Time-Varying Risk Aversion and Dynamic Portfolio Allocation

We study the implications of time-varying risk aversion for dynamic portfolio allocation under the framework of regime-switching models. In our model, both asset returns and investor risk aversion are regime dependent: In a bull regime, asset return is high, volatility is low, and risk aversion is low, and the opposite happens in a bear regime. We develop an efficient dynamic programming algorithm that overcomes the challenges imposed by regime-dependent preference in obtaining time-consistent portfolio policies. Empirically, we show that CBOE Volatility Index (VIX) is an important predictor of regime shifts and investors with regime-dependent risk aversion achieve better investment performance than those with constant risk aversion.