In this paper we revisit the mechanisms driving self-enforcement in quasi-voluntary compliance, thus contributing to an old puzzle in political economy: why do people pay taxes?. In particular, why do high-income citizens pay taxes even in the absence of auditing institutions? What explains the fact that actual tax compliance is higher than the level identified as a fair norm by experimental researchers? In addressing these questions, we propose a distributive logic of tax compliance. We argue that beyond arguments about the monitoring ability and the efficiency of state institutions and normative concerns for the worse-off in society, the distributive incidence of the public sector over the long run drives the level of compliance. We test the argument on the basis of a series of lab experiments in which subjects are randomly assigned to different fiscal treatments that approximate the variation in the design of national fiscal systems around the world. We find significant evidence that the distribution of net benefits (as determined by the progressivity of taxes and transfer) impacts levels of tax compliance.