Can Models with Idiosyncratic Risk Solve the Equity Premium Puzzle? Redux

Can idiosyncratic risk explain the equity premium? We revisit this question using a novel measure of imperfect risk sharing, implied by a large class of heterogeneous-agent models, constructed using household-level panel data. We identify a group of households – with relatively high income but low net worth – whose consumption is sufficiently volatile and risky to explain 94% of the observed U.S. Sharpe ratio. In contrast, the consumption dynamics of high net-worth individuals predict a negative Sharpe ratio and so do not constitute the relevant pricing factor, consistent with models featuring wealth motives.