The existence and role of group identities in societies is well established. In representative democracies with political competition, it is possible for parties to exploit social identities for electoral benefits. We ask, what happens when i) a group of voters with an underlying social identity unconditionally support one political party over the other, and ii) there is post-electoral conflict against the minority identity? We study political competition between two office motivated parties in which the citizenry comprises of three groups of voters. We allow parties to cultivate a vote bank -- an identity group affiliated with each party that votes en masse for that party. Critically, there is a wedge in the exogenous probability of identity formation, and therefore, ex-post conflict between majority and minority identities. This generates an asymmetry in conflict payoff for the majority identity that depends on the party in power. We show that this asymmetry in conflict payoff i) reduces bargaining power of both the vote bank groups; ii) increases political rents of the parties; and iii) decreases expected transfers to non-partisan voters. We test the theoretical predictions by constructing a novel dataset that combines data on Hindu-Muslim riots, household expenditures and state assembly elections in India for the period from 1983 to 2012.