Negative Selection and Peer Effects in Indiana’s Voucher Program

• In Event: School Choice, Student Mobility, and the Geography of Opportunity

Abstract

In the United States, 30 states have a school choice program, and 4 more have recently passed legislation for new programs. Since 2023, 5 states have expanded their school choice programs to be near universal. As school choice expands rapidly in the nation, it is important to understand how these programs affect students who use these programs as well as those left behind in public schools. Economic models of school choice come to three conclusions: new competition pressures public schools to improve performance; students gain from greater choice; and students positively select into these programs, decreasing the quality of public schools. The first two points are well documented, but we know less about how exiting peers affect their classmates who are left behind, especially in the context of school choice. 
In this paper, I use student-level data from the Indiana Department of Education to study Indiana's private school voucher program, one of the largest and most generous private school voucher programs in the nation. First, I document which students opt into the voucher program. Second, I estimate the effect of voucher student exit on their peers who are left behind in public schools. Contrary to most models on school choice, I find no evidence of positive selection on academic ability. Instead, I find students who transfer out of public schools with vouchers have higher rates of disciplinary infractions, specifically out-of-school suspensions and expulsions. These results are consistent across expansions in income eligibility as Indiana moves from targeting low-income students to becoming nearly universal. Next, I use a staggered event study design to compare students who had a student with a voucher exit their class to those who do not. I find bad apple students cause their left-behind classmates to have 2.7 fewer in-school suspensions and 1.7 fewer out-of-school suspensions per 100 students, which corresponds to a 14\% and 11\% reduction, respectively. In contrast, good apple students cause no change in their classmates who are left behind.
I find peers matter, and their reallocation in a voucher setting has meaningful impacts on students who are left behind in public schools. While most papers search for changes in test scores to prove peer effects, I find changes in behavioral outcomes, which point toward the importance of other dimensions in which students affect each other. While good apple peers do not appear to improve their classmates' behavior, bad apple peers worsen it, especially for older students and those in classes with high monitoring costs for educators. Taken together, my results suggest reallocation in a voucher setting improves the peer quality of public schools, rather than worsens it, contrary to school choice models.

Author

• Kendyl Guzman, Clemson University
  Presenting Author