The traditional income-leisure model treats income as a composite commodity; it is not appropriate for studying commodity subsidies which alter relative prices within the composite. I suggest reinterpreting the traditional model as a special case of a utility function weakly separable with respect to leisure and all other commodities. This interpretation allows the work incentive effects of any subsidy program to be inferred from the terms of the program and data on the work effort effects of any other subside program, most notably income maintenance experiments. I illustrate our approach by estimating the work incentive effects of public housing. The model implies that even if special complementarities between leisure or work and the subsidized good are neglected, in kind transfers will have different work incentive effects than equivalent cash transfers. In practice, in kind transfers will generally stimulate work efforts vis-a-vis equivalent cash grants.