Association for Public Policy Analysis & Management

Since 2011, 44 states have implemented major reforms to their teacher evaluation systems while five states have passed legislation that effectively eliminates tenure for teachers. While these teacher accountability reforms remain hotly debated, surprisingly little evidence exists about the effect of these reforms on the teacher labor market. Prior literature is largely limited to simulation-based analyses that point to potential gains in student achievement from dismissing low-performing teachers. These studies, however, explicitly assume accountability reforms will have no general equilibrium effects on future teacher labor supply.

In this paper, we conduct the first empirical test, to our knowledge, of how teacher accountability reforms affect an individual’s decision to enter the teacher labor force. We construct an original state-by-year panel dataset spanning from 2002 to 2015 to estimate the effect of revoking tenure and introducing high-stakes evaluations on several indicators for preparation to enter the teacher labor force. Our primary outcome of interest is the number of teaching licenses granted in a given state and year recorded by the Federal Department of Education. This outcome is well suited for our study because it focuses on public-sector teachers and links each teacher to their state of employment. Obtaining a license is often the last step prior to entering the labor market and this measure is likely to be immediately responsive to policy reforms. We complement these analyses with data from the Integrated Postsecondary Education Data System (IPEDS) on the number of traditional teacher program degree completers as well as Title II data on the number of alternative certification program completers.

We estimate the effect of teacher accountability reforms by exploiting variation in the timing of these reforms across states in a Difference-in-Differences (DD) framework. We begin with a simple DD model which we then extend to allow for differential linear trends pre and post policy reform in treated states. This allows us to explicitly test the parallel trends assumption as well as to model separately any immediate effects of the policy changes and any incremental effects on teacher labor supply over time. We extend these models with a completely flexible event-study specification which allows for a non-parametric examination of pre and post treatment trends.

Across models, we find consistent evidence that both repealing tenure and introducing high-stakes evaluation reforms reduced teacher labor supply. We estimate that number of licensures granted in states dropped by approximately 20% in states that adopted accountability reforms. In joint models, both tenure and evaluation reforms appear to contribute to this contraction of the labor supply. A range of robustness and falsification tests confirm these findings. Additionally, we find suggestive evidence that teacher accountability reforms increase the number of teachers who complete alternative certification programs. Analyses using IPEDS data on teacher completers allow for a range of sub-group analyses across gender, race, subject area specializations, and competitiveness of the institution attended by teacher candidates. These estimates do not reveal any definitive patterns although small differences cannot be ruled out due to imprecision.