OECD Tax Policy Studies No. 02: Tax Burdens: Alternative Measures

TAXATION OECD Tax Policy Studies

«

OECD Tax Policy Studies

Tax Burdens ALTERNATIVE MEASURES In response to growing demand by policy-makers, various measures to assess tax burdens of households, individual firms and the business sector as a whole have been developed. This study reviews some of the most common measures used to gauge tax burdens of households and corporations. In addition, it provides some illustrative numbers from various sources on tax rates and tax burdens in OECD countries. The study concludes that all current measures reviewed have at least some important shortcomings. Results based on these measures should therefore be interpreted with their limitations in mind, and judged with due caution when used to address policy questions.

Tax Burdens ALTERNATIVE MEASURES TAXATION

www.oecd.org

ISBN 92-64-17137-1 23 2000 02 1 P FF 150

No. 2

9:HSTCQE=V\VX\Y:

No. 2

 OECD, 2000.  Software: 1987-1996, Acrobat is a trademark of ADOBE. All rights reserved. OECD grants you the right to use one copy of this Program for your personal use only. Unauthorised reproduction, lending, hiring, transmission or distribution of any data or software is prohibited. You must treat the Program and associated materials and any elements thereof like any other copyrighted material. All requests should be made to: Head of Publications Service, OECD Publications Service, 2, rue Andr´e-Pascal, 75775 Paris Cedex 16, France.

OECD Tax Policy Studies No. 2

Tax Burdens: ALTERNATIVE MEASURES

ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT

ORGANISATION FOR ECONOMIC CO-OPERATION AND DEVELOPMENT Pursuant to Article 1 of the Convention signed in Paris on 14th December 1960, and which came into force on 30th September 1961, the Organisation for Economic Co-operation and Development (OECD) shall promote policies designed: – to achieve the highest sustainable economic growth and employment and a rising standard of living in Member countries, while maintaining financial stability, and thus to contribute to the development of the world economy; – to contribute to sound economic expansion in Member as well as non-member countries in the process of economic development; and – to contribute to the expansion of world trade on a multilateral, non-discriminatory basis in accordance with international obligations. The original Member countries of the OECD are Austria, Belgium, Canada, Denmark, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, Turkey, the United Kingdom and the United States. The following countries became Members subsequently through accession at the dates indicated hereafter: Japan (28th April 1964), Finland (28th January 1969), Australia (7th June 1971), New Zealand (29th May 1973), Mexico (18th May 1994), the Czech Republic (21st December 1995), Hungary (7th May 1996), Poland (22nd November 1996) and Korea (12th December 1996). The Commission of the European Communities takes part in the work of the OECD (Article 13 of the OECD Convention).

Publié en français sous le titre: Mesurer les charges fiscales QUELS INDICATEURS POUR DEMAIN ? N° 2

© OECD 2000 Permission to reproduce a portion of this work for non-commercial purposes or classroom use should be obtained through the Centre français d’exploitation du droit de copie (CFC), 20, rue des Grands-Augustins, 75006 Paris, France, Tel. (33-1) 44 07 47 70, Fax (33-1) 46 34 67 19, for every country except the United States. In the United States permission should be obtained through the Copyright Clearance Center, Customer Service, (508)750-8400, 222 Rosewood Drive, Danvers, MA 01923 USA, or CCC Online: http://www.copyright.com/. All other applications for permission to reproduce or translate all or part of this book should be made to OECD Publications, 2, rue André-Pascal, 75775 Paris Cedex 16, France.

FOREWORD This is the second issue in a new series of Tax Policy Studies launched by the OECD. The series aims to disseminate to a larger audience work undertaken by the OECD Fiscal Affairs Secretariat in the areas of tax policy and tax administration. Over the years, in response to growing demand by policy-makers, various measures to assess tax burdens have been developed. The present study reviews some of the most common measures used to gauge tax burdens of corporations and households. In addition, it provides some illustrative numbers from various sources on tax rates and tax burdens in OECD Member countries. The study concludes that all current measures reviewed have at least some important shortcomings. Results based on these measures should therefore be interpreted with their limitations in mind, and judged with due caution when used to address policy questions. Also, the study announces further work in this area to be undertaken by the Working Party on Tax Policy Analysis and Tax Statistics of the OECD Committee on Fiscal Affairs. This study was prepared by the Working Party on Tax Policy Analysis and Tax Statistics. The project was led by Flip de Kam and W. Steven Clark of the OECD Fiscal Affairs Secretariat. The study is published under the responsibility of the Secretary-General.

3

© OECD 2000

TABLE OF CONTENTS Chapter 1.

Measuring Tax Burdens ........................................................................................................................................ 7

Introduction.......................................................................................................................................................................... 7 Nominal tax rates................................................................................................................................................................. 7 Tax-to-GDP ratios ................................................................................................................................................................ 8 Average tax rates ................................................................................................................................................................. 9 Marginal effective tax rates .............................................................................................................................................. 10 Structure of the report ...................................................................................................................................................... 11 Chapter 2.

Nominal Income Tax Rates ................................................................................................................................ 13

Introduction........................................................................................................................................................................ 13 The structure of personal income tax rates.................................................................................................................... 13 “All-in” rates of taxes on personal income ..................................................................................................................... 15 High implicit tax rates for the poor ................................................................................................................................. 17 Nominal top rates by type of income ............................................................................................................................. 18 The structure of corporate income tax rates .................................................................................................................. 19 Corporate income tax and shareholder taxation ........................................................................................................... 22 Notes................................................................................................................................................................................... 25 Chapter 3.

Tax-to-GDP Ratios................................................................................................................................................ 27

Introduction........................................................................................................................................................................ 27 Tax expenditures versus direct expenditures ................................................................................................................. 28 Tax treatment of social security benefits........................................................................................................................ 29 The relationship between tax base and GDP, and economic cycle effects ............................................................... 30 Revisions to the measurement of GDP .......................................................................................................................... 31 Notes................................................................................................................................................................................... 32 Chapter 4.

Average Tax Rates and the Need for Micro-Data.......................................................................................... 33

Introduction........................................................................................................................................................................ 33 Two frameworks for assessing corporate tax burdens .................................................................................................. 34 Implicit ATRs ...................................................................................................................................................................... 35 Backward-looking (adjusted profit-based) ATRs........................................................................................................... 40 Notes................................................................................................................................................................................... 43 Chapter 5.

Marginal Effective Tax Rates............................................................................................................................. 47

Introduction........................................................................................................................................................................ 47 Defining marginal effective tax rates .............................................................................................................................. 48 Analysis of the no-tax case ............................................................................................................................................... 49 Analysis of a simple corporate income tax..................................................................................................................... 51 Analysis of targeted corporate tax instruments............................................................................................................. 53 Limitations of METR analysis........................................................................................................................................... 56 Notes................................................................................................................................................................................... 63

© OECD 2000

5

Tax Burdens: Alternative Measures

Chapter 6. Policy Relevance of Alternative Tax Burden Measures ................................................................................ 65 Introduction........................................................................................................................................................................ 65 Some illustrative results................................................................................................................................................... 65 Assessing the tax burden on “old” versus “new” capital.............................................................................................. 68 Assessing the corporate tax burden, from an equity perspective.............................................................................. 69 Assessing the impact of corporate taxation on investment incentives...................................................................... 71 Conclusion.......................................................................................................................................................................... 73 Notes................................................................................................................................................................................... 74

Annexes to Chapters 2.A. 2.B. 4.A. 4.B. 6.A. 6.B. 6.C.

Measuring the Overall Statutory Corporate Income Tax Rate in Japan..................................................................... 77 The effects of dividend taxation and integration relief .............................................................................................. 79 The OECD Classification of Taxes .................................................................................................................................. 81 Illustration of Application of Micro-Data....................................................................................................................... 83 Backward- and Forward-Looking Corporate Tax Rate Results ................................................................................... 87 Implicit Average Tax Rates – Illustration of the Effect of Treatment of Interest....................................................... 88 Statement on the Usefulness of METR Analysis for Tax Policy Purposes ................................................................. 89

References ................................................................................................................................................................................ 91

List of Boxes 1.A. 1.B. 1.C. 2.A. 2.B. 2.C. 2.D 3.A. 3.B. 3.C. 3.D

Why nominal rates may be deceptive............................................................................................................................. 8 When nominal rates matter .............................................................................................................................................. 8 Why tax-GDP ratios can be deceptive............................................................................................................................. 9 How OECD Countries Provide Basic Personal Tax Relief............................................................................................ 14 High implicit tax rates for the poor................................................................................................................................ 18 Franking Systems ............................................................................................................................................................. 23 Equalisation tax................................................................................................................................................................ 24 Tax expenditure, or not? ................................................................................................................................................. 28 How tax expenditures reduce the tax-to-GDP ratio .................................................................................................... 28 How the tax treatment of social benefits impacts on the tax-to-GDP ratio.............................................................. 29 How a revision of GDP can lower the tax-to-GDP ratio ............................................................................................... 31

List of Tables 2.1. Rate schedules of central government personal income tax (single person, no dependants), January 1998 ............ 15 3.1. Tax and spending ratios: an illustrative example (% of GDP)..................................................................................... 29

List of Charts

6

2.1. 2.2. 2.3. 5.1. 5.2. 5.3. 5.4. 6.1. 6.2.

Highest Tax Rates in Wage Income, 1998 ...................................................................................................................... 17 Highest “all-in” Tax Rates, 1998...................................................................................................................................... 19 Statutory Corporate Income Tax Rates (January 1, 1998) ............................................................................................ 20 Return on investment on the no-tax case ..................................................................................................................... 50 Return on investment with a simple corporate income tax........................................................................................ 52 Return on investment in the net capital import case.................................................................................................. 55 Return on investment in the net capital export case .................................................................................................. 55 Backward-looking Corporate Tax Rate Measures, 1995............................................................................................... 66 Forward-looking Corporate Tax Rate Measures, 1998 ................................................................................................. 67

© OECD 2000

Chapter 1

MEASURING TAX BURDENS Introduction In all OECD Member countries, the rapid integration of national economies forces policy-makers to critically re-examine existing tax systems. It is in the context of this policy debate that policymakers and others seek to better grasp the impact of taxation on domestic investment, employment growth and overall economic performance. Policy-makers are concerned that present tax systems may discourage economic activity and destroy jobs, but face limitations on the amount of tax relief they can deliver, given the need to finance government expenditure. Thus, efficiency, competitiveness and revenue concerns are at stake. At the same time, increased difficulty in taxing income from capital resulting from the increasing mobility of capital may create pressures for a gradual shifting of the tax burden away from capital onto labour and consumption, raising equity and efficiency concerns. In response to growing demand, policy-analysts have developed various measures to assess tax burdens and the impact of taxes on economic activity. This study reviews the most common measures to gauge tax burdens: i) nominal tax rates (Section 1.2); ii) tax-to-GDP ratios (Section 1.3); iii) average tax rates (Section 1.4); iv) marginal effective tax rates (Section 1.5). The study is factual and comparative. It does not offer judgement on specific tax systems or any particular tax reform proposals. In addition, the study provides some illustrative numbers on impacts of tax systems of OECD Member countries, drawing from recent academic research and supplemented by OECD Secretariat data and calculations. In the latter case, national authorities have provided the data on macroeconomic aggregates and on details of national tax systems used to prepare the tables concerned. Nominal tax rates The most basic and often-cited measures of tax burdens are nominal or “statutory” tax rates. The OECD annually updates its Tax Database containing detailed information on personal and corporate income tax systems of the twenty-nine Member countries, including data on the nominal rates of income taxes. Chapter 2 of this study discusses rates of taxes on personal and corporate income. In 1998, in OECD economies combined top nominal rates of all taxes on personal income ranged from 33 per cent (New Zealand) to 66 per cent (Belgium). Chapter 2 also documents the spread in nominal rates of the corporate income tax, ranging from 28 per cent (Finland and Sweden) to 57 per cent (Germany). Such nominal rates are relevant, because they have an important signal function, partly determine the value of tax concessions and are commonly an important factor in decision making on new investment. As many would be quick to point out, however, nominal tax rates tell an incomplete story, because effective tax rates will usually be lower than nominal rates suggest. The reason being that household income and corporate profits determined in accordance with standard accounting practices may be reduced by specific provisions in the tax legislation before nominal rates are applied to a significantly © OECD 2000

7

Tax Burdens: Alternative Measures

smaller tax base (taxable income, profits). For example, under most tax systems individuals can defer taxation on part of their income set aside for old age, and in many cases imputed rent of home-owners goes untaxed while interest expenses are partly and – in rather exceptional cases – even fully deductible. Similarly, corporate profits determined following standard accounting practices may be reduced by specific provisions in the tax legislation, such as generous depreciation schemes and tax incentives to promote investment aimed at R&D or particular regions, reserve provisions for tax purposes and because of international tax planning (see Box 1.A).

Box 1.A.

Why nominal rates may be deceptive

Assume accounting profit amounts to 100 units and the nominal corporate tax rate is 35 per cent. If taxable profit of this firm is only 60 units, the corporate tax bill of 21 units reflects the nominal rate of 35 per cent applied to taxable profit of 60. The effective average tax rate in this case is much lower, i.e., 21 per cent: 21 units paid in tax over 100 in accounting profits.

Clearly, nominal tax rates will more closely indicate effective tax burdens where taxpayers have limited opportunities to reduce the tax base below earned income and book profits using such specific tax reliefs. It follows that the effective tax burden of households or corporations – on an individual basis or for households or corporations as a group – can only be measured by expressing taxes actually paid as a percentage of some adjusted measure of income or profits.* Nominal rates as applied to taxable income or profit are relevant to tax planning incentives, since they generally determine the value of tax deductions – e.g., deduction of interest payments (see Box 1.B) – and income inclusions – relevant to transfer pricing incentives and thin capitalisation.

Box 1.B.

When nominal rates matter

If the nominal corporate tax rate in a country is 35 per cent, the value of deducting one additional currency unit of interest is 0.35 currency units. In the context of a multinational corporation considering where to book additional interest expense for tax purposes, the relevant comparison between jurisdiction A and B is thus a comparison of nominal corporate income tax rates in both jurisdictions.

Tax-to-GDP ratios A second way to approach the tax burden of the household or corporate sector of the economy is to express total revenue from taxes paid by that sector as a percentage of Gross Domestic Product (GDP). As discussed in Chapter 3 of this study, tax-to-GDP ratios must be interpreted with great caution.

8

* This alternative approach produces average tax rates, to be discussed in greater detail in Chapter 4. As noted in Section 4.3, certain adjustments to accounting or book profit are required for an accurate measure of the corporate tax burden (e.g., adjustments for inflation, losses, foreign profits).

© OECD 2000

Measuring Tax Burdens

Indeed, ratios of aggregate tax over GDP provide only limited information on the tax burden of the household or corporate sector, as may be demonstrated by a closer analysis of the ratio of corporate income tax (CIT) to GDP. First, such CIT-to-GDP ratios mask changes in corporate tax as a percentage of corporate profit. To see this, note that corporate tax relative to GDP is determined by the product of two ratios: 1. corporate tax divided by pre-tax corporate profit; and 2. pre-tax corporate profit as a share of GDP. The first ratio, which is an average corporate income tax measure, will vary with changes in the nominal corporate tax rate and with changes in the corporate tax base. Changes in this ratio therefore reflect primarily changes in tax policy, the efficiency of the tax administration, compliance and tax planning – the way corporations respond to existing legal provisions. The second ratio, pre-tax corporate profits relative to GDP, will vary with fluctuations in the share of corporate profits in aggregate value-added in the economy. Holding tax policy constant – i.e., assuming the rules determining corporate tax rates and the tax base are fixed – and assuming unchanged tax practice (administration, compliance), a drop in corporate profit over GDP would cause the corporate tax-to-GDP ratio to decline. This outcome could be mistakenly interpreted as indicating a reduction in corporate tax on corporate profits, when in fact this value is unchanged (see example in Box 1.C).

Box 1.C.

Why tax-GDP ratios can be deceptive

Assume the ratio of corporate income tax revenues to the value of GDP in countries A and B is 2.5 per cent and 5 per cent, respectively. This suggests that tax burdens are higher in country B. However, this need not be the case. If corporate earnings in countries A and B constitute 10 per cent and 20 per cent of GDP, respectively, the average effective tax rate for the corporate sector is 25 per cent in both cases since in country A corporations pay 2.5 per cent of GDP on their profits equal to 10 per cent of GDP, and corporations in country B pay relatively twice as much in taxes on profits that in relative terms are twice as high as in country A.

A second limitation of CIT-to-GDP ratios is that they include only one of several taxes corporations pay out of earnings. In a number of countries, other taxes paid by corporations may be important. In particular, taxes on capital, taxes on financial transactions and payroll taxes can be significant. A third problem is that such ratios include negative profits (i.e., losses) in the denominator (GDP), which should be excluded when measuring the effective average tax rate on profitable firms. Finally, CIT-to-GDP ratios may include corporate-level taxes on distributed earnings (i.e., equalisation taxes), which many would argue represent pre-payments of personal level taxes shareholders must pay and should therefore not be included in the measurement of corporate tax burdens. Average tax rates Average tax rates (ATR) are a third way to assess tax burdens of households or corporations. Taking recourse to ATR measures partially overcomes the limitations associated with a comparison of nominal rates and tax-to-GDP ratios. Their main advantage, relative to nominal tax rates, is that they take into account actual taxes paid. They also use a more narrowly defined (targeted) measure of tax base in comparison to tax-to-GDP ratios. © OECD 2000

9

Tax Burdens: Alternative Measures

To calculate an average tax rate for an individual household, all relevant taxes paid are divided by a measure of household income, defined as the sum of consumption plus the change in net wealth of that particular household during a given time period (i.e., an economic definition of income) or, alternatively, measured following some legal definition of income. ATRs are mainly used in practice to analyse tax burdens on corporations, or the business sector as a whole (including unincorporated business) where, as noted, included in the denominator is a more narrow measure of aggregate business surplus than is total GDP. Average corporate tax rates take into account the impact of special relief (e.g., investment tax allowances, credits, enhanced depreciation provisions, donations to reserves to cover future risks) on marginal and infra-marginal investment, taxplanning and other factors determining final tax liabilities. Thus, the limitations inherent to tax burden measures such as nominal rates and tax-to-GDP ratios are partly lifted. Important differences may be observed, however, in the choice of the relevant tax base (denominator of the average tax ratio). To calculate an average tax rate for a corporation, the standard approach is to relate total taxes actually paid out of corporate earnings to an adjusted measure of corporate financial profit, with adjustments to corporate financial (book) profit made to arrive at a measure of true (economic) income. To calculate an average rate for all corporations, total taxes actually paid out of corporate earnings are related, under the so-called “implicit” tax rate approach, to the “operating surplus” of the economy. The operating surplus of the corporate sector is a measure of domestic value-added at the corporate level accruing to suppliers of capital [gross output at producer’s prices less the sum of intermediate and fixed capital consumption, wage costs (including employer social security contributions), and indirect taxes net of subsidies]. Viewed another way, corporate operating surplus captures domestic source income of capital suppliers generated at the corporate level and realised in the form of interest, rents, royalties, dividends and retained earnings. National Accounts also report separately (for most countries) the operating surplus of unincorporated business. Total operating surplus of the economy will differ from aggregate commercial and true economic profits of corporations and unincorporated enterprises on account of a number of factors. One of the main differences is the inclusion in operating surplus of interest income paid (directly or indirectly) to households (savers). This renders implicit corporate tax rates of questionable relevance, and suggests that the implicit tax rate approach be limited to analyses of average tax rates on income from capital (which factor returns to debt and equity capital in the denominator, with both corporate and personal tax on these amounts included in the numerator). The implication is that more conventional average corporate tax rate measures relating corporate income taxes paid to corporate profit are called for. However, as discussed in Chapter 4, profit-based average tax rates derived from aggregate or firmlevel data may suffer from a number of shortcomings, for example the inclusion in the denominator of negative profits of loss-making firms, and the inclusion in the numerator of net domestic tax on foreign source income (excluded from the denominator). Ideally, detailed micro-data should be used to enable a proper match of taxes actually paid on adjusted corporate profit at the firm level for a representative sample of firms. Adjustments to corporate book profit figures are necessary to improve consistency between numerator and denominator amounts, and to move towards a true definition of economic income. This micro-data set would allow one to determine measurement errors in average tax rates derived using aggregate data, and to establish average tax rates by sector, by firm size (small and medium-sized enterprises versus multinational corporations), and so on. The OECD is currently doing some work in this area but it will take time to produce relevant results, part of the problem being that many countries are presently not able to provide the required data sets for a representative sample of firms. Marginal effective tax rates

10

Yet another way to analyse tax burdens is to calculate the “wedge” that taxes drive between pre-tax and after-tax rates of return at the margin, that is, on the last currency unit invested where the marginal benefit of the investment just covers it marginal cost. The resulting marginal effective tax rates (METRs) can be calculated assuming a specific type or mix of investments (buildings, machinery, inventories), a © OECD 2000

Measuring Tax Burdens

specific or mixed source of finance (retained earnings, new equity, debt), and with reference to historic rates of inflation. In sum, METRs are computed by deriving, for a representative group of investors and firms, the wedge that taxation creates between pre-tax and after-tax rates of return at the margin, taking into account nominal tax rates and basic rules determining tax deductions and tax credits. In theory, METRs measure the impact of taxation on required rates of return and thus investment incentives at the margin. METRs may be compared across investment projects, investor groups, methods of finance, and across countries. In 1991, the OECD published a study Taxing Profits in a Global Economy that included comprehensive work in this area. As stressed in Chapter 5, results of METR analysis – a highly theoretical construct – must be interpreted with due caution, bearing in mind the simplifying assumptions behind the neo-classical theory of investment upon which the methodology is based. Also, when calculating METRs only a part of the tax system is taken into account, for example possibilities to form reserves and common tax-planning techniques are neglected. And finally the METR calculations assume that investors pay tax according to the nominal rates. It was already explained that taxpayers often pay lower effective rates, largely for reasons not included in METR calculations. Structure of the report The structure of the report is as follows. Chapter 2 begins with nominal or “headline” statutory tax rates, which are the simplest tax burden measures, yet convey limited information about the overall impact of tax systems. Tax rate structures are reviewed, and figures are given for “all-in” personal and corporate statutory income tax rates in OECD countries. Integration considerations are also briefly addressed. Chapter 3 is devoted to a consideration of tax-to-GDP ratios, which are widely quoted in the popular press and used to make cross-country tax burden comparisons. Yet as the text explains, these measures are potentially misleading indicators, and the various sources of bias must be remembered when using such statistics to analyse the impact of taxation over time and across countries. Chapter 4 considers a number of “backward-looking” average tax rate measures, which may be constructed at the aggregate economy, industry, or firm level. The analysis of so-called implicit corporate tax rates, derived using National Accounts data, finds these measures to be highly imprecise indicators of tax burden. A preference is shown for average tax rate measures derived using economic profit as the relevant denominator (tax base), with a call for the need for micro- (firm-level) data to enable warranted tax base adjustments. Work in this area, underway by the Working Party No. 2 in collaboration with OECD countries, will serve to illustrate the varying importance of the adjustments across countries and thereby assist in making cross-country comparisons of tax levels. Chapter 5 then turns to consider “forward-looking” tax burden measures, which include average tax rates derived for hypothetical (discrete) investment projects, which many would argue are most relevant to the analysis of corporate locational decisions. A special focus is given in Chapter 5 to marginal effective tax rate analysis given the now widespread use of this framework, with the objective of laying out clearly its basic underpinnings and limitations for tax policy analysis purposes. Chapter 6 concludes with a discussion of the usefulness of these various measures in addressing common tax policy questions involving equity considerations (is the corporate sector paying its fair share?) and efficiency concerns with tax systems (is taxation discouraging corporate investment?) which are increasingly confronting policy makers. The final sub-section briefly concludes and points towards further work in this area by the Working Party on Tax Policy Analysis and Tax Statistics of the OECD Committee on Fiscal Affairs.

11

© OECD 2000

Chapter 2

NOMINAL INCOME TAX RATES Introduction Comparisons of different tax systems tend to begin by looking at nominal top rates of income taxes imposed by central government. However, concentrating on these “headline” rates can lead to misleading conclusions. This chapter first compares the structure of personal income tax rates in 1998, using information from the OECD Tax Database. The focus is on top marginal rates: the highest percentage of tax levied on one additional dollar, yen or franc of taxable income (Section 2.2). Most studies compare only the top rates of personal income tax imposed by central government. However, in twenty-two out of the twenty-nine OECD Member countries sub-central governments also levy taxes on personal income. Clearly, these should be taken into account to get the full picture of top rates on income (Section 2.3). Perhaps most surprisingly, it is in many cases not the rich who at the margin – that is, on one additional unit of income) are exposed to the highest rates (Section 2.4). In Section 2.5 it is shown that, within a given country, nominal top rates may vary significantly, depending on the type of income.1 The discussion then turns to a consideration of nominal corporate income tax rates. The statutory corporate tax rate structure is shown to exhibit a considerable degree of diversity across OECD countries (Section 2.6). Rate schedules may be flat or graduated and may include a temporary or permanent surcharge/surtax. In certain federal countries, an additional layer of taxation may be found at the state, provincial or cantonal level. Moreover, tax rates may differ depending on the type of business income, whether income is retained or distributed to shareholders, and even according to the capitalisation of the firm, although this distinction is rare. This chapter also examines the so-called “integration” problem that arises with the co-existence of both corporate- and shareholder-level income taxation (Section 2.7). Various techniques may be used to avoid the double taxation that otherwise occurs when income that is subject to tax at the corporate level is subject to tax again at the personal level when distributed to individual shareholders as dividends. This consideration has implications for the proper measurement of the tax burden on income from capital. If, however, imputation credits are recognised as providing relief to shareholder level taxation, they may be ignored in assessing corporate tax burdens, which for the most part are the focus of this study. The structure of personal income tax rates The fundamental structure of personal income taxes imposed by federal or central governments is highly similar in every OECD country. A certain amount of income may be exempted from tax; this is the personal exemption. In some countries, the first slice of income is not exempted, but instead taxed at zero per cent, to the same effect (zero band). The alternative is that all income is taxed, but that taxpayers are entitled to a reduction of their tax bill by means of a basic tax credit. To gauge how much these basic tax reliefs may reduce the tax bill, they can be expressed as a percentage of the gross wage of an average production worker in each OECD country. On this measure, Greece exempts only 3 per cent of the average worker’s wage, Korea 7 per cent, the Netherlands 14 per cent, France 20 per cent, the United Kingdom and the United States 24 per cent. In Sweden, an average production worker pays no income tax to the central government, because the tax allowance exceeds his © OECD 2000

13

Tax Burdens: Alternative Measures

wage by ten per cent. The associated tax reduction for individual taxpayers is determined by applying their marginal tax rate to the amount of the personal exemption or zero band (see Box 2.A). With the exception of Germany, which applies several tax formulae, income in excess of the personal exemption or zero-rated band is sliced into “brackets”. The number of brackets ranges from one (Sweden) or two (Iceland and Ireland) to eight or more (Luxembourg, Mexico, Spain and Switzerland). All of the taxable income within a bracket is taxed at the same rate. The rate applied to the income in successive brackets increases. The result is a progressive tax: as total taxable income rises, a growing share thereof is – at least in principle – paid in tax. Under a progressive income tax, the value to taxpayers of the personal exemption or a zero-rated first band increases as they move into higher taxed brackets. In contrast, the value of tax credits is independent of the taxpayers’ income level (see Box 2.A). In 1998, nine OECD countries used basic tax credits: Austria, Canada, Hungary, Iceland, Italy, Mexico, New Zealand, Poland and Portugal.

Box 2.A.

How OECD Countries Provide Basic Personal Tax Relief

Basic relief for income-taxpayers can be structured in different ways. Suppose that countries A and B have an identical rate structure consisting of four brackets. The first 20 000 units of income are taxed at 20 per cent, the next 20 000 units at 30 per cent, the next 20 000 units at 40 per cent and any income over 60 000 at the top rate of 50 per cent. In country A, taxpayers are entitled to a personal exemption of 10 000 (before the rates are applied). The tax bill of low-income earners in the first bracket is thus reduced by 2 000, since they save 20 per cent of 10 000 (their exemption) in tax. The tax bill for those in the highest taxed bracket is slashed by 5 000, since they forego 50 per cent of 10 000 in tax. The example serves to illustrate how tax relief increases with income and is determined by the marginal rate applicable to the last 10 000 units of income. In country B, all taxpayers can claim a credit of 3 000 against income tax due. Here, the value of the tax relief is the same for each individual, regardless of the level of income. If the credit is “non-wastable”, the tax collector hands out the excess of the credit over tax due. The result is a negative income tax.

How progressive a given rate schedule is depends not only on the amount of basic tax relief, but also on the width of the respective brackets, and the marginal rates applied to the income in each bracket. Income taxes imposed by central governments exhibit remarkable variety in bracket lengths and marginal rates, reflecting national views on what – given existing revenue needs – constitutes an equitable distribution of the tax burden. Income in the first bracket of personal income tax schedules is taxed at a low 0.8 per cent in Switzerland, at 25.8 per cent in Belgium, and at 27.4 per cent in Iceland. Top marginal rates of personal income tax levied by central government range from 25 per cent (Sweden) and 33 per cent (New Zealand) to as much as 60 per cent (the Netherlands). In Ireland and New Zealand, taxpayers at the income level of an average production worker are already exposed to the top marginal rate of 48 per cent and 33 per cent, respectively. In Austria, Belgium, Canada, Finland, France, Germany, the Netherlands and the United Kingdom workers must earn about twice the average before they start paying the top rate. On the other hand, Swiss and American employees are only confronted by the top rate if their salaries exceed ten times the average production worker’s wage. In Turkey the top rate does not apply until taxable income is twenty-nine times the average wage and more. Table 2.1 summarises the rate structure of the personal income tax levied by central governments in all OECD countries.

14

Before drawing any firm conclusions from this panoply of income tax schedules, three points should be borne in mind. First, the actual tax bill of individual taxpayers also reflects the impact of various deductions – for example, for mortgage interest and employee contributions to occupational pension plans – and exemptions – for example, for capital gains or interest received. This means that effective tax © OECD 2000

Nominal Income Tax Rates

Table 2.1.

Australia Austria Belgium Canada Czech Rep. Denmark Finland France Germany Greece Hungary Iceland Ireland Italy Japan Korea Luxembourg Mexico Netherlands New Zealand Norway Poland Portugal Spain Sweden Switzerland Turkey UK USA

Rate schedules of central government personal income tax (single person, no dependants), January 1998a Type of basic tax relief

Tax relief as proportion of APWb

PE/ZR TC PE/ZR TC PE/ZR PE/ZR PE/ZR PE/ZR PE/ZR PE/ZR TC TC PE/ZR TC PE/ZR PE/ZR PE/ZR TC PE/ZR n.a. PE/ZR TC TC PE/ZR PE/ZR PE/ZR PE/ZR PE/ZR PE/ZR

0.15 0.03 0.19 0.03 0.23 0.12 0.33 0.20 0.21 0.03 0.09 0.18 0.20 0.09 0.07 0.25

0.02c

0.08c

0.14 0.00 0.13 0.03 0.03 0.21 1.10 0.20 0.13 0.24 0.24

Lowest standard rate

20 10 25.75 17.51 15 8 6 10.5 formula 5 20 29.31 26 19 10 10 61 3 8.85 15 18.8 19 15 17 25 0.77 25 20 15

Number of tax brackets

4 5 7 4 5 3 6 6 4 4 6 2 2 5 5 4 7 8 3 3 3 3 4 8 1 10 7 3 5

Highest standard rate

47 50 56.65 31.3 40 29 38 54 53 40 42 34.31 48 46 50 30 46 35 60 33 32.5 40 40 47.6 25 11.5 55 40 39.6

Highest rate starts at (proportion of APW wage)

1.4 2.3 2.2 1.8 5.9 1.1 2.2 2.2 2.1 2.5 2.0 1.8 0.7 3.5 7 5.5 2.4 7.5 1.9 1 1.1 4.7 4.5 4.6 1.1 10.4 28.5 1.8 9.7

n.a. = not applicable. APW = average production worker. PE/ZR = personal exemption or zero rate band. TC = tax credit. a) Deductions or allowances related to specific income sources are not included. b) Figures in Columns PE/ZR and TC are not directly comparable (see Box A). c) The tax credit is a decreasing function of personal income. This percentage considers the level of the tax credit which corresponds to the APW income.

burdens in countries with lower nominal rates but little in the way of basic relief, deductions and exemptions, could well be higher than effective rates in countries which combine higher nominal rates with more generous exemptions and deductions. The second point is one made at the start of this Chapter, namely that in most OECD countries there are other taxes to pay on income, beyond that owed to central government. Finally, national tax systems are often characterised by minor peculiarities, which while perhaps complicating matters slightly, do not have a major bearing on the general picture outlined here. “All-in” rates of taxes on personal income For several reasons, households in most OECD countries are often confronted by higher marginal rates than is suggested by regularly cited headline or “standard” rates of the personal income tax as shown in Table 2.1. Central governments sometimes impose temporary increases on the income tax, such as the “austerity surcharge” in Belgium and the German “solidarity surcharge”. Such surcharges jack up the total income tax bill. Also, households may have to pay local, regional, provincial or state income taxes on top of the central government income tax. This is the case in Belgium, Canada, Iceland, Japan, Korea, the Nordic countries, Spain, Switzerland and the United States. Within a given country, rates of income taxes levied by © OECD 2000

15

Tax Burdens: Alternative Measures

local and state government often show substantial variety. The OECD Tax Data Base reports both the highest and the lowest state/provincial and local rates found anywhere in a given country, and also a “typical”, average value. This average rate of income taxes levied by sub-central levels of government is identical to that reported in the annual OECD publication on the Tax/Benefit Position of Employees.2 Here, these average rates are used. In a few OECD countries income taxes imposed by sub-central levels of government are quite significant. To illustrate, in Sweden the typical top rate of provincial and local income taxes is 31.7 per cent, which means it exceeds the 25 per cent income tax rate levied by central government. In some countries, state, regional or local income taxes paid constitute a deductible item, inasmuch as the amount paid in such taxes may be deducted when calculating taxable income for the central or federal government income tax. For the presentation of “all-in” tax rates in Chart 2.1, the deductibility of noncentral government income taxes of course has been taken into account. Another feature to note, particularly in Europe, is the tax some national governments impose on income on behalf of the state church. This church tax features in the tax system of Austria, Germany, the Nordic countries3 and Switzerland. In Table 2.1, the church tax is included only in the cases of Denmark and Switzerland. Some will question whether the church tax really is a “tax” as defined by OECD and other international organisations, in the sense of being a compulsory, unrequited payment to general government. Here, “unrequited” means that benefits provided by government to taxpayers are not normally in proportion to their payments. In fact, this should be decided on a case-by-case basis, taking into account specific institutions existing in the countries concerned. Likewise, it is sometimes difficult to define the status of social security contributions – are they really taxes or are they payments for some form of social protection? In part, the answer will depend on the degree to which these payments are directly linked to the value of the benefits they offer. Social security programmes come in two basic forms. In some the revenue is earmarked to finance programmes that essentially cover the whole population. Here, the tax base may be identical to – or closely resemble – that for the personal income tax. However, in contrast to the rate structure of the income tax, a ceiling or “cap” often applies and income above that ceiling is not subject to further contributions. In addition to programmes covering the whole population, most European countries run social insurance programmes which only protect workers, or at least important sections of them. The tax base to finance such employee social insurance is wage income, usually up to a ceiling, which in turn is related to the maximum amount of wages insured against the risks of unemployment and disability. Furthermore, in a few instances such payments can be made into individual accounts such as pension plans, in which cases the relatively strong tie between contributions and benefits makes them even less “tax-like”. Of course, when considering top marginal rates, social security contributions are only relevant if they are not “capped”. Contributions – where they are imposed – are usually deductible for personal income tax purposes, though this is not the case, for example, in Hungary, Norway and the United Kingdom. When calculating marginal “all-in” rates, the OECD Secretariat takes into account the deductibility of social security contributions – where applicable. Since the income tax and social contributions are quite similar in terms of the tax base they use and of their economic impacts, both are included in the calculations underlying Chart 2.1 which compares the standard top rate of central government personal income tax with “all-in” top rates. These include the combined effect of temporary increases of the central government income tax, income taxes levied at local, regional and state levels, the church tax4 and employee social security contributions. Not included here are social security contributions directly paid by employers, even though these may eventually be borne by labour through tighter wage deals. Conversely, when labour is much in demand, employees may be able to shift part of their income taxes and employee contributions back onto employers by successfully demanding additional wage increases. That said, this chapter – in only considering nominal rates of taxes, which the law requires employees to pay – neglects any shifting of taxes on wages. 16

One lesson from the previous analysis is that gaps at the margin between top income earners domiciled in various OECD countries are narrower than often imagined and certainly not as wide as the head© OECD 2000

Nominal Income Tax Rates

Highest Tax Rates in Wage Income, 19981

Chart 2.1.

“All-in” rate

Standard rate

20

20

10

10

0

0

ry

re

ga

Ko

un H

l ga

at

ni

te

d

St

ur

rtu

Po U

lic

bo

ub xe

m

ep R

ch ze C

Lu

d

ra

an

st Au

ay

el Ic

w

la

or N

ly Ire

la er

itz

Sw

Ita

a

y

ad

an

an C

ai

m

Sp

er G

en ed

ke

Sw

ce

Tu r

an Fr

k

an

Fi

nl

ar

n

m en

D

iu

pa Ja

lg Be

a

30

es

30

g

40

lia

40

nd

50

nd

50

n

60

y

60

d

% 70

m

% 70

1.

The chart shows only those (22) OECD countries where the “all in” rate of taxes on wage income differs from the highest standard rate of personal income tax imposed by central government. From left to right, countries are ranked by decreasing “all-in” rate. Source: OECD.

line rates of the personal income tax imposed by central governments seem to suggest. In fact, after including all taxes on personal income, the marginal top rate in most countries rises substantially, exceeding 60 per cent in France and Turkey (both 61 per cent), Denmark and Sweden (62 per cent), Japan (65 per cent)5 and Belgium (66 per cent). The highest all-in rates for taxpayers in the United States fall in the 40-48 per cent range, depending on the State where they are resident. That puts the gap with their counterparts in Sweden, which most people would see as the quintessential welfare state, in some cases at as low as nine percentage points. But it is important to point out that taxpayers in Sweden and most other OECD countries are already confronted with top rates at much lower income levels than are taxpayers in the United States. High implicit tax rates for the poor It is not necessarily individuals in the highest income groups who always pay the highest marginal rates. In some countries, the rate structure of the personal income tax may show one or more “humps”. In such cases, individuals in the low- or middle-income range are exposed to higher marginal rates than are the very rich. For example, “all-in” rate humps can arise when the combined marginal rate of income tax and capped social security contributions exceeds the income tax rate applicable to income above the ceiling set for those contributions. Under exceptional circumstances, the standard personal income tax rate may also feature a “camel back” structure. In the second half of the 1980s, the US federal income tax had such a rate structure for a few years. At the time, income in the first bracket was taxed at 15 per cent, income in the top bracket at 28 per cent. As a result, tax relief for high-income earners was almost twice the tax relief for low-income earners. To recoup the higher tax relief for well-off taxpayers, lawmakers enacted a 33 per cent bracket that was sandwiched between the first and the top bracket. The rates of the federal income tax in Switzerland still show a hump. © OECD 2000

17

Tax Burdens: Alternative Measures

Rather surprisingly, low-income earners may also be confronted with very high marginal rates, in exceptional cases even exceeding 100 per cent. The reason is that the poorly paid not only contribute more tax when their income goes up, but in many cases also lose part of their means-tested tax relief, subsidies and benefits, which has the effect of an “implicit” tax (see Box 2.B). Workers with an implicit marginal tax rate over 100 per cent are only rational if they reduce the number of hours worked. Their gross wage will now of course be lower, but they pay less tax and receive higher tax relief and benefits. As a result, their net disposable income may increase while they put in less hours.

Box 2.B.

High implicit tax rates for the poor

The following example illustrates how high implicit tax rates may go. If the head of a one-earner couple with two young children finds a low-paid job after five years of unemployment, net income in and out of work is the same in Finland and Sweden. In other words, the implicit tax rate is 100 per cent, because each unit of income earned is washed out by a unit of benefits lost. In the case of Denmark and the Czech Republic, the implicit rate is almost 100 per cent; in Germany and the United Kingdom it is about 80 per cent, in France and the United States it is about 50 per cent. Once in employment, low-wage earners may be exposed to similar implicit tax rates, as means-tested benefits are cut further with every (small) wage increase. Source:

OECD, Benefit Systems and Work Incentives, 1998 edition, p. 35

Nominal top rates by type of income As Chart 2.2 demonstrates, the “all-in” top rates of taxes on personal income may also vary by type of income. Generally speaking, labour income is more heavily taxed, especially so if it is subject to contributions earmarked to finance employee social insurance. On the other hand, over the past fifteen years a growing number of OECD countries introduced lower, flat rates for certain types of capital income, notably interest and dividend, e.g. Belgium, the Czech Republic, Greece, Hungary, Italy, Poland and the Nordic countries. The introduction of lower flat rates for capital income is often interpreted as a response to growing pressures from tax competition between nations. Financial capital, being highly mobile, tends to flow to those jurisdictions where it is taxed at the lowest rates. To address capital flight, tax policy-makers may decide to reduce the domestic tax burden on capital income. Flat rates for capital income can reduce the overall progressivity of the income tax and they may compromise the redistributive impact of the tax. In some countries, the moves were also part of a more general strategy designed to lower the efficiency costs of taxation by reducing rates, while at the same time taxing larger shares of the capital income tax base. Lastly, it is important to note that many high-income earners escape paying the top statutory rates shown in Chart 2.1. First, because in a number of OECD countries capital income – often an important income component of the well-off – is not taxed at progressive rates but subject to lower, flat rates (see Chart 2.2). Second, the self-employed tend to be over-represented among higher-income earners. Compared to other groups of taxpayers, the self-employed may be in a better position to limit their tax obligations, legally (by using tax reliefs for business) and illegally (by under-reporting income).

18

More generally, tax planning often takes the bite out of high rates. For example, in countries where the corporation income tax rate is substantially lower than the top personal income tax rate, the selfemployed may have a strong incentive to incorporate their business and pay themselves a limited director’s fee. To illustrate this point, in the Netherlands the gap between both rates is 25 percentage points © OECD 2000

Nominal Income Tax Rates

Highest “all-in” Tax Rates, 19981

Chart 2.2. Wage income

Interest from bank deposits

Dividend income

20

20

10

10

0

0

y

at

ke

St d

te ni U

la

Tu r

en Sw

itz

er

ed

Sw

ga

nd

rtu Po

la Po

o N

or

w

ic

a re

ex M

pa

Ko

Ita

Ja

d

la Ire

ry

an el

Ic

H

un

ga

ec

e

ce

re G

an Fr

k

an

ar

nl

m

ub

en D

Fi

m iu

ep

C

ze

ch

R

st

lg Be

Au

es

30

nd

30

l

40

ay

40

n

50

ly

50

nd

60

d

60

lic

% 70

ria

% 70

1.

The chart shows only those (21) OECD countries where the “all in” rate on wage income differs from the “all-in” rate on dividend or/and interest income. Source: OECD.

(35 instead of 60 per cent). In countries that do not tax capital gains, taxpayers will be advised to transform taxable capital income into tax-exempt capital gains. As a result of successful tax planning, highincome earners often see their taxable income shrink and tax bills reduced. The structure of corporate income tax rates For a number of reasons, OECD countries tax corporate profits at the corporate level, and not just at the personal level when profits are distributed to individual shareholders. One of the main reasons for this legal structure is to counteract tax deferral possibilities that would otherwise exist, an issue considered separately below. Chart 2.3 sets out the basic statutory corporate income tax rates applicable in OECD countries as of January 1, 1998. In taxing corporate profits, a number of approaches may be observed. First, differences exist in the determination of taxable income, as in the case of the taxation of business and investment income at the personal level. Depreciable assets (e.g., machinery, buildings) used to earn business income may be written-off or depreciated for tax purposes on a straight-line or declining-balance basis at rates exceeding rates used for financial accounting purposes (or more generally at rates exceeding physical depreciation), in order to encourage investment activity. Interest expense on debts may be deductible in full or only in part, subject to limits under thin-capitalisation rules.6 Furthermore, dividends received from resident corporations may be deductible, on the basis that corporate tax has been paid at source by the distributing company, or partial inclusion may be required. Foreign dividends may be exempt or taxed with a foreign tax credit, depending on whether the dividends represent a portfolio or significant (direct) equity interest, and whether the country of residence of the taxpayer operates a territorial (source) based tax system or not. International treaties to eliminate or limit “double taxation” will also impact on the tax finally paid. © OECD 2000

19

Tax Burdens: Alternative Measures

Chart 2.3.

Statutory Corporate Income Tax Rates (January 1, 1998) Per cent 36.0

Australia

34.0

Austria

40.2

Belgium

44.6

Canada 35.0

Czech Republic

34.0

Denmark 28.0

Finland

41.7

France

56.7

Germany 40.0

Greece Hungary

19.2 38.0

Iceland 32.0

Ireland Italy

37.0 46.4

Japan 30.8

Korea Luxembourg

37.5

Mexico

34.0

Netherlands

35.0

New Zealand Norway

33.0 28.0

Poland

36.0

Portugal

37.4

Spain

35.0

Sweden

28.0

Switzerland

27.8

Turkey United Kingdom United States

44.0 31.0 40.0

Notes: Chart shows “all-in” basic nominal corporate income tax rates, including central and sub-central taxes and surtaxes. Special targeted (as opposed to basic) corporate tax rates, applicable in certain countries to small business income (e.g., Canada, United Kingdom), manufacturing income (e.g. Ireland), financial services income, etc. are not shown (see main text). The German rate shown is that applicable to retentions, while the rate shown for Italy excludes reductions in the basic rate resulting from the dual income tax and also excludes the regional (expenditurebased) production tax IRAP (see text.) Source: OECD Tax Data Base.

Capital gains may or may not be taxed, and if taxed, may be subject to a reduced inclusion rate (i.e., less than full inclusion), or special rollover provisions may apply that defer corporate taxation of gains on assets if used in a similar business activity, for example. These and other provisions are relevant to the determination of taxable income from one country to the next in respect of various business activities. As noted below, some of these differences in the determination of the corporate tax base are captured in various measures of effective (as distinct from nominal) corporate income tax rates.

20

Second, the nominal rate structure applied to taxable profits shows a considerable degree of diversity across OECD countries. Rate schedules may be flat or graduated, with or without a surtax. Federal (as distinct from unitary) countries create the possibility of taxation at the state, provincial or cantonal level. And under scheduler approaches, different tax rates may apply to different categories of business income. Dis© OECD 2000

Nominal Income Tax Rates

tributed income may be taxed at a different statutory rate than retained income, and preferential rates have also been introduced in favour of income derived from equity as opposed to debt capital. While most countries have adopted a flat corporate income tax rate – meaning a fixed rate, invariant to the level of corporate taxable income – others have graduated tax structures in place, with the low bracket rate often coined the “small business” tax rate. In Canada, the first C$ 200 000 in active business income is subject to a reduced 12 per cent federal corporate tax rate as compared with a basic federal rate of 28 per cent, and similarly in the UK the first £300 000 in profits are taxed at a reduced “small business” tax rate of 21 per cent versus 31 per cent, as of January 1, 1999.7 The US federal corporate tax rate structure is also graduated with a top tax rate of 35 per cent applying to taxable income in excess of $10 million.8 On top of flat or graduated statutory rate structures, many systems include a surtax (i.e., a tax on corporate tax, as distinct from a tax on corporate taxable income) introduced often as a temporary measure, but in practice, with continued application year after year, taking on more or less a permanent status. Another important aspect is that a number of OECD Member countries have federal systems, creating the possibility of a second layer of taxation at the sub-central intermediate level (e.g., state, province or canton.) For example, most states in the US and each province in Canada levy corporate income tax. State income tax is deductible for federal tax purposes in the US as are cantonal (and municipal) income taxes in Switzerland. In Canada, provincial income taxes are not deductible from the federal base. Instead, the basic Canadian federal corporate tax rate is reduced by 10 percentage points to allow “tax room” to the provinces, and provide corporations with tax relief in respect of provincial income taxation. In Germany, local government imposes a business trade tax (Gewerbesteuer) on a base similar to taxable income (and deductible) for federal income tax purposes subject to certain adjustments.9 In contrast, in other federal countries including Australia, Austria, Belgium and Mexico, only the federal (central) government imposes income tax. Sub-central taxation is also observed in a number of unitary countries at the local level. Hungary and Luxembourg also impose a deductible local business tax at a fixed rate. In Japan, corporate income tax consists of the national (central government) corporation tax, the local government business tax and the prefectural and municipal inhabitant taxes. The local business tax rate is graduated, and while local governments typically adhere to a (common) standard rate structure, the rates applicable to each income band may be increased up to a maximum level (with the ceiling set by the central government, and with such increases leading to a reduction in federal transfers). The 46.37 per cent tax rate shown in Chart I is the overall standard rate, which takes into account the linkages between the applicable tax bases across years (as explained in Annex 2.A). A non-deductible expenditure-based local tax called the Regional Production Tax (IRAP) is found in Italy, replacing the previous income-based local corporate tax (ILOR). In certain cases, special corporate tax rates apply to different types of business income. Typically, this is effected by way of a special deduction from the corporate tax base, which generates a special nominal tax rate for the relevant income (some would prefer to call such a rate an effective rate). For example, while the basic federal corporate tax rate in Ireland is 32 per cent, a special 10 per cent rate applies to income from manufacturing industry and income of companies licensed to operate in the International Financial Services Centre (IFSC) in Dublin and in the Shannon Free Airport Zone, as well as income from other targeted activities.10 Profits derived from such activities are in the first instance subject to tax at the 32 per cent basic rate, and then relief is claimed to bring the effective rate down to 10 per cent. Similarly, a manufacturing and processing deduction in Canada lowers the applicable federal tax rate by 7 percentage points below the basic rate. A number of other OECD countries have similar regimes targeted at a range of activities, from manufacturing to finance activities. The tax rates in Chart 2.3 show the basic statutory tax rates applicable to non-targeted business activities. The application of nominal tax rates may also vary depending on the distribution policy of firms. In Germany, the basic central government corporate income tax rate is 45 per cent on retentions, while 30 per cent on distributions. This split-rate regime requires the separation of income into three main pools or reserves – income taxed at the retention rate (EK45), income taxed at the distribution rate (EK30), and finally tax-exempt income (EK0). Domestic and taxable foreign source income is allocated between EK45 and EK 30, or EK30 and EK0 depending on the effective corporate tax rate on the income © OECD 2000

21

Tax Burdens: Alternative Measures

being allocated. An equalisation tax is imposed at the rate (30/70) on distributions out of the tax-exempt pool EK0 (certain exceptions apply) to bring the effective tax rate up to 30 per cent, while refunds are provided at the rate of (15/55) on distributions out of EK45 to bring the effective tax rate down to the 30 per cent rate. This ensures that distributions are taxed at a uniform 30 per cent rate, which is used as the basis for the taxation of distributions at the personal shareholder level (including the granting of imputation tax credits, as discussed in Section 2.7). Last to consider is the application of a differential rate depending on the capitalisation of the firm, as one finds in the Italian system where the basic corporate income tax (IRPEG) rate is 37 per cent (a central government tax). A Dual Income Tax (DIT), aimed at encouraging the equity capitalisation of companies, reduces the basic CIT (IRPEG) to 19 per cent on the portion of taxable income corresponding to a notional return of 7 per cent on the net increase in equity of the company over the net equity shown in the companies balance sheet relating to the accounting period in force as of September 30, 1996. (i.e., a reduced IRPEG rate of 19 per cent applies to notional profit earned on a net increase in equity.) A minimum average IRPEG rate of 27 per cent applies (i.e., in any period, the average of the 37 per cent and 19 per cent IRPEG rates may not be less than 27 per cent (excluding IRAP)). The notional return (“ordinary yield”) is set each March 31 by the Ministry of Finance.11 Corporate income tax and shareholder taxation Among the possible roles of a corporate income tax, a central one is its withholding function, requiring that corporations pay tax on behalf of their shareholders on income earned at the firm level. Without a corporate income tax, individuals could retain earnings in a corporation as a way to defering, perhaps indefinitely avoiding being subject to tax at the personal level on (the distribution of) such income. However the existence of both corporate and shareholder level income taxation raises an “integration” problem. In the absence of mechanisms to integrate the corporate income tax and the personal income tax, income from equity capital invested in the corporate sector is subject to double taxation. That is, income that is subject to tax at the corporate level is subject to tax again at the personal level when distributed to individual shareholders as dividends. The double taxation of income can lead to a number of distortions in the economy. It may encourage individuals to forego the business advantages of incorporation, which can result in efficiency losses. It may also depress the level of savings and the level of corporate investment, with negative implications for job creation. Integration measures designed to alleviate the double taxation of income can operate to offset or remove these distortions (for a discussion of the effects of dividend taxation and relief from double taxation, see Annex 2.B). Proponents of integration argue that tax relief provided at the shareholder level in respect of corporate tax underlying a dividend distribution can lower the after-corporate tax rate of return that a corporation must earn to attract shareholders and provide them a competitive after-tax rate of return on their investment. The degree to which integration is successful in encouraging investment in a project may depend on a number of factors, including the design features of the integration system, the taxcharacteristics of the existing or new shareholders supplying the investment capital, and the type of funds used to finance the investment project (e.g., retained corporate earnings, capital from newly issued corporate shares, debt capital). The relative importance of these factors can differ according to one’s “view” of the impact of dividend taxation and measures designed to relieve the double taxation of distributed income.

22

Moreover, the double taxation of corporate distributions may distort a corporation’s decision of how to finance an investment project, and how to distribute earnings to shareholders. In general, the double taxation of dividend income may increase the cost of new share issues as a source of finance relative to the cost of retained earnings and debt. Integration relief may lower the cost of issuing new shares (i.e., can lower the after-corporate tax rate of return required by new shareholders) and thus operate to remove the tax-bias away from this method of finance. Integration methods that lower the effective tax rate on distributed income may also reduce the tax-induced bias to flow corporate earnings to shareholders by way of share redemption rather than by distribution.12 © OECD 2000

Nominal Income Tax Rates

Integration systems in OECD countries Integration systems can be differentiated on the basis of whether they partially or fully relieve the burden of corporate income tax at the corporate level, or at the shareholder level. An example of the former is a dividend deduction system that gives a corporate income tax deduction in respect of corporate tax on distributed earnings. Systems that provide relief at the shareholder level can be differentiated according to whether they simply exempt from personal tax distributed income (e.g., dividend exclusion system), or instead include distributed income in shareholders’ taxable income, but provide them with a tax credit that partially or fully offsets corporate-level tax (e.g., shareholder imputation or integration systems). Within the latter class of systems, a number of possibilities exist. As noted, the tax relief granted may be partial or full. Australia, Finland, France, Germany, Mexico, New Zealand and Norway all operate full imputation systems. Systems may be in place to ensure consistency between shareholder-level tax relief and corporate income tax actually paid on a distribution. Franking systems in place in Australia and New Zealand (Box 2.C) may be contrasted with equalisation tax systems found in Finland, France

Box 2.C.

Franking Systems

Under Australia's full imputation tax system, Australian companies are required to maintain a franked (i.e., tax-paid) income account to record income that has been subject to corporate-level tax – a company’s account is increased whenever it pays tax (or receives a tax-paid or “franked” dividend from another resident company), and is decreased whenever the company pays a franked dividend. Additions to a company's franking account in respect of Australian corporate income tax that the company has paid are determined by multiplying the amount of tax paid by a factor (1 – u)/u, where u = 0.36 is Australia’s (flat) corporate income tax rate. Upon the distribution of D units of dividends paid out of the franking account, individual shareholders must include in taxable income the amount of dividends grossed-up by the imputation tax credit accompanying the dividend. In other words, the inclusion in taxable income is given by: D(1 + (u/(1 – u)) = D/(1 – u) where u/(1-u) is the imputation tax credit rate measuring the amount of corporate tax underlying D units of dividends. The final personal tax liability is given by: m(D/(1 – u)) – D(u/(1 – u)) = D(m – u)/(1 – u) where m denotes the shareholder’s marginal personal income tax rate. The net effect is that the individual pays personal tax on the amount of distributed income measured before corporate tax, while receiving full credit for the amount of tax already paid on the distribution at the corporate level. Under the Australian system, the tax credit is fully refundable if there is insufficient personal income tax liability to fully absorb the credit. The Norwegian system is akin to the Australian system, with the difference that the shareholder’s (flat) nominal personal income tax rate at 28 per cent exactly equals the flat nominal corporate income tax rate (m = u), implying no additional tax liability at the personal level.

and the United Kingdom (Box 2.D). Other differences (e.g., whether shareholder tax credits are refundable or not) are also found. Other systems providing partial tax relief include systems in Portugal, Spain and Canada. These systems are also similar in that no equalisation tax applies to ensure corporate level taxation at the shareholder tax credit rate. Under the Portuguese system, withholding tax applies at a 25 per cent rate and the taxpayer has the option of treating the withholding tax as a final tax on the distribution, or including in taxable income the dividend gross of withholding tax and imputation relief and claiming a credit for both. A withholding tax system also applies in Spain. In Canada, no equalisation or withholding tax is imposed on distributions to resident shareholders who are granted a dividend tax credit at the rate of 25 per cent © OECD 2000

23

Tax Burdens: Alternative Measures

Box 2.D.

Equalisation tax

The integration systems in Finland and France also provide for full integration, but rather than relying on a franking account, rely on an equalisation tax (“compensatory” tax in Finland, and “précompte mobilier” in France) to ensure that distributions are taxed at the corporate level at a rate corresponding to integration credits granted at the individual shareholder level. The equalisation tax is creditable (deductible) against basic corporate income tax liability to avoid double taxation, while at the same time ensuring that distributed income has been taxed at the basic statutory corporate tax rate. The U.K. system provides for partial rather than full integration relief. In particular, the imputation tax credit per unit of dividends is 0.25, as determined by (us/(1 – us)) where us is the small company tax rate of 20 per cent. As under the above-noted systems, individuals are required to include in taxable income dividends measured gross of imputation credits. An individual liable to tax at the lower personal tax rate of 20 per cent would not pay any additional tax on a dividend (with personal tax liability offset exactly by the imputation tax credit), while taxpayers subject to tax at the top rate would bear an additional tax burden. An equalisation tax known as the Advance Corporate Tax (ACT) is payable at a 20 per cent rate on distributions, creditable against mainstream (basic) corporate tax, ensuring taxation at the corporate level at a rate matching the degree of imputation relief.

per unit of qualifying dividends. The credit rate corresponds to a 20 per cent notional combined federal and provincial small business tax rate. In contrast, other OECD countries including Austria, Belgium, Hungary, Japan, Luxembourg, the Netherlands, Poland, Sweden, Switzerland and the US operate classical tax systems that deny imputation relief. Under these systems, income is taxed at the corporate level, and distributions of after-tax income are then taxed again at the personal level with no direct relief for underlying corporate income tax on the distribution.13 Sometimes, a limited exemption for dividend income under the personal income tax applies, as in the case of the Netherlands. The advantages and disadvantages of providing full, partial or no integration relief are the subject of ongoing debate. Accounting for integration relief in tax burden measures As noted above, imputation tax credits provide for a reduction in tax at the personal level in respect of income tax paid at the corporate level on dividends distributed to shareholders. How imputation tax credits should be treated for tax burden measurement purposes depends on one’s view of the role of the corporate tax system. Under the view that corporate income tax has as its primary role a withholding tax function, then when measuring tax burdens on individuals versus firms, the amount of corporate tax on distributed income that is deemed to be a prepayment of personal tax – that is, the imputation credit amount – should be treated as part of the personal income tax burden. Thus personal income tax measured on a cash-flow basis net of imputation credits, plus imputation credits earned on distributed income, would together give the personal income tax burden. The corporate tax burden would then be measured by subtracting the same amount of imputation credits from corporate income tax revenues. If however one views corporate tax as a levy on corporations to pay for the public goods and services that corporations consume, then integration relief should be treated as a reduction in the personal tax burden. Under this view, corporate tax on distributed earnings is not to be regarded as a prepayment of personal tax, but a levy on corporations in its own right. The view is also consistent with providing relief for corporate tax to shareholders by way of imputation tax credits to mitigate the distortionary effects of double taxation (i.e., to lower the cost of capital, encourage savings).

24

When measuring the total tax burden on income from capital, which includes both corporate-and shareholder-level taxation, of course there is no need to label imputation credits as an offset to one level of tax or the other. However, when measuring the corporate tax burden separately, the issue does arise. Most approaches either ignore this complication, or implicitly treat all corporate tax as a levy on corporations in its own right, and make no offsetting adjustment to corporate taxes paid in respect of integration relief. © OECD 2000

Nominal Income Tax Rates

NOTES 1. 2. 3. 4. 5. 6.

7.

8.

9.

10.

11.

12.

13.

Sections 2.1-2.5 are based on De Kam and Bronchi (1998). See: OECD, The Tax/Benefit Position of Employees, 1998-1999 (Table 2.5). Paris, 1999. With the exception of Norway. Only in the cases of Denmark and Switzerland. In the course of 1998, the top rate in Japan was reduced to 50 per cent. Differences in statutory corporate income tax rates across countries encourage firms operating in more than one jurisdiction to book interest expenses (and other tax deductible amounts) in high-tax countries in order to minimize the amount of profit taxed at high rates. In response to this tax-planning incentive to shift taxable profit out of high-tax (into low-tax) countries, “thin-capitalisation” rules may be introduced which limit the amount of debt (in relation to equity or assets) that may be booked by a multinational in a given (high-tax) jurisdiction, with tax deductions denied for interest on any excess debt. The reduced Canadian rate, referred to as the small business rate, applies only to the first C$ 200 000 of active business income earned by Canadian-controlled private companies. Thus foreign-controlled companies and all public companies (those listed on prescribed stock exchanges) are denied the tax relief. The first $50 000 in taxable income is taxed at 15 per cent; the band $15 001 to $75 000 is taxed at 25 per cent; the band $75 001 to $10 million is taxed at 34 per cent, while taxable income in excess of $10 million is taxed at 35 per cent. A additional tax of 5 per cent is imposed on taxable income between $100 000 and $335 000 (implying a total rate of 39 per cent on taxable income in this band) which operates to phase out the benefits of graduated rates below 34 per cent for corporations with taxable income in excess of $100 000. The adjustments include certain expenses which may be deducted for federal corporate income tax purposes, but not for local business tax purposes, and vice versa (e.g., 50 per cent of interest payments on long-term loans may not be deducted for business tax purposes.) These targeted activities include a range of grant-aided export-oriented service industries; ship repairing, certain engineering design activities, grant-aided computer services including software development, and qualifying shipping activities. Net equity is increased by injections of capital in cash (i.e., excluding contributions in kind), capital contributions, transfers of current profits to reserves and increases of new reserves constituted by premiums paid in connection with the issuance of new shares. Net equity is reduced by voluntary reductions in net equity which result in a decrease of both the capital and/or the reserves (e.g., distribution of dividends); consequently, the losses of the financial periods would not reduce the amount of the capital increase. Integration measures may also influence cross-border portfolio savings decisions. A distinguishing feature of the tax systems of many OECD countries is that while the home country tax system offers resident individual shareholders some degree of double taxation relief in respect of corporate tax on distributed domestic source income, such relief is not provided by the home country in the case of portfolio foreign source dividend income. This would suggest that on the basis of tax factors alone, individuals in such countries would generally prefer to place savings in domestic shares, rather than foreign shares, even if they are able to completely offset foreign withholding tax by claiming foreign tax credits. However, this assumes that after-corporate tax (pre-personal tax) rates of return are the same on both investments. A discussion of this issue is beyond the scope of the current study. Classical tax systems limit tax relief to a tax deduction for corporate tax paid. That is, personal tax is imposed not on the full amount of pre-corporate tax income underlying a distribution, but rather on the after-corporate tax amount. Consider a distribution measured by Y(1 – u), where Y denotes pre-corporate tax earnings and u denotes the corporate income tax rate. Under classical tax systems, personal tax is imposed on Y(1 – u) and not on Y (i.e., a deduction for corporate income tax in the amount of uY is provided.

25

© OECD 2000

Chapter 3

TAX-TO-GDP RATIOS Introduction The tax-to-GDP ratio – showing the share of total tax revenues in gross domestic product – is the main aggregate indicator used in the annual OECD Revenue Statistics publication to measure the significance of taxes in Member countries. Despite the attraction of expressing tax revenues as a percentage of aggregate income (value-added) in a given economy, this indicator has some important limitations as a comparative measure of the tax burden and role of government across countries and over time. Similarly, the share of aggregate public outlays of general government in GDP is an indicator that must be interpreted with care. Factors that can affect the level and trend of tax-to-GDP ratios, and which may vary across countries and therefore affect comparability of results, include the following: i) the extent to which countries provide social or economic assistance via tax expenditures, rather than direct government spending;1 ii) whether or not social security benefits are subject to tax; iii) the relationship between the tax base and GDP, and the economic cycle; iv) the measurement of GDP; v) the amounts of tax evaded and the size of the underground economy, the latter being usually incompletely reflected in GDP figures; and vi) time lags between tax accruals and receipts, particularly in the case of corporate income tax. At the same time, when interpreting tax-to-GDP ratios it must be remembered that countries may adopt policies – for example, enact regulations – that have similar goals as tax-financed expenditure programmes and create similar economic incentives and disincentives as compared to taxes and spending, but that may have different implications for the size of the government sector as measured by the ratio of tax revenues or public outlays to GDP. Assessing tax burdens and the role of government in an economy also requires that a longer-term view be taken. This is especially the case in a situation with an ageing population, as a single year snapshot may not reveal the extent of future tax liabilities. Notably, as compared to countries with pay-as-yougo financed pensions, countries with funded pension plans in place as a rule defer taxation on deductible or exempted contributions, and on returns earned by the pension fund/insurer (resulting in lower current tax revenues) until the pension is paid into the hands of pensioners (resulting in higher future tax receipts). This chapter discusses how the first four factors noted above, listed under items i)-iv), can partially explain structural differences in tax-to-GDP ratios which need not represent genuine differences in the underlying burden of tax on the economy. The fifth factor is not discussed because it necessarily involves consideration of “dark numbers” where true values are uncertain. The sixth factor particularly affects single-year comparisons of tax levels. In addition to these factors, it is also important to recognise that different taxes have different economic effects, and no single aggregate indicator is likely to reflect the economic costs associated with tax. © OECD 2000

27

Tax Burdens: Alternative Measures

Tax expenditures versus direct expenditures In addressing a range of economic and social objectives, government may intervene in the market through public spending. An alternative instrument is tax concessions, as a substitute for direct public spending. “Tax expenditures” are defined as expenditures made through the tax system. The tax expenditure concept was developed in recognition of the fact that the tax system can be used to achieve similar goals as public spending programs, but that accounting for the costs and benefits of tax measures is often less rigorous, regular and observable than for direct expenditure.2 Currently, a tax expenditure list is included in the annual budget documents of more than ten OECD countries. Although the notion of tax expenditures is now well-accepted, definitions of exactly what constitutes a “benchmark” tax system – used to identify tax expenditures as deviations from the benchmark – are controversial. Indeed, the implicit or explicit benchmarks (the “normal” structure of the tax) against which countries measure tax expenditures vary considerably (See Box 3.A).

Box 3.A.

Tax expenditure, or not?

If a special deduction is offered which lowers the effective tax rate on income earned by small businesses below the basic nominal corporate income tax rate, should the small business tax rate be considered a tax expenditure or as being part of the benchmark tax rate structure? Different approaches to such questions mean that country practices in presenting tax expenditure accounts differ. This is one important reason why amounts involved in spending through the tax system are not directly comparable across countries.

Apart from problems associated with defining a “benchmark” tax system, there are several other reasons why cross-country comparisons of tax expenditure figures may be misleading, even if confined to a particular sector or activity. A full discussion of such pitfalls may be found in section V of the report on Tax Expenditures which the OECD released in 1996. As is explained in Box 3.B, a country that prefers tax expenditures over direct government expenditures will – all other things equal – have a lower tax-to-GDP ratio than countries opting for direct spending programmes.

Box 3.B.

How tax expenditures reduce the tax-to-GDP ratio

The following hypothetical example illustrates how the substitution of tax expenditures for direct public expenditure can affect the tax-to-GDP ratio. Assume that total public outlays in country A equal 50 per cent of GDP and that one-tenth of the budget is spent on investment incentives aimed at private firms. To stimulate private investment, instead of tax-exempt direct subsidies, country B uses tax incentives equally worth 5 per cent of GDP. As a consequence, in net terms investors in country B receive in financial assistance from the government the same share in GDP (5 per cent) as do investors in country A. However, the overall spending and tax levels in country B are measured as being 5 per cent of GDP below those in country A. It follows that country A could reduce its tax and spending ratios by moving from direct subsidy spending to introducing equivalent tax reliefs. Although direct spending on investment incentives is discontinued, investors would receive the same government assistance as before because they pay 5 per cent of GDP less in taxes on income and profits. So in both cases, private firms as a group receive exactly the same financial support from the government. 28

© OECD 2000

Tax-to-GDP Ratios

Tax treatment of social security benefits The tax treatment of social security benefits also impacts on tax-to-GDP ratios, with the result that differences in this policy area influence differentially tax-to-GDP ratios across countries. Tax concessions, as an alternative to transfer payments or subsidies, may be used to help cover private health costs, stimulate investment in housing, and promote educational opportunities. For example, exempting social security benefits from taxation, as some countries do, is one way of increasing the after-tax benefits provided by such programmes. But because not all countries adopt this approach, tax-to-GDP ratio comparisons are affected (see Box 3.C).

Box 3.C.

How the tax treatment of social benefits impacts on the tax-to-GDP ratio

The following example serves to illustrate how the tax treatment of social security benefits can affect the tax-to-GDP ratio. Assume that total public outlays in country A equal 50 per cent of its GDP and are fully tax-financed. The wages of public servants claim one-quarter of the public budget. Government spending on goods and services produced by the private sector absorbs a further one-quarter of budget allocations. The remaining half of the budget (equal to 25 per cent of GDP) is paid out as transfer income to benefit recipients in whose hands it is taxed at an average rate of 20 per cent. In this example, as a group, benefit recipients contribute one-tenth of aggregate tax revenues (5 per cent of GDP). Country B is recorded as having a smaller public sector, with tax and spending levels at 45 per cent of GDP. However, a closer look at its public budget and th.e relevant public programs reveals that in terms of the (net) impact of the public sector on the private sector, both countries are in exactly the same position. Country B spends 5 per cent of GDP less on income transfers, but it does not tax such income in the hands of recipients. As a consequence, in net terms benefit recipients living in country B receive the same share in GDP (20 per cent) as is the case in country A. But the overall spending and tax levels in country B are measured as being 5 per cent of GDP lower than in country A. It follows that country A could reduce its tax and spending ratios by 5 per cent of GDP, solely by switching to exempting transfer income from tax. The budget item for transfers would shrink by 5 per cent of GDP, without benefit recipients experiencing a reduction in their net disposable income. Gross benefit amounts would be lower, but since transfer income would now be tax-exempt, benefit recipients could spend the same amount on goods and services as before. Table 3.1 below summarises the preceding argument. The impact of the tax treatment of social security benefits can be illustrated by taking as an example government support to families with children. Such may be provided in four different ways: 1) as taxable child benefit; 2) as tax-exempt child benefit; 3) as a credit against personal income tax; and 4) as a personal income tax allowance. The tax-to-GDP ratio will be highest in the first case and lowest in the last case, with net disposable household income the same in all cases. Data suggest that significant amounts may be at stake. As an example, in the case of Germany, the change from child benefits to tax credits in 1996 involved DM 20 billion (about 0.5 per cent of GDP).*

Table 3.1. Tax and spending ratios: an illustrative example (% of GDP) Country

Public wages and purchases from the private sector Spending on transfer incomes Total spending, tax/GDP ratio Taxes on transfers Total spending, tax/GDP ratio after correction for taxes on transfers Net spending on transfer incomes after taxes

A

B

25 25 50 5 45 20

25 20 45 0 45 20

* Under the German reform the previously untaxed child benefit (DM 20 billion) and the child allowance (DM 18 billion) were merged into a more generous tax credit programme involving DM 49 billion (1997 figures). 29

© OECD 2000

Tax Burdens: Alternative Measures

In practice, the role of direct taxes and social contributions varies substantially between countries. For example, a recent study by the OECD showed that in 1995 income tax and social contributions paid by benefit recipients equalled roughly 5 to 6 per cent of GDP in Denmark, the Netherlands and Sweden, whereas in Australia, Ireland, the UK and the US they paid less than 0.5 per cent of GDP in such taxes.3 For Belgium, Germany, Italy and Norway taxes paid by this group lay in the middle range (0.5 to 3 per cent of GDP). These differences in direct taxes contributed by benefit recipients reflect the scale of social expenditures, the extent to which these expenditures are targeted on those with low incomes, and how far benefits are exempted from income tax and social contributions. Also, if benefits are tax-exempt – making for a smaller aggregate tax base – then social security contributions to finance the public programs concerned are in a number of countries not deductible – making for a broader income tax base than would otherwise be the case – and vice versa, although there are important exceptions to this rule. In the former case, the base erosion produced by the tax exemption of benefits may be largely offset by the non-deductibility of contributions to finance social security. The large increase in unemployment in a number of continental European countries together with the gradual ageing of populations has put upward pressure on the level of social security outlays. As the amount of income redistributed through the social security system has expanded, so probably have the budgetary and statistical impact of the different tax treatment of social security transfers. The relationship between tax base and GDP, and economic cycle effects Another key consideration is that differences in tax-to-GDP ratios do not necessarily reflect differences in tax policies, across countries or over time. To illustrate this point, first note that the aggregate tax-to-GDP ratio consists of a number of component taxes entering the numerator.4 Consider one of these component parts, for example the corporate income tax (CIT) to-GDP ratio. This ratio can be expressed as follows: (CIT/GDP) = (CIT/Π) x (Π/GDP)

(1.a) *

where Π denotes economic profit. If τ denotes the effective average corporate tax burden on corporate economic profit, so that CIT = τ* x Π, the ratio CIT-to-GDP may be written as follows: (CIT/GDP) = τ* x (Π/GDP)

(1.b)

This expression reveals that differences in measured values of CIT-to-GDP over time and across countries can arise not only due to changes in tax policy over time and between countries – as captured by differences in τ* – but also as a result of differences in the ratio of economic profit to GDP over time and across countries. It follows that differences in the CIT-to-GDP ratio, and more generally differences in the aggregate tax-to-GDP ratio over time and between countries, cannot be taken immediately as indicative of differences in tax policy. In particular, the contribution of capital to GDP as captured by the ratio (Π/GDP) can change over time, and thus (CIT/GDP) will be observed to change over time even with tax policy held constant. The share of GDP that is effectively subject to corporate income taxation may vary as an economy moves through the business cycle, because firms claim a varying amount of tax incentives, or as a consequence of intensified tax planning activity. The ratio (Π/GDP) can differ across countries at a given point in time, and thus (CIT/GDP) can be observed to differ across countries on account of similar considerations. Another important factor is that the build-up of corporate tax loss pools carried forward and used to offset corporate tax liabilities will differ both over time, and across countries at any given point in time. These differences will impact on the CIT-to-GDP ratio – (in terms of the breakdown shown above, will cause differences across countries and over time in τ*) – that are reflective more of past policy decisions (e.g., the prior use of tax incentives) than of current tax policy priorities.

30

More generally, aggregate tax-to-GDP ratios and other component ratios relying on aggregate tax and GDP data (e.g., corporate tax-to-GDP ratios) mask important distinctions between the tax treatment of the relevant tax bases, and the relationship of those bases with gross value-added in the economy as measured by GDP. The implication is that variations in tax-to-GDP ratios over time and across countries in fact may say little about differences in tax policy. Only a closer examination of the component parts of ratios derived from aggregate data can shed light on factors contributing to observed dynamics of such measures. © OECD 2000

Tax-to-GDP Ratios

Revisions to the measurement of GDP The measurement of GDP can differ across countries and time, making comparisons of tax-to-GDP ratios difficult. First, the degree of accuracy with which GDP is measured by the statistical agencies of different countries would appear to vary considerably. Second, the scale of the so-called “black” or underground economy, which is recorded in GDP only sketchily, differs between countries and is thought to be large in a number of countries. To the extent that no tax is collected on such activities, however, this may not be a large source of measurement error. Third, GDP figures are subject to numerous revisions, including the revision and updating of estimates, not necessarily for all countries at the same time, reflecting better data sources and estimation procedures. Generally, these revisions have also a rather limited impact on tax ratios. Occasionally, however, GDP figures may change in a more fundamental way when internationally agreed guidelines to measure the value of gross domestic product are structurally adjusted. This occurred in the mid-1990s, when the System of National Accounts 1993 (hereafter: 1993 SNA) began to gradually replace its predecessor, the System of National Accounts 1968 (1968 SNA). In computing their gross domestic product, the fifteen Member States of the European Union (EU) are bound to adhere to the European System of Integrated Economic Accounts (ESA), which is primarily an elaboration of System of National Accounts, though differing from it in several aspects.5 Following the current revision of the System of National Accounts, the 1979 ESA has been replaced by the 1995 ESA. Considering for example the 1999 edition of Revenue Statistics, eight out of the fourteen OECD countries that are not part of the European Union still report gross domestic product following the 1968 SNA. Australia, Canada and Norway, the Czech Republic and Hungary however report their GDP on the basis of the 1993 SNA. By mid-1999, eleven EU member States had implemented the 1995 ESA to measure their GDP. The GDP estimates available for four EU member countries – Austria, Greece, Luxembourg and Portugal – are still based on the 1968 SNA/1979 ESA framework, as is the case for Switzerland which is not a member of the EU. The impact on tax ratios in moving from “old” to “new” GDP measures can be significant. The case of Denmark offers a good illustration (see Box 3.D).

Box 3.D.

How a revision of GDP can lower the tax-to-GDP ratio

The recent revision implementing the revised SNA/ESA framework to compute the value of gross domestic product increased the Danish 1997 GDP by 5.0 per cent. The isolated effect of this GDP revision was to lower the tax ratio by roughly 2.5 percentage points.* Thus the GDP revision may explain to a large extent why Denmark in 1997 – different from the situation in 1994-1996 – no longer heads the list when OECD countries are ranked by decreasing tax-to-GDP ratio. * The differences are not only due to the move from the old to the new System of National Accounts. The 1968 SNA data is not revised so the differences partly reflect the move from 1968 SNA to 1993 SNA and partly the revisions to the 1997 data which would be expected to show up between the data reported on the questionnaires returned in 1998 and those sent back in the second quarter of 1999.

Over the next few years, as an increasing number of OECD Member countries implement the revised SNA/ESA framework to compute the value of gross domestic product, and make revisions for a period reaching back farther into the past, the comparability of tax ratios between countries and over time will improve. However, consequences of the recent structural one-off revision of GDP figures should be kept in mind when analysing tax ratios reported in the 1999 edition of Revenue Statistics and constitute one more argument why tax-to-GDP ratios should be interpreted with due caution.6 © OECD 2000

31

Tax Burdens: Alternative Measures

NOTES 1. See Annex II to Revenue Statistics in OECD Member countries 1965-1979 (Paris, 1980) for an analysis of the borderline problems involved. 2. If countries exempt certain income components from tax – for example, the first 5,000 currency units of interest income earned, transfer income received or the imputed rent of owner-occupied houses – or if countries allow certain deductions in computing taxable income – for example, for exceptional health costs and charitable donations – such exemptions and deductions typically qualify as tax expenditures, special provisions which represent government expenditures made through the tax system to achieve economic and social objectives. See: OECD, Tax Expenditures – A review of the issues and country practices (Paris, 1984) and OECD, Tax Expenditures: Recent experiences (Paris, 1996) for a general discussion of the conceptual issues involved. 3. Adema (1999) p. 30. 4. In particular, the aggregate tax-to-GDP ratio, which we can denote as T/GDP, can be broken down as follows: (T/GDP) = (PIT/GDP) + (CIT/GDP) + (X/GDP) where PIT measures personal income tax revenues, CIT measures corporate income tax revenues, and X measures all other tax revenues (i.e., other component taxes including social security contributions, taxes on payroll and workforce, taxes on property, consumption taxes, etc.). 5. These differences are not pertinent to tax/GDP comparisons as reported in OECD Revenue Statistics. 6. See the special feature S.3 in OECD (1999), pp. 34-38.

32

© OECD 2000

Chapter 4

AVERAGE TAX RATES AND THE NEED FOR MICRO-DATA Introduction As noted in Chapter 1, tax burden measures linking taxes paid by households and/or firms to economic concepts of income generally offer a much more informative indicator of the burden and impact of tax systems than a simple reliance on nominal (statutory) tax rates. Similarly, such measures offer greater insight than figures expressing tax revenue as a percentage of GDP. Continued interest in the calculation of average tax rates on capital income, labour income and other aggregates can be traced to at least four considerations. First, other tax measures of policy interest, including nominal tax rates and marginal effective tax rates, provide less than full consideration of the factors determining tax burdens and incentive effects. Nominal tax rates, while potentially relevant to investment and work incentives, do not give any account of other equally (possibly more) important tax considerations.7 Similarly, marginal effective tax rates (METRs), while taking into account a number of factors thought relevant to investment and work incentives, may in certain cases yield misleading indicators of the incentive effects of the tax system (see Chapter 5). Average tax rates (ATRs), on the other hand, by including the actual amount of tax collected in the numerator, implicitly take into account the combined effect of nominal income tax rates, tax deductions and tax credits. ATRs also take into account the effects on the domestic tax base of tax planning, as well as tax relief available from lax or discretionary administrative practice. Thus, ATRs, if properly measured, generally will give much better measures of overall tax burdens. Moreover, ATRs may be useful in explaining behavioural effects of taxation, and in certain cases may be better indicators than METR statistics.8 Second, there is a view, perhaps mistaken, that average tax rates analysis is intrinsically a simpler exercise than other types of tax analysis, for example METR analysis. In particular, there is no need to delve into the detailed rules that go into the determination of capital cost allowances, tax credits, the cost of financial capital, and so on, that must be encoded in a METR formula. The net effect of these and other rules on tax burdens is captured in the measurement of the final tax revenue amount. Third, policy makers often want to know how the tax system is impacting on the economy and its agents – for example, the take-up rate of special tax concessions by various socio-economic strata; the degree of income redistribution across households brought about by a progressive personal income tax rate schedule. Indeed, policymakers and the public alike are keenly interested in assessing fairness in the application of specific tax programmes as well as equity in the tax system in general. For example, questions arise concerning how the tax burden is shared among individuals and corporations, across labour and capital, and how this balance has changed over time. People also want to better understand whether the tax system impedes, encourages, or is neutral towards investment and job creation, with often a special interest in the tax treatment of small and medium-sized enterprises (SMEs). Equity and efficiency concerns are raised if small business profits are judged over-taxed, depressing after-tax retained earnings, the main source of finance for SMEs. Fourth, it is recognised that, whatever the relative strengths or weaknesses of ATR statistics, such numbers will be generated, quoted, interpreted and used to influence tax policy debate. Given this, an interest emerges in developing a better understanding of what goes into the making of such statistics, what can be made of them, and what can not. © OECD 2000

33

Tax Burdens: Alternative Measures

A focus on corporate taxation An area of interest that has dominated much of the recent public debate on the role of taxes concerns the taxation of business. Given the increased competitive pressures on business accompanying the liberalisation of trade and investment flows, multinationals often warn of the need to lower corporate tax rates to attract foreign direct investment and discourage domestic capital flight. On the other hand, a perceived gradual shifting of the tax burden onto less mobile factors of production, including low-skilled labour, and consumption has fuelled demands that corporate tax burdens be increased, in order to ensure that firms pay their fair share of taxes in support of public programs. This debate has heightened calls for reliable measures of corporate tax burdens to enable comparisons across factors of production, countries and time. Many would argue that such demands are off the mark as corporate tax burdens calculations are artificial, given that corporations are merely legal forms through which individuals may conduct business. Thus, the focus should be on measuring tax burdens on capital, taking into account not only corporate taxes on income received at the corporate level, but also personal tax on the returns to individual capital suppliers. As noted below, determining personal tax on investment income is not a straightforward exercise, and recourse must be made to micro-data for reliable estimates. However, aside from this, interest in separate corporate tax burden measures may not diminish, and in fact may grow, given the distinction made by many between taxes on individuals versus taxes on corporations, and the continued pressure by business to keep corporate rates low, and by labour to ensure corporations pay their fair share. As becomes clear from the discussion, none of the approaches reviewed to tackle these and other tax burden questions are without shortcomings, whether relying on publicly available aggregate or firmlevel data. The problems posed by aggregate data in general, namely those related to an inability to make necessary adjustments to ensure consistency in the approach – in this case, consistency between numerator and denominator amounts – are found here as well. And publicly-available firm-level data, while in principle offering this advantage, raises its own problems related to both the numerator (the amount of taxes paid) and the denominator (profit). Taxes paid may partly concern profits earned in other years and are thus not necessarily connected to profits of the current year. Also, published accounts of individual firms may not separately show all relevant taxes paid. As regards the denominator, profits reported in annual accounts reflect national accounting practices which hampers international comparability of tax rates based on firm-level data. Moreover, the sample of firms may cover only specific sectors of the economy, and some years. Average tax rates calculated for one sector of the economy need not be representative of the tax burden on other sectors. And one may be interested in examining the trend in tax burdens over a longer period than that for which detailed firm-level data are available. Absent a detailed, representative firm-level data set covering the main sectors of the economy over the time period of interest, policy-analysts must turn to other data and estimating techniques. Some of the main options available are reviewed below. Two frameworks for assessing corporate tax burdens Average tax rates (ATRs) derived for incorporated businesses may be calculated (with a lag, allowing for the gathering, compilation of required data) as the ratio of corporate income tax divided by some measure of pre-tax corporate profit or surplus. Typically, the focus is on domestic corporate income tax (i.e., foreign tax on foreign source income is ignored) imposed on resident corporations, regardless of (domestic versus foreign) ownership. In principle, actual corporate tax paid should be included in the numerator, however several studies deriving ATRs at the firm level, relying on data gathered from financial accounts, use tax liabilities as reported for book purposes. Economic profit (as opposed to book or taxable income) should enter the denominator, which may call for inflation adjustments to financial income, the exclusion of loss firms and special consideration of loss-carryovers, and possibly other adjustments to ensure a matching of numerator and denominator amounts.9 34

In assessing effective tax burdens, two types of frameworks can be distinguished – backward-looking measures, and forward-looking measures. © OECD 2000

Average Tax Rates and the Need for Micro-Data

Backward-looking corporate average tax rate measures assess the average rate of corporate tax on income derived from previously acquired (i.e., existing/installed) capital of firms. Such frameworks are “backward-looking” in the sense that they assess current tax liabilities on profit generated from capital stock accumulated in the past.10 Within the category of backward-looking ATRs, one can distinguish between approaches that that rely on aggregate (economy-wide) data, including so-called implicit tax rates, and those that rely on firm-specific(micro) data (publicly-available or confidential). Firm-level data offers the advantage that it can be aggregated to generate size- or sector-specific rates. Alternatively, one can consider forward-looking corporate average tax rates that assess corporate taxes as a percentage of pre-tax profit on a prospective investment. While backward-looking measures may be the most appropriate indices, if measured properly, to address the equity concerns with the tax system, they may have limited value for inferring how the current tax system is impacting on investments in new productive capacity. In order to address the latter concern, this second approach of measuring tax burdens on new investment is in order. Within the forward-looking category, one may distinguish between projectanalysis ATRs and marginal effective tax rates (METRs). Project-analysis ATRs consider a discrete investment project and measure corporate taxes payable as a percentage of economic profit for a given discrete (lump-sum) amount of capital invested. In contrast, METR analysis assesses tax burdens at the margin, on the last currency unit invested, using theoretical models that characterise optimal investment behaviour. These forward-looking measures are addressed in Chapter 5. We begin below by first considering so-called “implicit” tax rate analysis. Such measures have been the subject of much scrutiny in the recent past, at least in the European context. Their appeal can be explained to a large extent by the fact that they may be derived using publicly-available data (taken from the OECD Revenue Statistics, and National Accounts data). Yet they may be highly misleading indicators of tax burden, particularly in the case of corporate implicit tax rates, as discussed below. The following sections consider backward-looking average tax rates derived using aggregate data, and results derived using micro-data. As regards the use of micro-data, while results derived using financial accounting information are suggestive, we argue that recourse should be taken to use confidential data compiled by central governments. Only this data is sufficiently rich in detail to permit the types of refinements that should be made in principle to arrive at internally-consistent corporate ATRs, and measures of tax burden on income from capital and labour. Implicit ATRs Like profit-based ATR measures, implicit corporate tax rates derived for the economy as a whole include aggregate revenues from taxes on corporate income in the numerator. However, corporate operating surplus as reported in the National Accounts enters the denominator.11 Different from accounting definitions of business profit, operating surplus is measured gross of (including) interest, rent and royalties paid by corporations.12 For this reason, and several others, variations in corporate implicit tax rates over time and across countries cannot be readily interpreted as indicative of changes in tax policy or investment incentives, with implications for its usefulness as a policy tool. This following discussion addresses potential difficulties with the measurement of “implicit” average tax rates for the following categories of income and factors of production: 3. personal income (sub-section 4.3.1); 4. corporate income (sub-section 4.3.2); 5. capital (sub-section 4.3.3); and 6. labour (sub-section 4.3.4). The following section (4.4) suggests how micro-data might be used to address the problems identified.13 In considering these four implicit average tax rates, formulae are presented to facilitate the discussion. While it is recognised that variants to these formulae may be applied, the problem areas identified generally are common to all alternative formulations, allowing in most cases reference to a single representation. We choose here to use the formulae found in Mendoza et al. (1994), unless otherwise indicated. © OECD 2000

35

Tax Burdens: Alternative Measures

Personal income average tax rates The average tax rate on personal income may be expressed as follows: tP = PIT1100/(W + PEI + OSPUE)

(1)

where PIT1100 measures taxes on income, profits and capital gains of individuals (Revenue Statistics category 1100), W measures wages and salaries in cash and kind paid to employees gross of employee social security contributions, PEI measures property and entrepreneurial income received by households, and OSPUE measures the operating surplus of private unincorporated enterprises.14 Treatment of capital gains Category 1100 includes personal tax on (net) capital gains. However, the denominator excludes the corresponding tax base (realised net capital gains), implying an inconsistency between the measurement of the numerator and denominator of the personal ATR. This inconsistency could be eliminated in principle by subtracting from the numerator personal tax on net capital gains.15 However, this approach would remove a potentially significant personal tax component. Another option might be to add to the denominator realised net capital gains of resident individuals on the sale/disposition of domestic property. This information generally would be provided on individual tax returns in countries that impose this tax although information available at the Secretariat suggests that few countries have compiled this data. It seems unlikely that information on realised net capital gains would be available in countries that do not tax gains, and comparability across countries would require that the adjustment to the denominator be made for all countries. However, for countries that do tax individuals on net capital gains, it would be useful to determine the importance, in quantitative terms, of the omission of net realised capital gains from the denominator of the personal income ATR. Accounting for integration relief Imputation credits offering shareholders double taxation relief in respect of underlying corporate income tax on dividends received are usually captured in Revenue Statistics under category 1100. Some would argue that the personal tax burden on dividend income is properly measured gross of imputation relief, as corporate-level tax on distributions is, in effect, a pre-payment of personal tax liabilities on such amounts. This argument is particularly forceful in systems where an equalisation tax is imposed on distributions at the corporate level, and where the distributions tax is creditable against regular corporate income tax, with the rate providing a match with the imputation relief provided at the individual shareholder level. Adjusting ATRs in respect of integration relief would involve increasing aggregate personal income tax revenues PIT1100 by a notional amount equal to total imputation credits claimed at the personal shareholder level (and, in the measurement of the corporate income ATR (see below), reducing aggregate corporate income tax revenues (category 1200) by the same amount). Following the principle of cashbasis reporting, figures shown in Revenue Statistics are not adjusted in this way. Aggregate shareholder imputation credits are not reported separately in Revenue Statistics.16 However, this amount (or a rough estimate of it) could in principle be drawn from individual taxpayer-level data, in some cases requiring simplifying assumptions regarding the carry-over of unused credits. Information on the magnitude of the required adjustment would be another useful contribution of micro-data analysis. Personal tax on foreign source income

36

Both categories, wages and salaries paid to employees (W) and property and entrepreneurial income (receipts) of households (PEI) capture amounts paid to households by resident producers. For countries that tax resident individuals on their worldwide income, the personal income tax measure PIT1100 in the numerator of the personal ATR includes not only domestic tax on domestic source property, © OECD 2000

Average Tax Rates and the Need for Micro-Data

entrepreneurial and wage income, but also net domestic tax on foreign source property, entrepreneurial and wage income.17 This raises the question of how significant the distortion is to the ATR introduced by this measurement asymmetry. Where micro-data is available that identifies separately foreign source income, the net amount of domestic tax on these amounts could be estimated in principle and subtracted from the numerator. This would permit gauging how significant a factor this is. Corporate income average tax rates The implicit average tax rate on corporate income may be expressed as follows: tC = CIT1200 /OSC

(2a) C

where CIT1200 measures taxes on income, profits, capital gains of corporations, and OS measures the operating surplus of the corporate sector (equal to the total operating surplus of domestic producer units OS, less the operating surplus of private unincorporated enterprises, OSPUE). Concept of operating surplus too broad The surplus amount OSC in the denominator of (2a), while measured net of wages and salaries, is gross of interest expense, rents, royalties and other amounts deductible for book income or accounting purposes. The base in equation (2a) is therefore broader than what might be considered appropriate when considering a corporate income average tax rate. Therefore, it would be informative to consider an alternative base which nets from OSC amounts deductible for book purposes (including interest expense, rents, royalties) paid to the household sector and to non-residents. Defining this adjusted book income measure ABYC, the corporate income ATR could be measured alternatively as: tC^ = CIT1200/ABYC

(2b)

Figures generated by equation (2b) – adjusted to exclude loss firms and corporate tax on foreign source income (as discussed below) – could be more accurately identified as average corporate income tax rates.18 Treatment of losses National Accounts production and income data cover both profitable and non-profitable firms. The inclusion of non-profitable firms tends to lower aggregate operating surplus in the economy.19 As firms in a loss position typically do not pay tax, the inclusion of loss firms generally would have no impact on the numerator of the corporate ATR, while reducing its denominator. The net effect is an increase in the ATR over what would be observed if loss firms were excluded. Some would argue that an ATR that properly reflects the corporate tax burden would exclude firms in a loss position.20 Or in other words, the inclusion of such firms tends to overstate the correct corporate ATR reflecting the average tax burden on profitable firms. Using micro-data, non-profitable firms could be removed from the sample, allowing a corporate ATR to be derived for profitable firms. Variation in ATRs by industry One major drawback of corporate ATRs derived from aggregate data is that they are unable to reveal the variation in corporate income tax burdens across industries. This inability stems from the fact that Revenue Statistics data do not breakdown corporate income tax revenue by industry. Nor do National Accounts data provide the necessary breakdown.21 Average tax rates by industry may however be derived using micro-data on a representative sample of corporations categorised by industrial classification number. When deriving corporate ATRs by industry, the use of corporate OS in the denominator clearly would not be appropriate.22 Instead, an adjusted book (accounting) profit measure should be used, as discussed above and represented in equation (2b). With ATR figures derived using this approach, it would be interesting to observe whether relatively low ATRs are found in industries or sectors that involve business activities that are believed to be generally more mobile than others (e.g., financial institutions sector versus manufacturing sector).23 © OECD 2000

37

Tax Burdens: Alternative Measures

Corporate tax on foreign source income For countries taxing resident corporations on their foreign income – including foreign branch income, dividends received from foreign affiliates, and other foreign source investment income – the corporate income tax measure CIT1200 in the numerator of the corporate ATR will be broader in coverage than the denominator, which includes only the operating surplus generated by resident producer units. As in the personal ATR case, this raises the question of how significant the inconsistency between the measurement of numerator and denominator amounts is. It would also be interesting to know whether this measurement bias has increased over time, as one might expect given increased relative interest in cross-border investment activity. Again, micro-data could be usefully applied to address this question. Labour income average tax rates The implicit average tax rate on labour may be expressed as follows: tL = (tP(W) + SSC2000) + PAY3000)/(W + SSCF2200)

(3)

tP

where is the average tax rate on (overall) personal income as noted under item (a). SSC2000 measures the sum of social security contributions of employees (SSCE2100), employers (SSCF2200), the selfemployed and others such as benefit recipients (SSCPUE2300), plus contributions that are unallocable between these three categories (SSCOTH2400). Lastly, PAY3000 measures taxes on payroll and workforce.24 Personal tax on wage income The common approach when using aggregate data to measure an average tax rate on labour is to estimate the amount of personal tax collected on wage and salary income (W) using the average tax rate on personal income t P. Even where labour and capital income are pooled for tax purposes at the individual taxpayer level, such an approach may be criticised where aggregate labour income is believed to be subject – on average across taxpayers – to a significantly different average tax burden than capital income. Generally, capital income will tend to be concentrated in the hands of high-income individuals and therefore, under a progressive rate structure, be subject to higher marginal and average tax rates as compared to labour income. On the other hand, special concessions or tax breaks may apply to income from capital, so that the average tax rate for capital income might not be significantly different from that for labour income. Forcing this assumption on the exercise, as the ATR model does when based on aggregate data, is however a shortcoming of the framework. Using micro-level data – that is, tax data collected at the individual taxpayer level – it should be possible to generate more accurate measures of the personal average tax rates on separate items of income, including wages and salaries, taxable income from capital, as well as transfers (if taxable), and to examine how such measures compare with those derived using aggregate data. This should be possible where the data set includes, for individual taxpayers in the sample, separate reported figures for the items of income for which tax ratios are being calculated. Annex 4.B considers how micro-data might be used for this purpose, and considers three possible systems for illustrative purposes: i) progressive taxation of combined labour and capital income (global taxation); ii) system i) with a 50 per cent inclusion rate for capital income; and iii) a dual tax system with separate taxation of labour and capital income. As illustrated in annex 4.B, rather than using tP in equation (3), an average personal rate on total wage and salary income could be derived from a representative sample of taxpayers as follows: tW = ∑j(Wj/W)*(PITj/Yj) = ∑jwj*tPj

(4) th

38

where Wj measures the wage and salary income of the j taxpayer in a sample of n individuals (j=1,..,n) and where W = ∑jWj measures aggregate wages and salaries in the sample. PITj measures the final personal income tax liability of the jth taxpayer on his total net income of Yj. Equation (4) therefore measures the average personal tax rate on wage income tW as a weighted average of each individual taxpayer’s aver© OECD 2000

Average Tax Rates and the Need for Micro-Data

age personal tax rate tPj, with the weights wj = (Wj/W) attached to these individual tax rates reflecting the distribution of total wages and salaries across taxpayers. Using micro-data to generate a more precise estimate of the amount of average personal tax rate on wage and salary income would address a central defect with ATR analysis based on aggregate data. Capital income average tax rates The average tax rate on capital may be expressed as follows:25 tC = (tP(PEI + OSPUE) + CIT1200 + TIMP4100 + TFT4400)/OS

(5)

where TIMP4100 measures recurrent taxes on immovable property, TFT4400 measures taxes on financial and capital transactions, and with the other variables tP, PEI, OSPUE, OS and CIT1200 as defined above. Personal tax on property and self-employed income The common approach to generating an average tax rate for capital is to group corporations and unincorporated enterprises together and estimate personal tax on income from capital, including domestic source property income received by households and self-employment income, using an overall average tax rate on personal income (tP). This methodology has been criticised on at least two grounds. First, as noted in relation to the labour ATR, a potential measurement error is introduced when using an average tax rate on personal income derived using aggregate data to estimate personal tax on specific categories of personal income. Second, treating the operating surplus of unincorporated enterprises (OSPUE) as solely a return to capital, and the tax imposed on it as entirely a tax on capital, has been criticised as biasing the results, as this surplus amount is in fact a return to both capital and labour supplied by the self-employed.26 While an attempt could be made to apportion OSPUE and the tax imposed on it in part to labour and in part to capital, it is recognised that splitting OSPUE into labour and capital components will always remain a rather arbitrary exercise. An alternative approach would be to treat unincorporated enterprises separately and to calculate an average tax rate for the combined returns to owners of unincorporated enterprises. It then becomes unnecessary to ascribe OSPUE to the (unknown) combination of factor inputs.27 Under this option, and using micro-data, an average personal tax rate on OSPUE could be derived as a weighted average of (j=1,…n) individual average personal tax rates tPj with the weights zj = (Zj/Z) reflecting the distribution of total self-employment income Z = ∑jzj across taxpayers, as follows (see also equation (4)): tUE = ∑jzj*tPj

(6)

The following reasons can be given in support of treating unincorporated enterprises separately. First, the analysis along these lines avoids potentially biasing the results from rather arbitrarily splitting OSPUE into its component parts. To the extent that the final average tax rate results are more robust, their informational content may be increased rather than decreased. Second, where a policy interest in measuring average tax rates arises out of equity concerns over the sharing of the tax burden between workers and investors, it is unnecessary to artificially split the return to owners of PUEs into returns to labour and capital, as the worker and the investor in this case are one and the same. If this latter approach is adopted, a capital ATR could be measured excluding unincorporated enterprises, with operating surplus of incorporated enterprises OSC appearing in the denominator: tC = ((tPEI·PEI) + CIT1200 + TIMP4100 + TW4200 + TFT4400 + TOTH6000)/OSC or alternatively, using an adjusted book income figure for the corporate sector tC^ = ((tPEI·PEI) + CIT1200 + TIMP4100 + TW4200 + TFT4400 + TOTH6000)/ABYC

(7a) (ABYC

): (7b)

The capital ATRs shown in equation set (7) differ from the ATR shown in equation (5) in several respects. First, as noted, they exclude the unincorporated sector. Second, they include recurrent taxes on net wealth TW4200, taxes on financial and capital transactions TFT4400, and other taxes TOTH6000, which arguably should be included in a comprehensive capital ATR measure.28 Third, rather than using an over© OECD 2000

39

Tax Burdens: Alternative Measures

all average tax rate on personal income (tP) to estimate the amount of personal tax on income from property (PEI), equation (7) uses an average personal rate on income from property which could be derived as a weighted average of individual taxpayer’s average personal tax rate tPj, with the weights vj = (PEIj/PEI) reflecting the distribution of property income across taxpayers. tPEI = ∑jvj*tPj

(8)

Yet another possibility would be to create a new ATR for “business” that, like the capital ATR shown in (5), would group together incorporated and unincorporated enterprises. In the numerator one would include corporate income tax, as well as personal tax on self-employment income (the numerator would thus include personal tax on the returns to labour and capital provided by the self-employed). Treating OSPUE in this way avoids the difficulties associated with attempting to split the returns between labour and capital. A candidate business ATR measure might be: tB = (tUE·(OSPUE – SSCPUE2300) + CIT1200 + TIMP4120 + TW4220 + TFT4400 + TOTH6000)/(OS – SSCPUE2300) (9a) The business ATR is broader than the corporate ATR shown in equation (2), including OS rather than OSC=(OS-OSPUE) in the denominator, and including personal tax on OSPUE rather than just CIT1200 in the numerator. Moreover, it includes other taxes on business including recurrent taxes on immovable property TIMP4120, recurrent taxes on net wealth TW4220, taxes on financial and capital transactions TFT4400, and other taxes TOTH6000. A narrower business income base (denominator amount), measuring adjusted book income of incorporated and unincorporated enterprises denoted below by ABY, could also be considered: tB^ = (tUE·(OSPUE – SSCPUE2300) + CIT1200 + TIMP4120 + TW4220 + TFT4400 + TOTH6000)/ABY

(9b)

Finally, note that both equations benefit from the use of a tax rate on self-employment income tUE derived from micro-data using equation (6), rather than using an overall aggregate average personal tax rate tp. Corporate tax on foreign source income In countries with residence-based tax systems, CIT1200 will include some amount of net domestic tax on foreign source income, with the corresponding surplus excluded from the denominator, as noted above. Again, micro-data may reveal how important this consideration is. Non-resident withholding tax on domestic source income Another matter to investigate is the treatment of operating surplus paid to non-resident investors. Most countries include non-resident withholding tax on payments of interest, dividends, rents and royalties in category 1200. However, other countries report this amount separately in category 1300 (taxes on income, profits, capital gains that are not allocable between 1100 and 1200). For example, in Canada and Hungary, category 1300 reports non-resident withholding tax. In the case of New Zealand, non-resident withholding tax, which is included as one of several entries in category 1300, is significant – roughly onethird as large as total corporate income tax revenues reported under heading 1200. In Denmark and Greece, the components of category 1300 are not explicitly identified, but again the amounts are large, roughly one-half the size of total corporate income tax revenues. For those countries that report non-resident withholding tax in the Revenue Statistics, this amount should be included in the corporate ATR. Backward-looking (adjusted profit-based) ATRs

40

As is clear from the preceding sub-section, reliance on aggregate operating surplus data to measure an average tax rate for the corporate sector is fundamentally flawed. The construction of meaningful (interpretable) measures requires that corporate income taxes paid be determined as a percentage of economic profit, with consistency sought between numerator and denominator amounts as regards the treatment of interest, and other factors including the treatment of losses and foreign source income. Proper measurement therefore requires access to sufficiently detailed firm-level data. © OECD 2000

Average Tax Rates and the Need for Micro-Data

Similarly, calls for separate tax burden measures for different categories of taxpayer (e.g., low- versus high-income, small business versus large), different factors of production (e.g., capital versus labour), special business activities (e.g., research & development, manufacturing, financing activities) and across different regions and countries require more detailed information and analysis than that offered by nominal tax rates and rates derived using aggregate tax and surplus figures. Typically, such (confidential) data is only compiled by and accessible to government officials. This section briefly reviews a number of considerations relevant to measuring profit-based corporate ATRs, and then considers two recent studies that have relied on financial statement data to generate average corporate tax rates in the European Union. The section concludes by underlining the need for detailed micro-data to properly address the corporate tax burden question.29 Matching corporate (economic) profit and corporate tax paid Adjustments to arrive at a proper measure of economic profit for inclusion in the denominator could include consideration of the following points. First, in order to arrive at a measure of the tax burden on profitable firms, non-profitable firms (i.e., firms making an economic business or non-capital loss) should be excluded from the sample (and thus not enter the denominator, being excluded from the numerator). The consideration was raised in section 4.3 in the implicit tax rate context. Second, attempts could be made to establish whether business cycle effects are distorting results. As noted above, current period claims from large loss-carryforward pools (following a downturn in the economy) may give misleadingly low ATR figures. Similar considerations arise as regards current claims from large pools of other tax allowance and tax credits built-up over time and carried forward to the current year due to low prior-year taxable profits. One possibility in such cases is to derive, for a representative sample of firms and using micro-simulation models, notional corporate taxes payable under restricted loss claims. Many countries have such models at the central government (Ministry of Finance/ Revenue) level permitting an assessment of corporate income tax liabilities under alternative assumptions regarding, for example, revised tax rates, thresholds, and claims from discretionary tax pools.30 If trend/normal loss utilisation rates cannot be determined and factored into revised (notional) corporate tax estimates, ATRs arguably should be calculated over a multi-year period and averaged to arrive at results that smooth out business cycle effects, however the choice of which years to average over may not be obvious. Clearly, the usefulness of this estimate will be reduced when there is instability in the tax code over the sample period. Third, in adjusting financial (book) income to arrive at a measure of true economic income generated in the corporate sector, consideration may be given to adjusting profits for inflationary effects. These include a capital consumption adjustment (i.e., adjusting book profits down to reflect the fact that actual capital depreciation (replacement) costs will exceed those measured using historic purchase prices) and similarly an inventory valuation adjustment (tending to reduce profit). Other possible adjustments include those for corporate debt (i.e., reflecting the decline in real debt burdens accompanying (unanticipated) inflation), tending to increase the measure of economic profit), and similarly an adjustment for losses on non-interest-bearing financial assets including cash and demand deposits (tending to reduce profit).31 Fourth, book deductions for current and deferred taxes should be added back to financial profits, including any deductible state or local taxes. Dividend income (both domestic and foreign) should also be netted out, as should other foreign investment income and foreign branch profits. Domestic dividend receipts should be netted out so as to not double-count domestic profits that are distributed (as recognized under inter-corporate dividend deduction tax provisions). The inclusion of dividends in the denominator would result in an overestimate of domestic economic profit and an underestimate of the true average corporate tax rate. Similarly, foreign dividends should be netted out, as their inclusion would require including in the numerator foreign indirect (corporate) tax on the income underlying the dividend, for which information generally would not be compiled. Moreover, public interest in corporate tax burden measures generally is limited to domestic tax considerations. For a similar reason, foreign branch profits should be excluded. © OECD 2000

41

Tax Burdens: Alternative Measures

Recent corporate ATR results As noted above, ideally profit-based corporate ATRs should be measured using in the denominator adjusted corporate profits which may differ (potentially significantly) from book or financial profits on account of a number of adjustments in respect of inflation, losses, foreign source income and possibly other factors. In practice, available data typically is not sufficiently rich to permit these adjustments, and so approximate measures are derived. Examples are measures constructed with the help of Compustat data (for Canadian and U.S. firms) or the BACH database of the European Commission. Generally, such studies have found that corporate ATRs are lower than is indicated by nominal rates. A recent study commissioned by the Dutch ministry of Finance illustrates the gap between nominal rates of the corporate income tax and the average percentage of profits firms actually pay in tax. The study is based on the annual accounts of nearly 3 000, mainly listed, manufacturing companies located in all EU Member states. It shows that the EU manufacturing sector in 1990-1996 paid 27 per cent in tax on its profits, nearly ten percentage points below the average of statutory rates (36.5 per cent). For German companies, the study reports a gap of 11.5 points, and an effective rate of 38.5 per cent. This is still the highest effective rate found for any EU country, but the gap with other economies has become much smaller than a comparison of nominal rates would suggest: the average rate reported for Italy is 35 per cent, for France 33 per cent, for the Netherlands 32 per cent and for the UK 29 per cent. It should be noted that the study commissioned by the Dutch Ministry of Finance focuses on the manufacturing sector and does not cover other sectors of the economy, such as banking and insurance that are often thought to pay less tax. Also, earnings figures were taken from annual accounts as published by the firms concerned, reflecting differences in national accounting practices. This is especially relevant in the case of countries such as Germany where in many cases a significant part of earnings is added to (hidden) reserves, reducing the denominator which pushes up the average tax rate shown.32 Indeed, in a recent report the Bundesbank has warned that cross-country comparisons – even if based on harmonised annual accounts – are subject to considerable problems, notably stemming from institutional differences in corporate financing, different national accounting regulations as well as statistical and methodological discrepancies in the corporate balance sheet data.33 The Bundesbank publishes statistics on corporations and the taxes they pay, based on the annual accounts of some 60,000 firms. For recent years, this statistics shows an average tax rate of about 35 per cent.34 Since earnings figures used reflect German accounting principles, this average tax rate is much higher than in the case where profit taxes actually paid are related to aggregate business income earned in Germany. The same conclusion seems to apply to other research analysing “tax ratios” of large German companies, as cited in the recent study by the Bundesministerium der Finanzen.35 Current ATR project using micro-data In this chapter it is argued that the calculation of average tax rates using aggregate data drawn from Revenue Statistics and National Accounts raises several potentially significant methodological problems. The Working Party on Tax Policy Analysis and Tax Statistics of the OECD Committee on Fiscal Affairs has commissioned academic researchers with the calculation, verification and comparison of various average tax rates already published by Eurostat (1997), Mendoza et al. (1994) and Jarass and Obermair (1997). Their findings are to be published in a subsequent paper in the OECD Tax Policy Study series.

42

Delegates to the Working Party have also concluded that micro-level data on individual taxpayers might be usefully applied to assess some of the potential pitfalls or “problem areas” with average tax rate analysis based on aggregate data and firm-level data drawn from financial statements which does not contain sufficient detail to permit careful measurement of numerator and denominator amounts. At the time of writing of this report, a number of Delegates are working together with the Secretariat in developing a better understanding of the informational content of ATRs derived for various broad categories of income. © OECD 2000

Average Tax Rates and the Need for Micro-Data

NOTES 7. This problem is particularly acute in the case of capital, where tax base and/or tax credit adjustments tied to the use/purchase of capital may figure prominently in investment decisions and more generally in determining the overall tax burden on capital. Similarly, tax credits linked to earned income or, on the demand side, to payroll may be important considerations in determining employment activity and the tax burden on labour. 8. For example, corporate ATRs, if properly constructed, may be better indicators of the impact of corporate-level tax on investment activity (e.g., locational decisions) if economic rents are being earned. 9. Tax-to-GDP ratios are also backward-looking. Apart from their general limitations discussed in Chapter 3, ratios of corporate income tax (CIT) over GDP provide only limited information on the tax burden of the corporate sector. In particular, such ratios mask changes in corporate tax as a percentage of corporate profit, and changes in corporate profit as a percentage of GDP. See Section 1.3 where it is noted that corporate tax relative to GDP is determined by the product of two ratios: 1) corporate tax divided by pre-tax corporate profit; and 2) pre-tax corporate profit divided by GDP. The first ratio, an average effective corporate tax rate, will vary with changes in the nominal corporate tax rate and with changes in the corporate tax base. Changes in this ratio therefore reflect changes in tax policy, administration, tax planning and compliance. The second ratio, pre-tax corporate profits relative to GDP, will vary with fluctuations in the contribution of corporate profit to aggregate value-added in the economy. Holding tax policy constant (i.e., assuming the rules determining corporate tax rates and tax base are fixed), a drop in corporate profit over GDP would cause the corporate tax-to-GDP ratio to decline. This outcome could be mistakenly interpreted as indicating a reduction in corporate tax on corporate profits, when in fact this value is unchanged. 10. Backward-looking tax rate measures are “average”, as opposed to “marginal”, in the sense that they are not theoretically-based on equilibrium conditions equating marginal benefits and costs, and they consider tax effects on infra-marginal capital units (where economic rents may be earned), and not just tax effects at the margin (where rents are fully exhausted.) Furthermore, they are often termed “effective” as they capture a variety of provisions in addition to the basic statutory tax rate bearing on tax liability. 11. While a number of variants exist, a corporate implicit AETR is measured as: AETRc (OS) = CITAGG/OSC where CITAGG measures aggregate corporate income tax revenues collected in the year in the economy (as reported under category 1200 in Revenue Statistics, OECD), and where OSC measures the operating surplus of the corporate and quasi-corporate sector. 12. In the case of Germany, figures are only available for the operating surplus of the corporate sector plus the unincorporated sector. In this case taxes on profits of the whole business sector can be expressed as a percentage of the total operating surplus of the national economy. 13. Micro-data sets – in particular, personal and corporate income tax databases – which may be used to determine more precise estimates of ATRs on various forms of income, generally offer no guidance to refining estimates of the ATR on consumption. Given the focus in this Chapter on the application of these micro-data sets, we ignore consumption ATRs. 14. The remainder of this chapter follows the practice of indicating the relevant Revenue Statistics tax category number as a subscript to the tax variable. For a list of Revenue Statistics tax categories, see Annex 4.A. 15. For a number of countries, this adjustment could be made using Revenue Statistics data. In particular, Denmark, France, Hungary, Italy, Ireland, Korea, Netherlands, Sweden, Switzerland, the UK and the US all report separately under category 1120 personal tax on (net) capital gains. However, other countries that tax individuals on net capital gains (e.g., Canada) do not separately report this amount, implying that use of micro-data would be required. 16. This data is reported for certain countries in footnotes to the country tables. 17. We assume here an interest in measuring a “source-based” average tax rate, as opposed to a “residence-based” average tax rate. A source-based average tax rate would measure domestic tax on domestic source income, including income paid abroad. In countries that tax their residents on their world-wide income, in principle domestic tax on foreign source income should be netted out of personal and corporate income tax in the numerator of the average tax rate. Note that a source-based average tax rate would not take account of the total amount of tax imposed on domestic value added, as income paid abroad generally is subject to foreign tax. However,

© OECD 2000

43

Tax Burdens: Alternative Measures

18.

19.

20.

21.

22.

23.

24.

25.

26. 27. 28. 44

given the interest in average tax rate analysis in assessing domestic taxation, as opposed to overall taxation, ignoring foreign tax on domestic source income arguably is appropriate. (Note finally that under a residencebased average tax rate, one would want to include in the denominator domestic surplus net of amounts paid abroad, plus foreign surplus receivable by resident taxpayers; in the numerator, the corresponding entries would include domestic corporate tax net of the amount imposed on corporate profit paid/accruing to non-residents, plus domestic tax on foreign source income.) At the same time, results from equation (2a) may be useful to those interested in the average amount of income tax paid by corporations on the surplus generated in the corporate sector (with the bulk of the remainder of tax imposed directly on that surplus (flow) captured in the personal income ATR). Where, for example, the wages and salaries paid by a firm exceed its gross revenue, its operating surplus would be negative – adding this amount to the operating surplus of profitable firms would lower aggregate operating surplus. The general principle to follow is that losses should be factored out, unless the losses of one business activity can be transferred to offset tax on the profits of another business activity. This would include the exclusion, for ATR purposes, of corporations in an overall loss position (whose losses cannot be transferred to a separate corporate entity). Note that where a single firm has both a profitable business activity and a non-profitable one, and combines its income for tax purposes, thereby transferring its losses, the effect of those losses would enter the denominator (lowering OSC) and the numerator (lowering CIT1200) symmetrically, and on this basis the losses should not be carved out. Also note that, of the group of profitable firms (positive book/accounting profits), one may distinguish between tax-paying and non-tax-paying firms. The latter group may be non-taxpaying on account of accelerated write-offs or tax credits, for example. These firms should not be carved out of the sample. While National Accounts figures are reported showing the distribution of total operating surplus (including incorporated and unincorporated enterprises) across industries, the corresponding breakdown for OSPUE is in some cases not provided, implying that the allocation of corporate operating surplus cannot be backed out from these numbers. To see this, recall that operating surplus is measured gross of interest, rent and royalty payments (except for royalties on depreciable tangible property which are treated in the same fashion as depreciation costs and subtracted from gross revenues in arriving at operating surplus), and that such payments are taxed in the hands of the suppliers of the underlying (financial or real) property. Therefore, when measuring an ATR for industry A, if financial or real property used in industry A is supplied by industry B or by the household sector, the numerator of the ATR derived for industry A will exclude tax on surplus paid out in the form of interest/rent/royalty, included in the denominator of the ATR for industry A (with such income taxed in industry B, or in the household sector), implying a mismatch between numerator and denominator amounts. When deriving an aggregate ATR for the domestic corporate sector as a whole that aggregates surplus across all domestic corporations in the denominator, and all the corporate tax thereon in the numerator, this particular problem, at least in relation to rental and royalty payments, generally would not arise, assuming that the suppliers of the underlying property are other domestic corporations. However, the problem will arise in the context of interest payments to households, intermediated by the financial sector, as such income will be subject to personal income tax (excluded from the numerator of the corporate ATR). For countries with residence-based systems, it may be useful to consider an ATR for each sector that includes foreign source income and the domestic tax thereon (i.e., one where the denominator includes foreign branch profit, foreign dividends, interest, rents royalties, and the numerator includes domestic tax on such income). To the extent that such foreign source income bears little/no foreign tax, the ATR may be a close approximation to the overall ATR on such income. The inclusion of social security contributions (SSCF2200) and payroll taxes (PAY3000) paid by employers in the numerator of the labour ATR (an approach which rests on the assumption that such taxes are shifted onto labour) means that the ATR is not a pure “first incidence” measure (i.e., some tax shifting is incorporated). Arguably, categories 4210 and 4220 measuring “recurrent taxes on net wealth” (individual and corporate), which can be denoted by TW4200, and category 6000 measuring “other taxes” which can be denoted by TOTH6000 and which includes taxes on financial institutions and stamp fees, should be included in the capital tax ATR. Note that TOTH6000 is particularly important for countries including Austria, where the tax on financial institutions is roughly one-quarter of corporate income tax (1200); in Japan where TOTH6000 is roughly one-third of corporate income tax; and in Turkey where it is roughly two-thirds of personal income tax and three times corporate income tax. This approach is adopted by Mendoza et al. (1994) and in an internal study by the European Commission (1997). Note that the taxation of labour income of employees (as distinct from owners) of unincorporated enterprises would be captured in the average tax rate on labour income of employees. As a result of the inclusion of taxes on stocks (property and wealth) and certain transactions, these ATRs are not strictly average income tax rate measures, and for this reason, a preference for using an adjusted accounting income measure (ABYC) over a broader operating surplus measure (OSC) may not be as strong.

© OECD 2000

Average Tax Rates and the Need for Micro-Data

29. The use of micro- or taxpayer-level data collected by tax authorities is also not free of problems and limitations. For example, often information on tax-exempt income components is not compiled. Such amounts include taxexempt capital gains and returns earned by pension funds for participants in occupational pension plans. In such cases it is impossible to determine household income following an economic definition of income. Surveys to address this omission are vulnerable because households often misstate income and tax data, or because they are reluctant to provide such information to interviewers. 30. By the term “discretionary tax pools” we refer to pools (unclaimed balances) of prior-year losses, tax allowances (e.g., depreciation allowances) and tax credits, where the timing of the tax claims on these pools is at the discretion of the taxpayer (typically with restrictions limiting the amount of time such claims can be carried from one tax period to the next, before expiring). 31. For a discussion of adjusting financial (book) income for inflation effects, see Holland and Myers (1979). 32. Handelsblatt, 4 May 1999. 33. The Bank compared corporate profitability in Germany, France and the United States over the period 1990–1995, using the BACH database of the EU Commission, concluding that “the informative value of such cross-country comparisons of profitability is much more limited – for methodological reasons – than is commonly assumed”. See: Deutsche Bundesbank, Monthly Report, October 1997, 33-43. 34. Bundesministerium der Finanzen (1999), pp. 10-11. 35. Op. cit., 12-14.

45

© OECD 2000

Chapter 5

MARGINAL EFFECTIVE TAX RATES Introduction As reviewed in Chapter 4, an important distinction can be drawn between backward-looking and forward-looking tax rate measures. Within the forward-looking group, one may consider project-analysis corporate average tax rates and/or marginal effective tax rates which, in principle, are better suited than backward-looking measures in assessing the impact of taxation on investment incentives. This is because investment decisions are inherently forward-looking, based on expectations of future streams of after-tax distributed profits from investment, discounted at a rate reflecting the marginal shareholder’s opportunity cost of funds. Looked at another way, prior-year tax on profits generated on capital acquired in the past may not be representative of tax burdens on a prospective investment. This holds even where tax rules remain unchanged from one year to the next, given the dependency of tax payments on depreciation and loss carryover claims which in turn depend on business cycle and other (non-tax) events. Project-analysis average tax rates (ATRs) are determined by calculating, for a hypothetical investment project, pre- and post-tax corporate profits (losses) in each period over the life of that project.1 Alternative project-types and corresponding income streams may be analysed, including net additions to existing capital stock, and new capital projects involving initial start-up costs yielding negative income in early years. Different project assumptions draw out different tax attributes. For example, in the latter case the tax treatment of loss-carryforwards would be important. In both cases pre- and post-tax corporate profits are discounted to the current period (typically using a discount rate equal to a weightedaverage of the cost of debt and equity capital) to convert the future stream of net profits and taxes thereon into comparable present value amounts. Pre- and post-tax profits on a present value basis are differenced to give the net present value corporate tax burden. Dividing this amount by the present value of pre-tax profits yields a corporate ATR for the given project.2 The corporate tax system influences the corporate ATR through its determination of post-tax profits for a given pre-tax profit stream. Changes in tax law that reduce the present value of after-tax profits (i.e., increase the present value of corporate taxes payable) would increase the measured value of the ATR, indicating reduced investment incentives.3 Marginal effective tax rates (METRs), another forward-looking tax rate measure, assess the impact of taxation at the margin. In analysing the impact of capital taxation, this means the impact of taxation on the last unit of capital invested. In the context of analysing the impact of labour taxation, the focus is on the impact of taxation on the last unit of labour hired. Following the focus of the previous chapter, we focus here on corporate marginal effective tax rates (METR©). Corporate marginal effective tax rates (METR©) are used to assess the impact of corporate taxation on new investment by measuring the magnitude of the tax wedge driven between rates of return pre and post corporate tax.4 Corporate METRs factor in the nominal corporate tax rate applied to business profits, and in addition may factor in certain special incentives delivered through the tax law in support of the purchase of capital. METR© statistics also address the effect of taxation on the cost of finance (the cost of debt and/or equity funds used to purchase capital), taking into account the tax-deductibility of interest expenses, and incorporate either explicitly or implicitly assumptions regarding the influence (or not) of shareholder-level taxation (e.g., imputation relief, capital gains taxes) on the setting of the firm”s discount rate used to discount future income streams and net tax liabilities tied to the last unit of capital acquired. © OECD 2000

47

Tax Burdens: Alternative Measures

To those not familiar with the neo-classical investment paradigm from which METR© statistics are derived, the interpretation of METR statitistics may not be readily apparent. Indeed, too often studies using this form of tax rate analysis tend to obscure rather than clarify the underlying assumptions and model structure. This chapter attempts to explain METR© analysis in a step-by-step “user friendly” fashion in an effort to shed light on the basic concepts and address the “black box” reputation of such measures. The technical analysis is followed by a review of key underlying assumptions and data limitations, to be kept in mind when interpreting METR results, especially when used to influence policy debate. Defining marginal effective tax rates A marginal effective tax rate derived for income from capital is a measure of the distortion imposed by the income tax system on a “marginal investment”, defined as one that generates a return that is just enough to cover all the costs associated with the investment, and no more. In general, the marginal effective tax rate on income from capital used in production (e.g., plant, machinery, equipment) is derived as follows: METR = (Rg – Rn)/Rg

(1)

where Rg denotes the before-corporate tax real rate of return (net-of-depreciation) on a marginal investment, and Rn denotes the after-personal tax real rate of return on savings used to finance that investment. The term (Rg – Rn) is referred to as a marginal tax wedge (MTW.) This wedge measures the divergence between pre-tax and post-tax returns imposed by corporate and personal income taxation. More specifically, the wedge measures the difference between the real before-tax, net-of-depreciation rate of return earned by the firm on the last unit of capital installed, and the real after-tax rate of return to the saver measured after all corporate and personal income tax payable on that return. Dividing this wedge by the pre-tax rate of return determines the marginal effective tax rate. For example, if the pre-tax return on the last dollar of capital installed is $0.10, and the corresponding after-tax return to the saver is $0.06 implying a tax wedge of $0.04, then the marginal tax rate is 0.4. The METR measure in equation (1) assesses the combined impact of the corporate and personal tax systems on domestic investment incentives. In the closed-economy case, where domestic investment funds are provided by domestic savers alone, it is not (strictly) possible to measure separately the incentive effect of the domestic corporate tax system and the domestic personal tax system on domestic investment. This inability to isolate the incentive effect of the corporate tax system on investment arises because the cost of financing investment depends on the personal tax system in the closed-economy case, given that domestic savers are the sole source of finance – equilibrium is established only when domestic saving equals domestic investment. The small open-economy case In the small open-economy case, where firms have unfettered access to international capital markets, the investment incentive effect of the corporate tax system can be assessed independently of personal tax considerations (which affect savings alone). In particular, the marginal tax wedge (MTW) can be broken down into two parts: MTW = (Rg – R) + (R – Rn)

(2)

where Rg and Rn are defined as above, and R denotes the real “world interest rate” at which firms can access financial capital on international capital markets, taken to be the real before-personal-tax rate of return available to domestic savers. The first term in equation (2) calculates the tax-wedge created by the corporate tax system, which measures the investment incentive effect of the domestic tax system. The second term calculates the tax-wedge created by the personal tax system, which measures the savings incentive effect of the domestic tax system. This chapter focuses on the small open-economy case and reviews the determinants of the wedge created by the corporate income tax system (Rg – R), divided by the pre-tax return Rg, in measuring the corporate marginal effective tax rate: 48

METR© = (Rg – R)/Rg

(3) © OECD 2000

Marginal Effective Tax Rates

where the METR© value has the following interpretation:

METR© value

Impact on investment

METR© < 0 METR© = 0 METR© > 0

Encouraged No impact (neutral) Discouraged

In the case where METR© < 0, the tax system on balance subsidises capital investment. This occurs where the present value of tax deductions and credits earned on investment exceeds the purchase price of the asset. In the case of a non-zero METR©, the greater is the absolute value of the METR© (i.e., the greater the deviation of METR© from zero), the greater is the inferred impact of the tax system on investment. In principle, there are a large number of marginal effective tax rates that one may want to calculate, varying by type and size of firm, type of investment decision, source of finance, and type of saver. Since it is not feasible to calculate a separate rate for all the possible margins that may exist in the economy, some aggregation is inevitable. Firms are generally aggregated at least to broad industry categories (e.g., manufacturing, trade, services, etc.), capital goods are usually aggregated to machinery, building and inventories, while sources of finance may or may not be aggregated. Before examining the construct of a METR© that takes into account various corporate tax parameters impacting on the net benefits and costs of investment at the margin, it is useful to first consider the basic underpinnings of the model in the no-tax case. The analysis considered below calculates the METR© for depreciable capital (e.g., machinery, buildings) under the small-open economy assumption that domestic investment has no influence on the rates at which funds can be obtained, and the firm’s financial cost of capital is determined independently of personal tax rates. After analysing the no-tax case in Section 5.3, the behavioural impact of a simplified corporate income tax system is considered in Section 5.4 in order to isolate the basic underpinnings of the tax-inclusive METR model. In Section 5.5 a METR© is derived for a more generalised corporate income tax system that includes two stimulative corporate tax instruments. Analysis of the no-tax case Absent a corporate income tax, it can be shown that in the small open-economy case and under conditions of perfect competition, a profit-maximising firm will adjust its physical capital stock each period to ensure that net revenues from installed capital at the margin (as measured by Fk) are sufficient to cover the cost of financing that capital (measured by Rf) and the cost of replacing the fraction worn out or depreciated during production (measured by δ): Fk = (Rf + δ).

(4)

It is important to note that while the firm just breaks even on the last unit of installed capital, economic rent (profit) may be earned on infra-marginal units. Under the equilibrium condition given by equation (4), the firm would not choose to expand its capital stock beyond the break-even point (to avoid losses on further investment). In the no-tax case the firm’s real cost of funds Rf equals the real required rate of return on equity demanded by shareholders R, Rf = R = i – π.

(5)

The required rate of return that firms must pay shareholders in order to induce them to hold the firm”s equity, measured by R, is given by the nominal interest rate on an alternative investment (e.g., bonds) of equivalent risk, denoted by i, less the general inflation rate π. © OECD 2000

49

Tax Burdens: Alternative Measures

The equilibrium condition (4) is intuitive and states, as noted above, that it is optimal for a firm to invest in physical capital just up to the point where the marginal net revenue from an additional unit of installed capital (Fk) equals the marginal cost measured by (Rf + δ) and more formally referred to as the user cost of capital.5 The user cost of capital, which is the implicit rental cost of using a currency unit (e.g., dollar) of capital for one period consists of two parts – the first part is the real cost of finance Rf which results from the payments the firm must make on funds raised to purchase the physical asset, and the second is the capital consumption cost (δ) which results from the loss in value of the capital asset due to depreciation.6 Note that the required before-corporate tax, net-of-depreciation, real rate of return earned by a firm on a marginal investment, denoted by Rg, is determined by the following: Rg = Fk – δ

(6)

where Fk measures the increase in net revenues (revenues less operating costs) generated by a unit increase in its physical capital stock denoted by k. A key assumption in the analysis is that Fk declines as the capital stock increases. Fk is measured indirectly from theoretical conditions characterising the investment behaviour of a profit-maximising firm which suggest that investment is carried out just up to the point where the marginal benefit of a dollar’s worth of capital per period equals the cost of holding the dollar of capital for a period (see equation (4)). Part of Fk is absorbed to maintain the real value of the physical capital stock, which is assumed to depreciate (i.e., decline) in value due to wear and tear, and obsolescence7 – the symbol δ denotes the exponential (declining-balance) rate of economic depreciation. In the closed-economy case, part of the residual given by (Fk – δ) goes to the revenue authorities in the form of corporate and personal income tax (as measured by (Rg – Rn)), and the rest goes to the saver (as measured by Rn). In the open-economy case, part of the residual given by (Fk – δ) goes to the revenue authorities in the form of corporate income tax (as measured by (Rg – R)), and the rest goes to the domestic or foreign saver (as measured by R). Market equilibrium in the no-tax case is illustrated in Graph 5.1, at E0. The investment schedule Ia shows the net of depreciation rate of return (Fk – δ) at different investment levels. The schedule is downward-sloping under the assumption that Fk decreases as the capital stock increases. The investment schedule is derived from the neo-classical theory of the firm, which assumes that the firm”s managers will undertake an investment only if it increases the market value of the firm’s equity. This assumption is satisfied if the incremental unit of capital, when added to the firm’s productive capacity, provides a stream of real net returns that is sufficient to cover all of the costs associated with the investment.

Chart 5.1.

Return on investment in the no-tax case

Rg, R

Rg = R

E0

Ia

I0

50

Source:

I

OECD.

© OECD 2000

Marginal Effective Tax Rates

The supply of funds schedule, horizontal at the real world interest rate Rf, illustrates the small openeconomy assumption that capital requirements of the domestic economy are too small to influence the market rate R at which domestic firms acquire capital. Market equilibrium is established at E0 where Ia intersects the supply of funds schedule. The following tables summarise the rates of return at E0. As regards the calculation of the marginal effective tax rate, from equations (3)-(6): METR© = (Rg – R)/Rg = (Rf – R)/Rf = 0

(7)

The following table summarises the relevant rates of return and the obvious finding of a zero marginal effective corporate tax rate in the no-tax case.

Summary of statistics (no-tax case) (Rf + δ) where Rf = R

User cost of capital

Gross rate of return, net of depreciation (Rg) Saver’s required real rate of return (real cost of funds) METR©

Rf R 0

Analysis of a simple corporate income tax Consider now introducing a corporate income tax into the framework, where corporate profits are taxed at rate (u) and physical capital can be depreciated (written-off) for tax purposes at a decliningbalance rate (α) equal to the physical depreciation rate (δ) and depreciable costs are indexed to the general rate of inflation. In the presence of such a corporate tax system, it can be shown that the representative firm will adjust its capital stock each period up to the point where: Fk(1 – u) = (Rf + δ)(1 – uδ/(Rf + δ))

(8)

To understand the last term in this equation, consider the following table which shows the stream of capital cost allowances associated with a dollar investment in period t, depreciated for tax purposes at a declining-balance rate (α), assumed to equal the economic depreciation rate δ. In the table, UCC denotes the undepreciated amount of capital cost for tax purposes.

Illustration of value of capital cost allowance stream Period

t t+1 t+2 t+3 Etc.

Beginning period UCC

Capital cost allowance (CCA)

Tax value of CCA

Present value at beginning of t of stream of CCA

End of period UCC

1 (1 – α) (1 – α)2 (1 – α)3

α α (1 – α) α (1 – α)2 α (1 – α)3

uα uα (1 – α) uα (1 – α)2 uα (1 – α)3

uα/(1 + Rf) uα (1 – α)/(1 + Rf)2 uα (1 – α)2/(1 + Rf)3 uα (1 – α)3/(1 + Rf)4

(1 – α) (1 – α)2 (1 – α)3 (1 – α)4

Summing the cells in the fifth column of the above table gives the following expression for the present value to the firm, at the beginning of period t, of the stream of capital cost allowances associated with a dollar investment made in period t: PVt = uα Σt(1 – α)x – t/(1 + Rf)x – t + 1

(9a)

The summation term in equation (9a) reduces to (1/(Rf + α)). If the depreciation rate for tax purposes (α) equals the economic depreciation rate δ, the present value term equals: PVt = uδ/(Rf + δ) © OECD 2000

(9b)

51

Tax Burdens: Alternative Measures

The discount rate (Rf + δ) includes the real cost of funds (Rf) rather than the nominal cost of funds (Rf + π), where π measures the general rate of price inflation, under the assumption that the capital cost allowance system is indexed to inflation. If the capital cost allowance system is not indexed, the discount rate would be (Rf + π + δ) which exceeds (Rf + δ) by the amount of the inflation rate, capturing the fact that the present value to the firm of capital cost allowances in a non-indexed system falls as the rate of inflation increases. Note finally that the discount rate includes the capital cost allowance depreciation rate, which captures the fact that the pool of capital cost allowances generated by a dollar of investment declines over time, at the declining-balance rate, as capital cost allowances are claimed (used up). The investment equilibrium condition given by equation (8) differs from that given by equation (4) for the no-tax case, in two important respects. First, because the income from the sale of output is taxed at rate u, the after-tax benefit to the firm from employing an additional unit of capital falls from Fk to Fk(1 – u). Second, with the firm able to deduct from its taxable income, over future periods, a stream of capital cost allowances on its investment, the cost to the firm of an additional dollar’s worth of capital falls by (uδ/(Rf + δ)) dollar/cents, equal to the present discounted value of the capital cost allowance deductions associated with the capital expenditure. In Graph 5.2 which illustrates this case, the investment schedule Ia shows the pre-corporate tax rate of return (Fk + δ) at different investment levels, while the investment schedule Ib shows the aftercorporate tax rate of return (Fk + δ)(1 – u) at different investment levels. The METR© in this stylised case can be explained as follows. In order to pay shareholders their required rate of return R, the firm must earn a before-corporate-tax rate of return of Rf/(1 – u) where Rf = R, of which (u/(1 – u))Rf is paid in corporate income tax, leaving Rf = R dollars in the hands of the shareholders. With the introduction of the corporate income tax, only domestic investment is reduced. Domestic savings are unaffected under the small open-economy assumption that the capital market can clear without domestic investment being equal to domestic savings.

Chart 5.2.

Return on investment with a simple corporate income tax

Rg, R Rg

R

E0 E1

Ia Ib Il Source:

52

I0

I

OECD.

The following table summarises the relevant rates of return, and the corporate-tax wedge METR© at the new equilibrium E1 in which domestic investment is reduced, while domestic savings (not shown) remain unchanged. © OECD 2000

Marginal Effective Tax Rates

Summary of statistics (simple corporate income tax case) User cost of capital

(Rf + δ)(1 – uδ/(Rf + δ)) = Rf + δ(1 – u) where Rf = R

Gross rate of return, net of depreciation (Rg) Saver’s required real rate of return (real cost of funds) METR©

Rf/(1 – u) R u

Analysis of targeted corporate tax instruments Now consider an extended version of the small open-economy METR© model that includes both debt and equity finance, as well as two possible stimulative tax instruments – an investment tax credit, and an accelerated capital cost allowance. Special corporate tax instruments – an investment tax credit on purchases of physical capital, earned at rate ψ; – an accelerated declining-balance depreciation allowance, claimed at rate α > δ – the capital cost allowance system is not indexed to inflation. The real cost of financial capital is first derived below. Then the optimal investment condition is characterised in the presence of the investment tax credit and accelerated capital cost allowance. Measuring the cost of finance Rf In order to purchase a physical capital asset, a firm can raise financing from two main sources: debt finance (borrowing, issuing bonds), and equity finance (retained earnings, new equity shares.) Let the nominal rates of return payable to savers on corporate debt and equity, as determined on world capital markets, be denoted by i and ρ, implying a real weighted-average rate of return to savers of: R = βi + (1 – β)ρ – π

(10)

where the weights (β) and (1 – β) correspond to the fraction of capital raised through debt and equity respectively, and π is the general inflation rate. Given the deductibility of nominal interest payments from taxable income, the firm”s real after-tax cost of financial capital can be expressed as:8 Rf = βi(1 – u) + (1 – β)ρ – π

(11)

where u is the nominal corporate income tax rate. Analysis of investment equilibrium The availability of an investment tax credit and accelerated capital cost allowance reduce the cost of acquiring physical capital, and thus influence the value of Rg. The investment tax credit reduces the purchase price of a dollar of capital to (1 – ψ) dollars. Consider now th present value to the firm of tax deductions generated by a (non-inflation adjusted) accelerated capital cost allowance system: (1 – ψ)uα/(Rf + π + α)

(12)

where the stream is discounted at a nominal funds rate (Rf + π) rather than the real rate Rf, due to the nonindexation of the capital cost allowance system. The (1 – ψ) term in equation (12) reflects the assumption that the firm is only allowed to write-off the cost of the investment measured net of the tax credit earned on the purchase (on the principle that the investment tax credit reduces the effective acquisition price to the firm of a dollar’s worth of capital to (1 – ψ) dollars.) Where only the historic (purchase) dollar value of an asset can be depreciated for tax purposes (non-inflation adjusted tax depreciation system), the real value of the permitted deduction declines at the rate of inflation π. The combination of the investment tax credit and accelerated capital cost allowance reduces the effective acquisition price, measured in dollars (or other currency units), from one to: (1 – ψ – (1 – ψ)(uα/(Rf + π + α)) © OECD 2000

(13)

53

Tax Burdens: Alternative Measures

Given the cost of finance and the corporate income tax structure outlined above, it can be shown that the representative firm will adjust its capital stock k each period up to the point where: Fk(1 – u) = (Rf + δ)(1 – ψ)(1 – uα/(Rf + π + α))

(14)

The user cost of capital, shown on the right side of the expression, gives the cost to the firm of holding a (constant) dollar of capital for one period – this cost equals the real financial and economic depreciation cost of holding a dollar’s worth of capital for one period, given by (Rf + δ), multiplied by the after-tax cost of acquiring a dollar’s worth of capital, as given by (1 – ψ)(1 – uα/(Rf + π + α)). Dividing by (1 – u) and subtracting δ gives the required before-corporate-tax rate of return to capital: Rg = ((Rf + δ)(1 – ψ)(1 – uα/(Rf + π + α))/(1 – u)) – δ

(15)

The following table summarises the rates of return determining the METR© statistic that measures the corporate tax wedge determining the investment distortion in the extended small open-economy case.

Summary of statistics (extended small open-economy model) User cost of capital Gross rate of return, net of depreciation Saver”s required real rate of return METR©

(Rf + δ)(1 – ψ)(1 – uα/(Rf + π + α)) where Rf = βi(1 – u) + (1 – β)ρ – π ((Rf + δ)(1 – ψ)(1 – uα/(Rf + π + α))/(1 – u)) – δ R = βi + (1 – β)ρ – π (Rg – R)/Rg

The METR© framework incorporates into a single summary statistic the competing influences that these corporate tax considerations have on the incentive to invest, as summarised below.

Assessing the impact of corporate tax incentives targeted at investment Increase in tax parameter

Transmission mechanism

Impact on Rg and METR©

Impact on investment

Corporate tax rate (u)

Decreases net return on business profits Decreases cost of debt finance Increases value of capital cost allowance Decreases effective cost of physical capital Decreases effective cost of physical capital

Increases Decreases Decreases Decreases Decreases

Decreases Increases Increases Increases Increases

Capital cost allowance rate (α) Investment tax credit rate (ψ)

54

These METR results are dependent on the small open-economy assumptions that 1) investment decisions of domestic firms are influenced by corporate income tax provisions; 2) are not influenced by personal income tax provisions; and 3) that savings decisions of domestic householders are influenced by personal income tax, but not by corporate income tax. The level of savings and investment in the small openeconomy case are depicted in Graph 5.3 for the case where the economy is a net importer of capital. Graph 5.4 considers the net capital export case. In the net capital import case, and absent any tax with the level of investment in the economy given by I0 and the level of savings S0, the difference between domestic investment and domestic savings in the amount of (I0-S0) is financed by inflows of foreign capital. With the introduction of a corporate tax, investment declines from I0 to I1 (under the assumption that the pre-tax rate of return Rg exceeds the world interest rate R).9 However, domestic savings are not affected because domestic households can continue to earn the same rate of return on their savings. With the introduction of personal tax (not considered in Graph 5.3), domestic savings would decline below S0, but domestic investment would not be affected, since the reduction in savings could be made up through an increased net inflow of capital. © OECD 2000

Marginal Effective Tax Rates

Chart 5.3.

Return on investment in the net capital import case

S(R)

Rg E0

R

E1

I(Rg)a

I(Rg)b S0 = S1 Source:

I1

I0

I, S

OECD.

For the net capital export case illustrated in Graph 5.4, in the no-tax situation investment is again given by I0, savings is S0, and the excess of domestic savings over domestic investment (S0-I0) represents savings invested in foreign assets. Investment falls from I0 to I1 with the imposition of a corporate tax (assuming again that Rg exceeds R), and savings are not affected. Introducing a personal tax in this case (not shown) would reduce savings below S0, but domestic investment would not be affected, as firms could continue to acquire financing at international rates.

Chart 5.4.

Return on investment in the net capital export case

S(R) Rg

E0 R E1

I(Rg)b I(Rg)a I1 Source:

OECD.

© OECD 2000

I0

S0 = S1

I, S

55

Tax Burdens: Alternative Measures

Limitations of METR analysis This Section considers marginal effective tax rate analysis, with an eye towards critically assessing its usefulness as a guide to the formulation of tax policy. The focus here is on METRs applicable to (physical) capital investment.10 METRs have enjoyed considerable support as summary statistics of the combined interaction of a range of tax parameters linked to investments decisions, and as indicators of how tax provisions compare over time, across industries and across countries. This widespread interest has encouraged many specialists to devote time and effort to measuring, cataloguing and promoting METRs as benchmark statistics to guide tax policy making. Given this development, it is important to reflect on the many caveats linked to the underlying framework, in order that METR results and their use can be placed in proper perspective. The issues reviewed below suggest that METR analysis can be helpful if not pushed beyond its limits. METRs should be seen as rough proxy variables that summarise at a broad level the interaction of various tax rules relating to capital investment. METRs also provide a useful framework for identifying the various channels through which tax policy might be expected to influence investment behaviour – through the taxation of returns from investment; through the impact of tax deductions and credits on the effective purchase price of additional units of capital; and through the possible effects of corporate and shareholder-level taxation on the cost of funds (financial capital). However, METRs do not offer a definitive assessment of the impact of tax policies on actual investment flows or capital stocks. A number of the key assumptions typically invoked are untenable in many instances, thus making comparisons across sectors or countries difficult. Also, the measurement of METRs is typically fraught with data and aggregation problems, so that even if the underlying assumptions hold, the resulting statistics may be off the mark. Moreover, owing to the static partial equilibrium framework from which METR statistics are derived, METR analysis by itself is incapable of assessing investment and capital stock responses to tax changes, factor substitution possibilities, timing issues, distributional effects, tax-planning responses, and a host of other areas that must be addressed when assessing alternative tax policy settings. Careful consideration of the range of conceptual and data problems touched on in this chapter suggests that METR statistics and comparisons across sectors or countries cannot be used confidently as an indicator of the influence of taxation on investment or as a guide to the setting of tax policy. The goal of this Chapter is to provide background material on some of the caveats associated with METR analysis in order to assist in the development of a consensus on the usefulness of METR statistics as guides to assessing tax policy positions and options. The development of a consensus view is important to ensure that accurate and consistent weight be attached to such measures in discussions of tax policy within and outside the OECD. The following Subsection 5.6.1 reviews key assumptions that underlie conventional METR analysis, and questions whether these assumptions can be assumed to hold in all cases. To the extent that they cannot, caution must be exercised in relying on METR statistics as a basis for comparing the net influence of tax policy on investment incentives across sectors and countries. In Subsection 5.6.2 a range of problems imposed by data limitations and aggregation techniques is surveyed, which further caution against the use of METR statistics for tax policy analysis purposes. Robustness of underlying assumptions This subsection critically reviews six assumptions underlying METR analysis. Perfect competition and absence of economic rent

56

Most models used to calculate marginal effective tax rates assume that the representative firm operates in perfectly competitive markets, and takes output prices as given. While this assumption may hold in certain sectors, and in some countries, it will not hold in most cases most of the time. Firms often enjoy a degree of monopoly power, for example. This suggests inclusion in the denominator of the METR statistic of a measure of the elasticity of the representative firm’s investment demand schedule.11 However, © OECD 2000

Marginal Effective Tax Rates

data on this elasticity, and how it varies across sectors, countries and time would require access to detailed firm data which are generally unavailable to analysts. The model also assumes that firms invest in capital up to the point where the after-tax marginal benefit from the last unit of installed capital just equals the after-tax marginal cost. This result is based on the assumption first that the firm’s investment plans are generated by the managerial goal of maximising shareholder equity, and second that investment capital is infinitely divisible. In a number of cases, a firm’s investment plans may not be driven by value-maximising behaviour. Other management objectives may be at work, and/or a firm may be earning significant (pure) economic rents – owing to barriers to entry into the industry or other factors – such that investment plans are not pushed up to the margin where project net benefits equal net costs at the margin. Moreover, investment projects will usually be “lumpy” (i.e., not infinitely divisible). In such cases, investment in capital may proceed up to a point where economic rents are being earned at the margin (and these rents are not realised, as the acceptance of an additional investment project, requiring a large discrete increase in the capital stock, may push net project benefits below net project costs). In such cases where economic rents are earned at the margin, (relatively small) variations in METRs – generated by variations in income tax rates, investment tax credits, and/or other METR parameters – would not be expected to influence the level of investment as conventional METR analysis assumes. Declining marginal productivity of capital The neo-classical investment theory underlying METR analysis assumes that the marginal product of capital, that is the additional amount of output generated from an additional unit of installed capital, declines as the size of the overall capital stock increases. This yields an equilibrium result where firms invest just up to the point where the after-tax value of additional output generated by investment at the margin just equals its after-tax cost. The assumption of decreasing returns at the margin is adopted in most METR models examining investments in real physical capital, R&D and human capital. However, this assumption may not hold across all sectors, countries and time. New insights from recent work in growth theory, for example, characterise knowledge capital as generating increasing returns in the production of output at the margin.12 Inputs of physical and human capital generate knowledge, and knowledge helps produce additional human capital. The possibility that the rate of return on capital may increase rather than decrease with increases in the capital stock, in this case knowledge capital, suggests that the extension of METR analysis to investments in knowledge capital may be inappropriate and provide a misleading indicator of tax effects.13 Financial structure and market arbitrage conditions One of the most difficult and contentious areas in METR analysis involves the choice of the financial market arbitrage assumption to employ in the determination of the representative firm’s cost of funds (r). Complications arise because income tax systems treat different types of finance differently, and different types of savers differently. Moreover, empirical uncertainty surrounds the extent to which tax burdens (or tax relief) tied to financial returns are borne by firms versus savers, and the extent, manner and timing in which differential tax treatment of financial assets is arbitraged away at the corporate or individual investor level. These are very much unresolved empirical issues, and one would expect that the answers vary depending on the time period, the country and even the sector being examined. Many arbitrage assumptions have been used in the literature. While these will not be exhaustively reviewed here, a brief consideration of possible arbitrage assumptions under the so-called “fixed-r” approach is illuminating.14 The fixed-r approach assumes all projects earn the same after-corporate tax rate of return. For any given saver, this means that all (domestic) projects yield the same after-personal tax rate of return. Differences in the tax treatment of investment income across different investors imply that after-personal tax rates of return will vary across savers. Within the fixed-r framework, several possibilities exist. One approach is to assume that arbitrage at the level of the firm equates the after-tax cost of debt finance with the (non-deductible) cost of equity finance, without distinguishing between retained earnings and new share issues.15 Another is to assume © OECD 2000

57

Tax Burdens: Alternative Measures

that dividends have some intrinsic value to shareholders (perhaps offer a signalling function), and that firms trade off these intrinsic benefits against the tax cost of dividends compared with share repurchases, increasing dividend pay-out until they are indifferent at the margin between new share issues and retained earnings as a source of finance.16 Yet another approach is to assume that rates of return on retained earnings and new share issues are derived such that each alternative yields the same after-personal tax rate of return as that available on debt. Where tax treatment differs across shareholders, as for example between resident and non-resident shareholders, it is necessary to identify a “marginal shareholder” whose tax treatment is relevant in the determination of the after-corporate tax cost of funds. The identification of such a shareholder is not always evident. Having chosen a representative shareholder, one might appeal to the “new view” of dividend taxation (see Annex 2.B) which posits that dividend taxes are irrelevant to the determination of the cost of capital. Instead, one could use retained earnings to represent the marginal source of equity finance. Or one might choose to use a weighted average of the costs of retained earnings and new share issues. In these latter cases, one typically assumes that the after-tax return on bonds should be used as the benchmark opportunity cost. However, even under this assumption, one must choose a given bond and corresponding interest rate, recognising that bond rates will vary by term and risk, with theory offering little guidance. Another complicating factor is that firms, to varying degrees across different sectors and countries, are relying increasingly on new financial (derivative) products, while most METR studies account only for conventional bonds and equity, thus missing the most active part of financial markets. Moreover, the factors used to weight different costs of finance (either in the determination of an overall cost of funds, or in weighting individual METRs each based on a single financing instrument) are typically based on historic average data, which in certain cases may be unrepresentative for forward-looking marginal investment projects. Choosing a particular arbitrage assumption and holding it fixed across different METR calculations is not problematic if the exercise is one of attempting to summarise differences in the tax treatment of different investment projects, under the given set of arbitrage and other assumptions. However, as with the other assumptions and parameters entering a METR statistic, results that suggest that investment incentives in one project-type are more or less tax-distorted than those in another project-type are only as robust as the underlying assumptions themselves. This follows from the fact that the presence (or not) of various tax parameters in the cost of funds expression depends uniquely on the arbitrage assumption. Where the correct arbitrage condition varies across time, or by country or by sector, METR statistics that use a fixed arbitrage assumption across cases will produce inaccurate investment incentive comparisons across time, countries and sectors. Loss offsetting METR analysis implicitly assumes that tax systems treat revenues and losses symmetrically, or in other words provide full or perfect loss offsetting. For this symmetry to hold, governments must provide taxpayers with full refundability of negative tax liabilities (or some equivalent, such as a carry-forward of tax losses and unused tax credits with interest). In practice, this is never the case. Tax systems generally provide only imperfect loss offsetting. While most corporate tax systems contain carry-back and carry-forward provisions for losses, these typically have a limited duration, and the carry-forward of losses is without interest. Moreover, the corporate tax rate applicable to a carry-back or carry-forward may differ from the current corporate rate. In principle, capital cost allowance and investment tax credit parameters used to determine the net cost of physical capital should be adjusted to account for tax loss situations so that they reflect the (expected) present discounted value of the tax relief they represent. Similarly, interest deductions should be discounted in arriving at the cost of debt finance. On the revenue side, the corporate tax rate applicable to marginal revenues should be discounted in loss carry-forward situations.17 These adjustments work in opposite directions, and therefore a priori it is uncertain whether imperfect loss offsetting increases or decreases effective tax rates.

58

Failure to account for imperfect loss offsetting means that METRs will be biased, either positively or negatively, and to a degree which will vary across sectors, countries and over time as the tax loss position of firms (including the build-up of pools of tax losses) varies across sectors, countries and over time. Loss positions will be influenced by a host of factors including the percentages of firms in start-up versus mature states, differences in the timing of business cycles and exposure to economic shocks, and differences in current and past © OECD 2000

Marginal Effective Tax Rates

availability of investment tax incentives (which in many countries largely account for tax losses). Another source of bias stems from the fact that countries differ in terms of their rules governing loss transfers within corporate groups. The limited amount of information on such factors suggests that the net effect of these distortions could be significant. For example, in the mid-1980s, it is estimated that Canada had a much higher proportion of companies not paying taxes as compared with the United States.18 In particular, roughly 50 per cent of investment in Canada and 80 per cent of investment in the United States was conducted by firms that were fully taxpaying during that period, with the amounts varying significantly across different industry sectors. Treatment of risk (uncertainty) Closely related to the issue of loss-offsetting is the treatment of risk. METR analysis often employs static expectations and assumes perfect certainty. This assumes that firms expect that the future values of all parameters in the METR expression, including tax rates, will remain unchanged from their current values, and that these values are expected with certainty. Many would agree that these assumptions are counter-factual. Frequent changes to the tax code attest to the fact of the non-static nature of tax rates and parameters, and moreover future tax changes are often announced. Also, future movements of output and input prices, and interest rates and inflation rates are generally uncertain, suggesting that firms would consider a range of probable values for these parameters, rather than assigning a fixed single value to them, when assessing their investment incentives. This is particularly the case when the project is longer term and the capital commitments are not easily reversible; more on the issue of capital irreversibility in what follows. The literature suggests that METR analysis should distinguish between different types of risk, including income risk and capital risk.19 Income risk refers to uncertainty regarding future net revenues arising from the stochastic movement of output and current input prices or demand faced by the firm. Capital risk refers to uncertainty regarding the economic rate of depreciation of installed capital, due either to an unknown future purchase price of capital or to a stochastic rate of physical depreciation or obsolescence. If the tax system grants full loss offsets or its treatment approaches that, it may not be necessary to adjust METR expressions for income risk, as the government in this case shares equally in the profits and losses of the company – for example sharing 35 per cent of the profits and 35 per cent of the income risk, assuming a 35 per cent corporate income tax rate. The cost of bearing income risk is thus implicitly fully deducted under a full loss offset tax system, with no additional tax distortions being introduced for income-risky investments versus comparable riskless investments. The situation is however different for capital-risky investments. In most countries, tax depreciation allowances are based on the original cost of the asset, and thus do not change with unanticipated changes in the market value of installed capital, as can occur with unanticipated technological change or unanticipated tax changes. The implication is that, even where negative taxes are refunded, the tax system does not provide deductions of the full cost of bearing capital risk and thus the METR expression should take into account this tax distortion on risky assets. Capital risk may be accounted for in theory by increasing the economic rate of depreciation (without an equivalent increase in the tax depreciation rate.) However, in practice it is difficult to measure the risk premium associated with capital risk. Some have argued that a firm’s market value equals its asset value, and so fluctuations in market value reflect changes in the value of underlying assets.20 This view suggests that capital asset pricing models (CAPMs) could be used to assess capital risk premium. However, other work has shown that it is the correlation between the economic cost of depreciation and consumption that is relevant, and that this correlation could be negative, implying that CAPM estimates for capital risk premiums would be inappropriate.21 Capital irreversibility In addition to ignoring the complications introduced by uncertainty, most METR analysis implicitly assumes that capital investments are reversible in full and without cost. This characterisation is clearly inappropriate for most types of capital. Capital will often be task or industry specific, and where it is not, © OECD 2000

59

Tax Burdens: Alternative Measures

the costs of removing it from a given installation and transferring it to another may be significant. If the conversion of capital to alternative uses is extremely costly, if not impossible, then the capital investment is said to be irreversible. Relatively little research has been undertaken to determine the implications of irreversibility for the measurement of capital demand and METRs, despite the apparent importance of this factor.22 The available evidence examining the implications of capital irreversibility suggests that the METR on irreversible capital can be significantly higher than that on fully reversible capital, depending on the level and type of risk, and is generally increasing in income and capital risk.23 The implication is that ignoring this feature would tend to bias cross-sector or cross-country METR results, as a result of different risk characteristics across different investment cases. Data limitations The preceding subsection reviewed conceptual problems associated with METR analysis. This section discusses limitations to METR analysis arising from data problems. Perhaps the most central data problem encountered arises from the fact that many of the required variables are unobservable due to the forward-looking nature of investment decisions. Another is the modelling uncertainty over underlying production technologies of the “representative” firm. Yet another arises from the fact that aggregation is inevitable, given that firm-level data are generally not available. The following briefly reviews some of the difficulties in obtaining representative values for key parameters in standard METR equations for capital investment. Corporate-level tax parameters METR expressions for capital investment generally include a corporate income tax rate and a term measuring the present value of tax depreciation allowances claimed on a unit of investment at the margin. Investment tax credit rate(s) may also enter, if applicable. Typically, current nominal corporate tax rates, capital cost allowance and investment tax credit rates are applied, despite the fact that expectations over the future values of these rates may not be static, and full loss offsetting generally is not provided. The problem facing the model builder is that the tax (and price) parameter values entering the METR equation should be the ones expected by investment decision-makers. The difficulty is that underlying expectation formation processes (which one would expect to vary between investment cases) are generally unknown. Similarly, detailed information on the past and expected future loss positions of firms are unknown, thus making the task of adjusting parameter values to account for imperfect loss offsetting highly uncertain – and therefore often ignored. The appropriate rate at which to discount future capital cost allowances is also less than clear. In the case where tax depreciation is not indexed for inflation, there is agreement that a nominal rate should be used. But should the nominal rate correspond to the real cost-of-finance term entering the METR equation directly, or should it be the nominal rate on government bonds in light of the fact that, at least for a taxable firm, the stream of tax relief is more-or-less certain? Alternatively, investment managers may use some other (unknown) rate.

60

Another complication related to discounting tax parameters in recognition of imperfect lossoffsetting is how to account for tax-planning opportunities. Where incentives exist for taxpayers to shift domestic profits offshore, through transfer-pricing techniques and tax-motivated financial structures for example, the use of the nominal corporate income tax rate in the METR equation will tend to overstate the rate at which investment income is taxed at the margin. But the question arises of how to accurately account for this effect. In particular, to what extent should the corporate income tax rate be reduced below its statutory value? The answer might vary across sectors and countries, and would be expected to change over time as the percentage of firms in the domestic economy with global operations (and global taxplanning opportunities) increases, and as the array of sophisticated tax-planning techniques evolves. Moreover, the ability of firms to tax-plan around the nominal rate depends on the existence (or not) and © OECD 2000

Marginal Effective Tax Rates

the strength of provisions aimed at protecting the domestic income tax base (e.g., transfer pricing rules and enforcement, thin-capitalisation rules). Moreover, for certain (inbound) investment projects, foreign corporate tax rates will influence domestic investment plans, meaning that the use of domestic corporate tax rates to explain domestic investment incentives may be misleading. In general, foreign tax rates will matter where foreign (inbound) investors are taxed in their home country on their foreign source income at tax rates in excess of domestic (host country) tax rates. The importance of taking into account foreign investors and their tax position will vary across sectors, countries and time. Historical data on patterns of inbound investment would generally provide an unreliable indicator of the importance of inbound investment, and (forwardlooking) information on the domestic investment plans of offshore investors is typically unavailable. Moreover, modelling the interaction of domestic and foreign tax systems accurately would require detailed knowledge of the global operations of inbound investors in order to establish tax-planning opportunities. A simplifying approach that is often implicitly used is to assume that domestic (host) country tax rates exceed those in the home country of foreign investors. But even if this were the case, the domestic statutory tax rate may overstate the rate at which domestic source profits are taxed to the extent that significant amounts of domestic profit can be shifted to low or zero-tax rate jurisdictions. Shareholder-level tax parameters Personal income tax rates also enter METR analysis, both in the determination of the net return to savers and, depending on the arbitrage assumption, in the calculation of the firm’s cost of funds. Relevant personal tax rates generally include an (ordinary) income tax rate, a dividend tax rate, and an effective (accrual) capital gains tax rate. Different tax wedges may be measured for different groups of savers, and for different types of investment returns (interest, retained earnings, new share issues). Given the number of possible combinations, METRs are usually derived for broad groups of savers (e.g., domestic households, tax-exempt institutions). For the domestic (taxable) household group, a weighted-average income tax rate is generally called for. If available, information on the distribution in prior years of investment income across taxable income classes is used to derive the weights. These may be appropriate in the calculation of a historic (backward-looking) METR series, but may be misleading for the purpose of deriving a current period or forward-looking METR. Similarly, where the net return to savers is measured as a weighted average of returns on interest, retained earnings and new share issues, a weighting scheme that relies on historic data may misrepresent the actual distribution of returns on funds used to finance current or future investment at the margin. A number of expressions found in the literature to measure the net return to savers include a capital gains tax rate term. A data complication encountered here arises out of the fact that while in practice capital gains are subject to tax on a realisation basis, the METR model assumes that they are taxed on an accrual basis. Thus a representative effective accrual-equivalent capital gains tax rate must be derived. In order to convert a statutory (realisation-based) capital gains tax rate to its accrual equivalent, one must make assumptions regarding the expected holding period of equity shares. METR modellers typically must rely on historic information on the average holding period of some basket of shares, which may generate an unreliable estimate of the average expected holding period of shares in firms in a given sector, country and time. For example, optimal holding periods required for preferential tax treatment will vary between countries and over time. An appropriate discount rate must also be chosen. The approach is inevitably ad hoc and subject to measurement error. As noted above, shareholder-level tax rates may also factor into the formula for the firm’s cost of finance. Because different taxpayer groups – such as domestic households, financial institutions, taxexempts, non-resident investors – are subject to different tax treatment, a choice must be made to identify the “marginal” shareholder or “tax-clientele” whose tax treatment is relevant to the determination of the cost of funds (i.e., capitalised into share prices). One would not expect that the identity of the marginal shareholder group would hold consistently across different sectors, given the varying importance of investor groups (e.g., domestic households, financial institutions, tax-exempts and non-residents) across sectors, © OECD 2000

61

Tax Burdens: Alternative Measures

countries and time. Yet, typically, insufficient information is available to identify marginal shareholder groups, and an arbitrary choice is made and assumed to consistently hold across all (domestic) cases. Economic depreciation One of the key parameters entering METR expressions for depreciable capital is the (true) rate of economic depreciation.24 This rate generally cannot be observed and must be inferred. Often secondhand markets do not exist for many types of capital goods, and even where they do, information on the prices at which these goods are transacted is typically unavailable to METR modellers. The typical approach taken is to use reported (and often dated) information on service lives, and to make some arbitrary assumption about the depreciation profile (e.g., exponential). A key problem in measuring economic depreciation, and one that is often ignored, is that the term should take into account not only physical wear and tear, but also expected relative price changes (i.e., the change in the capital goods price index relative to the output price index). Expected relative price changes can occur as a result of a number of considerations. For example, news of the imminent release of new capital technology that will render currently installed capital obsolete would reduce the value of the latter. Similarly, an announcement of an investment tax credit targeted at capital expenditures made after some future date would depress the value of capital stock installed today. A range of such factors could influence investment decisions, yet information on these factors is generally unknown and not taken into account by the METR modeller. Required rates of return A number of conceptual problems associated with the choice of the appropriate arbitrage assumption and financial structure were already considered. At a minimum, it is generally necessary to choose a representative interest rate, and depending on the arbitrage (or non-arbitrage) assumption, some independent measure of the cost of equity may be required as well. In measuring the cost of debt finance, the choice of an appropriate interest rate is generally arbitrary. Interest rates vary according to the term of maturity of debt, and the market’s risk assessment, which generally will be a function of the borrowing firm’s financial position and the value of its property. In principle, some weighted-average interest rate should be used, but typically very little information is available to guide the choice over which interest rates to include and what weights to use. Where equity rates of return are measured independently (rather than being assumed to be determined according to some arbitrage condition), one of a number of estimating approaches may be taken. For example, one may use (reciprocals of) adjusted price-earnings ratios, which are reported for various stock indexes. However, this approach and others suffer from the fact that they are essentially backwardlooking, reflecting past earnings performance rather than expected future performance, and therefore may provide unreliable indicators of expected required rates of return. Inflation rate Another key term in the METR equation is the expected inflation rate, which is subtracted from the nominal cost of funds to arrive at a real cost of funds estimate. As with other parameters, it is the expected value of this rate, and not necessarily its current period value, which is relevant. However, the manner in which investment managers form expectations over this rate is unclear. Static expectations may apply in some cases, or alternatively various forecasting techniques may be used. An average METR statistic would in principle then involve some average of these estimation techniques, which of course are unknown.

62

Often METR analysis proceeds by simply assuming an expected inflation rate, and showing through sensitivity analysis that the METR results will differ significantly depending on the expected inflation rate chosen. If the intention is simply to aggregate into a summary statistic the combined effect of various tax parameters thought to influence investment, and to compare this effect across different investment cases as a useful indicator of differential tax treatment, then such an approach generally would not be problematic. © OECD 2000

Marginal Effective Tax Rates

NOTES 1. In general, pre-tax profit (losses) are derived on the basis of estimated revenues, wage, depreciation and interest costs associated under fixed or variable capital structures. Tax liabilities are derived by applying statutory tax rate and rules governing the determination of tax base, tax credits, treatment of losses, and other relevant tax provisions. 2. A project-analysis corporate ATR is measured by the ratio ATRc (PROJ) = PV(CIT)/PV(Y) where the present value at time (t) of net corporate tax liabilities is given by PV(CIT) = PV(Y) – PV(Y*), where PV(Y) measures the present value of pre-tax corporate profit PV(Y) = Σns = 1(Ys/(1 + Rf)s), PV(Y*) measures the present value of after-tax corporate profit PV(Y*) = Σns = 1(Y*s/(1 + Rf)s), where Ys and Y*s denote pre- and post-tax corporate profits in periods (s ≥ t), and Rf denotes the firm”s cost of funds (discount rate). 3. Note that reduced investment incentives can also be captured in corresponding internal rate of return (IRR) calculations. In particular, a reduced internal rate of return (Rf) would accompany an increase in the present value of corporate tax payments (i.e., the Rf value in the IRR calculation {NPV = 0 = –X + Σs(Y*)/(1 + Rf*)s} would decline, with NPV measuring the net present value of the project with an initial investment of X currency units). 4. Tax wedges and corresponding marginal tax rates may also be derived taking account of shareholder-level taxation, measuring the percentage wedge between pre-corporate tax and after-personal tax rates of return (an overall METR.) Note also that one can derive average marginal effective tax rates. An average METR is a weightedaverage of separate METRs, with each individual METR derived for an investment of a given type – characterised by type of asset (machinery, building, inventories, etc.), type of finance (debt, retained earnings, new equity), or type of saver (individual, parent company, tax-exempt, and so on). 5. Some use the term cost of capital to refer to the user cost of capital – this can create confusion, since the term cost of capital is also used loosely to refer to the financial cost of capital, which is the cost to the firm of obtaining financing (note that the financial cost of capital Rf is a component of the user cost of capital). 6. More generally, the capital consumption cost also includes the change in the value of the capital asset due to changes in the price of the capital asset relative to the general price index – we ignore this consideration in our analysis. 7. This study ignores changes in the real value of the capital stock due to changes in its relative purchase price (changes in its purchase price relative to changes in the general price index (i.e., real capital gains on holding physical capital)). 8. Throughout this paper, it is assumed that the representative firm always has sufficient corporate taxable income against which to deduct interest and capital cost allowances. 9. If investment incentives (e.g., investment tax credits, capital cost allowances) in the corporate tax system are sufficiently rich that the value of Rg in the presence of the tax system is below the value of Rg in the no tax case, I1 would exceed I0. Savings would continue to be unaffected. 10. While taxes of various sorts impinge upon a variety of resource allocation decisions that firms and households take, this note focuses on the limitations of METR analysis as applied to investments in capital (of which there are a number of types, including physical, R&D and human), given that most METR efforts and complexities are concentrated in this area. 11. In particular, assuming the monopolist maximises profits, the term (1-(1/h)) should appear in the denominator, where h is the elasticity of demand (positive for a downward sloping demand curve). 12. Knowledge capital itself is taken to be the product of research investment which exhibits diminishing returns at the margin (i.e., doubling amounts invested in research in the pursuit of knowledge, holding other factors constant, will not double the amount of new knowledge produced). However, increments in knowledge capital generate increased amounts of output at the margin (due to partially appropriatable spillover effects). 13. See for example works by Romer (1989, 1994) who treats knowledge capital as non-rivalrous (i.e., can be shared with others at zero opportunity cost) and as being at least partly excludable. 14. As described in the text, the “fixed-r” approach treats the after-corporate tax rate of return as fixed across all investment projects. In contrast, the “fixed-p” approach treats pre-corporate tax rates of return as fixed across

© OECD 2000

63

Tax Burdens: Alternative Measures

15.

16. 17.

18. 19. 20. 21. 22. 23.

24.

projects. The latter approach is not viewed as an arbitrage assumption of how financial markets work, but rather as one technique for assessing how a particular project would be taxed under alternative tax and financing regimes. Arbitrage assumptions are still required under the fixed-p approach. Where METRs are calculated for each source of finance separately, for example as in King and Fullerton (1984), the fixed-p approach assumes that firms face the same after-tax cost of debt and equity finance (equal to the common cost of funds (r)). This might occur, for example, if the interest rate on debt rises with the amount borrowed (due, for example, to increased bankruptcy risk), so that firms continue to rely on cheaper debt finance (with deductible interest) up to the point where the costs of debt and equity are equated. See for example, Bradford and Fullerton (1981). See Poterba (1987). In the case of a loss carry back, no adjustment generally would be required. The value of the reduction in the loss carry-back caused by the additional revenue from a marginal investment is determined by the current corporate tax rate (assuming that the current corporate tax rate and that applicable in the carry back period are the same). See Mintz (1988) and Altshuler and Auerbach (1990). See Auerbach (1983), Gordon (1985), Bulow and Summers (1984), and Gordon and Wilson (1989). See Bulow and Summers (1984). See Gordon and Wilson (1989). Exceptions include Bertola and Caballero (1991) and McKenzie (1992). Pindyck (1991) provides a review of earlier research. Bertola and Caballero (1991) present a model which generalises the derivation of the Jorgenson user cost of capital to an environment with irreversible capital and risk. MacKenzie (1992) uses the same approach to examine the implications of irreversibility and different sources of risk for the measurement of METRs. This depreciation parameter, along with the real cost of finance, determines the flow cost of holding a currency unit’s worth of capital for one period.

64

© OECD 2000

Chapter 6

POLICY RELEVANCE OF ALTERNATIVE TAX BURDEN MEASURES Introduction This final chapter compares alternative measures of corporate tax burdens and offers some guidance on their suitability for policy analysis. The two policy questions addressed are the tax burden on the stock of capital held in the corporate sector, important to assessing equity in the tax system and in particular the sharing of the tax burden; and an assessment of the tax burden on newly acquired capital, relevant to assessing investment incentives, questions regarding efficient resource allocation and related policy goals. In making this assessment, the chapter compares: – nominal (statutory) corporate income tax rates, including surtaxes and profit/business taxes imposed by sub-central levels of government), denoted STAT; – corporate income tax-to-GDP ratios, denoted ATR(GDP); – corporate implicit average tax rates, denoted ATR(implicit); – average (historic) corporate tax rates derived using firm-specific data, denoted ATR(firm); – project-analysis average corporate tax rates, denoted ATR(proj); and – corporate marginal effective tax rates, denoted METR. For illustrative purposes, Section 6.2 presents values for each of the above measures, with coverage limited to ten OECD countries that are Member states of the European Union. The figures exhibit considerable variation, both across EU countries and across the various measures calculated for each country. As explained below, unlike the nominal tax rates which are fixed by statute, and the tax-to-GDP ratios which are also derived ex-post from fixed numbers, the values for the other measures will depend critically on the assumptions made and, of course, on the taxes included. This highlights the importance of thoroughly checking tax rate figures against these considerations when comparing values across countries or sectors, particularly when using tax rate comparisons for policy purposes. In making the assessment of corporate income tax burdens on firms operating in a given country, a key distinction is between the tax burden on existing capital assets, as compared to the burden on prospective investment. The measured tax burden on existing capital may be a highly misleading indicator of the tax burden on prospective investment, and vice versa (Section 6.3). Of potential measures, in general the best indicators of the corporate income tax burden on existing capital are average tax rates, derived using firm-level or aggregate data. Such figures require certain adjustments and should be interpreted taking into consideration that they are “first incidence” measures that do not account for possible shifting of corporate tax liabilities (Section 6.4). Section 6.5 considers the usefulness of forward-looking tax rates, with a focus on the commonly-used marginal effective tax rate framework for tax policy analysis purposes. Section 6.6 briefly concludes. Some illustrative results Chart 6.1 shows nominal corporate income tax rates and three backward-looking corporate tax burden measures. The nominal tax rates, the corporate tax-to-GDP ratios, and the corporate implicit average tax rates are for 1995, while firm-level (historic) average tax rate figures are an average for the period 1990-1996. Given the reliance of backward-looking tax rates on actual (measured) tax liabilities, such © OECD 2000

65

Tax Burdens: Alternative Measures

Chart 6.1. Backward-looking Corporate Tax Rate Measures, 1995 STAT

ATR (firm)

ATR (implicit)

ATR (GDP)

EU-average Sweden Spain Italy Ireland France Belgium Austria Netherlands United Kingdom Germany 0

10

20

30

40

50

60

Sources: STAT (1995); OECD Tax Data Base; ATR (firm) [average for 1990-1996]: Buijink, Janssen and Schols, 1999; ATR (implicit) [1995]: De Haan and Volkerink, 2000 (forthcoming) ATR (GDP) [1995]: Revenue Statistics, OECD, 1998.

measures become available with a lag, in contrast to forward-looking tax rates based on current information on tax and financial parameters. Also, the implicit tax rate for Germany includes in the numerator not only corporate income tax, but also an estimate of personal tax and business tax on self-employment income.1 (The underlying figures for Charts 6.1 and 6.2 are reported in Annex 6.A.) Among the ten EU economies shown in Chart 6.1, Germany posted the highest nominal rate of corporate income tax, including the business tax (over 56 per cent), followed by Belgium and Ireland. At the same time, Germany is shown to have the lowest corporate tax-to-GDP ratio, at just over 1 per cent. This result is explained largely by the importance of the unincorporated business sector in Germany, in contrast to the other jurisdictions where the incorporated sector dominates. Thus while corporate tax on corporate profit is much lower than what the statutory tax rate would suggest for Germany, its position at the bottom of a corporate tax-to-GDP ranking is driven by the relatively lower contribution of corporate profit to German GDP. For the other countries as well, the corporate-tax-to-GDP ratio is not representative of the actual tax burden on income from capital at the corporate level, for the reasons set out in Section 3.4. The historic (backward-looking) average corporate tax rates derived using manufacturing firmspecific data serve to illustrate the effect of special allowances, deductions, and tax credits, as well as tax planning, in lowering the effective rate of corporate tax below that indicated by the statutory tax rate. This measure also illustrates that the importance of these considerations varies considerably across countries. In particular, the spread between the nominal and profit-based tax rates differs markedly, ranging from just half of one percentage point in Sweden, to 24 percentage points in Ireland, and 19 and 18 points in Belgium and Germany respectively. In the case of the UK and France, a 4 point spread is found while in the Netherlands the gap is just over 3 points. The unweighted average spread between the nominal and profit-based tax rates in the EU countries is 9.6 points. 66

It is also striking that both in the case of Italy and the UK, the implicit tax rate is shown to be higher than the statutory rate. This result highlights the potential for inconsistencies between the numerator and © OECD 2000

Policy Relevance of Alternative Tax Burden Measures

denominator amounts of the sort explained in Section 4.4 that render such measures of questionable relevance, a point to which we return below. Chart 6.2 shows nominal corporate tax rates for the ten countries in 1998, and compares these with two forward-looking measures, project-analysis average corporate tax rates and corporate marginal effective tax rates respectively, also computed for 1998. For most countries, with the exception of France and Spain, the marginal effective tax rates are shown to be well below statutory rates, reflecting the existence of special tax rules providing preferential tax relief. Similarly, for new investment assessed on a (discrete) project basis, the average tax rates are below statutory rates, with German and French firms subject to the highest average corporate tax rates (over 45 per cent). For certain countries, the project-based tax rates are well below the statutory rates, reflecting the existence of special tax rules providing preferential tax relief. The exception is France, where the projectbased rate is above the statutory corporate income tax rate. The reason is that the project-based tax rate calculation factors in a local business tax in France with a different base than that used for corporate income tax purposes.2 Indeed, the spread in possible tax rates in both charts indicates the scope for interest groups and policy-makers to select those tax burden measures that best reflect their assessment of the current tax system and how it should evolve. This remainder of this chapter addresses two main policy questions in relation to the taxation of income from capital at the corporate level, and considers the relative strengths and weaknesses of the measures reviewed in the preceding section when addressing these issues. The two policy areas that we focus on are: – the corporate tax burden on existing capital, relevant to equity concerns (Section 6.3); and – the corporate tax burden on newly acquired capital, relevant to investment behaviour and efficiency concerns (Section 6.4).

Chart 6.2. Forward-looking Corporate Tax Rate Measures, 1998 STAT

ATR (proj.)

METR

20

30

40

EU-average

Spain

Ireland

Belgium

Netherlands

Germany 0

10

50

60

Sources: Statutory rates: OECD Tax Data Base; ATR (proj.); Price Waterhouse Coopers 1998; METR: Baker and McKenzie, Amsterdam 1999.

© OECD 2000

67

Tax Burdens: Alternative Measures

Assessing the tax burden on “old” versus “new” capital The assessment of the corporate income tax burden on firms operating in a given country essentially involves an analysis of the taxation of income earned on assets held at the corporate level – including financial capital (cash, bonds, stocks), physical capital (buildings, machinery and equipment), land, inventory capital, and intangible capital (patents, trademarks).3 In making this assessment, a key distinction is between the income tax burden on existing capital assets, as compared to the burden on a prospective investment. This distinction between tax burdens on “old” versus “new” capital is critical for policy analysis purposes on account of two considerations. First, an assessment of the tax burden on existing or “old” capital is particularly relevant to tax policy questions concerning equity in the tax system, whereas an assessment of the tax burden on “new” capital is particularly relevant to analysing investment incentives and related policy goals. Second, the measured income tax burden on the existing capital stock will often differ significantly from that on newly acquired capital. The differential tax burden on “old” versus “new” capital arises as the existing capital stock in the corporate sector consists of a mix of financial and non-financial assets of varying types, vintages and taxattributes acquired in the past. Consider first the fact that corporate tax owing in the current period on income derived from that existing capital stock will depend on the particular mix of assets held. Thus so too will the amount of corporate tax payable per unit of income generated. Therefore, historic average tax rates measured as corporate tax liabilities as a percentage of economic profit will differ from the effective corporate tax rate on capital acquired at the margin (or a weighted average of such marginal tax rates) to the extent that the prospective investment (or a weighted average of prospective investments) consists of a different asset mix subject to varying tax treatment, including particular tax subsidies. Tax depreciation rates vary across capital asset classes, certain types of income may be drawn into the tax base at different inclusion rates, different rules will typically apply to income earned on domestic versus foreignsource income, and so on. An assessment of the tax burden on the existing capital stock may provide a misleading indicator of the tax burden on newly acquired capital on account of other provisions that operate to link one tax period to the next. One of the most important factors in this respect is the tax treatment of losses. Most tax systems allow businesses to carry non-capital (business) losses forward to offset tax payable in future years, in recognition of the fact that the tax year (i.e., a 12-month assessment period) is an artificial construct.4 To illustrate, assume a firm has a business loss (negative taxable income) in one year, but records a profit in the following year. The firm pays no tax in the first year, but is taxable in the second. In principle, tax liabilities should be determined by allowing the loss incurred in the first year to be carried forward (with interest) to offset positive taxable income in the following year, to give a tax result similar to that which would have occurred had the tax period not been constrained to a single one-year period. In any given year, the existing stock of losses carried forward from prior years and available to offset current period taxable income will depend, among other factors, on the timing of that year over the economic (business) cycle. Loss carryforward pools would be relatively large following a downturn in the economy, for example. Therefore, the tax burden on existing capital measured in a year when relatively large loss carryforwards are claimed (i.e., in a year when corporate tax payments are relatively low), may provide an underestimate of the tax burden on newly acquired capital. A similar consideration is that systems that provide investment tax credits often allow unused credits to be carried forward to offset tax in future years. Research and development tax credits, for example, are often earned by firms that have not yet developed and taken a product to market, and so have no current tax liabilities on profits against which to claim a credit. Tax credit carryforwards may be introduced to ensure a stimulative effect. Patterns of tax credit carryover claims, like loss claims, will depend on business cycle effects, which tend to expand and contract profits and tax base. Therefore, in the presence of carryforward provisions, a tax burden measure based on current period corporate tax payable may be a misleading indicator of the tax burden on newly capital.

68

Another consideration is that corporate tax assessed on realised net capital gains, while relevant to the tax burden on existing corporate assets, may not be relevant to assessing the tax burden on investment at the margin. An adjustment to market interest (discount) rates or expectations over future earnings of exist© OECD 2000

Policy Relevance of Alternative Tax Burden Measures

ing capital, causing an adjustment in asset prices with capital gain or loss effects, will affect current period tax liabilities on dispositions of capital where such gains/losses are drawn into tax, as they are in many systems. While changes in asset prices may affect investment decisions, the potential impact of capital gains taxation on investment behavior may differ significantly from that captured by average corporate tax rates influenced by capital gains/losses on current dispositions of previously acquired capital. The above noted factors arise even where tax policy is held constant over time. Differences in tax burdens on old versus new capital tend to be even more pronounced where tax policy changes over time, as it often does. Consider for example the implications of a reform where accelerated depreciation schedules are replaced with rates that more closely reflect economic depreciation, in which case the tax reducing effects of the old depreciation regime would tend to understate the tax burden on new investment. The tax burden measurement for income derived from depreciable capital purchased in prior years, written-off for tax purposes at rates that differ markedly from depreciation rates applied to capital purchased in the current period, would not be representative of the tax burden on new investment.5 In short, the fact that the income tax liability of a corporation, or a group of corporations, in a given year is an amalgam of tax considerations relevant to income generated on existing capital stock – which may differ for a number of reasons from the tax considerations relevant to a prospective investment – means that corporate tax liabilities measured in a prior year (or even in the current year) relative to (adjusted) financial profit may be a highly misleading indicator of the tax burden on “new” capital, and vice-versa. These points are taken up and elaborated below in Sections 6.4 and 6.5, which address the choice among alternative tax burden measures for tax analysis purposes. Assessing the corporate tax burden, from an equity perspective The design of corporate tax policy often involves a balancing of revenue, equity, efficiency and “competitiveness” and perhaps other considerations (e.g., simplicity), with different groups holding different views over what the appropriate balance is, how that balance should be achieved, and how the resulting corporate tax burden should be measured. Virtually all would agree that corporations should bear at least part of the tax burden. Corporate-level taxation is necessary to avoid tax deferral possibilities that would otherwise exist. And equity considerations recognise that corporate entities, not just individuals, benefit from public expenditures including infrastructure development and costs in administering legal and regulatory frameworks. Furthermore, corporate taxation permits source country taxation in the case of taxexempt and non-resident investors who might otherwise avoid contributing their share towards public expenditure in support of business activities. In measuring the corporate tax burden to address equity concerns that the corporate sector is currently paying its “fair share”, the approach generally taken is to determine corporate income tax liabilities as a percentage of (adjusted) corporate-level profit derived from the existing corporate capital stock. The particular asset mix that gives rise to current period tax liabilities is generally irrelevant, with a focus only on the total amount of corporate tax paid in relation to corporate profit – in the sense that a “buck is a buck” raised from the corporate sector.6 Given the lags with which corporate tax and profit data are compiled and made available, reference generally must be made to data measured in prior years (e.g., the most recent year in which the requisite data is available). Turning to a consideration of possible tax measures, it is clear that the nominal corporate tax rate cannot be used in isolation to assess the tax burden on corporations. This stems from the basic fact that the tax liability of a corporate entity and of the corporate sector as a whole depends on a range of provisions impacting on the tax base, in addition to the nominal rate.7 Aggregate tax-ratios are a common yardstick by which tax systems are assessed. Since they were published for the first time in 1973, the OECD Revenue Statistics have consistently reported aggregate tax revenues as a percentage of gross domestic product (GDP.) Also shown are corporate taxes as a percentage of GDP, personal taxes as a percentage of GDP, and similar ratios for other broad tax aggregates. However, tax-to-GDP ratios may provide misleading indicators of tax burdens. As reviewed in Chapter 3, corporate tax-to-GDP ratios are problematic, as the ratio fluctuates with changes in the share of corporate profits in GDP, even with the share of corporate tax in corporate profit held constant (Section 3.4). More© OECD 2000

69

Tax Burdens: Alternative Measures

over, overall tax-to-GDP are misleading indicators of net tax burdens across countries, as they do not address tax expenditures and the fact that countries vary in their relative reliance on direct and tax expenditures (Section 3.2). Tax ratios are also under the influence of the tax treatment of social benefits (Section 3.3). Backward-looking (historic) average tax rates derived using actual firm-level or aggregate data on actual taxes paid and (adjusted) financial profits earned, however, may be robust indicators of the income tax burden on corporations. Such measures incorporate the effects of both current and past tax provisions, including nominal rate(s), depreciation allowances, the treatment of reserves, the treatment of losses (i.e., carryover provisions for non-capital, capital and tax incentive losses, complex group relief provisions), tax credits and related carryover provisions, other tax laws and regulations, tax-planning, and tax adjustments/relief provided on a discretionary basis by tax administrators. Proper measurement of average tax rates requires that firms with negative business income (loss companies) be excluded and inflation adjustments be made, as noted in Chapter 5. Profit-based measures using actual taxes paid in the numerator of course require access to such data, which typically is only available to Ministry of Finance or Revenue analysts. Taxes reported in financial statements may not give a true account of tax liabilities. The central reason for this is that taxes shown in financial statements reflect the amount of tax notionally owing on financial (book) profits, which may differ from taxable profits. One of the main reasons for this difference concerns the treatment of depreciation. Where capital is depreciated for tax purposes on an accelerated basis, depreciation claims for tax purposes will exceed depreciation charges for financial accounting purposes based on the estimated economic lives of assets. Also, taxpayers typically have discretion over the timing of their depreciation claims (as well as claims made from unclaimed loss carryforward pools and perhaps unclaimed investment tax credits.) To the extent that the offsets to tax taken into account in the calculation of taxes payable for financial reporting purposes differ from claims actually made for tax purposes, differences between financial accounting and actual tax payments will arise. This problem with relying on taxes payable as reported in financial statements has led policy analysts outside government to search for other methods. One such method is the implicit tax rate approach relying on aggregate figures linking taxes actually paid in the year, as available in Revenue Statistics, to National Accounts reporting of operating surplus. However, implicit average tax rates are highly uncertain indicators of the tax burden on capital at the corporate level, for mainly three reasons. First, operating surplus, which appears in the denominator of corporate implicit ATRs, includes interest, rent and royalty income taxed in the hands of individual savers. As the numerator only includes corporate income tax, there is a mismatch between numerator and denominator amounts.8 This mismatch renders implicit ATRs unreliable estimators of corporate tax burdens and corporate investment incentives. Second, in countries that tax resident companies on their foreign source income, an additional mismatch between numerator and denominator amounts is introduced, as operating surplus includes only domestic profit – i.e., foreign income is excluded from the denominator, while net domestic tax on this amount is included in the numerator. Third, the inclusion in operating surplus of firms with negative business income causes an upward bias in corporate implicit ATRs. An accurate corporate ATR measure should exclude such firms, but this adjustment is not possible when relying on National Accounts data.

70

Forward-looking project-based average tax rates and marginal effective tax rates (METRs) provide limited information on the tax burden on capital employed in the corporate sector, for primarily two reasons. First, both measures assess tax liabilities on newly-acquired capital only. The resulting tax rate measures will probably misrepresent the tax burden on previously installed or “old” capital, for the reasons noted in Section 6.3. Second, ATRs derived from project analysis and METRs derived from equilibrium conditions reflect the tax burden tied to a set of assumptions (e.g., relating to rates of return, financing and distribution policy, loss carryover/utilisation rates, expected inflation and interest rates) that may not be representative of actual values determining past (or current) period profit levels, investment patterns and taxes paid. Also, the choice over the weights used to obtain an average of individual project ATRs or METRs (each derived for a different mix of assets and industry), representative of the tax burden on capital at corporate level for the economy as a whole, would necessarily be an arbitrary exer© OECD 2000

Policy Relevance of Alternative Tax Burden Measures

cise. Moreover, such measures typically overlook possible tax-planning techniques that in practice may lower the tax burden far below that implied simple financial structure assumptions. Despite the fact that such models can be run under alternative sets of assumptions on key variables, they provide a highlystylised and therefore limited measure of the actual tax burden on firms, which can only be fully captured with reference to actual corporate taxes paid. In summary, of these potential measures, in general the best indicators of the income tax burden on corporations are backward-looking profit-based ATRs, derived using firm-level or aggregate data on actual taxes paid. The measures are more reliable where profits are adjusted to correct for accounting practices which open up a gap between profits reported in book accounts and true economic profits, and where profits are further adjusted to ensure consistency between numerator and denominator amounts (primarily in relation to the treatment of losses and foreign source income) and to correct for inflation. Such figures should be interpreted taking into consideration the fact that they are “first incidence” measures that do not account for the possible shifting of corporate tax liabilities onto consumers (through higher prices) and workers (through lower wages). Assessing the impact of corporate taxation on investment incentives Policy-makers have long been aware of the possible impediments to investment created by capital income taxation. Indeed, a main policy objective of tax reforms undertaken in a number of OECD countries during the 1985-1989 period was a reduction in tax-induced distortions to the allocation of capital across assets, industries and sources of finance. The 1990s witnessed a partial reversal of this trend with a number of countries introducing special investment tax incentives and preferential tax regimes to attract a larger share of increasingly mobile investment capital. While nominal corporate income tax rates are often cited as important in those cases where tax actually impacts on investment decisions, they are typically considered along with other tax factors (e.g., special investment incentives) determining final tax liabilities on a prospective investment.9 Indeed, it is often said that the nominal corporate income tax rate itself typically does not influence the location of real (i.e., productive) capital, but rather steers the choice over debt-versus equity financing (financing structure) and income-shifting strategies including transfer pricing. In particular, for investments in manufacturing facilities and other business activities where non-tax factors dominate locational choices, financing structures and transfer prices are increasingly being used to leave as little profit (i.e., as much cost) as possible in high-tax jurisdictions to minimise the firm’s global tax bill. Backward-looking (historic) ATRs may be imprecise indicators of investment incentives. As noted, implicit corporate tax rates are highly imprecise measures due largely to the fact operating surplus is measured gross of interest expense (see Annex 6.B for a discussion of the implications vis-à-vis judging investment incentives). As regards profit-based ATRs, last year’s tax as a percentage of last year’s income may be a good measure of the corporate tax burden in that year, but needs to say little about the impact of the tax system on newly-installed capital. Moreover, the same reasons that make backward-looking profit-based ATRs useful indicators of the tax burden on both “old” and “new” capital, may render such measures misleading indicators of the impact of taxation on new investment. This is because investment decisions, based on the expected present value of future after-tax revenues and costs, are fundamentally forward-looking. Even where tax rules have remained stable for a number of years, current year claims from pools of un-depreciated capital costs, which incorporate the effects of depreciation provisions and investment patterns in earlier years, may not be representative of current and future deprecation claims on planned investment. Similar considerations apply with respect to investment tax credits and other provisions that carryover tax reductions from one year to the next. Forward-looking project-based ATRs, which assess the after-tax benefits and costs of investment on prospective investment, in principle may be useful indicators of the drag that income taxes impose on investment. Recent work has emphasized the importance of ATR analysis in the context of foreign direct investment decisions, or more generally in analyzing investment behavior where investors must choose between two or more competing projects (due for example to financing or demand constraints limiting total capital commitments) and expect to earn economic rent (i.e., rates of return in excess of minimum © OECD 2000

71

Tax Burdens: Alternative Measures

required rates).10 Such measures, however, are typically derived under a simplified set of market and financing assumptions, and therefore omit various considerations that impact materially on actual tax liabilities – notably complex tax provisions, tax planning, discretion exerted by the tax administration that may not be accurate or representative, and therefore may not adequately capture the overall impact of taxation. Similar problems arise with the use of METRs. For example, financing structures assumed in both domestic and cross-border variants of METR models typically ignore the use of financing vehicles (e.g., holding companies) in offshore jurisdictions and treaty structures that are critical to the determination of the costs of funds. The task of determining the relevant financing structures that would underlie prospective investments in each sector of the economy of a given country is an enormous task, and for this reason is not incorporated in the reported figures. However, a gross mischaracterization of this central feature of investment renders the resulting statistics of little real relevance. This problem is becoming more important over time as an increasing number of firms and a growing part of total investment, both direct and portfolio, are structured offshore – both in the context of investment by domestic entities and foreign investors. Additionally, the relevance of other assumptions to industry estimates, typically held constant across countries and time, must be questioned. A discussion of the range of difficulties in modelling investment behaviour and addressing data measurement problems is found in Chapter 5 of this study. In summary, forward-looking measures of the tax burden on corporations derived from project/economic models cannot be used in isolation to analyse the impact of taxation on investment. Recourse must be made to other sources of information, including survey-based information. Chapter 5 reviewed the concept of marginal effective tax rates and discussed a number of problems associated with this particular tax measure as a reminder of the limits of METR analysis to offering useful tax policy guidance. Sections 5.5 and 5.6 present a rather long list of conceptual and data problems associated with the marginal effective tax rates framework. The present section rounds out the discussion by considering the usefulness of METR models for tax policy analysis purposes in cases where the underlying assumptions are believed to hold and where the required data is at hand. It turns out that even in this rather unlikely event, the ability of METR analysis to offer a useful guide to the setting of tax policy is limited, on at least three counts. First, METR analysis says nothing about the amount by which investment expenditures actually respond to tax incentives (or disincentives.) In order to gauge the investment response, an understanding of the underlying production technologies is required. In other words, the first-order conditions from which the METR equations are derived do not explain the elasticity or sensitivity of investment demand with respect to changes in the (user) cost of capital (a comparative static analysis involving specification of the production technology is required). A lower corporate income tax rate, or a higher capital cost allowance rate or investment tax credit rate will generate a reduced tax wedge. But the METR model is silent on the timing as well as the magnitude of the investment response, and thus the ultimate impact on the capital stock. This shortcoming is fundamental, as the impact of taxation on capital formation is the essential concept for assessing the influence of taxation on aggregate demand and economic growth. Second, when used to assess incentives for R&D and human capital investment – areas in which theory would suggest the use of positive tax incentives (negative tax wedges) to address market failure (i.e., under-investment due to unpriced spillover effects) – the METR framework provides no guidance on how large the tax incentive should be. This point (related to the one above) stems from the fact that the METR framework offers no explanation of the underlying determinants of the corresponding capital stocks, let alone their optimal values. Without this framework, too often METR modelers are tempted to point to the METR statistic in a given country that signals the most generous tax support, and to hold this value out as the benchmark or target METR to which other countries should strive. Such reasoning is hollow, however, as the degree of tax support may be excessive relative to the unmeasured optimum.

72

Third, the partial equilibrium framework is unable to address the impact of tax policy changes on other factor demands (e.g., labour) and prices. Nor does it consider distributional or transitional effects. Nor does it consider the linkage between the setting of tax parameters and revenues, and thus the bud© OECD 2000

Policy Relevance of Alternative Tax Burden Measures

getary impact of tax policies. In short, many of the key issues that confront tax policymakers when assessing tax policy are simply left unaccounted for. This fact is potentially the most problematic, given the need to address a wide range of issues when recommending tax policy change. This rather discouraging note begs the question: Why has METR analysis generated so much interest? One might argue that this interest grew out of the relative simplicity with which the framework could be used to incorporate the interaction of tax provisions thought to influence investment behaviour, and to capture this net influence in a single summary statistic. Tax rules and regulations are often a messy affair, particularly for non-tax experts, and the prospect of encapsulating their net effect into a single summary statistic is obviously attractive. However, in reality, neither the operation of tax systems, nor the determinants of investment behaviour (nor the interaction of the two) are so simple. Indeed, after reviewing the strengths and weaknesses of marginal effective tax rate analysis, the Working Party on Tax Policy Analysis and Tax Statistics of the Committee on Fiscal Affairs at its 58th Meeting in May 1999 formulated a statement emphasising the need to exercise caution when using METR analysis for policy purposes. The statement acknowledges that METR analysis offers a convenient framework for summarising at a broad level the interaction of tax rules relating to capital investment, and for identifying the various channels through which tax policy might be expected to influence investment behaviour. However, many difficult conceptual and data problems are encountered in the use of METRs. See Annex 6.C for a reproduction of the statement. Conclusion Over the years, in response to growing demand by policy-makers, various measures to assess tax burdens have been developed. The present study has reviewed some of the most common measures used to measure tax burdens of taxpayers, focusing primarily on corporations. In addition, it has provided some illustrative numbers from various sources on tax rates and tax burdens in ten OECD Member countries. More specifically, over the past fifteen years economic analysis of the economic impacts of taxation has increasingly relied on calculated marginal effective tax rates (METRs). However, careful consideration of the range of conceptual and data problems reviewed in Chapter 5 suggests that METR statistics and comparisons across sectors in countries have to be used cautiously as indicators of the influence of taxation on investment or as a guide to the setting of tax policy. The study concludes that all current measures reviewed have at least some important shortcomings. Results based on these and other tax burden measures should therefore be interpreted with their limitations in mind, and judged with due caution when used to address policy questions. Further work in this area will be undertaken by the Working Party on Tax Policy Analysis and Tax Statistics of the OECD Committee on Fiscal Affairs.

73

© OECD 2000

Tax Burdens: Alternative Measures

NOTES 1. As explained in Chapter 4, an implicit corporate average tax rate includes in the denominator corporate operating surplus, taken from National Accounts, and corporate income tax in the numerator. In the case of Germany, National Accounts data report only total operating surplus (of the incorporated and unincorporated business sector), making it necessary to include in the numerator an estimate of the personal tax collected on unincorporated business income (social security contributions paid by the self-employed have been excluded). The ATR for Germany may be referred to as a business income ATR. 2. The business tax is levied on a base which includes the rental value of buildings and equipment plus a percentage of payroll, but is capped to a certain percentage of value added which varies with the amount of turnover realised by the company. 3. Full consideration of the tax burden on income from capital includes not only an assessment of corporate taxation of corporate earnings, but also shareholder-level taxation of that income, taking account of provisions (if any) that integrate the two tax layers. Policymakers are however also interested in the narrower question of the corporate tax burden, the focus of this paper. 4. Certain countries also allow businesses to carry losses back to offset tax in previous years. Carryback relief is in general more advantageous than carryforward relief unless losses may be carried forward with interest (due to the time value of money). 5. In most tax systems, tax depreciation (or depletion) rates applicable to capital acquired in prior years continue to apply to undepreciated capital stocks even when new depreciation rates are introduced. This avoids unanticipated capital gains/losses on existing capital following the introduction of new tax treatment. 6. While the final incidence of the corporate income tax is uncertain (i.e., some uncertain partial shifting of the tax to labour via wage reductions, and to consumers through price increases), provided that shareholders actually bear some portion of the tax, increased corporate income tax will mean an increased percentage tax burden on corporate owners/shareholders. 7. Similarly, personal income tax comparisons limited to rate structures of central governments systems are misleading, because three out of four OECD countries levy additional taxes on personal income. Social security contributions and church tax may push marginal top rates of taxes on income still higher (Section 2.3). It is also shown that high-income earners are not lonely at the top. In fact, many low-income earners are exposed to marginal rates of up to 100 per cent and – in exceptional cases – even more, because after an increase of their income they not only pay more tax but also stand to lose means-tested benefits (Section 2.3). 8. The mismatch between numerator and denominator amounts in an ATR which uses operating surplus in the denominator can be addressed by including in the numerator, together with corporate income tax, an estimate of personal income tax on investment income (in which case the ATR is no longer a purely corporate tax measure). 9. Tax considerations including statutory tax rates and preferential tax incentives are often cited as determining location decisions in relation to highly mobile business activities (i.e., those that can locate at alternative sites at little differential non-tax cost), such as those of administrative and distribution centres. 10. See for example Devereux and Griffith (1998).

74

© OECD 2000

ANNEXES TO CHAPTERS

Annex 2.A.

MEASURING THE OVERALL STATUTORY CORPORATE INCOME TAX RATE IN JAPAN Consider the following approach to derive the effective overall corporate income tax rate in Japan. Define the following variables: t = effective overall corporate tax rate uc = national corporate income tax rate up = combined prefectoral and municipal surtax rate ub = (deductible) enterprise business tax (EBT) rate The computation considered below will use the following values in effect January 1, 1998:

National corporation income tax rate Prefectoral income tax rate (standard) Municipal income tax rate (standard) Combined prefectoral + municipal rate Enterprise business tax rate

uc up ub

0.345 0.051 0.123 0.173 0.111

Consider a ¥ 1 increase in business income in year t. This generates a tax burden in year t equal to the following: uc(1 + up) + ub (1) However, the enterprise business tax (EBT) is deductible in the following year, t + 1. This impact on taxable income in year t + 1, given by (–ub), reduces the tax bases of the national, prefectoral, municipal and EBT taxes in year t + 1. The causal effect spills over to subsequent years as well. In particular, since the EBT in year t + 1 is deductible in year t + 2, the reduction in EBT in year t + 1 causes an increase in taxable income in year t + 2 (i.e., a smaller EBT deduction), which in turn implies increased national, prefectoral, municipal and EBT taxes in year t + 2. The increased EBT in year t + 3 means reduced tax bases in the following year, and so on, with the effect in each subsequent year following year t declining over time. This effect can be illustrated as follows. Let u* denote the combined national, prefectoral and municipal income tax rate (i.e., u* equals uc(1 + up)) to simplify the notation. The overall tax effect of a ¥ 1 increase in business income in year t is given by the following: t = (u* + ub)∆Xt + (u* + ub)∆Xt + 1 + (u* + ub)∆Xt + 2 + (u* + ub)∆Xt + 3+(u* + ub)∆Xt + 4 … etc. (2) where ∆Xs measures the change in taxable income in a particular year, as given by ∆Xt = +1 ∆Xt + 1 = –ub∆Xt = –ub ∆Xt + 2 = –ub∆Xt + 1 = –ub(–ub) = (ub)2 ∆Xt + 3 = –ub∆Xt + 2 = –ub(ub)2 = –(ub)3 ∆Xt + 4 = –ub∆Xt + 3 = –ub(–ub)3 = (ub)4 ... etc. Substituting these values into equation (2) gives the following: t = (u* + ub)(1) + (u* + ub)(–ub) + (u* + ub)(ub)2 + (u* + ub)(–ub)3 + (u* + ub)(ub)4 + … etc which simplifies to the following t = (u* + ub)(1 – ub + (ub)2 – (ub)3 + (ub)4 + … etc. or more compactly,

© OECD 2000

(3) (4) 77

Tax Burdens: Alternative Measures

t = (u* + ub)Z (5) where Z = 1 – ub + (ub)2 – (ub)3 + (ub)4 + … etc. (6a) To solve for the value of Z, consider multiplying Z by the factor (–ub). This gives: Z(–ub) = –ub + (ub)2 – (ub)3 + (ub)4 + … etc. (6b) Now subtracting equation (6b) from equation (6a) gives: Z(1 + ub) = 1 (6c) Dividing by (1 + ub) gives the following value for Z: Z = 1/(1 + ub) (6d) Substituting this solution into (5) gives the following expression for the combined statutory national, prefectoral, municipal and enterprise business income tax rate : t = (u* + ub)/(1 + ub) (7) Substituting the values given above in the table/box yields a value of 0.4637.

78

© OECD 2000

Annex 2

Annex 2.B.

THE EFFECTS OF DIVIDEND TAXATION AND INTEGRATION RELIEF The policy question of the impact of dividend taxation on corporate investment, and conversely the impact of integration measures relieving the double taxation of distributed income, has been the subject of considerable analysis and debate. A number of competing theories have been advanced and tested, and the evidence suggests that the impact of dividend taxation (or alternatively, relieving dividend tax) is not uniform across all cases, but rather depends on a number of factors that may vary from one case to another. The “traditional” view Under the so-called “traditional” view, the double taxation of income from equity capital significantly reduces corporate investment levels, and the introduction of integration measures to alleviate double taxation can be expected to positively impact on capital formation and economic growth. This view rests on the assumption that the payment of dividends has some intrinsic value, apart from being a means to allocate earnings to shareholders. For example, dividends may serve as a signal to shareholders that the distributing company is performing well (Miller and Rock 1985). Moreover, shareholders who may be uncertain over whether managers act in their best interest may look to earnings distribution as a means to place a limit on managerial discretion over the use of the earnings of the corporation (Jensen 1986). Proponents of the traditional view argue that these considerations mean that shareholders will accept a lower after-tax return on shares in corporations with higher dividend payout ratios. However, an increased reliance on dividends, as opposed to share repurchases, as an alternative means to allocate corporate earnings to shareholders, imposes a tax cost on shareholders. This is because dividend income is generally taxed at a higher effective rate than capital gains income. Thus a corporation could be expected to increase its dividend payout ratio just up to the point where the intrinsic (non-tax) benefits from increased dividends at the margin are offset by the net tax burden from an additional one dollar distribution of earnings. This net tax burden is equal to the difference between the shareholder”s effective personal income tax rate on dividends and his effective (accrual) tax rate on capital gains. Therefore, under the traditional view, an increase in the effective tax rate on dividends would lead to a reduction in dividend pay-out and a resulting increase in the firm’s cost of equity finance. Dividend tax relief through the adoption of an integration measure would lower the firm’s cost of capital, leading to increased investment. The “new” view (or “tax capitalisation” view) Proponents of the so-called “new” view or “tax capitalisation” view argue that dividend taxation has no impact on the investment decisions of “mature” firms that finance their investment at the margin using retained earnings. The new view, unlike the traditional view, does not assume that the transfer of earnings to shareholders by way of dividend distribution has some intrinsic value (i.e., dividends and capital gains are perfect substitutes) and assumes that firms cannot repurchase shares. Dividend taxes do not discourage investment financed out of retained earnings because the cost of retained earnings is independent of the dividend tax rate, under the assumption that the dividend tax rate is expected to remain constant. This result follows from the (arbitrage) assumption that the shareholder cannot avoid dividend taxation on earnings within the firm – these earnings will be subject to dividend taxation if distributed in the period when earned and invested in an alternative asset, or if instead the earnings are retained in the firm and distributed later. The present value of the tax burden is the same under both options (assuming that the effective dividend tax rate remains constant). In this sense, equity is “trapped” within the corporation. The new view therefore posits that dividend taxes, which get capitalised into the price of equity shares, impact on the price of shares but have no impact on investment. Integration measures should not be expected to encourage corporate investment, but rather result only in a windfall gain to existing shareholders. It must be emphasised that the new view as summarised above applies to “mature” firms that are able to finance their investment expenditures entirely out of retained earnings, and need not rely on new share issues which generally are a more expensive form of equity finance. Dividend taxes, and relief therefrom, can however be expected to impact on the investment decisions of newly established or rapidly expanding “immature” firms whose investment capital needs outstrip available retained earn-

© OECD 2000

79

Tax Burdens: Alternative Measures

ings and require new equity infusion. Unlike retained earnings, the cost of new share capital increases with the rate of dividend taxation, as owners of capital outside the firm can avoid dividend taxation by investing this capital in an alternative asset, say bonds, not subject to this tax. Thus, in situations where investment is financed at the margin by new share issues, measures providing double taxation relief can be expected to lower the cost of equity finance, and thereby increase corporate investment, a result consistent with that under the traditional view. The “tax-clientele” view The “tax-clientele” view emphasises the variability of effective dividend and capital gains tax rates across different investors, with the result that investors with certain tax characteristics are more likely to hold certain assets compared with investors with other tax characteristics. Investors subject to relatively low dividend tax rates, for example, would tend to favour stocks with relatively high dividend payouts, whereas investors subject to high dividend tax rates would tend to favour “growth” stocks with low dividend payouts. Risk preferences and transaction costs may also vary from one investor group to another. In this setting, one group of investors that enjoys a relatively more favourable tax treatment on the returns from a particular equity investment, and may demonstrate a preference for that investment for non-tax (i.e., risk-preference) reasons, may outbid other investors and emerge as the “marginal” shareholder of the company”s equity. Importantly, it is the dividend tax rate applicable to the marginal shareholder that is factored into the determination of the cost of equity of the firm and thereby influences the level of investment that it undertakes, as well as the market value of its equity.

80

© OECD 2000

Annex 4

Annex 4.A.

THE OECD CLASSIFICATION OF TAXES 1000 Taxes on income, profits and capital gains 1100 Taxes on income, profits and capital gains of individuals 1110 On income and profits 1120 On capital gains 1200 Corporate taxes on income, profits and capital gains 1210 On income and profits 1220 On capital gains 1300 Unallocable as between 1100 and 1200 2000 Social security contributions 2100 Employees 2200 Employers 2300 Self-employed or non-employed 2400 Unallocable as between 2100, 2200 and 2300 3000 Taxes on payroll and workforce 4000 Taxes on property 4100 Recurrent taxes on immovable property 4110 Households 4120 Other 4200 Recurrent taxes on net wealth 4210 Individual 4220 Corporate 4300 Estate, inheritance and gift taxes 4310 Estate and inheritance taxes 4320 Gift taxes 4400 Taxes on financial and capital transactions 4500 Other non-recurrent taxes on property 4510 On net wealth 4520 Other non-recurrent taxes 4600 Other recurrent taxes on property 5000 Taxes on goods and services 5100 Taxes on production, sale, transfer, leasing and delivery of goods and rendering of services 5110 General taxes 5111 Value added taxes 5112 Sales taxes 5113 Other general taxes on goods and services 5120 Taxes on specific goods and services 5121 Excises 5122 Profits of fiscal monopolies 5123 Customs and import duties 5124 Taxes on exports 5125 Taxes on investment goods 5126 Taxes on specific services 5127 Other taxes on international trade and transactions 5128 Other taxes on specific goods and services 5130 Unallocable as between 5110 and 5120

© OECD 2000

81

Tax Burdens: Alternative Measures

5200 Taxes on use of goods, or on permission to use goods or perform activities 5210 Recurrent taxes 5211 Paid by households in respect of motor vehicles 5212 Paid by others in respect of motor vehicles 5213 Other recurrent taxes 5220 Non-recurrent taxes 5300 Unallocable as between 5100 and 5200 6000 Other taxes 6100 Paid solely by business 6200 Paid by other than business or unidentifiable

82

© OECD 2000

Annex 4

Annex 4.B.

ILLUSTRATION OF APPLICATION OF MICRO-DATA This annex illustrates how micro-data (data on individual taxpayers) could be used to generate more accurate measures of personal tax collected on separate items of income (e.g., wages and salaries, taxable income from capital, transfers). This information would allow one to examine how labour and capital ATRs (see Subsection 4.4.3 and Subsection 4.4.4) derived using micro-data compare with those derived using aggregate data. The exercise should be possible where the data set includes, for individual taxpayers in the sample, separate reported figures for the items of income for which tax ratios are being calculated. Case (1): Progressive taxation of combined labour and capital income (global approach) Consider a simplified case consisting of a sample of three taxpayers, two with identical earnings consisting mostly of wages and salaries, and the third earning substantially more investment income. Assume further that the domestic tax system applies the following progressive personal income tax (PIT) rate structure:

Statutory PIT rate structure Income band ($) Tax rate

0-5 0

5-25 0.20

25 and over 0.30

The following table provides micro-data showing the wages and salaries and investment income of the three taxpayers and their tax payable, and also computes their average and marginal tax rates. Aggregate figures are also shown. The shaded part of the table shows the aggregate data that would be available from National Accounts and OECD Revenue Statistics reports in this simplified example. The results show that roughly two-thirds of wages and salaries is found to be taxed in the hands of individuals with an average personal income tax rate of 15 per cent. Investment income is found to accrue mainly to taxpayers with a higher average tax rate of 21 per cent. Knowing the distribution of wages and salaries across taxpayers allows to compute a weighted average of these individual rates, with the weights reflecting the percentage distribution of labour income across taxpayers. Similarly, a weighted average rate for taxable investment income can be derived. Moreover, the micro-data allows to observe the rates at which these income flows are taxed in the hands of individuals at the margin, so that a weighted average marginal tax rate for each category of income may be calculated. Table 2 compares these findings with those available from aggregate data and serves to illustrate the increased information made possible by the micro-data.

83

© OECD 2000

Tax Burdens: Alternative Measures

Case (1) Micro- and aggregate data Wages and salaries

Individual 1 Income Tax Average tax rate Marginal tax rate

Combined income

2

20 (0.2)15 = 3 (3/20) = 0.15

18 0.15 0.20

Individual 2 Income Tax Average tax rate Marginal tax rate

0.15 0.20

18

2

0.15 0.20

Individual 3 Income Tax Average tax rate Marginal tax rate Aggregate Figures Income Tax Average tax rate Ave. marginal tax rate

Investment income

20 (0.2)15 = 3 (3/20) = 0.15

0.15 0.20

20

20

0.21 0.30

40 (0.2)20 + (0.3)15 = 8.5 (8.5/40) = 0.213

0.21 0.30

56

24

(36/56)0.15 + (20/56)0.213 (36/56)0.20 + (20/56)0.30

80 14.5 (14.5/80) = 0.181

(4/24)0.15 + (20/24)0.213 (4/24)0.20 + (20/24)0.30

(Shaded block shows available data at aggregate level.)

Case (1) Comparison of results

Average tax rates Wages and salaries Investment income Average marginal tax rates Wages and salaries Investment income

Aggregate data (mave)

Micro-data (mW and mI)

0.181 0.181

0.173 0.203

n.a. n.a.

0.236 0.283

The following case (2) (see below) is identical to the first, but assumes a 50 per cent inclusion rate for capital income, while case (3) (also below) considers a dual income tax system. In the results shown in cases (2) and (3), the average tax rate results derived using the aggregate data are found to be not significantly different from those derived using micro-data. These findings are of course only illustrative, and indicate that the exercise is intended only as a check on the accuracy of average tax rate analysis based on National Accounts and Revenue Statistics data. If progressive tax rates applied to labour income reached the 50-60 per cent range at relatively low levels of taxable income, the discrepancy between average tax rates determined using aggregate data versus micro-data would be significantly greater.

84

© OECD 2000

Annex 4

Case (2): Progressive taxation of combined labour and capital income (global approach), 50% inclusion rate for capital income

Statutory PIT rate structure Income band ($) Tax rate

0-5 0

5-25 0.20

Wages and salaries

Individual 1 Income Taxable income Tax Average tax rate Marginal tax rate

Combined income

2 1

0.14 0.20

20 19 (0.2)19 = 2.8 (2.8/20) = 0.14

0.14 0.10

18 18

2 1

0.14 0.20

Individual 3 Income Taxable income Tax Average tax rate Marginal tax rate Aggregate Figures Income Taxable income Tax Average tax rate Ave. marginal tax rate

Investment income

18 18

Individual 2 Income Taxable income Tax Average tax rate Marginal tax rate

25 and over 0.30

20 19 (0.2)19 = 2.8 (2.8/20) = 0.14

0.14 0.10

20 20

20 10

0.138 0.30

40 30 (0.2)20 + (0.3)5 = 5.5 (5.5/40) = 0.138

0.138 0.15

56 56

24 12

(36/56)0.14 + (20/56)0.138 (36/56)0.20 + (20/56)0.30

80 68 11.1 (11.1/80) = 0.139

(4/24)0.14 + (20/24)0.138 (4/24)0.10 + (20/24)0.15

(Shaded block shows available data at aggregate level.)

Comparison of Case (2) results:

Average tax rates Wages and salaries Investment income Average marginal tax rates Wages and salaries Investment income

Aggregate data (mave)

Micro-data (mW and mI)

0.139 0.139

0.139 0.138

n.a. n.a.

0.236 0.142

85

© OECD 2000

Tax Burdens: Alternative Measures

Case (3): Dual tax system (separate taxation of labour and capital income)

Statutory PIT rate structure Income band ($) Tax rate on labour Tax rate on capital

0-5 0 0.20

5-25 0.30 0.20

25 and over 0.30 0.20

Wages and salaries

Investment income

Combined income

Individual 1 Income Tax Average tax rate Marginal tax rate

18 (0.3)13 = 3.9 (3.9/18) = 0.217 0.30

2 (0.2)2 = 0.4 (0.4/2) = 0.20 0.20

20 4.3 (4.3/20) = 0.215

Individual 2 Income Tax Average tax rate Marginal tax rate

18 (0.3)13 = 3.9 (3.9/18) = 0.217 0.30

2 (0.2)2 = 0.4 (0.4/2) = 0.20 0.20

20 4.3 (4.3/20) = 0.215

Individual 3 Income Tax Average tax rate Marginal tax rate

20 (0.3)15 = 4.5 (4.5/20) = 0.225 0.30

20 (0.2)20 = 4 (4/20) = 0.20 0.20

40 8.5 (8.5/40) = 0.213

56 8.4 (12.3/56)=0.22 (36/56)0.30 + (20/56)0.30

24 4.8 (4.8/24)=0.20 (4/24)0.20 + (20/24)0.20

80 17.1 (17.1/80)=0.214 (36/80)0.3 + (20/80)0.3 + (4/80)0.2 + (20/80)0.2

Aggregate Figures Income Tax Average tax rate Ave. marginal tax rate

(Shaded block shows available data at aggregate level.)

Comparison of Case (3) results Aggregate data (mave)

Average tax rates Wages and salaries Investment income Average marginal tax rates Wages and salaries Investment income

Micro-data (mW and mI)

0.214 0.214

0.22 0.20

n.a. 0.2

0.3 0.2

86

© OECD 2000

Annex 6

Annex 6.A.

BACKWARD- AND FORWARD-LOOKING CORPORATE TAX RATE RESULTS The following tables provide the figures underlying Charts 6.1 and 6.2 found in Chapter 6.

Figures to Chart 6.1.

Backward-looking Corporate Tax Rate Measures, 1995 STAT

ATR (firm)

ATR (implicit)

ATR (GDP)

56.7 33.0 35.0 34.0 40.2 36.7 38.0 37.0 35.0 28.0 36.5

38.5 29.0 31.8 17.7 21.0 32.8 13.9 35.3 24.1 27.5 26.9

17.6 40.6 22.2 7.6 29.5 21.0 na 51.4 20.5 28.2 26.3

1.1 3.3 3.3 1.6 3.0 1.6 2.9 3.6 1.9 3.0 2.9

Germany UK Netherlands Austria Belgium France Ireland Italy Spain Sweden EU-average

Source: Statutory corporate income tax rates STAT are taken from the OECD Tax Data Base (1995). ATR rates based on firm-level data ATR(firm), showing averages over 1990-1996, are taken from Buijink, Janssen and Schols (1999). Implicit corporate tax rates ATR (implicit) are taken from De Haan and Volkerink (2000), while the corporate tax-to-GDP rates ATR (GDP) are derived using OECD Revenue Statistics, 1998.

Figures to Chart 6.2. Forward-looking Corporate Tax Rate Measures, 1998

Germany UK Netherlands Austria Belgium France Ireland Italy Spain Sweden EU-average

STAT (1998)

ATR (proj)

METR

56.7 31.0 35.0 34.0 40.2 41.7 32.0 41.3 35.0 28.0 36.4

47.2 30.3 31.8 35.0 31.8 47.5 21.8 40.6 33.1 21.0 32.1

37.0 22.3 23.2 27.0 23.5 40.7 22.3 17.7 32.8 17.2 24.3

Source: Project-based ATRs are taken from a 1998 study prepared by the firm PriceWaterhouseCoopers, commissioned by the Ministry of Finance, the Netherlands. The METR statistics are taken from a 1999 study prepared by the firm Baker & McKenzie, also commissioned by the Ministry of Finance, the Netherlands.

87

© OECD 2000

Tax Burdens: Alternative Measures

Annex 6.B.

IMPLICIT AVERAGE TAX RATES – ILLUSTRATION OF THE EFFECT OF TREATMENT OF INTEREST Implicit average effective tax rates (ATRs) include operating surplus in the denominator, which is measured gross of interest, rent and royalty payments. This annex illustrates the implications of including operating surplus, rather than profit (measured net of these expenses) in the denominator, with a focus on the implications of the treatment of interest. The two scenarios considered in the table below differ in that the assumed interest rate on debt is 10 per cent in the first case, compared with 15 per cent in the second (on the same level of debt of 400 currency units). Because interest expense is tax deductible, corporate income tax is lower in scenario 2 compared with scenario 1. Operating surplus is the same between the two cases, being measured gross of interest. Therefore, the corporate implicit ATR measured by corporate tax divided by operating surplus is lower in scenario 2. Observing corporate implicit ATRs alone, one might conclude that corporate tax policy differs between the two cases, and in particular that the effective corporate tax rate is lower in scenario 2, when in fact corporate tax policy is unchanged. One might also (erroneously) conclude that investment incentives would be greater in scenario 2 with the lower implicit ATR, when in fact a rate of return (ROR) calculation finding reduced profitability on account of higher interest expense in scenario 2, suggests that investment incentives are in fact reduced, not increased. These misconceptions stem from the reliance on operating surplus, taken from National Accounts data (which do not report separately profit figures.)

Assets Debt Equity Revenues Less wages, depreciation = Operating surplus Less interest = Profit CIT (@40%) After-tax Profit ROR on equity Corporate Implicit AETR Corporate Profit AETR

Scenario 1

Scenario 2

1000 400 600 200 (100) 100 (40) (@10%) 60 (24) 36 6% (36/600) 24% (24/100) 40% (24/60)

1000 400 600 200 (100) 100 (60) (@15%) 40 (16) 24 4% (24/600) 16% (16/100) 40% (16/40)

88

© OECD 2000

Annex 6

Annex 6.C.

STATEMENT ON THE USEFULNESS OF METR ANALYSIS FOR TAX POLICY PURPOSES Working Party No. 2 of the Committee on Fiscal Affairs at its 58th Meeting in May 1999 formulated the following statement which emphasises the need to use extreme care when using METR analysis for policy purposes. “An important area of economic analysis concerns efficiency considerations tied to capital income taxation and policy changes that might improve the allocation of factors of production – including physical, R&D and knowledge capital – and thus minimise the dead-weight loss or excess burden of taxation. Marginal effective tax rate (METR) analysis and the closely related concept of the user cost of capital have provided an important intellectual impetus to thinking in this area. METR analysis provides a useful framework for identifying the various channels through which tax policy might influence investment behaviour – through the taxation of returns from investment; through the impact of tax allowances and credits on the effective purchase price of capital; and through the possible effects of corporate and shareholder-level taxation on the cost of funds (financial capital). METRs also provide a convenient way of summarising at a broad level the interaction of key tax parameters relevant to the taxation of capital income, and for comparing the treatment under the tax code of different investment activities, and how this treatment compares across industries, countries and over time. Given the attractiveness of its ability to generate summary statistics for the tax-adjusted price of capital, METR analysis has been widely used by academics, private researchers and policy-makers alike to analyse the effects of taxation on investment. Indeed, METR analysis provided much of the intellectual support to tax reforms in many OECD countries introduced over the mid-to-late 1980s which broadened the tax base and lowered tax rates on income from various types of capital in order to secure efficiency gains. Moreover, many analysts continue to rely on this methodology, with ongoing interest, application and review leading to useful modifications that better capture and measure the effect of taxation on capital prices. While recognising the usefulness of METR analysis, many analysts advise caution in light of observations that for certain investment decisions, the results generated may provide unreliable indicators of investment incentives and the effect of taxation on those incentives. METR analysis, like other forms of economic analysis, makes a number of simplifying assumptions which may not hold in certain cases. These calls for caution highlight the need to assess the strength of the underlying assumptions when interpreting the results. This need is obviously critical where METR statistics are considered as a guide to the design of tax policy.”

89

© OECD 2000

REFERENCES Adema, W. 1999, Net Total Social Expenditure, OECD Labour Market and Social Policy Occasional Papers No. 39 (Paris). Altshuler, R. and A.J. Auerbach 1990, “The Significance of Tax Law Asymmetries: An Empirical Investigation”, Quarterly Journal of Economics, 105, 61-86. Auerbach, A.J. 1983, “Corporate Taxation in the United States”, Brookings Papers on Economic Activity, 2, 451-505. Baker & McKenzie 1999, Survey of the Effective Tax Burden in the European Union, Amsterdam. Report commissioned by the Ministry of Finance, the Netherlands. Bertola, G. 1988, Adjustment Costs and Dynamic Factor Demands: Investment and Employment Under Uncertainty. Unpublished Ph.D. thesis, Cambridge Mass., MIT. Bertola, G. and R. Caballero 1991, “Irreversibility and Aggregate Investment”, Princeton University, mimeo. Bradford, D.F. and D. Fullerton 1981, “Pitfalls in the Construction and use of Effective Tax Rates”, in Hulten, C.R. (ed.), Depreciation, Inflation and the Taxation of Income from Capital, Washington: The Urban Institute, 251-278. Buijink, W., B. Janssen and Y. Schols 1999, Corporate effective tax rates in the European Union, University of Maastricht (MARC). Study commissioned by the Ministry of Finance, the Netherlands. Bulow, J.I. and L.H. Summers 1984, “The Taxation of Risky Assets”, Journal of Political Economy, 92, February, 20-39. Bundesministerium der Finanzen 1999, Steuerbelastung deutscher Unternehmen, Volks- und Finanzwirtschaftliche Berichte, Bonn. De Haan, J. and B. Volkerink 2000, Effective Tax Rates: A Critical Survey, OECD, Paris (forthcoming). De Kam, Flip and Ch. Bronchi, 1998, [article in Tax Notes International]. Devereux, M.P. and R. Griffith, The Taxation of Discrete Investment Choices, Working Paper Series No. W98/16, The Institute for Fiscal Studies, London. Eurostat 1997, Structures of the Taxation Systems in the European Union, 1970-1995. Series 2A, Luxembourg. Gordon, R.H. 1985, “Taxation of Corporate Capital Income: Tax Revenues Versus Tax Distortions”, Quarterly Journal of Economics, 100, February, 1-27. Gordon, R.H. and J.D. Wilson 1989, “Measuring the Efficiency Cost of Taxing Risky Capital Income”, American Economic Review, 79, 427-439. Holland, D. and S. Myers, 1979, “Trends in corporate profitability and capital costs”, in R. Lindsay ed., The nation”s capital needs: Three studies (Committe on Economic Development, Washington.) Jarass, L., and G.M. Obermair 1997, “More Jobs, Less Tax Evasion, Cleaner Environment”, Wiesbaden. Report commissioned by the European Commission. King, M.A. and D. Fullerton 1984, The Taxation of Income from Capital, University of Chicago Press.

© OECD 2000

91

Tax Burdens: Alternative Measures

McKenzie, K.J. 1992, “The Implications of Risk and Irreversibility for the Measurement of Marginal Effective Tax Rates on Capital”, Discussion Paper No.137, University of Calgary. Mendoza, e.g., A. Razin, and L.L. Tesar 1994, “Effective tax rates in macroeconomics. Cross-country estimates of tax rates on factor incomes and consumption”, Journal of Monetary Economics, 34, 297-323. Mintz, J.M. 1988, “An Empirical Estimate of Corporate Tax Refundability and Effective Tax Rates”, Quarterly Journal of Economics, 103, 225-231. OECD 1998, Revenue Statistics 1965-1997, OECD, Paris. OECD 1999, Revenue Statistics 1965-1998, OECD, Paris. Pindyck, R.S. 1991, “Irreversible Investment, Capacity Choice, and the Value of the Firm”, Journal of Economic Literature, 24, 1110-1148. Poterba, J. 1987, “Tax Policy and Corporate Saving”, Brookings Papers on Economic Activity, 2, 455-515. PriceWaterhouseCoopers 1998, Study of Potential of Effective Corporate Tax Rates in Europe, Amsterdam. Report commissioned by the Ministry of Finance, the Netherlands. Romer, P. 1989, “Increasing Returns and New Developments in the Theory of Growth”, NBER Working Paper Series, Working Paper No. 3098, 1-31. Romer, P. 1994, “The Origins of Endogenous Growth”, Journal of Economic Perspectives, 8, 322.

92

© OECD 2000

OECD PUBLICATIONS, 2, rue André-Pascal, 75775 PARIS CEDEX 16 PRINTED IN FRANCE (23 2000 02 1 P) ISBN 92-64-17137-1 – No. 50901 2000