Sensitivity of Systematic Risk Estimates to the Return Measurement Interval under Serial Correlation.

This paper analytically and empirically investigates the sensitivity of the return measurement interval to the market beta estimate and suggests a market beta estimation method incorporating the investment horizon through a vector autoregressive (VAR) model when there is serial correlation in returns. The analytical relation between the beta estimate and the return measurement interval is obtained. Based on the analytical relation, a decision function for the intervalling effect is provided. It is found that the intervalling effect is mostly caused by January returns. Copyright 1999 by Kluwer Academic Publishers