The Employment Effects of Road Construction in Rural India ∗ PRELIMINARY - PLEASE DO NOT CITE WITHOUT PERMISSION

Despite a broad consensus that high transportation costs are a large barrier to economic development, many of the world’s poor live in rural communities without paved roads, impeding their access to outside markets and public services. Nevertheless, the impact of rural roads on economic activity has proven difficult for economists to assess. We first motivate our empirical approach by demonstrating the presence of large-scale manipulation of administrative records in order to improve the likelihood of road treatment. We then introduce a novel empirical strategy based on alternative government data. We use three estimation strategies that take advantage of the allocation rules of a large-scale rural road construction program in India to estimate the impact of village feeder roads on rural nonfarm economic activity. We first provide OLS estimates that are based on the time when rural roads were constructed. Second, we use a regression discontinuity around population cutoffs for program eligibility. Third, we instrument for road construction using the population rank of villages within districts, which dictates the order of road construction. We find that new paved roads lead to large increases in rural employment, as measured in both firm and demographic censuses. We then provide evidence that these effects differ based on the size of the village, with evidence suggesting that roads lower unemployment in smaller villages while facilitating structural transformation in larger ones. JEL Codes: O12/P16/D72. ∗We are thankful for useful discussions with Alberto Alesina, Josh Angrist, Lorenzo Casaburi, Shawn Cole, Ed Glaeser, Ricardo Hausmann, Richard Hornbeck, Lakshmi Iyer, Asim Khwaja, Michael Kremer, Sendhil Mullainathan, Rohini Pande, Andrei Shleifer and David Yanagizawa. We are grateful to Sandesh Dhungana for excellent research assistance. Mr. PC Mohanan of the Indian Ministry of Statistics has been invaluable in helping us use the Economic Census. This project received financial support from the Center for International Development and the Warburg Fund (Harvard University). All errors are our own. †Oxford University ‡Dartmouth College