What Do Preschool Vouchers Buy? Value-added and Lottery Estimates from India (with Peter Berg and Mauricio Romero)

Private providers dominate preschool provision in India, and vouchers are the standard tool to expand access to them. We study the productivity of preschools and the learning effects of vouchers from India's Right to Education (RTE) Act, which allocates free private-school seats by lottery. We combine a representative panel of ~6,300 children aged 3--8 with ~2,400 applicants to the 2022 and 2023 RTE lotteries. We report three sets of results. First, value-added estimates, validated against lottery-based instrumental-variable estimates, show that a year in private preschool raises test scores by 0.3--0.7s.d. (all $p<0.01$) relative to public preschool or no enrolment. This premium rises with fees in math and English, but not in Hindi. Second, winning a voucher raises English scores by 0.14s.d. ($p<0.01$) two to three years later, but we can rule out effects greater than 0.15s.d. in math and 0.12$\sigma$ in Hindi at the 95\% level. Vouchers to the two top fee tiers drive the English gain (0.23s.d., $p<0.01$), and vouchers to cheaper schools have no effect. Third, a model combining value-added, lottery-induced enrolment responses, and test-score persistence replicates the lottery estimates. The model attributes most of the gap between productivity and policy effects to enrolment responses, not fade-out. Winning a seat lowers underage primary-school enrolment by 32--43 percentage points, and vouchers to cheap schools crowd out more productive private options. If lottery losers always attended public options, the model predicts a 0.88s.d. math gain by age 5--6 from a nursery seat in the two top fee tiers, versus 0.11 s.d. under observed enrolment.