A Comprehensive Approach to Revealed Preference Theory

Richter's theorem and Afriat's theorem are two fundamental results underlying modern revealed preference analysis. In this paper, we provide a version of Richter's theorem that characterizes the rationalizability of a choice data set with a continuous utility function (rather than simply a complete preorder as in the original result) and extend Afriat's theorem so it becomes applicable in choice environments other than the classical setting of consumer demand. Furthermore, while standard treatments give very different proofs for these two results, we introduce a framework within which both results can be formulated and established in tandem. We also demonstrate how our generalized versions of these theorems can be used in empirical studies. In particular, we apply our results to devise tests for rationalizability in the context of choice data over lotteries, contingent consumption, intertemporal consumption, and positions in policy space. Some new results on the revealed preference theory of consumer demand (for instance, on the possibility of deriving utility functions from estimated Engel curves) are also reported.