Caminal focuses on banking and presents a simple but powerful model (chapter 3) from which the impact of different forms of taxation on equilibrium behavior can be predicted. The chapter presents some new results on the contrasting impact in competitive and monopolistic environments. Inducements to saving have been a motivation for many tax initiatives affect- ing financial intermediation. In chapter 4, Tullio Jappelli and Luigi Pistaferri survey theory and experience regarding the effectiveness of such incentives and provide some new cross-country evidence. The use of prudential and other forms of financial regulation as a sort of corrective tax, and their interaction with deposit insurance, is con- sidered in chapter 5. Philip Brock provides some new results on the interaction between these. The chapters in Part I are more technical than those in the remainder of the volume. The second part of the volume includes three contrasting chapters on empirical experience. Chapter 6, by Mattias Levin and Peer Rit- ter, reviews recent trends in relevant tax design in industrial coun- tries, highlighting the limited degree of convergence in approach that has occurred. Chapters 7 and 8 describe the rather extreme recent experiences of Brazil and Russia, with quasi-taxes affecting financial intermediation, including inflation and unremunerated reserve re- xiii xiv PREFACE quirements. These two case studies, by Eliana Cardoso and Brigitte Granville, respectively, serve as cautionary tales. The final part contains five essays on specific tax issues. Chapter 9, by Emil Sunley, discusses the principles of loan-loss provisioning and related matters. The following two chapters by Andrei Kir- ilenko with Victoria Summers and Karl Habermeier, respectively, look at financial transactions taxes. Chapter 10 assesses the recent experience with bank debit taxes in Latin America, providing esti- mates of deadweight losses. Chapter 11 illustrates the way in which even small securities transactions taxes can have large affects on the volume of trading of related financial instruments. The practical case for applying a form of value-added tax to financial services is assessed by Satya Poddar in chapter 12. Finally, in chapter 13, Patrick Honohan provides new empirical evidence on the inflation tax and discusses its interaction with other taxes in influencing the scale and profitability of financial intermediation. Much of the underlying research for this volume was funded by the World Bank’s research support budget. In addition several chap- ters have been contributed by officials of the International Mone- tary Fund. The draft chapters were discussed at a workshop in Washington, D.C., in April 2002. Special thanks are due to the dis- cussants at that workshop: Alan Auerbach, Ricardo Bebczuk, Ger- ard Caprio, Stijn Claessens, Liam Ebrill, Roger Gordon, Harry Huizinga, Kyung Geun Lee, Alberto Musalem, and Klaus Schmidt- Hebbel. Their comments greatly contributed to the quality of the final product. Authors also wish to acknowledge additional comments and as- sistance on individual chapters from: Giacinta Cestone and Jorge Rodriguez (chapter 3); Tea Trumbic (for research assistance, chapter 4); Ilan Goldfajn, Eustáquio Reis, Sergio Schmukler, Altamir Lopes, Eduardo Luís Lundberg, Sérgio Mikio Koyama, and Márcio Issao Nakane (chapter 7); Claudia Dziobek and Victoria Summers (chap- ter 9); IMF Staff for the data used in chapter 10; Stefan Ingves, Richard Lyons, and participants at the 2001 Australasian Finance and Banking Conference (chapter 11). Thanks are also due to Agnes Yaptenco for excellent administra- tive support. Patrick Honohan March 2003 Contributors Robin Boadway Queen’s University, Kingston, Ontario Philip L. Brock University of Washington, Seattle Ramon Caminal Institut d’Anàlisi Econòmica, CSIC, Barcelona, and CEPR Eliana Cardoso Georgetown University, Washington, D.C. Brigitte Granville The Royal Institute of International Affairs, London Karl Habermeier International Monetary Fund Patrick Honohan The World Bank and CEPR Tullio Jappelli Universita di Salerno and CEPR Michael Keen International Monetary Fund Andrei Kirilenko International Monetary Fund Mattias Levin Centre for Economic Policy Studies, Brussels Luigi Pistaferri Stanford University and CEPR Satya Poddar Ernst & Young, LLP Peer Ritter Centre for Economic Policy Studies, Brussels Victoria Summers International Monetary Fund Emil M. Sunley International Monetary Fund 1 Avoiding the Pitfalls in Taxing Financial Intermediation Patrick Honohan Because the financial sector keeps systematic accounts and acts as a gatekeeper of liquid resources, it provides many useful tax “han- dles” for the fiscal authorities. In many countries, a distorted struc- ture of financial sector taxation has evolved, reflecting both inertia and opportunism. Inertia, in that some very old taxation and ad- ministrative practices have survived in the financial sector, reflecting the convenience of collection and the fact that those liable to com- ply are a small and privileged group of regulated intermediaries that are unlikely to be politically vocal and that can, in any event, pass on much of the tax to their customers. Opportunism, in that a sud- den need for budgetary revenue can trigger an increase in reliance on this quick and reliable source. Growing awareness of the strategic importance of the financial sector in catalyzing economic growth, combined with the increasing global competition in financial services, has led to a substantial re- consideration of the domestic financial sector’s potential as a cash cow for the budget.1 Indeed, in some instances the pendulum may have swung to the other extreme, with special interests successfully arguing for exaggerated fiscal concessions in the name of improved financial sector performance. This volume provides the basis for considering proposals for fi- nancial sector tax reform in a comprehensive light. The focus takes account of empirical realities in middle-income developing coun- 1 2 PATRICK HONOHAN tries, though many of the principles have wider application. The dis- cussion is confined to domestic financial intermediation (thus ex- cluding issues of international cooperation or harmonization), though recognizing that application needs to take account of the interna- tional environment. Financial sector taxation is complex both in theory and practice and it needs to evolve in response to financial innovation and wider economic changes.2 The volume does not propose or attempt to offer a simple blueprint, but the discussion of theoretical, empirical, and practical considerations does lead to a number of guidelines for de- veloping what would constitute a good financial sector tax system. Proposals for financial sector tax reform typically come from one of two powerful perspectives. The reformer may be an enthusiast for a big simplification, usually some form of “flat tax,” including VAT on financial services, zero taxation on capital income, or a uni- versal transactions tax. Or the reformer may be an advocate of sub- tle corrective taxation designed either to offset some of the many market failures to which the financial sector is prone or to achieve other targeted objectives. In practice, just like the perennial conflict between simplicity in tax administration and economic efficiency of the tax rates, the two perspectives can conflict rather severely. Information and control requirements of much of corrective taxation tend to be poorly ac- commodated by the big simplifications. This tension has remained unresolved over the years. Elements of each approach have become embodied in the taxation, explicit and implicit, of the sector. At the same time, the ever-pressing demands of revenue intrude as a fur- ther influence on policy design. As a result, the tax systems in most countries often end up as a complex mixture defying any straight- forward rationalization. The big flat-tax ideas are diluted and mod- ified; the corrective taxes may misfire by conflicting with others in- troduced for different reasons. Meanwhile, even as simplification and correction continue their tug of war, policy design can all too often neglect the two distinc- tive traps into which financial sector taxation can fall: namely, the sector’s unique capacity for arbitrage and its sensitivity to inflation and thus to non-indexed taxes. This chapter argues that the practi- cal design of financial sector taxation should be guided by a defen- sive approach in which proposed taxes are assessed relative to their ability to resist arbitrage and their degree of inherent indexation. Although the defensive approach does not provide an adjudication between simplification and correction, it will protect against many of the worst distortions that have been observed. AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 3 As a working standard, a tax system is considered to be good for the financial system if it meets three main criteria. It minimizes the distortions it imposes, for a given amount of revenue collected, es- pecially by causing the formal financial sector to be bypassed through disintermediation to untaxed or differently taxed competitors. It is corrective of known distortions, such as those that result from im- perfect and asymmetric information. Finally, it does not push tax collection from the sector beyond the point where marginal distort- ing costs exceed those elsewhere in the economy. With these criteria in mind, the theory and experience described in the remaining chap- ters of this study do suggest some key practical guidelines. First, while reformers should not expect to find a complete and practical solution in any of the “big ideas,” each has lessons for a good system. • The notion of a value-added tax (VAT) on financial services— even if practicalities impede its introduction as such—represents a useful benchmark against which existing and proposed indirect taxes can be compared for their burden and impact. • Significant financial transactions taxes are hard to justify on theoretical grounds and should be resorted to only as a transitory device when fiscal revenue is under particular pressure. • Heavy emphasis on the taxation of income from capital should be avoided. Second, attempts at corrective taxation should be undertaken with extreme caution. History suggests that unintended side effects or deadweight losses may dominate the results. This implies that special tax-based schemes to encourage stock exchange listing, house- hold saving, and the like should be viewed with caution, bearing in mind the substantial opportunity cost in terms of lost revenue and the questionable gains. Third, while tax shifting is common throughout the economy, the potential for arbitrage is very high in finance. All financial sector taxes need to be designed in as arbitrage-proof a way as possible (the first defensive criterion). Fourth, inflation generally has a more pervasive effect in finance on the impact of taxation. All financial sector taxes need to be designed to be as inflation-proof as possible (the second defensive criterion). Fifth, approximating taxation of the financial sector to that of other sectors is a reasonable goal. The challenge lies in mapping the somewhat distinctive institutions and concepts of financial inter- mediation to that of the remainder of the economy, and distinguish- ing between its role as an intermediary and manager of the funds of 4 PATRICK HONOHAN others from those of a true principal. This allows one to relate or “map” the various tax rates and tax bases affecting financial firms to the more familiar concepts most closely corresponding, such as sales taxes, corporate income tax, and collection on account of cus- tomer income taxes. The remainder of this chapter discusses the background to each of these guidelines, referring as appropriate to the remaining chap- ters of the volume. The next section reviews the main forms of tax that are relevant. The section that follows describes the thrust of re- form ideas. The chapter then examines the effectiveness of the sev- eral corrective taxes that have been employed. After highlighting the two most distinctive relevant features of the financial system for de- signing tax structures, the discussion comes to the practical question of tax rates and how to calibrate these for comparability with non- financial taxes. The chapter concludes with a call for moderation in tax design and advocacy of the defensive approach, which should be the guiding principle in policy design. The Main Types of Explicit and Implicit Tax Governments have used financial intermediaries to relieve their budgetary pressures in three main ways. First, they have applied a variety of explicit taxes. Some are common to firms in other sectors of the economy. Some are special to the financial sector, such as fi- nancial transactions taxes, unremunerated reserve requirements, and deposit insurance premia. Some seem similar to those applied to other sectors, but in practice have a qualitatively different impact even if imposed at the same nominal rate. Additionally, differential application of mainstream explicit taxation to financial intermedi- aries, including different rates of tax, can be important, as in the treatment of loan-loss provisions in calculating taxable income, or in the application of sales taxes to interest received by banks. Sec- ond, they have imposed reserve requirements, which have had the effect of boosting the net revenue of the central bank and hence in- directly the government. Third, they have made regulations chan- neling funds to government or favored sectors and borrowers in ways that involve implicit subsidies, notably by imposing interest rate ceilings. Explicit Taxes Taxes may be levied on many different elements of a financial inter- mediary’s business. Net corporate income (profits), gross revenue (in- AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 5 terest and fees), and the value of payments made or received through the intermediary are the most important types. Interest paid by the intermediary to its creditors are also often taxed, and the interme- diary may be obliged to withhold this tax, thus making only net-of- tax payments to the creditors. Less commonly, elements of the bal- ance sheet of the financial institution (assets, liabilities, or net capital) could also form tax bases. Inasmuch as non-financial corporations are also liable to cor- porate income tax and to a variety of sales taxes, it is important to identify whether, and in what way, taxation of the financial in- termediary often differs sharply from the standard situation. The fi- nancial intermediary may be subject to special rules or rates. Or the way in which the standard tax is applied may have a distinct inci- dence on financial intermediaries because of characteristic ways in which their business differs structurally from that of non-financial businesses. For instance, the total value of payments made and received by a bank (credits and payment to customer accounts) is a large multiple of the total value-added of a bank. Furthermore, the value of pay- ments bears no stable relationship to the value-added or profits of a bank. As with the value of goods carried by shipping or airline companies, a tax on such payments, even at a low rate, could not be regarded as an approximation to a value-added tax on other companies. The same would be true of taxes levied on securities market transactions. On the other hand gross interest, insurance premium income, and fee receipts in a non-inflationary environment could be of the same order of magnitude: perhaps twice the value-added. Such a ratio would be equally characteristic of many non-financial compa- nies. However, in contrast to these, and unlike net interest, the gross interest is highly sensitive to the nominal level of wholesale interest rates and to expected inflation. In a volatile inflationary environ- ment, this too becomes a rather arbitrary tax base. The calculation of appropriate reserves against loan losses is an issue for the accounting of any company with receivables or other claims in its balance sheet. But it looms much larger for financial in- termediaries, where annual loan-losses even in good years can often be much larger than the profits earned. Therefore the tax treatment of loan-loss provisioning is relatively much more important for fi- nancial intermediaries, in that the timing of sizable tax payments can be at stake. The inertial element in explicit taxation of finance is reflected in stamp and registration duties. These have a long history in taxation, having been applied to the formal registration of legal documents, 6 PATRICK HONOHAN including those recording transfers of property ownership. They have their legacy in taxes on payments transaction and transactions in securities exchanges. Modern tax systems depend to a large ex- tent on approximations to a comprehensive income tax or expendi- ture tax. Stamp duties are poor approximations of either concept. Withholding taxes on interest paid to depositors and other forms of special treatment of income received by the customers of finan- cial intermediaries can also be distorting. This is especially the case when they apply at different rates to different categories of income, such as on local currency and dollar-denominated deposits.3 Reserve Requirements and Seigniorage The inflation tax and related taxes4 deserve a section by themselves because of their historical importance, the scale of potential rev- enue, and the ease with which they can be collected. Requirements that banks should hold a certain fraction of their deposits in the form of liquid reserves, whether in cash, at the central bank, or at some analogous institution, dates at least to the early part of the 19th century. Initially, it represented a convenience to ensure the smooth completion of the daily clearing and to reduce the re- course of banks to central bank borrowing. Since reserves placed with the central bank were often unremunerated, reserve require- ments boosted net income of the central bank, which is usually passed to the fiscal authority as a dividend payment in due course and recognized as a nontax revenue item in the budget. In this way the banks were implicitly taxed and the budget relieved. The fiscal el- ement was at first not considered especially important, but it became so as bank margins narrowed, especially where nominal interest rates were rising. Some central banks responded by introducing remuner- ation on required reserves; others tolerated avoidance through sub- stitution by banks of nonreservable categories of instrument. Nowadays, reserve requirements are generally seen as an exten- sion of the base of seigniorage, inasmuch as substitution of deposits for cash holdings had reduced the base of seigniorage as a tax. Secondary liquidity reserve requirements have also been imposed in several countries. Often, secondary reserves have had to be held in the form of designated government securities. Sometimes the se- curities were sold directly to the banks with off-market yields, and as such embody a fairly obvious implicit tax. Such requirements have often also been imposed on insurance companies and other nonbank intermediaries. These types of requirements thus shade into directed credit and interest ceiling arrangements. AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 7 Directed Credit and Interest Ceilings Governments seek control over where the loanable funds mobilized by the financial system will be applied for somewhat different rea- sons to those motivating reserve and liquidity requirements, but again there is a clear fiscal dimension. This kind of mechanism has been operated in nearly all countries over the years and takes many forms. Sometimes there is a requirement to place a special deposit amounting to a specified proportion of the bank’s mobilized re- sources in the central bank or another public agency charged with on-lending these to borrowers in preferred sectors. Sometimes there is a requirement to lend a certain fraction of the bank’s resources to specified sectors, or failing that, to deposit an equivalent amount with a specialized bank that can do the lending. Whether or not there is an explicit interest rate ceiling on these sectoral require- ments, the diversion of funds has the effect of lowering the market- clearing rate for them. This will act as if there were a tax on the in- terest income from this part of the lending (partly compensated by a higher market-clearing rate on non-favored sectors). Except where the government is the borrower, the benefit of this tax does not di- rectly go to it. Nonetheless, it is appropriate to see the budget as a hidden beneficiary, in that, absent the directed credit, subsidization of the preferred borrowers would have to have been done through other means, including direct budgetary allocations. System-wide interest rate ceilings, much rarer now than in the past, and capital controls have the effect of lowering local interest rates and this too can be seen as a tax affecting financial intermedi- ation. The government’s budget is almost always the largest single borrower, and as such the biggest direct beneficiary of system-wide interest ceilings and their equivalents. The Big Reform Ideas One general approach to financial sector taxation is to attempt a great simplification on the theory that low rates and a wide base with few exemptions are likely to generate relatively low distortions. This approach holds out the prospect not only of minimizing the incentive for complex schemes of financial engineering designed to avoid tax, but also of making such schemes relatively difficult to develop. The three main handles for taxation—income, expenditure, and transactions—have each been the subject of prominent and exten- sively discussed grand and simple schemes. These are: the proposition 8 PATRICK HONOHAN that capital income should not be taxed at all; the proposal that value added by the financial services industry should be subject to a uni- form tax; and the idea that a tax on all financial transactions at a very low rate could generate large revenues with negligible distortion. These are considered one by one in the sections that follow. Capital Income: Should It Be Taxed at All? The underlying basis for the argument that it might be optimal not to tax income from capital at all is the insight that this involves a form of double taxation on future consumption. Shifting the per- spective from the statutory base of the tax, capital income, to a vari- able more closely relevant to economic policy, namely utility based on household consumption, raises serious questions. A constant nominal or statutory tax rate on capital income implies an effective rate on consumption that may increase without bound for con- sumption far into the future. Because future consumption depends on the reinvestment of after-tax capital income, the more remote the date of future consumption, the higher the effective tax rate; and this effective tax rate may increase without bound. Optimal tax pol- icy can improve on a situation with infinitely high effective tax rates. Accordingly, this reasoning points to the optimality of capital income taxation converging to zero (see chapter 2). Many subtle qualifications can be made to the implicit models of utility, income, and consumption underlying this analysis, and the precise prescription for zero taxation is not very robust. Nonethe- less, it retains some force and serves as an important counterweight to proposals for high rates of capital income taxation designed to achieve other goals. One such goal is that of ensuring the socially optimal rate of national saving (since private markets cannot gener- ally be relied upon to do this and may result in over-saving). An- other is redistribution. Yet even if households differ in their wage- earning capacity and tax policy is being used for redistributional goals, these can best be achieved by a tax on wage income alone— at least in simple models of intertemporal preferences. Once again the use of capital income taxation would be suboptimal because of the compound interest effect. If income from capital is not to be taxed, then it might seem to follow that the income of financial intermediaries ought not to be taxed either. But in practice some corporate income—perhaps a large portion—represents pure profit or economic rent. Neglected in the models that generate the result of no tax on capital income, pure profit may be taxed without distortion, and this argument is another important qualification. Where financial markets are uncompetitive— AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 9 and the scale economies that are involved in parts of finance make this relevant, especially in financially closed economies—this could be an empirically important factor.5 A stronger line of attack on the zero capital income tax proposi- tion comes from practical issues of enforcement and informational deficiencies. If capital income goes completely untaxed, this may provide an easy loophole for high-earning households to camouflage their earnings by transforming or laundering them into capital in- come. A tax on capital income may be an important practical expe- dient to close such loopholes.6 If so, withholding the tax at source, or taxing corporate income as a form of implicit withholding, may further help to overcome the tax authorities’ informational disad- vantage and administrative collection costs. However, these consid- erations tend to be swamped in many developing countries by the need to consider the impact on foreign-owned firms. Given the fact that many capital exporting countries allow credit to their residents for tax paid in the host country, this can pave the way for host coun- tries to tax foreign-owned companies in the knowledge that they will not be discouraged from investing to the extent that the tax paid to the host country simply reduces their home country tax. The elegant simplicity of the theoretical argument against capital income tax thus ultimately fails, though it points to a need to jus- tify such taxation—and the taxation of the income of financial and other companies—on grounds other than those of simple consis- tency with taxation of wage income. Taxing Financial Services: Can a VAT Work? About 70 percent7 of the world’s population live in countries with a VAT and the tax is a key source of government revenue in more than 120 nations (Ebrill and others 2001). So if a VAT is the way forward for the bulk of (indirect) taxation on expenditure, to what extent should it also be the model for financial services? The first observation has to be that in practice, most financial services are “exempt” in virtually all countries employing a VAT. This does not mean that these financial services wholly escape the VAT, as the status of “exempt” does not allow financial service providers to recover VAT paid by their taxable suppliers and built into the price of their inputs. Indeed, taxable firms that use finan- cial services as inputs cannot recover the VAT paid by the suppliers of financial service firms either, with the result that there is tax “cas- cading.” But value that has been added by the exempt financial sec- tor firms is not directly captured in the tax. Whether aggregate tax receipts would increase or fall if the exemption were removed is an 10 PATRICK HONOHAN unresolved empirical issue, which depends not only on the degree to which financial services are used by tax-liable firms, but also on the different rates of VAT that may be in effect. The exemption of most financial services from VAT appears to be a historical inheritance without much political or economic ration- ale. While it is not difficult to measure the value-added of a bank (profits plus wages), there is the practical difficulty of deciding how much credit taxable firms that use financial services would be enti- tled to claim. The charge for many financial services is an implicit one bundled with others in, for example, the spread between de- posit and lending rates. Determining how much of the spread should be attributed to depositor services and how much to borrower services is not straightforward. Thus it is not obvious how much credit each should receive for VAT already paid on inputs. Yet it is not impossible to devise simple rules of thumb that can provide a reasonable approximation. Thus, for example, the cash flow method where VAT is paid on all net cash receipts (including capital amounts) could be adequate in a static environment. How- ever, start-up problems and treatment of risk may not be adequately resolved by this method, and changing tax rates also presents diffi- culties for the approach. A variant of the cash flow method, using suspense accounts and an accounting rate of interest to bring trans- actions at different dates to a common standard, could help ease the transition problems and has been shown to be workable by detailed pilot studies in the European Union (see chapter 12). The lack of any clear potential revenue gain, and fears about the practical complexity and possible hidden distortions or loopholes, have inhibited any significant move to bringing financial services into the VAT net.8 The resulting distortions are quite serious in some cases. For one thing, there is a clear incentive to self-supply inputs. Second, there are distortions at the margin. Financial services such as factoring, which can represent a particularly effective form of lend- ing to small- and medium-scale enterprises—low cost and low risk— could be severely tax-disadvantaged by falling within the VAT net in many jurisdictions for which other forms of lending are exempt. The grand simplification offered by the VAT thus fails, not on theoretical grounds, but on the grounds of administrative and prac- tical difficulties or uncertainties. Nevertheless, it does point in the direction of what might be desirable for substitute indirect taxes.9 Transactions Taxes: Panacea or Pandora’s Box? Because of their loose connection with consumption and utility, and their potential for generating significant distortions in the organiza- AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 11 tion of production and distribution, transactions taxes (including trade taxes) have lost favor as a tool of general tax policy over the years relative to income and expenditure taxes. But the vast scale of financial sector transactions has presented itself to some scholars and some governments as a convenient base for rapidly generating substantial revenue. There is a paradox here. Critics of transactions taxes point to the potentially serious distortions that it causes, while advocates argue that, because of the large base, sizable revenues can be realized with low nominal tax rates. To the extent that the deadweight cost of a tax is often supposed to be proportional to the square of the tax rate, introducing a low-rate financial transactions tax in order to reduce the much higher rates of labor income or other taxes might reduce total deadweight in the tax system as a whole. At the most extreme, a recent proposal suggests that what seems at first sight to be an administratively trivial and quantitatively tiny 0.15 percent rate of tax on all automated payments could raise enough revenue (in the United States) to replace the entire existing tax system (Feige 2000). Feige shows that existing automated pay- ments amounted (in 1996) to somewhere in the region of US$300– 500 trillion, or of the order of 50 times the value of GDP. How, he asks, could anyone argue that a tax rate of 0.15 percent, even ap- plied to such a large base, be considered seriously distorting by com- parison with the existing tax regime? Analysis of the payments that would be affected reveals that about 85 percent relate to financial transactions (purchase or sale of stocks, bonds, foreign exchange or other money changing transac- tions, and so on). To a large extent, then, the initial burden of a uni- versal payments tax would fall on the financial sector. Of course, if the perspective shifts (as before, with the capital in- come tax) from the statutory or nominal base to the more econom- ically relevant concept of consumption, this argument takes on a different light. Then it becomes clear that the average good or ser- vice in the typical consumption bundle must be “hit” by the tax not once, but dozens of times, as it works its way through financing, de- sign, production, and distribution. Criticisms of this proposal fall into two main groups. First, the tax would not collect as much revenue due to the sizable elasticities involved.10 Financial sector transactions in particular would be arbi- traged in such a way as to drastically reduce the number of recorded transactions. What are now sequences of linked transactions carried out for little more than bookkeeping convenience at negligible cost would be collapsed into a single more complex transaction. Portfolio readjustments would be made with reduced frequency without sub- 12 PATRICK HONOHAN stantially altering expected return and risk. Microeconomic studies of the precise mechanisms that are at work to generate gross trans- actions of such a high multiple of GDP in wholesale financial mar- kets are not plentiful so that reliable estimates of these effects are not yet available.11 Furthermore, the scope for avoiding such a tax through offshore financial transactions must be taken seriously. The second main objection is that, even if the tax did collect the expected revenue, the distortion costs would not necessarily be any smaller than with the existing system. This objection relies on either of two observations. First, the financial system would bear the main brunt, and as such that the tax would in fact be more concentrated, not less. Or second, in terms of final consumption, the tax would effectively cascade to cumulative rates comparable to those observed at present. No country has seriously considered replacing its tax system with a universal payments tax, but there are numerous examples of par- tial transactions taxes, applying for example to bank debits or to se- curities transactions.12 Bank debit taxes introduced in half a dozen Latin American countries in the past 15 years or so in a bid to raise revenue have been successful in that goal, at least for a while, with revenues ranging from about 0.5 percent of GDP to as much as 3.5 percent in one case for one year. It is fair to say that revenue from these taxes held up unexpectedly well over three to four years. That revenue would fall off after the first year was predicted by many, and it did occur on average, though the effect did not prove to be statis- tically significant in regression of the available data. Nevertheless, many of the schemes had to be adapted administratively in the course of their operation, to exempt some transactions that would otherwise have been too distorting (and probably also to capture others that had escaped the net). The distortions of these and of se- curities transactions taxes have been discussed in the literature. They certainly are distorting, yet applied in moderation, these transactions taxes have been less distorting than many observers expected (see chapters 10 and 11, this volume). Thus such transactions taxes have been surprisingly resilient— despite expectations that they would not only distort financial mar- kets and drive out capital but would quickly lose their revenue- raising ability. But they are far from being a panacea, and indeed have little to recommend them beyond their ability to deliver rev- enue speedily and with low direct administrative costs. Corrective Taxes It is not just taxation that distorts financial markets. Information deficiencies, monopoly power, and other factors push most financial AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 13 markets away from the ideal of the atomistic market with fully in- formed participants competing on a level basis. Under these circum- stances, the nonrevenue side effects of taxes and tax-like measures can be turned to advantage and form part of the corrective policy structure in this area. Indeed, many measures of this type may have regulation and market efficiency as their primary objective, with revenue seen as a side effect.13 But as discussed below, the effectiveness of many such measures in their supposedly corrective role has been challenged and remains controversial. Deposit Insurance The most complex and contentious of these debated corrective quasi- taxes is deposit insurance. That it is a tax is fairly clear from the contributions or levies that are generally imposed on participating banks, especially given that these are typically compulsory and that the rate of tax usually bears at best an imperfect relation to the “fair premium.” Indeed, the anticipated gross revenue from the levy is typically small and in many cases is calculated to be insufficient to cover even the expected payout costs as calculated using option-pricing formu- lae (Laeven 2002). Furthermore the probability distribution of net payout costs is severely skewed. Systemic banking crises entailing fiscal costs of up to 50 percent of a year’s GDP are never matched by a corresponding deposit insurance fund accumulation in lucky, crisis-free countries.14 For many advocates, the perceived corrective role of deposit in- surance is essentially one of reducing the likelihood of depositor panic. By protecting depositors against the risk that their deposits will be unpaid if a bank proves to be insolvent, it is hoped that a self-fulfilling panic—including contagion to other banks triggered by the insolvency of one bank—can be avoided.15 On the other hand, by lowering the vigilance of potentially informed depositors, the moral hazard of heightened risk-taking by the bankers, unpun- ished by market discipline, could in theory result in heightened risk to the system as a whole. Although early deposit insurance schemes entailed a uniform in- surance premium per dollar of deposit, there have been moves in several countries to differentiate the rate of premium in accordance with some measure of the perceived riskiness of the participating bank’s portfolio. This dimension of such taxes is designed to reduce the moral hazard potential, but it depends to some extent on the in- formation available to the deposit insurer as to the accuracy of the ex ante risk assessment (Honohan and Stiglitz 2001). About a quar- 14 PATRICK HONOHAN ter of schemes have some risk-differentiation, but the differentials are small and are not always systematically imposed (Demirgüç- Kunt and Sobaci 2001).16 Econometric estimates of how financial system performance varies across countries with the existence and characteristics of de- posit insurance systems suggest that countries whose socio-political institutions are generally rated as strong need not fear that the moral hazard side effect will outweigh other beneficial effects. Al- though deposit insurance weakens market discipline even in such countries, the effects seem to be offset by better official oversight. However, for countries with less well-developed institutions (along the dimensions of rule of law, governance, and corruption), the es- tablishment of a formal deposit insurance scheme17 does appear to present a heightened risk of crisis (Demirgüç-Kunt and Detragiache 2002; Demirgüç-Kunt and Kane 2002) and does not even promote deposit growth (Cull, Senbet, and Sorge 2002). Having risk-based deposit insurance premia does not appear to mitigate the systemic risk, thus the potential for introducing a corrective structure of the deposit insurance tax may be limited. Deposit insurance, with or without risk-based premia, may not be a very effective corrective mechanism. It clearly needs to be sup- plemented in this role by strong administrative or other controls, in- cluding supervision of minimum capitalization ratios (see chapter 5). Moreover, it may interact with other taxes. For instance, a tax on bank gross receipts will reduce the expected after-tax return to a risky investment, though chapter 5 shows that there would be some offset to this inasmuch as the government (deposit insurer) is coin- suring the risk to a greater extent in the presence of such a tax. On the other hand, chapter 5 also shows that a marginal reserve re- quirement (see below) could be more likely to reduce the moral haz- ard effect on bank risk-taking behavior. All in all, though, the un- certain strength and reliability of such effects argue for blunter and more reliable instruments in restraining bank risk-taking, a matter that lies beyond the scope of the current volume. Provisioning and Capital Adequacy The amount of loan-loss provisioning that is allowable to banks as a deduction against income for tax purposes can be a very signifi- cant factor in arriving at the net tax liability—and is often sufficient to shelter the entire tax bill. By the same token, this can be a mat- ter of considerable revenue significance for the authorities. But it has long been acknowledged that there is a potential corrective role for the treatment of loan-loss provisions. This argument hinges on AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 15 the arbitrariness that inevitably arises in arriving at a reasonable provision that would ensure that the banks’ accounts represent a true and fair picture of the business. If the fiscal rules have the ef- fect of biasing company accounting, this could be damaging for the transparency of the financial system and for good decisions on risk management. Recent accounting scandals have focused attention on the difficulty of seeing through valuation procedures used in non- financial company reporting procedures. Bank accounts can be ar- guably even less clear-cut, especially in times of economic turbu- lence or change. To the extent that equity capital represents a cushion protecting depositors and other claimants against the consequences of a de- cline in the value of the bank’s loan portfolio and other assets, the equity holders of a lightly capitalized bank at risk of failure—and the directors, to the extent that they are acting as the equity hold- ers’ agents—will have an incentive to minimize the amount of cap- ital that they truly have at risk. In this way, they will transfer risk to other claimants. They will minimize their capital at risk provided they can do this without inducing an increase in the required return on their other liabilities. If the fiscal authority disallows the de- ductibility of reasonable loan-loss provisions (for example, provi- sions that can be justified on the basis of a reasonable objective forecasting model), that reinforces the incentive to understate pro- visions and thereby to overstate capital, potentially misleading reg- ulators and the market. On the other hand, a well-capitalized bank may be more at- tracted by the advantages of advancing tax deductibility, and may use the range of uncertainty to increase loan-loss provisioning, thereby reducing revenue. Balancing the pressures of revenue needs with the risk of losing transparency is thus a constant tug-of-war and different countries adopt different rules (see chapter 9 in this volume and Laurin and Majnoni 2003). The preferred goal here would seem to be a move away from mechanical rules (such as disallowing general provisions but allowing specific provisions) toward a more realistic, forward- looking accounting that allows predictable but not yet identified losses to be adequately provisioned, so long as these are accepted by the institutional regulator.18 Promoting Saving A widespread explicit goal of corrective tax measures affecting the financial sector is the promotion of saving. The goal is driven partly by fiscal needs, in an attempt to ease the financing of government 16 PATRICK HONOHAN deficits; partly by a perception (colored by an earlier generation of macroeconomic theories and, because of new research findings, no longer generally accepted by economists) that aggregate economic growth is, in the long-run, driven by national saving; and partly by a desire to ensure that households do not under-save, particularly for retirement, but also for housing and education.19 In practice, such measures tend not to affect all savings media equally. Hence their sometimes substantial impact on the structure and performance of the financial system, which, in certain cases at least, can far outweigh the net impact of the policy on the goal of increasing household saving (OECD 1994; Honohan 2000). For practical reasons, measures that operate by modifying in- come tax schedules tend to be relevant only in middle-income coun- tries, or at least in countries that have achieved a certain minimum level of the effectiveness of the income tax system. Furthermore, there is widespread skepticism among experts as to the effectiveness of mandatory saving for housing in achieving the goal of improving access to housing for the targeted low-income groups. On the other hand mandatory retirement saving programs appear to increase na- tional saving by a significant amount on average, especially perhaps where they are tax-advantaged (chapter 4). Other Dimensions of Corrective Financial Taxation In other cases, supposedly corrective financial sector taxation comes more in the form of a vague and unthinking encouragement of what are seen as social “goods.” This is not unique to the financial sec- tor: finance ministers are typically bombarded with proposals to ex- empt from taxation items or activities thought to be meritorious. Except where tax relief appears to be the most effective way of cor- recting some market distortion that is resulting in an undersupply of the item or activity in question, the ministers are usually advised to resist such special pleading. But lobbying of this type does appear to be notably successful in finance. For example, consistent with the observation that most countries feel that their financial system is unduly bank-dominated, there is constant advocacy of tax conces- sions targeted at companies with a stock exchange listing.20 This is at best a crude instrument, especially if the underlying reason for the underdevelopment of the stock exchange lies in an insufficiently de- veloped information and legal infrastructure, as is often the case. It is much better to direct policy attention to correcting these infra- structural deficiencies. Another much used quasi-tax often thought of as corrective, in a sense, is the unremunerated reserve requirement. The sense in which AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 17 this might have been thought of as corrective is that it provides a lever on which monetary policy can operate. Actually, as is now ac- knowledged by authorities on monetary policy, the perceived need for unremunerated reserve requirements was based on a misconcep- tion. Monetary policy does not require unremunerated reserve re- quirements or any other quasi-tax for its effectiveness (see chapter 5). Two Distinctive Elements for Financial Sector Tax Design If there are two key features of the financial sector that distinguish it from other sectors when it comes to designing taxation, these must surely be the system’s capacity for arbitrage and its sensitivity to inflation and thus to non-indexed taxes. The System’s Capacity for Arbitrage Whether mainly flat or mainly corrective, the actual impact of most financial sector taxes depends crucially on the extent to which they have been constructed so as to be insulated from the high elastici- ties that prevail in the sector. Tax design in this area is confounded by arbitrage between functionally equivalent contracts or institu- tional forms. One important illustration of this can be seen when one consid- ers how the incidence of bank taxes can be shifted. Because of sub- stitutability and the possibility of arbitrage and near-arbitrage, the full incidence of taxation imposed on one component of the inter- mediation process (deposits, loans, intermediary profits) may very well be fully shifted to another component. Ramon Caminal (chap- ter 3, this volume) has developed a formal model of intermediation, taking account of the provision of liquidity as well as inter- mediation services by banks in order to examine the influence of various bank taxes on the volume and cost of intermediation ser- vices provided to depositors by banks. Several striking results are obtained. For instance, the ability of at least some borrowers to sub- stitute alternative sources of funding implies a tendency for the im- position of a VAT on banking services to be passed back to deposi- tors.21 Furthermore, the conditions under which a tax on bank loans falls not on the cost of funds, but instead on the return to bank shareholders, are also plausible, including a range of assumptions on competitive conditions. (However, if regulatory capital require- ments are likely to be binding in the sense that banks hold more capital than they would freely choose to, a tax on banks’ profits 18 PATRICK HONOHAN may in contrast fall wholly on lending interest rates.) In contrast to general models of production, then, plausible modeling of the de- gree of substitutability in banking involves such high elasticities that predicting the incidence of a tax to fall wholly on a class of agents not directly the subject of the taxation can be plausibly predicted. On the other hand, recognizing that the services provided to savers by investment funds may be highly substitutable for some of the ser- vices obtained from bank deposits, Caminal has also shown how, under reasonable circumstances, the presence of untaxed investment funds implies that taxation of deposits will affect only the monitor- ing and transaction service provision by banks, and not the provi- sion of liquidity. These contrasting cases suggest the heightened risks involved in imposing taxes under the assumption that the taxpayer who is liable will be the one incurring the incidence of the tax. Just what the in- cidence will be can be worked out in theoretical cases (to a greater extent than is the case for taxes on non-financial sectors). However, the task of matching these theoretical cases to the real world is a striking challenge for the empirical policy analyst, given the diffi- culty of estimating many of the relevant behavioral relationships— as is evident from their relative absence from the literature, even for industrial countries. Along with the shifted incidence can be a large behavioral effect. This may not be socially costly in equilibrium (if the substitute truly is functionally equivalent). However, short-term disruption and costly incurring of new sunk capital to support the substitute activity could be quite severe. Even more acute problems of the same general type are associ- ated with the taxation of new financial instruments. At the heart of financial innovation is, in the words of Boadway and Keen (chapter 2), the creation of new instruments by repackaging the cash flows generated by others. Arbitrage is here the mechanism, not just an outcome. The reasons for this repackaging are manifold: to better align the instruments with the liquidity and maturity preferences of different classes of investors; or to shift particular risks between in- vestors who have different appetites for them, whether based on in- formation or on correlations with the remainder of their portfolio. If the rebundled instruments are differently treated by taxation, this can block the repackaging and inhibit the risk-sharing that is in- volved.22 Furthermore, of course, differential tax treatment (for ex- ample of debt and equity, or of income and capital) can be a pow- erful driver of innovation designed for no better reason than to repackage cash flows into a less heavily taxed form. AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 19 Boadway and Keen note that many of these issues have been dealt with in a piecemeal way by tax authorities in advanced economies. Theoreticians have been exploring ways of rationalizing the taxa- tion of new financial instruments, both by devising unambiguous decompositions of the instruments into fundamental components, and by determining the timing at which the amounts are crystallized for the purpose of calculating the tax (accrual versus realization accounting). To date, no general agreement among theoreticians, let alone practitioners in advanced economies, has yet emerged. This rules out, for the present, the possibility that tax authorities in a de- veloping country could piggyback on a pre-packaged solution. In- deed, for market participants, the tax situation is even less satisfac- tory in developing countries, where the likely tax treatment of new instruments is often undetermined or disputed. Sensitivity to Inflation Although inflation has pervasive effects throughout the economy and in particular has been shown to be negatively correlated with growth, at least for sufficiently high rates, banking and other parts of the financial sector that extensively employ nominal financial contracts can be more directly and deeply affected than most. High and variable rates of inflation induce significant substitution away from non–interest-bearing monetary assets in favor of assets offer- ing higher real returns and inflation hedges. This can, on the one hand, shrink the size of the banking system’s intermediation. On the other hand, the financial system’s capacity to provide the instru- ments to insulate economic agents from the inflation will tend to expand this side of its activities. Indeed, empirically, the balance sheet size of the banking system is found to shrink with inflation, whereas inflation is found to be positively associated with prof- itability and the value-added of the banking system (chapter 13). Inflation also has a strong influence on the government’s finances. The term “inflation tax” is well chosen, even though there is no per- fect correspondence between the implicit inflation tax rate as mea- sured by the opportunity cost of holding interest-free base money (which will be related to the expected inflation rate) and the flow of financing to the budget from money creation (Honohan 1996). The interaction between inflation and a non-indexed tax sys- tem can have sizable and unexpected effects, even in a country with single-digit inflation (Feldstein 1983, 1999). As inflation increases, the double distortions of inflation and taxation can be multiplicative rather than additive, with severe consequences. For financial sector 20 PATRICK HONOHAN firms, the impact of inflation on the scale and activity of financial services firms needs to be considered alongside its impact on their tax-inclusive cost structures. The effective tax rate of several com- monly employed financial sector taxes, such as taxes on gross inter- est receipts of banks, or unremunerated reserve requirements, rise almost in proportion to the rate of inflation. In the case of nominal interest rate ceilings, the effective rate of tax rises more than in pro- portion to the rate of inflation. Given that inflation rates can be high, volatile, and unplanned, this degree of sensitivity to inflation in the effective rate of tax is generally quite undesirable (chapter 13). Good tax policy needs to take account of these sensitivities. A de- fensive approach is needed that ensures that the tax structure is not vulnerable to severe distortion along these lines. Calibrating Different Types of Tax Where a defensive approach has not been followed, poorly con- structed tax systems—whether the consequence of a drive for rev- enue, or of misdirected sophistication—have often had sizable un- expected side effects. Some of the most dramatic cases are discussed in the companion chapters by Cardoso (chapter 7) and Granville (chapter 8), both of which—not coincidentally—refer to periods of high inflation. Part of the problem in many difficult cases has been that the fi- nancial sector taxes and implicit or quasi-taxes have not been seen for what they are. Thus very high effective tax rates have emerged in cases where legislators would not have conceived of imposing comparable nominal tax rates.23 On the other hand, lobbyists are prone to finding ways of exag- gerating the tax burden on financial intermediaries by adding up taxes that touch the sector only slightly and expressing these as a percentage of the sector’s profits. Is there some simple way of approximating the burden of a given tax—or better, the impact of reform in a particular tax? This section looks at how this question might be addressed in respect to the main types of tax or quasi-tax that most often raise such questions. The relevant taxes include unremunerated reserve requirements, tax on intermediary interest receipts, withholding tax on interest payments by intermediaries, stamp tax on bank debits, and stamp tax on bank loans. One practical approach to calibrating these taxes and judging their appropriateness is to map each tax into its closest non-financial AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 21 analog. Thus one decides whether the tax is more nearly an income or a sales tax. If an income tax, is it more a tax on the intermediary’s shareholders or on the intermediary’s fund-providing customers? If a sales tax, what is the product that is being taxed and what is its net- of-tax price? As with most issues of incidence, these questions cannot always easily be answered. Nevertheless, even an approximate answer can clarify the issues significantly. The discussion in several chapters of this volume can help. Market power and substitution possibilities are central. In many countries, the market power of banks is being eroded by inter- national competition for depositor services and from alternative sources of industrial funding, as well as by liberalization of entry. Taxes and quasi-taxes that might hitherto have been assumed to fall on the shareholders of banks in a manner analogous to an income tax may now be more likely to be passed on to those customers who have few alternatives, notably small borrowers whose creditworthi- ness is costly to determine. Caminal (chapter 3) models these issues in some detail and Cardoso (chapter 7) presents interesting evidence that pass-through has been very high in Brazil. Under such conditions, the taxes described fall into three groups: those that are best seen as a tax on lending services, those on trans- actions services, and income taxes on suppliers of funds. Both unremunerated reserve requirements imposed on banks and special taxes on interest receipts of banks are best seen (under these circumstances) as similar to sales taxes on the provision of lending services (such as credit appraisal and monitoring) to small borrow- ers. The effective tax rate can be approximated by comparing the tax paid (or, in the case of unremunerated reserve requirements,24 the opportunity cost of the reserved funds) per dollar lent to the net of tax cost of the service. High effective tax rates often result. Offi- cial estimates for Brazil in 2001 can be read, in this perspective, as implying an 85 percent effective tax rate on average for lending (chapter 7). Furthermore, because the tax base—the cost of inter- mediation services—is not sensitive to the nominal rate of interest whereas the tax paid is, the resulting effective rate can be very sen- sitive to the nominal rate of interest and thus to the rate of inflation (see chapter 13). The stamp duty on bank loans, typically proportional to the loan size but not to its maturity, can be analyzed in much the same way. In this case the effective tax rate may increase sharply as maturities shorten, allowing the methodology to reveal the obvious technical deficiency in such a tax.25 22 PATRICK HONOHAN Transactions taxes and the stamp tax on checks likely fall mainly on the user of the transactions involved. The relevant tax rate is thus computed as if it were a sales tax on the relevant service. This allowed the stamp tax on checks in Chile in 2002 to be computed as a sales tax rate of 100 percent, compared to a rate of about 20 percent for a similar tax imposed in Ireland. Judging the appropriate treatment of the withholding of income tax on deposit interest requires careful consideration of the effec- tiveness of the remainder of income tax. If income tax on the rev- enue from competing capital assets is collected effectively, then the fact that tax due on deposit interest is withheld at the source can best be thought of as chiefly an administrative convenience, rather than as an additional imposition affecting the withholding interme- diaries and their other customers. The empirical judgment here will often depend crucially on the degree of international capital mobil- ity (see Huizinga and Nicodeme 2001). Guarding Against Arbitrage and Inflation It would be hard to justify a dogmatic approach to reforming fi- nancial sector taxation on the basis of the review presented here. None of the extreme blueprints that have been proposed for the op- timal structure of financial sector taxation can be fully endorsed. Nonetheless, some important lessons emerge. Both empirical ex- perience and theoretical propositions suggest that some financial sector taxation can have large and damaging effects that were not anticipated. Therefore the main message proposed here is one of moderation. There is no reason why the financial sector should not pay its share of needed tax revenue. However, design should be de- fensive in the sense of guarding against the two major vulnerabili- ties to which the sector is prone: sensitivity to arbitrage and sensi- tivity to inflation. Starting with the overall question of capital income tax, the case for exempting capital income from all taxation (the first “big idea” floating in the field) is too narrow to justify action. All that we would propose in this regard is to take the literature as a caution against excessively high capital income tax rates, both in absolute terms and in comparison with income tax. Attempting to “soak the rich” through a high capital income tax rate may be an unwise way to proceed. The second big idea, that of extending VAT to financial services, seems much better based. It certainly seems a more desirable goal AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 23 than attempts to rely heavily on transactions taxes (whose distort- ing effects must soon dominate, even if they have shown their rough and ready merits as transitory revenue-spinners in times of fiscal pressure). Here, too, excessive enthusiasm should be tempered. The VAT might be the way to go in a world where administrative collection costs were low. In practice, there may be less to gain from it than appears at first sight. Pioneering its adoption could be a diversion for a developing economy—unless the authorities are sure of their administrative capacity to be pioneers in the area. For most, the VAT should best be seen as a benchmark against which to compare other indirect taxes in regard to their neutrality and their burden (effective tax rate equivalent on relevant value-added). Given the propensity for inflation rates to be high and volatile, taxes and quasi-taxes that are super sensitive to inflation should be avoided. Taxes with such features should be replaced with others that are more inflation-neutral. Above all, each tax proposal should be considered in terms of the possible substitutes that exist for the activity, asset, return, or trans- action being taxed and that could both magnify the behavioral ef- fect of the tax and place its incidence far from where it was origi- nally expected. This must be the key defensive mechanism. In making judgments on the overall burden of taxation on finan- cial intermediation, which inevitably requires some assessment of incidence, particular efforts need to be made to isolate those taxes that, it can reasonably be assumed, are passed through to deposi- tors. Income tax on deposit interest collected at source, for exam- ple, is not borne by the interest payer if income tax on other sources of capital income is also efficiently collected. Notes 1. To be sure, the enormous fiscal cost of banking crises means that in many countries, the financial system has been a net charge on the budget in recent years. 2. The survey of recent developments by Levin and Ritter (chapter 6, this volume) documents the considerable variety that continues to exist among industrial country systems. 3. Differential treatment of taxation of dividends of listed companies can also be seen as an implicit negative tax on the use of formal stock markets. 4. Seigniorage is the term applied to the profit gained by the issuer of currency. A part of it comes from the inflation tax, which arises because 24 PATRICK HONOHAN currency holders must constantly increase their nominal holdings of cur- rency in inflationary times. 5. Caminal (chapter 3, this volume) explores the implications for tax in- cidence of market power in banking. As he and others have noted, though, leaving banks with some untaxed economic rent (or franchise value, as it tends to be called in the banking literature) can reduce the propensity, po- tentially strong among insured banks, to assume socially excessive risks (Stiglitz 1994; Caprio and Summers 1996). Indeed it has been argued that the credibility of banks as delegated monitors for the depositors depends on their being able to earn profits in most countries (Diamond 1996). 6. Differentiating the rate of withholding tax as between income from high-risk (equity) and low-risk (debt, deposits) assets could help achieve progressivity, even absent information on the income of the recipients, as- suming diminishing risk-aversion with wealth (Gordon 2000). 7. The largest countries, by population, without a VAT are India, the United States, Iran, Ethiopia, Congo DR, Myanmar, Afghanistan, North Korea, Iraq, and Malaysia. 8. A few countries have introduced substitute taxes based on applying a rate to the estimated value-added of banks obtained by summing the wage and profits. 9. Compare the Chinese “business tax,” which has recently been ap- plied at the rate of 8 percent to the income from a bank’s trading portfolio, but is not applied to the income from the bank’s investment portfolio. This strongly discourages trading by banks in government securities, thereby sharply diminishing liquidity in that market. A VAT-like tax would not cre- ate such a distortion. (A further distortion in China comes from the ex- emption from income tax of interest from government bonds, but the in- clusion of capital gains from trading such bonds at a rate of 33 percent. Such problems are avoided in some other countries by treating capital gains from trading activities of financial intermediaries on par with income for the purposes of direct taxation.) 10. This consideration needs to be kept in mind by those who would see the proposal as socially progressive in that payments in which they are di- rectly or indirectly involved likely represent a much higher multiple of the income of prosperous people than of the poor. After all, if such a tax did not raise the hoped-for revenue, the consequence might have to be cutbacks in public services needed by the poor. 11. See Lyons (2001) for the foreign exchange market. 12. Tobin taxes, imposed at a low rate on international capital flows, are much more focused and do not typically have revenue as the main ob- jective. Instead they are seen as corrective taxes intended to reduce volatile speculative capital flows. They have generated an enormous literature that will not be summarized here. AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 25 13. The revenues are not always explicitly accounted for, as when unre- munerated reserve requirements augment the central banks net revenue but are nowhere accounted for explicitly as a revenue source. 14. Even the relatively much smaller fiscal costs of the U.S. banking crises of the 1980s were more than enough to empty the insurance funds. 15. Protection of the small depositor is another goal. This is quite a dis- tinct role, of course, as runs only by small depositors do not threaten sys- temic liquidity. 16. For example, the U.S. premia currently vary according to two criteria, capitalization and supervisory assessment, from zero (for a well- capitalized bank that is highly rated by the supervisors) to 0.27 percent of deposits (for an undercapitalized bank, which is seen by supervisors as pos- ing a substantial probability of loss to the insurer unless corrective action is taken). Argentina has charged a basic rate of 0.36 percent, subject to being doubled where banks were paying high interest rates for deposits (and those with very high rates were not covered at all). Cameroon and other franco- phone African countries impose 0.15 percent plus 0.5 percent of net non- performing loans. Other risk-based formulations, including ex post assess- ments, are levied in other countries. 17. This, despite the consideration that a degree of implicit protection may be assumed by depositors even when no formal scheme exists. 18. Various formulations are possible. One is to have the tax authori- ties accept provisioning that had been agreed upon by the prudential regu- lators. A tougher approach would be to require the bank to defend its pro- visioning on the basis of an objective forecasting method. However, it needs to be recognized that such forecasts are often more an art than a science. 19. Tax incentives are also widely used to favor health and life insur- ance (chapter 4). 20. In Egypt, very favorable tax treatment of listed firms from the 1980s induced a widespread listing of firms on the Cairo and Alexandria Stock Exchange, to which the lenient listing requirements provided no dis- couragement. Over 500 companies were listed by 1990; 10 years later, the figure had grown to more than 1,000. However, most of the firms were closely held and the free-float of shares is little more than one-tenth of the total capitalization. The market capitalization of the actively traded free- float was only about 1 percent of GDP. In this case the supposedly correc- tive tax exemption had a heavy deadweight in that it failed to achieve the presumed objective of a deep and liquid equity market. Recent listing rule reforms are expected to dramatically cut the number of listed firms in 2003. 21. At least under the plausible assumption that the marginal borrower is VAT-liable while the marginal depositor is not (see chapter 3). 22. For example, the existence of withholding taxes on gross interest re- ceipts can stifle the market in interest rate swaps. 26 PATRICK HONOHAN 23. The case of Russia in the 1990s, described by Granville in chapter 8 of this volume, is a cautionary example. 24. Or reserves remunerated below market rate. A simple break-even calculation implies that an addition of λ to the loan interest rate will be re- quired to recover an interest penalty of φ applied to reserve requirements of θ, where λ = φ θ/(1– θ). 25. In Egypt, the application of a constant stamp tax independent of loan maturity hampered the development of short-term bridging finance. References Boadway, Robin, and Michael Keen. 2003. “Theoretical Perspectives on the Taxation of Capital Income and Financial Services” (chapter 2, this volume). Brock, Philip L. 2003. “Corrective Taxes and Quasi-Taxes for Financial In- stitutions and Their Interaction with Deposit Insurance” (chapter 5, this volume). Caminal, Ramon. 2003. “Taxation of Banks: Modeling the Impact” (chap- ter 3, this volume). Caprio, Gerard Jr., and Laurence H. Summers. 1996. “Financial Reform: Beyond Laissez Faire.” In Dimitri Papadimitriou, ed., Financing Pros- perity into the 21st Century. New York: Macmillan. Cardoso, Eliana. 2003. “Seigniorage, Reserve Requirements, and Bank Spreads in Brazil” (chapter 7, this volume). Corvoisier, Sandrine, and Reint Gropp. 2002. “Bank Concentration and Re- tail Interest Rates.” Journal of Banking and Finance 26(11): 2155–89. Cull, Robert, Lemma Senbet, and Marco Sorge. 2002. “Deposit Insurance and Financial Development.” The Quarterly Review of Economics and Finance 42(4): 673–94 Demirgüç-Kunt, Aslı, and Enrica Detragiache. 2002. “Does Deposit Insur- ance Increase Banking System Stability?” Journal of Monetary Econom- ics 49 (7): 1373–1406. Demirgüç-Kunt, Aslı, and Edward J. Kane. 2002. “Deposit Insurance around the Globe: Where Does It Work?” Journal of Economic Perspec- tives 16 (2): 175–95. Demirgüç-Kunt, Aslı, and Tolga Sobaci. 2001. “Deposit Insurance around the World: A Database.” World Bank Economic Review 15 (3): 481–90. Diamond, Douglas W. 1996. “Financial Intermediation as Delegated Mon- itoring: A Simple Example.” Federal Reserve Bank of Richmond Eco- nomic Quarterly 82 (3): 51–66. Ebrill, Liam, Michael Keen, Jean-Paul Bodin, and Victoria Summers. 2001. The Modern VAT. Washington, D.C.: International Monetary Fund. AVOIDING THE PITFALLS IN TAXING FINANCIAL INTERMEDIATION 27 Feige, Edgar. 2000. “Taxation for the 21st Century: the Automated Pay- ment Transaction Tax.” Economic Policy 31: 473–511. Feldstein, Martin. 1983. Inflation, Tax Rules, and Capital Formation. Cambridge, Mass.: National Bureau of Economic Research. ———. 1999. “Capital Income Taxes and the Benefit of Price Stability.” In Martin Feldstein, ed., The Costs and Benefits of Price Stability. Chicago: University of Chicago Press. Gordon, Roger H. 2000. “Taxation of Capital Income vs. Labour Income: An Overview.” In Sijbren Cnossen, ed., Taxing Capital Income in the European Union: Issues and Options for Reform. Oxford: Oxford Uni- versity Press. Granville, Brigitte. 2003. “Taxation of Financial Intermediaries as a Source of Budget Revenue: Russia in the 1990s” (chapter 8, this volume). Habermeier, Karl, and Andrei Kirilenko. 2003. “Securities Transaction Taxes and Financial Markets” (chapter 11, this volume). Honohan, Patrick. 1996. “Does It Matter How Seigniorage Is Measured?” Applied Financial Economics. 6 (3): 293–300. ———. 2000. “Financial Policies and Household Saving.” In Klaus Schmidt-Hebbel and Luis Servén, eds., The Economics of Saving and Growth. Cambridge: Cambridge University Press. ———. 2003. “The Accidental Tax: Inflation and the Financial Sector” (chapter 13, this volume). Honohan, Patrick, and Joseph E. Stiglitz. 2001. “Robust Financial Re- straint.” In Gerard Caprio, Patrick Honohan, and Joseph E. Stiglitz, eds., Financial Liberalization: How Far, How Fast? New York: Cam- bridge University Press. Huizinga, Harry. 2002. “A European VAT on Financial Services.” Eco- nomic Policy 35: 499–534. Huizinga, Harry, and G. Nicodeme. 2001. “Are International Deposits Tax-Driven?” Economic Paper 156. Economic Commission, Brussels. Huizinga, Harry, and Soren Bo Nielsen. 2000. “The Taxation of Interest in Europe: A Minimum Withholding Tax?” In Sijbren Cnossen, ed., Tax- ing Capital Income in the European Union: Issues and Options for Re- form. Oxford: Oxford University Press. Jappelli, Tullio, and Luigi Pistaferri. 2003. “Tax Incentives for Household Saving and Borrowing” (chapter 4, this volume). Kirilenko, Andrei, and Victoria Summers. 2003. “Bank Debit Taxes: Yield Versus Disintermediation” (chapter 10, this volume). Laeven, Luc. 2002. “Pricing Deposit Insurance.” Policy Research Working Paper 2871. World Bank, Washington, D.C. Laurin, Alain, and Giovanni Majnoni, eds. 2003. “Bank Loan Classifica- tion and Provisioning Practices in Selected Developed and Emerging Countries.” World Bank, Washington, D.C. 28 PATRICK HONOHAN Levin, Mattias, and Peer Ritter. 2003. “Taxation of Financial Intermedia- tion in Industrial Countries” (chapter 6, this volume). Lyons, Richard. 2001. The Microstructure Approach to Exchange Rates. Cambridge, Mass.: MIT Press. OECD. 1994. Taxation and Household Saving. Paris: Organisation for Economic Co-operation and Development. Poddar, Satya. 2003. “Consumption Taxes: The Role of the Value-Added Tax” (chapter 12, this volume). Stern, Nicholas H., and Robin Burgess. 1993. “Taxation and Develop- ment.” Journal of Economic Literature 31(2): 762–830. Stiglitz, Joseph E. 1994. “The Role of the State in Financial Markets.” In Michael Bruno and Boris Pleskovic, eds., Proceedings of the World Bank Annual Conference of Development Economics 1993. Washington, D.C.: World Bank. Sunley, Emil. 2003. “Corporate Income Tax Treatment of Loan-Loss Re- serves” (chapter 9, this volume). Part I Theoretical and System-wide Issues 2 Theoretical Perspectives on the Taxation of Capital Income and Financial Services Robin Boadway and Michael Keen The financial sector is the proximate source of most capital income, and earns its keep by providing financial services of various kinds. Thus the tax treatment of capital income and financial services shape the environment within which financial intermediation oc- curs. The level and manner in which capital income is taxed is clearly likely to affect the extent and form of financial intermedia- tion, for instance, and the treatment of financial services will impact directly the provision of such services. The purpose of this chapter, therefore, is to provide a broad background for the wider discussion of financial intermediation in this book by reviewing what is known about—at least some central aspects of—the optimal taxation of capital income and financial services. While the taxation of capital income and (perhaps to a lesser, but nevertheless increasing, extent) of financial services are recurrent policy concerns in many countries, formal theoretical analyses have often focused more on the positive effects of taxation than on the design of optimal policy. Thus the territory of this chapter is at once dauntingly large and largely unexplored. For both reasons, the treatment is selective and, perhaps, eclectic. If there is a theme run- ning through the selection of topics here, it is that of addressing two strands of thought to be found in the optimal tax literature which, 31 32 ROBIN BOADWAY AND MICHAEL KEEN if accepted, would have profound implications for the tax treatment of financial activities. These are the views that capital income should not be taxed, nor should financial services. The chapter thus starts by reviewing the lessons of theory for the optimal taxation of capital income, the central issue being whether capital income should be taxed at all. This is clearly a pivotal ques- tion in thinking about the taxation of financial activities: if it is not optimal to tax (or subsidize) capital income, then the main ration- ale for taxing the income associated with financial transactions be- comes the practical one of bolstering the taxation of labor income (or consumption). The chapter then addresses the issues raised for the optimal design of taxes on capital income—assuming that such taxes are indeed desired—that have been highlighted by financial in- novation in recent years. Next, the discussion addresses the more particular, but policy relevant, question of whether, and if so at what rate and (more briefly) how, the services of financial interme- diaries should be subject to commodity taxation. The chapter con- cludes with a discussion of some of the caveats and limitations of the discussion here, some of which are addressed elsewhere in this volume. The Optimal Taxation of Capital Income Whatever its underlying source, capital income generally arises as the return on a financial asset—as interest paid on a bank deposit, for example, or as capital gains on share in a company. The tax treatment of capital income—a central feature of any tax system— thus plays a potentially key role in shaping the fiscal environment within which the financial system operates. This section sets out some of the key issues, and continuing controversies, related to the taxation of capital income. The discussion starts by considering the taxation of capital income at household level, and then turns to its taxation at business level. Two caveats common to all the models discussed should be stressed at the outset. First, all assume that the government can commit to the future path of tax rates and thus ignore the funda- mental time inconsistency problem that arises in the taxation of capital income, and so ignore a time consistency problem that is in- herent to tax design in the presence of capital accumulation. This point is taken up below. Second, it is also assumed throughout that the government is un- restricted—other than by its intertemporal budget constraint—in its THEORETICAL PERSPECTIVES 33 ability to issue debt. This effectively uncouples the path of capital accumulation from private savings decisions. If the latter would otherwise imply too rapid an addition to the capital stock, the gov- ernment can soak up the excess by selling debt; if it implies exces- sively slow accumulation, the government itself can supply funds to finance investment. The central concern in taxing capital income is to bring about the minimum distortion of lifetime consumption and labor supply decisions consistent with the government’s intertem- poral revenue needs. The Optimal Taxation of Household Capital Income There is a large literature on the optimality of taxing capital in- come—or of not taxing it. Results depend on the type of model being used and the assumptions underlying the instruments and in- formation that are available to the government. The analysis is nec- essarily dynamic in nature, since capital income is generated from assets that by definition last for more than one period. This in itself not only makes the analysis more complicated, but also raises fun- damental issues about how to model dynamic economies. Especially important in this regard is the treatment of bequests. The extreme cases of relevance are the infinite-lived dynastic model, in which be- quests are fully operative and agents are perfectly farsighted, and the overlapping-generations (OLG) model, in which there are no be- quests. An important distinction between these two cases lies in the ease with which they can accommodate heterogeneous households. In the dynastic model, each cohort is generally and naturally (though not inescapably) formulated as being a single representative agent, so the optimal tax problem adopts the classical Ramsey formula- tion. The OLG model is better able to handle heterogeneity of the sort that is most relevant for introducing redistributive concerns. To address these issues, this section begins by considering the proper role of capital income taxation in the three main cases con- sidered in the literature: an economy comprising only one type of in- dividual, who lives for only two periods; the Ramsey growth model of a single infinitely lived individual or dynasty; and an OLG model. A large part of the purpose here is to clarify the links between, and common intuition behind, the results from these different models. The discussion then turns to the case in which agents differ in their earnings capacity (so that capital income taxation may have a re- distributional role across income groups), to the effects of restric- tions on the instruments and information available to government and, finally, to the time consistency issue. 34 ROBIN BOADWAY AND MICHAEL KEEN Homogeneous households The discussion below summarizes the results for the two extreme types of models with homogeneous households—the infinite-lived representative agent model of Ram- sey, and the OLG model with a representative agent in each cohort. The discussion draws on the recent review by Erosa and Gervais (2001, 2002). BENCHMARK TWO-PERIOD CASE It is useful to begin with the simple case of optimal tax design in the context of a single individual (or a number of identical individuals) who lives for only two periods, consuming and supplying labor in each. Though artificial—the gov- ernment could in this case employ a nondistorting poll tax—this proves useful in understanding the results from (somewhat) more realistic contexts below. Suppose then that the consumer seeks to maximize lifetime util- ity U(C1,C2,L1,L2), and Ct , Lt , denote respectively consumption and labor in period t. Allowing for a full range of tax rates, this opti- mization is subject to a lifetime budget constraint: (1 + τc2 ) p2C2 (1 – τ w 2 )w2L2 (2.1) (1 + τc1 ) p1C1 + = (1 – τ w1 )w1L1 + (1 + (1 – τ r )R) (1 + (1 – τ r )R) where R is the pre-tax interest rate, the pt and wt are pre-tax com- modity prices and wage rates, and, importantly, both consumption tax rates τct and labor tax rates τwt may vary over time. The gov- ernment chooses these tax rates to maximize the consumer’s utility subject to raising some fixed present value of tax revenue (so that debt policy is implicitly assumed to be unconstrained other than by long-term solvency). A number of observations follow from this simple structure. First, although equation 2.1 allows for five tax instruments—two on consumption, two on labor, and the capital income tax—there is no loss of generality in restricting attention to any three of these, setting the other to zero. This is because there is no loss of general- ity in tax models of this kind in taking some good to be untaxed numéraire. Moreover, since the effect of the capital income tax is to distort intertemporal prices, this can be achieved by instead induc- ing appropriate time variation in the other taxes. Intuitively, with only four commodities, there is no gain in deploying any more than three taxes. Thus there is no loss of generality in supposing, for ex- ample, that the government can deploy only the two labor taxes and the capital income tax, so that the consumer’s budget constraint is simply:1 THEORETICAL PERSPECTIVES 35 p2C2 (1 – τ w 2 )w2L2 (2.2) p1C1 + = (1 – τ w1 )w1L1 + (1 + (1 – τ r )R) (1 + (1 – τ r )R) Clearly the distinctive role of the capital income tax is in its effect on prices across periods, having no effect on relative within-period prices. Second, having chosen instruments in this way, the question of whether or not capital income should be taxed reduces to the ques- tion of whether or not a particular pair of commodities should be taxed at the same rate. In equation 2.2, for instance, it can be seen that the key effect of the capital income tax is to distort the relative price of consumption in the two periods. If, therefore, it is optimal to tax consumption in the two periods at the same rate, then it is optimal not to tax capital income. The usefulness of this is that it enables one to invoke established results on the desirability of uni- form taxation within subsets of groups. Writing the expenditure function for the consumer’s optimization as E(p1 *,p *,w *,w *,u), where 2 1 2 the asterisks indicate tax-inclusive present value prices, it is optimal to tax both consumption goods at the same rate—and hence not to tax capital income—given optimal taxation of labor income in both periods, if the expenditure function takes the implicitly sepa- rable form2 E(A(p1 ⋅ *,p *,u),w *,w *,u) for some function A( ).3 A suffi- 2 1 2 cient condition for this is that the direct utility function take the form U(φ(C1,C2), L1,L2), with φ(⋅) homogeneous of degree one. Third, an implication of the preceding observation is that the an- swer to the question of whether or not it is optimal to tax capital income—or, more generally, the optimal value of tr—depends on the assumption made as to which other taxes may be levied. If, for in- stance, one instead allows for taxes on consumption in each period as well as on capital income, the optimal rate of capital income tax- ation will be zero if it is optimal to tax labor in the two periods at the same rate. For this, it is sufficient—whatever the pattern of pre- tax prices (meaning, in particular, whatever the pre-tax interest rate)—that the expenditure function takes the form E(p1 *,p *,B(w *, 2 1 * w2 ,u),u). It may seem surprising that whether or not one wishes to tax capital income turns on what is in effect merely a normalization, since clearly the allocation of resources brought about by the opti- mal tax system does not depend on what is an essentially arbitrary choice of instruments. The point is, however, that the treatment of capital income required to bring about that allocation depends on the range of instruments deployed. Suppose, for instance, that pref- erences are such that uniform taxation of consumption in the two periods is optimal, but that the only instruments available are the 36 ROBIN BOADWAY AND MICHAEL KEEN two taxes on consumption and the capital income tax. If the capital income tax rate is then set to zero, uniform taxation of C1 and C2 effectively leaves the relative prices of labor in the two periods undistorted—which is not, in general, optimal. Fourth, the optimal tax problem has a particularly simple solu- tion if preferences are additively separable over time, so that U(C1, C2,L1,L2) = U(C1,L1) + βU(C2,L2), (with β > 0 being a discount factor)—a form of separability quite distinct from the implicit sep- arability of the second observation above. Moreover, the discount factor happens to equal the pre-tax interest rate, so that β = 1/(1 + R). In this case it is optimal—whatever the choice of instrument set—not to tax capital income. Both consumption and labor supply are constant over time at the optimum, and those taxes that are de- ployed (on first and second period consumption, for instance) are optimally time invariant.4 A converse observation is also useful: if preference is of this intertemporally separable form and the opti- mum is a steady state in consumption and labor, then that optimal tax system has capital income untaxed and other tax rates time- invariant. (All this is proved in the appendix.) The intuition is sim- ply that, as is well known, these preferences—combined, critically, with an equality between the discount and interest rates—imply that in the absence of tax, the consumer would keep both con- sumption and labor constant over the life cycle. In effect, consump- tion and labor are essentially identical commodities in absence of taxation, and there is then no gain, when it comes to designing an inherently distorting tax system, from treating them differently within periods or to distort decisions between them over time. It should be noted that the sufficient conditions just described for it to be optimal not to tax capital are more restrictive than the implicit separability condition above in the sense that it also requires a for- tuitous coincidence between pre-tax prices (the interest rate) and preferences (the consumer’s discount rate). Nevertheless, this case has a particular importance in this literature: once one moves out of the two-period case into the realm of growth theory, attention nat- urally focuses on optimal taxation in steady states. THE RAMSEY GROWTH MODEL A natural extension of the two- period case just addressed is to that of the infinitely lived individual. In the simplest version of this case, the representative agent’s in- tertemporal utility function is of the additively separable form: ∞ (2.3) ∑ β U(C , L ). t t t t =0 THEORETICAL PERSPECTIVES 37 Suppose, as in the classic paper by Chamley (1986), that the govern- ment uses wage and capital income taxation only. Note that, since there are in effect infinitely many goods, this involves some loss of generality; the usefulness of adding consumption taxation to the policy menu is treated below. The household faces a per period bud- get constraint: (2.4) Ct + At +1 = wt Lt + (1 + rt )At where w – are the after-tax wage and interest rate, and A is the –- and r t t t level of household assets. The representative household maximizes equation 2.3 subject to equation 2.4, taking the future path of prices to be given. This yields the stream of household choices Ct(w –, – r ), – – – – Lt(w, r ), and thus At(w, r ). Per-period aggregate production is given by a costant returns to scale production function Yt = F(Kt ,Lt ), where Kt is the capital stock in period t, which depreciates at rate δ. De- noting the (exogenous) stream of government expenditures by Gt , the economy’s resource constraint is: (2.5) Ct + Kt +1 − (1 − δ)Kt + Gt = Yt . Before-tax factor prices are determined competitively by the rele- vant marginal productivity conditions: Fkt – δ = rt , Fᐉt = wt . Letting Bt be the public debt at time t, the per period government budget constraint is given by: (2.6) Bt +1 + τ rt At + τ wt Lt = (1 + Rt )Bt + Gt – and τ = w – w where τrt = rt – r –- are the per unit capital and labor t wt t t income tax rates.5 By Walras’ Law, either equation 2.3 or 2.4 is redundant. Delet- ing the former, the planner’s problem is then to: ∞ (2.7) Max τ wt , τ rt , bt ∑ β U[C (w , r), L (w , r)] s.t. t =0 t t t Chamley (1986) shows that the solution to this problem has a strik- ing property, and one that has become extremely influential:6 if the Ramsey solution converges to the steady state with constant con- sumption and labor supplies, then the optimal tax on capital income will go to zero. Intuitively, this can be seen as a generalization of the result for the two-period case noted above: if consumption and la- bor do not change over time, there is no gain from distorting the in- tertemporal dimension of choice between them. 38 ROBIN BOADWAY AND MICHAEL KEEN Even stronger results are obtained by restricting the form of the utility function. If the per period utility function is additively separa- ble with a constant intertemporal elasticity of consumption, so that (2.8) U (Ct , Lt ) = Ctσ − V (Lt ), σ ∈ (0,1) then capital income taxation is zero not merely in the limit but from the first period on. In period zero, capital is effectively fixed so is taxed at a confiscatory rate. After that, however, there is no need to tax capital income. The reason why it is not optimal to tax capital income after the first period can again be seen from the two-period case. For these preferences can be shown to imply that the expendi- ture function (over any finite number of periods) has the implicit separability of the form7 E(A(p1, p2 *, . . .), w *, w *, . . . . u), so that— 1 2 given optimal taxation of wage income in each period—there is no gain from taxing capital income. Though remarkable, it should be stressed that—even apart from the assumption that the government is able to commit to the future path of taxes, and is unconstrained in its debt policy—the general result of capital income taxation being zero in the long run only ap- plies in the steady state. If there are ongoing changes in the struc- ture of the economy that preclude it from converging to a steady state, the result does not apply. This analysis also assumes that only wage and capital income tax- ation are used as government policy instruments, which is a real re- striction, since there are here more than two consumption goods. As Coleman (2000) has recently pointed out, adding consumption taxa- tion can have fairly dramatic effects in terms of reinforcing the argu- ment for zero capital taxation. Adding more instruments to the gov- ernment’s policy set can only make it less likely that it will be optimal to tax capital income, since there is then less potential need to deploy capital income taxation as a surrogate for missing instruments.8 THE OVERLAPPING-GENERATIONS MODEL Erosa and Gervais (2001, 2002) take the alternative approach of analyzing labor and capi- tal income taxation in an OLG model of identical representative agents, each of whom lives for J+1 periods. In obvious notation, the lifetime utility function for a cohort born in period t is now: J (2.9) ∑ β U(C j tj , Ltj ) j =0 Labor productivity may vary over the life cycle. Letting zj be the productivity (measured in units of effective labor supply) at age j, THEORETICAL PERSPECTIVES 39 assumed to be the same for all cohorts, the per period budget con- straint is now: (2.10) Ctj + Atj +1 = wtj z j Ltj + (1 + rtj )Atj where, note, tax rates are allowed to be both time- and cohort- specific. As before, the representative household of cohort t maxi- mizes equation 2.7 subject to equation 2.8 to determine Ctj (w –-,– r ), –- – –- – Ltj(w ,r ), and thus Atj(w ,r ). With a constant population growth rate, the share of households of age j in the population, µj, will be con- stant over time. The per period budget constraint of the government can then be written: J J (2.11) Bt +1 + ∑τ j =0 rtj µ j Atj + ∑τ j =0 ltj µ j Ltj = (1 + rt )Bt + Gt The objective function of the government is assumed to be the dis- counted sum of lifetime utilities, where the discount rate is denoted γ. Denoting by Wt(w–-,– r ) the lifetime utility of cohort t, the problem of the government is then to: ∞ (2.12) Max τ wt , τ rt ,b t ∑ γ W (w , r ) s.t. t t t =0 The solution to this problem is given in Erosa and Gervais (2001, 2002). In this setting—and in sharp contrast to the Ramsey growth set- ting of Chamley (1986)—the zero capital income tax result in the long run does not apply except in special circumstances. The reason is that now consumption and leisure generally vary over the indi- vidual’s life cycle even in the steady state. There are, however, two cases in which the optimal rate of capi- tal income taxation is zero—echoing again the two cases identified in the two-period case above. One is that in which consumption and leisure are constant over the life cycle. This will be the case if the planner’s discount rate is the same as the households’ (γ = β) and pro- ductivity is constant over the life cycle (zj = z∀j ) (This latter feature ensures that not simply hours worked but effective labor supply—a distinction that did not exist in the two-period model—is constant over the life cycle). The second case is that in which the utility function takes the form of equation 2.8. In this case—as one would now expect—the opti- mal tax on capital income will again be zero for all periods beyond the first (that is, even if the consumer is not in a steady state).9 40 ROBIN BOADWAY AND MICHAEL KEEN Note too that these results presume that the government is able to deploy cohort-specific tax rates. If it cannot do so, then some tax on capital income will generally be desirable even in the circum- stances just described. (See also Alvarez and others 1992.) More precisely, Erosa and Gervais (2001) are able to show that in this case capital income taxation will be positive if and only if labor supply is decreasing over the life cycle. Of course, if consumption taxation were allowed, no capital income taxes would be required if the optimal consumption tax is constant over time, which is the case under utility functions of the type given in equation 2.6. Heterogeneous households Once households differ, the analysis becomes even more complicated. The existing analyses focus on the case in which households differ by income-earning ability, or pro- ductivity, and rely on a celebrated result of Atkinson and Stiglitz (1976). Consider an atemporal economy in which all households have identical utility functions defined on labor and a vector of n consumption goods, U(C1, . . . , Cn ,L). Households differ, however, in their wage rates wi , where wi+1 > wi , reflecting their different abil- ities or productivities. Following Mirrlees (1971), the government is unable to observe wage rates, but can observe incomes yi ≡ wi Li and so can apply a nonlinear income tax. Also, the government cannot observe individual commodity demands but can observe transac- tions anonymously, so it can also apply linear (only) excise taxes. The government is assumed to maximize a quasi-concave objec- tive function, which ensures that redistribution goes from higher to lower wage households. Its problem is to choose a set of tax policies to maximize its objective function subject to a resource constraint, and a set of incentive compatibility (“self-selection”) constraints. The constraints ensure that higher-wage households prefer their own incomes (both earned and disposable) to that of lower-wage house- holds; in other words, the constraints state that the greater amount of leisure that the high-ability types could take if they were to earn the same gross income as the low-ability types would be less valuable to them than the consumption they would forego. The Atkinson- Stiglitz theorem says that if the utility function is weakly separable in goods and labor, so that U(C1, . . . , Cn ,L) = U(g(C1, . . . , Cn ),L), then differentiated commodity taxes should not be used. The intuition for this is straightforward.10 The nonlinear income tax is deployed to achieve the optimal amount of redistribution. At the optimum, incentive constraints will be binding on all but the lowest-wage households; otherwise more redistribution could be accomplished. Differential commodity taxation will thus be useful THEORETICAL PERSPECTIVES 41 only to the extent that it weakens the incentive constraints. Under weak separability, the preference ordering over goods is indepen- dent of labor supply. Therefore higher-wage “mimickers” would choose the same goods bundle as the lower-wage households they are tempted to mimic. But then differential commodity taxation cannot make it less attractive to them to mimic, and so has no ef- fect other than the adverse one of distorting consumer choice. By the same token, if the Atkinson-Stiglitz separability condition does not apply, the self-selection constraint will be weakened by impos- ing higher commodity taxes on goods that are more complementary with leisure—which are the ones that will be more heavily demanded by the mimickers. The Atkinson-Stiglitz result can be applied to the problem of cap- ital income taxation by reinterpreting the distinct goods as con- sumption at different dates. Differential commodity taxation of the atemporal interpretation then translates into a distortion of in- tertemporal prices, and so corresponds to the taxation of capital in- come. Noting this analogy, Stiglitz (1987) shows that if household utilities are weakly separable between goods and labor, then all that is needed is an optimal nonlinear tax within each period, with no tax or subsidy applied to interest income. The reasoning is essen- tially the same as in the atemporal case: being a differential tax on future consumption, an interest tax would only be used if future consumption were more complementary with leisure than is current period consumption, and an interest subsidy in the case of relative substitutability.11 If leisure and future consumption are comple- ments, an indirect tax on future consumption weakens the self- selection constraint: higher-wage mimickers will enjoy more leisure, and their consumption bundle will include a higher proportion of future consumption than low-ability persons. A tax on future con- sumption will make it relatively more difficult for the former to mimic the latter. Though this analysis is suggestive, there are a number of caveats and restrictions that should be noted. First, if the labor supplies of households of different productivities are not perfect substitutes, in- terest income taxation or subsidization is called for, depending on the substitutability-complementarity relationships among the differ- ent types of labor and capital, as Stiglitz (1987) shows. Second, and more profoundly, the “trick” of reinterpreting commodities as con- sumption at different dates is more problematic than it may at first appear, and really works only in the two-period case. Once one goes to the more realistic world of multiple periods with labor supplied in several of them, the Atkinson-Stiglitz logic no longer applies. For 42 ROBIN BOADWAY AND MICHAEL KEEN one thing, as noted in the earlier discussion of taxation with homo- geneous households, allowing labor supply to vary for more than one period makes the case for taxing or subsidizing interest income more likely. Even more fundamentally, once households are subject to nonlinear income taxation in successive periods, the nature of their optimization problem and of the government’s information constraint becomes much more complicated. For instance, the gov- ernment may use consumers’ first period decisions on earnings to make inferences about underlying ability; and consumers’ aware- ness of this may then affect those earning decisions. To date, there is no fully satisfactory analysis of intertemporal nonlinear income taxation, making it difficult to say anything even vaguely definitive about the case for taxing or subsidizing capital income in a world of heterogeneous households. Restrictions on tax instruments and information The above mod- els of optimal taxation assume that the government is both well in- formed (though not quite fully) and unrestricted in its policy instru- ments. But the policy environment may be more restrictive than that. Take, for instance, the Ramsey growth context and suppose that while the government can levy consumption, labor income, and capital income taxes, the labor income tax rate is constrained to be nonnegative. Coleman (2000) demonstrates two remarkable results. First, suppose that at least some of the government revenue raised through taxes is transferred lump sum to the household, whose members then consume it. Since this lump-sum income is spent on taxable consumption goods, a consumption tax is somewhat more general than a labor income tax. If one now constrains labor taxa- tion to be nonnegative as above, labor income should never be taxed, while capital income should be subsidized in the long run. In effect, the consumption tax is used to raise revenue and cannot be replicated by labor income taxation. On the other hand, if all tax rates must be nonnegative, in the long run it is optimal to tax only consumption, while in the intermediate run, all three tax sources might be used. Second, suppose that all government revenue is used to purchase public goods or services and none is transferred to households as cash. If labor income taxation is precluded from being negative, so a first best cannot be achieved, then a consump- tion tax is no longer needed beyond the first period: its effects can be replicated by combinations of the other two taxes. One is back in the world of Chamley (1986) discussed above, along with all his results. In particular, capital income taxes converge to zero as the economy converges to a steady state. Restrictions on instruments may also be important in the context of taxing heterogeneous households. One possibility, for instance, is THEORETICAL PERSPECTIVES 43 that the government can use only linear income taxes, perhaps be- cause of restricted administrative capacity. Deaton and Stern (1986) show that in this case differential commodity taxes can be dispensed with if the utility function is not only weakly separable but, more- over, if the sub-utility function of goods is quasi-homothetic so that all Engel curves are linear. In this case—which is clearly more re- strictive than that of the Atkinson-Stiglitz result above—commodity taxes do not add anything to the ability of the government to dif- ferentiate among households of different income levels. Applying the same intertemporal interpretation as above, the implication is that the circumstances under which capital income taxation can be dispensed with are even more restrictive than those that apply when nonlinear taxes can be used. Again, the results are agnostic with re- spect to whether taxation or subsidization of capital income is op- timal in the event that the required restrictions do not apply Another line of inquiry, and one that is potentially especially rel- evant for financial markets, assumes that while the government can observe household labor income (as above), it cannot observe capi- tal income. At best it can tax capital income anonymously and in- directly using a constant tax rate. According to the above analysis, no such capital income tax would be needed if the conditions of the Atkinson-Stiglitz theorem or its extensions were satisfied. Two sets of circumstances have been identified in which capital income might nevertheless be taxed even if the required separability conditions were to hold. The first arises because of the possibility of unob- served bequests. As argued by Boadway, Marchand, and Pestieau (2000) and Cremer, Pestieau, and Rochet (2003), if for some reason the size of bequests is not statistically independent of wage rates, the taxation or subsidization of interest income can be a useful supple- ment to an optimal nonlinear labor income tax. In particular, a tax (subsidy) should be used if bequests are positively (negatively) cor- related with wage rates. Gordon (2001) has identified a second circumstance. He con- structs a model in which households choose a portfolio of assets as well as an amount of savings. The portfolio combines a risky asset (equity) and a risk-free asset (debt). He argues that if risk aversion falls with wealth, higher-income households will hold relatively more equity in their portfolios and lower-income households will hold more debt. In these circumstances, distributional consideration means that a subsidy on debt will be optimal. The time consistency issue As noted at the outset, all the analyses reviewed above assume that the government can costlessly commit to the future path of tax rates. When it cannot so commit, however, there is an inherent temptation to renege on previously announced 44 ROBIN BOADWAY AND MICHAEL KEEN plans. The point here is that at any given time, the stock of previ- ously accumulated capital is large relative to the annual flow of new investment. In these circumstances, rational governments will be unable to avoid imposing high taxes on that capital and the income it generates, despite the fact that in the long run these taxes decrease capital accumulation and lead to inefficiencies (Fischer 1980). That is, even though governments would wish to commit, if they cannot do so then they will wish to impose heavier taxes on capital once savings decisions have been made than they intended ex ante. This may be one reason why rates of capital income taxation observed in the real world tend to be much higher than optimal tax analysis of the kind above would advocate (though the fact that effective rates of capital income tax are often lower than statutory rates, and sometimes even negative—reflecting the availability of deductions— rather dulls the point.)12 This time consistency problem has potentially profound implica- tions for capital income taxation. In politically stable countries, gov- ernments may carry sufficient credibility for the problem to be of only limited significance. In others, however—especially, it would seem, many developing countries—the problem may be more profound. Faced with this problem, governments, recognizing that they are un- able to avoid excessive taxes on capital income—and that the private sector will also recognize this, and amend its behavior accordingly— can take measures to undo the worst effects of excessive taxation. They may find it desirable, for instance, to allow capital to move to lower tax jurisdictions (Chari, Kehoe, and Prescott 1989) or even to tolerate some degree of evasion of capital income taxes (Boadway and Keen 1998). Other measures to overcome the time consistency prob- lem involve up-front subsidies to investment, such as tax holidays and investment tax credits. Such subsidies may also be based on political economy arguments, such as common agency arguments for political parties providing favors in return for financial support (Marceau and Smart 2003). Designing policies to overcome inefficiencies arising from the political process is, of course, problematic since it is not clear how governments could be persuaded to undertake them. Optimal Taxation of Corporate Income There are two main types of arguments for taxing corporate income as part of an optimal tax system. The first—which is the one typi- cally stressed in policy discussion—sees the corporate tax as a with- holding device against capital income at source. The second sees the corporate tax as a tax on economic rents earned by corporations. This section considers these in turn. THEORETICAL PERSPECTIVES 45 The withholding role of the corporation tax Corporations are legal entities, distinct from their owners, in which incomes earned on be- half of shareholders are pooled. If shareholders could be taxed di- rectly on the income they earn through corporate ownership, the withholding role would disappear. Distinct considerations arise, however, in relation to domestic and foreign shareholders. WITHHOLDING AGAINST THE INCOME OF DOMESTIC SHAREHOLDERS The case for using the corporate tax as a withholding device against domestic shareholders is based on two presumptions: that it is de- sirable to tax capital income from all sources at the personal level, and that it is difficult to attribute all income earned through corpo- rations to individual shareholders as it accrues. The key issue here is the difficulty of taxing capital gains on an accruals basis at the personal level (an issue that is discussed later). If, as is almost in- variably the case in practice, capital gains are taxed on a realization basis, then income that is retained and reinvested in the corporate sector can be sheltered from personal tax until it is taken out as ei- ther sales of shares or future dividends. Taxing corporate income as it is earned is a way of bringing this income into the tax system at the appropriate time. Designing a corporate tax system for this purpose involves a number of considerations, some of which are difficult to implement perfectly. The base for the corporate tax should be only retained earnings, excluding both interest payments and dividends (since these can readily be taxed at the personal level). All accrued current and capital costs should be deductible, including true depreciation of depreciable assets and the imputed cost of inventory use, natural resources, and intangible assets—and there are well-known difficul- ties in properly accounting for these. However, as discussed shortly, there are reasons to include all eq- uity income earned by the corporation to satisfy the other with- holding role—that with respect to the income of foreign sharehold- ers, whose dividends escape domestic taxation. To the extent that the corporate tax is intended to be a with- holding device, credit for this withheld tax should be given to share- holders when they eventually take their income out of the corpora- tion. This requires some form of integration applied at the personal level: for example, a tax credit could be given against dividends and capital gains earned in corporations that had paid the corporate tax. This is conceptually straightforward for dividends but problematic for capital gains, since it would require identifying that part of gains that is due to taxed retentions, a daunting task.13 Special provisions would need to be applied to deal with some types of ownership. First, corporations might themselves hold shares 46 ROBIN BOADWAY AND MICHAEL KEEN in others, either subsidiaries or as minority owners. To avoid double taxation of corporate source income, the inter-corporate flows of capital income should be excluded from tax. Thus if the corporate tax applies to corporate retained earnings, capital gains earned by domestic corporations on their shares in other corporations should be tax-exempt.14 If the tax applies to all corporate equity income, both capital gains and dividends earned by domestic corporations should be exempt. This has obvious implications for financial cor- porations. Second, careful attention should be paid to the taxation of small corporations. The aim should be to ensure that they are treated on a par with unincorporated businesses so that there are no pure tax incentives for incorporation. Note that domestic households might still be able to shelter their capital income in foreign corporations, and the domestic corporate tax cannot do anything to mitigate that. If the foreign country has very low corporate tax rates, or if households are able to avoid pay- ing domestic tax by simply failing to report their incomes, house- holds will be able to earn tax sheltered income abroad and there is little the corporate tax system can do to avoid that. WITHHOLDING AGAINST THE INCOME OF FOREIGN SHAREHOLDERS Fur- ther considerations arise in an open economy. Since the personal in- come tax is generally levied on a residence basis, and so does not apply to foreign shareholders, the withholding role applies only, in general, to a subset of shareholders.15 More fundamentally, if in- vestors in a corporation are able to earn some more or less fixed after-tax rate of return on the world market, any tax levied on their capital income will ultimately end up being shifted back to domestic residents through induced capital outflows, and hence a reduced rate of return to immobile domestic factors. In these circumstances, one might not want to tax corporations at all (although time-consistency problems may nonetheless lead to taxation, as mentioned earlier). But other withholding arguments can be made for taxing foreign- owned corporations. If the country has some power in world capi- tal markets, so that it need not take the after-tax return on world markets as given, then a tax on foreigners will indeed to some ex- tent be borne by them. Most countries, however—and especially de- veloping countries—are not in this position, except perhaps in rela- tion to some natural resources. More important in practice is the observation that, to the extent that foreign-owned corporations can credit tax paid in the host country against corporate income tax in their home country, such a tax will serve mainly to transfer tax rev- enues from foreign treasuries to the domestic one. This argument THEORETICAL PERSPECTIVES 47 seems to have some weight in debtor countries, such as Canada, and is a major consideration in tax-setting in many developing coun- tries. Why creditor countries might allow crediting for taxes paid abroad by their domestic corporations is a matter of continuing re- search interest.16 But, given that major investing countries like the United States have such a system, the case for capital-importing countries using a corporate tax is strengthened. The design of a tax to exploit the tax crediting practices of cred- itor nations is somewhat different from that whose role is to with- hold against domestic shareholders. There is some incentive to mimic the tax structure and rate of the main creditor nations to exploit most effectively the tax revenue transfer. Since tax crediting might be jeopardized by discriminatory treatment of foreign versus do- mestic corporations, that implies that the tax treatment of corpora- tions of all types will partly be driven by a desire to exploit the transfer of revenues from foreign treasuries. The need to treat foreign and domestic corporations on a par has a further important implication. Provisions for integrating corpo- rate and personal taxes must be implemented at the personal level rather than the corporate in order to target them to domestic share- holders alone. That is, credit must be made to resident households against corporate taxes paid by domestic corporations in which they have shares rather than, say, credit being paid to corporations that pay out dividends to domestic residents. However, in this case, the rationale for integration is no longer obvious, as Boadway and Bruce (1992) argue. If the marginal investor in a corporation is a nonresident, the degree of integration will affect the domestic resi- dent’s incentive to save, while the firm’s incentive to invest will be determined by the corporation tax. Thus integrating via the personal tax undoes any tax that might apply at the personal level, rather than undoing the corporate tax; and the corporate tax will effec- tively be borne by the firms themselves, causing a distortion in in- vestment. Reflecting in part increased awareness of the relative inef- fectiveness of integration in open economies, a number of economies have in recent years moved away from integration schemes toward a “classical” form of corporation tax, under which dividends are taxed without allowance for underlying corporation tax paid.17 Thus the case for corporate tax itself comes into question in an open economy, except to the extent that it is effective at transferring income from foreign treasuries. Even so, the benefits of that tax- transfer must be offset against resulting investment distortions, which considerably weaken the case for integration. 48 ROBIN BOADWAY AND MICHAEL KEEN The corporation tax as a rent collector The second possible role of the income tax, and one that has traditionally been stressed in the academic literature, is that of a tax on economic rents (or “pure” profits), meaning the excess of revenues over the full opportunity costs of earning them. The argument is that most economic rents earned in the economy will be earned in the corporate sector, and so are conveniently identified and taxed there. An appropriately de- signed corporate tax—one that is neutral, in the sense of not dis- torting corporate decisions—will divert some of the rents from the private sector to the public sector. Such a tax will be fully efficient, and thus economically ideal, in that it does not distort production decisions by the taxed firm. Whether it is equitable to tax rents at the level of the firm is another matter. The tax can be viewed as a confiscatory tax on the current owners of assets that give rise to fu- ture rents, which may be regarded as horizontally inequitable to the extent that other sources of asset wealth are not so taxed. Rents can come from a variety of sources. One important source might be the ownership of fixed factors, such as natural resources, that have no alternative use. Other sources of rent are locational or informational advantages. Yet another might be the monopoly prof- its that come with market influence. The economic modeling of neu- tral corporate taxes has tended to focus on the fixed factor story, using a dynamic model of the firm (see King 1977; Atkinson and Stiglitz 1980; Auerbach 1983; Auerbach and Kotlikoff 1987). The firm is viewed as having a strictly concave production function with possibly both current and capital inputs. The strict concavity gives rise to economic rents at the profit maximizing (as well as the social) optimum. It is straightforward to extend the analysis to take ac- count of other sources of rents, such as monopoly of renewable or nonrenewable natural resources. As well, the focus of the literature has tended to be on firms that earn business incomes from the sales of goods and services rather than those that earn financial income. International considerations again matter, however. To the extent that the source of the rents is not specific to a particular location— access to a particular mineral deposit, for instance, or to a protected local market—a tax on rents may drive investments to other juris- dictions offering a lower rate of tax. There are various ways, in principle, to design a corporate tax so that it is neutral. Two benchmarks are an economic profits tax and a cash flow tax (Boadway 1980). An economic profits tax has as its base the imputed current-period economic profit of the firm. This requires accounting for all revenues less costs on an accrual basis. Revenues include those from all sales of goods and services in the THEORETICAL PERSPECTIVES 49 accounting period, all measured on an accrual (rather than cash flow) basis. Costs are more complex. The costs of using current in- puts within the accounting period are deducted on an accrual basis. The costs of using capital inputs must be appropriately imputed to each accounting period. For depreciable capital, this involves both true economic depreciation and the full cost of financing the depre- ciable asset holdings. For depletable assets, it involves the value of resources depleted in the period as well as the financing costs. For inventory, it is the value of goods taken out of inventory plus the cost of financing. For intangible assets, it includes the depreciation of those assets, and so on. Needless to say, there are difficult mea- surement costs involved in measuring economic profits: appropriate market prices—such as for intangibles—often do not exist. The true values of economic depreciation and depletion of assets are difficult to measure. The costs of financing assets must take account of both debt and equity financing. Costs of depreciation and finance must be adjusted for inflation, and a cost of risk-taking must be imputed to reflect uncertainty about future input and output prices. These difficulties make the implementation of an economic prof- its tax virtually unfeasible. It is much easier to apply the cash flow alternative, which takes advantage of the fact that the present value of economic profits is also the present value of cash flows of the firm. A cash flow tax on business income takes as its base the cash flow of revenues from the sale of output less the cash flow cost of inputs as they accrue. There are two main variants of such a tax. Under an “R-base” cash flow tax—in the terminology of the Meade Committee (1978)—net financial inflows related to debt (interest payments and repayments of interest and principal) are excluded from the base. Under an “R+F”-base,” they are included; such a tax is equivalent (as a consequence of the identity between the firm’s sources and uses of funds) to an “S-based” tax levied directly on net distributions to shareholders (dividends less new equity issues). Under a cash flow tax, the difficulties of accrual accounting no longer apply. Capital purchases are simply expensed in the ac- counting year in which they are made, and no allowance need be made for the costs of financing. A cash flow tax avoids the need to measure true depreciation and depletion, the true costs of equity fi- nancing, and the costs of risk. Moreover, no inflation accounting is required. This simplicity does come with a cost, at least from the policymaker’s perspective. First, neutrality of the cash flow tax re- quires that the tax rate be constant over time, whereas a pure in- come tax retains neutrality even if the tax rate changes year by year.18 Second, it will typically be the case that cash flows are neg- 50 ROBIN BOADWAY AND MICHAEL KEEN ative for some firms, especially young, growing ones that are en- gaged in large investments. For the tax to be fully neutral, tax losses must be refundable in the year in which they occur (or be carried forward at an appropriate rate of interest), something that policy- makers are notoriously averse to allowing (perhaps reasonably enough, given the potential danger of firms disappearing once they have claimed tax rebates on losses). This problem can be overcome, and the advantages of cash flow accounting maintained, under a modified cash flow tax (Boadway and Bruce 1984). Under this alternative, not all investments are fully expensed in the year in which they occur. Instead any arbitrary amount of them can be added to an account—the undepreciated cap- ital stock for tax purposes. The firm is allowed to draw down the un- depreciated capital stock at some rate each year, and a deduction is given to the firm for the financial costs associated with holding that undepreciated capital stock. This scheme retains the virtues of a cash flow tax, especially the use of cash rather than accrual accounting and the avoidance of inflation indexing. It can also avoid the main problem of the cash flow tax, the need for refundability of losses. The ability to draw down the undepreciated capital stock can be re- stricted to ensure that negative taxable income is not generated. The main difficulty with implementing the tax is to choose an appropri- ate cost of finance to apply to the undepreciated capital stock.19 One form of modified cash flow tax that has attracted particular interest is the “Allowance for Corporate Equity” (ACE) scheme, un- der which companies are allowed to deduct against tax an amount equal to some notional rate of return on its invested equity (calcu- lated retentions plus new equity issues in the current period). Such a scheme—whose properties are developed in detail by Devereux and Freeman (1991)—was implemented by Croatia in the latter 1990s.20 Similar schemes—but with the imputed equity taxed at a reduced rate rather than fully exempted—have been adopted in Austria, Brazil, and (until 2001) Italy. These are perhaps the most deliberate attempts that have yet been made to craft the general corporation tax as, in effect, a tax on rents. The case for a tax on economic rents depends on their quantita- tive importance. One would expect that some sectors are more likely to generate economic rents than others. Thus rent-type taxes are most often found in relation to the resource extraction.21 Whether it is useful to apply a rent tax to the entire corporate sector is an open question. Treatment of financial institutions It might be argued that there is a possibility of economic profits being a feature of the financial sec- THEORETICAL PERSPECTIVES 51 tor, given the usual dominance of a few large firms. In principle, a cash flow-type tax can also be designed to apply to financial income. Two alternatives are possible. One approach would be to treat fi- nancial assets accumulated by the firm for the purposes of earning income analogously to the treatment of real capital in the non- financial firm. The costs of holding them would be deductible and the revenues they generate taxable. In a pure cash flow system, the tax base applying to the financial earning of the firm would include financial income less the net acquisition of assets less the real costs associated with intermediation. More formally, denoting by A the ⋅ income-earning financial assets of the firm, by A its rate of change, ⋅ and by r the interest rate on these assets, the base is rA – A – C, where C includes all the non-financial costs incurred on a cash flow basis. (Modified cash flow taxation would create an account into which fi- nancial assets could be put, similar to the undepreciated stock of capital under the modified cash flow system.) The other alternative, which is analogous to the S-base described above, would also include net financial transaction with nonshare- holders on the liability side. These would include debt transactions as well as liabilities in the form of deposits into the intermediary. For these transactions, the firm would add to its tax base the cash flow transactions arising from the acquisition of these liabilities: net changes in the stock of such liabilities less the payment of interest on them. Denoting the stock of debt by B, the base in this case is ⋅ ⋅ rA – A + B – ρB – C.22 The case for including these elements of fi- nancing arises because of the possibility of the financial institution earning pure profits because of the services it provides and the mar- ket power it might have. In the end, the case for imposing a cash flow tax on financial corporations, and the form that tax should take, depend on a judgment as to whether the rents that are earned are significant enough to warrant the administrative costs of im- posing the tax. Summary The literature on the optimal taxation of capital income is large, dis- parate, and sometimes difficult. It has focused, to a large degree, on the question of whether capital income should be taxed at all. The answer to this depends on what other instruments are available to the government. If it is unrestricted in its ability to tax consumption and labor income, for instance, then the capital income tax is a re- dundant instrument, since its effects can be replicated by introduc- ing appropriate time variation in these other taxes. Even when the choice of instruments is such that the capital income tax is not re- 52 ROBIN BOADWAY AND MICHAEL KEEN dundant, however, there are important benchmark cases in which it should not be used. This is the case if the optimum happens to have the feature that each individuals’ consumption and labor supply re- main constant over time—as in the steady state of a Ramsey growth model—or in which intertemporal preferences have the feature that there is no gain from distorting intertemporal prices. Clearly though, these are both restrictive sets of circumstances. It is a great weakness of the literature—reflecting in part the an- alytical complexity of the issue—that it gives relatively little firm guidance on the rates at which capital income should be taxed (or even whether the tax should be positive) outside these special cases. All these results are subject, moreover, to the fundamental time con- sistency problem: even if the government would wish not to tax cap- ital income, it may be forced to do so by the expectation of the pri- vate sector that, ex post, some such tax will be optimal. Attention then focuses on the devices that government might use to mitigate this problem, a line of inquiry that is proving fruitful in under- standing why it is, for instance, that many developing countries con- tinue to find it attractive to offer tax holidays and other up-front in- centives to foreign investors. Taxing income at the corporate level should be seen as a means to the end of taxing households, and to that extent is ultimately mo- tivated by administrative rather than theoretical concerns. There is potentially an important role for the corporate tax as a means of withholding against personal capital income, and a corporate level tax may be a particularly convenient way of levying a nondistorting tax on rents. In this area at least, theory has provided quite clear guidelines as to proper tax design. New Financial Instruments and the Capital Gains Problem The proliferation of new financial instruments over the last two decades or so has raised profound conceptual and practical problems for tax design. These issues—which remain largely unresolved—are not just ones for the most advanced economies to grapple with. They affect emerging market economies with sophisticated financial sec- tors, and even many relatively low-income countries must address the difficulties they may consequently face from the financial activi- ties of some of their largest (and often foreign-owned) taxpayers. This section briefly reviews the main issues arising from, practice in respect of, and potential policy responses to continuing financial innovation.23 THEORETICAL PERSPECTIVES 53 Financial Innovation: The Tax Issues The essence of financial innovation is the creation of new assets by repackaging the cash flows generated by others. This can be done either by disaggregating the cash flows from some asset into a series of constituent parts or by combining the flows associated with a set of distinct assets. Examples of the former would include the strip- ping of interest and principal payments from debt for sale as sepa- rate assets; the use of share options to isolate upside and downside risk in price movements; and the use of notional principal contracts, such as swaps, to create assets with payoff equal to the difference between the payoffs on existing assets. An example of the latter form of repackaging is the convertible debt contract, combining debt obligations with an option to convert the claim into equity at a pre-specified exercise price. Such repackaging implies, of course, equivalence between the values of the related financial assets (since otherwise costless arbi- trage profits could be made by selling some subset and buying an- other). Holding both a share (with current price S) and a put upon it (current price P) gives the same pattern of payoffs, for example, as holding a zero-coupon bond with principal amount equal to the exercise price of the put (current price B) with principal repayment of S and a call option also with the same exercise price of the put (C).24 It must therefore be the case that: (2.13) S + P = B + C. Similarly, holding a forward contract on some asset (being obliged to sell in the future at some specified price) is equivalent to holding a call on the underlying asset and selling a put at the same exercise price, so that F = C – P, where F denotes the value of the forward contract. There are other equivalences, of course. A con- vertible bond, for example, is equivalent to the combination of a standard debt contract and a call option on the company’s equity. The massive financial innovation of recent years has been driven much more by commercial considerations—especially the better management of risk—than by the avoidance of tax. Nevertheless, it clearly holds the potential for facilitating the avoidance in so far as repackaging can bring about a change in the pattern of tax pay- ments. By the same token, differential tax treatment can distort the forms of packaging chosen, and perhaps even prevent the emer- gence of socially useful financial instruments. The use of financial transactions to reduce taxation is of course long established. The tax preference for debt over equity, for in- 54 ROBIN BOADWAY AND MICHAEL KEEN stance, has long figured centrally in the analysis of corporate finan- cial decisions. Clearly, however, the world has become more com- plex than that: even the distinction between debt and equity has be- come less clear-cut with the development of such instruments as the monthly income preferred shares (treated in the United States as debt for tax purposes, but equity for accounting purposes). Perhaps more fundamentally—since the asymmetry between debt and equity is one that could in principle be quite easily removed25—financial innovation may increase the scope for exploiting the differential taxation of some items as “income”—in the legalistic sense of being taxable on accrual at normal rates of income tax—and others as “capital,” with the latter benefiting from the advantage of taxation on realization rather than accrual and, in many cases, a statutory tax rate below that applied on ordinary income. (Deductions, con- versely, are more advantageously taken against ordinary income.) Any difference in statutory rates is, in principle, easily remedied. In this sense the more fundamental issues are those arising from the de- ferral of tax on capital gains. Exploiting this feature of most systems has long been a central concern of tax planners. One such device is the use of the “strad- dle”: simultaneously holding and selling short the same asset, then closing the losing side at the end of the tax year (so taking any de- duction available for capital losses) and offsetting the position car- ried forward in a similar way during the next tax year (Stiglitz 1983). Another is the zero-coupon (or “original issue discount”) bond: one paying no interest but only a fixed amount on redemp- tion, so yielding all its return in the form of capital gains rather than immediately taxable interest income. The new financial instruments amplify the scope for taking returns as capital rather than income and exploiting the benefits of deferral. Suppose, for instance, that capital gains are tax-exempt if the underlying asset is held for more than five years, but taxable as ordinary income if held for less. Con- sider the situation of an investor who has held an asset for four years and would like to realize its value but will face a tax charge if she does. Instead of selling the shares now, she might bor- row an amount equal to their current value S and simultaneously sell a call and buy a put upon it at exercise price S. The proceeds of the loan can be used to finance consumption of S in year 4; when year 5 comes long, the share itself provides exactly the resources needed to repay the loan and offset the position on the option.26 At no risk, the investor is thus able to entirely avoid bringing the cap- ital gain into tax. There are other ways of achieving much the same effect. One would be by “bed and breakfasting”—selling the share THEORETICAL PERSPECTIVES 55 on the last day of the tax year and repurchasing the next—though in this case risk is not entirely eliminated. The point of the example is rather to illustrate ways in which new instruments can expand the range of avoidance devices available to investors. This further exposure of the problems posed by the differential treatment of capital and income items is not the only tax issue raised by financial innovation. Commercially advantageous interest rate swaps, for instance, may be rendered unprofitable if withhold- ing taxes are levied on the gross interest payments associated with the swap rather than with the net flows.27 The proper treatment of stock options given to employees has remained unclear. Should the value of the option at award be treated as employment income, for instance, or should tax be levied only at the time of any realized gain subsequent to the exercise of the option? But it is the issue of deferral that is the most profound. Current Practice Many tax systems address the most obvious devices to exploit de- ferral rules. The United States, for example, seeks to undo straddles by requiring gains on any position offsetting a loss to be treated as realized; and rules are commonly provided for treating accrued gains on zero-coupon bonds as interest. But such rules are ad hoc responses to particular difficulties rather than solutions to the gen- eral problem. Straddle rules may pick up investors who simultane- ously buy and sell short shares in the same companies, but it may be more difficult to apply them to offsetting positions in different companies, perhaps in different sectors, whose returns are closely correlated. Indeed the adoption of such rules may itself provide an incentive to use more sophisticated financial instruments: the adop- tion of bed and breakfasting rules, for instance—ignoring the sale of an asset that is repurchased shortly after—may increase the attrac- tions of the option-based device above. More generally, however—and especially in developing countries— the tax treatment of new instruments is often unclear. Many countries make no explicit provision for such instruments in their tax legisla- tion. Their treatment then becomes a matter of interpretation, prece- dence, and practice in such matters as the degree to which tax treat- ment follows accounting rules. Even where explicit rules are in place, they differ significantly. Within the OECD, for instance, some coun- tries treat share options as instruments in their own right (taxing the premium in the hands of the issuer at the time of issue, and the holder on the gain in value at the time of exercise). Others look rather to the 56 ROBIN BOADWAY AND MICHAEL KEEN transaction in the share (taxing the premium, if at all, only on exer- cise, and the holder only on subsequent sale of shares acquired at ex- ercise).28 This variation of treatment is in itself a source of difficulty in an international context,29 since it creates scope for the exploitation of inconsistencies in, for example, the timing of a taxable event. More fundamentally, it reflects the lack of a coherent intellectual framework for the taxation of financial instruments. Policy Responses To a large degree the tax problems posed by financial innovation are not new in themselves. The issue, rather, is that the repackaging that is their essence exposes still more clearly weakness in the pre- existing tax system: most noticeably, the distinction between items taxable as ordinary income and others taxable as capital gain, and most fundamentally, the problems posed by the usual practice of taxing the latter on realization. Financial innovation has thus lent renewed importance to the search for coherence in the taxation of alternative forms of capital income. Coherence in this context means, in the first instance, neutrality: the principle that taxation should not distort the pattern of finan- cial transactions that investors, both individuals and firms, choose to undertake. Arguments are sometimes made in support of provid- ing tax preference to some kinds of financial income over others. In particular, the preferentially low tax rates on capital gains that many countries offer, often with rates lower the longer an asset is held, are sometimes rationalized as a counter to an inherent short- termism of investors. Most of these arguments, however, have little force. For instance, charging long-term gains at a reduced rate may give further encouragement to holding assets for longer periods— something that the ability to defer taxable gains does in itself, of course. However, it also gives an incentive to realize losses sooner rather than later, in order to take any associated deduction at a higher rate. In any event, the discussion that follows focuses on the question of how to achieve the key benchmark of neutrality. What is needed, in principle, is some structure that treats identi- cally for tax purposes all conceivable repackaging of any stream of cash flows. Holding a share and selling a call, for instance, should have exactly the same tax consequences, in each state of nature, as borrowing and selling a put. Three main approaches to this prob- lem have been suggested. One is “bifurcation,” meaning the decomposition of an asset into its basic components and taxation by reference to the latter. This is a natural response to some composite assets, such as the convertible THEORETICAL PERSPECTIVES 57 bond. More generally, however, it requires the identification of some set of fundamental assets that are the basic objects of the tax rules, to- gether with agreement on a unique decomposition. Nor does this in it- self address the difficulties that arise from the taxation of some assets on a realization basis. It is to this problem—“fixing realizations”— that the remaining approaches are centrally addressed. The second strategy is to “mark-to-market,” meaning that tax is charged on the increase in market value of net assets over some ref- erence period (most obviously the start and finish of the tax year). This, in effect, is taxation on accrual. This has the merit of treating all assets uniformly and eliminating any incentive for selective real- ization. Marking-to-market is indeed the norm for market makers, and there is in principle no difficulty in extending it to the transac- tions in marketed securities of major companies and wealthy indi- viduals. The method is not easily applied, however, to nonmarketed assets, and indeed one effect of its adoption would be to distort the choice between the two. A further common objection to this ap- proach is that it may also present taxpayers with liquidity problems, requiring them to dispose of some part of the asset in order to pay their tax bill. It is hard to see how this could be a significant problem, however, for most marketed assets to which marking-to- market could plausibly be applied. A more real concern for many governments is that marking-to-market may also increase the volatility of their tax revenues, since gains and losses would all be taxed as they accrue without any smoothing from those who would otherwise choose to defer a gain or hold a loss. The third approach is to retain taxation on realization but to charge that tax in such a way as to mimic taxation on accrual. Since valuations do not need to be observed continuously, this approach could be applied to a wider set of assets than marking-to-market, and is likely too to imply less volatility of revenues. Three ways of mimicking accrual taxation have been suggested. One form of mimicry is the method proposed by Vickrey (1939) and the Meade Committee (1978), and it is simply to apportion the realized gain over the holding period and charge tax on realization equal in present value to the tax applicable to the implied series of gains. Consider for instance an asset acquired for price P(0) and sold T periods later for P(T). Denoting the risk-free interest rate by i, the tax charge at realization under this scheme is T TV = ∫ 0 τge i(1− τ)(T − s)P(T )e − g(T − s)ds (2.14) τg = (1 − e[i(1− τ)− g]T )P(T ) g − i(1 − τ) 58 ROBIN BOADWAY AND MICHAEL KEEN where g = (1/T )ln[P(T)/P(0)] denotes the average rate at which gains cumulate over the holding period. Any such allocation is essentially arbitrary, however, so that some distortion of holding periods will remain. It remains tax advantageous to defer selling assets holding accrued gains, since, all else equal, those gains will then be allocated over a longer horizon. An alternative form of mimicry is the suggestion by Auerbach (1991) that taxation at realization be calculated on the assumption that value has accumulated at the risk-free rate of return through- out the holding period, so that liability at realization is T (2.15) TA = ∫0 τie i (1− τ)(T − s)P(T )e −i (T − s)ds = τ(1 − e −iT )P(T ) the rationale being that by effectively excluding unexpected gains from tax, this eliminates any tax-induced distortion of realization decisions. One implication of the scheme that may prove hard to sell politically, however, is that investors may have a significant tax lia- bility even on assets on which they have made an ex post loss. Finally, Bradford (1995) proposes a mimicry system that also has the feature of leaving realization decisions undistorted but, unlike Auerbach’s, brings into tax any gain or loss relative to the risk-free re- turn. This works by specifying, when an asset is acquired, a “capital gains tax date” (D) at which any gain is deemed to acquire, the work- ing assumption being that the pre-tax risk-free return is earned at all other times. This gain is brought into tax at a pre-specified tax rate (θ), and tax at the “regular” rate (τ) charged to offset the gain implied by the ability to earn the pre- rather than post-tax return while hold- ing the asset. This implies a total tax charge at realization30 of:31 −iτ(T − D) (2.16) TB = (1 − (1 − θ)e )P(T ) − e −i (1− τ)T (1 − (1 − θ)e iτD )P(0) Though seemingly complex, the intuition behind the scheme is straightforward: by pre-specifying the date at which gains or losses are deemed to arise and taxing away the potential gains from de- ferring realization relative to that date, the realization decision is left undistorted. And the tax θ, bearing only on unexpected gains or losses, is also nondistorting. All these schemes are evidently rather complex. They are not im- practicable. Italy, for example, has over the last few years used a variant of the Vickrey scheme (as discussed by Alworth, Arachi, and Hamaui 2002). But the recent elimination of the Italian scheme highlights the continuation of unresolved issues in this difficult area of tax policy. THEORETICAL PERSPECTIVES 59 The Optimal Taxation of Financial Services Financial companies sell financial services. These can take many forms: the service of intermediating between borrowers and lenders, the risk-pooling service of providing insurance, advice on financial arrangements, and so on. To the company, the sale of these services is a source of income to which the general tax principles discussed in the previous section apply. For the purchaser, they are purchases of a commodity fundamentally like any other. The question then arises as to how such purchases should be treated for tax purposes. A complicating feature is that the price paid for financial services may sometimes not be immediately evident. The price paid for in- surance, for example, is not the premium itself but the premium less the expected payout. In relation to intermediation, the total price paid by both parties is represented by the difference between bor- rowing and lending rates. Allocating that price between the two parties is less straightforward, the standard approach being to con- ceive of these as defined relative to a hypothetical “pure” interest rate. If a borrower pays 15 percent and the lender receives 10 per- cent, the total price paid by the two of 5 percent might be allocated relative to a pure interest rate of 13 percent, so that the borrower pays 2 percent and the lender 3 percent. Thus the analysis of the preceding section can be thought of as applying to transactions at the pure rate of interest, ignoring the whole question of the associ- ated purchase of financial services. Leaving aside these difficulties, the first fundamental question is conceptually clear: how should the purchase of financial services be incorporated into the wider indi- rect tax system? The second question is more closely related to the problems of identifying the price paid for financial services: how, in practice, can such services be taxed? Should Financial Services Be Taxed? A fundamental guiding principle of tax policy design is that tax should be levied on final consumption, not on intermediate trans- actions. The intellectual underpinning of this prescription is the pro- duction efficiency theorem of Diamond and Mirrlees (1971). If pure profits can be taxed at any rate, and if there are no restrictions on the distorting tax instruments that can be deployed, then a Pareto- efficient tax structure has the feature that intermediate transactions are not distorted. The intuition behind this result—which has proved immensely powerful in addressing a wide range of issues—is straight- forward. By definition, a tax reform that eliminates a production in- 60 ROBIN BOADWAY AND MICHAEL KEEN efficiency increases aggregate output, and so must be desirable so long as the government has enough tax instruments at its disposal to ensure that this increased output is shared (in a weak sense) across all consumers. The question then is how this result is to be applied to financial services. For purchases by businesses, the answer is straightforward: assuming the other conditions of the Diamond-Mirrlees theorem to be satisfied—a point discussed further below—these should be un- taxed. The proper treatment of purchases by consumers has proven more contentious.32 Purchases by consumers The proper tax treatment of purchases of financial services by consumers turns out to be a somewhat subtle issue, with the literature pervaded by one fallacy and one half (at best) truth. The fallacy is the argument that since financial services them- selves are obviously not an object of final utility to consumers but merely a form of intermediate transaction to serve more fundamen- tal objects of desire, they should be untaxed. Grubert and Mackie (1999) begin from such a premise, for example, noting that their ar- gument “starts from the view that consumption goods directly yield the consumer utility. They are the arguments of his utility function [whereas] many financial services . . . seem to us unlikely to yield him utility” (p. 25). Chia and Whalley (1999) are also sometimes cited in support of this line, though they are careful not to make quite such an unqualified argument.33 The half-truth, however, is the view—most widely found in the strand of the literature most concerned with the practical issues of how one might tax financial services, and the positive effects of doing so—that financial services are a commodity like any other, and so should be taxed like any other. To see that the first view—financial services should be untaxed because they do not yield utility—is a fallacy, it is enough to pursue its logic still further. For there are many other items that are com- monly taxed, or recommended to be so, even though they too are clearly not final objects of utility (except perhaps for a few strange people). Pet food, cutlery, cars, and many other such items are clearly, for most, means to an end—well being of their pets, getting food into their mouths efficiently, transport—rather than ends in themselves. But if these too are intermediate transactions, this argu- ment would imply that they too should be exempt. In the limit, in- deed, one could argue—along the lines, for instance, of the notion of capabilities of Sen (1985)—that essentially all commodity pur- THEORETICAL PERSPECTIVES 61 chases are merely inputs to some deeper sense of well being. But it would clearly be nonsense to conclude from this that no commod- ity purchases by consumers should ever be taxed. Indeed the view that the final objects of utility cannot in all cases be observed and taxed is the starting point for classical optimal tax theory, which is built upon the problems posed by the unobserv- ability of potential earnings capacity and, hence, of leisure. Optimal tax theory does not generally suppose that the final objects of a household’s utility are observed but only its net market trades, and it is the taxation of these trades that is the concern of tax policy de- sign. At risk of belaboring the point, suppose, for example, that a household has utility u(z) defined over some objects z that are un- observable in themselves but produced from observed net market trades x according to some production relation f(x). Then individu- als behave as if they maximized a utility function u*(x), where u* is the composition u[f(.)], and standard tax theory—including the Diamond-Mirrlees production efficiency theorem—applies with utility so characterized: the key attributes for tax design and reform in this framework, recall, being the patterns of demands for the ob- servable net trades x.34 Thus the first question to ask is not whether financial services ap- pear in the utility function—any argument that turns on this is in- herently untrustworthy as a guide to optimal tax policy—but how they enter into the pattern of observable net trades. Consider then the example of a consumer who lives for two pe- riods, with consumption C1 in the first and C2 in the second (corre- sponding prices being P1 and P2), with income of Y (only) in the first. Imagine too that the only way to transfer income between the two periods is to obtain a bank account at a fixed cost PF in order to lend money at the world interest rate R, with the bank also charging for its intermediation services by reducing the return to the lender by charging a spread of PS . Savings are then (2.17) S = Y − P1C1 − PF and second-period consumption is (2.18) P2C2 = (1 + R − Ps )S Combining the two to eliminate S, the lifetime budget constraint can be written as P2C2 (2.19) P1C1 + PF + = Y. 1 + R − Ps 62 ROBIN BOADWAY AND MICHAEL KEEN This has an important implication, stressed by Jack (2000) and Auerbach and Gordon (2002). Imposing a uniform tax at rate τ on consumption in both periods and the fee for opening the account— but, crucially, leaving that “spread” charge PS untaxed—the budget constraint becomes (1 + τ)P2C2 (2.20) (1 + τ)P1C1 + (1 + τ)PF + = Y, 1 + R − Ps which, dividing both sides by (1+ τ), is equivalent to the budget constraint P2C2 (2.21) P1C1 + PF + = (1 − τ*)Y 1 + R − Ps where τ* = τ/(1 + τ). Thus a uniform tax on consumption and the bank fee, but not on the spread charge, is equivalent to a simple tax on income. In particular, focusing for the moment on the simple case in which Y is exogenous, such a tax structure is equivalent to a lump-sum tax, and hence fully nondistorting. Taxing consump- tion, the bank fee and the spread, however, it is easily seen that the tax would not be lump sum. This provides a simple benchmark for policy. Financial services charged for as a fixed fee should be taxed; those charged for as a spread should not be. (This is one sense in which it is a half-truth to think of financial services as being like any other commodity.) Ap- pealingly enough, this benchmark is not far from current normal practice under the value-added tax (VAT), which is to fully tax fee- type services but exempt (though not zero-rate) implicit charges. The rationale is quite different, however: above it emerges as a mat- ter of principle; in practice it is seen as a matter of administrative practicability. It should also be stressed, given the remarks above, that these ar- guments have nothing to do with the structures of the consumer’s utility function or preference map. It may be natural to think of the consumer having a standard utility function U(C1 ,C2 ) defined only on the consumption good. But that is immaterial: the arguments above apply even if the consumer happens to enjoy the act of open- ing a bank account, or even has preferences defined over the spread they have the misfortune to pay.35 These results do, however, need to be interpreted with some care. First, it is natural to think of the result in terms of additivity or oth- erwise of the budget constraint in items related to financial services, THEORETICAL PERSPECTIVES 63 and of leaving the relative prices of consumption items unchanged— the key feature of the mechanics above being that PF enters the bud- get constraint (equation 2.19) additively while PS does not. Notice, however, that one can also combine equations 2.17 and 2.18 to write the lifetime budget constraint as P2C2 PS S (2.22) P1C1 + PF + + = Y. 1+ R 1+ R The budget constraint can thus be written in such a way that an ex- plicit expenditure item PS S does enter in the “usual” additive way, with the consumer neatly shown to spend their income on four items: the two consumption goods, the bank fee, and services embodied in the spread. With the present value price of P2 interpreted as defined relative to the “world” interest rate rather than the lending rate R–PS , the same argument as above might now seem to lead to the quite different conclusion that a uniform tax that included the spread charge would be lump sum. The reason this is incorrect is because the consumer’s optimization problem—whatever their preferences hap- pen to be—must take into account not only the budget constraint (equation 2.22) but also equation 2.17, the “production function” for savings; and in equation 2.17 the imposition of a uniform tax on all four items would not be equivalent to a tax on Y. Thus the issue is not quite whether the budget constraint is additive in expenditure on financial services, or even, arguably, whether it affects the relative price of consumption items. The issue, rather, is whether the budget constraint is additive when full account has been taken of the links between net trades that those services imply. It is in this sense that the nature of financial services as an intermediate good (and, potentially, of many services more generally)—not in terms of their utility rele- vance—makes them different from other items. A second and related point is that things might not be as simple as this illustrative example suggests. It may be that neither fees PS nor the spread PF are fixed. For example, it could be that the spread falls with the level of savings. In this case large spreads on infra- marginal savings have the character of fixed fees and ought to be taxed. By the same token, bank fees PF may vary with the level of savings, in which case—by the previous arguments—they are like a spread and should be taxed. The implication is that distinguishing between fixed fees and spreads related to saving volumes poses dif- ficult administrative challenges. The third point to stress is that the results above, from manipu- lating budget identities, only establish a benchmark of a kind of 64 ROBIN BOADWAY AND MICHAEL KEEN neutrality—rendering an indirect tax system equivalent to a simple labor income tax; they do not establish any optimality properties. If Y is fixed, and also observable, then a lump-sum tax could be im- posed directly. The more interesting case for tax design—the start- ing point of optimal tax theory—is that in which income is itself a choice variable, reflecting intensity of work effort. As is well known, there is then no presumption that a uniform tax on con- sumption would be optimal. For that problem of tax design, the de- sirability of a tax on PS cannot be ruled out a priori as part of a tax structure that maximizes welfare subject to some government rev- enue constraint.36 The example cited by Chia and Whalley (1999) is instructive in this context. They consider two individuals who acquire identical cars, one purchasing by cash and the other by leasing. Since they each possess the same car, and it is the car that enters their utility func- tions, Chia and Whalley argue, they should pay the same tax. But their economic situations could be quite different. The individual who purchased in cash may have had to spend a morning collecting the money from the bank, while the individual who is leasing may face binding credit constraints that adversely affect other aspects of their situation, such as the ability to acquire education. These under- lying differences between the two, unless dealt with directly by other instruments of public policy, may rationalize treating the two ways of acquiring car services quite differently for tax purposes. Accepting then that financial services purchased by consumers are in some circumstances a proper subject for taxation, the ques- tion then is: At what rate? The answer is likely to be context-specific, depending not least on the range of other instruments available to government. In a general sense, the appropriate rate will be governed by the same range of con- siderations that apply to the determination of optimal tax rates more generally: the extent to which the use of financial services serves as a substitute for leisure, and the degree to which taxing financial services eases the self-selection problem by making it less attractive for high- ability types to masquerade as low-ability types.37 In the example above, for instance, to the extent that payment in cash requires using up time that could otherwise be spent in the market, relatively favor- able tax treatment of leasing would be called for. There is, perhaps, a general presumption that insofar as financial intermediation services serve as an alternative to time-consuming activities they should be taxed, if at all, at a relatively low rate. The issues here are clearly subtle ones, but the broad conclusion seems clear: the fact that financial services are not in themselves an THEORETICAL PERSPECTIVES 65 object of utility is in itself of no relevance, and certainly does not imply that their purchase by consumers should not be taxed. While their appropriate treatment is likely to be context-specific, there are reasonable grounds to suppose that the optimal rate of taxation for such services will often be relatively low. Purchases by business The appropriate treatment, in principle, of business sales and purchases of financial services is less contentious, there being widespread agreement that these—like all other business inputs—should ideally be untaxed. The most compelling rationale for this is again the Diamond-Mirrlees production efficiency theorem. That theorem does, however, admit exceptions. It fails to apply, in general, if there are restrictions on the government’s ability to tax pure profits or if it is in some way constrained in its ability to de- ploy distorting domestic taxes. In either case, the taxation of inter- mediate transactions might serve some role as a surrogate for these missing tax instruments. Might either of these considerations have any special force in relation to financial services? Certainly the financial sector is sometimes popularly supposed to generate excessive profits and/or excessive payments to those work- ing in the sector, and this has been used to garner support for taxes on financial activities. But if it is felt necessary to levy special taxes on these agents, this can often be better done by means of more di- rectly targeted profit or income taxes. The United Kingdom, for instance, levied a windfall profits tax on banks in the early 1980s, while Canada and other countries have annual capital taxes on fi- nancial institutions. There may be cases, however, in which taxing businesses’ finan- cial transactions can serve as a partial substitute for missing tax in- struments, the most obvious being those in which this can serve as a partial defense against tax evasion. It may be easier to monitor transactions through the banking system than it is to enforce proper payment of taxes on income or wealth, so that some tax on those transactions—perhaps creditable against taxes on income or wealth— may serve as a poor but worthwhile surrogate. For whatever reason they are imposed, taxes on business use of financial services are liable to generate production inefficiency by distorting input choices away from the use of those services. Firms may take out less insurance, or undertake less investment, than is appropriate in terms of underlying social costs. The consequent dis- tortion of such firm’s output prices then cascades through the pro- duction system insofar as those outputs become other firms’ inputs. Quite how significant these deadweight costs are is a matter for em- 66 ROBIN BOADWAY AND MICHAEL KEEN pirical study. Both experience and common sense, however, suggest that businesses are likely to prove adept in ways of restructuring their financial operations to avoid such taxes. With this high re- sponsiveness of the base, the presumption must be that the dead- weight cost (per dollar of revenue raised) of taxes on the business use of financial services is likely often to be high. Can Financial Services Be Taxed? Accepting then the general principles that financial services pur- chased by consumers ought to be taxed (albeit perhaps at a low rate) while business purchases should not be, the question arises: How can this be done in practice? In particular, how should financial services be incorporated into the VAT, now the most common form of indi- rect tax even in developing economies?38 These issues are discussed in chapter 12, but an overview here may also be helpful. For financial services provided on a fee-paying basis, such as safekeeping services and financial advice, there is no particular dif- ficulty: VAT (or any other form of sales tax) can be charged in the usual way. The difficulty arises for services charged for in the margin be- tween the return paid to lenders and that charged to borrowers.39 The problem is typically not in identifying the aggregate value-added created by intermediation, but in allocating it between consumers (to be taxed) and producers (not to be taxed). This is difficult. Suppose, for example, that a bank pays its depositors 5 percent and charges its borrowers 15 percent. Clearly the value-added by the bank is 15 – 5 = 10 percent of deposits (less any material inputs, and assuming too there is no risk of default). This should be taxed. If all loans were to final consumers, this would be the end of the matter. Assume instead that the borrower is a registered firm. How much of the 15 percent should be creditable? The standard concep- tual approach on this issue has been to imagine a hypothetical “pure” interest rate at which the lender could have lent (but with- out enjoying the ancillary services such as clearing offered by the bank) and at which the borrower could have borrowed (had they been able to find suitable lenders without the help of the intermedi- ary). If this pure rate is 12 percent, for instance, then the value- added provided to the borrower is 15 – 12 = 3 percent of the loan, and the remaining 12 – 5 = 7 percent is value-added provided to the lender. On a loan of $1,000 and at a VAT rate of 10 percent, the ap- propriate outcome is thus for the borrower to be charged VAT of $3 and the lender VAT of $7, the total payable thus being 10 percent THEORETICAL PERSPECTIVES 67 of the aggregate value-added on 10 percent of $1,000. These VAT payments would be creditable, in the usual way, if lender or bor- rower is registered. But this outcome has generally been regarded as impossible to bring about, since it would appear to require, in particular, identi- fying a “pure” interest rate. This has led most countries using the invoice-credit VAT (which means almost all countries with a VAT) to “exempt” financial intermediation, meaning that tax is not charged on outputs of the financial services sector but nor can tax paid on inputs be reclaimed. Both businesses and consumers thus bear some tax on their purchases of the financial sector, in the indirect form of unrecovered tax paid on the inputs—office equipment, software— used in their production. Some countries (notably Israel) have instead taxed value-added in financial services by the addition method: that is, by levying tax directly on the sum of wages and profits.40 A potential alternative is afforded by the subtraction method (as was at one point proposed in Canada), under which tax is simply levied on the excess of a firm’s outputs over its inputs. Either method is capable of taxing aggregate value-added in this sector, and indeed either would in principle be perfectly adequate within a wider VAT system based comprehensively on the addition or subtraction method. Neither method, however, sits well with the application of the invoice-credit method in the rest of the VAT system. Neither enables the identifi- cation of embodied VAT on a transaction-by-transaction basis, and hence neither allows the systematic crediting of financial services provided to registered traders. In principle, these difficulties can be circumvented by applying VAT on a “cash flow” basis.41 Under this system, all inflows of funds, including the receipt of a loan, and all interest receipts, would be treated akin to sales, and be taxable if the recipient is registered. All outflows, including the repayment of loans, or payment of interest, would attract credit if the payer is registered. For example, consider a loan of $1,000 to a registered trader (at 15 percent) financed by a deposit (paid 5 percent) from a consumer. Assume as before that the tax rate is 10 percent. Under the cash flow VAT: • The bank is liable to pay $100 on the deposit, but this is ex- actly offset by a credit of $100 on the loan itself. When the loan is repaid, the bank has a net inflow of 10 percent of $1,000 (its spread on the loan) and so owes tax of $10. • When the loan is made, the business pays tax of $100. When it is repaid, it receives a credit of 10 percent of $1,150 (principal plus interest). 68 ROBIN BOADWAY AND MICHAEL KEEN If the government is able to earn the pure rate of interest of 12 percent on its receipt from the business of $100, it is left with net revenue of $112 + 10 – 115 = $7, which is exactly 10 percent of the services of $70 provided to the consumer. It may be useful to make the general argument algebraically. De- note the amount of the loan by L, the borrowing and lending rates by rB and rL respectively, and the tax rate by τ. Also define λB and λL to take the values 1 and 0 if the borrower (respectively, lender) is registered or not; thus the example above has λB = 1 and λL = 0. In the first period of the loan (assumed for simplicity to last only two periods), the net liability of the bank is zero; that of a borrower is λBτL; and the lender receives a credit of λLτL. In period 2, the net li- ability of the bank is τ(rB – rL)L; a registered borrower is due a credit of τ(1 + rB)L, and a registered lender is liable for tax of τ(1 + rL)L. Assuming that the government obtains a rate of return p on its net re- ceipts in the first period of λBτL – λLτL, net revenue in period 2 is: (1 + p)(λB – λL)τL + τ(rB – rL)L – λBτ(1 + rB)L + λLτ(1 – rL)L (2.23) = τL[(1 – λB)(rB – p) + (1 – λL)(p – rL)] Thus tax is ultimately collected only on that part of the margin (rel- ative to the government’s discount rate) that reflects the value-added enjoyed by nonregistered traders. So long as the interest rate available to the government is identi- fied with the pure interest rate, the cash flow approach achieves the theoretically correct allocation of value-added, and allows a proper crediting of input tax. Intuitively, by crediting inflows and outflows at the same rate, the system ensures that the present value of the rev- enue raised from registered traders is zero. Revenue is ultimately raised only to the extent of that part of the margin that falls on un- registered traders. The treatment of pure insurance—insurance with no savings ele- ment—is also straightforward under the cash flow VAT. An exam- ple is given in box 2.1. Outside the case of pure insurance, however, the cash flow scheme is cumbersome administratively. Some measures can be taken to al- leviate this (for example, by suspending the payment of tax, and re- funds, associated with the initial receipt of loans and deposits). But even within the European Union, where the scheme has been closely considered, there are doubts as to its practicability. For developing countries, it seems likely to remain overly complex for some time yet. It should be emphasized that it is by no means certain that mak- ing financial services fully taxable will generate an increase in VAT THEORETICAL PERSPECTIVES 69 Box 2.1 Treatment of Pure Insurance under a Cash Flow VAT Consider the case in which an insurance company receives premiums of $100 and pays $80 (exclusive of VAT) in claims, so that value added is $20. The tax rate is 20 percent (this being the tax-exclusive rate: that is, the rate charged on values not including tax). The insurance company is liable to VAT of $20 on its premiums, but allowed a credit of $16 in respect of the claims it pays. Thus the insurer pays net tax of $4, which is 20 percent of the value-added of $20. The credit enables the insurer to send the insured a check for $96 should the insured event occur. If the insured is a consumer, $96 is just enough to purchase goods to the tax-exclusive value of $80: the credit included in the claim pays the VAT on the replacement goods bought. Total tax collected is thus exactly 20 percent of value-added. If the insured is registered for VAT, they take a credit of $20 on the premium. When the claim of $96 is received, output tax of $16 is charged, which is exactly offset by an input tax credit of $16 on the replacement property. Total tax collected by the government is zero. Consistent with the logic of the invoice-credit VAT, the tax thus “sticks” only on final sales to final consumers. revenues. If financial services were fully taxable, revenue would be collected only on sales to final consumers. When they are exempt, in contrast, tax is collected on inputs into the sector.42 Which will lead to greater revenue is an empirical question. For the EU, Huizinga (2002) sees a substantial potential revenue gain. For many developing countries, however, purchases of financial services by domestic consumers may be quite limited, so that subjecting finan- cial services to VAT is unlikely to be the fiscal panacea for such cash-strapped governments that it may sound. Concluding Comments Financial activities are ubiquitous in a market economy, and the policy issues pervasive. This chapter has necessarily focused on some key issues concerning the optimal taxation of capital income generally, and financial services more specifically. The perspective taken in this chapter has been the taxation of financial activities as part of a system of optimal taxation for revenue-raising purposes. 70 ROBIN BOADWAY AND MICHAEL KEEN In fact, the tax (and subsidy) system might also be used for correc- tive purposes—to correct for inefficiencies resulting from market failures. The possibility of market failure is particularly germane in financial markets because of the information asymmetries that are endemic to the sector. Understanding the proper role of tax policy in such circumstances remains under-developed. Needless to say, this chapter has not been able to cover all aspects of the taxation of financial activities. Some of these issues, notably several distinct policy problems raised by the distinct functions of fi- nancial institutions, are at the interface between tax and regulatory policies and are taken up in subsequent chapters; for instance the appropriate tax provision for bad loans in the case of banks, the imposition of reserve requirements, and the provision of deposit in- surance (see in particular chapters 5, 7, and 9). Money is itself a key financial asset, so that seigniorage will typically have a role to play in the wider tax system (see, for instance, chapter 13 in this volume, and Poterba and Rotemberg 1990). Some of the other more important omissions from this coverage— other than the key issue of market failure just mentioned—bear spe- cial emphasis: Financial markets are sometimes seen as being excessively vola- tile, presumably—the argument is not always well articulated—in the sense that they adjust more rapidly than other markets, creating adjustment problems. This then leads some to advocate taxes on fi- nancial transactions as a corrective device. Even in its own terms, however, the argument is not convincing. By thinning the market, such taxes could well increase volatility (Hubbard 1993). An issue with wider ramifications than finance is the impact of taxation on risk-taking, which has received considerable attention (see for instance Sandmo 1985). The associated issues of optimal policy design, however, have been less studied, and are not ad- dressed here: whenever there is uncertainty in the models developed here and in chapter three, there will be no aggregate uncertainty and all agents will be risk-neutral. In general terms, of course, when risk markets are incomplete there may be scope for welfare-improving interventions. This will depend in large part, however, on whether governments are better able to bear risk than is the private sector, which is far from obviously being the case. Additionally, the recent literature on corporate governance (sur- veyed by Becht, Bolton, and Röell 2003) points to inefficiencies (in re- lation to takeovers, for instance) that tax policy might be used to ad- dress. These, however, appear to have received almost no attention. THEORETICAL PERSPECTIVES 71 Many countries, especially it would seem developing ones, adopt tax and other measures intended to nurture the financial sector. Though such policies are often criticized as misplaced, the question of appropriate tax design in the presence of under-developed finan- cial markets seems to have received little attention. Finally, international considerations raise issues of increasing im- portance, currently heated aspects including the desirability or oth- erwise of restricting international tax competition. These issues have generated a large literature, still growing rapidly, that is simply too large to be reviewed here. These omissions are daunting. Understanding the issues that are addressed here, however, seems fundamental to addressing this wider set of policy problems. Appendix Suppose that the government’s instruments are taxes on wage in- come in the two periods and a capital income tax. Maximizing U(C1,C2,L1,L2) = U(C1,L1) + βU(C2,L2) subject to the budget con- straint (equation 2.2), written more compactly as: (A2.1) *C = w *L + w *L C1 + p2 2 1 1 2 2 (where the price of C has been normalized to unity, and asterisks are used to indicate tax-inclusive prices), the familiar necessary condi- tions for the consumer’s optimization problem are: (A2.2) 1 = λ UC (A2.3) 2 = λp * βUC 2 (A2.4) 1 = –λw * UL 1 (A2.5) 2 = –λw * βUL 2 The government’s present value revenue constraint is then, in obvi- ous notation, (A2.6) 2C + T 1 L + T 2 L = R TC 2 w 1 w 2 where taxes are taken to be defined in specific form: in particular, 2 = p – p * , so that capital income taxation is zero iff T C = 0. TC 2 2 2 72 ROBIN BOADWAY AND MICHAEL KEEN The optimal tax problem, in primal form,43 is then addressed by forming the Lagrangean L = U(C1,L1) + βU(C2,L2) + α(w1L1 + w2L2 – C1 (A2.7) 1C + βU 2C + U 1L + βU 2L ) – p2C2 – R) + µ(UC 1 C 2 L 1 L 2 where the second term (with multiplier α) captures the govern- ment’s revenue constraint, obtained from (A2.1) and (A2.3), and the third (with multiplier µ) incorporates the incentive compatibility constraint by combining the consumer’s necessary conditions and budget constraint. The first-order conditions are then 1 1 1 1 (A2.8) UC – α + µ(UC + UCC C1 + ULCL1) = 0 2 2 2 2 (A2.9) βUC – αp2 + µβ(UC + UCC C2 + U LCL2) = 0 1 1 1 1 (A2.10) UL + αw1 + µ(U L + UCL C1 + ULLL1) = 0 2 2 2 2 (A2.11) βUL + αw2 + µ(UL + UCL C2 + ULLL2) = 0 To establish the claims in the text, suppose that C1 = C2 and L1 = 1 2 L2 at an optimum. Then from (A2.8, 9), UC = βUC /p2. Hence from * C 1 2 (A2.2, 3), p2 = p2 , implying T 2 = 0. Moreover, since UC = UC (by time-invariance of C and L), it must be that β = p2 = 1/(1 + R). Then from (A2.10, 11) w1 * /w = w * /w . 1 2 2 Notes 1. This can be seen by noting that equation 2.1 is equivalent to equa- tion 2.2, with τr, τw1, and τw2 in the latter replaced by 1  (1 + (1 − τ r )r)(1 + τ c1)  τ∗ r = 1−  − 1 , τ r 1 + τc 2  ∗ 1 − τ w1 1 − τw2 w1 = 1− , and τ∗ w2 = 1 − . 1 + τ c1 1 + τc 2 2. This is essentially a generalization of that of King (1980) and Atkin- son and Sandmo (1980) for the special case of a two-period life cycle with labor supply only in the first period. These, in turn, are essentially an ap- plication of Corlett and Hague (1953): if leisure is equally substitutable for first and second period consumption, and if the government is free to issue THEORETICAL PERSPECTIVES 73 debt as well as tax labor and capital income, there is no role for capital in- come taxation. 3. See Deaton (1979). Besley and Jewitt (1995) derive a more general sufficient condition on preferences for uniform taxation to be optimal. 4. More precisely, the first-order conditions characterizing an optimal tax system are satisfied along a growth path of the type just described. The claims in this paragraph are proved in the appendix. 5. There is a slight abuse of notation here, in that these tax rates were specified in ad valorem form above rather than, as here, specific form. The two are equivalent from the perspective of the consumer, so that which one uses is essentially a matter of modeling convenience. 6. See also Atkeson, Chari, and Kehoe (1999); Erosa and Gervais (2001, 2002). 7. This can be seen by noting that lifetime preferences are in this case of the form 1/ σ   ∑ 2 ⋅ = u(φ(C1 ,C2 ), L1 , L2 ), where φ() U() ⋅ = Cσ   t =1 t  is homogenous of degree one. This form satisfies the sufficient condition for implicit separability mentioned in the text above. 8. Coleman also notes that if the government is unrestricted in its choice of tax instruments, and assuming that the representative household has positive initial assets, a tax scheme that taxes consumption proportionately and subsidizes labor proportionately at the same tax rate in all periods is effectively a tax on initial wealth, and so is lump-sum. This can be seen by adding a term representing initial assets to the left of equation 2.1. Assum- ing initial asset wealth is large enough to pay for future tax requirements, a first best is achieved and the optimal tax on capital income would be zero, which would be distortionary. This is essentially the time consistency point in another guise. 9. For the case in which preferences are not separable over time, Erosa and Gervais (2002) show that it will be optimal not to tax capital income from period 1 onward if preferences are of the form U(G(C1,C2 ),L1,L2 ), with G homothetic. This of course is the sufficient condition for implicit separability of the expenditure function mentioned in the third observation of the two-period case. 10. See Edwards, Keen, and Tuomala (1994); Nava, Schroyen, and Marchand (1996). 11. This discussion presumes that interest income is taxed at a constant rate, which would be appropriate if individual capital income could not be observed. If it can be, presumably one might want interest income in the second period to be taxed in a nonlinear fashion. 12. This has long been the case in Sweden, for instance; in the United States, see Gordon and Slemrod (1988). 74 ROBIN BOADWAY AND MICHAEL KEEN 13. Nevertheless, the task is not hopeless: Norway attempts to do ex- actly this. 14. This has not been standard practice but, following the German tax reform of 2000, appears to be becoming more common. 15. One could conceive of international cooperation to this effect—each country uses its corporation tax as a withholding device against other’s per- sonal income tax system—but this is not observed in practice. 16. Gordon (1992), for instance, argues that a capital exporting coun- try acting as a Stackleberg leader (and as such, taking account of the likely responses of the others) may find it desirable to offer a credit in order to in- duce a higher tax abroad and thereby stem capital flight. 17. See Fuest and Huber (2000) for an argument that such a system is indeed optimal in an open economy setting. 18. With a falling rate of cash flow taxation, for instance, there is an in- centive to bring forward investment in order to take the up-front deduction at a higher rate than the later returns will be taxed. The point is neatly made by Sandmo (1979). 19. Bond and Devereux (1999) argue for the use of the risk-free rate of return. 20. Keen and King (2002) assess the Croatian experience. 21. In fact, in the case of natural resources, there are alternatives to rent taxes as a means of ensuring that some share of resource rents accrue to the public sector, including royalty systems, the sale of leases, and joint public- private ventures (see Boadway and Flatters 1993). 22. In principle, these net liability transactions could also be included in the cash flow tax base of non-financial corporations, as well. As shown by Boadway, Bruce, and Mintz (1983), the neutrality properties of alternative cash flow tax bases varies depending on what one assumes about the de- terminants of the financial structure of the firm. 23. For more extensive accounts, see Alworth (1998), Edgar (2000), and Warren (1993). 24. A put (call) option is the right, but not obligation, to sell an asset at some pre-specified exercise (or “strike”) price. The example assumes Euro- pean options with expiry date coincident with that on the zero-coupon bond. 25. By, for instance, the adoption of a cash flow or ACE-type corporate tax along the lines discussed above. 26. Denote the price in year 5 by S*, the net worth of the investor is then S* – S + max[S–S*,0] – max[S*–S,0] (this being the value of the share less the obligation on the debt plus the payoff on the put less the cost of the call), which is exactly zero. 27. Under a straight interest rate swap—the exchange of fixed and floating rate obligations—each party makes a net payment equal to the ex- THEORETICAL PERSPECTIVES 75 cess of one interest charge over another. If withholding is levied only on the first component of this net payment, tax payments could exceed the under- lying gain associated with the net flows. For an example of this, see appen- dix II of OECD (1994). 28. OECD (1994). 29. These are discussed in Thuronyi (2001). 30. Auerbach and Bradford (2001) note that this scheme could also be implemented as a generalized cash flow tax, with a deduction for asset ac- quisitions at a tax rate corresponding to the coefficient on P(0) in equation 2.16 and full taxation of proceeds at the rate corresponding to that on P(T). 31. As explained in Auerbach and Bradford (2001), this is the sum of two charges. The first is that on the “deemed” gain at D, cumulated to time T, given the working assumption that the asset earned the pre-tax risk-free rate at all other dates: g(P(T)e – i (T–D) – P(0)e iD)e i(1–τ)(T–D) The second is the cumulated value of the tax value of accumulating at the pre-tax rather than the post-tax interest rate: P(0)(e iD – e i (1–τ)D)e i(1–τ)(T–D) + P(T)e – i (T–D)(e i (T–D) – e i(1–τ)(T–D)). 32. For brevity—there being enough issues to deal with—this chapter leaves aside the considerable practical difficulties that arise in distinguish- ing between purchases of financial services for personal and business use, a much wider problem that affects a whole range of commodities. 33. Chia and Whalley (1999) argue that “No utility per se flows from the use of intermediation services. Since financial services do not enter pref- erences directly, it may be inappropriate to include them in broadly based indirect taxes such as a VAT” (p. 705). Note, however, the careful and quite correct use of the word ‘may,’ and the equally careful and correct implicit recognition that it may be optimal to tax financial services at a rate other than prevailing on commodities in general. Indeed they are quite explicit in recognizing that the optimal rate of tax on financial service may not be zero (p. 718). 34. The composition u* may not have the regularity properties usually required of a utility function. However, one may follow the tradition of op- timal tax theory in being somewhat cavalier in the treatment of second- order conditions. 35. Conversely, suppose that one knew for sure that preferences were of the form U(C1 ,C2 ). Then this example shows that even if one were unre- stricted in the ability to tax what appears in the utility function, it may be desirable to tax other items too (the bank fee). 76 ROBIN BOADWAY AND MICHAEL KEEN 36. It is worth noting that the simulation results reported in Chia and Whalley (1999) do not show the optimal rate of tax on intermediation ser- vices to be zero. They merely show that a situation in which consumption in each period and those services are taxed at a uniform rate may be infe- rior to one in which only consumption is taxed at a uniform rate. As they themselves are at pains to emphasize, this does not imply that the full opti- mum is characterized by nontaxation of financial services. 37. These are the same considerations as arose above in the context of taxing capital income with heterogeneous individuals. 38. Recall that the basic principle of the usual invoice-credit VAT is that all inputs are taxed but with businesses able to credit this input tax against the tax charged on their outputs. See Ebrill and others (2001) for a recent treatment; this section draws on chapter 7. 39. The discussion here focuses mainly on simple loan transactions. Similar issues apply to insurance contracts with a savings component and to other more complex forms of financial intermediation. 40. Profits for this purpose should in principle be defined on a cash flow basis, with investment immediately expensed. 41. See Poddar and English (1997) and European Commission (undated). 42. This is a simplification: revenue will be less than this to the extent that financial services are exported (hence zero-rated) and greater to the ex- tent that a higher price of financial services leads to higher prices of taxed commodities produced with their help. 43. See, for instance, Atkinson and Stiglitz (1980, pp. 376–80); Deaton (1979). References Alvarez, Yvette, John Burbidge, Ted Farrell, and Leigh Palmer. 1992. “Op- timal Taxation in a Life Cycle Model.” Canadian Journal of Economics 25 (1): 111–22. Alworth, J. S. 1998. “Taxation and Integrated Financial Markets: The Chal- lenges of Derivatives and Other Financial Innovations.” International Tax and Public Finance 5 (4): 507–34. Alworth, J. S., G. Arachi, and R. Hamaui. 2002. “Adjusting Capital Gains Taxation: Lessons from the Recent Italian Tax Experience.” University of Bocconi. Processed. Atkeson Andrew, V. V. Chari, and Patrick J. Kehoe. 1999. “Taxing Capital Income: A Bad Idea.” Federal Reserve Bank of Minneapolis Quarterly Review 23 (1): 3–17. Atkinson, Anthony B., and Agnar Sandmo. 1980. “Welfare Implications of the Taxation of Savings.” The Economic Journal 90 (359): 529–49. THEORETICAL PERSPECTIVES 77 Atkinson, Anthony B., and Joseph E. Stiglitz. 1976. “The Design of Tax Structure: Direct vs. Indirect Taxation,” Journal of Public Economics 6 (1): 55–75. ———. 1980. Lectures on Public Economics. New York: McGraw-Hill. Auerbach, Alan J. 1983. “Taxation, Corporate Financial Policy and the Cost of Capital,” Journal of Economic Literature 21 (3): 905–40. ———. 1991. “Retrospective Capital Gains Taxation.” American Economic Review 81 (1): 167–78. Auerbach, Alan J., and David F. Bradford. 2001. “Generalized Cash Flow Taxation.” University of California, Berkeley. Processed. Auerbach, Alan J., and Roger Gordon. 2002. “Taxation of Financial Ser- vices Under a VAT.” American Economic Review, Papers and Proceed- ings 92: 411–16. Auerbach, Alan J., and Laurence J. Kotlikoff. 1987. Dynamic Fiscal Policy. Cambridge: Cambridge University Press. Becht, Marco, Patrick Bolton, and Ailsa Röell. 2003. “Corporate Gover- nance and Control.” In George Constantinides, Milton Harris and Rene M. Stulz, eds. Handbook of the Economics of Finance. Volume 1A. New York: North Holland. Besley, Timothy, and Ian Jewitt. 1995. “Uniform Taxation and Consumer Preferences.” Review of Economic Studies 58 (1): 73–84. Boadway, Robin. 1980. “Corporate Taxation and Investment: A Synthe- sis of the Neo-Classical Model.” Canadian Journal of Economics 13 (2): 250–67. Boadway, Robin, and Neil Bruce. 1984. “A General Proposition on the Design of a Neutral Business Tax.” Journal of Public Economics 24 (2): 231–39. ———. 1992. “Problems with Integrating Corporate and Personal Taxes in an Open Economy.” Journal of Public Economics 48 (1): 39–66. Boadway, Robin, and Frank Flatters. 1993. “The Taxation of Natural Re- sources: Principles and Policy Issues.” Policy Research Working Paper WPS No. 1210. World Bank, Washington, D.C. Boadway, Robin, and Michael Keen. 1998. “Evasion and Time Consistency in the Taxation of Capital Income.” International Economic Review 39 (2): 461–76. Boadway, Robin, Neil Bruce, and Jack Mintz. 1983. “On the Neutrality of Flow-of-Funds Corporate Taxation.” Economica 50 (1): 49–61. Boadway, Robin, Maurice Marchand, and Pierre Pestieau. 2000. “Redistri- bution with Unobservable Bequests: A Case for Taxing Capital Income.” Scandinavian Journal of Economics 102 (2): 253–67. Bond, Stephen R., and Michael P. Devereux. 1999. “Generalised R-Based and S-Based Taxes Under Uncertainty.” Working Paper No. 99/9. Insti- tute for Fiscal Studies, London. 78 ROBIN BOADWAY AND MICHAEL KEEN Bradford, D. F. 1995. “Fixing Realization Accounting: Symmetry, Consis- tency and Correctness in the Taxation of Financial Instruments.” Tax Law Review 50 (4): 731–85. Brock, Philip L. 2003. “Corrective Taxes and Quasi-Taxes for Financial In- stitutions and Their Interaction with Deposit Insurance” (chapter 5, this volume). Caminal, Ramon. 2003. “Taxation of Banks: Modeling the Impact.” (chap- ter 3, this volume). Chamley, Christophe. 1986. “Optimal Taxation of Capital Income in Gen- eral Equilibrium with Infinite Lives.” Econometrica 54 (3): 607–22. Chari, V. V., Patrick Kehoe, and Edward C. Prescott. 1989. “Time Consis- tency and Policy.” In Robert Barro, ed., Handbook of Modern Business Cycle Theory. Cambridge, Mass.: Harvard University Press. Chia, N. C., and J. Whalley. 1999. “The Tax Treatment of Financial Inter- mediation.” Journal of Money, Credit and Banking 31 (4): 704–19. Coleman, Wilbur John II. 2000. “Welfare and the Optimum Dynamic Tax- ation of Consumption and Capital Income.” Journal of Public Econom- ics 76 (1): 1–39. Corlett, W. J., and D. C. Hague. 1953. “Complementarity and the Excess Burden of Taxation.” Review of Economic Studies 21 (1): 21–30. Cremer, Helmuth, Pierre Pestieau, and Jean-Charles Rochet. Forthcom- ing, 2003. “Capital and Labor Income Taxation in an Overlapping- Generations Growth Model.” Journal of Public Economics. Deaton, Angus. 1979. “The Distance Function in Consumer Behavior with Applications to Index Numbers and Optimal Taxation.” Review of Economic Studies 46 (3): 391–405. Deaton, Angus, and Nicholas Stern. 1986. “Optimally Uniform Commod- ity Taxes, Taste Differences and Lump-Sum Grants.” Economics Letters 20 (3): 263–66. Devereux, M. P., and H. Freeman. 1991. “A General Neutral Profits Tax.” Fiscal Studies 12 (1): 1–15. Diamond, P., and J. A. Mirrlees. 1971. “Optimal Taxation and Public Pro- duction, I: Production Efficiency.” American Economic Review 61 (1): 8–27. Ebrill, L., M. Keen, J. P. Bodin, and V. Summers. 2001. The Modern VAT. Washington: International Monetary Fund. Edgar, T. 2000. The Income Tax Treatment of Financial Instruments: The- ory and Practice. Toronto: Canadian Tax Foundation. Edwards, Jeremy, Michael Keen, and Matti Tuomala. 1994. “Income Tax, Commodity Taxes and Public Good Provision: A Brief Guide.” Finan- zarchiv 51 (4): 472–97. Erosa, Andrés, and Martin Gervais. 2001. “Optimal Taxation in Infinitely- Lived Agent and Overlapping Generations Models: A Review.” Federal Reserve Bank of Richmond Economic Quarterly 87 (1): 23–44. THEORETICAL PERSPECTIVES 79 ———. 2002. “Optimal Taxation in Life-Cycle Economies.” Journal of Economic Theory 105 (2): 338–69. European Commission. Undated. Value-Added Tax: A Study of the Meth- ods of Taxing Financial Institutions. Brussels: European Commission. Fischer, Stanley. 1980. “Dynamic Inconsistency, Cooperation and the Be- nevolent Dissembling Government.” Journal of Economic Dynamics and Control 2 (1): 93–107. Fuest, Clemens, and Bernd Huber. 2000. “The Optimal Taxation of Divi- dends in a Small Open Economy.” Working Paper Series No. 348. Mu- nich: CESifo Gordon, Roger. 1992. “Can Capital Income Taxes Survive in Open Econ- omies?” Journal of Finance 47 (3): 1159–80. ———. 2001. “Taxation of Interest Income in an Open Economy.” Paper prepared for the Research Center for Economic Policy (OCFEB) Confer- ence on Tax Policy in the European Union, October 17–19. The Hague: Ministry of Finance http://www.few.eur.nl/few/research/ocfeb/congreseu/. Gordon, Roger, and Joel Slemrod. 1988. “Do We Collect any Revenue from Taxing Capital Income?” Tax Policy and the Economy 2: 89–130. Grubert, Harry, and James Mackie. 1999. “Must Financial Services be Taxed under a Consumption Tax?” National Tax Journal 53 (1): 23–40. Honohan, Patrick. 2003. “The Accidental Tax: Inflation and the Financial Sector.” (chapter 13, this volume). Hubbard, Glenn R. 1993. “Securities Transactions Taxes: Tax Design, Rev- enue and Policy Considerations.” Tax Notes (November 23): 985–1000. Huizinga, Harry. 2002. “A European VAT on Financial Services.” Eco- nomic Policy 35: 499–534. Jack, William. 2000. “The Treatment of Financial Services under a Broad- Based Consumption Tax.” National Tax Journal 53 (4): 841–51. Keen, Michael, and John King. 2002. “The Croatian Profits Tax: An ACE in Practice.” Fiscal Studies 23 (3): 401–18. King, Mervyn A. 1977. Public Policy and the Corporation. London: Chap- man and Hall. ———. 1980. “Savings and Taxation.” In C. A. Hughes and G. M. Heal, eds. Public Policy and the Tax System. London: Allen and Unwin. Marceau, Nicolas, and Michael Smart. 2003. “Corporate Lobbying and Commitment Failure in Capital Taxation.” American Economic Review, forthcoming. Meade Committee. 1978. The Structure and Reform of Direct Taxation. London: Allen and Unwin. Mirrlees, James A. 1971. “An Exploration in the Theory of Optimal In- come Taxation.” Review of Economic Studies 38 (2): 175–208. Nava, Mario, Fred Schroyen, and Maurice Marchand. 1996. “Optimal Fis- cal and Public Expenditure Policy in a Two-Class Economy.” Journal of Public Economics 61 (1): 119–37. 80 ROBIN BOADWAY AND MICHAEL KEEN Organisation for Economic Co-operation and Development. 1994. Taxa- tion of New Financial Instruments. Paris: OECD. Poddar, Satya. 2003. “Consumption Taxes: The Role of the Value-Added Tax” (chapter 12, this volume). Poddar, Satya, and Morley English. 1997. “Taxation of Financial Services under a Value-Added Tax.” National Tax Journal 50 (1): 89–111. Poterba, James M., and Julio J. Rotemberg. 1990. “Inflation and Taxation with Optimizing Governments.” Journal of Money, Credit and Banking 22 (1): 1–18. Sandmo, Agnar. 1979. “A Note on the Neutrality of the Cash Flow Cor- porate Tax.” Economic Letters 4: 173–76. ———. 1985. “The Effects of Taxation on Savings and Risk-Taking.” In Alan J. Auerbach and Martin Feldstein, eds. Handbook of Public Eco- nomics, Vol. 1 Amsterdam: North-Holland. Sen, Amartya. 1985. Commodities and Capabilities. Amsterdam: North- Holland. Stiglitz, Joseph. E. 1983. “Some Aspects of the Taxation of Capital Gains.” Journal of Public Economics 21 (2): 257–94. ———. 1987. “Pareto Efficient and Optimal Taxation and the New New Welfare Economics.” In Alan J. Auerbach and Martin Feldstein, eds. Handbook of Public Economics, Vol. 2. Amsterdam: North-Holland. Sunley, Emil. 2003. “Corporate Income Tax Treatment of Loan-Loss Re- serves” (chapter 9, this volume). Thuronyi, V. 2001. “Taxation of New Financial Instruments.” Tax Notes International (October 15): 261–73. Vickrey, W. 1939. “Averaging Income for Income Tax Purposes.” Journal of Political Economy 47 (3): 379–97. Warren, A. C. Jr. 1993. “Financial Contract Innovation and Income Tax Policy.” Harvard Law Review 107 (2): 460–92. 3 Taxation of Banks: Modeling the Impact Ramon Caminal In most countries banking activity is subject to general taxation (personal and corporate income taxes), but often banking services are treated differently by tax authorities. In some cases, they enjoy favorable treatment. For instance, in the European Union most fi- nancial services are exempt from value-added tax (VAT), ostensibly for technical reasons (although the issue is currently under consid- eration). In other cases, banking services are subject to special taxes, like unremunerated reserve requirements. Unremunerated reserve requirements are an implicit form of bank-specific taxation that work in combination with inflation. Taxation of banks is of particular interest for various reasons. First, banks are financial intermediaries that perform unique and crucial functions—although in many countries they are subject to increasing competition from investment funds and security markets. Second, banks are heavily regulated and monitored, which reduces the administrative costs of some forms of taxation; at the same time they are subsidized through under-priced deposit insurance and bailouts of insolvent banks. Third, banks often enjoy some monop- oly power, especially in the household and small business sectors. The goal of this chapter is to develop a theoretical framework to analyze the impact of various forms of taxation. The model is based on the modern theory of the banking firm and integrates in a uni- fied framework the most important aspects of banking activity; in 81 82 RAMON CAMINAL particular, monitoring, transaction services, and asset transforma- tion. It is important to understand how each of these functions is af- fected by taxation. Banks reduce informational asymmetries with their loan applicants by ex ante screening, interim monitoring, and ex post verification of financial returns. Moreover, banks develop long-term, customer-specific relationships that allow them to reuse the information acquired in previous transactions. Bank deposits play an important role in the payment system. They can easily be converted into cash or directly used in transactions through checks, credit, and debit cards. Depositors can also set up automatic pay- ments. Some of these transaction services may be separately priced, but to a large extent they are implicitly paid by accepting a rate of return on bank accounts below those of alternative assets. Finally, banks and other financial intermediaries perform an important asset transformation function. In particular, bank assets are riskier and less liquid than their liabilities. The model presented here ab- stracts from risk diversification and (as for example with Diamond and Dybvig 1983) focuses exclusively on liquidity insurance. The model developed in this chapter focuses on how efficiently savings are channeled into various investment opportunities. Thus it takes as given the amount of funds. In this sense it is a partial equilibrium model. The simplest version of the model is presented in the second section and analyzed in the third and fourth sections, and assumes perfect competition in both the deposit and the loan market. A crucial determinant of the incidence of bank taxation is whether the deposit and loan segments are separable: that is, whether deposit and loan interest rates are independently deter- mined. Under conditions of separability, a tax on deposits does not affect lending, and vice versa a tax on bank loans leaves the level of deposits unchanged. The framework developed in this chapter makes heavy use of the separability hypothesis. In particular, it is as- sumed that banks can invest in a safe asset with an exogenous rate of return. The chapter revisits the separability hypothesis and ar- gues that in the real world the necessary conditions to obtain sepa- rability may be violated quite often. However, simultaneous devia- tions may partially compensate one another, and as a result a model assuming separability may still be a useful benchmark. Recent improvements in information technology and innovations in financial contracting have increasingly challenged traditional banking activities.1 The chapter considers explicitly the effects of increasing competition from investment funds and more efficient se- curity markets. Investment funds are characterized as perfect sub- stitutes of banks in the asset transformation function. This is ex- TAXATION OF BANKS: MODELING THE IMPACT 83 treme, but it captures the idea that the implications of asset pooling by any large financial intermediary are similar. Another caveat is that some investment funds are already providing a limited amount of transaction services in some countries. If investment funds man- age to develop such services, then bank deposits and investment fund holdings will become almost perfect substitutes. In some less developed countries the main source of competition for banks comes from the informal credit sector. Some of the effects may be analogous to those discussed in the chapter, although the normative implications are likely to be very different. The chapter then turns to imperfect competition. Banks’ market power introduces two new effects. First, the distortionary effects of taxes and market power are compounded, and as a result even a small tax rate is likely to have a negative first-order effect on wel- fare. However, the distortions associated with market power de- crease with the ability of banks to price discriminate, which is likely to be significant in the loan market. Second, taxes may at least par- tially cut into banks’ economic profits, and hence reduce the tax burden of investors (which, in a more general model, would reduce the distortion on savings decisions). However, economic profits may prevent banks from taking excessive portfolio risk. The chapter then discusses whether taxes may induce higher risk-taking by al- tering the rewards to prudent behavior. Taxation and regulation are shown to interact in various ways. On the one hand, corporate income taxation cannot be exclusively a tax on pure economic profits if capital requirements are binding. On the other hand, if new taxes induce excessive risk-taking behav- ior, then existing solvency regulation must be strengthened in order to preserve the stability of the banking system. Some of these issues are analyzed in the chapter. The Benchmark Model The economy is populated by four types of agents: investors, entre- preneurs, banks, and nonbank financial intermediaries (investment funds). Transactions take place in three consecutive periods, in- dexed by t, t = 0, 1, 2. There is a continuum of ex ante identical in- vestors with mass equal to N, which is assumed to be sufficiently large. Each investor has an endowment of one unit at time t = 0. Next, at time t = 1 she finds out whether she is impatient—that is, she must consume then and obtain a utility of u(c 1)—or she is patient—that is, she must consume at t = 2, and obtain utility u(c2). 84 RAMON CAMINAL Investors’ type (whether patient or impatient) is assumed to be the agent’s private information. The probability of the two events is, for simplicity, one-half.2 Ex ante preferences on consumption of real goods are given by expected utility: U = u (c1) + ρu (c2) where ρ is a positive discount factor and u is a concave (and twice differentiable) function. Investors have access to the following safe investment technologies: • Short-term investment: Each unit of investment at t = 0 yields one unit at t = 1. The same technology can be operated between pe- riods 1 and 2. • Long-term investment: Each unit of investment at t = 0 yields R > 1 at t = 2. The investment project can be liquidated at t = 1, in which case the unit return is Z. ASSUMPTION 1: R–1 < Z < 1 In order to simplify the presentation, the time discount factor is fixed: ASSUMPTION 2: ρ = R–1 There is a continuum of heterogeneous entrepreneurs, with mass equal to 1, who have no funds but have access to two types of proj- ects (safe and risky). They can invest one unit in the safe project at t = 0 and obtain X at t = 2. Alternatively, they can invest one unit at t = 0 in the risky project and obtain at t = 2 a random financial return {X with probability p 0 with probability 1 – p plus a non-financial return (private benefit) of φ (1 – ρ). Entrepreneurs differ only in their probability of success of the risky project, ρ, which is common knowledge. Entrepreneurs are distributed over the interval [0,1] according to the probability den- sity function h (p). Let H (p) denote the cumulative distribution. Fi- TAXATION OF BANKS: MODELING THE IMPACT 85 nally, entrepreneurs are assumed to be risk-neutral and wish to con- sume only at time t = 2. In other words, their objective function is the expected net return at time t = 2. ASSUMPTION 3: R+φ>X>R>φ As shown below, the first inequality implies that in the absence of monitoring entrepreneurs prefer the inefficient (risky) project. From the second inequality it follows that the safe project has a positive net present value (NPV), which is always higher than the NPV of the risky project (third inequality). However, if p is sufficiently close to one then the risky project also has a positive NPV. Banks are assumed to perform three functions: monitoring, trans- action services, and liquidity transformation. That is, first, they can monitor entrepreneurs, which reduces agency costs in the credit mar- ket. Second, they provide transactions services to investors, such as routine payments and check writing. Third, they offer investors li- quidity insurance. More precisely, banks can monitor an entrepreneur, which in- volves a cost m > 0 (measured in time t = 2 units) independently of p, and induce the entrepreneur to choose the efficient project (the safe project). Thus bank monitoring sets an upper bound on the size of the agency cost of all entrepreneurs. This is a bit drastic, since in equilibrium all entrepreneurs get credit independently of market con- ditions. The purpose here is to focus on the choice between bank and nonbank financing rather than on entrepreneurs’ credit availability. Banks’ liabilities are also special. As discussed below, banks can provide a better time structure of returns than direct investment. Moreover, by incurring a cost µ per unit of deposit they provide transaction services. These are valued by investors according to the function v(D), where D is the level of bank deposits and v is a con- cave (and twice differentiable) function, with v (0) = 0, v' (0) = ∞, and v' (D) > 0 if and only if D < 1. The utility derived from trans- action services enters additively in investors’ utility function. Thus investors’ ex ante preferences can be written as:3 (3.1) U = u (c1) + ρu (c2)+ v(D) Notice that under such a formulation the supply of deposits will de- pend not only on the relative return of deposits with respect to al- ternative assets, but also on investors’ income. This is analogous to 86 RAMON CAMINAL the interest rate and income effects in the money demand function, which has broad empirical support. The Walrasian Outcome in the Absence of Taxes and Investment Funds This section analyzes the equilibrium of a Walrasian system with and without banks. Direct Investment First consider the Walrasian equilibrium under direct investment. That is, suppose that there exist perfectly competitive markets to trade the claims on various investment projects, both at t = 0 and at t = 1. For simplicity, assume that investors pay a cost d at time t = 1 (measured in current units) if they sell a claim on one unit deliv- ered at time t = 2, but do not incur any transaction cost when pur- chasing assets at time t = 0. This is meant to capture the idea that the cost of acquiring an asset is relatively small but frequent trans- actions in the secondary market are relatively more costly. One im- portant implication of the above assumption is that at time t = 0 in- vestors can perfectly diversify their portfolios and hence act as risk-neutral agents when lending to a particular entrepreneur. Of course, this is extreme. The goal here is to concentrate on the term structure of assets (liquidity insurance) and completely abstract from asset risk. On the other hand, the entrepreneur incurs a non- pecuniary cost (measured in time t = 2 units) of f (< m), when col- lecting one unit of funds in the market for securities. Given the structure of returns and because of limited liability, se- curity design is not an issue in this model. Thus without loss of gen- erality one can assume that entrepreneurs can issue a bond that promises to repay r b in case of success. If they invest in the risky asset they obtain p(X – r b) + (1 – p)φ, and if they invest the safe asset they get X – r b. Given Assumption 3 and since investors require r b ≥ R, then entrepreneurs of all types prefer to invest in the risky project. Competition among investors implies that the net benefit from lending to entrepreneurs instead of investing in long-term technol- ogy is zero: that is, R rb = p TAXATION OF BANKS: MODELING THE IMPACT 87 Because of limited liability, r b ≤ X. As a result, only those entrepre- neurs with p ≥ po can obtain financing in the securities market, where po is given by:4 R =X po Given that entrepreneurs’ projects have a two-period maturity, those with p ≤ po will obtain financing—the market will clear—provided investors wish to invest a sufficient amount in assets with such ma- turity. Similarly, since the liquidation value of entrepreneurs’ projects at time t = 1 is zero, it is necessary to check that the secondary mar- ket works properly. These issues are discussed in the appendix. Turn now to investors’ decisions. At time t = 0 investors must de- cide how to distribute their unit endowment between short-term as- sets, 1 – I, and long-term assets, I. At time t = 1 there are potential gains from trade. Patient consumers would like to use the proceeds from short-term assets to buy assets that mature at time t = 2, and impatient consumers may be willing to sell their long-term assets. Let q be the price paid by patient consumers at time t = 1 for a claim on one unit at time t = 2. Hence sellers obtain q(1 – d) per unit. At time t = 1 patient consumers are willing to participate in the sec- ondary market provided the implicit return on those assets is not below their alternative investment opportunity (the short-term in- vestment technology). Thus patient consumers are willing to buy the claims sold by impatient consumers provided q ≤ 1. Similarly, impatient consumers are willing to sell their claims on second pe- riod returns only if they find it profitable: that is, if q(1 – d) R ≥ Z. The following assumption simplifies the analysis considerably. ASSUMPTION 4: (1 – d)R = Z Under Assumption 4, impatient consumers trade if and only if q ≥ 1. As a result, the equilibrium price is q = 1. At time t = 0 investors are able to anticipate the equilibrium price of the secondary market. Hence their consumption profile as a function of their portfolio is given by: c1 = 1 − I + q(1 − d)RI = 1 − (1 − Z)I (3.2) 1− I c2 = + RI = 1 + (R − 1)I q 88 RAMON CAMINAL Investors choose I in order to maximize equation 3.1 subject to equation 3.2. The solution is fully characterized by the first-order condition: u ′(c1) 1 − R −1 = u ′(c2 ) 1− Z Given Assumption 1, c1 < c2 , which implies that I > 0. The assumption about positive transaction costs in the secondary market is not innocuous. In case d = 0, then the equilibrium in the secondary market is q = R–1 and as a result c1 = 1 and c2 = R. It turns out, as pointed out by Jacklin (1987), that in this case banks cannot improve the market allocation (they cannot provide liquidity insur- ance). Diamond (1997) and Holmstrom and Tirole (1998) present more sophisticated models of liquidity with implications analogous to the current formulation. The Role of Banks Now introduce banks into the picture. They intermediate between investors, on the one hand, and safe investment technologies and entrepreneurs, on the other. Thus when considering lending to en- trepreneurs, their opportunity cost of funds is R. The distinctive fea- ture of banks on the asset side is that they can monitor entrepre- neurs (at a cost) and induce the efficient project selection. In other words, bank monitoring eliminates any potential inefficiencies orig- inated by entrepreneurs’ moral hazard problem. Alternatively, a large literature has focused on a different type of informational fail- ure in the credit market: adverse selection (hidden types), instead of moral hazard (hidden actions). Thus this characterization of the role of banks in reducing informational asymmetries is clearly lim- ited, although it suffices to illustrate how various forms of taxation may reduce banks’ contribution to economic efficiency in the credit market (through the disintermediation effect). Competition among banks implies that a loan contract will re- quire the entrepreneur to choose the safe project and pay back at time t = 2 an interest rate: (3.3) rl = R + m Clearly, one must assume that monitoring is economically feasible: ASSUMPTION 5: X>R+m TAXATION OF BANKS: MODELING THE IMPACT 89 As a result, an entrepreneur with a probability of success on the risky project p, will apply to a bank loan instead of issuing bonds provided the following condition holds: X – r 1 ≥ p(X – r b) + φ(1 – p) – f The left-hand side is the expected return of borrowing from a bank. In such a case the entrepreneur expects to be forced to choose the safe project. The right-hand side is the expected return of borrow- ing on the securities market (which is feasible if and only if p ≥ po). Given the equilibrium values of r1 and rb such a condition can be written as p ≥ p*, where p* is given by: m−f P∗ = 1 − X−φ For simplicity, assume that the entrepreneur who is indifferent be- tween the two sources of credit is not constrained in the securities market: that is ASSUMPTION 6: p* > p0 Thus entrepreneurs with a low value of p (high agency cost in the securities market) choose to borrow from banks, while entrepre- neurs with a value of p close to one (low agency cost) prefer to bor- row on the securities market. Notice that banks make credit feasi- ble for entrepreneurs with p < p0 (those who are excluded from the bond market). Now turn to the liability side of a bank’s balance sheet. Bank de- posits provide transaction services and on top of this they are able to offer a better time profile of financial returns than direct investment. In particular, a bank deposit could offer investors a return r1d if funds d are withdrawn at time t = 1, and a return r2 if they are withdrawn at time t = 2. Thus a patient consumer withdrawing at time t = 1 can invest the funds withdrawn in the short-term investment technology (or buy claims in the secondary market). Hence, banks must take into account investors’ incentive compatibility constraint: (3.4) d ≤ rd r1 2 Banks do not need to transact in the secondary market because there is no aggregate uncertainty and they can exploit the law of large 90 RAMON CAMINAL numbers. Since banks anticipate that half of their depositors will withdraw their funds at time t = 1, they need to invest a proportion β‚ of their deposits in long-term assets, and a proportion 1 – β‚ in short-term assets, in such a way that: 1 d (r1 + µ) = 1 − β 2 (3.5) 1 d (r2 + µ) = βR 2 Eliminating β in the above equations produces the set of all possi- ble combinations of r1d and r d that can be offered to investors: 2 d r2 1+ R (3.6) r1d = =2−µ R R If investors decide to deposit in a bank a fraction D of their en- dowment and to invest directly a fraction 1 – D, then they can enjoy the following consumption profile: d D + (1 – D)[1 – (1 – Z)I] c1 = r1 (3.7) d D + (1 – D)[1 + (R – 1)I] c2 = r2 Thus in equilibrium, bank deposit contracts will be part of the so- lution to the following optimization problem: choose (r1 d, r d, D, I) 2 in order to maximize equation 3.1 subject to equations 3.4, 3.6, and 3.7 and the feasibility constraint D ≤ 1. Notice that the market allocation implies a higher return in the second period than in the first. However, in the absence of the in- centive compatibility constraint, banks would offer a structure of returns such that consumption in both periods is the same, which implies that the deposit rate in the first period is higher than in the second. Clearly, this violates the incentive compatibility constraint (equa- tion 3.4). Hence, in equilibrium: 2R (3.8) d r1d = r2 ≡ rM − µ ≡ −µ 1+ R The first-order condition that characterizes the optimal amount of deposits can be written as: R−Z M (3.9) u ′(c1) (r − µ − 1) + v ′(D) = 0 R −1 TAXATION OF BANKS: MODELING THE IMPACT 91 By pooling assets, banks can in principle offer investors a time pro- file of financial returns that dominates that of direct investment, since banks do not need to trade in the secondary market or liqui- date projects too early. Moreover, deposits provide transaction ser- vices. On the negative side, those services are costly, which implies that the overall return on deposits must be lower. Thus one must distinguish two cases depending on the costs of providing transac- tion services: • If r M – µ – 1 > 0, then the optimal amount of deposits is D = 1. In this case, deposits dominate direct investment. • If r M – µ – 1 < 0, then D < 1. In this case, deposits do not dom- inate direct investment. The cost of providing transaction services is sufficiently large to reduce the level of financial returns, which leaves room for a positive amount of direct investment. Bank Taxation in the Absence of Investment Funds This section examines the effects of taxation under perfect competi- tion among banks but in the absence of nonbank financial interme- diaries. It looks at five types of taxation: on deposits, loans, value- added, investors’ capital income, and banks’ corporate income. By focusing on one type of taxes at a time, one can study the effects of “differential” taxation. Since this chapter pays little attention to normative issues, it does not discuss in detail how to achieve tax neutrality with respect to portfolio allocation. A Tax on Deposits The structure of the model involves separability between loans and deposits. In particular, loan interest rates or credit availability is not affected by changes in the deposit market. The reason is that banks’ opportunity cost of funds in the loan market is the (exogenous) re- turn on the long-term investment technology, and not the deposit rate. Suppose that the government sets a proportional tax on the gross interest rate on deposits, τ d.5 Clearly, such a tax will not change the structure of pre-tax deposit rates: r1 d = r d = r M – µ. How- 2 ever, investors will take into account such a tax when deciding the amount of deposits, since the after-tax return is now (1 – τ d )rt d, t = 1, 2. One must distinguish between two cases: • If r M – µ – 1 > 0, then deposits dominate direct investment, and as a result a small τ d is nondistortionary. If the tax rate is sufficiently large, then the effects are analogous to those of the second case. 92 RAMON CAMINAL • If r M – µ – 1 < 0, then investors face a trade-off between a higher (composite) return of direct investment and the transaction services provided by bank deposits. In this case, the optimal level of deposits is given by the first-order condition of the representative in- vestor’s optimization problem (adaptation of equation 3.9): u ′(c1) = R−Z R −1 [ ] (1 − τ d )(r M − µ) − 1 + v ′(D) = 0 Clearly, dD <0 dτd Taxes affect the supply of deposits through two different channels. A higher tax rate increases the opportunity cost of deposits (the differ- ence between the rates of return of deposits and direct investment). It also reduces investors’ disposable income. Both effects reduce the sup- ply of deposits. Notice that the effect of τ d on deposits depends on the second derivative of functions u (⋅) and v (⋅) This allows one to sign such a derivative, but also suggests that it is very difficult a priori to determine how changes in various parameters influence the distor- tionary effect of taxes. The next proposition summarizes these results. PROPOSITION 1. Under perfect competition in banking and in the absence of investment funds, a tax on deposits has no effect on the loan market. If the costs of providing transaction services are suffi- ciently small, so that bank deposits dominate direct investment, then a small tax on deposits does not affect the amount of deposits. If the costs of providing transaction services are sufficiently large so that bank deposits do not dominate direct investment, then a tax on deposits reduces the amount of deposits and increases direct invest- ment. Thus two functions of banks are jeopardized: provision of transaction services and liquidity insurance. It is immediately clear that even in the second case (deposits do not dominate direct investment), a small tax rate on deposits has only a second-order effect on total welfare, since the market alloca- tion is efficient. A Tax on Bank Loans Suppose that the government sets a proportional tax on the gross re- turn on bank loans, τ l. Let r l denote the after-tax loan rate. The competitive rate is independent of taxation and hence entrepreneurs TAXATION OF BANKS: MODELING THE IMPACT 93 must pay r l/(1 – τ l). Therefore, the threshold value of p, p*, is now given by: m + τl R − f (1 − τl ) (3.10) p∗ (τ) = 1 − (1 − τl )(X − φ) Thus dp *(τl ) <0 dτl Under separability, a tax on bank loans does not have any effect on the deposit rate or the provision of transaction services and liquid- ity insurance. However, it raises the cost of loans, which discourages entrepreneurs away from information-intensive financing and into unmonitored financing. PROPOSITION 2. Under a perfectly competitive banking system and in the absence of investment funds, a proportional tax on bank loans raises loan rates and induces some entrepreneurs to switch from bank to nonbank financing (the level of bank monitoring de- creases). It does not have any effect on the deposit interest rate or the level of deposits. It has been argued that bank monitoring creates a positive infor- mational externality on the securities market (Besanko and Kanatas 1993).6 This would not alter the qualitative results, but it would amplify the distortionary effects of taxation.7 A Tax on Banks’ Value-Added Consider a tax on banks’ value-added under the cash flow approach.8 Suppose that entrepreneurs are VAT-registered businesses, but depos- itors are not engaged in commercial activities. At times t = 0 and t = 1 banks have zero net cash flow and hence they pay no tax. At time t = 2 their net cash flow is: 1 d (3.11) CF = r l L + RB − (r + µ)D 2 where L is the amount of loans and B is the amount of investment in the long-run technology (or bonds). Banks’ feasibility constraint can be written as:  1  (3.12) L + B = 1 − (r d + µ) D  2  94 RAMON CAMINAL Using equation 3.12, equation 3.11 can be rewritten: R + 1 M  (3.13) CF = (r l − R)L +  (r − µ − r d ) + µ D  2  In the absence of taxes, r l and r d are given by equations 3.3 and 3.8. Substituting these equations into equation 3.13, it can be established that CF = mL + µD Notice that such a measure of banks’ value-added captures only mon- itoring and transaction services, but disregards liquidity provision. This is because it has been assumed that asset transformation is costless. Suppose that the government sets up a tax τν on banks’ value- added. More specifically, value-added throughout the economy is already taxed at the rate τν, and the government lifts the exemption of banking services from general VAT. Under perfect competition after-tax profits in both loans and deposits must be zero. This im- plies that the equilibrium loan rate is given by: m (3.14) rl = R + 1 − τν Similarly, the deposit rate in equilibrium can be written as: 2τ ν µ rd = r M − µ − (R + 1)(1 − τ ν ) Since the assumption has been made that depositors are outside the VAT system, this implies that lifting the exemption on banking ser- vices is equivalent to a tax on deposits. However, since the tax paid on loans reduces the borrowers’ tax bill, the effect on the loan mar- ket is null. In order to check this last statement, simply notice that under the exemption, those entrepreneurs financed by loans obtain: πE = (X – r l) – τνX = (1 – τν)X – R – m If the exemption is lifted, then entrepreneurs get: πE = X – r l – τν(X – r l + R) The reason is that now entrepreneurs can reduce their tax bill by the amount τ(r l – R).9 Using equation 3.14, one can check that entre- preneurs’ profits are identical in both regimes. TAXATION OF BANKS: MODELING THE IMPACT 95 PROPOSITION 3. If borrowers are VAT-registered firms and depos- itors are not engaged in commercial activities, then lifting the ex- emption of banking services in the base of the general VAT is equiv- alent to setting a tax on deposits, with no impact on the loan market. A Tax on Gross Interest Income Consider a proportional tax on investors’ gross interest income, τY. The tax base of such a tax includes gross interest from deposits as well as the gross return of direct investment. Clearly, the relative re- turn of deposits and direct investment remains unaffected. How- ever, such a tax reduces the supply of deposits through the income effect. More specifically, if (r M – µ) – 1 < 0, then the supply of de- posits is given by equation 3.9: R−L M u ′(c1) (r − µ − 1) + v ′(D) = 0 R −1 The effect of such a tax comes through cl. In particular, a higher tax rate reduces disposable income, which implies a reduction in both consumption of real goods and the demand for banks’ transaction services (reduces the supply of deposits). Thus we have the follow- ing result: PROPOSITION 4. Provided that deposits do not dominate direct in- vestment, a tax on interest income reduces the amount of deposits. Thus as long as the income effect in the supply of deposits is sig- nificant, general taxes like the personal income tax may also have an effect on financial intermediaries.10 Banks’ Corporate Income Taxes A complete analysis of the impact of corporate taxation on banking activity requires a satisfactory theory of banks’ capital structure. Unfortunately, such a theory is not yet available. The role of banks as intermediaries implies that the bank will tend to have little inside capital. However, if bankers are subject to moral hazard and perfect diversification is not feasible, then equity capital held by insiders provides an incentive for the insiders to do monitoring (Holmstrom and Tirole 1997). In any case, in the real world banks tend to hold relatively low levels of capital11 and as a result capital requirements are frequently binding. 96 RAMON CAMINAL This model is not able to deal with security design issues, and hence debt and equity are indistinguishable. Nevertheless, it can still provide a preliminary analysis of the effects of corporate income taxation whenever capital requirements are binding. Suppose that banks hold a ratio of capital to loans equal to η, K = ηL.12 Since ag- gregate risk or corporate control is not an issue in the present framework, and equity must have the same maturity as loans, po- tential shareholders will require a rate of return at time t = 2 equal to R.13 Finally, let banks’ corporate tax rate be τc. Since banks offer depositors a constant interest rate and invest in the short-run tech- nology the exact amount in order to pay impatient depositors, then banks’ economic profits can be written as: rd + µ R (3.15) π B = (r l − m)L + RB − D− ηL 2 1 − τc Notice that the cost of equity for the bank is R/(1 – τc). Banks’ fea- sibility constraint is: rd + µ (3.16) D= D + B + (1 − η)L 2 The first term of the right-hand side is the amount invested in the short-run technology. The rest is invested in the long-run technol- ogy, B, or in loans, (1 – η)L (an amount ηL of loans is financed by equity). Plugging equation 3.16 into equation 3.15 and rearranging:   η  R+1 M π B = L r l − m − R1 − η + c +D (r − r d − µ)     1− τ   2 Under perfect competition (and because of separability) banks make zero profits in both deposits and loans. As a result, deposit rates are un- affected by corporate income taxes, r d = r M – µ, but loan rates are not:  η  r l = (1 − η) + R + m  1 − τc  This leads to the next proposition. PROPOSITION 5. If capital requirements are binding, and under per- fect competition, a tax on banks’ corporate income is equivalent to TAXATION OF BANKS: MODELING THE IMPACT 97 a tax on bank loans. Hence, it raises the loan interest rate (and re- duces the amount of loans) but does not affect the deposit rate.14 The result that a corporate income tax is a tax on loans may be more general than it appears. Suppose that capital requirements are not binding but banks’ inside capital provides positive incentives to monitor borrowers (as in Holmstrom and Tirole 1997). Then loans and capital are also complementary. Similarly a tax on corporate in- come would also fall on loans. Competition from Alternative Financial Arrangements This section asks to what extent the presence of alternative financial arrangements affects the previous analysis. It focuses on the one hand, on the appearance of other financial intermediaries (mutual and pension funds) that offer investors imperfect substitutes of bank deposits; and, on the other hand, on the impact of more efficient markets for securities, which allow an increasing number of firms to consider issuing bonds as an alternative to bank loans. Investment Funds In principle, any large intermediary can supply investors the same asset transformation function offered by banks. Asset pooling allows intermediaries to issue financial contracts with lower risk and higher liquidity. This is the main role of mutual funds and pensions funds. In this dimension, banks are no longer special. Accordingly, in the context of the current model, investment funds are characterized as follows: they can supply the same liquidity services as banks (in par- ticular, they can offer a constant return) but they cannot monitor en- trepreneurs or offer investors transaction services. Such a character- ization overlooks various observations. First, money market funds in the United States have begun to provide some transaction services.15 Second, investment funds have been reported in some instances to monitor the management of corporations where they hold stock. Third, various intermediaries seem to specialize in issuing different varieties of financial contracts. Therefore, the proposed characteri- zation must be considered only as a first approximation. In particular, it is assumed that perfectly competitive investment funds can hold the same portfolio in terms of maturity structure as banks and thus offer investors liquidity insurance. In principle, in- vestment funds (under perfect competition) should offer contracts 98 RAMON CAMINAL that allow investors a unit return of r1 m if they withdraw their funds m at time t = 1, and r2 if they wait until time t = 2. The previous analy- sis of equilibrium deposits leads to the conclusion that perfectly competitive investment funds will offer: 2R r1M = r2M = r m ≡ 1+ R As discussed above, such a structure of returns dominates direct in- vestment. As a result, banks face tighter competition. In the absence of taxes, investors can achieve the following consumption structure as a function of their portfolio: c1 = r1d D + r M (1 − D) (3.17) d c2 = r2 D + r M (1 − D) Thus the equilibrium deposit contract will be part of the solution to the following optimization problem: choose (r1 d, r d, D) to maximize 2 equation 3.1 subject to equation 3.17. As expected, deposit interest rates are also constant over time: 2R d r1d = r2 = rm − µ = −µ I +R Thus consumption is the same in both periods, c1 = c2 = c, and the optimal supply of deposits is given by: 1+ R (3.18) −u ′(c) µ + v ′(D) = 0 R where c = rM – µD Now the level of deposits can be compared in the absence and in the presence of investment funds (equations 3.9 and 3.18). If µ < (R – 1)/ (R + 1) then bank deposits dominate direct investment, but do not dominate investment fund holdings. Hence competition from invest- ment funds unambiguously reduces the level of deposits. The reason is that without direct investments, investors are at a corner solution (D = 1), but competition from investment funds place investors in an interior solution (D < 1). However, if µ > (R – 1)/(R + 1), then the ef- fect of investment funds on bank deposits is ambiguous. If the first- TAXATION OF BANKS: MODELING THE IMPACT 99 order condition equation 3.18 is evaluated at the level of deposits without investment funds, then c1 is larger, which implies that u'(c1) is lower. However, the relative price of deposits in terms of con- sumption is higher. That is, 1+ R R− Z µ> (1 − r M + µ) R R−1 Hence in this case, the effect of investment funds on deposits is am- biguous. The intuition is straightforward. On the one hand, the presence of investment funds raises the opportunity cost of holding deposits (the price effect). On the other hand, investment funds ex- pand the set of feasible financial returns (income effect). The price effect works against deposits, but the income effect works in the op- posite direction.16 The effect of investment funds on bank deposits is summarized in the following proposition: PROPOSITION 6. The presence of investment funds unambiguously reduces the level of bank deposits, if the latter dominates direct in- vestment. The effect on bank deposits is ambiguous, if the latter does not dominate direct investment. Having analyzed the effect on the tax base, now turn to the im- pact of investment funds on the distortionary effects of taxation. Consider again the effect of a tax on deposits. Under competition from investment funds, the amount of deposits will be given by: 1 + R  d 2R  u ′(c1) τ + (1 − τ d )µ  = v ′(D) R  1+ R  where 2R  2R  c1 = − Dτd + (1 − τd )µ  1+ R  1 + R  Clearly, dD / d τ d < 0. Notice that the size of the distortionary effect depends among other things on the degree of convexity of the utility functions—on the sign of uٞ(⋅) and vٞ(⋅). Hence it is rather difficult to make state- ments based on theoretical arguments about how competition from investment funds influences the effect of taxes. Because of separability, competition from investment funds does not affect the impact of a tax on loans. 100 RAMON CAMINAL PROPOSITION 7. In the presence of investment funds, a tax on de- posits unambiguously reduces the level of deposits. Provided that bank deposits do not dominate direct investment, the effect of a de- posit tax may be higher or lower than in the case where bank de- posits only compete with direct investment. Bank-specific taxes can reduce only monitoring and provision of transaction services, since liquidity insurance is also provided by banks’ competitors. More Efficient Security Markets On the asset side, banks in some industrial countries are increas- ingly facing tighter competition from security markets (development of commercial paper market). This could translate into a decrease in the transaction costs incurred by entrepreneurs when they bor- row on security markets (a reduction in f ). In order to study the effect of a change in f on the credit market equilibrium, refer again to equation 3.10: m + τl R − f (1 − τl ) p∗ (τl ) = 1 − (1 − τl )(X − φ) Thus the effect of a reduction in transaction costs on the entrepre- neurs’ choice of the source of funds is simply: dp∗ (τl ) 1 = >0 df X−φ Unsurprisingly, a reduction in transaction costs will reduce the num- ber of entrepreneurs who choose to apply for a bank loan instead of issuing securities. What is less obvious is the effect of f on the im- pact of taxes. Notice that: d 2 p∗ =0 dτl df That is, the level of f does not affect the impact of taxes on the threshold level p*. Hence the impact of f on the size of the distor- tion caused by taxes depends exclusively on the distribution of en- trepreneurs: that is, on the shape of h (p). It could be reasonable to expect that, in the relevant range, h'(p*) < 0. That is, if the securities market is inefficient (f large), then p* is close to 1, and hence the number of firms borrowing on the securities market is small. Moreover, a small tax would reduce the TAXATION OF BANKS: MODELING THE IMPACT 101 threshold but would induce only a few additional entrepreneurs to switch from bank lending to unmonitored lending. However, if trans- action costs are very low, then the threshold level p* could fall in a re- gion where the density of entrepreneurs is substantially higher. As a result, a tax would have a larger impact on bank lending. Although reasonable, this was only a conjecture on the shape of the distribution of entrepreneurs on the relevant dimension. To summarize: PROPOSITION 8. As security markets become more efficient, the level of bank loans falls as more firms choose to obtain their funds in the security markets. The distortionary effect of taxes may in- crease or decrease, although the former appears more likely. Taxing Imperfectly Competitive Banks Banks seem to enjoy some degree of market power, at least in the household and small business sectors.17 Monopoly power creates distortions in the allocation of resources, provided banks can neither perfectly price discriminate across customers nor offer nonlinear prices. Price discrimination seems particularly likely in the loan mar- ket, since rates are individually tailored. This section examines how market power and taxes interact. It considers two polar cases. The first assumes that banks enjoy market power in the deposit market but cannot offer nonlinear prices (depositors are identical). The sec- ond assumes that banks enjoy market power in the loan market and can perfectly price discriminate (the loan size is exogenous). Monopoly Power in the Deposit Market This subsection extends the model above to allow for monopoly power in the deposit market. Thus banks compete not only with one another but with a perfectly competitive investment fund industry. Suppose there are two banks, A and B (given the static nature of the game, it can easily be generalized to n banks). Preferences of the representative investor are given by: U = u(c1) + ρu(c2) + v (DA, DB) where Di is the amount of deposits in bank i, i = A, B. For simplic- ity, following Matutes and Vives (2000), the examination restricts itself to a quadratic function: 1 2 2 v(DA , DB ) = α(DA + DB ) − (DA + 2λDA DB + DB ) 2 102 RAMON CAMINAL where λ represents the degree of substitutability between deposits at the two banks, 0 < λ < 1. As λ → 0 then the two goods become in- dependent and banks become monopolists. In the opposite extreme, as λ → 1 then the two goods become perfect substitutes and (under interest rate competition) the market for deposits becomes perfectly competitive. The timing of the game is as follows. First, banks simultaneously set their deposit interest rates; next investors choose how much to deposit into each bank and in investment funds. It is easy to show (see appendix) that if bank owners face the same liquidity shocks as investors, then in equilibrium they offer deposit contracts with con- stant interest rates, rid d d 1 = r12 = ri . Mutual funds also offer a constant M interest rate: r = 2R/(1 + R). As a result, investors’ consumption profile is also constant over time: c1 = c2 = c. Thus given (rA d, r d), in- B vestors choose (DA, DB) in order to maximize 1+ R U= u(c) + v(DA , DB ) R subject to (3.19) c = r M − (r M − i d A )DA − (r M − id B )DB where i d denotes the after-tax deposit rate: that is, iid = (1 – r d)rid. The supply of deposits is obtained by inverting the first-order con- ditions. For i, j = A, B, i ≠ j. Di = a − bΩ(r M − iid ) + kΩ(r M − i d j ) α a≡ 1+ λ 1 b≡ (3.20) 1 − λ2 λ k≡ 1 − λ2 1+ R Ω≡ u ′(c) R For convenience it is assumed that banks are relatively large to be able to affect interest rates, but sufficiently small so as not to affect investors’ consumption.18 Thus the supply of deposits to bank i in- creases with the deposit rate set by bank i and decreases with the de- posit rate set by the rival bank. In case of identical deposit rates, the TAXATION OF BANKS: MODELING THE IMPACT 103 aggregate supply of deposits decreases with the difference between the return on investment funds, r M, and the return on deposits, i d. Such a margin depends on three factors: intermediation costs, µ, taxes, and monopoly power. If the margin is zero, then investors put all their funds in bank deposits. As the margin increases, the aggre- gate supply of deposits decreases. It is assumed that banks need to set a positive deposit rate in order to attract a positive amount of deposits: that is,19 ASSUMPTION 7: a – (b – k)r MΩ < 0 If one plugs equation 3.20 into equation 3.19, then one gets con- sumption as a function of the deposit rates. In case of equal deposit d = i d ≡ i d, rates, iA B c = r M – 2(r M – i d)[a – (b – k)Ω(r M – i d )] Notice that a higher deposit rate may increase or decrease con- sumption. A higher deposit rate, on the one hand, reduces the op- portunity cost of deposits and induces investors to demand more transaction services at the cost of lower consumption (substitution effect). On the other hand, it expands the feasible set that tends to increase both the demand for transaction services and the demand for real consumption (income effect). The appendix also shows that if bank owners are subject to the same liquidity shocks as investors (and have the same preferences on financial returns), then the banks’ expected return from deposits is equal to r M = 2R/(1 + R). Thus given the interest rate set by other banks, rjd, and the tax rate on deposits, τ d, bank i chooses rid in order to maximize: πi = [r M – µ – rid]Di (rid, rjd ) where Di (rid, rjd ) is given by equation 3.20. Evaluating the first-order condition at the symmetric equilibrium: b(r M − µ)(1 − τd ) + (b − k)r M − a Ω (3.21) rd = (2b − k)(1 − τd ) Notice that Ω depends on c, and hence on r d. Therefore, the effect of τ d on r d and i d is complex. The direct effect of τ d on i d is clearly 104 RAMON CAMINAL negative. However, if a higher tax implies lower consumption, and hence higher Ω, the indirect effect has a positive sign. Nevertheless, the appendix shows that the direct effect always dominates the in- direct effect and hence: di d <0 dτd A direct implication of this result is that a tax on deposits reduces the level of deposits: dD <0 dτd Similarly, because of Assumption 7, the direct effect of τ on r d is positive, but if a higher tax implies higher consumption the indirect effect is negative. In this case, the appendix shows, that for some pa- rameter values the indirect effect dominates. In general, the direct effect dominates, provided the difference between the return on in- vestment funds, r M and the return on deposits, i d, is not too large. In other words, it is more likely that dr d/dτ d > 0 if intermediation costs and the tax rate are not too large, and banks do not have too much monopoly power.20 Deposit taxes unambiguously reduce investors’ utility, since they reduce the after-tax deposit rate and hence investors’ feasible set shrinks. The effect of taxes on banks’ profits is less straightforward. In the case that dr d/dτ d > 0, clearly taxes reduce bank profits since the supply of deposits faced by each bank shifts downward. In the case dr d/dτ d < 0, there is the possibility that softer rate competition more than compensates the decrease in the aggregate supply of deposits, and as a result bank profits increase rather than decrease with taxes. However, this is highly unlikely, since taxes reduce deposit rates only if banks have sufficient monopoly power, so that the strategic com- plementarity effect (the reduction in rate competition) is not an im- portant determinant of bank profits. In other words, for most pa- rameter values a tax on deposits falls on both investors and banks. PROPOSITION 9. A tax on deposits unambiguously decreases the net interest rate received by investors and hence it reduces the amount of deposits. If the interest margin is not too large (which oc- curs as a combination of low intermediation cost, low tax rate, and low monopoly power) then it increases the deposit rate paid by banks. As a result, the burden of the tax is shared by investors and bank owners. TAXATION OF BANKS: MODELING THE IMPACT 105 Notice that in this case a small tax has a first-order effect on wel- fare since in the absence of taxation, the level of deposits is too low because of banks’ monopoly power. However, the model may be ab- stracting from other important distortions. For instance, Whitesell (1992) argues that since one of the alternatives to bank deposits, cash, is typically taxed by inflation, in the absence of bank-specific taxes, the level of deposits may actually be too high. Now turn to corporate income taxation. The analysis is similar to the case of perfect competition. In particular, monopoly rents after taxes are given by:   η  R+1 M π B = (1 − τc )L r l − m − R1 − η +  + D (r − r d − µ)    1 − τ c    2 Therefore the incentives to set deposit rates are unchanged. Hence the corporate income tax is a tax on pure economic profits. At the same time, as discussed earlier, corporate income taxation distorts loan rates. In other words: PROPOSITION 10. If capital requirements are binding, a tax on banks’ corporate income is equivalent to a tax on both economic profits and loans. Thus it raises the loan rate and reduces bank profits. Similarly, a value-added tax under the same conditions described above is a tax on deposits and hence it distorts the deposit rate. The burden of the tax tends to be shared by bank owners and investors. Monopoly Power in the Loan Market Suppose that all entrepreneurs have access to the securities market but the loan market is segmented in such a way that each entrepre- neur can borrow only from a particular bank.21 Thus banks do not interact with one another and the only limit to their market power comes from the securities market. As discussed, provided p ≥ p0 en- trepreneurs can get credit in the securities market at the rate R/p and will select the risky project. Hence monopolistic banks will find it optimal to set an interest rate r l that leaves the entrepreneur (with p ≥ p0) indifferent between borrowing from the bank or from the se- curities market: rl  R (3.22) X− = p X −  + φ(1 − p) − f 1− τ  p 106 RAMON CAMINAL That is, rl = (1 − p)(X − φ) + R + f 1− τ provided such loan rate is above the average cost, r l ≥ R + m.22 In other words, a bank lends to entrepreneurs at the rate given by equation 3.22, if and only if p0 ≤ p ≤ p*, where p* is given by τR + m − (1 − τ)f p∗ = 1 − (1 − τ)(X − φ) (3.23) Notice that equations 3.10 and 3.23 are equivalent. Since banks can price discriminate (to the same extent that markets do) then monop- oly power does not involve any additional distortion and affects only the distribution of surplus between banks and entrepreneurs. Taxes have the same distortionary effects analyzed for the case of perfect competition above. Also, notice from equation 3.22 that the loan rate by entrepreneurs, r l/(1 – τ ), is independent of the tax rate. Hence only the marginal entrepreneurs are affected by taxation (they are in- duced to switch to market financing) but the bulk of the tax falls on bank owners. Consequently, the following result is obtained: PROPOSITION 11. The effect of loan taxation on entrepreneurs’ choice of credit is the same under perfect competition and monop- oly. Thus a small tax rate has a second-order effect on welfare and, moreover, the burden of the tax falls exclusively on bank owners. The above proposition may have important implications for the design of optimal taxes.23 Under perfect competition and constant returns to scale, taxing intermediaries is dominated by other forms of taxation (Diamond and Mirrlees 1971). However, if banks are local monopolists and able to price discriminate, then a tax on bank loans may be part of the optimal tax system, since it is a tax on pure economic profits and a small rate has only second-order effects on welfare (Caminal 1997). The relationship between monopoly power and the allocation of monitoring effort could be affected by a double moral hazard problem (Holmstrom and Tirole 1997; Besanko and Kanatas 1993). If banks cannot commit ex ante to monitor borrowers, then they need to be given incentives to exert efficient levels of monitoring effort. Inside capital is one possible source of incentives. Monopoly power is an- other one (Caminal and Matutes 1997). In the latter case if taxation falls on bank owners, then incentives to monitor may be jeopardized. TAXATION OF BANKS: MODELING THE IMPACT 107 Bank Solvency It is commonly agreed that banks’ moral hazard is one of the rele- vant factors behind banking failures. Because of limited liability and amplified by risk-insensitive deposit insurance, bank owners wish to take excessive risk. Direct supervision is not sufficient, and solvency requirements force banks to take on more capital, which involves a higher cost. It has been suggested that an efficient complementary measure is to protect banks’ profits either by relaxing competition policy in banking (Perotti and Suarez 2001) or by setting deposit in- terest rate ceilings (Hellmann, Murdock, and Stiglitz 2000). The re- lationship between market structure and risk-taking has been well documented empirically.24 When banks’ moral hazard problem is binding, then taxation is likely to affect incentives to prudent be- havior. However, different types of taxes may have different effects on risk-taking attitudes. To illustrate this point, consider what happens when certain changes are introduced to the monopoly power version of the model. First, deposit interest rates are set by the authorities. Second, banks can invest in two types of long-run assets. A safe asset yields a rate of return equal to R with probability one at time t = 2. In- stead, a risky asset yields at time t = 2 a rate of return equal to γ with probability θ, and 0 with probability 1 – θ. ASSUMPTION 8: γ > R > θγ Thus the risky asset yields a higher return in case of success— although from an ex ante point of view it is dominated by the safe asset (lower expected return and higher risk). Investors are not con- cerned about banks’ investment policies because deposits are as- sumed to be protected by a flat-fee deposit insurance system. For simplicity (and this is the third departure from the basic monopoly model above), bank owners are risk-neutral and care only about consumption at t = 2. Suppose that the government — sets a binding deposit rate ceiling, which is constant over time, r d < r M – µ. If at — time t = 1, banks do not pay back r d to impatient consumers, then the bank is intervened and owners obtain zero. Hence it is a domi- nant strategy for banks to invest in short-term assets such that they can meet their payment obligations at time t = 1. 1 d 1− β = (r + µ) 2 108 RAMON CAMINAL Banks can still choose whether to allocate the rest of the funds to the safe or to the risky asset. In case of prudent behavior, the amount of profits per unit of deposits is given by: rd + µ R +1 d π P = Rβ − =R− (r + µ) 2 2 If a bank chooses to invest in the risky asset, then expected profits per unit of deposits is:  rd + µ    π R = θ  γβ −  = θ  γ − γ + 1 (r d + µ)  2   2    Hence banks choose prudent behavior if and only if πP ≥ πR: that is, 2(R − θγ ) rd ≤ − µ ≡ r − µ < rM − µ R − θγ + 1 − θ Thus if the interest rate ceiling is below the threshold level, – r – µ, then the bank chooses to invest in the safe asset. Otherwise they take excessive risk. Since the threshold level is strictly below the competitive return on deposits, this implies that the regulator faces a tradeoff: either to grant banks a certain level of profits at the cost of lowering deposit rates and transaction services, or to induce banks to take excessive risk. Now consider the effect of a tax on deposits. Taxes and rate ceil- ings can be combined in two different — ways. First, the ceiling can be set on the rate that banks pay: r d, in our notation. In this case, clearly a tax falls entirely on investors and it does not affect banks’ –- Second, the ceiling can be set on the rate that risk-taking incentives. investors receive, i d = (1 – τ d)r d. In this case, the tax falls entirely on banks and if the following condition holds: id > r − µ ≥ id 1 − τd then taxes induce risk-taking. In the case that deposit rates are freely determined, a tax on de- posits would tend to fall on both investors and bank owners. As in Hellman, Murdock, and Stiglitz (2000), the reduction in profits in- duces banks to take excessive risk,25 although the analysis becomes more tedious because of the income effect in the supply of deposits. TAXATION OF BANKS: MODELING THE IMPACT 109 Now consider a tax on banks’ corporate income. In the absence of capital requirements banks will choose to hold zero capital26 and the returns from prudent and risky behavior will be respectively:  R +1 d  π P = (1 − τ c )R − (r + µ)  2   γ + 1  π R = (1 − τ c )θ  γ − (r d + µ)  2  Clearly, risk behavior is unaffected by taxation of banks’ corporate income.27 The next proposition summarizes the main message of this section. PROPOSITION 12. Whenever bank solvency requires positive eco- nomic profits, taxation may induce excessive risk-taking. For in- stance, a tax on deposits that increases banks’ costs of funds will make riskier portfolios relatively more attractive. However, a pro- portional tax on pure economic profits is neutral with respect to risk. Thus if the main priority of the government is to preserve bank solvency, a tax on deposits must be accompanied by a change in reg- ulation. There are two main options: higher capital requirements and additional market power. Capital requirements may not be ef- ficient and sometimes may even induce higher risk-taking.28 Alter- natively, the government can grant banks more monopoly power (through additional entry and branching restrictions, softer compe- tition policy, rate regulation, and so on). In other words, if the so- cial costs of banking failures are sufficiently large, then it may be optimal to allow banks to make a certain amount of profits, which implies that a tax on deposits must go along with additional regu- latory changes. As a result, the burden of taxation falls exclusively on banks’ customers. Some Reflections on the Assumptions and Wider Implications This section considers the extent to which relying on the assumption of separability weakens the empirical relevance of the model, and also discusses empirical evidence and macroeconomic implications. Separability between Loans and Deposits Revisited The incidence of bank taxation depends on whether there is sepa- rability between loans and deposits: that is, whether conditions in 110 RAMON CAMINAL one of the markets affect equilibrium prices in the other. The model developed in this chapter makes heavy use of the separability hy- pothesis. In particular, in the model banks have access to safe in- vestment technology, which represents simultaneously the expected return of the funds collected through deposit contracts, and the op- portunity cost of the loans to entrepreneurs. This is a convenient ar- tifact that requires further discussion. The literature tends to identify separability with the following set of conditions:29 • Banks are able to borrow and lend in a perfect bond market at an exogenous interest rate. This is the case, for instance, if there is an organized interbank lending market, whose interest rate is fixed either by monetary authorities or by arbitrage with international capital markets. • The costs of granting loans and capturing deposits are addi- tively separable. In other words, the marginal cost of loans is inde- pendent of the level of deposits, and vice versa. • The supply of deposits and the demand for loans are independent. • The probability of failure for banks is zero. Clearly, these conditions are quite restrictive. If the banking in- dustry is an oligopoly and the economy is closed (or capital mobil- ity is imperfect) then the first may not hold. Also, there may be economies of scope and the second may fail. Finally, as Dermine (1986) shows, if the bank’s probability of failure is positive, there is deposit insurance, and banks enjoy limited liability, then the (mo- nopolistic) bank’s optimization problem ceases to be separable. A higher deposit rate increases the probability of default, which re- duces the cost of lending one additional unit.30 The above discussion suggests that many real world situations are likely to deviate in various dimensions from separability. Never- theless, a model assuming separability may be more useful than it might appear at first sight. The reason is that simultaneous devia- tions may compensate each other to some extent. Consider the interaction between loans and deposits through costs. In particular, assume that monitoring costs are reduced with the level of deposits: that is, m is a function of D, with m'(D) < 0. This may be explained by the fact that banks use information from deposits in their loan-granting decisions (Fama 1985).31 In this case, the effects of a tax on deposits would clearly spill over into the loan market. In particular, an increase in the deposit tax rate would de- press deposits (unless investment funds are absent and bank deposits dominate direct investment), which would increase monitoring costs. TAXATION OF BANKS: MODELING THE IMPACT 111 As a result, loan rates would increase and some entrepreneurs would turn to the securities market to get their funding. Now consider the case of bankruptcy risk. Because of limited li- ability banks perceive a lower cost of funds as the probability of bankruptcy increases. As a result, banks become more aggressive in the credit market, pricing their loans at lower rates. In this context a tax on deposits falls partially on bank profits (the ex ante deposit rate increases) and hence the probability of bankruptcy increases, which pushes loan rates downward. Thus if both channels operate at the same time, it is unlikely that a tax on deposits leaves loan rates unaffected, but the size of the ef- fect may be small and, more importantly, of ambiguous sign.32 Empirical Evidence on the Effect of Taxation The empirical evidence on the effects of bank taxation is rather frag- mented. One important cross-country study is Demirgüç-Kunt and Huizinga (1999). Unfortunately, their information on the shape of the implicit and explicit tax schedules is rather imprecise, particu- larly in the former case. They find that reserves reduce interest mar- gins and profits, especially in less developed countries, which may reflect the fact that the opportunity cost of reserves (the implicit tax) is higher in less developed countries. They also present evidence fa- vorable to a complete pass-through of corporate income tax to bank customers. In fact, their regression analysis shows that both interest margin and profitability increase with the tax on corporate income. There are some earlier empirical studies on the incidence of re- serve requirements. Bartunek and Madura (1996) study the effects of two unique reserve requirement adjustments in the early 1990s. The results indicate that not all of the benefit from the reduction in the implicit tax resulting from the lowering of reserve requirements was passed on to depositors and borrowers. The larger banks tended to experience strong favorable valuation effects.33 The existing evidence on the incidence of reserve requirements (Bar- tunek and Madura 1996) seems compatible with the results above. Provided size is positively related to market power, shareholders of large banks are expected to capture a larger share of the benefits from the reduction in reserve requirements. Depositors are also expected to benefit from such a reduction. However, the evidence on banks’ corporate income taxes (Demirgüç- Kunt and Huizinga 1999) does not fit well into the picture. First, the positive effect of taxation on profitability is difficult to reconcile with any sensible economic model. Second, the complete pass-through re- 112 RAMON CAMINAL sult suggests that banking is a perfectly competitive industry (other- wise such a tax would fall on economic profits), which is at odds with most of the evidence. Macroeconomic Implications The fact that reserve requirements are an implicit form of taxation has long been recognized.34 Such a requirement induces banks to hold a higher fraction of their deposits in the form of non-interest-bearing reserves, which reduces the average return of banks’ portfolios and increases their demand for monetary base. Thus the reserve require- ment operates as a tax by increasing seigniorage revenue (by expand- ing the base of the inflation tax). More specifically, under certain as- sumptions, a reserve requirement is equivalent to a proportional tax on deposits plus an open market sale of government bonds of an amount equivalent to the volume of resources kept captive by the re- quirement (Romer 1985; Bacchetta and Caminal 1994). Thus the non-monetary model was unable to accommodate a pre- cise description of reserve requirements and hence the interaction be- tween banking activity, inflation, and the government budget con- straint was missing. This is an important limitation for at least two reasons. First, in practice inflation is only imperfectly controlled by monetary authorities. As a result, the implicit tax is likely to have an arbitrary (random) component. Second, the inflation tax affects not only bank deposits but also one of their imperfect substitutes: cash. Thus a complete analysis of the effect of reserve requirements on the level of deposits should consider how inflation and reserve ratios af- fect the choice between cash and deposits; otherwise the welfare im- plications may be misleading (Whitesell 1992). In particular, if re- serve requirements are interpreted as a tax on deposits, then the model here is likely to overstate their distortionary effects. A related issue is the modeling of transaction services. This is bound to be important for a normative analysis but not so much for the positive analysis conducted in this chapter. This study treated transaction services provided by banks as an additional consump- tion good by including them as an argument of investors’ utility function. Alternatively, it could have been assumed that deposits re- duce transaction costs. Results on optimal taxation may strongly depend on such a choice.35 The model analyzed in this chapter has focused on the role of banks in the allocation of savings and abstracted from the substi- tutability between capital and labor and from the determination of the level of savings. In dynamic general equilibrium models, capital TAXATION OF BANKS: MODELING THE IMPACT 113 taxation creates a gap between the rate of return on savings and the marginal product of capital. The optimality of capital income taxa- tion depends on various circumstances. In infinitely lived representa- tive consumer models, it is well known that capital income taxes cannot be part of the optimal tax system in steady state (Chamley 1986).36 Bank taxes are also capital income taxes and on top of that they affect the efficiency of the allocation of funds to various invest- ment opportunities. Thus if our model is incorporated in a standard general equilibrium framework, then one would expect to obtain the same results as under standard capital income taxes, augmented by the additional distortionary effects that the study has identified.37 There is another body of literature, both empirical and theoreti- cal, that relates financial intermediation and long-run growth. Some of these models characterize banks as institutions that either pro- vide liquidity services or alleviate informational asymmetries.38 In some cases, the engine of growth is a capital externality, following Romer; in others, it is innovation effort. Once again, the role of banks can be described as reducing the gap between the return on savings and the marginal return on investment. Externalities in re- search and development or in capital accumulation transform level effects into growth effects. The empirical evidence is compatible with a positive relationship between financial development and eco- nomic growth, although it is more difficult to establish the direction of causality.39 Overall, this literature suggests that bank taxation could have negative implications for long-run growth.40 Concluding Remarks This chapter has developed a theoretical framework to analyze the impact of various forms of taxation. Banks have been characterized as intermediaries able to perform three main functions: asset trans- formation, provision of transaction services, and monitoring. In- vestors can directly access security markets, but the resulting port- folios are relatively illiquid. In addition, they may or may not have access to investment funds, which also perform an asset transfor- mation function. On the other side of market, some entrepreneurs may have access to security markets (unmonitored credit), but those entrepreneurs with high-risk projects prefer information-intensive credit (bank loans) over uninformed credit (commercial paper). The total amount of funds is exogenous, and hence the model focuses on the allocation of those funds among alternative financial arrange- ments. The main results can be stated as follows: 114 RAMON CAMINAL • Under certain conditions, the loan and deposit markets are sep- arable, and hence a tax on deposits does not affect either loan rates or monitoring activity. Similarly, a tax on loans does not affect either deposit rates or the provision of transaction services and liquidity in- surance. The benchmark case where all these conditions are met may be more useful than it appears at first sight since multiple deviations from the benchmark may partially compensate one another. Some of the results listed below presume that separability holds. • If borrowers are VAT-registered firms and investors are not, then a value-added tax is equivalent to a tax on deposits. If loans and bank capital are complementary (which occurs, for instance, when capital requirements are binding) then a tax on banks’ corporate in- come is equivalent to a tax on both economic profits and loans. • A tax on deposits tends to reduce the level of deposits and in- crease direct investment and/or investment fund holdings. As a re- sult the level of transaction services is reduced and, if investment funds do not exist, investors enjoy a lower level of liquidity insur- ance. In the absence of investment funds, it may be the case that de- posits dominate direct investment and a small tax rate does not cre- ate any distortion. Imperfect competition in the deposit market has two effects. First, it distorts the laissez-faire equilibrium and hence even a small tax rate has a first-order effect on total welfare. Sec- ond, taxes fall partially on economic profits, which alleviates the tax burden of investors (and potentially reduces the distortion on savings decisions). • A tax on bank loans reduces the amount of lending and mon- itoring effort and induces more firms to borrow on the securities market. The size of the distortion is likely to increase with the effi- ciency of the securities market. The effect of banks’ market power in the loan market is similar to that in the deposit market, except that banks’ ability to price-discriminate is higher in this market. In the extreme case that individual banks compete exclusively with capital markets, then a tax on bank loans may fall exclusively on bank owners. • A general tax on capital income reduces the level of deposits, since investors’ willingness to pay for banks’ transaction services in- creases with their disposable income. • A tax on deposits may induce banks to invest in riskier port- folios. If bank solvency is the top priority, then such a tax must be introduced only if at the same time banks are given more monopoly power. In this case the tax would fall entirely on banks’ customers. Summarizing, bank taxation increases the gap between the mar- ginal return on investment and the marginal return on savings (like general capital income taxation). On top of that, it tends to reduce TAXATION OF BANKS: MODELING THE IMPACT 115 the efficiency of the transformation of savings into investment by re- ducing banks’ contributions. However, if banks’ market power is excessive (that is, beyond the level required by the goal of preserv- ing the stability of the banking system) then the overall distortion associated with bank taxation may substantially shrink. One important limitation of the current discussion is that it has overlooked tax enforcement problems. In particular, if foreign fi- nancial intermediaries are available, then evasion could be relatively easy and the impact of taxing domestic financial intermediaries may be substantially altered. On the other hand, taxes on financial in- termediaries may induce more cash transactions, reduce financial records within intermediaries, and make monitoring firm income more costly for tax authorities. This may undermine the collectibil- ity of other taxes. These issues are beyond the scope of this chapter. Other extensions of the current framework appear to be fruitful. First, a complete welfare analysis of the disintermediation effects of taxation requires an explicit consideration of alternative means of payment (cash) and potential informational spillovers from bank monitoring to security markets. Second, it would be nice to trans- form the current model into a building block of a standard dynamic general equilibrium model in order to develop a more complete pic- ture of the aggregate effects of taxation. Appendix Market Clearing under Direct Investment At time t = 0 an amount of funds I N are invested in long-term assets. All entrepreneurs with p ≤ p0 obtain financing if the following condition on the probability distribution H holds: 1 – H(p0) = L < I N There is no need to worry about market clearing in the secondary mar- ket since both buyers and sellers are indifferent between trading in the sec- ondary market and in alternative assets. However, it is important that claims on entrepreneurs’ projects are not liquidated. In particular, the fol- lowing condition must hold: 1 1 R[1 − H(p0 )] < N(1 − I ) 2 2 The left-hand side is the amount of claims on entrepreneurs’ projects that must be sold in the secondary market and the right-hand side is the 116 RAMON CAMINAL total demand for claims. In the text it has been shown that I > 0. Also, if consumer’s degree of risk aversion is not too small, then I < 1. In other words, investors always have a diversified portfolio (they hold both short- term and long-term assets). Therefore, provided N is large enough the above two conditions will be satisfied. The Objective Function of Oligopolistic Banks Suppose that bank shareholders are subject to the same liquidity shocks as investors: that is, half the shareholders will turn out to be impatient and the other half will turn out to be patient. To accommodate such liquidity shocks, it can be arranged that patient shareholders buy the shares of those who claim to be impatient at time t = 1 at a prespecified price. Let π1 de- note the payment received by impatient shareholders per unit of deposits. Thus each impatient shareholder of bank i receives 2π1Di. Similarly, let π2 denote the payment received by patient shareholders per unit of deposits at time t = 2; that is, each patient shareholder receives 2π2Di. If bank i com- mits to pay depositors (rid 1, ri2), then it must invest a fraction 1 – β‚ on short- d term assets: 1 d (A3.1) 1− β = (ri1 + µ) + π1 2 Also, the budget constraint at time t = 2 determines residual profits: 1 d (A3.2) π2 ≡ βR − (ri 2 + µ) 2 Solving equation A3.1 for β and plugging it into equation A3.2, yields the shareholders’ intertemporal budget constraint:  rd rd  R+1 (A3.3) π2 = R1 − i1 − i 2 − π1  − µ  2 2R  2 Under the assumption that banks cannot affect investors’ consumption (see note 16) and if other banks and investment funds set constant interest rates, then the supply of funds to bank i is given by:  1  R+1 (A3.4) Di = a − bu′(c d ) ∆ i1 + ∆ i 2  + k u′(c d )∆j  R  R where cd is the constant level of consumption by investors, i ≠ j, and TAXATION OF BANKS: MODELING THE IMPACT 117 ∆it = r M – (1 – τ)rit d Finally, if ∆i1 = ∆i2 then such a constant level is denoted by ∆i. Thus given the constant interest rate set by the rival bank, rjd, bank i ’s shareholders choose rid1, ri2, π1 and π2 in order to maximize: d s 1 s U = u(c1 )+ u(c2 ) R where s = 2π D c1 1 i s = 2π D c2 2 i and subject to the budget constraint of equation A3.3 and Di given by equa- tion A3.4. From the first-order conditions it follows that: rid d d 1 = ri2 = ri π1 = π2 As a result, bank i’s shareholders also enjoy a constant level of consump- tion, given by: c s = (rM – µ – rid)Di This expression is the starting point of the analysis in the text. Notice that the above time structure of shareholder returns makes the plan incentive compatible, since patient shareholders have no incentive to pretend to be impatient. Taxation and the Oligopolistic Deposit Rate It has been assumed that individual banks cannot affect aggregate con- sumption, but clearly taxes do affect consumption. Thus the effect of taxes on the deposit rate cannot be computed by simply taking the partial deriv- ative in equation A3.4. Let vi(DA, DB) denote the partial derivative with respect to Di. Then: vi(DA, DB)= α – Di – λDj 118 RAMON CAMINAL where i ≠ j. The previous assumption can be adapted on the saturation point by assuming that in a symmetric allocation the marginal utility of in- vesting all the funds in deposits is zero:  1 1 1 vi  ,  = α − (1 + λ) = 0  2 2 2 Hence α = 1/2. The equilibrium level of consumption in a symmetric equilibrium can be written as: (A3.5) c = r M – 2(r M – i d)D where D is the level of deposits at each bank: that is 1+ R (A3.6) D = a − (b − k) u′(c )(r M − i d ) R Totally differentiating equations A3.5 and A3.6 and solving yields:  1+ R  2 a − 2(b − k) u ′(c)(r M − i d ) R =   dc (A3.7) di d 1+ R M d 2 1 − 2(b − k) u ′′(c)(r − i ) R The denominator is positive but the sign of the numerator is ambiguous. A change in the after-tax interest rate makes the relative price of deposits cheaper (substitution effect), which induces less consumption, but also ex- pands the feasible set (income effect), which induces higher consumption. From equation 3.21 in the text, it follows that a b(r M − µ)(1 − τd ) + (b − k)r M − (A3.8) d i = Ω (2b − k) Applying the implicit function theorem to this equation, one can compute the effect of taxation on the after-tax interest rate: di d −b(r M − µ) = dτ d au′′(c ) dc (2b − k) − Ω di d Using equation A3.7, one can show that the denominator has a positive sign: TAXATION OF BANKS: MODELING THE IMPACT 119  1+ R  (2b − k)1 − 2(b − k) u′′(c)(r M − i d )2  −  R  2 au′′(c) Ω [ M d a − 2(b − k)Ω(r − i ) > 0 ] The first three terms are positive and the fourth is negative. However, the sign of the last three terms is equal to the sign of: a2 1+ R M d 2 Γ≡ + (b − k)(2b − k) (r − i ) − 2a(b − k)(r M − i d ) Ω R Since 2 − λ 2b − k u′(c ) < 1 + λ < ≡ 1− λ b−k it follows that: 2  R 1+ R  Γ > a − (r M − i d ) (b − k)(2b − k) u ′(c)  > 0  1+ R R  Therefore it has been shown than a higher tax rate implies a lower after-tax deposit rate. Turn now to the effect of taxation on the pre-tax deposit rate, r d. Totally differentiating equation 3.21 with respect to r d, c, and τ yields: a a u′′(c) (b − k)r M − d Ω Ω u′(c) dr = dτ + dc (2b − k)(1 − τ d )2 (2b − k)(1 − τ d ) From the definition of i d, it follows that dc = dc di d [(1 − τ )dr d d − r d dτd ] where dc/did is given by equation A3.7. Hence dr d M0 d = dτ M1 where 120 RAMON CAMINAL a (b − k)r M − M0 ≡ Ω − a u ′′(c) r d dc 1 − τd Ω u ′(c) di d a u ′′(c) dc M1 ≡ (2b − k)(1 − τ d ) − (1 − τ d ) d Ω u ′(c) di Notice that if dc/di d > 0, then dr d/dτ d > 0. However, in principle the sign of dc/di d is ambiguous. In fact, such ambiguity is easily translated into the sign of dr d/dτ . In general, from equation A3.8, it follows that: a b[τr M + (1 − τ)µ] + (A3.9) M r −i = d Ω 2b − k Now look at two extreme cases. First, suppose µ = τ = 0. If one were to plug equation A3.9 into equation A3.7, evaluated at µ = τ = 0, it can be checked that dc/di d > 0, and hence dr d/dτ d > 0. By continuity, provided µ and τ are not too large, then a proportional tax on deposits raises the before-tax deposit rate. Second, suppose that banks are monopolists: that is, λ = 0. Consider the case τ d = 0. This implies that b = 1 and k = 0. As a result, dc −2Ωµ d = 2 <0 di 1+ R  1  1− u ′′(c) µ +  2R  2Ω  The numerator of dr d/dτ can be written as follows:  1  u ′′(c) µr d M0 = r M 1 − +  4u ′(c)  u ′(c) 1+ R  1  2 1− u ′′(c) µ +  2R  2Ω  For any value of µ > 0, there exists a utility function such that 4u'(c) is higher but arbitrarily close to one. As a result, the first term is positive but arbitrarily small, while the second term is negative. In fact, as 4u'(c) goes to one, the first term goes to zero, while the second goes to a strictly nega- tive number. Finally, the denominator of dr d/dτ can be written as follows: u′′(c ) µ M1 = 2 − u′(c ) 1+ R  1  2 1− u′′(c ) µ +  2R  2Ω  TAXATION OF BANKS: MODELING THE IMPACT 121 Therefore, provided 4u'(c) is close enough to one and I is positive but not too large, then M1 is positive and M0 is negative. In this case, a higher tax rate induces a lower pre-tax deposit rate. Notes 1. Recent changes in the size and composition of bank activities are dis- cussed, for instance, in Mishkin (1996) and Allen and Santomero (2001). 2. Thus as in Diamond and Dybvig (1983), investors’ preferences are extreme: the utility of consumption is positive either at t = 1 or at t = 2. 3. It makes sense to assume that transaction services are enjoyed the same period that consumption takes place (impatient consumers at time t = 1, and patient consumers at time t = 2). Banks pay the costs of providing those services accordingly. Finally, the function v represents the expected value of deposits. 4. It is easy to check that, provided f is not too large, entrepreneurs’ par- ticipation constraint is not binding. In other words, all entrepreneurs that can get financing are willing to accept it. 5. A reserve requirement is equivalent to a proportional tax on deposits, except for the timing of revenues. If banks can price transaction services separately, then it is easy to show that a proportional tax on deposits is equivalent to a proportional tax on transaction services. That is, if both taxes are required to raise the same amount of revenue, then they must cause the same distortion on the level of deposits. 6. Empirical support is provided by James (1987). 7. In the absence of the externality, a small τ l has only a second-order effect on welfare. However, if banks create a positive externality to the se- curities market, even a small tax would have a first-order effect on welfare. 8. See Poddar and English (1997). 9. Under the cash flow method, borrowers should pay τ ν on their capi- tal inflow at time t = 0, and then get a credit of τ νr l on their capital outflow at time t = 2. This discussion assumes that the tax payment associated with a cash inflow of a capital nature can be carried forward to the period dur- ing which the capital transaction is reversed, although the deferral is sub- ject to interest charges at the market rate, R. This is the cash flow method with Tax Calculation Account (Poddar and English 1997). 10. One may wonder about the effect of a general tax on firms’ profits. In the present model, the demand for credit is inelastic and hence such a tax would create no distortion. However, in a richer model, the amount of en- trepreneurial projects may also depend on their net return. In that case, profit taxes are likely to reduce bank loans. 11. One reason why banks find equity financing expensive (sometimes prohibitively expensive) is adverse selection (Stein 1998). Boyd and Gertler 122 RAMON CAMINAL (1993) observe that for the U.S. banking industry, the equity to assets ratio has been steadily declining since 1980. 12. Solvency ratios usually take the form of a minimum ratio of capital to a risk-weighted measure of assets. The present formulation gives a posi- tive weight only to loans to entrepreneurs. 13. Investors will be willing to buy banks’ equity at an implicit rate of return R only if deposits do not dominate direct investment: that is r M – µ – 1 < 0. This subsection deals only with this case. 14. In case capital requirements are computed in such a way that all bank assets get a positive weight, and provided deposits and equity are the only source of funds for banks, then it can readily be shown that separa- bility breaks down and corporate taxes also affect deposit rates. 15. However, Whitesell (1992) argues that money market funds may pro- vide efficient services for very large transactions, but that cash and checks are more efficient for small and medium-size transactions, respectively. 16. The empirical evidence seems to indicate that the emergence of pen- sion and mutual funds has mostly affected directly held assets, but not so much bank assets. See the discussion in Allen and Santomero (2001). 17. Market power may be partly created by regulation (entry and branch- ing restrictions, interest rate ceilings, and so on). It has been shown that deregulation and liberalization have increased competition and reduced bank profits. See Keeley (1990) and Demirgüç-Kunt and Detragiache (1999). Other traditional factors may be crucial to explain banks’ market power in the household and small business sector, including geographic differentiation, specialization, and customer-specific investments. 18. The representative investor can be thought of as a coalition of mul- tiple agents, each one in charge of a small fraction of total savings and in- structed to maximize the coalition’s objective function, anticipating that all returns are pooled before consumption takes place. This is somewhat anal- ogous to the celebrated assumption in Lucas (1990). 19. This is actually an assumption on the marginal utility of consump- tion, since this is equivalent to 4u'(c) > 1 – λ. 20. In the limit as λ goes to one (Bertrand competition), the deposit rate converges r M – µ (the perfectly competitive deposit rate). Hence drd/dτ d = 0 and taxes fall entirely on investors. 21. One possible justification could be that banks and entrepreneurs have been engaged in a long-run relationship and at a point in time the in- cumbent bank has an advantage in monitoring previous customers, which is sufficiently large to imply complete segmentation of the loan market. 22. For those borrowers with p < po the monopolist will set a loan rate r l = (1 – τ)X. 23. As in the case of perfect competition, in this context a tax on banks’ corporate income, provided capital requirements are binding, is equivalent to a tax on bank loans. Also, a tax on banks’ value-added, under the same conditions, is exclusively a tax on deposits. TAXATION OF BANKS: MODELING THE IMPACT 123 24. Keeley (1990) blames the decline of charter values due to liberaliza- tion and deregulation for the increase in failures in the U.S. banking indus- try since the 1980s. Demirgüç-Kunt and Detragiache (1998) also point at liberalization as one of the factors that explain banking crises in a large set of countries. Inappropriate regulation accompanying liberalization seems to aggravate crises. 25. However, there is a countervailing effect. Lower bank rents may dis- courage bank monitoring and increase credit rationing. As a result, banks’ portfolios may be less exposed to aggregate risk. See Caminal and Matutes (2002). The relation between market structure, regulation and financial fragility is also studied in Matutes and Vives (1996 and 2000). 26. The argument below is robust to the introduction of capital re- quirements. More specifically, capital requirements are likely to reduce risk- taking, but their presence does not alter the neutrality of a tax on profits. 27. In general the treatment of loss offsets may affect banks’ risk-taking behavior. However, in this particular model, if the tax law allows a rebate in case of default, nothing changes since bank owners cannot appropriate the rebate. 28. See Hellmann, Murdock, and Stiglitz (2000) and references quoted there. 29. See Freixas and Rochet (1997) for a more detailed discussion. 30. If banks offer “tied-up contracts” (consumers can obtain credit from a bank only if they deposit their cash in the same bank, or they get a lower loan rate if they do so) then the third condition fails. Chiappori, Pérez-Castrillo, and Verdier (1995) show in the context of the Salop model that in an unregulated banking sector this type of contract would never emerge, but it does if deposits are subject to interest rate ceilings (in this case banks are willing to subsidize credit). 31. See also Mester, Nakamura, and Renault (1998) for empirical evidence. 32. If the interest rate on the safe asset is determined by domestic sup- ply and demand conditions, and in the absence of government intervention, then a tax on deposits will tend to increase the loan rate. The size of this ef- fect depends on the elasticity of savings with respect to their average return and on the weight of deposits in total savings. 33. For similar evidence, see Osborne and Zaher (1992). The earlier studies on the incidence of reserve requirement are discussed in Demirgüç- Kunt and Huizinga (1999). 34. See, for instance, Fama (1980). 35. See Kimbrough (1989) and Chia and Whalley (1999). 36. Results are different if agents have finite horizons (Erosa and Ger- vais 2002). 37. Some of the literature on banking in general equilibrium has fo- cused on the role of banks on the propagation of shocks and on the trans- mission of monetary policy. See, for instance, Smith (1998) and Fuerst 124 RAMON CAMINAL (1994). From an international perspective, the ability to tax the banking in- dustry may be drastically reduced after the liberalization of the domestic fi- nancial system. See Bacchetta and Caminal (1992). 38. An example of the first type is Bencivenga and Smith (1991). Green- wood and Jovanovic (1990) and De la Fuente and Martin (1996) belong to the second group. 39. Levine (1997) provides an excellent survey of the main theoretical arguments as well as of the empirical evidence. 40. See Roubini and Sala-i-Martin (1995) for a model with this type of result. References Allen, Franklin, and Anthony Santomero. 2001. “What Do Financial In- termediaries Do? Journal of Banking and Finance 25 (2): 271–94. Bacchetta, Philippe, and Ramon Caminal. 1992. “Optimal Seigniorage and Financial Liberalization.” Journal of International Money and Finance 11 (6): 518–38. ———. 1994. “A Note on Reserve Requirements and Public Finance.” In- ternational Review of Economics and Finance 3 (1): 107–18. Bartunek, Ken, and Jeff Madura. 1996. “Wealth Effects of Reserve Re- quirement Reductions in the 1990s on Depository Institutions.” Review of Financial Economics 5 (2): 191–204. Bencivenga, Valerie, and Bruce Smith. 1991. “Financial Intermediation and Endogenous Growth.” Review of Economic Studies 58 (2): 195–209. Besanko, David, and George Kanatas. 1993. “Credit Market Equilibrium with Bank Monitoring and Moral Hazard.” Review of Financial Studies 6 (1): 213–32. Boyd, John, and Mark Gertler. 1993. “US Commercial Banking: Trends, Cycles and Policy.” In Olivier Blanchard and Stanley Fischer, eds., NBER Macroeconomics Annual. Cambridge Mass.: MIT Press. Caminal, Ramon. 1997. “Financial Intermediation and the Optimal Tax System.” Journal of Public Economics 63 (3): 351–82. Caminal, Ramon, and Carmen Matutes. 1997. “Can Competition in the Credit Market Be Excessive?” CEPR Working Paper 1725. London: Centre for Economic Policy Research. ———. 2002. Market Power and Banking Failures. International Journal of Industrial Organization 20 (9): 1341–61. Chamley, Christophe. 1986. “Optimal Taxation of Capital in a General Equilibrium with Infinite Lives.” Econometrica 54 (3): 607–22. Chia, Ngee-Choon, and John Whalley. 1999. “The Tax Treatment of Fi- nancial Intermediation.” Journal of Money, Credit, and Banking 31 (4): 704–19. TAXATION OF BANKS: MODELING THE IMPACT 125 Chiappori, Pierre-André, David Pérez-Castrillo, and Thierry Verdier. 1995. “Spatial Competition in the Banking System, Localization, Cross-subsi- dies and the Regulation of Interest Rates.” European Economic Review 39 (5): 889–919. De la Fuente, Angel, and Jose Marín, 1996. “Innovation, Bank Monitoring, and Endogenous Financial Development.” Journal of Monetary Eco- nomics 38 (2): 269–301. Demirgüç-Kunt, Aslı, and Enrica Detragiache. 1999. “Financial Liberaliza- tion and Financial Fragility.” In Boris Pleskovic and Joseph E. Stiglitz, eds., Proceedings of the 1998 World Bank Conference on Development Economics. Washington, D.C.: World Bank. Demirgüç-Kunt, Aslı, and Harry Huizinga. 1999. “Determinants of Com- mercial Bank Interest Margins and Profitability: Some International Ev- idence.” World Bank Economic Review 13 (2): 379–408. Dermine, Jean. 1986. “Deposit Rates, Credit Rates and Bank Capital: The Monti-Klein Model Revisited. Journal of Banking and Finance 10 (1): 99–114. Diamond, Douglas. 1997. “Liquidity, Banks, and Markets.” Journal of Po- litical Economy 105 (5): 928–56. Diamond, Douglas, and Philip Dybvig. 1983. “Bank Runs, Deposit Insur- ance, and Liquidity.” Journal of Political Economy 91 (3): 401–19. Diamond, Peter, and James Mirrlees. 1971. “Optimal Taxation and Public Production, Part I: Production Efficiency, Part II: Tax Rules. American Economic Review 61 (1): 8–27 and 61 (2): 261–78. Erosa, Andrés, and Martin Gervais. 2002. “Optimal Taxation in Life-Cycle Economies.” Journal of Economic Theory 105 (2): 338–69. Fama, Eugene. 1985. “What’s Different about Banks? Journal of Monetary Economics 15 (1): 29–40. Freixas, Xavier, and Jean-Charles Rochet. 1997. Microeconomics of Bank- ing. Cambridge, Mass.: MIT Press. Fuerst, Timothy. 1994. “Monetary Policy and Financial Intermediation.” Journal of Money, Credit, and Banking 26 (3): 362–76. Greenwood, Jeremy, and Boyan Jovanovic. 1990. “Financial Development, Growth, and the Distribution of Income.” Journal of Political Economy 98 (5): 1076–1107. Hellmann, Thomas, Kevin Murdock, and Joseph E. Stiglitz. 2000.“Liberal- ization, Moral Hazard in Banking, and Prudential Regulation: Are Capi- tal Requirements Enough?” American Economic Review 90 (1): 147–65. Holmstrom, Bengt, and Jean Tirole. 1997. “Financial Intermediation, Loan- able Funds, and the Real Sector.” Quarterly Journal of Economics 112 (3): 663–91. ———. 1998. “Private and Public Supply of Liquidity.” Journal of Politi- cal Economy 106 (1): 1–40. 126 RAMON CAMINAL Jacklin, Charles. 1987. “Demand Deposits, Trading Restrictions, and Risk Sharing.” In E. Prescott and N. Wallace, eds., Contractual Arrangements for Intertemporal Trade. Minneapolis: University of Minnesota Press. James, Christopher. 1987. “Some Evidence on the Uniqueness of Bank Loans.” Journal of Financial Economics 19 (2): 217–35. Keeley, Michael. 1990. “Deposit Insurance, Risk, and Market Power in Banking.” American Economic Review 80 (5): 1183–1200. Kimbrough, Kent. 1989. “Optimal Taxation in a Monetary Economy with Financial Intermediaries.” Journal of Macroeconomics 11: 493–511. Levine, Ross. 1997. “Financial Development and Economic Growth: Views and Agenda.” Journal of Economic Literature 35 (2): 688–726. Lucas, Robert. 1990. “Liquidity and Interest Rates.” Journal of Economic Theory 50 (2): 237–64. Matutes, Carmen, and Xavier Vives. 1996. “Competition for Deposits, Fra- gility, and Insurance.” Journal of Financial Intermediation 5 (2): 184–216. ———. 2000. “Imperfect Competition, Risk-taking, and Regulation in Banking.” European Economic Review 44 (1): 1–34. Mester, Loretta, Leonard Nakamura, and Micheline Renault. 1998. “Check- ing Accounts and Bank Monitoring.” Working Paper 98/25. Federal Re- serve Bank of Philadelphia. Mishkin, Frederic. 1996. “Bank Consolidation: A Central Banker’s Per- spective.” NBER Working Paper 5849. Cambridge, Mass. Osborne, Dale K., and Tarek S. Zaher. 1992. “Reserve Requirements, Bank Share Prices and the Uniqueness of Bank Loans.” Journal of Banking and Finance 16 (4): 799–812. Perotti, Enrico, and Javier Suarez. 2001. “Last Bank Standing: What Do I Gain If You Fail?” University of Amsterdam. Processed. Poddar, Satya, and Morley English. 1997. “Taxation of Financial Services under a Value-Added Tax: Applying the Cash-Flow Approach.” Na- tional Tax Journal 50 (1): 89–112. Romer, David. 1985. “Financial Intermediation, Reserve Requirements, and Inside Money: A General Equilibrium Analysis.” Journal of Mone- tary Economics 16 (1): 175–94. Roubini, Nouriel, and Xavier Sala-i-Martin. 1995. “A Growth Model of Inflation, Tax Evasion, and Financial Repression.” Journal of Monetary Economics 35 (2): 275–301. Smith, R. Todd. 1998. “Banking Competition and Macroeconomic Perfor- mance.” Journal of Money, Credit, and Banking 30 (4): 793–815. Stein, Jeremy. 1998. “An Adverse-Selection Model of Bank Asset and Lia- bility Management with Implications for the Transmission of Monetary Policy.” Rand Journal of Economics 29 (4): 466–86. Whitesell, William. 1992. “Deposit Banks and the Market for Payment Media.” Journal of Money, Credit, and Banking 24 (4): 483–98. 4 Tax Incentives for Household Saving and Borrowing Tullio Jappelli and Luigi Pistaferri Modern theories of intertemporal consumption choice emphasize that individuals may save for a variety of motives: to smooth life- cycle fluctuations in income (the retirement, or life-cycle motive), to face emergencies arising from income or health risks (the precau- tionary motive), to purchase durable goods and housing, and to ac- cumulate resources for one’s heir (the bequest motive) (Browning and Lusardi 1996). Individual choice may be affected by the government policies that, in virtually all countries, target private saving. Government targeting is selective, and tends to affect not only the overall level of saving but also the allocation of saving among its many different forms. Raising the overall level of saving is often viewed as an effective way to raise investment and growth. Many forms of government in- tervention thus aim at simply increasing the volume of saving, but leave the ultimate decision about the allocation of saving to the in- dividual. In other cases government intervention mandates individ- uals to save in specific forms or for specific purposes. For instance, in almost all countries governments promote retirement saving, be- cause having insufficient resources during retirement entails a high burden not only for the elderly lacking these resources, but also for society as a whole. Promoting saving for housing and other goods to which policymakers assign high priority (education, health, or life protection) is also a popular goal. This chapter reviews the literature 127 128 TULLIO JAPPELLI AND LUIGI PISTAFERRI on these tax incentives, with special focus on long-term saving, housing, and household liabilities. In very poor countries households rely on informal markets for credit transactions, so government intervention has a limited role in shaping household saving and portfolio allocations. The chapter therefore places special emphasis on the importance of tax incen- tives and saving instruments available in middle-income countries with relatively developed financial markets (as in several Latin American and East Asian countries). It is precisely in middle-income countries that mandatory saving instruments are more widespread, and often the only effective way of raising the overall level of sav- ing and shaping household portfolios. In this area, the most devel- oped countries have accumulated a wide experience in designing and implementing various tax incentives schemes. This experience can be used to evaluate the effectiveness of tax incentives and to draw lessons for middle-income countries. A careful review of the international tax codes reveals that in most middle-income countries the tax system targets long-term, re- tirement saving instruments. This is hardly surprising given the role of retirement saving as the most important and widely available household financial asset. The tax features of pension funds are of special interest, given the recent wave of reforms of the social secu- rity system in Latin America and East Asia. Almost invariably, man- dated contributions to pension funds are tax exempt, and very often voluntary contributions to long-term saving instruments are also heavily favored by the tax code. This chapter therefore concentrates mainly on mandated contributions to pension funds, although it also devotes some space to the tax treatment of other, more “sophisti- cated” assets available in industrialized countries. The second area of widespread government intervention in middle-income countries is incentives to save for housing accumulation plans. These pro- grams are absent in industrialized countries, but quite common in several middle-income countries. The chapter is divided into five parts, each addressing a particular saving instrument and area of policy intervention. The first section begins by reviewing the interest rate effect on personal saving. The specific question addressed is whether public policies affecting the real rate of interest impact the overall level of saving. The second section examines the effect of tax incentives on long-term manda- tory saving programs. In the absence of tax incentives, mandated as- sets are a substitute for private accumulation and should not affect national savings. However, the tax deductibility of mandated con- tributions present in almost all countries can be an effective way of influencing not only the composition of wealth, but also the overall level of saving. On this front, the chapter presents international evi- TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 129 dence based on regression analysis showing a positive association between the national saving rate and the stock of mandated assets in household portfolios. The third section examines government programs that target sav- ing for home purchase. Direct subsidies to home mortgages, de- ductibility of mortgage interest payments, and reduced loan rates are just a few examples of government intervention in this area. But the most striking instruments are mandatory contributions to prov- ident funds designed to accumulate resources for a down payment against home purchase. The interplay between saving for retirement and saving for housing accumulation is also of special interest in this area. The fourth section explores government programs that target saving for health and education, while the fifth section ana- lyzes the effect of tax incentives to borrow, rather than to save. For each of these five important issues, the chapter examines em- pirical evidence on the main characteristics of government programs, with a special focus on middle-income countries. It also addresses a number of issues that should be of interest to policymakers. First, on which grounds should government policy target some assets rather than others? Second, do tax-sheltered assets and liabilities lead to substitution away from more heavily taxed savings instruments or do they affect the overall level of saving? And finally, is there any les- son that can be drawn from the experience of developed countries for the design of saving and borrowing incentives in middle-income countries? Three simple models in the appendix summarize the effect on saving of the interest rate, mandated pension contributions, and tax incentives to borrow. Incentives for Voluntary Retirement Saving The interest rate effect on personal saving has attracted a long tra- dition of research, both theoretically and empirically. The specific question that this literature addresses is whether public policies af- fecting the real rate of interest affect the overall level of saving. Al- though taxing the return to saving can have a strong effect on household asset selection and allocation, theoretical models, such as the standard two-period model of intertemporal choice presented in the first part of the appendix, are ambivalent regarding the effect on the overall level of saving, even when one considers models with un- certainty and precautionary saving (Bernheim 2002).1 Nor has the empirical literature been able to pin down this effect. The empirical literature on saving and taxation has grown tremendously in recent years. Bernheim (2002), Besley and Meghir (2001), Honohan (2000), and Poterba (2000) provide surveys of the 130 TULLIO JAPPELLI AND LUIGI PISTAFERRI most recent developments, and interested readers are encouraged to refer to these excellent contributions. This chapter confines itself to summarizing the main empirical strategies that have been used to test the interest rate effect on saving. They can be broadly divided into three groups: estimation of saving function with time-series or cross-country data, Euler equation estimates of the intertemporal elasticity of substitution, and analysis of specific tax reforms. The earliest approach consists of specifying a saving function and estimating the interest rate elasticity controlling for other determi- nants of saving. Many such studies have been performed with time- series data from individual countries, and with cross-country or panel data from both the industrialized and developing countries. Honohan (2000) reviews several studies based on individual country time-series or cross-country data and concludes that “more studies have found a positive interest rate elasticity than a negative one, but the coefficients have generally been small and often insignificant” (p. 83). Possible reasons why this approach has been inconclusive stem from aggregation problems (some investors might have a positive elasticity of saving, others a negative one) and endogeneity of inter- est rates and other variables introduced in saving regressions. The Euler equation for consumption states that consumption growth depends on the difference between the real rate of interest and the intertemporal rate of time preference. The sensitivity of con- sumption growth with respect to the real interest rate is the elastic- ity of intertemporal substitution in consumption. Since Hall’s (1978) seminal contribution, the possibility of estimating structural prefer- ence parameters with the Euler equation has attracted enormous in- terest among applied economists. More recently, the approach has shifted attention from studies using aggregate data to analyses based on household panel data, which allow rigorous treatment of aggre- gation issues. After more than two decades of empirical studies, however, the approach has also been largely inconclusive, at least as far as pinning down the interest rate elasticity is concerned. First of all, it is now clear that estimation of Euler equations poses extremely difficult econometric problems, particularly in the presence of short panel data, omitted variables due to precautionary saving and liquidity constraints, and non-separabilities between con- sumption and leisure (Browning and Lusardi 1996). Second, the sen- sitivity of consumption growth to the real interest rate is hard to es- timate with panel data on households, because at any given point in time households face the same real rate of interest. Researchers must therefore rely on the variability of marginal tax rates affecting the after-tax return to saving. Third, most empirical studies find that the elasticity of substitution is positive and generally less than one, but TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 131 estimates vary considerably over this range (Bernheim 2002). Fi- nally, even high values of the elasticity do not necessarily imply a positive interest rate elasticity of saving. Even though a high value of the elasticity makes it more likely that the substitution effect dom- inates the income effect, the Euler equation delivers information about the shape of the consumption profile, not about the level of consumption and saving. A third generation of studies has approached the issue at hand studying the portfolio and saving effect of specific tax reforms. Table 4.1, drawn from Poterba (2001), reports information on sav- ing incentives for major industrial nations’ voluntary retirement plans. With the exception of France and Japan, these programs are widely available in industrialized countries. The specific tax provi- sion and generosity of saving incentives vary considerably, but the basic features are common. Households can contribute up to a spe- cific limit to retirement saving accounts (such as Individual Retire- ment Accounts (IRAs) and 401(k) plans in the United States, Indi- vidual Saving Accounts in the United Kingdom, and life insurance plans in Italy), using pre-tax dollars. These plans have provided the ground for empirical research trying to assess to what extent con- tributions to tax deferred saving accounts represent “new saving,” or merely a substitution between tax-favored and other assets. The advantage of this approach is that it is based on clear experiments, and on rigorous econometric methods. At the theoretical level, however, it is not clear whether tax- deferred saving accounts should increase the overall level of saving. This may be the case even in situations in which the interest effect on saving is positive (that is, the substitution effect dominates the income and wealth effects).2 This point is made most clearly by Besley and Meghir (2001). Following their example, assume that individuals can invest in only two assets, both of which have a gross rate of return r and one of which is tax-free up to a certain level of investment L. An individ- ual can invest up to L in the tax-favored asset, yielding r. Any sav- ings above L yield a net of tax return r(1– θ). In the absence of uncertainty, the effect of the tax incentive on the allocation of wealth is clear: investors put as much wealth as they can in the tax-favored asset, up to the limit L. The effect on the level of saving, however, is ambiguous. If desired saving is less than L, all wealth is allocated in the tax-favored asset and, under the assump- tions of the appendix, an increase in r increases saving. If instead de- sired saving is higher than L, individuals would have saved more than L even in the absence of incentives. In such a situation, the ef- fect is reversed. Besley and Meghir show that the effect of the tax Table 4.1. Saving Incentives in Voluntary Retirement Funds in Major Industrial Nations 132 Retirement Contributions Country saving accounts? Contribution limit deductible? Special notes Canada Yes $9,400 ($15,500 Canadian), indexed Yes Limits on foreign stock; carry forward unused contributions France No — — — Germany Yes Vermogensbildungsgesetz, Yes Investment in “long term funds”; other limit $2,200 programs to accumulate housing down payments Italy Yes 2% of wages or $1,414 Yes __ Japan No — — Universal maruyu postal saving accounts were phased out in 1986 Netherlands Yes 1,700 Guilders, or approximately $850 per Yes “Employee saving scheme” and year for employee saving scheme “premium saving scheme”; four year vesting period before withdrawal United Kingdom Yes Personal pensions, contributions of Yes ISAs face restrictions on investment 17.5–40 percent of earnings; individual choices; total contribution limits were Saving Accounts (ISAs), limit of £5000/ higher in years before 2000 year contribution starting in 2000 United States Yes $2,000 for Individual Retirement Yes Other variants include Roth IRAs and Accounts, $10,500 for 401(k) plans 403(b) plans Source: Poterba (2001). TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 133 incentive depends on the relation between investors’ wealth and the limit L, and on the intertemporal elasticity of substitution. In prac- tice, then, the same incentive can generate a wide range of hetero- geneity of individual responses. Engen, Gale, and Scholz (1996) point out another feature of tax- deferred saving accounts: that is, that they are less liquid than con- ventional saving. The saving incentive might reduce savings of those who do not behave according to the intertemporal choice model, but have a fixed saving target. For these individuals, the incentive and the higher return on saving will make it easier to reach the tar- get, because they must give up fewer resources for the same target. A further criticism of the approach based on the analysis of tax re- form is that none of the studies is able to estimate the interest rate elasticity of saving and thus provide information valuable to poli- cymakers. Finally, tax incentives reduce tax revenues and therefore public saving. In order to conclude that tax incentives increase cap- ital accumulation, one must therefore consider not only the effect of tax incentives on private saving, but also the revenue loss associated with the incentives. Most of the studies in this area, however, adopt a partial equilibrium framework. As mentioned, the main question addressed by the literature is whether tax-deferred saving accounts such as 401(k)s and IRAs— introduced in the United States in 1978 and 1986, respectively— represent “new saving” or a substitution between tax-favored and conventional assets. Despite these “clean” tax experiments, a con- sensus has yet to emerge. On the one hand, Poterba, Venti, and Wise (1996) argue that IRAs and 401(k)s increase saving.3 Engen, Gale, and Scholz (1996) conclude instead that they produce mainly a re- allocation effect of household portfolios, and little effect on in- tertemporal choice. If a scheme does have a total saving effect, then one would ex- pect the saving of the participants to be higher than that of non- participants, as indeed it is on average. But participants are also on average wealthier and nearer to retirement than non-contributors. So perhaps the preferences of participants are different as well: for instance, that the wealthy contribute to IRAs because they have a higher taste for saving than non-contributors. In short, cross- sectional studies suffer from an identification problem caused by unobserved heterogeneity and self-selection issues. Panel data, track- ing the behavior of investors over time, can help the researcher to control, at least to some extent, for fixed individual heterogeneity. In principle, one can establish if investors have chosen to save more or less after the introduction of a tax-advantaged scheme. But even in this case it is difficult to attribute the decision to increase saving 134 TULLIO JAPPELLI AND LUIGI PISTAFERRI to the introduction of the scheme. An investor might choose to save more for reasons that are totally unrelated to the existence of the scheme (for instance, because the investor is reaching retirement age, or because he would like to purchase a home). To summarize, of the many studies that have analyzed the inter- est rate effect on saving, none has found convincing evidence of a systematic relation between the two variables, so that the emerging consensus is that rate-of-return effects on saving are at best small. Time-series data suffer from aggregation problems. After the initial enthusiasm, researchers have realized how difficult it is to tackle the econometric and identification issues in structural Euler equations for consumption. The analysis of tax reform, and particularly of the incentives provided by IRAs and 401(k)s in the United States, has delivered conflicting and inconclusive evidence, despite the great number of high-quality empirical investigations spurred by two major U.S. tax reforms. On a balanced reading of this literature, as summarized by Poterba, Venti, and Wise (1996); Engen, Gale, and Scholz (1996); Bernheim (2002); and Besley and Meghir (2001), there is broad consensus (at least in the United States) that tax- deferred saving accounts have induced massive portfolio shifts to- ward tax-favored assets, but much less consensus on whether sav- ing incentives have actually increased saving. Promoting Mandatory Saving Besides creating a fiscal wedge between pre-tax and after-tax re- turns, governments affect household saving and borrowing in a variety of ways. The most important programs in middle-income countries promote saving through mandated contributions to pen- sion funds and housing saving accounts. These contributions are mandatory, and crowd out almost automatically conventional sav- ing. However, contributions are generally granted preferential treat- ment by the tax code, and might therefore be a far more effective way of promoting the overall level of saving than tax-deferred sav- ing accounts based on voluntary contributions. There are several reasons for setting up mandatory saving pro- grams. First, policymakers often target domestic saving, on the as- sumption that increasing domestic saving promotes investment, job creation, and growth. Another reason is that mandatory saving pro- grams might be the only effective way of providing the elderly with adequate resources to be spent during retirement, particularly when people are myopic or do not have enough information to plan for relatively long horizons. Moreover, if saving for retirement were left TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 135 only to the discretion of the individual, free-riding behavior might impose a high burden to society as a whole. After all, the wide im- plementation of these programs and the fact that a considerable portion of household wealth is locked in mandatory saving pro- grams indicates the social approval of schemes designed to ensure people with adequate reserves to be spent during retirement. Apart from the effect on domestic saving and capital stock, a fur- ther motivation for promoting pension funds and other institutional investors is the desire to encourage so-called popular capitalism through mandatory or voluntary contributions to saving accounts earmarked for retirement. The development of the contractual sav- ing industry is widely thought to have a favorable impact on the deepening and diversification of the financial system. Moreover, in- dividuals might have greater incentives to perform on the job when they have a direct stake in the performance of the economy through the holding of risky assets.4 This section describes the characteristics of these programs, par- ticularly in Latin American and East Asian countries. The second part of the appendix shows, in the context of a simple model of the impact of mandated contributions on private and national saving, that the existence of tax concessions on mandatory savings can re- sult in higher overall savings. Empirical evidence based on cross- country data is presented showing that reserves of mandatory pen- sion systems and private pension fund assets do indeed increase national saving, as predicted by the model. The focus is mainly on pension funds, although several remarks apply to unfunded plans as well. Portfolio Effects of Mandatory Contributions In the simple model outlined in the second part of the appendix, mandatory contributions displace private accumulation one-for- one. In more realistic cases, this need not be the case. A first im- portant issue is how to measure mandatory saving and pension wealth. Then one needs to distinguish between the portfolio effect of mandatory assets and their effect on national saving. The first two issues are analyzed in this section, while the next section pro- vides empirical evidence on the impact of mandatory retirement saving on national saving. In general, individual pension assets are defined as the difference between the present discounted value of future pensions and the present discounted value of contributions. Clearly, these depend on legislation (the contribution rate and the pension award formula), expected inflation, expected retirement age, expected rates of re- 136 TULLIO JAPPELLI AND LUIGI PISTAFERRI turn, and mortality. Computation of pension wealth is therefore dif- ficult, for both defined benefits and defined contribution plans. Individual expectations are usually hard to measure. In the case of social security wealth, the problem is exacerbated by the fact that expectations depend on long-run phenomena (for instance, income growth and rates of return many years from now) and on the pos- sibility of pension reforms changing individual retirement options and pension award formulae. For this reason pension wealth can be estimated only with a high degree of approximation. Since Feldstein’s (1976) seminal contribution, the empirical litera- ture has tried to pin down the wealth replacement effect by regress- ing private assets on pension wealth. With individual-level data, a widely adopted specification is: a = f(t) + αy p + βX + σPW + ε where a is private wealth, f(t) an age polynomial, y p permanent in- come, X a set of demographics, and PW the present discounted value of pension benefits. Permanent income can be estimated us- ing the method proposed by King and Dicks-Mireaux (1982). The parameter σ measures the wealth replacement effect, which should equal –1 in case of complete crowding out. Most empirical studies find some, but less than full offset of pen- sion wealth on private wealth or saving: that is, negative values of σ but higher than –1. This suggests a degree of substitution between the two sources of wealth that is substantially lower than predicted by the theory (Mackenzie, Gerson, and Cuevas 1997). There are many potential explanations for this finding. Since there is no space to discuss all of them in detail, this discussion limits itself to a broad summary. Gale (1998) shows that the parameter σ might be a downwardly biased estimate of the displacement effect. The bias is induced by the use of a measure of disposable income that is net of pension contri- butions, rather than the appropriate measure, which instead treats contributions as mandatory savings. Gale also shows that the bias is likely to be higher for the young. Once the correction is implemented, his estimate of σ (based on U.S. Survey of Consumer Finances data) ranges from –0.52 to –0.77. There are other theoretical reasons that can explain a low esti- mate of σ. Pension wealth is illiquid, and cannot be used as a col- lateral. If the consumer is liquidity constrained, an increase in pen- sion wealth is not necessarily followed by an increase in current consumption. Precautionary saving can have similar effects. An increase in pri- vate wealth offset by a decline in pension wealth can reduce uncer- TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 137 tainty and thus increase current consumption. Also in this case, the offset is less than complete. A related issue is that consumers are un- certain about the overall solvency of the social security system and the prospect of future pension reforms. If this is the case, they may revise their social security wealth expectations downward and save more in the current period. Another factor that may have a bearing on the wealth replace- ment effect is the extent of financial education of the household. Many households have short-run horizons, either because of liquid- ity constraints or myopia. They will therefore prefer one dollar of financial wealth rather than one dollar of pension wealth. Indeed, Gale shows that the displacement effect is stronger for individuals who contribute to IRAs and for college graduates. This discussion has so far neglected the possibility that consumers respond to an increase in pension wealth by retiring earlier (a fact that cannot be captured in the simple two-period model of the appendix). But according to the life-cycle hypothesis, saving and wealth increase with the length of retirement. Feldstein (1976) shows that this induced retirement effect can potentially invert the sign of the relationship between private and pension wealth (the estimate of σ could be smaller in absolute value or even turn positive).5 The Effect of Mandatory Saving on National Saving Tax incentives for pension funds exist in virtually all countries. Most tax codes allow contributions to pension funds to be made out of pre-tax income, or to attract a tax rebate. Investment income of pen- sion funds is often allowed to accumulate free of tax. Many coun- tries even allow some part of the benefits to be paid out tax-free or at a concessionary tax rate. A regime in which contributions and returns are exempt is a clas- sical example of an expenditure tax, treating saving as any other form of consumption. On the other hand, a regime in which contri- butions are taxed and benefits are exempt corresponds to a regime in which accruals to both earnings and saving are taxed. In practice, the various tax regimes can have different effects, depending on the marginal tax rate during the working span and the marginal tax rate at retirement. Whitehouse (2000) provides a description of the tax treatment of pensions in the OECD countries. With the exception of Australia, France, Iceland, and Japan, pension contributions are made out of pre-tax income or attract a tax rebate. There is always a limit on de- ductibility of contributions. In the majority of OECD countries in- come accruing in the pension fund accumulates tax free, but there are exceptions. Australia, Denmark, and Sweden tax, partially or 138 TULLIO JAPPELLI AND LUIGI PISTAFERRI totally, the real return of the fund. On the other hand, all countries except New Zealand tax withdrawals, although at different degrees. For instance, some countries allow withdrawal of a tax-free lump sum, while others apply tax penalties to early withdrawals. Outside the OECD, the tax treatment of pension funds follows similar principles: contributions and returns are usually exempt, while benefits are taxed. Table 4.2 describes the main features of the tax treatment of mandatory contributions to pension funds for a selected group of Latin American and East Asian countries that have recently reformed their pension systems. In Argentina, Chile, Colombia, Costa Rica, Mexico, and Uruguay contributions and investment in- come are exempt, while benefits are taxed. In Latin America the ex- ception is Peru, where contributions and benefits are taxed but in- vestment income is exempt. The situation is more heterogeneous in East Asia, where all but one of the countries listed in table 4.2 (the Philippines) feature ex- empt contributions. In all but one also (Indonesia) pension benefits are also tax-free, while investment income is taxed in the Philippines and Thailand, and partly in Indonesia. Malaysia and Singapore op- erate a national, publicly managed, provident fund system, while In- donesia has a private management structure. Figure 4.1 plots histograms of the GDP share of the sum of re- serves of mandatory pension systems and private pension fund as- sets in selected countries of Latin America, Asia, and Africa. Chile, Malaysia, Singapore, and South Africa stand out, with a GDP share of 50 percent or higher. Pension fund assets are positively associated with the ratio of national saving to GDP, as shown in figure 4.2. One possible interpretation of the positive correlation is that the tax provisions attached to mandatory saving programs are associated with higher national saving, as discussed. Since the positive relation between pension fund assets and na- tional saving might be driven by other variables, the discussion now turns to regression analysis. The following reduced form for na- tional saving is estimated: S Sg PFA = α o + α1ρ + α 2 + α 3DEP + α 4 +ε Y Y Y where Y denotes GDP, S national saving, ρ the growth rate of GDP, Sg government saving, DEP the dependency ratio (defined as the ratio of those aged less than 15 or more than 65 and total popula- tion), and PFA the sum of reserves of mandatory pension systems and private pension fund assets as a ratio to GDP.6 All variables are Table 4.2. Tax Treatment of Mandatory Retirement Funds in Latin America and East Asia Year Latin America implemented Contributions Investment income Benefits Argentina 1994 Exempt Exempt Taxed Bolivia 1997 n.a. n.a. n.a. Chile 1981 Exempt Exempt Taxed Colombia 1994 Exempt Exempt Exempt up to a ceiling Costa Rica Exempt Exempt Taxed El Salvador 1998 n.a. n.a. n.a. Mexico 1997 Exempt Exempt Taxed Peru 1993 Taxed Exempt Taxed Uruguay 1996 Up to 20% of earnings Exempt Taxed are exempt East Asia Brunei Exempt Exempt Exempt Indonesia 1997 Exempt Funds bank deposits and returns on Taxed listed local securities are exempt. Returns on open-ended mutual funds, unlisted securities and property are taxed. Malaysia 1991 Exempt Exempt Exempt Philippines 1998 Employees’ contributions are Taxed Exempt taxed. Employers’ contributions are exempt for qualified occupational plans. Singapore Exempt Exempt Exempt Thailand 1998 Exempt Exempt Exempt 139 Sources: Whitehouse (2000); Holzmann, MacArthur, and Sin (2000). 140 TULLIO JAPPELLI AND LUIGI PISTAFERRI Figure 4.1. Pension Fund Assets in Developing Countries (percent of GDP) Latin America Chile Brazil Belize Costa Rica Jamaica Bolivia Honduras Argentina Colombia Peru Uruguay El Salvador Paraguay Ecuador 0 10 20 30 40 50 60 Asia and North Africa Malaysia Singapore Philippines Indonesia India Nepal Egypt Jordan Morocco Tunisia 0 10 20 30 40 50 60 Sub-Saharan Africa S. Africa Sri Lanka Kenya Mauritius Gambia Tanzania Ghana Swaziland Senegal Ethiopia Zambia Uganda Chad Namibia 0 10 20 30 40 50 60 Source: See note to table 4.3. TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 141 Figure 4.2. National Saving and Pension Fund Assets in Developing Countries (percent of GDP) 50 45 40 National Saving/GDP 35 30 25 20 15 10 5 0 0 10 20 30 40 50 60 Pension Fund Assets Source: See note to table 4.3. averaged over the 1985–95 period. Data sources and variables def- inition are reported in the data appendix. The purpose of the regression is to summarize the main determi- nants of national saving and, in particular, the correlation between tax-favored pension fund assets and national saving, once other variables affecting national saving are controlled for.7 The reason for using national saving as the dependent variable is that national saving is measured as national income less total (pri- vate plus public) consumption, a measure that does not rely on the definition of mandatory saving. On the other hand, private saving definitions are largely arbitrary depending, among other things, on the way mandatory contributions and pension withdrawals are treated. Furthermore, national saving is independent from inflation, while conventional definitions of private saving require a measure of private sector income, which is affected by the loss incurred from the depreciation of nominal assets due to inflation. Several studies have estimated versions of this equation with time series data on individual countries, international cross-sections, or panels of countries (Modigliani 1993; Masson, Bayoumi, and Samiei 1998). According to the life-cycle model, saving should be positively related to the growth rate of income, and negatively affected by the dependency ratio (α1 > 0 and α3 < 0). Models with finite horizons also suggest that national saving increases with government saving (α2 > 0). However, according to the Ricardian equivalence pro- position, public and private saving are perfect substitutes, so that 142 TULLIO JAPPELLI AND LUIGI PISTAFERRI an increase in government saving should not affect national savings (α2 = 0). The proposed specification can be used to assess the impact of mandatory and contractual saving on the national saving rate. The discussion above highlights that in the absence of tax incentives, pension fund assets are a substitute for private accumulation and should not affect national saving (α4 = 0). If instead contributions are favored by the tax code, there should be a positive effect on na- tional saving (α4 > 0). Table 4.3 reports the regression results. The first regression relies on a cross-section of 60 countries for which complete records were available. The GDP growth rate and the government saving rate are associated with higher national saving, while an increase in the de- pendency ratio reduces national saving. All coefficients are statisti- cally different from zero at the 1 percent level. These findings are in agreement with previous studies and provide strong evidence for the life-cycle model. The effect of pension fund assets is positive and statically different from zero in the less developed countries (LDC).8 To check the robustness of these results, a sensitivity analysis was performed allowing for changes in the estimation method, the defi- nition of government saving, the sample used, and the omission of other variables potentially affecting national saving. The second specification re-estimates using a robust method to control for the impact of influential values.9 The third and fourth regressions rely on a definition of inflation-adjusted government saving, which is available for a subset of the countries in our sample. The last two regressions report OLS and robust estimates based on the 38 LDC countries in the sample. In no case is the general pat- tern of results affected. In particular, the coefficient of pension fund assets ranges from 0.14 to 0.18 percent, and is statistically different from zero at the 1 percent level in all specifications, providing sup- port for the proposition that tax-favored pension contributions raise national saving, at least in LDC countries. In the sample of LDC countries, the first quartile of the distribution of pension fund assets is 1.6 percent, and in the third quartile is 11.2 percent. The estimates indicate that the thought experiment of increasing pension fund assets from the first to the third quartile is associated with an increase in national saving of 1.3 to 1.7 percentage points of GDP. The baseline specifications also include a number of additional variables that could potentially affect the national saving rate: the real interest rate (measured as the difference between the nominal interest rate on government bills and the actual inflation rate), per capita in- come, the Gini coefficient of income inequality, the GDP ratio of so- cial security expenditures (taken from Palacios and Pallarès-Miralles Table 4.3. National Saving Regressions Total sample, inflation-adjusted Total sample government saving LDC sample Robust Robust Robust Variable OLS regression OLS regression OLS regression GDP growth 1.389 1.470 2.232 2.298 1.478 1.562 (2.44) (2.66) (4.61) (5.07) (2.05) (2.26) Dependency ratio –0.520 –0.497 –0.566 –0.669 –0.493 –0.502 (–2.92) (–2.87) (–3.00) (–3.79) (–2.29) (–2.43) Government saving / GDP 0.453 0.582 0.239 0.249 0.195 0.339 (2.06) (2.73) (1.66) (1.85) (0.65) (1.18) Pension fund assets / GDP –0.005 –0.026 –0.041 –0.049 — — (–0.16) (–0.80) (–1.16) (–1.51) (Pension fund assets / GDP) ϫ 0.167 0.179 0.143 0.173 0.168 0.156 LDC (2.54) (2.79) (2.31) (3.00) (2.67) (2.58) LDC dummy –4.333 –4.724 –3.852 –2.625 — — (–1.47) (–1.65) (–1.40) (–1.02) Constant term 36.666 36.231 35.733 38.878 31.443 31.976 (5.33) (5.42) (4.98) (5.79) (2.91) (3.08) Number of countries 60 60 48 48 38 38 Note: The table reports national saving regressions using a sample of 22 OECD countries and 38 developing countries. The third and fourth regres- sions use the inflation-adjusted definition of government saving. The last two regressions are based on a sample of 38 developing countries. T statistics are reported in parenthesis. 143 Source: See data appendix. 144 TULLIO JAPPELLI AND LUIGI PISTAFERRI 2000), and a dummy variable for each continent. The coefficients of these variables are not statistically different from zero. Fostering Accumulation of Housing When governments extend incentives for retirement saving, they change household portfolio allocations, but not necessarily the in- tertemporal allocation of consumption. If instead the tax code tar- gets specific goods, such as housing, education, or health, it changes not only the margins of intertemporal choice, but also consumption allocations between different goods. This section reviews govern- ment intervention in the promotion of saving targeted to housing. The next section examines tax incentives to save for education, health, and other “merit” goods. Programs whose aim is to foster the accumulation of housing take various forms, ranging from direct government investment, subsidies, and contributions to households or firms, to indirect sup- port for housing accumulation through tax incentives. One ration- ale for government intervention in this area is the notion that hous- ing is an investment good, hence tax-deductible capital costs are in principle offset by imputed rental values. In practice the offset is modest, as the market value on which imputed rents are computed is underestimated (Duebel 2000). So the main reason is to shift the portfolio allocation of wealth toward goods to which society as- signs an important weight in alleviating poverty and in raising liv- ing conditions, much like targeting retirement saving is a remedy to household myopia and potential free-riding problems. This chapter aims at a broad description of tax incentives to save and to borrow, so it does not discuss direct government intervention in the housing market. Instead, it focuses on instruments designed to raise household saving for housing finance: mandatory contributions to specially designed housing funds, contractual saving for housing, subsidies to financial intermediaries providing housing funds and, in the fifth section, mortgage interest rate relief schemes. In some industrialized countries (notably Austria, France, and Germany), housing finance has traditionally relied on contractual saving (Impavido, Musalem, and Tressel 2002). According to these contracts, households save an agreed amount over an extended pe- riod at pre-specified terms. At the end of this period, households are eligible for a loan to purchase or renovate an owner-occupied house. The saving schemes are characterized by fixed, below-market rates on savings and subsequent loans. The advantages of contract saving for housing are threefold. First, they provide adequate information TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 145 to potential borrowers. Second, they make funds available for down payments. Finally, they offer protection against volatile financial markets (Lea and Renaud 1995). The government has a relevant role in these programs, by providing saving premiums and tax benefits. The rationale for these programs is that they allow poor house- holds to access mortgage loans for home purchase that would be otherwise unavailable. This might be even more important in less developed countries, where many households lack the income to qualify for a loan and/or financial institutions tend to screen out these households because mortgages require regular payments over the long run, as well as income stability—conditions that are un- likely to be met by poor households. Limited competition among fi- nancial institutions amplifies these market failures. In the German Bauspar scheme, the principle is one of reciproc- ity. Contractual deposits are mobilized by specialized institutions and are available only to make loans to participants. Savers fulfill- ing the terms of the contract receive a government interest premium proportional to the amount saved (up to a maximum). The French system is traditionally more integrated with capital markets and contractual saving for housing works like an account with subsidized return. In contrast to the German model, savers can keep their savings in the housing fund without obtaining the loan if they find the deposit yield attractive. Deposit institutions can use the deposits to finance housing loans or purchase bonds in the second- ary mortgage market. As in Germany, French savers enjoy a gov- ernment interest premium on contractual saving over the banks’ de- posit rate. Lea and Renaud (1995) provide a thorough discussion of the pros and cons of contractual saving schemes for housing, and conclude that the success of these programs critically depends on fi- nancial and price stability, two conditions that are seldom met in LDC countries. Housing banks enjoy a wide range of subsidies, including direct tax-based funding; direct deposits with tax-preferred or regulation- preferred status or government guarantees; and exemption from in- come taxation, stamp duties, and lien registration costs. They play a prominent role in several developing countries, such as Argentina, Brazil, India, Indonesia, Jordan, Korea, Thailand, and Venezuela (Duebel 2000). But the most striking programs to foster the accumulation of housing are mandatory contributions to housing funds, widespread in several Latin American and East Asian countries. Table 4.4 reports the main features of these programs. They differ quite widely in the amount of contributions, eligibility, and loan characteristics. Contri- butions to housing funds are usually based on payroll tax, with one 146 TULLIO JAPPELLI AND LUIGI PISTAFERRI Table 4.4. Mandatory Housing Funds Country Program Main features Argentina FONAVI Prior to 1991 mandatory contributions were 5 percent of the wage tax. Since 1991, mandatory contributions have been 40 percent of the fuel consumption tax. To qualify for a loan, household monthly income must range between $500 and $1,000. The household then assigns 25 percent of income to monthly repayment installments. Terms are up to 25 years. Bolivia FONVIS (phased The program featured deductions from out in 1998) public sector salaries earmarked for so- cial housing improvement and construc- tion. The current system is based on tax incentives for long-term saving. Brazil FGTS The fund is made up of mandatory monthly deposits by employers into accounts held in the names of their employees at the Federal Savings Bank. The contribution rate is 8 percent. Loans with FGTS funds are targeted to low-income families with monthly income of up to 12 minimum wages (approximately US$933). Average loans amount to approximately R$20,000 (US$11,500). Colombia CAV Mandatory contributions amount to one month of wages. Withdrawal is allowed for home purchase, as well as for work- ers who go back to school, if they quit or are dismissed. Jamaica National Housing Established in 1976. It is funded by Trust statutory deductions from employers, employees, and self-employed individu- als. Contributors must earn more than the minimum wage. The contribution rate is 5 percent (2 percent from employ- ees, 3 percent from employers). The self- employed contribute 3 percent. Funds are available for first home purchasing after a minimum of 3.5 years of contri- butions, for a maximum amount of JA$800,000 (per contributor). TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 147 Table 4.4. Mandatory Housing Funds (continued) Country Program Main features Mexico INFONAVIT Mexican employers contribute 5 percent of their payroll to the fund, up to a maximum of 10 times the statutory min- imum wage. The contributions accumu- late in the national housing fund, used by INFONAVIT to award 30-year mort- gage loans, with an average interest rate of 6 percent plus an adjustment for the annual increase in the minimum wage. Peru FONAVI (phased Peruvian employers contributed 5 per- out in 2000) cent of total wages (salary plus bonuses) to the housing fund. The scheme (con- verted to an “emergency solidarity tax”) was phased out in January 2000. Philippines PAG-IBIG Home Mandatory contributions are 1 percent Development for employees earning not more than Mutual Fund 1.5 million pesos per month; 2 percent for those earning more than 1.5 million. All employers contribute 2 percent of the monthly compensation for all em- ployees covered by the social security system and the Government Service Insurance System. Venezuela Ahorro Mandatory contributions for public and Habitaçional private employees are 1 percent of wages for employees and 2 percent for employ- ers. The self-employed can participate in the program on a voluntary basis with a contribution rate of 3 percent. Home- owners and individuals over 60 are exempt from the contribution. Source: Duebel (2000). notable exception: in Argentina, contributions are no longer from the income payroll tax but from the fuel consumption tax. Many of these programs were not originally earmarked for hous- ing, but have evolved to allow contributors to withdraw funds for home purchase. For instance, in Brazil the FGTS (Fundo de Garan- tia de Tempo de Serviço) was established in 1967 as a severance pay fund in the case of termination without just cause. It requires em- ployers to deposit a (tax-deductible) 8.3 percent of the worker’s pay in a restricted saving account in this fund, managed by the National 148 TULLIO JAPPELLI AND LUIGI PISTAFERRI Housing Bank, in the names of their employees. The account earns interest of 3 percent a year, in addition to a monthly adjustment to compensate for inflation.10 Employees may withdraw the balances in these bank accounts if they have been dismissed without just cause, if the firm is liquidated, at retirement, or to purchase a per- sonal residence under a government-approved housing financing scheme. In case of dismissal without just cause, the employer is obliged to pay the employee an additional amount equivalent to 40 percent of the accumulated balance in the employee’s FGTS bank account.11 Mandatory saving is sometimes targeted to both retirement and housing. Table 4.5 lists some selected examples around the world. In Malaysia, for instance, accumulated Employees Provident Funds can be used to serve retirement purposes (60 percent), home purchase (30 percent), and even medical expenses (10 percent). Similar provi- sions, subject to limits, exist in Colombia and some Asian countries. The stated targets of most mandatory savings programs for hous- ing are low-income households. However, experts have lamented that most of the social/public housing projects have in practice fa- vored middle-income rather than poor households. In Venezuela only 8 percent of households contributing to the Ahorro Habita- çional fund have actually received a housing unit (under the Ley of Politica Habitaçional). In Mexico, INFONAVIT has assisted little more than 11 percent of households contributing to the fund. In Peru, the FONAVI fund has been in effect no more than an addi- tional payroll tax imposed on employers.12 Experts have also noted that mandatory contributions to housing funds tend to distort credit markets and inhibit their development. Targeting Merit Goods: Education, Health, and Life Protection Virtually all industrialized countries provide some form of tax in- centives for saving toward education, and tax deductibility of edu- cational expenses, health and life insurance premiums, and out-of- pocket medical expenses. Education, health, and life protection can be regarded as merit goods and, in the case of education and health, long-term growth enhancing factors. These programs are absent or of negligible importance in less developed countries. Table 4.6 de- scribes the features of some of these programs and tax provisions for selected countries. A striking addition to the wide range of tax concessions and other government interventions already influencing the college tu- TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 149 Table 4.5. The Interaction between Mandatory Pension Funds and Housing Finance Country Program Main features Colombia CAV Mandatory contributions equal to one-month wages are made to the CAV. If the pension fund exceeds 110 percent of minimum wages, it can be used for home purchase. Malaysia Employees Contributions to the EPF are apportioned Provident Fund into three accounts: 60 percent is for retire- (EPF) ment, 30 percent is for housing or pre-retire- ment plans (to be eligible for pre-retirement withdrawal the account holder must be over 50 years old), and 10 percent is for medical expenses. The withdrawal can be used to buy or build a house, or to pay off a housing loan. A member is eligible to apply for withdrawal if he has not made any previous withdrawals to buy or build a house and has not attained the age of 55 at the time of application. Philippines Reserve funds Borrowing for housing against a share of the of the social accumulated pension fund is possible. security system Singapore Central Borrowing for housing against a share of the Provident Fund accumulated pension fund in CPF is possible. (CPF) In fact, over half of withdrawals in recent years have been for housing purposes. Sri Lanka Employees The EPF is the largest mandatory saving pro- Provident Fund gram in Sri Lanka. The contribution rate is (EPF) 12 percent for employers and 8 percent for employees. Contributions are taxed (that is, they are paid out of after-tax income), but income from accumulated funds and benefits are exempt. It is possible to withdraw funds from retirement accounts to purchase a home. Sources: Whitehouse (2000); Holzmann, MacArthur, and Sin (2000). ition motive for saving in the United States (see Feldstein 1995; Souleles 2000) is the Education IRA, introduced by the Taxpayer Relief Act of 1997. Under this scheme an annual, non-deductible contribution of up to $500 per year per child (tapering off for high- income households) may be invested until the child reaches age 18. Withdrawals to pay for college expenses for the child are tax-free. 150 TULLIO JAPPELLI AND LUIGI PISTAFERRI Table 4.6. Tax Incentives for Education, Health, and Life Protection Main features Education United States Education IRA, introduced in 1997. An annual, non- deductible contribution of up to $500 per year, per child may be invested until the child reaches age 18. Withdrawals to pay for college expenses for the child are tax-free. Single taxpayers earning an adjusted gross income above $95,000 ($150,000 on a joint return) are not eligible. United States Tax deductibility of up to $2,500 on qualified student loans; tax credits of up to $1,500 of undergraduate education and up to $1,000 for any post-high school education. Germany Tax deductibility of 30 percent of tuition fees of government-approved private schools. Health United States Major out-of-pocket health expenses deductible (excluding private health insurance premiums). Italy Major out-of-pocket health expenses deductible (including private health insurance premiums). Argentina Private long-term health insurance tax-deductible, subject to limits. Taiwan Private long-term health insurance tax-deductible. Germany Private long-term health insurance tax-deductible, subject to limits. India Private long-term health insurance tax-deductible up to a cap. Japan Private long-term health insurance tax-deductible up to a cap. Life protection United States Investment income is not taxed but withdrawals and contributions are. Employer-provided group term policies up to $50,000 are not taxed. Canada Investment income and withdrawals are tax-free if they provide only an insurance, rather than an investment, element. United Kingdom Withdrawals are tax-free, but returns are taxed. Tax relief on premiums phased out in 1984. TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 151 Table 4.6. Tax Incentives for Education, Health, and Life Protection (continued) Main features Life protection, continued Germany Contributions are tax-deductible subject to a cap; returns are not subject to tax. Italy Contributions are tax-deductible subject to a cap and a time limit for life insurance products specifically designed to supplement social security pensions; withdrawals are taxed. Japan Contributions are tax-deductible subject to a cap and a time limit; withdrawals are partially tax-exempt. France Contributions are tax-deductible subject to a cap and a time limit; withdrawals are taxed. Argentina Contributions are tax-deductible subject to a cap; returns are tax-free. India Contributions are tax-deductible; returns are taxed. Singapore Contributions are tax-deductible subject to a cap; returns are taxed. Taiwan Contributions are tax-deductible subject to a cap; returns are taxed. Sources: Whitehouse (2000); Holzmann, MacArthur, and Sin (2000). In most European countries schools are public and higher educa- tion is funded by general taxation with limited tuition fee expenses, and accordingly a lower need to save to meet this capital outlay. Nevertheless, in some countries private education expenses receive some limited tax advantage (as in Germany and Italy). The tax code also influences the demand for health insurance. Major health expenses are tax deductible in a handful of countries, particularly where the government-provided healthcare system is absent or deemed to be of poor quality. In the United States, out- of-pocket medical expenditures above a threshold (currently 7.5 per- cent of adjusted gross income) can be deducted from adjusted gross income. This excludes health insurance premiums paid by employees. In Italy both private health insurance premiums and a limited amount of out-of-pocket medical expenses are tax deductible. Among devel- oping countries, Argentina, India, and Taiwan offer tax deductibility of private long-term health insurance subject to some limits. Finally, several countries’ tax systems contain incentives to en- courage life insurance contributions.13 Many life insurance con- 152 TULLIO JAPPELLI AND LUIGI PISTAFERRI tracts are structured to embody a significant element of saving with only a modest insurance component. Typically, the tax advantages are granted only if the investors commit to long-term contracts, which can be seen as a substitute for retirement saving.14 Tax de- ductibility tends to be operative on the financial component of the contract, rather than on the pure insurance component—although Canada provides an exception in this respect. In the United States accumulated reserves in life insurance con- tracts are relatively unimportant because administrative costs lower the net returns offered below the pre-tax return on the assets that back the policies. In Italy, life insurance premiums were partly tax- deductible between 1987 and 2000, but the tax system did not dis- criminate between the saving and the insurance component. The 2000 tax reform maintains incentives only for the financial compo- nent.15 In developing countries (such as Argentina, India, Singapore, and Taiwan) tax systems feature partial tax deductibility of contri- butions to life insurance policies.16 From a public policy perspective, it would be interesting to know how the schemes discussed above compare with non-financial based incentives that achieve the same broad goals; for instance, com- paring the effectiveness of tax incentives for education with direct subsidies, or tax deductibility of health-related expenses with de- ductibility of health insurance premiums. Which types of scheme are most effective in enhancing education or improving health stan- dards; which are prone to abuse; and what are the redistributive im- pacts and macroeconomic consequences of the different incentive regimes? So far the evidence is scant, particularly in developing countries, preventing clear-cut policy advice in this area. Incentives to Borrow In many developed countries tax incentives to save coexist with tax incentives to borrow in the form of tax deductibility of interest pay- ments. Such allowances may have important implications for the structure of household portfolios and for the overall level of saving. For instance, in almost all EU countries and in the United States a portion of mortgage interest rate payments is tax-deductible (Maclennan, Muellbauer, and Stephens 1998). On the other hand, few countries offer tax incentives for consumer credit and educa- tional loans (see the discussion above for details on education). Table 4.7, drawn from Poterba (2001), summarizes the type of borrowing incentives for house purchase and consumer credit that exist in nine major OECD countries. Because of the low level of de- TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 153 Table 4.7. Tax Treatment of Borrowing, Selected OECD Countries Tax treatment of Country Is mortgage interest deductible? consumer borrowing Canada No Not deductible France Yes Not deductible Germany No Not deductible Italy Only for first-time homebuyers Not deductible Japan No, but new homebuyers enjoy Not deductible a tax credit for up to six years Netherlands Yes Deductible subject to a cap Sweden Yes Fully deductible until 1991. Not deductible afterward. United Kingdom No (effective April 2000) Not deductible United States Yes, subject to a rarely Not deductible. Home binding limit equity loans up to $100,000 are tax- deductible. Source: Poterba (2001). velopment of credit markets, such programs are seldom present in less developed countries. Only one of the nine nations, the Netherlands, currently allows tax deduction for consumer credit, subject to a limit (in Sweden, tax deductibility of consumer credit was phased out in 1991). House- holds can deduct mortgage interest payments in five out of nine na- tions. In the United Kingdom the tax deduction was phased out in April 2000 (see below). In Japan taxpayers are not allowed to deduct mortgage interest payments, but enjoy a special tax credit for first- time home purchase, subject to a time limit. Three countries—France, the Netherlands, and the United States—allow relatively unrestricted deductions for mortgage interest. A fourth, Italy, allows mortgage interest deductions only for first-time homeowners. In the United States, households cannot deduct interest on more than $1 million of mortgage debt; in practice, this constraint rarely binds. In light of this cross-sectional heterogeneity in tax rules, one should expect house- holds to allocate a greater share of their portfolios to housing assets in France, the Netherlands, and the United States, and to rely more on mortgage debt in these countries, than in other nations. 154 TULLIO JAPPELLI AND LUIGI PISTAFERRI Tax deductibility of loan payments reduces the after-tax interest rate paid on the loan. Theoretically, for net borrowers the effect and substitution effects discussed in the first part of the appendix go in the same direction: an increase in the interest rate increases interest payments (the income effect), as well as the relative price of current consumption with respect to future consumption (the substitution effect). For both reasons, borrowing and current consumption de- cline. Thus on theoretical grounds the removal of tax incentives to borrow should be accompanied by a decline in debt (that is, an in- crease in saving). Yet tax incentives for borrowing have been the subject of much less investigation than incentives to save. Ideally, such investigation should be carried out at the microeconomic level, exploiting the variability induced by tax reforms that change the in- centives of different household groups to borrow. A good example in this respect is the U.K. MIRAS (Mortgage In- terest Relief at Source) experiment. Before MIRAS, there were two sources of potential cross-sectional variability. Mortgagers received tax relief on the interest on the first £30,000 of a mortgage; the tax relief increased with the marginal tax rate. Moreover, until 1988 the £30,000 limit applied on single mortgagers rather than the property: that is, unmarried people could each receive relief on loans up to £30,000, including more than one on the same property. In 1988 the government eliminated such distortion. According to Hills (1990) this tax change, advertised five months in advance, was a contribu- tory factor to the overheating of the U.K. housing market in 1988. The model in the last section of the appendix shows that quan- tity constraints on borrowing generally distort the intertemporal consumption profile. In practice, policymakers might still want to discourage household borrowing, because they might target saving and investment at the expense of consumption. The experience of several developed countries is consistent with this argument. For in- stance, in South Korea and Taiwan, selective credit ceilings have been placed on mortgage loans with the explicit aim of fostering indus- trial investment.17 Conclusion This survey began with the consideration that there are good reasons for encouraging household saving. First of all, there is considerable evidence that even in closed economies or economies with imperfect capital mobility, higher saving leads to more productive investment and ultimately wider economic development. A second compelling argument for promoting pension funds and the life insurance indus- try is that the contractual savings industry can have a favorable im- TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 155 pact on the diversity and efficiency of the financial system. Propo- nents argue that they can contribute to the establishment of so-called popular capitalism, enhancing the incentives to perform of individu- als with a direct stake in the holding of risky financial assets. While the arguments in favor of promoting saving are widely ac- cepted and theoretically compelling, there is less consensus on the ways to achieve this goal and on the interpretation of the empirical evidence. A first issue is whether governments should rely on tax in- centives to encourage voluntary retirement saving—but leave the ul- timate decision to save up to the individual consumer—or should governments rely on mandatory retirement programs. Tax incentives to save can be justified on the ground that they do not distort the intertemporal choice between consumption and sav- ing: a regime in which contributions and returns are tax-exempt treats saving as any other form of consumption. However, in prac- tice this is not the case, as shown in the second section above, be- cause in most countries the tax code does not achieve neutrality. And there is considerable empirical debate as to the effectiveness of tax incentives in promoting saving: most studies conclude that tax incentives affect the allocation of household portfolios, but the ef- fect on the amount saved is less clear-cut. Tax incentives also raise serious distributional issues, particularly in developing countries, as only individuals in the upper tail of the income distribution have the resources and the financial information required to take advantage of the incentives. Finally, tax incentives have a cost in terms of rev- enue losses. To the extent that losses are shared more equally than gains, tax incentives to save might adversely affect the distribution of resources in the economy. For these reasons, promoting volun- tary saving might not be the right instrument for achieving the goals outlined above. However, mandatory saving programs avoid the free-riding problems that would arise if individuals were free to choose the amount of saving; they force even myopic individuals to set aside re- sources to be spent during retirement. If the policy goal is to in- crease investment and growth and promote popular capitalism, mandatory saving programs are more likely to succeed than volun- tary saving schemes, even at the cost of distorting the intertemporal allocation of resources that households would choose otherwise. Besides retirement saving, there is a wide range of instruments that governments apply to promote saving for specific goods (hous- ing, education, and health). The widest area of intervention is hous- ing, where the most striking instruments are mandatory contribu- tions to provident funds designed to accumulate resources for a down payment against home purchase. Several countries also en- courage saving for education and medical expenditures, while still 156 TULLIO JAPPELLI AND LUIGI PISTAFERRI others encourage (or discourage) borrowing. At the moment, the interplay between these different saving instruments, their distor- tionary impact on household portfolios and, ultimately, their effec- tiveness in achieving sectoral and macroeconomic goals have been subject to many fewer empirical studies. A priori, one can expect considerable portfolio effects from these programs, but in the ab- sence of detailed studies it is difficult to evaluate their overall effec- tiveness. For instance, mandatory saving programs for housing clearly raise the share of wealth allocated to housing. However, the effectiveness in raising the overall homeownership and saving rates can be undermined by displacement effects on other forms of wealth, bureaucratic costs in managing the provident funds, government revenue costs, and the associated distributional impacts. There is ample ground for detailed studies on these and related issues, but at the moment the evidence is scant, particularly in de- veloping countries, preventing clear-cut policy advice in this area. The studies would require microeconomic surveys to sort out indi- viduals affected and not affected by the tax incentives, and detailed data on the composition of household portfolios to track the re- sponse of the individual wealth components to the presence of in- centives to housing, health, education, and borrowing. While the discussion above applies to developed and developing countries alike, this study concludes by singling out two issues that are likely to be of particular importance in developing countries. The evidence for developing countries indicates that pension cover- age is far from complete. Extending pension coverage, and the associated tax benefits, might be far more important than introduc- ing sophisticated saving incentives. Besides affecting national sav- ing, pension coverage can have a number of other effects, such as in raising the attachment of workers to firms and labor productivity. Needless to say, financial transaction costs are high in many devel- oping countries.18 These costs are compounded by judicial and bu- reaucratic costs, which in many countries represent a hidden tax on business and financial transactions (Djankov and others 2002). This implies that the return and incentive to save can be increased by making the environment more competitive and efficient without re- sorting to explicit tax instruments. Appendix. Analytical Framework for Assessing the Impact of Tax Incentives on Saving This appendix sets out a simple model of saving and examines the impact of tax incentives. TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 157 A Two-period Model of Saving To provide a background to some of the issues analyzed in the second sec- tion, it is useful to consider the effect of changes in the after-tax return to saving in the simplest intertemporal choice model. Consider the standard two-period consumption model, where the individual maximizes utility max u(ci, c2) subject to an intertemporal budget constraint: c21 y2 c1 + = a1 + y1 + = a1 + h1 1 + r (1 − θ) 1 + r (1 − θ) where c1 and c2 denote, respectively, first and second period consumption, a1 indicates beginning-of-period wealth, and h1 represents human capital: that is, the present discounted value of current and future (after-tax) in- come. The individual receives r(1– θ) on resources carried on from the first period to the second, where r is the real rate of interest and θ is the tax rate on asset income. The standard assumption is that the utility function is concave, and that the marginal utility of consumption is positive and decreasing: that is, ∂u(⋅)/∂cj >0, ∂ 2u(⋅)/∂c2 j < 0, for j=1, 2. The first-order condition of the max- imization problem is: ∂u(.) / ∂c1 = 1 + r (1 − θ) ∂u(.) / ∂c2 If first period consumption is reduced by one unit, second period con- sumption increases by 1+r(1– θ). Thus 1+r(1– θ) represents the price of first period consumption in terms of second period consumption. Using the in- tertemporal budget constraint, the optimization problem delivers two de- mand functions: ∗ c1 = f [r (1 − θ), a0 + h0 ] ∗ c2 = g[r (1 − θ), a0 + h0 ] Given the assumptions about the utility function, an increase in total wealth raises both c* * 1 and c 2. The after-tax real interest rate has three effects on first period consumption. If c* 1 < a1 + y1—that is, if the individual is a net saver in the first period—an increase in the after-tax return increases asset income received in the second period, raising both c* * 1 and c 2 (the in- come effect). The same increase in r(1– θ) raises the price of first period con- 158 TULLIO JAPPELLI AND LUIGI PISTAFERRI sumption, increasing c* * 2 at the expense of c 1 (the substitution effect). Fi- nally, an increase in r(1– θ) reduces human capital and consumption in both periods (the wealth effect). While the substitution effect and the wealth ef- fect reduce c*1 , the income effect tends to raise it. The overall impact of the interest rate on first period consumption and saving is therefore ambiguous. On the other hand, if c* 1 > a1 + y1—that is, the individual is a net bor- rower in the first period—the income effect also tends to reduce c* 1 because the borrower would have to pay higher interest in the second period. Thus an increase in the interest rate for a net borrower always reduces first pe- riod consumption (increases saving). The effect of a change in the return to saving therefore depends on the distribution of the endowments over the life cycle. Even when the effect can be signed, the strength of the effect de- pends on the concavity of the utility function: that is, on the intertemporal elasticity of substitution. As indicated in the second section above, the ef- fect is ambiguous not only theoretically, but also empirically. The Tax Treatment of Mandatory Contributions This appendix studies the effect of the deductibility of mandated contribu- tions on the composition of household wealth and on the overall level of sav- ing. Consider a standard overlapping generations model, where households maximize utility and are subject to a general income tax rate τ and to a mandatory contribution to a fully funded pension fund µ. In the first period, income taxes and mandated contributions are a fixed proportion of wages. In the second period, capital income and pension benefits are taxed at the same general income tax rate τ. Further consider two cases. In a first sce- nario, pension contributions are levied on first period income and are not tax- deductible. In the second scenario, contributions are fully tax-deductible.19 Assume that the contribution rate does not exceed the propensity to save that would prevail in the absence of pensions. The effect of mandated pension contributions on national saving arises from the tax treatment of contributions, and does not require that saving responds positively to changes in interest rates. Consider the log-utility case, in which income and substitution effects cancel each other out. Since second period income is zero by assumption, there is no wealth effect. Con- sumers solve the problem: max lnct, t +βln ct, t+1 In the first scenario, with a mandated pension fund yielding the same return as private savings, the budget constraints in the two periods are: TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 159 y p c t + s t = wt (1 – τ – µ) o t + (1 + rt+1) wt µ](1 – τ) = (1 + rt+1)(1 – τ)[s t + wt µ] c t+1 = [(1 + rt+1)s p p The intertemporal budget constraint is: o ct +1 cty + = wt (1 − τ) (1 + rt +1 )(1 − τ) In the absence of population and income growth, national saving is the sum of mandated and private (or discretionary) saving, s m p t and s t respectively: β st (nd) = stp + stm = wt (1 − τ) 1+ β  β(1 − τ)  Mandated savings s m t = µwt and private savings is s t =  − µ w. p  1 + β  The first point to note in the expression for national saving is the nega- tive effect of τ on after-tax wages and therefore saving. The interest rate channel is not operative here, because log utility has been assumed, and be- cause it has also been assumed that second period income equals zero. The second point to note is that a change in µ does not affect st(nd). Na- tional saving would be the same even in the absence of forced saving. How- ever, an increase in mandatory savings (s m p t ) will reduce private savings (s t ) one-for-one (a wealth replacement effect). This poses important definitional issues when measuring and comparing saving across countries that differ in the extent of mandatory programs. In particular, mandatory contributions to pension funds, paid either by employees or employers, should be counted as part of income and therefore household saving. On the other hand, the pension paid in the second period, [(1 + rt+1) wt µ](1 – τ), should not be con- sidered as part of income because it is matched by an equal reduction in pension wealth. Thus first period saving is st(nd) and second period dissav- ing is just –ct+1. In this simple closed-economy framework, national saving translates into investment, thereby increasing the capital stock. In fact, appending a capital market equilibrium condition and assuming that the production function is Cobb-Douglas, y = Akα, one obtains a closed form solution for the steady-state capital stock: 1  β(1 − τ)(1 − α)A  1− α k(nd) =    1+ β  160 TULLIO JAPPELLI AND LUIGI PISTAFERRI As with the national saving, the steady-state capital stock is decreasing in τ and unaffected by the contribution rate µ. Consider now a situation in which contributions to mandatory pension funds are tax-exempt, so that the income tax is computed on income net of the contribution. As discussed in the chapter, this is the standard case in most industrialized and developing countries. The budget constraints in the two periods are: y t = (wt – µwt )(1 – τ) = wt(1 – τ)(1 – µ) ct + s p o t + wt µ) c t+1 = (1 + rt+1)(1 – τ)(s p and the resulting intertemporal budget constraint is: o ct + ct +1 (1 + rt +1 )(1 − τ) [ = wt (1 − τ) + τµ ] National saving is then: β st (d) = 1+ β [ wt (1 − τ) + τµ ] By comparing the two expressions for national saving, one sees that saving is higher in the economy where contributions are tax-deductible: β st (d) − st (nd) = µτwt 1+ β The difference in national saving depends on the contribution rate and on the tax rate. For reasonable values of µ and τ, saving can be substantially higher in an economy with tax-deductible contributions. By appending a capital market equilibrium condition, one can also derive the steady state capital stock 1 [ ]  β (1 − τ) + τµ (1 − α)A  1− α k(d) =    1+ β    that is clearly higher in the economy with tax-deductible contributions. Often only a fraction of the workforce contributes to pension funds. De- noting this fraction by λ, national saving is a weighted average of the sav- ing of contributors and non-contributors. While the fraction does not affect national saving in the economy with no deductions, when contributions are tax-deductible national saving increases with pension fund coverage: TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 161 st = β 1+ β { [ ]} wt (1 − τ)(1 − λ) + wt (1 − τ) + τµ λ One can extend this simple framework in several directions. In endoge- nous growth models (for instance, models where output is a linear function of the capital stock), the growth rate of the capital stock and of national in- come, rather than the levels of the two variables, increases with pension coverage and with the mandatory contribution rate. It has implicitly been assumed here that the tax rate finances government consumption that is simply wasted. But in more realistic setups, government expenditure might impact the productivity of the private sector. If the gov- ernment budget is balanced, one can show that in an economy with tax- deductible contributions the required tax rate is lower than in an economy without deductions. Equivalently, for given tax rate, the capital stock is higher in the economy with deductions. However, during the transition to a tax-favored regime, even though the tax treatment of mandatory contribu- tions promotes pension saving, it might defer government revenues. Finally, only dynamic efficient economies have been considered here. In dynamically inefficient situations eliminating tax deductions is one way to approach the golden rule of capital accumulation. The Effect on National Saving of Tax Incentives for Borrowing To see the implications on national saving of changes in incentives to bor- row, consider a simple overlapping generations model with two types of in- dividuals: a fraction λ with endowments in the first period and nothing in the second (net savers), and a fraction (1 – λ) with endowments only in the second period (net borrowers). Assume for simplicity that the proportion of the two types in the population is exogenous. Net borrowers borrow from net savers in the first period, and repay their debt in the second period. This framework requires only minimal change in assumptions and notation with respect to the model in the previous section of this appendix. Net borrowers face an interest rate on borrowing given by (1 + rb) = (1 + r)(1 + φ – θ), where φ is the cost of financial intermediation (assumed exogenous) and θ is the tax incentives (say, deductibility of consumer credit or mortgage interest). For simplicity, it is assumed that φ is wasted during the intermediation process. Net savers face an interest rate on lending equal to r. Assume also that rb > r (that is, φ > θ). First period saving of the “net savers” is: l β st = wt 1+ β 162 TULLIO JAPPELLI AND LUIGI PISTAFERRI First period saving of “net borrowers” is negative: b 1 wt +1 st =− 1 + β (1 + r )(1 + φ − θ) Clearly, raising the incentive θ will reduce the interest rate on borrow- ing, increase debt, and therefore reduce saving. Note that u1 = β(1 + ri)u2 where for a net borrower ri is the borrowing rate and for a net saver it is the lending rate. Thus the Euler equation implies that individual consump- tion growth is higher for net borrowers than for net savers. Aggregate wealth in period t is the sum of the saving of the savers and the dissaving of the borrowers: l b β 1 wt +1 (1 − λ)Nt λNt st + (1 − λ)Nt st = wt λNt − 1+ β 1 + β (1 + r )(1 + φ − τ) An increase in θ increases borrowing and first period consumption of net borrowers and therefore lowers aggregate saving. Thus aggregate wealth falls with θ. Note also that reducing the cost of financial intermedi- ation φ will increase first period borrowing and reduce saving. The capital market equilibrium condition is Kt+1 = λNt st l + (1 – λ)N sb. t t Suppose that population is stationary. With a Cobb-Douglas production function, the marginal product of labor is w = w = A(1 – α)kα and the mar- ginal product of capital is (1 + r) = αAkα –1. The steady state capital stock is: 1 λγ (1 − α)A  1−α k=  (1 − λ)γ (1 − α) 1 +   α(1 + φ − θ)  where γ = [β/(1 + β)]. The capital stock also falls with θ. While this model shows that incentives to borrow generally depress sav- ing, the growth effect of tax incentives to borrow is less clear-cut, particu- larly in models with endogenous growth. The model takes for granted that there are two types of individuals in the economy. But suppose that indi- viduals with no endowment in the first period would like to borrow to fi- nance human capital investment. In turn, this might prompt a higher wage rate in the second period. Then, introducing tax incentives to borrow will tend to foster human capital accumulation and, in models in which human capital is an engine for growth, the steady-state growth rate (De Gregorio 1996). TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 163 Data Appendix. Data Used in the Estimation Data used in the regressions for national saving can be downloaded from ftp://monarch/worldbank.org/pub/prddr/outbox. National saving is defined inclusive of all external transfers. Public sector data refer to central gov- ernment. The definition of the public sector is that of consolidated central government: that is, budgetary central government plus extra-budgetary central government plus social security agencies. This definition thus ex- cludes local and regional governments. A detailed description of the data set can be found in Schmidt-Hebbel and Servén (1999). The original data- base includes 150 countries and spans the years 1960 to 1995. For each country, this chapter takes the 1985–95 average. Each series consists of an Excel file; the files used are given in the table below. Data for pension fund assets and public expenditure on pensions are drawn from Palacios and Pallarès-Miralles (2000). Variable File Gross national saving as a share of GDP FORMU333.XLS Government saving as a share of GDP FORMU342.XLS GDP at current prices FORMU16.XLS GDP at 1987 prices FORMU17.XLS Total population CONST263.XLS Population younger than 15 CONST264.XLS Population older than 65 CONST265.XLS Dependency ratio (CONST264.XLS+ CONST265.XLS)/ CONST263.XLS Pension fund assets Sum of reserves of mandatory as a share of GDP pension systems and private pension fund assets as a share of GDP. Source: Palacios and Pallarès-Miralles (2000). Public expenditure on Includes all government expenditures on pensions as a share of GDP cash transfers to the elderly, the dis- abled, and survivors, and the adminis- trative costs of these programs. Source: Palacios and Pallarès-Miralles (2000). Countries Used in Estimation for National Saving Argentina, Australia, Austria, Belgium, Belize, Bolivia, Brazil, Canada, Chad, Chile, Colombia, Costa Rica, Denmark, Ecuador, Egypt, El Salvador, 164 TULLIO JAPPELLI AND LUIGI PISTAFERRI Ethiopia, Finland, France, The Gambia, Germany, Ghana, Greece, Hon- duras, India, Indonesia, Ireland, Italy, Jamaica, Japan, Jordan, Kenya, Lux- embourg, Malaysia, Mauritius, Mexico, Morocco, Namibia, Nepal, Nether- lands, Norway, Paraguay, Peru, Philippines, Portugal, Senegal, Singapore, South Africa, Spain, Sri Lanka, Swaziland, Sweden, Switzerland, Tanzania, Tunisia, Uganda, United Kingdom, United States, Uruguay, and Zambia. Notes 1. The aggregate tax elasticity of saving depends not only on preference parameters, but also on the distribution of wealth. Cagetti (2001) shows that life-cycle models of wealth accumulation in which the precautionary motive is quantitatively relevant also imply extremely low intertemporal elasticities, in contrast to models without uncertainty (Summers 1981). Such models thus imply small effects for tax-favored saving instruments on the saving of the median household. But if the richest 5 percent of the pop- ulation behaves differently (for example, being less risk-averse, or being more affected, as entrepreneurs, by capital income taxation), aggregate wealth may still exhibit a large interest rate elasticity. Thus the distribution of wealth is a crucial parameter to evaluate the impact of tax incentives. This is particularly relevant in developing countries where wealth inequal- ity tends to be higher than in most industrialized nations. 2. Behavioral theories suggest that people do not optimize in the neo- classical sense and fail to provide adequately for their retirement because of lack of self-control (Thaler and Shefrin 1981; Thaler 1990). Targeted sav- ing instruments may therefore serve as a focal point for inconstant people to actually accomplish their savings goals. For recent evidence on this point, see Madrian and Shea (2001). 3. They estimate that about half the contributions to IRAs represent new saving, 30 percent is financed by the tax saving allowed by the same IRAs, and the remaining 20 percent is a pure portfolio effect. 4. For instance, Moscowitz and Vissing-Jorgensen (2002) find that in the United States, households with private equity ownership invest on av- erage almost two-thirds of their private holdings in a single company in which they have an active management interest. 5. Feldstein notices that such effect tends to be particularly relevant just after the introduction of a pay-as-you-go system, when contributions bear little relation to annuities and workers revise their expectations concerning the age of retirement. 6. Using the stock of pension fund assets in a flow (saving) regression might be criticized on the grounds that the stock reflects not only the gen- erosity of the mandatory scheme but also its maturity. Nevertheless, intro- ducing stock variables in aggregate saving or consumption regressions is a well-established practice in applied economics. For instance, in tests of the TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 165 life-cycle hypothesis it is quite common to regress saving on wealth, and many tests of the Ricardian equivalence proposition rely on regressions of private consumption on the stock of public debt. 7. The purpose of the regressions is descriptive, so this study does not attempt to account for the possible bias induced by the endogeneity of some of the right-hand side variables (such as GDP growth), the role of omitted variables (such as corporate savings), and time aggregation issues. See Loayza, Servén, and Schmidt-Hebbel (2000) for a discussion of these and related issues. 8. Bailliu and Reisen (1997) also find that mandatory funded pensions promote national saving in developing economies. Their findings are con- sistent with a model based on mandatory contributions and credit con- straints faced by low-income workers. 9. The robust estimation method performs an initial OLS regression, calculates the Cook’s distance, eliminates the gross outliers for which the Cook’s distance exceeds 1, and then performs iterations based on Huber weights followed by iterations based on a biweight function (a routine pro- grammed in the STATA econometric software). 10. In the early 1970s the Brazilian government also introduced a housing saving account that was exempt from income tax and indexed to inflation. 11. At the time of writing, the Brazilian Parliament is discussing reduc- ing the contribution rate to FGTS from the current level of 8.3 percent to a 2 percent level. Moreover, in case of lay-offs the employer will not have to give severance pay or pay the 40 percent penalty to the FGTS. 12. In the late 1990s the Fujimori government transformed the housing fund into an “Emergency Solidarity Tax” and, after January 2000, the new government phased out the mandatory contribution scheme because of its unpopularity. 13. See Poterba (1994) for an overview of saving through insurance contracts in G7 countries. 14. For instance, in France, Italy, and Japan, tax deductibility of premi- ums is available only for contracts 5 years or longer. 15. Jappelli and Pistaferri (2002) find that there is little responsiveness of contributors’ decision to invest to changes in the return of the policy in- duced by changes in the tax deductibility of the premiums. 16. Many countries tax bequests. Among developing countries, Brazil, Chile, and Singapore do; India, Indonesia and Mexico do not; and Ar- gentina levies a transfer fee on the beneficiary. Critics of the estate tax claim that it has an adverse effect on investment and saving, as it raises the cost of capital and it makes it harder for family businesses to survive the deaths of their founders. 17. This is witnessed by country studies. Park (1991, p. 41) argues, “One advantage of the repressive financial system in Korea, and to a lesser degree in Taiwan, may have been its ability to supply long-term finance. . . . 166 TULLIO JAPPELLI AND LUIGI PISTAFERRI Without government intervention the profit-oriented behavior of the com- mercial banks would have resulted in the dearth of long-term finance.” 18. One example are migrants’ remittances, an important source of sav- ing for developing countries. The Multilateral Investment Fund (2001) presents evidence that a typical Latin American migrant in the United States remits 15 percent of her wages; however, transfer costs (which in- clude transfer fees, exchange rate commissions, check cashing fees, and other charges at the point of receipt) absorb about 15 percent of any sum transferred. 19. A situation in which contributions are not tax-deductible but bene- fits are tax-exempt is not examined because it is rarely observed in practice. References Bailliu, Jeanine, and Helmut Reisen. 1997. “Do Funded Pensions Con- tribute to Higher Savings? A Cross-country Analysis.” Development Center Technical Papers 130, Paris: OECD. Processed. Bernheim, Douglas B. 2002. “Taxation and Saving.” In Alan J. Auerbach and Martin Feldstein, eds., Handbook of Public Economics. Amster- dam: Elsevier Science Publisher. Besley, Timothy, and Costas Meghir. 2001. “Tax Based Saving Incentives.” London School of Economics, Department of Economics. Processed. Browning, Martin, and Annamaria Lusardi. 1996. “Household Saving: Micro Theories and Micro Facts.” Journal of Economic Literature 34 (4): 1797–1855. Cagetti, Marco. 2001. “Interest Elasticity in a Life-Cycle Model with Pre- cautionary Saving.” American Economic Review Papers and Proceed- ings 91 (2): 418–21. De Gregorio, José. 1996. “Borrowing Constraints, Human Capital Accu- mulation and Growth.” Journal of Monetary Economics 32 (1): 79–104. Djankov, Simeon, Rafael La Porta, Florencio Lopez-de-Silanes, and Andrei Schleifer. 2002. “Courts: The Lex Mundi Project.” NBER Working Paper 8890. Cambridge, Mass. Processed. Duebel, Achim. 2000. “Separating Homeownership Subsidies from Fi- nance.” Land and Real Estate Research Paper 14. World Bank, Wash- ington, D.C. Processed. Engen, Eric M., William G. Gale, and John Karl Scholz. 1996. “The Illu- sory Effects of Saving Incentives on Saving.” Journal of Economic Per- spectives 10 (Fall): 113–38. Feldstein, Martin S. 1976. “Personal Taxation and Portfolio Composition: An Econometric Analysis.’’ Econometrica 44 (4): 631–49. ———. 1995. “College Scholarship Rules and Private Saving.” American Economic Review 85 (3): 552–66. TAX INCENTIVES FOR HOUSEHOLD SAVING AND BORROWING 167 Gale, William G. 1998. “The Effects of Pensions on Household Wealth: A Reevaluation of Theory and Evidence.” The Journal of Political Econ- omy 106 (4): 706–23. Hall, Robert E. 1978. “Stochastic Implications of the Life Cycle-Permanent Income Hypothesis: Theory and Evidence.” The Journal of Political Economy 86 (6): 971–87. Hills, John. 1990. Unraveling Housing Finance. Oxford: Clarendon Press. Holzmann, Robert, Ian W. MacArthur, and Yvonne Sin. 2000. “Pension Systems in East Asia and the Pacific: Challenges and Opportunities.” So- cial Protection Discussion Paper 14. World Bank, Washington, D.C. Processed. Honohan, Patrick. 2000. “Financial Policies and Household Saving.” In Klaus Schmidt-Hebbel and Luis Servén, eds., The Economics of Saving and Growth. Cambridge: Cambridge University Press. Impavido, Gregorio, Alberto R. Musalem, and Thierry Tressel. 2002. “Contractual Savings, Capital Markets and Firms’ Financing Choices.” In Shanayanan Devarajan and F. Halsey Rogers, eds., World Bank Econ- omists’ Forum Volume 2. Washington, D.C.: World Bank. Multilateral Investment Fund. 2002. “Remittances to Latin America and the Caribbean.” Inter-American Development Bank, Washington, D.C. Processed. Jappelli, Tullio, and Luigi Pistaferri. 2002. “Tax Incentives and the Demand for Life Insurance: Evidence from Italy.” Journal of Public Economics. Forthcoming. King, Mervyn A., and Louis Dicks-Mireaux. 1982. “Asset Holdings and the Life-Cycle.” Economic Journal 92 (2): 247–67. Lea, Michael J. and Bertrand Renaud. 1995. “Contractual Savings for Housing: How Suitable Are They for Transitional Economies?” World Bank Policy Research Working Paper 1516. Washington, D.C. Loayza, Norman, Luis Servén, and Klaus Schmidt-Hebbel. 2000. “What Drives Private Saving Across the World?” The Review of Economics and Statistics 82 (2): 165–81. Mackenzie, George A., Philip Gerson, and Alfred Cuevas. 1997. “Pension Regimes and Saving.” Occasional Paper 153. International Monetary Fund, Washington, D.C. Processed. Maclennan D., John Muellbauer, and M. Stephens. 1999. Asymmetries in Housing and Financial Market Institutions and EMU. Discussion Paper 2062. CEPR, London. Processed. Madrian, Brigitte, and Dennis F. Shea. 2001. “The Power of Suggestion: In- ertia in 401(k) Participation and Savings Behavior.” Quarterly Journal of Economics 116 (4): 1149–87. Masson, Paul, R. Tamim Bayoumi, and Hossein Samiei. 1998. “Interna- tional Evidence on the Determinants of Saving.” World Bank Economic Review 12 (3): 483–501. 168 TULLIO JAPPELLI AND LUIGI PISTAFERRI Modigliani, Franco. 1993. “Recent Declines in the Savings Rate: A Life Cycle Perspective.” In Mario Baldassarri, ed., World Saving, Prosperity and Growth. New York: St. Martin’s Press. Moscowitz, Tobias J., and Annette Vissing-Jorgensen. 2002. “The Returns to Entrepreneurial Investment: A Private Equity Premium Puzzle?” American Economic Review. Forthcoming. Multilateral Investment Fund. 2001. Remittances to Latin America and the Caribbean. Washington, D.C.: Inter-American Development Bank. Palacios, Robert, and Montserrat Pallarès-Miralles. 2000. “International Patterns of Pension Provision.” World Bank, Washington, D.C. Processed. Park, Yung Chul. 1991. “The Role of Finance in Economic Development in South Korea and Taiwan.” Paper presented to the CEPR Conference on Finance and Development in Europe, Santiago de Compostela, December. Poterba, James M. 1994. Public Policies and Household Saving. Chicago: University of Chicago Press. ———. 2000. “Taxation, Risk-Taking, and Household Portfolio Behav- ior.” In Alan J. Auerbach and Martin Feldstein, eds., Handbook of Pub- lic Economics. Amsterdam: Elsevier Science Publisher. ———. 2001. “Taxation and Portfolio Structure: Issues and Implications.” In Luigi Guiso, Michalis Haliassos, and Tullio Jappelli, eds., Household Portfolios. Cambridge, Mass.: MIT Press. Poterba, James M., Steven F. Venti, and David A. Wise. 1996. “How Re- tirement Saving Programs Increase Saving.” Journal of Economic Per- spectives 10 (Fall): 91–112. Schmidt-Hebbel, Klaus, and Luis Servén. 1999. “Aggregate Saving and In- come Distribution.” In Klaus Schmidt-Hebbel and Luis Servén, eds., The Economics of Saving and Growth. Cambridge: Cambridge University Press. Souleles, Nicholas. 2000. “College Tuition and Household Savings and Consumption.” Journal of Public Economics 77 (2): 185–207. Summers, Lawrence H. 1981. “Capital Taxation and Accumulation in a Life Cycle Growth Model.” American Economic Review 71 (4): 533–44. Thaler, Richard H. 1990. “Saving, Fungibility and Mental Accounts.” Jour- nal of Economic Perspectives 4 (1): 193–205. Thaler, Richard H., and H. M. Shefrin. 1981. “An Economic Theory of Self-control.” Journal of Political Economy 89 (2): 392–406. Whitehouse, Edward. 2000. “The Tax Treatment of Funded Pensions.” London: Axia Economics. Processed. 5 Corrective Taxes and Quasi-Taxes for Financial Institutions and Their Interaction with Deposit Insurance Philip L. Brock Prudential and monetary policy instruments act as potentially cor- rective quasi-taxes on financial intermediaries and they interact with other taxes. Among the most important of these instruments are de- posit insurance programs, which have become increasingly widely used in the past two decades.1 Concern about prudential regulations to accompany deposit insurance has grown with the realization that deposit insurance may create a potentially large contingent liability for the government when banks become insolvent. In response to this concern, there has been an outpouring of books and articles on pru- dential bank regulation and on the design of a financial safety net.2 Informational asymmetries between borrower and lender under- pin the existence of financial intermediaries. The informational dis- parities lead to problems with moral hazard and adverse selection that may limit the scope of direct lending or cause it to dry up. Banks and other financial intermediaries can reduce the problems of moral hazard and adverse selection by holding a diversified loan portfolio that permits them to reduce the per borrower cost of mon- itoring. That is, they reduce the costs of lending by serving as dele- gated monitors for many depositors. 169 170 PHILIP L. BROCK While banks can frequently improve upon direct lending between lenders and borrowers, informational asymmetries also face deposi- tors in their dealings with banks. These asymmetries have histori- cally been an important element in bank runs. Governments have often cited the instability of banks as a reason for intervening in the regulation of the financial system. Governments attempt to stabilize and improve the efficiency of financial intermediaries by the creation of deposit insurance and central bank lender-of-last resort powers. The implicit and explicit guarantees associated with the govern- ment’s policies require accompanying corrective taxes and quasi- taxes to minimize regulatory moral hazard. Corrective taxation fre- quently falls short of the mark, with the result that government deposit guarantees may encourage risk-taking and the misallocation of society’s resources. This chapter examines the interaction between deposit insurance, taxation, and prudential regulation (viewed as a form of quasi- taxation). The analysis of deposit insurance, capital requirements, interest rate controls, and credit ceilings that follows is done from the perspective of corrective taxation in the presence of a deposit guarantee. The analysis of an indirect tax highlights the impact of the tax on the size of the government’s contingent liability associ- ated with deposit insurance. The analysis of a reserve requirement on capital inflows demonstrates the offsetting prudential and risk- inducing incentives created by reserve requirements.3 Why Have Deposit Insurance? In economies characterized by incomplete information and incom- plete markets, competitive market equilibria will generally not be Pareto-efficient. Greenwald and Stiglitz (1986) have shown that tax interventions and subsidies can often be designed that are poten- tially Pareto-improving when information is incomplete. Finance is especially prone to incomplete markets stemming from incomplete information. In financial transactions, asymmetric information gen- erally leads to the need for costly verification of a borrower’s re- ported returns. Townsend (1979) and Gale and Hellwig (1985) have shown that debt contracts economize on verification costs since verification is required only when default takes place. Debt contracts, like other in- surance contracts, encounter problems related to moral hazard and adverse selection.4 These problems are essentially tied to the con- vexity of returns associated with a debt contract. At the point of de- CORRECTIVE TAXES AND QUASI-TAXES 171 Figure 5.1. Debt Contracts with Alternative Liability Rules Change Change in net worth in net worth a 0 –E b Limited liability –2E c Double liability –W d Unlimited liability D D+E Value of assets (A) fault, a borrower has an incentive to add risk to the project. If the additional risk pays off, the borrower can avoid default. If not, the borrower is no worse off in default than before. Figure 5.1 illustrates how the convexity in returns for a bank varies under different liability arrangements. Bank owners con- tribute an amount (E) of equity. At point (a), where deposits plus equity equal the value of the bank’s assets (A = D + E), the net worth of bank owners is their paid-in capital E. Under limited liability, the face value amount of equity E represents the maximum bank own- ers can lose if the bank goes bankrupt. In figure 5.1, the kink in the profit line for bank owners occurs at point (b) because bank own- ers cannot lose more than E. It is this convexity in the profit line as- sociated with the kink at point b that accentuates moral hazard and adverse selection in banking. Alternative liability arrangements shift the point at which the profit line has a kink, but do not eliminate the convexity of returns associated with the use of debt contracts. Double liability was the norm in U.S. banking prior to the Great Depression. Under double liability bankers were personally liable for bank losses up to an amount equal to their equity stake in the bank. Such an arrangement produces a kink in returns at point (c). Unlimited liability places a banker’s pledgeable wealth at stake, so that the kink takes place at point (d). Both unlimited and double liability have drawbacks that have re- stricted (or eliminated) their current use in banking legislation. As Carr and Mathewson (1988) emphasize, unlimited liability creates a moral hazard among bank owners who must monitor one an- 172 PHILIP L. BROCK other’s personal wealth, since all owners are responsible for bank losses. The costs of this monitoring may limit the size of banks (by limiting the number of owners) and the complexity of loans (more complex loans are more difficult to monitor). Carr and Mathewson (1988) demonstrate that owners and depositors may prefer limited liability because it reduces the bank’s costs of controlling moral haz- ard among bank owners. In addition to moral hazard problems, the use of personal wealth as collateral for unlimited or double liability creates special collection problems, which, for example, greatly de- layed the payoff of depositors during the Great Depression in the United States (Kane and Wilson 1998). Limited liability has emerged as a preferred institutional form for financial institutions to handle problems of moral hazard and ad- verse selection associated with asymmetric information. Limited lia- bility for banks has well-known drawbacks, including nontrivial monitoring costs by depositors and free riding in monitoring that can lead to bank runs and the inefficient liquidation of banks. There thus arises the question of whether the Greenwald and Stiglitz (1986) hypothesis of Pareto-improving taxes and subsidies can be applied to limited liability financial institutions. Along these lines, Banerjee and Besley (1990) and John, John, and Senbet (1991) have shown that a profit tax and creditor subsidy can reduce the convex- ity in payoffs to bank owners associated with limited liability. The variance of after-tax profits is smaller than before-tax profits, so such a tax mitigates moral hazard. If the tax proceeds are used to subsidize the cost of deposits, then the combination of tax and sub- sidy can reduce moral hazard. In both papers, the combination of taxes and subsidies creates a Pareto-improving insurance mechanism (relative to limited liability) that lowers the cost of credit.5 As a practical matter, direct government subsidies to deposits are rarely seen. An alternative approach to the question of potentially Pareto-improving government interventions is to examine observed subsidies and determine whether there are taxes that would result in Pareto improvements relative to laissez faire (that is, limited liabil- ity). One of the most important subsidies to banks—or more pre- cisely, a tax-and-subsidy scheme—is government deposit insurance. Deposit insurance lowers the cost of deposits to banks (as potential depositors no longer require as high a default-risk premium), but it may also accentuate moral hazard and adverse selection by reduc- ing or eliminating the incentive of depositors to incur monitoring costs.6 Whether or not deposit insurance is Pareto-improving de- pends on the monitoring and corrective tax policies that are imple- mented in conjunction with the insurance. CORRECTIVE TAXES AND QUASI-TAXES 173 This view of deposit insurance as one element of a potentially Pareto-improving tax and subsidy package is the starting point for this chapter’s model of corrective taxation of deposit insurance. In the model, deposit insurance lowers the cost of credit to banks but also transfers the control of moral hazard from depositors to the regulatory authority. The model details regulatory and tax inter- ventions that attempt to control moral hazard in the presence of de- posit insurance. The model has two periods.7 The first period is the investment period and the second period is the consumption period. There is a countable infinity of risk-neutral agents in the economy. Bankers are able to undertake risky indivisible projects but have insufficient funds to undertake the projects on their own. Depositors cannot un- dertake projects. A representative investment project requires one unit of funds and has a state contingent payoff in period 2 of ai(s). The state of the world is assumed to have a uniform distribution over the interval [0, – s ]. The cost of funds is the risk-free gross return (r *) given by world capital markets, so that bankers are willing to undertake projects whose expected returns equal or exceed r *: s ∫ a (s)f (s)ds ≥ r . ∗ (5.1) 1 0 Lenders, however, can only verify the returns by incurring a mon- itoring cost (γ). As in Townsend (1979) and Diamond (1984, 1996), monitoring in the case of default will provide the banker with the incentive to tell the truth about the project’s outcome. Deposit con- tracts for any individual deposit di < 1 will take the following form, where r is the contracted (gross) interest rate, (1 – γ)a1(s)di is the payment (net of monitoring costs) when the banker defaults, and r*di is depositors’ required rate of return:8 s ∫ min[rd ,(1 − γ )a (s)d ]f (s)ds = r d . ∗ (5.2) i 1 i i 0 In equation 5.2 depositors bear the downside risk of the banker’s project in exchange for a high promised return (r > r*) when the project turns out well. High costs of monitoring by depositors will preclude the financing of some projects that meet the criterion of equation 5.1. A deposit guarantee transfers the cost of monitoring from de- positors to the government by placing the default risk with the gov- 174 PHILIP L. BROCK ernment. In this model, having the government monitor on behalf of the lenders eliminates the duplication of monitoring costs.9 A guarantee to depositors also requires the government to undertake the monitoring by placing the default risk with the government. With deposit insurance in place, depositors contract with the bank at the international rate r*, as shown in equation 5.3: s ∫ min[r a (s) + v (s)]f (s)ds = r * ∗ (5.3) , 1 1 0 where (5.4) v1(s) ≡ max[0, r ∗ − a1(s)]. Equation 5.3 states that depositors receive the contracted payment on deposits (r *) in nondefault states. In default states, depositors re- ceive the cash flow from the risky project (ai(s)) plus the amount of the deposit guarantee, v1(s). Equation 5.4 states that resources from the government’s guarantee fund will always meet the bank’s pay- ment shortfall [r* – a1(s)]. As Merton (1977, 1978) first demonstrated, government deposit insurance, which is a third-party guarantee to depositors made by the government, can be analyzed equivalently as a put option on the value of a bank’s assets. The government’s deposit insurance elimi- nates default risk for depositors and creates rents for bankers by lowering the interest rate that they must pay to attract deposits. Equally important, it eliminates duplication in monitoring by de- positors, thereby reducing the resource costs associated with finan- cial intermediation. The lower costs of intermediation, in turn, per- mit an expansion in the number of projects that are profitable to undertake. The Fiscal Cost of Deposit Insurance To the extent that deposit insurance protects the payments system by preventing bank runs, the existence of an insurance program may not involve actual outlays of funds. This is the message of the well- known Diamond-Dybvig (1983) model of bank intermediation, where government-provided deposit insurance eliminates bank runs at no cost to the government. In practice, deposit insurance is often expensive to the government. Government payouts associated with deposit insurance have often been very large, as documented by Lind- CORRECTIVE TAXES AND QUASI-TAXES 175 gren, Garcia, and Saal (1996) and Caprio and Klingebiel (2002). Honohan and Klingebiel (2003) analyze a sample of 40 countries in which governments spent an average of 12.8 percent of GDP in connection with explicit and implicit deposit insurance commit- ments. In a number of countries, including Indonesia, Chile, Thai- land, and Uruguay, the fiscal costs of banking system intervention during the last 20 years exceeded 30 percent of GDP. A large part of the fiscal cost is a transfer from taxpayers to bank borrowers, owners, and lenders that will involve a transfer of wealth but not necessarily the misallocation of resources (Boyd, Chang, and Smith 1998). However, a deposit guarantee also subsidizes risk- taking. Returning to this chapter’s model, with deposit guarantees the banker may undertake projects whose expected return is less than r*. Privately profitable projects may not be socially profitable projects in the presence of deposit guarantees, since the presence of the guarantees lowers the privately required rate of return to below r* (by the amount of the expected value of the guarantees): s s (5.5) ∫ 0 ∫ a1(s)f (s)ds ≥ r ∗ − v1(s)f (s)ds. 0 Equation 5.5 shows that whenever there is an expected payment to s depositors associated with deposit insurance, ∫ v (s)f (s)ds > 0, 0 1 the expected transfer may result in the incentive to invest in socially un- profitable projects. Controlling the Fiscal Costs of Deposit Insurance Controlling the fiscal costs (more accurately, the contingent fiscal liability) associated with deposit insurance is a way of preventing banks from investing in socially unproductive projects. Honohan and Klingebiel (2003) provide empirical evidence that failure to im- plement prudential controls of various types has a significant im- pact on the escalation of fiscal costs. The most direct way to limit the fiscal cost is to charge a deposit insurance premium that taxes away the expected value of the gov- ernment’s deposit guarantee. In the model this would amount to set- ting the insurance premium equal to v1 for project a1(s). However, there is a growing literature that questions the ability of govern- ments to set actuarially fair-priced deposit insurance in the presence of adverse selection.10 Kaufman and Wallison (2001) note that even 176 PHILIP L. BROCK though the U.S. Federal Deposit Insurance Corporation (FDIC) now has the authority to set risk-based insurance premiums, it has not implemented a truly risk-based system. In any case, governments have chosen to charge deposit insurance premiums that do not tax away the expected value of the guarantee (see Laeven 2002). In order to limit the fiscal cost associated with deposit insurance, governments have more actively engaged in the use of bank capital requirements. In the model it will be convenient to express the cap- ital requirement as a ratio of bank capital to liabilities (e). Given the capital requirement, d = 1/(1 + e) of deposits are required to finance a project a1(s). Equation 5.6 indicates that the capital requirement will reduce the government’s payment when the bank defaults, al- though it will not eliminate the payment for low enough realizations of the project.  r∗  (5.6) v1(s) ≡ max 0, − a1(s)  1+ e    The capital requirement works as an instrument of prudential regulation in the presence of a deposit guarantee. But as Black, Miller, and Posner (1978, p. 387) note, the government and banks will be in a natural conflict over the amount of capital in the pres- ence of the deposit guarantee. This conflict tends to weaken the reg- ulatory impact of the capital requirement, since banks will take measures to overstate regulatory capital. Equally important, banks can choose to expand their loan portfolios in a way that adds risk rather than diversifies away risk. Banks will do this even in the pres- ence of a capital requirement, as the following example demon- strates. Suppose that the representative bank is now assumed to have a new investment opportunity (which may or may not be so- cially profitable) that requires one unit of resources and generates a state-contingent return a2(s): s ∫ a (s)f (s)ds > ∗ (5.7) 2 r ∗ − ∫ [v (s) − v (s)]f (s)ds. 0 2 1 Equation 5.10 indicates that bankers may have the incentive to un- dertake new socially unprofitable projects (where the expected re- turn is less than r*) whenever the new projects raise the overall risk- iness of the bank’s portfolio and, hence, the expected value of the government’s deposit guarantee. A high capital requirement on the bank will not prevent the bank from choosing a new project that s increases the value of the government’s guarantee  v2 (s) − v1(s)  0 ∫[ ]  f (s)ds > 0 . In fact, to the extent that new projects have a low ex-   s  ∫ ∗ pected return  a (s)f (s)ds ≈ r  , projects that are chosen by the  2  0  178 PHILIP L. BROCK banker will be those that increase the riskiness of the portfolio. This result is known as asset-substitution moral hazard. Gennotte and Pyle (1991), Besanko and Kanatas (1996), and Calem and Rob (1999) find that raising capital requirements may also increase risk- taking by banks due to managerial moral hazard. Peltzman (1970) notes that although bank regulators can either require problem banks to reduce portfolio risk or to increase capital, they rarely intervene directly in banks’ portfolio decisions, since they lack the private information that bankers use when making loans. As a result, most efforts are spent on capital regulation. By 1988 these efforts resulted in the Basel Accord’s risk-weighted capital require- ments. A number of economists have noted that since the risk- weighted capital requirements do not address the overall riskiness of bank portfolios, they will not generally solve the problem of socially unproductive investments by banks in the presence of deposit guar- antees. Risk-weighted capital requirements under the Basle Accords have given rise to what Jones (2000) and Mingo (2000) call “regu- latory capital arbitrage” that has lowered the effective risk-based capital requirements of many banks to well below the 8 percent stan- dard set by the Accords. Apart from the difficulties associated with capital requirements (whether risk-weighted or not) in richer countries, there are opera- tional difficulties in transferring these requirements to the banking environments of many developing countries. Kane (1994) concludes that the transfer of the standards of the Basle Accords to these coun- tries was “economically inappropriate and politically infeasible” due to insufficient information-gathering capacity, deficiencies in the legal systems, and lack of arms-length regulatory enforcement. Caprio and Honohan (1999) also note that in many developing countries bank capital is impaired by failure to declare loans non- performing and by relying on financially related firms to buy a bank’s stock, thereby lowering the true capital of the bank. Given the difficulties associated with the use of risk-based deposit insurance premiums or risk-based capital requirements, a number of economists have called for the use of more “blunt instruments” to lower the size of the contingent liability associated with deposit in- surance, especially for developing countries. One of these instru- ments is deposit rate ceilings. A number of recent studies all present findings that indicate the potential drawbacks of deregulated inter- est rates for bank risk.11 Hellman, Murdock, and Stiglitz (2000) have recently argued that (possibly nonbinding) interest rate ceilings in combination with capital requirements can eliminate excessive risk-taking by banks that is connected to deposit insurance. CORRECTIVE TAXES AND QUASI-TAXES 179 An alternative view of deposit rate ceilings comes from econo- mists who stress the role of market discipline (allowing depositors potentially to incur capital losses) as an alternative to bank regula- tion. In this view, placing depositors at risk not only lowers the gov- ernment’s contingent deposit liability, but it disciplines banks by re- quiring them to pay a risk premium based on their asset portfolio (Demirgüç-Kunt and Huizinga 2000). Martinez Peria and Schmuk- ler (2001) show that depositors did in fact exert market discipline (resulting in higher interest rates and withdrawal of deposits) during financial crises in Argentina, Chile, and Mexico during the 1980s and 1990s. These empirical results correspond to periods of gener- alized financial distress in which the ability of the government to make good on its deposit guarantees was in question. On the other hand, Billett, Garfinkel, and O’Neal (1998) and Jordan (2000) ques- tion the usefulness of depositor discipline on an individual bank basis. They note that when bank risk increases, banks find it easy to substitute insured deposits for departing uninsured funds. The em- pirical evidence for the United States is that by offering small pre- miums over the market rate on insured deposits, banks at risk have been able successfully to neutralize market discipline. In essence, the problem with deposit rate ceilings to promote sta- bility (as in Hellman, Murdock, and Stiglitz 2000) or with freely de- termined deposit rates to promote market discipline is that deposit insurance tends to create a highly elastic supply of deposits to indi- vidual banks. In terms of this chapter’s analytical model, depositor discipline at the bank level requires that depositors expend re- sources on monitoring banks, as in equation 5.2. Deposit insurance weakens the incentive to expend these resources when the deposit insurance system is well capitalized. Depositor discipline will occur at a system-wide level if there is an upper limit (v –) on government insurance payments that would result in depositor losses in some states of the world. In this chap- ter’s model, the supply of deposits to an individual bank is perfectly elastic, so that deposit rate ceilings will never be binding and de- positor discipline will never apply to a single bank. If the govern- ment has a ceiling (v –) on its deposit guarantee funding, the guaran- tee will be a risky one for depositors. Equation 5.12 states that the government will pay out no more than (v –) even if the shortfall in the promised payment to depositors [r/(1 + e) – a1(s)] exceeds (v –). Higher deposit rates will reflect the possible government default on its deposit guarantees. Equations 5.11 and 5.12 with guarantor risk are the counterparts to equations 5.3 and 5.4 without guarantor risk. As a result of the cap (v –) on payments to depositors in equa- 180 PHILIP L. BROCK ˆ1 < v1), so that depositors tion 5.12, the guarantee is less valuable (v respond by demanding a higher contractual interest rate (r > r*). s  r  r∗ (5.11) ∫ 0 min  1 + e , a1(s) + v ˆ 1(s)f (s)ds =  1+ e where   r  (5.12) ˆ 1(s) ≡ max 0, min v − a1(s), v  .    1 + e  If the aggregate investment of banks in new projects increases the government’s expected liability associated with the deposit guaran- tee, then the deposit rate will exceed the riskless rate. Depositor dis- cipline will respond to the potential insolvency of the government’s deposit insurance fund rather than to the portfolio choices of any single bank. Another “blunt” instrument is quantitative restrictions on the ex- pansion of credit by banks. These restrictions have often taken the form of credit ceilings, which are effectively 100 percent marginal reserve requirements on new deposits (so new deposits cannot be used for loans). Another form of quantitative restrictions is capital controls that limit the amount of borrowing a bank can undertake (thereby limiting the expansion of loans). A major problem with capital controls is that they eliminate the funding of both good loans and bad loans, by good banks and bad banks. Quantitative restric- tions give rise to lobbying pressures by bankers on those bureaucrats who are in charge of the quantitative controls. Black, Miller, and Posner (1978) note that this kind of regulation suppresses competi- tion beyond the level needed for regulatory purposes and also tends to expand its reach into other areas (such as controls over which sec- tors of the economy can be financed). Barth, Caprio, and Levine (2003) have expanded upon this critique of government regulatory failure with what they call the “grabbing hand” approach to gov- ernment banking regulation (as opposed to the “helping hand” view of a benign government that corrects market failures and limits the negative side effects of deposit insurance). Becker (1983) and others have shown that efficiency-enhancing actions on the part of the government are more apt to be under- taken than those actions that are purely redistributive. In their model, deposit insurance is efficiency-improving because it econo- mizes on total monitoring costs of banks. Along with the gain in efficiency is an added opportunity for “rent seeking” by interest CORRECTIVE TAXES AND QUASI-TAXES 181 groups. Guo (1999) provides convincing evidence of rent seeking in the political, bureaucratic, and legal delays surrounding the Federal Savings and Loan Insurance Corporation’s resolution of failed sav- ings and loan institutions in the United States in the 1980s. Krosz- ner and Strahan (2001) discuss legislative strategies to implement efficiency-enhancing bank regulatory reforms while simultaneously blunting rent seeking by the private sector and “grabbing hand” be- havior by the government. Taxation of Banks for Non-Prudential Reasons There is virtually no literature on the interaction of the taxation of banks with the government’s deposit guarantee. The single excep- tion appears to be Santomero and Trester’s (1994) analysis of the in- teraction between different national taxation practices and a com- mon deposit insurance fund in a unified Europe. Since there is little literature to cite, this section examines the implications of taxation within the model developed in this chapter. In this model, the incidence of taxation will fall either on bank owners or on the government as provider of deposit insurance.12 Sup- pose that the government taxes the gross return on banks’ projects at the rate (t). The banker’s expected profits (π) are the following: s  r∗  (5.13) 0 ∫ π ≡ max (1 − t)a1(s) −   , 0f (s)ds. 1+ e   In states where the banker pays off depositors, the tax lowers banker’s income and expected profits. In states where the banker defaults, equation 5.14 indicates that the tax raises the size of the government’s contingent liability:  r∗  (5.14) vt (s) ≡ max 0, − (1 − t)a1(s).  1+ e    In one limiting case (where banks never default), the tax just reduces bank profits. In the other limiting case (where banks always de- fault), the tax results in a one-for-one increase in the government’s deposit insurance liability. When the bank defaults in some (but not all) states of the world, the tax reduces bank profits and simultane- ously increases the government’s expected deposit insurance liabil- ity. This result is similar to Caminal’s (2003) result that greater tax- ation of deposits raises the probability of bank failure, although Caminal does not explicitly consider the impact of deposit taxation on the government’s deposit insurance liability. 182 PHILIP L. BROCK Demirgüç-Kunt and Huizinga (2001) analyze the ability of for- eign banks to engage in profit shifting as a way of minimizing the impact of taxation. Banks engage in a different type of profit shift- ing when the incidence of bank taxes is ultimately borne by the de- posit insurance fund and taxpayers who finance the fund. To the extent that the shifting is intertemporal, the resources that the gov- ernment gets from taxing banks will ultimately be paid for partially by future taxpayers. In that sense, what appears to be tax revenue from banks may approximate state-contingent government debt (since the size of the deposit guarantee payout will generally be state- contingent). The tax also alters the banker’s incentive to invest in projects. Equation 5.14 indicates the conditions under which a bank will find it privately profitable to undertake a project in the presence of a project tax and a deposit guarantee: s s ∫ (1 − t)a (s)f (s)ds ≥ r − ∫ v (s)f (s)ds. ∗ (5.15) 1 t 0 0 The direct effect of the tax on projects’ returns may reduce the ex- pected after-tax return to less than the world interest rate, thereby decreasing the attractiveness of investment. But the incidence of the tax on the deposit insurance fund also raises the value of the deposit guarantee, thereby increasing the attractiveness of marginal projects with high default probabilities. The direct and indirect effects work in opposing directions. Reserve Requirement Interactions with Deposit Insurance Reserve requirements are a form of bank taxation when they are nonremunerated or remunerated at less than market interest rates. Yet at the same time reserve requirements affect the portfolio com- position of a bank, reducing the amount of loans for a given equity ratio. Reserve requirements are therefore a hybrid between an indi- rect tax on banks and an additional equity instrument. As such, they have an ambiguous impact on the size of the contingent liability as- sociated with deposit insurance. One use of reserve requirements is to control capital inflows. A number of authors have analyzed the welfare impact of reserve re- quirements as capital controls.13 Chile and Colombia each raised re- CORRECTIVE TAXES AND QUASI-TAXES 183 serve requirements in response to increased capital inflows at various periods during the 1980s and 1990s. These higher reserve require- ments were therefore marginal requirements that affected only new sources of external funding for banks. The following analysis as- sumes a similar use of reserve requirements as they affect the mar- ginal investment decisions by banks. Unlike other previous analyses, the model highlights the interaction between the corrective and dis- torting aspects of reserve requirements on new investment decisions in the presence of deposit guarantees. Suppose that the bank has invested in a project a1(s) and now wishes to invest in a project a2(s). The government wishes to curtail excessive investment in new projects and decides to subject all new deposits to a reserve requirement (k). The government pays a return ˆ) on required reserves, where 0 ≤ r (r ˆ ≤ r*. In order to finance the second project the bank must now set aside required reserves (kd) as well as add equity (ed) to bank capital. Solving the balance sheet equation 1 + kd = d + ed shows that the bank will require d = 1/(1 + e – k) of new deposits to finance the new project. The reserve requirement will alter the expected value of the gov- ernment’s deposit guarantee in two ways. First, to the extent that the government remunerates required reserves at a rate less than the deposit rate, the reserve requirement will be a tax on banks. Second, the reserve requirement acts as a higher marginal equity require- ment on the bank. The equity requirement ed involves e/(1 + e – k) of new bank capital, which is greater than the equity requirement e/(1 + e) on the old project. The greater equity stake of the banker reduces the banker’s incentives to undertake a risky project. The offsetting tax-like and equity-enhancing effects of the reserve requirement are tied to the term (θ) capturing the bank’s promised payment to depositors associated with the second project r*/(1 + e – k) net of any remuneration on required reserves r ˆ k/(1 + e – k): r ∗ − rk ˆ (5.16) θ≡ . 1+ e − k The expression θ in equation 5.16 is the net cost to the bank of debt financing of the second project (that is, the cost of deposits r* minus whatever the government pays on required reserves r ˆ k). To the ex- tent that required reserves pay less than the international interest ˆ < r*), the government’s policy acts like a tax to raise the cost rate (r of debt financing above the riskless rate. The reserve requirement will affect the expected value of the gov- ernment’s deposit guarantee. In comparison with the guarantee v2(s) 184 PHILIP L. BROCK without reserve requirements (equation 5.9), the guarantee will now take the following form:  r∗  (5.17) ˆ 2 (s) ≡ max 0, v + θ − a1(s) − a2 (s).  1+ e    Increasing the reserve requirement will have an ambiguous effect on the size of the guarantee.14 On one hand, if reserves are remuner- ated at the international interest rate, the value of the guarantee is decreasing in the size of the reserve requirement: s ∂θ −er ∗ ∂ (5.18) ∂k|r ˆ =r∗ = (1 + e − k)2 <0 ⇒ ∂k|r ˆ =r∗ ∫ vˆ (s)f (s)ds < 0. 0 2 On the other hand, if required reserves are not remunerated (rˆ = 0), then the government’s guarantee is increasing in the size of the re- serve requirement: s ∂θ r∗ ∂ (5.19) ∂k|r ˆ =0 = (1 + e − k)2 >0 ⇒ ∂k|r ˆ =0 v ∫ ˆ 2 (s)f (s)ds > 0. 0 Non–interest-bearing reserve requirements will only lower the gov- ernment’s deposit guarantee if they tax the project enough to keep the bank from financing it. The tax effect is captured in equation 5.20 by the term (r * – rˆ )k/(1 + e – k) ≥ 0, which is the opportu- nity cost to the bank of holding required reserves. The bank will un- dertake the new project if the following condition holds: s s (r ∗ − r ∫ [vˆ (s) − v (s)]f (s)ds, ˆ)k ∫ ∗ (5.20) a2 (s)f (s)ds > r + − 2 1 1+ e − k 0 0 where v1(s) ≡ max[0, r*/(1 + e) – a1(s)] is the government’s contin- gent liability associated with the original project. If required reserves are paid the international rate, the tax term disappears and the reserve requirement then acts to offset the mar- ginal risk-taking incentives associated with the deposit guarantee that are shown in equation 5.10. With the payment of interest on reserves at the international rate, equation 5.20 differs from equa- tion 5.10 only by the higher marginal equity requirement. If re- quired reserves are not remunerated, then there are two offsetting CORRECTIVE TAXES AND QUASI-TAXES 185 factors affecting investment in the second project. As in equation 5.15, the tax term discourages investment in all new projects, but the increase in the expected size of the guarantee associated with the tax encourages risky project selection for those investments that are undertaken. Other Topics There are several other issues related to the corrective taxation of banks due to government deposit guarantees. The first is charter value. One way to mitigate the moral hazard associated with deposit insurance is to give banks something valuable that they will lose if they become insolvent. Marcus and Shaked (1984) showed at a the- oretical level the importance of charter value in lowering the incen- tive of banks to gamble for resurrection when their underlying loan portfolio is weak. Keeley (1990) provided strong evidence that in- creased competition in the U.S. banking industry in the 1980s caused an erosion of charter values that caused banks to raise the riskiness of the loan portfolios and to reduce their capital. Other more recent papers, such as Matutes and Vives (2000), also find links between the degree of competition in the banking industry and the level of risk-taking. Limiting competition to reduce moral hazard from de- posit insurance is a form of taxation on borrowers and depositors. The use of regulation to restrict competition in banking is often counter-productive, partly because sheltered banks may become “too big to fail” as well as politically powerful enough to “capture” their regulators. The second issue is liquidity requirements. In the presence of de- posit guarantees or a lender of last resort, banks will have dampened incentives to hold liquid assets to meet liquidity shocks. Liquidity re- quirements then become a corrective quasi-tax on banks. Since emergency liquidity provision is traditionally the role of the central bank, liquidity requirements on banks involve the combined func- tions of deposit insurance and lender of last resort. Recent papers by Williamson (1998), Repullo (2000), and Sleet and Smith (2000) have addressed liquidity provision by a lender of last resort in the presence of deposit insurance. In addition to liquidity requirements as a corrective quasi-tax, penalty rates on discount windows func- tion as a corrective tax to deter overly illiquid asset portfolios. In these papers coordination between the deposit insurance agency and the central bank is an important aspect of the corrective measures. A final issue is state-contingent taxation of banks. Dewatripont and Tirole (1994), Nagarajan and Sealey (1998), and Marshall and 186 PHILIP L. BROCK Prescott (2001) suggest that taxation of banks contingent on the macroeconomic environment provides a way of separating idiosyn- cratic risk of bank failure (for which banks should be held account- able) from macroeconomic risk (which may be beyond the control of banks). If banks are subject to ex post state-contingent penalties, then these penalties can be an effective deterrent to overly risk- taking behavior. A drawback is that with limited liability the state- contingent penalties will have a dampened impact on banks’ behav- ior. Kaufman and Wallison (2001) have pointed out that the current U.S. deposit insurance system is essentially one with state-contingent deposit insurance premiums on the banking industry as a whole. Conclusion This chapter has developed a conceptually straightforward model to pull together the large literature on corrective taxation of financial intermediaries in the presence of government deposit insurance. Al- though deposit guarantees can improve economic efficiency by pro- tecting the payments system and by reducing duplication of moni- toring costs by the private sector, they can also create regulatory moral hazard by increasing banks’ incentives to invest in socially unproductive projects and to add risk to asset portfolios. These un- wanted byproducts of deposit insurance can be mitigated by the use of deposit insurance premiums, bank capital requirements, interest rate ceilings, credit ceilings, and reserve requirements. Each of these corrective instruments has limitations and unfavorable side effects, so that there is no “one size fits all” set of prudential regulations to correct the moral hazard associated with the government’s provi- sion of deposit insurance. Perhaps the most surprising finding of the chapter is the scarce attention that has been paid to the impact of noncorrective bank taxes on the contingent liability associated with deposit guarantees. To the extent that the incidence of bank taxes is shifted onto a deposit insurance fund, tax revenue may in certain instances (especially when banks are close to insolvency) be es- sentially equivalent or isomorphic to the proceeds from the sale of state-contingent government debt. Notes 1. Countries adopting these programs have included Bahrain, Chile, Colombia, Congo, Denmark, Ecuador, El Salvador, Greece, Ireland, Italy, Jamaica, Kenya, Korea, Mexico, Nigeria, Peru, Portugal, Sri Lanka, Swe- CORRECTIVE TAXES AND QUASI-TAXES 187 den, Switzerland, Tanzania, Turkey, the United Kingdom, and Venezuela (Demirgüç-Kunt and Detragiache 2000). 2. Among the books are Benston and others (1986); Brock (1992); De- watripont and Tirole (1994); Lindgren, Garcia, and Saal (1996); Calomiris (1997); Eichengreen (1999); and World Bank (2001). Some recent articles include Bhattacharya, Boot, and Thakor (1998); Brock (2000); Kane (2000); Barth, Caprio, and Levine (2001); Claessens and Klingebiel (2001); Freixas and Santomero (2001); Stiglitz (2001); and Kuritzkes, Schuermann, and Weiner (2002). 3. Although the use of corrective taxes and quasi-taxes to offset the im- pact of deposit insurance forms the vast bulk of the literature (especially the recent literature) on bank regulation, there are other potential areas for cor- rective taxes. These may include, with the usual caveats, market concentra- tion, macroeconomic instability, and income distribution. This chapter is concerned solely with corrective taxation as it applies to deposit guarantees and the associated problems of informational asymmetries. 4. Debt contracts are a type of insurance contract, with a borrower’s collateral serving as a deductible and the loan cost acting as the premium for the policy. Lenders are like insurers who receive interest and fees for the loan, but who also bear the risk of loss if the borrower defaults. 5. In related work, de Meza and Webb (1999) show that a balanced budget tax-and-subsidy scheme may be Pareto-improving in the presence of adverse selection in credit markets. In earlier work, de Meza and Webb (1987) showed that taxation in the presence of adverse selection in finan- cial markets may enhance efficiency but did not address the issue of Pareto- improving tax/subsidy schemes. 6. Calomiris (1989) provides convincing evidence that prior to 1933 state deposit insurance systems in the United States had a similar purpose “to insulate the economy’s payments system from the risk of bank failures.” Deposit insurance can also protect small savers. Bradley (2000) documents that insured U.S. postal savings bank accounts for small savers both pre- ceded and partially motivated federal deposit insurance coverage in 1933. As Rochet (1999) has pointed out, banks are different from other firms be- cause their creditors are also their customers. The creditors of other firms are larger, more sophisticated, and better able to assess firm risk than small depositors of banks. For banks, the government acts as the representative of small depositors in collecting information and establishing norms of prudential regulation. Dewatripont and Tirole (1994) have made this rep- resentation hypothesis the core concept in their theory of prudential reg- ulation of banks. Benston (2000) also stresses that “government-provided deposit insurance can achieve economies of information for consumers, who need not incur the costs of investigating and monitoring banks or the deposit insurance contracts that they offer.” 188 PHILIP L. BROCK 7. The model is patterned loosely after James’s (1988) analysis of “off- balance-sheet” banking. Deposit guarantees are a type of off-balance-sheet activity. 8. Assume for the moment that the project is entirely debt financed. 9. Assume that these monitoring costs on a per depositor basis are neg- ligible, so that γ / n ➞ 0 as the number of depositors (n) per project becomes large. 10. Chan, Greenbaum, and Thakor (1992); Giammarino, Lewis, and Sappington (1993); Nagarajan and Sealey (1998); and Freixas and Rochet (1998). 11. Bhattacharya (1982); Smith (1984); Aharony, Saunders, and Swary (1988); Bundt, Cosimano, and Halloran (1992); and Honohan (2001). 12. This is because depositors must be paid an expected gross return of r* and banks own their projects (so that there are no borrowers). 13. These include Chinn and Dooley (1997); Aizenman and Turnovsky (1999); De Gregorio, Edwards, and Valdés (2000); and Herrera and Valdés (2001). 14. The algebraic expression for the ambiguous effect is: ∂θ/∂k = r* – (1 + e)rˆ/[1 + e – k]2. References Aharony, Joseph, Anthony Saunders, and Itzhak Swary. 1988. “The Effects of DIDMCA on Bank Stockholders’ Returns and Risk.” Journal of Banking and Finance 12 (3): 317–31. Aizenman, Joshua, and Stephen Turnovsky. 1999. “Reserve Requirements on Sovereign Debt in the Presence of Moral Hazard—on Debtors or Creditors?” NBER Working Paper 7004. Cambridge, Mass. Banerjee, Abhijit, and Tim Besley. 1990. “Moral Hazard, Limited Liability and Taxation: A Principal-Agent Model.” Oxford Economic Papers 42 (1): 46–60. Barth, James, Gerard Caprio, Jr., and Ross Levine. 2001. “Banking Systems around the Globe: Do Regulation and Ownership Affect Performance and Stability? In Frederic S. Mishin, ed., Prudential Regulation and Su- pervision: Why It Is Important and What Are the Issues. Cambridge, Mass.: National Bureau of Economic Research. ———. 2003. “Bank Regulation and Supervision: What Works Best?” Journal of Financial Intermediation. Forthcoming. Becker, Gary. 1983. “A Theory of Competition Among Pressure Groups for Political Influence.” The Quarterly Journal of Economics 98 (3): 371–400. Benston, George. 2000. “Is Government Regulation of Banks Necessary?” Journal of Financial Services Research 18 (2–3): 185–202. CORRECTIVE TAXES AND QUASI-TAXES 189 Benston, George, Robert Eisenbeis, Paul Horvitz, Edward Kane, and George Kaufman. 1986. Perspectives on Safe and Sound Banking. Cam- bridge, Mass.: MIT Press. Besanko, David, and George Kanatas. 1996. “The Regulation of Bank Cap- ital: Do Capital Standards Promote Bank Safety?” Journal of Financial Intermediation 5 (1): 160–83. Bhattacharya, Sudipto. 1982. “Aspects of Monetary and Banking Theory and Moral Hazard.” Journal of Finance 37 (2): 371–84. Bhattacharya, Sudipto, Arnoud Boot, and Anjan Thakor. 1998. “The Eco- nomics of Bank Regulation.” Journal of Money, Credit, and Banking 30 (4): 745–70. Billett, Matthew, Jon Garfinkel, and Edward O’Neal. 1998. “The Cost of Market versus Regulatory Discipline in Banking.” Journal of Financial Economics 48: 333–58. Black, Fisher, Merton Miller, and Richard Posner. 1978. “An Approach to the Regulation of Bank Holding Companies.” Journal of Business 51 (3): 379–412. Boyd, John, Chun Chang, and Bruce Smith. 1998. “Deposit Insurance: A Reconsideration.” Working Paper 593. Federal Reserve Bank of Min- neapolis. Bradley, Christine. 2000. “A Historical Perspective on Deposit Insurance Coverage.” FDIC Banking Review 13 (1): 1–25. Brock, Philip. 1992. If Texas were Chile: A Primer on Banking Reform. San Francisco: ICS Press. ———. 2000. “Financial Safety Nets: Lessons from Chile.” The World Bank Research Observer 15 (1): 69–84. Bundt, Thomas P., Thomas F. Cosimano, and John A. Halloran. 1992. “DIDMCA and Bank Market Risk: Theory and Evidence.” Journal of Banking and Finance 16 (6): 1179–93. Calem, Paul, and Rafael Rob. 1999. “The Impact of Capital-Based Regu- lation on Bank Risk-Taking.” Journal of Financial Intermediation 8 (4): 317–52. Calomiris, Charles. 1989. “Deposit Insurance: Lessons from the Record.” Federal Reserve Bank of Chicago Economic Perspectives, 13 (May/ June):10–30. ———. 1997. The Postmodern Bank Safety Net: Lessons from Developed and Developing Economies. Washington, D.C.: The AEI Press. Caminal, Ramon. 2003. “Taxation of Banks: Modeling the Impact” (chap- ter 3, this volume). Caprio, Gerard, and Patrick Honohan. 1999. “Restoring Banking Stability: Beyond Supervised Capital Requirements.” Journal of Economic Per- spectives 13 (1): 43–64. Caprio, Gerard, and Daniela Klingebiel. 1997. “Bank Insolvency: Bad Luck, Bad Policy, or Bad Banking?” In Michael Bruno and Bruno 190 PHILIP L. BROCK Pleskovic, eds., Annual World Bank Conference on Development Eco- nomics 1996. Washington, D.C.: World Bank. ———. 2002. “Episodes of Systemic and Borderline Financial Crises.” In Daniela Klingebiel and Luc Laeven, eds., Managing the Real and Fiscal Effects of Banking Crises. World Bank Discussion Paper No. 428. Wash- ington D.C. Carr, Jack L., and G. Frank Mathewson. 1988. “Unlimited Liability as a Barrier to Entry.” Journal of Political Economy 96 (4): 766–84. Chan, Yuk-Shee, Stuart I. Greenbaum, and Anjan Thakor. 1992. “Is Fairly Priced Deposit Insurance Possible?” The Journal of Finance 47 (1): 227–45. Chinn, Menzie, and Michael P. Dooley. 1997. “Financial Repression and Capital Mobility: Why Capital Flows and Covered Interest Rate Differ- entials Fail to Measure Capital Market Integration.” Monetary and Economic Studies 15 (4): 81–103. Claessens, Stijn, and Daniela Klingebiel. 2001. “Competition and Scope of Activities in Financial Services.” World Bank Research Observer 16 (1): 19–40. De Gregorio, José, Sebastian Edwards, and Rodrigo Valdés. 2000. “Con- trols on Capital Inflows: Do They Work?” Journal of Development Eco- nomics 63 (1): 59–83. de Meza, David, and David Webb. 1987. “Too Much Investment: A Prob- lem of Asymmetric Information.” Quarterly Journal of Economics 102 (2): 281–92. ———. 1999. “Wealth, Enterprise, and Credit Policy.” Economic Journal 109 (1): 153–63. Demirgüç-Kunt, Aslı, and Enrica Detragiache. 2000. “Does Deposit Insur- ance Increase Banking System Stability? An Empirical Investigation.” World Bank, Development Research Group, Washington, D.C. Processed (April). Demirgüç-Kunt, Aslı, and Harry Huizinga. 2000. “Market Discipline and Financial Safety Net Design.” World Bank, Development Research Group, Washington, D.C. Processed (April). ———. 2001. “The Taxation of Domestic and Foreign Banking.” Journal of Public Economics 79 (3): 429–53. Demirgüç-Kunt, Aslı, and Edward J. Kane, 2002. “Deposit Insurance Around the Globe: Where Does It Work?” Journal of Economic Per- spectives 16 (2): 175–95. Dewatripont, Mattias, and Jean Tirole. 1994. The Prudential Regulation of Banks. Cambridge Mass.: MIT Press. Diamond, Douglas. 1984. “Financial Intermediation and Delegated Moni- toring.” Review of Economic Studies 51 (3): 393–414. ———. 1996. “Financial Intermediation as Delegated Monitoring: A Sim- ple Example.” Federal Reserve Bank of Richmond Economic Quarterly 82/3: 51–66. CORRECTIVE TAXES AND QUASI-TAXES 191 Diamond, Douglas, and Philip Dybvig. 1983. “Bank Runs, Deposit Insur- ance and Liquidity.” Journal of Political Economy 91 (3): 401–19. Eichengreen, Barry 1999. Toward a New International Financial Architec- ture: A Practical Post-Asia Agenda. Washington, D.C.: Institute for In- ternational Economics. Freixas, Xavier, and Jean-Charles Rochet. 1998. “Fair Pricing of Deposit Insurance. Is it Possible? Yes. Is it Desirable? No.” Research in Eco- nomics 52 (3): 217–32. Freixas, Xavier, and Anthony Santomero. 2001. “An Overall Perspective on Banking Regulation.” University of Pennsylvania. Processed. Gale, Douglas, and Martin Hellwig. 1985. “Incentive-Compatible Debt Contracts: The One-period Problem.” Review of Economic Studies 52 (4): 647–63. Gennotte, Gerard, and David Pyle. 1991. “Capital Controls and Bank Risk.” Journal of Banking and Finance 15 (4–5): 805–24. Giammarino, Ronald, Tracy Lewis, and David Sappington. 1993. “An In- centive Approach to Banking Regulation.” The Journal of Finance 48 (4): 1523–42. Greenwald, Bruce C., and Joseph E. Stiglitz. 1986. “Externalities in Economies with Imperfect Information and Incomplete Markets.” Quarterly Journal of Economics 101 (2): 229–64. Guo, Lin. 1999. “When and Why Did FSLIC Resolve Insolvent Thrifts?” Journal of Banking and Finance 23 (6): 955–90. Hellmann, Thomas, Kevin Murdock, and Joseph E. Stiglitz. 2000. “Liber- alization, Moral Hazard in Banking, and Prudential Regulation: Are Capital Requirements Enough?” American Economic Review 90 (1): 147–65. Herrera, Luis O., and Rodrigo Valdés. 2001. “The Effect of Capital Con- trols on Interest Rate Differentials.” Journal of International Economics 53 (2): 385–98. Honohan, Patrick. 2001. “How Interest Rates Changed Under Financial Liberalization: A Cross-Country Review.” In Gerard Caprio, Patrick Honohan, and Joseph E. Stiglitz, eds., Financial Liberalization: How Far, How Fast? New York: Cambridge University Press. Honohan, Patrick, and Daniela Klingebiel. 2003. “The Fiscal Cost Impli- cations of an Accommodating Approach to Banking Crises.” Journal of Banking and Finance. Forthcoming. James, Christopher. 1988. “The Use of Loan Sales and Standby Letters of Credit by Commercial Banks.” Journal of Monetary Economics 22 (3): 395–422. John, Kose, Teresa A. John, and Lemma W. Senbet. 1991. “Risk-shifting In- centives of Depository Institutions: A New Perspective on Federal De- posit Insurance Reform.” Journal of Banking and Finance 15 (4-5): 895–915. 192 PHILIP L. BROCK Jones, David 2000. “Emerging Problems with the Basel Capital Accord: Regulatory Capital Arbitrage and Related Issues.” Journal of Banking and Finance 24 (1): 35–58. Jordan, John S. 2000. “Depositor Discipline at Failing Banks.” New Eng- land Economic Review March/April: 15–28. Kane, Edward. 1994. “Difficulties of Transferring Risk-Based Capital Re- quirements.” Policy Research Working Paper 1244, World Bank, Wash- ington, D.C. ———. 2000. “Designing Financial Safety Nets to Fit Country Circum- stances.” Boston College. Processed (May). Kane, Edward, and Berry Wilson. 1998. “A Contracting-Theory Interpre- tation of the Origins of Federal Deposit Insurance.” Journal of Money, Credit, and Banking 30 (3): 573–95. Kaufman, George, and Peter Wallison. 2001. “The New Safety Net.” Reg- ulation 24 (2): 28–35. Keeley, Michael. 1990. “Deposit Insurance, Risk, and Market Power in Banking.” American Economic Review 80 (5): 1183–1200. Kroszner, Randall S., and Philip E. Strahan. 2001. “Obstacles to Optimal Policy: The Interplay of Politics and Economics in Shaping Bank Super- vision and Regulation Reforms.” In Frederic S. Mishkin, ed., Prudential Supervision: What Works and What Doesn’t. Chicago: University of Chicago Press and NBER. Kuritzkes, Andrew, Til Schuermann, and Scott Weiner. 2002. “Deposit In- surance and Risk Management of the U.S. Banking System: How Much? How Safe? Who Pays?” Financial Institutions Center Working Paper 02- 02, University of Pennsylvania, Wharton School, Philadelphia. Laeven, Luc. 2002. “The Pricing of Deposit Insurance.” Policy Research Working Paper 2871. World Bank, Washington, D.C. Lindgren, Carl-Johan, Gillian Garcia, and Matthew Saal. 1996. Bank Soundness and Macroeconomic Policy. Washington, D.C.: International Monetary Fund. Marcus, Alan, and Israel Shaked. 1984. “The Valuation of FDIC Deposit Insurance Using Option-Pricing Estimates.” Journal of Money, Credit, and Banking 16 (4): 446–60. Marshall, D., and E. S. Prescott. 2001. “Bank Capital Regulation with and without State-Contingent Penalties.” Carnegie-Rochester Conference Series on Public Policy 54 (1): 139–84. Martinez Peria, Maria Soledad, and Sergio Schmukler. 2001. “Do Deposi- tors Punish Banks for Bad Behavior? Market Discipline, Deposit Insur- ance, and Banking Crises.” The Journal of Finance 56 (3): 1029–51. Matutes, Carmen, and Xavier Vives. 2000. “Imperfect Competition, Risk Taking, and Regulation in Banking.” European Economic Review 44 (1): 1–34. CORRECTIVE TAXES AND QUASI-TAXES 193 Merton, Robert C. 1977. “An Analytic Derivation of the Cost of Deposit Insurance and Loan Guarantees: An Application of Modern Option Pricing Theory.” Journal of Banking and Finance 1 (1): 3–11. ———. 1978. “On the Cost of Deposit Insurance When There are Surveil- lance Costs.” Journal of Business 51 (3): 439–52. Mingo, John. 2000. “Policy Implications of the Federal Reserve Study of Credit Risk Models at Major US Banking Institutions.” Journal of Bank- ing and Finance 24 (1): 15–33. Nagarajan, S., and C. Sealey. 1998. “State-Contingent Regulatory Mecha- nisms and Fairly Priced Deposit Insurance.” Journal of Banking and Fi- nance 22 (9): 1139–56. Peltzman, Sam. 1970. “Capital Investment in Commercial Banking and Its Relationship to Portfolio Regulation.” Journal of Political Economy 78 (1): 1–26. Repullo, Rafael. 2000. “Who Should Act as Lender of Last Resort? An In- complete Contracts Model.” Journal of Money, Credit, and Banking 32 (3): 580–605. Rochet, Jean-Charles. 1999. “Solvency Regulations and the Management of Banking Risks.” European Economic Review 43 (4–6): 981–90. Santomero, Anthony, and Jeffrey Trester. 1994. “Structuring Deposit In- surance for a United Europe.” Financial Institutions Center Working Paper 94-22. University of Pennsylvania, Wharton School, Philadelphia. Sleet, Christopher, and Bruce D. Smith. 2000. “Deposit Insurance and Lender-of-Last-Resort Functions.” Journal of Money, Credit, and Bank- ing 32 (3): 518–75. Smith, Bruce D. 1984. “Private Information, Deposit Interest Rates, and the ‘Stability’ of the Banking System.” Journal of Monetary Economics 14 (3): 293–317. Stiglitz, Joseph E 2001. “Principles of Financial Regulation: A Dynamic Portfolio Approach.” World Bank Research Observer 16 (1): 1–18. Townsend, Robert. 1979. “Optimal Contracts and Competitive Markets with Costly State Verification.” Journal of Economic Theory 21 (2): 265–93. Williamson, Stephen. 1998. “Discount Window Lending and Deposit In- surance.” Review of Economic Dynamics 1 (1): 246–75. World Bank. 2001. Finance for Growth: Policy Choices in a Volatile World. New York: Oxford University Press. Part II Practical Experience 6 Taxation of Financial Intermediation in Industrial Countries Mattias Levin and Peer Ritter This chapter presents a broad overview of the recent changes that have occurred in the industrial countries in taxation of income from capital and other aspects of tax affecting financial intermediation, to- gether with an interpretation of the reform thinking underlying these changes. The chapter first focuses on the overall question of how to tax capital income. It notes how this has become a central issue as rates have declined and tax bases have broadened, and highlights the emergence of dual tax systems as an increasingly popular way for- ward. The chapter then examines the practical issues in the taxation of different financial instruments, noting in particular the problems that arise by retaining for tax purposes a distinction between debt- type and equity-type instruments. It further considers the treatment of new instruments that can blur this distinction, as well as that be- tween income and capital gains, and points out the importance of timing issues in this context. The chapter follows with a brief review of some of the special issues surrounding the taxation of intermedi- aries, documenting the diversity of special taxes, implicit and explicit, to which they can be subject, despite a trend towards elimination. Two general but often conflicting tax criteria have been central in shaping the voluminous body of law on the taxation of financial in- termediation in industrial countries. The ability-to-pay principle rep- 197 198 MATTIAS LEVIN AND PEER RITTER resents a normative judgment about the equitable distribution of the tax burden across individuals. The economic efficiency principle aims to keep the allocative distortions from taxation at a minimum. In aiming at the goals of equity, efficiency, and simplicity, tax sys- tems must also adapt to changing economic circumstances. These changes include the increase in the elasticity of capital flows with re- spect to price, and hence to taxation, following liberalization of capital flows. However, tax systems are difficult to reform. This is partly because of the political consequences of changing the distri- bution of the tax burden, partly because of dogmatism regarding principal features in the tax law, and partly because of the complex interrelationships between the various taxes raised—just to mention a few reasons. Overall, the same principles apply to the taxation of financial in- stitutions as for other companies. However, some issues become much more important, including the timing of realized gains, short- term versus long-term investments, and the valuation of income streams. The treatment of capital income differs sharply across countries. Some countries tax capital income as part of total income, often at progressive rates. Other countries have particular tax rates for cap- ital income, often on a proportional basis. Still other countries do not tax capital gains. Most countries do not tax all dividends equally. Exceptions are often made for shares traded on or off exchanges. Furthermore, in- terest income is often taxed differently depending on the instrument (public bonds are often tax-exempt) or institution (bank deposits are often exempt).1 Numerous special regimes and deductions apply to financial in- vestments, as encouraging these investments often is regarded as an important policy objective: to encourage risk capital or to stimulate pension savings, for example. This chapter describes how industrial countries have grappled with these challenges in reforming their taxation of financial inter- mediation. Despite the inherent difficulties of tax reform, most countries have undergone significant reforms since the 1980s. The main drivers for tax reform since the 1980s have been an eagerness to reduce the distorting effect of high marginal tax rates (rate cut- ting) while preserving financing commitments (base broadening). But taxation of financial intermediation has been affected by a num- ber of additional concerns:2 • Location of financial activity (competition between financial centers) FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 199 • Disintermediation (a growing array of financial institutions) • Demographics (the need to stimulate private savings to fund pensions) • Financial innovation (new instruments shaped to circumvent traditional taxes) Sometimes these reforms have been forward-looking responses to the general challenges outlined above. In most cases, however, reform has been more reactive and provisional, at a minimum lead- ing to a decrease in rates and refinement of systems dealing with globalization (including imputation systems, transfer pricing, anti- avoidance measures, and information exchange). Thus overall, the tax systems in place still bear the same contours as before reform started in the 1980s. But it is increasingly doubt- ful whether that kind of reactive response is sufficient in the face of current challenges. An alternative systematic response to the chal- lenges faced by industrial countries are dual income tax systems, which tax all returns from capital at a lower, proportional rate com- pared to ordinary income, which remains taxed at progressive and higher rates. A more fundamental approach would be to opt for a consumption-based tax system to ensure neutrality toward financ- ing and intertemporal allocation. Taxation of Capital and Other Income A central issue in the taxation of financial intermediation is the treatment of capital income for tax purposes (see also chapter 2). Before looking at the major reform trends in this regard, it will be convenient to review the relevant concepts of income. Definition of Income As a measure of the ability-to-pay, the Haig-Simons (H-S) concept of income—broadly, the sum of current consumption and changes in net worth—has remained at center stage in taxation in the indus- trial countries.3 In principle this concept is meant to be independent of the source (labor or capital). As regards changes in net worth, they can be taxed as they accrue or when they are realized. The debate around this concept of income is centered on three is- sues that are relevant to financial instruments. The economic sense of the H-S concept has been called into question: for example, by stressing that it may lead to an intertemporal misallocation of re- sources through the taxation of changes in wealth. Even if one ac- 200 MATTIAS LEVIN AND PEER RITTER cepts the H-S concept and thus taxation of changes in wealth, the choice between taxation on the accrual or realization basis remains. First, should changes in net wealth be taxed? Proponents defend the H-S concept by claiming that it corresponds to the principle of “ability to pay.” A change in wealth would constitute a change in the amount that could possibly be spent on consumption and should thus be taxed. Critics argue that this view is static; it ignores that wealth is deferred consumption and hence the savings that build up wealth are out of current income that is already taxed. These critics have instead suggested tax systems built on current consumption or on cash flows, exempting the opportunity cost of capital from tax. Second, should capital income and other income be taxed at the same rate? Many countries distinguish between income from capital assets and other income by applying different rates. This distinction also relates to the differential treatment of debt and equity in most tax systems. At the root of this distinction is the belief that owner- ship should be taxed differently from other sorts of income (includ- ing the lending of capital). Financial instruments can be designed to avoid such a distinction. In contracts, dual income tax systems sep- arate labor from other income with the allocative argument that in order to minimize tax distortions, factors with the lower elasticity of supply should be taxed higher. The schedular approach of the dual income tax departs systematically from the H-S approach. Third, should the taxation of the net change in wealth be on ac- crual or on realization? The H-S dictum that changes in net wealth are to be taxed is still prevailing in industrial countries, although in practice capital gains are often not taxed. Many suggestions for re- form of financial instruments taxation focus on the measurement of changes in net wealth. The traditional way, taxation on realization, has been defended mostly on practical considerations; in particular that valuation is often difficult. These considerations may be valid for traditional assets like land or paintings by old masters. Financial instruments however can be used to defer realization and also often do not even encompass an asset but only payment flows. Hence fi- nancial instruments may require a reconsideration of the traditional approach to measure capital gains. Tax Reform in the 1980s: Reducing Distortions A steady increase in tax rates (to help finance growing welfare com- mitments) and deteriorating economic performance in the 1970s helped trigger a more critical assessment of the distorting effects of FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 201 Table 6.1. Cutting Tax Rates Personal income tax Corporate income tax Top marginal rate (no. of brackets) Top marginal rate 1970s 1989 2001 1970s 1980s 2002 US 70 (14) 28 (2) 38.6 (5) 46 34 40 DE 56 (4) 54.5 (4) 48.5 (4) 56 50 38.4 UK 83 (11) 40 (3) 40 (3) 52 35 30 FR 64.7 (13) 56.8 (13) 53.3 (6) 50 34 34.3 IT 72 (32) 50 (6) 45 (5) 25 36 40.3 NL 72 (9) 72 (3) 52 (3) 46 35 34.5 Sources: Knoester 1993; European Commission 2001; IBFD, 2001; Van den Noord and Heady, 2001. taxation in many countries during the 1980s (Messere 1999, 2000; Duisenberg 1993; Knoester 1993; Steinmo 1995). Existing systems were seen as deficient in each of the major dimensions. • Efficiency: Statutory marginal rates were high—for example, marginal personal income tax rates were over 80 percent in Sweden, and 70 percent in the United States and the United Kingdom— which led to substantial tax avoidance. • Equity: Tax systems were not as redistributive as expected, considering the high marginal rates. The reason was that the rates were effectively lowered by numerous exemptions and the regres- sive characteristics of social security contributions. • Simplicity: Decades of using the tax system as a means to achieve objectives of a nonfiscal nature had resulted in complex sys- tems with numerous exemptions, allowances, regimes, and so on. As a result, compliance costs and administration costs had soared (Duisenberg 1993). In the reforms of the early 1980s, countries generally tried to ren- der their tax systems more efficient by cutting marginal income tax rates. Countries also made systems simpler by cutting the number of tax brackets (table 6.1). With little reduction in expenditure commitments, the tax changes had to be revenue neutral: that is, not affect the overall tax revenue. Countries therefore broadened the tax base, partly by elim- inating exemptions, special regimes, and so on, and partly by tax- ing more forms of income. This is of particular importance for the focus of this chapter, as countries have turned to taxing capital in- come and capital gains to a larger extent than before in order to compensate for the decrease in tax rates (table 6.2). Some countries 202 MATTIAS LEVIN AND PEER RITTER Table 6.2. Examples of Base-Broadening Measures in the 1980s Country Personal Corporate US • Abolition of favorable treat- • Abolition of Investment Tax ment of capital gains Credit • Repeal of tax-sheltered • Reduction in the deductibil- investments ity of R&D expenditure • Elimination of special business reductions UK • Increase in taxation of earn- • Abolition of 100% first-year ings (for example, benefits allowance for investments in kind such as cars) in plant and machinery • Reduction of tax subsidy on • Depreciation allowances owner-occupied housing for investment brought into • Capital gains brought in line with true economic under income tax depreciation FR — • Elimination of investment tax credits • Elimination of favorable depreciation allowances (actual life) NL • Taxable compensation — allowance • No deduction for social security taxes • Lower personal exemptions Source: Knoester 1993. also changed the composition of their tax base, for example by re- lying more on indirect taxes, primarily VAT (Messere 2000). Reform Since the 1990s: Contours of an Ideal System? Reform since the 1990s has aimed at simplifying the tax system by continuing to rein in exemptions, thus broadening the tax base, while at the same time decreasing the tax rates. Reforms have also tried to deal with increasing capital mobility, attempting to find ways of taxing capital without provoking capital flight (see, for example, OECD 1998). The motives for reforming personal capital income taxes are largely the same as for overall tax reform: • Efficiency: Faced with decreasing levels of gross savings in general and household savings in particular, and as a high level of private savings has become increasingly desirable from a public FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 203 point of view (to provide risk capital, or to cater to individual pen- sion needs), countries have increasingly tried to stimulate financial savings by tax cuts (see Bosworth and Burtless 1992, Feldstein 1995). In some countries this has taken the form of general tax cuts. In others, however, it has taken the form of an increase in special regimes (Gordon 2000). Few countries have, however, radically changed their tax systems in order to stimulate savings. • Equity: Traditionally, some countries have tried to use taxes to redistribute capital wealth (for example, wealth taxes, inheritance taxes) for the sake of vertical equity. In order to achieve horizontal equity, taxes on dividends have in many cases been reduced or abol- ished in order to bring them in line with taxes on interest income (Messere 1999).4 • Simplicity: While the number of income categories and brack- ets has often been reduced, in those countries that have started to tax capital gains the complexity of the tax system has increased rather than decreased. Three general approaches have been employed in the treatment of capital income. Including capital income in personal income taxes Some countries have brought in capital income under the ordinary personal income tax, thus widening the tax base. Under such a system, capital income flows are taxed progressively under the personal income tax. This would be in accordance with the logic of the Haig-Simons approach. On that front, personal income tax rates in the EU remain higher than in other industrial countries. However, tax rates have decreased substantially since the 1980s, with the average top rate falling from about 56 percent in 1983 to 47 percent in 2001.5 Some countries have moved faster and further than others. Belgium and the United Kingdom have cut more than France and Germany, for example. However, in other countries, such as Denmark and Luxembourg, top rates have actually increased.6 While some countries continue to tax capital gains separately, most countries have either incorporated capital gains under income taxation or, as will be further detailed next, started to tax all capital income (gains, dividends, and wealth) at a separate proportional rate. Taxing capital income separately: dual income tax systems The re- forms outlined above, no matter how voluminous, have remained piecemeal rather than comprehensive. Therefore in most cases so far, reform has not offered a coherent response to the challenges posed by globalization. Faced with the increasing mobility of capital, some 204 MATTIAS LEVIN AND PEER RITTER Table 6.3. Dual Income Tax Systems Denmark Finland Norway Sweden Year introduced 1987 1993 1992 1991 Capital income tax rates (percent) • Corporate 32 28 28 28 • Other 39.7–59 28 28 30 Personal income tax rates 39.7–59 22.5–54.5 28–41.5 31–56 Elimination of double taxation of corporate profits • Distributions Yes Yes Yes No • Retentions No No Yes No Withholding taxes on nonresidents • Dividends Yes Yes Yes Yes • Interests No No No No • Royalties Yes Yes No No Sources: Cnossen 2000; IBFD 2001. countries have chosen an alternative path by imposing dual, if not multiple, income taxes. The aim has been to tax capital income at a lower rate in order to prevent tax evasion (Huizinga and Nicodème 2001). In the late 1980s to early 1990s, the Nordic countries (Den- mark, Finland, Norway, and Sweden) adopted such dual income tax (DIT) systems. Under dual income tax systems all income is sched- uled into two types: • Capital income: This includes business profits (return on eq- uity), dividends, capital gains, interest, rents, and rental values. Capital income is taxed at a proportional rate. • Personal income: This includes wages and salaries, fringe ben- efits, pension income, and social security benefits. Personal income is taxed at progressive rates. This led to significant cuts in capital income tax rates and often a decrease in the progressiveness of the personal income tax rate (table 6.3). Why did the traditionally social-democratic Scandina- vian countries decided to launch such a reform? The reasons were a willingness to reduce the distortionary effect of progressive income taxes, to strengthen incentives for private sav- ings, and to eliminate the numerous possibilities for tax arbitrage and tax avoidance deriving from the vast array of exemptions and deductions available on capital income (Nielsen and Sørensen 1997). FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 205 An associated benefit was that reform would lead to higher tax rev- enues as a result of the widening of the base. The Nordic countries have not levied withholding taxes on non- residents’ interest income, or in some cases on royalty income, how- ever (Cnossen 2000). This may give rise to tax avoidance. One rea- son may be that due to the collective dynamics in the EU, such taxes could contribute to capital flight. The attractiveness of dual income systems lies in their relative pragmatism and simplicity. A low and flat rate on capital income has been heralded as a pragmatic way of dealing with the greater elas- ticity of supply of capital, the difficulties of verifying capital income, and maintaining international competitiveness. It is in that respect not surprising that dual income systems have become popular in the Nordic countries, which are small and open economies faced with the significant risk of capital outflow if their capital income taxation is too high. Moreover, by applying the same rate to all sources of pri- vate capital income, the tax system also becomes simpler. The main criticism regards equity. Dual tax rates may give rise to tax shifting, which negatively affects horizontal equity. As capital in- come holders are often wealthier, it also affects vertical equity com- pared to the Haig-Simons concept in place.7 This was indeed the main political problem in implementing these reforms. For example, at the time of the reform, the trade union leader in Sweden com- plained to the social democratic government that he would never be able to defend the reform in front of his members, as it made him (a relatively high-income earner) much better off than under the old tax system. Nevertheless, as the broadening of the tax base and the elim- ination of numerous exemptions from capital income tax led to calls for continued tax relief for special interest groups, the governments were able to defend the reforms in the name of general interest. In 2001 the Netherlands introduced a tax system similar to DIT. Instead of a capital income tax, a wealth tax was established.8 In the beginning of 2003 Germany was debating to move to a system sim- ilar to a DIT. The main attraction of DIT is that while it reduces distortions and takes into account the different elasticities of supply, by keep- ing progressive taxes on personal income and removing numerous exemptions on capital income, policymakers continue to hold in- struments to maintain the tax system as reasonably progressive. Allowance for corporate equity The tax system known as the al- lowance for corporate equity (ACE) has its origin in the cash flow approach to taxation. For a cash flow corporation tax, the idea is to take the difference between sales revenue and expenses as the tax 206 MATTIAS LEVIN AND PEER RITTER base. Expenses would include purchases of capital goods (gross in- vestment). Neither distributed earnings nor interest would be de- ductible. The main advantage is neutrality with respect to financing of investment. In the ACE system, the tax base equals the accounting profit in a given period (net of depreciation and interest payments) minus the allowance for corporate equity (shareholder equity times the “pro- tective rate of interest”—the market rate of interest) at the begin- ning of the period. Deducting the market rate of interest from the corporate equity makes investment choices neutral to whether they are financed by retained earnings, debt, or equity. Positive corporate equity would be equivalent to a deferred payout of investment in- come from the perspective of the cash flow tax, and this return is taxed (net of the market rate of interest) each period. The account- ing profit thus consists of this change in corporate equity plus the dividends paid (as the distribution of generated earnings) minus new equity. The ACE implies that only pure profits are taxed. A major advantage in implementation of the ACE over the cash flow tax is that the ACE would retain usual accounting practices. The ACE system has been implemented in Croatia.9 At the per- sonal level, wage income and the gains from the disposal of real es- tate and other property titles are taxed. There are no capital gains or savings taxes; the taxation of distributed profits and interest (ex- ceeding the return over the market rate of interest) at the company level is final. Since the market rate of interest is tax-free, the system is nondistortive toward the intertemporal allocation of consumption. Leaving capital gains untaxed at the personal level may look like a violation of vertical equity compared to the Haig-Simons concept in place in most countries. However, from an intertemporal per- spective, the ACE system taxes all consumption fully. Those who save more will be taxed when they consume their savings, but will be credited for the opportunity cost of saving. Since individuals are the owners of companies after all, the ACE taxes pure profits only once at the corporate level. As will become clear shortly, the two features of the ACE system, namely intertemporal neutrality and neutrality toward the choice of financing, play an important role in the taxation of new financial in- struments. By and large it is the violation of one or both of these characteristics that makes the taxation of financial instruments so complex. To sum up, in general the tax reforms of the 1980s and 1990s lowered statutory rates and broadened the base. As a result, the ef- fective marginal tax rate at the corporate level has remained stable over this period (Devereux, Griffith, and Klemm 2002). FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 207 Taxation of Financial Instruments A tax system is neutral against financing choices when a given pres- ent discounted pre-tax flow of profits yields the same discounted after-tax income, irrespective of the means of finance. Furthermore, taxation affects the opportunity cost of investment and hence its level by affecting the intertemporal allocation of expenditure. The Traditional Approach In almost all tax systems one or more of the three sources of finance for investment (retained earnings, debt, and equity issuance) experi- ences different tax treatment from the others. As discussed in detail below, the tax treatment of interest often gives an incentive to debt financing, while the distribution of earnings to shareholders is often discouraged by the tax system via the corporate tax rate. This not only favors a deferral of dividend distribution but also opens room for tax arbitrage via new financial instruments that can replicate any given payment pattern, thereby undoing the traditional distinction between debt and equity. To uphold the categorization, such tax ar- bitrage is often countered with anti-avoidance legislation. Most tax systems make a distinction between the returns payable to a financial instrument (that is, between interest and dividend in- come). Moreover, these returns are often taxed at rates different from those applicable to capital gains. However, new financial in- struments can be used to blur the distinction between each of these categories and hence be used to transfer returns from one category to another. Here too, the resulting tax avoidance is often countered by the authorities with anti-avoidance rules and classification of new financial instruments into either the debt or equity category. Even when these distortions are removed and all means of financ- ing carry the same statutory tax rate, a further distortion can arise. This is the timing: the point in time at which a capital gain is con- sidered taxable. In particular, when capital gains are taxed upon re- alization (that is, either the sale or the payment at the terminal date of a particular instrument) there may be an incentive to defer the re- alization of gains (see chapter 2). One reason for this distinction between interest and dividends may be a “proprietary view of the corporation”: interest is paid to outsiders, dividends to owners.10 According to this line of reasoning, dividends are seen as a non-deductible distribution of profits to those who hold control. This has been the rationale for a preferen- tial treatment of debt. If the Modigliani and Miller (1958) suggestion 208 MATTIAS LEVIN AND PEER RITTER that debt and equity are equivalent from a financing point of view were valid, a differential treatment of debt and equity might have consequences on tax revenue but little on efficiency. However, recent corporate governance literature stresses the difference in control rights of the two instruments in determining the corporate financial structure (“financial contracting”) and derives conditions for an op- timal debt-equity ratio. Even when financial structure through em- bedded control rights matters for an efficient allocation of funds, it is not clear whether the government should try to influence the al- location of funds with differential taxation, since this would again necessitate a tenable distinction between debt and equity taxation. Another argument for government intervention in the financing de- cisions could be financial market failures (see chapter 2). Taxation of Interest Income Granting household loans favorable deductions Most industrial countries have traditionally granted households deductions from tax where the underlying loan is for business purposes. However, in- dustrial countries have increasingly granted favorable tax treatment for other investment purposes as well. Some examples can be found in table 6.4. Corporate funding: the beneficial tax treatment of debt As for cor- porate funding decisions, most tax systems in industrial countries favor debt financing. Accordingly, the interest expenses to service the loan is deductible from taxable income (trading profits).11 In contrast, income already included in corporate income taxes is taxed a second time when distributed as dividend income and when the instrument is sold. As a result, tax wedges on debt are much lower than the wedges on new equity or retained earnings (table 6.5). As illustrated below, some studies have found that tax wedges for companies’ funding decisions in the manufacturing sector may be higher for equity than for either retained earnings or debt. Meanwhile, the taxation of interest income has decreased. One reason, apart from an eagerness to stimulate (a particular form of) savings, is the deregulation of capital flows in general and in Europe in particular. The liberalization of capital flows in the EU in the 1980s led countries to decrease taxation on interest income, as can be observed in the figure 6.1, which shows a decline from over 50 percent on average to just 30 percent (Huizinga and Nicodème 2001). For example, following the removal of the last exchange con- trol measures in 1990, France decreased the tax on long-term bond interest income from 27 to 18.1 percent and later to 15 percent (Maillard 1993). Following reforms in 1997–98, Italy’s capital in- FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 209 Table 6.4. Tax Exemption of Interest Expenses Depending on the Purpose of the Loan Business Nonbusiness Principal Secondary Country purposes investment residence residence Other BE ߛ ߛ ߛ — — DK ߛ P P P P DE ߛ ߛ — — — ES ߛ P P — — (if not taxable (up to gross (deductible under objective income arising up to exemption, then from renting certain sum no exemption) immovable and other property) conditions) FR ߛ N P — — (excluding rental real property) IE ߛ ߛ P — — IT P — — — — LU ߛ ߛ P P P NL ߛ ߛ ߛ ߛ — (only if interest is attributable to taxable income) AT ߛ — P P — (only (only construction) construction) UK ߛ — P — — (limits on rate of relief and ceiling on amount of loan) US ߛ ߛ P P — (home acqui- (home acqui- sition debt) sition debt) JP ߛ N N N N Notes: ߛ = Full exemption granted. P = Partial exemption granted. N = Not deductible. AT Austria; BE Belgium; DE Germany; DK Denmark; ES Spain; FR France; IE Ireland; IT Italy; JP Japan; LU Luxembourg; NL Netherlands; UK United Kingdom; US United States. Source: Lee 2002. come taxation is among the lowest of industrial countries, perhaps reflecting Italian authorities’ attempt to lure capital back onshore (OECD 2001). Cnossen (1996) argues that “in reality, most interest is not taxed at all due to the symbiosis between interest deductibil- ity at company (and personal) level and the existence of capital-rich tax-exempt investors, such as pension funds, life insurance compa- nies and social security funds.” 210 MATTIAS LEVIN AND PEER RITTER Table 6.5. Marginal Effective Tax Wedges in Manufacturing, 1999 Retained Sources of financing Standard Country earnings New equity Debt deviationa BE 1.36 2.54 –0.60 1.29 DK 1.89 2.43 2.49 0.27 DE 0.89 2.53 1.28 0.70 ES 3.20 2.23 1.65 0.64 FR 3.58 7.72 0.67 2.89 IE 1.52 4.12 0.69 1.46 IT 1.27 1.27 0.39 0.41 LU 3.57 2.37 1.62 0.80 NL 0.46 5.33 2.46 2.00 AT 0.74 2.65 0.06 1.10 UK 2.88 2.40 1.55 0.55 US 1.66 4.79 1.42 1.54 JP 3.30 5.50 –0.09 2.30 OECDb 2.02 4.03 1.09 1.23 EUb 1.95 3.24 1.01 0.91 Notes: The entries in the table show the “degree to which personal and corporate tax systems scale up the real pre-tax rate of return that must be earned on an in- vestment, given that the household can earn a 4 percent real rate of return on a de- mand deposit” (OECD methodology based on the method in Fullerton and King 1984). Calculations are based on top marginal tax rates and a 2 percent inflation rate. Weighted average of different types of investment: machinery 50 percent, build- ings 28 percent, inventories 22 percent. See table 6.4 for country names. aThe standard deviation of the entries in the other three columns. bWeighted average across available countries (based on 1995 GDP and PPPs). Source: Van den Noord and Heady 2001. Figure 6.1. Average Taxes on Interest Income on Residents in the EU, 1983–2000 55 50 45 40 35 30 25 1983 1985 1987 1989 1991 1993 1995 1997 1999 Source: Huizinga and Nicodème 2001. FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 211 Despite an overall fall in rates, significant divergences remain (table 6.6). Some countries have high rates (exceeding 30 percent in Germany and Switzerland), some have low rates (15 percent in Bel- gium and France), while some countries impose no taxation at all (Denmark, Luxemburg, and the Netherlands). However, these rates must be seen in conjunction with the whole tax system. For exam- ple the German tax rate is basically a withholding tax, which be- comes relevant in personal income taxation only after high exemp- tions. For the majority of savers, this tax would be effectively zero. In the beginning of 2003 Germany was discussing a move to taxing capital income at source at a uniform rate, thus abandoning capital gains taxation under the personal income tax rates. Taxation of Equity Dividends Countries have also tried to bolster portfolio investments in general and the use of equity in particular. This has the associated benefit that it increases the amount of risk capital available to companies. The United Kingdom was a European precursor in this area, with the in- troduction in 1983 of a Business Expansion Scheme aimed at stimulat- ing investments in new firms and the establishment of Personal Equity Plans in 1986 (Leape 1993). Other countries have since followed suit. Reforms aimed at stimulating equity financing Industrial countries have increasingly come to regard the unfavorable tax treatment of equity as problematic. This is especially the case in the EU, where following the launch of the euro, the European Commission, the Eu- ropean Parliament, and member states are actively trying to achieve full integration of capital markets in order to extend financing sources to companies and decrease their cost of capital. Tax rules that punish equity financing are thus increasingly anachronistic. Often the tax system distorts the decision whether to pay out the returns on equity to shareholders or to retain earnings. The distri- bution of corporate returns by dividends is often discouraged by taxing capital gains at a personal level at a lower rate than divi- dends (the case in Austria, Belgium, Germany, Greece, the Nether- lands, Spain, and Switzerland; see table 6.7). Reducing double taxation There are three main alternative ap- proaches to dealing with double taxation of equity (see table 6.8): • Classical system: Countries with this system do not allow the shareholder any credit for corporate income tax paid when dividends are being taxed. Thus there is an element of double taxation. Most such 212 MATTIAS LEVIN AND PEER RITTER Table 6.6. Taxation of Interest Income in Some Industrial Countries Top tax rate on Interest Interest interest income from received withholding tax government bonds Country Gross Net Resident Nonresidenta Resident Nonresident BE — ߛb 15 0–15 15 0 DK ߛ — 0 0 60.5c 0 DE ߛ — 31.65d 0 53.8c 0 HE — — 15–20 10–45 15 7.5 ES ߛe — 18e 0 48c 0 FR — ߛb 15 0–15 15 0 IE — ߛ 24f 0–15 24g 0 IT — ߛ 27 0–15 12.5 0 LU ߛ — 0 0 47.15c 0 NL ߛ — 0 0 60c 0 AT — ߛb 25 0 25 0 PT — ߛb 20 10–20 20 20 FI — ߛ 29 0 29 0 SE — ߛ 30 0 30 0 UK ߛ — 20h 0 20i 0 US ߛj — 0 0 n.a. n.a. CH ߛ — 35 0–35 n.a. n.a. JP — ߛ 20 20 n.a. n.a. —Gross or net interest income not taxed. n.a. Not available. Note: See table 6.4 for additional country names: HE Greece; PT Portugal; FI Fin- land; SE Sweden; and CH Switzerland. aRate depends on double tax treaty. bInterest received net of withholding tax, which is either final or creditable against income tax liability depending on choice of taxpayer. cCapital income included in personal income tax. Reported rate thus top marginal income tax rate. d30% federal withholding tax plus 5.5% solidarity surcharge. eInterest income normally part of ordinary income taxation. If interest generated for longer than two years, only 70% of income subject to income tax. fA reduced rate applies for Special Saving Accounts. Tax refunded in special cir- cumstances (such as charitable organizations). gTax creditable against income tax liability. hInterest from bank and building societies paid net of tax (20%), but may be paid gross for nontaxpayers who register. Some National Saving products exempt, but in- terest on others received gross automatically. Withholding tax on interest creditable against income tax liability. iInterest on some government securities paid net of tax (20%) but may be received gross in certain circumstances. Withholding tax on interest creditable against income tax liability. jInterest from federal securities exempt from state and local taxation; state and local securities exempt from federal tax. Sources: Haufler 2001; Joumard 2001; Lee 2002. FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 213 Table 6.7. Tax Treatment of Dividends and Capital Gains on Shares, 1998, Resident Taxpayers Taxation of dividends Taxation of capital gains Rate Rate Country (%) Rules (%)a Rules BE 15 Withholding tax that 0 Capital gains by individuals can be final (taxpayer’s not engaged in business activity option) not taxable. DK 25 Final withholding tax 40 Rate applies to a taxable base arising from the disposal of shares exceeding DKr 35,000. DE 48.47 Taxed as ordinary gross 0 Capital gains realized through income or as ordinary private transactions of resident income with creditable individuals generally not sub- withholding tax. ject to income tax. (Under review at the time of printing.) HE — — 0 Gains derived from sale of movable property (other than nonlisted companies with lim- ited shares and limited liability companies) are not taxed. ES 28.57 Several possibilities: 56 Treated like ordinary income. treated as ordinary in- For holding periods longer than come, exempt, cred- 2 years, net gain is reduced by itable withholding tax. 25% for each additional year. FR 33.33 Taxed as ordinary in- 26 Capital gains on securities are come with withholding taxed at this flat rate (compris- tax, always creditable ing basic rate of 16% plus so- against ordinary income cial surcharges). tax. IE — Treated as ordinary in- 40 For gains on the disposal of come, 21% dividend im- shares in nonquoted trading putation credit which is companies held for at least 3 always creditable against years, 26%. ordinary income tax. IT 10–12.5 Withholding tax, fully 12.5 Net capital gains on shares and creditable against ordi- other securities are subject to a nary income tax. substitute tax that replaces in- dividual income tax. LU 25 Treated as ordinary in- 46.6 No separate capital gains taxed come. Creditable with- in Luxembourg. holding tax. NL 25 Treated as ordinary in- 0 In general capital gains are not come, creditable with- included in taxable base. holding tax. AT 25 Withholding tax that 0 In general capital gains are not can be final at taxpay- included in taxable base. ers’ option. (Table continues on next page.) 214 MATTIAS LEVIN AND PEER RITTER Table 6.7. Tax Treatment of Dividends and Capital Gains on Shares, 1998, Resident Taxpayers (continued) Taxation of dividends Taxation of capital gains Rate Rate Country (%) Rules (%)a Rules PT 25 Withholding tax that 10 Net annual gains from the dis- can be final at taxpay- posal of shares are in principle ers’ option. subject to a tax at a final rate of 10% unless the transferor opts for its inclusion in his tax- able income. FI 28 Taxed as ordinary in- 28 Income from capital is subject come with creditable only to a national income tax. withholding tax. SE 30 Taxed as capital 30 In general, all capital gains re- income. alized by an individual are in- cluded in the category income from capital. Income from cap- ital is taxed separately nation- ally (no municipal taxes apply). UK — Taxed as ordinary gross 40 Capital gains of an individual income, 20% dividend are aggregated with his income imputation credit and are taxed at income tax which is creditable rates. against ordinary in- come tax liability. US — Taxed as ordinary gross 25 Assets must be held for more income. (Under review than one year, otherwise gains at the time of printing.) are taxed as ordinary income. CH 35 Treated as ordinary in- 0 Capital gains are exempt. come. Creditable with- holding tax. JP 20–35 Depending on amount 21.2 For listed companies, a central of dividend paid by a rate augmented by a local rate. single company: ordi- If sale of asset is trusted to nary income with 20% securities company, separate creditable withholding withholding tax. tax, 35% final with- holding tax, or 20% op- tional withholding tax. —Dividends not taxed. Note: See tables 6.4 and 6.6 for country names. aTop personal tax rate. Source: Van den Noord and Heady 2001. countries levy a withholding tax at source (company) when dividends are being paid. The majority of the EU’s member states, the United States, Switzerland, and Japan operate under this system. (In January 2003 the U.S. administration proposed the exemption of dividends). Table 6.8. Ways of Dealing with Double Taxation Country Classical system Imputation system Other systems BE ߛ Shareholder has option of paying with- — — — — holding tax as final tax on dividend income. DK ߛ Withholding tax final and replaces per- — — — — sonal income tax. DE — — — — ߛ New system from 2002: earlier imputa- tion system replaced with half-rate sys- tem under which half the dividends re- ceived from German corporations taxed under personal income tax (applies to foreign shares as well). HE — — — — ߛ Exempt: dividends exempt from per- sonal income tax. ES — — ߛ Full imputation system, but coupled — with withholding tax, which is cred- itable against personal income tax. FR — — ߛ Full imputation system. — IE ߛ Since April 1999 a classical system. — — — — Prior to that, partial imputation (divi- dend treated as normal income, but 21% imputation, creditable against ordinary income tax). IT — — — — ߛ Choice between imputation credit and a reduced flat tax rate on dividends. LU ߛ Withholding tax credited against per- — — — — sonal income tax liability. 215 (Table continues on next page.) Table 6.8. Ways of Dealing with Double Taxation (continued) Country Classical system Imputation system Other systems 216 NL ߛ Withholding tax credited against — — — — personal income tax liability. AT ߛ Shareholder has option of paying — — — — withholding tax as final tax on divi- dend income. PT — — — — ߛ Shareholder is entitled to a tax credit, coupled with a withholding tax on dividends. Both set off against personal income tax liability. FI — — ߛ Full imputation system, but coupled — — with withholding tax, which is cred- itable against personal income tax. SE ߛ Withholding tax final and replaces — — — — personal income tax. UK — — ߛ Partial imputation system with — — non-withholding tax on dividends. US ߛ Under recently tabled proposals (Janu- — — — — ary 2003), the United States would move to an exemption system, as equity dividends would be entirely exempt from tax. CH ߛ Withholding tax credited against — — — — personal income tax liability. JP ߛ Withholding tax credited against — — — — personal income tax liability. ߛ Classical system. —No imputation or other system. Note: See tables 6.4 and 6.6 for country names. Sources: Lee 2002; Van den Noord and Heady 2001. FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 217 • Imputation system: This system also typically imposes a with- holding tax, but imputes full or partial adjustment to the share- holder’s taxable income based on the tax rate already applied at com- pany level. Six of the EU’s member states operate such a system. France is currently debating the scope of the dividend imputation, more particularly whether to confine “dividends” only to regular div- idends (as agreed by the general shareholder meeting) or also to exceptional distributions of revenues. It has been claimed, however, that imputation systems cannot easily take into account cross-border shareholdings. For this reason, Germany abolished its full imputation system in 2001. The corporate tax rate was decreased and cannot be imputed anymore at the personal level. The resulting double taxation of dividends is mitigated by the provision that the personal income tax rate applicable to dividends will be halved. Preferential treatment of retained earnings was an explicit goal of this reform. In Italy there are also proposals to move from the imputation system to the classi- cal system. One argument is to avoid granting the imputation credit in financial operations between taxable and tax-exempt companies. • Exemptions: Finally, a few countries (the United States if re- cent proposals are adopted, and Greece) operate a system under which no credit is given for corporate tax but dividends themselves are entirely exempt from taxation. Several suggestions have been made for reform.12 On the one hand, the advantageous treatment of debt through interest de- ductibility could be removed, so that neither interest nor dividends would be eligible for relief at the corporate level. One proposal in this direction is the comprehensive business income tax (CBIT), discussed in the United States. On the other hand, both interest and dividends could be granted the same relief at the corporate level and then taxed at the personal level. This is usually known as dual (or full) imputa- tion. However, even a full imputation system can favor retained earn- ings against outside finance, if the corporate tax rate is below the (personal) income tax rate of dividends and interest income. The proposal known as ACE, described above, is intended to re- move this effect. This would grant a tax relief at the level of the nom- inal rate of interest for the company’s total equity capital. Only prof- its in excess of the normal return on investment (“pure profits”) would be taxed. The taxation at the level of the company is final; div- idends or interest are not taxed again at the personal level. The ACE would be a move toward an expenditure-based system of taxation. Capital Gains Taxes Capital gains are the taxable gains (and losses) that are realized on the disposition of a (financial) asset. The tax rate often depends on 218 MATTIAS LEVIN AND PEER RITTER the time the asset was held. This reflects a distinction between asset trading (which is seen as an ordinary income-generating activity) and investing. Since the intention of the asset holder is not always obvious to tax authorities (unless the asset holder is specifically li- censed as an asset trader), legal distinctions often rely on ad hoc de- marcations, such as the time during which an asset was held or a set of criteria concerning the asset holder. Such date rules are arbitrary, however, as illustrated by a recent tax reform in Germany where the period was increased from six to twelve months (and abolished al- together for final taxation at the corporate level). Capital gains are often more lightly taxed than interest and divi- dend income. For example, in Belgium while interest and dividend income are taxed at 15 percent, capital gains from shares are un- taxed. Similar favorable treatments of capital gains apply in many other industrial countries, including Italy and Switzerland. Between 1965 and 1982 the relative importance of capital gains taxes de- clined in most OECD countries except France and Japan. Capital gains taxes generate little revenue: less than 1 percent of total tax revenues in the great majority of countries. If capital gains are sub- ject to a lower tax than dividends, there is an incentive for a com- pany to retain earnings and defer dividend payments or to engage in so-called dividend stripping: that is, for an investor to sell shares be- fore profits are paid out and subsequently buy them again. Countries apply different methods in determining the taxable capital gain. The most common principle of timing is taxation on realization. This is the point in time when a party to a transaction has the unconditional legal right or obligation to an amount. Hence under realization valuation, changes in value become tax relevant when amounts are due under a contract (payable or receivable). Even if capital and income are taxed at the same rate, taxation on realization provides an incentive to an early realization of losses and a deferral of gains. For the two other methods unrealized gains and losses also be- come tax-relevant. One model of capital gains taxation is the so- called “accretion taxation.” This model is also referred to as mark- to-market. Financial instruments are taxed according to their market value. All changes in value are taken into account. If they were taxed at the same rate as other forms of income (such as labor), this would correspond to the Haig-Simons ideal of comprehensive income taxa- tion, according to which all net changes in wealth should be taxed.13 A variant is to tax only those returns that are expected ex ante. This is “accrual taxation,” where the change in value at the end of the fis- cal year is calculated using either the risk-free interest rate or meth- ods of accounting to calculate the expected growth in value of the in- FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 219 strument (such as yield-to-maturity) at the time of purchase. Actual changes in value that occur after the resolution of uncertainty fol- lowing the issue (that is, the difference to their expected value) thus remain untaxed.14 This can lead to a tax benefit if such losses can be deducted against other forms of income. A further variant is to tax only the change in value of the asset while exempting the market rate of interest. This last form would be a step to expenditure taxation, where only pure profits are taxed. The ACE described above is an ex- ample. Since the market rate of interest is exempt from taxation, any unrealized expected gains do not lead to tax distortions. Chapter 2 describes some ways to accomplish formulary accrual taxation. The taxation of unrealized gains has been criticized on two grounds. First, the tax valuation of assets may be quite costly for firms and also for tax authorities when monitoring compliance. In order to reduce compliance costs, it has been suggested (Alworth 1998) that financial accounting rules, using market valuations of as- sets, should be applied for tax purposes, at least for large firms. Such marking-to-market may also imply low monitoring costs on the side of the tax authority. If unrealized capital gains are very high, the asset holder may be forced to sell the asset in order to pay the tax. In response to the first criticism, it has been suggested that accretion taxation be applied only to liquid instruments. Further- more, financial accounting may require the company to discount its debt obligations using its credit rating. If a borrower got into fi- nancial difficulties and revised its outlook on future profits down- ward, the resulting increase in the discount rate of the liabilities could lead to a taxable gain.15 Recent developments in the United States indicate that some financial accounting systems may also leave too much judgmental leeway to the individual accountant. Second, the marking-to-market taxation is criticized as leading to liquidity problems. This can probably be solved, however. Even if an asset is taxed only at the time of its disposal, there are ways to include the implicit tax credit of unrealized gains into the final tax payment ex post.16 Taxation of Personal Financial Wealth Net wealth taxes are raised once a year on the stock of certain as- sets subtracting associated debts. One of the aims of wealth taxes is vertical equity: taxing the wealthy and redistributing income. How- ever, wealth taxes are at odds with the Haig-Simons concept. A number of OECD countries apply them (Finland, France, Iceland, Luxembourg, the Netherlands, Norway, Spain, Sweden, and Swit- zerland) (Van den Noord and Heady 2001). 220 MATTIAS LEVIN AND PEER RITTER Figure 6.2. Average Tax on Financial Wealth in the EU, 1983–2000 0.6 0.55 0.5 0.45 0.4 0.35 1983 1985 1987 1989 1991 1993 1995 1997 1999 Source: Huizinga and Nicodème 2001. Most of the countries imposing wealth taxes have levied them for a long time (the Netherlands started in 1892, for example) although some countries have imposed them more recently (for example, Spain in 1978; France in 1982, subsequently withdrawn and later reimposed in 1989). Over time, taxing wealth has become rather complex and today involves significant administrative work to dis- cover and valuate wealth. In addition, the tax incidence is under- mined by tax planning (for example, by inflating liabilities or in- vesting in under-assessed assets). In some cases, the administrative burden has become disproportionate to the yield. Germany, for ex- ample, therefore decided to abandon them. In the EU, many coun- tries have eliminated wealth taxes following the liberalization of capital flows in the 1980s (Austria in 1994; Denmark and Germany in 1997) (Huizinga and Nicodème 2001) (figure 6.2). Moreover, in 2001 the Netherlands abolished their capital gains taxation and replaced the existing net wealth tax. Net wealth at the personal level is now taxed presumptively. Changes in wealth— which incorporates savings deposits, bonds, and stocks and some real estate but excludes pension wealth—are presumed to rise in value by 4 percent a year, irrespective of their actual return. A 30 percent tax rate is applied on this presumptive amount, leading to a net wealth tax of 1.2 percent (Bovenberg and Cnossen 2001). New Financial Instruments The taxation of new financial instruments has three basic features around which the legislation has evolved.17 First, most countries FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 221 retain a distinction between interest and dividends. Second, they distinguish between income and capital amounts. Third, the taxa- tion depends on the method of accounting for the payments from the instrument. Hybrid instruments combine features of both equity and debt. Taxation of hybrids often depends on the distinction between debt and equity through debt relief at the corporate level but also on dif- ferent treatment at the personal level (except for a full imputation system). The traditional approach to these instruments is to allocate each instrument into either the debt or equity category according to sets of criteria. These vary between countries. For example, in Ger- many, in order to classify as debt, instruments may carry cash flow rights, but must avoid rights in liquidation. Preferred shares, for ex- ample, fall under equity. In France, the tax authorities have not taken a definite position as to the classification of many of such instru- ments (including preferred shares) in order not “to be bound by a position it would have taken” (David 1996, p. 56). In the United Kingdom, the list of derivatives taxed equivalently to debt instru- ments was broadly extended in 2002, excluding equity-linked in- struments held for nontrading purposes. In the United States, the bi- furcation technique is applied to many of these instruments, whereby hybrid instruments are decomposed into debt and equity compo- nents, which are taxed accordingly. The drawback is that there are multiple ways to decompose hybrids, also necessitating a set of clas- sification rules. The distinction between changes in capital values and income18 carries over to most financial instruments. In the United States, for example, the tax treatment of options depends on whether the un- derlying property is a capital asset. Hedging in the United States does not depend on any underlying asset, allowing transferring gains and losses between categories. The United Kingdom has long distin- guished between income and capital amounts, but since 1994 debt and some derivatives (the list was extended in 2002) are attributed to income. In Germany an option is treated as two separate trans- actions. The premium is taxed as income to the option writer on the sale of the option; the purchase as expenditure. The second transac- tion depends on the contingency the option specifies. If the option is exercised, the gain or loss on both sides is taken as a capital gain. If the option is not exercised, then the capital account transaction does not become tax relevant. If the option was not worth exercising but the holder sold it at a price just above zero, the holder’s loss would become tax-relevant. Still, for a private individual, as long as the capital gain from exercising an option falls outside the period in which capital gains are taxable, there is no tax. Rules are different 222 MATTIAS LEVIN AND PEER RITTER for businesses. Although Germany taxes capital gains (within the time limit) at the same rates as income, there are limits as to set off gains and losses between capital and other income. The above-mentioned timing rules for the tax recognition of gains and losses are significant for the use of new financial instru- ments. Taxation upon realization implies that increases in value do not become relevant as long as no amounts are due. This is equiva- lent to granting the taxpayer an interest-free loan over the unreal- ized increase in value (or denying him a loss relief). For some well- known cases (such as zero coupon debt obligations), anti-deferral measures are used, which amount to some form of accrual taxation. As explained above, taxation on accrual still leads to ex post incen- tives (after the resolution of uncertainty) to selectively realize losses and defer gains, in particular if gains and losses are not treated sym- metrically. So-called “straddle” transactions can be used to exploit this. The solutions available are marking-to-market or the accrual taxation with limitation on deductibility of losses. 19 Most countries apply different timing rules to different instru- ments, under different criteria. The rules may vary according to the category of taxpayer; for example, financial intermediaries are more often subject to mark-to-market valuation. France applies marking- to-market to financial instruments that are traded on an organized exchange. Some countries, primarily in continental Europe, treat gains and losses asymmetrically as a consequence of their financial accounting principles, with recognition of unrealized losses, while gains are only recognized upon realization. In the United Kingdom, almost all instruments except for equity are now taxed as they are reflected in the accounts: that is, either on an accruals basis or mark- ing-to-market. Derivatives are mostly taxed marking-to-market. As explained above, accrual taxation does not tax changes in value be- yond those expected at the time of purchase. Some derivatives are for risk management only; they are pure bets with an expected value of zero ex ante. Examples include fu- tures and swaps. In Germany, such instruments are valued marking- to-market, but only if delivery or close-out are within the capital gains taxation limit of one year. Hence unrealized gains or losses after one year remain untaxed.20 In New Zealand risk-management instruments are taxed on realization. The United States determines timing mostly on the basis of capital-income distinction, using the bifurcation approach already mentioned (fixed element taxed as accrual, residual unpredictable element taxed on realization) to approximate the accrual approach (Warren 2001). The major practical difficulty here is choosing among the multiple ways of decomposing the financial instrument: continuing financial innovation precludes a conclusive categoriza- FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 223 tion. The United States also permits marking-to-market taxation for some instruments, such as futures and currency contracts and secu- rities held by traders. To summarize, at the root of most timing issues are the debt- equity and the capital-income distinctions. New financial instru- ments can be used to replicate a given pattern of payments tran- scending the borders between such tax rate distinctions. Most tax systems respond to this by lists of classifications of instruments, while trying to maintain the class (and hence tax rate) distinctions. Other countries apply uniform tax rates across all returns that are not labor income, along dual income tax lines. New Zealand (see below) tries to approximate accrual taxation for all types of what it calls “financial arrangements.” In other words, if a tax system were designed to respect neutral- ity toward the means of financing an investment as well as neutral- ity toward the intertemporal allocation of expenditure, the com- plexity surrounding the taxation of (new) financial instruments would be radically reduced. Hedging The use of financial instruments to eliminate the risk of an underly- ing asset is called hedging. Although most countries apply “separate transactions valuation” if financial instruments are used for hedging purposes, such instruments often receive a special tax treatment. In many countries they are “integrated” to be taxed as a unit, so that offsetting positions are taken into account; usually the tax treatment is matched to that of the underlying asset. The rationale is that under a regime taxing financial positions when they are realized, the “separate transactions principle” could lead to asymmetries in the timing of the taxation of gains and losses across instruments. Tax authorities face the problem of determining what constitutes a hedge. In the United States the taxpayer must identify a hedge in his financial and tax books on the day it is concluded. With the sep- arate taxation of capital and income amounts, hedge accounting can serve to transfer the losses of the hedge to the income account if not tied to the underlying asset (Edgar 2000). In France, the dec- laration is at the end of the year. Hedges are subject to three condi- tions: the transaction must be for the forthcoming year, the under- lying position must be “on a different market,” and the variations of all positions should be correlated, leaving open the question how large the covariance should be (David 1996). Under an ideal accrual regime, the integration of instruments to a hedge for tax purposes might be unnecessary, since the whole port- folio would be taxed and offsetting positions essentially matched 224 MATTIAS LEVIN AND PEER RITTER (Edgar 2000). However, even in that case, if the underlying position were nontraded debt or shares or non-financial assets, hedge ac- counting may be justified. France uses a mark-to-market approach to risk-managing instruments if they are traded on an organized market. However, when they are tied to a transaction that is not traded on an organized market, then the instruments can constitute a hedge and their taxation is tied to the realization of the underly- ing transaction. New Zealand, with its accrual legislation, in general does not allow for the integration of hedging positions, nor does the United Kingdom. Recent Policy Developments The tax policy response to new financial instruments has often been to sort new products into existing categories of debt or equity; in- come or capital amounts; and realization or marking-to-market timing. Since financial instruments can be used to transcend the boundaries between these categories for a given payment pattern, such tax legislation is often accompanied by classification plus anti- avoidance rules. More substantial reforms attempt to tackle these categorizations. These reforms also depend on the assumed per- spective about the ideal concept of income. The respective view de- termines the rate of capital gains taxation, the tax accounting rule of capital gains, and the interest (and equity) deductibility. In 1987, New Zealand implemented a system that uses accrual rules to come close to the Haig-Simons ideal of taxation. A wide def- inition of “financial arrangements” covers instruments that are debt or have debt components and involve contingent payments. This cat- egory seems to encompass instruments where an advance of funds gives the right to receive a return, be it fixed or contingent. Shares are excluded. For such financial arrangements accrual taxation is at- tempted, as well as for those instruments for which marking-to-mar- ket is not possible. The accrual takes a company’s financial account as a basis, though the law tries to limit deferral allowed by account- ing practices. Where possible, tax per financial instrument is calcu- lated using yield-to-maturity accounting. For new financial instru- ments that involve only risk management—and thus for which internal rate of return methods are not possible—a “base price ad- justment” is made, which amounts to taxation at the time the asset is realized. Accretion taxation is limited to traded futures and forward contracts for foreign exchange. However, the historical distinction between interest and dividends was retained. Australia is discussing a similar system, with a much broader approach to accretion taxation. Since 2002 the United Kingdom has broadly acknowledged marking-to-market taxation while the returns are largely attributed FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 225 away from capital income in some cases to revenue income in al- most all, thus removing a possible distortion. Such accretion taxa- tion is approximated by the U.S. practice of bifurcation, where fi- nancial instruments are disaggregated into a fixed-income and a contingent component, which are taxed on accrual and on realiza- tion, respectively, albeit at different rates. The dual income tax (DIT) countries of the Nordic area adopted a schedular approach. This eliminated the distortions between in- terest and dividends, as well as the distinction between income and capital amounts. The timing issue from realization-based recogni- tion of capital gains is partially solved through a mix between ac- crual and realization systems (Edgar 2000). The tax system of Italy was overhauled in 1998. Business income is now calculated in a value-added fashion, the tax base being the an- nual change in the business accounts (sales revenue minus cost of in- termediate goods), which are taxed at a flat rate. Neither labor costs nor interest payments are deductible. This removes the tax distor- tions relating to the debt-equity distinction (including tax avoidance strategies). In addition, an allowance for new total equity (issues and retained earnings starting from 1996) was introduced. Since ACE taxation narrows the corporate tax base and would in principle lead to higher statutory tax rates (Bond 2000), normal profits were not fully exempt from the Italian tax. Thus the Italian system has been characterized as a mixture between DIT and ACE (Bordignon, Gi- annini, and Panteghini 2001). The valuation of capital gains at a per- sonal level was designed to counter the incentives to deferral (Al- worth, Arachi, and Hamaui 2002). The applied valuation method depends on the way financial instruments are held. If they are held in a “managed portfolio” through an authorized intermediary, taxa- tion is on an accrual basis, and taxes are paid through the interme- diary. Private individuals can alternatively list their holdings in their tax declaration (“tax return method”) or have taxes deducted on a transactional basis through an intermediary (“administered portfolio method”). For the latter two methods, taxation was on realization, where an “equalizer” was applied to make it correspond to accrual taxation. The “equalizer” was calculated by retrospective taxation methods.21 This avoided the aforementioned liquidity problem with accrual taxation of unrealized gains. Traded securities with observ- able prices were taxed on realization while capitalizing accrued gains with the risk-free interest rate. Securities not traded on exchanges were capitalized ex post with the risk-free interest rate using the av- erage price increase over the holding period. Foreign mutual funds were capitalized with their internal rate of return. There were limits as to offsetting gains and losses between several categories of capital income. The “equalizer” was introduced in 2001 and repealed be- 226 MATTIAS LEVIN AND PEER RITTER cause of interest group pressure and a change in government in the same year, so it never came into effect. While the Italian reforms taxed capital gains at a lower rate than other income (including div- idends and interest), at the personal level the approach resembles Haig-Simons more at the personal level than at the corporate level. Croatia adopted the allowance for corporate equity system in 1994 (Rose and Wiswesser 1998; Wenger 1999). Business profits were calculated as the difference in total equity over the financial year to which a standardized rate of interest (to approximate a nominal market return) was credited. Hence only pure profits were taxed. Unrealized revaluations of assets would increase the com- pany’s total equity. The imputation of the nominal risk-free interest rate to corporate equity implies that the company is fully compen- sated for an early realization of gains. A creditor is taxed on the in- terest received, whereas the debtor can deduct the interest paid. Since the market rate of interest is deducted from the company ac- counts, the investor would be indifferent between the choices of financing and any interest in excess of the market rate would be taxed only once. At the personal level, capital gains tax was zero and neither interest nor dividends were taxed as income. Thus busi- ness profit taxation was final at the company level. Distortions be- tween debt, equity, and retained earnings as well as incentives to se- lective realization of gains and losses were avoided. Following a change in government, Croatia abandoned the allowance for the market rate of interest in 2001. However, similar tax reform is now under preparation in Bosnia. Taxation of Intermediaries Banks and other financial institutions generate significant value- added, both in terms of profits and high revenue employees. Finan- cial intermediaries also have a pivotal role in the economy, as they contribute to the smooth functioning of credit flows and thus eco- nomic growth. Just as the taxation of financial instruments may dis- tort saving and investment decisions, getting taxes wrong on the intermediary level may affect the smooth functioning of financial intermediation. Problems of Defining the Tax Base In principle, the taxable income of financial intermediaries is deter- mined in the same way as the income of other companies (Fitzgib- bon and Walton 1990). However, a number of issues (such as inter- FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 227 est income, trading income, and risk management income) make it more difficult to determine what constitutes taxable income for fi- nancial companies than for other industries. These difficulties arise for many reasons: • Difficulties of valuation: While the sophistication of financial markets means that market prices exist for many services and prod- ucts of financial institutions, valuation for tax purposes is neverthe- less not straightforward. For example, what is the outstanding value of assets and liabilities for tax purposes of loans? How should the value of new financial instruments be assessed? To what extent should financial accounting practices be used for tax purposes? • VAT: Unlike other goods and services, it is difficult to attach an explicit price to many financial services. As a result, financial ser- vices are exempt from VAT in most countries.22 This is not without problems, however. It gives rise to conflicts between tax authorities and financial intermediaries over what line to draw between a fi- nancial intermediary’s financial activities—which are VAT-exempt— and non-financial activities—for which VAT is due. Moreover, as fi- nancial intermediaries do not receive any credits on the VAT-exempt services they sell, they have to pay more VAT on their inputs. It also gives rise to complex apportionment systems whereby intermedi- aries are supposed to apportion input credits between taxable (input-creditable) or exempt (non-creditable) activities (Wurts and Fenton 2002). Even within the EU the apportionment methods vary widely between member states. Overall, there is no easy way to deal with financial services and VAT. Exemption is a second-best solu- tion, but currently there are no workable alternatives. Recent re- search suggests that advances in information technology and the changed relationship between banks and their clients may overcome some of these obstacles (Huizinga 2002). • Cross-border groups: The problems of determining taxable income are magnified in the case of groups with permanent estab- lishments abroad. The most acute problem in this respect is the al- location of income between the permanent establishments and the home office. This problem is more pronounced for permanent es- tablishments in the field of finance than for other industries. First, financial permanent establishments are increasingly offering ser- vices from the full value chain, such as front-office and back-office trading services. Second, it is more difficult to divide the tax base in financial services, as a financial transaction involves no physical movement of goods, rather just accounting entries (IFA 1996). • Tax evasion: Given the problems of correctly valuing the assets and liabilities of financial intermediaries and given banks’ extensive 228 MATTIAS LEVIN AND PEER RITTER international networks of branches and subsidiaries, it is not sur- prising that banks are well-placed to make good use of the current practices of dealing with the taxation of international groups (such as manipulating transfer pricing) (Demirgüç-Kunt and Huizinga 2001). None of these issues is unique for financial intermediaries, but the associated problems are more pronounced. Taxation of Funds While investment in funds is often induced by tax systems at the in- vestor and fund level, most tax administrations regard the company managing a fund as a taxable entity and thus subject to corporate income tax. In order to capture the overall tax burden, it is necessary to look at tax at the management company level, fund level, and investor level. Generally, the tax treatment of fund management companies in industrial countries takes one of the following forms: • Not subject to tax: Funds are not considered a “taxable per- son,” and thus are not taxed (Belgium), or are exempt from taxes (Finland). • Subject to tax, but exempt: In some countries, a fund normally subject to tax may be exempt if it fulfills certain conditions (for ex- ample, on basis of activity, as in Luxembourg). • Eligible for a special tax base: In some countries, funds are subject to tax but the taxable income (base) is much lower than for other companies (investment companies in Belgium) or reduced (paid distributions to investors subtracted from taxable income in Sweden and the United Kingdom). • Subject to a special tax rate: In some countries, funds are sub- ject to tax but at special (low) rates (such as in the Netherlands, where the rate can sometimes reach 0 percent). • Fully subject to tax but compensation is made available at the investor level: In some countries, the fund is fully subject to tax, but investors are normally compensated for the tax paid at fund level via reduction or exemption of tax at investor level (the imputation system in the United Kingdom, for example) (IFA 1997). Moreover, the overall tax level differs between different kinds of funds. Generally, pension and insurance funds receive more favor- able tax treatment than investment funds, with mutual funds falling in between. The reason for this more favorable treatment is the gains to society from citizens catering to their own pension needs and risk exposure. One example of taxation in the field of funds is provided below: namely insurance companies (table 6.9). As can be Table 6.9. Taxation of General Insurance Companies Country Income tax (%) VAT Premium taxes (%) Other taxes BE 39 Exempt 3–9.25 • 0.17% on value of goods held by nonprofit pension funds (Compensatory funds: • 9.25% tax on profits on policies 0.25–10) • 4.5% on 190% of payroll DK 30 Exempt 1–50 • 0.6–4% on deeds DE 25 Exempt 2–15 ES 35 Exempt 6 • 1% tax on capital formation FR 33.33 Exempt 7–30 • 1% financial institution overhead tax (Compensatory funds • 4.25–13.6% social contribution tax 1.9–8.5) • 0.1% turnover tax on value of shares, stocks, real estate, etc. • Stamp duties • Profit sharing plan • Professional tax IE 10 Exempt 2 • 1% on share capital IT 36 Exempt 2.5–21.25 • Special tax on securities (Compensatory funds LU 30 Exempt 1–6.5) • Municipal business tax (varies, 10% Luxembourg) 4–6 • 0.5% wealth tax • 0.2% net wealth tax on capital • 1% capital investment tax NL 35 Exempt 7 • 1% of share capital AT 34 Exempt 1–10 • 1% of capital contributions • 0.04–0.15% stock turnover tax (bonds, shares) PT 32 Yes, but recovered 0.45–12 • License fees on pro rata basis. FI 29 Exempt 22 — SE 28 Exempt 15 — UK 30 Exempt 5 — US 35 Exempt 3–4 • States’ capital and franchise taxes CH 35 Exempt 2.5–5 • 0.5% on shareholder equity 229 Note: AT Austria; BE Belgium; CH Switzerland; DE Germany; DK Denmark; ES Estonia; FI Finland; FR France; IE Ireland; IT Italy; LU Luxem- bourg; NL Netherlands; PT Portugal; SE Sweden; UK United Kingdom; US United States. Sources: OECD 1999; Lee 2002. 230 MATTIAS LEVIN AND PEER RITTER seen, insurance companies are subject to corporate income tax in all industrial countries but pay no VAT (except Portugal). Moreover, insurance companies must pay taxes on the premiums they distrib- ute. There are also wide divergences between industrial countries re- garding other taxes an insurance company may be subject to (such as payroll taxes, transaction taxes, and wealth taxes). An additional problem is that defining the precise taxable base becomes difficult when the fund management service is bundled with other financial services. This is the case when, for example, an insurance company holds a fund’s assets and claims on its own books instead of just managing them. Reduced tax rates for funds open the opportunity for tax avoidance through accelerated losses for non– tax-exempt entities, as a tax-exempt fund can hold an offsetting po- sition with accelerated gains on a financial instrument without (or with fewer) tax consequences. The tax picture of funds would not be complete without looking at the taxation of the fund holder. Individuals are often encouraged to save via tax incentives. Such incentives are particularly common in the field of pension savings. Favorable tax treatment of pension savings is one of the most important expenditures in many OECD countries. The tax treatment has taken many forms. The most com- mon is to grant tax allowances for private pension contributions and exempt returns on fund assets while benefits remain taxed (the so-called EET approach). But even within the EU, three member states operate an ETT-system (Exempt contributions, Taxed invest- ment income and capital gains of the pension institution, and Taxed benefits) and two member states operate a TEE-system (Taxed con- tributions, Exempt income, Exempt benefits). Taxes on Raising and Transferring Capital There are other forms of taxation on equity capital that burden the use of equity relative to debt. The trend here is one of reduction of such taxes, and in certain cases, elimination. Nevertheless, six EU member states impose a stamp duty of 1 percent on capital issues, and eight levy a stamp duty on transactions in shares or bonds (table 6.10). The transactions duties were abolished in Germany in 1991, but were reintroduced in the United Kingdom in 1997 at 0.5 per- cent on stock exchange transactions.23 Implicit Taxes: Reserve Requirements The original motivation for requiring credit institutions to hold re- serves was to ensure that they had sufficient liquidity to meet de- FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 231 Table 6.10. Stamp Duties, 2001 Stamp duty on issues Stamp duty on transactions Country Rate (%) Comments Rate (%) Comments BE 0.5 — 0.2 — DK 0 — 0.5 Shares only DE 0 — 0 — HE 1 — 0.5 Bonds — 0.3 Shares ES 1 — 0 — FR 0 — 0.15–0.3 Shares only: 0.3% up to a turnover of 153,000. More than 153,000: 0.15%. IE 1 Joint stock 1 Bond and public loans companies not taxed IT 1 0% on bonds 0.14 Shares traded outside recognized stock exchange 0.009 Government bonds LU 1 0.5% on family 0 — companies NL 0.55 0 — AT 1 0% on bonds 0 — PT 0 — 0.002–0.015 — FI 0 — 0 — SE 0 — 0 — UK 0 — 0.5 — CH 1 0.06–0.12% 0.15 Domestic securities on bonds 0.3 Foreign securities Note: See table 6.9 for additional country names. HE Greece; AT Austria. Source: Swiss Federal Tax Administration. mands for withdrawal of deposits. A more recent motivation has been the contribution of reserves in smoothing short-term money market interest rates and the long-term money creation process by its effect on bank’s lending, credit creation, and other activities. A reserve requirement means that banks must hold a specific level of reserves at the central bank. Most countries that impose re- serves allow averaging: that is, banks have to hold the specified level of reserves on average over a specific time (table 6.11). There are, however, drawbacks with reserve requirements: • First, the requirement to hold reserves diminishes the amount of capital available for relending, thus reducing banks’ ability to create deposits and thus the overall money creation process. • In addition, if the reserves are remunerated at below-market rates, the effect of compulsory reserves can be compared to the ef- fects of a tax. 232 MATTIAS LEVIN AND PEER RITTER • As reserve requirements impose an additional cost on banks, they may cause arbitrage incentives if reserve requirements are not equally imposed in all countries. That has indeed been the case (in the EU, for example) and has been one of the reasons why reserve rates have been decreased or have become remunerated in recent years, as further described below. • In many countries, requirements are imposed only on certain credit institutions. This creates distortions. Those distortions have been another reason why requirements recently have been brought down. In light of these drawbacks, most industrial countries have re- duced the reserve requirement ratio since the 1990s. France lowered requirements between 1990 and 1992, the United States lowered theirs in 1992, and Germany followed suit in 1994 and 1995 (Davies 1998). Nevertheless, most industrial countries, with a few notable exceptions (such as the United Kingdom and Canada) continue to impose reserve requirements. Considering the negative effects of reserve requirements and a few notable examples of countries that operate a monetary policy without such requirements, it may perhaps be surprising that the countries participating in the European Monetary System decided Table 6.11. Reserve Requirements in Selected Industrial Countries US EURO UK SE DK JP CA Reserve 3–10a 2b 0 0 0 0.05–1.3 0 requirement (% of liabilities) Averaging Yes Yes — — — Yes Yes Length of 2 1 — — — 1 4–5 period weeks month month weeks Penalties Discount — — — — Discount Bank rate rate rate +2% +3.75% n.a. Remuneration No Rate of — — — — — refinancing operations aInterest-bearingand non–interest-bearing checking accounts. bOvernight deposits; deposits with agreed maturity up to two years; deposits re- deemable at notice up to two years; debt securities issued with agreed maturity up to two years; and money market paper. Zero percent reserves required for deposits with agreed maturity or period of notice over two years, repos, debt securities issued with maturity over two years. Source: Davies 1998; European Central Bank. FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 233 to maintain the practice at the launch of the euro. However, one reason might have been that most of the euro-members had reserve requirements in place before joining, including France, Germany, and Italy. In a 1998 press release, the European Central Bank (ECB) stated that “without the use of a minimum reserve system, the ESCB [European System of Central Banks] would be faced with a rela- tively high volatility of money market interest rates, which would re- quire the frequent use of open market operations for fine-tuning purposes.” The required rate of reserves in the euro-system is 2 percent of li- abilities, calculated as an average over a month. The reserve base is composed of overnight deposits, deposits with agreed maturity up to two years, deposits redeemable at notice up to two years, debt se- curities issued with agreed maturity up to two years, and money market paper. In addition, the euro-system provides two standing fa- cilities, enabling credit institutions to deposit excess liquidity or ac- cess additional liquidity if they so wish. As a result, the ECB is able to intervene less frequently in the market (currently once a week). The ECB introduced two features aimed at reducing the compet- itive disadvantage of those credit institutions that are based in the euro-area and thus falling under the reserve obligation. First, credit institutions are allowed to deduct a lump sum of 100,000 from their reserve liability base. As a result, credit institutions with a small reserve base do not have to hold reserves. Second, the reserve holdings are remunerated. The applicable rate is the rate of ECB re- financing operations. Some countries, however, do not have reserve requirements, such as the United Kingdom. The Bank of England instead requires its settlement banks, on a daily basis, not to let their accounts go into overdraft. In order to prevent volatile money market interest rates, the Bank of England tries to forecast the demand for reserves, re- sorting to frequent open market operations (normally twice a day). However, such forecasting is difficult and as a result, the volatility of UK overnight interest rates is high by international standards (Davies 1998). In the United States, all deposit institutions (commercial banks, saving banks, savings and loan associations, and credit unions) must maintain reserves on certain types of deposits (transaction de- posits—that is, interest-bearing and non–interest-bearing checking accounts). The Monetary Control Act allows the Federal Reserve to impose rates of between 8 and 14 percent. The reserve requirements are imposed at a sliding upward scale. For the first $7.1 million of a bank’s transactions, a rate of 0 percent is imposed. For the next $41.3 million of a bank’s transaction accounts, a 3 percent reserve 234 MATTIAS LEVIN AND PEER RITTER rate applies. Beyond that, a rate of 10 percent applies. These rates are adjusted annually according to different formulas. No reserves are imposed on personal time deposits and euro-currency liabilities. The average reserve requirement is imposed over a two-week pe- riod. Despite persistent lobbying, no remuneration is paid on com- pulsory reserves. Such a change would have to be approved by Con- gress, which so far has been reluctant because of the revenue loss it would entail. Overall, there is thus a trend to decrease reserve requirements, as alternative means to achieve the end (operational efficiency of mon- etary policy, solvency of financial institutions) exist. In addition, in light of increasing awareness of the tax-like features of the reserves system, some countries (including those in the euro-system) have de- cided to remunerate compulsory reserves. Conclusion Since the 1980s industrial countries have engaged in numerous tax reforms. The aim has been to make tax systems more efficient, fair, and simple. Reform has generally operated within given structures by cutting rates and broadening tax bases. These general trends are mirrored in the taxation of financial intermediation. Countries have tried to improve efficiency by cutting rates and have broadened the bases by reducing certain exemptions, while creating new tax fa- vored savings policies. Globalization and the increasing elasticity of supply of capital in- crease the difficulties of taxing capital income. It is therefore sur- prising that countries continue to tax capital income to such a large extent. Political considerations and bureaucratic and legal inertia go some way toward explaining why industrial countries have not moved further toward reducing the taxation of capital income. But perhaps the reason is simpler: Tax competition is not as strong as is usually thought. Reforms have in general been reactive to the development of fi- nancial instruments. The distinction between interest and dividends has been maintained in many countries, while the rules to sort par- ticular financial instruments into these categories have been refined. The distinction between income and capital amounts has been given up in some countries. The tax accounting of gains and losses often follows financial accounting principles. Since financial accounting currently differs across countries, tax treatment of the timing of gains and losses does too. It remains to be seen whether the change in financial accounting practices (perhaps their harmonization) will affect the taxation of financial instruments. FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 235 As regards the taxation of financial instruments, the two areas of concern are efficiency and tax evasion. While in industrial countries financial instruments are used to minimize tax liability, taxpayers in developing countries may follow less sophisticated routes. General compliance is probably more of an issue in developing countries. This should not be neglected in discussing ways to tax capital gains. Accrual taxation may require some sophistication on the part of tax accountants; marking-to-market also requires sophisticated capital markets. With underdeveloped capital markets and the need to favor investment, developing countries may gear their tax policy to focus more on efficiency. This could imply avoiding following the historical peculiarities of developed tax systems. Rather, tax legisla- tion on financial intermediation might be embedded in a broader perspective on taxation. Developing countries, with less complex tax systems in place, may therefore perhaps be bolder. There is no need for developing countries to fall into the same traps of tax system design. Indeed, the example of Croatia illustrates that it is practically possible to in- troduce and operate a consumption-based system. However, as il- lustrated by the fate of the Italian tax reform, there are limits to po- litical acceptance of bold reform initiatives. Notes 1. Generally, nonresidents are treated differently from residents. Often what constitutes taxable income differs. Countries have generally tried to reduce the complexity of taxing nonresidents by imposing withholding taxes at source. In most cases, these taxes are set at a rate aimed at equalling the tax conditions that residents experience. In some cases, however, non- residents receive more beneficial tax treatment than residents. Moreover, withholding rates mirror the underlying complexity of taxing residents. As a result, withholding rates often differ between types of income (dividends vs. interest) and types of instrument (for example, equity vs. bonds, public vs. private). 2. As well as those mentioned, within the European Union the need to avoid discrimination against residents of other EU member states has been an important driver of change in recent years. Important as this has been, it is rather specific to the case of an economic union and thus is not dis- cussed here. 3. This section draws on Van den Noord and Heady (2001); see also Devereux (2000). 4. Many academic papers stress the particularly distorting effects and efficiency losses of this understanding of “equity.” Not taxing the accumu- 236 MATTIAS LEVIN AND PEER RITTER lation of capital (at its opportunity cost) would not violate interpersonal eq- uity, because capital accumulation reflects the intertemporal allocation of consumption. As long as in this dynamic sense expenditure is taxed, inter- personal equity would be safeguarded. 5. This may be compared with a reduction in the average rate of cor- porate income tax in the EU from 44 to 34 percent. 6. This and other factual material in the chapter has been drawn from IBFD (2001). 7. Van den Noord and Heady (2001). For a discussion on the effi- ciency/equity aspects of dual income taxation, see Nielsen and Sørensen (1997), who defend DIT on efficiency grounds. Lower taxation of wealth, in fact, means a lower taxation of foregone consumption. 8. Bovenberg and Cnossen (2001). 9. Rose and Wiswesser (1998) and Keen and King (2002) give a favor- able assessment of the implementation that began in 1994. The system was abandoned in 2001 for political reasons. 10. Edgar (2000). The study summarizes how financial instruments in their analytical form can be separated into three building blocks. Credit-ex- tension instruments give the right to a principal in return for the lending of funds, the classical case being debt. Price-fixing instruments give rise to the purchase or sale of an asset at a specified price, such as futures, forwards, and swaps. Price-insurance instruments give the holder the right, but not the obligation, to buy or sell an asset. Among the latter are classical com- mon shares, which can be seen as a call option on the assets of the firm. 11. See Michielse (1996), Valkonen (2001), and Joumard (2001), for example. 12. For a survey, see Cnossen (1996). 13. Although in a dual income system the income from capital and from labor are taxed at a different rate, this in principle leaves open the question whether capital gains are taxed on accrual or on realization. 14. Known as “unanticipated deferral” in the tax literature (Edgar 2000, pp. 80–83 and 218–39). Since their expected value is zero (if losses and gains are treated symmetrically), their taxation would affect risk-tak- ing. If the financial instrument is capitalized with the risk-free interest rate, the diversifiable and undiversifiable risk will remain tax-free. 15. Muray (2001). 16. See chapter 2, this volume, and Alworth, Arachi, and Hamaui (2002). 17. The examples in the text are intended to illustrate the diversity of national legislation. For a comprehensive survey, see Plambeck, Rosen- bloom, and Ring (1995). For a summary of the issues, see Alworth (1998); Thuronyi (2001). FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 237 18. In dual income tax system countries, the distinction between capital and income is different. Income basically means labor income, while capi- tal means the return on assets other than human capital. 19. Such risk arbitrage is explained in Edgar (2000, pp. 89 and 219). 20. Until 1999 this was the case only if a taxable yield could be derived from the provision of an underlying capital asset for a fixed period, so that price-fixing instruments—like index-linked options and futures—were not taxable. 21. See chapter 2 for an analytical description. 22. In the EU, for example, the 6th VAT directive (77/388/EEC) ex- cludes VAT for financial services. 23. For further discussion of securities transactions taxes, see chapter 11. References Alworth, Julian. 1998. “Taxation and Integrated Financial Markets: The Challenges of Derivatives and other Financial Innovations.” Interna- tional Tax and Public Finance 5 (4): 507–34. Alworth, Julian, Giampaolo Arachi, and Roni Hamaui. 2002. “Adjusting Capital Income Taxation: Lessons from the Recent Italian Tax Experi- ence.” Bocconi University, Milan. Processed. Boadway, Robin, and Michael Keen. 2003. “Theoretical Perspectives on the Taxation of Capital Income and Financial Services” (chapter 2, this volume). Bond, Stephen. 2000. “Levelling Up or Levelling Down? Some Reflections on the ACE and CBIT Proposals, and the Future of the Corporate Tax Base.” In Sijbren Cnossen, ed., Taxing Capital Income in the European Union: Issues and Options for Reform. Oxford: Oxford University Press. Bordignon, Massimo, Silvia Giannini, and Paolo Panteghini. 2001. “Re- forming Business Taxation: Lessons from Italy?” International Tax and Public Finance 8 (2): 191–210. Bosworth, Barry, and Gary Burtless. 1992. “Effects of Tax Reform on Labour Supply, Investment and Saving.” In Journal of Economic Per- spectives 6 (1): 3–25. Bovenberg, Lans, and Sijbren Cnossen. 2001. “Fundamental Tax Reform in the Netherlands.” International Tax and Public Finance 8 (4): 467–80. Cnossen, Sijbren, 1996. “Company Taxes in the European Union: Criteria and Options for Reform.” Fiscal Studies 17 (4): 67–97. ———. 2000. Taxing Capital Income in the European Union: Issues and Options for Reform. Oxford: Oxford University Press. 238 MATTIAS LEVIN AND PEER RITTER David, Cyrille. 1996. “General Introduction.” In Geerten Michielse, ed., Tax Treatment of Financial Instruments. The Hague: Kluwer Law Inter- national. Davies, Haydn. 1998. “Averaging in a Framework of Zero Reserve Re- quirements: Implications for the Operation of Monetary Policy.” Bank of England Working Paper No. 84. London. Demirgüç-Kunt, Aslı, and Harry Huizinga, 2001. “The Taxation of Do- mestic and Foreign Banking.” Journal of Public Economics 79 (3): 429–53. Devereux, Michael P. 2000. “Issues in the Taxation of Income from Foreign Portfolio and Direct Investment” In Sijbren Cnossen, ed., Taxing Capi- tal Income in the European Union: Issues and Options for Reform. Ox- ford: Oxford University Press. Devereux, Michael P., Rachel Griffith, and Alexander Klemm. 2002. “Cor- porate Income Tax Reforms and International Tax Competition.” Eco- nomic Policy 35: 451–95. Duisenberg, Willem F. 1993. “Tax Reform in Perspective.” In Anthonie Knoester, ed., Taxation in the United States and Europe: Theory and Practice. New York: St. Martin’s Press. Edgar, Tim. 2000. “The Income Tax Treatment of Financial Instruments: Theory and Practice.” Canadian Tax Paper 105, Canadian Tax Founda- tion, Toronto. European Commission. 2001. “Company Taxation in the Internal Mar- ket.” European Commission Staff Working Paper 1681. Brussels. Feldstein, Martin. 1995. “Fiscal Policies, Capital Formation, and Capital- ism.” European Economic Review 39 (3–4): 399-420. Fitzgibbon, Christopher P. J., and Miles Walton. 1990. Taxation and Bank- ing. London: Sweet & Maxwell. Fullerton, Don, and Mervyn A. King. 1984. The Taxation of Income from Capital: A Comparative Study of the United States, the United King- dom, Sweden and West Germany. Chicago: The University Press of Chicago. Gordon, Roger H. 2000. “Taxation of Capital Income vs. Labour Income: An Overview.” In Sijbren Cnossen, ed., Taxing Capital Income in the European Union: Issues and Options for Reform. Oxford: Oxford Uni- versity Press. Haufler, Andreas. 2001. Taxation in a Global Economy. Cambridge: Cam- bridge University Press. Huizinga, Harry. 2002. “A European VAT on Financial Services.” Eco- nomic Policy 35: 499–534. Huizinga, Harry, and Gaëtan Nicodème. 2001. “Are International Deposits Tax Driven?” DG ECFIN Economic Paper, No. 152. Brussels: European Commission. FINANCIAL INTERMEDIATION IN INDUSTRIAL COUNTRIES 239 IBFD (International Bureau of Fiscal Documentation). 2001. The Taxation of Private Investment Income. Amsterdam. IFA (International Fiscal Association). 1996. Principles for the Determina- tion of the Income and Capital of Permanent Establishments and Their Applications to Banks, Insurance Companies and Other Financial Insti- tutions (Cahiers de Droit Fiscal International Vol. 81a). The Hague: Kluwer Law International. ———. 1997. The Taxation of Investment Funds (Cahiers de Droit Fiscal International Vol. 82b). The Hague: Kluwer Law International. Joumard, Isabelle. 2001. “Tax Systems in European Union Countries.” Economics Department Working Paper 301, OECD, Paris. Keen, Michael, and John King. 2002. “The Croatian Profit Tax: An ACE in Practice.” Fiscal Studies 23 (3): 401–18. Knoester, Anthonie, ed. 1993. Taxation in the United States and Europe: Theory and Practice. New York: St. Martin’s Press. Leape, Jonathan I. 1993. “Tax Policies: The United Kingdom.” In Anthonie Knoester, ed., Taxation in the United States and Europe: Theory and Practice. New York: St. Martin’s Press. Lee, Kyung Geun. 2002. “Taxation of Financial Intermediation.” Back- ground note for Workshop on Taxation of Domestic Financial Interme- diation, April 8. World Bank, Washington, D.C. Processed. Lodin, Sven-Olof. 2001. “International Taxes in a Rapidly Changing World.” International Bureau of Fiscal Documentation Bulletin 55 (1): 1–7. Maillard, Didier. 1993. “Tax Policies: France.” In Anthonie Knoester, ed., Taxation in the United States and Europe: Theory and Practice. New York: St. Martin’s Press. McLure, Charles E. Jr. 2001. “Globalization, Tax Rules and National Sov- ereignty.” International Bureau of Fiscal Documentation Bulletin 55 (8): 328–41. Messere, K. C. 1993. Tax Policy in OECD Countries: Choices and Con- flicts. Amsterdam: IBFD Publications BV. ———. 1999. “Half a Century of Changes in Taxation.” International Bu- reau of Fiscal Documentation Bulletin 53 (8–9): 340–65. ———. 2000. “20th Century Taxes and Their Future.” International Bu- reau of Fiscal Documentation Bulletin 54 (1): 2–29. Michielse, Geerten, ed. 1996. Tax Treatment of Financial Instruments. The Hague: Kluwer Law International. Modigliani, F. and M. H. Miller. 1958. “The Cost of Capital, Corporation Finance and the Theory of Investment.” American Economic Review 48 (3): 261–97. Muray, R. 2001. “Changes to the Taxation of Financial Instruments.” In- ternational Tax Review (April) (www.internationaltaxreview.com). 240 MATTIAS LEVIN AND PEER RITTER Nielsen, Søren Bo, and Peter Birch Sørensen. 1997. “On the Optimality of the Nordic System of Dual Income Taxation.” Journal of Public Eco- nomics 63 (3): 311–29. OECD (Organisation for Economic Co-operation and Development). 1998. Harmful Tax Competition—An Emerging Global Issue. Paris: OECD. ———. 1999. ”Taxing Insurance Companies.” OECD Tax Policy Studies No. 4. Paris. ———. 2001. Economic Survey—Italy 2001. Paris. Plambeck, Charles, David Rosenbloom, and Diane Ring. 1995. “Tax As- pects of Derivative Financial Instruments: General Report.” In Interna- tional Fiscal Association, ed., The Taxation of Financial Derivative In- struments (Cahiers de Droit Fiscal International Vol. 80b). The Hague: Kluwer Law International. Rose, Manfred, and Rolf Wiswesser. 1998. “Tax Reform in Transition Economies: Experiences from the Croatian Tax Reform Process of the 1990s.” In Peter Birch Sorensen, ed., Public Finance in a Changing World. Cambridge: Cambridge University Press. Steinmo, Sven. 1995. “The Politics of Tax Reform.” In Cedric Sandford, ed., More Key Issues in Tax Reform. Bath, England: Fiscal Publications. Thuronyi, Victor. 2001. “Taxation of New Financial Instruments.” Tax Notes International 24 (3): 261–73. Valkonen, Tarmo. 2001. “The Finnish Corporate Income Tax Reform and the Financial Strategy of Firms: A General Equilibrium Approach.” Em- pirica 28 (2): 219–39. Van den Noord, Paul, and Christopher Heady. 2001. “Surveillance of Tax Policies: A Synthesis of Findings in Economic Surveys.” OECD Eco- nomics Department Working Paper 303, Paris. Warren, Alvin. 2001. “U.S. Income Taxation of New Financial Instru- ments.” Harvard University, School of Law, Cambridge, Mass. Processed. Wenger, Ekkehard. 1999. “Taxes on Business Profits.” In Manfred Rose, ed., Tax Reform for Countries in Transition to Market Economies. Stuttgart: Lucius & Lucius. Wurts, Brian, and Frankie Fenton. 2002. “Indirect Taxes on Financial Services.” International Tax Review (April) (www.internationaltax review.com). 7 Seigniorage, Reserve Requirements, and Bank Spreads in Brazil Eliana Cardoso In 2001, nominal bank lending interest rates in Brazil reached an an- nual average of 44 percent for business loans and 73 percent for per- sonal loans. With the inflation rate (measured by the consumer price index) no more than 8 percent, such rates act as a serious constraint on borrowing, especially for longer terms. No wonder then that bank credit was just 28 percent of GDP: not because of a lack of sophisti- cated credit analysis or even of lending capacity, but essentially from the effect of high interest rates on demand. The cost of funds to banks is high. Money market and wholesale deposit rates averaged almost 18 percent in 2001, and the intermediation margins above these rates must cover the cost of taxes and other impositions— including the cost of high reserve requirements—and government- directed lending, as well as the costs of non-performing loans. This chapter examines the role of bank-captured seigniorage as well as explicit taxation in influencing spreads. Bank seigniorage revenue depends on the interaction between inflation, the market for demand deposits, and the rate of reserve requirements imposed by the central bank. Bank seigniorage revenue increased with infla- tion in Brazil until 1989, declined when inflation accelerated above 1,000 percent per year after 1992, and turned negative with the government stabilization Real Plan in 1995. 241 242 ELIANA CARDOSO Despite the lack of competition otherwise observed in Brazilian banking, it will be seen that any increase in seigniorage collected by commercial banks has tended to reduce the spread between interest rates on deposits and loans. As a corollary, it can be inferred that reductions in the cost of reserve requirements and directed credit programs can drive down bank spreads and net margins—if sup- ported by sound fiscal and monetary policy. Since inflation sta- bilized in the mid-1990s, the role of explicit taxation on financial intermediaries has become relatively more important for the expla- nation of the behavior of bank spreads. The first section briefly summarizes the topic of financial liberal- ization in Brazil and follows with a review of the macroeconomics underlying the country’s historically high interest rates. The chapter then discusses reserve requirements and bank seigniorage collection. It continues with evidence on the impact on bank spreads of seignior- age collection, inflation, explicit taxation, operational costs, and pro- visions for non-performing loans. The chapter concludes with some suggestions for strengthening Brazil’s banking system. Financial Liberalization in Brazil Like many other countries, Brazil has taken extensive steps toward financial liberalization during the past 20 years. Successive govern- ments reduced credit controls, rationalized reserve requirements, re- moved all interest rate ceilings on deposits in 1979, reduced barriers to entry after 1991, and liberalized controls on international capital flows during the 1990s. The path of reforms accelerated after 1995, when the government closed or privatized 10 state banks. Between mid-1994 and mid-2001, the share of public banks in total assets of the banking sector fell by 65 percent, while the share of foreign in- stitutions increased fourfold. Loans from the financial system to the public sector practically disappeared (figures 7.1 and 7.2). A variety of tax and quasi-fiscal instruments affect financial insti- tutions and financial intermediation in Brazil. Some are explicit taxes included in the tax code, such as the income tax, the tax on pre-tax corporate income (CSLL), the tax on financial operations (IOF), taxes on gross revenues (PIS and COFINS), and the tax on bank deb- its (CPMF). Table 7.1 describes the structure of taxes and contribu- tions in Brazil at the end of the 1990s, including the base of the tax, its destination, and the share of revenues in the relevant variable. Other taxes on financial intermediation are not defined explicitly and are not treated as taxes in budget accounting. These include nonremunerated reserve requirements on demand deposits and di- SEIGNIORAGE AND BANK SPREADS IN BRAZIL 243 Figure 7.1. Loans from the Private Financial Sector as a Percent of GDP, Brazil, 1989–2000 25 To private sector 20 To public sector 15 10 5 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Source: Table 7.2. Figure 7.2. Loans from the Public Banking Sector as a Percent of GDP, Brazil, 1989–2000 45 40 To private sector 35 To public sector 30 25 20 15 10 5 0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 Source: Table 7.2. rected credit at subsidized interest rates. Implicit taxes (discussed later) secure credit for the government itself or for preferred sectors at lower than market interest rates. Brazil has a shallow intermediation system, whether measured by the ratio of liquid liabilities or private credit to GDP—no better than the Latin American averages and well below that of upper-middle Table 7.1. Taxes and Contributions at the End of the 1990s, Brazil 244 Value in Share in Taxes and contributions Base 1997 (R$) relevant variable Destination Federal Taxes Federal, 53%; states, 21.5%; Income tax Personal and business income 35.6 4.6% of GDP municipalities, 22.5%; development funds, 3%. Import tariffs CIF imports 5.1 8.3% of imports Federal Industrial products (IPI) Industrial value-added 16.6 9.8% of manufacturing GDP Federal, 43%; states, 21.5%; municipalities, 22.5%; development funds, 3%; export fund, 10%. Financial operations (IOF) Loans, insurance, 3.8 6.7% of financial GDP Federal Foreign investment Rural land Land values 0.2 Federal, 50%; local, 50% Fees Various 0.3 Federal Federal Contributions Social security Private wage bill 44.1 20.1% of non-government Social security benefits wages COFINS Gross revenues 18.3 2.4% of GDP Social security benefits CPMF Check debits 6.9 12.1% of financial GDP Health CSLL Pre-tax corporate income 7.2 2% of operational surplus PIS/ Pasep Gross revenues 7.3 0.9% of GDP FAT: Deposits at federal financial institutions Federal employee Federal wage bill 2.6 6.1% of federal wages Payment of federal pensions contributions Other social contributions Various 0.6 5.9% of non-government Severance benefits, housing wages FGTS Wage bill 12.9 Economic contributions Various 0.9 0.8 of wages FNDE Salário-Educação Wage bill 2.8 State Taxes Value-added (ICMS) Value added 59.8 7.7% of GDP States, 60%; local, 25%; FUNDEF, 15% Vehicle (IPVA) Vehicle value 3.8 States, 50%; local, 50% Inheritance (ITCD) Inheritance value 0.3 States Social security contributions Wage bill 1.5 1.9% of state government States wages Municipal Taxes Urban property (IPTU) Assessed value 3.1 2.6 of rental income Municipalities Transfer of fixed assets Value of asset 0.8 0.9% of services GDP Municipalities (ITBI) Services (ISS) Value-added 4.4 Municipalities Fees Various 2.0 Municipalities Other Various 0.1 Municipalities Source: Carrizosa 2000. 245 246 ELIANA CARDOSO income countries (see Beck, Demirgüç-Kunt, and Levine 1999). The loans made by private commercial banks have short maturity and even short-term credit is scarce. In 2000, the average stock of credit by the private financial sector to the private sector represented just 14.6 percent of GDP (table 7.2). The wide bank spreads are un- doubtedly a contributing factor, associated as they are with the extremely high lending rates noted above. One causal factor in these wide spreads is certainly the low ability of creditors to enforce claims.1 But it is the other tax-like factors on which this chapter focuses. The Macroeconomics behind Brazil’s High Interest Rates This section looks at some of the major macroeconomic develop- ments that formed the backdrop against which bank interest rates have been determined. Real Interest Rates in the 1980s and 1990s In Brazil, the negative real interest rates that characterized the fi- nancial repression in the 1960s and 1970s are long gone. On aver- age over the ten years between 1975 and 1984, real interest rates on deposits were negative because of interventions, unexpected changes in inflation, and imperfect indexation (table 7.3). Thereafter, with increasing financial liberalization, real interest rates on time deposits rose sharply averaging 10 percent between 1985 and 1989, though with huge fluctuations (table 7.4). With the acceleration of inflation in the mid-1980s, indexation intervals became shorter and a share of deposits was held in accounts linked to the daily behavior of over- night interest rates. Between 1984 and 1994, the annual rate of inflation exceeded 100 percent in all years except 1986. Yet though confidence in the finan- cial system was damaged by a spectacular series of failed stabilization plans, involving six monetary reforms in ten years,2 together with a moratorium on external debt in 1987 and the 1990 deposit freeze,3 inflation did not destroy the Brazilian economy. Indexation, the adaptive policy response, became pervasive throughout the economy and its capacity to accommodate inflation may partially explain Brazil’s failure to engage in serious structural change before 1995. By 1992, when President Fernando Collor was ousted from power in a corruption scandal, inflation reached 1,000 percent per year and exceeded 2,000 percent in December 1993. With daily indexation of Table 7.2. Loans as Percent of GDP, Brazil, 1989–2000 Loans from the Loans from the public financial private financial sector sector Loans from public and private financial sectors To the To the To the To the To the Total to public private public private public Personal Business the private Total Year sector sector sector sector sector loans loans sector loans 1989 14.7 25.7 1.6 16.5 16.3 1.5 40.8 42.3 58.6 1990 7.1 11.2 0.7 9.3 7.9 0.6 19.9 20.5 28.4 1991 8.0 14.3 0.8 11.3 8.8 0.9 24.7 25.6 34.4 1992 9.4 18.2 0.8 15.1 10.2 1.4 31.9 33.3 43.5 1993 11.3 23.4 0.7 22.3 12.0 2.3 43.4 45.7 57.8 1994 4.3 12.2 0.3 12.2 4.5 2.3 22.1 24.4 28.9 1995 4.6 13.8 0.4 14.1 4.9 2.4 25.5 27.8 32.8 1996 4.8 12.6 0.4 13.6 5.2 2.4 23.8 26.2 31.4 1997 3.7 12.0 0.3 13.3 4.0 3.2 22.0 25.3 29.2 1998 2.3 13.3 0.2 13.3 2.5 3.7 22.9 26.7 29.1 1999 1.8 13.6 0.1 13.5 2.0 3.9 23.2 27.1 29.1 2000 1.2 11.9 0.1 14.6 1.4 4.6 21.8 26.4 27.8 Note: Loans as percent of GDP were calculated by dividing the average stock of credit (between December of current year and December of previous year) by GDP. Source: Central Bank of Brazil. 247 248 ELIANA CARDOSO Table 7.3. Real Interest Rates, Brazil, 1970–2001 (percent per year) Passive Active real rates real rates Certificate of Commercial Working Average of Period time deposit paper capital active rates 1970–74 3 24 — — 1975–79 –3 14 10 — 1980–84 –6 20 20 — 1985–89 10 * * — 1990–94 19 — 42 — 1995–98 22 — 74 92 1999–2001 6 — — 45 — Not available * Information for active rates between 1985 and 1989 from different sources is inconsistent. Note: Real interest rates are defined as r = [(1 + i)/(1 + π)] – 1 where i is the an- nualized average monthly interest rates and π is the general price index (IGP-DI). Calculations use the general price index because current consumer price indices are not available for earlier periods. Real interest rates between 1999 and 2001 are higher when consumer price indices are used in place of the general price index. Sources: Central Bank of Brazil, Institute of Economic Research (IPEA), Andima, and Broadcast. Table 7.4. Spreads between Active and Passive Annual Interest Rates, Brazil, 1970–2001 (percent) Between rates Between rates Between on commercial on working average paper and capital and active rates certificate of certificate of and certificate Period time deposit time deposit of time deposit 1970–74 20 — — 1975–79 18 13 — 1980–84 28 28 — 1985–89 — — — 1990–94 — 19 — 1995–98 — 43 57 1999–2001 — — 37 — Not available Note: Spreads are defined as s = (1 + rl)/(1 + rd) –1, where rl is the real interest rate on loans and rd is the real interest rates on time deposits, as defined in table 7.3. Source: Table 7.3. SEIGNIORAGE AND BANK SPREADS IN BRAZIL 249 Figure 7.3. Passive Real Interest Rates before the Real Plan, Brazil, January 1970–June 1994 (percent) 10 8 6 4 2 0 –2 –4 –6 –8 –10 0 2 4 6 8 0 2 4 6 8 0 2 4 197 197 197 197 197 198 198 198 198 198 199 199 199 n. n. n. n. n. n. n. n. n. n. n. n. n. Ja Ja Ja Ja Ja Ja Ja Ja Ja Ja Ja Ja Ja Note: Gray = actual monthly rates; black = moving average. Source: Instituto de Pesquisa Econômica Aplicada (IPEA), Rio de Janeiro. financial assets, real interest rates on time deposits jumped to 19 per- cent on average in the first half of the 1990s (table 7.3, figure 7.3). Launched in 1994, the Real Plan combined a brief fiscal adjust- ment, a monetary reform, and the use of the exchange rate as a nominal anchor. Stabilization was supported by very tight monetary policy: real deposit rates averaged 22 percent per year between June 1995 and December 1998 and the authorities sharply increased re- serve requirements (see discussion below). The plan brought infla- tion under control with remarkable speed: measured by consumer prices, it fell from four digits in 1994 to two digits in 1995 and to less than two percent in 1998. Nevertheless, the real exchange rate appreciated sharply. The difference between domestic and foreign interest rates re- sulted in increased external borrowing and helped finance the cur- rent deficit resulting from real appreciation, providing apparent stability. To avoid a monetary expansion induced by capital flows, inflows were partly sterilized, and this entailed sizable fiscal costs, given the international interest differential that had opened up. Banking Problems By increasing the cost of debt servicing, high real interest rates not only complicated fiscal adjustment but also contributed to the dete- 250 ELIANA CARDOSO rioration of bank portfolios, particularly those of public banks, fur- ther straining the fiscal resources needed for restructuring. Between mid-1994 and mid-1997, the central bank intervened in 51 banks and 140 other financial institutions. The failure of two big banks (Banco Econômico and Banco Nacional) prompted the creation of a program providing assistance to private banks, known as PROER. Public banks had also undergone restructuring before the collapse of the real. In August 1996, a program called PROES (a sister pro- gram to PROER) was introduced to reduce the role of state govern- ments in the banking system and curb credit expansion to states and municipalities by allowing the central government to finance the re- structuring of state banks. State bank claims on impaired assets were exchanged for central government bonds, with the state govern- ments becoming, in turn, debtors to the central government. The state governments had to liquidate, privatize, or ensure that state banks would be run on a commercial basis. Alternatively, they could be transformed into non–deposit-taking development agencies. After the restructuring, the share of state banks assets in the fi- nancial system fell sharply, though the largest financial institutions in the country are still the federally owned Banco do Brasil and Caixa Econômica Federal. In 2000, the average stock of credit from public banks to the private sector was 12 percent of GDP, compared to 14.6 percent from private banks (table 7.2). After the Real Plan During these years, the general lack of confidence in the ability of the regime to sustain the exchange rate anchor and to meet its obliga- tions was reflected in the increasing use of dollar-denominated and floating rate debt. By early 1999, 21 percent of domestic public debt was dollar-denominated and 70 percent was indexed to the over- night interest rate. Moreover, maturities fell; the interest due on do- mestic debt in January 1999 alone exceeded 6 percent of GDP. Given the lack of fiscal consolidation, external international shocks to con- fidence in 1997 and especially in 1998—combined with strong re- sistance by the domestic business community to the record high in- terest rates that were being employed in an attempt to stem capital outflows—forced the government to float the real on January 15, 1999; six weeks later it had depreciated by 35 percent.4 Charting the appropriate course of monetary policy in subse- quent months required balancing the risk of a return to the old story of persistent inflation (if interest rates were left too low) against the danger of pushing the economy into a severe recession—not only costly in itself but a threat to the government’s counter-inflationary SEIGNIORAGE AND BANK SPREADS IN BRAZIL 251 Figure 7.4. Real Interest Rate, Brazil, 1992–2001 (Selic rate deflated by the consumer price index, percent per year) 60 Before After the Real Plan Inflation targeting 50 stabilization (semi-rigid rate) (floating rate) Russian 40 Mexican crisis crisis (August 1998) (December 30 1994) Asian Floating the real crisis (January 1999) (July 1997) 20 10 0 Jan. 1992 Jan. 1994 Jan. 1996 Jan. 1998 Jan. 2000 Source: BBV Bank. resolve. In any event, the timing and scale of the interest rate in- creases in the early months of the float were successful in shifting the economy from a potentially explosive situation to a path of steadily declining inflation, allowing real interest rates to fall grad- ually. A formal inflation-targeting policy was adopted in June 1999 and succeeded in meeting the stringent targets of 8 and 6 percent for 1999 and 2000. Interest rates and reserve requirements were also reduced (table 7.3, figure 7.4). Despite this success, lending rates and bank spreads continue to be very high (tables 7.4, 7.5). The central bank calculates that op- Table 7.5. Spreads between Active and Passive Monthly Interest Rates and Net Margins of Commercial Banks, Brazil, 1995–2000 (percent) Spread between the average Commercial bank margin net active interest rates per month of administrative expenditures, and the rate on certificate expenditures with bad loans, Period of deposits per month and explicit taxes 1995 5.3 0.90 1996 3.6 0.77 1997 3.1 0.76 1998 3.3 0.90 1999 3.2 1.01 2000 2.5 1.01 Source: Central Bank of Brazil. 252 ELIANA CARDOSO Figure 7.5. Monthly Spread between Loan and Deposit Rates and Net Margin of Banks, Brazil, 1995–2001 (12-month moving average, percent) 6 1n(1+spread) 5 1n(1+margin) 4 3 2 1 0 95 95 96 96 96 96 97 97 97 97 98 98 98 98 99 99 99 00 00 00 00 01 99 O ly 2 1 . 1 01 01 Ju 200 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 19 ct 0 20 ly ct. an. ril uly ct. an. ril uly ct. an. ril uly ct. ril ly ct. n. ril ly ct. n. n. il u J O J Ap J O J Ap J O J Ap J O p J O Ja p Ju O Ja u pr Ja A A A Source: Central Bank of Brazil. erational expenses and provisions for non-performing loans account for 35 percent of the total spread between deposit and lending rates; direct and indirect taxes explain an additional 30 percent and the net bank margin the remaining 35 percent (Central Bank of Brazil 2001). The sharp decline in spreads between 1996 and 2001, shown in figure 7.5, derives in part from the decline in operational costs; the end of inflation forced banks to merge and reduce the numbers of branches and staff. Other factors also contributed to this decline: the tax on financial operations (IOF) was cut from 6 to 1.5 percent in October 1999. Reserve requirements also declined between 1996 and 2000, as discussed in the next section. Reserve Requirements and Commercial Bank Seigniorage The central bank requires each bank to hold a minimum amount of specified reserve assets, including cash deposits, with the central bank, in proportion to deposit liabilities (tables 7.6, 7.7). The dif- ference between the market interest rates on short-term securities and the interest rate paid on required reserves represents a tax. As in the case of currency, the government is borrowing at below mar- SEIGNIORAGE AND BANK SPREADS IN BRAZIL 253 Table 7.6. Rates of Required Reserves before the Real Plan Brazil, 1969–93 (percent of deposits) Demand depositsa Time deposits Month and year Region Ab Region Bb Region Ab Region Bb April 1969 27 18 9 4.5 May 1971c 27 18 9 4 January 1973 27 18 0 0 March 1973d 27 18 0 0 July 1973e 27 18 0 0 July 1974f 27 18 0 0 February 1975g 27 18 0 0 July 1975h 27 18 0 0 April 1976 33 18 0 0 July 1976 35 18 0 0 October 1977 40 18 0 0 July 1979i 40 18 0 0 December 1984 40 18 22 22 June 1985 40 18 20 20 January 1992 40 18 0 0 November 1993 40 40 0 0 aThe periods and formula for calculating average deposits on which required re- serves were to be based changed many times between 1969 and 1993. bRegion B: Acre, Alagoas, Amazonas, Bahia, Ceará, Espírito Santo, Goiás, Maranhão, Mato Grosso, Pará, Paraíba, Pernambuco, Piauí, Rio Grande do Norte, Sergipe. Region A: All other states. cA reduction of one-half percentage point in required reserves will make up re- sources from demand deposits destined to loans to small and medium enterprises. dA reduction of 2 percentage points in required reserves will make up resources from demand deposits to be destined to loans to exporters. ePercentage of demand deposits destined for rural credit increased from 10 percent to 15 percent. fAmount to be destined to loans to small and medium enterprises increased to 4 percent of demand deposits. gMandatory loans for working capital of small and medium enterprises increases to 8 percent of demand deposits. h55 percent of required reserves to be held in government bonds (LTN or ORTN). iPercentage of demand deposits destined for rural credit increased from 15 percent to 17 percent. Source: Central Bank of Brazil. ket interest rates. Commercial banks collect seigniorage (or an in- flation tax) on non–interest-bearing demand deposits (Brock 1989). Non–interest-bearing reserve requirements reduce this revenue. Commercial banks can pass this loss of revenue on to depositors, who will receive lower interest rates on deposits, and to borrowers, 254 ELIANA CARDOSO who will face higher interest rates on loans. The spread between de- posit and loan rates will increase. How much the deposit rate will fall and how much the loan rate will increase depends on the elas- ticity of demand for loans compared with the elasticity of supply of deposits, assuming that both markets clear. The inflation rate will also interact with reserve requirements to increase the spread be- tween the two rates, depending on how depositors allocate their money holdings between currency, demand deposits, and time de- posits (McKinnon and Mathieson 1981). For most of the period between 1969 and 1993, reserve require- ments on time deposits in Brazil were zero (table 7.6). In the poorest regions of the country (Acre, Alagoas, Amazonas, Bahia, Ceará, Es- pírito Santo, Goiás, Maranhão, Mato Grosso, Maranhão, Pará, Paraíba, Pernambuco, Piauí, Rio Grande do Norte, and Sergipe), re- quired reserves on demand deposits were 18 percent from 1969 to 1993, when they were increased to 40 percent. In the richest regions, reserve requirements on demand deposits increased from 27 percent in 1969 to 40 percent between 1977 and 1993. Reserve requirements were seen as a way of taxing the profits that would accrue to the banks during periods of high inflation. Re- stricted competition prevented interest competition for deposits, al- lowing banks to earn high profits on non–interest-bearing demand deposits. Reserve requirements represented a tax on these profits. But after 1975 and until mid-1994, 55 percent of reserve require- ments on demand deposits could be held in government securities. Between 1969 and 1993, a percentage of demand deposits were earmarked for rural credit, loans to exporters, and loans to small and medium enterprises (see notes to table 7.6). Although the situ- ation has improved since 1995, banks are still required to allocate 25 percent of average demand deposit balances to rural credit and 60 percent of savings deposits to real estate finance. In principle, the impact of forced investments on spreads is similar to reserve re- quirements. But the interest paid on government directed lending contributes to meeting the interest cost of deposits. Currently there are no ceilings on the rates commercial banks can charge for loans mandated for the rural sector. For real estate lending, the gross yield matches the cost of funding of savings deposits. In mid-1994, the Real Plan increased reserve requirements on de- mand deposits to 100 percent (60 percent on other liquid resources, and 20 percent on time deposits (table 7.7). The required reserves- to-deposit ratio rose from an average of 26 percent during Janu- ary–June 1994 to 64 percent during November 1994–April 1995. This increase in reserve requirements and the decline of inflation led to a substantial loss of seigniorage revenue for deposit banks. SEIGNIORAGE AND BANK SPREADS IN BRAZIL 255 Table 7.7. Rates of Required Reserves after the Real Plan, Brazil, 1994–2001 (percent) Demand Time Saving Credit FIF-30 Period deposits deposits deposits operations FIF-CP days Before the Real Plan 40 0 15 0 0 0 July 1994 100 20 20 0 0 0 Aug. 1994 100 30 30 0 0 0 Oct. 1994 100 30 30 15 0 0 Dec. 1994 90 27 30 15 0 0 April 1995 90 30 30 15 0 0 May 1995 90 30 30 12 0 0 June 1995 90 30 30 10 0 0 July 1995 83 30 30 10 35 10 Aug. 1995 83 20 15 8 40 5 Sept. 1995 83 20 15 5 40 5 Nov. 1995 83 20 15 0 40 5 Aug. 1996 82 20 15 0 42 5 Sept. 1996 81 20 15 0 44 5 Oct. 1996 80 20 15 0 46 5 Nov. 1996 79 20 15 0 48 5 Dec. 1996 78 20 15 0 50 5 Jan. 1997 75 20 15 0 50 5 Dec. 1998 75 20 15 0 50 5 March 1999 75 30 15 0 50 5 May 1999 75 25 15 0 50 5 July 1999 75 20 15 0 50 5 Aug. 1999 75 20 15 0 0 0 Sept. 1999 75 10 15 0 0 0 Oct. 1999 65 0 15 0 0 0 March 2000 55 0 15 0 0 0 June 2000 45 0 15 0 0 0 Sept. 2001 45 10 15 0 0 0 Source: Central Bank of Brazil. Bank seigniorage revenue (or inflationary revenue) is the increase in non–interest-bearing demand deposits (∆DD) minus the increase in non–interest-bearing required reserves (∆RR).5 Figure 7.6 shows commercial bank seigniorage revenue divided by loans. Between 1970 and 1989, bank seigniorage was high rela- tive to more recent periods. It reached a peak immediately after the Cruzado Plan in 1985, when prices were frozen and money growth increased ahead of inflation. It turned temporarily negative in 1987, when inflation accelerated ahead of money growth. Between 1990 and mid-1994, a period of extremely high infla- tion, bank seigniorage declined as people economized on their hold- 256 ELIANA CARDOSO Figure 7.6. Seigniorage Collected by Commercial Banks as Share of Loans, Brazil, 1971–2001 (12-month moving average, percent) 2.0 1.5 1.0 0.5 0.0 –0.5 70 73 76 79 82 85 88 91 97 00 94 19 19 19 19 19 19 19 19 19 20 Source: Central Bank of Brazil and IPEA. 19 ings of nonremunerated real demand deposits. In 1995, bank seig- niorage revenue turned negative. The Real Plan’s sharp increase in reserve requirements reduced bank seigniorage. Under the Real Plan, the share in total seigniorage seized by the central bank increased from an average of 60 percent in the first half of 1994 to 84 percent a year later. As a consequence, the share in GDP of seigniorage seized by deposit banks fell from 2 percent to close to zero and seigniorage collected by the central bank rose from 1.8 percent of GDP in 1993, the peak inflation year, to 3 percent in 1994, the year of the Real Plan (Cardoso 1998). This appropriation of seigniorage from the banking sector to the central bank helped finance government spending as inflation ebbed, but it also put the banking sector at risk. Lending interest rates and bank spreads increased sharply, as did non-performing loans. As already mentioned, these elements exposed the weak- nesses of the Brazilian banking sector in the mid-1990s. After 1998, as reserve requirements declined, bank seigniorage recovered. The data suggest that there is a Laffer curve for bank seigniorage in relation to inflation. Seigniorage increases with inflation but as inflation continues to increase, the demand for real money, includ- ing interest-free demand deposits, declines more than proportion- ally with the increase in inflation, and bank revenue from seignior- age declines (table 7.8). SEIGNIORAGE AND BANK SPREADS IN BRAZIL 257 Table 7.8. Ratio of Bank Seigniorage to Loans, Inflation Rate, and Required Reserves, Brazil, 1970–2000 Average Range of required Bank seigniorage inflation rate reserves on divided by loans (IGP-DI) demand deposits Period (percent) (percent per year) (percent) 1970–74 0.485 21 18 to 27 1975–79 0.223 46 18 to 40 1980–84 0.364 143 18 to 40 1985–89 0.500 507 18 to 40 1990–June 1994 0.189 1,660 18 to 40 July 1994–1998 –0.118 16 60 to 100 1999 0.086 20 60 to 65 2000 0.225 10 45 to 55 Sources: Central Bank of Brazil and IPEA. The objective of the next section is to determine the impact of seigniorage on bank spreads and margins. Empirical Evidence on the Link between Seigniorage and Spreads Bank spreads and margins reflect the cost of intermediation. Ac- cording to Ho and Saunders (1981), banks are risk-averse dealers in loan and deposit markets where loan requests and deposit funds ar- rive randomly. Bank interest margins are fees charged by banks for the provision of liquidity. Allen (1988) extends the model to ac- count for cross-elasticity of demand between bank products and Angbazo (1997) extends it for default risk. Wong (1997) confirms the results of the earlier models and predicts that bank interest mar- gins are positively related to the banking sector’s market power, op- erating costs, credit risk, and interest rate risk. Explicit and implicit taxes on financial intermediation can also raise spreads. Taxes on financial transactions drive a wedge between what borrowers pay and lenders receive, thus increasing the spread (Chamley and Honohan 1993). The wedge will reduce the total amount of resources passing through the financial system and raise the rate paid by borrowers or lower deposit rates, depending on the elasticity of demand for credit and on the elasticity of supply of de- posits (Hanson and Rocha 1986). Reserve requirements can also drive a wedge between borrowing and lending rates and thus act as an implicit tax on financial inter- mediation, if the interest rate on required reserves is lower than the 258 ELIANA CARDOSO interest rate on deposits. The wedge develops because the reserve requirement allows only a fraction of the deposits to be loaned. Therefore the lending rate must exceed the deposit rate in order to cover the total interest due on deposits. In the case where both deposits and required reserves are interest- free, an increase in required reserves would transfer seigniorage rev- enue from commercial banks to the central bank and act as an in- crease of taxes on bank profits. Koyama and Nakane (2001) find a positive long-run relationship between bank spreads in Brazil and operational costs, indirect taxes, and country risk. They use the Johansen co-integration test and find a positive but nonsignificant relationship between bank spreads and the required rate of reserves. But the co-integration test might not be appropriate, inasmuch as the rates of required reserves and indirect taxes are not known to be nonstationary. This section takes a different approach to test the impact of re- serve requirements on bank spreads. The effect of reserve require- ments on bank spreads depends on the interaction with inflation and deposit demand and its ability to affect bank seigniorage rev- enue and spreads. Thus this section tests the hypothesis that a gain in bank seigniorage revenue reduces bank spreads and margins. Table 7.9 reports results of regressions of the spread between average lending rates and the interest rate on time deposits after the Real Plan. The lagged inflation rate, lagged operational costs, lagged provisions for overdue loans, and lagged direct and indirect taxes are included in regressions 3 and 4 in table 7.9. The ratio be- tween seigniorage revenues and loans appear with different lags in all specifications. Table 7.10 reports results of regressions of the bank margin net of explicit taxes, operational costs, and provisions for bad loans. Table 7.11 reports results of regressions of the spread between the interest rate on loans to working capital and the inter- est rate on time deposits from 1990 to 2000. Definitions of vari- ables and data sources are listed in the appendix. The estimation technique is ordinary least squares. In all the regressions, bank seigniorage revenue has the expected negative sign and is significant. An increase in seigniorage collected by commercial banks reduces the spread between deposit and loan rates. The variable measuring bank seigniorage revenue is defined as a single-month value of bank seigniorage relative to loans with four lags in regressions 1 and 4. It is defined as the six-month moving av- erage of bank seigniorage relative to loans in regressions 2 and 4, again with a lag of four months. As expected, the effect of the aver- age seigniorage is bigger than the effect of a single month. A 100 percent increase in the six-month average bank seigniorage four months earlier reduces spreads between 15 and 30 percent in differ- SEIGNIORAGE AND BANK SPREADS IN BRAZIL 259 Table 7.9. Determinants of the Spread between Active and Passive Rates after the Real Plan Regression 1 Regression 2 Regression 3 Regression 4 (1994:12– (1994:12– (1995:01– (1995:01– Variable 2001:09) 2001:09) 2001:09) 2001:09) Constant 0.001 0.001 0.012 0.013 (1.08) (1.23) (15.97) (20.01) Bank seigniorage –0.074 — –0.070 — revenue (t–4) (–2.05) (–2.53) 6-month average — –0.148 — bank seigniorage (–1.64) — revenue (t–2) 6-month average bank seigniorage — — — –0.305 revenue (t–4) (–4.45) Inflation rate 0.047 0.053 0.058 0.041 (t–1) (1.54) (1.80) (2.83) (2.49) Direct taxes — — 0.97 0.93 (t–1) (23.42) (19.57) Indirect taxes — — 0.997 0.94 (t–1) (8.08) (8.28) Operational costs — — 0.517 0.543 (t–1) (4.39) (5.15) Provisions for overdue loans — — 0.785 0.749 (t–1) (7.88) (8.37) Spread 0.95 0.94 — — (t–1) (29.24) (25.98) Number of observations 82 82 81 81 Adjusted R2 0.95 0.95 0.95 0.96 Durbin-Watson 1.90 1.88 1.66 1.83 Notes: All variables expressed as natural logarithms (see appendix for definitions). Estimation by OLS with Newey-West HAC Standard Errors and Covariance (t statistics in parentheses) (lag truncation = 3) Source: Author’s calculations. ent specifications, while a 100 percent reduction in bank seigniorage four months earlier reduces the spread by 7 percent. Regressions 9 and 10 reported in table 7.11, including years be- fore the Real Plan, show a bigger impact of bank seigniorage on spreads. A 100 percent increase in the six-month average of the ratio between bank seigniorage and loans two months earlier re- 260 ELIANA CARDOSO Table 7.10. Determinants of the Commercial Bank Net Margin after the Real Plan Regression 5 Regression 6 Regression 7 Regression 8 (1994:12– (1994:10– (1994:11– (1994:11– Variable 2001:09) 2001:09) 2001:09) 2001:09) Constant 0.0025 0.0024 0.002 0.0025 (2.94) (2.93) (1.72) (3.11) Bank seigniorage –0.023 — — — revenue (t–3) (–2.25) — — — Bank seigniorage –0.027 — — — revenue (t–4) (–1.95) — — — 6-month average of bank seigniorage — –0.056 — — revenue (t–2) — (–1.34) — — 6-month average of bank seigniorage — — –0.094 –0.063 revenue (t–3) — — (–2.57) (–1.82) Inflation (t–1) 0.011 — 0.025 — (0.81) — (2.02) — Inflation (t–2) 0.021 — — — (0.96) — — — February 95 dummy 0.001 0.001 0.0015 0.001 (Mexico’s contagion) (3.76) (2.67) (2.40) (3.75) February 99 dummy 0.002 0.003 0.0023 0.003 (the real floats) (2.65) (15.59) (11.48) (17.71) Margin (t–1) 0.687 0.734 0.662 0.722 (6.98) (8.34) (7.90) (8.39) Trend — — 0.0003 — — — (1.05) — Number of observations 82 84 83 83 Adjusted R2 0.63 0.61 0.62 0.61 Durbin-Watson 1.99 2.03 2.04 2.00 Notes: All variables expressed as natural logarithms (see appendix for definitions). Estimation by OLS with Newey-West HAC Standard Errors and Covariance (t statistics in parentheses) (lag truncation = 3) Source: Authors’ calculations. duces the current bank spread by 73 percent. Results in table 7.11 have to be looked at with a grain of salt because data for interest rates before 1995, compiled from different sources, are not based on large samples such as the information published by the central bank for the period after September 1994. SEIGNIORAGE AND BANK SPREADS IN BRAZIL 261 Table 7.11. Determinants of the Spread between Interest Rates on Working Capital Loans and Interest Rates on Time Deposits Regression 9 Regression 10 Variable (1990:03–2000:12) (1990:03–2000:12) Constant 0.0195 0.0196 (10.23) (8.48) Bank seigniorage revenue –0.268 — (t–2) (–1.84) — 6-month average of bank — –0.727 seigniorage revenue (t–2) — (–2.99) Inflation –0.070 –0.061 (–6.69) (–6.11) March 1990 Dummy 0.171 0.166 (Collor Plan) (12.87) (28.31) July 1994 Dummy 0.192 0.192 (Real Plan) (15.81) (17.96) Spread (t–1) 0.244 0.236 (5.40) (3.34) Number of Observations 130 130 Adjusted R2 0.79 0.80 Durbin-Watson 1.91 2.03 Notes: All variables expressed as natural logarithms (see appendix for definitions). (t statistics in parentheses) Source: Authors’ calculations. Inflation Costs The reasons usually given for intermediation costs to rise with in- flation are the following: Inflation decreases the maturity of con- tracts and thus requires more frequent interest rate transactions per unit of assets. Chronic inflation leads to an expansion of the branch network as banks compete for low-cost deposits by offering more services and branches. The variable measuring operational costs in regressions 3 and 4 probably captures both effects better than the inflation rate variable. In regressions 1 to 8 in tables 7.9 and 7.10, the coefficient of lagged inflation rate, following the Real Plan, is positive but small. In regressions 9 and 10 in table 7.11, covering a period of high and volatile inflation rates, the inflation rate has a negative and signifi- cant coefficient that perhaps reflects inflationary revenue not fully captured by the bank seigniorage revenue. 262 ELIANA CARDOSO It is also worthwhile observing that in regressions 3 and 4, cov- ering the period of low inflation after the Real Plan, the response of spreads to changes in inflation and bank seigniorage revenue is smaller than the response to other variables such as explicit taxes on financial intermediation and the costs of provisioning for non- performing loans. Explicit Taxes The results in table 7.9 (regressions 3 and 4) suggest that both indi- rect and direct taxes pass through completely to bank customers. Thus the evidence does not support the notion that corporate in- come taxes, as opposed to indirect taxes on financial intermedia- tion, are not a distorting tax on bank profits. The complete pass-through of income taxes in a context of grow- ing capital mobility is also consistent with the assumption that in- ternational investors demand a net-of-tax return on capital invested in the country. Non-Performing Loans Non-performing loans are often seen as a source of upward pressure on bank spreads and real lending rates because bankers try to off- set their losses on their non-performing loans by charging higher in- terest rates to their performing borrowers. The evidence in table 7.9 is consistent with this hypothesis. A 100 percent increase in the ratio of provisions for overdue loans to credit increases the spread between deposit and lending rates by 80 percent. The evidence for Brazil is consistent with the evidence from in- dustrial countries where non-performing loans are associated with higher spreads. It contradicts the evidence for Latin America in Brock and Rojas-Suarez (2000). It is also meaningful to observe that while average bank spreads between monthly deposit and loan rates fell between 1996 and 2000, the ratio of the interest rate on working capital loans to non- prime borrowers relative to the interest rate on working capital loans to prime borrowers increased (figure 7.7). High lending rates may reflect financial distress among non-prime borrowers.6 Operational Costs Figure 7.8 shows a dramatic decline in operational costs of banks between 1996 and 2001. According to regressions in table 7.9, only SEIGNIORAGE AND BANK SPREADS IN BRAZIL 263 Figure 7.7. Risk Spread and Average Intermediation Spread, Brazil, 1995–2000 (percent) 1.0 5.5 Spread nonprime-prime borrower (lhs) 5.0 Spread loan-deposit rates 0.9 (rhs) 4.5 4.0 0.8 3.5 3.0 0.7 2.5 0.6 2.0 9 95 996 996 9 97 9 97 9 98 9 98 999 9 99 0 00 .1 .1 .1 .1 .1 .1 .1 .1 .1 .2 ug Fe b ug Fe b ug Fe b ug Fe b ug Fe b A A A A A Source: Published interest rates. Figure 7.8. Operational Costs Relative to Credit, Brazil, 1995–2001 (percent) 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 5 5 96 6 97 7 98 8 99 9 00 0 01 99 99 99 99 99 99 00 19 19 19 19 20 20 1 .1 .1 .1 .1 .1 .2 ne ne ne ne ne ne ne ec ec ec ec ec ec Ju Ju Ju Ju Ju Ju Ju D D D D D D Source: Central Bank of Brazil. 264 ELIANA CARDOSO half the decline in operational costs is passed through to a decline in bank spreads. This points to less-than-perfect competition in the Brazilian banking sector. Net Margin Table 7.10 reports the results of the regression of commercial bank net margin. Bank net margin is calculated as a residual by the Central Bank of Brazil (2001). (It deducts from interest rates on loans the deposit interest rate and the effect of direct and indirect taxes, operational costs, and provisions for overdue loans.) The co- efficient for bank seigniorage has the expected negative sign and is significant. A 100 percent increase in the six-month average of bank seigniorage lagged two months reduces the net margin between 6 and 9 percent. Lagged inflation also has a small and positive impact on the net margin. Conclusion Brazil’s financial development is still lacking: it has all the basics, but depth and term finance are absent. Term credit in the free seg- ments of the market does not exist because of very high real inter- est rates and uncertainty linked to a history of high inflation. Recent demonstrations of commitment to low inflation, an important ele- ment of the inflation targeting adopted since 1999, will help build confidence. Domestic deposit rates remain high even in the absence of ex- pected large exchange rate depreciation because high government debt coupled with political uncertainty raises concerns among in- vestors. Persistent fiscal adjustment will contribute to the credibility of macroeconomic policies and to a sustained reduction in interest rates. Recently adopted technology for credit scoring and credit in- formation will also help improve the reach of the financial system and access for small entrepreneurs. Brazil’s banking system can benefit from more competition. There is still a significant share of the banking sector in state hands and further privatization will help in this regard. It is also suggested that promoting competition from new, low-cost banks, especially foreign banks, can reduce spreads. Brazil’s government has carefully managed the entry of new foreign banks, aware that their cost ad- vantage could derive from a low-cost labor force with little senior- ity, rather than any difference in real efficiency. Whereas the open- SEIGNIORAGE AND BANK SPREADS IN BRAZIL 265 ing of the domestic banking sector spelled disaster for local banks in other emerging economies, the major Brazilian banks have in- creased their market share and margins. Concentration of the banking sector continued to increase in 2001. To judge from the data, the banks seem to have chosen to earn high profits rather than compete vigorously. Year-end results for 2001 shows that Brazilian banking sector profitability con- trasted sharply with the modest results of the vast majority of non- financial enterprises. The net profits of the 31 largest banks more than doubled between 2000 and 2001. The analysis in this chapter suggests that reducing reserve re- quirements and directed credit could reduce bank spreads and net margin. Yet experience suggests that these measures can succeed only if supported by adequate fiscal policy. Appendix. Definition of Variables The spread between active and passive rates is defined as the natural loga- rithm of the ratio between one plus the average interest rate on loans in the segment of free loans and one plus the interest rate on time deposits of 30 days. Operational costs are defined as the natural logarithm of one plus the ratio of operational costs and the volume of credit (calculated from a sam- ple of 17 large banks, Central Bank of Brazil 2001, annex I). Provisions for overdue loans are defined as the natural logarithm of one plus the ratio of provisions for overdue loans and the volume of credit (based on a sample of 17 large banks in Central Bank of Brazil 2001, annex I). Direct taxes are defined as the natural logarithm of one plus the ratio between the burden of the Income Tax and the Pre-Tax Corporate Income (the CSLL) in a 30-day loan financed by a 30-day time deposit based on tax rates and simulations (Central Bank of Brazil 2001, annex I). Indirect taxes are defined as the natural logarithm of one plus the indi- rect tax rates. Indirect taxes include the tax on financial operations (IOF), taxes on gross revenues (PIS and COFINS), and the tax on bank debits (CPMF). Commercial bank seigniorage revenue is defined as the natural loga- rithm of one plus the ratio of bank seigniorage and the credit of banks to the private sector. Bank margin is the natural logarithm of one plus the net margin calcu- lated as a residual after deducting from lending rates: the passive rate and the expenses with administration, bad loans, and taxes. 266 ELIANA CARDOSO Data and Sources Interest rates on time deposits are from the Central Bank of Brazil and from the Institute of Applied Economic Research (IPEA) in Rio de Janeiro, Brazil. Consistent interest rates on loans do not exist for the whole period be- tween 1970 and 2001. For the period between September 1994 and September 2001, the Cen- tral Bank of Brazil has published the average of interest rates on different instruments calculated from large samples. Interest rates on loans between 1990 and 2000 are interest rates on working capital loans from Andima until 1991, from DIESP until 1993, and from the Central Bank between 1994 and 2000. Interest rates on loans between 1970 and 1980 are interest rates on com- mercial paper (ao mutuário) from the website of Instituto de Pesquisa Econômica Aplicada (IPEA). Data on credit, monetary base, and demand deposits are from the IPEA data bank. The general price index, IGP-DI, from Fundação Getúlio Vargas, Rio de Janeiro, is also available on the IPEA website. Notes 1. Courts are inefficient: Dakolias (1999) shows that there are 2,975 cases pending per judge in Brasilia and 3,129 in São Paulo, compared to 58 in Singapore and 244 in Hungary. The time taken to resolve a case (number of cases pending at the start of the year divided by the number of cases re- solved during this year) is 1.9 years in Brasilia and 1.6 years in São Paulo, compared to 0.04 years in Singapore and 0.5 years in Germany. Pinheiro and Cabral (1999) estimate that a judicial execution to recover a creditor claim can take between one and ten years. The government is working to improve these conditions. Congress is examining the creation of a Bank Credit Bill (cédula de crédito bancário), a credit instrument that allows the collection of debt under commercial law instead of civil law and thus in- creases the speed with which a loan claim can be executed. Among measures adopted to reduce costs of financial intermediation, the central bank mod- ernized the payment system and introduced a Credit Risk Data Center. The central bank now makes available on its website standardized information on credit operations, including interest rates for each type of operation, de- gree of arrears, and average term differentiated by financial institution. 2. In the 20th century, Brazil had eight monetary reforms that removed zeros from the previous currency and changed the name of the currency, as follows: Mil-Réis (1900–42), Cruzeiro (1942–66), Cruzeiro Novo (1967– 69), Cruzeiro (1970–86), Cruzado (1986–89), Cruzado Novo (1989–90), Cruzeiro (1990–93), Cruzeiro Real (1993–94), and Real (1994–2000). SEIGNIORAGE AND BANK SPREADS IN BRAZIL 267 3. In early 1990, when inflation reached close to 3,000 percent per year, the Collor Plan of March 1990 drastically cut liquidity. An arbitrary freeze was imposed for 17 months on nearly two-thirds of the money supply (M4), broadly defined to include demand deposits, mutual funds, federal bonds, state and municipal bonds, saving deposits, and private bonds. Although Brazilians eventually managed to circumvent some of these controls, the fi- nancial freeze took over personal assets and was wildly unpopular. 4. Because the PROER program had restored bank balance sheets to health, and because many had anticipated the devaluation and positioned themselves to benefit from it through the holding of dollar-denominated gov- ernment bonds and financial derivatives, the banking system survived the de- valuation and did not become a destabilizing factor, as in other countries. 5. Observe that ∆DD – ∆RR ≡ ∆M1 – ∆H: that is, the difference be- tween total seigniorage and the seigniorage collected by the central bank. ∆M1 is total seigniorage: the sum of the increase in currency, ∆C, and ∆DD. ∆H is the seigniorage collected by the central bank: ∆C + ∆RR. Thus ∆DD – ∆RR ≡ ∆DD + ∆C – ∆C + ∆RR ≡ ∆M1 – ∆H In figure 7.6, the ratio between commercial banks seigniorage and loans is (∆M1 – ∆H)/L, where ∆M1 is the increase in the monthly average stock of M1; ∆H is the increase in the monthly average stock of high powered money; and L is outstanding credit to the private sector, average between current and previous month. 6. Patrick Honohan suggests that the widening spread on non-prime over prime rates could instead reflect changing competitive conditions in banking, with the top firms getting increased access to competitively sup- plied finance so that banks have to squeeze poorer risks harder to cover their costs. References Allen, Linda. 1988. “The Determinants of Bank Interest Margins: A Note.” Journal of Financial and Quantitative Analysis 23 (2): 231–35. Angbazo, Lazarus. 1997. “Commercial Bank Net Interest Margins, Default Risk, Interest-rate Risk, and Off-Balance-Sheet Banking.” Journal of Banking and Finance 21 (1): 55–87. Beck, Thornsten, Aslı Demirguç-Kunt, and Ross Levine. 1999. “A New Database on Financial Development and Structure.” Policy Research Working Paper 2146. World Bank, Washington, D.C. Brock, Philip. 1989. “Reserve Requirements and the Inflation Tax.” Jour- nal of Money, Credit and Banking 21 (1): 106–21. Brock, Philip, and Liliana Rojas-Suarez. 2000. “Understanding the Behav- ior of Bank Spreads in Latin America.” Journal of Development Eco- nomics 63 (1): 113–34. 268 ELIANA CARDOSO Cardoso, Eliana. 1998. “Virtual Deficits and the Patinkin Effect.” IMF Staff Papers 45 (4): 619–46. Carrizosa, Maurício. 2000. Brazil: Structural Reform for Fiscal Sustain- ability. Washington D.C.: World Bank. Central Bank of Brazil. 2001. Juros e Spread Bancário no Brasil: Avaliação de Dois Anos do Projeto. Brasília, DF: Central Bank of Brazil. Chamley, Christophe, and Patrick Honohan. 1993. “Financial Repression and Bank Intermediation.” Savings and Development 17 (3): 301–08. Dakolias, Maria. 1999. Court Performance Around the World: A Compar- ative Perspective. World Bank Technical Paper 430. World Bank, Wash- ington, D.C. Hanson, James, and Roberto Rezende Rocha. 1986. “High Interest Rates, Spreads, and the Costs of Intermediation.” Industry and Finance Series 18. World Bank, Washington, D.C. Ho, Thomas S., and Anthony Saunders. 1981. “The Determinants of Bank Interest Margins: Theory and Empirical Evidence.” Journal of Financial and Quantitative Analysis XVI (4): 581–99. Koyama, Sérgio Mikio, and Márcio I. Nakane. 2001. “Os determinantes do Spread Bancário no Brasil.” In Central Bank of Brazil, ed., Juros e Spread Bancário no Brasil. Brasília, DF: Central Bank of Brazil. McKinnon, Ronald, and Donald Mathieson. 1981. “How to Manage a Re- pressed Economy.” Princeton Essays in International Finance 145. Prince- ton University, International Economics Section, Princeton, N.J, Pinheiro, Armando Castelar, and Célia Cabral. 1999. Credit Markets in Brazil: The Role of Judicial Enforcement and Other Institutions. En- saios BNDES 9. Rio de Janeiro: Banco Nacional de Desenvolvimento. Wong, Kit Pong. 1997. “On the Determinants of Bank Interest Margins under Credit and Interest Rate Risks.” Journal of Banking and Finance 21 (2): 251–71. World Bank. 2001. Finance for Growth: Policy Choices in a Volatile World. New York: Oxford University Press. 8 Taxation of Financial Intermediaries as a Source of Budget Revenue: Russia in the 1990s Brigitte Granville Control over the instruments of monetary policy, conventionally seen as a means of maintaining price stability, can also be employed as a means of providing implicit subsidies and imposing implicit taxes. The degree to which—and the manner in which—this can happen de- pends on how the banking system is structured, and on the political and statutory role of the central bank. In turn, the effectiveness of both the banking system and of monetary policy in achieving its con- ventional goal is strongly affected by the degree to which government spending, tax, and deficit policies are operated through the use of monetary policy instruments, such as the minimum reserve require- ment and the interest rate on central bank lending. Russia provides a dramatic case study of a dysfunctional quasi- fiscal role for finance. Russia began its reform efforts in January 1992 not only with budget and banking systems inherited from the centrally planned economy, but also in the wake of de facto insol- vency. After all, in December 1991, the Soviet Vneshekonombank (which had been responsible for the country’s external and foreign currency obligations) had defaulted on all its liabilities, including short-term trade debt. 269 270 BRIGITTE GRANVILLE The subsequent decade can be divided into three main periods for the purposes of this study. For most of the 1992–95 period, seigniorage coming from the nominal growth in the monetary base (including required and excess reserves) provided financial support to fiscal and quasi-fiscal expenditures. At the same time, high infla- tion in 1992–94 allowed taxpayers to exploit time lags between the calculation and payment of tax liabilities. This in turn adversely af- fected real tax revenues, already depleted by the adverse effects on compliance of the erosion of private income through recession, as well as through inflation itself. The revenue decline increased the fis- cal pressure for monetary financing of the budget deficit, given the lack of any alternative. Seigniorage declined between 1995 and 1997, reflecting efforts to reduce and stabilize inflation. The primary budget deficit was re- duced and there was a shift in the financing of the budget deficit away from central bank credits in favor of Treasury bills.1 In addi- tion, the independence of the Central Bank of Russia (CBR) was in- creased2 and the floating exchange rate was replaced in July 1995 with a (crawling) peg regime. Nevertheless, the ratio of domestic debt to GDP and the burden of servicing that debt grew so fast that financing the debt became unsustainable, resulting in the August 1998 default. Considerable progress has been achieved in the fiscal position since 1998, reducing government recourse to the CBR, as well as to foreign borrowing and other forms of domestic financing. Thanks to high oil prices, the CBR has been able to build up its foreign ex- change reserves. The monetary base expanded—now reflecting a growth in real money demand rather than inflation. Thus seigniorage again became an important source of revenue, though it is now being used to ensure external debt servicing capacity, and hence solvency. The next section describes several features of the system, espe- cially the directed credits of the 1992–95 period and the high bill yields of 1995–98. The banking system benefited from these con- ditions, at least for a while, though others were taxed, whether through the inflationary consequences of the credit expansion or the collapse of 1998, which was the culmination of the high-yield bill pe- riod. The chapter then analyzes the extent and role of two important implicit financial sector taxes: the inflation tax on currency (which fell disproportionately on lower-income households), and unremu- nerated reserve requirements (which were largely passed through to banking customers in the extremely high banking spreads). The chapter concludes by discussing the persistent lack of depth in the Russian banking system, partly a consequence of the implicit taxes discussed in the previous sections. TAXATION OF FINANCIAL INTERMEDIARIES 271 The Central Bank and the Banking System The Soviet State (Central) Bank was formally liquidated on Decem- ber 25, 1991 and the Central Bank of Russia (CBR)—subsequently called Bank of Russia (BOR)—took over its responsibilities. Until 1994, the CBR remained a passive institution charged with financ- ing a budget deficit that was large but at the same time difficult to measure, due the extent of fiscal and quasi-fiscal expenditures. At least until 1994, the financing of explicit fiscal expenditures, including the servicing of external debt, accounted for a relatively low proportion of money creation. In contrast, a large proportion was attributable to subsidies or transfers to the enterprise sector: in other words, to quasi-fiscal expenditures. This is reflected in the balance sheet of the CBR by the fact that the sizeable growth in domestic assets (NDA) was dominated until mid-1993 not by credits to government but by credits to commer- cial banks. These credits were designed not only to provide liquid- ity to banks through the normal refinance mechanism; they also in- cluded subsidized credits funded by the budget and the central bank and channelled through commercial banks to state enterprises. Only after mid-1993 did credits to government—that is, explicit mone- tary financing of the budget deficit—become the main source of do- mestic asset growth. Directed Credits The quasi-fiscal activities of the central bank were financing activi- ties that were not purely monetary in nature.3 Their typical form was the provision of subsidized credit. The legacy of the central planned economy was visible here in the difficulty of distinguishing between budget and credit financing, with both banks and enter- prises being state-owned. Enterprises would receive subsidies either from the government budget or from a bank in the form of sub- sidized credits. These credits were designated “directed credits” or centralized credits because they originated from the center—whether from the central bank or the government—but were channelled through commercial banks. A Federal Treasury Department was not introduced until 1994 and became operational only in 1998.4 Mean- while, budgetary funds were being distributed partly by the CBR and partly by designated commercial banks. Directed credits were used both to provide liquidity to banks and support specific enter- prises, sectors, and regions in Russia and in the Former Soviet Re- publics (FSRs). 272 BRIGITTE GRANVILLE One of the reasons that state enterprises had the political weight to obtain credits at subsidized interest rates was that they tradition- ally undertook a large amount of the social expenditures (housing, kindergartens, medical care, and so on) benefiting their employees, as well as implicit unemployment insurance (Boycko and Shleifer 1994). These credits were directly funded by the CBR and did not appear in the budget. This confusion of fiscal and monetary policy made the system of transfer to enterprises complex and nontransparent. A principal justification offered for this kind of support to enter- prises was the need to keep the level of employment stable. It was difficult to re-deploy workers, not least because of repressive policies preventing workers from moving to where they could gain employ- ment (migration between cities was not allowed without a residence permit). Unemployment was kept at artificially low levels through such “directed” credits and other subsidies. During the period 1992–98, registered unemployment remained between 1.1 and 2.6 percent of the labor force.5 Even under the more rigorous Interna- tional Labour Organization (ILO) definition, it varied between 4.8 and 11.9 percent for the same period. Low unemployment figures, however, had their counterpart in a dramatic fall in labor produc- tivity. Indeed, the nonpayment of wages to workers meant that many jobs were in fact fictitious (Commander and Tolstopiatenko 2001). This policy aggravated the shrinking of the tax base. It would have been cheaper—because more effectively directed—to have fi- nanced direct provision of welfare benefits to workers. Directed credits were allocated following requests from enter- prises to the Supreme Soviet (or legislature), the government, and even in some cases directly to the CBR.6 The commercial banks were left as little more than the passive instruments of such deci- sions, reminiscent of their position under the old command-admin- istrative system. The credits themselves were rarely reimbursed.7 As for supplying liquidity to the banks, faster change in the com- mercial banking system could have been promoted by introducing a straight discount window. Instead, credit auctions were introduced only in February 1994 and even then represented a small share of refinancing to banks. (The first Lombard credit auctions were not until April 1996.) The Pattern of Credit Outlays in Detail Central bank credits to enterprises fell into two categories (table 8.1), those channelled through commercial banks, and those that began as credits to the Ministry of Finance but were on-lent to enterprises. TAXATION OF FINANCIAL INTERMEDIARIES 273 Table 8.1. Central Bank of Russia Credit Allocations, 1992–93 1992 1993 Allocation In Rbn % of GDP In Rbn % of GDP Total CBR creditsa 5,703 31.6 24,790 15.3 Budget 1,189 6.6 11,276 6.9 Enterprises 3,608 20.0 11,111 6.8 Via commercial banks 2,804 15.5 8,150 5.0 Agriculture and Roskhleboprodukt 1,300 7.2 3,616 2.2 Energy 400 2.2 193 0.1 Northern Territories 300 1.7 2,134 1.3 Industry 500 2.8 466 0.3 Other 304 1.7 421 0.3 Regions with urgent need — — 1,320 0.8 Via Ministry of Finance 804 4.5 2,961 1.8 Working capital 600 3.3 — — Investment 105 0.6 700 0.4 Military conversionb 77 0.4 — — Roskhelboprodukt 22 0.1 1,566 1.0 Other budget loans — — 695 0.4 Other republics 906 5.0 2,403 1.5 — Not available. aThe ratio of CBR directed credit over the total stock of CBR gross credit to banks amounted to 99 percent in 1992 and in 1993. b In 1993, conversion credits for military-industrial enterprises classified as subsi- dies in fiscal accounts. Sources: Based on data from CBR, Ministry of Finance, and IMF (1995, table 35, p. 96). The credits channelled through commercial banks totalled Rbs 2,804 bn or 15.5 percent of GDP in 1992, and Rbs 8,150 bn or 5 percent of GDP in 1993. They were directed or “centralized” cred- its in the sense that through its regional branches, the central bank informed the commercial banks (usually former state banks special- ized in the sector chosen for the credit) which state enterprises were to receive the credit and at what interest rate. The credits were con- centrated on the agriculture8 and energy sectors. Most of these cred- its were to compensate for the price controls that stayed in force in 274 BRIGITTE GRANVILLE those sectors. For instance, bakery enterprises were subsidized for any producer price above Rbs 12/kg. The sizeable energy subsidy represented, broadly speaking, the difference between the value of potential energy exports restricted through quotas and domestic prices of primary energy sources.9 The Central Bank’s credits directed to enterprises via the Ministry of Finance amounted to Rbs 804 bn or 4.5 percent of GDP in 1992, and 1.8 percent of GDP in 1993. Such credits were allocated off- budget to enterprises. They enabled the Ministry of Finance to bor- row a greater amount from the CBR than the ceiling imposed by the Supreme Soviet, since special credit lines had been secured for these operations separate from the overall credit ceiling. These credits in- cluded military conversion and working capital credits. Military conversion credits were subsidies to the Military- Industrial Complex (MIC) used to maintain employment in a sector that would otherwise have been left crippled by the collapse of state defense procurement orders. Civilian products represented only some 44 percent of the output of the MIC enterprises in 1988. Rbs 77 bn were allocated during 1992.10 The origin of working capital credits lay in the central planning system, when enterprises had automatically obtained working cap- ital from the budget. During the second half of 1992, the govern- ment faced an increasing number of complaints from firms claiming that they could not borrow from commercial banks, the interest rate being too high (although it was negative in real terms) and the term too short (usually three months or less). The government agreed to extend working capital loans through the central bank. These loans were usually for two years at subsidized interest rates. The sum al- located—Rbs 600 bn—was meant to restore the 1992 real value of working capital. Scale and Subsidy Element By the end of 1992, credits to state enterprises allocated either by the Central Bank or by the government amounted to almost 20 per- cent of GDP; while another 25 percent of GDP had been allocated in explicit subsidies through the budget (table 8.2). Financial trans- fer to state enterprises thus amounted to 45 percent of GDP in 1992. In 1993 and 1994, directed credits programs were decreased, mostly in the areas of social expenditures and unemployment insur- ance. Centralized credits were officially ended by a presidential de- cree (no. 244) in February 1995. The problem of credits was not confined to their size. They were, in addition, allocated largely at concessional interest rates substan- TAXATION OF FINANCIAL INTERMEDIARIES 275 Table 8.2. Explicit Subsidies to State Enterprises, Russia, 1992 Value (Rbn) % of GDP Explicit subsidies 4,454 24.7 Agriculture 308 1.7 Coal 180 1.0 Local budgets 585 3.2 Other 30 0.2 Interest rates of which 630 3.5 CBR credits 495 2.7 Government credits 135 0.7 Centralized imports 2,721 15.1 Source: IMF (1993, table A1, p. 139). tially below the CBR refinance rate (which was already highly neg- ative in real terms), the difference being paid by the federal budget. For instance at the beginning of 1993, agricultural credits were al- located at an annual rate of 25 percent (plus a 3 percent commis- sion for the commercial bank), while the refinance rate stood at an annual 100 percent (as of March 30, 1993). The subsidy covered the 75 percent difference, which was supplied by the federal budget at the end of the fiscal year. Total subsidies on interest rates—whether from the CBR or from the government—amounted to 3.5 percent of GDP in 1992. Theoretically, the responsibility for these credits and their repay- ment lay with the commercial banks, which became the distributing intermediaries. The credits were provided for one year, although normal commercial credits were granted for no more than three months. Firms supposedly had an incentive to repay them in order to get more funds. But the negative interest rates effectively meant that directed credits amounted to grants. And no action at least until the end of 1994 was taken against a bank or a firm that failed to re- imburse such a loan. Banks for their part showed no reluctance to handle such credits, especially because banks usually did not imme- diately channel the money to the earmarked firm.11 On October 1, 1993, subsidized credits were cancelled. Rouble Zone Another important counterpart to the expansion of the monetary base during 1992 and 1993 was credits to Former Soviet Republics. The CBR provided both non-cash and cash credits to these coun- 276 BRIGITTE GRANVILLE tries of the “near-abroad” to allow their enterprises to continue trading with Russian enterprises. The credits financed these coun- tries’ imports from Russia and thus had much the same effect as di- rect subsidies to Russian industries. The main difference was that with pure subsidies, the products might not even have been sold but instead might simply have piled up at the factory. The Rouble zone was especially costly in 1992. Central bank credits to the Former Soviet Republics alone reached 8.5 percent of Russian GDP if delivery of cash is excluded, and 11.6 percent if cash is included (Granville 1997). At the same time, the FSRs’ own central banks were themselves able to issue rouble credits to be spent in Russia, thus further contributing to the growth of Russia’s money supply and inflation. In 1992–93, conflict between President Boris Yeltsin and the Su- preme Soviet was reflected in conflicting policy stances of the Min- istry of Finance and the central bank on the question of resolving the rouble zone issue. The Executive wanted to end the Rouble zone, a policy opposed by the central bank, which was constitu- tionally subordinate to, and politically protected by, the Supreme Soviet. While the Ministry of Finance managed in July 1992 to reg- ulate the expansion of non-cash credit, cash transfers remained be- yond its control. The CBR exploited the failure to bring cash trans- fers under Ministry of Finance control by making transfers in cash, in the form of pre-1993 banknotes: that is, without the Russian flag. This loophole was stopped abruptly in July 1993 with the removal from circulation in Russia of the pre-1993 roubles, an action that precipitated the collapse of the Rouble zone. The Market for Government Bills If banks benefited from their role as passive intermediaries in the process of directed credits, they profited even more—at least for a time—from treasury operations in foreign exchange and govern- ment bills. Whereas private banks actively speculated in foreign exchange gains during the earlier phase from 1992 to 1995, they changed their strategy after the floating exchange rate was replaced with a (crawl- ing) peg regime in July 1995. From 1995 to 1998 they switched their holdings from foreign exchange to rouble Treasury bills (GKOs), the market for which had been introduced in May 1993. Central bank data show that income on GKOs accounted for 41 percent of the earnings of the top 100 banks as of September 1997. Thanks to the high yield on GKOs, Sberbank (Russia’s oldest bank), with about 35 percent of the market, returned pre-tax profits of $US2 billion in TAXATION OF FINANCIAL INTERMEDIARIES 277 1996. Since around two-thirds of Sberbank’s liabilities consisted of household deposits, Sberbank acted as a conduit for channelling housing liquid assets into government debt. In contrast to Sberbank, the large commercial banks relied for their build-up of GKO hold- ings on the profitability of GKOs themselves and on various forms of foreign borrowing. The balance sheet position that resulted was a mirror image of that which had prevailed in the period up to mid- 1995; instead of large foreign currency holdings matched by rouble liabilities, there were now GKO holdings matched by foreign cur- rency liabilities.12 During this period, the majority of banks derived their profit- ability from the spread between deposit rates and government bond yields. In other words, instead of being taxed, banks were subsi- dized and protected by the CBR policy by artificially maintaining high yields on GKOs. This policy had several elements. First access to foreigners was restricted until 1996; then foreign investors were forced under the terms of the “S-account” system to purchase dol- lars forward at artificial prices in order to cap the covered GKO yields.13 De facto full liberalization came only months before the de- fault, in January 1998. In August 1996, foreign investors were allowed to enter the mar- ket by opening a specially created new category of accounts, known as “S” accounts, in authorized Russian banks. The bank would then transfer the funds to its account at the Moscow Interbank Currency Exchange (MICEX), where it would buy and sell GKOs according to its foreign client’s instructions. Foreigners investing in the GKO-OFZ market (OFZ were Treasury bonds) through “S” accounts could thus trade freely in the secondary market. To prevent rapid outflows of foreign money, however, the central bank required foreign investors to tie up their investment for three months before repatriating their returns. At the beginning of this three-month period, foreigners had to enter a forward contract with CBR, with the rouble/dollar ex- change rate being artificially set to ensure a fixed dollar return of 15 percent.14 Such a massive transfer to the banks was economically harmful across the board, not least to the fiscal position and the sound development of the Russian banking system itself. Forms of Taxation The private sector (households, enterprises, and banks) suffers from inflation because of the heightened costs attached to holding non– interest-bearing monetary assets, specifically currency and bank re- serves. Currency is mainly held by households, while other central 278 BRIGITTE GRANVILLE bank liabilities are held by the banking industry in the form of re- quired reserves. Inflation Tax on Currency In terms of its distributional effect, inflation is clearly regressive. This is not only because of the limited ability of pensioners and low- wage earners to shelter their income against inflation, but also be- cause access in practice to foreign currencies or other inflation-proof assets is generally restricted to higher-income groups and more mo- bile groups. The wealthiest sections of the population have easier access to income-bearing, indexed, or otherwise protected assets. In Russia although the minimum sum required to open a deposit ac- count at the saving bank Sberbank15 was, until October 1, 1993, just 10 roubles—and as such, a small barrier to this form of saving—this minimum was increased in dramatic steps; by August 1995, it was 300,000 Rbs—equivalent to more than half the then-prevailing monthly average wage rate. By effectively precluding deposits by small savers, this increased the regressivity of the inflation tax.16 Tables 8.3 and 8.4 present some data on the inflation tax on cur- rency (C) and bank reserves. Seigniorage on currency is measured as a flow (∆C). Seigniorage has been divided into an inflation compo- nent measured on a stock basis (πt MBt) and a residual reflecting changes in velocity of currency. In both cases, revenue from the in- flation tax on currency was initially very important, but declined steeply over the period. From table 8.4 it can be seen that seignior- age from bank reserves was at a higher level during the more infla- tionary quarters: for instance, from 1994Q I to 1995QII, 1998QIV to 1999QII, and 2000QI to 2000QIII. Minimum Reserve Requirements From the beginning of 1992 minimum reserve requirements were increased sharply: from 2 to 15 percent on short-time deposits (term less than one year) and 10 percent on long-term deposits. In April 1992 the requirements were further raised to 20 percent and 15 per- cent for short-term and long-term deposits, respectively.17 They stayed at this high level until May 1996, when they were decreased to 18 percent and 14 percent, respectively (table 8.5). Note, however that these rather high ratios were not achieved in practice. The actual requirement in practice worked out at a much lower proportion of deposits. Thus minimum required bank reserves as reported by the BOR at the end of December 1992 amounted to no more than 11 percent of total deposits reported for the same date, TAXATION OF FINANCIAL INTERMEDIARIES 279 Table 8.3. Seigniorage on Currency, Russia, 1993–2001 (percent) Inflation Inflation Seigniorage component Residual consumer from currency from currency component Currency Year and prices % GDP %GDP % GDP %GDP quarter (1) (2) (3) (4) = (2) – (3) (5) 93 IV 68.9 — — — 13.4 94 I 46.6 5.0 8.0 –3.1 17.2 94 II 25.2 5.8 4.4 1.4 17.4 94 III 18.5 3.2 3.1 0.1 16.7 94 IV 42.3 2.9 6.5 –3.6 15.3 95 I 50.0 –0.7 6.4 –7.1 12.7 95 II 27.6 5.5 4.0 1.5 14.6 95 III 17.8 2.7 2.5 0.3 13.8 95 IV 13.8 2.8 2.1 0.8 14.9 96 I 10.9 1.3 2.0 –0.8 18.7 96 II 6.4 3.5 1.3 2.1 20.4 96 III 2.1 –1.6 0.3 –1.9 16.2 96 IV 3.1 1.3 0.5 0.8 15.1 97 I 4.9 0.2 0.9 –0.7 17.4 97 II 4.0 5.2 0.9 4.3 21.8 97 III 1.8 –0.3 0.3 –0.7 19.2 97 IV 0.6 –0.6 0.1 –0.7 17.6 98 I 3.2 –1.5 0.7 –2.2 21.1 98 II 1.5 1.9 0.3 1.6 20.5 98 III 16.1 3.9 3.6 0.3 22.1 98 IV 39.9 4.8 8.8 –4.0 22.2 99 I 22.9 –1.6 4.6 –6.2 20.1 99 II 8.6 4.9 1.7 3.2 19.5 99 III 6.2 –0.3 1.0 –1.8 15.7 99 IV 4.0 4.0 0.8 3.2 18.7 00 I 4.5 –1.1 0.8 –1.8 17.2 00 II 3.7 4.8 0.7 4.1 19.6 00 III 5.3 1.8 0.9 0.9 17.5 00 IV 4.9 3.4 1.1 2.4 21.4 01 I 6.7 –1.0 1.4 –2.4 21.2 01 II 5.6 4.0 1.3 2.7 22.5 —Not available. Source: Calculated from IMF, International Financial Statistics, March 1998 and January 2002. as compared with the 17.5 percent average of nominal required ra- tios. There is no one obvious explanation for the difference between minimum percentage reserve requirements imposed by law and the actual minimum reserves held by banks. Various hypotheses may be advanced. 280 BRIGITTE GRANVILLE Table 8.4. Seigniorage on Bank Reserves, Russia, 1993–2001 (percent) Seigniorage Inflation Seigniorage from bank consumer Bank from reserves as % Year and prices reserves bank reserves government quarter % per year as % deposits as % GDP revenue 93 IV 68.9 71.1 n.a. n.a. 94 I 46.6 60.1 0.9 n.a. 94 II 25.2 51.5 2.7 20.3 94 III 18.5 60.5 5.1 44.7 94 IV 42.3 49.8 1.2 8.7 95 I 50.0 49.9 2.9 21.3 95 II 27.6 46.4 3.1 22.7 95 III 17.8 39.3 –0.6 –3.9 95 IV 13.8 35.3 0.9 6.8 96 I 10.9 35.2 1.1 9.9 96 II 6.4 33.4 0.0 –0.2 96 III 2.1 35.3 1.4 11.4 96 IV 3.1 33.9 0.2 1.1 97 I 4.9 33.7 0.8 10.1 97 II 4.0 32.7 0.4 3.4 97 III 1.8 28.5 –0.8 –6.8 97 IV 0.6 33.6 2.2 13.1 98 I 3.2 29.7 –1.3 –16.0 98 II 1.5 27.3 –1.1 –9.2 98 III 16.1 27.6 –1.5 –14.9 98 IV 39.9 31.3 3.0 19.7 99 I 22.9 41.3 4.6 44.3 99 II 8.6 44.8 3.6 22.9 99 III 6.2 42.3 0.4 3.1 99 IV 4.0 41.6 1.6 9.8 00 I 4.5 50.1 4.7 29.9 00 II 3.7 51.3 2.8 14.6 00 III 5.3 52.0 2.4 14.2 00 IV 4.9 45.6 0.0 0.1 01 I 6.7 41.5 –0.9 –5.4 01 II 5.6 37.3 –0.2 –1.0 01 III 2.3 35.6 — — n.a. Not applicable. — Not available. Source: Calculated from IMF, International Financial Statistics, March 1998 and January 2002. A first hypothesis is that the method of calculating minimum re- serve requirements permitted a certain amount of manipulation. Banks were allowed to choose the deposit base on which reserve re- quirements were calculated. Two different methods were available: the average of the balances at the end of consecutive five-day periods Table 8.5. Minimum Reserve Requirements, Russia, 1995–98 (percent) Type of Sberbank Less than account Over 3 individual Dates in effect: 1 montha 1–3 monthsb monthsc FXd accountse Feb. 1–April 30, 1995 22 15 10 2 — May 1, 1995–April 30, 1996 20 14 10 1.5 — May 1–June 10, 1996 18 14 10 1.25 — June 11–July 31, 1996 20 16 12 2.5 — Aug. 1–Oct. 30, 1996 18 14 10 2.5 — Nov. 1, 1996–April 30, 1997 16 13 10 5 10 May 1–Nov. 11, 1997 14 11 8 6 9.5 Nov. 12, 1997–Nov. 30, 1998 14 11 8 9 9.5 Dec. 1, 1997–Jan. 31, 1998 14 11 8 9 8 Feb. 1–Aug. 23, 1998 11 — — — 8 Aug. 24–Aug. 31, 1998 10 — — — 7 — Not available. aDemand accounts and fixed-term liabilities with terms of 30 days and less. bFixed-term liabilities falling due from 31 days to 90 days. cFixed-term liabilities falling due from 91 days and more. dForeign currency accounts. eIndividuals’ rouble accounts regardless of term (Sberbank). Source: Central Bank of Russia. 281 282 BRIGITTE GRANVILLE each month; and the average on daily balances for all days in the month. In addition, all banks were required to provide deposit bal- ances on the first day of the month. Provision of deposit balances re- lating to the sixteenth day of the month was optional. This method allows for two or more banks to collude in shifting deposits in order to obtain a more favorable (lower) reserve requirement.18 The fact that the calculation took place only once a month gave added op- portunity for collusion between banks. This method of calculation was eventually changed in 1995. A second hypothesis is that banks may have been exempted from reserving against an amount of deposits equal to total centralized credits, which would have reduced the reserve requirement by be- tween one third and one half. There was a return to the use of high reserve requirements in the immediate aftermath of the August 1998 crisis. The BOR relied again on an increase in (unremunerated) reserve requirements, though this time the intention was mainly to control liquidity, using increased non–interest-bearing reserve requirements to keep control over re- serve money and therefore pass the cost of sterilization to commer- cial banks. From late 1998, reserve requirements were raised on four occasions from 5 percent (for both rouble and foreign currency de- posits) to 10 percent on foreign currency deposits and corporate rouble deposits and to 7 percent on household rouble deposits. In 1999, commercial banks’ holdings of (unremunerated) free reserves at the CBR increased to the equivalent of more than 10 percent of rouble broad money, well above their pre-1998 crisis level. This in- crease reflected the fact that the collapse of the GKO market had de- prived banks of alternative liquid instruments and that the interbank market remained dormant, as well as the reluctance among banks to increase credits to the economy.19 This policy has allowed the BOR to build up external reserves as inflows increased on the back of the upswing in the commodity price cycle. Excess Reserves In addition to the required reserves, Russian depositories also hold substantial excess reserves (also non–interest-bearing): that is, re- serves above the amount actually reported as minimum reserve re- quirements. This is especially surprising given that excess reserves are deposited by commercial banks at the central bank on a non– interest-bearing account; they are voluntary and can be withdrawn at any time. The scale of excess reserves is interpreted as being in TAXATION OF FINANCIAL INTERMEDIARIES 283 part due to the inefficiency of the inter-bank payment system ef- fected through the CBR inter-branch clearing system. The high level of excess reserves in either nominal or real terms through 1992 and 1993 created a risk of a new acceleration of the money supply, even in the absence of increased credits. In August 1993, for example, compulsory reserves (that is, minimum reserve requirements) stood at Rbs 1,615.1 billion, while funds held in CBR correspondent accounts (that is, excess reserves) amounted to Rbs 4,407 billion. If all the banks with excess reserves had decided to use them at the same time to expand credit, the money supply could have increased by four times. Impact of Financial Intermediary Taxation The inflation tax resulting from the consequent increase in the rou- ble money base has been largely paid by bank depositors. Lately, the abundant rouble liquidity has enabled the government to maintain highly negative real yields on its domestic debt. This, along with the structural distortions caused by tax evasion already mentioned, continues to inhibit the expansion of the long-term deposit base— which, in turn, would enable the banks to reorient their balance sheets to the real sector, thereby evading this latest incidence of the inflation tax. The Burden of High Reserve Requirements Raising the rate of required reserves on deposits is a tempting pol- icy tool. By increasing the monetary base and thereby counter- balancing a decrease in the money multiplier, the result is higher government revenues. These additional revenues are ultimately fi- nanced by depositors and investors. The borrowers and depositors pay for the extra expenditure of the budget financed this way. The cost for the economy is the dis- tortion in the level and structure of real interest rates (that is, high real interest rates on borrowers and low or negative real interest rates on depositors). Table 8.6 presents the actual nominal interest rates on loans and deposits. High reserve requirements impose a tax on bank interme- diation in the absence of reserve remuneration at market rates. This tax results in a widening of the spread between lending and deposit rates. While real lending rates were quite high, real deposit rates were negative until the fourth quarter of 1995 and again after Au- gust 1998. 284 BRIGITTE GRANVILLE Table 8.6. Nominal Interest Rates on Loans and Deposits, Russia, 1994–2001 (percent per year) Deposit Lending Annualized Real Real Year and rate rate inflation deposit lending quarter (id) (il) Spread rate rate (rd) rate (rl) 94 III 181.3 301.6 120.3 97.05 84.25 204.55 94 IV 125.2 297.3 172.1 309.92 –184.72 –12.62 95 I 142.9 416.3 273.4 406.52 –263.62 9.78 95 II 122.4 378.5 256.1 165.26 –42.86 213.24 95 III 72.9 251.1 178.2 92.83 –19.93 158.27 95 IV 69.6 231.9 162.3 67.60 2.00 164.30 96 I 61.7 187.8 126.1 51.10 10.60 136.70 96 II 55.4 176.4 121 28.26 27.14 148.14 96 III 60.1 142.8 82.7 8.84 51.26 133.96 96 IV 43 80.2 37.2 12.94 30.06 67.26 97 I 29.1 63.9 34.8 21.13 7.97 42.77 97 II 18.2 48.6 30.4 17.12 1.08 31.48 97 III 11 38.4 27.4 7.40 3.60 31.00 97 IV 7.4 33.8 26.4 2.58 4.82 31.22 98 I 11.7 32.8 21.1 13.30 –1.60 19.50 98 II 12.6 42.4 29.8 6.14 6.46 36.26 98 III 18.8 46.8 28 81.94 –63.14 –35.14 98 IV 25.1 45.2 20.1 282.52 –257.42 –237.32 99 I 22 45.1 23.1 128.14 –106.14 –83.04 99 II 13.4 39.8 26.4 38.94 –25.54 0.86 99 III 10.4 38.4 28 27.20 –16.80 11.20 99 IV 9 35.5 26.5 17.03 –8.03 18.47 00 I 9.6 31.7 22.1 19.43 –9.83 12.27 00 II 6.6 25.9 19.3 15.69 –9.09 10.21 00 III 5.4 21.4 16 22.76 –17.36 –1.36 00 IV 4.4 18.8 14.4 21.27 –16.87 –2.47 01 I 4.4 18.8 14.4 29.57 –25.17 –10.77 01 II 4.8 17.9 13.1 24.31 –19.51 –6.41 01 III 5.0 18.0 13.0 9.61 –4.61 8.39 Source: Calculated from IMF, International Financial Statistics, March 1998 and January 2002. The Lack of Monetary Depth: A Problem of Trust A long-term consequence of the erosion of savings through the in- flation tax (especially the destruction of real savings balances in the Sberbank during the inflationary burst of 1990–92) and the wide in- termediation spreads is the failure of the banking system to grow (table 8.7). The abrupt destruction of savings in August 1998 added to the malaise by creating a widespread popular distrust of private banks after the events of August 1998. Adding this to the state’s TAXATION OF FINANCIAL INTERMEDIARIES 285 Table 8.7. Deposits in the Banking System, Russia, 1993–2000 (percent of GDP) Demand Time and Foreign currency Total deposits saving deposits deposits deposits 1993 7.3 2.9 7.0 17.3 1994 5.3 3.9 6.1 15.4 1995 4.5 4.5 3.6 12.6 1996 4.1 4.4 3.2 11.8 1997 6.6 3.2 3.2 13.0 1998 5.5 3.5 7.1 16.1 1999 5.5 3.7 6.4 15.5 2000 6.3 3.7 5.9 15.9 Source: Calculated from IMF, International Financial Statistics, March 1998 and January 2002. long track record of monetary confiscation through inflation and (in recent years) by successive devaluations of the rouble, it is easy to explain why Russian citizens prefer to save in foreign currency (in practice, U.S. dollars) rather than roubles. It does not fully explain the weak deposit base, however. Foreign currency deposit accounts have been legal since 1992. Sberbank deposits carry an implicit state guarantee, which was honored in practice after the financial crash of August 1998. It should follow that a large volume of U.S. dollar savings should be kept in the Sberbank.20 Instead, such savings amount to only 6 percent of GDP, with at least a further 15 percent of GDP kept in the form of cash U.S. dollar savings at home. The reason for this is that the distrust of the state extends to fear of state expropriation of wealth, given that virtually all personal wealth has been accumulated without full payment of tax, and so is vulnerable. Russian households therefore prefer to hide cash dollars under the mattress rather than place them in a bank deposit account where they might be discovered by law enforcement officials. With the personal tax reform (including a low 13 percent flat rate of personal income tax) in force from January 2001, the state aims to encourage households and businesses to abandon the gray economy. With time, it is hoped that people will obtain a track record of legality and gradually become inclined to place their sav- ings in deposit accounts. Turning to the assets side of the banking system’s balance sheet, long-term lending to enterprises is constrained by the weak deposit base. Lending to the private sector was about 9 percent of GDP in 1997 and 12 percent in 2000. Twenty-five percent of total banking system credit to the banking sector is accounted by Sberbank. Only about 4 percent of private sector investment is financed by domes- tic borrowing, while 63 percent is financed by the enterprise’s own 286 BRIGITTE GRANVILLE funds. But the absence of intermediation also stems from structural failings that extend beyond those of the banking system itself. Chief among these is the lack of functioning commercial land and hous- ing markets. As a result, firms and households have been deprived of the most obvious source of collateral for borrowing. Just as tax reform is a key to restoring a normal deposit base, so another key structural reform can help solve this problem of collateral: namely, land reform. Important steps were taken in 2001 with the adoption of the Land Code (signed into law on November 8, 2001), and then in 2002, when a separate law on agricultural land was adopted by the federal parliament. This code gives enterprises the right to buy freehold title to the land on which they sit, and it defines a reason- able price scale to prevent the exercise of this right falling victim to predatory local officialdom. Notes 1. 1995 Federal Budget Law. 2. The “Central Bank Law” enacted in April 1995. 3. This section draws on materials contained in Granville (1995). 4. Diamond (2002, pp. 12–13). 5. The Russian definition excludes all job seekers who have an alterna- tive income, such as students and pensioners. 6. Freinkman (1994, p. 7) reports that “the CBR itself initiated special subsidized directed credit programs in 1992 (about 30 percent of total CBR directed credits), which were targeted at expanding working capital of en- terprises and reducing the burden of the arrears crisis.” 7. According to The Wall Street Journal Europe (June 6, 1991), “In 1990, the Soviet government wrote off Rbs 93 bn of bad loans to the agri- cultural sector alone.” 8. Amelina, Galbi, and Uspenskii (1993) show that the volume of grain credits from the CBR in 1992 was equivalent to about Rbs 620 bn or 3.4 percent of GDP, which was about half the credits issued to the agricultural sector in 1992. 9. World Bank and International Monetary Fund (1993, p. 7). 10. Gavrilenkov and Koen (1995, p. 11). 11. Freinkman (1994, p. 10) reports that “Three groups of banks are participating in such programs: a/former state specialised banks (Promstroi and Rosselkhoz) are main channel banks for a wide set of enterprises. b/sec- toral banks (e.g Neftekhim, Gasstroi, Electro, etc.) founded by enterprises in particular subsectors of industry and being under the control of enter- prise associations and concerns, which emerged in place of former line min- istries. c/some new commercial banks founded by truly private sector (e.g. TAXATION OF FINANCIAL INTERMEDIARIES 287 Menatep), which established strong links to the government and are the most successful in extracting various privileges.” 12. See Dmitriev and others (2001). 13. See Willer (2001, pp. 239–260). 14. See Granville (2001, pp. 93–130). 15. The Sberbank is the oldest bank in Russia, founded in 1842. It sur- vived the 1917 Revolution by virtue of five years’ dormancy, which were followed by resuscitation in 1922 as a national savings institution for the masses. Following passage of the Russian law on banks and banking activ- ities of December 2, 1990, the Sberbank became a joint stock commercial bank on March 22, 1991, registering its license on June 20, 1991. It has the advantage of having inherited an immense geographic coverage in terms of a national branch network. The Sberbank’s broad geographical coverage and visibility meant that it enjoyed a degree of trust among ordinary peo- ple, far greater than any of the more modern commercial banks. 16. See Granville and Shapiro (1996). 17. In May 1995, minimum reserve requirements were changed as fol- lows: on deposits up to 30 days, 20 percent; between 30 and 90 days, 14 percent; over 90 days, 10 percent. 18. International Monetary Fund (1995, p. 202). 19. International Monetary Fund (2000, p. 13). 20. A draft law on compulsory insurance of household deposits in pri- vate banks is under consideration at the time of writing, to supplement the implicit guarantee of deposits in the CBR-owned Sberbank (which helps ex- plain why about 85 percent of household deposits are held in Sberbank). The proposed law would endow a government deposit insurance agency with an initial Rbs 6.5 billion transfer from the government, to be supple- mented with contributions from the banks at a rate of 0.15 percent of their deposits. Deposits would be insured up to a ceiling. References Amelina, Maria, Douglas Galbi, and A. Uspenskii. 1993. The Distribution of Central Bank Credits for Grain Procurements. Macro and Financial Unit (MFU), Moscow. Boycko, Maxim, and Andrei Shleifer. 1994. “The Russian Restructuring and Social Assets.” Paper presented at the conference on Russian Eco- nomic Reforms in Jeopardy, Stockholm School of Economics, June 15–16, Stockholm School of Economics, Stockholm, Sweden. Commander, Simon, and Andrei Tolstopiatenko. 2001. “The Labour Mar- ket.” In Brigitte Granville and Peter Oppenheimer, eds., Russia’s Post- Communist Economy. Oxford: Oxford University Press. Diamond, Jack. 2002. “Budget System Reform in Transitional Economies: The Experience of Russia.” IMF Working Paper WP/02/22. Interna- tional Monetary Fund, Washington, D.C. 288 BRIGITTE GRANVILLE Dmitriev Mikhail, Mikhail Matovnikov, Leonid Mikhailov, Ludmilla Sycheva, and Eugene Timofeyev. 2001. “The Banking Sector.” In Brigitte Granville and Peter Oppenheimer, eds., Russia’s Post-Communist Econ- omy. Oxford: Oxford University Press. European Bank for Reconstruction and Development (EBRD). 1997. “Transition Report.” European Bank for Reconstruction and Develop- ment, London. Freinkman, Lev. 1994. “Government Financial Transfers to the Enterprise Sector in Russia: General Trends and Influence on Country Macroeco- nomic Performance.” World Bank, Washington, D.C. Processed. Gavrilenkov, Evgeny, and Vincent Koen. 1995. “How Large was the Out- put Collapse in Russia? Alternative Estimates and Welfare Implica- tions.” In Staff Studies for the World Economic Outlook, Washington, D.C., International Monetary Fund. Granville, Brigitte. 1995. The Success of Russian Economic Reforms. London and Washington, D.C.: Royal Institute of International Affairs, International Economic Programme. Distributed by the Brookings Institution. ———. 1997. “Farewell, Rouble Zone.” In Anders Åslund, ed., Russia’s Economic Transformation in the 1990s. London: Pinter. ———. 2001. “The Problem of Monetary Stabilization.” In Brigitte Gran- ville and Peter Oppenheimer, eds., Russia’s Post-Communist Economy. Oxford: Oxford University Press. Granville, Brigitte, and Judith Shapiro. 1996. Less Inflation, Less Poverty, First Results for Russia. Discussion Paper 68. International Economics Programme, The Royal Institute of International Affairs, London. Granville, Christopher. 2001. “The Political and Societal Environment of Economic Policy.” In Brigitte Granville and Peter Oppenheimer, eds., Russia’s Post-Communist Economy. Oxford: Oxford University Press. International Monetary Fund. 1993. World Economic Outlook. Washing- ton, D.C. ———. 1995. Russian Federation. IMF Economic Review 16. Washington, D.C. ———. 2000. Russian Federation: Staff Report for the 2000 Article IV Consultation and Public Information Notice Following Consultation. Washington, D.C. Willer, Dirk. 2001. “Financial Markets.” In Brigitte Granville and Peter Oppenheimer, eds., Russia’s Post-Communist Economy. Oxford: Ox- ford University Press. World Bank and International Monetary Fund. 1993. Subsidies and Di- rected Credits to Enterprises in Russia: A Strategy for Reform. Wash- ington, D.C. Part III Particular Taxes 9 Corporate Income Tax Treatment of Loan-Loss Reserves Emil M. Sunley The tax treatment of bank loan-losses has been a contentious issue in a number of developing and transition countries. Banks and bank regulators generally want the tax rules for recognizing loan-losses to conform closely to regulatory accounting in order to encourage banks not to under-provision for loan-losses and to ensure a current tax benefit from loss provisioning. Tax officials often are wary of regulatory accounting, and fear that accepting it for tax purposes will significantly reduce income taxes paid by banks. The treatment of loan-losses is the central tax policy issue relat- ing to the taxation of banks, given the importance of loans in bank assets and the cost of bad debts. In the United States, for example, loans and leases represented 60 percent of bank assets in 2000, and loan-loss provisioning represented 21 percent of net income before taxes and provisioning (Bassett and Zakrajs ek 2001). In many de- veloping and transition economies, loan-losses are an even larger share of net income before taxes and loss provisioning. This chapter addresses primarily when and how loan-losses should be recognized as an allowable expense for tax purposes. To address these issues, financial and regulatory accounting of loan- losses first must be considered. 291 292 EMIL M. SUNLEY Financial and Regulatory Accounting for Loan-Losses Loan-losses are an inevitable cost that banks incur in order to earn income, and these losses should be recognized as an expense for both financial and tax purposes.1 For financial accounting, loans are recorded at their face value until they become fully worthless and are written off. However, a provision or reserve account is es- tablished for potential losses present in the portfolio of loans. On a bank’s financial statement, the value of loans is often shown on a net basis (that is, the face value of the loans corrected by the esti- mated loss). In particular, International Accounting Standard IAS 30 (IAS 30 2002) provides that: The amount of losses which have been specifically identified is recognized as an expense and deducted from the carrying amount of the appropriate category of loans and advances as a provision for losses on loans and advances. The amount of potential losses not specifically identified but which experi- ence indicates are present in the portfolio of loans and ad- vances is also recognized as an expense and deducted from the total carrying amount of loans and advances as a provision for losses on loans and advances. Thus IAS 30 recognizes both specific and general reserves.2 Spe- cific reserves are linked to specific loans and the amount of reserve required usually depends on the length of time that payments of in- terest and principal have been past due, the value of any pledged collateral, and the financial soundness of the borrower. The general reserve is not linked to specific loans but reflects losses that experi- ence indicates are present in the portfolio of loans but not yet specif- ically identified. Any amounts set aside for future losses should be accounted for as appropriations of retained earnings: that is, not recognized as a current expense.3 Regulatory accounting for bank loan-losses is very similar to fi- nancial accounting required by IAS 30.4 The Basel Committee on Banking Supervision has outlined sound practices for loan account- ing and disclosure (Basel Committee 1999). In particular, A bank should identify and recognize impairment in a loan or a collectively assessed group of loans when it is probable that the bank will not be able to collect, or there is no longer rea- sonable assurance that the bank will collect, all amounts due according to the contractual terms of the loan agreement. The impairment should be recognized by reducing the carrying amount of the loan(s) through an allowance or charge-off and CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 293 charging the income statement in the period in which the im- pairment occurs. A bank should measure an impaired loan at its estimated re- coverable amount. The aggregate amount of specific and general allowances should be adequate to absorb estimated credit losses associ- ated with the loan portfolio. Like IAS 30, the Basel Committee recognizes both specific and general reserves. The general reserve is to cover latent losses, which are not yet identified but which are known to exist. The general re- serve is not supposed to cover future losses.5 The assessment of both specific and general reserves “should be performed in a systematic way, in a consistent manner over time, in conformity with objective criteria and be supported by adequate documentation” (Basel Committee 1999). That said, the setting of the appropriate level of allowances “necessarily includes a degree of subjectivity” (Basel Committee 1999). The additions to specific and general reserves both reduce re- ported profits. However, they are often accounted for differently on the balance sheet. In Slovenia, for example, the required provision for impaired loans is accounted for on the active side of the balance sheet as an adjustment for doubtful accounts; the provision for per- forming loans is accounted for on the passive side of the balance sheet as part of “other long-term liabilities.” Moreover, general re- serves for performing loans are often counted as part of bank capi- tal, generally tier II capital (discussed further below), as these re- serves are not “pledged” to cover specific loans that are already impaired. The bank regulatory agency in many countries specifies a scheme for classifying loans and setting minimum reserves. The guidelines for the various loan categories—for example, special mention, sub- standard, or doubtful—are often set in terms of past due payments. More forward-looking criteria to reflect expected probability of de- fault (such as the creditworthiness of the borrower) are still un- common (Laurin and Majnoni 2003). For example, in broad outline, Turkey requires specific provision of: • 20 percent of loans with limited potential to be recovered or 90 days in arrears • 50 percent of loans unlikely to be recovered or 180 days in arrears • 100 percent of loans deemed irrecoverable or with arrears over one year. 294 EMIL M. SUNLEY These required provisions are reduced to the extent that doubt- ful or bad debts are covered by guarantees or collateral. The amount of this reduction varies according to the quality of the guar- antee or collateral. In addition, there must be a general provision of 0.5 percent of cash loans and 0.1 percent of contingent liabilities. The Central Bank of the Philippines requires specific reserves ranging from 5 to 100 percent: • 5 percent for “loans especially mentioned” (loans that have potential weakness—past due for 30 to 90 days) • 25 percent for “substandard loans that are unsecured” (loans that involve a substantial and unreasonable degree of risk to the in- stitution because of unfavorable record or unsatisfactory character- istics—past due more than 90 days) • 50 percent for “doubtful loans” (loans that have the weakness inherent in substandard loans, with the added characteristics that the existing facts, conditions, and values make collection or liqui- dation in full highly improbable and in which substantial loss is probable) • 100 percent for “loss” loans (loans considered uncollectible and worthless, and that are past due for a period of at least six months). In addition to the specific reserve, the Central Bank of the Philip- pines requires a general reserve equal to 2 percent of a bank’s un- classified loan portfolio. The rationale for reserve accounting is straightforward. In a port- folio of loans, certain loans are non-performing6 or otherwise im- paired. Some of these loans will ultimately be uncollectible. If, based on past experience, 25 percent of the loans past due for 6 to 12 months are likely to be uncollectible, a reserve should be established for loan-losses equal to 25 percent of the amount of these loans. These losses are already present in the portfolio of loans past due for 6 to 12 months and should be recognized as an expense for this period, even though it is not possible at this time to identify just which loans will ultimately be worthless. The reserve for loan-losses is not for future losses (that is, losses not yet present or latent in the portfolio of loans).7 While banks and other companies need to maintain sufficient equity capital to cover unantic- ipated future losses, additions to the company’s equity should not be treated as a current expense (for either financial or tax purposes). Under a normal reserve method, regardless of how the actual re- serve is calculated, any recoveries for loan-losses are credited to the reserve, and any loans charged off the books are debited to the re- serve. The expense item for loan-losses for any year is equal to the CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 295 amount necessary to bring the reserve up to the end-of-year level, after the beginning-of-year reserve has been adjusted for recoveries and charge-offs. The expense item is computed as follows: Reserve at end of year8 Less: Reserve at beginning of year9 Plus: Loans written off Less: Actual recoveries for loan-losses previously written off Equals: Expense item for loan-losses Determining the expense for loan-losses based on a reserve method appears to provide a double deduction for losses. Each loss, however, reduces profits only once. There is no double deduction because the actual charge-off is debited against the reserve,10 and once a loan is charged off, there would be no end-of-year reserve with respect to that loan.11 The end-of-year reserve relates only to loans outstanding on the bank’s books at the end of the year. There is a closely related issue to the treatment of loan-losses: the treatment of unpaid interest. Most banks, at least large banks, are on the accrual method of accounting12 and thus accrue interest in- come on loans as the claim arises and not when the income is re- ceived. The Basel Committee recommends that a bank should cease accruing interest when a loan is identified as being impaired. For countries that use a loan classification scheme, a loan would be con- sidered impaired when a specific reserve is required for the loan. In addition, when interest ceases to accrue on a loan, uncollected in- terest that had been previously accrued should be reversed. This treatment of unpaid interest seems reasonable, as the bank’s claim on unpaid interest may be of lower quality than the bank’s claim on the principal amount of the loan. Banks, like other businesses, need to maintain sufficient capital to provide a cushion to cover large, unanticipated future losses. To ensure this, the bank regulatory authority in each country requires banks to meet capital adequacy standards. Under the 1988 Basel Accord, tier I capital (basic equity) must be at least 4 percent of risk- weighted assets. Tier II capital (undisclosed reserves, revaluation re- serves, subordinated debt, and general loan-loss provisions) must be at least 8 percent of risk-weighted assets. The general loan-loss pro- vision, however, cannot be greater than 1.25 percent of risk-weighted assets. This, in part, ensures that the regulatory authority does not set a high mandatory general provision to offset the tendency of cer- tain banks to under-provision non-performing loans. To the extent that the prudential rules for loan-loss provisioning and the capital adequacy standards require banks to maintain more 296 EMIL M. SUNLEY equity capital than is truly needed, the capital cost of the bank is increased and nonbank financial institutions may have a competi- tive advantage.13 For example, insurance companies may guarantee bank loans. This may reduce the bank’s need for specific reserves as the guarantee will reduce the reserve required for non-performing loans (see the prudential rules for Turkey described above). In addi- tion, the insurance companies may not have to meet capital re- quirements that are as stringent, giving them a lower cost of capital than banks. Though beyond the scope of this chapter, this all sug- gests that when prudential rules for loan-loss provisioning and cap- ital adequacy standards are too conservative, banks can be put at a competitive disadvantage. The Tax Treatment of Loan-losses: An Introduction The tax treatment of loan-losses varies widely across countries.14 Some countries (the United States for large banks—those with more than $500 million in assets,15 Australia, Korea, Malaysia, and the Philippines) allow only the charge-off method, under which loan- losses are recognized only when loans become worthless. In deter- mining whether a loan is worthless, all pertinent evidence, including continual non-performance, adequacy of collateral, and the finan- cial condition of the debtor should be considered. Under the charge- off method, if an amount previously charged off as uncollectible is later recovered or the loan again becomes performing, the amount previously written off is restored to income. In the Philippines, loan-losses are allowed only for worthless and uncollectible loans that have been charged off the books of a bank’s account as of the end of the taxable year. The tax authority allows a debt to be written off for tax purposes once it has been written off by the bank with the approval of the Central Bank of the Philip- pines. Some countries (Japan and Thailand) set limits on the tax de- duction for loan-losses. In Thailand, banks can deduct loan-loss provisions from taxable income up to 25 percent of net income or 0.25 percent of total outstanding loans, whichever is less. Loan- losses may be written off for tax purposes only when civil action has been brought against the debtor, the debtor has declared bank- ruptcy, or died. Many countries allow a reserve method (that is, provisioning) for accounting for loan-losses for tax purposes, in addition to requiring it for regulatory purposes. However, only a few countries attain full conformity between financial and tax accounting for loan-losses. Many countries (such as Canada, France, Kazakhstan, and the United CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 297 Kingdom) grant tax deductibility to specific allowances or charge- offs in the year they occur, but not for general allowances. Serbia al- lows a tax deduction only for the specific allowance, but gives the allowance a “haircut”—that is, reduces its effective value. Under the new Serbian income tax, banks are allowed a tax deduction equal to 90 percent of the addition to the loan-loss provision re- quired by the national bank for non-performing loans. The Russian Tax Code establishes its own reserve rules (related to the rules of the national bank). A reserve is allowed for loans past due only 45 days, and the total reserve cannot exceed 10 percent of the gross re- ceipts of the tax year. In the Kyrgyz Republic, banks may establish a reserve based on the experience of the leading banks of the world. The reserve shall not exceed 10 percent of the loans outstanding. A few countries allow general provisioning for tax purposes (for ex- ample, based on a percentage of eligible loans), but limits are usu- ally placed on the general provision. Germany requires that the gen- eral provision for tax purposes not exceed 60 percent of average loan-losses over the past five years. Singapore limits the general provision to 3 percent of the amount of qualifying loans (Laurin and Majnoni 2003). Given this wide diversity in tax treatment, there clearly is no generally accepted international standard as to the appropriate tax treatment of loan-losses. In determining the tax treatment of loan- losses, a country should weigh several considerations. Does the charge-off or the reserve method best measure the income of the bank? Should there be full or partial conformity between the regu- latory and tax treatment of loan-losses? Should a tax deduction be allowed for general reserves? If the tax law is changed, for example, to allow a reserve method, how should the transition be treated for tax purposes? Charge-off vs. Reserve Methods Both the charge-off and reserve methods recognize that bad debts are costs of earning income, and thus their cost should be a de- ductible expense for financial and tax purposes. In general, the re- serve method accelerates the recognition of the expense compared to the charge-off method. The fundamental question, however, is which method results in a better matching of income and expenses. This is an empirical question. A test of whether a method for recognizing bad debts results in a proper matching of income and expense is to determine whether it results in the effective tax rate on a portfolio of loans being equal to 298 EMIL M. SUNLEY the nominal tax rate. If the effective tax rate—measured as the per- centage reduction in the rate of return due to taxes—is less than the nominal tax rate, the recognition of the losses is too accelerated. If the reverse is true, the recognition of the losses is too delayed.16 Consider first the following simple example. A bank makes 1,000 loans for $1,000 each at the end of year 0 (table 9.1). Each year 2 percent of the loans default. If the interest rate on these loans is 12.2449 percent, the interest income will cover the 2 percent loss and provide a 10 percent before-tax return.17 If each year’s loss— the principal amount of the loans that defaulted during the year— is allowed as an immediate tax deduction and the nominal tax rate is 30 percent, the after-tax rate of return is 7 percent. Thus the after- tax rate of return is 30 percent lower than the before-tax rate of re- turn, implying an effective tax rate of 30 percent, which is equal to the 30 percent nominal tax rate. In this example, the true economic loss18 in value of the loan portfolio would be permitted as a tax-deductible expense each pe- riod. This is the necessary and sufficient condition to ensure that the effective and nominal tax rates are equal (Samuelson 1964). The key assumption is that the loans default at a constant per- centage rate per year (declining balance assumption).19 Suppose, in- stead, that all loans remain fully performing until the end of the fifth year (table 9.2). At that time, 9.6 percent of the loans default, which is just equal to the cumulative amount of defaults in the first example. If the bad debts were deductible for tax purposes instan- taneously at the end of year 5, the effective tax rate would be 31.4 percent. For the effective tax rate to equal the nominal tax rate, a reserve for future losses would have had to have been established in the earlier years, even though no loans are past due. The annual ad- dition to this reserve would need to be just equal to the decline in the value of the loan portfolio, and this addition would need to be deductible for tax purposes. Over the five-year period, the portfolio would decline in value from $1,000,000 to $903,921 as the end of the five-year period approaches. Put another way, the value of the portfolio at the beginning of year 5 would not be $1,000,000—even though no loans have defaulted as yet. In contrast to the assumption that loans go bad only in the final year, a bank could originate loans at the end of year 0 and certain loans would instantaneously go bad.20 All the other loans would re- main performing for five years, when the principal amount would be paid in full. Both the charge-off method and the reserve method, by permitting the loan-losses to be written off in year 1, would provide too generous tax treatment of the bad debts in that the resulting effective tax rate would be less than the nominal tax rate. The loan- Table 9.1. Income and Cash Flow from Loan Portfolio (constant rate of default) Present Present Before-tax value Taxable Tax After-tax value Year Principal Interest cash flow (r = .10) income (30%) cash flow (r = .07) 0 1,000,000 — –1,000,000 –1,000,000 — — –1,000,000 –1,000,000 1 980,000 120,000 120,000 109,091 100,000 30,000 90,000 84,112 2 960,400 117,600 117,600 97,190 98,000 29,400 88,200 77,037 3 941,192 115,248 115,248 86,588 96,040 28,812 86,436 70,558 4 922,368 112,943 112,943 77,142 94,119 28,236 84,707 64,623 5 903,921 110,684 1,014,605 629,990 92,237 27,671 986,934 703,670 Total — — — 0 — — — 0 Note: At the end of year 0, the bank makes $1,000,000 of five-year loans. At the end of each year, 2 percent of the loans default. The interest rate is 12.2449 percent per year on the outstanding principal amount at the end of each year. The portfolio yields a 10 percent before-tax rate of return (fifth column). Taxable income is equal to interest income for the year less the amount of loans that defaulted during the year. The after-tax rate of return is 7 percent (last column). The effective tax rate is 30 percent. Source: Author’s calculations. 299 300 Table 9.2. Income and Cash Flow from Loan Portfolio (default at term) Present Present Before-tax value Taxable Tax After-tax value Year Principal Interest cash flow (r = .104958) income (30%) cash flow (r = .072034) 0 1,000,000 — –1,000,000 –1,000,000 — — –1,000,000 –1,000,000 1 1,000,000 122,449 122,449 110,818 122,449 36,735 85,714 79,955 2 1,000,000 122,449 122,449 100,291 122,449 36,735 85,714 74,582 3 1,000,000 122,449 122,449 90,765 122,449 36,735 85,714 69,571 4 1,000,000 122,449 122,449 82,143 122,449 36,735 85,714 64,896 5 903,921 110,684 1,014,605 615,982 26,270 7,881 1,006,724 710,997 Total — — — 0 — — — 0 Note: At the end of year 0, the bank makes $1,000,000 of five-year loans. At the end of year 5, 9.6 percent of the loans default. The interest rate is 12.2449 percent per year on the outstanding principal amount at the end of each year. The portfolio yields a 10.4958 percent before-tax rate of return (fifth column). Taxable income is equal to interest income for the year less the amount of loans that defaulted during the year. The after-tax rate of re- turn is 7.2034 percent (last column). The effective tax rate is 31.4 percent. Source: Author’s calculations. CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 301 losses in year 1 are a cost of earning income not only in year 1 but also over the life of the portfolio. Thus these losses should be spread over the five-year period and not deducted solely in the first year. These examples are admittedly artificial, but they allow one to determine the effective tax rate on a portfolio of loans, given the pat- tern of loan-losses. One conclusion from these examples is that if the loans were expected to go bad at a constant rate per year, the charge-off method would, in theory, produce the appropriate match- ing of income and expenses. There is an important proviso, however. These examples assume that all past due loans are ultimately worth- less and there is no recognition lag between when past due payments occur and when the charge-off is allowed. The reserve method, by accelerating the tax deduction compared to the charge-off method, may provide a reasonable solution for the recognition lag and take into account that all past due loans do not become worthless. If banks are required to go to court or await the bankruptcy or death of the debtor before writing off debts that clearly are not re- coverable, then the charge-off method is overly restrictive. Depend- ing on the bankruptcy laws and court practices, the charge-off de- lays in some countries could go on indefinitely. At minimum, bad debts should be charged off for tax purposes in the year they are classified as worthless for regulatory purposes. The Case for Conformity A major advantage of having a high degree of conformity between loan-loss provisioning for financial and tax purposes is that the tax authority would not have to assess the reasonableness of the provi- sion. Instead, the tax authority could rely on the bank regulatory au- thority to “audit” the loan-loss provision. This would provide banks with greater certainty by reducing disputes between the banks and the tax authority. However, a high degree of conformity does not necessarily require full conformity. Administrative simplification would still be obtained if only specific provisions are deductible for tax purposes or if the specific provision is given a “haircut,” as in Serbia. Conformity between financial and tax accounting for loan-losses also would ensure that the tax system does not provide a disincen- tive for banks to adequately provide for loan-losses.21 Each dollar added to the reserve for financial or regulatory purposes would re- duce taxable profits by a dollar and provide a current tax benefit, so long as the bank has positive taxable income. Once a bank is in a tax loss position, additional tax deductions from additional re- 302 EMIL M. SUNLEY serving only increase the loss carry forward, and this may provide no tax benefit if the bank ultimately fails. The tax treatment of loan-loss provisioning also affects capital adequacy ratios. If a bank sets a specific provision of 100, tier I cap- ital and profits are reduced by 100, absent taxation. As a result, stockholders’ equity is also reduced by 100. However, if the provi- sion is deductible for tax purposes, after-tax profits, stockholders’ equity, and tier I capital are all reduced by (1– t)100, where t is the corporate tax rate. Thus a tax deduction for the loan-loss provision cushions the effect of the provision on the amount of the bank’s tier I capital. All other things equal, this should reduce the disincentive for banks, already constrained by the need to meet capital adequacy ratios, to adequately provide for loan-losses. If there is no current tax deduction for the loan-loss provision (for example, the country is on the charge-off method for tax purposes), the effect on tier I capital may still be cushioned if a deferred tax asset22 is recognized for financial accounting, and this asset is counted toward tier I capital. Admittedly, this asset is a non-earning asset, and therefore not as valuable as an asset that can generate income. The Case against Full Conformity There are several considerations that can be used to argue against full conformity between financial and tax accounting in this context. Financial vs. Tax Accounting Both financial and tax accounting are based upon the premise of measuring income, but their goals are somewhat different. Financial accounting—and prudential rules, in particular—is based on conser- vatism: that is, to delay recognition of income as long as possible and to anticipate expenses and losses as soon as possible. Notwithstand- ing the bankruptcy of Enron, such a system is designed to ensure that the profits and the net worth of the company are not overstated. The objective of tax accounting is the opposite: that is, to ensure that income is not understated. Therefore, income is taxed as soon as it belongs to the taxpayer. Thus it would be normal to tax pre- payments of rent as soon as they are received, regardless of the fact that they relate to a period beyond the tax year. Similarly, tax ac- counting defers deductions until it is clear that the liability will ac- tually be incurred. Various reserves allowed for financial accounting generally are not allowed for tax accounting.23 CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 303 General Reserves From the tax accounting point of view, the general reserve for loan- losses looks quite suspect. Although the general reserve is for “pres- ent” or “latent” losses, it is, by its nature, “speculative,” and in- volves a great deal of judgment by bank managers (Laurin and Majnoni 2003). The fact that the general reserve is often included as part of tier II capital suggests that this reserve may be for future losses. At the very least, it is for current losses that have not yet been identified. Consistent with tax accounting, generally, a liability should not be recognized for tax purposes until it is certain (that is, it can be identified) and can be reasonably estimated. Moreover, rec- ognizing general provisions for tax purposes can be very expensive in terms of foregone tax revenue, as many banks would have general provisions that are larger than their specific provisions.24 It is not surprising, therefore, that tax authorities are wary of allowing banks to deduct general reserves for unidentified, but arguably “present” losses. Specific Reserves There is also a suspicion, at least among tax officials, that the spe- cific reserves are set on the high side to be conservative, to protect the capital of the banks, and to reduce overly risky behavior.25 If the bank regulator fears that banks may not properly classify loans, the required provisions for each category may be set at a higher rate in order to partially offset the tendency to misclassify loans. Moreover, required reserves for regulatory purposes are minimums, as the bank regulatory authority wants to make certain that banks do not understate their losses.26 The tax authority, however, wants to en- sure that banks do not overstate reserves. Instead of being mini- mums, the regulatory reserves should be maximums, if allowed for tax purposes. To counteract the bias that regulatory reserves are too conserva- tive, it may be appropriate to reduce the provision level for tax pur- poses—say to 80 or 90 percent of the regulatory rate. If the loans actually do go bad, they will ultimately be completely written off; this is therefore a timing issue. Tax Treatment of Borrowers vs. Bank Lenders From the tax policy perspective, there is an additional concern re- garding a reserve method for accounting for loan-losses: namely, the asymmetric treatment of banks and their business borrowers. In gen- 304 EMIL M. SUNLEY eral, when interest is an expense of the borrower, it is at the same time income of the lender, just as when a business pays wages and deducts the expense, the worker is taxable on the wage income. Similarly, when a bank has a loan-loss, the borrower has income— forgiveness-of-indebtedness income. However, this income is recog- nized only when the debt is forgiven: that is, written off. If a bank recognizes the loan-loss through provisioning, it may be able to claim the loss in an earlier tax year than when the borrower must recognize income. This timing difference between the expense of the bank and the income of the borrower works to the disadvantage of the fiscal authorities. This is probably more of a theoretical problem than a real one, however. When borrowers are unable to make the debt payments, they likely are in economic difficulty and have tax losses. In this situation, it may not matter in which year the forgive- ness-of-indebtedness income is recognized for tax purposes, as no tax liability will accrue in any event. Transition Rules If a country is going to switch from the charge-off to the reserve method, one needs to be concerned about how the transition is han- dled. Suppose, for example, the effective date of the switch is Janu- ary 1, 2003. Suppose further that for a particular bank, the beginning- of-the-year reserve is $100, and the required end-of-year reserve is $120. If the country had always been on the reserve method, the tax deduction for 2003 would be $20, ignoring any write-offs and re- coveries. However, the beginning-of-the-year reserve of $100 has never been allowed as a tax deduction. How should this $100 be treated for tax purposes? One possibility would be to allow an additional deduction of $100 in 2003 on the grounds that $100 would have been an ex- pense in earlier years if the country had been on the reserve method. This could be quite expensive in terms of foregone tax revenues. Also, a special rule (such as a longer loss carryover period) may be needed to cover the situation where the one-time deduction creates a large loss carryover that likely will not be used during the carry- over period. An alternative and the preferred approach for handling the tran- sition would be to treat the switch from the charge-off to the reserve method as a change in the method of accounting. The effect of the change—an additional deduction of $100—would be spread over three to five years.27 For countries that are concerned about the rev- enue cost of switching to the reserve method, the allowable deduc- CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 305 tion could be further limited in the first year to, say, 25 percent of taxable income before the reserve deduction. This percentage would increase to 100 percent over several years. Any amount disallowed in one year as a result of this limit would be carried over to subse- quent years until allowed. Conclusions and Recommendations There is no standard international practice as to the treatment of bank loan-losses for tax purposes. Some countries use the charge- off method; other countries allow provisions for loan-losses along the lines of the provisions required for regulatory accounting. Only rarely do countries have full conformity between book and tax pro- visions for loan-losses. It is much more common for countries to disallow general provisions for performing loans. The widespread practice of under-provisioning in many develop- ing countries, and the importance of ensuring that tax rules do not unduly discourage the establishment of needed provisions, means that there is much to be said for ensuring an adequate tax de- ductibility for loan-loss provisions. Some conclude that all provi- sions should be tax-deductible. But on the assumption that banks are following prudent provisioning practice, this chapter takes the position that the tax rules for loan-losses should be closely tied to (but not necessarily the same as) the prudential rules for provision- ing. However, no tax deduction should be allowed for a general pro- vision for performing loans. Also, it may be appropriate to reduce the specific provisioning allowed for tax purposes to 80 or 90 per- cent of the regulatory rate, in order to reflect the real income of banks as accurately as possible. When countries using the charge-off method for tax purposes switch to a reserve method, this change should be treated as a change in the method of accounting and the effect of the change spread over three to five years. Appendix A9.1. Soviet-style Accounting Whether loan-losses should be an allowable expense was an issue under Soviet-style accounting, which was developed to provide fiscal data to the state, rather than information for shareholders, creditors, and managers. In practice, an enterprise was required to offset all its expenses with revenues it earned from the sale of its products and services (Ash and Strittmatter 1992). There were no private banks in the Soviet economy before a transition period just before the break-up of the Soviet Union. All enterprise loans (or 306 EMIL M. SUNLEY cash transfers) came from the State Bank (Gosbank) after the Ministry of Fi- nance had approved the budget, which contained a breakdown by individ- ual enterprise. State-owned companies did not go bankrupt, in part, because the plan for each enterprise was set to ensure that revenues (including trans- fers from the State Bank) offset expenses. Once private banks appeared, loans were guaranteed by the state. If a business defaulted on its loan, the guarantee would be called. Bank bad debts generally were unknown. The income taxes first adopted by various states of the former Soviet Union reflected Soviet-style accounting: revenue offsetting allowable ex- penses. Thus private banks were not allowed to claim a tax deduction for loan-losses, which were not considered an allowable expense. Instead, these costs were accounted for as direct charges against the bank’s equity. Western-style accounting for the earnings of an enterprise is premised on a matching of expenses to the revenues to which they relate. With respect to the accounting treatment of bad debts, it is recognized that when a bank originates a portfolio of loans, it is not known which loans in the portfolio will default, but some inevitably will. A loan-loss is considered a cost that a bank incurs in order to earn income on performing loans. Loan-losses therefore should be a deductible expense. The various states of the former Soviet Union have all adopted income tax laws that now recognize bad debts as costs of earning income. For these countries, the major policy issue—and the primary focus of this chapter— is when and how the cost of loan-losses should be recognized for tax purposes. Appendix A9.2. Insurance Reserves and Investment Income The financial and tax treatment of insurance companies has been a con- tentious issue in a number of developing and transition countries. If the reg- ular corporate income tax is going to apply, two major policy issues need to be addressed: the appropriate reserves to allow for tax purposes, and the allocation of investment income.28 Insurance companies operating in a country are usually subject to regu- lation by an insurance authority that requires the maintenance of reserves against the contingent liabilities that are insured. These contingent liabili- ties may require expenditures to be made in the current year or in future years. In determining the appropriate reserve, the regulatory authority should require discounting of expected future expenditures, using a pre-tax discount rate, in order to reflect the time value of money. The reserves appear to be an ordinary and necessary cost of doing in- surance business, and additions to these required reserves are commonly al- lowed as deductions in determining taxable income. These reserves involve CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 307 quite complex actuarial computations, which cannot be effectively audited by the tax authority in most countries. They may exaggerate the contingent liabilities—for example, assuming high mortality or accident rates or low discount rates—reflecting the concern of the insurance regulators to give policyholders confidence in the company’s solvency. If the reserves are too conservative, then allowing a tax deduction for the full amount of the addition to reserves results in an understatement of the company’s income. One possible counteraction would be to give the re- serves required for regulatory purposes a “haircut” (or a reduction in their effective valuation) before allowing them as a deductible expense for tax purposes. In developing and transition countries, insurance policies are often writ- ten by international companies. The accumulated reserves may be invested offshore, as the domestic capital market is quite limited. The associated in- vestment income, however, would normally be foreign-source income and thus not taxable in the country in which the policies are written. This mis- matching of income and expenses can lead to insurance companies report- ing little or no taxable income in the country where the policies are written. A possible counteraction would be to require that the investment income of an insurance company operating across international borders be allocated based upon liabilities insured in each country. Because of the complexity of determining income of an insurance com- pany and the appropriateness of its reserves, some authorities recommend a gross premium tax in lieu of imposing the regular income tax on insur- ance companies (Hussey and Lubick 1996). Non-insurance activities such as financial leasing would be treated as conducted in a separate entity sub- ject to the regular corporate income tax. Notes 1. That bank loan-losses should be recognized as an expense for both financial and tax purposes is not controversial, at least under Western market-oriented accounting. However, under Soviet-style accounting, banks were not allowed to claim a tax deduction for loan-losses. Soviet- style financial and tax accounting and whether loan-losses should be al- lowed as an expense are discussed further in appendix A9.1. 2. In this chapter, the terms “reserves” and “provisions” will be used interchangeably. 3. Reserves of an insurance company recognize current and future ex- penditures that are necessary to settle current obligations. According to IAS 37 (2000), where the effect of the time value of money is material, insur- ance reserves (or provisions) should be discounted using a pre-tax discount rate. Appendix A9.2 elaborates on insurance reserves. 308 EMIL M. SUNLEY 4. By excluding the word “future,” the 1991 amendments to the Basel Accord aligned regulatory accounting with financial accounting. See Beat- tie and others (1995, p. 20). 5. The definition of specific and general provisions and their uses vary across countries (Laurin and Majnoni 2003). 6. The term “non-performing loans” most commonly refers to loans on which the payments are past due by 90 days or more, but practices vary across countries. Non-performing loans are usually subject to a specific re- serve. Loans that are past due for only a short period of time are not con- sidered to be non-performing (Cortavarria and others 2000). 7. It may well be that next year the oil sector will go sour and bank loans to the oil sector will become impaired. These losses are a future ex- pense and they should be recognized in future periods. 8. Based on the classification of loans at the end of the current year. 9. Based on the classification of loans at the end of the prior year. 10. In some countries, worthless loans are charged directly against in- come (and not debited against the reserve). When this approach is followed, the necessary addition to the reserve to bring the reserve up to the end-of- year balance would be correspondingly reduced. 11. If 75 percent of the loan had been reserved at the end of the prior year, then charging the loan against the reserve will give a tax deduction in the current year for 25 percent of the loan. The tax deduction would be equal to the reserve at the end of the year minus the reserve at the begin- ning of the year plus the bad debt written off (0 –75 + 100). 12. Under the International Accounting Standards (IAS), the accrual method is a fundamental accounting policy. Transactions and events are recognized when they occur and they are recorded in the accounting records and reported in the financial statements of the periods to which they relate. Expenses are recognized in the income statement on the basis of a direct association between the costs incurred and the earning of specific items of income (matching). Under the accrual method, the timing of the ac- tual payment or transfer of consideration is not relevant. 13. Also, differences in loan-loss provisioning and the capital adequacy standards across countries will impact the competitiveness of banks (Beat- tie and others 1995). 14. For an international comparison of the tax treatment of loan-losses in developed countries, see Escolano (1997) and Beattie and others (1995). 15. Although the United States requires large banks to use the charge- off method, it allows banks a partial write-off of debts, mitigating the harsh effects of the charge-off method (Dziobek 1996). 16. Calculation of effective tax rates on theoretical portfolios of loans buttressed the case for the United States switching from the reserve method to the charge-off method (Joint Committee on Taxation 1985). CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 309 17. (100)(1 – .02)(.122449) – 2 = (100)(.10). 18. The economic loss in the value of the portfolio would be measured by comparing the present value of the future before-tax cash flows, dis- counted at 10 percent, at the end of year, compared to the value of the end of the prior year. 19. Another key assumption is that the loan-loss is charged off immediately. 20. By analogy, when oil companies drill dry holes in seeking oil re- serves, the cost of these unsuccessful wells could be viewed as capital costs incurred in order to find producing wells. The costs of the dry holes would be capitalized and written off over the life of the producing wells. Most countries, however, permit dry holes to be expensed for tax purposes. 21. This may be particularly important in times of fiscal stress when banks have high rates of non-performing loans and defaults. It is somewhat ironic that during periods of fiscal stress, countries, mainly in Latin Amer- ica, have often adopted bank debit taxes, which tend to encourage disin- termediation (Coelho, Ebrill, and Summers 2001). 22. Some countries, including the United States, determine the income tax expense for financial accounting purposes based on the reported prof- its for financial purposes adjusted for permanent differences between fi- nancial and taxable income (such as tax credits and exempt income). If fi- nancial income is greater than taxable income due to timing differences, the tax on this income, which is not currently payable, is considered a deferred tax asset on the balance sheet. Thus even though the provision does not re- sult in a current tax benefit, after-tax financial profits and stockholders’ eq- uity are reduced by only (1– t) times the amount of the provision. To illus- trate, if financial profits are $100 and the tax rate is 20 percent, the tax expense for financial purposes would be $20. If taxable income is $150 due to greater provisioning allowed for financial purposes than for tax pur- poses, the $10 tax liability payable now on the $50 difference would be treated for financial purposes as a deferred tax credit. Thus the additional provision of $50, not recognized for tax purposes, reduces the net worth of the firm by only $40, as the $10 future tax savings generated by the provi- sioning is considered an asset on the balance sheet. 23. The two exceptions would be loan-loss provisions for banks—the focus of this chapter—and the required reserves of insurance companies. In both cases, the tax authority can rely on the regulatory authority for banks and insurance companies to determine the reasonableness of the reserve. 24. For example, if a bank had classified loans equal to 3 percent of total loans and had set specific reserves averaging 40 percent of the classi- fied loans, the specific provision would be only 1.2 percent of total loans— less than a general provision of 1.25 percent applying to the other 97 per- cent of loans. 310 EMIL M. SUNLEY 25. Even if this is so, bank managers (with the concurrence of regula- tors) may under-provide for reserves at times of distress, whether or not ad- ditions to reserves are tax-deductible. Tax policy probably cannot correct this. 26. The regulatory authority is also concerned that banks do not have “hidden reserves” in order to smooth income. Any reserve for future losses should be accounted for as an allocation of retained earnings: that is, below the line. 27. Spreading the effect of a change in the method of accounting is quite common practice when a business switches from cash to accrual account- ing or changes its method of accounting for inventories. When the United States went from a reserve method to the charge-off method for large banks, the existing reserve at the end of 1986 was brought back into in- come over four years. 28. There are also important issues related to how policyholders and beneficiaries should be treated. Generally, businesses should deduct premi- ums accrued or paid to insure business risks. Any benefits paid under the insurance policies to businesses would be included in taxable income. If the benefits are paid to employees or other persons, they generally are not tax- able. However, if the benefit payment replaces income that otherwise would have been taxed (such as lost earnings), the benefit probably should be tax- able. In the case of life insurance provided by an employer, it would be un- seemly to tax life insurance proceeds but it would be appropriate to tax em- ployees on the value of company-provided insurance. In the United States, for example, employees are required to include in income the cost of com- pany-provided group life insurance in excess of $50,000 of such insurance. References Ash, Ehiel, and Robert Strittmatter. 1992. Accounting in the Soviet Union. Westport, Conn.: Greenwood. Basel Committee on Banking Supervision. 1988. International Convergence of Capital Measurement and Capital Standards. Basel, Switzerland: Bank for International Settlements. ———. 1999. Sound Practices for Loan Accounting and Disclosure. Basel: Bank for International Settlements. Bassett, William F., and Egon Zakrajs ek. 2001. “Profits and Balance Sheet Developments at U.S. Commercial Banks in 2000.” Federal Reserve Bul- letin 87 (June): 367–93. Beattie, Vivien A., Peter D. Casson, Richard S. Dale, George W. McKenzie, Charles M. S. Sutcliffe, and Michael J. Turner. 1995. Banks and Bad CORPORATE INCOME TAX TREATMENT OF LOAN-LOSS RESERVES 311 Debts: Accounting for Loan-Losses in International Banking. New York: John Wiley & Sons. Coelho, Isaias, Liam Ebrill, and Victoria Summers. 2001. “Bank Debit Taxes in Latin America: An Analysis of Recent Trends.” IMF Working Paper WP/01/67. International Monetary Fund, Washington, D.C. Cortavarria, Luis, Claudia Dziobek, Akihiro Kanaya, and Inwon Song. 2000. “Loan Review, Provisioning, and Macroeconomic Linkages.” IMF Working Paper WP/00/195. International Monetary Fund, Washington, D.C. Dziobek, Claudia. 1996. “Regulatory and Tax Treatment of Loan-Loss Provisions.” IMF Paper on Policy Analysis and Assessment, 96/6. Inter- national Monetary Fund, Washington, D.C. Escolano, Julio. 1997. “Tax Treatment of Loan-Losses of Banks.” In William E. Alexander, Jeffrey M. Davis, Liam P. Ebrill, and Carl-Johan Lindgren, eds., Systemic Bank Restructuring and Macroeconomic Pol- icy. Washington, D.C.: International Monetary Fund. Hussey, Ward M., and Donald C. Lubick. 1996. Basic World Tax Code and Commentary. Arlington, Virginia: Tax Analysts. International Accounting Standards Committee. 2002. “IAS 30: Disclo- sures in the Financial Statements of Banks and Similar Financial Institu- tions.” In International Accounting Standards 2002. London: Interna- tional Accounting Standards Board. ———. 2002. “IAS 37: Provision, Contingent Liabilities and Contingent Assets.” In International Accounting Standards 2002. London: Interna- tional Accounting Standards Board. Joint Committee on Taxation. 1985. Tax Reform Proposals: Taxation of Financial Institutions. Report JCS 38-85, U.S. Congress, Washington, D.C. Laurin, Alain, and Giovanni Majnoni, eds. 2003. Bank Loan Classification and Provisioning Practices in Selected Developed and Emerging Coun- tries. Washington, D.C.: World Bank. Samuelson, Paul A. 1964. “Tax Deductibility of Economic Depreciation to Insure Invariant Valuations.” Journal of Political Economy 72 (6): 604–06. 10 Bank Debit Taxes: Yield Versus Disintermediation Andrei Kirilenko and Victoria Summers In the past 15 years, a number of countries, mostly in Latin Amer- ica, have imposed taxes on banking transactions. These taxes are usually levied on withdrawals from or other debits to bank ac- counts, including check clearance, cash withdrawals, and payments of loan installments. Since 1988, bank debit taxes have been introduced in Argentina, Brazil, Colombia, Ecuador, Peru, and Venezuela. With the exception of Brazil, bank debit taxes were introduced at the time of crisis with the objective of quickly generating a burst of revenue. These taxes have the support of many policymakers and virtually no organized popular opposition. Collection and administration costs of these taxes are minimal. In addition, the government gains an immediate and continuous revenue stream, since the taxes are collected from transactions in real time. However, since bank debit taxes are levied on intermediated fi- nancial transactions, their imposition is likely to result in financial disintermediation. Broadly speaking, disintermediation is the with- drawal of funds from financial intermediaries, with the payments being made in some other way (for example, in cash, by barter, or through accounts not subject to tax). Disintermediation results not only in the reduction of the tax base, but also in a possible misallo- cation of financial resources. 313 314 ANDREI KIRILENKO AND VICTORIA SUMMERS This chapter presents a first formal attempt to estimate the scale of disintermediation resulting from the introduction of a bank debit tax. Drawing on a model of financial intermediation by Kirilenko and Summers (2001), this study derives a relationship between the disintermediation resulting from introduction of a bank debit tax and the deadweight welfare loss attributable to the tax. Using data from four Latin American countries, this study esti- mates the scale of disintermediation attributable to the tax and de- duces the welfare loss. The calculations rely on two key assumptions: namely that the entire burden of the bank debit tax falls on bank borrowers, and that the impact of the tax on financial intermedia- tion does not yet arise in the first full month after its introduction. The study finds that bank debit taxes have coincided with signif- icant welfare losses, especially at higher rates. Expressed as a per- centage of revenue, deadweight losses following the introduction of the tax reached up to 30 percent in Venezuela, up to 35 percent in Colombia, and up to 45 percent in Ecuador. However, the study does not find significant deadweight losses in Brazil. Introduction of a bank debit tax resulted in disintermediation of up to 28 percent in Venezuela, up to 41 percent in Colombia, and up to 47 percent in Ecuador. This means that for every dollar in rev- enue raised by a bank debit tax, there was a loss of financial inter- mediation equal to 28 cents in Venezuela, 41 cents in Colombia, and 47 cents in Ecuador. Again, no significant disintermediation was found in Brazil. Both the deadweight losses and disintermedia- tion effects cumulate as the taxes remain in place. While bank debit taxes can be used as a quick and effective way to generate revenue, pending the implementation of improvements in the arrangements for collecting more efficient taxes (Tanzi 2000), this study confirms that, as time passes, and especially at higher rates of tax, debit taxes may lead to significant welfare losses and financial disintermediation. The rest of the chapter is organized as follows. After a brief de- scription of bank debit taxes and a discussion of their productivity, the chapter points out the relationship between deadweight loss and disintermediation. It then describes the data and estimation results and presents selected descriptive evidence. The chapter concludes that at higher rates or over an extended period of time, bank debit taxes may lead to significant financial disintermediation. Bank Debit Taxes in Latin America Bank debit taxes are currently in effect in Argentina, Brazil, Colombia, and Venezuela, and were in place previously in Ecuador BANK DEBIT TAXES: YIELD VERSUS DISINTERMEDIATION 315 and Peru. With the exception of Brazil, bank debit taxes were in- troduced at a time of, and in response to, general economic crisis, as an emergency means of raising government revenue. In all cases the tax was explicitly introduced on a temporary basis, though in some cases it was then extended. • In Argentina when a bank debit tax was introduced as a tem- porary measure in 1988, tax revenues were declining dramatically because of hyperinflation, increased evasion, and depressed eco- nomic activity. The tax was reintroduced under similar circum- stances in April 2001. • In Peru, when the Impuesto a los Débitos Bancarios y Fi- nancieros was introduced as a temporary and extraordinary rev- enue measure in 1989, the nation was immersed in a deep economic crisis and central government revenues had fallen from 14.9 percent of GDP in 1985 to 6.1 percent of GDP in 1989. • In Colombia, when a temporary Impuesto a las Transacciones Financieras (known as the “dos por mil”) was adopted in November 1998, the health of the financial sector had already deteriorated markedly and the government had declared an economic emergency.1 • In Ecuador, the tax was introduced in 1999, at a time when the economy plunged into a major economic and financial crisis. • In Venezuela, a temporary bank debit tax was collected from May to December 1994 and from May 1999 to May 2000. The tax was reintroduced in Venezuela for 12 months beginning March 2002. There is a considerable diversity in the design of these taxes from country to country.2 Transactions subject to or exempt from the tax as well as tax rates differ significantly. Tax rates have ranged be- tween 0.2 and 2.0 percent.3 In all cases, the rates have varied both across countries and across time, without any discernible trend. The taxes have been imposed on debits to (withdrawals from) checking, savings, and term accounts in banks and other financial institutions, and on loan withdrawals (in Brazil, Colombia, and Venezuela). In Colombia, the tax is also imposed on credits of bank interest to accounts and on repurchase agreements (repos). In Ecua- dor, the base of the tax was somewhat different; the tax was im- posed not only on withdrawals but also on credits to checking, sav- ings, term, loan, and other accounts at financial institutions, as well as on remittances abroad and on payments abroad by exporters and importers. Thus both deposits to and withdrawals from the same accounts were subject to the tax. The taxation of both debits and credits is also a feature of the tax introduced in Argentina in 2001. Most countries provided exemptions for transactions by certain types of institutions (for example, government agencies) and some 316 ANDREI KIRILENKO AND VICTORIA SUMMERS specific transactions, including repos and transactions with the cen- tral bank. In Argentina (through 1992 and between April and De- cember 2001) and in Ecuador, a portion of the bank debit tax has been creditable against the income tax or the VAT. Tighter anti-avoidance measures have been introduced in the more recent taxes. The measures included restrictions on the use of cash for settlements, prohibition of multiple endorsements of checks, and application of the tax to all but the first endorsement upon final settlement. Revenue Productivity As can be seen from table 10.1, the short-term revenue performance of transactions taxes, particularly in Brazil, Colombia, and Ecuador, has been quite strong. In Brazil and Colombia, the taxes have pro- duced revenues in the range of 0.60 to 1.45 percent of GDP for ad valorem tax rates in the range of 0.20 to 0.37 percent. The revenue performance of the tax in Ecuador was exceptionally strong in its first year, reflecting its application to a broader base including both debits and credits, though part of the gross revenues were creditable against other taxes. The taxes imposed in Argentina and Peru in the late 1980s and early 1990s were significantly less productive, as gauged by the ratio of revenues as a percent of GDP to the average statutory rate. Overall, the more recent taxes have been more productive than those introduced a decade ago. In Brazil, a high revenue yield has been sustained over several years. However, in Colombia and Ecua- dor, monthly real revenues from the tax were on a declining trend. In Venezuela, revenues held up through the end of 1999 from the tax’s introduction earlier in that year, but declined rapidly in 2000. It should also be noted that revenue productivity appears to decline with higher tax rates. For example, while the tax base in Ecuador was much broader than in the other cases, revenue productivity was considerably lower in Ecuador than in Brazil and in Colombia, where the tax rates are lower. Calculating Disintermediation and the Deadweight Welfare Loss Kirilenko and Summers (2001) formulate a model that focuses on the intermediation role of banks and postulates that the debit tax with the rate τ is perceived by the depositor as, in effect, a reduction BANK DEBIT TAXES: YIELD VERSUS DISINTERMEDIATION 317 Table 10.1. Bank Debit Taxes, Selected Latin American Countries, 1989–2002 Country and Effective Gross year introduced tax rate revenue a Productivity b Argentina 1989 0.70 0.66 0.94 1990 0.30 0.30 0.99 1991 1.05c 0.91 0.86 1992 1.20 0.92 0.77 2001 0.99c 1.08 1.10 Brazil 1994 0.25 1.06 4.24 1997 0.20 0.80 4.00 1998 0.20 0.89 4.44 1999 0.24c 0.83 3.46 2000 0.33c 1.34 4.02 2001 0.37c 1.45 3.97 Colombia 1999 0.20 0.69 3.45 2000 0.20 0.60 3.00 2001 0.30 0.75 2.50 Ecuador 1999 2.00 3.38d 1.69 2000 1.60 2.37d 1.48 Peru 1990 1.42c 0.69 0.49 1991 0.81c 0.57 0.70 Venezuela 1994 0.75 1.30 1.73e 1999–2000 0.50 1.12 2.24 2002 0.85 1.57 1.85 aGross revenue in percent of GDP. bGross revenue in percent of GDP divided by average statutory rate. cAverage of rates, adjusted for the period tax was in effect. dTax was levied on both debit and credit transactions. eAdjusted for the period tax was in effect. Source: National authorities and authors’ estimates. in the yield of bank deposits from rb, say, to rb/(1 – π).4 The model predicts that after the introduction of the tax, the equilibrium amount of funds intermediated by the banks declines by a fraction β of their initial value N, where β is a function of the tax rate τ. While disintermediation is an intuitive economic concept, it is quite difficult to estimate. At the same time, since Harberger (1964), there exist standard techniques to estimate a deadweight welfare 318 ANDREI KIRILENKO AND VICTORIA SUMMERS loss due to taxation.5 The deadweight welfare loss L caused by the tax is the area of the “Harberger triangle” that can be calculated as one half of the tax times the change in the tax base βN: 2L = τβ(τ)N This can be compared with tax revenues T, which are equal to the tax rate times the new tax base: T = τ[1 – β(τ)]N The deadweight loss as a fraction of revenues is thus L 1  β(τ)  l= =  , T 2  1 − β(τ)  or equivalently, β = 2 l/(1 + 2l). A zero deadweight loss implies that the bank debit tax has no im- pact on financial intermediation. Alternatively, infinite deadweight loss corresponds to complete disintermediation. Data and Estimation This study uses monthly series of bank debit tax revenues for four countries: Brazil (June 1999–December 2001); Colombia (January 1999–December 2001); Ecuador (January 1999–December 2000); and Venezuela (May 1999–May 2000). The revenue data comes from the authorities of these countries. Indices of bank debit tax revenues were constructed by express- ing the real value of each month’s revenue as a percentage of the rev- enue flow in the first full month of the operation of the tax. The real value is obtained by deflating each month’s nominal revenues by the corresponding month’s consumer price index (CPI) drawn from In- ternational Financial Statistics, a monthly publication of the Inter- national Monetary Fund. The base month for the indices is January 1999 for Colombia, January 1999 for Ecuador, and June 1999 for Venezuela. For Brazil, two base dates are used: July 1999 and May 2000, thereby treating the sizable change in tax rate from the latter month as if it were a new tax.6 The revenue indices are then divided by the statutory tax rate to obtain a measure of trends in the revenue productivity or yield of the tax. Thus while the revenue index for Colombia in December 2000 is 85, jumping to 120 in January 2001, the fact that the tax BANK DEBIT TAXES: YIELD VERSUS DISINTERMEDIATION 319 rate increased from 0.2 percent to 0.3 percent between the same two months means that the productivity index declined from 85 to 80 (=120 ϫ 0.2/0.3). In order to calculate the deadweight loss, one needs an estimate of the revenue that would have resulted had there been no disinter- mediation. The model assumes that during the first month after the introduction of the tax, there is minimal change in the behavior of borrowers and depositors. This assumption allows one to use actual bank debit tax revenues collected during the first full month fol- lowing the introduction of the tax as an estimate of τN.7 Estimates of deadweight loss l for each country and each month after the first full month for which data are available are expressed as a percentage of tax revenues and shown in figure 10.1 as three- month moving averages. Deadweight losses following the introduc- tion of the tax reached up to 30 percent in Venezuela, up to 35 per- cent in Colombia, and up to 45 percent in Ecuador. The study does not find significant deadweight losses in Brazil. However, because the tax was already being collected in Brazil be- tween February 1997 and February 1999, it is likely that most of the behavioral changes in response to the tax had already taken place before July 1999. As a result, the method used in this model, which assumes no behavioral response before July 1999, would likely result in an underestimate of the impact of the tax. In con- trast, Albuquerque (2002) estimates the deadweight loss in Brazil to be 21.7 percent of the net tax revenue in 2000. Using the relationship between the deadweight loss and disinter- mediation derived above, the estimates for deadweight losses as a fraction of revenues, l, can be translated into measures of disinter- mediation β for each country and month, also expressed as a per- centage of tax revenues and shown in figure 10.2 as three-month moving averages. Introduction of a bank debit tax resulted in disin- termediation of up to 28 percent in Venezuela, up to 41 percent in Colombia, and up to 47 percent in Ecuador. Again, the study does not find significant disintermediation effects in Brazil. It does find that both the deadweight losses and disintermediation effects in- crease as the taxes remain in place. Descriptive Evidence Anecdotal evidence suggests that bank debit taxes are distortionary and have contributed to significant financial disintermediation. First, following the introduction of the tax, individuals and busi- nesses substitute away from bank-intermediated transactions into Figure 10.1. Deadweight Loss from Bank Debit Taxes, Selected Latin American Countries, 1999–2001 320 (3 month moving average) (3 month moving average) 14 40 12 35 10 30 8 BRAZIL 6 25 4 20 2 15 0 10 COLOMBIA –2 –4 5 –6 0 October March August January June November April September February July December November October 1999 2000 2000 2001 2001 2001 1999 1999 2000 2000 2000 2001 2001 (3 month moving average) (3 month moving average) 50 35 45 30 40 25 35 VENEZUELA ECUADOR 20 30 25 15 20 10 15 5 10 5 0 0 –5 April July October January April July October September November January March May 1999 1999 1999 2000 2000 2000 2000 1999 1999 2000 2000 2000 Source: Authors’ calculations. Figure 10.2. Disintermediation Caused by Bank Debit Taxes, Selected Latin American Countries, 1999–2001 (3 month moving average) (3 month moving average) 20 45 15 40 BRAZIL 35 10 30 5 25 0 20 15 COLOMBIA –5 10 –10 5 –15 0 October March August January June November April September February July December May October 1999 2000 2000 2001 2001 2001 1999 1999 2000 2000 2000 2001 2001 (3 month moving average) (3 month moving average) 50 35 45 30 40 25 35 ECUADOR 20 VENEZUELA 30 15 25 10 20 5 15 0 10 5 –5 0 –10 April July October January April July October September November January March May 1999 1999 1999 2000 2000 2000 2000 1999 1999 2000 2000 2000 321 Source: Authors’ calculations. 322 ANDREI KIRILENKO AND VICTORIA SUMMERS cash. In Brazil, Colombia, and Ecuador where taxes were in effect from 1998 to 2000, the ratio of currency outside banks to narrow money increased by between 15 and 150 percent.8 Second, in order to avoid the tax, individuals and enterprises conduct a greater proportion of their bank transactions offshore. For example, in order to avoid paying bank debit taxes, Argentini- ans opened bank accounts in Uruguay and Ecuadorians used Aquas Verdes, a town on the border with Peru. Third, economic agents create new instruments and practices to minimize the impact of the tax. Multiple endorsement of checks is among the most common practices that emerge following the intro- duction of the bank debit tax. For example, in Colombia, the volume of cleared checks was cut in half from an average of about 60,000 per month to about 30,000 per month within days after the intro- duction of the tax. Another popular practice is to set up separate clearing and settlement bank accounts so that intraday payments be- tween customers can be aggregated and debited on a net basis at the end of the day, with only that transaction being subject to the tax. In some cases, such as in Colombia, banks deposited the net payment into their tax-exempt accounts with the central bank, thus avoiding the tax altogether. In addition, financial institutions in Brazil offered investment and privatization funds in which an investor pays the tax only at the time of the initial transaction. Subsequent transactions done on the behalf of the investor are not taxed, because money transfers between financial institutions are tax-exempt.9 Finally, trading volume in the domestic Treasury bill, foreign ex- change, equity, and interbank money markets may fall if transac- tions in these markets are subject to the tax. For example, in Colombia, immediately after the introduction of the tax in Novem- ber 1998, the volume in interbank foreign exchange and money markets declined to about 20 percent of the average pre-tax level, while the volume in the Treasury bill market declined to about 10 percent of the average pre-tax level. In Venezuela, according to the Annual Report of the Caracas Stock Exchange, after the tax was in- troduced in 1999, trading volume on the exchange dropped by 47 percent compared to the previous year. Conclusion This chapter has focused on the incidence of taxes on banking trans- actions. In the last 15 years, a number of countries in Latin America repeatedly implemented and revoked these taxes, which are usually levied on debits to bank accounts. These taxes are currently in ef- BANK DEBIT TAXES: YIELD VERSUS DISINTERMEDIATION 323 fect in Argentina, Brazil, Colombia, and Venezuela, where they have been quite effective in generating revenue in the short run. This has been the first formal attempt to estimate deadweight losses and disintermediation following the introduction of bank debit taxes, using data from Brazil, Colombia, Ecuador, and Venezuela. The study finds that, especially at higher rates, these taxes have co- incided with significant welfare losses and financial disintermedia- tion. Deadweight losses following the introduction of the tax reached up to 30 percent in Venezuela, up to 35 percent in Colombia, and up to 45 percent in Ecuador. The study also derives a relationship between deadweight welfare losses and disintermediation. Using this relationship, it shows that the introduction of a bank debit tax resulted in disintermediation of up to 28 percent in Venezuela, up to 41 percent in Colombia, and up to 47 percent in Ecuador. The analysis does not find significant disintermediation or deadweight losses in Brazil, for which the data series is not really long enough to allow effective use of the study’s method. The study also finds that both the deadweight losses and disintermediation effects cumulate over time as the taxes remain in place. These findings support a view that at low rates and for a limited time, bank debit taxes can be used as a quick and effective way to generate revenue, while the implementation of more traditional taxes is being improved. However, at higher rates or over an extended period of time, the taxes may lead to significant financial disintermediation. Notes 1. In December 2000, the tax was made permanent and its rate was in- creased to 0.3 percent. 2. See Coelho, Ebrill, and Summers (2001) for a comprehensive descrip- tion of bank debt taxes in Latin America. 3. The 2 percent tax rate was applicable in Peru from April to Septem- ber 1990. 4. This insight comes from Caminal (1997). 5. See Hines (1999) for a discussion of Harberger’s contribution to the estimation of deadweight welfare losses. 6. In Brazil, the tax was not collected during the first half of 1999, while the tax rate remained at 0.38 percent. In June 2000, the tax rate was low- ered to 0.3 percent. To compensate for a potential bias in the estimates be- cause of a reduction in distortions followed by the lowering of the tax rate, this study re-based the index of tax revenues to May 2000. 324 ANDREI KIRILENKO AND VICTORIA SUMMERS 7. For example, if the tax became effective on May 15, 1998, this study uses tax revenues for the month of June 1998. 8. This substitution can sometimes lead to a systemic crisis. For exam- ple, the introduction of a 1 percent tax on any bank transaction in Ecuador in December 1998 led to a widespread preference for cash, which seriously exacerbated the ongoing liquidity crisis in the banking system. A blanket deposit insurance guarantee, introduced at the same time as the financial transaction tax, was not sufficient to restore confidence in the banking sys- tem. In the three months after the introduction of the tax, amidst a run on the currency, six small banks and the second largest bank had to be closed. In March 1999, fearing a run on the whole banking system, the government froze all demand and savings deposits for six months and all time deposits for one year. 9. Such funds were authorized by the central bank in order to minimize the disintermediation effects of the tax. References Albuquerque, Pedro H. 2002. “How Bad Is BAD Taxation? Disintermedi- ation and Illiquidity in a Bank Account Debits Tax Model.” Presented to the 2002 North American Meeting of the Econometric Society, Los Angeles. Processed. Caminal, Ramon. 1997. “Financial Intermediation and the Optimal Tax System.” Journal of Public Economics 63 (3): 351–82. Coelho, Isaias, Liam Ebrill, and Victoria Summers. 2001. “Bank Debit Taxes in Latin America: An Analysis of Recent Trends.” Working Paper 01/67. International Monetary Fund, Washington, D.C. Harberger, Arnold C. 1964. “The Measurement of Waste.” American Eco- nomic Review 54 (3): 58–76. Hines, James R., Jr. 1999. “Three Sides of Harberger Triangles.” Journal of Economic Perspectives 13 (2): 167–88. Kirilenko, Andrei A., and Victoria Summers. 2001. “Bank Debit Taxes: Productivity vs. Financial Disintermediation.” International Monetary Fund, Washington, D.C. Processed. Tanzi, Vito. 2000. “Taxation in Latin America in the Last Decade.” Work- ing Paper 76. Stanford University, Center for Research on Economic De- velopment and Policy Reform, Stanford, Calif. 11 Securities Transaction Taxes and Financial Markets Karl Habermeier and Andrei Kirilenko This chapter argues that transaction taxes can have negative effects on price discovery, volatility, and market liquidity in securities mar- kets. These effects can lead to a reduction in market efficiency and may contribute to increased asset price volatility. Financial markets transform latent demands of investors into re- alized financial transactions. Securities transaction taxes (STTs) alter this transformation. Proponents of STTs argue that such taxes can reduce market volatility, help prevent financial crises, and re- duce excessive trading.1 Opponents believe that STTs are difficult to implement and enforce and that they can do great damage to finan- cial markets. This chapter considers the impact of transaction taxes on finan- cial markets in the context of four broad questions. How important is trading? What causes price volatility? How are prices formed? How valuable is the volume of transactions? These questions are at the core of the debate on the role of transaction taxes. The argu- ments here draw on research on market microstructure, asset pric- ing, rational expectations, and international finance. Market microstructure studies suggest that trading is essential for price discovery—the process of finding market clearing prices. A large number of markets rely on dealers to provide price discovery, as well as liquidity and price stabilization. Levying STTs on the This chapter is printed by permission of IMF Staff Papers, where an earlier version appeared. 325 326 KARL HABERMEIER AND ANDREI KIRILENKO dealers inhibits their ability to assist investors with the transforma- tion of latent demands into realized transactions. The literature also finds that much of the volatility is caused by informed traders as their information is aggregated into transaction prices. Taxing fi- nancial transactions does not reduce the volatility due to “noise” trading.2 Rather, it introduces additional frictions into the price dis- covery process. The literature on option pricing under transaction costs shows how frictions on the trading in one asset affects prices and volumes of that and other assets. Using a simple framework based on this lit- erature, this chapter demonstrates how volume can migrate to the assets that are not subject to the tax. That is to say, there will be an increase in the use of these assets for the purpose of risk or liquid- ity management. The chapter also argues that it is very difficult to design and implement a tax that does not favor one portfolio of as- sets over another portfolio with exactly the same payoff. Recent studies on rational expectations question the traditional view that volume is just an outcome of the trading process and is not valuable per se. These studies find that volume can play an informa- tional role. Consequently, if transaction taxes cause volume to mi- grate, then they can hamper the informational efficiency of markets. International finance provides other interesting examples of mar- ket fragmentation, where essentially the same security is traded in distinct markets, as well as of market segmentation, involving per- sistent price differentials for different groups of purchasers. Volume fragmentation can occur because of restrictions on trading of sub- stitutable securities, such as different classes of company shares. This leads to market segmentation and inefficient price discovery. This chapter is organized as follows. After reviewing the literature on STTs, the chapter provides a brief description of the Swedish ex- perience with such taxes. It then deals with the four broad questions specified above. The next-to-last section reviews international fi- nance evidence on market segmentation and execution costs in dif- ferent markets. The final section observes that transaction taxes can have a substantial effect on investment. Overall, the finding is that se- curities transaction taxes can create unexpectedly large distortions. Literature on Securities Transaction Taxes Opinion is divided on the merits of securities transaction taxes. Many proponents of STTs advance the following propositions.3 • The contribution of financial markets to economic welfare does not justify the resources they command. During a given time SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 327 period, the resources that change hands in financial markets far ex- ceed the value of the underlying or “real” transactions. • Many financial transactions are highly speculative in nature, and may contribute to financial or economic instability. • Market instability, including crashes, enriches insiders and speculators, while the costs are borne by the general public. • Financial market activity increases inequalities in the distribu- tion of income and wealth. From this perspective, some argue that governments ought to tax financial transactions to discourage destabilizing speculation that can threaten high employment and price stability, as well as to raise revenue. The massive volume of financial transactions in well- developed modern markets would—they reason—allow substantial revenue to be raised by imposing low tax rates on a broad range of transactions. Higher rates, it is argued, should be levied on short- term transactions, since these seem to benefit primarily market in- termediaries and not “real” users. Opponents of STTs have more faith in the ability of markets to allocate resources efficiently without direct intervention from pub- lic policy. However, the opponents also lack a convincing argument to justify the volume of resources flowing through financial mar- kets. In addition, numerous documented anomalies, as well as a his- tory of market crashes, do not lend themselves easily to the idea that financial markets are fully efficient. Neither does the fact that mar- ket participants devote considerable resources to analyzing previous transaction prices and volumes. Thus instead of showing that the al- location of resources to the financial sector is justified on efficiency grounds, or that observed market volatility is optimal, the oppo- nents of STTs have focused on practical shortcomings of the taxes themselves.4 There are two dimensions to the difficulties in implementing STTs. First, if an STT is applied in one financial market but not in others, the volume of transactions tends to migrate from the market that is taxed to markets that are not. Effective enforcement of STTs thus requires either a cross-market and perhaps even a global reach or measures to segregate markets. For example, tax authorities in one country may attempt to require payment of the tax on transac- tions made by their residents not only in financial markets within their own borders, but in other markets as well. Alternatively, they may impose controls on cross-border financial transactions. Second, financial assets promise certain payoffs at different times and under different relevant circumstances (the latter often summa- rized as “states of the world”). Since the composition of the assets used in financial transactions matters less than the distribution of 328 KARL HABERMEIER AND ANDREI KIRILENKO payoffs over time and in different states of the world, the tax base must be defined as a function of the final payoff, rather than the assets employed. A securities transaction tax would be considered neutral if it did not favor one portfolio of assets over another port- folio with exactly the same pattern of payoffs. Since payoffs can be replicated by portfolios consisting of different types of assets, the imposition of an STT can create a greater distortion than it is try- ing to mitigate. Instead of trading less because of the tax, investors may transact more in assets that are taxed at a lower rate or not taxed at all. As a result, real resources devoted to financial transac- tions may in fact increase rather than diminish following the impo- sition of an STT. Given the lack of a consensus on the theory, there have been many attempts to resolve the debate empirically. However, empiri- cal studies undertaken so far have not been able to decisively resolve the debate on the effects of transaction taxes on financial markets.5 Empirical research has encountered three major problems. First, the effects of taxes on prices and volume are hard to disentangle from other structural and policy changes taking place at the same time. Therefore, estimates based on the assumption that everything else in the economy is held constant are potentially biased. Second, it is difficult to separate transaction volume into stable (or “fundamental”) and destabilizing (or “noise”) components. Thus it is hard to say which part of the volume is more affected by the tax. Third, it is hard to differentiate among multiple ways in which transaction taxes can affect asset prices. These ways include changes in expectations about the impact of the taxes, the cost of creating and trading in close substitutes not covered by the tax, and changes in market liquidity. Empirical studies seek answers to three main questions. The first question is whether transaction taxes have an effect on price volatil- ity. Roll (1989) studies stock return volatility in 23 countries from 1987 to 1989. He finds no evidence that volatility is reliably related to transaction taxes. Umlauf (1993) studies equity returns in Sweden from 1980 to 1987, before and during the imposition of transaction taxes on brokerage service providers. He finds that the volatility did not decline in response to the introduction of taxes. Saporta and Kan (1997) study the impact of the U.K. stamp duty on volatility of se- curities’ prices. They also find no evidence of a relationship between the stamp duty and volatility. Jones and Seguin (1997) examine the effect on volatility of the introduction of negotiated commissions on U.S. national stock exchanges in 1975, which resulted in a perma- nent decline in commissions. They argue that this event is analogous SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 329 to a one-time reduction of a tax on equity transactions. They reject the hypothesis that the lowering of commissions increases volatility. Hu (1998) examines the effects on volatility of changes in transac- tion taxes that occurred in Hong Kong, Japan, Korea, and Taiwan from 1975 to 1994, and does not find significant effects. Finally, Hau and Chevallier (2000) examine the effect on volatility of mini- mum price variation rules in the French stock market. They argue that minimum price variation rules resulted in a doubling of trans- action costs for stocks priced above a certain threshold (500 francs). They argue that this is analogous to the application of a transaction tax on the stocks above the threshold. They find that the increase in transaction costs results in “a statistically significant, but economi- cally insignificant” reduction in the volatility of returns on a daily, weekly, and monthly basis. The second question is whether transaction taxes affect trading volume. Umlauf (1993) reports that after Sweden increased its trans- action tax from 1 to 2 percent in 1986, 60 percent of the volume of the 11 most actively traded Swedish stocks migrated to London. The migrated volume represented over 30 percent of all trading volume in Swedish equities. By 1990, that share increased to around 50 percent. According to Campbell and Froot (1995), only 27 percent of the trading volume in Ericsson, the most actively traded Swedish stock, took place in Stockholm in 1988. Hu (1998) examines 14 tax changes in four Asian markets and finds that differences in turnover before and after changes in the tax level are not statistically significant. Third, empirical studies seek to find out whether transaction taxes have an impact on the prices of securities. Umlauf (1993) re- ports that the Swedish All-Equity Index fell by 2.2 percent on the day a 1 percent transaction tax was introduced and again by 0.8 percent on the day it was increased to 2 percent. Saporta and Kan (1997) find that on the day stamp duty in the United Kingdom was increased from 1 to 2 percent, the stock market index declined by 3.3 percent. Hu (1998) finds that on average the return on the an- nouncement date is –0.6 percent in Korea and –1.6 percent in Tai- wan, with the result for Taiwan being highly statistically significant. One of the main reasons for the dispersion and inconclusiveness of results is the lack of appropriate data. Since the questions are es- sentially of the market microstructure-type, an ideal dataset would consist of transaction frequency data for individual financial instru- ments. In order to take revisions in expectations into account, the data should start well before the announcement of the transaction tax and include a sufficient number of observations following its im- 330 KARL HABERMEIER AND ANDREI KIRILENKO position. Furthermore, in order to separate volume into meaningful categories, the data should be broken down according to the type of investor: for example, institutional investors, hedge funds, and mu- tual funds. In contrast, most empirical studies rely on weekly equity index returns. The Swedish Experience To illustrate the arguments that follow, this section presents a brief description of the Swedish experience with STTs. The Swedish ex- periment lasted for more than eight years. The first measure was an- nounced in October 1983 and the last one was abolished in De- cember 1991. The analysis in this section is based on the studies by Umlauf (1993) and Campbell and Froot (1995). The initiative to impose financial transaction taxes came from the Swedish labor sector in 1983. The labor sector did not claim that trading in financial markets led to inefficient outcomes. Rather, ac- cording to Umlauf (1993), in the opinion of the labor sector, “the salaries earned by young finance professionals were unjustifiable . . . in a society giving high priority to income equality,” especially given the seemingly unproductive tasks that they performed. On this basis, the Swedish labor sector proposed to levy taxes directly on domestic brokerage service providers. Despite the objections of the Swedish Finance Ministry and the business sector, popular support led to the adoption of taxes by Par- liament. The taxes became effective on January 1, 1984. They were levied on domestic stock and derivative transactions. Purchases and sales of domestic equities were taxed at 0.5 percent each, resulting in a 1 percent tax per round trip. Round-trip transactions in stock options were taxed at 2 percent. In addition, exercise of an option was treated as a transaction in the underlying stock and thus was subject to an additional 1 percent round-trip charge. The tax cover- age and rates reflected a popular perception about the “usefulness” of transactions in different financial instruments, with those involv- ing equity options being the least “useful.” Continuing pressure from the labor sector compelled the Parlia- ment to double the tax rates in July 1986 and broaden its coverage in 1987. Furthermore, following large losses in interest futures and options (most notably by the City of Stockholm, which lost SEK 450 million), the tax was extended to transactions in fixed-income secu- rities, including government debt and the corresponding derivatives in 1989.6 The maximum tax rate for fixed-income instruments was set at 0.15 percent of the underlying notional or cash amount. In ad- SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 331 dition, the tax was designed to be “yield-neutral,” with longer ma- turities instruments being taxed at progressively higher rates. The revenue performance of the tax was disappointing. Accord- ing to the Finance Ministry of Sweden, the government collected SEK 820 million in 1984, SEK 1.17 billion in 1985, and SEK 2.63 billion in 1986. This accounted for 0.37, 0.45, and 0.96 percent of the total revenue for the corresponding years. After doubling the tax rates the government was able to collect SEK 3.74 billion in 1987 and SEK 4.01 billion in 1988. This accounted for 1.17 and 1.21 percent of the total revenue.7 Thus a 100 percent increase in the tax rate resulted in a 22 percent increase in revenue. Widespread avoidance was one reason for the weak performance of the tax. Foreign investors avoided the tax by placing their orders with brokers in London or New York. Domestic investors avoided it by first establishing offshore accounts (and paying the tax equal to three times the round-trip tax on equity for funds moved offshore) and then using foreign brokers. The scale of avoidance was manifested by a massive migration of stock trading volume from Stockholm to other financial centers. Since the brokerage business is highly competitive, finding a close substitute for brokerage services offshore was not costly. According to Umlauf (1993), following the doubling of the tax, 60 percent of the volume of the 11 most actively traded Swedish stocks migrated to London. The migrated volume represented over 30 percent of all trading volume in Swedish equities. By 1990, that share increased to around 50 percent. However, the market does not necessarily move off shore, if close substitutes are available domestically. For example, trading in bonds did not move offshore, but shifted to debentures, forward contracts, and swaps. Application of the tax to fixed-income instruments did result in a sharp drop in trading volume in Swedish government bills and bonds and in fixed-income derivatives contracts. Campbell and Froot (1995) estimate that during the first week of the tax, bond trading volume dropped by about 85 percent from its average during the summer of 1987 and trading in fixed-income derivatives essentially disappeared. This significantly undermined the ability of the Bank of Sweden to conduct monetary policy, made government borrow- ing more expensive, and eroded both popular and political support for the tax. Taxes on fixed-income instruments were abolished in April 1990. Taxes on other instruments were cut in half in January 1991 and abolished altogether in December 1991. Following the abolition of the tax, some trading volume came back to Sweden. According to Campbell and Froot (1995), 41 per- 332 KARL HABERMEIER AND ANDREI KIRILENKO cent of the trades in Ericsson took place in Stockholm in 1992. Over- all, the proportion of the trading volume in Sweden increased for al- most all equities in 1992. That year, 56 percent of all trading volume in Swedish equities took place in Stockholm. The most striking features of the Swedish experience are the ex- tent to which investors avoided the tax by finding or creating close substitutes, the extent to which market activity suffered, and the re- versibility of the effect once the tax was removed. How Important Is Trading? The Swedish labor sector believed that trading in financial markets is an essentially unproductive task. Just how important is trading? The answer to this question depends on how the trading is con- ducted. In Sweden, investors had to carry out financial transactions mostly through dealers. However, trading does not have be conducted exclusively through dealers. It can be done through other mechanisms. For example, in continuous electronic auctions, buyers and sellers trade directly with each other, bypassing the dealers. Why didn’t such an auction de- velop in Sweden? In fact, under the law, transactions executed with- out dealers were exempt from taxes. According to the market microstructure literature, under some circumstances, dealers offer services that cannot be provided by other types of market designs at lower cost. It is especially true for infrequently traded assets, such as most of the Swedish stocks. Per- haps for that reason the order flow migrated not to another trading design, but to dealers in London and New York. Dealers provide several important services. They provide liquid- ity and assume substantial risks by contributing their own capital. Accordingly, they demand adequate compensation for the provision of liquidity and the capital that they put at risk. The dealer’s com- pensation is higher for illiquid assets. In addition, dealers who act as market makers in particular secu- rities must furnish competitive bid and offer quotations on demand and be ready, willing, and able to effect transactions in reasonable quantities at the quoted prices. In other words, a buyer does not have to wait or look for a seller, but can simply buy from a dealer who sells from his inventory. According to Pagano and Roell (1990), “this implies that, in contrast with what happens on auction markets, traders are insured against execution risk, i.e., the risk of finding few SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 333 or no counterparties to trade.” The dealer’s compensation is higher for assets with a higher execution risk. This highlights another important function that dealers play: namely, the provision of price stability. According to Madhavan (2000), “the presence of market makers who can carry inventories imparts stability to price movements through their actions relative to an automated system that simply clears the market at each auc- tion without accumulating inventory.” The provision of liquidity, price discovery, and price stabilization requires inventory management. Inventory management is achieved through the buying and selling of securities. Hasbrouck and Sofianos (1993) examine a set of quote, trade, and inventory data for market makers (specialists) on the New York Stock Exchange (NYSE). Ac- cording to their data, the market maker’s activity (both purchases and sales) averages about 26 percent of the total transaction flow (also both purchases and sales). For the most frequently traded stocks, this number is 20 percent, while for the least frequently traded stocks, it rises to 38 percent.8 Thus dealers become much more im- portant as liquidity providers in less frequently traded stocks. Inventory management can involve both customer and inter- dealer trading. When a competitive inter-dealer market is available, dealers can adjust their inventory without waiting for a public order flow to arrive. According to the empirical evidence, dealers trade in the inter-dealer market when they want to manage large inventory positions. Lyons (2001) suggests that inter-dealer trading in the for- eign exchange market currently accounts for about two-thirds of the total volume. Hansch, Naik, and Viswanathan (1998) show that the average size of an inter-dealer trade on the London Stock Ex- change is much larger than the average size of a trade with the gen- eral public. They also show that inventory levels at which dealers trade among themselves is about twice as large as those at which they trade with the general public. They find that 38 percent of the variation in inter-dealer trading is explained by variation in inven- tory levels. They conclude that “inter-dealer trading is an important mechanism for managing inventory risks in dealership markets.” Thus trading is important. It helps manage risks. Dealers demand compensation for the services that they provide and the risks that they take. If trading becomes costly as a result of transaction taxes, dealers cannot manage their risks effectively. Accordingly, they be- come less willing to put their own capital at risk in order to provide liquidity. Investors cannot carry out their desired trades, their latent demands are not fully satisfied, and resources are not allocated to their best uses. 334 KARL HABERMEIER AND ANDREI KIRILENKO What Causes Volatility? The previous section argued that trading is important. But can it also be the cause of volatility? French and Roll (1986) conduct an empirical study of the vari- ability of stock returns over trading and nontrading periods. Using data for all stocks listed on the NYSE and American Stock Ex- change (AMEX) for the period 1963 to 1982, they find that on an hourly basis, the variance of stock returns is between 13 and 100 times larger when markets are open for trading than the variance when the markets are closed, depending on the definition of non- trading period. They investigate three possible causes for the higher volatility during trading hours. First, higher volatility may be caused by the arrival of more public information during trading hours. Second, it may be caused by informed investors as their private information is incorporated into prices. Finally, higher volatility may be caused by the process of trading itself as prices fluctuate because of market frictions and transaction costs. They also find that the process of trading accounts for at most 12 percent of the daily return variance. The rest of the variance is attributable to the arrival of public and private information during trading hours. While they cannot directly decompose the effects of public and private information on volatility, they conduct a test that suggests that most of the variability in stock returns can be attrib- uted to the arrival of private information during trading hours. Later studies have relied on much more refined transaction-level data to further decompose transaction price volatility. Madhavan, Richardson, and Roomans (1997) develop a stylized, reduced-form model of price volatility and use transaction-level, intraday data on 274 NYSE-listed stocks during 1990 to estimate it. They argue that price volatility can be explained by the variabil- ity of four components: public information, private information, transaction costs, and other market frictions (price discreteness). They estimate that the impact of public information accounts for 46 percent of volatility at the beginning of the trading day and 35 per- cent at the end. The impact of private information (including the in- teraction between cost and private information effects) drops from 31 percent in the morning to 26 percent at the closing of trading. Variability in transaction costs increases from 22 percent at the opening to 35 percent at the end of the trading day. Finally, price dis- creteness accounts for the remaining 1 to 4 percent at the beginning and the end of the trading day, respectively. SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 335 Transaction costs in the Madhavan, Richardson, and Roomans (1997) model capture dealer costs for supplying liquidity on de- mand. They include compensation for inventory costs, putting their capital at risk, and other transaction costs. The model implies that other things being equal, higher transaction costs increase volatility. If transaction costs also include transaction taxes, then introduction of STTs can result in higher rather than lower volatility of transac- tion prices. How Are Prices Formed? In perfect, frictionless markets, asset prices immediately reflect all available information. As the new information arrives, investors re- balance their portfolios of assets. The rebalancing results in an up- dated set of prices. In the absence of transaction costs, the rebalanc- ing can be done continuously and price discrepancies are eliminated instantaneously. However, in real markets, agents face transaction costs. The presence of even small transaction costs makes continu- ous rebalancing infinitely expensive. Therefore, valuable informa- tion can be held back from being incorporated into prices. As a re- sult, prices can deviate from their full information values. The dissatisfaction with the assumption of continuous portfolio rebalancing was the starting argument for the literature on the repli- cation of assets under transaction costs. The literature recognizes that continuous rebalancing is not feasible and formulates discrete rebalancing under transaction costs. This section presents a simple theoretical framework based on the literature on option pricing with transaction costs. It is assumed that securities taxes are a source of transaction costs. The frame- work studies the impact of STTs on portfolio rebalancing and price formation. A Simple Example Consider a simple two-period example (following Hull 1985). There are three assets in the market: a risk-free bond yielding 3 percent per year, a non–dividend-paying stock, and a call option on the stock. The starting price of each share of stock is equal to $20. After a year, assume that the stock price will either have increased to $22 or have fallen to $18, with equal probability. The strike price of the option at the end of the year is taken to be $21. 336 KARL HABERMEIER AND ANDREI KIRILENKO Simple option pricing theory can be employed to compute in what proportions a call option and a risk-free bond must be held in order to be equivalent to 100 shares of stock. As shown in the ap- pendix, on the assumptions given, this portfolio requires exactly 400 options (worth $0.63 each) and $1,747 of the bond. But a 1 percent transaction tax on buying or selling the stock greatly lowers the value of the option, as the tax of $0.22 must be incurred twice if the option is exercised and the stock then sold. Working through the arithmetic reveals that the option is only worth $0.39 and that now 694 options must be bought (along with $1,728 worth of bonds) to match 100 shares. If the transaction tax is also levied on option transactions or on bonds, there is a further change in the required number of options in the portfolio to replicate the shares—but in these cases, the changes are very small. Thus extending the transaction tax to all three assets certainly does not restore neutrality. Note that even in this simple example, it is quite difficult to de- sign and even more difficult to implement a tax that does not favor one portfolio of assets over another portfolio with exactly the same payoff (such as a stock versus a bond and a call option). A uniform transaction tax is not payoff-neutral. For a tax to be payoff-neutral, the tax rates must be such that a change in the value of a replicat- ing portfolio is exactly equal to the change in the price of the un- derlying asset. In other words, the tax rates must depend on what is known as the “delta” of the replicating portfolio.9 But it is known that, in practice as in theory, “delta” changes as more information is revealed about the (unknown) underlying stochastic process. Therefore, a payoff-neutral tax would have to be frequently ad- justed. This would make it difficult to implement. A Generalized Model Boyle and Vorst (1992) have generalized the simple two-period ex- ample to a multi-period case. They use a method proposed by Cox, Ross, and Rubinstein (1979), who assumed a dynamic price process according to which, during each subperiod of length ∆t, the stock price increases by a factor θ = exp{δ√∆t }, with probability p; other- wise it decreases by the same multiplicative factor. Boyle and Vorst show that, if this binomial multiplicative price is itself unaffected by transaction costs—that is, if the response of agents to transaction taxes does not lead to a change in θ—the call option can still be priced after the introduction of transaction costs by increasing the variance by an amount that is positively related to the rate of the transaction cost or tax and inversely related to the SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 337 length of the rebalancing period.10 Specifically, if δ2 is the original ˆ 2 is given by, variance, the modified variance, δ (11.1) ˆ 2 = δ2 1 + k 2  δ    δ ∆t  where k is the rate of transactions cost.11 How Valuable Is the Volume of Transactions? According to the example presented in the previous section, demand for assets changes following the introduction of a transaction tax on a stock. The demand for derivatives goes up and the demand for both stocks and bonds decreases. Changes in demand translate into changes in the volume of realized transactions. Was anything lost as a result of this change in volume? Does it matter if transaction vol- ume migrates to other instruments, markets, or countries? It does not, if the volume is not valuable. But how valuable is the volume of realized transactions? According to standard rational expectations models with supply uncertainty, trading orders have both informational (or “signal”) and “noise” components. Without the noise, aggregate supply uncertainty is resolved, and prices adjust to their full information level. Other- wise, the informational component is aggregated into prices and the “noise” is left in volume. Consequently, volume is just an outcome of the trading process. It does not have any information about the fun- damentals or the trading process and, therefore, lacks value. According to this view, the migration of volume to other instru- ments, markets, or countries does not result in any loss of value or efficiency. It just means a reallocation of supply uncertainty. In other words, if transaction volume moves from Stockholm to Lon- don, investors in Stockholm become exposed to less uncertainty as- sociated with “noise” trading and investors in London to more of it. Thus after a transaction tax is imposed, if volume migrates away from the taxed asset, the policymakers should perhaps just change their revenue projections and not worry about any fundamental market effects. The long-held view that volume is not valuable per se has recently come under scrutiny. Blume, Easley, and O’Hara (1994) investigate the informational role of volume. In their model, the source of “noise” is not supply uncertainty, but the precision of private in- formation about the signal. Prices aggregate information about the 338 KARL HABERMEIER AND ANDREI KIRILENKO average level of private information. Trading volume contains infor- mation about the precision of individual private signals. Thus vol- ume does not just contain “noise,” but has a nontrivial informa- tional role to play. Price-volume sequences are more informative than prices alone. This role becomes especially important for infre- quently traded stocks that often do not get much analyst coverage. In addition, Easley, O’Hara, and Srinivas (1998) investigate the in- formational role of transaction volume in options markets. They de- velop a model where informed traders can trade in stock or options markets. They empirically test the model and find that option volume data contain information about future stock prices. Thus they con- clude that “volume plays a role in the process by which markets be- come efficient.” Consequently, a migration of volume from the de- rivative market may also result in the loss of informational efficiency. This new view represents a fundamentally different perspective on the role of volume. It can be summarized as saying that “volume matters.” The migration of volume results in lower informational efficiency of instruments and markets from which it migrated. If transaction taxes cause the volume to migrate, then they do affect the ability of markets to aggregate information and prevent a more efficient allocation of resources. Evidence from International Finance The international finance literature provides examples of market seg- mentation and execution costs in different markets. Market segmen- tation can result from direct restrictions on foreign ownership, ex- change and capital controls, and regulatory and accounting aspects including disclosure rules, settlement practices, and investor pro- tection rights. Bekaert (1995) studies 19 emerging markets and finds that exchange and capital controls (and taxes that have a similar ef- fect), as well as regulation and accounting practices, are significant in explaining market segmentation. Restrictions on foreign ownership are apparently being circumvented by the closed-end country funds. Domowitz, Glen, and Madhavan (2000) use a comprehensive database of execution costs (including transaction taxes) for 42 countries from September 1996 to December 1998. They use panel data techniques to study the interaction between cost, liquidity, and volatility across countries and through time. They find that except for North America, explicit equity trading costs such as brokerage commissions, taxes, and fees account for about two-thirds of total execution costs. In the United States, average explicit one-way trad- SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 339 ing costs are the smallest for the countries in their study, accounting for 8.3 basis points or a fraction of 2.2 percent of mean return (374 basis points) for the period 1990–98. In other words, a complete re- balancing of the portfolio once a year results in an average explicit cost of 2.2 percent of its annual mean return. The largest explicit cost of 106 basis points is in Ireland, which has a stamp duty of 1 percent. In Ireland, the explicit costs of turning over a portfolio of equities just once a year accounts for a full 25 percent of the annual mean return. They also find that over time, with the exception of transition economies, costs have generally declined, and that higher trading costs are positively related to increased volatility and lower volume. Summary and Conclusions This chapter examines recent research relevant to assessing the im- pact of securities transaction taxes on financial markets. This re- search includes work on market microstructure, asset pricing, ra- tional expectations, and international finance. The study concludes that in most circumstances, transaction taxes can have negative ef- fects on price discovery, volatility, and liquidity and lead to a reduc- tion in market efficiency. The arguments made in this chapter may be summarized as fol- lows. First, in dealership markets, trading facilitates the provision of liquidity, price discovery, and price stabilization. Trading also helps to manage risks. If investors cannot carry out their desired trades, their latent demands are not fully satisfied and resources are not al- located to their best use. Second, price volatility can be explained by the variability of four components: public information, private information, transaction costs, and other market frictions. Other things being equal, higher transaction costs increase volatility. Consequently, the introduction of STTs can increase the volatility of transaction prices. Third, a simple theoretical framework based on the literature on option pricing with transaction costs shows that following the in- troduction of a transaction tax, the demand for derivatives can in- crease substantially. Moreover, it is difficult to design and implement a tax that does not favor one portfolio of assets over another port- folio with exactly the same payoff. Fourth, if transaction volume has an informational content, then a migration of volume would result in lower informational effi- ciency of instruments and markets from which it migrated. If trans- 340 KARL HABERMEIER AND ANDREI KIRILENKO action taxes are the cause of volume migration, then they can inhibit the informational efficiency of markets. Finally, the international finance evidence on market segmenta- tion and execution costs in different markets suggests that except for North America, explicit equity trading costs such as brokerage commissions, taxes, and fees account for about two-thirds of total execution costs. The study concludes that higher trading costs, some of which are due to STTs, are positively related to increased volatil- ity and lower volume. Transaction taxes can thus have a substantial effect on the trans- formation of investor demands into transactions. STTs can obstruct price discovery and price stabilization, increase volatility, reduce market liquidity, and inhibit the informational efficiency of finan- cial markets. Appendix. Working through the Numerical Example In order to compute the portfolios presented in the fifth section, begin by choosing a number of shares δ so that holding that number of shares and selling 100 call options provides a risk-free portfolio: that is, one that has the same value whether the share goes up or down. Since the value of op- tion at maturity when it is “in the money” is exactly 1 (since then the op- tion allows the share to be bought at the strike price of 21 and sold at 22), δ must satisfy: (A11.1) 22 δ – 100 = 18 δ. The solution to this equation is δ = 25. The value of this portfolio at the end of the year will be 18 δ = 450, which equals $437 discounted to the present at 3 percent per year. This, then, must be the value of the risk-free portfolio at the outset. Therefore since the 25 shares will then cost $500, one can conclude that the price of the 100 options is $500 – $437 = $63. Rearranging, one can conclude that a portfolio consisting of 100 call op- tions and $437 of bonds will exactly replicate the payoff on 25 shares. Equivalently, to replicate 100 shares requires exactly 400 options and $1,747 of bonds. Suppose now that a transaction tax of 1 percent is introduced on all pe- riod-one transactions in the stock. Once more, when the stock price is equal to 22, the option gives a right to buy the stock at 21 and sell it at 22. But now this round-trip transaction is subject to transaction taxes. To buy the stock, the option’s holder must pay an additional $0.21 when buying the stock and $0.22 when selling it. Accordingly, the net terminal value of an in-the-money option is now just $0.57. SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 341 Let δ* be the amount of stock in the risk-neutral valuation portfolio ad- justed for the transaction tax. Then, subtracting 1 percent transaction tax from the price of the share in each case, (A11.2) 21.78 δ* – 57 = 17.82 δ*. The solution is now δ* = 14.4 and the value of the portfolio at the end of the year will be 17.82 δˆ = $257. The present value of this amount is equal to $249. Since the 14.4 shares will the cost $288, the price of the 100 options is $288 – $249 = $39. Rearranging, one can conclude that a portfolio consisting of 100 call op- tions and $249 worth of bonds replicates just 14.4 shares. Replicating 100 shares requires 694 options (plus $1,728 of bonds). Notes 1. For example, Eichengreen, Tobin, and Wyplosz (1995) argue that “transaction taxes are one way to throw sand in the wheels of super- efficient financial vehicles.” 2. By this term is meant trading by those who have poor information, or whose trades are based on liquidity needs. It could also include trading automatically generated by portfolio balancing programs. 3. See, for example, Tobin (1984); Summers and Summers (1989); Stiglitz (1989); and Eichengreen, Tobin, and Wyplosz (1995). 4. See, for example, Campbell and Froot (1995). 5. Hammond (1995) reviews most of the empirical research on financial transaction taxes. Empirical studies since 1995 have sought to address sim- ilar issues by using other data sets. 6. Officially, the extension of the tax to fixed-income instruments was supposed to achieve “neutrality” with the tax on equity transactions. See Campbell and Froot (1995). 7. By contrast, tobacco taxes accounted for 1.26 and 1.37 percent of the total revenue collected in 1987 and 1998, respectively. 8. The statistics are calculated by taking the participation rates reported in the paper as a fraction of 50 percent, the rate which implies that the mar- ket maker is a counterparty to all trades. 9. “Delta” is conventionally defined as the response of the value of a de- rivative portfolio with respect to the price of the underlying asset. 10. Reinhart (2000) argues that the introduction of STTs may also make asset prices more variable in the general equilibrium setting. 11. Leland (1985) develops an extension to the Black-Scholes continu- ous-time model and shows how to modify the variance to price call options in the presence of transaction costs. In Leland’s model, the variance in- creases in the presence of transaction costs, reflecting the discontinuous re- balancing of portfolios necessitated by transaction costs. 342 KARL HABERMEIER AND ANDREI KIRILENKO References Bekaert, Geert. 1995. “Market Integration and Investment Barriers in Emerging Equity Markets.” World Bank Economic Review 9 (1): 75–107. Blume, Lawrence, David Easley, and Maureen O’Hara. 1994. “Market Sta- tistics and Technical Analysis: The Role of Volume.” Journal of Finance 69 (1): 153–81. Boyle, Phelim P., and Ton Vorst. 1992. “Option Pricing in Discrete Time with Transaction Costs.” Journal of Finance 47 (17): 271–94. Campbell, John Y., and Kenneth A. Froot. 1995. “Securities Transaction Taxes: What About International Experiences and Migrating Markets?” In Suzanne Hammond, ed., Securities Transaction Taxes: False Hopes and Unintended Consequences. Chicago: Catalyst Institute. Cox, John C., Stephen A. Ross, and Mark Rubinstein. 1979. “Option Pric- ing: A Simplified Approach.” Journal of Financial Economics 7 (3): 229–63. Domowitz, Ian, Jack Glen, and Ananth Madhavan. 2000. “Liquidity, Volatility, and Equity Trading Costs Across Countries and Over Time.” International Finance Corporation, Washington, D.C. Processed. Easley, David, Maureen O’Hara, and P. S. Srinivas. 1998. “Option Volume and Stock Prices: Evidence on Where Informed Traders Trade.” Journal of Finance 53 (2): 431–65. Eichengreen, Barry, James Tobin, and Charles Wyplosz. 1995. “Two Cases for Sand in the Wheels of International Finance.” Economic Journal 105 (428): 162–72. French, Kenneth, and Richard Roll. 1986. “Stock Return Variances: The Arrival of Information and the Reaction of Traders.” Journal of Finan- cial Economics 17 (1): 5–26. Hammond, Suzanne, ed. 1995. Securities Transaction Taxes: False Hopes and Unintended Consequences. Chicago: Catalyst Institute. Hansch, Oliver, Narayan Y. Naik, and S. Viswanathan. 1998. “Do Inven- tories Matter in Dealership Markets? Evidence from the London Stock Exchange.” Journal of Finance 53 (5): 1623–56. Hasbrouck, Joel, and George Sofianos. 1993. “The Trades of Market Mak- ers: An Empirical Analysis of NYSE Specialists.” Journal of Finance 68 (5): 1565–93. Hau, Harald, and Anne Chevallier. 2000. “Evidence on the Volatility Effect of a Security Transaction Tax.” INSEAD, Fontainebleau Cedex, France. Processed. Hu, Shing-yang. 1998. “The Effects of the Stock Transaction Tax on the Stock Market: Experiences from Asian Markets.” Pacific-Basin Finance Journal 6 (3–4): 347–64. Hull, John C. 1985. Options, Futures, and Other Derivatives. Upper Sad- dle River, N.J.: Prentice Hall. SECURITIES TRANSACTION TAXES AND FINANCIAL MARKETS 343 Jones, Charles M., and Paul J. Seguin. 1997. “Transaction Costs and Price Volatility: Evidence from Commission Deregulation.” American Eco- nomic Review 87 (4): 728–37. Leland, Hayne E. 1985. “Option Pricing and Replication with Transaction Costs.” Journal of Finance 60 (5): 1283–1301. Lyons, Richard K. 2001. The Microstructure Approach to Exchange Rates. Cambridge, Mass.: MIT Press. Madhavan, Ananth. 2000. “Market Microstructure: A Survey.” Journal of Financial Markets 3 (3): 205–58. Madhavan, Ananth, Matthew Richardson, and Mark Roomans. 1997. “Why Do Securities Prices Change? A Transaction-level Analysis of NYSE Stocks.” Review of Financial Studies 10 (4): 1035–64. Pagano, Marco, and Alisa Roell. 1990. Auction Markets, Dealership Mar- kets and Execution Risk. London: CEPR Financial Market Paper Series. Reinhart, Vincent R. 2000. “How the Machinery of International Finance Runs with Sand in its Wheels.” Review of International Economics 8 (1): 74–85. Roll, Richard. 1989. “Price Volatility, International Market Links, and their Implications for Regulatory Policies.” Journal of Financial Services Research 3: 211–46. Saporta, Victoria, and Kamhon Kan. 1997. “The Effects of Stamp Duty on the Level and Volatility of UK Equity Prices.” Working paper. Bank of England, London. Stiglitz, Joseph E. 1989. “Using Tax Policy to Curb Speculative Short-term Trading.” Journal of Financial Services Research 3: 101–15. Summers, Lawrence H., and Victoria P. Summers. 1989. “When Financial Markets Work Too Well: A Cautious Case for a Securities Transaction Tax.” Journal of Financial Services Research 3: 261–86. Tobin, James. 1984. “On the Efficiency of the Financial System.” Lloyd’s Bank Review 153: 1–15. Umlauf, Steven R. 1993. “Transaction Taxes and the Behavior of the Swedish Stock Market.” Journal of Financial Economics 33 (2): 227–40. 12 Consumption Taxes: The Role of the Value-Added Tax Satya Poddar This chapter considers the application of expenditure or consump- tion-type taxes to the financial sector, with special emphasis on the role of the VAT. Most financial services are exempt from value- added taxation (VAT) in virtually all the countries employing this form of broad-based consumption tax. This means that financial in- stitutions do not charge tax on the supply of exempt financial ser- vices, but they do pay tax on their own purchase of taxable goods and services acquired for use in making the exempt supplies. The de- cision to exempt financial services from VAT has revolved around conceptual and administrative difficulties associated with measuring the value of financial services on a transaction-by-transaction basis, rather than social or economic policy reasons. The nonneutral treat- ment of financial services relative to other taxable goods and ser- vices leads to a number of economic distortions and administration and compliance difficulties. While historically, tight regulation of most such services limited the impact of these nonneutralities and complexities, globalization and deregulation have significantly in- creased the importance of those situations where the exemption sys- tem does not perform well. Confronted with growing problems, policymakers are seeking al- ternative approaches that treat such services in a more neutral fash- ion. This chapter looks at the issues raised by the exemption system, and describes and evaluates alternative approaches that have been 345 346 SATYA PODDAR identified and, in some cases, tried in certain countries. Following a discussion of how financial services can be categorized for the pur- poses of taxation, the chapter turns to the current tax treatment of financial services and its drawbacks. The next section pauses to con- sider some theoretical considerations as to the ideal tax base. The chapter then looks at the main alternatives that are available, whether under the rubric of the VAT, or in terms of compensatory taxes. The study concludes with observations specifically relevant to developing countries. Categories of Financial Services To understand the current treatment of financial services under consumption tax systems and to undertake tax policy analysis of potential options, five categorizations of financial services are use- ful. These are by type, by provider, by economic function, by the form of service provided, and by the nature of the consideration. An obvious place to start in looking at categories of financial ser- vices is by reference to the type of the services provided. Financial services are also usually identified in this manner in the market place. They are referred to as being banking services, life insurance, and so on. Table 12.1 sets out one classification of the main types of financial services on this basis. The taxing statutes generally use Table 12.1. Five Main Categories of Financial Services 1. Deposits, borrowing, and lending a. Banking operations b. Credit card operations 2. Purchase, sale, and issuance of financial securities a. Bonds, shares, options, guarantees, and foreign currencies b. Gold and precious metals 3. Insurance a. Life b. Property and Casualty 4. Brokerage and other agent services a. Buying and selling of financial securities b. Underwriting and other transactions where agents act as principals 5. Advisory, management, and data processing a. Asset management and investment advice b. Administrative and information services, incidental or supplementary to financial services c. Other CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 347 type of service as the means of identifying financial services in order to specify the tax treatment applicable. For example, the Sixth VAT Directive of the European Community (EC) exempts insurance and reinsurance from VAT under article 13(B)(a), and a list of other fi- nancial services under article 13(B)(d). The types of services cited may be broad, as in table 12.1, or much more detailed and disag- gregated into many subcomponents, as is done in some countries under their tax regulations or administrative guidelines.1 Second, in some instances financial services may be associated with a particular type of provider, such as an insurance company or a bank. Some types of indirect taxes on financial services may spec- ify that the tax is to be collected by or imposed on only a particular type of business. For example, premium taxes may be collected only on services provided by companies that are registered as insurance companies. This approach may be workable only in cases were there is regulation of the activity involved and the service being taxed is available only from a financial institution of a particular type. De- regulation and disintermediation make it difficult to link the appli- cation of tax to a service by type of provider. Third, financial services can also be categorized by the economic function that is being performed that leads to creation or addi- tion of value. Financial services can be categorized under functions such as: • Intermediation between borrowers and lenders • Pooling of savings • Pooling of risks • Provision of liquidity • Transaction clearing services • Creation and making of markets, and • Agency services. These are functions that consumers of financial services are will- ing to pay for. Categorization by economic function is primarily useful as an analytical tool in addressing fundamental questions about what should be the proper indirect tax base for a financial service. They also draw attention to certain characteristics of finan- cial services that may be important in designing and analyzing tax structures. For example, management of risk is a fundamental in- gredient in many types of financial services. Under Canadian leg- islation, presence of financial risk is used as an important deter- minant for distinguishing financial services from non-financial services. While useful analytically, services are not typically pro- vided on a function-by-function basis, and this type of categoriza- tion is of limited use in the design of tax structures. 348 SATYA PODDAR A fourth categorization of financial services is according to the role played by the service providers. They may be involved in finan- cial transactions as principals or as agents arranging for transactions between principals. When providers of financial services act as prin- cipals, they will be entering into financial contracts under their own name and assuming any risks in the transaction (although the risks may be hedged). For example, an insurer will always be acting as a principal. When providers of financial services act as arrangers for transactions, they act as agents or in some other facilitating role for third parties that are the principals in the transaction. An asset man- ager or adviser would always be playing the role of arranging for transactions. As discussed later, this distinction has some important implications in understanding the current systems and designing al- ternative approaches. It is also a prime factor in terms of the fifth categorization of financial services: that is, by the nature of consid- eration. While the distinction between the role of the provider as a principal and as an agent or arranger is relatively clear in most cases, the same institution may act in both capacities. For example, an in- vestment dealer may arrange for transactions of clients as a broker and may also be involved in the market as a principal in a market- making capacity. The fifth categorization of financial services of relevance is the nature of consideration. Consideration for the supply of a financial service can take two basic forms, an explicit fee or commission, or a financial margin. In the case of explicit fees or commissions, fi- nancial services are charged for in the same way as other services, such as those of an electrician or a barber. The fee or commission measures the consideration for the supply of the service. It is rela- tively straightforward to apply the tax to such fees or commissions. This is not the case where the consideration is implicit, in the form of a financial margin. In a loan and deposit business of a bank, for example, financial transactions give rise to a series of cash flows representing principal, payment of interest, and payment of risk premiums. The consideration for the service in these cases is the margin that remains with the bank, after all cash inflows and out- flows have been taken into consideration. Such a margin represents a consideration for a bundle of transactions (such as deposits and loans) and cannot be readily attributed to individual transactions. Therefore it is difficult to identify the appropriate tax base in such cases, an issue that is returned to in detail later. It was mentioned earlier that the categorization by the role of the service provider was closely associated with the categorization by the nature of consideration. This is because an arranger for a finan- CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 349 cial transaction is constrained to collecting consideration in the form of a readily observable fee or commission. On the other hand, where the provider is a principal, it will often obtain consideration for the service in the form of a margin. However, this is not neces- sarily the case. For example, a bank may charge explicit fees for ser- vices to depositors and borrowers, or it may earn its consideration in the form of the margin between interest received on loans and paid on deposits, or by a combination of fees and commissions, and margin. Therefore there is an element of substitutability between the two basic forms of considerations that will be seen to have had important implications for the current manner in which financial services are treated in consumption tax systems. Current Approaches to the Taxation of Financial Services The most common approach to financial services under a VAT or other general consumption taxes is to exempt them from tax, for reasons shortly to be considered. However, the scope of the exemp- tion may vary somewhat from one jurisdiction to another. In addi- tion, certain jurisdictions have adopted compensatory taxes to ob- tain additional revenues where the exemption applies. The exemption system in the European Union provides a useful case for discussing the rationale for the exemption. Article 13 of the Sixth VAT Directive provides for certain exemptions from VAT for supplies of financial services made within the territory of a member state.2 The exemptions apply for supplies within the member states or to residents of other member states. Two broad rationales have been suggested for VAT exemptions, one dealing with the economic or social characteristics of a good or service, the other with opera- tional issues in the application of the tax to the good or service. The economic rationale for special treatment is the perception, whether factual or not, that it will make a consumption tax system less regressive. Exemptions (and lower tax rates) for certain goods and services supplied by public bodies and nonprofit organizations would fall into this category. The social reason for exemption would be that the good or service in question is so meritorious that it de- serves to be tax-free. The treatment of medical care or education would be examples of the social rationale. It is important to note that neither of these two arguments applies in the case of financial services. Indeed, attempts to tax specific financial services in devel- oping countries often appear to be founded on the view that such 350 SATYA PODDAR taxes would be progressive. There is even a view that some activi- ties, such as currency trading, may be related to undesirable specu- lation and thus constitute demerit transactions. The rationale for exemption of financial services thus must be found in the operational aspects of applying the tax to these types of transactions. At the advent of the VAT in Europe, there was not al- ways conceptual clarity and consensus about the appropriateness of some transactions being subject to VAT and about the definition of the base, if those transactions were to be taxable. Issues could in- volve whether a transaction was related to consumption or invest- ment, or what the conceptually correct measure of consideration was. As was noted earlier in the discussion of the nature of consid- eration for supplies of financial services, the consideration for finan- cial services may be of an implicit form that is hidden in the margin left after a series of monetary flows. In such cases, there is no read- ily available measure of consideration on a transaction-by-transac- tion basis that can serve as a basis for the application of VAT. While one can only speculate, as there is no explicit statement of the ra- tionale for exempting financial services, the main reason for the ex- emption was likely this lack of an explicit measure of consideration.3 While the discussion above dwelled on operational difficulties of applying VAT where consideration is received in the form of mar- gin, it would have been possible to apply VAT to fees and commis- sions, and restrict the exemption solely to margin activities. How- ever, full use of this approach has been rejected in the past because, to some extent, consideration for financial services provided by principals to a transaction can be in the form of either explicit fees or commissions, or margin. For example, as noted earlier, charges for deposit account services may take the form of either low or zero interest being paid on deposits, or explicit fees per transaction or per month. It was believed that there would be too much of an in- centive to substitute margin charges for explicit fees and commis- sions, and too many competitive distortions created under this ap- proach. As a result, many financial services for which consideration is typically in the form of explicit fees and commissions are also ex- empted. However, there are some exceptions. For example, safety deposit box fees are taxable in almost all countries with a VAT. Financial services are thus exempted under virtually all general consumption tax systems around the world, with some taxation of such services that involve explicit fees and consideration. The extent of taxation varies from country to country, but presumably reflects judgments about the substitutability of margin for fees and commis- sions in given instances. As a broad generalization, the more recently designed VATs tend to tax a somewhat broader selection of finan- cial services than the older systems. For example, Australia, Canada, CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 351 New Zealand, and Singapore, which have more recently introduced VATs, extend the taxation of financial services in various ways be- yond the European coverage of financial services. This may reflect, in part, the fact that the degree of substitutability between fees and margins may be declining under the pressures of certain economic developments. This issue is returned to later as the options available for extending the base for taxation are identified and evaluated. Once the operational difficulty of taxing financial services is accepted, the options that are fully consistent with a credit-invoice VAT approach are exemption and zero-rating. Under zero-rating, no tax is charged on the supply of financial services, but the service provider is allowed to recover input tax credits related to the sup- ply. Given this choice, exemption can be justified as somewhat of a compromise solution that applies some tax (by denying input cred- its) and thus collects some revenue. From these perspectives, ex- emption is a less extreme approach than zero-rating, which would be more appropriate in situations where regressiveness or merit ar- guments are the key policy rationales. Finally, it needs to be acknowledged that the current system in many countries places considerable reliance on compensatory indi- rect taxes applied to financial services or financial institutions that operate outside broad-based consumption tax structures. These taxes may be introduced primarily for revenue reasons, but may also be ra- tionalized as being responses to the incomplete taxation under the VAT due to exemption. For example, the compensatory taxes on the payroll of financial institutions in France and the Province of Quebec (Canada), both discussed later, are directly linked to the exemption (or zero-rating) for financial services and are an alternate form of tax on value-added in financial transactions. However, it should be rec- ognized that, in some instances, compensatory taxes on financial ser- vices have also been introduced in response to perceived weaknesses in the taxation of the income of financial service providers. Com- pensatory taxes in these cases are viewed as being in the nature of a minimum tax for income tax purposes. For example, the Canadian federal government and several Canadian provinces have taxes on the capital of financial institutions that owe their existence to a per- ceived failure to tax the income of such institutions appropriately in the past. The relationship of compensatory taxes to the exemption for financial services under VAT is thus often obscure or complex. Assessment of the Current System The discussion now turns to the problems and distortions arising under the current treatment of financial services. These can be 352 SATYA PODDAR grouped under three broad headings: economic, definitional, and al- location of input tax credits. It should be noted that, in general, these problems would have counterparts for any type of an exempt supply and thus are not unique to financial services. However, the significance of the result- ing economic problems caused by the exemption may be greater in the case of financial services than for other supplies. The financial sector plays an important role in national economies and has link- ages with other sectors throughout the economy. The impact of the distortions can thus extend to almost all other registrants. In addi- tion, competitive pressures in the financial sector are quite signifi- cant as a result of globalization and increasing international trade, meaning that distortions may translate into significant shifts in eco- nomic behavior and relative market position of individual products and their suppliers. Economic Distortions There are three types of economic distortion created by the exemp- tion structure. First, the effective tax rate falls below the statutory rate where the services are received by consumers. Second, there is tax cascading (a tax on a tax) where the services are received by businesses. Third, there is an incentive for financial institutions to produce inputs by themselves that would more efficiently be pro- duced by external suppliers (the self-supply bias) to take advantage of the exemption. There are also international trade implications and government revenue effects. Underlying each of these is the effect of exemption on the effec- tive tax rate of financial services. An exemption under a VAT is re- ally a two-edged sword, providing for a nil tax rate on supplies of the exempt good or service, but denying the deduction of input taxes related to the supply. This can have dissimilar results depend- ing upon where in the chain of supply the financial service is pro- vided. If the exempt good or service is supplied to a final consumer, the effective tax rate must take into account the input taxes paid by financial institutions that are not credited to them. In general, the effective tax rate is less than the standard rate for supplies to con- sumers, but it is not “zero.” For the effective tax rate to be truly zero, the supply would need to zero-rated, which allows for the full deduction of input tax credits. It is this fact that explains the choice of exemption as a compromise solution in the case of financial ser- vices, where the traditional rationales for nontaxation of regres- siveness and merit do not apply, but full taxation is not feasible on administrative grounds. CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 353 The effect is qualitatively different where the exempt good is sup- plied at some point in the supply chain before the final sale to con- sumers. In this case, the blocked input tax credits become embed- ded in the price paid by business purchasers, and tax cascading or tax on tax occurs. The prices of goods and services sold to final con- sumers would reflect not only the tax applicable at the point of final sale, but also the embedded, non-creditable input taxes. In this case, the exemption, rather than providing any relief, results in a net tax penalty. If full taxation is the desired objective, the system over- shoots this result, and the extent of the bias is determined by the size of the non-creditable input taxes. It also violates a fundamental tenet of VAT that tax should apply only on final consumption, not on intermediate transactions.4 The exemption applicable to financial services creates an incen- tive for financial institutions to self-supply inputs. This self-supply bias extends not only to direct labor services, but also to goods that require significant labor input. For example, in-house production of computer software by the financial institution will avoid the VAT that would otherwise apply if the software (treated as a good in cer- tain jurisdictions) were acquired from another supplier. This bias creates inefficiencies in the production and delivery of services by the financial sector. Imported financial services are generally not taxed by the desti- nation country, while the same service is usually zero-rated by the country of origin. Where the service is provided from a country without a VAT, it could effectively be zero-rated if there is no tax on the inputs to the financial service.5 As a consequence, imported sup- plies of financial services are typically free from VAT altogether. Be- cause domestic supplies of exempt financial services have embedded VAT, competitive inequities and economic distortions take place in the destination country. In other words, domestic financial institu- tions providing similar services will be less competitive than foreign suppliers, as their cost structure includes VAT. This creates a tax- based incentive for foreign supply of financial services that is unde- sirable from the domestic perspective. For example, financial insti- tutions in Europe are concerned about the effect on competition from the United States, which does not impose the burden of non- creditable input taxes on its financial institutions. While this may not have been perceived as a major problem in the past, it is an increasing concern as innovations in the delivery of services make foreign supply more feasible. The financial sector is a growth industry whose structure is changing through institutional consolidation such as mergers, new forms of alliance, and outsourc- ing. To accommodate these structural changes and support growth 354 SATYA PODDAR in the sector, many countries have taken steps to lower regulatory barriers to improve market access to international supplies of such services (for example, through foreign branch banking). Market de- velopments such as telebanking and internet banking have further facilitated this development. The supply of financial services with- out a physical presence in an economy is thus becoming common in many countries. Financial institutions are increasingly outsourcing the processing of their financial transactions to large service providers that can re- alize significant economies of scale by pooling the processing of transactions for several institutions. The service providers have suc- cessfully argued before the courts that these processing services in themselves constitute financial services and should not be subject to tax (for example, in the Danish case Sparekassernes Datacenter v. Slatteministeriet, and the U.K. case U.K. Customs and Excise Com- missioners v. FDR Ltd ). This exemption would be of limited conse- quence where the processing services are provided locally. The ser- vices would continue to incur non-creditable input taxes. However, where they are provided from a foreign jurisdiction without a VAT or from one where they are zero-rated, they will enjoy a tax advan- tage, relative to domestic services, which, assuming a VAT rate of 15 to 20 percent, could be as large as 5 to 10 percent of the total value of the services. The EC system of treating supplies of financial services between member states as exempt, as opposed to zero-rated, minimizes such competitive distortions, but only for trade within the Community. Moreover, even the trade within the Community is affected by the lack of uniformity in the tax rates and application of the exemption across the Community. As the Sixth VAT Directive is not directly applicable to member states, each member state has been required to define the boundaries of certain key terms and phrases within the confines of its tax legislation—which is closely bound up with na- tional civil and commercial law while respecting the Community law. The European Commission has been steadily trying to harmo- nize the categories of transactions that are exempt since the Sixth VAT Directive was implemented, but with only mixed success. Financial institutions operating at the Community level must con- tend with practical application of the VAT legislation in different member states. This creates legal uncertainty and barriers to intra- Community trade, while at the same time introducing potential plan- ning opportunities that distort behavior. In addition, tax authorities in individual member states must try to ensure that the tax is cor- rectly applied and collected with this distortionary force at work. CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 355 An important policy consideration in moving away from the ex- emption system is the potential impact on government revenues. Under a reform that introduced full taxation of financial services, the price of financial services to households would rise and the price to business registrants would fall. The net effect on revenues is thus an empirical question reflecting the extent of these offsetting effects. Huizinga (2002) has prepared estimates for the EC of the effect of introducing full taxation of financial services in the banking sector. This analysis found that revenues would rise by some 9.5 billion euros to 15.0 billion euros (about 0.1 percent of GDP), depending upon certain assumptions relating to potential economic responses to the change and the current level of taxation. An important con- sideration in this result for banking in the EC is that the tax col- lected on business services appears to be significantly lower than would be surmised by looking only at the size of the current tax base in respect of business services. Huizinga (2002, p. 516) ob- serves that, “The main reason must be that tax administrations cur- rently provide banks with almost full VAT input credits—against the spirit of the exemption system.” Overall, the indications are that under the exemption system the reduction in tax on consumer ser- vices is larger than the additional tax on business services. This means that there is a government revenue loss relative to full taxa- tion. However, this result may not necessarily hold in countries that currently levy significant compensatory taxes, which would be elim- inated upon the introduction of full taxation. Definition of Financial Services For the exemption system to operate, there must be a definition of what constitutes a financial service. There are number of difficulties and ambiguities involved in setting out and applying the necessary dividing lines. • Exempt versus taxable financial services: The VAT base typi- cally includes most supplies of goods and services on a broad basis and then excludes financial services as defined in the legislation. Whatever the definitions chosen, the actual system must make fine distinctions between taxable and exempt services in a large number of potential situations. For example, the Canadian Customs and Rev- enue Agency lists some 240 types of transactions for trust companies alone and divides them into taxable and exempt categories. This issue has become even more of a problem area because of deregulation and globalization. Traditional products are being un- bundled into their component parts and bundled into new products. 356 SATYA PODDAR As part of this process, financial transactions are being out-sourced to third parties, creating difficult questions as to whether these sup- plies are taxable or exempt. The dividing line between exempt fi- nancial services and taxable activities is thus becoming more prob- lematic and blurred. It is useful to note that virtually all the problematic cases are those in which the fee for services is explicit, rather than implicit in the form of the margin. The latter are always financial services that are exempted from tax, with the possible ex- ception of property and casualty insurance, which is taxable in cer- tain jurisdictions (such as New Zealand). The former can include a wide variety of services, including financial advisory services, wealth management services, trustee services, safekeeping and custodial, share registry, data/transaction processing, appraisals, debt collec- tion, financial counseling, actuarial, legal, accounting services, and insurance evaluation. Many of these services are taxable when pro- vided in isolation (financial advice), but exempt when provided in conjunction with a financial transaction (such as financial advisory services during the course of purchase or sale of financial securities). It is this dual treatment of these services that leads to complexity in the determination of their status. • Financial services supplied by non-financial businesses: Non- financial businesses often provide financial services as part of their activities. For example, retail stores may operate credit operations. For neutrality, such operations should receive the same treatment as similar operations carried out by financial institutions. However, to avoid complexity and allocation problems, some de minimis thresh- old will usually be used to ignore some incidental financial service activities, such as payment of interest and dividends. In applying the trade-off between neutrality and compliance simplification objec- tives, there will inevitably be some problems. • Services incidental or supplementary to financial services: The provision of a particular financial service will often have attached to it services of a taxable nature, or vice versa. For example, banks will often provide a combination of taxable payroll services, and exempt checking and other banking services. VAT systems must have rules in place to deal with these types of situations as they create diffi- culties in compliance or administration and provide opportunities for tax planning. • Mixed supplies: Financial services and non-financial services may be provided together for a single consideration. For example, credit card services, in addition to the basic credit authorization ser- vices, may include such features as “air mile” entitlement for re- demption on travel or goods, extended warranties for goods pur- chased, and rental car collision insurance. For comparable treatment CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 357 to the underlying components being purchased separately, there would need to be an allocation of the price among the components of the supply. Lawmakers and administrators must decide where to strike a balance in such situations between complex allocation re- quirements and formulaic allocations that leave open possibilities for tax planning. Input Tax Credit Allocation Wherever there is a co-mingling of exempt financial services with ei- ther taxable or zero-rated activities, registrants for the tax will need to allocate sales as to type in order to determine the extent of input tax credit entitlement. The two broad approaches used are direct attribution, in which the actual use of inputs is used to determine eligibility for input credits, and formula allocation, in which some readily available measure of exempt versus taxable or zero-rated rev- enues is used to determine right of deduction. However, there is con- siderable variation in the specifics of the input tax credit allocation, even among countries using the same broad approach. Whatever the rules used, allocation is the source of considerable complexity. In order to carry out the allocations, financial businesses may need to collect information and modify accounting systems that would not otherwise be required. In addition, planning opportuni- ties will inevitably exist that may cause changes in behavior. Change- of-use rules to handle changes in the use of fixed assets between ex- empt versus taxable activity also necessitate considerable additional record-keeping. Sales between related parties become problematic when the tax collected may not be creditable. The incentive is to minimize prices in sales between related parties and there are nor- mally arm’s length pricing rules to prevent this practice. However, as with transfer pricing issues under the income tax, the application of such rules introduces difficult administrative issues. Problems with non-deductible input tax credits on sales among affiliates within a corporate group provide an incentive for in-house supply. This may lead to tax-induced reorganizations or tax authorities may allow some form of group relief or consolidation in response to the issue. Financial Services: What Should the Tax Base Be? Like any other business, the provision of financial services requires the input of purchased goods and services, labor, and capital. A value-added tax is designed to tax final consumption expenditures that reflect the value-added through labor and capital inputs at each 358 SATYA PODDAR stage in the production of a final consumer good or service. It is im- portant to be very clear as to what this tax base consists of, in the case of financial services. As noted, where the consideration for a financial service does not take the form of explicit fees or commissions, it is hidden in a set of cash flows and is not readily observable. The cash flows under a fi- nancial transaction can have four elements: the principal amount of a financial instrument, the pure time value of money, the risk pre- mium, and the charge for intermediation services. The principal amount represents the capital value in the transac- tion. For example, it would be the amount deposited by a customer in a bank account or the amount loaned by a bank to the borrower. It is an amount that does not represent a permanent inflow or out- flow to the intermediary. It is a cash flow that is expected to reverse eventually at the conclusion of the transaction (for example, the de- posit will eventually be withdrawn or the loan repaid). The pure time value of money is that component of cash flows that reflects the opportunity cost of funds in the absence of any credit, market, or other such risks. It is the element in financial transactions that must be excluded to maintain the fundamental property of a consumption tax: namely, that the pre-tax and post- tax ratio of current to deferred consumption be maintained. The risk premium is a charge reflecting compensation to the in- termediary for the risks inherent in the transaction (such as an in- terest charge on a loan over and above the pure time value of money, reflecting the expected value of default by the borrower in loan repayments) or an outflow on account of risks (such as claim payments by an insurer). This risk premium should also be excluded from the base for a consumption tax, as it does not represent a net value added in the economy. To the extent that the expected risk premium reflects actual future loan losses or insurance claims, the financial flow associated with the risk premium will not be realized or will be paid out later. They thus do not represent consideration that will be retained by the financial service provider as a payment for value-added. It is a form of redistribution of wealth, either be- tween the financial institution and its customers, or among various participants in a risk-pooling arrangement. The intermediation charge is the residual component of any cash flows from a financial transaction. As the name suggests, it is only this last component that represents a consideration for financial ser- vices and that should attract VAT. The challenge lies in isolating this component for each individual transaction, in a manner that makes it practical to apply the VAT to it, and that is consistent with the application of VAT to other goods and services. Methods for doing this and further observations about the pure time value of money CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 359 and measurement of risk in this context are presented in the later section on methods of full taxation. The discussion in the previous section identified a variety of eco- nomic, definitional, and input tax credit allocation problems that can be associated with the current exemption system. Financial in- stitutions have raised concerns about the detrimental impact of the exemption on their competitiveness because of the large volume of blocked input tax credits associated with financial services provided to business customers. If the rationale for the exemption relates to difficulties in application of the tax rather than considerations of regressivity and social policy needs, and the current exemption sys- tem is perceived to be distortionary, it would appear that taxation would be desirable—if a practical approach to taxation could be identified. Indeed the European Commission has been seeking to identify and analyze taxation options that could replace the current system. The objective is to identify solutions that move toward full taxation of financial intermediation services as much as possible, consistent with other policy and operational concerns. However, some of the underlying premises in the rationale for full taxation of financial services are being debated, as implied in the above discussion. This debate is diverse and contains several strands. Before getting into options for revised approaches to taxation of fi- nancial services, it is useful to identify the underpinnings of these ar- guments, if only briefly. Since the considerations involved are varied, complex, and open to dispute, no attempt is made to provide a com- plete description or firm conclusions. However, the issues raised do provide some background relevant for the discussion of options. The most extreme form of challenge to full taxation as an option is the argument that financial intermediation services do not affect the consumer utility function, acting only to facilitate a change in the time profile of other consumer goods and services that do enter the utility function.6 Under this viewpoint, if only final consump- tion goods should be taxed under a VAT, and financial services are intermediate goods that do not provide any utility in themselves, fi- nancial services should be exempt from tax. This view has been challenged by Auerbach and Gordon (2001), who argue that a VAT is equivalent to a tax on the income of all primary factors that enter into the production of final consumer goods. For neutrality under this tax base, all income of primary factors, such as labor, that enter into the provision of financial services must also be taxable. An argument reaching a similar conclusion as to the fallacy of non- taxation based on financial services not entering directly into the utility function is provided in the discussion of aspects of optimal taxation of financial activities by Boadway and Keen (chapter 2, this volume). 360 SATYA PODDAR Jack (2000) has argued that the requirement that the price of all consumer goods should increase by the same percentage under a VAT may not require imposition of VAT on certain financial ser- vices. If the prices of other goods rise by the percentage of the VAT, and the prices of financial services are proportional to the size of nominal transactions, neutrality would require no direct taxation in such cases. On the other hand, where the financial services are of a fixed amount, taxation is appropriate. Building on this viewpoint, it can be argued that a modified system of exemption or zero-rating for margin services coupled with taxation of fees or commissions may be more appropriate than full taxation.7 Auerbach and Gordon show that this is erroneous if the tax is viewed as being equivalent to one on all primary factors of income. However, the discussion by Boadway and Keen (chapter 2) lays out the argument that, if the objective is an optimal consumption tax, the approach suggested by Jack may have some validity. In line with this, a final set of arguments expressed in Chia and Whalley (1999) and discussed by Boadway and Keen (chapter 2) is that an optimal consumption tax might apply low tax rates to fi- nancial services, but not necessarily zero tax rates. As this argument is really one for a different consumption tax base than contemplated under a VAT, it goes beyond the scope of the discussion here. Support for retaining the exemption system may also be based on the view that certain costs involved in the operation of full taxation may more than offset any benefits gained by elimination of the dis- tortions discussed earlier in this chapter.8 The discussion in the remainder of this chapter is based on the more conventional view that the VAT is designed to be a tax con- sistently applied to all the inputs that contribute to value-added. In this case, the objective of the tax is to fully tax financial services supplied to nonregistrants, and to ensure that taxation of supplies of financial services to business registrants does not result in tax cas- cading. The issue then is the identification of options that achieve the desired results without excessive administration or compliance costs, and the introduction of new distortions that may offset any benefits from improvements on the exemption structure. Taxation of all financial services to nonregistrants is consistent with the practice for non-financial goods and services. It does not matter if the nonregistrant is using these goods in some productive activity. As long as the person is not registered, the supply is taxable and there is no recovery of input tax credits available. The fact that the financial supply is supporting rearrangement of consumption patterns should not matter. Other services of a real nature such as transportation and storage costs also help in rearranging consump- CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 361 tion patterns, but are subject to tax. The key identifier is that sup- plies to households that are not registered as businesses are taxable, whatever their nature or subsequent use.9 This is a fundamental characteristic of credit-invoice VAT systems. Policy Options Against this background, three general approaches to policy options to replace the current exemption system can be considered. The first, the full taxation approach, would attempt to isolate the financial in- termediation consideration where it is hidden in financial margins and apply tax to it, as well as to any explicit fees and commissions charged in respect of financial services. The second approach in- volves alternatives modifying the exemption system by extending full tax to selected financial services. The final option attempts to apply the tax to value-added in financial services through compensatory taxes by use of the addition method of value-added taxation.10 The options that will be reviewed here include the following full taxation options: • Basic cash flow tax • Tax Calculation Account (TCA) system • TCA system with zero-rating of business transactions Also considered will be a variety of modified exemption systems: • The option system • Taxation of explicit fees and commissions • Taxation of agency services only • New Zealand system • Exemption with input credits • Australian system Finally, the French, Israeli and Quebec systems of compensatory taxes using the addition method will also be discussed. In modifying the exemption system, a question arises as to how one can evaluate whether an extension of coverage is “good” or “bad.” It is possible to enumerate some principles or benchmarks that can aid in this process. These principles and benchmarks are also relevant in assessing ad hoc systems of indirect taxation being adopted by developing countries for financial services for their rev- enue needs. First, the base for the tax should be no more than the financial in- termediation service. Ad hoc levies are particularly prone to violate this benchmark. For example, a premium tax on gross insurance pre- 362 SATYA PODDAR miums has an aggregate base of total premiums, rather than the in- termediation service base, which is premiums less claims. Second, modifications to the exemption system should lessen the tax on busi- ness inputs (that is, avoid tax cascading). This is one of the princi- ples that the exemption system and certain of the options being con- sidered fail to satisfy. Third, new approaches should lead to broader coverage of financial services. The narrowing of the tax base is one important source of the economic distortions and other problems with the exemption approach. Fourth, any new approach will raise compliance/administration issues of its own and may introduce new types of distortions. Only new approaches that are improvements in this respect should be adopted. Finally, new approaches must not be susceptible to tax planning or behavioral changes that effectively undo any improvements in respect to the other objectives. For ex- ample, extending taxation to a particular fee or commission may ap- pear to broaden coverage and not run afoul of any of the other guidelines. Nonetheless, it is undesirable if the tax can easily be cir- cumvented by converting the fee into a nontaxable financial margin. Full Taxation The earlier discussion described the fundamental problem with the application of VAT to financial services: the difficulty in identifying the appropriate consideration hidden in the margin created by a set of four different types of cash flows (principal, time value of money, intermediation services, and risk premium). The consideration for financial services can still be measured by the net cash flows re- ceived by a financial institution in respect of any given transaction. However, for purposes of applying a VAT, this value must be com- puted on a transaction-by-transaction and customer-by-customer basis to allow input credits where claimable. Under the normal VAT system, there is no apparent mechanism for allocating this aggregate measure of the value of financial services to individual transactions. It has generally been accepted, following from the discussion in the Meade Committee Report (Institute for Fiscal Studies 1978), that the basic cash flow method provides a means of measuring the value of margin services on a transaction-by-transaction and customer- by-customer basis. This report, which fully developed the so-called F-base for taxation of business services, is considered to supply the necessary conceptual definition of a tax base for transactions in fi- nancial instruments. The Tax Calculation Account (TCA) methods described below are operational approaches to achieving the results of the basic cash flow method, while avoiding certain problems as- sociated with it. CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 363 Basic cash flow tax Under the basic cash flow method, all cash in- flows from financial transactions (whether of income or capital na- ture) received by a financial institution would be treated as consid- eration in respect of taxable sales, on which VAT must be remitted to the government by the financial institution. All cash outflows (whether income or capital) paid by a financial institution would be treated as taxed purchases, for which the financial institution would be entitled to a refund of VAT from the government. Under this approach, transactions with nonresidents can be read- ily zero-rated simply by applying the zero rate of tax to cash inflows and outflows from and to nonresidents. Where the customer is a business that is a VAT-registrant, all cash inflows received by the business customer would be treated as taxable sales, on which VAT would be remitted to the government by the customer. All cash out- flows paid by the business customer would be treated as taxed pur- chases, for which the customer would be entitled to a refund of VAT from the government. The cash flow method automatically arrives at an appropriate treatment of risk on an ex post basis. For example, on a loan, the cash outflows generate a refund of VAT. If the loan leads to a bad debt, there is never any inflow of cash and there is no offsetting tax collected. In effect, there is full loss offsetting in respect of the actual losses associated with credit risk under the cash flow method. Inter- est payments on the loan are fully included in the tax base, even though they include a risk premium. This system of recognition of risks obviates the necessity of having to estimate the risk premium included in cash inflows and taken out of the tax base. Risks are re- moved from the base by giving a deduction for the losses or cash outflows (for example, in the form of insurance claim costs) suffered by the financial institution because of risks. If actual risks turn out to be smaller than the anticipated risks built into the pricing of the financial services, the excess is taxable as profit, which is a compo- nent of the value-added by the financial institution. In the same manner, any excess of actual risks over anticipated risks gives rise to a reduction in profits and is treated as a reduction in value-added. The basic cash flow method would result in VAT being applied to financial services in a manner that is consistent with the normal VAT system and would correctly allocate the VAT base between de- positors and borrowers. However, several problems have been iden- tified with respect to the application of the basic cash flow method to financial services, as noted below: • Since the principal amount of a loan would be subject to tax at the time the loan is made, borrowers would have to be able to 364 SATYA PODDAR find financing for the tax, in addition to the original loan require- ments. This could create additional borrowing requirements and present cash flow problems. • Significant transition problems would arise under the basic cash flow method upon implementation of the system and whenever the tax rate changed. The correct amount of tax is levied under the basic cash flow method only where all of the cash flows associated with a loan or deposit are subjected to tax. If cash flows arising be- fore the date of implementation are not subject to tax, then the appropriate amount of VAT will not have been collected. For ex- ample, if a deposit is received before the implementation date, no VAT would be remitted in respect of the deposit. If the deposit is withdrawn after the date of implementation, then VAT would be re- funded on the deposit even though VAT was not paid when the de- posit was accepted. • Due to the volume of cash inflows and outflows between fi- nancial institutions and their customers, applying VAT to all capital and income cash flows could be burdensome to both financial in- stitutions and their customers. These problems have generally been perceived as rendering the basic cash flow method nonoperational, despite its theoretical at- tractions. Despite these difficulties, the cash flow method is a rela- tively simple way of applying VAT to a variety of financial transac- tions, such as insurance and financial derivatives. It is also the most akin to the normal VAT for non-financial goods and services. If full taxation were introduced, it is likely that financial institutions would choose to use this method, if available, for certain of their transactions that are not subject to the transitional issues: that is, those transactions that do not straddle the commencement of the system or a change in the tax rates. Tax Calculation Account system The Tax Calculation Account (TCA) system was developed as an option that would mirror the re- sults of the cash flow system, while providing mechanisms to sur- mount the transition and liquidity problems associated with the full cash flow system.11 It essentially works via a tax suspension account for margin transactions, with all the necessary calculations handled by financial institutions. Under the TCA system, the collection or crediting of tax on cash inflows or outflows of a capital nature is suspended. The amount of the tax or credit suspended accrues interest at a rate (referred to as the indexing rate or the cost-of-funds rate) reflecting the pure time value of money. The suspended amount of tax or credit generally re- verses at the end of the transaction when the capital flows reverse: CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 365 for example, when a loan is repaid or the deposit is withdrawn. The net tax payable at the end of the transaction is thus the tax (credit) applicable on the interest received (paid) minus (plus) the interest at the indexing rate on the deferred tax (credit) amount. This system is equivalent to defining the consideration for the principal financial services as interest (or other forms of income or expense flows) re- ceived (or paid) minus the interest calculated at the indexing rate, which is the pure time value of money. The tax calculation account provides a means to make the neces- sary adjustments to avoid the transition problems referred to earlier. Under this system, an opening entry is created in the TCA for the tax (credit) amount for any balances outstanding for principal transac- tions (for loans and deposits, for example) at the time of commence- ment of the system. The opening entry is equal to the tax that would have been collected or credited on the principal amount, had that transaction taken place at the commencement of the system. Once this opening entry is created, all subsequent calculations for the ex- isting transactions would be identical to those for new transactions. The TCA calculations are needed for only those activities in which the financial institution is acting as a principal and the consideration takes the form of margin. Where the consideration takes the form of an explicit fee or commission, the tax could be applied to it directly, in a manner identical to that for other goods and services. Risks under the TCA system are handled in the same manner as under the cash flow system: that is, they are recognized on an ex post basis when they occur. The taxable margin for a loan is re- duced by any bad debts. As and when a loan becomes a bad debt, the taxable margin becomes a negative amount, for which the fi- nancial institution claims a tax credit. During 1995–96, the TCA system was tested at ten large finan- cial institutions in six countries in Europe, under a project spon- sored by the European Commission. The pilot tests involved a re- view of virtually all the major financial products and services offered by the institutions, their input tax credit allocation systems, and administration and compliance costs of the current and the TCA system.12 A number of design issues received special consider- ation during the testing of the system, and two key ones can be summarized here. First, under the TCA system, the indexing rate plays a critical role in the measurement of consideration for financial services. Should the indexing rate be a single indexing rate for financial transactions of all maturities or one that is matched to maturity? After extensive deliberations, use of a single, short-term, interbank rate was found to be appropriate in all cases. This allowed tremen- 366 SATYA PODDAR dous simplification of the system, which would have otherwise been unworkable. The second issue related to the impact of risk on the measure- ment of consideration. As was indicated earlier, an element included in the margins of financial institutions is a risk premium, and a pure risk premium should not attract a value-added tax. The risk pre- mium does not reflect any added value in an economy; rather it re- flects a redistribution of value among the market participants as market risks play out. While the existence of a risk consideration in pricing is not unique to financial services, it is a much more funda- mental and pervasive consideration in pricing for many such ser- vices than it is for real goods and services. The appropriate treat- ment of the risk premium can be achieved under the TCA system by including all interest charges in the tax base, and allowing the fi- nancial institution a tax credit for the bad debts realized by it. This gives the same loss offsetting result as is provided under the full cash flow method that was discussed earlier. The pilot tests confirmed that negative balances in the TCA account should be treated sym- metrically to positive ones in applying the VAT: that is, they should give rise to a VAT refund or credit.13 Overall, the tests confirmed that the TCA system was conceptu- ally robust and resulted in proper application of VAT to all finan- cial products and services. Financial institutions, however, did iden- tify a number of concerns and apprehensions about the system. The main concern identified in the tests was the newness of the concept and the time and effort that would be required by financial institu- tions to familiarize their staff and customers with it and to modify their computer system for the TCA computations. Another major concern was that the computed tax amount on a given transaction would reveal information on margins earned by financial institu- tions. This information would affect market competition, be subject to misinterpretation, and lead to time-consuming explanatory dis- cussions with customers. TCA system with zero-rating of business transactions To address the concern noted above about the implementation costs of the TCA system, an option was identified that would involve zero-rating of principal financial transactions with other VAT registrants. This ap- proach obviates the necessity of any TCA calculations for transac- tions with business customers, who account for the vast bulk of transactions for many financial institutions. For the remaining trans- actions with final consumers (or VAT nonregistrants), the TCA com- putations could be done on a global basis, assuming that such cus- tomers would not need any statement of the VAT included in their CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 367 charges to the financial institution. The method would maintain the overall conceptual integrity of the TCA system, while resulting in simplification of compliance and administration. The zero-rating of financial transactions with business customers would not result in revenue loss to the government, assuming that any tax otherwise collected on such transactions would have been fully deductible by the customers. The principal compliance issue beyond the calcula- tion of the necessary TCA in respect of services provided to nonreg- istrants would be the need to determine the tax status of customers in order to assign the appropriate tax status. This approach would also mean that pricing would need to be done on a tax-exclusive basis for business customers and a tax-inclusive basis for nonregis- trants, which could lead to some confusion among customers. Modified Exemption Systems Another general approach that could be taken would be to maintain the exemption system, but to modify it in ways that are designed to improve its operation. The two principal avenues for improving the operation of VAT for financial services are to reduce the degree of taxation of related business inputs and to broaden the range of fi- nancial services to consumers that are taxable. Several options in both these regards can be identified. One possibility to reduce the taxation of business inputs would be to make the “option” that is in place in certain countries in the European Union more widely avail- able. The tax coverage can also be broadened by extending the tax to agency services—as is done in Singapore—or to all explicit fees and commissions. Another type of approach is to make selected use of cash flow tax methods to tax selected types of financial services most amendable to this treatment, as is done for property and casu- alty insurance in New Zealand. Ad hoc methods to reduce the level of blocked input tax credits in respect of business customers are also being tried in Singapore and elsewhere. The Australian system, under its Goods and Services Tax (GST), is briefly described at the end of the section as a system that has adopted a set of these approaches. The option system In the EC, member states are allowed to extend an option to financial institutions to be taxable. The option is po- tentially of interest to financial institutions that deal largely with commercial clients where the customers will be able to claim input tax credits. Currently, Belgium, France, and Germany provide for this op- tion. The details for the option vary from country to country. In France, banks, financial institutions, and persons performing finan- 368 SATYA PODDAR cial transactions may elect to be subject to VAT. The election is ir- revocable and covers all revenues that fall into the scope of the elec- tion. The election mainly covers revenues from credit transactions (other than interest), such as filing charges; revenues from securities transactions undertaken for the benefit of the client; and particular commissions and brokerage fees. The option has not been widely used. It also has not been devel- oped systematically. For example, the concept of the base has not been well defined. Lacking a measure of financial intermediation in margin transactions, the base is usually considered to be some gross amount of cash flow. The option does meet the criteria of removing tax from business inputs, but the conditions for its use have tended to be restrictive and it is attractive only to firms with mainly business customers. Its uneven availability and variations in its structure have been a source of intra-Community distortions. In its current form, it is thus only a modest and partial solution to the problem of taxation of business inputs, but it does move in the right direction. Taxation of explicit fees and commissions Another approach re- sults from a reexamination of the issue of substitutability of consid- eration in the form of margin for explicit fees and commissions. This substitutability was noted at the outset as a reason for exempting many services primarily priced as a fee or commission. If concerns about substitutability are no longer valid, taxation of all explicit fees and commissions for financial services would be feasible. The two key factors forcing financial institutions into operating on a fee or commission basis are disintermediation and deregula- tion. At one time, a bank would offer a deposit account with low or zero interest rates to customers. The low interest rate would yield a margin, which would cover the costs of a set of financial services to customers. Under disintermediation, customers who do not require certain services are reluctant to sacrifice interest on their deposits. They choose to earn interest via financial products offering more competitive rates and pay for only those services they need on a sep- arate basis. Similar pressures exist to unbundle services related to borrowing with lower interest rates charged and specific services paid for separately. Less regulated markets have added to the pres- sures on the use of margin pricing by fostering competition and al- lowing the development of efficient, stand-alone services provided by third parties that can be priced only on an explicit fee basis. As a result of disintermediation and the unbundling of prices for fi- nancial products and services, fees and commissions account for an increasing proportion (as much as 80 percent) of total revenues of financial institutions. These same forces mean that there may be a CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 369 more limited risk of substitution of margin for fees than was as- sumed to be the case in the past. If one accepts that such substitution has become a less serious threat, alternative approaches can be identified that reduce the over- all distortions in the system markedly. One such approach would be the taxation of all explicit fees and commissions. This would signif- icantly increase the breadth of coverage of taxation of financial ser- vices, and, to the extent the preceding observations are correct, would lead to little tax-induced switches to margin pricing. Under this approach, exemption for financial services would be limited to consideration received in the form of a principal amount, interest, dividends, and other similar amounts (such as blended cash flows from financial derivatives) from transactions undertaken by a prin- cipal. Exemption would thus be strictly limited to consideration re- ceived in the form of margin. This option can be seen to score very well in relationship to the principles for assessing alternative approaches enunciated earlier. It applies only where the value of the service is identified in explicit form and thus applies to the appropriate base. It would reduce tax cascading where financial services are provided to businesses on a fee or commission basis. Indeed, it could encourage the trend to conversion of financial service margins into fees for business cus- tomers, which would lead to less cascading. It would provide a sub- stantial increase in the taxable coverage of financial services under VAT. Of course, there would be an incentive to shift consideration for services supplied to consumers into still-exempt margins. However, this will be no worse than the status quo, although it would mean that revenues that might be anticipated from the tax on fees and commissions would not materialize. It would also be possible to ex- clude specific types of transactions from tax where conversion to margins is most problematic, such as foreign exchange dealings. Al- ternatively, the conversion of fees to margin could be discouraged by deeming the margin earned on certain types of transactions (such as securities bought and resold to fulfill prior customer orders) to be fees. It would also provide significant benefits in reducing or eliminat- ing several of the distortions under the exemption system identified earlier. Assuming that explicit fees account for a significant portion of total revenues of financial institutions, the quantum of input taxes blocked as a result of the remaining exempt supplies should be reduced to a modest fraction. As a result, the self-supply bias for fi- nancial institutions would not be as serious under this approach. With the taxation of all explicit fees and commissions, a simplified formula for input tax allocations could be devised for financial in- 370 SATYA PODDAR stitutions that would be less complex and distortionary than the current structure. In the case of the EC, intra-Community biases would be reduced, as the list of exempt services and the input tax allocation formulae would be more consistent across member states. Another advantage of this system is that it would eliminate virtually all disputes about the taxable or exempt status of certain services, such as for contracting-out arrangements for financial transaction processing services. The potential tax in individual out- sourcing contracts can amount to tens of millions of dollars annu- ally. Examples of contracting out of financial services by financial institutions to specialized providers of services can be found for bro- kerage, advisory, and processing and management services, such as credit card processing data centers. This issue, as illustrated in the court cases Sparekassernes Datacenter v. Slatteministeriet and U.K. Customs and Excise Commissioners v. FDR Ltd. has assumed con- siderable importance. Before these cases were filed, outsourced ser- vices were generally held to be subject to VAT. European jurispru- dence in these cases invoked the principal that the characteristic of the service should not be changed by the contracting out of the ser- vice to a third party. While this removed the bias against outsourc- ing to local suppliers, treating them as exempt creates a huge incen- tive to outsource them from countries without VAT or in countries where the supply of exported services are zero-rated. Because such contracting out arrangements invariably involve the use of agents, the consideration is inevitably in the form of fees and commissions. Therefore taxation of all fees and commissions would address this issue. Taxation of explicit fees and commissions could be supplemented with an option given to financial institutions to apply the tax to the margin services as well. The application of tax to the margin ser- vices could be based on the TCA system. This approach could be viewed as an intermediate step to transi- tion to the full taxation system, as under the TCA system. Again, to minimize compliance burden of TCA calculations for each individ- ual transaction, principal financial transactions with VAT registrants could be zero-rated under this option-to-tax approach. This would allow financial institutions to compute the tax on their retail trans- actions on a global basis at the end of a tax period. In return, they would be able to claim full deduction of their input taxes. For those financial institutions that do make use of the option, this system will result in elimination of virtually all economic distortions (including tax cascading) that arise under the current exemption system. Once financial institutions become familiar with the TCA sys- tem, consideration could be given to making its use mandatory. This CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 371 would be a nonintrusive way of bringing financial services fully within the VAT system. Taxation of agency services only Singapore has adopted a variant of the previous approach under which it taxes only agency services. Examples of taxable services are brokerage for executing transac- tions for the sale and purchase of securities on behalf of customers; brokerage for life or general insurance; premiums of general insur- ance; and merchant banks’ fees on corporate restructuring. All prin- cipal services remain exempt, regardless of whether consideration is explicit or implicit. As far as agency services are concerned, this approach merits many of the same comments as the taxation of all fees and services. Where it applies, it employs the correct tax base and broadens the coverage. It also reduces certain distortions. For example, it ad- dresses disputes about the status of certain contracting out arrange- ments. Finally, it restricts additional taxation to those areas where substitutability of margins for fees is least likely. The main drawback of this approach is its limited coverage; it leaves the exemption system in place for all principal transactions. In these cases, it does not respond to tax cascading, broaden cover- age, or deal with the distortions of the exemption system. This ap- proach is an improvement on the current system and would be at- tractive for countries where it is believed that substitutability is likely to be a serious problem in the case of principal transactions. Relative to the taxation of all fees and commissions, the desirability of this approach depends upon whether the distortions associated with substitutability outweigh the new distortions arising from the differential treatment of principals and agents identified in the pre- vious paragraph. New Zealand system New Zealand applies its Goods and Services Tax to a comprehensive base, but again excluding financial services. The tax applies to general insurance but does not extend to life in- surance, creditor protection policies, and other financial intermedia- tion services. For general insurance, the New Zealand system is in most respects modeled after the basic cash flow structure.14 Overall, the New Zealand system for financial services is not much different from the basic exemption system prevailing in Eu- rope and other jurisdictions, and is subject to the same economic distortions and concerns as elsewhere. Exemption with input credits Conceptually, input credits are ap- propriate in a VAT where a good or service is taxed at the normal 372 SATYA PODDAR rate or a zero rate. However, in order to eliminate or alleviate the cascading effect of the exemption method, countries such as Singa- pore and Australia are allowing input credits in certain cases even under the exemption system. As has been described, fee-based financial services provided by financial agencies in Singapore are taxed, while most principal fi- nancial services remain exempt from VAT. However, in order to re- duce the cascading effect of the exemption method, Singapore al- lows financial institutions to claim input tax credits under either of two approaches: the “special method” or “fixed input tax recovery method.” Under the special method, financial services that were provided to VAT registrants are treated as if they were eligible for application of a zero rate. Therefore, in order to get this benefit, financial insti- tution needs to segregate the eligible financial services out of the total services that have been provided to all customers. This ap- proach is equivalent to the zero-rating of financial services provided to registrants, which was described earlier as an adjunct to the TCA method to make it less of a compliance burden. Under the other method, the fixed input tax recovery method, a financial institution can claim a credit for a fixed percentage of total input taxes. The recovery percentages are differentiated according to the type of financial institution involved. While this method does provide some relief for tax cascading in respect of services provided to business customers, it moves the coverage of the tax even further from full taxation in respect of services provided to nonregistrants. This fixed input recovery method is not seen as being a permanent part of the system. Presumably, it is intended that as experience is gained with the special method, input credit recovery may eventu- ally be limited to that approach. Australian system The Australian GST is a broadly based value- added tax that provides an exemption for financial services. How- ever, it extends taxation to financial agency services and to non-life insurance and provides for some recovery of input credits in regard to financial services. The definition of financial services is restricted to transactions performed by principals. Therefore such services as brokerage not undertaken as a principal are subject to tax. Insurance, other than life insurance, is also taxed. The strict exemption approach is modified somewhat by allow- ing a 75 percent credit for GST paid on a defined list of services ac- quired for use in making exempt financial supplies. This was done in an attempt to counter the self-supply bias created through the ex- CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 373 emption of financial services. The percentage for the credit allowed was chosen so that the noncreditable tax on acquired services was approximately equivalent to the noncreditable tax that would have applied on self-supplied services. Conceptually, the Australian system does address many of the problems of an exemption system. However, it is yet to be deter- mined how the complexity of its design (attributable primarily to the definition of services that are eligible for the 75 percent credit) would affect its operation. Compensatory Taxes Using the Addition Method The aggregate financial intermediation services of a financial insti- tution can be measured readily. This can be done by using any of the addition, subtraction, or cash flow systems. However, only the cash flow or the TCA systems can function on a transaction-by-transac- tion basis and be compatible with a credit-invoice VAT. Neverthe- less, in the absence of an operational method to tax financial ser- vices fully, it is possible to use one of the aggregate methods to bring some or all elements of such services into the tax net. Systems of this type, employing different forms of an addition tax, exist in France, Israel, and the Province of Quebec in Canada. This section looks at the features of these types of compensatory taxes as adjuncts to a VAT in the real sector. Compensatory taxes using the addition method are designed to tax the full value of financial services, or at least, a larger propor- tion of the full value than an exemption system. A major objective of compensatory taxes is to raise revenue in respect of activities that are seen as not being subject to full taxation. Conceptually, if non- labor inputs are already taxed under the exemption system, then supplementary taxation of labor used to supply financial services and the profits from supplying such services will achieve full taxa- tion of the value of financial services.15 However, these compensa- tory tax systems have major implications for tax cascading, incen- tives for self-supply, and many of the other issues that have been discussed in respect of the taxation of financial services. The fol- lowing discussion highlights the structure of three such taxes and briefly explores key implications of the approaches taken. French system Employers established in France must pay a payroll tax if at least 90 percent of their turnover (including any turnover that is outside the scope of VAT) is not subject to VAT. Therefore most financial institutions are subject to this payroll tax on all or a part of their gross salaries. The taxable portion of their salaries is 374 SATYA PODDAR equal to the difference between 100 percent and the percentage of the turnover that is either subject to VAT or zero-rated. Rates of payroll tax vary from 4.25 to 13.6 percent. The payroll tax is designed to bring the labor component of value- added into the tax base. As far as financial services supplied to households are concerned, the combination of blocked input taxes and the payroll tax on labor means that the tax base moves closer to the full taxation of financial services. It also eliminates the self- supply bias for financial institutions. A major disadvantage of the approach is that it increases the ex- tent of tax cascading in respect of business inputs, because the quan- tum of tax collected at the financial institution level now includes both blocked input taxes and the payroll tax on labor inputs. The French system does have the “option” in place that can provide re- lief in certain cases, but its reach is limited. Another undesirable feature of compensatory taxes is that they are applied and collected at the time the costs are incurred and not when the consumption takes place. If the labor being taxed is used to support financial services that will be delivered to consumers in future periods (as for example when it relates to manufacture of capital goods, such as computer software), tax is collected in the current period, rather than at the future time when consideration is actually received for delivering the service to the final consumer. This is tantamount to collecting tax on the time value of money, a characteristic of an income tax, not a consumption tax. Israeli system Israel applies an addition-method VAT to insurers and deposit-taking institutions. These institutions cannot claim a credit for input VAT. Credits are also not available to registered cus- tomers purchasing financial services from financial institutions. The Israeli system is thus similar to the French system, but also includes the profit component of value-added in the tax base. The comments about the implications of the Israeli system largely mirror those of the French system, but apply with even greater force because of the taxation of profits. A further issue that arises in this form of addition-method com- pensatory tax is the compatibility of the profit measure with a con- sumption tax base. Taxation of the time value of money will occur if an income tax measurement of profits is used to determine the base. To prevent this, the profit component must be adjusted in a number of ways to reflect consumption tax principles. Most no- tably, it must be recomputed on the basis of immediate expensing of capital. CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 375 Quebec system The Province of Quebec in Canada has adopted a different approach for taxation of financial services under its provin- cial VAT. Quebec zero-rates financial services, but applies supple- mentary taxes on labor and capital of financial institutions (and on insurance premiums). Revenue considerations are certainly an important rationale for this approach. In Quebec, the government explicitly stated that the taxes were being imposed to offset, at least in part, the revenue loss from zero-rating. The resulting system is a unique one with unusual features. The design neither tries to capture the full value of finan- cial services (due to zero-rating) nor does it attempt to fully imple- ment a zero-rating system. As far as the taxation of labor and capital is concerned, the com- ments are similar to those for the French and Israeli systems. How- ever, the combination of zero-rating of financial services and the ap- plication of the payroll tax has the unique effect of creating a strong incentive for outsourcing of internal processing and administration functions of financial institutions. By outsourcing these functions, they incur no additional non-creditable VAT, and avoid the payroll tax. Overall, the Quebec system is essentially one that zero-rates fi- nancial services. However, the compensatory payroll and capital taxes could be viewed as replacing partial taxation under the ex- emption system with another form of partial taxation. The Choices for Developing Countries Developing countries that introduce VAT will want to adopt mod- els that draw on experience elsewhere, and that do not incorporate untried features and approaches. In the case of financial services, this means that an exemption system will need to apply in the case of some financial services. The question at this point becomes how broad the exemption should be. Should it be quite inclusive, as in the EC, or be limited to principal activities, as in Singapore, or apply only to margin activities? Should compensatory taxes be employed? In answering these questions, policymakers will have to balance the demands of certain competing objectives, and take into account cer- tain characteristics of their financial system and economy. Developing countries will typically have an objective of applying taxation to financial services as a progressive source of revenue, as well as to avoid the distortions associated with an exemption de- scribed earlier. Taxation of financial services is viewed as progres- 376 SATYA PODDAR sive because such services as banking, brokerage, property and ca- sualty insurance, and foreign exchange transactions are connected closely with those with income and wealth. This has certainly been a major consideration in the decision of the Government of India to extend its service tax to a variety of financial and other services, in- cluding share brokerage and insurance. Generally, where an exemp- tion is in place under a VAT, there will likely be less revenue than under full taxation. The progressive revenue objective thus dictates as wide an application of VAT to financial services as possible. It will also lead countries to consider compensatory taxes where an exemption must be provided and even additional ad hoc taxes for revenue purposes. In deciding how far taxation can proceed, certain characteristics of the economy and the financial sector will need to be taken into consideration. Some factors found in developing countries may make it easier to apply taxes to financial services without adverse effects, while others may make it more difficult. The relevant con- siderations will vary from country to country and careful consider- ation will be needed to determine what specific approach should be taken among the alternatives described earlier. Factors that may dictate a conservative approach relying heavily on exemption are the importance of not retarding development of the financial sector, and not diverting financial activity into the in- formal sector where tax is not collected. It is paramount to recognize that financial markets are crucial for a well-functioning economy. A characteristic of developing economies may be that consumers and businesses make relatively little use of financial services. Developing the sector allows transaction costs to be reduced and capital markets to develop. If taxation is likely to cause financial activity to move un- derground or offshore at the expense of local institutions, there may be little revenue gain from taxation and significant economic cost. Blockages to the development of the sector as a result of excessive or badly designed taxes can do significant damage that cannot be justi- fied by short-term revenue considerations. If significant levels of financial services to households are pro- vided by small, largely unregulated, and untaxed entities in the in- formal sector, taxation may drive consumers toward this segment of the economy. Attempts to apply full taxation on financial services in this case will not increase tax revenues and may not be desirable on economic development grounds. Yet, in some instances, characteristics of financial services in de- veloping countries may make it possible to adopt models that tax a broader range of financial services more fully. If the financial sector is concentrated and highly regulated, approaches that tax most fees CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 377 and commissions or agency services could be feasible. Regulatory controls may mean that there is little behavioral change associated with these approaches. However, substitution of margin for fees and commissions is likely to be an issue in respect of at least some services even in these cases. For example, attempts to tax fees and commissions on foreign exchange transactions have proven unsuc- cessful in some countries such as Russia, as the form in which con- sideration was earned was rapidly changed to margins or moved into the untaxed, informal economy. Where substitution is likely to be feasible, it will probably be better to introduce an exemption for the service in question. Moving further along the spectrum to full taxation is also likely to be more feasible where competition is limited in the financial sector. This would be the case if the potential for international com- petition from competing sources of supply is minimal because of foreign exchange controls, or if these services are provided by gov- ernment monopolies, or both. In making the choice, it is important to consider compliance and administration issues and the extent to which taxation versus ex- emption will create competitive distortions and significant behav- ioral changes. Some factors that are problematic in developed economies may raise less difficulty in developing countries. For ex- ample, outsourcing may not be as feasible as in developed countries. If revenue and other policy considerations dictate further taxation of the financial sector than would be available under a broad system of exemption, then the best option would appear to be an extension of tax to those financial services for which an explicit fee is charged. Compensatory taxes would be a less satisfactory alternative because, as noted, they raise their own distortions and difficulties. Notes 1. One type of financial service identified in table 12.1 that is somewhat unusual is the purchase and sale of gold and precious metals. Gold and cer- tain other precious metals play a dual role in the economy in that they are traded both as consumer and industrial commodities, and also for invest- ment purposes. As a result, gold bullion, in particular, has often been sin- gled out for treatment that to some degree mirrors the treatment of finan- cial products under consumption tax systems. 2. Other exemptions specified include goods and services supplied by certain public bodies, sale of land and buildings, and leasing of immovable property. For useful perspectives on the current system of exemptions see European Economic Commission (1995) and OECD (1988). McLure 378 SATYA PODDAR (1987) provides a description and analysis of VAT exemptions in the Euro- pean Community, as well as other OECD-member countries. 3. For another discussion of the rationale for exemption, see Ebrill and others (2001, pp. 94–97). 4. See the discussion of this issue in chapter 2. 5. One response to importation biases, self-assessment, is not very effec- tive in the case of financial services. Another approach, requiring registration by foreign institutions, has been more effective where it has been applied be- cause, in an industry that requires considerable trust to operate, registrants and consumers are often reluctant to deal with nonregistered parties. 6. See Grubert and Mackie (2000). 7. See chapter 2 for a more complete discussion of the underlying issue. Edgar (2001) also surveys this issue and considers alternatives to full taxa- tion that might be based on the arguments advanced by Jack. 8. See Cooper and Vann (1999) for an argument of this type—in favor of retaining exemption—until a clear, internationally agreed method of tax- ation is settled upon. 9. VAT systems and income tax systems differ fundamentally in this re- gard. Individuals under an income tax are typically taxable on each source of income, consisting of labor income, business income, investment income, rents, and capital gains. There is no registration requirement and the sole test is whether the activity in question is undertaken with a profit motive. In the case of the latter four income sources, associated expenses may be de- ducted in determining income subject to tax. The allowance of a deduction for investment expenses that are attributable to investment activity means that the income tax does contain a dividing line between deductible and non-deductible expenses, including for financial services. This is a necessary feature of arriving at an appropriate tax base including, of course, the time value of money. 10. The discussion of the options includes a brief description of exam- ples of alternative approaches used in selected countries. However, these ex- amples are illustrative; they do not constitute a broad survey of country-by- country approaches. For a more inclusive discussion of systems of taxing financial services, see Tait (1988). 11. The TCA system was developed by Ernst & Young in a set of stud- ies on cash flow taxation done for the Commission of the European Com- munities. It is described in Ernst & Young (1996a). The system as it would apply to deposit or loan transactions is also described in Poddar and En- glish (1997). The system was further developed and analyzed in a series of pilot studies carried out at selected European financial institutions under the auspices of Commission of the European Communities. A complete de- scription of the system indicating how it would apply to various types of fi- nancial services drawing on the results from these pilot projects is available in the Customs section of the European Commission website (http://europa. CONSUMPTION TAXES: THE ROLE OF THE VALUE-ADDED TAX 379 eu.int/comm/taxation_customs/publications/reports_studies/report.htm #financial). 12. See previous note. 13. A detailed discussion of this issue can be found in Ernst & Young (1996b). The actual occurrence of risk is recognized for each individual transaction separately. This leads to situations in which the net cash flows or margins earned from certain transactions become negative. Even though each individual contract is priced to yield a positive anticipated margin (a positive ex ante price) to the financial institution, the actual margin (the ex post price) for a subset of transactions would necessarily be smaller or neg- ative because of the occurrence of risk. There is no operational mechanism that allows an ex ante isolation of the pure risk premium for tax purposes with precision or objectivity. 14. The exact structure adopted does vary in some ways from the ideal cash flow system. See, for example, Ernst & Young (1996a, pp. 156–57). 15. Value added should also include financing costs for capital: that is, interest on funds borrowed to acquire business assets. In the case of finan- cial institutions, one needs to draw a distinction between this interest com- ponent and the interest costs for deposits and other funds acquired during the course of borrowing and lending business. The former are a component of value added, but not the latter. A portion of profits, which represents a pure rate of interest on shareholders’ funds used in lending business, should also be excluded from the VAT base. This is achieved under the TCA sys- tem by defining the base as interest received less an indexing adjustment that represents cost of funds at the pure rate of interest. References Auerbach, Alan, and R. H. Gordon. 2001. “Taxation of Financial Services under a VAT.” University of California, Berkeley, Department of Eco- nomics. Processed. Boadway, Robin, and Michael Keen. 2003. “Theoretical Perspectives on the Taxation of Capital Income and Financial Services” (chapter 2, this volume). Chia, N. C., and J. Whalley. 1999. “The Tax Treatment of Financial Inter- mediation.” Journal of Money, Credit and Banking 31 (4): 704–19. Cooper, Graeme S., and Richard J. Vann. 1999. “Implementing the Goods and Services Tax.” Sydney Law Review 21: 337–436. Ebrill, Liam, Michael Keen, Jean-Paul Bodin, and Victoria Summers. 2001. The Modern VAT. Washington, D.C.: International Monetary Fund. Edgar, Tim. 2001. “Exempt Treatment of Financial Intermediation Services under a Value-Added Tax: An Assessment of Alternatives.” Canadian Tax Journal 49 (5): 1133–1219. 380 SATYA PODDAR Ernst & Young. 1996a. “Value-Added Tax: A Study of Methods of Taxing Financial and Insurance Services.” Prepared for the European Commis- sion, Directorate General XI, Customs and Indirect Taxation. Brussels. Processed. ———. 1996b. “Negative TCA Balances under the TCM/TCA Cash-flow Method.” Prepared for the European Commission, Directorate General XI, Customs and Indirect Taxation. Brussels. Processed. European Economic Commission. 1995. “Options for a Definitive VAT Sys- tem.” Working Paper E-5. Directorate-General for Research, Brussels. Grubert, Harry, and James Mackie. 2000. “Must Financial Services be Taxed Under a Consumption Tax?” National Tax Journal 53 (1): 23–40. Huizinga, Harry. 2002. “A European VAT on Financial Services?” Eco- nomic Policy 35: 499–534. Institute for Fiscal Studies. 1978. The Structure and Reform of Direct Tax- ation. Report of a Committee Chaired by Professor J. E. Meade (The Meade Committee Report). London: George Allen & Unwin. Jack, William. 2000. “The Treatment of Financial Services under a Broad- based Consumption Tax” National Tax Journal 53 (4): 841–51. McLure, Charles E. 1987. The Value-Added Tax, Key to Deficit Reduc- tions? Washington, D.C.: American Enterprise Institute for Public Policy Research. OECD (Organisation for Economic Co-operation and Development). 1988. Taxing Consumption. Paris. Poddar, Satya, and Morley English. 1997. “Taxation of Financial Services under a Value-Added Tax: Applying the Cash Flow Approach.” National Tax Journal 50 (1): 89-111. Tait, Alan. 1988. Value-Added Tax: Practices and Problems. Washington, D.C.: International Monetary Fund. 13 The Accidental Tax: Inflation and the Financial Sector Patrick Honohan Although first and foremost a monetary phenomenon, inflation has wider implications for macroeconomic stability, competitiveness, and contracting, notably because many contracts, especially wage contracts, are fixed in terms of money. Two particular features of in- flation—and their interaction—are the focus of this chapter: the po- tential to ease the government’s budget constraint and the impact on financial sector performance. Inflation is nowadays often an “accidental tax,” with surpris- ingly little use of direct monetary financing of the government. In- flation has two contrasting impacts on the financial sector. On the one hand, by increasing the risk and cost of payments and main- taining liquid transactions balances, it increases the demand for cer- tain financial services, swelling the value-added and profitability of banks. On the other hand, the interaction of inflation with a non- indexed tax system often results in an effective rate of taxation on financial intermediation that is super-sensitive to the inflation rate. The high and volatile effective tax rates are associated with wide in- termediation margins and a reduced scale of intermediation. For society, both effects are costly distortions. Even if banking activity and profits are increased by the additional demand for effi- cient payments services, this is a socially costly diversion of re- sources from more productive activities. The costs of tax-inflation interactions can be reduced by avoiding particular tax designs, in- 381 382 PATRICK HONOHAN cluding taxes on gross interest receipts, off-market interest ceilings, and unremunerated reserve requirements. This chapter is organized as follows. The next section reviews the underlying theory of inflation as a tax, asking whether it could form a part of the optimal set of taxes, especially bearing in mind its im- pact on the financial system. The section that follows examines the mechanics of how inflation generates revenue for the government, uncovering evidence suggesting that, in most countries recently, most of the fiscal benefit flows through the profits of the central bank, rather than in the form of direct financing. The chapter then presents empirical evidence of the impact of inflation on banking activity and profits, showing that, while the financial system shrinks with infla- tion—stock markets, apparently, more than the banking sector—in- flation tends to be associated with a boost to bank profitability and value-added. The chapter shows how highly sensitive the effective rate of other taxes (especially those on interest) can be to inflation, and proposes a measure of that sensitivity. After offering some re- marks on incidence, the chapter concludes with a call to improve the indexation of the tax system so far as the computation of interest is concerned. Could Inflation Be a Good Tax—Even for the Financial Sector? If inflation is always and everywhere a monetary phenomenon, then the inflation tax too is inherently a monetary phenomenon. But, as with the causes of inflation itself, it is often necessary to look behind the money creation in order to understand the processes that create and sustain the conditions for monetary expansion. Rather than being the consequence of a measured policy decision to impose an “inflation tax,” inflation is more often the result of quite different policy dynamics. For example, it may result from monetary accom- modation of a wage bargaining process, or of exchange rate depre- ciation caused by non-monetary factors. Genuine Tax or Analytical Construct? To some extent, the concept of inflation tax is an analytical construct rather than a recognized and managed source of revenue for the state. It is clear enough that inflation and the associated money cre- ation can represent a transfer of resources to the state, but measur- ing the resources so transferred, or even identifying the tax base, is a matter of analysis on which authors differ. No government budget THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 383 contains a line entitled “inflation tax” or “seigniorage.” That is not to say that the implicit revenue comes as a surprise at the end of the year; estimates of the revenue and expenditure of the various chan- nels through which inflation will affect the budget will normally be taken into account in budgetary forecasts. Thus only if the inflation is unexpected will there be surprises. It is the increased demand for holdings of nominal base money associated with inflation that will create the conditions for the budget to profit from the government’s monopoly pricing of base money. It is not even the case that inflation will always be associated with a net gain to the budget. Existing conventions and contracts may greatly erode, or even reverse, the gains. Thus as was famously pointed out by Tanzi (1977), if government wage rates are indexed or quickly adjust to changing price levels while tax receipts, struck in nominal terms, arrive at the Treasury in arrears, an engineered expansion in the money supply may not provide enough resources to cover the additional net outlays that are required because of the change in prices. Few governments, then, set out consciously to exploit the infla- tion tax. If there is monetary financing of the budget, it is seen as a financing, rather than as a tax device, albeit one that is likely to have the politically costly side effect of inflation. Nevertheless, once governments are benefiting from the implicit revenue of a steady rate of inflation, they will quickly feel the loss of this revenue from a stabilization, as was pointed out in respect of some potential Ex- change Rate Mechanism and European Monetary Union members (Grilli 1989a, b; Repullo 1991). Should an Inflation Tax Be Used? Some have argued that perhaps governments should consciously ex- ploit the inflation tax. Of course, Bailey (1956) and Friedman (1953, 1971) argued that in an essentially partial equilibrium steady state context, the optimal rate of inflation tax was zero, to be achieved (absent any way of paying interest on currency notes) by a steady proportionate contraction in the money supply and the associated de- flation. But, as noted in this context by Phelps (1973), taxes without distortions are not generally available, so that in order to finance so- cially desirable government expenditure, the optimal set of taxes might easily include the inflation tax. An unresolved debate has raged since over this issue. Among theorists, the question has been whether the inflation tax is one of those taxes (following the argument of Di- amond and Mirrlees 1971) that should optimally be omitted from 384 PATRICK HONOHAN the set of taxes, in preference to taxes that can achieve the desired im- pact on consumption without distorting production structures. An increasingly refined literature has pinpointed that, in order to sustain the original Friedman proposition, it must be the case that the role of money is essentially reducible to that of an intermediate good.1 If higher inflation induces less reliance on money balances in household portfolios, instead channeling more of savings into capital formation, the result could be more rapid growth. This proposition, advanced long ago by Tobin (1978) and retaining theoretical support in some models of endogenous growth, remains controversial and rather heterodox.2 Certainly, empirical studies cited below show a negative cross-country correlation between inflation and growth. The existence of non-trivial collection costs for formal taxes may provide another rationale for use of the inflation tax. Some authors have gone so far as to suggest that the presence of currency substi- tution may in this respect be harmful, in that it limits the scope for using the inflation tax for such reasons (Sibert and Liu 1998).3 The base of the inflation tax includes not only currency, but also unremunerated or partly remunerated required banking reserves. It is hardly disputed nowadays that the primary rationale for sizable unremunerated reserves is fiscal. (Even if some techniques of mone- tary stabilization involve the imposition of required reserves, little if any fiscal penalty is needed to ensure their effectiveness in this role.)4 Noting that bank depositors are more prosperous than those whose liquid assets are wholly in the form of currency notes, some scholars have argued that the major political economy motivation for using partially remunerated reserve requirements instead of relying on seigniorage from currency is only to discriminate between income groups. In these models, a partly remunerated reserve requirement will be preferred (to an unremunerated one) by a government sup- ported by the rich.5 As to the distinct question of whether reserve re- quirements or a tax on deposit interest is a preferable steady state way of raising government revenue, the majority view among theo- reticians, since the work of Freeman (1987) and Brock (1989), has been that the two are equivalent. Some authors argue that the de- posit tax is preferable, to the extent that it allows the banks to make more use of any special abilities to earn higher returns on loanable funds, thereby better facilitating the transfer of needed fiscal re- sources to the government.6 Some have taken this reasoning to imply that the rate of inflation tax, if optimally chosen, should be correlated with other rates of tax (the tax smoothing argument).7 While the tax smoothing argument may seem plausible in theory, the difficulty in practice of controlling inflationary psychology and expectations means that a given target THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 385 for the rate of inflation tax may not be as easily hit as with other taxes. Without doubt, recourse to monetary financing is often much easier and quicker than adjusting tax rates and schedules in order to raise additional revenue.8 Therefore, on a time-series basis, after taking account of adjustment costs, one would expect optimally chosen inflation financing to be correlated, not with other tax rates, but with unanticipated fluctuations in spending or in the govern- ment deficit (the residual financing argument). This does appear to be the case, but in a systematic way only for high inflation countries (Fischer, Sahay, and Vegh 2002; Boyd, Levine, and Smith 2001).9 Response of the Financial System to Inflation The monetary and financial system does not remain passive in re- sponse to inflation. Countless studies document the substitution away from non–interest-bearing monetary assets in favor of interest- bearing or indexed assets, or to those denominated in foreign cur- rencies, or to non-monetary assets. It is this substitution that has the most important impact on the performance and functioning of the financial system. On the one hand, the volume of funds that it is able to mobilize and intermediate may shrink. On the other hand, its capacity to provide the instruments to insulate economic agents from the effects of high or volatile inflation is an important aspect of its social contribution. The average rate, the variability, and the predictability of infla- tion are three key elements contributing to the impact of the infla- tion tax. These in turn are linked to the rate of monetary expansion, but not in a neat or mechanical way. Over the long term, the rate of inflation can be expected to equal the rate of base money growth less the rate of growth in the real economy, with some allowance for technical change in the demand for base money. In practice, this re- lation is a good predictor only at high rates of money growth (and it is very good for hyperinflation). This means that there is a dis- connect between the rate of tax as measured by the opportunity cost of holding interest-free base money (which will be related to the ex- pected inflation rate) and the flow of financing to the budget from money creation (Honohan 1996). Costs of Inflation What are the social costs of the inflation tax and how big are they? This broad and much discussed question takes us beyond the sphere 386 PATRICK HONOHAN of the financial sector (Cukierman 1984; Feldstein 1999). Broad, re- duced form calculations have compared how growth rates vary with inflation across countries.10 But such calculations do not consider the channels of effect and what aspects of the financial system may be involved. The early study by Bailey (1956) adopted a simple ap- proach by measuring the lost consumer surplus under an estimated demand for money function. Several more specific channels of effect have been studied theoretically, including the considerations that, in the face of steady inflation, agents will over-economize on the hold- ing of transactions balances (Cooley and Hansen 1991)11 and that they will hold too few precautionary balances, resulting in unneeded fluctuations in consumption (Ïmrohoro˘ glu and Prescott 1991).12 It is often noted that—as agents will adjust only to the expected part of inflation, with perhaps some allowance for the variability— unexpected inflation is, to some extent, like a lump-sum tax. How- ever, it could contribute to the market’s future allowance (risk- premium) for surprises entailing a long-lasting deadweight loss effect—which could even be larger than if each period’s inflation were fully anticipated in advance. How Does the Government Get Hold of the Inflation Tax Revenues? In simple textbook models, the role of money creation in influenc- ing the government’s budget constraint is typically expressed in terms of a balance sheet identity that must be satisfied at all times, such as: (13.1) G – T = B – (1 + r)B–1 + M – M–1 where G and T represent spending and taxation, B is the stock of bonds (perpetuities), r the rate of interest paid on the bonds, and M the stock of base money. This is all fine and perfectly consistent, but do the implicit institutional arrangements correspond to reality? Curiously, this question does not appear to figure prominently in the theoretical literature, although it is important in empirical analyses, especially for particular countries.13 Almost all countries now have an articulated accounting system that clearly distinguishes the central bank from the government. An increase in the central bank’s monetary liabilities (currency and the deposits placed, mainly by banks and other financial institutions— known as the money base) need not have as its counterpart an in- THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 387 crease in central bank credit to the government. (The empirical facts on this are examined later.) If the central bank does not “lend” newly created money to the government, is there still seigniorage, and if so how does the gov- ernment receive the seigniorage? The short answer to this is that the seigniorage does exist, and comes through the assets that are ac- quired by the central bank in return for issuing money base. These are either foreign exchange, claims on the banking system, or claims on other sectors of the domestic economy. These assets in turn are not usually handed over to the govern- ment to be liquidated in aid of the budget.14 Instead they are typi- cally held on the central bank’s balance sheet, where they earn inter- est until redeemed, and that interest accrues to the benefit of the central bank’s income statement. Most central banks are owned by the government or, if they are not, the government nevertheless is usually entitled to the bulk of the central bank’s profits in due course. In this case, it is through the distribution of the central bank’s profits that the government’s budget finally receives the seigniorage in the form of cash. Other demands on the revenues of the central bank can inter- vene, with the result that some of the seigniorage never reaches the government’s budget. For one thing, the operational expenses of the central bank will often be paid out of its interest revenues; also, the central bank may undertake lending or similar activities at below- market interest rates.15 Such schemes are, in effect, off-budget sub- sidies; although they reduce the flow of seigniorage to the budget, they do relieve the budget from alternative spending programs that would have been required to achieve the same goals as the subsi- dized lending by the central bank. Improved transparency and fiscal control argue against such hidden subsidies paid for out of seignior- age before it is transferred to the government, but they do continue to be observed, although less often than before. The often delayed and opaque nature of the link between mone- tary expansion and the budget in regimes where relatively little of the base money liabilities of the central bank are backed by its lend- ing to government contrasts sharply with the consolidated budget identity of equation 13.1. An expansion in non–interest-bearing base money that is fully backed by an increase in foreign bills, for exam- ple, will yield only the interest rate in the first year. There may even be a further delay in transmitting this to the government, depending on the accounting and dividend procedures of the central bank.16 Dividend and accounting policies of central banks are changeable and often rather opaque. Nonetheless, the balance sheet structures 388 PATRICK HONOHAN can be examined to detect which central banks back their base money liabilities with net claims on the government, which with net foreign assets, and which with other net claims on domestic sectors. The institutional arrangements where mostly government backing is used (corresponding to equation 13.1) may be termed “Mode I,” and where mostly foreign exchange is used may be termed “Mode II.” An analysis along these lines was carried out for 153 central banks using data in International Financial Statistics (IFS).17 A caveat needs to be added in regard to this data inasmuch as the stated composition of assets reported by central banks to IFS could differ from the effective composition after account is taken of swaps and other derivative activities.18 Strikingly, the mean (and median) percentage of foreign exchange backing of base money in the world’s central banks (in 2000) was more than 100 percent (table 13.1). Far from the picture implied by equation 13.1, which implicitly allocates all of the annual increase in base money to advances to government, the overall picture suggests a predominance of Mode II-type insti- tutional arrangements. Clearly there is also a wide variation from Table 13.1. Backing of the Money Base by Foreign Exchange or Claims on Government Change 2000 1998–2000 Proportion backed by: Quartile % % Lower quartile 45.0 28.8 Foreign exchange Median 100.4 115.5 Upper quartile 154.5 237.2 Lower quartile –12.8 –70.4 Government Median 11.7 –1.1 Upper quartile 53.8 63.7 Lower quartile –1.6 –13.0 Government less central Median 21.3 14.1 bank capital Upper quartile 60.7 71.4 Lower quartile 42.3 30.8 Currency share in money Median 56.7 54.8 base Upper quartile 72.9 81.2 Note: Based on calculations for 153 countries of the end-2000 level (and for 130 countries of the 1998–2000 change) in money base as a percentage of the end-2000 level (and the 1998–2000 change) in foreign exchange reserves, net claims on gov- ernment, and net claims on government less central bank capital. Also currency as a percentage of the money base. Source: Author’s calculations, based on International Financial Statistics. THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 389 country to country, with some countries relying heavily on net credit to government as the main backing for base money. The calculations were also carried out for 1998 and 1999 and for the change from 1998 to 2000. While the mean foreign exchange backing was lower in those years, it was still very high. Analysis of the change shows even greater reliance on foreign ex- change backing at the margin in 1998–2000. Of the 130 countries that increased their money base between 1998 and 2000, 80, or 38 percent, fully backed all of the new money base with foreign ex- change. All but 20 had at least 25 percent foreign exchange cover at the margin. Far from routinely adding to the government’s indebt- edness to the central bank, in well over half the countries (72 cases), central bank credit to government actually declined. In only 30 cases did such credit increase by as much as half the increase in the money base. An alternative calculation (the third row in table 13.1) nets out the capitalization (and other items) of the central bank on the grounds that an increase in net credit to the government that is funded by an increase in central bank capitalization does not repre- sent monetary financing. The results of the calculation are qualita- tively similar and indicate even less of a link between monetary fi- nancing of the government and increases in base money. If this analysis reveals that the timing and mechanisms of budg- etary finance through money creation and the inflation tax are most commonly closer to Mode II than Mode I—and as such, quite dif- ferent from the textbook model—it is natural to ask whether there is a clear statistical link between inflation and the institutional mode of seigniorage transmission, or between monetary depth and the in- stitutional mode, in each case measuring the mode by the degree to which money base increase is backed by foreign exchange or by net credit to government. Figure 13.1 presents the relevant scatterplots. Because of the wide range of values for the explanatory variable, no strong relationship is evident. Regression analysis confirms the theoretically plausible negative correlation between inflation and the degree of foreign backing (and a positive correlation with net government credit), but the finding is not statistically significant, even after excluding the largest outliers. The Impact of Inflation on the Financial Sector How does inflation affect the scale and profitability of financial in- termediation and real rates of return? In addition to reviewing some cross-country evidence from the literature, this section contributes Figure 13.1. Alternative Sources of the Monetary Base and Their Correlation with Monetary Depth and Inflation Monetary financing and financial depth Foreign backing and financial depth 390 (10 outliers removed) (10 outliers removed) Financial depth % GDP Financial depth % GDP 140 140 120 120 100 100 80 80 60 60 40 40 20 20 0 0 –500 –400 –300 –200 –100 0 100 200 300 400 500 –500 –400 –300 –200 –100 0 100 200 300 400 500 Net government financing as % money base Net CB foreign reserves as % money base Monetary financing and inflation Foreign backing and inflation (9 outliers removed) (9 outliers removed) Inflation average Inflation average 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 –10 –10 –500 –400 –300 –200 –100 0 100 200 300 400 500 –500 –400 –300 –200 –100 0 100 200 300 400 500 Net government financing as % money base Net CB foreign reserves as % money base Source: Author’s calculations, based on International Financial Statistics. THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 391 some new empirical evidence on these issues. Inflation is found to be positively associated with profitability and especially with the value-added of the banking system. But the balance sheet size of the banking system shrinks with inflation, though the effect may be smaller at higher rates of inflation. Stock market liquidity also di- minishes with inflation, apparently to a greater extent. The Fisher- ian link between inflation and nominal interest rates—only weakly evident in the cross-country data at low rates of inflation—is clear at higher rates. Since the rate of inflation is endogenous to macroeconomic poli- cies and financial sector structures, it would be unwise to assert too strongly that the relationships detected are causal ones. Furthermore the impact of inflation on financial sector magnitudes is highly de- pendent on administrative, legal, and tax characteristics for which no good statistical controls are available. Therefore the statistical as- sociations should be regarded as indicative of general tendencies, rather than immutable laws. Bank Profitability First, consider the impact of inflation on bank profitability. This question was addressed by the major studies of Demirgüç-Kunt and Huizinga (1999) and Claessens, Demirgüç-Kunt, and Huizinga (2000). Although exploring the role of inflation was not the main focus of these papers, the authors concluded that the impact of in- flation on profitability, while not very significant, is positive. Look- ing again at this issue with more recent data confirms and reinforces their finding. In addition, this study finds that inflation has a stronger and more consistent positive association with the value- added of the banking system. The above-mentioned authors used data from the income state- ments of about 3,000 banks in 80 countries from 1988 to 1995. They examined a large range of potential determinants, both bank- specific and country-specific, of bank profitability. The bank-specific determinants included equity capitalization (which not surprisingly increases the ratio of before tax profits to total assets),19 the aver- age tax rate paid (estimated to pass-through 100 percent to before tax profits), and a dummy for foreign ownership (helps a bank’s profitability by as much as 50 basis points in low-income countries, and hinders it in large). A bank with high overhead costs as a share of total assets does not, on average, recover these fully, although the point estimate suggests that the shortfall is very slight.20 The other bank-specific explanatory variables represent the shares in total as- sets represented by loans and non-interest earning assets, respec- 392 PATRICK HONOHAN tively, and of deposits in total liabilities. Relative to the excluded balance sheet categories, a heavy reliance on non-interest earning assets hits the bottom line, especially in rich countries; access to sources of funding other than deposits boosts profitability. Loans contribute to profits somewhat more than the excluded category. These are the only bank-specific variables used in the Demirgüç- Kunt and Huizinga (D-H) analysis, but they are augmented by a se- ries of macro, tax, and institutional variables. It is here that the ef- fect of inflation is measured. In fact, inflation is nowhere significant at the 10 percent level, but the coefficient is positive, its size imply- ing that a 10 percent rate of inflation boosts bank profitability by about 10 basis points of total assets. The significant macro variables in these regressions are per capita income and the real wholesale in- terest rate, both of which are estimated to add to profits.21 GDP growth is not significant. Demirgüç-Kunt and Huizinga include the economy-wide average reserve holdings as a proxy measure for this form of quasi-taxa- tion.22 Interestingly, average reserve holdings (as a share of total as- sets) in poorer countries sharply reduce before-tax profits, whereas they increase them in richer countries. One would expect that unre- munerated reserve requirements might have an especially severe im- pact where inflation is high, but this was not tested by the authors. If the focus is on the overall impact of inflation, it needs to be borne in mind both that inflation may be jointly determined with some of the other variables included in the regression (notably the real interest rate). Also, the wide variation of inflation rates (and in particular the inclusion of Brazil, which in the sample period reached 2,300 percent inflation) and the use of country dummies may mask a genuine impact of inflation on bank profitability. This prompts a revisit to the data and a closer look at inflation effects (though without retaining the micro aspects). With cross-sectional data covering some 70 countries, then, it seems appropriate to pur- sue a loose specification search to try to assess the scale and ro- bustness of any link between inflation and average profitability. As shown in figure 13.2, a bivariate scatterplot of the country- average values of the two variables suggests a positive relationship. This is confirmed by regression analysis on these averaged figures (table 13.2). The reported regressions use log-inflation instead of the level and include all of the other cross-country variables that were significant in the D-H study: namely, per capita GDP, real interest rates, and the level of reserves. Even including all these variables, in- flation remains highly significant when the equation is estimated in log form or, if Brazil is excluded, even in level form.23 Upon further examination, the interesting fact emerges that in- flation strongly interacts with reserve holdings—not to reduce prof- THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 393 Figure 13.2. Bank Profitability and Inflation, 1988–95 log-inflation 4.00 3.00 BRA 2.00 1.00 RUS 0.00 ROM JAM –1.00 EST –2.00 –3.00 –4.00 –5.00 –1.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 Profit before tax change Notes: See note to table 13.3; BRA Brazil; EST Estonia; JAM Jamaica; ROM Ro- mania; RUS Russia. its, but instead to increase them! Rather than the reserve holdings being involuntary, in countries with high reserve holdings and high inflation the banks are likely finding ample remuneration at least on their marginal reserve holdings. A look at some of the high-profit countries in the scatterplot shows Russia and Romania to be prom- inent, no doubt pointing to special features of these transition econ- omies’ systems in the early 1990s.24 For the second half of the 1990s, figure 13.3 presents a similar scatterplot of average country banking profits (measured here by the rate of return on assets) against log-inflation. The same ap- proach to equation specification is adopted. Once more, a clear upward-sloping relation appears in the simple regression (table 13.3, equation 1). In this case, a 10 percentage point increase in in- flation from the sample median of 6 percent per year to 16 percent is associated on average with an increase in return on assets of 0.4 percentage points—compared with a median value of 1.1 percent.25 The relationship appears strongly significant, and survives the addi- tion of per capita income and real interest rates (neither of which are significant in this later data set, as shown in table 13.3, equation 2). Once again, the pattern of outliers (more than three standard er- rors away) is interesting: Moldova, with high measured profitabil- ity and inflation; and Thailand, with very low profitability (reflect- ing the crisis of 1997–78) and moderate inflation. These outliers do 394 Table 13.2. Inflation and Bank Profitability, 1988–95 Equation 1 2 3 4 Exclude Brazil Log form Variable Estimate (t stat) Estimate (t stat) Estimate (t stat) Estimate (t stat) Constant 2.262 (9.6) 1.941 (8.4) 3.518 (11.8) 3.544 (12.1) Inflation 0.066 (1.1) –0.972 (3.9) 0.685 (5.0) 0.727 (6.9) GDP per cap ϫ 10–4 –0.863 (3.5) –0.689 (3.0) –0.122 (0.5) — — Real interest 0.001 (0.1) 0.001 (0.1) — — — — Countries All Not Brazil Not Brazil Not Brazil Functional form level level log log Method/no. obs OLS 67 OLS 66 OLS 72 OLS 72 RSQ/DW 0.188 1.46 0.338 1.82 0.409 1.91 0.741 1.76 Note: Dependent variable is profit before tax as a percentage of total assets (country average). Equation 1 includes the same variables as in D-H’s equation 1. Table 13.2. (continued) Inflation and Bank Profitability, 1988–95 Equation 5 6 7 8 Exclude Brazil Variable Estimate (t stat) Estimate (t stat) Estimate (t stat) Estimate (t stat) Constant 3.217 (11.8) 1.748 (3.2) 1.475 (3.1) 2.657 (6.9) Inflation 0.609 (6.2) –0.074 (1.2) 0.976 (4.2) 0.526 (5.1) GDP per cap ϫ 10–4 — — –0.503 (0.7) –0.384 (0.6) — — Real interest — — 0.000 (0.0) 0.002 (0.1) — — Reserve requirements — — 3.165 (1.6) 2.986 (1.8) 2.152 (1.8) Res. requirements ϫ GDP/cap — — –2.780 (0.6) –2.070 (0.5) — — Tax rate — — –0.364 (0.4) –0.643 (0.7) — — Tax rate ϫ GDP/cap — — 0.016 (0.1) 0.191 (0.1) — — Countries All All Not Brazil Not Brazil Functional form log level level log Method/no. obs OLS 73 OLS 55 OLS 54 OLS 58 RSQ/DW 0.355 1.92 0.265 0.86 0.409 1.91 0.560 1.77 Notes: Dependent variable is profit before tax as a percent of total assets (country average). Equation 6 includes the same variables as in D-H’s equa- tion 2. For each equation is shown the number of observations, the method (ordinary least squares), and the R-rated and Durbin-Watson statistics. Source: Author’s calculations. The banking data is from Demirgüç-Kunt and Huizinga (1999). 395 396 PATRICK HONOHAN Figure 13.3. Bank Profitability and Inflation, 1995–99 log-inflation 6 BLR 5 4 GHA NGA 3 MDV 2 BTA THA KOR 1 0 –1 JAP –2 –3 –2 –1 0 1 2 3 4 5 6 Return on assets Notes: See note to table 13.3. BLR Belarus; BTA Botswana; GHA Ghana; JAP Japan; KOR Korea; MDV Moldova; NGA Nigeria; THA Thailand. not, however, strongly influence the simple regression with log- inflation. Here the link with reserve ratios no longer applies. As a control against under-specification, the analysis also checked to see whether some of the other variables employed by D-H but not significant for the earlier period remain insignificant here. The only one that is consistently significant is the bank concentration ratio, as shown in table 13.3, equations 5, 6, and 7. It should be borne in mind that this variable is available only for about three-quarters of the countries. Overhead as a share of total assets is only marginally significant, but when included tends to reduce the significance of inflation (equation 5). However this effect is diminished when the main outliers, Moldova and Thailand, are removed. Overall then, in- flation does seem to be positively associated with bank profitability in the more recent period, as well. We conclude that although the link is far from mechanical, in both the early and the late 1990s, higher inflation has tended to yield substantially greater profit opportunities in at least some countries. Banking Value-Added Widening the focus to include all the value-added of the banking sys- tem reinforces the message that inflation tends to offer possibilities for the financial sector to generate more value-added per unit of total Table 13.3. Inflation and Bank Profitability, 1995–99 Equation 1 2 3 4 Dependent Variable ROA ROA VA VA Estimate (t stat) Estimate (t stat) Estimate (t stat) Estimate (t stat) Constant 0.529 (2.6) 0.212 (0.4) 2.422 (6.8) 2.234 (5.9) Inflation 0.410 (4.1) 0.513 (2.9) 1.462 (8.3) 1.648 (7.4) GDP per cap ϫ 10–4 — — –0.065 (0.4) — — — — Real interest — — 0.023 (1.1) — — — — Res. reqt. ϫ inflation ϫ 103 — — — — — — –0.510 (1.4) Countries All All All All Functional form log log log log Method/no. obs OLS 71 OLS 67 OLS 71 OLS 71 RSQ/DW 0.194 2.23 0.217 2.23 0.501 2.25 0.514 2.30 Note: Dependent variable is percent return on assets (country average), except equations 3 and 4, which use value-added as a percentage of total assets. (Table continues on next page.) 397 398 Table 13.3. (continued) Inflation and Bank Profitability, 1995–99 Equation 5 6 7 Variable Estimate (t stat) Estimate (t stat) Estimate (t stat) Constant –0.590 (1.6) 0.269 (0.9) –0.328 (1.1) Inflation 0.092 (0.7) 0.251 (2.8) 0.287 (1.5) Overheads 0.151 (1.6) — — 0.085 (1.1) Concentration 1.638 (3.5) 1.535 (3.3) 1.456 (3.8) Countries All — All — Not MDA, THAa Functional form — — log — log — Method/no. obs OLS 54 OLS 53 OLS 52 RSQ/DW 0.360 1.00 0.310 0.923 0.409 1.35 Notes: Dependent variable is percent return on assets (country average). For each equation is shown the number of observations, the method (ordi- nary least squares), and the R-rated and Durbin-Watson statistics. aMoldova and Thailand. Source: Author’s calculations. The banking data is updated from that in Demirgüç-Kunt and Huizinga (1999). Thanks to Luc Laeven for assembling and making available the 1995–99 data, which is also used in Demirgüç-Kunt, Laeven, and Levine (2003). THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 399 Figure 13.4. Bank Value-Added and Inflation, 1995–99 log-inflation 6 5 4 3 2 1 0 –1 –2 0 2 4 6 8 10 12 14 Value-added Note: See note to table 13.3. assets. Figure 13.4 shows that the simple correlation between value- added26 and inflation is even stronger than with just profits. Once again, inflation is the major variable of those discussed above that remains significant in the analysis under different specifications.27 A number of alternative specifications parallel to those dis- cussed above for profitability were explored, with value-added as the major explanatory variable. The cross-product of reserve hold- ings and inflation (also mentioned above for bank profitability) proved to be the most consistently significant variable in the spec- ifications that included a lot of explanatory variables. Removing the least significant variables one by one however, left the (log of the) inflation rate as the only significant variable, with a very high t statistic of over 8 (table 13.3, equation 3). The significance of the cross-product term between reserves and inflation progressively weakens as other explanatory variables are removed, to the point where it is no longer significant when included only with inflation (table 13.3, equation 4). Finding a strong relation between value-added and inflation should not be any surprise. Banks offer liquidity and transactions services that may be more highly valued in a period of high inflation or in countries where inflation is often high. In addition, these prof- itability and value-added figures are calculated as a share of total as- sets, and these, as is well-known, are prone to shrink in real terms in inflationary times. 400 PATRICK HONOHAN Bank Asset Size What then, of the impact of inflation on the overall balance sheet size of the financial sector and its major components? Here the pic- ture is unambiguous. As clearly documented by Boyd, Levine, and Smith (2001), inflation reduces both the size of the banking sector and measures of stock market activity (value traded and turnover). According to their estimates, however, beyond an inflation rate of 15 percent or so, the financial sector does not shrink any further. It appears that inflation has done “all the damage it can” by the time it reaches 15 percent. Actually, the suggestion that “all the damage has been done by 15 percent” requires close scrutiny in the context of the inflation tax.28 For one thing, it may seem to fly in the face of a long-held view that reliance on the inflation tax is limited by substitution away from money, and that there is a maximal rate of inflation tax29 (that is, that there is an inflation tax Laffer curve). The idea of an inflation tax Laffer curve can be made consistent with the finding that the size of the financial sector does not shrink much as inflation increases beyond 15 percent if at high rates of inflation the sector switches to reliance on interest-bearing instruments structured in such a way as to insulate the participants from fluctuations in inflation. Indeed, the same authors, as well as Barnes, Boyd, and Smith (1999), find that nominal financial asset yields tend to be much more strongly correlated (across countries) with inflation at high rates of inflation (again they use the 15 percent cutoff). Stock Market Activity Measures of stock market activity also decline with inflation, ac- cording to the estimates presented by Boyd, Levine, and Smith. Once again, they identify a slowing of the decline around 15 percent inflation, though here the cutoff is much less distinct and may rea- sonably be questioned.30 As both components are hit by inflation, it is important to know which of the two declines by more. For in- stance, if one takes the ratio of bank assets to market capitalization, does this decline with inflation or not? Somewhat surprisingly, the data and estimates assembled by Boyd, Levine, and Smith strongly suggest that the bank-to-market ratio tends to increase with infla- tion, at least if market capitalization or value-traded are used as market indicators (figure 13.5).31 That the effect of inflation on the stock market would be greater than on banking is particularly surprising when one considers that stocks are commonly considered a hedge against inflation (though a most imperfect one, as the THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 401 Figure 13.5. How Bank-to-Market Ratios Change with Inflation Bank-to-market ratios and inflation (a) Using market capitalization 3.0 5% inflation 2.5 15% inflation 2.0 1.5 1.0 0.5 0.0 Bank assets Liquid liabilities Private credit (b) Using value traded 14 5% inflation 12 15% inflation 10 8 6 4 2 0 Bank assets Liquid liabilities Private credit (c) Using turnover 1.6 5% inflation 1.4 15% inflation 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Bank assets Liquid liabilities Private credit Note: Shows the ratio of various measures of the size of the banking system rela- tive to the size or activity of the stock market. Source: Based on data in Boyd, Levine, and Smith (2001), using the interquartile gradient for each component. 402 PATRICK HONOHAN evidence shows). It seems likely that here especially a common path effect is apparent, with weak macroeconomic policy conditions re- sulting both in high average inflation and in weak stock market de- velopment. In other words, the endogeneity of inflation is especially problematic for the interpretation of this empirical finding. Interactions between Inflation and the Rest of a Non-indexed Tax System This study now introduces considerations relating to the interaction between inflation and the remainder of the explicit and implicit tax system. The analysis shows that certain types of tax affecting the fi- nancial system have effective rates that are highly sensitive to the rate of inflation. A simple index is proposed that can be used to measure this sensitivity. The study concludes that low values of the sensitivity index are desirable, especially in countries that have proved susceptible to episodes of inflation. Feldstein (1983, 1999) has aptly observed that the interaction be- tween inflation and a non-indexed tax system can have sizable and unexpected effects—even in a country with single-digit inflation. As inflation increases, the double distortions of inflation and taxation can be multiplicative rather than additive, with severe consequences. The effects are mainly through two channels. First, nominal in- terest is treated as income (or an expense), without any adjustment for inflation. Thus it receives a tax charge that depends on the in- flation premium that may be built into the interest rate. Alterna- tively, it might be said that the capital gain on nominal liabilities due to inflation is not chargeable to income tax. Second, depreciation al- lowances are calculated according to historical cost and not to a re- alistic replacement cost in line with rising prices. There is a further effect of taxation in the case where the tax schedule is progressive (with lower rates of tax imposed on lower values of the base). If the thresholds of the progression are not in- dexed to the price level, rising prices will push the average rate of tax on any real sum higher. Among the behavioral effects of a non-indexed tax system: • The real after-tax rate of return to investors in real projects is generally lower. This is because the second effect, usually, and on average, outweighs the first. Accordingly, investment in productive activity is penalized at the expense of investment in land, consumer durables, and other assets not yielding a taxable nominal return, such as gold. THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 403 • There may be a shift in the relative advantage of bond and eq- uity finance, even though, for the individual tax-paying shareholder, the fact that bond interest is tax deductible for the corporation may be offset by the fact that it is taxable as shareholder’s income. • Judging monetary policy is complicated. In an inflationary environment, a given gross-of-tax nominal rate of interest may look to be high enough not to be judged expansionary. But for the in- vesting company, the deductibility of nominal interest payments may mean that the net-of-tax real interest rate is very low or nega- tive, and that the stance of monetary policy is not tight enough to dampen spending. Feldstein notes an important distinction between steady inflation and changes in inflation. Comparing two different steady rates of in- flation, equity prices can be expected to rise faster in the higher in- flation environment. But when inflation jumps from one steady rate to a higher one, equilibrium price-earnings ratios are damaged by the higher effective tax rate to which they are subject. Accordingly, equity prices will fall at first, before beginning to increase at the faster rate. What about the influence on the financial services industry? Clearly, the effects can be significant here, too. Some of the same considerations arise, but the relative importance is altered. In addi- tion, there are some new considerations. For one thing, the financial services industry is obviously affected by shifts in the relative re- liance on different financing instruments. Thus while the impact of inflation on demand and supply conditions for other industries will be chiefly affected through whatever overall impact there is on the economy at large, inflation will have a first-order or direct influence on the demand for different financial services. The impact of infla- tion on the scale and activity of financial services firms needs to be considered alongside the effect on their tax-inclusive cost structures. The degree to which the effective tax rate on financial institu- tions varies with inflation differs as between different non-indexed taxes. Three degrees may be distinguished: • tax burdens that increase, but not in proportion to the rate of inflation (first-order non-indexation); • an increase in the effective tax rate that is approximately pro- portional to the rate of inflation (“second-order non-indexation,” or “supersensitivity”); and • an even greater (third-order) sensitivity to inflation. First-order non-indexation is associated, for example with a sim- ple failure to index the thresholds (such as the tax-free allowance and the point at which a higher rate of tax applies) in a progressive 404 PATRICK HONOHAN tax on non-interest income. In this case, a rise in prices has the ef- fect of pushing more of the tax base into the higher rate of taxation. But to the extent that the rates of taxation are fixed, there is a ceil- ing on the rate of taxation regardless of the rate of inflation or of the degree to which adjustment of the nominal thresholds lags inflation. When it comes to taxes on interest income, however, supersensi- tivity can often arise, implying a potentially volatile inflation rate. Where this arises will obviously depend on the precise specification of the taxes and quasi-taxes involved. This study cannot hope to model every possible tax structure and thus will focus on some sim- ple canonical cases. If a financial intermediary is thought of as adding value to in- vestable funds by repackaging them for users of funds—thereby in- creasing both the risk-return profile and liquidity of the provider and the cost and availability of funds for the users—then it is relative to that value-added that the impact of taxation can best be measured. The three forms in which non-indexed taxes arise most frequently for financial intermediaries are taxes on nominal interest income, nominal interest ceilings, and unremunerated reserve requirements. One way of looking at the role of inflation in increasing the bur- den of these taxes is to express the tax taken as a percentage of the real interest income that would otherwise be involved. Thus a fixed rate of taxation t on nominal interest income, a re- serve requirement at a fixed percentage θ of deposits remunerated at rate r a, or a fixed nominal interest ceiling r–, all represent forms of non-indexed taxation. Their effective rate varies with inflation. All three have higher effective rates as inflation increases. But the degree of variation differs as between the three taxes. As will be shown below, the effective rate for the first two rises rapidly but less than in proportion to inflation (figures 13.6–13.8). The effective tax rate corresponding to the interest ceiling increases more than in pro- portion to the inflation rate; a doubling of inflation more than dou- bles the effective rate of tax. As such, the interest ceiling may be de- scribed as a third-degree, non-indexed, inflation-supersensitive tax. These assertions are now substantiated under the familiar as- sumption that the nominal wholesale interest rate equals a fixed real interest rate ρ plus the expected inflation rate π.32 Fixed Tax Rate on Nominal Interest Income Then if the nominal deposit interest rate r d equals the nominal wholesale rate less a provision µd for deposit-related services, the nominal interest income per dollar deposited will be: rd = ρ + π – µd THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 405 Figure 13.6. Tax Rates at Different Rates of Inflation Rate of tax 1,000 100 Interest ceiling Reserve requirement Tax on interest 10 0 10 20 30 40 50 60 70 Rate of inflation Source: Based on formulas in text. Figure 13.7. Elasticity of Tax Rates at Different Rates of Inflation Rate of tax 300 Double tax 250 Interest ceiling Reserve requirement ceiling Tax on interest 200 150 100 50 0 0 5 10 15 Rate of inflation Source: Based on formulas in text. 406 PATRICK HONOHAN Figure 13.8. Net Interest Rates and Inflation, 1995–99 log-inflation 6 5 4 3 2 1 0 –1 –2 0 2 4 6 8 10 12 14 Net interest margins Note: See note to table 13.3 whereas the real interest income will be ρ – µd. Thus the tax col- lected expressed as a share of the pre-tax real interest income is: ρ + π − µd z=t . ρ − µd The elasticity of the rate of tax with respect to π is: π η= ρ + π − µd It is easily seen that η < 1 if ρ > µd and that η ➝ 1 as π ➝ ∞. Thus the tax on interest income is supersensitive and non-indexed of the second degree.33 Reserve Requirement For the reserve requirement, the nominal opportunity cost of the funds raised in deposits is the nominal wholesale interest rate r m = ρ + π. Instead of receiving this amount, the bank receives only the reserve remuneration rate r r on a fraction θ of the funds raised.34 THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 407 Thus the implicit tax can be written as θ (r m – r r), which, expressed as a share of the real opportunity cost of the funds is: ρ + π − rr z=θ ρ The elasticity of this tax rate with respect to the rate of inflation π is: π η= . ρ + π − rr For π = 0, η = 0; and η ➝ 1 as π ➝ ∞. Tax-through Interest Ceiling In this case the basic assumption is that the uncontrolled interest rate on lending, r a, exceeds the wholesale rate by a fixed provision, µa, for credit appraisal and loan-losses. By controlling the lending interest rate at – r , the government is providing an implicit transfer (tax from lender, subsidy to borrower) per dollar lent of r a – – r , if this is nonnegative. Expressed as a share of what would be the share of the pre-tax real interest income, the implicit tax rate can be written:  ρ + π + µa − r   z = max  a , 0,   ρ+µ   and the elasticity of this tax rate with respect to π is:   π   z = max  , 0 .  ρ + π + µ − r  a  In this case, for –r > r a, η > 1 and η ➝ 1 as π ➝ ∞. This is a su- persensitive tax with non-indexation of the third order, given that the elasticity of the effective tax rate with respect to inflation re- mains greater than one: much greater, over a certain range of values of inflation. These patterns are illustrated in figures 13.6 and 13.7 for partic- ular parameter values: ρ = 0.05; t = 0.25; µd = 0.01; µa = 0.03; θ = 0.25; –r = 0.1; r r = 0.03. It can be seen clearly from these plots that the effective rate of taxation expressed as a percentage of the real rate of return (or opportunity cost) of the funds involved can quickly 408 PATRICK HONOHAN exceed 100 percent. Indeed for the parameters chosen, which are not unrealistic, this point is passed for quite modest inflation rates in the range of 10 to 12 percent. Where the providers of funds to the intermediary are also taxed on the nominal interest received, without deduction for tax already paid by the intermediary, the burden is increased once again. This typically just increases the rate of tax without much altering the elasticity substantially. If the depositor has the alternative of receiv- ing non-intermediated interest income that is not taxed, it may be more appropriate to see the base of the double tax as being simply the intermediation margin. For the numerical example used in fig- ure 13.6, a double interest tax (already representing a very high rate of tax on the intermediation margin even at zero inflation) also in- creases rapidly, and, though its elasticity with respect to the infla- tion rate is lower, this also converges to unity. The calculations suggest summary ways of quantifying the degree of non-indexation of the various financial sector taxes with respect to inflation. The increase in the effective rate of tax as inflation goes from zero to ten percent could be one measure, which can be called the tax’s inflation gradient. The limiting elasticity of the effective tax rate with respect to inflation as inflation becomes very large could be the other. All taxes in a perfectly indexed tax system would have a zero gradient and zero limiting elasticity. The three interest taxes discussed have, for the parameters shown, gradients of 63, 80, and 100 percent, respectively. The double taxa- tion (expressed as a percentage of intermediation margin) has a gra- dient of 125 percent. All have limiting elasticities of 1. The gradients are obviously high. A value of 100 implies that inflation fluctuating between zero and ten percent results in an effective tax rate that fluc- tuates by 100 percent of the base. The limiting elasticity of 1 implies that the tax rate is unlimited as inflation grows. Actually any limit- ing elasticity greater than zero has the same implication (and even if the limiting elasticity is zero, the effective tax rate might still increase without limit, though more slowly as inflation increases). The chaotic conditions caused by an unlimited rate of inflation tax as hy- perinflation kicks in strongly suggests that any economy prone to surges of high inflation should ensure that its tax system has a limit- ing elasticity with respect to inflation no greater than zero. Optimal Degree of Supersensitivity Supersensitivity implies a high variation of effective tax rates where inflation is variable. Could this be good? It seems unlikely from the THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 409 point of view of a stable and healthy development of human and or- ganizational capital in the financial services industry. If monetary fi- nancing is primarily used in response to unanticipated deficits (the residual financing argument), then an induced volatility in the tax pressure on financial intermediation can hardly be a desired side ef- fect. Equally, if fluctuations in inflation tax are seen as optimally programmed in line with fluctuations in the burden of other taxes (tax smoothing arguments), an induced amplification of some of those other taxes seems equally unlikely to be optimal. (After all, such arguments imply that inflation tax may optimally take up some of the pressure when other taxes are also high and as such causing distortions. Increasing a tax that exacerbates the distortions of existing taxes is not helpful in that context.) Above all, it can be assumed that governments do not anticipate the indirect impact on financial intermediation that this study is considering. Thus even if setting zero as the optimal value of the gradient may be asking for a lot, a figure near 20 percent might be a realistic pol- icy ceiling. Incidence of Inflation Tax Interactions Absent behavioral responses, the increase in the tax take will fall on the legal person liable to pay. In the case of the reserve requirement tax, for example, the shareholders of the bank will bear the entire burden. Especially with such high effective tax rates, however, the assumption of no behavioral response is hopelessly unrealistic. For banks, it may be assumed that the main burden of the tax will be divided among depositors, borrowers, and bank sharehold- ers. There could also be a burden on employees and suppliers of the banks. In a reasonably full employment context with smooth ad- justment of the labor market, however, they would be affected only to the extent that they had sector-specific human capital deployed in the industry. Whatever behavioral assumptions are made, it is clear that siz- able behavioral responses can be expected to supersensitive taxes at high inflation. Naturally, the effects will be larger the closer the un- taxed substitutes. Thus in particular, the extent to which dollariza- tion has developed ready onshore alternatives to local currency banking products will be an important factor in the scale of the behavioral response (and the revenue impact of a particular super- sensitive tax). Also crucial is whether there are untaxed near-bank financial intermediaries competing with the taxed banking sector. 410 Table 13.4. Inflation and Net Interest Margins, 1995–99 Equation 1 2 3 4 Variable Estimate (t stat) Estimate (t stat) Estimate (t stat) Estimate (t stat) Constant 1.105 (2.4) 3.658 (4.0) –0.000 (0.0) 0.008 (0.0) Inflation 0.011 (2.3) 0.011 (2.7) 0.418 (2.7) 0.101 (2.5) Inflation^2ϫ103 — — — — — — –1.932 (2.8) Inflation^3ϫ106 — — — — — — 6.630 (2.9) log (GNP/Cap) –0.426 (3.1) –0.360 (4.1) — — — — Return on assets 0.456 (3.7) 0.570 (4.4) 0.559 (4.4) Overheads 1.099 (11.2) 0.950 (11.1) 0.937 (8.7) 0.970 (9.1) Res. requirement ϫ inflation — — — — — — — — Countries All All All All Functional form level level log log Method/no. obs OLS 68 OLS 68 OLS 71 OLS 71 RSQ/DW 0.813 2.07 0.872 2.23 0.845 2.22 0.864 2.09 (Table continues on next page.) Table 13.4. (continued) Inflation and Net Interest Margins, 1995–99 Equation 5 6 7 Variable Estimate (t stat) Estimate (t stat) Estimate (t stat) Constant 4.716 (4.9) 5.573 (3.1) 2.076 (5.9) Inflation 1.324 (5.6) 1.637 (9.4) log (GNP/Cap) –0.463 (5.0) –0.351 (2.0) — — Return on assets — — — — — — Overheads 1.055 (11.9) — — — — Res. requirement ϫ inflation ϫ 103 0.473 (2.5) — — — — Countries All All All Functional form log log log Method/no. obs OLS 68 OLS 68 OLS 71 RSQ/DW 0.844 2.05 0.587 2.42 0.563 2.35 Note: Dependent variable is net interest margin of banks as a percentage of total assets (country average). For each equation is shown the number of observations, the method (ordinary least squares), and the R-rated and Durbin-Watson statistics. Source: Author’s calculations. See note to table 13.3. 411 412 PATRICK HONOHAN Offshore finance, non-depository onshore finance, and informal fi- nance are also obviously important substitutes for banking, though to a lesser extent. If inflation merely passed through to all nominal interest rates on banking assets and liabilities on a one-for-one basis, then there would be no impact on net interest margins and no inflation tax at all. The interaction with the tax system is likely the major reason why instead one observes a clear sizable impact of inflation on bank net interest margins expressed as a share of total assets (figure 13.8, table 13.4). Here, as with the other tables, a selection only of the most significant results is shown. (For example, the table does not report a regression including both inflation and the interaction of inflation with reserve requirements as they are not both significant when included together.)35 The point estimate in the simplest of the regression equations re- ported in table 13.4 suggests that a doubling of inflation (say, from 5 to 10 percent, or from 10 to 20 percent) can widen the net inter- est margin by 115 basis points. Considering that the median net interest margin in the 70 countries used was just over 400 basis points, this a sizable effect. Not all of this need be through a tax ef- fect—for instance, it is easy to think of reasons why the average risk of a bank’s portfolio would increase with inflation—but the results indicate the potential magnitudes involved. Even if the tax interaction with inflation can be passed through to the intermediary’s customers through the net interest margin, this contributes to the reduced scale of intermediation and the resulting fluctuations in real intermediation activity discourage the develop- ment of specialized human capital in loan appraisal skills. Improving the Indexation of the Tax System The fiscal authorities are rarely set up to consider, in an explicit way, the impact of inflation on the budget or the interactions between an imperfectly indexed formal tax system and the rate of inflation. Understanding the fiscal impact of inflation and how it impacts the financial system is important for developing good policies. Nowadays, fewer and fewer governments rely on printing money; the fiscal benefits of inflation come in more indirect forms. Although inflation shrinks the financial system, it can be associ- ated with increased bank profitability. Yet because interest forms the bulk of their gross revenue, the distortions caused by the inter- action between inflation and other non-indexed taxes can be par- ticularly severe for banks and other financial intermediaries. THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 413 Our proposed measures of non-indexation (the tax gradient and the limiting elasticity) can help capture some of the most potentially damaging aspects of tax—inflation interaction. Improving the in- dexation of the tax system so far as the computation of interest is concerned would reduce the sensitivity of effective tax rates on intermediation. Notes 1. For instance Chari, Christiano, and Kehoe (1996) show that, whether the role of money is characterized in terms of “cash or credit in ad- vance of payment” or in terms of “money in the utility function,” the opti- mality of a zero inflation tax depends on homotheticity and separability properties of the relevant utility functions. Essentially, the assumptions re- quired are those that relegate the role of money to being an intermediate good in the production of utility. See also Correia and Teles (1997). 2. See Alogoskoufis and van der Ploeg (1994), which presents condi- tions under which inflationary financing of an increase in government spending will be better than tax-financing for growth. 3. The existence of an otherwise untaxed underground economy could be another justification, although perhaps not a strong one (Nicolini 1998). 4. According to Goodfriend and Hargreaves (1983), even in the United States, revenue was the original motive for introducing reserve requirements. 5. See Espinosa-Vega (1995); Espinosa-Vega and Russell (1999, 2001); and Chang (1994). 6. See Espinosa-Vega and Yip (2000) and Mourmouras and Russell (1992). Looking at it from another perspective, the deposit tax equals a re- serve requirement plus an open market operation (Bacchetta and Caminal 1992). 7. The positive correlation between seigniorage and conventional taxes predicted, for example, by Trehan and Walsh (1990), is based on the idea that the shocks come from spending. But other patterns of shock are possi- ble. Click (2000) shows empirically how interaction between exogenous shocks to any debt, seigniorage, or taxation feed through to the others con- temporaneously and over time. See also Mankiw (1987) and Poterba and Rotemberg (1990). 8. This is especially the case where the efficiency of the tax system is underdeveloped—a consequence perhaps of a polarized society where each side is reluctant to improve the permanent arrangements for tax collection in case the other side misuses the revenue. See Cukierman, Edwards, and Tabellini (1992), who provide some evidence that reliance on seigniorage is higher in politically polarized societies. 414 PATRICK HONOHAN 9. The volatility of inflation tax revenues or, more specifically, the de- gree of unpredictable fluctuation, does appear to be similar to that of deficits (Calvo and Guidotti 1993). 10. See Sarel (1996); Bruno and Easterly (1998); Fischer, Sahay, and Vegh (2002); and Khan and Senhadji (2000). 11. For Cooley and Hansen’s calibration, the welfare costs of this over- economizing on transactions balances is less than the saving made by being able to reduce income tax rates. 12. Using a calibrated model of precautionary balances, Ïmrohoroglu ˘ and Prescott deduce that the key impact of inflation or other financial sec- tor taxes is their impact on the real rate of return on deposits. As it is, this influences the degree to which agents will over-economize on precautionary balances. A tax that lowers the real rate of return by 5 percentage points is estimated to be equivalent to a loss of about 0.5 percent of average con- sumption. The model is subject to the criticism that precautionary savings may in the real world be held in other forms not subject to inflation tax. In- terestingly, despite fully simulating the stochastic dynamic programming problem of the household, this model does not predict any adverse effect of variations in inflation: only the mean effect on rate of return matters. 13. See Anand and Wijnbergen (1989); Fry, Goodhart, and Almeida (1996); and Goff and Toma (1993). See also Drazen (1985); Honohan (1991); and Robinson and Stella (1993). 14. Sometimes, as with the United Kingdom’s Exchange Equalization Account, the foreign exchange received may be transferred to the owner- ship of the government. 15. Discount window lending by the Deutsche Bundesbank was, until recently, a prominent example. The below-market interest rates charged on this reflected the low remuneration on banks’ deposits with the Bundes- bank. 16. The process is seen in sharp outline in the European Monetary Union as different member central banks are faced with the question of how to account for unredeemed legacy currency notes. One approach is to recognize that some of these notes are “dead” (lost or destroyed) and as such will never be presented. Of course making an assumption on these lines gives an accounting windfall to the central bank. Should this windfall be transferred as a special dividend to the government, or would that be inflationary? 17. The analysis requires considerable regrouping of categories in IFS, as the breakdown of the accounts of the monetary authority are presented in widely differing ways for different countries. 18. I am indebted to Klaus Schmidt-Hebbel for stressing this point. This could be especially relevant where a central bank wishes to conceal the true scale of its support to the market. Certainly, some central banks have been found in several celebrated instances to have undertaken future commit- ments of one sort or another that had the effect of reducing their usable for- THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 415 eign exchange reserves. And some central bank lending to local banks will have supported onlending to government. It seems less likely that much di- rect lending to government has been concealed in such ways, and it is this direct lending that our data purport to measure. 19. The effect is small, however: about 5 basis points in additional profit for each percentage point of total assets backed by equity, seemingly implying a marginal rate of return on equity of just 5 percent. 20. For a country with a per capita GDP of $1,000, for example, all but about 3 to 5 percent of overhead costs are recovered, on average. For a country ten times richer, however, the estimated recovery rate is much lower; in rich countries, high overheads mean lower profits. 21. Per capita income becomes wholly insignificant if an institutional dummy measuring the quality of contract enforcement in the economy (BERI) is included. Another factor included is bank concentration, which could well influence the banking system’s ability to capture some of the in- flation tax (see Baltensperger and Jordan 1997). 22. This contrasts with the approach of Saunders and Schumacher (2000), who instead use each bank’s non-interest earning assets as an ap- proximation to required reserves. Evidently both are imperfect proxies. Overall reserves include non-compulsory reserves, and in some countries some compulsory reserves are interest-earning even if at an off-market rate. 23. Equations 1-5 in table 13.2 display the cross-country correlation of return on assets (profit before tax) with inflation in the 1988–95 period. Using the significant macro indicators of D-H (1999), equation 1 finds (as they did) an insignificant, though positive, coefficient on the rate of infla- tion. Brazil is an outlier in the inflation data, however, and removing it (equation 2) uncovers a strong and significant positive relationship. Actu- ally, a more plausible functional form is to use the log-inflation (as the im- pact of a 1 percentage point change in inflation is unlikely to be the same at high inflation rates). Substituting this results in the per capita income variable becoming insignificant (equations 3 and 4). Inflation is significant with this functional form even if Brazil is included (equation 5). 24. Equations 6 and 7 of table 13.2 explore the tax variables examined in D-H, again using country averages. Dropping Brazil again allows infla- tion to become significant. The significance of the reserve holdings is am- plified if interacted with inflation (equation 8). This result appears to be driven at least partly by the high reported profitability of Russian and Ro- manian banks. 25. This impact is somewhat smaller than obtained in table 13.2, equa- tions 4 or 5 in the earlier period, but is in line with equation 8. 26. Measured as return on assets plus overheads as a percentage of assets. 27. This is in line with the theoretical predictions of Aiyagari, Braun, and Eckstein (1998), who emphasize the function of the banking system in supplying transaction services in times of inflation. 416 PATRICK HONOHAN 28. It should be noted that the alternative harmonic functional form for inflation, which embodies a more gradually slowing influence of inflation on financial sector size, actually fits the data better than the threshold re- gression emphasized by Boyd, Levine, and Smith. 29. Estimates by Bali and Thurston (2000); Easterly, Mauro, and Schmidt-Hebbel (1995); Kiguel and Neumeyer (1995); and others—of the rate of inflation that maximes the inflation tax—tend to be much higher than 15 percent. 30. Actually, the piecewise linear regressions estimated by Boyd, Levine, and Smith for market size and activity have significant discontinuities at the imposed break-point. Thus this study prefers to rely on the harmonic re- gressions they report, or simply on the interquartile differences in the data sorted by inflation. 31. There is some ambiguity here, depending on which estimates are used. Figure 13.5 is based on simple interquartile differences. Using the re- gression results suggests that market capitalization does not decline as fast, but that turnover declines more quickly. This study prefers to use the in- terquartile differences, as the regression estimates risk being extrapolated beyond the range where they can be regarded as trustworthy. 32. Actually, this Fisher relationship is itself quite controversial in the tax context. A partial equilibrium argument can be made to the effect that the nominal interest rate may increase more than one-for-one with the rate of inflation because of the need to compensate savers for the fact that nom- inal interest income is fully chargeable to personal income tax. On the other hand, as has been stressed by Feldstein, the ability of borrowers to pay in- terest, and hence their demand for funds, may be reduced in times of infla- tion by other aspects of non-indexation of the tax system (notably lack of indexation of depreciation allowances). Empirically there is no strong evi- dence of such a tax effect, at least for industrial countries where (because of the greater effectiveness of income tax collection) one would expect the effect to be strongest. 33. An alternative, suggested by Klaus Schmidt-Hebbel (personal com- munication), would be to specify these expressions in terms of the inflation factor π/(1 + π), which has the advantage of being bounded by unity and its limiting elasticity with respect to π is zero. 34. Many authors have made the simplifying assumption that all re- serve holdings of banks are unremunerated, but this is far from being the case. In many countries the central bank remunerates excess reserves (those in excess of the compulsory requirement). In some, even the required re- serves are remunerated (see Fry, Goodhart, and Almeida 1996). 35. Of course, even in an indexed system, such as that of Chile, unre- munerated reserve requirements would be expected to affect margins (see Basch and Fuentes 2000). THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 417 References Aiyagari, S. Rao, R. Anton Braun, and Zvi Eckstein. 1998. “Transactions Services, Inflation and Welfare.” Journal of Political Economy 106 (5): 1274–1301. Alogoskoufis, George, and Frederick van der Ploeg. 1994. “Money and Endogenous Growth.” Journal of Money, Credit and Banking 26 (4): 771–91. Anand, Ritu, and Sweder van Wijnbergen. 1989. “Inflation and the Fi- nancing of Government Expenditure: An Introductory Analysis with an Application to Turkey.” World Bank Economic Review 3 (1): 17–38. Bacchetta, Philippe, and Ramon Caminal. 1992. “Optimal Seigniorage and Financial Liberalization.” Journal of International Money and Finance 11: 518–38. Bailey, Martin. 1956. “The Welfare Costs of Inflationary Finance.” Journal of Political Economy 64 (2): 93–110. Bali, Turan G., and Thom Thurston. 2000. “Empirical Estimates of Infla- tion Tax Laffer Surfaces: A 30-country Study.” Journal of Development Economics 63 (2): 529–46. Baltensperger, Ernst, and Thomas J. Jordan. 1997. “Seigniorage, Banking and the Optimal Quantity of Money.” Journal of Banking and Finance 21 (6): 781–96. Barnes, Michelle, John H. Boyd, and Bruce D. Smith. 1999. “Inflation and Asset Returns.” European Economic Review 43 (4–6): 737–54. Basch, Miguel, and Rodrigo Fuentes. 2000. “Macroeconomic Influences on Bank Spreads in Chile, 1990–95.” In Philip Brock and Liliana Rojas- Suarez, eds. Why So High? Understanding Interest Rate Spreads in Latin America. Washington, D.C.: Inter-American Development Bank. Boyd, John H., Ross Levine, and Bruce D. Smith. 2001. “The Impact of Inflation on Financial Sector Performance.” Journal of Monetary Eco- nomics 47 (2): 221–48 Brock, Philip. 1984. “Inflationary Finance in an Open Economy.” Journal of Monetary Economics 14 (1): 37–53. ———. 1989. “Reserve Requirements and the Inflation Tax.” Journal of Money, Credit and Banking 21 (1): 106–21. Bruno, Michael, and William Easterly. 1998. “Inflation Crises and Long- run Growth.” Journal of Monetary Economics 41 (1): 3–26. Calvo, Guillermo A., and Pablo E. Guidotti. 1993. “On the Flexibility of Monetary Policy: The Case of the Optimal Inflation Tax.” Review of Economic Studies 60 (3): 667–87. Chang, Roberto. 1994. “Endogenous Currency Substitution, Inflationary Finance and Welfare.” Journal of Money, Credit and Banking 26 (4): 903–16. 418 PATRICK HONOHAN Chari, V. V., Lawrence J. Christiano, and Patrick J. Kehoe. 1996. “Opti- mality of the Friedman Rule in Economies with Distorting Taxes.” Jour- nal of Monetary Economics 37 (2–3): 203–23. Claessens, Stijn, Aslı Demirgüç-Kunt, and Harry Huizinga. 2000. “How Does Foreign Entry Affect the Domestic Banking Market?” In Stijn Claessens and Marion Jansen, eds. The Internationalization of Financial Services: Issues and Lessons for Developing Countries. Dordrecht, Hol- land: Kluwer. Click, Reid W. 2000. “Seigniorage and Conventional Taxation with Multi- ple Exogenous Shocks.” Journal of Economic Dynamics & Control 24 (8): 1447–79. Cooley, Thomas F., and Gary D. Hansen. 1991. “The Welfare Costs of Moderate Inflations.” Journal of Money, Credit and Banking 23 (2–3): 483–503. Correia, Isabel, and Pedro Teles. 1997. “The Optimal Inflation Tax.” Dis- cussion Paper 123. Federal Reserve Bank of Minneapolis. Cukierman, Alex. 1984. Inflation, Stagflation, Relative Prices and Imper- fect Information. Cambridge: Cambridge University Press. Cukierman, Alex, Sebastian Edwards, and Guido Tabellini. 1992. “Seigniorage and Political Instability.” American Economic Review 102 (3): 537–55. Demirgüç-Kunt, Aslı, and Harry Huizinga. 1999. “Determinants of Com- mercial Bank Interest Margins and Profitability: Some International Ev- idence.” World Bank Economic Review 13 (2): 379–408. Demirgüç-Kunt, Aslı, Luc Laeven, and Ross Levine. 2003. “The Impact of Bank Regulations and Concentration on Bank Efficiency.” World Bank, Washington, D.C. Processed. Diamond, P. A., and J. A. Mirrlees. 1971. “Optimal Taxation and Public Production I: Production Efficiency and II: Tax Rules.” American Eco- nomic Review 61 (1): 8–27 and (2): 261–78. Drazen, Allen. 1985. “A General Measure of Inflation Tax Revenues.” Eco- nomic Letters 17 (4): 327–33. Easterly, William R., Paolo Mauro, and Klaus Schmidt-Hebbel. 1995. “Money Demand and Seigniorage-maximizing Inflation.” Journal of Money, Credit and Banking 27 (2): 583–603. Espinosa-Vega, Marco A. 1995. “Multiple Reserve Requirements.” Journal of Money, Credit and Banking 27 (3): 762–76. Espinosa-Vega, Marco A., and Steven Russell. 1999. “A Public Finance Analysis of Multiple Reserve Requirements.” Working Paper 99-19. Federal Reserve Bank of Atlanta. ———. 2001. “Stability of Steady States in a Model of Pleasant Monetarist Arithmetic.” Working Paper 2001-20. Federal Reserve Bank of Atlanta. Espinosa-Vega, Marco A. and Chong K. Yip. 2000. “Government Financ- ing in an Endogenous Growth Model with Financial Market Restric- tions.” Working Paper 2000-17. Federal Reserve Bank of Atlanta. THE ACCIDENTAL TAX: INFLATION AND THE FINANCIAL SECTOR 419 Feldstein, Martin. 1983. Inflation, Tax Rules, and Capital Formation. Cambridge, Mass.: National Bureau of Economic Research. ———. 1999. “Capital Income Taxes and the Benefit of Price Stability.” In Martin Feldstein, ed. The Costs and Benefits of Price Stability. Chicago: University of Chicago Press. Fischer, Stanley. 1982. “Seigniorage and the Case for a National Money.” Journal of Political Economy 90 (2): 295–313. Fischer, Stanley, Ratna Sahay, and Carlos A. Vegh. 2002. “Modern Hyper- and High Inflations.” Journal of Economic Literature 40 (3): 837–80. Freeman, Scott. 1987. “Reserve Requirements and Optimal Seigniorage.” Journal of Monetary Economics 19 (2): 307–14. Friedman, Milton. 1953. Essays in Positive Economics. Chicago: Univer- sity of Chicago Press. ———. 1971. “Government Revenue from Inflation.” Journal of Political Economy 79 (4): 846–56. Fry, Maxwell J., 1981. “Government Revenue from Monopoly Supply of Currency and Deposits.” Journal of Monetary Economics 8 (2): 261–70. Fry, Maxwell J., Charles A. E. Goodhart, and Alvaro Almeida. 1996. Cen- tral Banking in Developing Countries. London: Routledge. Goff, Brian L., and Mark Toma. 1993. “Optimal Seigniorage, the Gold Standard and Central Bank Financing.” Journal of Money, Credit and Banking 25 (1): 79–95. Goodfriend, Marvin, and Monica Hargreaves. 1983. “A Historical Per- spective on the Rationales and Functions of Reserve Requirements.” Fed- eral Reserve Bank of Richmond Quarterly Review (April/May) http:// www.rich.frb.org/pubs/wpapers/pdfs/wp83-1.pdf Grilli, Vittorio. 1989a. “Seigniorage in Europe.” In M. de Cecco and A. Giovannini, eds., A European Central Bank? Cambridge: Cambridge University Press. ———. 1989b. “Exchange Rates and Seigniorage.” European Economic Review 33 (2–3): 580–87. Honohan, Patrick. 1991, “Inflationary Effects of Deficit Financing.” Jour- nal of Policy Modeling 13 (2): 229–40. Honohan, Patrick. 1996. “Does It Matter How Seigniorage Is Measured?” Applied Financial Economics 6 (3): 293–300. ˘ Ïmrohoroglu, Ays˛ e, and Edward C. Prescott. 1991. “Seigniorage as a Tax: An Empirical Investigation.” Journal of Money, Credit and Banking 23 (2–3): 462–75. Khan, Mohsin S., and A. S. Senhadji. 2000. “Threshold Effects in the Re- lation between Inflation and Growth.” IMF Working Paper WP/00/110. International Monetary Fund, Washington, D.C. Kiguel, Miguel A., and Pablo Andres Neumeyer. 1995. “Seigniorage and In- flation: The Case of Argentina.” Journal of Money, Credit and Banking 27 (3): 672–82. 420 PATRICK HONOHAN Mankiw, N. Gregory. 1987. “The Optimal Collection of Seigniorage: The- ory and Evidence.” Journal of Monetary Economics 20 (2): 327–41. Mourmouras, Alex, and Steven H. Russell. 1992. “Optimal Reserve Re- quirements, Deposit Taxation and the Demand for Money.” Journal of Monetary Economics 30 (1): 129–42. Nicolini, Juan Pablo. 1998. “Tax Evasion and the Optimal Inflation Tax.” Journal of Development Economics 55 (1): 215–32. Phelps, Edward S. 1973. “Inflation in the Theory of Public Finance.” Swedish Journal of Economics 75 (1): 67–72. Poterba, James H., and Julio J. Rotemberg. 1990. “Inflation and Taxation with Optimizing Governments.” Journal of Money, Credit and Banking 22 (1): 1–18. Repullo, Rafael. 1991. Financing Budget Deficits by Seigniorage and Im- plicit Taxation: The Cases of Spain and Portugal. CEPR Discussion Paper No. 583. Centre for Economic Policy Research, London. Robinson, David J., and Peter Stella. 1993. “Amalgamating Central Bank and Fiscal Deficits.” In Mario I. Blejer and Adrienne Cheasty, eds., How to Measure the Fiscal Deficit: Analytical and Methodological Issues. Washington, D.C.: International Monetary Fund. Sarel, Michael. 1996. “Nonlinear Effects of Inflation on Economic Growth.” IMF Staff Papers 43 (1): 199–215. Saunders, Anthony, and Liliana Schumacher. 2000. “The Determinants of Bank Interest Rate Margins: An International Study.” Journal of Inter- national Money and Finance 19 (6): 813–32. Sibert, Anne, and Lihong Liu. 1998. “Government Finance with Currency Substitution.” Journal of International Economics 44 (1): 155–72. Tanzi, Vito. 1977. “Inflation, Lags in Collection and the Real Value of Tax Revenue.” IMF Staff Papers 24: 154–67. Tobin, James. 1978. “A Proposal for International Monetary Reform.” Eastern Economic Journal 4 (3–4): 153–59. Trehan, Brian, and Carl E. Walsh. 1990. “Seigniorage and Tax Smoothing in the United States, 1914–86.” Journal of Monetary Economics 25 (1): 97–112. Index Note: n indicates note (nn more than one note), and italicized numbers indicate figures or tables. ability-to-pay principle, 197–98 allowance for corporate equity accidental tax. See inflation (ACE), 50, 74n25, 205–6, 236n9 accounting, 15 Italy, 225 accrual, 295, 308n12, 310n27 and profits, 217, 219 capital gains tax, 219 Alogoskoufis, George, 413n2 of central banks, 387–88, Amelina, Maria, 286n8 414n16 American Stock Exchange, 334 corporations, 45, 48–49 Angbazo, Lazarus, 257 financial vs. tax, 302–04, arbitrage, 2, 3, 207 309nn21–24, 310nn25–26 as element in tax design, 17–19 insurance companies, 306–07, financial sector transactions, 310n28 11–12 for loan-losses, 292–96, guarding against, 22–23 301–04, 307–8nn1–13, incentives, 232 309nn21–24, 310nn25–26 regulatory capital, 178 for revenues, 48–49 Argentina Soviet-style, 305–06, 307n1 debit taxes, 314–16 switching from charge-off to deposit insurance premium, reserve method, 304–05, 25n16 310n27 education, health, and life yield-to-maturity, 224 protection tax incentives, accretion taxation, 218, 224, 225 150, 151 accrual accounting, 295, 308n12, health insurance, 151 310n27 housing funds, 145, 146, 147 accrual taxation, 57, 218–19, 222, life insurance premiums, 224, 225, 236n13 165n16 ACE. See allowance for corporate revenue productivity, 316, 317 equity (ACE) role of market discipline in Africa, pension fund assets, 138, financial crisis of, 179 140, 141 tax treatment of, 138, 139, 140 agency services, taxation of, 371 Asia, pension fund assets, 138, 140, agents, 348, 370 141 agriculture sector, 173, 286nn7–8 asset managers, 348 Aiyagari, S. Rao, 415n27 assets, 49–50 Allen, Linda, 257 impact of inflation on, 400, allocations, exempt vs. taxable 416nn28–29 activities, 357 risky vs. safe, 107–08 421 422 INDEX asymmetric information, 88–91, Boadway, Robin, 18, 19, 31–80, 359, 169–70 360 Atkinson, Anthony B., 40–42, 43, Bolivia, 146, 147 72–73n2 bonds, 403 Auerbach, Alan J., 58, 62, BOR. See Bank of Russia (BOR) 75nn30–31, 359, 360 borrowers and borrowing, 169–70 Australia, 367 expected return of, 89 taxation system, 372–73 incentives to, 152–54, tax policy developments, 165–66n17 224–25 model for effect on national tax treatment of pensions, saving of tax incentives for, 137–38 161–63 VATs, 350–51 tax treatment of, 303–04 Austria, ACE scheme, 50 Bosnia, 226 automated payments, 11 Boyd, John H., 121–22n11, 400, averaging, 231 416n28, 416n30 Boyle, Phelim P., 336–37 Bailey, Martin, 383, 386 Bradford, David F., 58, 75nn30–31 Bali, Turan G., 416n29 Bradley, Christine, 187n6 Banerjee, Abhijit, 172 Braun, R. Anton, 415n27 Bank of England, 233 Brazil, 21, 50, 266n1, 313 bank failures, 181, 187n6 banking problems in, bank fees, 62, 63 249–50 bankruptcy risk, 111 debit tax study data and Bank of Russia (BOR), 271, 278–79, evidence, 318–22, 282 323–24n6 bank-to-market ratio, 400, 401 explicit taxes, 242, 244–45, Barnes, Michelle, 400 259, 262 Barth, James, 180 financial liberalization in, Bartunek, Ken, 111 242–46, 266n1 Basel Accord, 178, 292–93, 295, housing funds, 145, 146, 308n4 147–48, 165nn10–11 Basel Committee, 295 impact of seigniorage on Becker, Gary, 180 spreads and margins, Bekaert, Geert, 338 257–64, 267n6 Belgium, 203, 218, 367–68 and inflation, 261–62, 392, Bentson, George, 43, 88, 185, 187n6 393, 415n23 bequests, 33, 43 life insurance premiums, Bernheim, Douglas, 129–30, 134 165n16 Besanko, David, 178 overview of banking in, Besley, Timothy, 129–30, 131, 134, 241–42 172 real interest rate, 251 bifurcation practice, 56–57, 225 reserve requirements and Billett, Matthew, 179 seigniorage, 252–57, 267n5 Black, Fisher, 180 revenue productivity, 316, 317 Blume, Lawrence, 337 Brock, Philip L., 169–93, 384 blunt instruments, 178–80 brokerage services, 331 INDEX 423 Bruce, Neil, 47, 74n22 capital importing countries, 47 budget deficits, 270 capital income, 8–9, 22, 32–33, budgets, 1, 6, 7, 23n1 199–206 business tax, 24n9 and corporate income, 44–52 and dual income tax systems, Cabral, Célia, 266n1 203–05, 236n7 Cagetti, Marco, 164n1 household capital income, Cairo and Alexandria Stock 33–44, 73n5, 73nn7–8 Exchange, 25n20 tax treatment of, 198, 203–05, Calem, Paul, 178 236n5, 236n7 call options, 53, 74n24 capital income tax, 52, 113, 114 Calomiris, Charles, 187n6 heterogeneous households, Cameroon, 25n16 40–42, 73n11 Caminal, Ramon, 17–18, 24n5, homogeneous households, 81–126, 181 34–40, 72–73nn1–2, Campbell, John Y., 329, 330, 331–32 73nn4–5, 73nn7–9 Canada, 65, 351 restrictions on tax instruments and corporate tax, 47 and information, 42–43 education, health, and life time consistency issues, 43–44 protection tax incentives, capitalization, 389, 391, 415n19 150 capital requirements, 95–97, 109, financial services, 347 121–22nn11–12, 176–78 retirement savings, 132 capital to loans ratio, 96, 122n12 tax treatment of borrowing, Caprio, Gerard, 175, 178, 180 153 Cardoso, Eliana, 20, 21, 241–68 VATs, 350–51 Carr, Jack L., 171–72 Canadian Customs and Revenue cash, 112 Agency, 355 cash accounting, 310n27 capital cash flows, 10, 18, 56, 121n9, 362 depreciable, 49 basic cash flow tax, 363–64 substitutability of, 112–13 and capital gains, 53 tier I, 295, 302 and income from loans, 298, tier II, 293, 303 299, 300, 301 capital adequacy, 14–15, 25n18, modified cash flow tax, 50 295–96, 308n13 and profits, 67–69, 76n40 capital adequacy ratios, 302 R-based tax, 49 capital controls, 180, 182–83 and TCA, 364–66 capital exporting countries, 74n16 under financial transaction, capital flows, 208, 210 358–59 capital gains, 206 cash flow tax, 41, 48–51, 74n18, and new financial instruments, 74n22, 74n25, 75n30 52–58, 74–75nn24–27, cash transfers, Russia, 276 75nn30–31 CBR. See Central Bank of Russia taxing of, 45–46, 74n13, (CBR) 201–2, 207, 217–19, Central Bank of Brazil, 264 236nn13–14 Central Bank of the Philippines, 294, tax rate of, 211, 213–14 296 424 INDEX Central Bank of Russia (CBR), 270, housing funds, 146, 148, 149 282 reserve requirements, 182–83 credit outlays, 272–74 revenue productivity, 316, 317 credits to FSRs, 275–76 tax treatment of pension directed credits, 271–72, funds, 138, 139, 140 286n5–7 commercial paper, 113 and excess reserves, 283 Commission of the European and GKOs, 177 Communities, 378–79n11 overview, 271 commissions, taxation of, 368–71 refinance rate, 275 commodities, tax issues, 40–41, 43 subsidies, 274–75, 286–87n11 compensatory taxes, 351, 373–75, central banks, 386–89, 390, 379n15 414–15nn14–18, 416n34 competition Chamley, Christophe, 37, 39, 42 for banks, 82–83 charge-off accounting method, 296, and input tax credits, 359 297–301, 304–05, 308nn15–16, international, 21 309nn17–20, 310n27 and project selection, 88 Chari, V.V., 413n1 suppression of, 180 charter value, 185 and taxation on commissions, checks, endorsement of, 322 368 Chevallier, Anne, 329 compliance costs, 219 Chia, N. C., 60, 64, 75n33, 76n36 comprehensive business income tax, Chiappori, Pierre-André, 123n30 217 Chile, 416n35 consideration, 348–49, 350 fiscal costs of banking system, of risk, 366 175 and taxation of fees and life insurance premiums, commissions, 368, 369 165n16 consumer credit, 153 pension funds, 138, 139, 140, consumption, 8, 59, 62, 102–04, 141 107, 122nn18–19 reserve requirements, 182–83 and ACE system, 206 role of market discipline in and capital accumulation, financial crisis of, 179 235–36n4 stamp tax, 22 Euler equation for, 130–31 China, business tax, 24n9 growth of, 130 Christiano, Lawrence J., 413n1 profile of, 87–88 Claessens, Stijn, 391 taxation model, 121n3 Click, Reid, 413n7 and transaction taxes, 11–12 Cnossen, Sijbren, 209 and wealth, 200 Coleman, Wilbur John II, 38, 42, 73n8 See also utility functions collateral, 172, 294 consumption tax, 62, 349, 358, 374 Collor, Fernando, 246 and financial services, 351–52 Collor Plan, 267n3 heterogeneous households, Colombia 40–42, 73n11 deadweight losses, 314 homogeneous households, debit taxes, 313, 314–15, 323n1 34–40, 72–73n2, 73nn4–5, debit tax study data and 73nn7–9 evidence, 318–22 optimal, 360 INDEX 425 restrictions on tax instruments restrictions on expansion of, and information, 42–43 180 time consistency issues, 43–44 Russia, 271, 272–74 consumption utility, 84, 121n2, sources of, 89 122n19 to FSRs, 275–76 contracting-out arrangements, 370, working capital, 274 371 credit card processing data centers, contracts, tied-up contracts, 123n24 370 convertible bonds, 53 credit operations, 356–57 Cook’s distance, 165n9 Cremer, Helmuth, 43 Cooley, Thomas F., 414n11 Croatia, 50, 206, 226, 236n9 Corlett, W.J., 72–73n2 cross-border issues, 130–31, 217, corporate governance, 208 227 corporate income tax, 5, 105 Cruzado Plan, 255 and ACE, 205–06, 236n9 currency, 270, 414n16 bank, 95–97, 109, 111–12, Brazil, 252–54, 266n2 121–22nn11–14 foreign exchange, 275–76 decrease in, 201 and inflation tax, 278, 283, EU, 236n5 284, 384 fund management companies, and seigniorage, 23–24n4 228 overview of, 44–45 Dakolias, Maria, 266n1 tax on dividends, 208–09, deadweight losses, 314, 316–18, 319 211, 213–14, 215–16, 217 dealers, financial instruments trade, and withholding role, 45–47, 331–32, 341n8 74nn13–16 Deaton, Angus, 43 corrective taxes, 2, 3, 70, 187n3 debits, 12 and deposit insurance, 13–14, debit taxes 25n14–17, 173–74 and deadweight losses, 319, 320 loan-loss provisioning, 14–15, disintermediation and, 321 25n18 in Latin America, 314–16, and savings promotion, 15–16 323n1, 323n3 and social issues, 16 overview, 313–14 Tobin taxes, 24n12 and revenue productivity, 315, and unremunerated reserve 317 requirement, 16–17 study data and evidence, Costa Rica, tax treatment of pension 318–22, 323–24nn6–9 funds, 138, 139, 140 debt, 43, 96 cost of capital, 211 debt contracts, 170, 171 costs, 51, 334–35, 385–86 debt to GDP ratio, 270 Cox, John C., 336 and equity, 54, 208 credit, 233, 236n10, 254, 315 and financial innovation, auctions, 272 53–54 CBR allocations, 273 floating rate debt, 250–52, cost of, 172, 187n5 267n4 demand for, 121n10 tax treatment of, 208–09, 211, directed credits, 6, 7, 270, 304 271–72, 275, 286n5–7 debt-equity ratio, 208 426 INDEX default, 110, 170–71, 173–74 deregulation, 347, 368 deferral rules, 55 derivatives, 221, 222, 330, 388 deferred tax assets, 302, 309n22 Dermine, Jean, 110 delta (of option price), 336, 341n9 Detragiache, Enrica, 123n24 de Meza, David, 187n5 developed countries, contract saving Demirgüç-Kunt, Aslı, 111, 123n24, for housing, 145 182, 391, 392 developing countries Denmark, tax system, 137–38, 203, capital requirements, 178 204, 205 pension fund assets, 138, 139, deposit freeze, 246, 267n3 140, 141 deposit insurance, 13–15, 25n14–17, transaction costs, 156 107, 324n8 and VAT system, 375–77 fiscal cost of, 174–81 wealth and saving, 164n1 model of, 181–82, 188n12 Devereux, M.P., 50 necessity for, 170–74, Dewatripont, Mattias, 185, 187n6 187–88nn4–9 Diamond, Douglas, 88, 121n2, 173, and reserve requirement, 174 182–85, 188n14 Diamond, Peter, 59, 61, 65 Russia, 285, 287n20 Dicks-Mireaux, Louis, 136 deposit interest, 22, 107 differential commodity tax, 40–41, 43 depositors, 17, 25n21, 173–74 directed credits, 6, 7, 270–72, 275, deposits 286n5–7 and investment funds, 98–100 direct investments, 122n13 monopoly power in, 101–05, and deposits, 98–100 122nn17–20 market clearing under, 115–16 nominal interest rates on, 283, and Walrasian system, 86–88 284 direct taxes, 265 non-interest bearing, 253–54, disintermediation, 347, 368 255, 267n5 calculation of, 316–18 rates of, 97, 117–21, 122n14, overview, 313–14 179–80, 252 study data, 319, 320, 321 return on, 104 DIT. See dual income tax systems in Russian banking system, (DIT) 284–86, 287n20 dividends spread between interest rates and central banks, 387–88 on loans and deposits, taxation of equity, 211, 217 258–60, 261 tax on, 23n3, 198, 208–09, spread between loan and 211, 235n1 deposit rates, 252, 254, tax rate compared to capital 258, 262, 263 gains tax, 218 deposit tax, 82, 95, 99, 108–09, views of, 207–08, 236n10 121n5, 384 Domowitz, Ian, 338–39 effect on loan market, 110–11 double taxation, 211, 215–16, 217 and loan rates, 111, 123n32 dual imputation system, 217 model of, 91–92 dual income tax systems (DIT), summary of results of, 113–14 203–05, 225, 236n7, 237n18 depreciation, 49, 402 Dybvig, Philip, 121n2, 174 INDEX 427 earnings and debt, 54 retained, 45–46 debt-equity ratio, 208 and zero capital income tax, 9 equity to assets ratio, See also income 121–22n11 Easley, David, 337, 338 and tax systems, 205, 211 East Asia, 128, 145–48 equity financing, 121–22n11 Easterly, William R., 416n29 Ericsson, 329, 332 Eckstein, Zvi, 415n27 Ernest & Young, 378–79n11, economic development, and saving 379n13 incentives, 154 Erosa, Andrés, 34, 38, 39, 73n9 economic distortions, 352–55, 362, estate tax, 165n16 371, 378n5, 402 Euler equation for consumption, economic efficiency principle, 198 130–31, 134 economic profits tax, 48–50 euro, 233 Ecuador European Central Bank, 233 deadweight losses, 314 European Commission, 354–55, 359, debit taxes, 313, 314–16 365–66 debit tax study data and European Monetary System, 232–33 evidence, 318–22, European Monetary Union, 414n16 323–24n8 European Union, 69, 203, 235n2 revenue productivity, 316, 317 corporate income tax, 236n5 Edgar, Tim, 236n10 exemption system, 349, 350 education, 155–56 option system of taxation, savings for, 148–52, 367–68 165nn14–16 tax treatment of equity, 211 efficiency-enhancing actions by tax on wealth, 220 government, 180–81 VAT, 81, 227, 237n22, 354–55 Egypt, 25n20, 26n25 Europe, competition with U.S. from employment, 272, 274 financial institutions, 353 energy sectors, 273–74 exchange rate, 270 enforcement, 9, 115, 178, 246, exemption issues 266n1 corporation, 46 Engel curves, 43 for developing countries, Engen, Eric M., 133, 134 375–77 entrepreneurs dividends, 217 loan market, 93, 105–06 economic distortions created markets for, 113 by exemptions, 352–55, and role of banks in lending, 378n5 88–89 financial services, 345, 355–56 taxation model, 84–85 modified exemption systems, and Walrasian system, 86–87, 361–62, 367–73 121n4 policy options overview, equalizer, 225 361–62, 378–79nn10–11, equity, 15, 43, 96, 201, 203, 379n13, 379n15 235–36n4 and tax base, 357–61, 378n9 allowance for corporate equity, See also value-added tax (VAT) 50, 74n25, 205–06, 236n9 expected return of borrowings, 89 428 INDEX expenses, 304 current approaches to taxation loan losses, 207n3, of, 349–51, 377–78n2 208nn7–11, 292, 294–95, definition of, 355–57, 372 305–06, 307n1 for developing countries, explicit fees, taxation of, 368–71 375–77 explicit taxes, 4–5, 111 and economic changes, 2, and bank spreads, 257 23n2 Brazil, 242, 244–45, 259, 262 economic distortions created by tax exemptions, 352–55, factoring, 10 378n5 F-base for taxation of business exempt vs. taxable, 356 services, 362 foreign supply of, 353 Federal Deposit Insurance importance of volume of, Corporation (FDIC), 176, 187n6 337–38 Federal Reserve, 233–34 imported, 353, 378n5 Federal Savings and Loan Insurance and inflation, 385–86, 402–09, Corporation, 181 414nn11–12, 416nn33–34 Feige, Edgar, 11 mixed supplies, 356–57 Feldstein, Martin, 64n5, 402, 403, purchases by businesses, 65–66 416n32 purchases by consumers, financial instruments, 322 60–65, 75–76nn33–37 and capital gains, 52–58, services incidental or 74–75nn24–27, 75nn30–31 supplementary to, 356 defining tax base, 226–28, structural changes in, 353–54 237n22 supplied by non-financial fixed-income, 330–31, 341n6 businesses, 350 innovations in, 18–19 taxation of, 10–12, 24n10, principal amount of, 358 taxation of, 18–19, 207–08, 59–69, 66–69, 76n42, 220–23, 228–30, 236n10, 76nn38–40, 327, 362–67, 237n18, 237n20 379n11, 379n13 trading volatility, 334–35 and tax base, 357–61, 378n9 financial sector taxation, 10–12, 24n10 and TCA, 364–67, 378–79n11 approaches to reform of, 7–12 treatment of, 50–51 and capital income, 8–9, unbundling of prices for, 32–44, 199–206 368–69 criteria and guidelines for, 3–4 zero-rating of, 363, 372 tax design elements, 17–19 See also exemption issues; and VAT, 9–10, 24nn7–8, transactions and 345–77 transaction services financial services and transactions Finland, dual income tax system, assessment of current system 204, 204, 205 of taxation, 351–57 Fisher relationship, 404, 416n32 and cash flows, 51, 74n22, fixed factors, ownership of, 48 358–59, 362, 363–64 fixed fees, 62 categories of, 346–49, 377n1 fixed-income instruments, 330–31, compensatory taxes using 341n6 addition method, 373–75, fixed input tax recovery method, 379n15 372 INDEX 429 flat tax, 2 Gale, William G., 133, 134, 136, 137 floating rate debt, 250–52, 267n4 Garcia, Gillian, 175 forecasting, and provisioning, 15, Garfinkel, Jon, 179 25n18 Gennotte, Gerard, 178 foreign exchange, 275–76, 285, Germany, 74n14, 221, 266n1 414n14 borrowing, tax treatment of, backing of money base, 388–89 153 interbank, 322 capital gains tax, 218 foreign-owned companies, 9, 46–47 contract saving for housing, 145 Former Soviet Republics (FSRs), education, health, and life 275–76 protection tax incentives, formula allocation, 357 150 forward contracts, 53, 224 loan-losses, 297 401k plans, 131, 133–34 new financial instruments, France, 202, 203, 222 221, 222 borrowing, tax treatment of, option system of taxation, 153 367–68 capital gains tax, 218 reserve requirements, 232, 233 and compensatory taxes, 351, retirement savings, 132 373–74 tax system, 203, 205, 211, contract saving for housing, 367–68 145 wealth taxes, 220 dividend imputation, 217 Gertler, Mark, 121–22n11 education, health, and life Gervais, Martin, 34, 38, 39, 73n9 protection tax incentives, GKOs. See rouble Treasury bills 151 (GKOs) hedging, 223, 224 Glen, Jack, 338–39 life insurance premiums, gold, 377n1 165n14 goods and services, 349, 352, 353 new financial instruments tax, Gordon, Roger, 43, 62, 74n16, 359, 221 360 option system of taxation, government bonds, 212, 276–77 367–68 government securities, 6 pensions, 137 Granville, Brigitte, 20, 269–88 reserve requirements, 232, 233 Greenwald, Bruce C., 170, 172 retirement plans, 131, 132 gross domestic product (GDP) stock market, 329 and credits in Russia, 273–74 Freeman, Harold, 50 credits to FSRs, 276 Freeman, Scott, 384 loans as percent of in Brazil, Freinkman, Lev., 286n6, 286n11 243, 246, 247, 250 French, Kenneth, 334 and pension fund assets, 138, Friedman, Milton, 383, 384 140, 141, 142–43, Froot, Kenneth A., 329, 330, 331–32 164–65n6 fund management services, 228–30 ratio of domestic debt to, 270 futures, 222, 224 ratio relationship to savings in Russia, 285 Galbi, Douglas, 286n8 ratio of revenues as percent of, Gale, Douglas, 170 316 430 INDEX gross domestic product (GDP) policy environment of, 42–43 (continued) time consistency issues, 43–44 relationship to seigniorage households seized, 256 low-income, 148 and revenues in Peru, 315 panel data, 130 and subsidies in Russia, 274, housing, 155–56 275 saving for, 16, 128, 129, 134, growth, inflation and, 384, 413n2 144–48, 149, 165nn10–12 Grubert, Harry, 60 Huber weights, 165n9 Guo, Lin, 181 Huizinga, Harry, 69, 111, 182, 355, 391, 392 Habermeier, Karl, 325–43 Hungary, claims cases, 266n1 Hague, D.C., 72–73n2 Hu, Shing-yang, 329 Haig-Simons (H-S) concept of income, hybrid instruments, 221 199–200, 205, 218, 219, 226 Haik, Narayan Y., 333 Iceland, 137 Hall, Robert E., 130 implicit subsidies, 4 Hammond, Suzanne, 341n5 implicit taxes, 10, 62, 111 Hansch, Oliver, 333 and bank spreads, 257 Hansen, Gary D., 414n11 Brazil, 243 Harberger, Arnold C., 317 reserve requirements, 230–34 Hasbrouck, Joel, 333 Russia, 270 Hau, Harald, 329 on stock markets, 23–24n4 health insurance, 151 imputation system, 215–16, 217, 226 health, savings for, 148–52, Ïmrohorodlu, Ays e, 414n12 165nn14–16 incentives, 40–41, 90 hedging, 221, 223–24 capital, 106 Hellman, Thomas, 108, 178 for education, health, and life Hellwig, Martin, 170 protection, 150, 151 heterogeneous households, 40–42, for saving, 127 73n11 income hidden reserves, 310n26 and capital, 3, 221–22 Hills, John, 154 and cash flow from loans, 298, Holmstrom, Bergt, 88 299, 300, 301 homogeneous households, 34–40 definition of, 199–200 Hong Kong, 329 and reserve requirements, 384 Honohan, Patrick, 1–28, 129–30, treatment of, 6, 23n3 175, 178, 267n6, 381–420 See also capital income Ho, Thomas S., 257 income taxes, 262, 378n9 household capital income indexation, 246, 249, 412–13 heterogeneous households, indexing rate, 365–66 40–42, 73n11 India homogeneous households, education, health, and life 34–40, 72–73nn1–2, protection tax incentives, 73nn4–5, 73nn7–9 150, 151 overview of taxation issues, health insurance, 151 33–34 housing banks, 145 INDEX 431 life insurance premiums, and reserve requirements, 254, 165n16 405–07, 416n34 indirect taxes, 265 response of financial system Individual Retirement Accounts to, 385–86, 414nn11–12 (IRAs), 131, 133–34, 164n3 Russia, 270 Indonesia and seigniorage, 255–56 fiscal costs of banking system, sensitivity to, 19–20 175 steady vs. changing, 403, 405 housing banks, 145 as a tax, 6, 19, 23–24n4, life insurance premiums, 382–85, 413nn1–4, 165n16 413nn6–9 tax treatment of pension and tax revenues, 386–89, funds, 138, 139, 140 390, 414–15nn14–18 industrial countries, 199–200 and tax-through interest and equity, 211 ceiling, 407–08 reserve requirements, 232 inflation rates, 23, 257, 385 saving incentives, 132 Brazil, 392, 393, 415n23 tax of interest income in, 212 15% theory, 400, 416n29 infinite-lived dynastic model, 33, inflows of funds, 67–68 36–38, 73n5, 73nn7–8 information inflation, 3, 19–20, 22–23, 112, 141 asymmetric, 88–91, 169–70, and banking value-added, 396, 187n3 399 and capital income, 9, 24n6 Brazil, 241–42, 246, 249, 257, and deposit insurance, 13 261–62 restrictions on, 42–43 costs of, 261–62, 385–86 and stock trading, 334, 337–38 government use of inflation information technology, 82 tax, 383–85, 413nn1–4, input tax credits, 357, 359, 360, 413nn6–9 371–72 impact on bank asset size, 400, insurance 416nn28–29 funds, tax treatment of, 228, impact on bank profitability, 229, 230 391–96, 397–98, liquidity, 97 415nn19–23 policies, 307 impact on stock market premiums, 361–62 activity, 400–02, pure insurance, 68–69 416nn30–31 taxation of, 371 inflation-targeting policy, insurance companies, 228, 229, 230, 251–52 296, 307n3, 309n23 inflation tax on currency, 278, financial and tax treatment of, 283, 284 306–07, 310n28 interaction with tax systems, as providers of financial 409–12, 416n35 services, 347 and net interest margins, inter-dealer markets, 333 410–11 interest, 5–6 and non-indexed tax system, taxing of corporate funding, 402–09, 416nn33–34 208–09, 210, 211 432 INDEX interest (continued) investors, 88 taxing of household loans, taxation model, 83–85, 208, 209 121nn2–3 tax on, 95, 212 and Walrasian system, 86 views of, 207–08, 236n10 Israel, 373, 374 interest margin, 104 Italy, 58 interest rates, 21, 68, 102, 110, 241 capital gains tax, 208–09, 218 below market, 387, 414n15 education, health, and life ceilings on, 6, 7, 20, 407–08 protection tax incentives, elasticity of, 164n1 150, 151 inflation and, 410–11 life insurance premiums, on loans and deposits, 254, 165n14 283, 284 retirement savings, 132 and non-performing loans, tax policy development, 259, 262, 263 225–26 pure rate, 59 tax treatment of borrowing, on safe assets, 123n32 153 and savings, 129–34, 133 and spreads in Brazil, 251, Jacklin, Charles, 88 258, 259 Jack, William, 62, 360 subsidies on, 275 Jamaica, 146, 148, 149 See also real interest rates James, Christopher, 188n7 interest rate swaps, 25n22, 55, Japan 74–75n27 capital gains tax, 218 International Accounting Standards, education, health, and life 292, 293, 307n3, 308n12 protection tax incentives, international cooperation, 46, 74n15 150, 151 international finance, 326 life insurance premiums, International Financial Statistics 165n14 (IFS), 314n17, 388, 390 retirement plans, 131, 132 International Labour Organization, tax treatment of borrowing, 272 153 international tax codes, 128 tax treatment of pensions, 137 intertemporal choice, 129, 144 transaction taxes and price inventory management, 333 volatility, 329 investments, 82–83, 97–100, Jappelli, Tullio, 127–68 122nn15–16 John, Kose, 172 banks taxation in absence of, John, Teresa A., 172 91–97, 121n5, Jones, Charles M., 328 121–22nn9–11 Jones, David, 178 expenses of, 378n9 Jordan, 145 and insurance reserves, Jordan, John S., 179 306–07, 310n28 return on, 104 Kanatas, George, 178 taxation model, 84–85 Kane, Edward, 178 and Walrasian system, 86–91, Kan, Kamhon, 328, 329 121n4 Kaufman, George, 175, 186 INDEX 433 Keeley, Michael, 123n24, 185 leisure, 40, 41 Keen, Michael, 18, 19, 31–80, Leland, Hayne E., 341n11 236n9, 359, 360 lenders Kehoe, Patrick J., 413n1 lender of last resort, 185 Kiguel, Miguel A., 416n29 tax treatment of, 303–04 King, John, 236n9 less developed (LDC) countries, King, Mervyn A., 72–73n2, 136 pension fund assets, 142, 143, Kirilenko, Andrei, 313–24, 325–43 165n9 Klingebiel, Daniela, 175 Levine, Ross, 180, 400, 416n28, Korea, 145, 329 416n30 Koyama, Sérgio Mikio, 258 Levin, Mattias, 197–240 Kroszner, Randall, 181 liability, 170–72, 178 Kyrgyz Republic, 297 liberalization, 123n24, 123–24n37 in Brazil, 242–46, 266n1 labor See also reform heterogeneous households and life insurance, 151–52 taxation, 40–42, 73n11 life protection, savings for, 148–52, homogeneous households and 165nn14–16 taxation, 34–40, Lindgren, Carl-Johan, 174–75 72–73nn1–2, 73nn4–5, liquidity, 17–18, 85, 185 73nn7–9 liquidity insurance, 92, 97, 100 and restrictions on tax loan-losses, 5 instruments and accounting for, 292–96, information, 42–43 307–8nn1–13 substitutability of, 112–13 as future expense, 294–95, time consistency issues, 43–44 308n7 used in financial services, 373, overview, 291 374 Russia, 305–06 Laffer curve, 256, 400 tax treatment of, 296–304, Latin America, 128 308nn15–16, 309nn17–24, deadweight welfare loss, 310nn25–26 316–18 loan-loss provisioning, 14–15, debit taxes in, 12, 309n21, 25n18, 296–97 314–16, 318–22, 323n1, loans, 88, 113, 114, 169–70, 185, 323n3, 323–24nn6–9 265 mandatory savings to housing and accrual accounting, 295, funds, 145–48, 308n12 165nn10–11 agricultural sector in Russia, pension fund assets, 138, 140, 273, 286nn7–8 141 Brazil, 243 revenue productivity, 316, 317 and cash flow tax, 363–64 tax treatment of, 138, 139, 140 charge-off vs. reserve See also names of specific accounting method, countries 297–301, 308n16, Lea, Michael J., 145 309nn17–20 legislation, new financial and debt contracts, 187n4 instruments, 220–21 deductions for, 208, 209 434 INDEX loans (continued) market power, 21, 101–06, and deposit insurance, 176–78 122nn17–23 and deposits separability, market segmentation, 338 109–13, 123n30, 123n32 market structure, 107, 123n24 deposits tax, 91–92 marking-to-market, 57, 218, 219, income and cash flow from, 222–25 298, 299, 300, 301 Marshall, David, 185 interest rates on, 91, 254, 283, Martinez Peria, Maria Soledad, 179 284 Mathewson, G. Frank, 171–72 maturities of, 21, 26n25 Matutes, Carmen, 101, 123n24, 185 monopoly power in, 105–06, Mauro, Paolo, 416n29 122nn21–23 Meade Committee, 49, 57–58, 362 non-performing, 259, 262, medical expenses, 155–56 294, 308n6, 309n21 Meghir, Costas, 129–30, 131, 134 and reserve requirements, Merton, Robert C., 174 293–95 Mexico risk and, 176–78 housing funds, 146, 148 Russia, 285 life insurance premiums, and security markets, 100–01 165n16 and seigniorage, 255, 256, 257 role of market discipline in and spreads in Brazil, 252, financial crisis of, 179 258–60, 261, 262, 263 tax treatment of pension taxation of, 66, 76n39, 82, funds, 138, 139, 140 92–93, 96, 97, 106, 114, MIC. See military-industrial complex 122n23 (MIC) value of, 292 middle-income countries, saving London Stock Exchange, 333 incentives, 128 long-term assets, 87, 90 migrants, savings of, 166n18 low-income households, 148 military conversion credits, 274 Lucas, Robert, 122n18 military-industrial complex (MIC), lump-sum income, 42 274 Luxembourg, tax rates, 203 Miller, Merton, 176, 180 Lyons, Richard K., 333 Mingo, John, 178 Mintz, Jack, 74n22 Mackie, James, 60 Mirrlees, James A., 40, 59, 61, 65 macroeconomics, and taxation, models 112–13, 123–24n37 deposit insurance, 173–74, Madhavan, Ananth, 333, 334, 335, 181–82, 188n12 338–39 effect on national saving of tax Madura, Jeff, 111 incentives for borrowing, Malaysia, 138, 140, 141, 148, 149 161–63 management interest, 135, 164n4 heterogeneous households, manufacturing sector, 208, 210 40–42 Marcus, Alan, 185 homogeneous households, margin pricing, 368–69 34–40, 72–73nn1–2, margins, 257–64, 267n6 73nn4–5, 73nn7–8 market access, 354 investors taxation model, market discipline, 179 83–85, 121nn2–3 INDEX 435 overlapping-generations monopoly rents, 105 model, 33, 38, 73n9 moral hazard, 88, 107, 170 tax on deposits, 91–92 asset-substitution, 178 tax treatment of mandatory control of, 173 contributions to saving, and deposit insurance, 13–14, 158–61, 166n19 185 two-period model of saving, and liability, 171–72 157–58 mortgages, 129, 144–45, 152–53 modified cash flow tax, 50 Moscow Interbank Currency Moldova, 393 Exchange, 277 Monetary Control Act, 233–34 Moscowitz, Tobias J., 164n4 monetary policy, 123–24n37 Multilateral Investment Fund, as corrective quasi-tax 166n18 instruments, 169, Murdock, Kevin, 108, 178 186–87n1, 187n3 mutual funds, 102, 122n16 and fiscal policy, 272 goals of, 269 Nagarajan, S., 185 and inflation, 403 Nakane, Marcio I., 258 post Real Plan period, 250–52 natural resources, 46, 48–50, 74n12 money Netherlands central banks and government retirement savings, 132 role in, 386–89, tax system, 205 414–15nn14–18 tax treatment of borrowing, foreign exchange backing of 153 base money, 388–89, 390 wealth taxes, 220 rate of growth compared to net margins, 251, 252, 258, 260, inflation, 385 264, 265 role of, 384, 413n1 net-of-tax return on capital, 262 money market funds, 97, 122n15, Neumeyer, Pablo Andres, 416n29 322 New York Stock Exchange (NYSE), money transfers, 322 333, 334 monitoring New Zealand, 138, 222, 223, 224, of bank loans, 114 367 by depositors, 173–74 taxation system, 371 costs, 110, 219 VATs, 351 and deposit insurance, 172, noise trading, 337–38 187n6 nominal interest, 402 of entrepreneurs, 85, 88 fixed tax rate on, 404–06, and liability, 171–72 416n33 and monopoly power, 106–07 rates, 283, 284, 412 of previous customers, 122n21 nonbank financing, 93 rents and, 123n25 non-indexed tax system, 19, 402–09, monopoly power 416nn32–34 in deposit market, 101–05, non-prime rates, 262, 263, 267n6 122nn17–20 nonregistrants, taxation of financial in loan market, 105–06, services to, 360–61, 378n9 122nn21–23 nonresidents and monitoring efforts, 106–07 and cash flow tax, 363 436 INDEX nonresidents (continued) per capita income, 392, 393, 415n21 interest income, 205 Pérez-Castrillo, David, 123n30 tax treatment of, 235n1 personal income Nordic countries, 204–05, 225 decrease in tax on, 201 Norway, 74n13, 204, 205 including capital income in, notional principal contracts, 53 203–05 NYSE. See New York Stock Peru Exchange (NYSE) debit taxes, 313, 314–15, 323n3 OECD countries housing funds, 146, 148, borrowing incentives, 152–53 165n12 tax treatment of financial revenue productivity, 316, 317 instruments, 55 Pestieau, Pierre, 43 off-balance-sheet banking, 188n7 Phelps, Edward S., 383 offshore transactions, 12, 322, 331 Philippines O’Hara, Maureen, 337, 338 housing funds, 147, 148, 149 oil sector, 308n7, 309n20 loan-losses, 296 OLG model. See overlapping- tax treatment of pension generations (OLG) model funds, 138, 139, 140 O’Neal, Edward, 179 Pinheiro, Armando Castelar, 266n1 operational costs, 259, 262, 263, Pistaferri, Luigi, 127–68 264, 265 Poddar, Satya, 345–80 optimal consumption tax, 360 policy making, 224–26 options, 54, 221, 330, 335–37, and compensation taxes, 341nn9–11 373–75, 379n15 option system of taxation, 367–68 full taxation options, 361, outflows of funds, 67–68 362–67, 379n11, 379n13 out-sourcing, 356, 370, 375 modified exemption systems, overhead costs, 391, 415n20 361–62, 367–73 overlapping-generations (OLG) overview of options, 361–62, model, 33, 38, 73n9 378–79n10 oversight, 14 tax choices for developing ownership, 135, 164n4, 200, 338 countries, 375–77 popular capitalism, 155 Pagano, Marco, 332 Posner, Richard, 176, 180 Park, Yung Chul, 165–66n17 Poterba, James M., 129–30, 131, pay-as-you-go system, 164n5 133, 134, 152 payout costs, 13 precautionary balances, 386, 414n12 payroll tax, 374, 375 Prescott, Edward C., 414n12 peg regime, 270, 276 Prescott, Edward S., 186 Peltzman, Sam, 178 price discrimination, 101, 106 pension funds, 122n16, 128, 134, 135 price-fixing instruments, 236n10, developing countries, 138, 237n20 140, 141 price-insurance instruments, 236n10 effects of mandatory savings prices to, 135–37, 164n5 of consumer goods, 360 tax treatment of, 137–38, 139, controls in Russia, 273–74 228, 230 margin pricing, 369–70 INDEX 437 option pricing with transaction real interest rates costs, 335–37, 341nn9–11 Brazil, 246–49, 251, 266n2, sales between related parties, 267n3 357 and consumption growth, 130 stability of, 333 See also interest rates and transaction taxes, 329 realization valuation, 218, 222, 225, volatility of, 328–29, 334 236n13 prime rates, 262, 267n6 Real Plan, Brazil, 249 principals, 348 and inflation costs, 261–62 processing services, 370 passive real interest rates production efficiency theorem, 65 before, 249 productivity, 40–42, 73n11, 272 post Plan, 250–52, 258, PROER, Brazil, 250, 267n4 267n4 PROES, Brazil, 250 reserve requirements, 254, 255 profits, 104, 105, 293 and seigniorage seized, 256 ACE and, 206, 219 reforms, 2 and cash flow tax, 51, 74n22, in the 1980s, 200–02, 74n25 235–36nn4–5, 236n7, on currency, 23–24n4 236nn8–9 impact of inflation on, in the 1990s, 202–06, 391–96, 397–98, 235–36n4, 236n5, 236n7 415nn19–23 approaches to, 7–12 and liability, 171 Brazil, 246, 266n2 and loan portfolios, 177 concerns of, 198–99 and reserve requirements, cutting marginal income tax 254 rates, 201–02 and risk-taking, 108, 123n25 motives for, 201, 202–03 Sberbank, Russia, 276–77 Russia, 285–86 shifting of, 182 for stimulating equity tax on, 8–9, 24n5, 48–50, 67, financing, 211 76n40, 111, 121n10 stock exchange listings, 25n20 and VAT base, 379n15 registration duties, 5–6 progressive income taxes, 204 registration requirements, 360–61, project selection, 85, 86, 88–89, 113, 378n5, 378n9 173, 182 Reinhart, Vincent, 341n10 pure insurance, 68–69 Renaud, Bertrand, 145 pure interest rate, 59 rents, 74n21, 105, 123n25 puts, 53, 74n24 rent seeking, 180–81 Pyle, David, 178 Repullo, Rafael, 185 repurchase agreements, 315 quasi-taxes, 20, 21 reserve accounting method, and loan Quebec, Canada, 351, 373, 375 losses, 296, 297–301, 304–05, 308n16, 309nn17–20, 310n27 R+F-based tax, 49 reserve requirements, 4, 6, 180, 392, Ramsey growth model, 33, 36–38, 412, 416n35 42, 73n5, 73nn7–8 burden of high rate of, 283 rate of return, 50, 96, 122n13, 402 and deposit insurance, 182–85, R-based tax, 49 188n14 438 INDEX reserve requirements (continued) and cash flow tax, 51 effects of, 111 central bank, 386–89 and implicit taxes, 230–34, and compensatory taxes, 373 257–58 from tax on securities, 331 and income, 384 and loan-loss provisioning, 15 and inflation, 406–07, 416n34 production of in Latin non-interest bearing, 253–54, America, 316, 317 255, 267n5 relationship to financial relationship to bank spreads, services tax in EC, 355 258 Russia, 270 Russia, 278–82, 287n17 study data and evidence, and seigniorage, 252–57, 267n5 318–22, 323–24nn6–9 as a tax, 121n5, 126, 257–58 and tax rates, 385, 413n8, unremunerated, 16–17, 21, 414n9 26n24, 270 Ricardian equivalence proposition, reserves, 257 141–42, 164–65n6 excess reserves in Russia, Richardson, Matthew, 334, 335 282–83 risk, 173, 303, 310n25 general reserves, 303, 309n24 and capital adequacy, 15 hidden reserves, 303, 310n26 and capital gains tax, 236n14 and IAS, 30, 208n5, and corporate income, 109, 208nn10–11, 292–94, 307n2 123nn26–27 and inflation, 392–93, and deposit insurance, 13–14, 415nn22–23 25nn16–17, 175, 178 insurance, 306–07, 310n28 and financial innovations, loan losses, 197 53–54 prior to Real Plan in Brazil, 253 impact of on consideration and profits, 293 measurement, 366 secondary, 6 investments in risky projects, specific, 303, 310nn21–26 86, 113 restructuring of banks, Brazil, 250 and loan portfolios, 176–78 retail stores, 356 market structure and, 107, retained earnings, 217 123n24 retirement savings, 127, 128 and new financial instruments, incentives for, 129–34, 18, 25n22 164nn1–3 risky assets vs. safe assets, in industrial countries, 132 107–08, 135 life insurance as substitute for, spreads in Brazil, 263 152 under the TCA system, mandatory, 134 365–66, 379n13 tax deferred, 131 risk aversion, 43 tax incentives for, 155 risk premium, 358, 363, 366, 379n13 return on deposits, 104 risk-weighted assets, 295 return on investments, 104, 113, Ritter, Peer, 197–240 114–15 Rob, Rafael, 178 return on savings, 113, 114–15 Rochet, Jean-Charles, 43, 187n6 revenues, 13, 25n13 Roell, Alisa, 332 accounting for, 48–49 Roll, Richard, 328, 334 INDEX 439 Romania, 393, 415n24 national saving, 161–63 Romer, David, 113 overview, 127–29 Roomans, Mark, 334, 335 promotion of, 15–16 Rose, Manfred, 236n9 relationship to taxation, Ross, Stephen A., 336 129–30 rouble, 275–76 retirement saving, 127, 128, rouble Treasury bills (GKOs), 129–34, 164nn1–3 276–77, 282 Russia, 278, 284–85 Rubinstein, Mark, 336 stimulation of, 208 rural credit, 254 two-period model of, 157–58 Russia, 377 and wealth, 131, 133 accounting style, 305–06, S-based tax, 49, 51 307n1 Sberbank, Russia, 276–77, 278, 285, banking system, 271–75, 287n15 286–87n11, 286nn5–8 Schmidt-Hebbel, Klaus, 414–15n18, bank profits, 393, 415n24 416n29, 416n33 deposits in banking system, Schmukler, Sergio, 179 284–86 Scholz, John Karl, 133, 134 excess reserves, 282–83 Schumacher, Liliana, 415n22 inflation tax on currency, 278, Sealey, C., 185 283, 284 secondary markets, 89, 91 overview of 1990s financial secondary reserves, 6 history, 269–70 securities, 87, 89, 100–01, 105–06, Russian Tax Code, 297 113, 121n4 seigniorage, 278, 279, 280 and tax on loans, 114 Soviet-style accounting, transactions, 12 305–06 securities transaction taxes (STT), 335, 341n10 Saal, Matthew, 175 empirical studies of, 328–30, sales taxes, 5 341n5 Salop model, 123n30 implementation difficulties, Sandmo, Agnar, 72–73n2 327–28 Santomero, Anthony, 181 opponents of, 327 Saporta, Victoria, 328, 329 proponents of, 326–27 Saunders, Anthony, 257, 415n22 Sweden, 330–32, 341nn6–7 savings, 8, 132 Seguin, Paul J., 328 for education, health, and life seigniorage, 6, 23–24n4, 112, 265 protection, 148–52, on bank reserves, 278, 280 165nn14–15 and conventional taxes, 413n7 elasticity of, 123n32, 133 on currency, 278, 279, 384 government intervention in, factors dependent on, 241 127–28, 129 government receipt of, 387 for housing, 16, 128, 129, impact on spreads and margins 134, 144–48, 149, in Brazil, 257–64, 267n6 165nn10–12 and loans in Brazil, 255, 256 mandatory programs, 134–44, Russia, 270 158–61, 164–65nn4–9, self-supply bias, 352, 353, 369, 166n9 372–73 440 INDEX Senbet, Lemma W., 172 spread charge, 62, 63–64 separability between loans and spreads deposits, 91, 109–13, 123n30, between active and passive 123n32 interest rates, 251, 258, 259 separate transactions principle, 223 between deposit rates and Serbia, 297 bond yields, 276–77 Shaked, Israel, 185 between interest rates on loans shareholders and deposits, 258–60, 261 and corporation tax, 45–47, between loan and deposit 74nn13–16 rates, 252, 254, 258, 262, cross-border, 217 263, 283 and liquidity, 116–17 definition, 265 share options, 53, 55 impact of investments on, 254 short-term assets, 87, 90 impact of seigniorage on, short-term bridging finance, 26n25 257–64, 267n6 simplicity in tax systems, 201, 203 and non-performing loans, Singapore, 367, 372 259, 262, 263 claims cases, 266n1 risk, 263 life insurance premiums, Sri Lanka, 149 165n16 Srinivas, P. S., 338 pension fund assets, 138, 140, stamp duties, 5–6, 21, 22, 26n25, 141 231, 328 taxation of agency services, state-contingent taxation of banks, 371 185–86 VATs, 351 state enterprises, Russia, 275, 306 Sleet, Christopher, 185 Stern, Nicholas, 43 Slovenia, 293 Stiglitz, Joseph E., 40–42, 43, 108, Smith, Bruce D., 185, 400, 416n28, 170, 172, 178 416n30 stock markets, 328 social issues dealers role, 331–32, 341n8 and corrective taxes, 16 impact of inflation on, expenditures on, 272, 274 400–02, 416nn30–31 as reason for tax exemption, listings and tax treatment, 16, 349 25n20 social security system, 137 and seigniorage, 23–24n4 Sofianos, George, 333 Sweden, 331–32 software, 353 straddles, 54, 55, 222 solvency Strahan, Philip E., 181 of banks, 107–9, 123nn24–27 STT. See securities transaction taxes ratios, 122n12 (STT) social security system, 137 subsidies, 172, 274–75, 286–87n11 South Africa, pension fund assets, substitutability 138, 140, 141 in banking, 17, 18, 19, 21, 23 South Korea, mortgage loans, 154, between capital and labor, 165–66n17 112–13 Soviet State (Central) Bank, 271 of consideration, 368–69 Sparekasseernes Datacenter (SDC) v. in currency, 384 Slatteministeriet, 354, 370 elasticity of, 130–31, 133 INDEX 441 from non-interest bearing to tax-deferred accounts, 131, 133 interest-bearing assets, 385 tax evasion, 227–28 and taxation of agency tax on financial operations (IOF), services, 371 Brazil, 242, 252 Summers, Victoria, 313–24 tax incentives Sunley, Emil M., 291–311 for education, health, and life supersensitive taxes, 408–09 protection, 148–52, swaps, 25n22, 55, 74–75n27, 222 165nn14–16 Sweden for saving, 16, 25n19 dealers as traders, 331–32, tax instruments, restrictions on, 42–43 341n8 tax planning, 220 dual income tax system, 204, tax policy, financial instruments, 204, 205 56–58 marginal personal income tax tax rates, 20, 21, 26n23, 352 rate, 201 correlation of rates, 384–85, and securities transaction 413n7 taxes, 330–32, 341nn6–7 decrease of, 201 tax treatment of borrowing, elasticity of, 405, 407–08 153 and inflation, 404, 405, 406 tax treatment of pensions, marginal income tax rates, 137–38 201–02 transaction taxes and price on nominal interest income, volatility, 328, 329 404–06, 416nn33 Switzerland, capital gains tax, 218 tax shelters, 14–15, 46 tax shifting, 205 Taiwan, 329 tax wedges, 208, 210 education, health, and life TCA. See Tax Calculation Account protection tax incentives, (TCA) 150, 151 Thailand, 145, 296, 393 health insurance, 151 fiscal costs of banking system, mortgage loans, 154, 175 165–66n17 tax treatment of pension Tanzi, Vito, 383 funds, 138, 139, 140 tax-and-subsidy scheme, 172–73, Thurston, Thom, 416n29 187n5 tied-up contracts, 123n24 tax avoidance, 207 time consistency issues, 43–44, 52, 61 tax base, 357–60, 361–62 time-series studies, 130–31, 134 changes in, 201–02 timing rules, 222–23, 225, 237n20 defining, 226–28, 237n22 Tirole, Jean, 43, 88, 185, 187n6 Tax Calculation Account (TCA), tobacco taxes, 311n7 methods, 362, 364–67, 370–71, Tobin, James, 384 378–79n11, 379n15 Tobin taxes, 24n10 tax cascading, 9, 352 Townsend, Robert, 170, 173 and compensatory taxes, 373, trading, equity costs, 338–39 374 trading orders, 337–38 and input credits, 372 transactions and transaction services, and taxation of agency 89, 91, 126 services, 371 balances, 386, 414n11 442 INDEX transactions and transaction services new financial instruments, (continued) 221, 222 costs of, 156, 334–35, 335–37, reserve requirements, 233 341nn9–11 retirement savings, 132 and deposit tax, 92 risk capital, 211 and inflation, 399, 415n27 tax policy, 202, 224–25 and money market funds, 97, tax treatment of borrowing, 122n15 153, 154 pricing of, 121n5 United States, 47, 151, 152, 291, 328 stamp duties, 231 bifurcation practice, 225 taxes on, 313, 315–16 charge-off accounting, 308n15 See also financial services and competition from financial transactions institutions, 353 transaction taxes, 3, 10–12, 23, deferred tax income, 309n22 24n10 deposit insurance, 25n16 effect on price volatility, 328–29 equity ownership, 164n4 effect on trading volume, 329 marginal personal income tax impact on securities’ prices, 329 rate, 201 “noise” trading, 325, 344n2 new financial instruments, payment of, 22 221, 222–23 Sweden, 330–32, 341nn6–7 reserve accounting method, taxation model, 85, 121n3 308n16 See also securities transaction reserve requirements, 232, taxes (STT) 233–34 transparency, 15 retirement savings, 132, 134 Trehan, Brian, 413n7 savings for education, health, Trester, Jeffrey, 181 and life protection, 149, 150 Turkey, 293 switch from reserve to charge- two-period life cycle, 34–40, 72–73n2 off accounting method, 310n27 U.K. Customs and Excise tax reform measures, 202 Commissioners v. FDR Ltd., 354, tax treatment of borrowing, 370 153 Umlauf, Steven R., 328, 329, 330, 331 universal payments tax, 12 unanticipated deferral, 219, 236n14 unrealized gains or losses, 219, 222, unbundling of services, 368 226 undepreciated capital stock, 50 Uspenskii, A., 286n8 unemployment, 272, 274, 286n5 Uruguay, 138, 139, 140, 175 uniform tax, 63–64, 76n36 utility of consumption, 84, 121n2 United Kingdom, 65, 203, 328 utility functions, 60–61, 65, education, health, and life 74nn34–35, 85–86, 121n3, 359, protection tax incentives, 413n1 150 Exchange Equalization valuation, 227 Account, 414n14 value-added tax (VAT), 2, 3, 62, 105, hedging, 224 345, 360 marginal personal income tax applied to financial services, rate, 201 350–51 INDEX 443 and basic cash flow tax, 363–64 Walras’ Law, 37, 86–91, 121n4 and compensatory taxes, Walsh, Carl E., 413n7 373–75, 379n15 wealth, 131, 164n1 and depositors, 17, 25n21 as collateral, 172 as deposit tax, 114, 122n23 European Union, 220 and developing countries, and H-S concept, 199–200 375–77 redistribution of, 203 Europe, 81, 237n22, 354–55 taxation of, 205, 219–20 exemption issues, 227, and tax incentives, 131, 133 237n22, 349, 351, 367–73, wealth replacement effect, 377–78n2 135–37 and financial services, 9–10, Webb, David, 187n5 22–23, 24nn7–8, 66–69, wedge, between borrowings and 76n38, 375–76 lending rate, 257–58 on foreign processing services, welfare loss, 314 354 Whalley, J., 60, 64, 75n33, 76n36, and imported financial 360 services, 353, 378n5 Whitehouse, Edward, 137 and inflation, 396, 399 Whitesell, William, 105, 122n15 and intermediation charge, Williamson, Stephen, 185 358–59 Wise, David A., 133, 134 tax on banks’ value-added, Wiswesser, Rolf, 236n9 93–95, 121n9 withdrawals, tax on, 315 and TCA system, 366–67 withholding of taxes, 9, 22, 24n6, and value of bank payments, 5 211, 212 van der Ploeg, Frederick, 413n2 and corporation tax, 45–47, Venezuela, 145 74nn13–16 deadweight losses, 314 on gross interest receipts, debit taxes, 313, 314–15, 317 25n22 housing funds, 147, 148 imputation system, 215–16, Venti, Steven F., 133, 134 217 Verdier, Thierry, 123n30 interest income, 205, 212 Vickrey, W., 57–58 residents vs. nonresidents, Vissing-Jorgensen, Annette, 164n4 235n1, 205 Viswanathan, S., 333 Wong, Kit Pong, 257 Vives, Xavier, 101, 123n24, 185 working capital credits, 274 Vneshekonombank, Russia, 269 Vorst, Ton, 336–37 Yeltsin, Boris, 276 yield-to-maturity accounting, 224 wages heterogeneous households, zero-coupon bonds, 53, 54, 55, 40–42, 73n11 74n24 tax on, 67, 71–72 zero-rating, 351, 363, 366–67, 372, Wallison, Peter, 175, 186 375 Wall Street Journal Europe, 286n7 zero taxation, 8–9