Title: "Representation and development across groups: Evidence from a large-scale government program in India."
Abstract: Affirmative action skeptics often point to the self-undermining effects of quotas in the administration of public poverty alleviation programs. We examine how Scheduled Areas in India -- which reserve half of locally elected councils and their leadership positions for the historically disadvantaged Scheduled Tribes -- affect the implementation of the world's largest employment program, the National Rural Employment Guarantee Scheme (NREGS). A new dataset of nearly a quarter million villages allows us to characterize the overall quality of NREGS implementation and assess its distributional impacts across minorities and non-minorities, a substantial advance on existing work. Using a regression discontinuity design, we find that quotas yield significant benefits to targeted minorities. However, contrary to the expectations of affirmative action skeptics, the primary losers are the relatively privileged rather than other minorities. Overall, quotas work better than intended by yielding net positive program implementation.