處置效果與過度自信之研究─以臺灣期貨市場為例 - 政大學術集成

全文

(1)國立政治大學商學院金融學系碩士班碩士學位論文 Graduate Institute of Money and Banking National Cheng Chi University Master Thesis. 治. 政處置效果與過度自信之研究大. 立. ‧ 國. 學. ─以臺灣期貨市場為例. A Study of Disposition Effect and Overconfidence. Nat. n. al. er. io. sit. y. ‧. in Taiwan Futures Markets. Ch. engchi. i Un. v. 研究生：倪詩荏撰指導教授：陳威光博士林靖庭博士. 中華民國一○四年○六月.

(2) 處置效果與過度自信之研究 ─以臺灣期貨市場為例摘要過去的實證研究顯示交易人行為並非完全理性，因此，為了研究臺灣期貨市場投資人的交易行為，本篇論文利用期貨合約日內交易資料，探討臺灣期貨交易人的處置效果及過度自信之關係，並將期貨交易人身分類別分為國內法人、散戶、自營商及外資。此外，本文更深入探討投資人的交易行為是否受交易量及市場情況影響，因. 治政大此進一步將資料做不同方向的分類。首先，將散戶依照交易量再分群，以探討不立同交易量的散戶族群的交易行為是否不同，另外，我們也將資料區間分為金融海. ‧ 國. 學. 嘯前與金融海嘯期間，藉此觀察金融海嘯是否對交易行為有所影響。. ‧. 本篇論文研究結果顯示，（1）國內法人及散戶有較強烈的處置效果，而自營. sit. y. Nat. 及外資則較弱。（2）依據投資績效結果顯示，自營商的交易量大，卻損失慘重，. io. al. er. 相較之下，明顯地具有過度自信。（3）交易量大的散戶會有較低的處置效果，但. n. 會較具過度自信。（4）機構投資人於金融海嘯期間會降低其處置效果及過度自信，而散戶則具有相反的表現。. Ch. engchi. i Un. 關鍵字：處置效果、過度自信、期貨市場、金融海嘯. v.

(3) A Study of Disposition Effect and Overconfidence in Taiwan Futures Markets ABSTRACT Previous literatures indicate that investors don’t exhibit rationally. To understand investors’ behaviors in Taiwan futures markets, we use intraday trading records from TAIFEX to examine the disposition effect and overconfidence across investor types. We focus on four investor types including domestic institutions, individuals,. 治政大 we further discuss whether To shed some light on trading behaviors of investors, 立. proprietaries and foreigners.. investors’ behaviors are influenced by trading volume or market conditions. We further. ‧ 國. 學. separate the individuals by their trading volumes to analyze whether trading behaviors. ‧. are different. Meanwhile, we also divide our sample period into two sub-periods─. y. sit. io. al. er. investor types.. Nat. before and during financial crisis, to investigate the change of trading behaviors across. n. The main contributions of this study are that (1) the retail investors, namely domestic institutions and individuals,. Ch exhibit. iv n effect emore n g disposition chi U. on futures contracts.. However, the professional institutions, including proprietaries and foreigners, display less disposition effect. (2) Proprietaries trade more, but they lose enormously. It appears that proprietaries demonstrate overconfidence. (3) Individual investors with higher trading volume have the lower disposition effect, but higher overconfidence. (4) Overall, institutions reduce disposition effect and overconfidence on most of contracts during financial crisis. In contrast, individual investors perform worse and display more overconfidence. Keywords: disposition effect, overconfidence, futures markets, financial crisis..

(4) Contents 1.. INTRODUCTION ................................................................................................ 1. 2.. LITERATURE REVIEW .................................................................................... 3 2.1 Disposition Effect................................................................................................ 3 2.2 Overconfidence ................................................................................................... 6. 3. DATA and METHODOLOGY ............................................................................... 9 3.1 Data ...................................................................................................................... 9 3.2 Methodology ..................................................................................................... 11. 立. 政治大. 4. EMPIRICAL ANALYSIS ..................................................................................... 15. ‧ 國. 學. 4.1 All Trading Activities ....................................................................................... 15 4.1.1 Descriptive Statistics.................................................................................. 15. ‧. 4.1.2 Disposition Effect ....................................................................................... 18. sit. y. Nat. 4.1.3 Performance ............................................................................................... 21. io. er. 4.2 Trading Activities of Individuals by Groups by Volume ............................. 24 4.2.1 Descriptive Statistics.................................................................................. 24. n. al. Ch. i Un. v. 4.2.2 Disposition Effect ....................................................................................... 27. engchi. 4.2.3 Performance ............................................................................................... 29. 5. Robustness .............................................................................................................. 32 5.1. Descriptive Statistics ....................................................................................... 32 5.2 Disposition Effect.............................................................................................. 37 5.3 Performance ...................................................................................................... 40. 6. CONCLUSION ...................................................................................................... 43. REFERENCES ........................................................................................................... 45.

(5) List of Tables Table 3.2-1. An Example of Simple Round Trip ............................................... 11. Table 3.2-2. An Example of Complex Round Trip ........................................... 12. Table 4.1.1 Descriptive Statistics of Investors, Trading Volumes, and Round Trips in Different Investor Types ...................................................................... 16 Table 4.1.2 Disposition Effect of Different Investor Types ............................. 20 Table 4.1.3 Performance of Round Trips and Net Profit ................................ 22. 政治大. Table 4.2.1 Descriptive Statistics of Investors, Trading Volumes, and Round Trips for Different Individual Groups.............................................................. 25. 立. ‧ 國. 學. Table 4.2.2 Disposition Effect of Different Individual Groups ....................... 28. ‧. Table 4.2.3 Performance of Round Trips and Net Profit for Different Individual Groups .............................................................................................. 30. y. Nat. sit. n. al. er. io. Table 5.1 Descriptive Statistics of Investors, Trading Volumes, and Round Trips Before or During Financial Crisis........................................................... 34. Ch. i Un. v. Table 5.2 Disposition Effect of Different Investor Types Before or During Financial Crisis ................................................................................................... 39. engchi. Table 5.3 Performance of Round Trips and Net Profit Before or During Financial Crisis ................................................................................................... 42.

(6) 1. INTRODUCTION In the past, the financial theories assumed that investors are rational and maximize expected utility. However, there are many empirical researches describe that investors do not always behave as the expected utility theory. They are not as rational as assumption. Kahneman and Tversky (1979) propose prospect theory, which indicates that people are risk-averse when they are in gains, but risk-seeking when facing losses. A lot of researchers claim that people sell winners too early and hold losers too long, which Shefrin and Statman (1985) call the disposition effect.. 政治大 stock market but also in futures market. Odean (1998) examined the investors account 立 Previous researchers have found that investors revealed disposition effect not only in. ‧ 國. 學. obtained from discount brokerage house in U.S. stock market. Frino, Johnstone and Zheng (2004) examined the local traders and the matched non-local traders in futures. ‧. markets. Shefrin and Statman (1985) do a further research on disposition effect and. Nat. sit. y. provide some positive explanations for the disposition effect. They suggest that main. al. n. regret and self-control.. er. io. elements of theories are prospect theory, mental accounting, seeking pride and avoiding. Ch. engchi. i Un. v. To test whether the disposition effect exists in Taiwan futures markets, we obtained the trading records of futures contracts from May, 2006 through December, 2008 for all futures investors. According to the data type is time series, we apply an appropriate method about round trip duration. It is totally different from the common method from Odean (1998), which measure disposition effect by calculating the proportion of winners or losers in the portfolio. Besides, because numerous researches indicate that institutions and individual investors perform differently on trading, we separately examine whether investors in different investor types exhibit different disposition effect. 1.

(7) In addition to disposition effect, we also discuss the overconfidence in different investor types. We apply the idea of Glaser and Weber (2007) that people lose money with excessive trading activities. In this study, we will link trading volume with performance to investigate the overconfidence. To shed some light on trading behaviors of investors, we further discuss whether investors’ behaviors are influenced by trading volume or market conditions. According to the statistic result, individual investors are the primary trading group in Taiwan. Moreover, some previous studies suggest that some individual investors trade like professional institutions. Therefore, we are interested in their trading behaviors so. 治政 that we further separate the individual group by their trading 大 volumes. 立 Meanwhile, we are also curious about if the financial crisis have impact on the trading ‧ 國. 學. behaviors of different investor types. As a result, we utilize a date of happening. sit. y. Nat. conditions.. ‧. financial crisis to divide our sample period into two periods including different market. io. er. The main contributions of this study are that (1) the retail investors, namely domestic. al. institutions and individuals, exhibit more disposition effect on futures contracts.. n. iv n C However, the professional institutions, proprietaries and foreigners, display h eincluding ngchi U lower disposition effect. (2) Proprietaries trade more, but they lose enormously. It appears that proprietaries demonstrate overconfidence. (3) Individual investors with higher trading volume have the lower disposition effect, but higher overconfidence. (4) Overall, institutions reduce disposition effect and overconfidence on most of contracts during financial crisis. In contrast, individual investors perform worse and display more overconfidence. In the following sections, we review the literatures in Section 2. Then data and methodology are introduced in Section 3. Section 4 demonstrates the empirical analysis. Section 5 is robustness, and conclusion in Section 6. 2.

(8) 2. LITERATURE REVIEW 2.1 Disposition Effect The main purpose of this paper is to know whether investors exhibit the disposition effect in Taiwan futures markets. I refer the literatures about disposition effect. Kahneman and Tversky (1979) propose the prospect theory to describe investor’s behaviors, but these behaviors violate the theoretical financial theories. They find an important result demonstrating that people’s attitudes toward risks concerning gains are different from their attitudes toward risks concerning losses. According to the results,. 政治大. they use a value function to illustrate this phenomenon. The function curve is concave. 立. for gains and convex for losses. It means that investors are risk-averse as they gain,. ‧ 國. 學. while risk-seeking as they lose. Since the abnormal behaviors had been discovered, many researchers worked on further discussions.. ‧. Shefrin and Statman (1985) focus on prospect theory and do an extended research.. y. Nat. io. sit. They propose the disposition effect which shows that people have a tendency to sell. n. al. er. winners quickly and hold losers for a long time. They explain the disposition effect with. Ch. i Un. v. prospect theory, mental accounting, seeking pride and avoiding regret and self-control.. engchi. Odean (1998) is the first paper to quantify the disposition effect by a methodology which includes two ratios: Proportion of Gains Realized (PGR) and Proportion of Losses Realized (PLR). This paper indicates that the disposition effect are not motivated by a desire to rebalance portfolios, or to prevent the higher trading costs from low priced stocks. Even nor is it justified by subsequent portfolio performance. This paper finds that the disposition effect really exists for entire year but not December because of tax-motivated selling behaviors. Weber and Camerer (1998) conduct a questionnaire to calculate the disposition coefficient for the disposition effect. The experiment of the disposition effect involves 3.

(9) buying and selling six hypothetical stocks in the course of fourteen trading rounds. The results indicate that subjects are about fifty percent more likely to realize gains compared to losses. Grinblatt and Keloharju (2001) use a regression model for assessing the disposition effect. This allows them to control for investor characteristics and market conditions. Besides, they find that different types of investors are likely to react to past returns in different ways. The regression analysis indicates that the marginal effect of distance is less for firms that are more nationally known, for distances that exceed one hundred kilometers, and for investors with more diversified portfolios.. 治政 Shapira and Venezia (2001) use method of round 大 trips duration to examine 立 disposition effect. The results show that both professional and independent investors ‧ 國. 學. exhibit the disposition effect, but independent investors display more strongly. In. ‧. addition, they compare trade frequency, volume and profitability between independent. sit. y. Nat. and professionally managed accounts. The results demonstrate that professionally. io. er. managed accounts were more diversified and round trips were both less correlated with. al. market and slightly more profitable than those of independent accounts.. n. iv n C Feng and Seasholes (2005) assume initial portfolio diversification and number h ethat ngchi U. of trading right are sophistication proxies and obtain similar results. These authors further demonstrate that a combination of financial sophistication and trading experience can help investors to correct their behavioral mistakes. They propose that sophistication and trading experience eliminate the reluctance to realize losses and reduce the propensity to realize gains. Dhar and Zhu (2006) analyze the disposition effect at individual level and contribute to the following consequences: First, trading frequency has negative relationship with the degree of the disposition effect. It means that the more experience, the less disposition effect exhibit. Second, wealthier and more professional investors display 4.

(10) lower disposition effect. Barber, Lee, Liu, and Odean (2007) analyze all trading activity on the Taiwan Stock Exchange. They find that in aggregate, investors are about twice likely to sell a stock if holding for a gain than a loss. Most of investors reflect the disposition effect. The authors also categorize investors into individuals, corporations, dealers, mutual funds and foreigners. Individuals, corporations, and dealers are reluctant to realize losses, while mutual funds and foreigners are not. Jhang (2008) examined the Electronic Index futures and Financial Index Futures in Taiwan futures markets and linked the disposition effect and the profitability of. 治政 individual trader. The results show that losers have more 大 disposition effect. Instead, 立 there is no disposition effect among winners. Besides, volume is the most dominant ‧ 國. 學. determinant of profitability. Trading volume is positively related to profits.. ‧. Unprofitable traders suffer more losses as they trade more suggests that traders need to. sit. y. Nat. know themselves well and be disciplined enough to quit when they are losing money.. io. er. Li, Lin, Cheng and Lai (2013) observe the different types of investors behave in. al. Taiwan futures markets. The authors find that the disposition effect truly exists in the. n. iv n C futures market. In addition, foreignhinstitutional investors e n g c h i U have less disposition effect than retail investors or proprietaries investors. Moreover, they also demonstrate that the lower disposition effect, the higher average returns.. Wang (2014) use Weber and Camerer (1998) method to estimate the disposition effect in Taiwan stock market during January 2007 to December 2010. The empirical results show that the disposition effect of individual investors during the financial crisis is less than other periods. Additionally, the financing investors’ disposition effect is larger than the short selling investors during financial crisis.. 5.

(11) 2.2 Overconfidence In addition to disposition effect, this study also considers whether investors demonstrate overconfidence. Odean (1998) tests whether the trading profits of discount brokerage customers are sufficient to cover their trading costs. The surprising finding is that not only do the securities that these investors buy not outperform the securities they sell by enough to cover trading costs, but on average the securities they buy underperform those they sell. This is the case even when trading is not apparently motivated by liquidity demands,. 政治大 Barber and Odean (2000) focus on individual investors. They find that individuals 立. tax-loss selling, portfolio rebalancing, or a move to lower-risk securities.. ‧ 國. 學. hold common stocks directly pay a tremendous performance penalty for active trading. Those that trade most earn an annual return of 11.4 percent, while the market returns. ‧. 17.9 percent. Overconfidence can explain high trading levels and the resulting poor. Nat. sit. y. performance of individual investors. Their central contribution is that they prove. n. al. er. io. trading is hazardous to investors’ wealth.. i Un. v. Barber and Odean (2001) use theoretical models predict that overconfident investors. Ch. engchi. trade excessively. The authors test this prediction by partitioning investors on gender. Psychological research demonstrates that, in areas such as finance, men are more overconfident than women. Thus, theory predicts that men will trade more excessively than women. They analyze the common stock investments of men and women. They document that men trade 45 percent more than women. Trading reduces men’s net returns by 2.65 percentage points a year as opposed to 1.72 percentage points for women. Dorn and Huberman (2005) ask a sample of individual investors whether they judge their past investment successes to be mainly due to their personal skills, though they do 6.

(12) not test whether these investors indeed attribute good returns to their skills and bad returns to other factors, which forms an essential component of self-attribution bias. This paper examines some of the causes for the apparent failure to buy and hold a welldiversified portfolio. The subjective investor attributes gleaned from the survey help explain the variation in actual portfolio and trading choices. Self-reported risk aversion is the single most important determinant of both portfolio diversification and turnover. Glaser and Weber (2007) indicate that theoretical models predict that overconfident investors will trade more than rational investors. Therefore, the authors directly test this hypothesis by correlating individual overconfidence scores with several measures of. 治政 trading volume of individual investors. They created 大 an internet questionnaire which 立 was designed to measure various facets of overconfidence (miscalibration, volatility ‧ 國. 學. estimates, better than average effect). The authors find that investors who think that. ‧. they are above average in terms of investment skills or past performance (but who did. sit. y. Nat. not have above average performance in the past) trade more. Measures of miscalibration. io. er. are, contrary to theory, unrelated to measures of trading volume.. al. Moore and Healy (2008) distinguish three varieties of overconfidence: (a). n. iv n C overestimation of one's actual performance, overplacement of one's performance h e n g (b) chi U relative to others, and (c) excessive precision in one's beliefs. Experimental evidence. shows that reversals of the first two, when they occur, tend to be on different types of tasks. On difficult tasks, people overestimate their actual performances but also mistakenly believe that they are worse than others; on easy tasks, people underestimate their actual performances but mistakenly believe they are better than others. The authors offer a straightforward theory that can explain these inconsistencies. Overprecision appears to be more persistent than either of the other two types of overconfidence, but its presence reduces the magnitude of both overestimation and overplacement. Graham, Harvey and Huang (2009) find that people are more willing to bet on their 7.

(13) own judgments when they feel skillful or knowledgeable. The authors investigate whether this "competence effect" influences trading frequency and home bias. They find that investors who feel competent trade more often and have more internationally diversified portfolios. In addition, they also find that male investors, and investors with larger portfolios or more education, are more likely to perceive themselves as that they are more competent than are female investors, and investors with smaller portfolios or less education. This paper also contributes to understanding the theoretical link between overconfidence and trading frequency. Besides, this paper offers a potential mechanism for the "better-than-average" aspect of overconfidence to influence trading frequency.. 治政 Overall, overconfident investors tend to perceive themselves 大 to be more competent, and 立 thus are more willing to act on their beliefs, leading to higher trading frequency. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 8. i Un. v.

(14) 3. DATA and METHODOLOGY 3.1 Data This study based on datasets from Taiwan Futures Exchange ( TAIFEX). TAIFEX provides several kinds of commodities including index futures, index options, stock options, gold futures and so on. If people want to trade these contracts, they should request brokers to deliver their orders. Due to electronic trading system, the market can trade quickly and smoothly during 5-hours transaction period. The market open from 8:45AM to 01:45PM.. 政治大. Investors can put their orders at 8:30AM. They should prepare margin and save at. 立. broker. Then the brokers would provide trading service.. ‧ 國. 學. During transaction period, in order to let the investors learn the potential tendency of the market, TAIFEX will update the best five bid and ask prices, quantities, open. ‧. prices, highest prices, lowest prices, and the latest market prices. If the margin is lower. y. Nat. io. sit. than the maintenance margin, the investors will be notified to refill the margin to the. n. al. er. level of initial margin, or the TAIFEX will automatically close the positions of the accounts.. Ch. engchi. i Un. v. The TAIFEX use electronic system to record all trades information at any time. Owing to the data of trading records obtained from TAIFEX, we can do analyst on disposition effect and performance of investors. According to time series data, each of trades contains trading date, futures commission merchant number, account number, identity code, order type, buy or sell code, strike price, quantity, flag of market, cancellation, or revision and so on. In this study, we mainly focus on the active futures contracts in the TAIFEX. The datasets consist of the futures contracts including the Taiwan Stock Index Futures (TXF), the Mini Taiwan Stock Index Futures (MXF), the Taiwan Electronic Sector 9.

(15) Index Futures (EXF), and the Taiwan Financial Sector Index Futures (FXF). Both TXF and MXF target on the Taiwan Stock Exchange Capitalization Weighted Stock Index. Most of investors would select these two kind of futures so that they have the most frequent trading of all kinds of futures. Even though TXF and MXF have the same tracing index, they still have difference between their tick sizes. Each of points on TXF is larger than MXF four times. However, although these two products are quite similar on basement, they attract different preference investors. EXF tracks the Electronic Industry Weighted Average Index in the stock market. In Taiwan, electronic industry plays an important role on GDP or export and represents. 治政 main economic situation. These companies are competitive 大 in global electronic 立 industries. As the results, when the prices of electronic stocks change, the Electronic ‧ 國. 學. Industry Weighted Average Index, even the stock market or the futures market, would. ‧. have a great impact.. sit. y. Nat. The last contract we consider is FXF. FXF follow the Financial Industry Weighted. io. er. Index in the stock market. Taiwan Stock Exchange started in 1961. After years, Taiwan. al. Futures Exchange was set up in 1996. Now, there are so many banks in Taiwan,. n. iv n C including forty banks, thirty local branches and the mainland Chinese banks h e n gofcforeign hi U. and sixty-two overseas banking units. In view of this, the financial industry in Taiwan is pretty mature and flourishing. Hence, FXF is also a crucial reference. The time period of data consist of the trading date of contracts which include May 2006 to December 2008. As a result of time series data, we could learn how the positions of each investors sometime. Moreover, investors could be broadly classified into domestic institutions, individuals, proprietaries, foreigners and others. We mainly focus on domestic institutions, individuals, proprietaries and foreigners in this study. 10.

(16) 3.2 Methodology Among numerous researches which have contributed to the development of examining the disposition effect, there are two primary methods in empirical analysis: (1) using questionnaire to collect data (e. g., Weber and Camerer (1998)) and (2) using secondary transaction data (e.g., Odean (1998)and Grinblatt and Keloharju (2001)). To analyze the disposition effect on different futures by time series data, we follow the idea of methodology of Shapira and Venezia (2001), but we revise the method to be simple and easy to understand.. 政治大 means the net positions of an investor from non-zero to zero. The following Table 3.2立. Before we discuss further, we should introduce definition of round trip. A round trip. ‧ 國. 學. 1 is a simple example.. Table 3.2-1. ‧. An Example of Simple Round Trip. sit. y. Nat. The positive net position means the investor belongs to buy side. Contrarily, if the net. io. al. n. round trip.. er. position is negative, it means the investor belongs to sell side. This example is a complete. ni C h Time Buy or U Sell engchi. Account number. Date. 12345678. May 4, 2015. 08：45：00. 12345678. May 4, 2015. 12345678 12345678. v. Quantity. Net position. Buy. 5. 5. 11：00：00. Sell. 2. 3. May 5, 2015. 10：00：00. Buy. 1. 4. May 5, 2015. 13：45：00. Sell. 4. 0. If the net position pass from positive to negative, it would be more complex. In this case, we would regard as that one round trip have completed and another round trip starts. The following Table 3.2-2 is another example of round trip.. 11.

(17) Table 3.2-2 An Example of Complex Round Trip. The positive net position means the investor belongs to buy side. Contrarily, if the net position is negative, it means the investor belongs to sell side. The round trip started at 9 o'clock on May 1, 2015. However, the net position turns out to be negative at 13：00 on May 2, 2015. For dealing with this problem, we will separate the order into two orders. The quantity of one order is to make the net position be zero. The quantity of the other one is the remaining quantity of original order. Panel A Original Trading Data. Account Number. Date. 12345678. May 4, 2015. 08：45：00. 201505. Buy. 5. 5. May 4, 2015. 11：00：00. 201505. Sell. 2. 3. May 5, 2015. 10：00：00. 201505. Sell. 1. 2. May 5, 2015. 13：45：00. 201505. Sell. 4. -2. Contract. Buy or Sell. ‧ 國. y. sit. Modified Trading Data. io. al. Date. 12345678. May 4, 2015. 08：45：00. 12345678. May 4, 2015. 11：00：00. 201505. 12345678. May 5, 2015. 10：00：00. 12345678. May 5, 2015. 12345678. May 5, 2015. n. Account Number. Time. er. Panel B. Net position. Nat. 12345678. Quantity. ‧. 12345678. Buy or Sell. 學. 12345678. Contract. 立. Time. 政治大. v ni. Quantity. Net position. 5. 5. Sell. 2. 3. 201505. Sell. 1. 2. 13：45：00. 201505. Sell. 2. 0. 13：45：00. 201505. Sell. 2. -2. Ch. e n 201505 g c h i UBuy. In addition, if the round trip does not complete until the closing date, we will regard the end of round trip as the point at 13:30 on the closing date. Next, we discuss how to calculate the duration of round trip. The duration of round trip is the time period between the beginning of round trip and the end of round trip. 12.

(18) Besides, if the round trip cross one date to another. Because of transaction period, we define that there is five hours in one day. For the example of Table 1, the duration of round trip is ten hours. However, we do research on not only disposition effect but also performance. The performance can help us to define the round trip is a winner or a loser round trip. We define winner round trip as round trips earning positive profit and loser round trip as a round trips getting losses. Moreover, we can know how professional the investors are by their net profits. The most important of assumption is that we don’t add the transaction cost in the process of analyzing the performance. To understand easily how. 治政 many points the investors earn or lose, the performance 大is reported on the results by 立 points, not dollars. ‧ 國. 學. If the average duration of winner round trips is smaller than average duration of loser. ‧. round trips, we will call that as disposition effect. Proprietaries and foreigners are. sit. y. Nat. professional investors, while domestic institutions and individuals are retail traders. As. io. To discuss whether. al. iv n C trading volume market condition h e nand gchi U. n. be different.. er. a result, we expect that the trading behaviors between these two kinds of group would. have influence on. disposition effect, we further separate individual investor group and our sample data. Because of individuals are the primary group of investors in Taiwan Futures Exchange. They trade most and provide liquidity to the market. We want to know whether individuals who have the higher trading volume will display professionally and earn more money. Therefore, we separate the individuals into three groups by their trading volumes. There are low, medium and high group. The high group means the investors in this group are the top third trading volumes of individuals. So, we expect that the high group would be professional like institutions. They are likely to exhibit less disposition effect and earn positive profit. Some researchers have suggested that 13.

(19) investors trade too much so that profit cannot cover trading cost and be negative, called as overconfidence ( Odean (1998), Barber and Odean (2000, 2001)). In addition, we are interested in the effect of financial crisis. The data we obtained supports us to compare the difference. In order to do further research, we separate the sample period by the happening date of financial crisis. According to the history, some researchers deem the liquidity crisis on August 9, 2007 as a signal of the following financial crisis. This liquidity crisis caused the Dow falls 387 points, or 2.8 percent. The decline began at the opening bell after BNP Paribas suspended operations of three of its funds in the wake of turmoil in the American market for home loans.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 14. i Un. v.

(20) 4. EMPIRICAL ANALYSIS In this section, there are two parts. One is overall trading activities. We discuss four investor type, including domestic institutions, individuals, proprietaries, and foreigners. The other one just focuses on individual trading activities. We separate individuals by their trading volumes. Each part consists of descriptive statistics, disposition effect and performance.. 4.1 All Trading Activities 4.1.1 Descriptive Statistics. 政治大. Table 4.1.1 illustrates the descriptive statistics of investors, trading volumes and. 立. round trips in different types of investors on TXF, MXF, EXF and FXF.. ‧ 國. 學. In Panel A of Table 4.1.1, the number of investors trades TXF are the largest among the other three contracts, followed by MXF, FXF and EXF. The reason why investors. ‧. tend to trade TXF and MXF is that both TXF and MXF trace the Taiwan Stock. y. Nat. io. sit. Exchange Capitalization Weighted Stock Index, which is core index in Taiwan stock. n. al. er. market. Moreover, because MXF is much cheaper than other contracts, the number of. Ch. investors participate in MXF is large.. engchi. i Un. v. From the Panel B of Table 4.1.1, we can learn that individuals are the main investors in futures market, followed by proprietaries, foreigners, and domestic institutions. The trading volume of individuals are the most on each of contracts. Especially, the percentage of trading volume number of individuals on MXF is about eighty-three percent. Besides, we notice that foreigners relatively seldom trade MXF.. 15.

(21) Table 4.1.1 Descriptive Statistics of Investors, Trading Volumes, and Round Trips in Different Investor Types Table 4.1.1 reports the descriptive statistics of investors, trading volumes and round trips in different investor types on TXF, MXF, EXF and FXF from May, 2006 to December, 2008.The investor types include domestic institutions, individuals, proprietaries, and foreigners. We also add the all of investors as a reference. All of investors means the total investors during sample period. In the Panel A, B and C, there are two parts every panel. One is for the number, the other is for the percentage. The number of investors is the accounts in each of investor types and the number of trading volume is the sum of the quantities of orders for every investor type. We count the number of round trips and categorize the round trips into winner round trips (Winners) and loser round trips (Losers) in different investor type. Additionally, we calculate the percentage for investors and trading volume. The percentage of investor number for each of investor types is the number of investors divided by the number of all investors. Likewise, the percentage of trading volume number for each of investor types is the number of trading volume divided by the number of all trading volume. There are. 政治大. two ways of calculating the percentage of round trip number. For the investor types, it is the number of round trip. 立. divided by all of round trips. However, for the round trip type, the percentage is that winner or loser round trips. 學. ‧ 國. account for the total round trips in one investor type. In the Panel D and E, there are average trading volume and average round trips for investors in different investor types. Panel A Number of Investors. Percentage of Investor Number. EXF. FXF. TXF. 501. 138. 138. 0.44%. Individuals. Nat. 230. 111,638. 109,942. 31,826. 34,423. 97.49%. Proprietaries. 52. 36. 44. 42. 0.05%. Foreigners. 115. 73. 78. All. 114,513. Ch. n. Panel B. al. 20. 111,799. i e32,966 n g c h35,432. EXF. FXF. 0.21%. 0.42%. 0.39%. 98.34%. 96.54%. 97.15%. 0.03%. 0.13%. 0.12%. 0.02%. 0.22%. 0.22%. 100.00%. 100.00%. 100.00%. er. Domestic Institutions. MXF. sit. MXF. io. TXF. y. Number of Investors. ‧. Investor Type. iv 0.10% n U. 100.00%. Number of Trading Volume. Investor Type. Number of Trading Volume. Percentage of Trading Volume Number. TXF. MXF. EXF. FXF. TXF. MXF. EXF. FXF. Domestic Institutions. 445,721. 31,283. 19,343. 18,580. 0.59%. 0.12%. 0.29%. 0.34%. Individuals. 49,212,486. 21,175,409. 4,092,798. 3,456,358. 65.45%. 83.31%. 62.08%. 63.54%. Proprietaries. 14,128,684. 3,803,705. 941,078. 700,808. 18.79%. 14.97%. 14.27%. 12.88%. Foreigners. 6,780,976. 71,589. 1,175,836. 1,035,865. 9.02%. 0.28%. 17.84%. 19.04%. All. 75,192,645. 25,416,432. 6,592,850. 5,440,060. 100.00%. 100.00%. 100.00%. 100.00%. 16.

(22) Panel C. Number of Round Trips. Investor Type. Number of Round Trips. Percentage of Round Trip Number. TXF. MXF. EXF. FXF. TXF. MXF. EXF. FXF. Domestic Institutions. 23,245. 7,284. 2,019. 1,495. 0.31%. 0.12%. 0.24%. 0.21%. Winners. 12,878. 3,958. 1,221. 770. 55.40%. 54.34%. 60.48%. 51.51%. Losers. 10,367. 3,326. 798. 725. 44.60%. 45.66%. 39.52%. 48.49%. Individuals. 7,533,136. 5,874,480. 843,248. 706,070. 99.14%. 99.72%. 98.43%. 98.28%. Winners. 4,367,320. 3,339,351. 526,853. 436,633. 57.97%. 56.85%. 62.48%. 61.84%. Losers. 3,165,816. 2,535,129. 316,395. 269,437. 42.03%. 43.15%. 37.52%. 38.16%. Proprietaries. 30,768. 8,856. 8,925. 8,384. 0.40%. 0.15%. 1.04%. 1.17%. Winners. 16,267. 5,332. 4,861. 4,213. 52.87%. 60.21%. 54.46%. 50.25%. Losers. 14,501. 3,524. 4,064. 4,171. 47.13%. 39.79%. 45.54%. 49.75%. Foreigners. 11,606. 159. 2,509. 2,510. 0.15%. 0.00%. 0.29%. 0.35%. Winners. 6,022. 86. 1,232. 1,224. 51.89%. 54.09%. 49.10%. 48.76%. Losers. 5,584. 1,277. 1,286. 48.11%. 45.91%. 50.90%. 51.24%. All. 7,598,755. 5,890,779. 856,701. 718,459. 100.00%. 100.00%. 100.00%. 100.00%. 4,402,487. 3,348,727. 534,167. 442,840. 57.94%. 56.85%. 62.35%. 61.64%. 3,196,268. 2,542,052. 322,534. 275,619. 42.06%. 43.15%. 37.65%. 38.36%. Winners Losers. 73. 學. ‧ 國. 立. 政治大. ‧. Panel D Average Trading Volume. Average Trading Volume. Foreigners All. al. n. Proprietaries. io. Individuals. 889.66. 136.01. 440.82. 192.61. 271,705.46. 105,658.47. C 58,965.01 3,579.45 U hengch i 656.63 227.34. 17. EXF. y. MXF. sit. Nat. Domestic Institutions. TXF. v ni. FXF. 140.17. 134.64. 128.60. 100.41. 21,388.14. 16,685.90. 16,107.34. 13,280.32. 199.99. 153.54. er. Investor Type.

(23) Panel E. Average Round Trips Investor Type. Average Round Trips TXF. MXF. EXF. FXF. Domestic Institutions. 46.40. 31.67. 14.63. 10.83. Winners. 25.70. 17.21. 8.85. 5.58. Losers. 20.69. 14.46. 5.78. 5.25. Individuals. 67.48. 53.43. 26.50. 20.51. Winners. 39.12. 30.37. 16.55. 12.68. Losers. 28.36. 23.06. 9.94. 7.83. Proprietaries. 591.69. 246.00. 202.84. 199.62. Winners. 312.83. 148.11. 110.48. 100.31. Losers. 278.87. 97.89. 92.36. 99.31. Foreigners. 100.92. 7.95. 34.37. 32.18. Winners. 52.37. 4.30. 16.88. 15.69. 48.56. 3.65. 17.49. 16.49. 66.36. 52.69. 25.99. 20.28. 38.45. 29.95. 27.91. 22.74. 立. Losers. Losers. 學. Winners. ‧ 國. All. 政治大. 16.20. 12.50. 9.78. 7.78. ‧. In the Panel C of Table 4.1.1, individuals still have the largest number of round trips,. sit. y. Nat. followed by proprietaries, foreigners, and domestic institutions. Overall, winner round. io. er. trips are more than loser round trips. This result satisfies our expectation. The reason is. al. that when investor face winners, they have more probability to realize than losers.. n. iv n C In Panel D and Panel E of Table h e4.1.1, n gaccording c h i Uto that both proprietaries and. foreigners are professional institutions, they have higher average trading volume and average round trips than the other two investor types. Therefore, the results don’t conflict with our imagination.. 4.1.2 Disposition Effect Table 4.1.2 shows the statistics on duration of winner round trip, duration of loser round trip, and disposition effect of different investor types on TXF, MXF, EXF and FXF from May, 2006 to December, 2008. In the Panel A of Table 4.1.2, we compare the duration of every investor type. Overall, 18.

(24) foreigners have the biggest amount of duration, followed by proprietaries, domestic institutions, and individuals. On account of that institutional investors possess more funds, once they set up their target (short or long position), they will hold on until the next clear reversal news. According to the results, the duration of institutions are roughly about 1200 minutes or 4 days. Speaking of foreigners, both their winner and loser duration on MXF are much smaller than the other three contracts. We infer that foreigners are not familiar with MXF and seldom trade this contract. So, they will realize gains or losses at a rapid rate to prevent the unexpected event.. 治政 In contrast, individuals tend to realize their gains or losses 大 so that amount of duration 立 would be the lowest among four investor types. ‧ 國. 學. In the Panel B of Table 4.1.2, no matter on TXF, MXF, EXF or FXF, we can observe. ‧. that foreigners don’t display the disposition effect. In addition, the proprietaries exhibit. sit. y. Nat. a little bit of disposition effect on TXF, EXF and FXF, but not MXF. The results show. io. er. that On the other hand, domestic institutions and individuals clearly exhibit the. al. disposition effect. They are reluctant to realize their losses than gains. Especially, we. n. iv n C can find that individuals reveal thehhighest disposition e n g c h i U effect in all of investor types.. The results are consistent with previous researches, indicating that foreigners have the lowest disposition effect and individuals exhibit disposition effect most. The findings confirm the expectation that proprietaries and foreigners are more professional.. 19.

(25) Table 4.1.2 Disposition Effect of Different Investor Types Table 4.1.2 illustrates the disposition effect of different investor types on TXF, MXF, EXF and FXF from May 2006 to December 2008. In the Panel A, we calculate the mean winner round trip duration and mean loser round trip duration for every investor. Then, we separately average the mean winner round trip duration and the mean loser round trip duration by investor account for different investor types. To simplify the name of items, we use winner duration to represent the average the mean winner round trip duration by investor account. Similarly, loser duration represents the average the mean loser round trip duration by investor account. The unit of duration is minute. In the Panel B, we provide two method to evaluate whether the investors exhibit the disposition effect. If the winner duration is smaller than loser duration, it will show that the investor display the disposition effect. According to this, we use the difference duration between winner and loser and the duration ratio of winner to loser to test the existence of disposition effect. As a result, if the difference duration between winner and loser is negative, we will. 政治大. say that the investor exhibit disposition effect. Likewise, if the duration ratio are smaller than one, we will regard. 學. Investor Type. ‧ 國. Panel A Round Trip Duration. Winner Duration (by investor account). (by investor account). ‧. (1) EXF. FXF. TXF. Domestic Institutions. 1,388.46. 1,334.69. 1,178.34. 1,157.81. 1,729.94. Individuals. 595.10. 613.85. 661.81. 652.35. 1,036.88. Proprietaries. 2,184.89. 4,008.31. 1,883.60. 1,732.96. 2,214.93. Foreigners. 4,169.90. All. 622.02. MXF. EXF. FXF. 1,383.03. 1,721.45. 1,539.23. 1,061.95. 1,097.38. 1,127.63. 1,987.37. 1,966.25. 1,951.75. 885.75. 3,697.85. 3,508.54. 1,064.82. 1,133.95. 1,158.61. er. io. 4,052.13 3,865.30 3,095.62 a1,914.98 iv l620.95C 715.36 689.68 1,068.14 n hengchi U. n. Disposition Effect. (2). sit. MXF. Nat. TXF. Panel B. Loser Duration. y. 立. that as the existence of disposition effect.. Disposition Effect Difference Duration between Winner and Loser. Duration Ratio of Winner to Loser. (3). (4). = (1) - (2). = (1) / (2). Investor Type. TXF. MXF. EXF. FXF. TXF. MXF. EXF. FXF. Domestic Institutions. -341.48. -48.33. -543.11. -381.42. 0.80. 0.97. 0.68. 0.75. Individuals. -441.78. -448.10. -435.56. -475.28. 0.57. 0.58. 0.60. 0.58. Proprietaries. -30.05. 2,020.94. -82.65. -218.79. 0.99. 2.02. 0.96. 0.89. Foreigners. 1,074.28. 1,029.23. 354.28. 356.77. 1.35. 2.16. 1.10. 1.10. All. -446.12. -443.86. -418.58. -468.93. 0.58. 0.58. 0.63. 0.60. 20.

(26) 4.1.3 Performance Table 4.1.3 shows the performance of trading activities on TXF, MXF, EXF and FXF in different investor types from May, 2006 to December, 2008. The results from Panel A of Table 4.1.3 indicates that round trips belong to institutional investors, including domestic institutions, proprietaries, and foreigners, have larger amount of profits or losses than individuals on average. Because the institutions own more funds, their positions would be larger. It causes their profits or losses of round trips are bigger than individuals’. In addition, we find that domestic institutions, individuals, and proprietaries prefer to hold larger losers than winners.. 治政 When facing losses, they are likely to hold for a long大 time. On the other hand, when 立 facing gains, they would realize quickly. In contrast, foreigners are totally different ‧ 國. 學. from the other three investor types. We infer the reason is that foreigners are more. sit. y. Nat. quickly.. ‧. rational and professional so that they will admit their faults and close their positions. io. er. Panel B of Table 4.1.3 presents that the average net profit by investor account in. al. different investor types. The retail investors, including domestic institutions and. n. iv n C individuals make losses on most h futures contracts. U e n g c h i This finding is consistent with previous studies.. 21.

(27) Table 4.1.3 Performance of Round Trips and Net Profit Table 4.1.3 primarily describes the performance of different round trip types and different investor types on TXF, MXF, EXF and FXF from May, 2006 to December, 2008. In the Panel A, we calculate the mean profit of winner round trip and the mean loss of loser round trip for every investors. Then, we average the mean profit of winner round trip and mean loss of loser round trip by investor account for different investor types. To simplify the name of items, we use winner profit to represent the average the mean profit of winner round trip by investor account. Similarly, loser loss represents the average the mean loss of loser round trip by investor account. The unit of profit or loss is point. In the Panel B, we calculate the average net profit by investor account for different investor types. Panel A Round Trip Performance Winner Profit. Loser Loss. (by investor account). (by investor account). Investor Type. MXF. Domestic Institutions. 6,665.81. Individuals. 965.76. TXF. MXF. EXF. FXF. 2,468.46. 104.41. 450.47. -7,967.95. -2,655.44. -314.61. -1,031.44. 644.30. 43.19. 111.23. -1,382.87. -898.65. -53.79. -224.36. 1,314.09. 2,255.98. -1,914.10. -3,484.56. 102,807.89 200,149.01. -131,839.10 -251,711.47. 782,165.06 393,001.05 21,145.07 62,301.00 -762,955.37 286.18. Nat. Investor Type. 121.36. -2,695.80. Net Profit (by investor account). io. TXF. al 4,213.79. MXF. n Domestic Institutions Individuals. 1,064.82. Ch -3,053.86. -3,067.65. e n g1,560.73 chi. -16,130.95 -39,943.27 -106.08. -357.37. y. Performance. 620.95. er. 2,746.27. -80,658.21. ‧. All Panel B. FXF. 立. ‧ 國. Foreigners. EXF. 學. Proprietaries. 政治大. sit. TXF. EXF. i Un. v -248.99. FXF -1,750.41. -75.31. -447.08. -5,594,991.61. 1,100.91. -102,889.01. Proprietaries. -6,002,352.15. Foreigners. 894,009.30. 1,403,090.55. -21,333.89. 231,250.17. All. 0.00. 0.00. 0.00. 0.00. There is something special to speak of. We can learn that the foreigners earn the appreciable profits in most futures contracts, except for EXF. However, compared with foreigners, proprietaries lose enormous amount of money on trading of TXF, MXF and FXF. As we know from previous researches, although both proprietaries and foreigners are professional and trade a lots on futures, their net trading profits are completely opposite. It appears that proprietaries obviously exhibit the overconfidence The 22.

(28) possible reason is that the proprietaries and foreigners frequently hold opposite positions at the same time, but foreigners win at most of time. We also think that the points they hold long or short positions may cause different results. Because our sample period contains the period of financial crisis, the index probably go down sharply within a day and affect their net profits of positions. Therefore, we further discuss about whether investors behave differently owing to financial crisis.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 23. i Un. v.

(29) 4.2 Trading Activities of Individuals by Volume 4.2.1 Descriptive Statistics Table 4.2.1 illustrates the descriptive statistics of investors, trading volumes and round trips for individual investors on TXF, MXF, EXF and FXF. To examine whether part of individuals trade professionally like institutions, we further divide individual investors into three groups by trading volumes. There are low, medium and high group. The last third of individual investors who have the lowest trading volumes are defined as low group. The second third of individual investors who. 政治大 individual investors who have the highest trading volumes are defined as high group. 立. have moderate trading volumes are defined as medium group. The top third of. ‧ 國. 學. So, in the Panel A of Table 4.2.1, the number of investors of each of three groups is about equivalent.. ‧. As we expect, from Panel B of Table 4.2.1, high group has the highest number of. Nat. sit. y. trading volume and accounts for a large proportion of all individual trading volume.. n. al. er. io. The trading volume of high group is so far away from the other two groups. Due to this. i Un. v. finding, we consider that high group trade more like institutions so that we expect that. Ch. engchi. their trading behaviors will be more professional.. 24.

(30) Table 4.2.1 Descriptive Statistics of Investors, Trading Volumes, and Round Trips for Different Individual Groups Table 4.2.1 reports the descriptive statistics of investors, trading volumes and round trips for individuals on TXF, MXF, EXF and FXF from May, 2006 to December, 2008. We categorize individuals into three levels of groups by their trading volumes. High group means their trading volumes are the top third in all individual investors. The other two groups are medium and low. We also add the all of investors as a reference. All of investors means the all of individual investors during sample period. In the Panel A, B and C, there are two parts every panel. One is for the number, the other is for the percentage. The number of investors is the accounts in each of individual groups and the number of trading volume is the sum of the quantities of orders for every individual group. We count the number of round trips and categorize the round trips into winner round trips (Winners) and loser round trips (Losers) in different individual groups. Additionally, we calculate the percentage for investors and trading volume. The percentage of investor number for each of individual groups is the number of investors divided by the number of all individual investors. Likewise, the. 政治大. percentage of trading volume number for each of investor types is the number of trading volume divided by the. 立. number of all individual trading volume. There are two ways of calculating the percentage of round trip number. For the individual groups, it is the number of round trip divided by all of individual round trips. However, for the. ‧ 國. 學. round trip type, the percentage is that winner or loser round trips account for the total round trips in one individual group.. individual groups.. ‧. In the Panel D and E, there are average trading volume and average round trips for investors in different. y EXF. al. 10,610. Percentage of Investor Number FXF. sit. MXF. io. TXF. Number of Investors TXF. 37,214. Medium. 37,212. High. 37,212. iv n 36,647 C 10,608 11,474 33.33% hengchi U 36,647 10,608 11,474 33.33%. All. 111,638. 109,942. Panel B. n. Low. 36,648. 11,475. 31,826. 34,423. MXF. EXF. FXF. 33.33%. 33.33%. 33.33%. 33.33%. 33.33%. 33.33%. 33.33%. 33.33%. 33.33%. 100.00%. 100.00%. 100.00%. er. Group. Nat. Panel A Number of Investors. 33.33%. 100.00%. Number of Trading Volume. Group. Number of Trading Volume. Percentage of Trading Volume. TXF. MXF. EXF. FXF. TXF. MXF. EXF. FXF. Low. 4,974,725. 3,859,325. 553,927. 463,859. 10.11%. 18.23%. 13.53%. 13.42%. Medium. 5,821,299. 3,916,320. 562,166. 470,714. 11.83%. 18.49%. 13.74%. 13.62%. High. 38,416,462. 13,399,764. 2,976,705. 2,521,785. 78.06%. 63.28%. 72.73%. 72.96%. All. 49,212,486. 21,175,409. 4,092,798. 3,456,358. 100.00%. 100.00%. 100.00%. 100.00%. 25.

(31) Panel C Number of Round Trips Group. Number of Round Trips. Percentage of Round Trips. TXF. MXF. EXF. FXF. TXF. MXF. EXF. FXF. 127,834. 113,337. 16,357. 16,950. 1.70%. 1.93%. 1.94%. 2.40%. Winner. 63,095. 56,407. 8,178. 8,629. 49.36%. 49.77%. 50.00%. 50.91%. Loser. 64,739. 56,930. 8,179. 8,321. 50.64%. 50.23%. 50.00%. 49.09%. Medium. 815,385. 671,088. 67,248. 64,130. 10.82%. 11.42%. 7.97%. 9.08%. Winner. 459,834. 375,138. 39,185. 37,714. 56.39%. 55.90%. 58.27%. 58.81%. Loser. 355,551. 295,950. 28,063. 26,416. 43.61%. 44.10%. 41.73%. 41.19%. High. 6,589,917. 5,090,055. 759,643. 624,990. 87.48%. 86.65%. 90.09%. 88.52%. Winner. 3,844,391. 2,907,806. 479,490. 390,290. 58.34%. 57.13%. 63.12%. 62.45%. Loser. 2,745,526. 2,182,249. 280,153. 234,700. 41.66%. 42.87%. 36.88%. 37.55%. All. 7,533,136. 5,874,480. 843,248. 706,070. 100.00%. 100.00%. 100.00%. 100.00%. Winner. 4,367,320. 3,339,351. Loser. 3,165,816. 2,535,129. I Low Round Trip Type. 立. Panel D Number of Trading Volume. 政治大 526,853. 436,633. 57.97%. 56.85%. 62.48%. 61.84%. 316,395. 269,437. 42.03%. 43.15%. 37.52%. 38.16%. FXF. 105.31. 52.21. 40.42. 156.44. 106.87. 52.99. 41.02. 1,032.37. 365.64. 280.61. 219.78. 440.82. 192.61. y. 133.68. 100.41. 128.60. Number of Round Trips. io. Group. n. al. Low Winner. er. Panel E. Nat. All. MXF. sit. High. EXF. ‧. Medium. TXF. 學. Low. Number of Trading Volume. ‧ 國. Group. Number of Round Trips. i n C 3.44 U h e n g c h 3.09 i TXF. MXF. v. EXF. FXF. 1.54. 1.48. 1.70. 1.54. 0.77. 0.75. Loser. 1.74. 1.55. 0.77. 0.73. Medium. 21.91. 18.31. 6.34. 5.59. Winner. 12.36. 10.24. 3.69. 3.29. Loser. 9.55. 8.08. 2.65. 2.30. High. 177.09. 138.89. 71.61. 54.47. Winner. 103.31. 79.35. 45.20. 34.02. Loser. 73.78. 59.55. 26.41. 20.45. All. 67.48. 53.43. 26.50. 20.51. Winner. 39.12. 30.37. 16.55. 12.68. Loser. 28.36. 23.06. 9.94. 7.83. 26.

(32) We observe a reasonable consequence from Panel B and Panel C that there is a positive relationship between trading volume and round trip. Moreover, the results from Panel C of Table 4.2.1 indicate that difference of number between winner round trips and loser round trips increase as trading volume increases. For high and medium group, their winner round trips are obviously more than loser round trips. So, in the Panel D and E, we can obtain the same conclusion that the higher trading volume, the more round trips.. 4.2.2 Disposition Effect Table 4.2.2 shows the statistics on duration of winner round trip, duration of loser. 治政 round trip, and disposition effect of different individual 大groups by trading volume on 立 TXF, MXF, EXF and FXF from May, 2006 to December, 2008. ‧ 國. 學. On average, we can learn from Panel A of Table 4.2.2 that the average durations of. ‧. these three individual groups are about or lower than two days. Additionally, high group. sit. y. Nat. have the lowest amount of duration, followed by low and medium. This result means. io. er. that investors in high group realize their positions quickly, no matter these positions are. al. winners or losers. The possible reason is that most of individual investors in high group. n. iv n C choose day trade strategy so that thehaverage durationU e n g c h i is relatively low.. 27.

(33) Table 4.2.2 Disposition Effect of Different Individual Groups. Table 4.2.2 illustrates the disposition effect of different individual groups by trading volumes on TXF, MXF, EXF and FXF from May 2006 to December 2008. In the Panel A, we calculate the mean winner round trip duration and mean loser round trip duration for every individual investor. Then, we separately average the mean winner round trip duration and the mean loser round trip duration by investor account for different individual groups. To simplify the name of items, we use winner duration to represent the average the mean winner round trip duration by investor account. Similarly, loser duration represents the average the mean loser round trip duration by investor account. The unit of duration is minute. In the Panel B, we provide two method to evaluate whether the investors exhibit the disposition effect. If the winner duration is smaller than loser duration, it will show that the investor display the disposition effect. According to this, we use the difference duration between winner and loser and the. 政治大 duration between winner and loser is negative, we will say that the investor exhibit disposition effect. 立 Likewise, if the duration ratio are smaller than one, we will regard that as the existence of disposition. duration ratio of winner to loser to test the existence of disposition effect. As a result, if the difference. ‧ 國. 學. effect.. Panel A Round Trip Duration. Loser Duration. (by investor account). (by investor account). MXF. EXF. FXF. TXF. MXF. EXF. FXF. 669.62. 480.77. 470.11. 1147.47. 1165.36. 820.38. 793.36. Medium. 665.92. 712.80. 757.40. 717.91. 1150.45. 1236.44. 1236.77. 1253.67. High. 487.29. 459.14. 1234.95. 1335.86. All. 595.10. 613.85. 1,097.38. 1,127.63. Disposition Effect. sit. er. n. Panel B. a l747.26 661.81 Ch. y. 632.09. io. Low. Nat. TXF. Winner Duration. ‧. Group. v784.07 i 652.35 1,036.88 n 1,061.95 engchi U 769.04. 812.72. Disposition Effect Difference Duration between Winner and Loser. Duration Ratio of Winner to Loser. (3). (4). = (1) - (2). = (1) / (2). Investor Type. TXF. MXF. EXF. FXF. TXF. MXF. EXF. FXF. Low. -515.38. -495.74. -339.61. -323.25. 0.55. 0.57. 0.59. 0.59. Medium. -484.53. -523.63. -479.38. -535.76. 0.58. 0.58. 0.61. 0.57. High. -325.44. -324.93. -487.69. -566.82. 0.60. 0.59. 0.61. 0.58. All. -441.78. -448.10. -435.56. -475.28. 0.57. 0.58. 0.60. 0.58. 28.

(34) The results of Panel B of Table 4.2.2 demonstrate that the individual investors in high group show less disposition effect on TXF and MXF. In contrast, the individual investors in low group exhibit less disposition effect on EXF and FXF. This consequence tell us that the individuals in high group are more rational on TXF and MXF, however, the investors in low group display more rationally on EXF and FXF. Overall, the duration ratios of different individual groups are very close. It represents that trading behaviors of individuals are similar. They have tendency to realize losses than gains about 1.7 times. Even so, we still can discover that there is a relationship between trading volume and disposition effect. Comparatively, individual investors. 治政 possessing higher trading volume would have weaker 大disposition effect, or more 立 rational trading behaviors. ‧ 國. 學. 4.2.3 Performance. ‧. Table 4.2.3 describes the performance of trading activities on TXF, MXF, EXF and. sit. y. Nat. FXF in different individual groups from May, 2006 to December, 2008.. io. er. As the previous results of tables, because there is a positive relationship between. al. number of round trip and trading volume, we can know that there would be a higher. n. iv n C volume per round trip for each of the h investors i Ugroup. So, as we expect, we can e n g cinhhigh obtain in Panel A of Table 4.2.3 that the profit or loss per round trip and trading volume have the similar positive relationship. Furthermore, the average size of profit or loss per round trip for high group is over several times than the other two individual groups. It appears that the investors who have higher trading volume will realize their positions as they accumulate a larger size of profit or loss.. 29.

(35) Table 4.2.3 Performance of Round Trips and Net Profit for Different Individual Groups Table 4.2.3 primarily describes the performance of different round trip types and different individual groups on TXF, MXF, EXF and FXF from May, 2006 to December, 2008. In the Panel A, we calculate the mean profit of winner round trip and the mean loss of loser round trip for every investors. Then, we average the mean profit of winner round trip and mean loss of loser round trip by investor account for different individual groups. To simplify the name of items, we use winner profit to represent the average the mean profit of winner round trip by investor account. Similarly, loser loss represents the average the mean loss of loser round trip by investor account. The unit of profit or loss is point. In the Panel B, we calculate the average net profit by investor account for different individual groups.. Panel A Round Trip Performance Winner Profit. Loser Loss. (by investor account). (by investor account). Group. Low. 346.16. 459.18. Medium. 516.12. 539.73. High. 2,034.97. All. 965.76. TXF. MXF. EXF. FXF. 10.12. 24.93. -539.50. -694.00. -18.31. -64.76. 18.34. 44.57. -844.36. -872.19. -33.02. -123.73. 933.99. 101.11. 264.18. -2,764.72. -1,129.74. -110.04. -484.56. 644.30. 43.19. 111.23. -1,382.87. -898.65. -53.79. -224.36. 立. Net Profit. MXF. EXF. FXF. Low. -396.07. -10.21. -47.27. Medium. -1,214.34. High. -7,633.88. All. -3,053.86. -313.36. n. al. Ch. -1,344.92 6,423.04 e n1,560.73 gchi U. er. TXF. io. Nat. (by investor account). y. Group. FXF. sit. Performance. 政治大. ‧. Panel B. EXF. ‧ 國. MXF. 學. TXF. v ni. -24.90. -145.81. -190.80. -1,148.10. -75.31. -447.08. In the Panel B of Table 4.2.3, there are a little different consequence from four kinds of futures contracts. Most of futures contracts, including TXF, EXF, and FXF, perform worse and worse when trading more frequently. There is a negative relationship between performance and trading volume. That is what we call “overconfidence”. However, the performance of individuals is totally different on MXF. Their trading performance get more profitable with high trading volume. According to the findings, we can learn that the individual investors who trading a lot on MXF are professional like institutions. Even though both TXF and MXF target on the same index, the primary 30.

(36) group trading them are completely different. It represents that MXF attracts more professional and institution-like individual investors. The reason why MXF appeal to most of individual investors is perhaps that the margin is comparatively low with other futures contracts so that they can reduce exposure risk of futures positions.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 31. i Un. v.

(37) 5. Robustness To understand whether financial crisis indeed cause the behavioral bias of investors, we further separate our sample period into two periods by the happening date of financial crisis, August 9, 2007. The earlier period shown on table is before, then the later one is after. Therefore, we will focus on the heterogeneity of behaviors in different investor types before and during financial crisis in this section.. 5.1. Descriptive Statistics Table 5.1 illustrates the descriptive statistics of investors, trading volumes and round. 政治大. trips for different investor types on TXF, MXF, EXF and FXF before or during financial. 立. crisis.. ‧ 國. 學. Because the financial market in Taiwan steadily develop, we can observe on Panel A of Table 5.1 that the number of investors in most of investor types, except for. ‧. proprietaries, relatively grow up. In the Panel B of Table 5.1, the percentages of investor. Nat. sit. y. number in individuals and foreigners almost keep the same. However, the percentage. n. al. er. io. of investor number in domestic institutions increase visibly. Instead, the percentage in proprietaries goes down apparently.. Ch. engchi. i Un. v. The results from Panel C and D of Table 5.1 are that trading volumes of different investor types are larger on most of contracts, including TXF, MXF and FXF. However, EXF is relatively unattractive for retail investors like domestic institutions and individuals. Maybe the reason is the crowding out effect of investment. Because of the budget limit, when you put more on some of contracts, the other contracts would be crowed out so that trading volume decrease. Overall, it appears that investors actively participate in trading during financial crisis. Due to the change of participation on contracts, there are a little different percentage allocations on four contracts. Most significant change of all are on EXF and FXF. Individual investors put more weight on 32.

(38) FXF, but less on EXF. In contrast, proprietaries and foreigners put more weight on EXF, but less on FXF. For the descriptive statistic of round trips in Panel E and F, we can find that all of round trips in four investor types grow up. Because of the financial crisis, all investors are afraid of any fluctuations in the market so that they trade frequently and create more round trips. Additionally, there are some change on the relationship between winner and loser round trips during financial crisis. We notice that the percentage of winner round trips on FXF in domestic institutions is reversely higher than loser round trips. So do TXF in proprietaries and both MXF and EXF in foreigners. These results mean. 治政 that there are performance biases caused by financial crisis 大 in different investor types. 立 In the Panel E of Table 5.1, there are a clear increase on average trading volume in ‧ 國. 學. proprietaries and foreigners. This confirm our imagination of that relatively large. ‧. institutions would have more trading volume during financial crisis. Because they. sit. y. Nat. believe their professional and have more funds to support, large institutions are likely. io. er. to put more in the market. But for retail investors, namely domestic institutions and. al. individuals, they have more concern about their investments so that their trading. n. iv n C volume might increase or decrease h on different futuresUcontracts. engchi. Due to the relationship we find previously between trading volume and round trips, we can obtain the similar result that most of investors trade more and have more round trips during financial crisis.. 33.

(39) Table 5.1 Descriptive Statistics of Investors, Trading Volumes, and Round Trips Before or During Financial Crisis Table 5.1 reports the descriptive statistics of investors, trading volumes and round trips in different investor types on TXF, MXF, EXF and FXF before or during financial crisis. In order to analyze the effect of financial crisis, we divide our sample period. The point we separate our sample period into two periods is August 9, 2007. We use before and after to stand for earlier period and later period. The investor types include domestic institutions, individuals, proprietaries, and foreigners. We also add the all of investors as a reference. All of investors means the total investors during sample period. In the Panel A to F, we calculate both number and percentage for different investor types in two periods. The number of investors is the accounts in each of investor types and the number of trading volume is the sum of the quantities of orders for every investor type. We count the number of round trips and categorize the round trips into winner round trips (Winners) and loser round trips (Losers) in different investor type for two sample periods. If he round trips cross over August 9, 2007, we will put them to the earlier period. The reason is that we analyze the. 政治大. percentage of round trip which cross over is so low that we can neglect the effect. Additionally, we calculate the. 立. percentage for investors and trading volume. The percentage of investor number for each of investor types is the number of investors divided by the number of all investors. Likewise, the percentage of trading volume number for. ‧ 國. 學. each of investor types is the number of trading volume divided by the number of all trading volume. There are two ways of calculating the percentage of round trip number. For the investor types, it is the number of round trip divided. ‧. by all of round trips. However, for the round trip type, the percentage is that winner or loser round trips account for the total round trips in one investor type.. y. Nat. In the Panel G and H, there are average trading volume and average round trips for investors for different investor. n. al. er. io. Panel A Number of Investors. sit. types in two sample periods.. Investor Type. TXF Before. v i EXF n U. Number of Investors. Ch. After. MXF. e hi n g cAfter Before. FXF. Before. After. Before. After. Domestic Institutions. 271. 332. 77. 184. 72. 89. 72. 88. Individuals. 67,902. 77,072. 54,144. 86,209. 19,242. 19,637. 20,065. 22,302. Proprietaries. 51. 41. 31. 30. 41. 38. 38. 39. Foreigners. 83. 92. 6. 16. 51. 55. 55. 62. All. 69,655. 78,985. 55,106. 87,551. 19,925. 20,354. 20,678. 22,948. Panel B. Percentage of Investor Number Percentage of Investor Number. Investor Type. TXF. MXF. EXF. FXF. Before. After. Before. After. Before. After. Before. After. Domestic Institutions. 0.39%. 0.42%. 0.14%. 0.21%. 0.36%. 0.44%. 0.35%. 0.38%. Individuals. 97.48%. 97.58%. 98.25%. 98.47%. 96.57%. 96.48%. 97.04%. 97.18%. Proprietaries. 0.07%. 0.05%. 0.06%. 0.03%. 0.21%. 0.19%. 0.18%. 0.17%. Foreigners. 0.12%. 0.12%. 0.01%. 0.02%. 0.26%. 0.27%. 0.27%. 0.27%. All. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 34.

(40) Panel C Number of Trading Volume Number of Trading Volume Investor Type. TXF. Domestic Institutions. MXF. EXF. FXF. Before. After. Before. After. Before. After. Before. After. 222,888. 222,833. 6,347. 24,936. 12,865. 6,478. 6,817. 11,763. Individuals. 18,771,551 30,440,935 4,829,244 16,346,165 2,101,263 1,991,535 1,446,669 2,009,689. Proprietaries. 5,700,507. 8,428,177 868,624. Foreigners. 2,389,504. 4,391,472. All. 2,935,081. 468,541. 472,537. 327,159. 373,649. 70,857. 531,112. 644,724. 456,659. 579,206. 18,019. 28,684,788 46,507,857 5,776,673 19,639,759 3,316,266 3,276,584 2,323,612 3,116,448. Panel D Percentage of Trading Volume Percentage of Trading Volume Investor Type. TXF Before. MXF. EXF. 政治大. After. Before. After. FXF. Before. After. Before. After. 0.78%. 0.48%. 0.11%. 0.13%. 0.39%. 0.20%. 0.29%. 0.38%. Individuals. 65.44%. 65.45%. 83.60%. 83.23%. 63.36%. 60.78%. 62.26%. 64.49%. Proprietaries. 19.87%. 18.12%. 15.04%. 14.94%. 14.13%. 14.42%. 14.08%. 11.99%. Foreigners. 8.33%. 9.44%. 0.31%. 0.36%. 16.02%. 19.68%. 19.65%. 18.59%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. Number of Round Trips. ‧ FXF. y. EXF. Before. After. Before. After. Before. After. 9,063. 14,182. 1,532. 5,652. 996. 1,023. 715. 780. 7,867. 858. 3,100. 620. 399. 371. 403. 316. 409. al. 5,011. sit. After. n. Winner. MXF. Before. io. Domestic Institutions. TXF. Nat. Investor Type. Number of Round Trips. v601 i n 395. C6,315 h e n g674c h i2,552U 4,900,441 1,313,728 4,560,752. Loser. 4,052. Individuals. 2,632,695. Winner. 1,544,017. 2,823,303. 763,688. Loser. 1,088,678. 2,077,138. Proprietaries. 11,584. Winner. er. Panel E. ‧ 國. All. 立. 學. Domestic Institutions. 418,157. 425,091. 288,704 417,366. 2,575,663. 262,513. 264,340. 178,065 258,568. 550,040. 1,985,089. 155,644. 160,751. 110,639 158,798. 19,184. 3,716. 5,140. 3,314. 5,611. 3,038. 5,346. 5,706. 10,561. 2,316. 3,016. 1,792. 3,069. 1,445. 2,768. Loser. 5,878. 8,623. 1,400. 2,124. 1,522. 2,542. 1,593. 2,578. Foreigners. 2,428. 9,178. 9. 150. 1,065. 1,444. 1,032. 1,478. Winner. 1,220. 4,802. 4. 82. 494. 738. 488. 736. Loser. 1,208. 4,376. 5. 68. 571. 706. 544. 742. All. 2,655,770. 4,942,985 1,318,985. 4,571,694. 423,532. 433,169. 293,489 424,970. Winners. 1,555,954. 2,846,533. 766,866. 2,581,861. 265,400. 268,767. 180,397 262,443. Losers. 1,099,816. 2,096,452. 552,119. 1,989,833. 158,132. 164,402. 113,092 162,527. 35.

(41) Panel F Percentage of Round Trips Percentage of Round Trips TXF. MXF. EXF. FXF. After. Before. After. Before. After. Before. After. Domestic Institutions. 0.34%. 0.29%. 0.12%. 0.12%. 0.24%. 0.24%. 0.24%. 0.18%. Winner. 55.29%. 55.47%. 56.01%. 54.85%. 60.34%. 60.61%. 55.80%. 47.56%. Loser. 44.71%. 44.53%. 43.99%. 45.15%. 39.66%. 39.39%. 44.20%. 52.44%. Individuals. 99.13%. 99.14%. 99.60%. 99.76%. 98.73%. 98.14%. 98.37%. 98.21%. Winner. 58.65%. 57.61%. 58.13%. 56.47%. 62.78%. 62.18%. 61.68%. 61.95%. Loser. 41.35%. 42.39%. 41.87%. 43.53%. 37.22%. 37.82%. 38.32%. 38.05%. Proprietaries. 0.44%. 0.39%. 0.28%. 0.11%. 0.78%. 1.30%. 1.04%. 1.26%. Winner. 49.26%. 55.05%. 62.33%. 58.68%. 54.07%. 47.56%. 51.78%. Loser. 50.74%. 44.95%. 37.67%. 41.32%. 45.93%. 45.30%. 52.44%. 48.22%. Foreigners. 0.09%. 0.19%. 政治大. 54.70%. 0.00%. 0.00%. 0.25%. 0.33%. 0.35%. 0.35%. Winner. 50.25%. 52.32%. 44.44%. 54.67%. 46.38%. 51.11%. 47.29%. 49.80%. 49.75%. 47.68%. 55.56%. 45.33%. 53.62%. 48.89%. 52.71%. 50.20%. All. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. 100.00%. Winners. 58.59%. 57.59%. 58.14%. 56.47%. 62.66%. 62.05%. ‧. 61.47%. 61.76%. 41.41%. 42.41%. 41.86%. 43.53%. 37.34%. 37.95%. 38.53%. 38.24%. io TXF. Average Number of Trading Volume. al. iv nBefore. MXF. n. Investor Type. sit. Average Number of Trading Volume. er. Panel G. Nat. Losers. y. Loser. 立. 學. Before. ‧ 國. Investor Type. EXF. FXF. Domestic Institutions. 822.46. C h Before After U e82.43n g c135.52 h i 178.68 671.18. Individuals. 276.45. 394.97. 89.19. 189.61. 109.20. 101.42. Proprietaries. 111,774.65. 205,565.29. 28,020.13. 97,836.03. 11,427.83. 12,435.18. 8,609.45 9,580.74. Foreigners. 28,789.20. 47,733.39. 3,003.17. 4,428.56. 10,413.96. 11,722.25. 8,302.89 9,342.03. All. 411.81. 588.82. 104.83. 224.32. 166.44. 160.98. Before. After. 36. After. Before. After. 72.79. 94.68. 133.67. 72.10. 90.11. 112.37. 135.80.