Disposition Effect: Why Investors Sell Winners and Keep Losers

A purchase price is a fact about your own past rather than about a company's prospects, and it is nonetheless the number that decides most sales. Once a position is filed as a winner or a loser, the two sides of that line get treated differently: the winner offers a gain that can be banked today, and the loser offers a loss that only becomes real at the moment of sale. Holding on keeps it on paper.

Hersh Shefrin and Meir Statman put four mechanisms together in 1985. Prospect theory supplies the shape: measured from the purchase price, the value curve is concave in the region of gains, which makes a certain gain attractive, and convex in the region of losses, which makes a gamble on recovery attractive. Mental accounting supplies the ledger, because each holding gets its own account rather than being judged as part of a portfolio. Regret aversion supplies the motive to wait, since selling converts a bad decision into an admitted one. Self-control supplies the rest: people describe the rule they cannot keep, which is to cut losses.

Later work added another ingredient. On the realization utility account of Nicholas Barberis and Wei Xiong, the pleasure comes at the moment a gain is booked rather than from the wealth itself, which predicts the trading pattern observed and also predicts that investors will sell winners even when they expect them to keep rising. Market consequences follow: Andrea Frazzini showed in 2006 that the reluctance to sell at a loss slows the way prices absorb bad news, which is part of why stock prices drift after earnings announcements.

The pattern is not a laboratory artifact, and it is not explained away by tax. Martin Weber and Colin Camerer ran the trading game in a laboratory in 1998 and found the same preference for selling winners, with one instructive exception: when their design sold every position automatically at the end of each period, so that holding on required an active repurchase, the effect was much smaller. That is the clearest evidence that the behavior depends on who has to act, not on a view about the assets.

The name comes from Hersh Shefrin and Meir Statman's 1985 paper in the Journal of Finance, "The disposition to sell winners too early and ride losers too long: Theory and evidence." It is a theory paper: the authors drew together prospect theory, mental accounting, regret and self-control into a single account of why investors trade this way, and assembled the evidence then available, including the awkward fact that the tax-efficient strategy in the United States is the opposite of what investors do.

The decisive test came in 1998, when Terrance Odean published "Are investors reluctant to realize their losses?" in the same journal. He had the trading records of 10,000 accounts at a large discount brokerage from 1987 to 1993, which let him do something earlier work could not: count not just the sales but the opportunities. For each day an investor sold anything, he counted how many of their holdings were showing a gain and how many were showing a loss, then measured what fraction of each was actually sold.

The proportion of gains realized was 0.148 and the proportion of losses realized was 0.098. Investors were selling their winners at about one and a half times the rate at which they sold their losers, and this was not a reflection of a shrewd view about the stocks: the winners they sold outperformed the losers they kept by 3.4 percent over the following year. Loss realizations rose in December, the month in which the tax benefit of booking a loss is most salient, which is the pattern a tax-driven investor would show all year.

Odean's method, usually written as the gap between the proportion of gains realized and the proportion of losses realized, became the standard measure, and the same gap has since been reported in other countries, in professional as well as retail accounts, and in assets from houses to employee stock. Where it has been found to shrink is in accounts run with rules: automatic rebalancing, scheduled contributions and designs like Weber and Camerer's, where keeping a position takes an action.

The effect needs only a purchase price and the option to wait. The rows below are illustrations, not findings from the studies cited on this page.

Every workable counter removes the purchase price from the decision. This page describes research findings and is not advice about your own money.

• Ask the buying question instead of the selling question. Would you put money into this holding today at today's price? If not, the gain or loss you are sitting on is not a reason to keep it.
• Write the rules before the prices move. Rebalancing bands, a schedule for selling vested shares and a position limit all decide the sale in advance.
• Make holding the action. In Weber and Camerer's experiment the effect shrank when positions were closed automatically and had to be bought back deliberately.
• Judge the portfolio, not the position. Mental accounting is one of the four mechanisms in the original theory, and the counter to it is to look at the total rather than at each line.
• Check the tax arithmetic rather than assuming it supports you. In the United States, realizing a loss can offset realized gains, which is the opposite of what the behavior does for eleven months of the year.

The disposition effect has three close neighbors, and the difference is in what the decision is anchored to.

• It is not the sunk cost fallacy, which is about continuing to spend because of what has already been spent. The disposition effect needs no further investment: holding a losing position is a decision not to sell, and the costs are the returns given up.
• It is not loss aversion, which is the asymmetry that losses weigh more than equal gains. Loss aversion is one ingredient in the explanation; the disposition effect is the trading pattern that results once a purchase price defines where losses begin.
• It is not the endowment effect, the higher value placed on what you already own. That raises the price you would accept for anything you hold, whether it has risen or fallen; the disposition effect treats risen and fallen holdings differently.