Escaping Pay-for-Performance

How do regulators respond to performance pay? We study this question by exploiting the staggered adoption of performance pay by U.S. financial regulatory agencies between 1981 and 2006. We manually assemble employee-level data on 30,000 financial regulators from public and private payroll records, rulemaking documents, and employee surveys, allowing us to track their careers, incentives, and productivity inside and outside government. Using a stacked difference-in-differences design, we find that performance pay increased voluntary exits from the government by 43-57%. A separate reform affecting top federal executives produced similar effects, reinforcing external validity. Additional analysis highlights two broad channels underlying this response. Performance pay increases effort and thus potential private sector salaries. It simultaneously weakens public sector attachment via greater income volatility, a more competitive work environment, and distrust toward performance evaluations. Consequently, regulators with better outside options and a weaker public sector attachment - leave. Taken together, we highlight a fundamental trade-off in incentive design: performance pay raises productivity but reshapes the composition of the public workforce. We develop and estimate a structural model that quantifies this trade-off and evaluates counterfactual pay policies balancing incentives and retention.