Asset Prices in Turbulent Markets with Rare Disasters

I propose a parsimonious econometric model for the stochastic process governing the evolution of per capita consumption and stock market dividend over time. The model features stochastic volatility of consumption and dividend growth rates, and time-varying likelihood of rare disasters. I embed this time-variation of risk in an endowment economy with a representative agent and estimate the parameters from U.S. stock market data using Maximum Likelihood. Allowing for time-varying likelihood of rare disasters improves the model's performance. My model successfully explains a number of empirical puzzles: the high equity risk premium, excessive volatility of equity return, predictability of market returns through the price-to-dividend ratio, and the cyclical patterns observed in the term structure of the yield on dividend strips. In addition, the model-implied correlations between equity premium, variance risk premium, and the implied volatility of deep OTM put options are consistent with empirical findings in the literature.