DECOMPOSING WAGE VARIATION - A COMMENT ON KEANE,MICHAEL,P. INDIVIDUAL HETEROGENEITY AND INTERINDUSTRY WAGE DIFFERENTIALS

Keane (1993) uses panel data to control for the effects of time-invariant individual characteristics when estimating the effects of industry on wages. He concludes that these individual effects can account for 84 percent of the industry-associated variation found in typical cross-section studies. I argue that this conclusion is based on a misleading wage variance decomposition that would tend to overstate the importance of individual effects. A reconsideration of Keane's results shows that his estimates are of a similar magnitude to those of earlier studies that attempt to control for individual ability.