Repeated interaction and the public provision of private goods

The literature suggests that governments can use in-kind transfers to design efficient and targeted redistribution schemes if individual incomes are not directly observable. We investigate the extent to which the self-selection property of in-kind transfers carries through if redistributive transfers are made repeatedly. In a two-period setting, the government may gain information about the individuals' incomes in the first period and exploit this information for making targeted transfers in the second-period. This, however, also triggers changes in the individuals' behavior. If the government can commit to its future policy, the least cost policy may involve randomization between cash and in-kind transfers. Without commitment, the dynamic setting works against the government's interest. It may no longer be able to use inkind transfers to generate information about the individuals' types.