The basic result of this paper is that when government provides either public goods, private goods with externalities, or transfers motivated by altruism, public provision is large enough to completely crowd out private provision. In contrast, when government provides private goods, public provision redistributes resources without completely crowding out private provision. Thus the coexistence of public and private provision indicates a private good; public provision that completely crowds out private provision indicates a public good or the presence of externalities.
The model argues that national defense, aid to the poor, aid to the poor elderly, and municipal activities such as street-cleaning are publicly provided for efficiency. Public fire and police protection are also consistent with the efficiency argument. Goods and transfers which are publicly provided for purely redistributive reasons are: aid to farmers, social security to the middle and upper class, postal service, and education.
The model allows public provision to transfer resources to indirect demanders. Even when public activity is motivated by efficiency considerations, it structures provision to redistribute some of the gains. Perhaps direct demanders receive their preferred level of public goods and altruistic transfers without indirect demanders receiving rents. This leaves unexplained the excess applications to become postal workers and school teachers, for example, and the intense lobbying efforts on the part of defense contractors, welfare recipients, the Gray Panthers, and the local teachers union.
As usual, many simplifying assumptions were made in this paper in order to get results. For example, treating the political process as a sequential process ignores interactions among different types of goods. In addition, I assumed fixed endowments, that individuals interact in a Cournot-Nash fashion, and that no citizens are indifferent to the public good or dislike it. Relaxing these assumptions may weaken the predictions or yield additional implications.