Comparing micro-evidence on rent sharing from two different econometric models

The extent to which employers share rents with their employees is typically assessed by estimating the responsiveness of workers' wages on firms' ability to pay. This paper compares rent-sharing estimates using such a wage determination regression with estimates based on a productivity regression that relies on standard firm-level input and output data. Using a large matched firm-worker panel data sample for French manufacturing, we find that the respective industry distributions of the rent-sharing estimates are correlated and slightly overlap, but are significantly different on average. Precisely, if we only rely on the firm-level information, we obtain an average rent-sharing estimate of roughly 0.30 for the productivity regression and 0.17 for the wage determination regression. When we also take advantage of the worker-level information to control for unobserved worker ability in the model of wage determination, we find as expected a lower average value of 0.10.