Person:

Khwaja, Asim

Learning persistence plays a central role in models of skill formation, estimates of education production functions, and evaluations of educational programs. In non-experimental settings, estimated impacts of educational inputs can be highly sensitive to correctly specifying persistence when inputs are correlated with baseline achievement. While less of a concern in experimental settings, persistence still links short-run treatment effects to long-run impacts. We study learning persistence using dynamic panel methods that account for two key empirical challenges: unobserved student-level heterogeneity in learning and measurement error in test scores. Our estimates, based on detailed primary panel data from Pakistan, suggest that only a fifth to a half of achievement persists between grades. Using private schools as an example, we show that incorrectly assuming high persistence significantly understates and occasionally yields the wrong sign for private schools’ impact on achievement. Towards an economic interpretation of low persistence, we use question-level exam responses as well as household expenditure and time-use data to explore whether psychometric testing issues, behavioral responses, or forgetting contribute to low persistence—causes that have different welfare implications.

The current banking crisis highlights the challenges faced in the traditional lending model, particularly in terms of screening smaller borrowers. The recent growth in online peer-to-peer lending marketplaces offers opportunities to examine different lending models that rely on screening by multiple peers. This paper evaluates the screening ability of lenders in such peer-to-peer markets. Our methodology takes advantage of the fact that lenders do not observe a borrower’s true credit score but only see an aggregate credit category. We find that lenders are able to use available information to infer a third of the variation in creditworthiness that is captured by a borrower’s credit score. This inference is economically significant and allows lenders to lend at a 140-basis-points lower rate for borrowers with (unobserved to lenders) better credit scores within a credit category. While lenders infer the most from standard banking “hard” information, they also use non-standard (subjective) information. Our methodology shows, without needing to code subjective information that lenders learn even from such “softer” information, particularly when it is likely to provide credible signals regarding borrower creditworthiness. Our findings highlight the screening ability of peer-to-peer markets and suggest that these emerging markets may provide a viable complement to traditional lending markets, especially for smaller borrowers.