PUBLICATIONS

“Voter Response to Peak and End Transfers: Evidence from a Conditional Cash Transfer Experiment.” (with Sebastian Galiani, Nadya Hajj, Patrick J. McEwan, and Pablo Ibarrarán)

American Economic Journal: Economic Policy, 11 (3): 232-60.

 Additional coverage: “Electoral Reciprocity in Programmatic Redistribution: Experimental Evidence”, VoxEU Post, 10/22/2016  

Abstract: In a Honduran field experiment, sequences of cash transfers to poor households varied in amount of the largest (“peak”) and last (“end”) transfers. Larger peak-end transfers increased voter turnout and the incumbent party’s vote share in the 2013 presidential election, independently of cumulative transfers. A plausible explanation is that voters succumbed to a common cognitive bias by applying peak-end heuristics. Another is that voters deliberately used peak-end transfers to update beliefs about the incumbent party. In either case, the results provide experimental evidence on the classic non-experimental finding that voters are especially sensitive to recent economic activity.

WORKING PAPERS

“Missing and Fired: Worker Absence, Labor Regulation, and Firm Outcomes” [Job Market Paper]

Additional coverage: World Bank Development Impact Blog, 12/9/2019

Abstract: Estimates from labor market surveys across various developing countries suggest that anywhere between 5-25% of scheduled worker days are lost to absence. High worker absenteeism may adversely impact firm outcomes as firms bear large costs in coping with unanticipated absences. Simultaneously, common labor regulations implemented for workers’ welfare could have the unintended consequence of dampening firms’ coping strategies. Guided by a model of workers’ decisions about absence and firms’ decisions about worker turnover, I examine the impact of absenteeism on firm outcomes, and its dependence on labor regulations. I use a large panel dataset from manufacturing firms in India, where worker absenteeism remains a severe problem despite strong economic growth in recent years. First, I show, using both a fixed effects and instrumental variables approach, that firms' coping mechanisms cannot fully make up for workers' absences: firms lose about 0.14% of mandays worked for a 1% increase in absenteeism, primarily from the workers who work directly on the production lines. Correspondingly, firm revenue and profits fall by 0.21% and 0.30% respectively, for every percent increase in absenteeism. These effects are even greater in magnitude than other inputs risks faced by the firm, including credit and import shocks. Then, interacting absence with spatial variation in the regulatory structure that governs hiring and firing of workers, I find that firms fire more workers and hire new workers in response to absenteeism, particularly in states with less stringent labor regulations. Firms in states with less stringent regulations are able to mitigate between a third and half of the negative impact of worker absence.

"Propping Up the Wage Floor: Collective Labor Supply without Unions" (with Emily Breza and Supreet Kaur)

NBER Working Paper #25880 [Link to Paper] [Online Appendix]

Additional coverage: VoxDev, 10/07/2019, VoxEU, 10/12/2019

Abstract: Social norms can serve as a powerful force for conformity, producing collective behaviors among decentralized individuals. We test for this force in the labor market: whether norms prevent workers from supplying labor at wage cuts, generating cartel-like behavior in the absence of explicit collusion. We partner with 183 existing employers, who offer jobs to 502 workers in informal spot labor markets in India. Unemployed workers are privately willing to accept jobs below the prevailing wage, but rarely do so when this choice is observable to other workers. In contrast, social observability does not affect labor supply at the prevailing wage. Workers give up 38% of average weekly earnings in order to avoid being seen as breaking the social norm. In addition, they are willing to pay to punish anonymous laborers who have accepted wage cuts - indicating that collective labor supply behavior is reinforced through the threat of social sanctions. Finally, consistent with the idea that social conformity could have aggregate implications, measures of social cohesion correlate with downward wage rigidity and business cycle volatility across India.

“At What Price? Price Supports, Agricultural Productivity, and Misallocation”  [Online Appendix]

Abstract: Agricultural price support policies are a popular way to alleviate the risk inherent in volatile prices, but, at the same time, may distort input allocation responses to agricultural productivity shocks across multiple sectors. This could reduce productivity in the agricultural sector in developing countries. I empirically test for misallocation in the Indian agricultural setting, with national price supports for rice and wheat. I first motivate the setting using a two-sector, two-factor general equilibrium model and derive comparative statics. I then use annual variation in the level of the national price supports for rice and wheat relative to market prices, together with exogenous changes in district-level agricultural productivity through weather shocks, in a differences-in-differences framework. I derive causal effects of the price supports on production patterns, labor allocation, wages, and output across sectors. I find that rice area cultivated, rice area as a share of total area planted, rice yields, and rice production all increase, suggesting an increase in input intensity (inputs per unit area) dedicated to both staple crops. Wheat shows a similar increase in input intensity. The key input response is a reallocation of contract labor from the non-agricultural sector during peak cultivation periods, which results in an increase in wages in equilibrium in the non-agricultural sector (especially in response to price supports for the labor-intensive crop, rice, of 23%). The reallocation of labor reduces agricultural productivity by 82% of a standard deviation, and simultaneously reduces gross output in non-agricultural firms by 8.5%. I also find that rice- and wheat-producing households do not smooth consumption more effectively in response to productivity shocks in the presence of price supports.

SELECTED RESEARCH IN PROGRESS

“Demand for Flexible Work and Contract Choice” (with Suanna Oh and Yogita Shamdasani)

Abstract: The unskilled labor market in rural India is characterized by both high levels of unemployment and high rates of worker absenteeism. To explain the coexistence of these two features, we propose that workers have substantial demand for flexibility in work arrangements. We first build a model of a segmented labor market with two main sources of employment - regular employment with structured schedules and casual daily-wage employment with flexible schedules - to illustrate that worker demand for flexibility can drive both high absenteeism in the former and high unemployment in the latter. Next, we use an incentive-compatible choice experiment with unskilled workers to elicit preferences over pairs of contracts that offer varying levels of flexibility for a month-long employment opportunity. Using these elicitations and survey data, we aim to measure the following: 1) the share of workers who have demand for flexibility that exceeds what is permitted by structured schedules offered in the market; 2) the correlates of worker demand for flexibility including measures of poverty; and 3) worker valuation of flexibility in terms of foregone potential earnings. These measures would indicate how worker demand for flexibility affects labor supply decisions, both in terms of absence and selection into particular work arrangements. We discuss the potential implications of our findings for income inequality in an economy that is transitioning into more structured modes of production.

“Economic Opportunity and Motivation for Crime: Theft From Oil Pipelines in Nigeria"

Abstract: Nigeria is Africa's largest (and the world's thirteenth-largest) producer of oil. At the same time, the country loses at least 4% of daily production (and likely even more) to theft, largely spearheaded by militia groups. These groups recruit local youth for help with stealing oil efficiently. In this paper, I show that access to crime (proxied by households’ geographic proximity to pipelines) is used as a mechanism to hedge against agricultural risk. Using data on oil spills (since precise data on theft alone is not available), and a detailed map of oil pipelines in the country that I digitized, I find an increase of 0.66 spills per month (0.13 when restricted to large spills of over 1000 barrels – suggestive of militia activity) when percentage losses in anticipated agricultural yieldss are above 50% for households within 15km of pipelines. Non-food expenditures are the hardest hit in times of shock to agricultural output, for an average impact of about 28K Naira. I find that households close to pipelines do not face decreases in non-food expenditure when they face agricultural shocks (in fact, they make up for this shortfall by almost 200% (about half of median non-food expenditure in the sample)). Future work on this topic will ask whether access to crime can have a long-lasting effect on the educational outcomes of at-risk youth.