Association for Public Policy Analysis & Management

A common criticism of charter schools is that they experience high attrition by their lowest-performing students. These students tend to re-enroll into the traditional public school system (TPS), thus making the average traditional public school student more difficult-to-educate than would be the case in absence of charters.

A growing body of research has evaluated the relationship between student test score performance and the probability that they exit out of charter schools (see Winters et al. (2017) for a review of this literature). To date, however, no prior research of which we aware has followed the progress of these students—herein called leavers—over time after they enter TPS. Do leavers continue to fall behind? Do they perform better in TPS than when in charter schools? How well do leavers perform relative to students who never attended a charter school?

We evaluate these questions using longitudinal student-level data from Denver, Colorado. The Denver context is interesting because prior research finds that on average students benefit academically from attending a charter school in the city (Abdulkadiroglu, Angrist, Narita, & Pathak, 2015; Center for Research on Education Outcomes, 2009). Our approach incorporates a student fixed-effect in order to compare student performance in years before and after exiting a charter school for a TPS.

Preliminary findings suggest that leavers perform significantly and substantially better on math and reading tests after moving into a TPS then when in a charter school. We fail to find similar gains for students who move from one TPS to another. Further, we show that students who leave charter schools perform better in TPS than students who had attended TPS during the entire period.

Our results have a variety of implications for both policy and the academic literature on charter school effects. First, our results are at least inconsistent with the claim that attrition from charter schools harms the ability of TPS to succeed. Rather than pulling down TPS performance, charter leavers appear to perform better than the average student enrolled in a TPS. Second, the prior finding suggests parents are capable of identifying when their children would benefit from a different schooling environment. This finding is crucial for the consideration of school choice policies, which rely on the ability of parents to identify the most effective schooling option for their child.

Finally, our results are interesting to consider alongside estimates of the impact of attending a charter school on academic performance. Lottery-based estimates can truly estimate the causal effect of the offer of attending a charter school. Estimating the effect of actually attending a charter school requires a two-stage procedure and additional assumptions. Our results that students who exit appear to benefit from the move into a TPS suggests that prior treatment-on-the-treated effects for charter attendance might actually understate the effect of enrolling in a charter school.