Job Tenure, Wages and Technology: A Reassessment Using Matched Worker-Firm Panel Data

This Paper presents new estimates of the impact of job tenure on wages using a new French matched worker-firm dataset. We develop an identification strategy that relies on one specific feature of the French labour laws. They stipulate that firms, when firing workers, must include as one of their criteria the number of dependent children of their employees. Our dataset confirms that workers with a relatively large number of dependent children at the entry in the firm are, ceteris paribus, less likely to be laid-off and have on average higher job tenure than their co-workers with a relatively small number of dependent children. Within this framework, the relative number of children at entry in the firm represents a plausibly valid instrumental variable for identifying the impact of job tenure on wages. Our new IV estimate of the return to job tenure (3.1% per year) is much larger than the OLS estimate (1%). This result holds true regardless of whether we focus on educated or non-educated workers, men or women. The downward bias which affects OLS estimates suggests that workers who receive relatively high wage offers tend to change firms more rapidly: they tend to have relatively high wages and low job tenure. Regarding trends, our new IV estimator suggests that the returns to job tenure have declined over the 1990s in the industries where the share of educated workers is the largest. The technologies that complement highly skilled labour seem to drive a decline in the incentive to keep workers over long periods of time and, as a consequence, a decline in the impact of tenure on wages.