Volume 226, Issue 4

Forward-looking measures like the well-known effective marginal tax rate developed by King and Fullerton (1984) rely strongly on standardized assumptions on the effective use of depreciation deductions. This paper derives a method of assessing these assumptions empirically and of quantifying the bias resulting from simplifying assumptions. Using a balanced panel of German firms we calculate the path of depreciation deductions that should be observed if the assumptions on depreciation allowances were correct. These hypothetical deductions are compared to the observable ones in the data. We find that observable deductions slightly exceed the hypothetical ones. Moreover, we find a considerable degree of heterogeneity in the actual use of depreciation deductions across firms. On the basis of these results we calculate an adjusted effective marginal tax rate. It turns out that even small estimation biases in determining the tax deductions have a large impact on the effective tax rates for marginal investment projects. We conclude that forward-looking measures can and should be assessed empirically and that future research should emphasize the heterogenous impact of tax-related disincentives on business investment.

Since attrition in the European Community Household Panel (ECHP) has cumulated to a considerable extent, there is concern that attrition biases empirical analysis. In this paper we compare the performance of four different strategies for estimating an earnings equation in the presence of panel attrition. By splitting the completely observed sample in one wave according to the response behavior of the following wave, we assess empirically the bias of an un-weighted, an inverse probability weighted, a Heckman and a matching estimator through bootstrap methods. Our findings lead us to several conclusions. First, for the example of Mincerian earnings equations, attrition is no matter of great concern when using the ECHP data. Second, the three estimation strategies, which correct for attrition based on estimated response probabilities, reduce the number of significantly biased parameters. Third, the correction strategies strongly increase the variance of the estimates through relying on estimated response probabilities and increase the relative bias. Hence, the reduction of significant biases is rather due to increased variance than due to lower biases. This result is confirmed when comparing the mean square error of the different estimation techniques. Therefore, for the estimation of income equations the uncorrected estimation based on respondents is suggested as the superior estimation strategy.

After conceptual clarification of “international business cycle” and a review of the literature, a new indicator is proposed. This indicator refers to two time series only and allows for an internationally comparable quantification of a country’s position in the business cycle. We then calculate times series of this indicator for 30 countries from 1970-2000. After some plausibility checks, we refer to these series to test a number of hypotheses. Cross correlations reveal a high degree of interconnectedness. Moreover, the number of highly positive correlations has increased over time, whereas the number of low and moderate correlations has decreased. A principal components analysis yields a first component that can be interpreted as the world business cycle. The further components suggest the existence of a Scandinavian-Anglo-Saxon business cycle as well as of another, smaller group of Anglo-Saxon countries that move together. This finding is replicated by a hierarchical cluster analysis, which in addition suggests a closely integrated group of non-Scandinavian and non-English speaking European countries plus Japan and Israel. Furthermore, there is indication for some, albeit weak business cycle integration in Southeast Asia and in South America. The international business cycle is thus found to have a hierarchical structure.

The paper shows that the assessment of the speed of productivity growth crucially depends on how one chooses to measure value added. According to a widely held view, the growth rate of labour productivity has increased significantly in the U.S. since the mid-90s. The U.S. is perceived to outperform most European countries in this respect by a wide margin. Comparing the U.S. with Switzerland, we show this view to rely - at least in part - on statistical artefacts.

Many studies’ regulatory aim is to have non-insurance entitlements of the social insurance system funded by taxes. Since most studies use (macro)econometric methods, changes in household behaviour are not shown and the allocative effects resulting from tax funding are not sufficiently taken into consideration. It is the intention of the present study to fill this gap. A numerical equilibrium model of overlapping generations with a disaggregated household sector is used in the study. Differentiation is made according to occupational groups and income brackets. Furthermore, the relevant interdependences between tax and social law are modelled at the household level. The study examines the impact of partial tax funding of benefits of the state health and pension insurance. Since the resulting efficiency effects are explicitly derived, it is shown that from an allocative point of view the combined funding from higher income and consumer taxes, which is often favoured for distributional purposes, is inferior to a pure funding from consumer taxes. Combined funding is not needed, since the primary assumption of a regressive burden on contributors resulting from a consumer tax increase does not arise in the simulation study. In addition, it is shown that partial funding of health care benefits by consumer tax increases is not associated with an excess burden. Of all the examined funding alternatives, such a partial funding seems to be superior due to allocative neutrality and the resulting distributional effects.

The pass-through of market rates to retail interest rates is generally found to be particularly slow in Germany compared to other countries. One popular explanation is the organisation of the banking system in three strictly segregated “pillars”: savings banks, credit cooperatives and private banks, and the low competitiveness of the first two of them. In this paper we analyse the differences of the interest rate pass-through between these banking groups. We employ a dataset covering (roughly) 30 banks’ retail interest rates of four standard banking products (mortgages, consumer credit, savings accounts and time deposits). In a panel ECM we first estimate reference models of the interest pass-through for the four products. In a second step they are augmented by dummies representing the respective banking group. We find remarkable differences in the interest rate pass-through: in general it is the big banks and savings banks reacting significantly quicker to changes in the market than regional banks and credit cooperatives. Hence, in contrast to the “common knowledge” of sluggish reactions of state banks, the savings banks take full part in competition. The credit cooperatives however, smoothing their retail rates, shield their customers from interest rate change risks.