Income-Tax Avoidance and Tax Equity The Case of Germany by lht10255

VIEWS: 13 PAGES: 15

									Income-Tax Avoidance and Tax Equity: The Case of Germany
Salvatore Barbaro∗ Ashok Kaul† September 21, 2006

Abstract This paper analyzes the distributional impacts of the German income tax system. It is based on a large representative sample of the 1998 income tax files of the German population. The tax system is highly progressive even when accounting for the numerous exclusions the German tax law provides. These are often presumed to be designed for richer households. Most of the loss in tax revenue compared to a system that allowed for no exclusions (i.e. taxed gross income) stems from few standard deductions and personal exemptions that are difficult to manipulate. Measures of income mobility are derived to substantiate the claim that rich households do not systematically erode their tax bases. Taxable income as a fraction of gross income and tax payments as a fraction of gross (and taxable) income strongly increase in the income level. Also, within any gross income decile, there is not much movement when subtracting quantitatively important tax preferences. We conclude that there is no large discrepancy between the legislated and effective progressivity of the tax system.

Keywords: Tax Evasion, Tax Avoidance, Income-Tax Exemptions JEL Classification: H26
Corresponding author. Johannes-Gutenberg University and Ministry for Economic Affairs, Transportation, Agriculture and Viniculture Rheinland-Pfalz, Stiftsstr. 9, 55116 Mainz, Germany; email: salvatore.barbaro@mwvlw.rlp.de. † Johannes-Gutenberg University Mainz, Department of Law and Economics (FB03), 55099 Mainz, Germany; and IZA, Bonn, Germany; email: ashok.kaul@uni-mainz.de. The authors are grateful to Anna Rosinus and Silke Rath for helpful comments and to the Minister of Economic Affairs of the German State Rheinland-Pfalz, Hendrik Hering, for providing time and ressources to complete the present paper. The opinion expressed in this paper may differ from the government’s opinion.
∗

1

1

Introduction

The German income tax system applies a highly progressive tax schedule to a tax base that ought to reflect a comprehensive measure of income adjusted for tax provisions that are designed to reflect a taxpayer’s ability to pay taxes. Still, the system is perceived to be unfair by many taxpayers, in particular those in the low income deciles. It is often argued that richer households engage in tax avoidance by making use of legal tax provisions that erode their tax bases. In a nutshell, the claim is that the pretended progressivity of the German tax system is counteracted by presumably intended possibilities to narrow the tax base as income increases. In fact, the main goal of the 1999 and 2000 tax reform acts was to increase the fairness of the tax system by closing several loopholes that legally allowed to narrow one’s tax base (German Ministry of Finance, 2000). This paper asks whether the numerous intended or unintended possibilities of tax deductions, exemptions and tax write-off opportunities the German tax system offers are quantitatively important (in terms of tax revenue) and what their distributional effects are. Our analysis is based on detailed administrative income tax data for Germany in 1998 (FAST98). There are two issues. First, does the system systematically violate vertical equity? This would e.g. be the case if households with a high ability to pay were able to declare a low taxable income and then be treated like households with lower incomes who do not (or cannot) reduce their tax base. Second, does the system ensure horizontal equity, i.e. the equal treatment of equals? This would e.g. be violated if groups of households with the same ability to pay would end up with very different taxable incomes. This could happen because some households are capable of exploiting the write-off opportunities the tax system offers or because these opportunities are tailored to certain groups of households independent of their abilities to pay. Figure 1 shows that the difference between gross income and taxable income rises in income. Each box plot1 depicts the difference for each gross-income quintile. As can be seen, both the level and the dispersion rises in gross income. One may therefore conclude that richer households are more likely to manipulate their tax base and this fact might call into question the intended distributional effects of progressive income taxation.2
A box plot, also known as a box and whisker diagram, provides an excellent visual summary of many important aspects of a distribution. The box stretches from the (end of the first quartile to the third quartile. The median is shown as a line across the box. Therefore a quarter of the distribution is between this line and the top of the box and a quarter of the distribution is between this line and the bottom of the box. 2 Barbaro and Suedekum (2006, 2005) analyze the effect of tax exemptions on the
1

2

0

5000

10000

15000

20000

25000

30000

1

2

3

4

5

Figure 1: Difference between gross income and taxable income among the income quintiles. Data from the German taxpayers survey 1998 (FAST98); own calculation. Our results are in contrast to conventional wisdom. We show that the fraction of gross income that is subject to taxation substantially increases for higher gross income quantiles. For any gross income decile, we look at the dispersion of gross income and taxable income, as measured by the coefficient of variation. Taxable income is much more dispersed than gross income in every decile, by a factor of 4 on average. We show that about 4 this increase can be attributed to standard deductions (employee stan5 dard deduction (Arbeitnehmerpauschbetrag) sum of the standard deductions for other income sources, child allowances (Kinderfreibetrag), churchcontribution deduction (Kirchensteuer), (old-age) precautionary deduction (Vorsorgepauschale), lump-sum expenditure deduction (Sonderausgabenpauschale)). We will call these exemptions and deductions the “extended standard exclusions” and henceforth refer to gross income minus these extended standard exclusions as “standard-deduction-adjusted (SDA) income”. Third, we show that mobility across income deciles is quite low, i.e. only a small fraction of
political feasibility of different tax reform proposals.

3

households that are in a certain gross income decile can be found in a different taxable-income decile. For the two top deciles this fraction is particularly high, namely 72% for the ninth and 89% for the tenth decile. Finally, we show that in terms of loss in tax revenue due to deductions only a very small fraction can be attributed to the numerous opportunities the German tax system offers. The tax revenue that accrues from applying the German tax schedule to taxable income is lower by 21.2% than the tax revenue that would accrue (everything else being equal) if gross income were taxed. The tax loss due to the extended standard exclusions already explains a loss of 19.8%. So only the difference of about 1.4% can be attributed to the hundreds of deductions and write-off opportunities incorporated in the German tax system. The results in our paper are in contrast to the bottom line in Lang et al. (1997) who conduct a similar analysis for (West) Germany in 1983. Their main result is that richer households systematically exploit the possibilities the German tax systems offers to reduce their taxable income given gross income. The analysis of different years (1983 versus 1998) and the changes in the tax system as well as our use of German data for East and West Germany can certainly explain some of the differences in the results. Also, they make use survey data and while we use administrative tax data. Therefore their analysis partly captures illegal tax fraud since their measure of gross income may encompass income not reported to tax authorities. However, the statistics that can be calculated for both data sets are not that different. Therefore, we believe that the main differences in our results stem from the fact that we have—for our purposes—better data and could therefore decompose the difference between gross income and taxable income in detail. Lang et al. (1997) had to make use of proxies like socioeconomic and demographic characteristics to explain this difference. We discuss the relationship between their paper and our analysis in detail throughout the paper. The paper is organized as follows. Section 2 describes the data set on which our analysis is based. Section 3 presents descriptive statistics for (gross) income deciles. Section 4 considers how tax deductions affect the dispersion of income and mobility across income deciles. Section 5 offers some concluding remarks.

4

2

Some descriptive statistics

Data are taken from the official micro survey of German tax declarations, “FAST98”. This highly reliable data set is drawn from some 3 million3 tax declarations for the determination of the annual income tax 1998 in Germany, which accounts for a representative 10% survey of all tax payers. For each person (i.e., for each tax declaration), total taxable income is known, as well as the deduction amount for all tax exemptions. The data set, therefore, allows for a comprehensive explanation for the difference between gross and taxable income. Our starting point is a comprehensive income-tax measure, namely gross income. It includes seven income categories like wages and compensation, interest, dividends, capital gains (or loss), business income (or loss), pensions, farm income (or loss), rents, social security benefits, etc. We calculate gross income for every household by adding special allowances for each income source to adjusted gross income (Gesamtbetrag der Eink¨ nfte), which is the broadest income measure in our data set. Taxu able income is given in our data set. It can be derived from gross income by taking into account several exemptions and deductions. We have comprehensive information about all deductions and exclusions that households made use of. We first construct distribution tables by dividing the entire population of households into income quintiles by gross income levels. Taxes paid are then calculated for each group. The distribution of gross income and contributed taxes across quintiles is shown in Figure 2. It shows that the top 20 percent of households earn about 50 percent of all income and the bottom 20 percent of all households earn about 2 percent of all income. Most taxes are paid by upper-income groups. Given the progressive nature of our tax system, such a result can be taken for granted for taxable income. That it also holds for gross income is a first indicator for effective progressivity of the German tax system. Taxpayers in the top 20 percent of the gross income distribution pay 65% of all income taxes, while taxpayers in the bottom 20 percent contribute almost nothing to total revenue. Roughly 50% of federal taxes are paid by taxpayers in the top 10% of the distribution and more than one fith by the top 1%, as shown in Table 1. Here, “gi share” denotes the percentage of gross income earned by a particular income bracket and “tp share” denotes the share of contributed taxes.
3 Due to this extremly large number, we abstain from reporting any measure of statistical significance throughout the paper.

5

0.5

0.6

share gross income share tax payment

share 0.0 0.1 0.2 0.3

0.4

1

2

3 Income Quintile

4

5

Figure 2: Percentage of gross income earned and contributed income taxes for each gross income decile

3

The distribution of income and tax payment across income deciles

This section analyzes the question where most of the difference between gross income and taxable income comes from. We calculate an intermediate measure of income. Starting from gross income,we subtract the employee standard exemption, the special-expense lump-sum deduction (Sonderausgabenpauschale), the child exemption, the (old-age) precautionary exemption (Vorsorgepauschale) and the church-contribution deduction. We also subtract the standard exemptions from income sources other than employee income. For all these deductions, the households essentially have no discretion and therefore it is hardly possible to manipulate any of these items. The deductions so far will be called the extended standard exclusions. Our focus will be on the 6

< median top 10% top 5% top 1%

gi share 0.19 0.33 0.22 0.10

tp share 0.07 0.49 0.37 0.21

Table 1: Percentage of gross income earned and contributed income taxes for different income groups difference between taxable income and income after the extended standard exclusions have been subtracted from gross income. Note that we did not include any itemized deductions where households have discretion to manipulate their taxable income. Before we do these calculations, let us summarize important measures of gross income, taxable income and tax payments across income deciles. Table 2 refines the information in figure 1. Since our data are not censored, the mean income in the first decile is negative. This also explains the outlierbehavior of the coefficient of variation in the first decile. We will mostly focus on the top 9 deciles in our analysis. Since our data are uncensored as well as the fact the income is heavily skewed to the right also explains why the mean income of the tenth decile is higher by an order of magnitude than the mean in the ninth decile. Also, the dispersion, as measured by the coefficient of variation is much higher in the tenth decile than in the second to ninth decile. Interestingly, the coefficient of variation of taxable income (calculated for each gross income decile) is higher by a factor of about 2 (ninth decile) to 4 (fifth decile) in comparison to the coefficient of variation of gross income. In the tenth decile, there is only a small difference. Lang et al. (1997) argue that this is first evidence for the claim that some households succeed in eroding their tax base.

7

1 Decile 1 2 3 4 5 6 7 8 9 10

2 Limit 4391 10581 16728 22091 26514 31293 37583 46528 61592 6639001

3 4 5 6 mean.gi cv.gi mean.gbe cv.gbe 2233.63 −37.97 −1992.99 −10.53 7874,78 0.24 5996.15 0.34 13903.86 0.13 11569.27 0.21 19490.28 0.08 17164.05 0.15 24191.70 0.05 21998.45 0.11 28525.40 0.05 26402.15 0.09 33702.53 0.05 31537.46 0.09 40689.55 0.06 38730.68 0.09 50938.84 0.08 49586.46 0.10 74346.13 1.59 103243.14 1.67

7
gbe ti

0.51 1.96 1.55 1.36 1.28 1.26 1.22 1.18 1.16 1.12

8 9 mean.ti cv.ti −4880.00 −5.02 3435.00 0.76 7697.00 0.40 12895 0.28 17276 0.21 21126 0.19 25896 0.17 32785 0.15 42755 0.15 92596 1.81

10
ti gi

6.13 0.41 0.54 0.65 0.70 0.73 0.75 0.78 0.80 0.85

11 12 13 tax tax mean.tax gi ti 89.01 −0.01 −0.00 270.50 0.01 0.02 807.29 0.04 0.07 1863.70 0.08 0.12 2680.59 0.10 0.14 3522.18 0.12 0.16 4587.29 0.13 0.17 6379.73 0.15 0.19 9188.98 0.17 0.21 28869.93 0.26 0.31

8

Table 2: Summary statistics for gross income, gbe and taxable income, and tax payment across the income deciles

Concerning tax payments, a few observations are remarkable. First, the first two income deciles essentially pay no income taxes. This can be explained by the low taxable income in the first two deciles and the fact that positive tax rates apply to taxable income only after a personal exemption of about 6.000 Euros has been deducted per (adult) person in a household. Second, taxable income as a fraction of gross income increases monotonically from 41% in the second decile to 80% and 85% in the last two deciles. Also, tax payments as a fraction of gross income strongly and monotonically increases from gross income decile to decile, from 0 up to 26% for the top decile. This shows that - on average - the tax system is highly progressive, not only if tax payments are related to taxable income (see the last column of table 2) but also if gross income is considered. The question, thus, is how the difference between gross income and taxable income can be explained within an income decile.

4

Income dispersion and mobility

The previous section showed that on average the fraction of households’ gross income that is subject to taxation increases in income. Also, on average tax payments, average (and marginal) tax rates increase in income. This section reveals what is going on within every income decile. Also, we analyze how the extended standard exclusions help to explain the difference between gross income and taxable income across deciles. Finally, we consider a measure of income mobility when comparing gross income and taxable income.

4.1

From gross income to taxable income: differences across and within income deciles

The boxplots in figure 3 summarize three income measures for each of the income deciles. The red boxes depict the distribution of gross income. Since gross income is the order criterion for the deciles, the red boxplots do not overlap for any gross income level. The blue box plots depict the distribution of taxable income. As can be expected from table 2, on average taxable income is much lower than gross income in every decile. At first glance, the dispersion when considering taxable income instead of gross income does not increase much. However, a box plot is not perfectly suited to talk about dispersion and we will take up this question subsequently. The green boxes show for how much of the difference between gross and taxable income the extended standard exclusions account. They depicts the 9

0

20000

40000

60000

80000

100000

120000

b

y

z

b

y

z

b

y

z

b

y

z

b

y

z

b

y

z

b

y

z

b

y

z

b

y

z

b

y

z

Figure 3: Distribution of gross income (red), gross income minus standard allowances (green) and taxable income (blue) distribution of the SDA-income. As can be seen, from the first to the ninth decile an overwhelming fraction of this differences can be attributed to the extended standard exclusions. Even for the tenth decile more than 2 of the 3 difference is explained by these deductions. An important question for fiscal policy is whether the myriads of deductions and exemptions cause high losses in terms of tax revenue. We therefore compare actual tax revenue (based on taxable income and the income tax schedule) with the hypothetical situation where we apply the income tax schedule to gross income. The result is that actual tax revenue is 21.2% lower than in the hypothetical situation. The (hyothetical) tax loss due to the extended standard exclusions already explains a loss of 19.8%. So only the difference of about 1.4% can be attributed to the various itemized tax preferences embedded in the German tax system.

10

4.2

Mobility across gross income deciles

The boxplots hide movements within and across the boxes. In principle, when moving from SDA income (e.g. gross income after subtracting the extended standard exclusions) to taxable income, some households may be able to deduct a lot while others have no additional deductions. Gross Income 1 2 3 4 5 6 7 8 9 10 1 0.80 0.16 0.03 0.01 0.00 0.00 0.00 0.00 0.00 0.00 2 0.20 0.62 0.14 0.03 0.01 0.00 0.00 0.00 0.00 0.00 3 0.00 0.22 0.56 0.16 0.04 0.01 0.00 0.00 0.00 0.00 Taxed 4 5 0.00 0.00 0.00 0.00 0.26 0.00 0.46 0.34 0.18 0.42 0.07 0.18 0.02 0.05 0.00 0.01 0.00 0.00 0.00 0.00 Income 6 7 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.00 0.35 0.00 0.45 0.28 0.16 0.55 0.03 0.15 0.00 0.02 0.00 0.00 8 0.00 0.00 0.00 0.00 0.00 0.00 0.22 0.63 0.14 0.00 9 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.17 0.75 0.08 10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.09 0.91

Table 3: Percentage of taxpayers of taxed income decile within gross income decile We now take a look at income mobility across gross income deciles due to deductions and exemptions. Table 3 orders households in a transition matrix. It shows that—given that households are ordered according to their gross-income decile (the lines in the table)—the taxable-income decile (the columns) where they end up is typically not very different. The diagonal in the table is very dominant. This means that—after correcting for exclusions—many households remain in the same income decile. This fraction is particularly high in the ninth and tenth decile. Less than 1% of the households in the top gross-income decile end up in the 8th (or lower) taxable-income decile. Also, 84% of the household in the 9th gross income decile remain in the 9th taxable-income decile or even move to the 10th decile. Finally, if one reads the table columnwise, the picture shows little mobility. Very few households make a transition to a taxable-income decile from a gross-income decile that is not neighboring. Indeed, the Bartholomewmeasure see (see Bartholomew, 1982) of mobility yields about 4, which indicates relatively little relative mobility.

11

−0.2

0.0

0.2

0.4

0.6

0.8

1.0

1

2

3

4

5

6

7

8

9

10

Figure 4: Share of gross income which is liable to tax across the income deciles

4.3

Dispersion within a gross income decile

An interesting number in table 2 was the relationship between the coefficient of variation of gross income and the coefficient of variation of taxable income (given the gross income deciles). The latter was on higher by a factor of 2 to 4. This could be some indication for large movements within an income decile due to the use exclusions. In this section, we compare the coefficients of variation of gross income, standard-deduction-adjusted (SDA) income as well as taxable income. Table 4 shows that from the 2nd to the 8th decile about 2 of the increase in the coefficient of variation from gross income to 3 taxable income are due to the extended standard exclusions. The increase in the coefficient of variation for the two top deciles is explained to a larger extend by the deduction of the aforementioned deductions. The reason for this counterintuitive result is the following. We have seen in Figure 1 that the difference between gross income and taxable income rises in income. However, these difference was measured in absolute terms. In assessing horizontal tax equity, the difference in relative terms is a useful measure. Thus, we compute the taxable income in relation to gross income. 12

Decile 1 2 3 4 5 6 7 8 9 10

cv.gi cv.v cv.ti −37.97 −7.55 −5.02 0.24 0.48 0.76 0.13 0.27 0.40 0.08 0.19 0.28 0.05 0.14 0.21 0.05 0.12 0.19 0.05 0.11 0.17 0.06 0.10 0.15 0.08 0.11 0.15 1.59 1.75 1.81

cv.v cv.gi

cv.ti cv.gi

0.20 2.00 2.09 2.41 2.73 2.49 2.05 1.60 1.32 1.10

0.13 3.17 3.04 3.50 3.98 3.86 3.20 2.43 1.83 1.14

Table 4: Coefficient of variation for gross income (gi), adjusted income (v) and taxable income (ti) This measure indicated the share of gross income which is liable to tax. Figure 4 depicts that both, the level and the dispersion fall in income. This shows that only a moderate part of the increased dispersion within income deciles is due to itemized tax provisions. Furthermore, it is difficult to find evidence that the use of tax preferences is more important for high income households than for moderate-income or low-income households.

5

Concluding remarks

Designing an income tax system consist of two steps. First, a tax base has to be defined. Second, a tax schedule has to be applied to the tax base. Whether a tax system is considered to be fair depends on both steps. This paper has taken gross income as reported in the tax files as given. We have then asked what the distributional impacts of the several exemptions and deductions in the German tay systems are. We have seen that most of them are quantitatively unimportant. So why do we have them then? In some cases, these tax provisions are in place to encourage certain kinds of economic activity, such as the purchase of homes. In other cases, these preferences are in place to generate a certain kind of social behavior, such as charitable giving. In still other cases, these preferences are in place to provide assistance to low-income households. Finally, some tax preferences are designed to reflect a taxpayer’s ability to pay taxes. Tax preferences have varying effects and success in achieving their goals. Since the sum of the tax provisions make the system very complicated it is 13

therefore interesting to know what their costs and benefits are. In particular, what are the total costs for complying with the individual income tax? Compliance costs represent the time and resources expended by taxpayers to interact with the income tax system. These costs include the value of time spent learning about the tax law, maintaining tax records, completing and filing tax forms, and responding to any correspondence from the tax authorities. Compliance costs also include amounts paid to others to conduct any of these tasks on behalf of an individual or a business like expenditures on out-of-pocket costs for the services of tax professionals and software purchases. Total yearly compliance costs are difficult to estimate, in part because estimating the value of the time people spend on their tax returns is difficult. It would be interesting to make a cost-benefit analysis of the several tax provisions in the German tay system. This would complement the usual excess-burden calculations of distortionary taxation. These calculations typically do not account for compliance or administrative costs of taxation.

References
Alm, J., Sanchez, I., de Juan, A., 1995. Economic and noneconomic factors in tax compliance. Kyklos 48. Altig, D., Auerbach, A. J., Kotlikoff, L. J., Smeters, K. A., Walliser, J., 2001. Simulating fundamental tax reform in the United States. American Economic Review 91, 574–595. Andreoni, J., Brian, E., Feinstein, J., 1998. Tax compliance. Journal of Economic Literature, 818–860. Barbaro, S., Suedekum, J., 2005. The interaction of tax exemptions and individual tax reform preferences. iza Discussion Paper Series 1543. Barbaro, S., Suedekum, J., 2006. Reforming a complicated income-tax system: The political economics perspective. European Journal of Political Economy 22, 41–59. Bartholomew, D. J., 1982. Stochastic Models for Social Processes. Wiley, London. Elffers, H., Weigel, R. H., Hessing, D. J., 1987. The consequences of different strategies for measuring tax evasion behaviior. Journal of Economic Psychology 8, 311–337. 14

German Ministry of Finance, 2000. Steuerpolitik der Bundesregierung. BMF, Berlin. Kaplow, L., 1992. Income tax deductions for losses as insurance. American Economic Review 82, 1013–1017. Lang, O., N¨hrbaß, K.-H., Stahl, K., 1997. On income tax avoidance: the o case of Germany. Journal of Public Economics 66, 327–347.

15


								
To top