Many countries make special institutional arrangements to guarantee the political representation of minorities. It is often assumed or hoped that such measures will translate into substantive representation: that minority representatives will act in the interest of their group. In this paper I explore the effects of reserved seats for the Scheduled Castes (SCs, often also referred to as untouchables or Dalits) in Indian state assemblies 1974-2001. Using census variables from 1971 and 2001, I create a unique set of development indicators at the constituency level for 15 Indian states. This new data makes it possible to control for the selection bias inherent in quotas being non-randomly assigned in the 1970s, by matching more than 3000 constituencies on variables from 1971. Comparing these matched constituencies over time, I find that 30 years of quotas had neither a positive nor a negative effect on the literacy rate or the percent non-working people among SCs or non-SCs in reserved constituencies. Interviews with politicians and bureaucrats in India in 2010 and 2011 show that the no-impact findings can be explained by the fact that SC politicians face the same constraints as other politicians in the system, such as electoral incentives and party control of candidate nominations. The findings do not suggest that SC quotas have been a failure, only that SC politicians tend to be responsive to their voters - irrespective of caste - in much the same way as other politicians, and that they tend to act as agents of their parties rather than as agents of their group.
