Abstract
We propose a theoretical framework for the heteroscedasticity, and in particular for the asymmetric volatility of asset returns. Our model is based on the assumption that some investors are subject to mental framing in a dynamic setting. The analysis of individual trading data confirms that, in line with our model, investors tilt their portfolio towards riskier (less risky) assets subsequent to losses (gains). Based on their behavior, we derive a volatility process that accounts for the asymmetry thoroughly investigated in previous empirical studies: the parameter estimation of our volatility model yields the predicted negative relationship between abnormal returns and ensuing volatility.