Theories of turnover and wage dynamics have studied the impact of wage levels on turnover, but they have failed explicitly to model the role of wage growth in predicting turnover. This article presents a theory of turnover that explains why within‐job wage growth reduces the likelihood of worker‐firm separations. The model determines the evolution of value among jobs that differ systematically in permanent rates of wage growth and shows that the value of high wage‐growth jobs increases faster. With additional assumptions about the search process, this proposition implies that high wage‐growth jobs are less likely to end.

Turnover falls with tenure, but wages do not always rise (and sometimes fall) with tenure. We reconcile these findings by revisiting an old issue: how gains from firm‐specific training are split between workers and firms. The division is determined by a stationary distribution of outside offers. The lower the wage a firm pays to a specifically trained worker, the more profit it makes but the more likely the employee is to leave. The optimal time paths of wages and turnover show that, if marginal product is increasing, wages need not be increasing but it always implies a falling turnover rate.

Referrals can improve screening and self-selection of applicants during the hiring process. We model and estimate how referral information affects the selection of employees through job offers, acceptances, and turnover. Using data from a call center company, we show that referrals help employers attract applicants of superior performance. Yet performance differences between referred and nonreferred workers diminish with tenure through selective turnover. Our estimates reveal that referrals allow employers to screen on hard-to-observe but performance-relevant attributes for employees of high performance and high propensity to stay. Thus, referred applicants complete much of the sorting during the hiring process.

The past century and a quarter has seen frequent improvements in track and field records. We attempt to estimate what proportion of the speed of record breaking is due to globalization (competitors from more countries) and what proportion is due to technological progress (better equipment and training techniques). It appears that technological change is the chief driving force but that technological progress is improving the performance of seasoned elite athletes faster than it is improving the performance of adolescents. Both our results and our methods may have wider application.