Some financial analysts have postulated that the Dow-Jones Industrial Average (DJIA) is biased as a result of its treatment of stock splits. Any question of bias must be evaluated on an empirical basis, since the bias, if it exists, results from the interaction of the way the split adjustment is made, the subsequent price changes of the split stock, and the subsequent price changes of the unsplit stocks.
This paper develops two alternative tests for detecting the presence of bias in the DJIA. Seventeen stock splits over a 15-year period are then evaluated by the tests, with each split investigated on a daily basis for 12 weeks after the effective date of the split. In addition, the 17 stocks are studied to see whether stocks with dividend increases around the time of the split are more likely to cause bias in the DJIA.
Empirical results from both of the tests indicated no significant or consistent bias introduced in the DJIA either by the occurrence of a split or the particular method of adjusting for the split by the DJIA. In addition, a study of the split stocks with regard to dividend increases indicated no significant or consistent bias caused either by the split or the adjustment procedure. Thus, for the period studied and the tests used, no bias was found in the Dow-Jones Industrial Average as a result of stock splits.