Attracting and Retaining High-Quality Teachers in Low-Performing Schools

Susanna Loeb

Differences across schools in the qualifications of teachers are likely contributors to the substantial gaps in academic achievement among income and racial/ethnic groups of students. Such sorting of teachers across schools and districts is the result of a range of decisions made by individual teachers and school officials. These include the decisions of individuals and school officials that determine initial job matches and subsequent decisions that affect job quits, transfers and terminations. This paper focuses on the initial matching of teachers to jobs and introduces an empirical model that employs the method of simulated moments to estimate the effect of various characteristics of teachers and schools on this matching. The paper models the allocation of teachers to schools based on the preferences of both employers and employees. While the long-term goal is to identify policies that are effective for attracting and retaining teachers in low-performing schools, such identification has many difficulties, not the least of which is the endogeneity of any policy we observe. The goal of this paper is more limited. It will introduce a model for the matching of teachers to schools and estimate this model with a limited set of school and teacher measures. The data comes from administrative records in New York State that allow us to follow all teachers in the state over the past 30 years. The breadth of the data (all teachers in all public schools) allows analytical flexibility not possible with smaller datasets. For example, we can look at the sorting of teachers separately for each region of the state. In addition, we can see how the preferences and decisions of one teacher in a labor market affect the matching of other teachers to jobs. The length of the data will allow us to assess how sorting and preferences have changed over time and also, eventually, will allow us to use fixed-effect approaches to identify policy effects. The data is richer in its descriptions of teachers than other administrative datasets used to date, including teachers' test scores and undergraduate institutions. It also allows us to match teachers to characteristics of the schools in which they teach in a way that most national longitudinal surveys, such as High School and Beyond or the National Longitudinal Survey of Youth, do not. The paper starts with a description of the data and then provides background on the teacher labor market in New York State -- including the extent of systematic sorting of teachers across schools; the importance of initial job matches, compared to transfers and quits, in determining this sorting; and evidence concerning the geographical extent of teacher labor markets. Section IV discusses traditional models for assessing the sorting of teachers across schools within regions, including the difficulties that arise from the two-sided matching of teachers to employers. Section V presents our model of the initial job match and our estimation strategy. The final sections present empirical results and conclude.

About Susanna Loeb

Susanna Loeb is an assistant professor in the School of Education at Stanford University. She earned her doctoral degree in Economics from the University of Michigan in 1998. Susanna’s work addresses the impact of federal, state and local policies on educational equity. In particular, her work has focused on school finance, welfare reform and the interaction of teacher labor markets with education policy initiatives.