Chapter 2 Literature Review
2.3 Disposition Effect Testing Models
There are three kinds of models to test whether the disposition effect exists, we will discuss them as follows:
Model 1: Using the Abnormal Volume to Explain the Disposition Effect.
Most authors use the reference point effect to explain investment situations.
Disposition effect was initiated by Shefrin and Statman (1985) who described the tendency to “sell winners too early and ride losers too long” which is related to normative theory of investment behavior pattern. When investors buy a stock and the price of this stock in the future is higher than the current price to get capital gains in stock which is called winner.
Conversely, a “loser” is a stock with capital losses that the price in the future is lower than
the initial purchase price.
Mental accounting is used to explain why an investor acts likely to restrain the reference point be readjusted for a stock. Regret aversion could interpret why investors may have difficulty in realizing gains as well as losses. Regarding the self-control, it could provide a rational explanation why investors exhibit reluctantly to realize losses. Tax considerations press investors to realize short-term losses and long-term gains. However, the disposition effect is not able to be explained by any single element alone, it may be combined with the four elements to conjecture the effect.
Lakonishok and Smidt (1986) analyzed the relationship between abnormal volume and past prices for listed companies on the American Stock Exchange (ASE) and New York Stock Exchange (NYSE) during the years 1968-1982. The price should be classified in four segments for five-, eleven-, twenty-three-, and thirty-five-month. Comparing with the price in every month between these segments, there are three situations. At the beginning, during the first three segments, investors will have positive rewards and higher volume for winners, butthey will have negative return in thirty-five-month.
Second, for taxable investments predict that would behave the behavioral pattern of the abnormal volume of winners was higher than normal in January and the abnormal volume of losers was higher than normal in December. Third, winners were prone to have higher abnormal volume than losers in December.
Ferris, Haugen and Makhija (1988) chose very small U.S. stock that the volume at a give point in time can be predicted with historic volume at differential price. The result showed that the volumes were small at higher price regions and large at lower price regions.
They combine cross-sectional with vertical-sectional analyzed by multiple regression model exhibits that the abnormal volume is significant when the turnover of trading volume is higher.
Kao (2003) conferred the relationship between the disposition effect and information
release by the F. H. M. model. They found that: if there is bad news announced, the losers tend to sell their position. The other way, if there is good news announced, the winners tend to hold their position.
Tseng (2003) found four phenomena. (1) Disposition effect existed in Taiwan stock markets. (2) In general, individual investors’ behavioral patterns were stronger than institutional investors. (3) When investors got the capital gains, invested the small-scale companies had stronger disposition effect than extensive companies. (4) When investors got the capital gains, invested the non-electronic industry had stronger disposition effect than electronic industry. (5) When investors got the capital gains, bull markets had stronger disposition effect than bear markets.
Model 2: Using the Ratio of Gain and Loss to Confer on the Disposition Effect
Odean took PGR and PLR as a measurement and tried to observe if there exists disposition effect at the time of keeping and selling stocks for investors. The two ratios were calculated as follows:
PGR: Proportion of Gains Realized PLR: Proportion of Losses Realized RG: Realized Gains
RL: Realized Losses
PG: Paper (unrealized) Gains PL: Paper (unrealized) Losses
Then, there are two hypotheses to be tested:
Hypothesis 1:
The null hypothesis is H : 0 PGR≤PLR The alternative hypothesis is H1: PGR>PLR Hypothesis 2:
The null hypothesis is H : 0 PLR−PGR in December≤PLR−PGR in January through November.
The alternative hypothesis is H1: PLR−PGR in December >PLR−PGR in January through November.
Odean (1998) selected 10,000 customer accounts from a nationwide discount brokerage house of all Finnish stock market investors from January 1987 through December 1993. In the annual return ratio, the PGR is 23.3% and the PLR is 15.5%. It was clear to know that the proportion of sell winners was higher than the proportion of sell losers, and thereby the disposition effect was existed. In addition, the test of the hypothesis 2 had shown that the amount of investors realizing losses in December was higher than realizing gains, because of tax consideration. So the disposition effect did not exist in December.
Lee (2003) used the weighted average absolute return instead of the numbers of losses and gains. They found the disposition effect existed in Taiwan investors. But there was no significant difference of the disposition effect in lunar New Year.
Model 3: Creating a Disposition Coefficient to Test the Disposition Effect.
Weber and Camerer (1998) made portfolio decisions to illustrate the disposition effect, the subject included six risky assets during 14 periods. They had derived four hypotheses to conjecture the behavioral pattern and the fluctuation in the prices.
H1: (Purchase price reference point) Subjects sell more shares when the sale price is above the purchase price than when the sale price is below the purchase price.
H2: (Last period reference point) Subjects sell more shares when the sale price is above the last period price than when the sale price is below the last period price.
H3: Disposition effects are smaller when assets are automatically sold than when selling is deliberate.
H4: Trading volume is positively correlated with the size of price changes.
In order to describe the size of disposition effect, they created a disposition coefficient (α ) as following: result shows that these four hypotheses are all significant. In other words, it is consistent with disposition effect. Because investors prefer to risk-avoid in the gain regions than the loss regions. It would produce the mean-reversion if the prices are rebounded. The mean-reversion means that the winner stocks would fall and losers would rise.
Lin (2003) used the disposition coefficient from Weber and Camerer to test the trading behavior of individual investor. In that article, the different psychological effects where analyzed by the biggest 30 and the smallest 30 listed companies, and then separated into three benchmarks such as return-rate ranges, bull-bear periods and different months.
The empirical result showed that there was not significant difference in the Chinese New Year Period and the end of year contrast with the other months. The disposition effect existed within each segment but not obviously when the return-rate lies between 0%~
+-2% after considering the return-rate.
CHAPTER 3
RESEARCH METHODOLOGY
The purpose of this study is to verify whether the disposition effect exists in major developed markets and emerging markets. The research methodology is derived from the model constructed by Ferris, Haugen and Makhija (1988). In this chapter, we will discuss the research steps as follows, including research design, research period, subjects and sources of data and research limitations.
3.1 The Research Design
In the previous chapter, we introduced three kinds of disposition effect models, model 1 and model 2 must use the investors’ private trading data in brokers to test the disposition effect. In this paper, we use the public data to test the disposition effect between the developed markets and emerging markets that quoted from model 3.
3.1.1 Model and Variable Definition
The so-called disposition effect is to test whether investors prefer to sell assets that have gained value and keep assets that have lost value. When investors sell assets, it will lead to increase the trading volume, and when investors keep assets, it will lead to decrease the trading volume. The model is following Ferris, Haugen and Makhija (1988) as stated below.
1. First, we estimate the daily abnormal turnover for each country relative to Market Portfolio ( All Countries world Index, ACWI ):
Vit =Ai+BiVim +eit ………(4)
The weight of a country in an MSCI index is calculated in the following way:
Weight2 = 100
Following Beaver (1968), Dyl (1977), and Lakonishok and Smidt (1986), equation (3) allows us to estimate e , the fraction of the market value of country i traded on day t after it removing the effects of market-wide events.
2. The relation of the abnormal turnover, e , for the weighted price index of each it country, P , to its past turnover is formulated in the following fashion: it
a. Eight price ranges around P are considered to classify past price for the stock, it P , where n is a trading day in the past. These ranges are in
2 FIF is the Foreign Inclusion Factor.
Range 1 : Pit <Pin ≤(1+x)Pit
Range 2 : (1+x)Pit <Pin ≤(1+2x)Pit Range 3 : (1+2x)Pit <Pin ≤(1+3x)Pit Range 4 : (1+ )3x Pit <Pin
Range 5 : Pit ≥Pin >(1−x)Pit
Range 6 : (1−x)Pit ≥Pin >(1−2x)Pit Range 7 : (1−2x)Pit ≥Pin >(1−3x)Pit Range 8 : (1− )3x Pit ≥Pin
=
x 0.025, 0.05, 0.075, and 0.1.
where
P : the weighted price index of each country i trading on day t . it
P : the weighted price index of each country i trading on day n. in
n : the past trading day, going back for one calendar year(365 calendar days) x : the rank of eight ranges to classify four kinds.
The main idea of this step is to classify the eight ranges to identify capital gain or capital loss. Each past trading day, going back for on calendar year (365 calendar days), if the price index (P ) is greater than the past price index (it P ), there is capital gain. in Conversely, there is capital loss. Range 1 is the maximum capital loss and from the range 2 to range 4 is decreasing progressively. Conversely, from range 5 to range 7 are increasing progressively, and range 8 is the maximum capital gain.
b. After classification, we will accumulate the trading volume of every range. The raw (now abnormal) trading volume for that day is assigned to that range and all the volumes assigned to a range are added. Therefore, we will have eight cumulative trading volumes,
each representing the total cumulative trading volumes on different range.
3. Then we run a time-serious regression model:
it
According the hypothesis, if the disposition effect is valid, we will expect that H1 :
β
k <0 for k =1、2、3、4 andβ
k >0 for k =5、6、7、84. Comparing the disposition effect degree between developed markets and emerging markets:
According to the step 3, when the loss range exists the disposition effect, the coefficient is negative. Conversely, the coefficient is positive.
First, we calculate the total numbers of significant level that match to the expectation to compare the degree of disposition effect with each market.
Second, we calculate the total score in order to compare the degree of disposition effect between developed markets and emerging markets. We set up three kinds of standards. First, we set the score as 1 when the symbol match to the expectation and significant. Second, we set the score as -1 when the symbol doesn’t match to the expectation and significant. Third, we set the score as 0 when the symbol is not significant and whatever the symbol match to the expection or not. We will sum the total score for the four ranks of x to compare the degree of disposition effect between developed markets
and emerging markets.
3.2 The Research Subjects, Sources of Data and Research Period
3.2.1 Research Subjects
Our subjects are to compare the disposition effect between major developed markets and emerging markets. When we estimate the abnormal turnover, the weighting price index of all countries is calculated using the 24 developed and 27 emerging markets according to global index of Morgan Stanley Capital International Inc.(MSCI) (see Table 1).
MSCI is a leading provider of global indices and benchmark related products and services for investors worldwide. It is headquartered in New York, and conducts business worldwide with operations in Geneva, London, Hong Kong, Tokyo, Singapore, Sydney, Frankfurt, Milan, Paris, Princeton and San Francisco. The business of MSCI is to provide benchmark products and services to the investment management community, to distribute index and company-level data and to license the MSCI indices to third parties for the purpose of creating derivatives and proprietary products.
For the global or institutional investors, the proportion of each country in the MSCI index is usually the capital allocation ratio on each country’s stock markets. The proportion of each contury in the MSCI index will be adjusted flexibly according to the stock markets condition around the world. Consequently, the MSCI index is important.
Table 1
Developed Markets and Emerging Markets of All Countries World Index of MSCI Developed Markets Emerging Markets
Australia
3.2.2 Sources of Data and Research Period
The data include: daily data of weighting price index, trading volume, market value for 24 developed and 27 emerging markets. However, some markets have data missing, such as Colombia, Egypt, Germany, Jordan, Morocco, and Sri Lanka. Therefore, our research subjects include 23 developed and 22 emerging markets. Our data period is from November 25, 2002 through November 25, 2005. Sources of these data are from the Datastream database.
3.3 Research Limitations
First, when we calculate the turnover ratio of global index, we ignore the time difference between various countries.
Second, the price limit is different in the stock markets of each country that may cause the disposition effect biased.
Chapter 4
Empirical Results and Analysis
In this chapter, we will show the empirical results of the disposition effect for investors’ behavioral pattern. First, we investigate the disposition effect for each market.
Second, we compare the degree of disposition effect between developed markets and emerging markets.
According to the regression analysis, we have the coefficients and t-value of eight ranges. If the disposition effect exists, the investors will hold the stocks continuously when there are capital losses and make the abnormal turnover reduce. Conversely, the investors will sell the stocks when there are capital gains and make the abnormal turnover rise.
Therefore, if the disposition effect exists, the coefficients
β
1 throughβ
4 should be significantly negative while the coefficientsβ
5 throughβ
8 should be significantly positive.4.1 Analysis The Disposition Effect for Each Market
There are four ranks of x to verify the disposition effect. We take the x=0.075 to illustrate the disposition effect in Table 2 (developed markets) and Table 3 (emerging markets). The similar empirical results of other ranks are presented in table 6~11 of appendix A~C.
Table 2
Empirical Results of Developed Market for x=0.075
it
volume for country i for all past days in the previous 365 days from day t in which it traded in range 1. (
E2it, etc. are defined similarly.)
(November 25, 2002 through November 25, 2005)
Names ofMarkets
α
0β
1β
2β
3β
4β
5β
6β
7β
8 R2(adj)Australia -0.127035 1.61E-09+++ 1.46E-09+++ 1.98E-09+++ 1.54E-09+++ 1.38E-09*** 1.38E-09*** 1.42E-09*** N/A 0.271934 t value (-13.247) (11.790) (10.815) (10.177) (2.606) (11.509) (15.948) (12.087) N/A
Austria -0.002852 4.81E-09+++ 1.69E-09+ 1.40E-08+++ -4.87E-08 1.84E-09** 7.64E-09*** 7.17E-09*** 9.55E-09*** 0.660304 t value (-16.649) (4.150) (1.862) (4.278) (-1.583) (2.364) (9.183) (9.103) (17.088) Belgium -0.017182 1.19E-08+++ 1.36E-08+++ 1.05E-08+++ 1.04E-08+++ 9.54E-09*** 1.08E-08*** 1.45E-08*** 1.23E-08*** 0.482508 t value (-20.804) (11.990) (11.517) (7.841) (14.042) (14.908) (15.088) (21.266) (11.010)
Canada -0.063948 1.71E-09+++ 2.12E-09+++ 1.90E-09+++ 3.16E-09+++ 2.29E-09*** 2.16E-09*** 3.17E-09*** 1.47E-09*** 0.223182 t value (-9.099) (5.971) (4.662) (5.014) (5.298) (8.953) (8.862) (10.315) (2.760) Denmark -0.017559 8.91E-09+++ 1.05E-08+++ 1.06E-08+++ 9.34E-09+++ 8.52E-09*** 1.05E-08*** 1.03E-08*** 1.04E-08*** 0.430558 t value (-14.005) (9.711) (7.911) (9.176) (10.768) (11.531) (14.144) (14.519) (12.828)
Finland -0.013909 2.69E-09+++ 6.22E-10 8.63E-11 1.61E-09++ -9.90E-10 2.48E-09*** 2.06E-09*** -1.78E-12 0.027062 t value (-2.230) (2.597) (0.641) (0.072) (2.200) (-1.175) (3.298) (2.751) (-0.001) France -0.092578 2.68E-09+++ 4.28E-09+++ 4.06E-09+++ 3.23E-09+++ 1.80E-09*** 1.72E-09*** 1.73E-09*** 5.22E-09*** 0.084725 t value (-9.450) (7.837) (5.021) (3.583) (8.772) (7.784) (7.203) (4.398) (4.588)
Greece -0.000648 -1.67E-09*** -1.87E-09 -2.88E-09* -1.14E-09 1.90E-09*** 1.13E-09** 2.14E-09*** -4.57E-11 0.238179 t value (-0.660) (-2.942) (-1.374) (-1.833) (-1.287) (3.017) (2.129) (3.151) (-0.054) Hong Kong -0.005196 -7.06E-11*** 1.50E-11 -6.62E-11* -9.74E-12 6.55E-11*** 6.40E-11*** 1.07E-10*** 4.49E-11** 0.244601 t value (-1.768) (-3.359) (0.580) (-1.928) (-0.388) (3.495) (3.440) (3.625) (2.132) Ireland -0.006611 9.96E-10+++ 9.14E-10++ 3.86E-10 1.97E-09+++ 1.19E-09*** 8.82E-10*** 1.02E-09*** 2.93E-09*** 0.175979 t value (-9.824) (4.976) (2.375) (0.976) (9.925) (7.258) (5.749) (4.937) (4.506) Italy -0.190210 1.24E-09+++ 2.58E-09+++ 6.18E-10++ 1.43E-09+++ 1.11E-09*** 1.33E-09*** 1.49E-09*** -3.45E-09+++ 0.288309 t value (-10.648) (9.683) (11.642) (2.569) (9.598) (9.328) (11.702) (12.983) (-3.194) Japan -0.240039 6.50E-10+++ 1.60E-09+++ 5.25E-10 3.48E-10 1.02E-09*** 1.51E-09*** 1.97E-09*** 1.66E-09*** 0.468718 t value (-8.946) (4.655) (5.100) (1.203) (1.446) (8.240) (10.634) (12.767) (5.655)
Luxembourg 2.10E-05 1.59E-09 -4.90E-09 -3.47E-09 -1.98E-09 -2.96E-09 1.47E-09** -5.33E-09++ 4.35E-09*** -0.002771 t value (0.787) (0.682) (-0.941) (-0.765) (-1.111) (-1.123) (2.483) (-2.423) (4.058)
Netherlands 0.024018 -5.49E-10* -2.02E-09*** -7.27E-10* -3.01E-11 -1.03E-09+++ -7.65E-10+++ -4.52E-10 5.04E-10 0.103781 t value (3.208) (-1.877) (-4.390) (-1.933) (-0.083) (-4.138) (-2.607) (-0.998) (0.210)
New Zealand 0.000669 -9.98E-11 4.04E-11 N/A N/A -2.61E-10 1.19E-10 -2.99E-10 N/A 0.021817
t value (0.615) (-0.501) (0.073) N/A N/A (-1.384) (0.619) (-1.157) N/A
Norway -0.028502 2.25E-09+++ 5.16E-09+++ 3.33E-09+++ 3.59E-09+++ 2.38E-09*** 3.51E-09*** 3.73E-09*** 3.09E-09*** 0.430362 t value (-16.772) (5.844) (7.517) (5.267) (11.915) (7.368) (13.228) (13.625) (14.669)
Portugal -0.001598 2.16E-10++ 9.11E-11 1.16E-10 -1.25E-10 2.57E-10*** -1.26E-11 1.09E-10 5.77E-09*** 0.357782 t value (-2.581) (2.121) (0.495) (0.587) (-0.928) (3.391) (-0.086) (0.667) (16.763)
Singapore 0.001161 -1.12E-10*** -7.42E-11 -2.53E-10*** -4.18E-11 -2.82E-11 7.44E-12 1.28E-10*** -6.68E-12 0.117185 t value (1.245) (-3.958) (-1.072) (-3.495) (-0.917) (-1.330) (0.309) (3.577) (-0.192) Spain -0.138819 4.13E-09+++ 4.95E-09+++ 4.26E-09+++ 4.04E-09+++ 4.31E-09*** 4.06E-09*** 3.55E-09*** 4.40E-09*** 0.311715 t value (-16.303) (13.387) (10.370) (11.299) (13.967) (15.494) (16.177) (12.148) (7.594)
Sweden -0.008185 -7.24E-11 2.72E-10 -4.48E-10* -1.61E-10*** 1.39E-10* 4.40E-10*** -2.04E-10 6.41E-10*** 0.175667 t value (-2.888) (-0.771) (1.239) (-1.727) (-2.084) (1.679) (5.648) (-2.320) (5.335)
Switzerland -0.038416 2.93E-09+++ 7.60E-09+++ 6.70E-09+++ 4.12E-09+++ 3.37E-09*** 5.49E-09*** 1.07E-09 2.31E-10 0.131442 t value (-10.414) (5.745) (6.990) (5.115) (7.347) (6.693) (8.384) (1.071) (0.164)
United Kingdom -0.711254 1.17E-09 1.58E-09 1.29E-09 1.45E-09+++ 1.24E-09*** 1.34E-09*** 1.13E-09*** 2.74E-09*** 0.256910 t value (-14.289) (11.773) (10.783) (9.789) (15.292) (14.958) (13.875) (7.378) (4.002)
USA 1.694063 -2.96E-09** -2.42E-09 -3.72E-09*** -1.51E-09 -2.55E-09++ -3.67E-09++ -4.33E-09+++ -3.63E-09++ 0.166477 t value (1.961) (-1.965) (-1.550) (-2.707) (-1.076) (-1.686) (-2.492) (-2.608) (-2.122)
When the symbol match to the expectation and significant, the significant levels are as follows: * means 10%, ** means 5%, *** means 1%.
When the symbol doesn’t match to the expectation and significant, the significant levels are as follows: + means 10%, ++ means 5%, +++ means 1%.
Table 3
Empirical Results of Emerging Markets for x=0.075
it
volume for country i for all past days in the previous 365 days from day t in which it traded in range 1. (
E2it, etc. are defined similarly.)
(November 25, 2002 through November 25, 2005)
Names ofMarkets
α
0β
1β
2β
3β
4β
5β
6β
7β
8 R2(adj)Argentina -0.004806 -2.01E-09*** 2.55E-09+++ -1.02E-09 2.40E-09+++ 1.14E-09*** 2.40E-09*** 2.16E-09*** 1.82E-09*** 0.197498 t value (-9.637) (-3.861) (4.266) (-1.256) (7.735) (2.948) (6.845) (6.731) (9.757) Brazil -0.056577 1.92E-09 2.63E-09 -9.51E-10 2.87E-09 8.33E-09*** 3.88E-09*** 6.00E-09*** 9.97E-09*** 0.100516 t value (-4.781) (0.847) (0.922) (-0.215) (0.492) (4.540) (2.255) (3.232) (5.463)
Chile -0.004226 -6.75E-11 9.97E-13 -3.03E-10 -3.59E-09*** 1.34E-10*** 1.21E-10*** 2.15E-10** 2.52E-10** 0.122321 t value (-4.509) (-0.984) (0.008) (-0.658) (-2.828) (3.749) (3.638) (2.350) (2.543)
China -0.109227 1.33E-10 1.03E-09+++ 1.60E-10 7.12E-11 6.22E-10*** 8.82E-10*** 1.39E-09*** 1.15E-09*** 0.428096 t value (-19.708) (1.101) (5.773) (0.781) (0.357) (5.019) (8.233) (7.036) (16.975) Czech Republic -0.023262 3.07E-08+++ 1.01E-07+++ -1.12E-07** -7.68E-08 2.37E-08*** 1.60E-08*** 1.22E-08 7.25E-08*** 0.446639 t value (-12.503) (3.781) (3.676) (-2.071) (-1.276) (3.755) (3.091) (1.568) (11.354)
Hungary -0.007259 1.37E-10 2.56E-08++ -4.06E-08** 1.49E-08 1.23E-08*** 9.22E-09* -7.49E-09 3.06E-08*** 0.189728 t value (-3.147) (0.019) (2.298) (-2.532) (0.669) (3.568) (1.917) (-1.558) (11.182)
India -0.008118 1.92E-09 -4.63E-10 1.06E-08+++ 2.07E-09++ 5.30E-11 -2.89E-09++ -8.42E-09+++ 8.85E-09*** 0.258323 t value (-1.011) (1.368) (-0.284) (4.899) (2.073) (0.037) (-2.180) (-6.469) (11.824)
Indonesia -0.007459 -9.80E-11*** 2.13E-10+++ -1.91E-10* -9.95E-12 1.47E-11 1.30E-10*** 9.78E-11*** 1.73E-10*** 0.231445 t value (-5.605) (-3.226) (3.737) (-1.866) (-0.301) (0.508) (3.986) (3.049) (9.336)
Israel -0.003810 -5.84E-09 7.31E-09 -1.53E-08*** 3.74E-09 3.15E-09*** 1.85E-09 -1.96E-09 5.91E-09*** 0.101257 t value (-1.330) (-1.255) (0.879) (-2.686) (0.510) (2.824) (1.624) (-1.300) (2.886) Korea -0.282780 2.24E-09 3.65E-09 5.49E-09++ 2.81E-09+++ 6.00E-09** 4.47E-09* -1.09E-09 1.26E-08*** 0.216138 t value (-17.082) (0.807) (1.280) (1.682) (3.360) (2.160) (1.661) (-0.283) (11.420) Malaysia 0.004021 -1.10E-09*** 1.02E-09++ -3.82E-09*** 4.50E-09++ -4.66E-10 7.40E-10** -4.74E-10 5.59E-09*** 0.149593 t value (0.919) (-3.395) (2.488) (-4.596) (2.063) (-1.918) (2.142) (-0.759) (4.384) Mexico -0.048459 1.35E-09+++ 2.73E-09+++ -2.85E-10 3.03E-09++ 3.11E-09*** 2.51E-09*** 1.71E-09*** 2.70E-09*** 0.106369 t value (-7.720) (2.879) (4.482) (-0.320) (2.481) (7.667) (6.396) (4.084) (6.396)
Pakistan -0.078827 -8.93E-10** -8.06E-10 -7.39E-09*** 1.15E-10 4.40E-10 2.12E-09*** 2.12E-09*** 4.06E-09*** 0.470240 t value (-11.799) (-2.143) (-0.927) (-4.603) (0.088) (0.974) (4.563) (5.753) (23.289) Peru 0.006653 -4.51E-09 2.37E-08 -2.93E-07 N/A -8.17E-09 -8.78E-09 -7.02E-09 -3.11E-09 0.002866 t value (1.294) (-0.603) (1.038) (-0.433) N/A (-1.429) (-1.193) (-1.116) (-0.296)
Philippines -0.001291 -3.65E-11 1.77E-10++ 4.38E-11 -7.27E-12 1.26E-11 4.39E-11 3.97E-10*** 1.65E-10 0.079600 t value (-4.515) (-0.889) (2.399) (0.690) (-0.186) (0.308) (1.203) (4.775) (1.594)
Poland -0.008848 -1.33E-08*** 6.58E-09++ 5.54E-09++ 3.31E-10 1.57E-08*** 4.45E-09 1.60E-08*** -3.33E-09 0.243035 t value (-5.178) (-4.122) (2.422) (2.326) (0.085) (4.883) (1.455) (6.243) (-0.887)
Russia 0.059130 -5.13E-09*** 1.21E-09 -9.48E-09*** -1.57E-09 -6.27E-09+++ -1.38E-10 -3.52E-09+++ -4.18E-09+++ 0.069709 t value (5.220) (-5.230) (0.924) (-4.504) (-0.522) (-6.176) (-0.145) (-4.480) (-5.310)
South Africa -0.272488 1.65E-08+++ 1.55E-08+++ 1.82E-08+++ 1.71E-08+++ 1.52E-08*** 1.53E-08*** 1.64E-08*** 1.25E-08*** 0.069770 t value (-8.031) (7.920) (6.110) (5.401) (5.798) (6.791) (6.698) (8.420) (6.122) Taiwan 0.270540 -1.40E09*** -4.39E-11 -3.34E-10 -3.78E-10 -2.08E-09+++ 1.29E-09*** -8.63E-10+ 1.44E-09*** 0.216011 t value (3.579) (-4.080) (-0.115) (-0.578) (-1.338) (-7.300) (3.014) (-1.945) (3.118)
Thailand -0.047909 -4.14E-10** 1.10E-10 -1.77E-10 4.75E-10 7.97E-10*** 3.13E-10** 1.27E-09*** 2.13E-09*** 0.483491 t value (-16.655) (-2.160) (0.326) (-0.311) (0.756) (5.774) (2.161) (5.906) (21.740)
Turkey 0.010593 -8.28E-12*** -5.87E-12** -4.45E-12 -4.30E-12* 2.46E-12 3.73E-12** -4.34E-12+++ 3.34E-12*** 0.172727 t value (1.094) (-4.346) (-2.075) (-1.153) (-1.855) (1.475) (2.281) (-3.464) (5.199) Venezuela -0.00650 1.98E-10 5.01E-11 2.36E-10 1.75E-10 4.12E-10 5.50E-10** -4.45E-10 7.01E-10*** 0.008824 t value (-3.083) (0.622) (0.161) (1.107) (1.998) (1.333) (2.436) (-0.972) (3.521)
When the symbol match to the expectation and significant, the significant levels are as follows: * means 10%, ** means 5%, *** means 1%.
When the symbol doesn’t match to the expectation and significant, the significant levels are as follows: + means 10%, ++ means 5%, +++ means 1%.
These markets are divided into three groups in terms of degree of disposition effect.
According to the definition, disposition effect is to sell winners too early and ride losers too long. Therefore, winners and losers must achieve the statistical significance simultaneously to conform the disposition effect.
1. The Numbers of Significance Level is over 5:
When the numbers of the significance level that match to the expectation is over 5 that indicate the winners and losers exists simultaneously. Therefore, we think that this market have exist the disposition effect. There are three arguments in this condition:
First, stock prices under-react to bad news when more current holders are facing a capital loss, and over-react to good news when more current holders are facing a capital gain. Therefore, they will cause the investors keep the losses and sell the gains.
Second, investors have tendence to loss aversion and regret aversion. Therefore, they don’t want to face the bad news and hold the stocks with capital loss continuously.
Third, investors believe that long-term investing can reduce the risk. That is, investors are impassive for the bad news. This effect is called “Time Diversification3”. But it can’t explain the phenomenon of realized gains.
Those markets include Argentina, Chile, Greece, Hong Kong, Indonesia, Pakistan, Sweden, Thailand, and Turkey.
2. The Numbers of Significane Level is Through 1 to 4:
When the number of the significance level that match to the expectation is between 1 and 4, we think that disposition effect is incomplete. It can be divided into
3 Fisher, K. L., and M. Statman, “A Behavioral Fraamework for Time Diversification”, Accosciation for Investment Management and Research.
two parts. First, when the significance level tends to the winner, we call that
“short-term trading” or “speculative trading”. In other words, whether it produces losses or gains in the short term, investors will sell the stocks. Those markets include Austria, Belgium, Brazil, Canada, China, Denmark, France, Hungary, Ireland, Japan, Mexico, South Africa, Spain, and United Kingdom. Second, when the significance level tends to the loser, we call that “long-term investing”. In other words, whether it produces losses or gains in the short term, investors will hold the stocks, but not to sell. There are no such markets found in this study.
In addition, when the numbers of significance level are between 1 and 4, which some markets neither prone to winner nor prone to loser yet. We think that the disposition effect is incomplete. Those markets include Australia, Czech Republic, Finland, India, Israel, Italy, Korea, Luxembourg, Malaysia, Netherlands, Norway, Philippines, Poland, Portugal, Russia, Singapore, Switzerland, Taiwan, USA, and Venezuela.
3. The Numbers of Significance Level is 0:
When the number of significance level that matches to the expectation is 0, we assert that the disposition effect doesn’t exist. Those markets include New Zealand and Peru.
In addition, we relatively care about USA, United Kingdom, Japan (which are world-wide major stocks markets), Taiwan, China, Korea, Singapore, and Hong Kong (which are Great China Economy Countries). The numbers of significance level of USA is 2, United Kingdom is 4, Japan is 4, Taiwan is 3, China is 4, Korea is 3, Singapore is 3, and Hong Kong is 6. However, among these markets, Hong Kong exists the disposition effect relatively obvious than other markets which the disposition effect is incomplete.
USA, United Kingdom, and Japan are more efficient than other markets, moreover, the disposition effect is incomplete.
Subsequently, because we estimated the disposition effect of weighted price index for each country, which included different sector of stocks, it may cause the disposition effect incomplete and disintegrate. It is why some emerging and speculative markets, such as Taiwan, China, Korea and Singapore, does not exist disposition effect.
Finally, in the Hong Kong, because of the foreign capital flow in and out frequently and rapidly which cause the dispositon effect more stronger.
4.2 Analysis The Disposition Effect Between Developed Markets and Emerging Markets
In this section, we will compare the degree of disposition effect between developed markets and emerging markets. For simplification, we have established a standard for each outcome in all markets in 3.1. Table 4 offer a taxonomy that reflects the degree of disposition effect for developed markets and Table 5 for emerging markets.
Table 4 The Total Score of Developed Markets
Developed Market x=0.1 X=0.075 X=0.05 X=0.025 Total score
Data Source: Collection of this study.
Table 5 The Total Score of Emerging Markets
Emerging Market X=0.1 X=0.075 X=0.05 X=0.025 Total Score
Data Source: Collection of this study.