Public Goods, Self-Selection and Optimal Income Taxation

Abstract

Using the self-selection approach to tax analysis, this paper derives a modified Samuelson Rule for the provision of public goods when the government deploys an optimal nonlinear income tax. This approach gives a straightforward interpretation of the central result in this area, generalises it, and provides a simple characterisation of optimal policy in a wide range of circumstances. The analysis also emphasises and clarifies the significance of the choice of numeraire for the optimality of decentralising public spending decisions. It is well-known that the famous Samuelson Rule for the provision of public goods may not be applicable when the tax revenues used to finance the public goods come from distorting taxes. The principle was recognized as long ago as Pigou (1947), long before the Samuelson Rule was devised. Pigou saw that in addition to the resource cost involved in transferring funds from the private to the public sector there was a deadweight loss involved. From this insight, he suggested that the criterion for providing public goods should be more stringent than the Samuelson Rule would later suggest. In particular, one might expect that the inclusion of the deadweight loss of taxes as part of the cost of providing a public good would lead to a reduction in its optimal provision relative to a situation in which lump-sum taxes were available. The problem was investigated more systematically by Atkinson and Stern (1974), who considered the case in which a public good is financed by a set of optimal commodity taxes. All individuals were assumed to be identical, so efficiency alone dictated the objective. They found that no prima facie case could be made for the Pigou position that the use of distortionary taxes called for a reduction in the supply of the public good. Their analysis clearly showed that there were two influences at work when the financing of a public good was by distorting taxes. On the one hand,