Association for Public Policy Analysis & Management

Teach For America (TFA) is an alternative certification program that places intensively selected recent college graduates and mid-career professionals into classrooms serving high-need students for a two-year commitment period.  Though a body of evidence documents the effectiveness of TFA corps members at raising the math test scores of their students, little is known about the program’s impact beyond the level of corps members' classrooms. TFA’s recent placement strategy in the Miami-Dade County Public Schools (M-DCPS), where large numbers of TFA corps members are placed as clusters into a targeted set of disadvantaged schools, provides us with an opportunity to evaluate the impact of the TFA program on broader school performance. This study examines whether the influx of TFA corps members in target schools led to productivity spillovers onto other teachers’ performance.

Using longitudinal data from M-DCPS spanning the 2008-09 through 2013-14 school years, we generate TFA impact estimates of effectiveness compared to similarly-experienced teachers within their same schools using student fixed effects. We estimate main TFA impacts of approximately 0.11 and 0.02 standard deviations of student learning gains in math and reading, respectively; estimates in both subjects are statistically significant. We then explore the relationship between the density of TFA corps members within schools and student achievement. Our findings suggest a significantly positive spillover effect in math onto non-TFA teachers associated with the dense clustering of TFA corps members. The spillover effect appears to be most significant in middle and high school grades, and is more closely related to the school-level density of TFA corps members (rather than grade- or subject-level densities of corps members). On balance, the results suggest a modest, though significant, effect on other teachers' performance from having large numbers of TFA in these targeted schools. We do not find any evidence that corps members’ performance was sensitive to any differences in clustering. Also, we do not find any evidence of a significant spillover effect in reading.

Finally, we observe that many of the schools chosen to participate in the cluster strategy experienced large subsequent gains in math achievement. Our results suggest these gains were driven primarily by the composition effect of having larger numbers of relatively effective TFA corps members staffing these schools. However, the evidence of a modest spillover effect appears to have contributed to the observed gains. In other words, the combined output of staffing dense clusters of TFA corps members is greater than the sum of the individuals' output in these district schools.