In many developing democracies, multiple local politicians have to collectively decide on the allocation of limited public spending. This paper explores the interplay between electoral competition and divided government in shaping these budget allocations. The paper puts forward a novel theory where divided local government can lead to higher welfare spending and lower taxes if both branches face strong electoral competition. Under unified government, both branches of government target spending to the same set of voters, whereas under divided government, they target spending to a different set of supporters. Given the same level of electoral competition, divided government can thus increase the net distributive pressure for a given polity. As a result, public spending will increase for those budget items where benefits can be shared among branches. The paper tests the proposition using a natural experiment in the Philippines where different branches of the local government are elected in separate first-past-the past elections. Exploiting the observational variation in electoral competition for different offices and as-if random variation in divided government, I show that divided government increases social welfare transfer and reduces taxes on property and businesses, but only if both actors face strong electoral competition. Spending for broad public goods remained unchanged. I find further support for the mechanism using qualitative interviews with local politicians.