Edith Cowan University
Place of Publication
Joondalup, Western Australia
Faculty of Business and Public Management
School of Finance and Business Economics
In this study, I test the robustness of size and book-to-market effects on average stock returns reported by Fama and French (FF, 1992, 1993) using a sample that is less subject to survivorship bias and a longer sampling period. Specifically, I exclude NASDAQ stocks in the sample to reduce survivorship bias that has largely been induced by Compustat during an expansion project in 1978. Survivorship bias exists when the sustaining and successful firms are included in the data and those firms that fail or merge with other firms are omitted. Since NASDAQ stocks tend to have relatively smaller stocks than the NYSE and AMEX stocks, they were not included to mitigate such bias. I also test whether size and BE/ME factors are proxied by growth in earnings as studies like FF (1995) and Harris and Marston (1994) reported strong correlation between the two factors and growth.
There are two major findings from this study. First, I find that Fama and French results are period and sample specific. Size and book-to-market equity are only significant in the earlier subperiod from 1973 to 1984 but fail to capture variations in average stock returns over the most recent years. Furthermore, the exclusion of the NASDAQ stocks also contributes to the insignificance of size and value premiums as survivorship bias is reduced. Overall, the FF findings are largely driven by the returns of NASDAQ stocks and the earlier sampling period.
Second, although growth in earnings is significant in both subperiods, its opposite signs in each subperiod cancel each other’s significance in the overall period. I also did not find that size and BE/ME are proxied by growth in earnings as it fails to absorb the significance of these factors. Finally, Beta is found to be insignificant in the presence of both size and value factors.