A central question in labor economics and macroeconomics is whether the textbook competitive model provides an adequate representation of the labor market. Using longitudinal data on companies and establishments, this article suggests that it may not. As predicted by rent-sharing models of the labor market, changes in profitability are shown to feed through into long-run changes in wages. These are not temporary wage effects and are not driven by the unionized workplaces in the data. The article's estimates imply that, for rent-sharing reasons alone, Richard A. Lester's (1952) 'range' of wages is approximately 16 percent. Copyright 1997 by University of Chicago Press.