Taxation for redistribution: Optimal tax structures along the utility possibility frontier

Abstract

Mirrlees (1971) examined taxation for redistribution with imperfect information - the government can observe earnings but not wage rates or hours worked. Mirrlees assumed a one-good model with a continuum of types. Stiglitz (1982) worked out the two-type example of the Mirrlees model. This paper extends the distance-function approach to the Ramsey (1927) problem in Burbidge (2025a) to study optimal tax structures along the utility possibility frontier, upf, with two types and two goods.