有效法定税比

Effective vs. Statutory Taxation: Measuring Effective Tax Administration in Transition Economies By: Mark E. Schaffer and Gerard Turley William Davidson Institute Working Paper Number 347 November 2000 1 Effective versus Statutory Taxation: Measuring Effective Tax Administration in Transition Economies* Mark E. Schaffer M.E.Schaffer@hw.ac.uk Centre for Economic Reform and Transformation Department of Economics, School of Management Heriot-Watt University Edinburgh EH14 4AS, UK Gerard Turley G.Turley@hw.ac.uk Department of Economics National University of Ireland Galway, Republic of Ireland November 2000 Abstract Wide differences between effective or realised average tax rates and tax yields that would result if statutory tax rates were strictly applied indicate tax compliance and collection problems. Due to the greater politicisation of tax systems in transition economies (TEs), we would expect the shortfalls in effective tax yields for TEs to be larger than a benchmark for the mature market economies where tax systems are well established, the administrative capacity is stronger and tax arrears are tolerated less frequently. The methodology involves calculating an effective/statutory (E/S) tax ratio. Initial results indicate that the leading TEs have E/S ratios similar to the EU average. We find a positive correlation between progress in transition and effective tax administration, as measured by our E/S ratio. For slow reformers, the effectiveness of tax collection appears to vary with the extent of state control. Those TEs that have maintained the apparatus of the state have done well in tax collection compared to those countries where there is evidence of state decay. This raises a number of broad policy issues relating to the speed of transition, the interaction of politics and economic reforms, the capacity of the state to govern and the need for market institutions to develop. Key words: statutory taxation, average tax rate, tax collection, effective administration, transition economies Journal of Economic Literature, Classification Numbers: H2, H32, H87, P5. 5 tables, 32 references. * The analysis in this paper builds on work conducted for the European Bank for Reconstruction and Development, to whom we are grateful for financial support. We are grateful also to Michael Alexeev, Alan Bevan, and the participants at the July 2000 Galway workshop on “Institutions in Transition Economies” for helpful comments earlier versions of this paper. All views expressed in this paper are, however, solely the responsibility of the authors. 1 1. Introduction This paper attempts to measure the effectiveness of tax administration in transition economies (TEs) and how it compares to a benchmark for the mature market economies. We measure the effectiveness of tax administration by comparing statutory tax rates with effective tax yields. This method of measuring the administrative capacity of tax systems has been alluded to in the taxation literature but not systematically pursued in cross-country comparisons. Alex Radian, in his inquiry into tax administration in poor countries, noted that effective tax rates are lower than legal tax rates (Radian, 1980). David Newbery raised the issue of differences between statutory and effective tax rates, again in the context of developing countries, when he stressed the importance of examining the effective tax system rather than the legally defined tax system (Newbery and Stern, 1987). In the same World Bank publication, Vito Tanzi suggested that the gap between the statutory tax system and the effective tax system might be large in developing countries (Newbery and Stern, 1987). Elsewhere, Burgess and Stern (1993) argued that the wedge between the statutory and effective tax systems can be reduced by improvements in administrative capabilities. In this paper, we take a methodology previously used for measuring fiscal or revenue capacity in federal states (ACIR 1962) and adapt it to enable inter-country comparisons of effective vs. statutory taxation. We then use actual fiscal and national accounts data from 25 TEs and, as a benchmark, the average for the 15 member countries of the European Union, to measure the effectiveness of tax administration. Tax exemptions, deferrals, write-offs and arrears that firms receive or extract from the state are widespread, not only in transition economies but in market economies as well. In a broader sense, these tax concessions are often manifestations of a tax system that is politicised. One possible result of this bargaining and general politicisation of the tax system is a low level of tax compliance combined with a high incidence of tax avoidance.1 Measuring the extent of this financial aid using firm-level information is difficult and faces obvious data difficulties, e.g., these concessions may not be widely known or may not show up in the government’s budget. By measuring the difference between effective and statutory taxation at the aggregate level, our 1 This paper deals with tax avoidance as distinct from tax evasion. Tax avoidance is defined as the use of the tax system to minimise tax liabilities or obligations. It is a legal activity as distinct from the illegal (and acutely difficult to measure) activity of tax evasion. One way that enterprises can reduce their tax burden is by transferring their tax liabilities abroad. Transfer pricing is common in market economies where multinational corporations use their foreign subsidiaries that operate in low tax jurisdictions to reduce their overall tax burden. A well-known case is that of Rupert Murdoch’s News Corporation, a company that 2 methodology enables us to obtain aggregate measures of the degree of effectiveness of tax collection that can be compared across countries, for different taxes, and over time. In this paper we develop indicators that allow us to measure how broadly and strictly valued added tax, payroll tax and corporate income tax are implemented and complied with in TEs. We use the average of the EU-15 countries as an appropriate benchmark for comparison. Our focus is on comparisons for the most recent year available; comparisons over time will be pursued in future work. 2. The Tax System in TEs The tax system that existed under the socialist command economy was different from a Western- style tax system. There was no corporate income tax system, in the usual sense of the term. State enterprises were subservient to the various ministries and any ‘profits’ made were expropriated by the state. Likewise, losses were made good by arbitrary pricing and subsidies. Often, tax rates were numerous and non-parametric, tax structures were complex and differentiated, and tax liabilities were discretionary and negotiable. The main sources of tax revenue were typically enterprise profit tax, turnover tax (with highly differentiated, product-specific rates) and payroll taxes; direct taxes on individuals were unimportant. Although taxation as a percentage of GDP was high in socialist countries, administrative costs were low and tax collection was straightforward as firms tended to be large and closely monitored. In the Soviet system, ‘taxes’ were collected from the enterprise, with the State Bank acting as the fiscal agent of the state. Once the socialist system collapsed, TEs, some lacking an explicit tax system (or culture), had to build a market-oriented, rule-based tax system (including a market-type tax administration) from scratch. The creation of a new tax system involved the introduction of a corporate income tax system. Not only did this involve changes in how enterprises were treated by the state in terms of taxation, but it was also introduced in conjunction with other policies, such as price liberalisation, demonopolisation and privatisation. A VAT system to replace the turnover tax was introduced in the early years of transition and was in place in most TEs by 1994. Tax on individuals accounted for a small proportion of the total tax take in socialist countries: the transition to a market economy meant that a personal income tax system as operates in market economies was also to be introduced. made ₤1.4 billion in profits in Britain but reportedly paid no corporation tax there (The Economist, 29/1/00). 3 In the context of TEs and taxation, we are interested in the hardening of budget constraints of firms. Hence, the firm is the unit of analysis in this paper. As for the different taxes paid by the firm, corporate income or profit tax, value-added tax and social security taxes are mostly linear, flat rate taxes.2 Neither sales taxes (because they are levied at one stage only) nor excise taxes (because they are product-specific) are considered in our study. We do, however, treat social security contributions as a payroll tax. As mentioned above, we will concentrate on three taxes paid by the firm: corporate income tax (CIT), value-added tax (VAT) and social security tax (SST). Although VAT is essentially a tax charged on final purchasers, it is imposed at different stages of production at the firm level. For these three taxes, we estimate the difference between effective average tax rates and tax yields that would result if statutory tax rates were strictly applied. The methodology is explained below. 3. Methodology Our methodology is based on one commonly used in measuring tax or fiscal capacity in federal states. The ACIR Representative Tax System (RTS) method was initially proposed in the early 1960s and has been modified on several occasions since then (see ACIR 1962; 1971; 1982). Essentially, by applying national average or representative tax rates to member-state tax bases, the RTS method shows the amount of revenue that could be collected by the individual member states of a federal country, i.e., their fiscal capacity. With some modifications, we apply this methodology to sovereign states (and transition countries, in particular) rather than to states within a federal system. We begin with some definitions. Statutory tax rates are the rates that taxpayers are required to pay by law. Effective tax rates are the realised average tax rates. These are the same average tax rates as employed by Whalley (1975), Lucas (1990) and Mendoza et al. (1994).3 2 A linear tax is a tax whose marginal rate is constant. 3 These are aggregate average tax rates as distinct from the effective marginal tax rates that are commonly used in studies of household income, income distribution and taxation. Aggregate tax rates are normally used in macroeconomic modeling and in the taxation, economic growth and supply-side economics debate. For the problems associated with measuring effective tax rates, see Fullerton (1984). 4 Let Y be the gross tax base and T be actual tax payments; hence income net of tax is Y-T. We denote by t the statutory tax rate applied to gross income. The effective tax rate e, also defined on a gross basis, is calculated by dividing actual tax payments T by the appropriate gross tax base, or Y Te ≡ (1) Using the statutory and effective tax rates thus defined, we calculate two indicators that measure the effectiveness of tax administration. The first indicator is the ratio of effective tax to statutory tax. The effective/statutory (E/S) ratio is defined as follows: tY T t eratiostatutoryEffective ≡≡/ (2) This indicator measures the extent of the wedge between the statutory tax rate and the realised average tax rate. A ratio close to 1 indicates that the effective tax rate is close to the statutory tax rate. A ratio below 1 indicates that the effective tax yield is falling short of what application of the statutory tax rate would yield. Differences across countries in the extent of this shortfall in revenue may be accounted for by tax breaks, tax arrears and tax avoidance measures.4 As approximations of the gross tax base Y, we use national accounts measures of income: for VAT, total national income (GDP); for SST, income from labour; and for CIT, income from capital. These are, of course, only rough approximations of the actual statutory tax bases. For example: even in market economies, large portions of the economy are exempt from VAT (e.g., public administration and education); corporate income tax applies only to corporations and the usual tax base is net of depreciation and interest; for all three taxes, entities must usually be over a certain size threshold before becoming liable for taxation. Furthermore, the national accounts statistics of transition countries are generally regarded as less reliable than those of developed market economies, including their statistics on the division of GDP into labour and capital income. National accounts measures of income do, however, have the important advantages of being both readily available and, in principle, readily comparable across countries. Moreover, 5 our focus is not on levels of effective taxation in countries but on comparisons of levels across countries, and as noted the reasons for these deviations between income measures and statutory tax bases are found in all countries. Our second indicator entirely avoids these possible problems with national accounts measures of income by simply not attempting to match tax payments to the appropriate tax base. The normalised tax yield (NTY) instead relates tax payments adjusted for cross-country differences in statutory rates to GDP, and is defined as follows: t b GDP TyieldtaxNormalised ≡ (3) where t is the statutory tax rate and b is a benchmark rate. Put simply, the normalised tax yield tells us what the tax yield (for a particular tax) would be for a specific country if the statutory tax rate were the same for all countries. In the definitions above, we have used the convention of a tax base that is gross of tax. In practice, statutory tax rates are sometimes defined relative to a tax base that is inclusive of tax and tax liabilities are paid out of gross income; and sometimes statutory rates are defined relative to a tax base that is net of tax. Corporate income tax (like personal income tax) is an example of the former; tax liability is calculated by applying the corporate tax rate to gross profit. VAT and SST are examples of the latter. Firms calculate their gross VAT liability by “adding on” VAT as a percentage of the pre-tax price, and SST is typically calculated as a percentage of wages and salaries paid. We use the former convention – the tax base is gross of tax – for our calculations for all three taxes considered. This requires adjustments to the statutory tax rate for both VAT and SST.5 4 Accounting for tax evasion will depend on whether the national accounts adjust for the hidden economy. If there is an evasion adjustment in the calculation of GDP, tax evasion is captured by our measure; how fully will depend on the accuracy of the adjustment. 5 We are therefore using gross income only and adjusting statutory rates as necessary. The alternative approach would be to use statutory rates in conjunction with gross or net income, as appropriate. This approach, however, encounters data availability problems with net income. As noted above, we use national accounts measures of gross income that are only approximations of the actual tax base. Calculating net income from these approximations and the actual tax yield introduces further measurement error into the tax base approximations. By contrast, our tax rate adjustment in equation (5) is an identity and introduces no such measurement error. 6 Denote by tN the tax rate applied to net income Y(1-t) that would yield the identical tax revenue as the tax rate t applied to gross income Y. We then have, by definition, [ ])1( tYttY N −≡ (4) Rearranging equation (4) yields N N t tt + ≡ 1 (5) Equation (5) is used to convert a tax rate defined by statute as applying to net income into the equivalent tax rate on gross income. We illustrate this adjustment by reporting the benchmark rates used to calculate our normalised tax yield indicator. We denote by bN the benchmark net of tax rate equivalent to the gross rate b. We take as our benchmark rates the approximate (rounded to the nearest five or ten per cent) average statutory tax rates in use in the 15 member states of the European Union: bN=20% for VAT, bN=40% for SST, and b=35% for CIT. The following table reports the equivalent gross and net rates for these benchmarks; the figures in bold are the statutory rates as legally defined by statute. Table 1: Benchmark Tax Rates, Gross and Net Equivalents Rates defined by statute in bold b bN VAT 16.7% 20% SST 28.6% 40% CIT 35% 53.8% 4. Data Coverage and Sources Our primary interest is in examining the administrative capacity of tax systems in TEs and how it compares to levels in well-established market economies. We use the mean of the EU-15 countries as a benchmark, taking 1996 as the benchmark year. 7 There are 25 ex-socialist countries in our study; the TEs that are not included are those where tax data are difficult to obtain (Mongolia, Vietnam), where the tax system is highly complicated (China) or where war has occurred (Serbia-Montenegro, Bosnia-Herzegovina). Ten of our 25 countries are CEE countries (Albania, Bulgaria, Croatia, Czech Republic, FYR Macedonia, Hungary, Poland, Romania, Slovak Republic and Slovenia) for which we use data primarily for the period 1991 – 1997. The remaining 15 are all FSU countries (Armenia, Azerbaijan, Belarus, Estonia, Georgia, Kazakhstan, Kyrgyzstan, Latvia, Lithuania, Moldova, Russia, Tajikistan, Turkmenistan, Ukraine and Uzbekistan) for which the relevant period is 1992 – 1997.6 All these countries have a corporate income tax system, of sorts. All 25 countries with the exception of Croatia, FYR Macedonia and Slovenia had a VAT system in place by 1996. All TEs in our study have a social security tax system mostly financed by payroll taxes, with contributions being made by employers and/or employees. For all countries in our study, tax coverage is for general government, comprising of central and state, regional or provincial units of government and local government. An exercise in calculating statutory versus effective taxation depends on data pertaining to tax rates, tax takes and tax bases. As for statutory taxation, the basic tax rates are taken primarily from international tax handbooks. In particular, we used the IBFD’s European Tax Handbook, Coopers & Lybrand’s International Tax Summaries and Ernst & Young’s Worldwide Tax Guides. We also used various EBRD Transition Reports. Tax payments were obtained from the governments’ fiscal accounts where taxes are reported on a cash basis, i.e., counting actual receipts rather than accrued liabilities. Where possible, tax payments data are from the IMF’s Government Finance Statistics Yearbook (GFSY) or the CIS Statistical Yearbook. Other publications used include the OECD’s Revenue Statistics, the IMF’s Staff Country Reports and statistical yearbooks for various countries. As already mentioned, the tax bases for the three types of taxes are taken from the national accounts. For VAT, we use GDP as a proxy for the VAT base. Although in all VAT systems