Individual Submission Summary
Share...

Direct link:

Voucher Effects on School Entry and Exit

Friday, November 14, 10:15 to 11:45am, Property: Hyatt Regency Seattle, Floor: 5th Floor, Room: 504 - Foss

Abstract

A recent series of events including Supreme Court decisions, the election of President Trump in 2017, and COVID-19 have suddenly changed the landscape of public education and vouchers in the U.S.  As of 2024, 14 states are scaling up voucher policies. These tend to be universal, large enough to cover tuition at a typical private school, and available for other educational expenses, including services and resources for homeschooling.

We focus on one key question: How are school vouchers changing the supply side of the educational market and its various segments? Specifically, to what degree do we already see changes in school openings and closures across sectors—public, charter, and private schools? How are enrollments shifting not only between public and private schools, but among different types of private schools as well as other school types? Also, what do statewide education leaders in those same sectors report about what is on the horizon?

We are currently collecting data on private schools from each state, combining the federal Private School Survey (PSS), the public Private School Review (PSR), state-released school lists, and other online information into a harmonized panel data set that includes school enrollment and other relevant characteristics of schools for the last 15 to 20 years. This data cleaning process is well under way. Once completed, we will estimate reduced reduced-form effect of these policies on schools’ entries and exits using two event-studies. First, we plan to compare entry rates (measured at the geographic school-district level) across states with and without voucher policies, estimating:

Entrydst = Σk=T0 → -2 βkVoucherst + Σk=0 → T1 βkVoucherst + ΓXdst + λd + γt + εdst

where Entrydst is the fraction of schools in district d that entered the market in year t. Voucherst is an indicator that equals one starting the year t in which state s implements the policy and thereafter (i.e., this is an absorbing treatment). xdst is a vector of pre-policy district-level determinants of entry (e.g., socioeconomic background). λd and γt account for district-specific time-invariant, and year-specific country-wide determinants of entry, respectively. We estimate separate versions of equation (1) by school sector.

Second, to study exits, we plan to estimate the following event study:

Exitist = Σk=T0 → -2 βkVoucherst + Σk=0 → T1 βkVoucherst + ΓXist + λi + γt + εist

where Exitist is an indicator variable that takes the value of one if school i in state s exits the market in year t. Voucherst is defined in the same way as above. The key difference of this event study relative to that in (1) is that we would use school-level data. While we could run an analogous one by computing exit rates at the district-year level, the use of individual level observations assumes that εist is unrelated to state-wide voucher policies more likely to hold, which in turn allows us to give βk a causal interpretation. Again, we estimate separate versions of equation (2) by school sector. Given the staggered adoption, we will also estimate using the equivalent Callaway and Sant’Anna (2021) method.

Authors