We study effects of wartime violence on social cohesion in the context of Nepal's 10-year civil war. We begin with the observation that violence increased levels of collective action like voting and community organization-a finding consistent with other recent studies of postconflict societies. We use lab-in-the-field techniques to tease apart such effects. Our causalidentification strategy exploits communities' exogenous isolation from the unpredictable path of insurgency combined with matching. We find that violence-affected communities exhibit higher levels of prosocial motivation, measured by altruistic giving, public good contributions, investment in trust-based transactions, and willingness to reciprocate trustbased investments. We find evidence to support two social transformation mechanisms: (1) a purging mechanism by which less social persons disproportionately flee communities plagued by war and (2) a collective coping mechanism by which individuals who have few options to flee band together to cope with threats. The effects of war on social cohesion are due to a combination of psychological change, changes in individuals' material conditions and the incentives they face, and death and displacement that permanently alter the composition of communities. From a historical perspective, understanding precisely how violence transforms social cohesion is crucial for understanding how war affects a country's political development. From a practical perspective, understanding how violence transforms social cohesion is necessary for anticipating the political consequences of different wartime strategies (e.g., insurgency or counterinsurgency strategies). Changes in the social cohesion landscape are also important features of the social and political context for designing postconflict reconstruction. Indeed, postconflict social and political programming has tended to assume that war fragments

When do politicians prompt bureaucrats to provide effective services? Leveraging the uneven overlap of jurisdictions in India, we compare bureaucrats supervised by a single political principal with those supervised by multiple politicians. With an original dataset of nearly half a million villages, we find that implementation of India’sNational Rural Employment Guarantee Scheme, the largest employment program in the world, is substantially better where bureaucrats answer to a single politician. Regression discontinuity estimates help increase confidence that this result is causal. Our findings suggest that politicians face strong incentives to motivate bureaucrats as long as they internalize the benefits from doing so. In contrast to a large literature on the deleterious effects of political interventions, our results show that political influence may be more favorable to development than is commonly assumed.

No abstract

Does political affirmative action undermine or promote development? We present the first systematic analysis of Scheduled Areas in India, home to 100 million citizens, where local political office is reserved for the historically disadvantaged Scheduled Tribes. A newly constructed dataset of 217,000 villages allows us to probe conflicting hypotheses on the implementation of the world’s largest workfare program, the National Rural Employment Guarantee Scheme. We find that reservations deliver no worse overall outcomes, that there are large gains for targeted minorities, and that these gains come at the cost of the relatively privileged, not other minorities. We also find improvements in other pro-poor programs, including a rural roads program and general public goods. Reservations more closely align benefits to each group’s population share, allaying concerns of overcompensation for inequalities. Contrary to the expectations of skeptics, results indicate that affirmative action can redistribute both political and economic power without hindering overall development.

This is the accepted manuscript (post-print version) of the article.Contentwise, the post-print version is identical to the final published version, but there may be differences in typography and layout.