Take-up: Of the 1,980 households selected to receive vouchers, 1,210 households used them, for a 61 percent take-up rate.
Teacher and school inputs: School and teacher inputs varied significantly between public and private schools. Most notably, private schools had longer school years, lower student-teacher ratios, better facilities, and lower rates of teacher absence. Government schools had teachers with more years of teaching experience who were more likely to have college degrees. However, these teachers also had salaries that were six times higher. Expenditures per student were also over four times higher in public schools.
Household inputs: Students who received vouchers spent more time in school (on average, 26 more minutes per day, a 7 percent increase above an average of 6 hours and 23 minutes among students in comparison villages), but they did not spend more time studying or doing homework. Households that received vouchers did not increase expenditures on education. This suggests that any changes in test scores were due to changes at school, not changes at home.
Test scores: There were no significant differences in Telegu and math test scores among voucher winners and losers after two and four years of the program. Voucher winners had slightly higher scores in English. However, researchers found that voucher winners performed significantly better than non-winners in other subjects such as social studies, science, and Hindi. Across all subjects, lottery winners scored 0.13 standard deviations higher. While government schools spent 70 percent of instructional time on math, Telegu, and English, these subjects accounted for less than 50 percent of instructional time in private schools. As private schools achieved similar results in math and Telegu with less instructional time and used the extra time to improve test scores in other subject areas, they were more productive than public schools.
Spillover effects: Researchers tested whether the program lowered test scores among students who did not win the lottery for a voucher, students whose parents did not apply for a voucher, and students who were already attending private schools before the start of the voucher program. They did not find any evidence that the program harmed students in any of these three groups.
Taken together, these results support policies that use government funds to send poor children to private schools. Private schools achieved similar test scores in basic subjects with less instructional time and used the remaining time to increase test scores in other subjects. Furthermore, the schools achieved these results at a much lower cost and without disadvantaging the students who remained in public schools.