• *I am grateful to Kim Border, Jim Jordan, John Ledyard and Preston McAfee for helpful comments and suggestions and Sung Wan Park for finding errors in the earlier version. I have also benefited from the many valuable suggestions of the anonymous referee. Research support from Seoul National University Development Foudation is gratefully acknowledged.


In this paper we present a simple game form implementing Lindahl allocations as Nash equilibrium outcomes, which has nice stability properties. we show that if the preferences of eaach consumer are representable by a utility function of the form a(y)xi+bi(y), where xi(y), where xi is the amount of private good and y, the amount of public good, then the Nash equilibrium of our geme is locally stable under the gradient adjustment process. This restriction on the preferences has been known in hte literature as the necessary and sufficient condition for the Pareto optimal amount of pukic goods to be independent of the private goods distribution. This type of preference includes quasi-linear preferences as a special case. but unlike quasi-linearity, this allows a non zero income effect of demand for public goods as well as private goods, which is often supported by empirical evidence. Our result shows how an equilbirium can be achieved over time by a decentralized strategy adjustement process for a fairly general class of environments, even in the absence of a dominant-strategy equilibrium.