The disposition effect is diminishing in time

(1)

The Disposition Effect is diminishing in Time

Author:

G. Assorgia

UNIVERSITY OF AMSTERDAM

FACULTY OF ECONMICS & BUSINESS

BSc Economics & Business

(2)

Statement of Originality

This document is written by Student Gianluca Assorgia who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.

The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

ABSTRACT

It is very probable that the disposition effect is diminishing. A sample of 30 stocks listed on the S&P 500 index is used to find evidence of that. This is done via measuring the disposition effect in three

consecutive periods. As expected the first period shows most evidence of the disposition effect, the second period less than the first and the last period shows no evidence.

Keywords:

(4)

LIST OF TABLES

(6)

LIST OF FIGURES

(7)

CHAPTER 1 Introduction

Market anomalies are deviations from the efficient market hypotheses, which state that individuals act rational. Market anomalies are often explained by irrational behaviour, that is a consequence of psychological factors. It could be that investors, that are made aware of these psychological

shortcomings, adapt their behaviour to come to a more desirable outcome. But this is not always the case. Investors may recognize the problem but still could have a lack of self-control (Shefrin & Statman, 1985, p. 782).

One of the most researched market anomalies is the disposition effect. According to the disposition effect, investors sell winning stocks to early and hold losing stocks for too long (Shefrin & Statman, 1985, pp. 777-778). This is in contrast with the efficient market hypothesis, which supposes that individuals act rational. Investors quickly sell the stock if they have made a small gain, but hold losing stock longer hoping it returns to its original value. If an investor suffered a loss or not should not influence his next investment decision, but it does according to the disposition effect.

The disposition effect is proven multiple times. By analyzing 10,000 at a large brokerage house (Odean, 1998, p. 1775), doing experiments (Weber & Camerer, 1998, p.181) or analysing market data (Ferris, Haugen and Makhija ,1988, pp. 686-693). Since the disposition effect is such a well-documented market anomaly, it is likely that there is some form of adaptation to it, especially by large institutional investors. That suggests that the disposition effect is decreasing in time, as more and more investors adapt to it. But if and when this has taken place remains unanswered.

In the next chapters I try to get closer to that answer. I will make estimates of the disposition effect in various time periods, to conclude if there is less evidence of the disposition effect in later time periods. Therefore, I will use a methodology similar to that of Ferris et al. (1988, pp. 681-683). I will do this by making an analysis of various stocks from the same market. Subsequently, I will test if the hypothesis that the disposition effect exists in the time-frame. If the disposition effect is

diminishing, the later time periods will have no strong evidence of it.

In chapter 2 I will make a review of the existing literature about the disposition effect. Also, explain the tax-loss selling hypothesis and how it influences the disposition effect. Chapter 3 will show the exact methodology and hypothesis. Subsequently, there will be explained how the hypothesis will be tested. Chapter 4 explains which data is used, were it was obtained and at which time span it has. In chapter 5 the results are shown and explained. In the last chapter, the results are drawn and recommendations for future research are made.

(8)

CHAPTER 2 Literature

2.1 Prospect theory

The disposition effect is originated from prospect theory developed by Kahneman and Tversky in 1979. Prospect theory is a model that describes decision making under risk (Kahneman & Tversky, 1979, p. 263). From prospect theory can be derived if an investor makes losses, it is more likely that

he will take more risk to earn back his losses, risk that he would not accept if he had no losses (Kahneman & Tversky, 1979, p. 287). In the graph below shows the utility curve under prospect theory, which deviates from the expected utility theory. The point in which there are no perceived gains or losses is called the reference point. In gains the utility function is concave, but in losses it is convex. This means if there is a discrepancy between the reference point and the actual asset position, it will influence the magnitude of gains and losses that the investor experiences. For example, an investor who starts below his reference point and receives a gain, gets a higher increase in utility than if he had started at his reference point. This violated the axiom of the expected utility theory, that individuals act rational (Shefrin & Statman, 1985, p. 777)

However, research by Hens and Flcek (2011, p. 154) concluded that prospect theory can not predict the disposition effect. “the model predicts that those investors who sell winning stocks too early and hold losing stocks toolong would not, in the first place, have invested in stocks” (Hens & Flcek, 2011, p. 154). The disposition behaviour can post be explained by their model, but not

ex-Figure 1: Utility function under prospect theory (Kahneman & Tversky, 1979, p. 279)

(9)

ante. This weakens the connection between the disposition effect and prospect theory, but the

disposition effect was discovered due to prospect theory and is in that sense vital for the development of the disposition effect.

2.2 Disposition effect

The term ‘disposition effect’ was used for the first time by Shefrin and Statman in 1985. They defined it as: “The tendency of investors to sell winners too early and ride losers too long” (Shefrin &

Statman, 1985, p. 778). What they meant by this is that investors hold financial assets that have dropped in value longer than financial assets that have risen in value, hoping that the asset regains its value. According to the efficient market hypothesis, it should not matter whether an asset had gone up or down for the length of the holding period.

Besides prospect theory Shefrin and Statman (1985, pp. 779-783) three other explanations for the disposition effect: mental accounting, seeking pride and avoiding regret and self-control. When a decision maker buys for example two stocks, a mental account is opened for each stock. Rather than seeing this as one investment, the decision maker sees it as two separate accounts. The prospect theory rules for decision making also applies here (Shefrin & Statman, 1985, p. 780). The reference points are set at the purchase price of the stock. It is more difficult for a decision maker to close a mental account at a loss rather than a gain (Shefrin & Statman, 1985, p. 780).

The third explanation is seeking pride and avoiding regret. According to this theory the feeling of pride encourages investors to quickly sell his winning stocks. The feeling of regret is not favoured by investors and occurs when a losing stock is sold. The feeling of regret could also be felt when a winning stock is sold to early and continues to increase in value, and therefore could discourage also selling winning stocks which is in contrast with the disposition effect (Shefrin & Statman, 1985, pp. 181-182).

At last, there is self-control. Investors acknowledge that they have a problem with small profit and large losses. If they have a profit leave the market quickly, but these small profits are often offset by a large loss. The investors see it as a problem that they do not realize there losses quickly enough, but they describe it a problem in self-control (Shefrin & Statman, 1985, p. 782).

2.3 Tax-loss selling hypothesis

(10)

preferable for investors to have an immediate tax deduction, then a postponed tax deduction (Dyl, 1977, p. 166). This makes that at the end of the year, it is optimal for investors to realize their losses but not their gains. This makes the abnormal trading volume for losing stocks higher than for winning stocks, the exact opposite as it would be suggested by the disposition effect.

it is proven that the tax-selling hypothesis still existed in 2006 (Starks, Yong and Zheng, 2006, p. 3049). Unlike the disposition effect, it is expected that the tax-selling hypothesis did not decrease in time, under the condition that the tax laws did not have changed. When examining the disposition effect the must be accounted for the Tax-loss selling hypothesis, otherwise it will influence the results.

(11)

CHAPTER 3 Methodology and Hypothesis

The methodology is largely based on the methodology used by Ferris et al. (1988, pp. 681-683). Most of the research on the disposition effect is done via experiments (Weber & Camerer, 1998, p.181) or broker data (Odean, 1998, p. 1775). The methodology of Ferris et al. (1988, pp. 681-683) allows to use data that can be obtained from most financial databases. The parameters that are needed: the price Pit, the trading volume and the total shares outstanding for an individual company. From these parameters the return (rit), turnover ratio (Vit) and market turnover ratio (Vmt) must be calculated.

(1) (2) (3)

There are two regressions needed to estimate the disposition effect. First, calculate the abnormal turnover ratio (εit). This is done by using a market-model regression of the turnover ratio (Vit) on the market turnover ratio (Vmt).

(4)

The abnormal return indicates how the actual turnover ratio deviates from the expected turnover ratio. The difference between these two can partially be explained by the disposition effect. To find evidence of that, a second regression must be used.

(5)

is a dummy variable and indicates if the stock is a winning stock or a losing stock. Its value is 1 if the > 0 and 0 otherwise. To account for the December effect, a second explaining variable is

(12)

b1 represents the disposition effect. it is expected that b1 is positive, because if the disposition effect is present b1 will have a positive effect on the abnormal turnover ratio. In other words, if the stock is a winning stock, more investors will sell the stock due to the disposition effect and therefore the abnormal turnover ratio would be positive. This results in the hypothesis: H0: b1 > 0. This hypothesis will be one-sided tested against a confidence level of 90% and a confidence level of 95%.

(13)

CHAPTER 4 Data

The data from stocks that are listed at the S&P 500 index is used. The advantage from using the stocks from the S&P 500 index is that there is a large liquidity and a high turnover ratio. Also, it is expected that if the hypothesis of a diminishing disposition effect is true, it will be visible in this market, due to the institutional investors that are active in the market.

Not all stocks at the S&P 500 index are used. They need to be listed at the index for a large period to see if the disposition effect is diminishing. The period 1990-2016 is chosen, because the disposition effect was persistent in 1998, even with institutional investors (Odean, 1998, p. 1797). The whole period is than split in three 9-year subperiods. The data from 1998-2007 and 2007-2016 is used to find evidence of a decrease in the disposition effect.

From the original 500 stocks, 144 stocks are listed on the S&P 500 in the period 1990-2016. Because 144 is still a large number of stocks to examine, a random sample of 30 stocks is taken. The total number of trading days in the period 1990-2016 is 6805, which is equal to the number of observations per company. With 30 companies that makes the total number of observations. From each company the price, trading volume and shares outstanding are taken. The data is obtained from the Center for Research in Security Prices (CRSP). A description of the data is given in the Appendix

(14)

CHAPTER 5 Results

The results of the estimation of the coefficient b1 from the regression equation (5) are given below. Also, the standard error of the coefficient and the significance level are given. * Indicates a 90 % confidence level significance of the coefficient one-sided tested against 0. ** Indicate a 95 % confidence level significance of the coefficient one-sided tested against 0.

Table 1 Coefficients and significance levels of b1

Company name 1990-1998 1999-2007 2008-2016 3M CO b1 st.error 0.0885** 0.0222 0.0891 0.0518 -0.0008 0.0646 A T & T INC b1 st.error 0.0336** 0.0120 0.0690 0.0396 -0.0474 0.0301 ABBOTT LABORATORIES b1 st.error 0.0233 0.0124 0.0457 0.0319 0.0075 0.0358 AIR PRODUCTS & CHEMICALS INC b1

st.error 0.0313** 0.0138 0.0442 0.0435 -0.0018 0.0540 AVERY DENNISON CORP b1

st.error 0.0345* 0.0162 0.0623 0.0681 -0.0966 0.0801 BANK OF AMERICA CORP b1

st.error 0.1579** 0.0535 0.0680 0.0470 0.8216 0.7086 C S X CORP b1 st.error 0.0266 0.0174 0.1533 0.1228 0.0511 0.1373 CENTERPOINT ENERGY INC b1

st.error 0.0129 0.0177 0.0549 0.0787 -0.0374 0.0685 CORNING INC b1 st.error 0.0669 0.0402 -0.2618 0.1618 -0.0010 0.0758 ENTERGY CORP NEW b1

st.error 0.0910** 0.0255 0.1235* 0.0611 -0.0733 0.0296 FEDEX CORP b1 st.error 0.1136** 0.0293 0.0340 0.0418 0.0316 0.842 GAP INC b1 st.error 0.0486 0.0303 0.0230 0.0884 0.00503 0.0704

(15)

ILLINOIS TOOL WORKS INC b1 st.error 0.0356** 0.0116 0.0022 0.0331 0.0064 0.0675 INTERNATIONAL FLAVORS & FRAG INC b1

st.error 0.0154 0.0192 0.1192 0.0656 -0.0090 0.0787 INTERNATIONAL PAPER CO b1 st.error 0.0690 0.0438 0.0253 0.0366 0.1225 0.1622 JOHNSON & JOHNSON b1

st.error 0.0452** 0.0156 0.0980** 0.0364 -0.0812 0.0441 KELLOGG CO b1 st.error 0.0001 0.0124 0.0619 0.0371 -0.0356 0.0318 KROGER COMPANY b1 st.error 0.1309** 0.0245 -0.0257 0.0567 -0.0655 0.0787 LOWES COMPANIES INC b1

st.error 0.1087** 0.0348 -0.0103 0.0483 -0.1118 0.0953 NEWELL COMPANY b1 st.error 0.0187 0.0204 0.0859 0.0478 -0.0806 0.1505 NORFOLK SOUTHERN CORP b1

st.error 0.0050 0.0116 0.0689 0.0843 -0.0314 0.0995 NUCOR CORP b1 st.error 0.0602* 0.0273 0.1777 0.1171 0.2665 0.1844 PACIFIC GAS & ELEC CO b1

st.error 0.0587* 0.0288 0.0708 0.0573 -0.0510 0.0421 PEPSICO INC b1 st.error 0.0882** 0.0330 0.0432 0.0241 -0.0401 0.0441 PERKINELMER INC b1 st.error 0.0412 0.0251 0.0477 0.0566 -0.1388 0.1048 PUBLIC SERVICE ENTERPRISE GP INC b1

st.error 0.0475 0.0330 0.0906 0.0588 -0.0589 0.0243 SHERWIN WILLIAMS CO b1 st.error 0.0352* 0.0178 0.1314 0.0820 0.0303 0.1253 SOUTHERN CO b1 st.error 0.0265 0.0157 0.1091** 0.0447 -0.0264 0.0362

(16)

WHIRLPOOL CORP b1 st.error 0.0679* 0.0320 0.1195 0.1168 -0.0194 0.2695

In the period 1990-1998, 17 out of the total 30 stocks from the sample have a b1 coefficient with at least a significance level of 90%. 11 out of that 17 stocks have a b1 coefficient with at least a significance level of 95%.

The following period, 1998-2007, 3 out of 30 stocks show a b1 coefficient with at least a significance level of 90%, whereof 2 with at least a significance level of 95%.

(17)

CHAPTER 6 Conclusion

The results of the second regression (equation 5) shows in the period 1990-1998 17 out of 30 samples with a significant positive b1 coefficient. Which makes it plausible that the disposition effect still exists in that period. The following 2 periods, 1998-2007 and2007-2016 show respectively 3 and 0 significant positive b1 coefficients. Evidence of the disposition effect is diminishing in time. What suggests that the disposition effect itself is diminishing.

Future research could make this conclusion more robust. It could check if the disposition effect is also diminishing in other markets, for instance in different countries or for smaller firms. The time could be made wider, by collecting data from before 1990. Also, the role of institutional and private investors in the diminishing of the disposition effect could be more explored by future researchers.

(18)

REFERENCES

Dyl, E. (1977). Capital gains, taxation and year-end stock market behavior. Journal of Finance, 32(1), 165-175.

Ferris, S., Haugen, R., & Makhija, A. (1988). Predicting Contemporary Volume with Historic Volume at Differential Price Levels: Evidence Supporting the Disposition Effect. Journal of Finance,

43(3), 677-697.

Hens, T., & Vlcek, M. (2011). Does Prospect Theory Explain the Disposition Effect? Journal of

Behavioral Finance, 12(3), 141-157.

Kahneman, D., & Tversky, Amos. (1979). Prospect theory an analysis of decision under risk.

Econometrica : Journal of the Econometric Society, an Internat. Society for the Advancement of Economic Theory in Its Relation to Statistics and Mathematics, 47(2), 263-291.

Odean, T. (1998). Are Investors Reluctant to Realize Their Losses? Journal of Finance, 53(5), 1775-1798.

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.

Starks, L., Yong, L., & Zheng, L. (2006). Tax‐Loss Selling and the January Effect: Evidence from Municipal Bond Closed‐End Funds. Journal of Finance, 61(6), 3049-3067.

Weber, & Camerer. (1998). The disposition effect in securities trading: An experimental analysis.

(19)

APPENDIX A Description of the data obtained from CRSP

company name Variable Mean Std. Dev. Min Max

3M CO volume 2201997 1862414 77600 2.44e+07 price 9520212 2830962 41.83 181.42 sharesout 521855.6 191114.2 214001 784883 A T & T INC volume 1.29e+07

1.36e+07 64700 1.31e+08 price 40.10361 12.79233 19.34 83.75 sharesout 3325487 2209507 299648 6259793

ABBOTT LABORATORIES volume 4374875 3681779 124000 5.45e+07 price 4.617.103 1.023.501 23 781.875 sharesout 1274060 411005.1 222180 1580668 AIR PRODUCTS & CHEMICALS

INC

volume 899881.2 783285.8 19900 1.39e+07 price 6.933.293 3.039.685 241.875 158.13 sharesout 182815.9 56844.94 54853 234224 AVERY DENNISON CORP volume 555502.8 547197 8000 7039100

price 4.603.172 1.457.176 15.625 78.84 sharesout 93735.87 23260.7 44161 120850 BANK OF AMERICA CORP volume 5.55e+07 1.01e+08 0 1.23e+09

price 4.296.145 2.371.962

-8.021.875 124 sharesout 4128116 4091796 100966 1.08e+07 C S X CORP volume 3056051 3844783 32200 4.03e+07

price 4.368.645 1.722.636 18.39 91.75 sharesout 386818 330560.5 97849 1102764 CENTERPOINT ENERGY INC volume 1987638 2074430 15200 3.22e+07

price 2.425.159 1.096.714 4.5 50.02 sharesout 295655.4 104393.1 126248 430682

(20)

sharesout 900069.4 598464.3 91855 1578668 ENTERGY CORP NEW volume 1027183 677180.8 32300 5863200 price 5.295.085 2.558.108 166.875 126.07 sharesout 205347.7 23881.26 174886 246895 FEDEX CORP volume 1480392 1358388 24100 1.51e+07

price 7.978.984 3.600.167 29.875 201.02 sharesout 223063.5 110332.4 52698 317218 GAP INC volume 4404223 4205372 23600 3.88e+07

price 3.113.035 1.358.588 8.84 76.625 sharesout 522777.6 289457.9 35138 903759 ILLINOIS TOOL WORKS INC volume 1412306 1498187 9200 2.15e+07

price 6.408.627 1.834.664 26.19 127.93 sharesout 304047.7 161222.3 52794 567887 INTERNATIONAL FLAVORS &

FRAG INC

volume 383292.7 332302.3 5500 5187799 price 58.221 2.900.681 15.125 142.97 sharesout 84642.11 22442.02 37721 113163 INTERNATIONAL PAPER CO volume 2796323 2456873 54200 2.42e+07

price 4.383.067 1.498.011 4.09 90.75 sharesout 357571.8 136565.4 108700 493108 JOHNSON & JOHNSON volume 6419332 5527159 99500 9.84e+07

price 7.086.998 1.932.168 35.625 125.4 sharesout 2023432 1015651 332917 3119843 KELLOGG CO volume 1179989 1039711 18100 1.09e+07

price 535.224 1.653.117 21 112.875 sharesout 333934.3 90732.35 120010 415572 KROGER COMPANY volume 3516681 3652860 19000 7.71e+07

price 2.749.595 1.246.808 10.875 77.24 sharesout 504397.8 291635.2 81592 974723 LOWES COMPANIES INC volume 5298111 5992297 7900 5.43e+07

price 4.142.639 1.580.552 13.39 85.25 sharesout 703131.4 510171.2 36412 1550758 NEWELL COMPANY volume 1722325 2001126 15400 4.24e+07

price 2.930.594 9.943.639 4.54 54.89 sharesout 225933.4 90750.43 58927 482400

(21)

NORFOLK SOUTHERN CORP volume 1613894 1600039 28200 2.27e+07 price 5.446.526 2.526.767 12 117.2 sharesout 300570.3 106209.6 125088 435135 NUCOR CORP volume 1901044 2324967 3300 2.21e+07

price 5.422.326 1.423.571 25.52 119.3 sharesout 170801.8 117973.4 21378 319615 PACIFIC GAS & ELEC CO volume 1710002 1462351 85900 3.08e+07

price 3.395.177 1.139.328 6.9 65.39 sharesout 408450.4 36584.16 345320 505667 PEPSICO INC volume 4323285 2880810 239500 3.50e+07

price 5.543.389 2.013.652 21.625 109.96 sharesout 1375023 375146.6 262921 1773043 PERKINELMER INC volume 643223.7 695455.3 5500 1.62e+07

price 2.947.784 159.017 4.37 119.5 sharesout 87369.49 36644.51 27713 130797 PUBLIC SERVICE ENTERPRISE

GP INC

volume 1527753 1544948 55200 2.23e+07 price 3.830.114 1.387.978 21.76 103.24 sharesout 321976.8 129733.3 205840 508 SHERWIN WILLIAMS CO volume 721853.3 782674.3 12100 2.25e+07

price 7.275.364 7.309.908 174.375 312.1 sharesout 117155.9 33590.77 43105 173 SOUTHERN CO volume 2568026 2303518 41200 2.32e+07

price 3.291.979 8.461.807 17.5 54.54 sharesout 691571 180179.1 315654 980798 STANLEY BLACK & DECKER INC volume 655632 778676.8 6000 1.75e+07

price 5.089.389 2.233.207 19.25 125.78 sharesout 94212.4 42136.02 41801 176903 WHIRLPOOL CORP volume 829286.5 828414.4 13700 1.31e+07

price 7.500.018 3.922.312 17.875 215 sharesout 73373.79 3.855.716 66000 79944

The disposition effect is diminishing in time