Panel Paper: Heterogeneous Charter Schools Impacts On Student Achievement

Saturday, November 9, 2013 : 10:05 AM
Scott (Westin Georgetown)

*Names in bold indicate Presenter

Hiren Nisar, Abt Associates, Inc.
Recent reforms in education emphasize the use of charter schools to improve student achievement. Charter schools are public schools operating with greater independence than traditional public schools, often structuring their own curriculum and school environment without outside influence. Charter school advocates claim this freedom allows charter schools to improve student achievement by increasing instructional hours, using innovative instructional methods or operating in other manners that traditional public schools either cannot or do not. Despite growing popularity, there is a lack of concrete evidence regarding charter school effectiveness. However, previous research oversimplifies the diverse atmosphere in which charter schools operate. Charter schools vary in characteristics and the demographics of students they serve. I hypothesize that these differences, specifically differences in autonomy from the school district and student population demographics, should affect charter school effectiveness.  More autonomous charter schools operate with more flexibility from local school districts in terms of budget, academic program content and hiring decisions (mostly non-unionized teachers). To test this hypothesis, I examine the effects of charter schools on student performance using longitudinal data from all Milwaukee public schools, a large urban school district with a long history of charter schools. A key challenge in estimating the effect of charter schools is that students self-select into them.  I use propensity score matching to reduce the bias due to the self-selection by taking into account the covariates that predict receiving the treatment and achieving balance on the observable covariates for the treatment and control group.  I find that charter schools, on average, have no significant effect on student achievement; however, I show that this average effect masks important heterogeneity in the effectiveness of charter schools across charter schools with different levels of autonomy from the school district. I estimate large, positive and statistically significant test score gains for students attending charter schools with more autonomy from the school district, evidencing their effectiveness in improving academic outcomes. In contrast, the test scores of students attending less autonomous charter schools actually declined over the same time. These findings are robust while controlling for conversion schools and demographic differences between student populations. As hypothesized, results also vary by student sub-population.  Specifically, results show that African-American students are well served in highly autonomous charter schools, which should give hope to charter school advocates. Furthermore, these findings make clear that careful design of charter school policies can improve student outcomes. Indeed, as charter school laws differ depending on geography, differing policies may partially cause the mixed estimates of charter schools’ effectiveness across states. In the future, charter school authorizers should be encouraged to promote school autonomy from the school district. Policy makers should investigate other features of high performing charter schools in order to help authorizers make informed decisions.