A simple model of how a ruler's religious identity affects public goods provision. * Rulers reduce public goods if private substitutes excludable by religion exist. * Theory tested by exploiting variation in rulers' religion in Indian Princely States. * Muslim-ruled states had lower Hindu literacy but had no impact on Muslim literacy. * Religion of ruler had no impact on railroad ownership or post office provision. This paper describes a simple model of how a ruler's religious identity affects public goods provision. Our primary insight is that rulers reduce public goods expenditures to a greater degree when there are privately-provided substitutes excludable by religion. The basic idea is that if the good is provided privately to the ruler's co-religionists, the ruler faces weaker incentives to provide this public good because his co-religionists receive lower marginal utility from its provision. Testing such a conjecture is an empirical challenge, however, since the religious identity of rulers rarely varies over time and place. We address this problem by exploiting variation in the religion of rulers in the Indian Princely States. Using data from the 1911 and 1931 Indian censuses, we find that Muslim-ruled states had lower Hindu literacy but had no significant impact on Muslim literacy. This result is consistent with our model, as Muslim religious schools provided a substitute for public schools that served both Hindus and Muslims. The model is further substantiated by the fact that the religion of the ruler had no statistically significant impact on railroad ownership or post office provision, neither of which had privately-provided substitutes.