97 extraordinary role of the tax rate cut to 25% in 2001, which was promoted by the government as an answer to international tax competition, can be seen as an indicator for an important role of tax competition as well. On the other hand, reforms in the local trade tax were stronger in the 1970s than in the 1980s and the 1990s. Here average annual fiscal effects of tax reductions were 1.8 times higher in the period from 1965 to 1986(especially driven by the abolishment of the payroll tax) compared to the period from 1987 to 2004.175 Therefore, the reform pattern in the local trade tax does not support the hypothesis of a “race to the bottom”. What are the implications for the analysis to follow? First, we conclude from our findings that we need to integrate tax competition as explanatory variable for the corporate profit but not for the local trade tax. If we take into account that the fiscal effects of corporate tax reforms from 1964 to 2004 account only for less than 4% of the total fiscal effects of all tax reforms covered in our data-set, we see that international tax competition had only a very limited influence on our data-set as a whole. Therefore, we are able to largely abstract from tax competition in our general analyses of tax reforms but control for a possible influence on a case by case basis. 2 Normative approaches While we are generally focusing on polit-economic analyses, we do not completely exclude main normative approaches which all largely abstract from political and institutional variables in tax policy. It is especially important for us to analyze, whether normative approaches can contribute to our understanding of tax reforms and are thereby limiting the importance of positive theories.176 We start with the “financing function” of taxation which argues that tax policy is driven by expenditure decisions and the financial needs of the state. The second approach is based on macroeconomic considerations: Are tax reforms scheduled to smoothen the business cycle? Then we discuss the linkage in between inflation and tax policy. Do we see a reaction of tax policy to high inflation and the resulting “cold progression” in progressive taxes? Finally, we shortly review the approach of tax smoothing (see Table 5 for an overview of normative hypotheses). Despite its importance in normative analysis, we do not integrate the theory of optimal taxation as it has only very indirect implications for the variables covered in our data-set.177 175 Please note that we have integrated here only changes on the federal level while changes of the municipal multipliers, which have shown a general upward trend, are not covered. 176 For an overview of discussed approaches see Table 4 on p.90. 177 For a textbook-treatment see Myles (1995). See also Ramsey (1927), Mirrlees (1971), Mirrlees (1976), Auerbach/Hines (2002) and Slemrod (2004). 98 Normative hypotheses on taxation and tax reform Hypothesis 2 Hypothesis 3 Hypothesis 4Hypothesis 1 Timing of tax reforms takes the business cycle into account Tax policy reacts to cold progression by correcting for „bracket creeping“ Tax policy seeks to minimize the excess burden of taxation and tax reform Taxation is only a mean to finance public good provision M ec ha ni sm Tax burden reductions are scheduled to economic downturns Tax burden increases are scheduled to economic upturns Inflation has a direct influence on tax policy, especially in the progressive wage and income taxes Tax burden reductions in the progressive wage and income tax are more frequent and larger in times of high inflation Tax burden changes are directly (and positively) linked to expenditure changes (e.g. expenditure increases are linked to tax burden increases) Changes in the tax burden are directly (and negatively) linked to fiscal deficits (e.g. deficits are linked to tax increases) Taxation is largely independent from transitory deficits and the business cycle Changes in tax rates and tax bases occur relatively rarely Pr ed ic tio ns „Financing function of taxation“ „Stabilization function of taxation“ „Correcting of cold progression by tax reform“ „Tax smoothing“ G en er al Ap pr oa ch Table 5: Normative hypotheses on tax reforms With respect to each approach we first shortly review the general argument, then derive hypotheses and finally test these hypotheses based on our data-set.178 We dispose of data on tax reforms by the date when they were adopted and when they became effective. It is not always easy to decide which data should be used for our empirical tests. This can be illustrated with respect to the stabilization function. If we analyze the relationship of the macroeconomic state of the economy and fiscal effects of tax reforms by the date of their adoption, we should find out whether the macroeconomic conditions during the legislative process have influenced the tax policy decision. On the other hand only data on implemented reforms reflect the timing of tax burden increases and reductions which affect the business cycle. For our empirical tests in this part we have decided to rely mainly on the data on adopted reforms as they reflect the timing of political decisions more directly than data on implemented reforms.179 178 For a description and analysis of the data-set see part III and IV. 179 However, we have pursued all empirical tests with the data on implemented reforms as well and have derived very similar results. 99 2.1 The financing and the stabilization function We start our analysis with a combined discussion and testing of the financing and the stabilization function as both refer to the same variables. 2.1.1 The financing function of taxation 2.1.1.1 General approach In many normative approaches the link in between taxation and expenditure is straightforward. If we take Samuelson´s classical contribution “a pure theory of public expenditures”, the optimal tax rate is just a results of the optimal level of public good provision (Samuelson 1954). Another example is the allocation function in Musgrave’s three classical function of public finance which sees taxation mainly as a mean to finance public good provision (Musgrave 1973). But how close is the relationship in between expenditure developments and tax reforms in our data?180 2.1.1.2 Hypotheses Based on the approaches which mainly focus on the financing function of taxation we can directly derive hypotheses that we can test based on our data-set. First, we focus on the relationship of expenditure developments and tax policy. Here we can derive two hypotheses. The first hypotheses is that tax increases facilitate spending increases. In this case we should find that expenditure increases follow tax reforms which increase the tax burden (financing hypothesis I).181 However, the timing of expenditure increases and tax increases could as well be reversed. In case that tax policy would react to spending decisions in previous periods, we should see that tax burden increases following expenditure increases in previous periods (financing hypothesis II).182 But expenditure increases are not the only indicator for a link in between tax reforms and expenditure developments. Tax policy might not be directly linked to 180 In the empirical literature the causal relationship in between tax policy and expenditure decisions is hard to determine as most studies need to rely on tax revenue data which is only indirectly linked to tax policy. Based on time-series data for the US, Sheffrin and Hoover (1991) found that taxes appear to have caused spending before the mid-1960s. After the late 1960s, taxes and spending have been causally independent in the US. See as well the discussion in Winer (1983). Based on our data-set we are able to study the relationship in between expenditure developments and tax policy more directly. 181 Sheffrin and Hoover (1991) would name this the “sand in the sandbox” hypothesis. 182 Sheffrin and Hoover (1991) would call this the “tax collector for the welfare state” hypothesis. 100 expenditure increases – which for example could as well be financed by tax revenue increases based on cold progression – but to fiscal deficits. In this case we should expect to see tax burden increases via tax reforms as a reaction to fiscal deficits (financing hypothesis III), eventually with one or more time lags to account for delays in political reactions to fiscal developments. 2.1.2 The stabilization function of taxation 2.1.2.1 General approach The second classical function of public finance is – according to Musgrave (Musgrave 1973) – the stabilization function of the state which includes the use of tax policy to stabilize business cycles. If tax reductions have c.p. a stimulating effect on economic growth while tax increases are cooling down the economy, tax policy should take the current state of the business cycle into account. Based on the stabilization function the government is expected to schedule tax burden reductions in times of recessions while tax burden increases should be scheduled to economic booms. As deficits are likely to occur in times of economic downturns (when revenues are reduced stronger than expenditure), deficits should – based on the stabilization function – be correlated to tax burden decreases as well.183 2.1.2.2 Hypotheses The most important indicator for the current state of the business cycle is real GDP growth. High real GDP growth indicates an economic boom and low or negative real GDP growth an economic downturn. Based on the stabilization function we should expect a correlation of real GDP growth and fiscal effects of tax reforms. In times of high GDP growth (eventually with one or two time lags to account for the slack of policy making) we should see tax burden increases by tax reforms, while there should be tax burden reductions in times of low growth (stabilization hypothesis I). With respect to deficits a similar relationship can be expected based on the stabilization function: high deficits should go hand in hand with tax reforms, which reduce the tax burden, while low deficits or even surpluses should be correlated with tax burden increases (stabilization hypothesis II).184 183 Here we do not refer to “automatic stabilization” by taxes as discussed in Auerbach and Feenberg (2000). Furthermore, we do not incorporate temporary tax policy measures to smoothen the business cycle which were only applied from 1967 to 1975. See excursus on temporary measures in part III.3. 184 This is contrary to the predicted relationship in between deficits and tax reforms of the financing hypothesis 3. 101 2.1.3 Empirical testing of the financing and the stabilization hypotheses In the following part we test the hypotheses derived from the financing and the stabilization function of tax policy empirically. We discuss both approaches together as their hypotheses rely partly on the same independent variables (e.g. fiscal deficits). We first present general data on expenditure developments, tax revenue developments and deficits, then perform tests based on annual data and finally analyze tax policy based on legislative periods. 2.1.3.1 Revenues, expenditure and deficits 15% 20% 25% 30% 35% 40% 19 50 19 52 19 54 19 56 19 58 19 60 19 62 19 64 19 66 19 68 19 70 19 72 19 74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 20 04 PUBLIC EXPENDITURES/GDP PUBLIC REVENUES/GDP TAX REVENUES/GDP -7% -6% -5% -4% -3% -2% -1% 0% 1% 2% 19 50 19 52 19 54 19 56 19 58 19 60 19 62 19 64 19 66 19 68 19 70 19 72 19 74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98 20 00 20 02 FISCAL DEFICITS/GDP REVENUE AND EXPENDITURE DEVELOPMENTS 1950-2004 Data source: Deutsche Bundesbank. Fiscal deficits according to national budget calculations. Figure 46: Revenues, expenditures and deficits in Germany 1950-2004 Figure 46 maps the development of tax revenues, expenditures, and deficits over GDP for all levels of government in the Federal Republic of Germany from 1950 to 2004. We see that public expenditure over GDP was relatively stable until the end of the 1960s while tax revenues over GDP expanded. Fiscal deficits, which depend not 102 always only on tax revenues and expenditures but on other revenues as for example privatization receipts or fees for services as well, started to increase in the early 1960s. From 1970 until 1982 expenditure showed a strong upward trend, while tax revenues increased only slowly. This led to high deficits. After 1982 expenditures over GDP showed a downward trend which led – as tax revenues over GDP stayed relatively stable – to a fiscal consolidation and decreasing deficits. Reunification in 1990 pushed deficits up again, but consolidation made progress again at the end of the 1990s based on increasing tax revenues. In 2000 high revenues from the auctioning of UMTS licenses had a strong upward effect on revenues over GDP and even led to a surplus. 2.1.3.2 Tests based on annual data How can we test the financing and the stabilization function? Most studies rely – based on a lack of alternative data – on revenue and expenditure developments. However, it is problematic to evaluate tax policy based on revenue and expenditure developments which are affected by many different factors. As we dispose of firsthand data on the timing and the fiscal effects of tax reforms, we can study far more directly whether tax policy is linked to expenditure developments (financing hypothesis I and II), to deficits (financing hypothesis III and stabilization hypothesis II), or to real GDP growth (stabilization hypothesis I). We start with an analysis of annual data. Our independent variable is the net fiscal effect of tax reforms for each year in between 1965 to 2003.185 To make fiscal effects comparable over different years, we have divided them by nominal GDP. Our independent variables are – based on the financing and stabilization hypotheses – the growth rate of expenditure (t, t+1, t-1, t-2), the deficit over GDP (t, t-1, t-2) and real GDP growth (t, t-1, t-2) in the same period as well as with several time lags. We have estimated the following equation: D tt10 ??? ++=? ITt With: 0? , x? Constant, coefficients ?Tt Fiscal effects of tax reforms (adopted)/GDP IDt/t-1/t-2 Independent variables (expenditure growth rate; deficit over GDP, real GDP growth rate) with different lags t? Stochastic error term 185 We haven taken out the years 1964 and 2004 for these analyses because our data-set does not cover all tax reforms in these two years. We use tax reform data adjusted for temporary measures for our tests (see part III.2 for details). However, we have tested the hypotheses with unadjusted data as well and have derived very similar results. 103 R2 (adjusted) Dependant Variable: Explanatory Variables: DW Notes: t-statistics of the estimated parameters in parantheses. * significant at the 10% level; ** significant at the 5% level; *** significant at the 1% level Observations Testing normative theories of tax reforms in Germany 1965-2003 Net fiscal effects of tax reforms/GDP Deficit/GDPt-2 Real GDP growtht Expenditure growtht Expenditure growtht-1 Expenditure growtht-2 Expenditure growtht+1 Real GDP growtht-1 Real GDP growtht-2 Deficit/GDPt-1 Deficit/GDPt Constant i. -0.002327 (-1.580312) -0.00 1.87 39 0.000185 (0.985183) ii. -0.002853* (-1.895796) 0.02 1.98 39 0.000263 (1.394108) iii. -0.003443** (-2.264282) 0.06 1.84 39 -0.000346* (-1.848676) iv. -0.001129 (-0.771050) 0.03 1.84 39 -0.0000189 (-0.099581) v. -0.005394** (-2.586195) 0.09 1.69 39 -0.157856** (-2.229007) vi. -0.003455 (-1.623795) 0.01 1.85 39 -0.085207 (-1.1702271) vi. 0.001072 (0.511539) 0.01 1.85 39 0.089965 (1.240629) iix. -0.000242 (-0.155715) -0.01 1.85 39 -0.000397 (-0.819680) ix. 0.00000645. (-0.004142) 0.00 1.76 39 -0.000466 (-1.010055) x. -0.004173* (-2.751364) 0.12 1.98 39 - 0.001081** (2.427889) OLS OLS OLS OLS OLS OLS OLS OLS OLS OLS Table 6: Testing the financing and the stabilization hypotheses In Table 6 we display the results of our statistical analysis for bivariate linear relationships applying OLS estimations.186 The coefficient for expenditure growth with two time lags is significant at the 10% level but points at the opposite direction of the financing hypotheses (high expenditure increases led to tax reductions). Nonetheless, we find some support for the financing function. The deficit in the same period is significant on a 5% level and supports the financing hypothesis III: high deficits are correlated with tax burden increases via tax reforms in the same year. The coefficient of -0.16 indicates, that an increase of the deficit over GDP by one percentage point would lead c.p. to a tax burden increase of 0.16 pecentage points in the same year. Finally, we find that real GDP growth with two time lags is significant on a 5% level with a positive coefficient (which supports the stabilization hypothesis). However, the explanatory impact of all specifications is disappointing.187 Altogether, there is only very limited evidence for any of our hypotheses based on the financing and the stabilization function of tax policy in the annual data. 186 The Durbin-Watson test statistics for our OLS regressions (reported in table 6) do not support the hypothesis of serial autocorrelation in the error terms. 187 Different combinations of the discussed variables do not improve the evidence for the financing or the stabilization function either. See for a discussion Koester (2005). 104 2.1.3.3 Tests based on legislative periods One could argue that annual data are not suited to analyze the relationship in between expenditure developments, deficits, macroeconomic developments, and tax reforms as tax policy needs to be studied based on legislative periods to cover governments´ reactions to expenditure needs or stabilization goals. Therefore, we have calculated average annual values for all relevant variables for the 10 legislative periods covered in our data-set. We display our findings graphically in the next three figures. EXPENDITURE GROWTH AND TAX REFORMS -0,8% -0,6% -0,4% -0,2% 0,0% 0,2% 0,4% 0,6% 0 2 4 6 8 10 12 14 EXPENDITURE (AVG. GROWTH RATE P.A. - LEGISLATIVE PERIOD) FI S C A L E FF E C TS O F TA X R E FO R M S (L E G IS LA TI V E P E R IO D S , A V G . P .A .) -1990 -1980 -1972 -2002 -1987 -1976 -1994 -1998 -1983 -1969 over GDP % Own calculations based on: Federal Ministry of Finance (2004)/Federal Statistical Office(2007).Data on adopted tax reforms. Figure 47: Expenditure growth and tax reforms (based on legislative periods) With respect to expenditure growth (see Figure 47) we see that the relationship predicted by the financing function (expenditure growth is positively correlated with an increase of the tax burden via tax reforms) is reflected in our data on legislative periods, but that the correlation is not strong. With respect to fiscal deficits (see Figure 48) we find that high deficits are correlated with high tax burden increases. This speaks in favor of the financing function and not of the stabilization function. But here again the correlation is not strong. 105 -0,8% -0,6% -0,4% -0,2% 0,0% 0,2% 0,4% 0,6% -5,0% -4,5% -4,0% -3,5% -3,0% -2,5% -2,0% -1,5% -1,0% -0,5% 0,0% FISCAL DEFICIT/GDP (NATIONAL DEFINITION) FI S C A L E FF E C TS O F TA X R E FO R M S (L E G IS LA TI V E P E R IO D , A V G . P .A .) FISCAL DEFICITS AND TAX REFORMS -1990 -1980 -1972 -2002 -1987 -1976 -1994 -1998 -1983 -1969 over GDP Own calculations based on: Federal Ministry of Finance (2004)/Federal Statistical Office (2007).Data on adopted tax reforms. Figure 48: Fiscal deficits and tax reforms (based on legislative periods) Finally, we see in Figure 49 that real GDP growth is negatively correlated with tax burden increases via tax reform, contradicting the stabilization hypotheses (which had received some support based on annual data). Our analyses based on legislative periods is therefore supporting the results of our previous analyses based on annual data: there is some evidence for a weak correlation in between deficits and tax reforms as predicted by the financing function (high deficits are correlated with tax increases). For the stabilization function we find only some very limited evidence in the annual data and no evidence in the data based on legislative periods. Altogether the results do not point at an important role of the financing or stabilization function for tax reforms. 106 -0,8% -0,6% -0,4% -0,2% 0,0% 0,2% 0,4% 0,6% -1 -0,5 0 0,5 1 1,5 2 2,5 3 3,5 4 4,5 REAL GDP GROWTH (AVG. P.A., LEGISLATIVE PERIOD) FI S C A L E FF E C TS O F TA X R E FO R M S (L E G IS LA TI V E P E R IO D S , A V G . P .A .) GDP GROWTH AND TAX REFORMS -1990 -1980 -1972 -2002 -1987 -1976 -1994 -1998 -1983 -1969 over GDP % Own calculations based on: Federal Ministry of Finance (2004)/Federal Statistical Office(2007).Data on adopted tax reforms. Figure 49: Real GDP growth and tax reforms (based on legislative periods) 2.2 Tax reforms as a reaction to cold progression 2.2.1 General approach and hypotheses From a normative point of view, inflation can be linked to tax reform as well. In the progressive wage and income taxes (which are not inflation-indexed in Germany) inflation pushes up the tax burden by “bracket creeping” or “cold progression”: the nominal and real tax burden of individuals increases without tax policy changes, as wage and income inflation increases the applicable tax rates. Based on this cold progression one could expect that many tax reforms are just a reaction to inflation, moderating or even offsetting inflationary effects on the tax burden in the progressive wage and income taxes. Based on the mechanism of cold progression we would expect that the overall tax burden in the progressive wage and income taxes is reduced stronger in times of high inflation (cold progression hypothesis 1). Furthermore, we would expect that especially tax reductions in wage and income taxes are larger in times of high inflation (cold progression hypothesis 2).