I will post working papers here as they become available. I will also post all data and code for publications. Feel free to email to ask about works in progress and to check on the possibility of sharing data for working papers.

Peer-reviewed Publications

Kumar, T., Post, A., and Ray, I. “Flows, Leaks, and Blockages in Informational Interventions: A Field Experimental Study of Bangalore’s Water Sector.” World Development 106: 149-160, 2018.

Data and code

Abstract Under what circumstances might providing citizens with information compensate for unreliable public services? We present a field-experimental evaluation of a program that provided households in Bangalore with advance notification of intermittently provided piped water. The implementers expected that increasing service predictability would reduce wait times for water, reduce costs related to waiting, and improve citizen-state relationships. As many citizens did not receive accurate information, our study detected no impacts on household wait times for water or state-citizen relations. Nonetheless, our study suggests that notifications about water timing reduced stress, especially among low income populations. These findings indicate that greater attention should be paid to both psychological outcomes and the information production and dissemination chain in information interventions. We introduce a causal framework for analyzing “information pipelines” to enable such efforts.

Working Papers

“The non-electoral political effects of welfare policies: Evidence from a natural experiment in Mumbai”(under review)


Abstract Welfare programs in developing countries reach more people every day, potentially changing behavior and political landscapes in the long term. How do they affect an unstudied outcome, namely recipients' non-electoral political behavior? In these contexts, citizens often interact with governments to access resources, suggesting that receiving benefits may decrease participation. Yet research on "policy feedback" in the US shows that resources from benefits increase recipients’ capacity for action and motivate them to protect these resources. I study the effects of a common policy, subsidizing homeownership, on demand-making in Mumbai, India. A natural experiment of multiple housing lotteries shows that winning increases reported demand-making and knowledge about local politics, even among those who rent out the homes. Mechanisms may include changes in attitudes and an increased interest in improving communities. This study shows that welfare policies can facilitate collective demand-making activity among groups of beneficiaries.

“The human capital effects of subsidized homeownership: Evidence from a natural experiment in Mumbai”(under review)


Abstract Governments in countries at all income levels subsidize homeownership and thereby transfer wealth to middle-class households. I use subsidized housing lotteries in Mumbai to identify the human capital effects of such a transfer and find that 3-5 years later, winners are more educated than non-winners, with effects concentrated among school-age youth. The intervention also increases rates of employment, particularly among older youth who have had the chance to complete their education. Effects are not likely to be driven by relocation, as winners live in neighborhoods with poorer school quality and lower rates of literacy and employment than non-winners.

“From public service access to service quality: The distributive politics of piped water in Bangalore” (under review) with Alison Post, Megan Otsuka, Francesc Pardo-Bosch, and Isha Ray

Abstract Infrastructure services such as water, electricity, and mass transit are central to urban livelihoods. While the political economy literature on local public goods provision has examined patterns of expenditure on and access to infrastructure, variation in service quality for those receiving networked services has received far less attention. In this paper, we examine the distribution of service intermittency, which detracts from service quality and imposes significant welfare costs. We disaggregate intermittency into four dimensions: predictability, frequency, duration, and throughput. We extend arguments from the distributive politics literature to predict the allocation of burdens associated with intermittency among households; we show that this literature has paid insufficient attention to how network structures affect the ability of state or city officials to differentially channel service flows. We illustrate the importance of different dimensions of intermittency and network structure through an analysis of the political geography of piped water supply in Bangalore, India. We find that variation occurs at the “valve area” level, or the smallest units at which water pressure can be distributed, and not at the household-level. Households in low-income valve areas receive more frequent and regular service than those in more affluent ones, contrary to predictions from the distributive politics literature. Our work suggests that the distributive politics of network access differ significantly from those affecting water flows within the network.

“Preferences for Descriptive Representation: Asymmetries Between Hindus and Muslims in India.” with Pradeep Chhibber and Jasjeet Sekhon

Abstract Do minorities have a preference for descriptive representation? We address this question in India, a deeply religious society that has experienced extensive conflict between its Hindu majority and Muslim minority populations. Existing studies of preferences in such settings tend to attribute vote choice to strategic behavior by voters and parties. But because an election is a strategic context, voting outcomes usually reveal not the ideal preferences of voters, but rather their preferences mediated through a political and institutional context. Our research instead seeks to reveal ideal preferences through multiple experiments in diverse strategic settings in India. We find that within and across state lines, Muslims express a preference for co-religious candidates, but Hindus do not. Our findings support the idea that minorities may have a preference for descriptive representation even when it is not strategic to vote for coethnic or co-religious candidates.