Association for Public Policy Analysis & Management

Growing teacher shortages, accelerating retirement rates, and declining enrollments in teacher preparation programs (Tucker, 2014; Westervelt, 2015) represent mounting challenges facing many school districts, and these challenges are particularly pronounced in recruiting high quality teachers to low-income schools (Boyd, Lankford, Loeb, & Wyckoff, 2005; Glazerman, Protik, Teh, Bruch, & Max, 2013; Isenberg et al., 2014). Given this, understanding the dynamics of teacher attrition is particularly important for understanding which policies might best encourage the retention of high-quality teachers.

To lend evidence to this question, this paper draws on human resource administrative data from all teachers and former teachers from San Francisco Unified School District (SFUSD) in the years 2003-2015 linked with Internal Revenue Service (IRS) data on subsequent income and employers. For all teachers, we have information about age, gender, race, educational attainment, teaching credentials, subject taught, tenure status, and years in the district (and at their current school). Importantly, we can also construct value-added measures (VAMs) of teachers’ effectiveness, allowing us to examine how patterns vary across teachers with different levels of effectiveness.

Taken together, these matched data provide us unique opportunities to investigate dynamics of teacher attrition, future salary profiles, employment opportunities, and how these factors vary across demographic and professional teacher characteristics. Examining these patterns across teachers in different subjects shed light on whether high-quality teachers are particularly hard to retain due to outside salary pressures.

Our work is related to research indicating that high-poverty and high-minority schools have the highest rates of teachers leaving for other schools and departing from the profession entirely (Boyd, Lankford, Loeb, & Wyckoff, 2005; Ingersoll, 2001; Lankford, Loeb, & Wyckoff, 2002). It also builds on Ingersoll and May’s (2011) investigations of teacher turnover across multiple national data sets suggesting that there is a net migration of a substantial number of STEM teachers from poor, majority-minority, and urban schools to affluent, low-minority, suburban schools each year. We build on this prior work by following teachers to their subsequent employers to determine the degree to which teachers truly move to more affluent districts or to higher-paid non-teaching positions.

Preliminary results indicate that individuals who left teaching earn more, but that this is largely due to fixed characteristics of teachers who leave. After conditioning on fixed teacher characteristics, we find relatively little change in average earnings after teachers separate from teaching. Hence, relative to their prior earnings trajectories, teachers do not appear to have marked improvements in earnings after leaving teaching. There is also some heterogeneity in this pattern. Earnings parity is consistent for individuals across a variety of post-teaching industries with the exception of former teachers who enter the fields of business, accounting, and insurance. Similarly, earnings change little for teachers from most demographic groups, but appear to be slightly higher on average for African-Americans who leave this district.