Trade and Development Seminar with Jishnu Das (Georgetown)

The rise of private schools implies that many children in low-income countries now live in villages and towns with substantial school choice. The success of any policy—from vouchers to school consolidation—will therefore depend on what this school choice entails and whether there are meaningful and stable differences in school quality both across and within public and private sector schools. Using rich panel data on educational outcomes from 820 schools in 112 villages in rural Pakistan, we provide causal estimates of each school’s value-added and establish that our estimates are valid, unbiased measures of school quality. We then establish three key features of educational markets in our context.
First, the distribution of school quality is highly variable, with significant overlap in the public and private school distributions. Moving to a 1sd better public school raises test scores by 0.32sd, while moving to a 1sd better private school raises test scores by 0.21sd. Half the variation in school quality is within villages so that there are better and worse performing schools in every village.
Second, our estimates are closely related to more standard approaches in the literature that estimate a single causal impact of private schools on test scores. Using multiple identification strategies, we show that private schools causally increase test scores and that these causal estimates can be recovered from the full SVA distributions in the presence of treatment heterogeneity.
Third, SVA predicts test scores 5-8 years later, even though the majority of teachers and head-teachers have been replaced through natural churn. Private schools with high SVA gain market share and are less likely to shut down. For public schools, there is no correlation between SVA and market share and during a period of school consolidation and closures, the government was no more likely to close down poorly performing public schools.

  Jishnu Das (Georgetown)

Seminar will be held Virtually. Please sign up for our seminars listserv to receive the details.