This paper shows that there is a presumption that Pareto efficient taxation entails a positive tax on capital. When tax and expenditure policies can affect the market distribution of income, those effects need to be taken into account, reducing the burden imposed on distortionary redistribution. The paper extends the 1976 Atkinson-Stiglitz results to a dynamic, overlapping generations model, correcting a misreading of the result on the desirability of a zero capital tax. That result required separability of consumption from labor and that the only unobservable differences among individuals was in (fixed) labor productivities. In a general equilibrium model, one needs to take into account the effects of policy changes on binding self-selection constraints; and with non-separability, capital taxation depends on the complementarity/substitutability of leisure during work with retirement consumption. The final section considers taxation when there are constraints on the imposition of intergenerational transfers (either political constraints or those derived from unobservability.) It constructs a simple two class model, capitalists who maximize dynastic welfare and workers who save for retirement, whose productivity can be enhanced by (publicly provided) education. It derives a simple expression for the optimal capital tax, which is positive, so long as the social welfare function is sufficiently equalitarian and the productivity of educational expenditures are sufficiently high.
↧