Abstract Mirrlees’ Nobel Prize-winning theory of optimal taxation has initiated a long series of studies aimed at its clarification and development. However, the following issue is still not resolved: the optimal scale of taxation in the Mirrlees paradigm based on maximizing aggregate utility (built from individually optimized utility functions) does not generally lead to progressive taxation (which contradicts the practice of developed economies) and often assigns minimum tax rates to highly paid segments of society. The first purpose of this paper is to confirm this claim by proving the theorem on optimal tax scale in the (practically most popular) piecewise linear setting for a simplest utility function. The second goal is to propose a new paradigm of optimal taxation, in which, in addition to the mean utility, a second parameter is introduced, viz., the standard deviation of utility. We demonstrate that this approach leads to transparent and easily interpretable criteria for optimal taxation.
Dranov et al. (Sun,) studied this question.