The intertemporal risk-return relation: A bivariate model approach

This paper examines the intertemporal risk-return relation using a more sensible empirical specification that is motivated by two concerns: the theoretical risk-return relation is an ex ante relation and the empirical method used to detect the relation should be reliable. We measure both the expected excess return and conditional variance jointly using the common information set based on a bivariate moving average representation of excess returns and variances. As a result, we can detect a significant positive relation between the expected excess return and the conditional variance. We also find that the positive relation is robust.