مالیات مطلوب و مالیات کار OECD

We derive simple expressions for optimal labor taxes under different assumptions about government bond markets. We use these to examine OECD labor taxes, estimate the excess burden of taxation and assess the ability of optimal tax models to match the data. Optimal labor taxes are driven by: (i) a term reflecting Ramsey considerations which makes labor taxes vary positively with employment and (ii) a martingale component, reflecting the excess burden of tax, which shows persistent responses to shocks to the government's intertemporal budget constraint. Under complete markets (when governments can issue a full set of contingent securities) only the first factor is relevant. We find substantial evidence that incorporating incomplete markets into the optimal taxation model is critical for empirical success. However, we find strongest support for the martingale component and only weak evidence for the Ramsey component.

As shown in Fig. 1, labor tax rates1 show substantial variation across time. To what extent does the canonical tax smoothing model (as outlined in Ramsey, 1927, Barro, 1979 and Lucas and Stokey, 1982) account for these variations? In attempting to answer this question recent research has stressed the importance of assumptions about the government bond market, as the stochastic properties of optimal taxes are not invariant to the nature of government financing (see Aiyagari et al., 2002). Therefore evaluating empirically the tax smoothing model involves assessing the relative importance of incomplete markets for fiscal policy. Full-size image (67 K) Fig. 1. Marginal labor tax rates. Figure options The purpose of this paper is twofold. Firstly, it builds on the work of Zhu (1992) and Aiyagari et al. (2002) to show clearly the determinants of optimal taxes under both complete and incomplete markets. Under complete markets governments can issue a full range of Arrow Debreu contingent securities whereas under incomplete markets this is not the case. Using conventional assumptions about preferences and production and extending the model of Aiyagari et al. (2002) to include capital accumulation and productivity shocks we provide simple analytical expressions which lay bare the intuition behind optimal taxes under different assumptions regarding the structure of bond markets. Our extensions to the specific models used by Lucas and Stokey (1982) and Aiyagari et al. (2002) enable us to provide a more general insight into the relationship between taxes, government expenditure and the deficit. The second aim of this paper is to use OECD tax data to assess the validity of optimal tax models and the relative success of complete and incomplete market models. The plan of the paper is as follows. Section 2 outlines our model and derives the behaviour of labor taxes under the assumption of complete markets. Section 3 introduces incomplete markets and shows how this alters the stochastic behaviour of optimal labor taxes. Section 4 uses simulations to illustrate the differences between complete and incomplete markets allowing for both capital accumulation and productivity and expenditure shocks. Section 5 examines the evidence for optimal taxes and incomplete markets in the behaviour of OECD labor taxes and a final section concludes. 2. Optimal taxes under complete markets In this section we follow Lucas and Stokey (1982), Zhu (1992) and Chari et al. (1994) in considering the case of complete markets where governments can issue a full set of Arrow Debreu contingent claims. Under this assumption the government can insure itself, and thereby its tax plans, against unexpected shocks to government expenditure.

Using a standard business cycle model we have derived simple approximate expressions which show clearly the influences on optimal fiscal policy. Optimal taxes have to change for two reasons: (i) Ramsey considerations aimed at minimising the distortionary costs of taxation and (ii) financing considerations which ensure that the government expects its intertemporal budget constraint to hold. When bond markets are complete only the former matters and labor taxes depend only on employment. The result is that labor taxes are mean stationary and subject to predictable fluctuations. Under incomplete markets, optimal taxes still depend upon employment but also an additional martingale component. This term further increases the volatility of taxes, makes labor taxes harder to predict and reduces the ability of the government to operate a countercyclical fiscal deficit. We show that these differences between complete and incomplete markets are robust across a range of different specifications including the addition of productivity shocks, persistent shocks and capital accumulation. Our simulations suggest that using unit root tests to detect optimal taxes and incomplete markets will suffer from low power. We use our theoretical expressions to examine OECD labor tax rates and find numerous features that suggest only the incomplete market version of optimal taxation can account for the data—namely, the persistence of labor taxes, the difference between the serial correlation properties of employment and taxes, the crosssectional connection between tax and expenditure volatility and the observed persistence in our measures of the excess burden. The implications of our incomplete market model are accepted by the data although we find only a very weak role for the Ramsey employment effect. It would appear that variations in the excess burden of taxation dominate tax rates rather than Ramsey related employment fluctuations.