Finite sample properties of tests of the Epstein-Zin asset pricing model

This paper investigates the small sample properties of Hansen and Singleton (1982, Econometrica 50, 1269-1286)-type GMM tests of asset pricing restrictions implied by Epstein and Zin (1989, Econometrica 57, 937-969) preferences. The Monte Carlo results suggest that tests of the Epstein and Zin (1989, Econometrica 57, 937-969) asset pricing model often have little size-adjusted power to reject asset pricing restrictions implied by simpler, time and state separable expected utility preferences, even when parameters are chosen to make the difference between the relative risk aversion parameter and the reciprocal of the intertemporal substitution parameter large. There is evidence that a Wald test has greater power than other tests and that use of the Hansen et al. (1996, Journal of Business and Economic Statistics 14, 262-280) continuous-updating GMM estimator improves the power of the tests.