36 III A new data-set of tax reforms in Germany The empirical literature on tax policy and tax reforms employs many different data sources. In this part we start with a discussion of the different kinds of data which are available and then focus on a new data-set of tax reforms that has not been analyzed in detail in the literature so far. The data-set covers tax reforms in Germany from 1964 to 2004 and consists of more than 1,000 new tax regulations (including their fiscal effects). First we present the new data-set and test its reliability part II while analyses of tax reforms based on the data-set follow in part IV. 1 Empirical approaches to the analysis of tax policy In the literature on tax policy and tax reforms we can distinguish seven different kinds of data available for empirical analyses (compare Table 2). Empirical analyses of tax policy: available data i i l l li il l Data-sets based on: t - t : Advantagest Disadvantagesi t Revenue data (including effective tax rates based on revenue data) t (i l i ff ti t x r t s s r v t ) Easily accessible; inclusion of tax rate and tax base changes il i l ; i l i f t x r t t x c Rates based on macroeconomic data cannot discriminate in between tax policy and other effects t r i t t i ri i t i t t li t r ff t Data-set on tax laws based on the „Bundessteuerblatt“ (including number of new laws/pages of new regulations) t - t t l s t „ t r l tt“ (i l i r f l / f r l ti ) Tax policy outcomes (timing, number of laws …) directly observable x licy tc s (ti i , r f l ) ir tl r l No quantitative indicator for the importance of different laws tit tiv i ic t r f r t i rt f iff r t l s Effective tax rates (simulation method) ff ti t r t ( i l ti t ) Inclusion of tax base and tax rate changes possible I l i f t t r t c i l Simulation methods include only special cases/constellations i l ti t i l l i l /c t ll ti Data-set on tax reforms (including fiscal effects of reforms) t - t t r f r (i l i fi l ff t f r f r ) Tax policy outcomes (extent and direction) directly observable; far-reaching quantitative analysis possible li y t ( xt t ir cti ) ir tl r l ; f r-r i tit ti l i i l Possible data bias based on political interests of governments i l t i liti l i t r t f r t Nominal tax rates i l t r t Easily accessible; direct study of tax policy outcomes il si l ; ir t t f t li t Limited informational value: Tax base changes are not included i it i f r ti l l : s s r t i l Tax laws l Direct study of tax policy is possibleir ct st y f t x licy is ssi l Limited to case studies; Quantitative analysis nearly impossible i it t t i ; tit ti l i rl i i l Legislative statisticsi l tiv t ti ti Direct study of the tax policy process is possible ir t t f t t li r s i s i l Tax laws need to be separated; no quantitative indicator for importance of tax reforms l t s r t ; tit ti i i t r f r i rt f t x r f r Table 2: Data for empirical analyses of tax policy 37 The first source of data are tax laws and tax law changes. The major advantage of these data is that tax policy can be studied directly: Changes in tax regulation can be analyzed in detail and be linked for example to political or macroeconomic developments. The main disadvantage is that such an approach usually has to be restricted to a limited number of reforms40 or to an overview.41 A quantitative analysis is impossible because of a lack of suitable indicators. The second source of data are nominal tax rates. The main advantages of nominal tax rates are that they are easily observable and accessible42 and do reflect policy outcomes. Furthermore, quantitative comparisons and analyses become possible.43 The major disadvantage is that the informational value of nominal rates alone is very limited. Most importantly, changes in the tax base, which directly affect the tax burden and are often more important than tax rate changes, are not integrated. To overcome the limited informational value of nominal tax rates alone, a third kind of data has been developed: the calculation of effective tax rates based on simulations. Examples are the effective average tax rates in corporate profit taxation, calculated by the Center for European Economic Research (ZEW).44 These simulations are able to integrate effects of changes in tax rates and tax bases on the tax burden in specified economic settings. The disadvantage of these approaches is that they are only available for a limited number of taxes, a limited number of specified settings, and for a limited time horizon which makes an analysis of developments in a longer timeframe impossible. A fourth kind of data to analyze tax policy are tax revenues. Examples are the effective tax rates for capital, labour and consumption as calculated for example by Mendoza et al. (1994) based on OECD revenue data. Advantages of these data are that they allow for quantitative analysis, are easily accessible for almost all taxes and long periods of time, and that tax rate and tax base changes are both reflected within revenue developments. The most important disadvantage of the data is that revenue developments reflect not only tax policies, but as well the macroeconomic developments and a large number of other factors.45 Therefore, it is hard to derive insights on tax policy based on revenue data. 40 See e.g. for Germany Findling (1995), Zohlnhoefer (1999) and Ganghof (2004) or for the US Ballentine (1992), Pechmann (1987), and Buchanan (1987). 41 See e.g. Muscheid (1986), Offerhaus (1997), and Johann (2006) or for an international overview Koren (1989). 42 For international comparisons a large data-set is compiled for example in the “World tax database” of the University of Michigan. See: http://www.bus.umich.edu/otpr/otpr/introduction.htm. 43 See e.g. Boix (1998) or Garrett (1998). 44 See Jacobs (2002) and Jacobs/Spengel (2003). 45 Our detailed analysis of revenue developments by kind of tax in part IV shows for example that substantial revenue changes, which are not caused by macroeconomic changes or tax policy are very frequent. 38 A fifth source of data are legislative statistics, which are available from the German parliament, the Bundestag.46 The advantage of these data is that they allow for quantitative analyses of the legislative process. However, these quantitative analyses make far-reaching transformations of the data necessary. While some studies based on these data exist,47 a separate study of tax legislation is still missing. A sixth source of is the “federal tax journal” (“Bundessteuerblatt”) which is subscribed to especially by tax lawyers and tax consultants. Most tax laws changes (with exception for example of changes in some consumption taxes) are published in the federal tax journal. Blasch et al. (2006) have compiled a data-set on tax laws based on the federal tax journal that includes quantitative indicators like number and extent of tax law changes (by number of pages).48 The advantage of this data-set is that is makes direct quantitative analysis of tax policy possible. The main disadvantage is that the data-set does not include information about the direction of tax policy changes (increases or reductions of the tax burden) and that the data-set does not include an indicator to compare the importance of different tax reforms Finally, the seventh possible source of data (which we concentrate on here) is data from the Federal Ministry of Finance on the most important tax policy changes and their fiscal effects. These data are published in the annual reports of the Ministry and summarized in a new data-set for the time from 1964 to 2004.49 Compared to studies based on data from the federal tax journal this data-set has the important advantage that we have a quantitative indicator (fiscal effects) which allows us to assess the importance of different reforms as well as their direction (reduction or increase of the tax burden). The disadvantage is that we cannot know per se whether the data on fiscal effects of tax reforms (which is calculated and published by the Federal Ministry of Finance) are reliable or strongly biased. Therefore, we first describe and transform the data-set (see part III.2) and then discuss its reliability (part III.3). 2 A new data-set on tax reforms in Germany The German Federal Ministry of Finance offers highly promising data for empirical research in tax policy. The fiscal reports – published annually by the German Federal Ministry of Finance – include the most important tax law changes and since 1964 the Federal Ministry of Finance has published estimates of the expected fiscal effects of tax reforms and new regulations for the first 12 months of being fully 46 See for example the data handbook of the Bundestag: http://www.bundestag.de /bic/dbuch/start.pdf. 47 See e.g. the discussion in Burkhart/Manow (2006). 48 Blasch et al. (2006) have analyzed the timing of new regulations, the development of complexity (as measured by number of pages) and the influence of divided government on tax policy in Germany from 1950 to 2005. 49 Analyses of this data-set are rare so far. An analysis with respect to income taxes is Corneo (2005) and some more general points can be found in Wagschal (2005).