Past research demonstrates that the estimated size of the federal government earnings differential shrinks substantially with the addition of detailed occupational controls. Possible explanations for this reduction are: controlling for the differing sectoral distributions of common occupations, and controlling for detailed occupations unique to each sector. While occupational detail does not eliminate the federal differential, more than two-thirds of the reduction in the federal differential comes from controlling for unique occupations and, moreover, this is equivalent to excluding all observations in unique occupations. This finding is contrasted with that for the local sector in which the differing distribution of common occupations largely explains the pattern of the differential. (C) 2004 by the Association for Public Policy Analysis and Management.