This paper devises an empirical methodology for discriminating between the median voter model and the oligarchy choice model when applied to the collective provision of a public good. In particular, an empirical methodology is engineered so that a nested test procedure can evaluate competing models. The authors apply this methodology to examine the demand for military activities of ten members of the NATO alliance. A two-stage least squares procedure, corrected for autocorrelation, is used to estimate the demand equations. Test results vary: some allies abide by the median voter model, others by the oligarchy model, and still others by neither.

This paper develops a correlated probit model to describe dichotomous choices that may contain a public-goods component or some other forms of interdependency. The key contribution of the paper is to formulate tests for interdependent behavior among agents. In particular, we examine the decisions by nations whether or not to ratify the Montreal Protocol on Substances that Deplete the Ozone Layer. Specifically, we reject free riding as a motive for not ratifying the Protocol, and we find little evidence that individual nations were influenced by the behavior of their largest trading partners. Hence, the data suggest that, with respect to the Montreal Protocol, most nations acted without regard for the actions of other nations.