This paper examines policy effectiveness as a function of leader identity. We experimentally vary leader religious identity in a coordination game implemented in India and focus on citizen reactions to leader identity, controlling for leader actions. We find that minority leaders improve coordination, and majority leaders do not. Alternative treatment arms reveal that affirmative action for minorities reverses this result, while intergroup contact improves the effectiveness of leaders of both identities. We also find that minority leaders are less effective in towns with a history of intergroup conflict. Our results demonstrate that leader and policy effectiveness depend on citizen reactions, conditioned by social identity and past conflict.