Wage and Productivity Dispersion: Labor Quality or Rent Sharing?

The method introduced in this paper contributes to the estimation of production functions by embedding them in a structural equation system involving worker, firm, time, and occupation effects from an individual wage decomposition and accounting for labor input components that are substitutes and complements, while accommodating stochastically varying factor productivity. The structural model allows for differences in input quality as well as rent sharing, both of which may potentially help explain the observations that wage and labor productivity are dispersed across firms, and that more productive firms tend to pay higher wages. From our empirical results, which focus on the manufacturing sector only, both input heterogeneity and intrinsic differences in total factor productivity across firms are important for dispersion. We find that 41% of the dispersion in log value added per worker within the manufacturing sector is attributable to cross-firm differences in the levels of capital per worker, while another 39% of the variation stems from intrinsic TFP differences across firms. Only a smaller portion observed dispersion in log value added per worker (5s%) is associated with quality differences in the labor input. These results suggest that that there are major gains to reallocation of labor from firms with low marginal labor productivity to firms with high marginal labor productivity. The same would not be the case if the dispersion in marginal labor productivity were due to differences in labor quality alone. Rent sharing provide a link between individual wages and firms' marginal labor productivity and thus ties the dispersion in labor productivity to the wage distribution. Hence, rent sharing is a potentially important vehicle for reallocation as wage dispersion motivates job search. We provide a decomposition of individual log wages in the manufacturing sector and find that 70% of the individual log wage variation is due to individual characteristics, whereas only 13% is attributable to firm differences (i.e. rent sharing). The relatively small contribution of firm heterogeneity to individual wage dispersion is thus consistent with the inefficient allocation of labor across firms.