When something is disagreeable, we try, perversions excluded, to avoid it. And where this is not possible, to at least limit the damage. Losses are rarely thought to be unavoidable events in the life of the investor. Because they often imply that an earlier decision was badly thought out, losses are disagreeable and difficult to accept. The temptation for the investor who has been disappointed by the performance of an investment is to finish on a high note by selling the security at a good price. Again, this is a strategy with a two-fold implication. If he does sell, the final loss is limited and the investor can at least be satisfied by the fact of selling well, rather than of having bought well. If he does not sell, the investor pays dearly for his sins of greediness and finds himself with a security on his hands which he could have already sold and reinvested in a better product.
In the case of real estate assets, an interesting study is provided by Genesove and Mayer (2001)1 who analyzed the behavior of sellers of apartments in Boston between 1990 and 1997. The apartment market had undergone an uneven progression in the 1980s and 1990s with a leap in nominal prices of 170 percent between 1982 and 1989, a drop of 40 percent between 1989 and 1992, and finally, a rebound of 80 percent between 1992 and 1998. This latter rise surpassed the previous peak, reached in 1989 (see Figure 5.1).
These price movements were accompanied by large changes in sales figures and in the asking prices of the sellers. At the minimum in 1992, the asking price for new listings exceeded the market price on average by 35 percent and only 30 percent of them were sold in less than 180 days. In 1997, the asking price for new listings exceeded the market price by 12 percent and more than 60 percent of the properties found a buyer in less than 180 days. The overvaluation of new listings in 1992 is even more surprising, given that the amount of available housing was much greater in 1992 than in 1997 and sellers were not in a position of strength. Genesove and Mayer hypothesized that sellers would suffer from loss aversion at the bottom of the cycle which would lead them to ask for a price much higher than the market price in order to lessen the pain caused by the loss. In other words, at the bottom of the cycle, many sellers have an original purchase price higher than the market price and so are headed for selling at a loss. At the top of the cycle the proportion is reversed such that the average seller has less incentive to ask an exaggerated price. The statistical regressions performed by the two investigators validate this hypothesis. It turns out that sellers whose property is evaluated by the market at less than their original purchase price have a tendency to ask a higher price than sellers not posting a loss. The “premium” asked by losing sellers is of the order of 25 percent to 30 percent of the difference between the market price and their purchase price. For them that decision translates into a smaller chance of selling the property, into a longer time on the market, and, in a more positive vein, into a higher selling price. The gain would be from 3 percent to 18 percent of the difference between the purchase price and the market price. The stubbornness of the losers would finally be rewarded.
Figure 5.1 Price of real estate in Boston (1982–1998)
Source: Genesove and Mayer (2001).
Another interesting point in the study by Genesove and Mayer is that it shows aversion to nominal losses and not to real losses (confirming the results on money illusion in experiment #10). As well, loss aversion would be twice as strong for owners who live in their apartments compared to those who rent them out. They would be then more reluctant to lose money on their residence than on simply a real estate investment. A matter of sentiment or just of calculation? If the owners who rented out their apartments add the rents received to the price of the sale, then their losses decrease and, with that, their incentive to sell at too high a price.
Another tactic for investors who refuse to lose money on investments is to keep a losing stock until it comes back to equilibrium. They, on the other hand, do not have the same desire to keep their winning securities. On the contrary, it is quite tempting to sell a winning stock in order to realize the gain and congratulate oneself on a good decision. These two asymmetrical ways of behaving result from what researchers call the “disposition effect,” the tendency to keep losing securities longer than winning ones. This behavioral characteristic has been documented by a multitude of experimental studies regarding both individual and professional investors.
Weber and Camerer (1998)2 chased down the disposition effect in a laboratory experiment on economics students at the University of Kiel (Germany) and engineering students at the Aachen University (Germany). They set these students to work on an experimental market over 18 periods on which were listed six stocks (A, B, C, D, E, F). In each period the price of each stock could increase or decrease according to probabilities set in advance and independent of past changes in price. The distribution laws were:
The students were aware of the six probability distributions without knowing to which stocks (A, B, … or F) they applied. To find out, the students had to address the question while noting the past changes in price. For this purpose, at the beginning of the experiment the students were given the changes over four periods numbered –3 to 0. Then, at the beginning of each period running from 1 to 14, they had to state if they were buying or selling the six stocks, knowing that they were forbidden to sell capriciously. The initial sum granted to them for investment was DM10,000. On completion of the experiment the students left with a tiny fraction of the value of their portfolio at the end of period 14. Weber and Camerer found three indicators which argued for a disposition effect in the experimental subjects. First, 60 percent of the shares sold were of stocks that had gone up compared to 40 percent which were of stocks losing value. Second, the stocks sold during the game showed an average gain of DM 1 higher than the average gain of the stocks present in the portfolio at the end of the game (DM 0.4 or 0.7, depending on the calculation method used). Third, among the securities sold, two-thirds had an increase in their price in the period just before the sale. Taken as a whole, the three results suggest that individuals prefer to sell securities which have gained value to stocks which have lost value and prefer to sell when the price has just increased, rather than when it has just gone down. The experimental design is such that this behavior cannot be explained by a belief in a rapid reversion to the mean. Because the six stocks are differentiated by the various distributions (from − − to + +), the past increases in prices should be factors more encouraging for buying than for selling. Moreover, it is interesting to know that the experiment, re-performed with a single difference in methodology—the stocks were automatically sold at the end of each session—resulted in a considerably diminished disposition effect, whichever experimental indicator was used. It is above all the reluctance to sell a position “in the red” that is responsible for the longer retention in the portfolio of losing stocks compared with winning stocks.
The disposition effect is not without consequences for the performance of portfolios. This behavior is particularly costly to the investor. Odean (1998)3 found from a study of 6,380 investment accounts of individuals over the period 1987 to 1993 that investors sell half again more easily stocks which are gaining rather than those losing (except in December, for tax reasons). He also has shown that the winning stocks sold outperformed the market on average by 2.3 percent in the following year, while the losing stocks which were kept had underperformed the market by 1.1 percent. Thus, if the investors studied had kept the securities sold and sold the securities kept, they would have increased their annual performance by 3.4 percent! Other studies confirm that it is the investors who are the least subject to the disposition effect who perform the best.
Individuals evaluate their income according to a value-function represented by an S-curve. Above a reference point, which is often “0,” the value function is increasing but with a smaller and smaller slope, and with the same behavior occurring below the reference point. What is considered a gain by an individual offers him an always positive utility, but which is less than proportional to the gain. In the same fashion, losses always diminish wealth, but by less and less. Thus, for the investor a gain of $2,000 on a security brings him less than twice the pleasure of a gain of $1,000. Conversely, he prefers a loss of $2,000 to two consecutive losses of $1,000.
A consequence of the particular form of the value-function is that it encourages what researchers call “hedonic framing.” If he is looking to optimize the pleasure that investment in the markets gives him, then an investor must integrate his losses and segregate gains. Lumped together, losses are less painful, while gains are more pleasant when they are separated. The hedonistic investor uses ad hoc mental accounting to feel the best he can. But this mental exercise is possible only if the timing of the transactions allows the realization of the required associations and dissociations. It is difficult to consider two transactions separated by two months as one. On the other hand, if they occur during the same day, the investor can integrate them easily in his mind and then be concerned only by the overall performance.
Lim (2004)4 tested this hypothesis by observing the timing of the disposal of securities of 78,000 American households, clients of a brokerage house, from January 1991 to November 1996. If temporal separation facilitates mental dissociation, while temporal proximity promotes integration, then investors should show a tendency to sell their losing securities in short time intervals and conversely to space out the selling of securities which are winning. This behavior is what Lim found. Investors are more inclined to sell losing stocks in blocks than to sell winning stocks in that manner (see Figure 5.2). The proportion of multiple sales (on the same day) compared to total sales is lower by 20 percent for winning stocks; these constitute 8.46 percent of the sales versus 10.44 percent of the sales of losing stocks. Although significant throughout the year, the effect is especially noticeable in December when, for income tax reasons, the block sales of losing securities are almost twice as frequent as the block sales of earning securities.
By selling losing investments on the same day, the investor avoids a succession of disagreeable moments. This hedonic framing seems to have been adopted by the directors of companies because it is apparent that they use all their maneuvering room in accountancy matters to smooth out their profits from one period to the next and to downplay as much as possible their deficit periods. This latter strategy is commonly called the “big bath.” It avoids a series of losing or downgrading periods and preserves as much as possible the image of the company in the eyes of the investors.
Figure 5.2 Proportion of multiple sales
Source: Lim (2004).
Since he has put much time, energy, money, and so on, into a losing investment, the investor has all the more difficulty in disposing of it. The more he has solidified his position, the more he has demonstrated his patience, the more he has attentively followed the changes in price on the exchange, then the more the selling of the asset is a heartbreaker. Rather than realizing the loss of capital and recognizing the futility of his efforts, the investor is persuaded that it is important to “go right to the end,” that is, to keep the security until its value has returned to the break-even point, even to reinvest. Because the French and British persevered in financing the project despite its evident economic failure, American authors sometimes refer to this behavioral bias as the “Concorde effect.” In its most serious form, this effort brings an escalation of commitment in the investment. For an investor in the stock exchange one manifestation would be buying more of a losing security in the portfolio in order to reduce its average purchase price and to make the return to equilibrium easier.
Odean, Strahilevitz, and Barber (2005)5 established such behavior through the study of 66,465 investors between 1991 and 1996 and 665,533 investors between 1997 and 1999. In addition to the periods, the two samples analyzed were also distinguished by the size of the portfolios. The average portfolio ($107,000) in the second sample was twice as large as in the first sample ($47,000) and more diversified as well. The authors were able to calculate the proportion of securities held in the portfolio which were repurchased over a period of one year. They were able to classify their calculations by the performances realized by the securities, that is, to separate the stocks that gained from those which lost. Their calculations were made on a daily basis and summed to give an annual average. The results show that investors increased holdings of the losing stocks more than of the winning ones. For winning stocks the probability that the investor reinforced his position over an interval of 12 months was 9.4 percent on average, while for the losing stocks, it was 14.6 percent in the first sample and 12.8 percent in the second. The differences between the proportions for the earning and losing securities are statistically significant for both samples (see Figure 5.3).
Figure 5.3 Proportion of stocks already held in portfolios repurchased during 12 months
The repurchasing of losing stocks to lower their average purchase price does indeed lead to a more rapid return to equilibrium. The flipside is that the position on losing securities increases in step as the investor proceeds to “average down,” thereby considerably increasing the risk in the portfolio. Such a procedure is even more absurd when it leads to the buying back of securities for which one is no longer convinced of their potential for gain. It is certainly not by watering the weeds that one gets a beautiful garden.
Realized performances have an effect on risk-taking inverse to that of current performances. Large losses that have taken place in the past are ingrained in the memory of the investor and can turn him firmly away from risky assets with which they are associated; this is the “snakebite” effect. In the same way that it is difficult to get behind the wheel again after a highway accident, it is a delicate situation for an investor to go back to a security which has bought him losses in the past. Everything being equal, of course. For the “traumatized” investor, the security in question is often completely off-limits. The manner in which other investments are approached can also be affected. Possible future losses raise their heads and are more feared at the time of an investment. The investor can then react by selecting less risky investments with limited potential for decline.
Odean, Strahilevitz, and Barber (2005)6 continued their study of the impact of past performance on choice of investments by observing whether investors showed a greater ease in buying back securities for which they had posted a gain in the past compared to securities which had resulted in losses. To do so, they calculated the percentage of securities sold by investors which had been repurchased over 12 months. They sorted the transactions according to whether the previous sales had resulted in gains or losses (see Figure 5.4).
Figure 5.4 Proportion of securities sold which are bought back over 12 months
The results show that, for the two samples (1991–1996 and 1997–1999), the securities sold with a gain are bought back twice as often as are securities which brought losses.
The literature on buying behavior with regard to securities which were sold in the past is still too sparse to be certain that the old adage “once bitten, twice shy” also applies to investors. In a laboratory experiment Weber and Welfens (2005)7, for example, do not find that behavior is different when a security has previously brought gains than when it resulted in losses. The experiment is nevertheless very unusual because the participants in it confront a single negotiable security. If there had been two, perhaps they would have turned away from the stock which had occasioned losses in order to position themselves on the other.
Loss aversion is only a special case of a broader behavioral bias: regret aversion. The prospect theory, developed by Kahneman and Tversky (1979),8 which merited the Nobel Prize in Economics in 2002, points to a greater sensitivity to the value-function below the reference point of the individual. Geometrically speaking, the curvature of the value-function shows a larger slope for losses than for gains. A loss compared to the reference point is thereby perceived on average 2.25 times more negatively than a gain of the same amount is perceived positively. In this theory the reference point is not specified. If this reference point is the price, then the individual clearly demonstrates a loss aversion. But the general case is that below the reference point the individual shows regret. Relative to the purchase price, it is the regret at having chosen a losing investment; relative to the market return, it is at having chosen an underperforming investment; relative to the highest price during a certain period, it is at not having sold at the right time, and so on. Several studies prove indirectly that it is not automatically the purchase price which is used as the reference point by the investor. It is this changing point that determines whether it is an occasion for regretting or rejoicing and therefore keeping the asset rather than selling it.
Grinblatt and Keloharju (2000)9 show in a statistical study on the behavior of investors in the Finnish market that individual investors have difficulty selling securities which are losing relative to the market and that the difficulty is even greater when the underperformance is more recent. After a month, the impact on these investors of low return compared to the market is no longer significant. The results suggest that it is not so much the total performance which is scrutinized as the market scenario. An investor is often reluctant to give up a winning stock (relative to the market) if it has recently underperformed. Thus, to compensate for the reluctance to sell caused by a loss on day one, it would be necessary to have achieved gains five times higher in the previous month. Conversely, the more recently a security has outperformed the market, the more it has a chance of being sold. The parameters found by the authors suggest that a recent outperformance can easily compensate in the mind of the investor for an even more serious loss. For example, for individual investors, an outperformance on day one urges them to sell as much as an underperformance ten times as large recorded in the previous month dissuades them from doing so.
The investor’s level of satisfaction depends not so much on his performance, even if very recent, as on the image it sends to him of himself. A very illuminating study of this distinction is provided by Nofsinger (2001).10 Specifically, the author analyzed the trades made between November 1990 and January 1991 by individual investors in 144 companies listed on the New York Stock Exchange. Nofsinger paid special attention to investors’ reaction to information on the company and to macroeconomic information. In accordance with the disposition effect, good news about a company increases its price on the Exchange and causes investors to sell its shares. Similarly, news about a company which makes its price go down encourages retention of the stock. Nofsinger finds, on the other hand, that general hikes in price following good macroeconomic indicators do not (or only slightly) encourage selling. This result implies that investors sell earning stocks especially if they can take personal pride in themselves for the realized performance. Conversely, a loss is less badly received when it is the whole market which is going down because then it does not elicit regret at not having invested as well as others. The investor can more easily sell when the loss does not impugn him personally.
It is more the performance relative to the market than the absolute performance which is decisive for ordering a sale. That means, if loss aversion is almost generally present among investors, that the reference point used also varies widely from investor to investor. The multiplicity of reference points (purchase price, gains, market return, and others) argues for a large natural heterogeneity among individuals or for a significant instability in the cognitive processes of decision making. It also encourages one to wonder whether this instability would be permanent or ad hoc. Does the investor voluntarily change his reference point in such a way to maximize the satisfaction he derives from his investments? In other words, is the investor of good or bad faith? Little debated in the literature, this question is far from being resolved.
Paralysis in the face of a decision results in what psychologists call status quo bias. It is the tendency that a large fraction of us have to leave things as they are. It seems that we consider the maintaining of the status quo as bringing less involvement than any decision urging change. Any other decision would run the risk of error and loss. But, is deciding not to choose not already a choice? Is leaving things unchanged not running the risk of failing to gain through other options? It is surely in this term “failing to gain” where all the subtlety lies. It appears that in a great number of situations we take the status quo as a legitimate reference. All departure from the status quo leads to a loss or a real gain. Conversely, when the status quo is chosen, alternative performances are perceived as being simply theoretical.
Samuelson and Zeckhauser (1988)11 measured this bias with an experiment on the behavior of investors after they received an inheritance. The subjects were students majoring in economics and in finance. The first group was put in the following situation:
“You are an avid reader of financial journals but up to recently you have not had much money to invest. You just learned that a great-uncle left you a considerable sum of money. Your thinking about the way to invest these funds leads you to consider four financial instruments.
From these four investments, which one would you finally choose?”
Among the students, 32 percent chose the shares XYZ, 32 percent the municipal bonds, 18 percent the risky shares, and 18 percent the government bonds. Another group of students was put in a situation identical in all respects but one. A bequest of the same value was this time already mostly invested in one of the four financial instruments. The results show that this initial allocation greatly influenced their choice because the most selected product was systematically that in which the inheritance was already largely invested. These individuals display a reticence to change the allocation of the bequest even if it did not conform to their financial profile or interests. Nothing says that the great-uncle would have supported this choice, or this lack of choice.
The status quo bias is even more obvious when the investor has chosen the initial investment himself. Changing leads to judging himself badly and admitting to a poor decision. Choosing the status quo avoids experiencing a disappointment. The status quo bias encourages retaining securities even when they present a poor potential. It explains also why money leaves underperforming funds less rapidly than it goes into overperforming funds (Goetzmann and Peles, 1997).12
The status quo bias leads to another concept which is prominent on the Exchange: the endowment effect. The endowment effect explains the difference, for the same product, between the minimum price at which we are prepared to sell and the maximum price that we are prepared to pay to acquire it. By economic theory this difference has to be zero because, to the rational individual, every product has one price and one price only: that for the well-being it brings. In reality though, this difference is positive and suggests that we assign higher values to our possessions than their objective values. The endowment effect has been proven by experiments which create fictional markets with buyers and sellers.
One of the first experiments in this area took place in the late 1980s. It was conducted by Richard Thaler13 on students at Cornell University to whom he randomly offered, free of charge, coffee mugs stamped with the crest of the university which sold for $6 in the campus bookstore. This extraordinarily simple experiment showed that students who received the gift declared themselves ready to part with it for an average price of $5.25, while students who had not benefited from the giveaway agreed to buy a mug at an average maximum price of $2.75! Almost a 100 percent difference! Since the experiment was performed on a large number of students, it is difficult to claim that the price difference is explained, not by an endowment effect, but by a greater passion of the students in the first group for coffee or for their university. Kahneman, Knetsch, and Thaler (1990)14 subsequently tested the robustness of the conclusion by repeating the experiment with different assets and different instructions. The endowment effect turned out to be significant in all cases. They were also able to point out that the sellers’ prices lay farther from the objective price than those of the buyers. This objective price was set by subjects who were involved in neither selling nor buying the asset. The endowment effect operated therefore more on the sellers than on the buyers. There is greater reticence to part with whatever object than with money. This is the general rule. Scrooge is the exception to it.
In matters of savings, you can be aware that you have some “bad” investments in your portfolio that you would not choose if you did not already have them and, yet, be reluctant to part with them. For example, it would be quite possible for an investor to have in his portfolio some securities X, which are listed at $90, for which he thinks the fair valuation is $85, but which he will refuse to sell at less than $95! Such behavior is not necessarily found only in the most eccentric of investors. Empirical results suggest that the endowment effect is common among beginners and among market professionals but it can be nonetheless reduced by training and experience.
It is important to point out that some authors, such as Nofsinger (2001),15 argue that investors are frequently emotionally attached to their stocks; such an attitude adds to the endowment effect and status quo bias. If this attachment offers the advantage of dissuading investors from buying and selling too frequently, it can also have negative effects. When investors are attached to the securities of a company, they show an optimistic bias in the evaluation of information regarding it and consequently are inclined to hold the securities too long. The resulting inertia adds to the inertia engendered by the endowment effect. Attachment can focus on the security. Such is the case where the investor has held it in the portfolio for a long time and considers it a “travelling companion” or when it has given him some nice performance in the past. The investor can also be attached to the company itself if he works there or is retired from it.
Endnotes
1 Genesove, D., and C. Mayer, “Loss Aversion and Seller Behavior: Evidence from the Housing Market,” Quarterly Journal of Economics, November, (2001): 1,233–1,260.
2 Weber, M., and C. Camerer, “The Disposition Effect in Securities Trading: An Experimental Analysis,” Journal of Economic Behavior and Organization, 33, (1998): 167–184.
3 Odean, T., “Are Investors Reluctant to Realize their Losses?” Journal of Finance, 53, (1998): 1,775–1,798.
4 Lim, S., “Do Investors Integrate Losses and Segregate Gains? Mental Accounting and Investor Trading Decisions” (European Finance Association 2003 Annual Conference Paper, 2004); and Burgstahler, D., and I. Dichev, “Earnings Management to Avoid Earnings Decreases and Losses,” Journal of Accounting and Economics, 24, (1997): 99–126.
5 Odean, T., M. Strahilevitz, and B. Barber, “Once Burned, Twice Shy: How Naïve Learning and Counterfactuals Affect the Repurchase of Stocks Previously Sold” (working paper, 2005); and Staw, B., “The Escalation of Commitment: An Update and Appraisal,” in Organizational Decision Making, ed. Z. Shaphiro (Cambridge: Cambridge University Press, 1997) 191–215.
6 Odean, T., M. Strahilevitz, and B. Barber, “Once Burned, Twice Shy: How Naïve Learning and Counterfactuals Affect the Repurchase of Stocks Previously Sold” (working paper, 2005).
7 Weber, M., and F. Welfens, “The Repurchase Behavior of Individual Investors: An Experimental Investigation” (working paper, University of Mannheim, 2005).
8 Kahneman, D., and A. Tversky, “Prospect Theory: An Analysis of Decisions under Risk,” Econometrica, 47, (1979):313–327.
9 Grinblatt, M., and M. Keloharju, “The Investment Behavior and Performance of Various Investor Types: A Study of Finland’s Unique Data Set,” Journal of Financial Economics, 55, (2000): 43–67.
10 Nofsinger, J.R., “The Impact of Public Information on Investors,” Journal of Banking and Finance, 25, (2001): 1,139–1,366.
11 Samuelson, W., and R. Zeckhauser, “Status Quo Bias in Decision Making,” Journal of Risk and Uncertainty, 1, (1988): 7–59.
12 Goetzmann, W., and N. Peles, “Cognitive Dissonance and Mutual Fund Investors,” Journal of Financial Research, 20, (1997): 145–158.
13 Kahneman, D., J.L. Knetsch, and R. Thaler, “Experimental Tests of the Endowment Effect and the Coase Theorem,” Journal of Political Economy, 98 (6), (1990): 1,325–1,348.
14 Ibid.
15 Nofsinger, J.R., The Psychology of Investing (Upper Saddle River: Pearson Prentice Hall, 2001).