On the testable implications of collective choice theories

We analyze collective choices in game forms from a revealed preference view point. We call the joint choice behavior of n agents Nash- (respectively, P areto-) rationalizable if there exist n preferences over the conceivable jo int actions such that the joint actions selected from each game form coinci de with the Nash equilibria (respectively, the Pareto optima) of the corres ponding game. In the two-agent case, we show that every deterministic joint behavior which is Nash-rationalizable is also Pareto-rationalizable. The c onverse if false. We further identify general necessary and sufficient cond itions for Nash-rationalizability of an n-agent joint choice behavior. We a lso define and characterize partial versions of the Nash- and Pareto-ration alizability requirements. (C) 2000 Academic Press.