Bias in estimating returns to tenure: The impact of monopsony power

This note revisits the bias in returns to tenure estimates (RTT) that arises when firm employment and wages comove through firm-specific (FS) wage shocks - shocks to wages that likely occur in the presence of monopsony power(MP). In an earlier paper, Snell et al. (2018), found that adding firm-year fixed effects to a Mincer regression for workers in a small sample of firms in German and Portuguese panel data raised the estimated RTT schedules by about 30%. Here we re-examine the sign and magnitude of this bias using the universe of workers in Portugal. We find the bias remains negative and large. We then split the data into low and high labour market concentration firms to examine the role MP plays in these results. We find that the RTT bias arising from FS wage shocks is larger in firms with high MP than low MP. The results reinforce the need to control for FS wage shocks in RTT estimation exercises especially where comparisons across different workers in different firms are made. • Firm specific wage shocks cause downward bias in traditional returns to tenure estimates. • Bias remains even when controlling for seniority and match quality. • The bias is far greater in firms with higher monopsony power. • Monopsony power is a likely source of the firm specific wage shocks causing the bias. • Traditional (Mincer) regressions must control for firm specific shocks to avoid bias.