Most behavioral findings in this hub have been dismantled by replication failures. The disposition effect has not. In 1985, two finance professors formalized a pattern that brokers had noticed for decades: ordinary investors sell their winning stocks too quickly and hold their losing stocks too long. Thirteen years later, Terrance Odean took the theory to 10,000 retail brokerage accounts and found exactly the pattern Shefrin and Statman had predicted. Eight years after that, Andrea Frazzini found it in mutual fund managers --- the supposed professionals. The disposition effect is one of the cleanest examples in this hub of what robust behavioral finance actually looks like when the empirical foundation holds.
If you have been reading through this hub, you have watched a long list of canonical behavioral findings collapse under scrutiny. Ego depletion failed to replicate. Power posing was recanted by one of its own authors. Money priming evaporated. The marshmallow test shrank dramatically once parental socioeconomic status was controlled for. The bystander effect’s foundational anecdote turned out to have been substantially fabricated by a New York Times reporter. Even some findings still taught in MBA programs --- mental accounting, hot-hand fallacy, anchoring at certain magnitudes --- have shrunk substantially or generalized poorly when tested rigorously. A reasonable skeptic by this point might conclude that behavioral economics as a field is simply unreliable.
That conclusion would be wrong, and the disposition effect is one of the cleanest cases for why. Unlike most of the findings dismantled elsewhere in this hub, the disposition effect was not generated by a clever lab paradigm that fails to generalize to real-world settings. It was generated by Hersh Shefrin and Meir Statman observing, in 1985, what investors actually do in real brokerage accounts with real money, and then writing down a behavioral theory that predicted the pattern. That theory has been tested, ruthlessly, in the largest available datasets: 10,000 retail accounts in Odean 1998, the universe of US equity mutual funds in Frazzini 2006, futures traders in Locke and Mann 2005, Finnish retail investors in Grinblatt and Keloharju 2001, Israeli professional traders, Chinese retail accounts, real estate transactions, and even sports-betting markets. The qualitative pattern --- investors realizing gains at substantially higher rates than they realize losses --- has held up in every dataset where it has been tested.
This is what makes the disposition effect a calibration anchor for the entire field. When you are evaluating a behavioral-economics or behavioral-finance claim, the disposition effect is the benchmark for what real evidence looks like. Not “a single underpowered laboratory study found a marginal effect at p = 0.048.” But: “the same behavioral pattern shows up in retail accounts, professional accounts, fund managers, real estate sellers, and sports bettors, across multiple countries and across multiple decades, motivating formal theoretical work that has been adopted into mainstream finance.”
Here is what holds up, why it survived, and what makes it one of the cleanest anti-examples in this hub.
The Pattern: What Retail Investors Actually Do
Walk into any retail brokerage office in 1980 and ask the brokers what they consistently observed about their clients. The answer would have been some version of: “They take their gains too quickly, and they refuse to take their losses.” A stock goes up 15 percent, the client calls and asks to sell. A stock goes down 15 percent, the client calls and says they are going to “hold on and wait for it to come back.” The pattern was so well-known that it was part of broker folk wisdom decades before anyone formalized it.
What Shefrin and Statman did in their 1985 Journal of Finance paper was give the pattern a name --- “the disposition effect” --- and a theoretical foundation grounded in prospect theory’s value function plus a mental-accounting framework. The behavioral story is straightforward. Each stock purchase opens what Shefrin and Statman called a “mental account,” anchored to the purchase price. The investor evaluates the position relative to that reference point: above the purchase price, the position is “in the gain domain”; below it, in the loss domain. Prospect theory’s value function says that decision-makers in the gain domain become risk-averse and prefer to lock in certain gains rather than continuing to expose themselves to volatility. In the loss domain, the same value function predicts that decision-makers become risk-seeking and prefer to “stay in the game” in the hope of avoiding having to realize a loss.
The result is a systematic asymmetry in selling behavior. Stocks above the purchase price get sold at high rates because the investor wants to lock in the gain. Stocks below the purchase price get held at high rates because the investor wants to avoid realizing the loss. The pattern is not driven by rational tax considerations (in fact, tax considerations predict the opposite pattern --- realize losses to take the tax deduction, defer gains to defer the tax bill). It is not driven by superior information about which stocks are likely to outperform (Odean would later show that the stocks investors sell tend to outperform the stocks they hold, the opposite of what an information-based story would predict). It is driven by the psychological architecture of how people evaluate gains and losses relative to a reference point.
Shefrin and Statman’s 1985 paper laid out the theory but did not have access to large-scale individual trading data. The smoking-gun empirical evidence came thirteen years later, when Terrance Odean got hold of 10,000 retail brokerage accounts and the disposition effect went from theoretical prediction to one of the most rigorously documented anomalies in finance.
Odean 1998: The 10,000-Account Replication
The data Terrance Odean used for his 1998 Journal of Finance paper, “Are Investors Reluctant to Realize Their Losses?”, was at the time the largest dataset of individual investor trading behavior anyone in academic finance had analyzed. He had the complete trading records of 10,000 accounts at a large discount brokerage from 1987 through 1993 --- every buy, every sell, every share, every price, every date. The dataset’s size and granularity made it possible to test the disposition effect not as a stylized observation but as a precise quantitative prediction.
The key methodological innovation in Odean 1998 was the construction of two diagnostic statistics that controlled for the “supply” of winners and losers in each investor’s portfolio. Naively, you might think you could test the disposition effect by counting whether investors sold more winners than losers in absolute terms. But that counts wrong: an investor who happens to have more winners than losers in their portfolio will sell more winners than losers even with completely unbiased selling behavior. Odean’s fix was to compute, for each day on which an investor sold any stock at all, two ratios.
The Proportion of Gains Realized (PGR) is the number of gains the investor realized that day, divided by the total number of gains in the investor’s portfolio that day (realized plus unrealized). The Proportion of Losses Realized (PLR) is the corresponding statistic for losses. Under the null hypothesis of no disposition effect, these two ratios should be equal: a stock’s status as a gain or a loss should not affect the probability that the investor sells it. Under the disposition effect, PGR should exceed PLR --- investors should sell gains at higher rates than they sell losses.
The empirical result was decisive. Across the 10,000-account sample, the aggregate PGR was 14.8 percent, and the aggregate PLR was 9.8 percent. Investors realized gains at approximately 1.5 times the rate at which they realized losses. The difference was statistically and economically enormous. It was robust across the years 1987 through 1993, across the December tax-loss-selling month and the other eleven months of the year, across stocks with different levels of trading volume, and across investors with different account sizes.
The 1.5x ratio is the empirical anchor that every subsequent disposition-effect paper has either replicated or modestly varied. The exact number depends on the sample, the time period, and the methodology, but the qualitative pattern --- PGR substantially exceeding PLR by a factor in the range of 1.3 to 2.0 --- has held up in dozens of subsequent datasets. The Odean 1998 paper has been cited over 7,000 times. Its methodology has become the standard way to test for the disposition effect, used in academic studies in the United States, Finland, Germany, Israel, China, Singapore, Brazil, and India.
What Odean’s data also showed --- and this is the part that should make any disposition-effect defender uncomfortable about whether the bias is “really” rational --- is that the stocks investors sold tended to outperform the stocks they kept. The winners they sold went on, in the subsequent year, to outperform the losers they held by approximately 3.4 percent. This is the opposite of what a rational, information-based story would predict. If investors were selling their winners because they had superior information that those stocks were about to underperform, the sold winners should have underperformed the held losers. They did not. The disposition effect is not a sophisticated information-processing strategy. It is a behavioral anomaly that costs investors real money on a documented, measurable scale.
The Mechanisms: Loss Aversion Plus Mental Accounting
The disposition effect is one of the cleanest empirical manifestations of two interlocking psychological mechanisms: loss aversion and mental accounting. Both mechanisms were independently documented and theoretically formalized in the same decade that Shefrin and Statman’s 1985 paper appeared, and the disposition effect is in part what made the broader theoretical case for those mechanisms.
Loss aversion is the property of prospect theory’s value function that losses loom larger than equivalent gains: the disutility of losing $X is, on Kahneman and Tversky’s estimates, roughly two to two-and-a-half times the utility of gaining $X. Loss aversion alone is not sufficient to generate the disposition effect, because loss aversion is a feature of preferences over outcomes, and the disposition effect is a pattern of behavior over the act of selling. The bridge between the two is the act of “realizing” a paper loss by selling: the moment of sale converts a paper loss into a realized loss, and that conversion is psychologically significant.
Mental accounting, formalized by Richard Thaler in 1985 (the same year as Shefrin and Statman’s paper), is the practice of treating each financial position as a discrete “account” that gets evaluated on its own terms, rather than aggregating across positions. Under standard portfolio theory, an investor should be indifferent about which specific stock they sell to raise cash, because all the gains and losses on individual positions are absorbed into the total portfolio value. Under mental accounting, each position has its own gain-loss status, and the investor is not indifferent: selling a winner closes a “winning account” and feels good; selling a loser closes a “losing account” and feels bad.
The combination of loss aversion and mental accounting is what generates the disposition effect’s specific empirical signature. The investor frames each position as a separate mental account anchored to the purchase price; loss aversion makes the prospect of closing an account at a loss substantially more aversive than the prospect of closing an account at a gain is attractive; the asymmetric aversiveness produces the asymmetric selling behavior that Odean documented. Both mechanisms are necessary. Loss aversion without mental accounting would not generate the position-by-position asymmetry: an investor who aggregated all gains and losses into a single portfolio mental account would not care which specific positions they sold. Mental accounting without loss aversion would not generate the asymmetry either: an investor who treated each position as a separate account but was not loss-averse would sell winners and losers at the same rate.
The disposition effect’s mechanism-level cleanness is what makes it theoretically valuable for the field. It is not just an empirical regularity --- it is an empirical regularity that comes out as a structural prediction of two independently motivated psychological mechanisms, and it can be used to test those mechanisms in domains far from the original brokerage-account application.
Frazzini 2006: The Professionals Are Not Immune
The disposition effect’s most uncomfortable extension came in 2006, when Andrea Frazzini published “The Disposition Effect and Underreaction to News” in the Journal of Finance. Frazzini’s question was simple: do the same patterns of behavior that Odean documented in retail accounts show up in the trading decisions of professional mutual fund managers --- the people whose job is to make rational portfolio allocation decisions and who are supposed to be insulated from retail-investor biases by their training, their incentive structures, and their institutional context?
The answer was yes. Frazzini analyzed the equity holdings of US mutual funds from 1980 through 2002 using the Thomson Financial database. For each fund, in each quarter, he computed the analog of Odean’s PGR and PLR statistics. He found that mutual fund managers, in aggregate, realized gains at substantially higher rates than they realized losses --- the disposition effect was present, with a magnitude smaller than Odean had found in retail accounts but still statistically and economically meaningful.
The magnitude was about half of the retail-investor magnitude, depending on the specification, which is the most honest way to read the result. The professionals were not as biased as the amateurs. But they were biased in the same direction, by mechanisms that prospect theory and mental accounting predict should operate on professionals just as they operate on amateurs. The training, the incentives, and the institutional context were not enough to eliminate the bias.
Frazzini’s paper went further than just documenting the bias in fund managers. He showed that the disposition effect contributes mechanically to one of the major asset-pricing anomalies in finance: post-earnings-announcement drift, the tendency for stocks with positive earnings surprises to keep going up and stocks with negative earnings surprises to keep going down for weeks after the announcement. The behavioral story is: a positive earnings surprise creates winners (relative to recent purchase prices) for many of the stock’s holders; the disposition effect makes those holders sell into the price increase, dampening the upward price move; the underreaction creates a continuing price drift in the same direction for several weeks as the disposition-driven selling pressure exhausts itself. Frazzini provided cross-sectional evidence consistent with this mechanism: stocks with a higher proportion of disposition-affected holders exhibited larger post-earnings drift, controlling for size, value, and momentum factors.
The Frazzini 2006 result is one of the cleanest examples in finance of a behavioral mechanism causing a documented asset-pricing anomaly. The disposition effect is not just a curiosity about how individual investors mismanage their portfolios. It is a mechanism that contributes to systematic, exploitable patterns in stock returns. The mechanism shows up in retail accounts (Odean 1998), in professional accounts (Frazzini 2006), and in the cross-section of stock returns (the post-earnings-drift evidence in Frazzini and elsewhere). The behavioral story is not a stylized observation --- it is a quantitatively predictive theory that explains both individual behavior and aggregate market patterns.
The Replication Record: A Tour Through The Datasets
The disposition effect has been replicated, in some form, in essentially every dataset where it has been seriously tested. This is one of the most consistent replication records in all of behavioral economics, and it is worth surveying the breadth.
Grinblatt and Keloharju (2001), in the Journal of Finance, used the universe of Finnish equity-trading data --- every trade, by every retail and institutional investor, on the Helsinki Stock Exchange from 1995 through 1996. The Finnish dataset is uniquely valuable because it captures the entire population of investors rather than a sample. The authors documented disposition-effect behavior in both retail and institutional traders, with retail investors exhibiting roughly twice the magnitude of institutional investors --- consistent with Frazzini’s later result that professionals are biased in the same direction as amateurs but less so.
Locke and Mann (2005), in the Journal of Financial Economics, analyzed the trading behavior of professional futures-pit traders in Chicago. These are people whose entire job is short-term trading, who are paid to be unemotional about positions, and who hold positions for minutes or hours rather than days or months. The disposition effect showed up even at those time horizons, with traders holding losing positions longer than winning positions by statistically significant margins. Notably, Locke and Mann also showed that the worst-performing traders exhibited the largest disposition effect, while the best-performing traders exhibited the smallest --- consistent with the bias being a real performance drag rather than a rational adaptation to market microstructure.
Genesove and Mayer (2001), in the Quarterly Journal of Economics, applied the disposition-effect logic to the Boston housing market in the 1990s, when prices had recently declined from their 1980s peak. They found that homeowners who would have to sell at a nominal loss --- below their purchase price --- set list prices substantially higher than otherwise-comparable homeowners selling at a gain, and they took substantially longer to sell their homes. The disposition effect’s reference-dependence shows up in real-estate transactions just as clearly as in stock transactions, and the mechanism’s prediction --- reluctance to realize a nominal loss, even when the loss is economically rational --- holds in a market where the “asset” is the family home rather than an equity share.
Weber and Camerer (1998), in the Journal of Economic Behavior and Organization, ran a controlled laboratory experiment using simulated stock-trading scenarios. The advantage of the laboratory setting is that the researchers could randomly assign reference points: in one condition, subjects’ “purchase price” was set such that a stock was a gain; in another, the same stock at the same current price was framed as a loss. The disposition-effect pattern --- higher willingness to sell in the gain condition than in the loss condition --- showed up as a structural feature of how subjects evaluated the same objective situation under different reference frames. This is the experimental complement to the Odean and Grinblatt-Keloharju observational results, and it confirms that the mechanism is reference-dependence as theorized rather than some unobserved characteristic of investors who happen to have gains versus losses in real accounts.
Kaustia (2010), in the Journal of Financial and Quantitative Analysis, conducted what may be the most direct test of whether prospect theory’s value function specifically generates the disposition effect. The puzzle Kaustia identified is that prospect theory’s specific functional form --- with its kink at the reference point and its specific curvature in the gain and loss domains --- makes more specific predictions about disposition behavior than just “PGR > PLR.” Specifically, prospect theory predicts that the propensity to sell should be a step function of the gain-loss status: discontinuous at the reference point, relatively flat as the position moves further into the gain or loss domain. Kaustia’s empirical results were consistent with the step-function shape but with caveats about whether prospect theory’s specific functional form is the right mechanism or whether other reference-dependent theories fit the data equally well. The result is that the disposition effect is robust as a pattern but the precise theoretical mechanism is still being refined.
These five studies plus Odean 1998 and Frazzini 2006 are not exhaustive. The disposition effect has been documented in Chinese retail investor data, in Israeli stock-trading data, in sports-betting markets (where bettors hold losing tickets longer than winning tickets), and in cryptocurrency-trading data. The cross-cultural, cross-market, cross-asset-class replication record is one of the strongest in all of behavioral finance. It is the kind of replication evidence that makes a behavioral claim a benchmark rather than a curiosity.
Generalization Beyond Stock Investing
The disposition effect’s mechanism --- reference-dependence plus loss aversion plus mental accounting --- predicts that the same behavioral pattern should appear in any setting where a decision-maker can either “realize a loss” by closing a position or “stay in the game” by keeping it open. This generalization has been tested in domains far from retail equity trading, and the prediction holds.
Project management and project sunsetting. Project managers exhibit disposition-style behavior when deciding whether to terminate underperforming projects. A project that has consumed $5 million of budget against an expected $10 million is in the “loss domain” relative to the original budget reference point, and the same loss-aversion mechanism that keeps stock-investors holding losing positions keeps project sponsors funding losing projects. This is the well-documented “sunk-cost fallacy” pattern in organizational decision-making, and it is the same psychological mechanism that generates the disposition effect in equity trading. The mechanism’s prediction is that pre-commitment to exit criteria --- specifying in advance the conditions under which a project will be terminated --- should reduce the bias, and the empirical evidence on stage-gate processes in corporate R&D management is broadly consistent with this prediction.
Product portfolio management. Product managers exhibit disposition-style behavior when deciding which products to retire. A product that is underperforming relative to its launch expectations is in the loss domain, and the same reluctance to realize a loss that keeps investors holding losing stocks keeps product managers continuing to invest in underperforming products. The corporate-finance literature on “zombie projects” --- products and lines of business that should have been killed years earlier but are kept alive by managerial reluctance to write off the accumulated investment --- is a direct application of the disposition-effect mechanism to product-portfolio decisions.
Real estate. Genesove and Mayer 2001 documented this for residential housing, but the same pattern shows up in commercial real estate. Sellers anchored to a previous purchase price set list prices that are systematically too high relative to current market value, leading to longer time-on-market and worse eventual sale prices. The mechanism is the same reference-dependence that drives stock-investor disposition behavior, and the consequences --- worse outcomes for the seller as a function of unwillingness to take a nominal loss --- are the same.
Sports betting and gambling. Bettors hold losing tickets longer (in markets that allow position changes) and chase losses with additional bets that they would not place from a neutral starting position. The “doubling down to chase losses” pattern in gambling is the same behavioral mechanism that the disposition effect documents in equity trading, and it has been replicated in horse-racing wagering data, sports-betting platforms, and online poker datasets.
Managerial career decisions. Executives who have invested years building a position in a declining business unit exhibit disposition-style reluctance to abandon that position, even when the rational career move would be to switch into a growing area. The reference point in this case is the implicit investment of accumulated career-specific capital, and the same loss-aversion mechanism that keeps stock-investors holding losing stocks keeps managers holding losing career bets.
The breadth of the generalization is what distinguishes the disposition effect from the typical behavioral finding. Most behavioral findings in the failed-replication parts of this hub were demonstrated in one lab paradigm and then proposed as general theories of behavior, only to fail when tested in other settings. The disposition effect went in the opposite direction: it was first observed in real-world behavior, formalized as a theory with mechanistic content, then tested across many different real-world domains where the mechanism predicted it should generalize, and found in essentially all of them. This is what robust generalization looks like.
Counter-Mechanisms: How To Reduce The Bias
The disposition effect’s mechanism is not just diagnostic --- it points directly to interventions that reduce the bias. Three families of counter-mechanisms have empirical support:
Pre-commitment to exit criteria. The disposition effect is a real-time bias triggered by the gain-loss status at the moment of the sell decision. If you commit in advance to specific exit criteria --- “sell if the stock drops 15 percent from its purchase price” --- and you commit before knowing whether the stock will drop, the real-time loss-aversion mechanism gets bypassed. The decision has already been made; the only question at the moment of the sell is whether the pre-specified condition has been triggered. Pre-commitment is psychologically equivalent to using rule-based rather than discretionary decision-making, and the empirical evidence on stop-loss rules in retail investing is broadly consistent with this prediction: investors who use stop-losses exhibit smaller disposition effects than those who do not. The same logic applies to project termination criteria and product-retirement rules in organizational settings.
Separating gain and loss decisions. The disposition effect is generated in part by mental accounting --- treating each position as a separate account anchored to its purchase price. One counter-mechanism is to deliberately mix gain and loss decisions in the same decision frame, so that the investor or manager is making them as a portfolio rather than as individual accounts. The behavioral-finance literature suggests that “year-end portfolio rebalancing” frameworks --- where all positions are evaluated against a target allocation rather than against their individual purchase prices --- reduce disposition behavior by changing the mental-accounting structure of the decision. The same logic applies to corporate portfolio reviews where multiple projects are evaluated jointly rather than individually.
Periodic forced reviews. The disposition effect is partly a status-quo bias: the default action is to hold each position, and the investor only sells if some trigger overrides the default. One counter-mechanism is to make holding the active decision and selling the default --- to require, every six months or every quarter, an explicit affirmative decision to hold each position. The framing inversion turns the disposition-favoring asymmetry into a disposition-neutral one: the loss-averse instinct to avoid realizing losses gets matched against a loss-averse instinct to avoid actively re-committing to losers. The empirical evidence on this is thinner than on pre-commitment, but the mechanism is theoretically clean and the intervention is operationally feasible in many corporate contexts.
These counter-mechanisms do not eliminate the disposition effect --- they reduce it. The bias is deep enough in the psychological architecture that no intervention completely removes it. But the reductions are economically meaningful, and the existence of empirically supported counter-mechanisms is part of what makes the disposition effect a productive area of behavioral-finance research rather than just a documented anomaly.
The Strategist’s Takeaway
If you are evaluating a behavioral-finance or behavioral-economics claim --- a vendor pitching a “behavioral nudge” intervention, a consultant citing a “cognitive bias” that should reshape your portfolio strategy, an academic paper claiming a novel decision-making effect --- the disposition effect is your benchmark for what real evidence looks like. Specifically:
Robust replication across data sources and asset classes. The disposition effect has been documented in 10,000 retail brokerage accounts in the United States (Odean 1998), the universe of Finnish equity trading (Grinblatt and Keloharju 2001), US mutual fund managers (Frazzini 2006), Chicago futures-pit traders (Locke and Mann 2005), Boston housing-market sellers (Genesove and Mayer 2001), Chinese retail stock investors, Israeli professional traders, sports bettors, real estate investors, and cryptocurrency traders. If a behavioral claim has only been demonstrated once, in a single sample, with a single asset class, by a single research group, you are looking at the kind of finding that the replication crisis has repeatedly shown does not generalize. The disposition effect does not look like that. It looks like the opposite of that.
Theoretical generativity. The disposition effect did not just sit there as an isolated empirical curiosity. It motivated formal theoretical work connecting prospect theory’s value function to mental accounting to specific quantitative predictions about selling behavior, and the theoretical framework has been tested both with predictive content (Frazzini’s post-earnings-drift prediction was a structural consequence of the disposition effect, not an ad hoc patch) and with mechanism-level depth (Kaustia’s tests of the step-function shape of selling propensity). If a behavioral claim is just “people sometimes do this surprising thing in this specific context,” and there is no formal framework that explains why it happens or predicts when it should generalize, you are looking at folk psychology, not behavioral finance. The disposition effect is generative. It produced new theory, new predictions, and new experimental paradigms.
Practical implications with operational counter-mechanisms. The disposition effect’s mechanism points directly to specific counter-mechanisms --- pre-commitment to exit criteria, separating gain-loss decisions, periodic forced reviews --- that have been empirically tested and shown to reduce the bias. If a behavioral claim has no clear mechanism, no testable predictions about when it should generalize, and no implementable counter-mechanisms, you should be skeptical that it is anything more than a stylized observation. The disposition effect has all three, and that is part of what gives the literature its scientific traction.
The disposition effect’s calibration is the same as the Allais paradox’s, the prospect-theory framework’s, and the small handful of other anti-examples in this hub. When you see a behavioral claim, ask: how many times has this been replicated? In how many datasets? Using how many different asset classes? Is there a formal theoretical framework that organizes this result with others? Are there operational counter-mechanisms that have themselves been tested? If the answers are “many,” “many,” “many,” “yes,” and “yes,” you are looking at evidence that meets the disposition-effect standard. If the answers are “once,” “one,” “one,” “no,” and “no,” you are looking at the kind of finding that this hub has spent dozens of articles documenting as not having survived scrutiny.
The disposition effect is one of the cleanest cases in behavioral finance of a finding that genuinely cleared the bar. It is the standard against which other portfolio-decision claims should be measured.
Sources
- Shefrin, H., & Statman, M. (1985). “The disposition to sell winners too early and ride losers too long: Theory and evidence.” Journal of Finance, 40(3), 777—790. DOI: 10.1111/j.1540-6261.1985.tb05002.x.
- Odean, T. (1998). “Are investors reluctant to realize their losses?” Journal of Finance, 53(5), 1775—1798. DOI: 10.1111/0022-1082.00072.
- Frazzini, A. (2006). “The disposition effect and underreaction to news.” Journal of Finance, 61(4), 2017—2046. DOI: 10.1111/j.1540-6261.2006.00896.x.
- Barber, B. M., & Odean, T. (2013). “The behavior of individual investors.” In Handbook of the Economics of Finance, Vol. 2, 1533—1570. DOI: 10.1016/B978-0-44-459406-8.00022-6.
- Kaustia, M. (2010). “Prospect theory and the disposition effect.” Journal of Financial and Quantitative Analysis, 45(3), 791—812. DOI: 10.1017/S0022109010000244.
- Grinblatt, M., & Keloharju, M. (2001). “What makes investors trade?” Journal of Finance, 56(2), 589—616. DOI: 10.1111/0022-1082.00338.
- Locke, P. R., & Mann, S. C. (2005). “Professional trader discipline and trade disposition.” Journal of Financial Economics, 76(2), 401—444. DOI: 10.1016/j.jfineco.2004.01.004.
- Genesove, D., & Mayer, C. (2001). “Loss aversion and seller behavior: Evidence from the housing market.” Quarterly Journal of Economics, 116(4), 1233—1260. DOI: 10.1162/003355301753265561.
- Weber, M., & Camerer, C. F. (1998). “The disposition effect in securities trading: An experimental analysis.” Journal of Economic Behavior and Organization, 33(2), 167—184. DOI: 10.1016/S0167-2681(97)00089-9.
- Thaler, R. (1985). “Mental accounting and consumer choice.” Marketing Science, 4(3), 199—214. DOI: 10.1287/mksc.4.3.199.
Related
- Sunk Cost Fallacy --- The organizational-decision-making analog: the same loss-aversion mechanism that keeps investors holding losing stocks keeps project sponsors funding losing projects.
- Loss Aversion --- The reference-dependent valuation mechanism that does the core psychological work in generating the disposition effect.
- Prospect Theory: The Behavioral-Economics Framework That Actually Replicates (Anti-Example) --- The 1979 theoretical framework whose value function provides the mathematical structure for the disposition effect.
- Mental Accounting --- Thaler’s framework for the position-by-position evaluation that combines with loss aversion to generate the asymmetric selling pattern.
- Endowment Effect --- The related reference-dependence anomaly: ownership of a good raises the price at which one is willing to part with it, through the same loss-aversion mechanism.
FAQ
Is the disposition effect rational? Could investors have good reasons to hold losers and sell winners? The most careful empirical work has tested the leading rational explanations and found that they cannot account for the disposition effect’s magnitude. Tax considerations predict the opposite pattern (realize losses to take the deduction, defer gains to defer the tax bill). Information-based explanations (investors selling winners because they anticipate those stocks will underperform) are contradicted by Odean’s evidence that sold winners outperform held losers by about 3.4 percent in the subsequent year. Portfolio-rebalancing motivations might explain some sales but cannot explain the systematic asymmetry between PGR and PLR. The disposition effect appears to be a real behavioral bias that costs investors real money, not a rational strategy in disguise.
How big is the disposition effect in dollar terms? Does it actually cost investors much? Odean’s 1998 paper estimated that the underperformance of the sold-winners-versus-held-losers spread cost the 10,000 retail investors in his sample approximately 3.4 percent per year in foregone returns. Compounded over a multi-decade investing career, this represents an enormous reduction in terminal wealth. The disposition effect is not a small bias --- it is a major drag on retail-investor performance and one of the documented reasons that individual investors underperform passive index strategies on average.
Does the disposition effect predict that I should buy stocks that other investors are reluctant to sell? Frazzini 2006 documented that the disposition effect contributes to post-earnings-announcement drift, which means that, in principle, a sophisticated investor could exploit the disposition effect by trading against the disposition-driven underreaction. In practice, the magnitudes are modest and the transaction costs of implementing the strategy are non-trivial, so it is not necessarily a profitable trading strategy after costs. But the mechanism is real, and it has been incorporated into some quantitative-equity strategies that try to exploit behavioral biases in the broader market.
Why didn’t Shefrin and Statman win a Nobel Prize for this finding? The Nobel Prize tends to be awarded for theoretical frameworks rather than individual empirical anomalies, and Shefrin and Statman’s 1985 paper is best understood as one of the early applications of the broader prospect-theory framework that Kahneman and Tversky developed and that won Kahneman the Nobel in 2002. Shefrin and Statman’s contribution is theoretically derivative of prospect theory but empirically foundational, and it is widely recognized as one of the most important applications of prospect theory to real-world financial markets. The lack of a Nobel does not reflect on the importance of the finding --- it reflects the structure of how the Nobel committee recognizes behavioral finance.
Has anyone failed to replicate the disposition effect? No major attempt to replicate the disposition effect in a large dataset of investor trading behavior has failed to find the qualitative pattern. The magnitudes vary across samples --- professionals exhibit smaller effects than retail investors, sophisticated investors exhibit smaller effects than unsophisticated ones, and the effect is moderated by various individual characteristics --- but the basic finding that PGR exceeds PLR has held up in every major dataset where it has been tested. This is one of the most replicable findings in all of behavioral finance.
What is the relationship between the disposition effect and the sunk-cost fallacy? The two are mechanistically related but operationally different. The sunk-cost fallacy is the tendency to continue investing in a decision based on cumulative past investment that should be irrelevant to the forward-looking decision. The disposition effect is the tendency to hold losing positions and sell winning positions based on a reference-dependent gain-loss framing. Both are generated by loss aversion plus reference-dependence, but the disposition effect operates over realized-versus-unrealized gain-loss status while the sunk-cost fallacy operates over total accumulated investment. In investment-portfolio settings, the two often produce overlapping behaviors; in project-management settings, the sunk-cost fallacy is usually the more directly applicable framework.