Highlights :

• A time-varying probability of ‘disaster’ is sufficient to generate a recession and an increase in risk premia in the literature.
• However, this result critically relies on the value of the elasticity of intertemporal substitution (EIS). Indeed, when the EIS is smaller than unity, the agents choose to save and invest more. In a real business cycle model, an increase in disaster risk thus creates a boom.
• The New Keynesian structure allows to conciliate the recessionary effects of disaster risk with a plausible value of the EIS through the sluggishness in price adjustments.
• The model then generates recession and deflation, as well as a decrease in consumption, investment and wages, in line with the preference shock literature, while it preserves the countercyclicality of risk premia.

 Abstract :
This paper analyzes the effects of a change in a small but time-varying “disaster risk” à la Gourio (2012) in a New Keynesian model. Real business cycle models featuring disaster risk have been successful in replicating observed moments of equity premia, yet their macroeconomic responses are highly sensitive to the chosen value of the elasticity of intertemporal substitution (EIS). In particular, we show here that an increase in the probability of disaster causes a recession only when imposing an EIS larger than unity, which may be arbitrarily large. Nevertheless, we also find that incorporating sticky prices allows to conciliate recessionary effects of the disaster risk with a plausible value of the EIS. The disaster risk shock causes endogenous shifts in preferences which provide a rationale for discount factor first- (Christiano et al., 2011) and second- (Basu and Bundick, 2014) moment shocks.

 Keywords : disaster risk | rare events | uncertainty | asset pricing | DSGE models | new Keynesian models | business cycles

 JEL : D81, D90, E20, E31, E32, E44, G12, Q54