131 tively. And the timing of adoption and implementation of reforms did not improve, but worsen the revenue position of governments in the short-run. Alltogether we found therefore no evidence for governments´ attempts to create fiscal illusion via the timing of tax reforms. 3.5 Opportunistic behavior 3.5.1 General approach and related literature Public choice has always stressed that self-interest cannot be excluded with respect to politicians. Politics should be seen “without romance” (Buchanan 2003). Holding an office has mutual benefits for politicians and creates strong incentives to focus on re-election. Government policies might therefore be opportunistic in trying to increase their re-election probabilities by economic policy before elections. Tufte (1978) and Nordhaus (1975) demonstrated in their classical contributions that governments increased transfers, decreased payments and tried to stimulate economic growth before elections to manipulate the electorate. Taxation does usually play only a minor role in the study of opportunistic behavior. But this results at least partly from the fact that reliable data on the timing and the fiscal effects of tax reforms has been scarce.235 3.5.2 Hypotheses We can directly derive three testable hypotheses based on the general argument of opportunistic government behaviour. If governments behave opportunistically, we would expect that they try to increase re-election probabilities by reducing the tax burden directly before elections while tax burden increases should take place directly after federal elections (opportunism hypothesis I). As citizens are more likely to react to tax burden changes and not just to tax policy decisions, governments should try to reduce the tax burden before elections by implemented and not so much by adopted tax reforms (opportunism hypothesis II). Finally, we would expect that governments focus on manipulation in the direct taxes (like wage and income taxes) where tax burden changes are more visible than in the indirect taxes (opportunism hypothesis III). 235 Whether such a strategy is promising and leads to an increase of re-election probabilities depends on the individual voting decisions. If information is complete and voters behave rational, they should be able to detect such a maneuver. However, as we focus here on an empirical observation, we primarily want to analyze whether opportunistic government behavior is reflected in the data. In the chapter on economic voting (see part V.4) we discuss whether we can find an effect of opportunistic behavior on electoral decisions. 132 3.5.3 Empirical analysis Our data-set from 1964 to 2004 covers 11 elections: 1965, 1969, 1972, 1976, 1980, 1983, 1987, 1990, 1994, 1998 and 2002. We have excluded the election in 1965, because we do not dispose of data on tax policy in the whole legislative period before this election. -1,0% -0,8% -0,6% -0,4% -0,2% 0,0% 0,2% 0,4% 0,6% 0,8% INCREASES REDUCTIONS BALANCE -1,2% -1,0% -0,8% -0,6% -0,4% -0,2% 0,0% 0,2% 0,4% 0,6% 0,8% INCREASES REDUCTIONS BALANCE TAX REFORMS AND THE ELECTORAL CYCLE, 1965-2004 FISCAL EFFECTS OF ALL ADOPTED REFORMS FISCAL EFFECTS OF ALL IMPLEMENTED REFORMS Election yearElection year +1 Election year +2 Election year +3 Fiscal effects of reforms/GDP Fiscal effects of reforms/GDP Own calculations based on: Federal Ministry of Finance (2004), Federal Statistical Office(2007). Election yearElection year +1 Election year +2 Election year +3 NOTE: adjusted for temporary measures; including 1991-1993; average values per year. YEAR YEAR Figure 59: Tax reforms and the electoral cycle How does the pattern of tax reforms look like in different years of the legislative periods? Do we find evidence for opportunistic motivations of governments in tax policy? Do governments try to manipulate re-election probabilities by the timing of tax policy or by implementation of tax reforms? Based on our data-set we have calculated the timing of adopted and implemented tax reforms within the legislative periods covered here.236 236 We coded the first calendar year of a new government as election year plus one so that the next election usually takes place in year four. Thereby the election year is (as elections usually take place between September and December) assigned to the incumbent government and it is assumed that tax policy of the new government starts in January after the federal election. The only exceptions to this coding are the elections in 1983 (March) and 1987 (January). As these elections took place in the first quarter of the year, 1983 and 1987 are coded as year 133 Figure 59 shows the timing of tax policy by fiscal effects of adopted reforms in the upper part and fiscal effects of implemented reforms in the lower part. In the aggregated data for the ten legislative periods, we see that the strongest average annual tax increases were adopted in the first year of a new government and are reduced then until the 4th year – the expected election year (see upper part of Figure 59). Adopted tax burden reductions are strongest in the second year followed by the first year of a legislative period. If we look at the balance of increases and reduction (reflected by the line in the figure), we see a tax burden relief in the second and the fourth year but an increase in the first and the third year. The lower part of Figure 59 shows the timing of implemented tax reforms. Here we see a less clear pattern of tax burden increases. The aggregated data shows that average annual increases, are high in the first three years and are not reduced before the fourth, the election year. Decreases on the other hand are growing from the first to the third year and still dominate in the fourth year. If we look at the balance of reductions and increases a very clear pattern comes out: The net tax burden is increased by implemented reforms in the first year of the legislative period (by on average 0.3% of GDP), then continuously reduced until the third year (the year before federal elections). After a reduction of the tax burden by on average 0.4% in the year before the elections, the net tax burden relief amounts to 0.3% in the election year. We observe this pattern not only in the aggregated average data, but very clearly in eight of the analyzed ten legislative periods.237 This clear pattern of the timing of tax reforms in the electoral cycle supports our opportunism hypotheses I and II. Governments are trying to increase their reelection probabilities by relieving the tax burden of voters in the election year and the year before the election. This strategy is pursued especially by the timing of implemented tax reforms and not so much by adopted tax laws. But what about our third hypothesis? Do we see different patterns depending on the kind of tax we analyze? In the discussion of our data-set, we found that reforms in the wage and income tax, in the VAT and in mineral oil and tobacco taxes account for 87% of the fiscal effects of all reforms. What do we see if we analyze the timing of reforms in these most important taxes? As we found in the overall analysis of fiscal effects of tax reforms, that opportunistic behavior was strongest with renumber one (election plus 1) of a new government. Of the ten legislative periods, eight lasted the regular four years. But the legislative period before the elections in 1972 lasted only three years (1970, 1971, 1972) and before the elections in March 1983 (1981 and 1982). We argue that the government expected to be in power for a “normal” legislative period in both cases. Therefore, we have coded 1982 as election year +2 and 1972 as election year +3. However, we have derived very similar results when we excluded the two “short” legislative periods before the elections in 1983 and 1972. 237 We found larger net tax burden increases in the election years than in the whole legislative period only before the elections in 1972 and 1994 – both times under extraordinary macroeconomic conditions. 134 spect to implemented reforms we focus here as well on data on implemented reforms. Figure 60 shows the effects of implemented reforms in wage and income taxes (upper part of the figure) as well as in VAT (lower part of the figure). In wage and income tax we see a clear pattern reflecting opportunistic behavior. Here, tax burden increases were high in the two years after the elections, reduced in the third year and were very small in the election year. Tax burden reductions were low in the first year after the election, but grew strongly in the second year after the election and were strongest in the year before the next elections. The resulting net fiscal reform effects were positive in the year after elections but decreased strongly until the year before the next elections. -1,00% -0,80% -0,60% -0,40% -0,20% 0,00% 0,20% 0,40% INCREASES REDUCTIONS BALANCE -0,04% -0,02% 0,00% 0,02% 0,04% 0,06% 0,08% 0,10% 0,12% 0,14% INCREASES REDUCTIONS BALANCE NOTE: adjusted for temporary measures; including 1991-1993; average values per year. TAX REFORMS AND THE ELECTORAL CYCLE, 1964-2004 IMPLEMENTED REFORMS: WAGE AND INCOME TAX IMPLEMENTED REFORMS: VAT YEAR YEAR Election yearElection year +1 Election year +2 Election year +3 Election yearElection year +1 Election year +2 Election year +3 Fiscal effects of reforms/GDP Fiscal effects of reforms/GDP Own calculations based on: Federal Ministry of Finance (2004), Federal Statistical Office(2007). Figure 60: Wage and income and VAT reforms and the electoral cycle With respect to VAT reforms, which were much smaller in total extent, the pattern looks different. Here the increases were strongest in the year before the elections and the balance of increases and reductions showed no evidence for opportunistic behavior.238 238 The VAT increases took place in the first year (1983, 1977), the second year (1992), the third year (1968) and the fourth year of an electoral cycle (1998). The fiscal effects of reforms over GDP have been strongest in 1968 when the VAT was introduced and the rate was increased 135 Fiscal effects of reforms of the other two important consumption taxes (mineral oil tax and tobacco tax) are shown in Figure 61. With respect to mineral oil taxes, where the tax burden has always been increased and never reduced, we find some indication for opportunistic behavior as increases in the year after elections dominate while increases are comparatively low in the other years. For the tobacco tax on the other hand (the lower part of the figure), there is no evidence for opportunism as the most important increases took place in the second and the third year of the legislative period and there were no increases in the election years or the years after the elections. -0,01% 0,00% 0,01% 0,01% 0,02% 0,02% 0,03% 0,03% 0,04% INCREASES REDUCTIONS BALANCE -0,02% 0,00% 0,02% 0,04% 0,06% 0,08% 0,10% 0,12% 0,14% INCREASES REDUCTIONS BALANCE^^^^^^^^ TAX REFORMS AND THE ELECTORAL CYCLE, 1964-2004 IMPLEMENTED REFORMS: TOBACCO TAX IMPLEMENTED REFORMS: MINERAL OIL TAX YEAR YEAR Election yearElection year +1 Election year +2 Election year +3 Election yearElection year +1 Election year +2 Election year +3 Fiscal effects of reforms/GDP Fiscal effects of reforms/GDP NOTE: adjusted for temporary measures; including 1991-1993; average values per year. Own calculations based on: Federal Ministry of Finance (2004), Federal Statistical Office(2007). Figure 61: Mineral oil and tobacco tax reforms in the electoral cycle (see our treatment in part IV.2.3.). If we exclude 1968 (or the whole reign of the grand coalition from 1966 to 1969), the fiscal effects of VAT reforms in the year before the elections are coming down to zero which would somewhat increase the evidence for opportunism. Nonetheless, the very similar effects in the first, the second and the fourth year of the legislative period do not speak in favor of opportunistic behavior in VAT. 136 3.5.4 Summary of results We found that governments are trying to increase their re-election chances by reducing the tax burden in the election year and even more in the year before the election. Governments are scheduling the implementation (and not so much the adoption) of tax reforms in an opportunistic way. These findings support our first two hypotheses on opportunistic government behavior. Furthermore, opportunistic behavior is strongest in the wage and income taxwhere tax reform effects are most directly visible to voters. Strong reductions in the wage and income tax are implemented especially in the year before the elections, while increases are concentrated in the two years after the elections. On the other hand we find only very limited evidence for opportunism in the less visible taxes. Only the mineral oil tax increases were scheduled after elections, while opportunistic behavior played no role with respect to the tobacco tax and the VAT. These findings support our third opportunism hypothesis. Altogether we therefore find strong support for all three opportunism hypotheses. 3.6 The partisan politics approach 3.6.1 General approach and related literature While all approaches discussed so far focus only on the re-election motives of governments, partisan approaches (inspired especially by the early work of Hibbs (1977)) argue that party governments and political parties can be characterized by distinctive ideological goals resulting from differences in their clienteles. Based on the partisan politics approach, left-wing governments are expected to pursue different economic policies than right-wing governments. The influence of partisan orientation on economic policy and economic outcomes is studied in the literature mainly with respect to monetary policy, government expenditures and deficit spending.239 The political business cycle theory240 focuses on partisan influence on macroeconomic outcomes via monetary policy and government expenditure.241 Another branch of literature restricts itself to analyzing how partisan preferences of governments are affecting public expenditures without discussing macroeconomic consequences.242 Finally following Cowart ?s 1978 study, 239 Applications of partisan theory to Germany can be found e.g. in Cusack (1995), Lehmbruch (2000), Renzsch (1998) or Schmidt (1996). 240 Inspired e.g. by Nordhaus (1975), Hibbs (1977), Tufte (1978), Paldam (1979) or Alesina, Roubini and Cohen (1997). 241 See for an overview Belke (1996). 242 See e.g. Swank (1988).