Role of Direct and Indirect Taxes in the Federal Reserve System: A Conference Report of the NBER and the Brookings Institution 9781400875931

Table of contents :
Contents
Preface
Introduction: The Issues
Taxation, Resource Allocation, and Welfare
Comment
E. Cary Brown
William Fellner
Allocation Aspects, Domestic and International
Comment
Carl S. Shoup
Lawrence B. Krause
Reply by Musgrave and Richman
Equity, Administration and Compliance, and Intergovernmental Fiscal Aspects
Comment
Harvey E. Brazer
Ronald B. Welch
Comparison of European and United States Tax Structures and Growth Implications
Comment
Fritz Neumark
Dan Throop Smith
Summary of Conference Discussion
Conference Participants
Index

Citation preview

THE ROLE OF DIRECT AND INDIRECT TAXES IN THE FEDERAL REVENUE SYSTEM

This volume is a report of the National Bureau of Economic Research and of the Brookings Institutior

Tke Role οι Direct and Indirect Taxes in tke Federal Revenue System

A CONFERENCE REPORT OF THE NATIONAL BUREAU OF ECONOMIC RESEARCH AND THE BROOKINGS INSTITUTION

PUBLISHED BY PRINCETON UNIVERSITY PRESS, PRINCETON

1964

Copyright © 1964, National Bureau of Economic Research, Inc. All Rights Reserved L. C. CARD: 64-16726

Printed in the United States of America

NATIONAL BUREAU OF ECONOMIC RESEARCH 1964

Albert J. Hettinger, Jr., Chairman Arthur F. Burns, President Frank W. Fetter, Vice-President Donald B. Woodward, Treasurer Solomon Fabricant, New York University Geoffrey H. Moore, Associate Director of Research Hal B. Lary, Associate Director of Research William J. Carson, Executive Director DIRECTORS AT LARGE

Robert B. Anderson, New York City Wallace J. Campbell, Nationwide Insurance Erwin D. Canham, Christian Science Monitor Solomon Fabricant, Director of Research Marion B. Folsom, Eastman Kodak Company Crawford H. Greenewalt, Ε. I. du Pont de Nemours & Company Gabriel Hauge, Manufacturers Hanover Trust Company A. J. Hayes, International Association of Machinists Albert J. Hettinger, Jr., Lazard Freres and Company Nicholas Kelley, Kelley Drye Newhall Maginnes ir Warren H. W. Laidler, League for Industrial Democracy Charles G. Mortimer, General Foods Corporation George B. Roberts, Larchmont, New York Harry Scherman, Book-of-the-Month Club Boris Shishkin, American Federation of Labor and Congress of Industrial Organizations

George Soule, South Kent, Connecticut Joseph H. Willits, Langhorne, Pennsylvania Donald B. Woodward, Α. IV. Jones and Company DIRECTORS BY UNIVERSITY APPOINTMENT

V. W. Bladen, Toronto Harold M. Groves, Wisconsin Francis M. Boddy, Minnesota Gottfried Haberler, Harvard Arthur F. Burns, Columbia Maurice W. Lee, North Carolina Lester V. Chandler, Princeton Lloyd G. Reynolds, Yale Melvin G. de Chazeau, Cornell Paul A. Samuelson, Massachusetts Frank W. Fetter, Northwestern Institute of Technology R. A. Gordon, California Theodore W. Schultz, Chicago Willis J. Winn, Pennsylvania DIRECTORS BY APPOINTMENT OF OTHER ORGANIZATIONS

Percival F. Brundage, American Institute of Certified Public Accountants Nathaniel Goldfinger, American Federation of Labor and Congress of Industrial Organizations

Harold G. Halcrow, American Farm Economic Association Murray Shields, American Management Association Willard L. Thorp, American Economic Association W. Allen Wallis, American Statistical Association Harold F. Williamson, Economic History Association Theodore O. Yntema, Committee for Economic Development DIRECTORS EMERITI

Shepard Morgan, Norfolk, Connecticut Ν. I. Stone, New York City Jacob Viner, Princeton, New Jersey RESEARCH STAFF

Moses Abiamovitz Gary S. Becker William H. Brown, Jr. Gerhard Bry Arthur F. Burns Phillip Cagan Joseph W. Conard Frank G. Dickinson James S. Earley Richard A. Easterlin Solomon Fabricant Albert Fishlow Milton Friedman

Victor R. Fuchs H. G. Georgiadis Raymond W. Goldsmith Challis A. Hall, Jr. Millard Hastay Daniel M. Holland Thor HuItgren F. Thomas Juster C. Harry Kahn Irving B. Kravis Hal B. Lary Robert E. Lipsey Ruth P. Mack

Jacob Mincer Use Mintz Geoffrey H. Moore Roger F. Murray Ralph L. Nelson G. Warren Nutter Richard T. Selden Lawrence H. Seltzer Robert P. Shay George J. Stigler Norman B. Ture Herbert B. Woolley Victor Zarnowitz

RELATION OF NATIONAL BUREAU DIRECTORS TO PUBLICATIONS REPORTING CONFERENCE PROCEEDINGS

Since the present volume is a record of conference proceedings, it has been exempted from the rules governing submission of manuscripts to, and critical review by, the Board of Directors of the National Bureau. It has, however, been reviewed and accepted for publication by the Director of Research. (.Resolution adopted July β, 19^8, as revised November 21, 1949)

THE BROOKINGS INSTITUTION In publishing a study, the Institution presents it as a competent treatment of a subject worthy of public consideration. The interpretations and conclusions in such publications are those of the author or authors and do not necessarily reflect the views of other members of the Brookings staff or of the administra­ tive officers of the Institution.

THE BROOKINGS INSTITUTION BOARD OF TRUSTEES Eugene R. Black, Chairman Robert Brookings Smith, Vice Chairman William R. Biggs, Chairman, Executive Committee Arthur Stanton Adams Dillon Anderson Elliott V. Bell Louis W. Cabot Robert D. Calkins Leonard Carmichael Thomas H. Carroll Edward W. Carter Colgate W. Darden, Jr. Marion B. Folsom Gordon Gray

Huntington Harris David M. Kennedy John E. Lockwood H. Chapman Rose Sydney Stein, Jr. Donald B. Woodward Honorary Trustees Daniel W. Bell Mrs. Robert S. Brookings Huntington Gilchrist John Lee Pratt

STUDIES OF GOVERNMENT FINANCE Studies of Government Finance is a special program of research and education in taxation and government expenditures at the federal, state, and local levels. These studies are under the supervision of the National Committee on Govern­ ment Finance appointed by the Trustees of the Brookings Institution, and are supported by a special grant from the Ford Foundation. MEMBERS OF THE NATIONAL COMMITTEE ON GOVERNMENT FINANCE Elliott V. Bell Howard R. Bowen Robert D. Calkins (Chairman) Marion B. Folsom Erwin Nathaniel Griswold

Alvin H. Hansen George W. Mitchell Don K. Price Jacob Viner William C. Warren Herman B Wells

Joseph A. Pechman, Executive Director

Contents PREFACE

Xl

3

INTRODUCTION: THE ISSUES

John F. Due TAXATION, RESOURCE ALLOCATION, AND WELFARE

25

Arnold C. Harberger COMMENT

E. Cary Brown William Fellner ALLOCATION ASPECTS, DOMESTIC AND INTERNATIONAL

81

Richard A. Musgrave and Peggy Brewer Richman COMMENT

Carl S. Shoup Lawrence B. Krause Reply by Musgrave and Richman EQUITY, ADMINISTRATION AND COMPLIANCE, AND INTERGOVERNMENTAL FISCAL ASPECTS

141

Douglas H. Eldridge COMMENT

Harvey E. Brazer Ronald B. Welch COMPARISON OF EUROPEAN AND UNITED STATES TAX STRUCTURES AND GROWTH IMPLICATIONS

217

Otto Eckstein assisted by Vito Tanzi COMMENT

Fritz Neumark Dan Throop Smith SUMMARY OF CONFERENCE DISCUSSION

295

Samuel B. Chase, Jr.

315 317

CONFERENCE PARTICIPANTS INDEX

ix

Preface ONE of the oldest issues in public finance concerns the relative ad­ vantages and disadvantages of sales, excise, and related taxes, often referred to as "indirect" taxes, on the one hand, and income and other levies, often called "direct" taxes, on the other. Current discussions have centered on the implications for economic growth and for the U.S. bal­ ance of payments of varying the relative emphasis on indirect and direct taxation. The French use of the value-added tax and the tentative acceptance of this form of sales tax by the Common Market countries as part of their tax harmonization efforts have led to emphasis on this form of tax in the suggestions for reformulation of U.S. tax policy. The arguments on the issue are for the most part old and familiar ones, repeated time and again, frequently without careful analysis or consideration of the arguments on the opposite side. For example, it is commonly argued that sales and excise taxes have less deterring effect on investment and the rate of growth than income taxes, but the validity of this point has not been carefully examined. In light of the renewed interest in the question, the slow progress in the analysis of the issues, and the pertinence of these issues to their studies and research in the field of taxation, the National Bureau of Economic Research and the Brookings Institution concluded that a conference of experts might help to clarify the issues, to ascertain the major points of agreement and disagreement, and to disclose the major aspects on which further analysis and empirical work are needed. The two sponsoring organizations established, early in 1963, a plan­ ning committee for such a conference, consisting of Herbert Stein, Di­ rector of Research for the Committee for Economic Development, Richard A. Musgrave of Princeton University, Joseph A. Pechman of the Brookings Institution, Norman B. Ture of the National Bureau of Eco­ nomic Research, and myself as chairman. The committee requested four persons to prepare papers on four main aspects of the question, and two persons to serve as formal discussants for each of these papers. A group of experts in the field were invited to participate in the conference. A total of forty-three persons attended the conference, which was held at the Brookings Institution in Washington on October 17 and 18, 1963. The participants are listed on p. 315. xi

PREFACE

This volume contains a statement of the issues prepared by the chair­ man of the committee, the four major papers, the comments of the formal discussants, and a summary of the discussion at the conference. The research conference was part of the National Bureau's program of research on Tax Policies for Economic Growth and of the Brookings Institution's Studies of Government Finance. These programs are being financed by funds provided to the National Bureau by the Rockefeller Brothers Fund and the Life Insurance Association of America and to the Brookings Institution by the Ford Foundation. JOHN F. DUE

THE ROLE OF DIRECT AND INDIRECT TAXES IN THE FEDERAL REVENUE SYSTEM

Introduction: The Issues JOHN F. DUE UNIVEKSITY OF ILLINOIS THE question of the optimum relative role of direct and indirect taxes in the federal tax structure has been a perennial one for forty years. The last year has seen revival of interest in the question, prompted in large measure by the emphasis given by the European Common Market countries to harmonization of their tax structures, by concern over United States exports, and by a continuing lag in investment spending in the country, with consequent failure of economic growth to attain the rate regarded as desirable. It is the purpose of this introductory paper to outline briefly the history of the controversy, and to state the issues as they stand at the present time.

Some Concepts To lessen confusion and the need for detailed discussion of concepts in the papers to follow, and to lessen analysis of questions which are incidental to the main issues of the conference, a statement of the mean­ ing to be given to certain terms and of assumptions on several key issues for purposes of the conference will be given in this section. Little is to be gained by seeking to establish definitions of direct and indirect taxes. Instead, we shall simply interpret the term "direct tax" as referring to personal and corporate income taxes and death taxes, and "indirect tax" as referring to levies upon the production or sale of commodities. The question of the classification of payroll taxes is not of major im­ portance for the conference as a whole, and will be raised only in the Eckstein paper. The same is true of property taxation. There are two principal categories of indirect taxes: excises, imposed upon the production or sale of particular commodities or related groups of commodities; and sales taxes, imposed upon the sale of all commodities except those specifically exempted. Since broadening of the base of the federal excise tax structure is not under serious consideration, the prime issue of concern is whether the federal government should introduce a sales tax to replace a portion of the revenue now gained from personal and corporate income taxes. There are four major possible forms of sales tax worthy of serious consideration: those imposed, respectively,

3

INTRODUCTION

upon the manufacturing level, upon the last wholesale transaction through which a commodity passes, upon the retail sale, and upon value added, either at the manufacturing level or at all stages in production and distribution. In turn, the value-added tax may take the income or consumption form, as discussed in the Eldridge paper. It will be assumed that excises and sales taxes are borne primarily in relation to consumption expenditures through increases in the prices of consumption goods relative to factor incomes. The validity of this assumption has been subject to question over recent years,1 but in the context of the present issue, which relates to the alternative choice of income and commodity taxes to finance a given level of government activities, the assumption would appear to be the most suitable one. It is recognized, of course, that there are exceptions to the rule of shift­ ing, a portion of the tax undoubtedly resting on the owners of specialized resources, and upon the recipients of excess profits. The issue of shifting is considered to some extent in the Musgrave-Richman and Eldridge papers. The Historical Development of the Sales Tax Issue

The United States federal government is one of the few major national governments that have never imposed a general sales tax.2 Possible use of such a levy became a major issue in four periods: the Civil War, the post-World War I period, the depression years of the thirties, and World War II. In addition, there has been sporadic interest on several 1 The criticism has taken two major forms. Earl Rolph argued that the analysis of incidence of a tax should not consider the use of the revenues, and that thus the tax brings about a decline in factor prices and is borne in the same fashion as a pro­ portional income tax. (See "A Proposed Revision of Excise Theory," Journal of Political Economy, April 1952, pp. 102-117.) J. M. Buchanan, and to some extent Rolph and George Break in their more recent work, have argued that an increase in the general price level cannot be attributed to a sales tax, since such increases can result only from monetary changes, and thus a sales tax cannot be "borne by consumers." (See J. M. Buchanan, Fiscal Theory and Political Economy, Chapel Hill, 1960; and E. R. Rolph and George Break, Public Finance, New York, 1961, p. 294.) Musgrave has demonstrated that a sales tax on consumption goods is borne in relation to consumption expenditures, regardless of whether the general price level increases or factor prices fall. (See The Theory of Public Finance, New York, 1959, Chaps. 10, 11, 16.) A summary of the conflicting points of view is to be found in Due, "Sales Taxation and the Consumer," American Economic Review, December 1963, pp. 1078-1084. 2 With the imposition of sales taxes in recent years in Sweden, Denmark, and Eire, the only countries in western Europe not imposing sales taxes are Spain and Portugal, although the British purchase tax is of restricted scope. In the Western Hemisphere, Canada, Mexico, Brazil, Chile, the Argentine, Uruguay, and some of the smaller countries use sales taxes. They are found also in Australia and New Zealand, and extensive use is made of them in India.

INTRODUCTION

occasions in the last fifteen years. Currently, while the proposal is not a major political issue, it has received renewed attention in various studies of tax revision. The earlier interest in the tax warrants brief review. CIVIL WAR

Early in the war, Congress imposed an income tax and a comprehensive system of excises, the first important internal revenues ever employed by the federal government. These were highly unpopular and not well administered, and proposals for a broad-based sales tax were made, even though sales taxation was not currently used in any country, out­ side of vestiges of the alcavala in Spain and in Latin America. In 1862 the New York Chamber of Commerce and other business groups pro­ posed to Congress the introduction of a sales tax, and in 1864 the proposal was endorsed by the Commissioner of Internal Revenue. But the Ways and Means Committee rejected the proposals, and the House did likewise when amendments for a sales tax were offered. THE POST-WORLD WAR I PERIOD

Not until 1920 did interest in a federal sales tax revive. During World War I, sharp increases were made in income taxes, and an excess profits tax was levied, as well as a number of excises. After the war, strong opposition to these levies developed, primarily among business groups, and a sales tax was widely proposed to allow reduction in other taxes. The movement was strengthened by the enactment in this period of sales taxes in France, Germany, Canada, and other countries. The pro­ posals were considered in the hearings of the Ways and Means Com­ mittee and the Senate Finance Committee in 1920 and 1921. In 1921 several bills and amendments, two calling for a turnover or multiplestage tax and one for a manufacturers' sales tax, were introduced by Senator Smoot, but all were defeated.3 THE DEPRESSION

A decade later, interest revived in a sales tax as a means of eliminating depression-induced federal deficits. The Hearst newspapers led the cam­ paign for the tax, with the Canadian sales tax as the model, and the Ways and Means Committee gave serious attention to the proposal, despite initial Treasury opposition (largely based on administrative 3 A detailed account of the 1920-22 sales tax movement is to be found in. the article by Κ. M. Williamson, "Literature on the Sales Tax," Quarterly Journal of Economics, August 1921, pp. 618-633.

INTRODUCTION

considerations). The committee included in the Revenue Bill a provision for a 2| per cent manufacturers' sales tax similar to the Canadian tax, and Secretary of the Treasury Ogden Mills (who in the early twenties had proposed a spendings tax) withdrew his opposition, but argued that the tax should be a temporary emergency measure only. However, the bill was amended on the floor of the House to strike out the sales tax provision. The Ways and Means Committee endorsement was the closest Congress has ever come to enactment of a sales tax. Except for the Townsend old-age-pension proposals, which were based upon a turn­ over tax, little was heard of a federal sales tax for another decade. Meanwhile, between 1932 and 1937, retail sales taxes had become im­ portant sources of state revenue. THE WORLD WAB II PROPOSALS

With the outbreak of World War II, interest in such a tax was imme­ diately renewed, primary support coming from business groups, and opposition from the Treasury, labor, and others. While most of the proposals favored a manufacturers' sales tax, the Treasury, after an extensive study, recommended the use of the retail form if the tax should be used.4 Despite considerable support in Congress, the proposals did not obtain approval of the Ways and Means Committee, and a supplement to the income tax called the Victory Tax was imposed instead. For the remainder of the war, little was heard of the sales tax issue. THE POST-WORLD WAR II PERIOD

The sales tax has received sporadic interest since 1945. While the issue has not been a major one, on the other hand it has never entirely died out. However, the emphasis has shifted over the period, both in terms of sponsorship and type of sales tax proposed, and in recent years the proposals have been influenced by European experience and a similar issue in Great Britain. In the early fifties, spurred in part by the realization that high defense spending was not going to permit over-all tax reduction, the National Association of Manufacturers took the lead in urging a federal sales tax at the manufacturing level. The primary argument used by the NAM was the need for elimination of the discriminatory system of excises, but stress was also placed upon the need for reduction of income taxes 4 The study was published under the title Considerations Respecting a Federal Retail Sales Tax, Hearings (on Revenue Revision of 1943) of Ways and Means Committee, 78th Congress, 1st Session, 1943, pp. 1097-1272.

6

INTRODUCTION

because of their adverse incentive effects.6 The greater stability of rev­ enue (an argument which is unacceptable in terms of modern fiscal policy) and the greater ease of administration and compliance were also advanced as arguments for the tax. Bills embodying the NAM proposals were repeatedly introduced into Congress by Representative Mason (Republican, Illinois), but were pigeonholed in the Ways and Means Committee. In the late fifties and particularly in the last three years, the source of support for increased relative reliance on indirect taxation, the nature of the arguments, and the specific types of proposals have changed in considerable measure. The Committee for Economic Development grad­ ually came to support a federal sales tax. In the mid-fifties the CED had repeatedly expressed support for a reduction in excises, with the state­ ment that if expenditures remained at high levels the excises should ultimately be replaced by a retail sales tax. After 1960, the CED shifted toward positive support of the introduction of a retail sales tax; note, for example, the statement in the December 1962 publication, Reducing Tax Rates for Production and Growth (p. 38): "The Federal government needs to raise more of its revenue from taxes on the use of income for consumption and less from taxes on the earning of income by work and investment. . . . Therefore the Federal government should have in its revenue system a general excise tax on a very broad base at a moderate rate." Sympathy for a federal sales tax in one form or another has come from such economists as William Fellner,6 Arnold Harberger,7 W. J. Baumol,8 H. M. Somers,9 D. T. Smith,10 and others. Continuing support has been expressed in the First National City Bank Monthly Economic Letter, for example, and in England by the Economist. Most of this recent sup5 See, for example, NAM News, Special Report, A Manufacturers Uniform Excise v. a Retail Sales Tax, May 5, 1951; A Tax Program for Economic Growth, New York, NAM, 1955; and the papers by H. L. Lutz, "Place and Role of Consumption Taxes in the Federal Tax Structure," Joint Committee on the Economic Report, Federal Tax Policy for Economic Growth and Stability, Washington, 1955, and R. Robey, "The Relative Role of Federal Taxes," in Tax Revision Compendium, U.S. Congress, House of Representatives, Committee on Ways and Means, 1959, pp. 215-219. 6 "Possibilities of Broadening the Tax Base, Reducing Tax Rates and Promoting Economic Growth," Tax Revision Compendium, pp. 193-200. 1 "The Corporation Income Tax; An Empirical Appraisal," Tax Revision Com­ pendium, pp. 231-248. 8 K. Knorr and W. J. Baumol, What Price Economic Growthf, Englewood Cliffs, N.J., 1961. 9 "Theoretical Framework of Sales and Use Taxation," Proceedings of the National Tax Assoiation for 1961, pp. 607-618. 10 Federal Tax Reform, New York, 1961.

INTRODUCTION

port is based upon much more careful argument than in the past and thus is much less doctrinaire. A retail sales tax or a value-added tax is favored, rather than the manufacturers' sales tax. The arguments for the proposals have been made primarily in terms of economic growth, and the support from economists of high reputation has without question been due in part to increasing emphasis on the lag in investment as the source of continuing unemployment and a slow rate of economic growth. In the last two years, concern with the United States balance of pay­ ments and its export position, together with the development of the Common Market in Europe and the stress given in Common Market countries to the significance of taxes for export, have increased support for a specific plan substantially different from the sales tax proposals of earlier years, namely, the replacement of the corporate income tax by a value-added tax. This point of view appears in the writings of Harberger and Smith, for example, and in the Economist proposals. OTHER DEVELOPMENTS

Since the late twenties, the relative reliance of the federal government on indirect taxes, namely, excises, has changed very little, ranging between 12 and 18 per cent of the total tax revenue, except for the depression years, when sharp declines in income tax yields and some increases in excises raised the figure to as high as 45 per cent in 1933 (Table 1). In the decade since 1953, the percentage yield has varied only in the narrow range between 12.6 and 13.8 per cent. Except for increases made in 1950 in liquor and tobacco and motor vehicle excises for a temporary period but continued down to the present year, and later increases in the motor fuel taxes, the trend has been to reduce and eliminate excises, as a product not of any deliberate policy but of the pressures of the various industry groups on Congress. The sharpest re­ ductions were made in 1954, when most of the excises outside of the liquor, tobacco, and motor vehicle fields were reduced in half. The excise tax on transportation of freight was eliminated in 1958 and that on passenger transport in 1962 (except on air transport). In each session of Congress there is strong pressure from various industry groups to make further changes. Further disintegration of the excise structure is almost inevitable, partly because of the lack of any logic underlying the selection of items for tax. Table 2 shows the major excises in 1963 and their revenue yield. At the same time, state sales taxation has grown steadily; the number using the tax as of January 1, 1964, was 37, and more are certain to be

8

INTRODUCTION TABLE 1

FEDERAL EXCISE TAX COLLECTIONS, SELECTED YEARS, 1902-63 (million dollars)

Excise Tax Collections as Percentage of Total Tax Collections

Year

Total Tax Collections

1902 1913 1922 1929 1933

513 662 3,371 3,541 1,871

244 302 834 540 839

47.6 45.6 24.7 15.1 44.8

1939 1949 1954 1955 1956

5,500 40,857 70,525 66,895 75,814

1,768 7,579 9,532 9,211 10,004

32.1 18.6 13.5 13.8 13.2

1957 1958 1959 1960 1961

80,926 80,778 80,746 92,898 95,409

10,634 10,814 10,760 11,865 12,064

13.1 13.3 13.3 12.8 12.6

1962 1963

100,549 105,917

12,749 13,410

12.7 12.7

Total Excise Tax Collections

Source: Treasury Bulletin; U. S. Bureau of the Census. Governmental Finances in the United States. 1902-57. aIncluding

employment taxes.

preliminary.

added to the list. In addition, extensive use is made of the tax by the municipalities in New York State, which itself does not use the tax, and in several other states. In the 1962-63 fiscal year the tax yielded 23 per cent of the total tax revenue of the states and 31 per cent in those states using the tax in that year, with a range from 48 per cent in Illinois to 19 per cent in Louisiana. Table 3 indicates the yield of the tax by state in the 1962-63 fiscal year, the tax rates as of January 1, 1964, and the year of introduction. The Issues With this brief historical sketch completed, attention will be given to the major issues involved in the question of relative reliance on income

9

INTRODUCTION TABLE 2

MAJOR UNITED STATES FEDERAL EXCISES

Type of Excise

Sumptuary Distilled spirits Wine Beer Cigarettes Cigars Other tobacco products Slot machines Wagers Total, sumptuary

Rate, January If 1964

$10,50 per gallon (proof) $ 0.17 to $2.25 per gallon, according to alcoholic content $ 9.00 per barrel (31 gallons) $ 0.08 per package of 20 $ 2.50 to $20.00 per 1,000 according to selling price $ 0.10 per pound $250 per year

10%

Highway Allocated to Highway Trust Fund Trucks and buses 10% Truck use tax $3.00 per 1,000 pounds Gasoline and diesel fuel $0.04 per gallon Tires, tubes, etc. $0.05 per pound, and others Total

Miscellaneous Business machines Stamp taxes on transfer of securities, etc. Total, miscellaneous Other Total

2,507 104 831

2,011 50

18 26 5,547

Luxury ^ Manufacturing level Radio, TV, etc. 10% Electrical appliances 5% (air conditioners 10%) Musical instruments, etc. 10% Sporting goods, etc. 10% (firearms 11%) Cameras, film, etc. 10% (projectors 5%) Light bulbs 10% Pens, etc. 10% Matches $0.02 per 1,000 Flaying cards (stamp) $0,13 per pack Retail level Luggage 10% Jewelry 10% Furs 10% Toilet preparations 10% Services, etc. 10% Admission and cabarets 20% Club dues 10% Telephone and telegraph Passenger transport (air) 5% 10% Safe deposit boxes Total, luxury

To general revenue Automobiles and parts Lubricating oil Total

Revenue Yields, 1962-63 Fiscal Year (million dollars)

10% (parts 8%) $0.06 per gallon

10% Various

212 130 20

42 25 36 9 4 9

74 182 29 158 83 71

881 234 7 2,206

303 99

2,610 399 3,411

1,784 74 1,858

75 140 215 173 13,410"

INTRODUCTION NOTES TO TABLE 2

Source of revenue data:

Treasury Bulletin,

a Percentage

wise noted·

of selling price of the firms subject to tax, except where other­ There are exceptions to some of the rates listed.

^The highway levies (except the truck use tax), the business machines tax, and the liquor and tobacco taxes are also imposed upon the manufacturer. c After

adjustments for depositary receipts.

and sales taxation at the present time. There is no attempt in this paper to analyze the issues or evaluate the arguments; this task is the function of the papers which follow. The issues can be classified into several major groups, relating, respectively, to economic growth; to resource alloca­ tion, including questions of international trade; to equity considerations; to administration and compliance; and to intergovernmental fiscal rela­ tions. The significance for United States policy of the experience in Europe and elsewhere with the two forms of taxation is also an issue. POTENTIAL ECONOMIC GROWTH

It is important to distinguish clearly between the potential and the actual paths of economic growth. The former describes the maximum possible output over time, given the changes occurring in technology; the changes in the supply and quality of resources, including labor; and the S/Y ratio, which determines the rate of capital formation possi­ ble at full employment. The actual path of growth cannot rise above the potential growth path, but will lie below it if full employment is not attained. The first major issue is the relative effect of income and sales taxation on the growth of potential output. There are several types of effects which must be considered. 1. Does the sales tax increase the percentage of national income saved at full employment by shifting a greater portion of the burden from those families which save high percentages of their incomes to those which save lower percentages and thus must absorb a larger amount of the tax from the portion of income which would otherwise be spent on consumption? This result will be attained if the use of a sales tax shifts a greater portion of the total tax burden to the lower income levels, provided that the over-all marginal propensity to save is lower in the lower income groups than in the higher ones. It is obvious that the average propensity to save is lower in the lower income groups, but it is by no means so obvious that the marginal propensity is lower. This is 11

TABLE 3

STATE RETAIL SALES TAXES, 1963

Yield, 1962-63 Fiscal Year3 Amount (thousand . dollars)

State

Illinois Michigan Hawaii Washington Georgia Arizona Mississippi Kansas Arkansas Tennessee Utah Kentucky Missouri Florida South Dakota California Alabama Pennsylvania Maine South Carolina Iowa Connecticut Ohio Nevada New Mexico Wyoming Rhode Island North Dakota Maryland Colorado North Carolina West Virginia Oklahoma Louisiana Texas Wisconsin Indiana Total

548,363 499,884 56,900 231,778 171,965 78,291 80,391 84,333 . 66,722 121,266 41,175 112,073 135,369 191,339 20,688 813,310 100,338 397,770 30,137 80,497 88,133 101,861 277,459 19,490 42,847 12,400 28,930 17,607 110,664 57,926 145,942 51,231 72,379 96,900 180,489 55,440 0 5,222,287

Percentage of Total Tax Revenue

50.8 43.7 42.7 42.2 38.9 37.6 36.3 35.3 35.2 34.4 33.7 33.3 32.7 32.3 31.9 31.8 31.4 31.4 30.8 30.5 30.4 30.3 29.9 29.7 28.8 27.5 27.4 25.7 25.4 25.0 24.8 22.7 22.5 19.0 17.3 9.2 0 31.6®

Year of Introduction

1933 1933 1935 1934 1951 1933 1932 1937 1934 1947 1933 1960 1934 1949 1933 1933 1937 1956 1951 1951 1934 1947 1935 1955 1934 1934 1947 1934 1947 1934 1933 1933 1933 1936 1961 1962 1963

Tax Rate Jan. If 1964 (per cent)

3 1/2 4 3 1/2' 4 3 3K 3 2 1/2 3 3b 3 3 3 3 2b 3b 4 5 4 3 2 3 1/2 3 2 3° 2 3 2 1/4 3 2 3 3 2 2 2 3 2

Source: U. S. Bureau of the Census, State Tax Collections in 1962 and in 1963. and Detail of State Tax Collections in 1962 and in 1963. aExcludes yield from nonretail portions of the taxes; includes yield from separate levies on automobile sales and transient accommodations.

kpius municipal taxes in many areas. cRate

on retail sales; other rates are on certain nonretail transactions.

^Not in operation during 1962-63 fiscal year. eOf

those states using the sales tax.

INTRODUCTION

an empirical question upon which studies of recent years have thrown additional light.11 Even if there is no significant difference in marginal propensities to consume at various income levels, however, the relative burden on fami­ lies within particular income groups will vary according to the percentage of income spent on taxable items; increased use of the sales tax would presumably, from this source alone, have some net influence on the over-all propensity to save. 2. Will use of sales taxation provide greater incentive to save? The sales tax can be avoided, at least currently, if income is saved; with the income tax, no such escape is possible. The significance of this influence depends, of course, upon the relative importance of various motives for saving; for example, if saving is made for saving's sake, the tax is effectively avoided; if funds are saved for the purchase of taxable goods in the ensuing period, when the tax will still be in operation, there is merely a delay in payment of tax.12 The basic issue is, therefore, What is the relative importance of various motives for saving, and what is the relative responsiveness of savings made for various purposes to the two forms of tax? 3. If a higher rate of saving is regarded as desirable, can this be obtained more easily by changes in the tax structure or, as suggested by Musgrave,13 by a higher over-all level of taxes and a budget surplus? 4. What are the relative effects of the two forms of tax upon the incentives to work, to gain greater education and skill, to take more responsible positions, and to undertake the development of business enterprise? The effects of the income tax on work incentives has been analyzed at some length in recent years, and is currently the object of a study by Daniel M. Holland for the National Bureau of Economic Research. Recognition of both income and substitution effects makes it clear that the answer cannot be attained by deductive reasoning, and that further empirical evidence is needed. Much less attention has been given to the effects of sales taxes on work incentives; the usual argument is that their nonprogressive nature and the ability to escape them by saving additional income received result in less effect than income taxa11 For a summary, see T. R. Beard, "Progressive Income Taxation, Income Re­ distribution, and the Consumption Function," National Tax Journal, June 1960, pp. 168-178. 12 Note the discussion of this question in the article by M. A. Willemsen, "The Effect upon the Rate of Private Savings of a Change from a Personal Income Tax to a Personal Expenditure Tax," National Tax Journal, March 1961, pp. 98-103. 13 See R. A. Musgrave, "Growth with Equity," Proceedings of the American Eco­ nomic Association for 196$, pp. 323-333.

INTRODUCTION

tion. But the answer to the question depends upon the relative reactions of persons to the two forms of taxation. It is by no means impossible that, for many taxpayers, increased cost of goods resulting from the tax may have the same effect as a tax on the income itself. This is more likely to be true in underdeveloped economies than in the United States, but the relative importance of this phenomenon in the U.S. must be explored as well. 5. What are the relative effects of the two forms of tax upon relative efficiency in production? The corporation income tax, it is argued, penal­ izes efficient producers since it is paid only by profitable enterprises. As a result, the tax shifts claims on resources from efficient to inefficient corporations.14 Moreover, it exerts a bias in favor of the use of labor inputs and against the use of capital; hence, it impedes optimum input combinations. A value-added tax, on the other hand, would shift more of any given business tax burden to inefficient firms. In addition, it is neutral with respect to the choice of factor combinations in production. The significance of this argument depends, in part, upon the nature of the "marginal" firms and the source of the higher rate of profit of the "supramarginal" firms. To what extent are low profits the result of other considerations than inefficiencies, such as recent establishment of the business? To what extent are high profits the result of efficiency or of monopolistic restrictions? Whether substitution of value-added taxation for profits taxation would result in better combinations of factors in production depends on the elasticity of substitution of capital for labor. Moreover, sales taxes themselves are capable of producing changes in methods of doing business which may have serious effects on efficiency; application of tax to goods used in certain techniques of production and not to those used in others, as is typical, may alter production processes. The turnover form of sales tax is capable of pro­ ducing serious distortions in production organization and methods. EFFECTS UPON LEVEL OF EMPLOYMENT AND ACTUAL PATH OF ECONOMIC GROWTH

As noted, the economy will fail to attain the potential growth path if full employment is not realized. Failure to attain full employment, in turn, is primarily the result of inadequate total spending in the economy, 14 Committee for Economic Development, Tax Reduction and Tax Reform,—When and How, New York, 1957; Dan T. Smith, "Capital Formation and the Use of Capital," Proceedings of the American Economic Association for 1962, pp. 314-322; Martin Norr, "The Value Added Tax in France," Report of the 1962 Conference, Canadian Tax Foundation, pp. 243-253.

INTRODUCTION

and thus is likely to be influenced more by the over-all level of govern­ ment spending and revenues and the relation between these two magni­ tudes (and thus the budget surplus or deficit) than it is by the types of tax employed. But the issue, so far as the question under discussion is concerned, and one of the major ones in the entire direct-indirect tax controversy, is that of the relative effects which the two forms of taxa­ tion have upon total spending and thus upon the extent to which the potential rate of economic growth is actually attained in the economy. This general issue can, in turn, be broken down into several elements: 1. What are the relative influences on the volume of investment—the element in total spending which has lagged in the last five years—of a tax on the earnings from investment, on the one hand, and inadequate consumption, on the other? Or, in other words, to what extent has investment been deterred by existing income taxes on individuals and corporations? To what extent would the collection of the same amount of revenue from commodity taxes reduce investment by curtailing con­ sumption? There are a number of elements which affect the answers to these questions: the extent to which the corporate income tax is shifted, the nature and location of investment decision making in the corporation, the portion of total investment which is sufficiently margi­ nal to be affected by the income tax, and the propensity to consume at various income levels. 2. To what extent will a shift from income to commodity taxation reduce total consumer spending? In part, of course, the answer depends on the nature of the reduction in income taxation; if this takes the form of a change in the corporate income tax, the net effect will be different from what it will be if the personal income tax is reduced. But, as noted earlier, the effect will also depend upon the marginal propensi­ ties to consume at various income levels and among families within particular income levels. 3. To what extent will a shift from income to commodity taxation increase United States exports relative to imports? This question is discussed below. 4. To what extent do adverse investment incentive effects of the income tax result from particular features of the structure of the income tax rather than the use of the income basis of taxation as such? May it not be possible to reduce the adverse effects by changes in the income tax more easily than by introduction of a sales tax? The interrelation of the relative effects of the two forms of tax upon the potential and actual growth paths must of course be noted. To the 15

INTRODUCTION

extent to which the shift to indirect taxation increases the potential rate of growth, it may aggravate the task of attaining the potential path. An increase in the propensity to save, while increasing the rate of capital formation possible at full employment, may bring about a reduction in total spending and thus an increased gap between the actual and poten­ tial growth paths. Whether or not this occurs will depend in large measure upon the extent to which the shift in tax, by reducing the direct burden on the gains from investment, will increase the volume of invest­ ment. RESOURCE ALLOCATION AND INTERNATIONAL TRADE

Recently attention has been given to an issue almost ignored in earlier years: the relative effects of the two types of tax in producing undesired distortions in resource allocation, with particular emphasis on the effect of the income taxes in adversely affecting United States exports. So far as domestic production is concerned, the issue has been raised most emphatically by Harberger,16 who charges that income taxation, particularly the corporate tax, produces a distortion of resource alloca­ tion in favor of those lines of production, such as housing and agriculture, not significantly subject to the corporate tax. Replacement of this levy by a uniform-rate value-added tax applied to all types of business would eliminate the distortion. There are several issues involved, however. What is the quantitative significance of this distortion? Harberger at­ tempts to provide an answer, but obvious questions can be raised about the reliability of his estimates. To what extent can the pattern of re­ source allocation attained in the absence of the corporate tax be regarded as the optimum? To what extent can a sales tax be devised which will in fact apply to all lines of production in a uniform fashion? No sales tax yet devised, value added or otherwise, has ever accomplished this goal, because of the difficulties, administrative and political, of reaching certain activities, and because of nonuniformity in shifting. The foreign trade aspect is the one which has received greatest em­ phasis, in part because of the attention given to tax aspects of the European Common Market and the significance of these aspects for American exports in a period in which the United States is suffering from payments deficits. The French value-added tax has been accepted tentatively as the standard form of sales tax for the Common Market countries; under this form of tax, exports are exempted and the entire 15 "The Corporation Income Tax: An Empirical Appraisal," Tax Revision Com­ pendium, pp. 231-248.

INTRODUCTION

amount of sales tax which has accumulated at various stages in the process of production and sale of a product is refunded at the time of export. Imports are subject to tax. Thus, it is argued, exporters sell free of any tax element. By contrast, American exporters receive no equivalent refund of tax which they pay as a result of their conduct of economic activity. This argument is presented simply and clearly in the CED's monograph, Reducing Tax Ratesfor Production and Growth (pp. 39-40): A major advantage of a general excise tax is that it would tend to im­ prove the ability of the United States to compete with others in world markets. France, Germany and several other countries already have such a tax and it is likely to be adopted by still others under the policy of har­ monizing taxation within the Common Market. The tax is not collected, or is rebated, on exports, so that countries having the tax are able to sell abroad at prices below the prices charged their domestic consumers. At the same time a compensating tax is levied on imports. Thus a country like the U.S. which does not have such a tax receives imports from a country that does at prices below the domestic price level of the exporting country. At the same time the prices at which we must try to sell abroad are increased by the compensating tax on exports. It will become increas­ ingly important for the United States to equalize this situation as tariffs between us and the Common Market are reduced. This argument would appear to be valid only to the extent to which the corporation income tax is now reflected in the prices of exported goods, or to the extent to which an indirect tax would be shifted back­ ward. If the corporate tax is not shifted forward, the imposition of a tax which would be reflected in prices and the subsequent rebate of this tax on exports would not benefit the American export industry. If the in­ direct tax were not shifted forward, the price level exclusive of tax would fall, and exports would benefit in the same fashion as from devaluation of the dollar. But given institutional wage and price rigidities, backward shifting would not appear to be significant. Thus the basic issue is the old and familiar one: to what extent is the corporation income tax shifted forward? There are also other ramifications relating to exchange rates and terms of trade which will be considered in the Musgrave paper. EQUITY CONSIDERATIONS

Traditionally, the primary argument against commodity taxation has been based on equity grounds. There are two major complaints: against regressivity in the distribution of tax burden relative to income, and against distribution within particular income levels which is not in

17

INTRODUCTION

conformity with usual standards of equity, such as the relatively heavy burden on large families compared to smaller families at a given income level.16 There are several issues involved in the question: To what extent is the actual burden distribution of a sales tax regressive relative to income? Empirical studies over the last decade suggest that a sales tax with food taxable is regressive, while one with food exempt is more or less proportional in most income ranges.17 Serious question has been raised about the appropriate basis for com­ parison; it has been argued that the permanent component of income, a concept developed by Milton Friedman, is a more suitable basis for comparison than actual income received during the period. If Friedman's conclusion that the percentage of permanent income saved is constant at all income levels is accepted, the sales tax will not be regressive rela­ tive to permanent income, and in fact is likely to be progressive.18 Thus the question resolves itself into two issues: 1. The validity of the Friedman hypothesis of the constancy of the ratio of consumption to permanent income. This has been questioned in various empirical studies. 2. The relevance of the permanent-component doctrine for appraising the equity of burden distribution. Under usual standards of equity, is it preferable to tax a person on the basis of the actual income received during the year, or this figure adjusted to omit transitory components? May the positive transitory elements be regarded as even more suitable as a basis for taxation than the permanent elements? May the negative transitory elements be regarded as significantly affecting taxpaying ability? Given the pattern of burden distribution which a sales tax produces, what significance should be attached to this pattern in the evaluation of the desirability of increased federal use of this form of tax? This is, of course, a value-judgment issue, which can be resolved only in terms of the pattern of distribution of real income which a person regards as the optimum. But there are certain aspects of the question which must not 16 This question has been explored by Reed R. Hansen; see "An Empirical Analysis of the Retail Sales Tax with Policy Recommendations," National Tax Journal, March 1962, pp. 1-13. 17 See, for example, the articles by D. G. Davies, "An Empirical Test of Sales Tax Regressivity," Journal of Political Economy, February 1959, pp. 72-78; and "Com­ modity Taxation and Equity," Journal of Finance, December 1961, pp. 581-590. 18 See D. G. Davies, "Progressiveness of Sales Taxes in Relation to Various In­ come Bases," American Economic Review, December 1960, pp. 587-595; D. 0. Morgan, Jr., "Reappraisal of Sales Taxation," National Tax Journal, March 1963, pp. 89-101.

INTRODUCTION

be overlooked. One is the need for considering the pattern of distribution in terms of the federal tax structure as a whole, not of a particular tax", this raises the question of the patterns of the other taxes, which have been studied extensively by Musgrave and others. Second, the same persons are both federal and state-local taxpayers; the most meaningful picture is one which considers the combined federal-state-local pattern. As noted above, the sales tax produces widely varying burdens among various persons within given income classes, on the basis of their cir­ cumstances—family size, age, urban or rural dweller, and so on.19 The burden may be regarded as perverse in some instances; the larger family will spend higher percentages of income on taxable goods and thus pay more tax, whereas its taxpaying ability may be regarded as less. The significance of this perversity is a value-judgment issue. Another equity consideration often raised is the fact that the sales tax ensures some payment from persons able to evade or avoid income tax. This is obviously true; the basic issues relate to the significance of such evasion and avoidance in reducing the equity of the income tax, and the relative suitability of the sales tax approach to the problem compared to that of improving the structure and enforcement of the income tax. A final problem arises in connection with proposals for substituting a sales tax for the corporate income tax. Such a substitution would greatly enhance the value of the corporation as a tax shelter for individ­ ual stockholders unless provision were made for full, current taxability to the stockholder of his share in the corporation's earnings. ADMINISTRATIVE AND COMPLIANCE CONSIDERATIONS

An argument which has frequently been advanced for a sales tax is the relatively greater ease of administration and lessened problems of tax­ payer compliance. The validity of this argument in the context of a federal sales tax is open to serious question, since the issue is not that of replacement of the income tax by a sales tax, but increased relative reliance on commodity taxation. The issues therefore center around the question of the relative administrative costs and effectiveness of the two programs: the present system, and one which includes a sales tax. It is difficult to argue that the latter would be less expensive for a given degree of effectiveness in enforcement. The precise answer depends on several considerations: 1. If a federal sales tax were introduced, would the exemption figure 19 See

Hansen, "An Empirical Analysis."

19

INTRODUCTION

of the income tax be raised? This would reduce both enforcement and compliance costs to a much greater extent than a lowering of income tax rates without a change in the number of taxpayers. 2. What form would a federal sales tax take? From a cost-of-collection standpoint, clearly the manufacturers' sales tax, on the Canadian pat­ tern, is the most inexpensive, and a value-added tax confined to the manufacturing sector would be comparable. But in many respects, which have been widely discussed in the literature, the retail sales tax, or a value-added tax extended through the retail level, is a much more satis­ factory form of levy. The collection costs would, however, be somewhat higher. State experience suggests that with a 3 per cent tax rate, effective collection requires administrative expenditures of roughly 1.75 per cent of revenues.20 Since the expenditures would not be significantly greater with higher rates, a 10 per cent tax would require a collection expenditure equal to perhaps .7 per cent. The collection costs of the 11 per cent Canadian manufacturers' sales tax is .37 per cent.21 INTERGOVERNMENTAL FISCAL RELATIONS

Another major issue is the significance of the introduction of a federal sales tax for federal-state fiscal relations. As noted, the sales tax has become the major source of state revenue, with a definite upward trend in the number of states using the tax, in rates, and in the coverage of the taxes. There are several major issues: 1. Would federal use of the tax retard further state usage; if so, where would the states turn in their quest for needed funds? It can be argued that state-local functions are now less adequately financed than federal functions, and that a federal sales tax would aggravate the situation. The net result might easily be an increased program of federal grants to the states, a trend to which there are obvious objections. 2. If state usage was not retarded, would the combined federal-state sales tax burden be regarded as unreasonable, even if the federal tax in itself was regarded as acceptable? 3. Would there be less interference with state use of the tax if the federal government levied its tax at the manufacturing level? This is the argument long advanced by the National Association of Manufac­ turers in its defense of the choice of the manufacturing level for the tax. 20 J.

F. Due, /State Sales Tax Administration, Chicago, 1963. supplied by Canada, Department of National Revenue, Septem­ ber 1963. 21 Information

INTRODUCTION

Or would this only complicate matters by creating two different types of sales taxes? In Canada, where the provinces tax at the retail level and the federal government at the manufacturing level, increased support has been developing for coordinating these two sets of taxes into a single levy at the retail level. 4. Would federal intervention in the field facilitate state enforcement of taxation on interstate transactions? This development is possible if the federal government should use the retail form of tax. The change would improve administration, but might lessen the tax autonomy of the states. The equity, administrative, and intergovernmental issues are considered in the Eldridge paper. EUROPEAN EXPERIENCE

The argument has been raised that the countries of western Europe have expanded more rapidly than the United States in the last decade and have experienced less unemployment because of the greater relative reliance on indirect taxes.22 Thus, it is argued, a major step necessary for more rapid growth in this country is the revision of the tax system in the direction of greater reliance on indirect taxation. This point of view has been repeatedly expressed in the First National City Bank Monthly Economic Letter, and is stated in moderate form in the Com­ mittee for Economic Development's Reducing Tax Rates for Production and Growth. This point is regarded as of sufficient importance to warrant extensive analysis, the results of which appear in the Eckstein paper. There are several major issues: 1. What have been the relative growth rates in western Europe and the United States? What has been the relative growth in the United States and such countries as Canada which have similar economic con­ ditions but rely more heavily on indirect taxes than the United States? 2. What is the actual relative reliance on the two forms of taxes in the various countries? The answer to this question is not as easy to find as it might appear to be. There are questions about the inclusion in the totals of various social security levies and the classification of property, payroll, and other taxes. There is a tendency to make the comparison only in terms of national government taxes; would a comparison includ22 Descriptions of European sales tax systems are to be found in the paper by Clara Sullivan, "Sales Taxes in the European Economic Community and Some Important Issues Involved in their Harmonization," Report of the 196% Conference, Canadian Tax Foundation, pp. 254-265; in J. F. Due, Scdes Taxation, Urbana, 111., 1957; and in OEEC, The Influence of Sales Taxes on Productivity, Paris, 1958.

21

INTRODUCTION

ing all levels of government not be more appropriate? This basis en­ counters the difficulty of varying tax structures of the subordinate levels within a country. 3. Is not the significant question the relative burden of the two forms of taxes in the various countries, rather than the percentage reliance on each form? A country might rely exclusively on income taxes, but employ much lower rates than those used in countries in which, despite extensive reliance on indirect taxes, high government expenditures rela­ tive to national income necessitate high rates. The tasks of comparing relative burdens of particular taxes among countries are themselves very serious. Mere comparison of tax rates alone is misleading, since differences in coverage, exemptions, special concessions, administrative effectiveness, and the like are significant but difficult to measure in quantitative terms. 4. Do not the patterns of public expenditures and the general attitude of the public toward the government, the tax structure, the effectiveness of tax administration, and other factors have significant but unmeasurable influences on the relative effects of the various tax structures on economic development? 5. To what extent can differences in the rates of economic growth be attributed to tax differences, rather than to other factors? Will not a country-by-country comparison be more revealing than a general com­ parison between the United States and western Europe or the Common Market countries as a group? OTHER ISSUES

There are a few other issues which are now regarded as being of minor significance. At one time the supporters of indirect taxation made ex­ tensive use of the stability-of-revenue argument: the yield of indirect taxes falls less rapidly in depressions than the yield of direct taxes. There are two questions to be raised: (1) What is the actual difference in the two patterns of behavior?23 (2) To what extent is stability of revenue over periods of fluctuating national income desirable? In terms of modern fiscal analysis, stability is an undesirable characteristic. Another issue is tax-consciousness: the desirability that taxpayers be aware of the tax burden which they are bearing. Awareness is maintained with the retail type of tax or the value-added tax extended through the 23 For a recent empirical study, see the article by D. G. Davies, "The Sensitivity of Consumption Taxes to Fluctuations in Income," National Tax Journal, Septem­ ber 1962, pp. 281-290.

INTRODUCTION

retail level; it is lost with the manufacturing or wholesale sales taxes, since the tax cannot be kept separate from the prices of the products with these forms of tax. But the basic issue is the importance to be attached to the tax-consciousness argument. A third consideration is the argument that indirect taxes, particularly general sales taxes, make all persons contribute to government, whereas the usual income taxes do not. This argument can be debated on both equity grounds and those relating to the attainment of optimum levels of government expenditures. Should all persons be made to pay some taxes, in order to keep them aware of the real costs of governmental services? Are such payments at the state and local level not sufficient to accomplish this purpose? Is the fact that low-income groups are almost certain to experience a net gain from governmental activities anyway, even if they must pay some tax, an effective answer to this argument?

Conclusion This statement of issues serves to emphasize one well-recognized aspect of the controversy over increased federal reliance on indirect taxation: there can be no scientific answer to the question. The resolution of many of the particular issues must be based upon value judgments and upon inadequate evidence about economic effects of the various taxes. Fur­ thermore, value judgments rather than scientific analysis must serve as the basis for weighing the relative importance of various conflicting considerations so far as the major issues are concerned: equity versus economic effects, for example. But it is hoped that the papers prepared on various aspects of the problem and the discussion of them may aid in the reaching of more intelligent judgments on the question. The con­ ference, therefore, should aid in the clarification of the issues and thereby contribute to public policy formulation in this area.

Taxation, Resource Allocation, and Welfare ARNOLD C. HARBERGER UNIVERSITY OF CHICAGO I. Introduction1 DURING the past decade the traditional solution to the issue of direct vs. indirect taxation has been subjected to serious challenge. Whereas earlier the theoretical superiority of direct over indirect taxation was accepted as virtually axiomatic, now it must be regarded as something that cannot be "proved," and that in principle has to be investigated for each particular case. The traditional case for direct taxation rested on its alleged "neutrality" among the alternatives open to consumers. Whereas an indirect tax on one item of consumption made that item artificially expensive to consumers, relative to other goods, and thus consumers were faced with relative prices that did not reflect the relative marginal costs of production of different goods, an income tax was generally held to be free of this defect. The income tax, it was held, was similar to a system of excise taxes striking all commodities with an equal percentage tax rate. Whatever might be the ratios between the marginal costs of producing different goods, the same ratios would (under competition) apply to the prices facing consumers. Marginal rates of substitution in consumption (given by the ratios of gross-of-tax prices between pairs of goods) would therefore be equal to marginal rates of substitution in production (given by the corresponding ratios of net-of-tax prices). Many economists realized, long before 1951, that this proposition 1 This introductory section is meant to serve, in addition to the usual functions of an introduction, as a guide to readers who might otherwise be put off by the amount of mathematics that appears in the main body of the paper. The principal motivation for and conclusions from each section are set out here in nonmathematical terms, so that readers may skip any offending section of the main text without sub­ stantial loss of continuity. For the further guidance of readers, let me here point out that Section IY contains the most mathematical manipulation, and can be skipped with the aid of the summary in the introduction without serious loss. Sec­ tion II, on the other hand, develops the fundamental framework that is used through­ out the paper. Skipping this section therefore entails substantial loss of content. I have tried, in Section II, to develop the basic argument with the aid of graphical analysis, so as to ease the burden on less mathematically oriented readers, and I hope, accordingly, that most readers will be able to work through the derivations presented there. In comparison with Sections II and IV, the remaining sections contain only a modest amount of mathematical manipulation.

TAXATION, RESOURCE ALLOCATION, WELFARE

would have to be modified in a situation in which the supply of labor was other than completely inelastic. But the extent of the required modi­ fication was not appreciated. In point of fact, the required modification is so great that it destroys any theoretical presumption of the superiority of direct over indirect taxation. This was pointed out by Little in his fundamental paper written in 1951.2 He presented a simple example in which labor was the only factor of production and in which there were only three "goods": say, bread, wine, and leisure. In this case, an excise tax on wine would distort the choices between wine and bread and between wine and leisure, but would leave the choice between bread and leisure undistorted. Similarly, an excise tax on bread would distort the choices between bread and wine and between bread and leisure, but would leave the choice between wine and leisure undistorted. However, an income tax (interpreted as an equal-rate excise tax on both bread and wine) would distort the choices between bread and leisure and between wine and leisure, leaving the choice between bread and wine undistorted. Each of the three cases distorts two of the three possible choices, while leaving the third choice undistorted. There is thus no qualitative differ­ ence between the nature of the effect of direct as against indirect taxa­ tion. Any preference for direct taxation over indirect must therefore be based on quantitative rather than qualitative comparisons, or else on grounds (like equity or political feasibility) that are not strictly economic. This challenge to the traditional preference for direct taxation pro­ vides the focus for most of the present paper. Once Little's argument is appreciated, we can no longer resolve the issue by saying that distortions are present in the indirect-tax case but absent in the direct-tax case. We must recognize that distortions are present in both cases, obtain relevant measures of their effects, and then compare these measures. The relevant measure, in this case, is the cost to the economy of the inefficiencies resulting from tax-induced distortions of choices. This cost, which I like to call the welfare cost of a tax system, has traditionally been labeled the "excess burden" of taxation. Section II of the paper is an attempt to expound the principles of measuring the welfare costs of a set of taxes in a case where all goods are produced at constant cost. The case chosen is a particularly simple one, because, with the simplification of constant costs, the argument 21.

M. D. Little, "Direct Versus Indirect Taxes," Economic Journal, September

1951, pp. 577-584.

TAXATION, RESOURCE ALLOCATION, WELFARE

cin be carried out in diagrammatic terms.3 From this exercise a general expression is derived, from which we can determine the welfare cost of any given pattern of taxes on final goods and services when production is governed by constant costs, when distortions other than taxes are absent, and when the amount of labor supplied by each individual to the market does not vary as a consequence of changes in tax rates. In Section III we generalize the result of Section II, incorporating the possibility of tax-induced changes in the supply of labor. Thus we face up directly to the problem posed by Little, for the formulation of Section III shows how it is possible to measure the welfare cost of an income tax and to compare this with the welfare costs induced by alter­ native patterns of indirect taxation. To emphasize the possibility of extracting useful numerical results from analyses of this type, an attempt is made at the conclusion of Section III to estimate the welfare cost of the U.S. personal income tax. It should be borne in mind that this exercise is intended mainly as an example of how the job might be done, and not as a definitive analysis of the U.S. income tax. Among other things, the measurement effort of Section III takes account only of the costs of the personal income tax arising from the labor-leisure choice, and not those that affect occupational choice, investment decisions, etc. Moreover, the measurement is based on crude and scanty evidence of a highly aggregative nature, whereas a more complete analysis would take into account the differential responses of different classes of labor supply to tax incentives (e.g., distinguishing between the effects of taxation on labor-force participation rates of women and men). Finally, the measurement in Section III (as distinct from the theory presented) is based on the assumption that the income tax is the only tax present in the system, or, perhaps more plausibly, on the assumption that the welfare costs of the personal income tax are not altered by the many other taxes that are present in the U.S. system. In spite of these qualifications, however, I feel that the exercise of Section III provides us with some useful insights. The estimated welfare cost of the U.S. personal income tax is about f1 billion per year—hardly a negligible figure. Yet the simplifying assumptions just alluded to prob­ ably have the effect of understating the true cost. It is hard indeed, in the light of this result, to hold to the traditional position that the personal income tax is truly neutral. 3 For a more general mathematical treatment, see my paper, "The Measurement of Waste," American Economic Review, May 1964.

27

TAXATION, RESOURCE ALLOCATION, WELFARE

In Section IV we attempt a direct comparison of the welfare costs of income and excise taxation. This entails nothing more than a straight­ forward application of the general formulation derived in Section III, but some relevant insights are obtained by carrying out the steps. As the first step in Section IV, we try to find the "best" way of raising a given amount of revenue by commodity taxation. We find, as others4 have previously shown, that an income tax (i.e., a tax striking all goods and services other than leisure at an equal rate) will be the "best" way only in very unusual circumstances. If, as seems to be the case in· the real world, we cannot tax leisure as such, then the "best" tax system we can achieve under this constraint is one that strikes at higher-thanaverage rates those goods that are complements to or poorer-than-average substitutes for leisure, and that strikes at lower-than-average rates those goods that are better-than-average substitutes for leisure. An income tax is "best," given the constraint that leisure cannot be taxed, only if all goods are equally good substitutes for leisure (i.e., only if a tax on leisure would lead to an equiproportional change in the consump­ tion of each and every good and service). This conclusion could indeed be shattering to the principle of direct taxation if it meant that the welfare costs of taxation could be greatly lowered by having widely different rates of taxation on different goods and services as against the equal rates implied by income taxation. It does not appear, however, that this is the case. One way of determining the answer to the question just posed is to compare the welfare cost of a particular indirect tax with the welfare cost of an income tax yielding equal revenue. This is the second step taken in Section IV. It recog­ nizes that the relevant substitute for an indirect tax is not "no tax at all" but rather some other manner of obtaining the same yield, and it con­ siders the particular alternative represented by income taxation. The result of this exercise is that, as long as the excise tax in question is not itself a very general tax (striking a large fraction of all goods and thus approaching an income tax in coverage), and as long as the elasticity of demand for the commodity or commodities subject to excise tax is not very small, the substitution of an income tax for the excise tax in ques­ tion will very likely result in a reduction of welfare cost. The third step in Section IY attempts to make the preceding compari4 See Little in Economic Journal, September 1951; W. J. Corlett and D. C. Hague, "Complementarity and the Excess Burden of Taxation," Review of Economic Studies, No. 54, 1953-54, pp. 21-30; J. E. Meade, Trade and Welfare, Vol. II, Mathematical Supplement, London, 1955; R. G. Lipsey and K. Lancaster, "The General Theory of Second Best," Review of Economic Studies, No. 63, 1956-57, pp. 11-32.

28

TAXATION, RESOURCE ALLOCATION, WELFARE

son a bit more realistic. One of the key propositions that emerges from the study of welfare costs in a general- equilibrium framework (such as that taken in this paper) is that the effects of a particular tax depend not only on that tax itself but on the other taxes that are present along with it. The comparison made in step two of Section IV considers an excise tax that is being used as the sole tax in the system, and is replaced by and compared with an income tax of equal yield. In step three of Section IV we alter this comparison, and instead consider an excise tax which, when present, stands side by side with an income tax of a given size. We compare this setup with an alternative one in which the excise tax is taken off and replaced by a sufficient rise in the rate of income tax to produce the same total yield as was previously obtained from the excise and income taxes combined. This comparison is slightly more complicated than that undertaken in step two, but does not greatly alter the conclusion. Broadly speaking, there is still a strong presumption in favor of replacing excise taxation with income taxation, as long as the excise taxation itself is not very general and as long as the taxed com­ modities do not have very low elasticities of demand. When one comes to consider excise taxes that have very low rates and/or very broad bases, however, the result is no longer so clear. In these cases, considera­ tions of the complementarity (or low substitutability) of an excise-taxed commodity with leisure might weigh the final judgment in favor of retaining the excise tax rather than replacing it by an adjustment of income tax rates. One might say, in short, that the theoretical case in favor of direct taxation has been forced into full retreat by the considera­ tions raised by Little and others in recent years; but that the practical case remains comparatively unscathed, particularly where the excise taxes in question have relatively high rates and do not strike a very high proportion of consumer expenditures. In Section V the effects of taxation on saving are considered. The treatment, far briefer than the subject deserves, focuses on the nonneutrality of ordinary income taxation with respect to the saving decision, and on the measurement of the welfare costs engendered by this nonneutrality. It is shown that income taxation has the effect of reducing the rate of saving, when the alternative to the income tax is a consumption tax yielding the same amount each year. A simple formula for estimating the welfare costs of the distortions introduced by income taxation through its influence upon the savings decision is then pre­ sented. In terms of this formula, a very rough judgment is reached on the possible order of magnitude of this class of welfare costs in the

29

TAXATION, RESOURCE ALLOCATION, WELFARE

United States economy. Finally, it is suggested that the welfare costs stemming from the differential tax treatment of different kinds of income from capital in the U.S. are likely to be substantially greater than the welfare costs arising from the influence of taxation on the saving decision as such. That is to say, the U.S. economy very likely suffers greater costs from tax-induced misallocations of its given capital stock than from the influence of taxation on the over-all size of that capital stock. Finally, in Section VI, an effort is made to draw some conclusions from the preceding analysis on the influence of taxation on economic growth. Since previous sections of this paper are concerned mainly with the labor-leisure choice and with the saving decision, the discussion of taxation and growth is focused in these terms. The conclusion is reached that the effects of taxation on the labor-leisure choice have a truly negligible effect on the economy's rate of growth, but that the effect of taxation on the saving decision may reduce the economy's annual rate of growth by as much as one-tenth or two-tenths of a percentage point. For those who are surprised by the small effect of taxation via the saving decision, Section YI reviews the evidence sup­ porting this result. To set the stage for the treatment which follows, I should first like to discuss certain key assumptions that will be used throughout this paper. 1. It is assumed that, under any two situations being compared, the economy's productive resources are fully employed. This assumption is the conventional one in studies of the allocative effects of taxation. The resource allocation question can be framed as "how will our re­ sources be used?" as distinct from the alternative income-and-employment question of "will some of our resources be idle?" The effect of this assumption, for the problems treated in this paper, is to restrict the set of possible resource allocations to that determined by the production frontier of the economy. Thus in Figure 1, the alternative bundles of goods Xi and X2 produced in any of the cases to be compared will lie somewhere along the production frontier AB. A production point like C, indicating that the economy is not producing up to its potential, is ruled out by assumption. 2. It is assumed that, under any two alternative taxes to be compared, the government will obtain just that amount of purchasing power needed to buy a given bundle of goods. One need not belabor the point that if we are to discuss the effects of a given tax on resource allocation, we must implicitly or explicitly compare the situation in which that tax

30

TAXATION, RESOURCE ALLOCATION, WELFARE

Figure 1

exists with some alternative situation. Once this is recognized, we have a variety of choices. We can compare the "tax situation" with a "no-tax situation" or with any possible combination of alternative taxes. But this is not the worst of it. If the total revenues of the government have different purchasing power in the cases being compared, we must either have differing amounts of deficit or surplus in the two cases, or differing amounts of government expenditure. If surpluses or deficits differ, de­ flationary or inflationary forces are set in motion by the tax change, and we are taken from the realm of allocative theory into that of income-andemployment analysis. If government expenditures differ in the two cases, we must inquire into how the government spends its added revenue or re­ duces its prior expenditures. This opens a Pandora's box of possibilities in an analysis of resource allocation effects, for there are an infinity of ways in which the government could allocate any increase or reduction in expenditures. To avoid having to guess what the government would do with its money, we simply assume that the government would do the same thing in both the situations being compared. This, together with the previous assumption, enables us to conceive of a "consumption frontier" or, perhaps better, a private-sector expenditures frontier, ob­ tained by deducting the fixed pattern of government purchases from the economy's production frontier. Thus, in Figure 2, every point on the "consumption frontier" CD is obtained by deducting the fixed amounts of government purchases Gi and G2 from the corresponding point on the production frontier AB. As long as the government is getting the bundle (G1, G2) of Xi and X2, and as long as the economy is operating on the production frontier, production will take place somewhere within the segment EF of AB, and the "consumption" of the private sector must lie somewhere along CD in any two cases being compared. 3. It is assumed that the changes in output dictated by the changes 31

TAXATION, RESOURCE ALLOCATION, WELFARE

Figure 2

in tax policy being considered are accomplished at (approximately) constant unit cost. Taken strictly, this assumption requires that the production frontier be linear within the relevant range. Taken less strictly, it requires only that the production frontier be reasonably well approximated by a linear function within the relevant range. I am pre­ pared to argue that this is likely to be the case for the changes in output which would be induced by plausible changes in the taxes now in force in the United States. But I must confess that this assumption is mainly dictated by considerations of convenience. The mathematical parts of subsequent sections of this paper would be greatly complicated if the assumption of constant costs were abandoned, and yet the principal qualitative conclusions of the analysis would be unaffected. 4. It is assumed that, where "welfare costs" are being measured, the tax setup being considered has as its alternative a tax system of similar over-all incidence. The alternatives being compared are assumed to have similar over-all incidence in order to isolate the allocative effects of taxation from its redistributive effects. For example, if we are concerned with the effects of eliminating a tax on jewelry, whose incidence among income brackets is as indicated in column 1 of Table 1, we might so adjust income-tax rates as to increase income-tax collections from each bracket by approximately these same amounts. Thus if income-tax col­ lections, by bracket, were originally as given by column 2 of Table 1, we would consider an income tax collecting the amounts given in column 3 to be an appropriate alternative to the tax setup given by columns 1 and 2. This is not to say that the real alternatives faced by policy-makers are likely to be as neatly comparable as those in Table 1. But if the practical alternative (say, 3) to alternative 1 is an income tax which is different in its incidence from alternative 2, we would simply analyze the move

32

TAXATION, RESOURCE ALLOCATION, WELFARE TABLE 1 TAX COLLECTIONS BY INCOME BRACKET, PER TAXPAYER

Alternative 1 Income Brackets

1 2 3 4 5

Excise Tax on Jewelry (1) 20 30 40 50 60

Income Tax (2) 50 100 200 400 1,000

Alternative 2 Income Tax (3) 70 130 240 450 1,060

from 1 to 3 as the sum of a move from 1 to 2, which we would evaluate using purely allocative criteria,6 plus a move from 2 to 3, which we would evaluate using purely distributive criteria. II. Measuring the Welfare Cost of Excise Taxes In this section, we begin by outlining the "textbook" treatment of welfare cost in the simplest terms, and proceed to extend the analysis to more complicated cases. The textbook case is illustrated in Figure 3. We are to compare a situation in which there is an excise tax of $1 per unit of Z (X = Xi; Px = $2.00) with a situation in which this tax does not exist, but in which the government is raising equivalent revenue by an income tax (X = X0; Px = $1). It is important to recognize at the outset that Dx is defined as holding government revenues constant. Thus at P4 we have government revenues equal to (Il)(X4), all obtained from the excise tax on X. As we move to Pi, the excise tax must be lowered to $.75, and government revenues from the excise tax alone are only ($.75) (X3)- However, since government revenues must be constant all along the curve, there must be, at point Ps, income taxation to yield (H)(X4) — ($.75)(Xa). At Pi income taxes must yield ($1)(X4) — 5 Strictly speaking, one can eliminate distributive criteria in evaluating the move from 1 to 2 only when each and every individual bears the same burden of tax in both cases. Actually, the move from 1 to 2 as described above would entail some redistribution of income in favor of jewelry purchasers and away from those who do not purchase jewelry or purchase only small amounts relative to the average of their income bracket. But this redistribution is entirely within income brackets, and thus is not likely in itself to significantly alter the pattern of demand. Moreover, if the equal treatment of equals is accepted as a goal of tax policy, one would have to regard the move from 1 to 2 as having beneficial effects on distribution (differing tastes of people with similar incomes should not make these people "unequal"). Thus, as long as there would be an allocative gain on this move, we can consider it as being slightly reinforced by distributive considerations.

TAXATION,

RESOURCE ALLOCATION,

WELFARE

Figure 3

they must yield and at P0 they must yield ($1)(X 4 ). The simplest exposition of the measurement of the welfare cost of excise taxation can be made by looking only at the left-hand diagram in Figure 3. As between the situation and the situation consumers have given up units of the good X . The value placed by them on the first unit given up was $1, but subsequent units had higher values. If we think of the tax being gradually raised from zero to $1, this is clearly evident. When the tax is raised from zero to $.01, the units of X given up have values between $1 and $1.01; when the tax is raised from $.01 to $.02, the units of X given up have values between $1.01 and $1.02. Finally, when the tax is raised from $.99 to $1.00, consumers value the units of X given up at between $1.99 and $2.00. The sum total of the value placed by consumers on all the X they have given up in the process of raising the tax from zero to $1 can be measured by the area But in the process of demanding less X as a consequence of the tax, consumers have released resources from industry X and increased their demand for other goods (F). The measure therefore overstates their real loss; from it we have to deduct the value of the extra Y they consume in the excise-tax case. Where there is no tax (or other divergence between the value placed by consumers on the marginal product of productive factors and the rewards received by those factors) on F, the value to consumers of the extra F they have in case ' compared to case ' I can simply be measured by the area This represents the amount which was originally paid in industry X to the factors transferred to F as a result of the tax. Deducting

34

T A X A T I O N , RESOURCE A L L O C A T I O N , W E L F A R E

from X o P 0 P i X 4 , leaves the triangle P 0 P 4 F as the measure of the welfare cost or "excess burden" of a $1 per unit excise tax on X . The right-hand diagram in Figure 3 may help readers to verify that X0P0FX4, does in fact measure the value to consumers of the extra Y they demand as a consequence of the tax on X. Again, considering a sequence in which the tax on X is gradually raised from zero to $1, we have first a slight displacement of the demand curve for Y as the price of X is raised from $1.00 to $1.01, and a little extra Y is bought at the price of $1.00. As the price of X is raised from $1.01 to $1.02, another slight displacement of the demand curve for Y occurs, and'again a little extra Y is bought at the price of $1. Summing up the values to consumers of the successive increments to Y generated by raising Px all the way to $2.00, we obtain the area Y0GHY4, which is equal to the area X0P0FX4 in the left-hand diagram. There is no contribution to welfare cost (corresponding to the triangle P0P4F) coming from the market for Y because, by assumption, no dis­ tortions are present in that market. The resource cost of an extra unit of Y is $1.00, and the moment consumers find that it is worth $1 to buy an extra unit of Y, they can freely do so. Thus the value to consum­ ers of each extra unit of Y they take is equal to the price they pay for it. To summarize the result obtained so far, when a tax on X of T x per unit is the only distortion present, the welfare cost of that tax can be measured by -^TxAX, where ΔΧ is the change in the consumption of X induced by the tax. If we define tx(=Tx/Px0) as the percentage rate of tax, we can note that (with constant costs) it also represents the percentage change in the price of X . We then can derive ΔΧ = i ) x x X t x , where ηχχ is the own-price elasticity of demand for X. Substituting this expression for ΔΧ, and Px0tx for Tx in the cost measure ( —|ΓχΔΧ), we obtain as an alternative expression for the welfare cost of an excise tax —\XP x 0 T) x x t% This expresses the welfare cost of a tax in terms of ( X P x 0 ) , the total net-of-tax receipts from sales of the product in question; ηχχ, the own-price elasticity of demand for the product (a negative number); and tx, the effective percentage rate of tax. Since the total tax yield, Rx, can be expressed as XPx0tx, we can also express the welfare cost of the tax as — \Rxr)xxtx. Since data on Rx and tx are usually available for any existing tax, only ηχχ need be estimated to obtain a measure of the welfare cost of such a tax. In the course of this paper we shall show that this simple textbook measure of welfare cost is not a bad approximation in the cases of most existing taxes. But in order to reach this conclusion, we must modify

35

T A X A T I O N , RESOURCE ALLOCATION, WELFARE

the above analysis to take account of the main ways in which the realworld situation diverges from that depicted in Figure 3. By far the most disquieting assumption underlying Figure 3 and the analysis of it is the assumption that no excise taxes or other distortions exist in the "rest" of the economy (industry F). It is possible, however, to modify the analysis so as to avoid making this assumption.' Figure 4 differs from Figure 3 only in that the initial price of Y is taken

^(Ρ,-ίΙΟΟ)

D^Px=S2.00)

$2.00 $1.75 $1.50

$1.50

$1.25 1.00

$1.00

Y Figure 4

to be $1.50, consisting of a unit cost of $1 plus a tax of $.50 per unit. Once again, as we envision the tax on X being raised in small steps from 0 to $1.00 per unit, we obtain the area X0PoPiXi as measuring the value to consumers of the (X0 — Xi) units of X that they are induced to forego as a consequence of the tax. The difference between this and the earlier case appears in the righthand diagram of Figure 4. Whereas, in Figure 3, consumers were initially paying $1 per unit of F, and continued paying that price after the tax was imposed on X, now, in Figure 4, consumers are initially paying $1.50 per unit of Y, and continue to pay that price throughout the exercise. Hence the value to consumers of the extra Y that they are induced to consume by the tax on X is F 0 ZJF 4 , rather than Y 0 GHY i . In the present case, therefore, the net change in the welfare of consumers as a result of the tax on X is measured by the difference between a loss of XoPoPiXi (on account of [X0 — Xi] units of X being foregone), and a gain of Y0IJYi (on account of [Yi — F0] more units of Y being con­ sumed). Since the areas X0P0FXi and Y0GHYi are equal as before (the cost of production of the addition to F is the same as the reduction in production costs of X), the net loss in the welfare of consumers can be measured by P0PiF minus GIJH. Since GIJH = (Iu) X (Y0GHYi), and

36

TAXATION,

RESOURCE ALLOCATION,

WELFARE

we can express This, in turn, can be expressed as Now so that the net loss in welfare from placing a tax of tx on X in the presence of an already existing tax of ty on Y can be expressed as: ~

"

From this it can be seen that there is a clear gain from placing a tax on X equal to the pre-existing tax on Y. The expression is always positive, and the expression is negative when Therefore the "cost" of moving tx from zero to ty is negative, indicating a gain in welfare. In the example chosen for Figure 4, tx was set at twice ty. Thus the expression above indicates that there is no gain or loss in welfare. This canbe verified by looking at Figure 4. The loss measured by the triangle j is just counterbalanced by the gain measured by the rectangle GIJH.6 The extension of this result to the case of three or more commodities is self-evident. In Figure 5, we assume that a tax of $.50 per unit on A%.

Figure 5

and one of $.25 per unit on Xs were already in existence before a tax of $1.00 per unit was levied on Xi. The reduction in welfare of consumers as a consequence of the tax on X\ is measured by the area ABC minus the sum of the areas DEFG and HIJK. We can now turn to the derivation of a general expression for the welfare cost of a set of excise taxes. Assume there are three commodities, all of which are taxed. The welfare cost of the whole set 6 Both the rectangle and the triangle have equal bases, but the height of the triangle is just twice that of the rectangle.

37

TAXATION, RESOURCE ALLOCATION, WELFARE of taxes can be measured by taking first the welfare cost of a tax on Xi, with no other taxes in existence; adding to this the addition to the welfare cost obtained by imposing a tax on X 2 , given that the tax on Xx already exists; and finally adding the increment to the welfare cost incurred by imposing a tax on X 3 , given that the taxes on Xt and X 2 already exist. Since the effects of the taxes will be analyzed in three steps, let us denote by Ai those changes taking place in the first step (when a tax is imposed on X i ) , and by A2 and A3, respectively, those changes taking place in the second and third steps. When a tax of 2\ is imposed on Xh with no taxes on X2 and X 3 , the welfare cost is simply i.e., the triangle

(a) in Figure 3. When a tax of T 2 is imposed on X 2

in the presence of a tax of Ti on Xi, the addition to welfare cost is (b) The first component is the triangle under the demand curve for X2 (corresponding to ABC in Figure 5), and the second is the offset (corresponding to DEFG in Figure 5) arising from the fact that a tax on X i already exists. (There is no offset corresponding to HIJK

in Figure 5,

because at this step we have not yet imposed any tax on X 3 .) When a tax of T3 is imposed on X 3 in the presence of taxes of on .

and rI\

the addition to welfare cost is

(c) The sum of

gives us the welfare cost of the set of taxes

I taken together. Let me here introduce Si,- as a general notation for defined to include just the substitution effect. We can then write:

38

TAXATION,

RESOURCE ALLOCATION,

WELFARE

Substituting into (a), (b), and (c), and summing, we obtain as the expression for the welfare cost of the set of taxes or alternatively Recognizing that

we can express this as

This is one of three fundamental forms in which the welfare cost of a set of taxes may be written. The second fundamental form may be derived by breaking up into expressions reflecting changes in the quantity of each commodity. Thus we have, from commodity 1,

Denoting the total change in

i, we

may express this as

(a') Likewise, we have for commodity 2,

which we can express as

(bO

Finally, we have for commodity 3,

which can be expressed as The sum of will give us the welfare cost of the (c') set of 7taxes The equality under consideration. But is one sixofofthe thefundamental nine termsproperties in the sum of demand cancel relationships reflecting the substitution effect only. See John R. Hicks, Value and Capital, p. 311. Hicks uses the notation Xy to refer to what we have called Sij. It is not easy to give an intuitive interpretation of why Sa should equal The best I can do is as follows: If the prices of both Xi and Xs rise by 1 per cent, no substitution effects will take place between them. This means that the substitution effect between X, and X 2 which takes place because of a rise of 1 per cent in Pi is annulled by the substitution effect taking place as a result of a 1 per cent rise in Pi. Now the transfer of purchasing power from Xi to Xi taking place as a result of a 1 per cent rise in the price of Xi is given by . And the transfer of purchasing power from Xi to Xi resulting from a 1 per cent rise in the price of X 2 i s equal to j . In order for the second39 effect to annul the first, must equal

TAXATION,

RESOURCE ALLOCATION,

WELFARE

each other o u t . Consider t h e t w o t e r m s that

W e know

and

T h u s these t e r m s c a n b e e x -

pressed as w h i c h m u s t equal zero since

B y the same token, e a c h of t h e

pairs of terms

and

must

equal zero. T h u s the s e c o n d f u n d a m e n t a l expression f o r welfare c o s t reduces t o

T o derive t h e third f o r m of the expression f o r welfare cost, consider the expression (first f o r m ) , letting C» = unit cost of c o m m o d i t y i, h t h e percentage rate of t a x o n c o m m o d i t y i, a n d

Now

t h e unit t a x o n i,

w e m a k e use of t h e w e l l - k n o w n relationship a m o n g t h e

Sa,

8 This relationship can be interpreted intuitively in two ways. First, when there is a linear constraint of the form then, as long as the constraint is satisfied,

for any definition of z. Thus setting;

we have

or

alternatively, we may start from the general property of substitution effects that "only relative prices count." This says that the effect on X< of a 1 per cent rise in its price would be annulled by a 1 per cent rise in all other prices. The effect on X i of a 1 per cent rise in its own price is given by

The effect on Xi of a 1 per cent rise in the price of on Xi of a 1 per cent rise in all other prices is

. The effect In order for the first effect,

SuPi, to be offset by the second, we must have Now if, as we do when measuring the total welfare cost of a set of distortions, we are measuring this cost as against an undistorted initial situation, we can characterize the undistorted situation as one in which, for all j, Pj = kC,; that is, prices bear the samerelationship to unit costs for all commodities. Thus we have See Hicks, Value and Capital, pp. 310-311. 40

TAXATION,

RESOURCE

ALLOCATION,

WELFARE

On the left-hand side of the equalities in the six equations above, we have the terms appearing within the bracket in the expression for welfare cost (first form). On the right-hand side, we have the expressions which will be used in deriving the third form. To do so, we simply collect terms in S12, S13, and S2i, yielding

Summing the three terms to the right of the equalities, we have an expression equivalent to that in brackets in the first form measure of the welfare cost. Since the bracket is multiplied by ), we have for our third fundamental form of the measure for the welfare cost:

The expression for the welfare cost of a set of taxes can be derived directly from utility functions as follows. (1) Taking a Taylor expansion up to quadratic terms, we have where

Now Ui is itself a function of the

and

(2)

so and

(3)

(4) Thus we can rewrite (2) as

(5) If consumers maximize utility subject to their money income and the prices facing them in any situation, we have, as a first-order condition of maximum, that (6) where X is a Lagrange multiplier representing the marginal utility of money. Obviously, from (6)

(7) Substituting (6) and (7) into (5), we have

(8) 41

TAXATION,

RESOURCE ALLOCATION,

WELFARE

Interpreting the P / s in (8) as representing the prices in an undistorted situation (i.e., as being proportional to unit costs), we have which must equal zero under our assumption of constant costs (linear transformation functions). Thus (8) reduces to

(9) To translate AU from utility terms into money terms, we must divide by the marginal utility of money. Thus money value of change in utility equals (10) which expressed as a cost is

When the only disturbances

between the two situations being compared are taxes, we have Thus from (10) we can derive ; as the measure of the welfare cost of a set of taxes . Moreover, since we require that changes in the all take place along a linear constraint, without first-order income effects, we can express Substituting this into

we obtain

as an alternative (equivalent) measure of welfare cost. This derivation is essentially the same as that given by H. Hotelling.9 Hicks derives the e x p r e s s i o n ( i n his n o t a t i o n , ) by a "

j

somewhat different route.10 III. Measuring the Welfare Cost of Income Taxation In this section we apply the apparatus derived in the previous section to the problem of direct taxation with particular reference to the choice between labor and leisure. We shall disregard problems connected with the choice between consumption and saving, which will be dealt with separately below, and shall accordingly assume that all goods are direct consumption goods, and that all income is spent on such goods. The income tax will be represented as a flat-rate excise tax striking all goods (other than leisure) at the same percentage rate. 9 "The General Welfare in Relation to Problems of Taxation and of Railway and Utility Rates," Ecmometrica, July 1938. 10 Value and Capital, p. 331.

42

TAXATION, RESOURCE ALLOCATION, WELFARE

To begin, let us briefly review the traditional case for preferring direct to indirect taxation. Here leisure is not considered as a good, and the supply of work is assumed to be unaffected by alterations in the pattern of taxation. The third fundamental form of the expression for welfare cost, i.e., IΣΣ

— tj)2CiCj,

} i