Reexamination Of The Intervalling Effect On The CAPM Using

The intervalling bias on the pattern of residual returns may safely be ignored when constructing an event study methodology. The same relative results can be maintained whether either the residual return interval or the parameter estimating return interval is altered. Any event-related effect in the experimental group of a sample of stocks will appear, regardless of the parameter differencing and residual return intervals chosen. This ability to ignore the intervalling bias holds if the securities in the test sample come from a representative range of trading frequencies or if the event data are nonclustered. An explicit control may be needed if the sample of securities chosen is characterized by low trading frequencies. This will be especially critical in studies that try to use market models to identify such anomalies as the small firm effect. The data used in the analysis cover 1,652 stocks continuously listed for the 1980-1984 sample period.