Governments distribute a variety of benefits to win votes. Why do some benefits have greater electoral impact than others? This paper provides descriptive evidence that a $10 LPG cylinder and $2000 house have comparable electoral impact in India. This motivates a typology of distributive decisions in which politicians face two key trade-offs: first, given a finite budget, politicians can distribute a cheap benefit to many voters or an expensive benefit to fewer voters; and second, they can either distribute the benefit through brokers or a government program. Clientelism skews distribution in favor of loyalists but provides effective credit claiming. Programmatic distribution provides better targeting but worse credit claiming. Using data from India's National Election Studies, I show that there is political targeting of the LPG cylinder but not the house. LPG cylinder recipients are also more likely to be contacted by the ruling party broker but not house recipients. I conclude that party elites pursue a mixed strategy while distributing benefits, using brokers to deliver cheap benefits and government programs to deliver expensive benefits. Ironically, excluding brokers from the expensive benefit can have political costs because brokers make up for the value deficit in a benefit through effective canvassing.