The dynamics of crises and the equity premium

There has been a considerable debate whether disaster models like Barro (2006) can rationalize the equity premium puzzle. This is because empirically disasters are not single extreme events, but tend to be long-lasting periods in which moderate negative consumption growth realizations cluster. Our paper proposes a novel way to explain this stylized fact. By allowing for consumption drops that can spark an economic crisis, we introduce a new economic channel that combines long-run and short-run risk. First, we document that our model can match consumption data of several countries. Second, we show that in a model with recursive preferences our new channel generates a large equity risk premium even if the consumption drops are assumed to be of moderate size.