A Reconsideration of the Optimal Income Tax.

This paper develops a formula for the optimal nonlinear income tax, the terms of which are familiar from the theory of linear income taxation. The development uses the idea of a perturbation of the optimal schedule and is based upon as assumption of differentiability. It is also shown that the introduction of non-differentiability, implying bunching of taxpayers, may be desirable and, in this case, the optimal schedule may be difficult to determine. The analysis can be applied to nonlinear tax reform