This paper presents empirical evidence that accounting for heterogeneity in financial market participation is important for evaluating the empirical performance of the Consumption-based Capital Asset Pricing Model (C-CAPM). Using the U.S. Consumer Expenditure Survey as a common testing ground, I re-assess three well-known characterizations of the “equity premium puzzle”: (i) the inconsistency of the representative agent's IMRS with Hansen and Jagannathan bounds; (ii) Mehra and Prescott's calibration of a large representative agent's risk aversion; (iii) Hansen and the Singleton's large structural estimates of the preference parameters based on aggregate data. In all three cases, the estimates of risk aversion conditional upon financial market participation are not as far from reasonable values as the corresponding unconditional ones. The differences suggest that part of the “equity premium puzzle” can be accounted for by the use of a “representative agent” assumption rather than a more appropriate “representative stockholding agent” assumption.

We decompose the wealth effect on consumption into two components. First, we distinguish between exogenous and endogenous wealth changes. Second, we distinguish between anticipated and unanticipated exogenous changes. We estimate the impact of exogenous components using data from the 2008—2010 panel of the Italian Survey of Household Income and Wealth. The wealth effect is about 3 cents per (unexpected) euro increase in wealth and driven by house price changes. The consumption response to anticipated changes in wealth is of similar magnitude and also driven by housing. We show that these findings are consistent with binding borrowing constraints.