Loss aversion

An investor holds a global tracker fund in her SIPP. Markets fall 8% over a few weeks, and she opens the app every morning to see the damage. The fall is within the range any equity investor should expect, and her plan assumed it. Yet each red number registers more sharply than the green ones did during the previous year’s rise.

By the end of the month she has moved a chunk of the portfolio into cash. Nothing about her goals or her time horizon has changed. What changed was the way the losses felt.

Loss aversion is the tendency for a loss to weigh more heavily in a decision than a gain of the same size. It sits at the centre of prospect theory, and it affects how investors hold, sell and size their risk.

What the research shows

Kahneman and Tversky (1979) built prospect theory as a critique of expected utility theory. They showed that choices under risk display several systematic patterns that the standard theory cannot accommodate. People judge outcomes as gains or losses relative to a reference point, not as final levels of wealth, and the two sides are not treated symmetrically.

Tversky and Kahneman (1992) put numbers on this. They argued that losses loom larger than gains, and that the asymmetry was too large to be explained by income effects or by ordinary risk aversion. Their median estimate of the loss aversion coefficient was 2.25, meaning a loss was weighted a little over twice as heavily as an equal gain.

That figure has been repeated widely, so it is worth checking how it has held up. Brown, Imai, Vieider and Camerer (2024) examined 607 empirical estimates of loss aversion from 150 articles. They put the mean coefficient at 1.955, with a 95 percent probability that the true value falls between 1.820 and 2.102. That is somewhat below 2.25 but clearly above 1. They also found that few study characteristics were substantially correlated with differences in the estimates.

The finding is contested. Gal and Rucker (2018) reviewed the literature and argued that it does not show losses to be systematically more powerful than gains. In their reading, the effect depends heavily on context. Mrkva, Johnson, Gächter and Herrmann (2020) replied with five samples totalling 17,720 participants. They found loss aversion in all of them, but moderated: it was lower among people with more domain knowledge and experience, although people at every level were still loss averse.

Yechiam and Zeif (2025) re-analysed the Brown et al. dataset, adding moderators drawn from the debate. They found no loss aversion in studies with unordered, symmetric gains and losses, and found it only where losses were smaller than gains or payoffs were presented in order. For the symmetric, unordered studies, they put the parameter at about 1.07, not significantly above 1. They conclude that loss aversion is not a robust phenomenon. They also note that stakes in the reviewed studies were typically low, so the meta-analysis may not capture loss aversion at high stakes. This is one research group’s reanalysis, and Brown et al. and Mrkva et al. reach more favourable conclusions from similar material.

A fair reading is that loss aversion appears in many designs, that the average estimate is close to 2, and that its size varies with task design, framing and the participant. Whether it is a stable trait or a product of how choices are presented is still argued over.

Evidence from investor behaviour

Odean (1998) analysed trading records for 10,000 accounts at a large US discount brokerage. Investors showed a strong preference for realising winners over losers, and neither rebalancing, trading costs nor subsequent performance explained it. In taxable accounts, the pattern was suboptimal and led to lower after-tax returns.

The pattern is not confined to US retail investors. Barber, Lee, Liu and Odean (2007), using the trades of nearly four million Taiwanese investors, found that investors were about twice as likely to sell a stock held at a gain as one held at a loss. Mutual funds and foreign investors, which together made up under 5% of trades by value, did not show the reluctance.

Loss aversion is only one candidate explanation for this “disposition effect”. Shefrin and Statman (1985), who named it, proposed four ingredients: loss aversion, mental accounting, regret avoidance and lack of self-control. The behaviour is well documented; its cause is less settled.

Benartzi and Thaler (1995) applied the idea to the equity premium. They argued that investors who are loss averse and who also evaluate their portfolios frequently would demand a large premium for holding shares. They called this combination myopic loss aversion. Their simulations suggested the size of the premium is consistent with earlier prospect theory parameters if investors evaluate their portfolios annually. The result has been challenged. Durand, Lloyd and Tee (2004) modified the methodology slightly and found the analysis was not robust.

Thaler, Tversky, Kahneman and Schwartz (1997) tested the mechanism in an experiment. Participants who received the most frequent feedback allocated less to the riskier asset than those who saw results less often. Haigh and List (2005) asked whether professionals escape this. They found that traders recruited from the Chicago Board of Trade showed behaviour consistent with myopic loss aversion to a greater extent than students.

How it shows up in portfolios

Holding losers, selling winners

A position that has fallen below the purchase price becomes a realised loss only when sold. Many investors therefore keep it, because selling would make the loss definite. Winners get sold early to lock in the gain. Odean’s results are the standard evidence for this pattern, though loss aversion is not the only candidate explanation.

Checking too often

Equity returns are volatile over short periods, so a daily or weekly view will contain many down periods. Frequent checking exposes the investor to more small losses. This is the “myopic” part of myopic loss aversion. In the experiments above, participants who saw results more often took less risk.

Holding too little in equities

If short-term dips weigh heavily, risky assets can look less attractive than their long-run record might justify. Lee and Veld-Merkoulova (2016), studying private investors, found that higher myopic loss aversion was associated with a lower share of total assets in stocks, with the effect clearest among investors who both evaluated frequently and traded regularly. Their study is observational and relies on survey responses, so it shows an association rather than a cause.

Selling after a fall

Large drops can turn paper losses into a wish to stop them. Moving to cash after a fall relieves the discomfort immediately. It also creates a second decision, about when to return, which a loss-averse investor is poorly placed to make.

Passing on favourable bets

Many people turn down a gamble with a positive expected value because the possible loss looks larger than the matching gain. For investors this can appear as reluctance to add to a position, or to rebalance into whichever asset has just fallen.

Why it happens

Part of the mechanism lies in how value is represented. In prospect theory, outcomes are measured against a reference point, and the value function is steeper below that point than above it. The reference point is often the purchase price, the last valuation or a recent peak, and it moves with circumstances. That is one reason measured loss aversion varies across tasks.

There is also an emotional component. Realising a loss can feel like admitting an error, and Shefrin and Statman’s inclusion of regret avoidance reflects that idea. Holding on keeps open the possibility of vindication.

An evolutionary story is often offered. For ancestors living close to subsistence, a serious loss could be fatal and a windfall merely helpful, so weighting threats more heavily may have paid. It is plausible, but it is a hypothesis about origins, and the evidence above does not depend on it.

Finally, Mrkva et al. found that loss aversion falls with knowledge and experience. One reading is that people with less familiarity construct their preferences on the spot, which leaves more room for framing to push judgement. That is an interpretation, not a tested finding.

Ways to counter it

None of these techniques removes the bias. They change the conditions under which it operates. Whether any of them suits a given person is a personal question.

• Written rules set in advance. A short investment policy statement covering target allocation, rebalancing triggers and the conditions for selling can be drafted while calm. Its value is that the plan exists before the loss does.
• Less frequent evaluation. The experimental evidence links less frequent feedback to greater risk acceptance. Some investors set a fixed review date, such as quarterly or annually, and avoid daily price views in between.
• Assessing the portfolio as a whole. Narrow framing makes each holding look like its own gain or loss. Looking at total portfolio value and its progress toward a long-term goal widens the frame.
• A forward-looking test. This asks whether an investor would buy the position today, at today’s price, if they held cash instead. It strips out the purchase price as a reference point.
• Separating the tax question from the psychological one. Tax rules differ by account and jurisdiction. Where a rule applies, it can be worked out as its own calculation, so the purchase price is not doing double duty as an anchor and a tax input. Odean’s finding of lower after-tax returns relates to taxable accounts.
• A decision journal. Recording what was expected, and why, before a fall makes it harder to reconstruct the reasoning afterwards.

Related biases

• Disposition effect: the pattern of selling winners and holding losers, for which loss aversion is one candidate explanation among several.
• Myopic loss aversion: loss aversion combined with frequent evaluation, the version most relevant to risk-taking inside a portfolio.
• Mental accounting: treating each holding as its own account, which makes individual losses easier to see and harder to accept.
• Endowment effect: valuing what you already own above its market price, often discussed alongside loss aversion in riskless choice.