Disasters, Recoveries, and Predictability

I review the disaster explanation of the equity premium puzzle, discussed in Barro (2006) and Rietz (1988). In the data, disasters are often followed by recoveries. I study how recoveries a¤ect the implications of the model. This turns out to depend heavily on the elasticity of intertemporal substitution (IES). For a high IES, the disaster model does not generate a sizeable equity premium. I also study whether the disaster model can \u85t the time-series evidence on predictability of stock returns, as argued by Gabaix (2007). I \u85nd that the model has di ¢ culties matching these facts. Finally, I propose a cross-sectional test of the disaster model: assets which do better in disasters should have lower average returns. There seems to be little evidence supporting this prediction.