“The Disposition Effect” Shefrin & Statman (1985)

by vonNudge

This paper is another classic in behavioural finance. The disposition effect refers to the tendency for investors to hold on to losers for too long and to sell winners too soon. And this is despite the quirks of the US tax code at the time, where short-term losses can be used for high tax deductions, while long-term gains are taxed at a lower rate than short-term gains. Rationally, therefore, investors should sell losers and hold on to winners. When in actual fact they display the exact opposite behaviour!

Why is this the case? The authors use a combination of prospect theory and mental accounting. From prospect theory, investors are risk-averse for gains and risk-seeking for losses. From mental accounting, investors tend to make decisions in isolation instead of integrating them into the bigger picture. Realising losses is psychologically painful: the investor has to admit to making a mistake. Holding on to the loss allows for the possibility of recouping the loss and closing that mental account out at a gain. This type of gambling behaviour is utility-maximising because investors are risk-seeking for losses. A stock that has increased above the purchase point leads to a mental account with a gain, where individuals are risk-averse. This means that larger gains lead to a proportionally smaller increase in utility than smaller gains. It therefore makes sense to close that account and “bank” the positive utility from the gain. Empirical evidence confirms that there is rather a lot of this going on from unsophisticated investors.

What else might be causing the disposition effect? I have a feeling that investors are once again looking at their financial decisions on a case-by-case basis and are failing to understand how the logic of portfolio theory requires them to integrate each decision into one overarching plan. Of course losing money hurts, but it is a perfectly acceptable consequence if you own a diversified portfolio of many different asset classes that tend to move in opposite directions.

Although in this context the logic of rebalancing requires adding to losing investments by selling winners. The situation is slightly different, however, as investors are rebalancing between asset classes (e.g. stocks, bonds, property), which is much less risky than individual assets (a given stock such as Apple). Investors who rebalance their portfolio can still use tax-loss harvesting (selling short-term losers) as a part of their strategy.

My feeling is that the disposition effect could be reduced with a wider framing of the investment problem. (See page 6 of the paper on some anecdotal evidence that the authors quote to this effect.)