An Intertemporal Risk Factor Model

Current factor models do not identify risks that matter to investors. To address this issue, we provide a factor model implementation of the ICAPM, which captures market risk and intertemporal risk (i.e., changes in long-term expected returns and volatility). We build our intertemporal risk factors as mimicking portfolios for changes in dividend yield and realized volatility and demonstrate that, ex-post, they capture news to long-term expected returns and volatility. Our estimated risk price signs are in line with the ICAPM and their magnitudes imply an average risk aversion around five. Moreover, the ICAPM performs comparably with (and mostly better than) previous factor models in terms of its maximum (out-of-sample and cost-adjusted) sharpe ratio as well as its pricing of the testing assets Lewellen, Nagel, and Shanken (2010) recommend: single stocks, industry portfolios, correlation-clustered portfolios, and bond portfolios