Working Papers
Abstract: Corruption in hiring for public sector jobs is common in developing countries, and has been assumed to have a detrimental effect on delivery of government services. This paper provides a framework for understanding this type of corruption and demonstrates that it need not have negative consequences. I collect original data from a hiring process for managerial positions in a developing country health bureaucracy, and find that hires paid large bribes, averaging 17 months' salary. I use data on bribe offers to characterize the structure of these markets, showing that job allocations are made as if via a first-price, winner-pay, sealed bid auction. To establish the consequences of corruption, I estimate a structural model of entry to determine the identity of hires under counterfactual hiring procedures, such as standardized testing, and compare them to actual hires. For this comparison, I identify causal relationships between a set of hire characteristics and better delivery of health services. Based on these characteristics, actual hires are of comparable or superior quality to the counterfactual hires, e.g. as compared to hires under a knowledge-based test, actual hires are 4.3-8.7 percentage points closer to the predicted optimal set of hires. This is explained by a novel mechanism in my model: although hiring decisions are based primarily on bribes, hires are high quality because applicant wealth and willingness to pay for the position are strongly positively correlated with quality. Applying this to a general model of hiring, I identify the environments in which corruption will lead to misallocation, discuss how anti-corruption efforts should be designed, and argue for a greater focus on hiring for mid-level government managers. 

While the vast majority of minimum wage research has focused on possible disemployment and labor market effects, this paper examines its effect on prices. Utilizing detailed transaction-level datasets, we find minimal pass-through of minimum wage changes onto retail and grocery prices. This finding is robust to a number of identification strategies, such as matching counties across state boundaries, and is precisely identified. Additionally, we do not find evidence for anticipation/retrospective adjustment in prices, heterogeneous responses based on the size of the minimum wage change, or heterogeneous effects on consumers of different income groups. Data from the Quarterly Census of Employment and Wages rationalize these findings. Since labor is a relatively small portion of total costs in this sector and the effect of the minimum wage on labor costs is modest, this leads to minimal response in prices. Estimates from other industries suggest that, aside from the restaurant and fast-food sectors, this low pass-through holds more generally. 

Dowry payments are an important part of household finances in India, and typically are between one to two years of household earnings. Yet there is little empirical evidence on determinants of dowry payments, with existing work relying on small and non-representative samples. In the first part of the paper, we leverage data on over 76,000 marriages to document seven stylized facts in Indian marriage markets between 1930 and 1999 . We show that although there has been little shift in some practices, such as intercaste marriage, there were large changes in dowry payment. Between 1930-1975, the proportion of marriages with any dowry paid increased from 35-40% to nearly 90%. Over the same period, median real dowry more than doubled, but decreased after 1975 in real terms as well as a fraction of household income. In the second half of the paper, we use this data to test major theories of dowry: (i) whether dowry serves as a bequest to female children or is a groom price; (ii) if the increase in dowry prevalence resulted from lower castes adopting high caste practices (Sanskritization hypothesis); (iii) how changes in sex ratios on the marriage market affect dowry (marriage squeeze hypothesis); (iv) if changes in dowry can be explained by cross-caste competition for grooms (Anderson, 2003). We find that the patterns in the data do not support these theories, but instead that the changes in dowry are explained by shifts in the quality (earnings/education) distribution of brides and grooms. This has implications for design of anti-dowry policies in India.

Works in Progress

Intrahousehold Spillovers of Incarceration: Evidence from Ohio and Pennsylvania (with Sam Norris and Matt Pecenco)

Incarceration, Health, and Death (with Sam Norris and Matt Pecenco)

Timing of Cash Transfers and Child Health (with Karthik Muralidharan, Paul Niehaus and Sandip Sukhtankar)

Designing Hiring Systems: Theory and Empirical Evidence

Beliefs and Bribes: Police Corruption in India (with Abhijit Banerjee, Esther Duflo, John Firth, Dan Keniston and Ben Olken) (first round of data-collection complete)