State dependence of aggregated risk aversion: Evidence for the German stock market

We propose a dynamic generalization of the Capital Asset Pricing Model (CAPM) that allows for a time-varying market price of risk (MPR) reflecting both cross market dependence and future investment opportunities. The realized volatility approach is employed to determine market risk. The advocated state space model takes autoregressive dynamics of the MPR and predetermined state variables into account. For the case of the DAX, the major German stock index, the empirical analysis strongly underpins time variation of risk compensation. The MPR is conditioned upon the EURIBOR, a national and an international term spread, returns of the Dow-Jones-Industrial-Average-Index (DOW), and a dummy variable hinting at excess activity of noise traders. Moreover, we document forecasting results based on a short horizon trading strategy. The proposed model is characterized by strong market timing ability.