Empirical Asset Pricing with Multi-Period Disasters and Partial Government Defaults

Abstract

According to the rare disaster hypothesis, the high excess returns that characterize U.S. postwar data resulted because investors ex ante demanded a compensation for possibly disastrous but very unlikely risks that they ex post did not suffer from. Empirical assessments of the rare disaster hypothesis are scarce, and the frequently used assumption that disasters shrink to one-period events is under suspicion of being the driving force behind the hypothesis' success in calibrations. This study uses the simulated method of moments to estimate a consumption-based asset pricing model that accounts for multi-period disasters, partial government defaults, and recursive investor preferences. The estimates are economically plausible, precise, and qualitatively insensitive with respect to different model specifications. Furthermore, the model-implied market Sharpe ratio, mean market return, and equity premium are comparable to the empirical data. The results thus help to restore the nexus between the real economy and financial markets that is implied by the consumption-based asset pricing paradigm.