An International Comparison of Tax Systems in Industrial Countries
Total Page:16
File Type:pdf, Size:1020Kb
V An International Comparison of Tax Systems in Industrial Countries Enrique G. Mendoza, AssafRazin, and Linda L. Tesar1 he precise measurement of tax rates that affect projections of real present values for investment T economic decisions at the aggregate level is projects in specific industries. However, as critical to the design of economic models that simu- Frenkel, Razin, and Sadka (1991) argue, the com- late the effects of fiscal policies. The extensive ana- plexity of tax credits and tax exemptions, as well as lytical work on the macroeconomic implications of the numerous equivalences that link broad catego- different tax systems produced during the last ries of taxes, makes constructing effective marginal decade, as reviewed in Frenkel and Razin (1987), tax rates that affect actual economic decisions at the emphasized the importance of modeling explicitly aggregate level extremely difficult. It is also diffi- the structure of incentives and constraints under cult to show that marginal tax rates that apply to which households and firms formulate optimal particular individuals in a household survey, or a plans in order to produce reliable assessments of the specific aggregation of incomes based on tax- effects of policies. This is particularly true in an bracket weights, are equivalent to the aggregate tax environment of increasing international economic rates that affect macroeconomic variables as mea- integration.2 The literature established that the sured in conventional national accounts systems. models designed to simulate the effects of changes Moreover, detailed time series and international in fiscal policy must consider a realistic description cross-sectional applications of methods for comput- of the rates of taxation prevailing in different coun- ing effective marginal tax rates are seriously limited tries before experimenting with alternative policies. by the available data. The issue is particularly relevant in the context of Lucas (1990) and (1991) and Razin and Sadka current discussions on the implications of the con- (forthcoming) have opted for an approach that pro- vergence of fiscal policies envisaged for the duces effective average tax rates, for the taxes that European Community under the Maastricht Agree- generate the majority of the government's tax reve- ment and the agreements to harmonize indirect nue, based on data on actual tax payments and taxes, and on the effects of significant changes in national accounts. Their analysis suggests that the tax regime announced in the recent deficit- these tax rates are useful approximations of the reduction plan for the United States. taxes that distort decisions by representative agents Unfortunately, it is difficult to measure tax rates in macroeconomic models. Their method focuses that are relevant for macroeconomic modeling. on the information that national accounts data pro- Many studies have been written on the measure- vide regarding post- and pre-tax prices and ment of effective marginal tax rates on labor and incomes, combined with figures that aggregate tax capital income.3 Each constructs estimates of mar- revenues by allocating them to taxes on consump- ginal tax rates by combining information on statu- tion and factor incomes. This method, although less tory tax rates, tax returns, and tax codes with data rigorous in the treatment of the tax laws, produces on income distribution, household surveys, and measures of the tax rates that are consistent with the representative agent assumption and, by looking at 'International Monetary Fund, Tel-Aviv University, and the aggregate data, it also takes into account the University of California-Santa Barbara, respectively. Work on effective, overall tax burden resulting from each of this project started while Assaf Razin and Linda Tesar were the major tax categories (i.e., taxes on capital and visiting scholars at the Research Department of the Interna- labor income and taxes on consumption). In addi- tional Monetary Fund. We gratefully acknowledge comments and suggestions by Peter Clark, David Coe, Robert Ford, tion, this method is easier to implement in multi- Robert Hagemann, Carmen Reinhart, and Peter Wickham, as country research projects because it exploits the well as the research assistance provided by Kote Nikoi. international consistency of available data sources 2See Frenkel, Razin, and Sadka (1991). on national accounts and revenue statistics. 3For the United States, see Auerbach (1987), Barro and This paper describes an application of this Sahasakul (1986), Joines (1981), and Seater (1985) and for other countries, King and Fullerton (1984), McKee, Visser, method to compute time series of the effective aver- and Saunders (1986), OECD (1991b), and Carey, Chouraqui, age tax rates on consumption, capital income, and and Hagemann (1993). labor income for the group of seven largest indus- 86 ©International Monetary Fund. Not for Redistribution Methodology for Computng Effective Average Tax Rates trial countries, using information publicly available (1992) study the extent to which these tax rates can from the Organization for Economic Cooperation explain the international convergence or divergence and Development (OECD). Comparing these tax of growth patterns. Finally, the proposed tax rates rates with some of the available estimates of effec- are also being incorporated into the multi-country tive marginal tax rates, we find that, despite differ- macroeconometric model of the International Mon- ences, the tax rates reported here are within the etary Fund (MULTIMOD) with the aim of produc- ranges of marginal tax rate estimates and display ing policy simulations in which the short-run and similar trends. We also show that our estimates of long-run interactions of the tax rates with other the tax rates are generally consistent with some key macro-aggregates of interest are taken into account. predictions of macroeconomic models. In particu- lar, in most countries, the savings rate is inversely related to the tax rate on capital income, the average A Methodology for Computing Effective number of hours worked is negatively correlated to Average Tax Rates the sum of the labor and consumption taxes, and the rate of unemployment is positively correlated with While the concept of the marginal tax rate that the labor income tax. The first two results are con- affects the decisions of economic agents is simple in sistent with the intertemporal equilibrium model of theory, and quite easy to quantify at a micro- savings in an open economy, as explained in economic level, computing effective marginal tax Frenkel, Razin, and Sadka (1991); the second is rates that apply at a national or international level is consistent with models of equilibrium unemploy- quite difficult. Within one country, computing these ment, or the "natural rate," as in Pissarides (1985) tax rates is problematic (1) because tax revenue data and Adams and Coe (1990). The investment rate is and the tax system itself do not conform to the also inversely related to the capital income tax, aggregate concepts of a macroeconomic model; (2) reflecting the well-known positive correlation because the many exemptions and credits make it between savings and investment, and suggesting difficult to extrapolate the information from the stat- that the rates of taxation affecting the returns on utory tax rates written in the law; (3) because equiv- foreign and domestic capital tend not to offset each alent effects may result from different types of other. taxes; and (4) because of the need to have available Comparing the data across countries, we find that data on the distribution of income be consistent when tax rates on capital income are above average, with systems of income tax and social security con- savings and investment rates tend to be below aver- tributions. At an international level, differences in age, and when labor income taxes are relatively the structure of the tax systems and the limitations high, the rate of unemployment tends to be higher of the information available on tax revenues and and the number of hours worked tends to be lower. income distribution further complicate the comput- The international cross-sectional and time-series ation. Following Frenkel, Razin, and Sadka (1991) information is combined in panel data tests to for- and Razin and Sadka (1993), we look at effective malize the evidence obtained from the inspection of average tax rates based on actual tax payments and correlation coefficients. Finally, the cross-country national accounts as a useful approach. analysis highlights important differences in the dis- This section of the paper describes our method tribution of the tax burden on consumption, labor for computing effective average rates of taxation on income, and capital income between North Amer- consumption and the income derived from capital ica, Japan, and Europe, which indicate the magni- and labor services for the group of seven largest tude of adjustment that policies of tax harmoniza- industrial countries. Using data from two publica- tion may require. The analysis indicates that tions by the Organization for Economic Coopera- consumption taxes in the United States are signifi- tion and Development—Revenue Statistics of cantly lower than in Canada and the European Com- OECD Member Countries, OECD (1990) and munity, but that increasing the U.S. consumption National Accounts: Volume II, Detailed Tables, tax could result in a higher natural rate of unem- OECD (1991a)—we compute time series of the ployment