Impact of Voucher Design on Public School Performance: Evidence from Florida and Milwaukee Voucher Programs

This paper examines the impact of vouchers in general, and voucher design in particular, on public school performance. It argues that all voucher programs are not created equal. There are often fundamental differences in voucher designs that affect public school incentives differently and induce different responses from them. It analyzes two publicly funded voucher programs in the U.S. The 1990 Milwaukee experiment can be looked upon as a "voucher shock" program with a sudden Government announcement that the low-income public school population will be eligible for vouchers. The 1999 Florida program, on the other hand, can be looked upon as a "threat of voucher" program where the failing public schools are first threatened with vouchers, and vouchers are implemented only if they fail to meet a government designated cutoff quality level. In particular, a school getting an "F" grade for the first time is exposed to the threat of vouchers but does not face vouchers unless and until it gets a second "F" within the next three years. To analyze and compare the impacts of these two alternative designs on public school incentives and performance, I construct a theoretical model that captures the basic features of the two programs. It is an equilibrium theory of public school and household behavior where public school maximizes net revenue and households care about public school peer group quality in addition to public school effort. I use the model to analyze three alternative scenarios--a simple public-private system without vouchers, a Milwaukee-type voucher shock system and a Florida-type threat of voucher system. Providing micro-foundations to the public school payoff function, the model endogenously determines public school effort and peer group quality at each of the program equilibria. It yields three predictions. First, the effects of a Milwaukee-type program on public school effort and quality are ambiguous (where public school quality is a composite of public school effort and peer group quality.) Second, a Florida-type program will lead to an unambiguous improvement in public school effort and quality and third, these improvements will exceed the corresponding improvements (if any) in a Milwaukee-type program. Using school-scores data and a difference-in-differences estimation strategy in trends, the second part of the paper shows that these predictions are validated empirically. There is significant evidence of large improvement of the treated schools in Florida, but no consistent evidence of improvement of the Milwaukee treated schools. The findings are reasonably robust in that they survive several robustness checks including correcting for mean reversion. The design of the Florida program also enables a regression discontinuity analysis to investigate the impact of the Florida program. Interestingly, the results from this analysis strongly mirror those obtained from the above analysis. Thus both theoretically and empirically, the paper presents strong evidence that voucher design matters and in particular, the public school response under a Florida-type program is much larger than that in a Milwaukee-type program. The findings have important implications for public school reform, which are all the more relevant in the context of the present concern over public school performance.