In this paper I readdress the result that capital income taxes are bad instruments for pure redistribution and should be zero in the long run. In a neoclassical growth model a capital income cum investment subsidy tax, which is not distorting accumulation, is considered to investigate if net capital income taxes used for pure redistribution are zero in a long-run optimum. I find that capital income taxes may be nonzero, depending on the political power of those who receive redistributive transfers, the distribution of pre-tax factor incomes, and the intertemporal elasticity of substitution.

JEL - Classification: O41, H21, D33