Article ID: | iaor201522748 |
Volume: | 6 |
Issue: | 4 |
Start Page Number: | 275 |
End Page Number: | 288 |
Publication Date: | Dec 1983 |
Journal: | Journal of Financial Research |
Authors: | Bey Roger P |
Keywords: | investment, stochastic processes, simulation, statistics: regression |
The objective of this research was to investigate whether the market model is an appropriate description of the stochastic process generating security returns. In contrast to previous studies which concentrated only on the stationarity of β, the emphasis in this research was to study the stationarity of the entire market model. Four statistical tests, cusum, cusum of squares, moving regression, and time‐trending regression, were used. The data base consisted of 200 securities and four time periods. The cusum of squares generally identified the largest number of nonstationary securities. However, the amount and composition of the nonstationary sets varied considerably both by statistical test and time period. The lack of consistency in the stationarity of the market model limits the predictability of stationarity and, hence, hinders the application of the market model. Serious questions are raised as to whether the market model is an appropriate description of the stochastic process generating security returns.