124 3.2 Polit-economic approaches to tax policy – overview There is not one unifying theory of the political economy of taxation but many different approaches that try to explain tax policy. Some of these approaches can be combined, but some are also competing. Furthermore, the approaches differ largely in their scope. While some are only focusing on one aspect of tax policy (for example the government’s motivation to try to influence re-election probabilities), others are far more encompassing. This makes the models are often hard to compare. Furthermore, there are barely any empirical tests which compare the theories and their empirical relevance directly. This is probably the decisive reason, why the field has stayed so fragmented. Therefore, the opportunities for empirical testing based on our new data-set are especially valuable for comparing the empirical relevance of different approaches. In this section we review five major approaches from which we can directly derive testable hypotheses. We start with the theory of inertia, followed by a discussion of the concept of fiscal illusion. Then we move on to approaches with opportunistic and partisan motivation in tax policy and analyze the effects of divided government within the German bicameral legislature. Finally, we test combinations of several hypotheses statistically (part V.3.8) and summarize our results. Overview – polit-economic theories of tax policy and tax reform Opportunistic behavior Governments try to increase their re-election probabilities by tax policy Tax burden increases will take place after, tax burden reductions before elections Can be tested directly based on tax reform data Alternative approaches Fiscal Illusion by timing of tax reforms Governments use timing of tax reforms to camouflage real tax burden changes A larger share of tax increases (compared to tax reductions) will be enacted retrospectively (and with a larger time-lag) Can be tested directly based on tax reform data Inertia/theory of reform Incentive to avoid tax reforms (and especially tax increases) because of high political costs Minimize political costs and maximize revenue by self-restriction to automatic tax revenue increases based on cold progression (inertia in tax policy) Can be tested directly based on tax reform data Partisan behavior Ideologically motivated redistribution via tax reforms Right wing: Reduce the tax burden and shift it to regressive taxation Left-wing: Increase the tax burden and shift it to progressive taxation Can be tested directly based on tax reform data Institutional factors (divided government) Divided government leads to political gridlock Extent and frequency of tax reform activity will be larger in times of undivided government M ec ha ni sm Pr ed ic tio ns Te st in g Can be tested directly based on tax reform data Models of bureaucracy Interest group models Probabilistic voting on taxation Median voter model Leviathan model Table 8: Polit-economic approaches to the study of tax reform 125 Our analysis leaves out important approaches from the scientific literature on the political economy of taxation as the median voter model,225 the Hettich-Winer model,226 interest groups models,227 approaches which analyze the role of the bureaucracy for tax policy,228 and the Leviathan model229. Our decision to leave these approaches out should not be mistaken as a statement against the relevance of these approaches. Instead we are only less able to derive hypotheses from these approaches which would be directly testable based on our data-base of fiscal effects of tax reforms. Before we start to discuss and test hypotheses, we shortly review some descriptive statistics of tax reforms depending on political variables. Table 9 shows descriptive statistics of tax reforms from 1965 to 2003230 depending on political variables. We see that the number of new regulations was lower in election years than in average years. This resulted largely from a lower number of tax increases. In case of undivided government the numbers look very similar to those in election years. With respect to partisan politics, the most important observations is that there was a lower number of tax burden increases in case of undivided right government. Furthermore, the grand coalition (governing from the end of 1966 225 See for the basic model Black (1948) or Downs (1957). For a general application to taxation see Meltzer/Richard (1981), for a special application Sjoquist (1981) and for an encompassing discussion of the model Congleton (2002). 226 See e.g. the comprehensive discussion in Hettich and Winer (1999). Hettich and Winer model tax policy based on a probabilistic voting model as the equilibrium outcome of a political and economic system. In their model tax policy leads to an equalization of marginal political costs and marginal political benefits for all citizens. This equalization predicts a complex tax system with a large number of tax bases, tax exemptions and group specific treatments. Tax reforms reflect responses of political decision makers to exogenous shocks (see Hettich/Winer 2004) affecting political influence of different groups, changes in administrative costs or changes in the Laffer-curves triggered by shifts in between tax bases or the occurrence of new prevention possibilities. Empirical tests of the model are still scarce. Hettich and Winer (Hettich and Winer 1999, chpt. 11) themselves demonstrate in a comparative study that the lower transaction costs in terms of lobbying and coordination in the Canadian compared to the US- American system allowed for a more rapid and faster adjustment of the tax system. Kenny and Winer (2001) apply the model in a cross-country study and try to explain tax structures based on variables like real GDP per worker, women in the labor force, oil extraction, magnitude of international trade, urbanization, population density, educational attainment, secondary enrollment and socialist government. However, with respect to tax reform the model is not sufficiently spelled out yet (see Hettich and Winer 2004). 227 For the interest group approach see Becker (1983) and with respect to taxation Holcombe (1997) or Witte (1985). For a textbook treatment see Mueller (2003), chpt. 20. 228 See for the general approach Niskanen (1971). The role of bureaucracy in tax reform is discussed in Bird (2003 and 2003a), administrative issues are analyzed in Sandford et al. (1989). See for the role of bureaucracy as well Hettich/Winer (2004) and for applications Findling (1995) or Mann (1987). 229 See for the approach Buchanan (1975), Nelson (1986/1987) and Oates (1989). Applications to taxation are Kau/Rubin (1981) and Breeden/Hunter (1985). 230 We have excluded 1964 and 2004 because our data covers only parts of these years. 126 to the end of 1969) was – based on the number of reforms and new regulations – less active in tax policy. However, these observations are only preliminary. In the following parts we analyze the developments in more detail based on the fiscal effects of reforms, which are far better suited for a comparative analysis of tax policy under different political constellations than just the number of reforms and regulations. Descriptive Statistics – Adopted German tax reforms (1965-2003) Constellation AVERAGE (All years) Election year* Right Right and undivided Partisan Left Undivided Government** Grand Coalition 5.2 # of reforms p.a. 5.6 6.5 4.5 4.9 6.7 6 5.9 10.3 # of reductions p.a. 12.8 10.6 15.3 11.4 11.4 6.7 14.1 2.8 # of increases p.a. 10.1 3.3 9.0 2.8 12.3 2.7 13.7 14.3 # of new regulations p.a. 26.2 16.1 25.9 15.8 29.3 9.3 32.2 13 # of years 39 11 18 10 18 3 26 Extent of increases/ GDP p.a. 0.20% 0.44% 0.19% 0.44% 0.18% 0.46% 0.28% 0.56% Total fiscal effects/ GDP p.a. -0.21% -0.13% -0.13% 0.09% -0.31% -0.20% 0.12% -0.08% Extent of Reductions/ GDP p.a. -0.42% -0.57% -0.32% -0.64% -0.53% -0.49% -0.66% -0.16% Note: Number of new regulations without fiscal effect not displayed. 1965, 1969, 1972, 1976, 1980, 1982 (election in March 1983), 1986 (election in January 1987), 1990, 1994, 1998, 2002;. ** Including grand coalition. Own calculations based on: Federal Ministry of Finance (2004)/Federal Statistical Office(2007). Divided Government Table 9: Descriptive statistics of tax reforms by political constellation 3.3 Status quo bias and inertia in tax policy 3.3.1 General approach and related literature How much activity are we expecting if we chose a polit-economic perspective including self-interested politicians. From a polit-economic point of view tax reforms, which increase the tax burden, are attractive for governments because of the additional public goods and redistributive programs which can be financed with the additional tax revenues. These might help to increase the re-election probabilities. On the other hand, almost all changes in the tax system and especially overall tax burden increases create not only winners but as well losers who have to bear the