Confusion of Confusions: A Test of the Disposition Effect and Momentum (Digest Summary)

Focusing within a specific investing context on a bias in behavioral finance—that is, the disposition effect (the predilection of investors to retain losers and sell winners prematurely) confronted by a stock split—the author notes the post-split absence of this effect among some individual investors because of their failure to adjust purchase price reference points. This confusion of winner versus loser status in their stock holdings leads to significantly higher ex-date returns for splits with higher gains. But the identified magnitude is small relative to momentum, which remains robust for stocks without the disposition effect. Although the disposition effect may slow the incorporation of news into stock pricing, the author concludes that the disposition effect alone does not explain momentum’s influence.

Individual investors who are uninterested in active trading could find this article relevant. Those clients seem most susceptible to overlooking any split-adjusted reference prices among their holdings. The author highlights his research’s significance in that it provides pioneering evidence that (1) individual investors are inattentive to nominal changes in share price unrelated to fundamentals and (2) any abnormal returns experienced at the ex-date are related to the disposition effect. Up to that point, the disposition effect appears to characterize individual investors’ behavior. Post-split, there is minimal difference in their realization of gains and losses.

The author’s other investigation related to momentum could have broader reach within behavioral theory. The proposition previously made by many researchers that the disposition effect offers a causal explanation for momentum is tested here by the disposition effect’s absence. He confirms that there is no disposition effect post–split date and stock prices converge to fundamental values, but momentum is tangible in the data. This presence thus supports the conclusion that additional influences underpin momentum, although the author does not speculate on their specifics.

The author uses a dataset that was first used in 2000 by other researchers and that covers the trading, with full transaction details, of 78,000 households holding accounts at a large discount broker between January 1991 and November 1996. Paralleling recent disposition effect study methodology, investors’ sell versus hold decisions are monitored daily by using a logit regression as a function of the unrealized gains and dummy variables measuring price appreciation, current price relative to the prior month’s high or low, and average squared return of stock versus market over the prior two months to control for volatility. Both forward and reverse splits are tracked, with 1,916 distinct split events in the sample and reverse splits tallying 1% of the total. The author attempts to isolate the effect of splits from other factors that may influence transactions. He claims that comparison is possible between two investors holding the exact same splitting stock on the same day but with a pre-split versus post-split nominal purchase price. Return predictability following any breakdown of the disposition effect is gauged by using weekly CRSP data covering stock splits of all NYSE and AMEX stocks between 1967 and 2011.

Three hypotheses are tested via regressions that model investor demand functions to determine whether post-split (1) the rate of taking gains versus losses is indistinguishable, (2) stocks with the largest gains have the largest ex-date price jumps, and (3) the absence of a disposition effect eliminates momentum. The author’s findings support the first and second hypotheses and reject the third.

The author extends research on the disposition effect rather than breaking new ground regarding concepts, datasets, or methodological design. But his thorough approach of investigating the analytical space related to the implications of the effects of stock splits gives the work credibility in the world of behavioral finance. Given the author’s concentration on individual investors, it would be informative to know whether the identified anomalies affect market pricing overall for the affected stocks as well as the related opportunities for associated trading strategies. His confirmation of the persistence of momentum indicates that further research could have productive applications.