I examine determinants of stochastic relative risk aversion in conditional asset pricing models. I first develop time-series specification tests with non-linear state-space models with heteroskedasticity based on Merton (1973)'s ICAPM. I then established the following facts. First, the surplus consumption ratio implied by the external habit formation model is the most important determinant of relative risk aversion. Second, the CAY of Lettau and Ludvigson (2001a) without a look-ahead bias explains part of relative risk aversion, and the short term interest rate has some explanatory power for hedging components. Finally, I show the selected models from extensive time-series analysis are at least comparable to or better than the Fama-French three-factor model in explaining the value premium and the cross-section of industry portfolios.
time-varying relative risk aversion hedging components return predictability the value premium nonlinear state-space model with GARCH