General properties of rational stock-market fluctuations

Which pricing kernel restrictions are needed to make low dimensional Markov models consistent with given sets of predictions on aggregate stock-market fluctuations ? This paper develops theoretical test conditions addressing this and related reverse engineering issues arising within a fairly general class of long-lived asset pricing models. These conditions solely affect the first primitives of the economy (probabilistic descriptions of the world, information structures, and preferences). They thus remove some of the arbitrariness related to the specification of theoretical models involving unobserved variables, state-dependent preferences, and incomplete markets.