پول خانگی و اثرات نقطه سر به سر برای انواع مختلف از معامله گران: شواهد از بازارهای آتی تایوان

Using a set of transaction records from the Taiwan Futures Exchange, we examine risk-taking behavior subject to prior outcomes and study the house money and break-even effects across various trader types. The empirical results show that the degree of morning gains/losses nonlinearly influences afternoon risk taking for all trader types, but the pattern is different for each type. Active individuals exhibit a house money effect after experiencing large gains and exhibit a break-even effect after large and small losses. Futures proprietary firms exhibit a break-even effect only after experiencing large morning losses. By contrast, foreign institutions exhibit only a house money effect after they experience small gains. The additional risk-seeking behaviors of futures proprietary firms and foreign institutions do not have a significant influence on market volatility or liquidity; only active individuals' risk-seeking behaviors when facing large morning losses impact both market volatility and liquidity.

To date, literature on behavioral finance greatly emphasizes the relation between prior profits and subsequent risk-taking behavior (Barberis and Xiong, 2009, Genesove and Mayer, 2001 and Heath et al., 1999). Using a large sample of students to study risk-taking behavior in an experimental setting, Thaler and Johnson (1990) find that under certain circumstances people are more likely to take risks after prior gains, known as the house money effect. On the other hand, Thaler and Johnson also note that prior losses lead to risk-seeking behavior in situations where future outcomes offering an opportunity to break even look particularly attractive. This result is termed the break-even effect. In this study, we attempt to examine the relation between risk taking and prior gains/losses in the context of Taiwan's stock index futures (TX futures) market. Recent literature shows that the relation between risk taking and prior gains/losses is significantly different across various types of investors. For example, a considerable body of research finds that retail individuals (Dhar and Zhu, 2006, Feng and Seasholes, 2005 and Odean, 1998) and professional investors (Frino et al., 2004 and Locke and Mann, 2005) ride their losses and quickly cash out their gains and are thus prone to the disposition effect. A similar behavior pattern can be recognized among fund managers. Motivated by career or reputation concerns, fund managers tend to increase risk taking following bouts of poor performance and cut risk following bouts of good performance (Chevalier and Ellison, 1997). By contrast, O'Conell and Teo (2009) suggest that fund managers aggressively control risks after losses and mildly increase risks after gains, which is consistent with dynamic loss aversion (Barberis et al., 2001). The sign of the effect differs from that observed for individual investors and professional traders. In other examples, Locke and Mann (2005) argue that individual investors are likely to exhibit irrational behavior, whereas disciplined market professionals are able to minimize potential behavioral influence. Likewise, Grinblatt and Keloharju (2001) investigate the disposition effect and find significant differences in trading styles between Finnish individual investors and foreign institutions. However, the experimental study of Haigh and List (2003) finds that professional Chicago Board of Trade (CBOT) futures traders show more myopic loss aversion than less experienced decision makers (i.e., students). As a result, O'Conell and Teo (2009) suggest that behavioral bias is sensitive to investor types. This study aims to conduct a detailed analysis of the trading behavior of different types of investors in the context of Taiwan's futures market. This study examines whether prior outcomes affect subsequent risk-taking behavior and which effect exists on the TX futures market. The main focus is to examine how prior outcomes affect risk-taking behavior for each of the investor types (including futures proprietary firms, foreign institutions, and active individual investors) at the account level and determine whether changes in investor risk attitude have any impact on the market. The contributions of this study are as follows. First, Coval and Shumway (2005) document that market makers engage in greater risk taking after prior losses to break even and reduce risk exposure after prior gains to stay ahead, consistent with the notion that investors try to avoid a sure loss. This effect was also determined by professional futures traders at the Chicago Mercantile Exchange during 1995 (Locke and Mann, 2005). Coval and Shumway (2005) treat profit as merely positive and negative values of a single psychological driver. However, Frino et al. (2008) argue that the value function of Kahneman and Tversky (1979) is built on the understanding that human decision makers react disparately to gains and losses.1 Accordingly, both a house money effect and break-even effect can occur at the same time since traders exhibit symmetric risk-taking behavior with respect to gains and losses. Following Frino et al. (2008), this study includes both gains and losses as separate explanatory variables and examines whether traders exhibit a house money effect and/or a break-even effect. However, even using the gains and losses disparately employed by Frino et al. (2008), one cannot effectively capture the nonlinearity of risk-taking behavior on various morning performances. Grinblatt and Keloharju (2001) and Kaustia (2010) suggest that the propensity to sell a stock is affected by different levels of past performance. Therefore studying in more detail how risk perception varies as a (possibly nonlinear) function of previous performance allows more powerful inferences for risk-taking behavior. To better understand the changes in risk attitude, this study examines whether the risk-taking behaviors of different types of TX futures traders are affected by different levels of past gains and losses. Second, institutional investors are believed to be more sophisticated than individuals. Locke and Mann (2005) argue that individuals display irrational behavior and that disciplined market professionals can minimize potential behavioral influences. Choe and Eom (2009) investigate the Korean futures market and find that individuals are much more prone to the disposition effect than institutional and foreign investors. The phenomenon is also found in Taiwan's futures markets (Chou et al., 2010). However, the experimental study of Haigh and List (2003) finds that professional CBOT futures traders demonstrate more myopic loss aversion than less experienced decision makers (students). As a result, behavioral bias may be sensitive to trader types. Most of the published evidence based on actual trading decisions in the futures market focuses on professional traders' behaviors but says little about the behavior of other trader types. To our knowledge, there is very little literature that simultaneously investigates the relation between prior outcomes and subsequent risk-taking behavior for various trader types. We attempt to examine whether one group of traders is more or less prone to risk taking following certain previous outcomes and, moreover, whether market professionals with sufficient discipline and/or sophistication can diminish the behavioral bias more than individuals in a controlled trading experience and activeness. Finally, to determine whether traders' changes in risk taking exert any influence on the market, this study examines the house money and break-even effects on market liquidity and volatility. Liu et al. (2010) also examine the relation between prior trading outcomes and subsequent risk taking in the Taiwan Stock Index options markets. Our study differs from theirs in three ways. First, theirs focuses on market makers' trading behavior in the option markets while ours examines three other investor types in the futures market. Second, their study treats profits as merely positive and negative values of a single psychological driver, while ours includes both gains and losses as separate explanatory variables and can therefore test for house money and break-even effects simultaneously. Finally, by including a squared morning profit term in the regression model, Liu et al. (2010) detect a nonlinear relation between afternoon risk taking and morning profits. Our study also finds a nonlinear relation but, rather than including a quadratic morning profit term, we rank traders' morning profits into different groups of profitability levels and include these different degrees of morning gains/losses as explanatory variables to explore further how investor risk-taking behaviors are affected by different degrees of past gains and losses. Specifically, Liu et al. only find a quadratic relation between afternoon risk taking and morning profits but do not show the pattern of this relation. By incorporating different levels of prior profits, we determine the patterns of the relation between afternoon risk taking and different degrees of morning gains/losses. In summary, by considering various types of investors and including gains and losses separately, as well as incorporating different degrees of morning gains/losses, we find that investors display both house money and break-even effects and such patterns vary between different types of investors. Moreover, traders' risk-taking activities depend on not only the signs of prior investment outcomes, but also their magnitudes. The remainder of this paper is organized as follows. Section 2 describes the data and the proposed method. Section 3 presents our empirical results and Section 4 discusses our conclusions.

The prospect theory of Kahneman and Tversky (1979) implies that traders exhibit different risk attitudes when facing gains and losses but says little about how the magnitude of gains and losses affects risk-taking behavior. As such, in the current study, we examine how risk-taking behavior is affected by the amount of prior gains and losses across different types of traders. Our empirical results show that for active individual investors in the TAIFEX both the house money and break-even effects coexist. Since our model 1 does not divide profits into gains or losses, their influence on subsequent risk taking weakens and only a break-even effect is found. However, when redefining profits as either gains or losses with measures of gains and losses included as separate explanatory variables in model 2, we find significant house money and break-even effects on active individual investors. When we further divide morning gains/losses into a larger half and a smaller half and include different levels of morning gains and losses as explanatory variables in model 3, we find a significant house money effect for larger morning gains and a significant break-even effect for both large and small morning losses for active individual investors. The results imply that the degree of morning outcomes nonlinearly influences traders' afternoon risk taking. For active individual investors, both gains and losses have positive effects on subsequent trading risk. When facing morning losses, large or small, investors show a tendency to take on greater risk, consistent with the break-even effect. However, only large morning gains drive active individuals to take on greater risk. Therefore active individuals show a tendency toward the house money effect only for larger morning gains and not for smaller morning gains. Foreign institutions only show a tendency toward the house money effect for small morning gains; therefore we can only find such an effect in model 3 but not in model 1 or 2. Futures proprietary firms show a tendency toward the break-even effect only for larger morning losses. Overall, the additional risk-seeking behaviors of futures proprietary firms and foreign institutions do not have significant influences on market volatility and liquidity. Only active individuals' risk-seeking behaviors when facing large morning losses impact both market volatility and liquidity.