Time series and cross-section parameter stability in the market model: the implications for event studies

• Published in: The European journal of finance 1997-09, Vol.3 (3), p.243-259
• Main Authors: Andrew Coutts, J., Mills, Terence C., Roberts, Jennifer
• Format: Article
• Language: English
• Subjects: Single-INDEX Market Model Parameter Estimates Time Series Cross-SECTION Stability
• ISSN: 1351-847X; 1466-4364
• DOI: 10.1080/135184797337462

This paper investigates the time series and cross-section stability of parameter estimates from the single-index market model, using a UK data set relating to the security prices of parent companies, divesting in the form of a management buyout. A battery of tests of structural stability are undertaken, and we find that instability exists in the vast majority of the fitted models, both in relation to changes in the estimation period, and also to changes in the cross-section sample of firms included in this analysis. The implications of instability for the event study method are clearly illustrated by the construction of recursive cumulative abnormal return series. Our results suggest that when the market model is used within the event study framework, the quantitative results are extremely sensitive to the chosen estimation period and cross-section sample of firms. We suggest that if event studies continue to be pursued in the applied finance literature, it is essential that tests of parameter stability are incorporated into this framework. In addition, 'sensitivity analysis', that is, changes to the estimation period and cross-section sample employed, should also be investigated, and conclusions should be limited to interpreting the patterns of the cumulative abnormal returns.