Developing democracies increasingly have welfare programs that use objective rules to identify beneficiaries, and directly transfer benefits to intended recipients, cutting out party intermediaries. This policy shift has potential to attract new voters, but runs the risk of alienating core supporters. To understand these electoral trade-offs, I leverage a natural experiment involving a large rule-based direct transfer program in India (a $2000 house), as well as survey experiments. Contrary to expectation, I find that core supporters do not punish their party for substantially benefiting out-groups. Among out-groups, receipt of benefits does not durably increase support for the ruling party. These findings shed new light on the impediments to distributive reform: it is ambivalent policy winners (out-groups), not angry losers (core supporters), that disincentivize politicians from reducing discretion and favoritism.