Association for Public Policy Analysis & Management

Improving teacher quality depends, primarily, on attracting and retaining high-quality teachers and, in practice, doing so with limited resources. Making the most of limited resources requires a detailed understanding of teacher preferences. Normally, a researcher could infer preferences from choice. Teacher choice sets, however, are not observable and neighboring schools do not offer differing compensation structures, implying real-world choice data would be largely uninformative even if attainable. To meet the challenge, I use a discrete-choice experiment among K-12 public school teachers to study preferences over compensation structure and working conditions in a setting in which teachers have incentive to reveal their preferences. I estimate teacher utility functions and use them to calculate teacher-welfare and student-achievement maximizing bundles, both of which suggests districts overpay in benefits and underpay in salary and merit compensation. If the best teachers have different preferences, policymakers can generate a separating equilibrium which differentially attracts and retains them. High value-add teachers broadly have similar preferences to other teachers but systematically prefer schools that offer merit pay.