Snakes and Ladders: Asset Pricing with Household Disasters

Abstract

Disaster models are typically agnostic as to what spawns tail events, but are often calibrated on labor shocks. Using a novel way to construct rare event factors using the CEX survey data, I study whether the type of idiosyncratic shock modeled matters empirically. I estimate an augmented consumption-based asset pricing model, and find that while labor shocks do come through, they surprisingly do not significantly improve the fit of the model when used on their own, compared to a standard model with no shock factors. I show that including labor shocks with other types of events, notably housing, health and demographic shocks, serves to enhance the goodness-of-fit, reduces the pricing error, and generates plausible parameters for the utility function. The results suggest that labor is indeed a channel of idiosyncratic tail risk for asset pricing models, but the variations of the risk premium are most effectively captured when demographic, housing and health shocks are also included.