with Abhijit Banerjee, Esther Duflo, and Nina Singh
Revise and resubmit, Econometrica
October, 2019 Draft
Should police activity be narrowly focused and high force, or widely-dispersed but of moderate intensity? Critics of intense "hot spot" policing argue it primarily displaces, not reduces, crime. But if learning about enforcement takes time, the police may take advantage of this period to intervene intensively in the most productive location. We propose a multi-armed bandit model of criminal learning and structurally estimate its parameters using data from a randomized controlled experiment on an anti-drunken driving campaign in Rajasthan, India. In each police station, sobriety checkpoints were either rotated among 3 locations or fixed in the best location, and the intensity of the crackdown was cross-randomized. Rotating checkpoints reduced night accidents by 17%, and night deaths by 25%, while fixed checkpoints had no significant effects. In structural estimation, we show clear evidence of driver learning and strategic responses. We use these parameters to simulate environment-specific optimal enforcement policies.
with Guilhem Cassan and Tatjana Kleinberg
Workers' social identity affects their choice of occupation, and through these choices the prosperity and equality of aggregate economy. We study this phenomenon in a setting where work and identity are particularly intertwined: the Indian caste system. Using a new dataset that combines information on caste, occupations, and wages with historical evidence of subcastes' traditional occupations we find significant evidence that caste members are greatly overrepresented in their traditional occupations. To quantify the effects of caste-level distortions, we develop a structural general equilibrium Roy model of occupational choice. High skilled individuals, who might otherwise select into high skill-return occupations, are induced to remain in their low-skill traditional occupations. We structurally estimate the model and evaluate counterfactuals such as removing caste-level distortions to quantify their effects on occupational distributions, productivity, and inequality.
with Dave Donaldson
The 1918 influenza epidemic struck India when the subcontinent was mired in its long-term Malthusian equilibrium of low population growth and stable per-capita con-sumption. Its terrible death toll left survivors with additional agricultural land, which we show they rapidly put to agricultural use with no decrease in yields. We explore the extent to which this increased per-capita wealth gave rise, over the ensuing decades, to heightened investments in both child quantity as well as child quality. Consistent with most Malthusian unified growth theories, we find that individuals in heavily affected districts had more children in the aftermath of the influenza. We also find that these children were taller and better educated. Our results suggest that the preference for child quality existed even in societies that appeared Malthusian both to contemporary observers and modern historians.
January 2011 Draft
Bargaining for retail goods is ubiquitous in developing countries, where traders spend substantial amounts of time haggling over purchases. Would welfare be higher if trade was conducted at fixed prices instead? The answer is theoretically ambiguous: if bargaining is a low cost form of price discrimination, it may lead to greater trade and welfare and even approximate the optimal incentive compatible outcome. However, if bargaining imposes large utility costs on the participants, then a fixed price may be preferable. I develop the tools to resolve this question, specifying a model of repeated trade with hidden valuations adapted to the context of bargaining, and developing a dynamic structural estimation technique to infer the underlying parameters of the market. I then apply these techniques to bargaining data I collected from the market for local autorickshaw transportation in Jaipur, India.
April 2011 Draft
This paper carries out the first comparison of production function parameters estimated by structural techniques with those estimated via randomized instrumental variables using a unique dataset and field experiment performed by . In the context of a simple model of a household firm, I discuss the coefficients that each approach estimates, and the assumptions necessary to interpret those coefficients as the structural parameters of the model. I find that the values of structural and experimental estimators that most plausibly estimate the same parameters are indeed statistically and economically similar, suggesting that in some contexts structural models of production functions may be effective in recovering the parameters of production functions in the context of developing markets. These parameters may then be used to address questions relating to firm productivity and capital allocation that are both central to the study of firms in development, and potentially difficult to identify using randomized variation alone.