Time-Varying Relative Risk Aversion: Mechanisms and Evidence

This paper explores theoretically and empirically the issue of time-varying relative risk aversion. We analytically solve a parsimonious life-cycle portfolio choice model with the preferences given by Greenwood, Hercowitz and Huffman (1988, GHH). Our analytical solution identifies four partial equilibrium effects in our model with GHH preferences on risky shares through two channels, and two net effects whose signs hinge on the value of a key structural parameter. With household-level micro data, our mean and quantile regression results show that wealth negatively affects risky shares and the estimated effects are statistically significant and robust. This finding provides strong evidence to support our theoretical prediction. Thus, we show successfully that our portfolio choice model with GHH preferences provides a plausible underlying mechanism in understanding the wealth effect on risky shares in the microdata. Furthermore, we conclude that such a mechanism alone is not sufficient in explaining how risky shares respond to labor income and labor income risks in the microdata.