Using household survey data on expenditures and incomes, we construct an objective measure of corruption in the public sector for a broad spectrum of countries. Specifically, we focus on the consumption-income gap for public sector workers relative to private sector workers to gauge the extent of hidden income (bribes) in the government. After validating our data and documenting properties of the consumption-income gap, we compare our measure with popular corruption perception indices. We find that i) the relationship between our measure and the alternatives is nonlinear; ii) available indices appear to be only weakly (and sometimes “wrongly”) correlated with the consumption-income gap at high frequencies; iii) the available indices appear to have a low weight on the relative consumption-income gap in the public sector.