Wage differentials across labor markets and workers: Does cost of living matter?

Wage differential studies rarely account for interarea differences in cost of living, owing both to data limitations and theoretical ambiguity. This s tudy develops a price index for 185 metropolitan areas comprising about 70% of the U.S. labor force. Current Population Survey data for 1985-95 and da ta on site-specific amenities are used to estimate earnings differentials b ased on nominal wages, wages fully adjusted for measured cost of living, an d a simple approximation of "real" wages with partial adjustment for price- level differences. Dispersion in approximate real wages across 185 labor ma rkets and differentials by region and city size are substantially lower tha n dispersion in nominal or full adjustment wages. Estimates of racial and e thnic differentials display moderate sensitivity to choice of a wage measur e, whereas other standard differentials do not. Both nominal wages and wage s fully adjusted for cost of living may provide misleading estimates of rea l wage differentials. Absent data on interarea prices and amenities, resear chers should include detailed controls for region and city size in nominal wage equations. (JEL J31, R23).