Budgetgames and the private and social demand for mixed public goods

Abstract

We have indicated how budgetgames played by a representative sample of voters can be used to estimate the marginal social demand for mixed public goods. A model has been developed that enables us to distinguish between private and social demand considerations of the players. Information on marginal social demand is obtained by comparing the actual outcome of the game with a simulated game solution based on the optimization of private benefits only. This optimization problem is shown to be approximately equivalent to a linear programming problem in which each respondent minimizes the loss of subsidies on his consumption of the mixed public goods considered. Empirical results show the largest marginal social demand for public services for the elderly, followed by mental health care, primary and secondary education and higher education. A negative marginal social demand is expressed for in-patient health care, police and justice, outpatient health care and culture and recreation, the latter service having the smallest social demand. An unexpected result is the relatively large marginal social demand for higher education. Although this service is severly cut in the actual game, this outcome mainly derives from small private benefits to the average voter. Therefore, the usual interpretation of budgetgames — that identifies heavy budget cuts with small social demand — underestimates the social demand for higher education. The reverse is true for general in-patient health care. Although this service is moderately cut in the actual game, this result mainly derives from the large private benefits to the average voter. In this case the usual interpretation of the game overestimates social demand. We have also studied the dependence of the marginal social demand pattern on the political orientation of the voters. Apart from typical pure or almost pure public goods such as defence, general government and police and justice services — clearly favoured by right-wing voters — we find no strong dependence on political orientation.