The Role of Temporal Aggregation in Asset Pricing Tests
Posted: 13 Jul 1998
Date Written: October 1995
This paper investigates one of the frequently advocated possible reasons for rejection of the Lucas Consumption Based Capital Asset Pricing Model (CCAPM). Empirical studies using US data have rejected the CCAPM, attributing the failure to the possible use of temporally aggregated data. This paper conducts a simulation based study to assess the significance of the "temporal aggregation bias" in CCAPM tests. The effects of the decision interval not matching the sampling interval of the system affects the estimates of the Relative Risk Aversion Coefficient and the Discount Factor as well as the critical J-statistic used in GMM Tests. I find a consistent (downward) bias in the CRRA coefficient which increases with the level of aggregation. The discount factor estimates decrease with the level of temporal aggregation. Surprisingly, the rejections of the model are more frequent at low levels of aggregation than at high levels. The paper is consistent with past studies regarding temporal aggregation effects for the "Permanent Income Hypothesis" and the "continuous version of the CAPM".
JEL Classification: G12, C1
Suggested Citation: Suggested Citation