Association for Public Policy Analysis & Management

Research clearly demonstrates that teachers are inequitably distributed across schools: less experienced, less educated teachers teach in lower income, lower achieving schools with higher proportions of minority students (Goldhaber, Lavery, & Theobald, 2015). With few exceptions, research observes only where teachers actually teach, rather than their expressed job preferences. We know little about the job preferences of novice teachers, beyond a tendency to teach closer to where they grow up or attend college (Boyd, Lankford, Loeb, & Wyckoff, 2005).

Sorting may exacerbate the inequitable distribution of teachers if districts produce graduates who are academically underprepared, and the same graduates return as teachers (Boyd et al., 2005). Rather than the draw of home, one alternative explanation for this sorting may be that as higher-achieving, wealthier schools tend to have earlier hiring timelines (Engel, 2009; Papay & Kraft, 2015) and, as teachers tend to be risk averse (Dohmen & Falk, 2010), more qualified teachers may take the first job that is offered, leaving lower-achieving schools with later hiring timelines a diminished pool of teachers from which to hire.

To better understand this selection process, we ask:

1) Do teacher job preferences vary by educational credentials?

2) Do teacher job preferences change over their time on the labor market?

3) Do teacher educational credentials and observed job search behaviors predict success on the job market?

            We draw data from 2 sources: a database of applications to open teaching positions in 317 of Wisconsin’s 425 school districts from January to October, 2014. These data include vacancy descriptions, date of vacancy posting, date of application, applicants’ work history, personal statements, and education. This application database is combined with a longitudinal state-level administrative dataset that contains hiring information for these vacancies and employee gender, ethnicity, work address, and salary information for all Wisconsin teachers. The combined dataset is the most comprehensive dataset of a teacher labor market to date.

First, we use OLS regression to predict variation in job preferences by educational credentials. Next, to address teacher job-seeking behaviors, we use descriptive graphical analysis, plotting the average number of applications per week per teacher during the hiring season by education credentials. Finally, to address success on the job market, we use observed search behaviors (e.g., timing of entry on to the market, length of time on the market, average number of jobs per week applied to) to predict whether or not teachers are hired at all.

Preliminary results indicate that novice teachers are on the market for 7.5 weeks and apply to an average of 16 jobs per week. Teachers who major in education apply to more jobs overall, and to schools with higher achievement and lower proportions of poor and minority students. Our results will enable us to make better policy recommendations for recruiting and hiring novice teachers, aimed at increasing equity in the distribution of quality teachers across schools.