The Effects of Institutional Risk Control on Trader Behavior
Garvey, Ryan, Wu, Fei, Journal of Applied Finance
We examine how institutional risk control mechanisms influence proprietary stock trader behavior. When traders are forced to liquidate their inventory at a pre-designated time, they often hold onto their losing trades until the very last moment. We find that the difference between losing and winning round-trip holding times systematically widens leading up to an inventory liquidation deadline and trading becomes less driven by trading practices and more induced by the firm's control mechanism as the deadline draws near. When trade price is heavily controlled yet trade size isn't, we find that the difference between losing and winning round-trip holding times systematically widens with trade size. This result suggests traders increase their risk-taking in areas where institutional control mechanisms are weaker. Our findings highlight the difficult balancing act firms face with getting market professionals to realize their losses without impeding their trading strategies.
A considerable amount of research has uncovered behavioral biases among financial market participants.1 In order to circumvent these biases and help employees make better decisions, financial institutions implement risk control mechanisms with their employees who make trade decisions with institutional capital. The contributions of this study are: 1) to examine how effective control mechanisms are at mitigating psychological trading biases, and 2) to examine how employees respond to control mechanisms.
While our results are useful for firms implementing or planning to implement risk control mechanisms, our results also provide a step forward for the academic literature. Much of the academic literature has been devoted to uncovering behavioral biases in various settings and among different types of market participants. Uncovering these biases is, of course, a necessary first step. Yet, some trading biases are well known and firms have been grappling with ways to control them for decades. Despite this, there is very little research that examines trader behavior in settings where prevention techniques are actively implemented by firms to get their employees to recognize and refrain from biases in their decisions.2 In our paper, we examine such a setting.
We analyze how proprietary stock traders, who work on behalf of a National Securities Dealer, react to institutional control mechanisms that are primarily intended to get them to realize their trading losses. It is well known that market professionals often have difficulty coming to terms with their losses. Consequently, they have a tendency to hold their losing trades too long because they want to recover from their losses. This desire to get even is quite persuasive in financial market settings, and it is inevitably ingrained in many of the everyday decisions that traders make. Indeed, some of the greatest trading losses of all time have occurred because traders were simply unwilling to take a loss, so they gambled in an attempt to recover from their loss.3
In order to help traders come to terms with their losses, our sample firm implemented several control mechanisms. The most binding of these control mechanisms was that they required traders to liquidate their inventory by the end of the trading day in order to ensure loss realization. The firm implemented other control mechanisms including an emphasis on price control. They implemented training sessions that often stressed the dangers of holding losses too long. For example, the firm cites in their training manual that a trader's inability to take a loss is the number one reason why traders fail. And they employed a trading manager who closely monitored trading activity throughout the day. The firm even hired an on-site psychologist who was readily available to meet with traders.4
Despite all of these control measures, traders still appear to have difficulty coming to terms with their losses. We find that, on average, traders hold their losing trades significantly longer than their winning trades, which is consistent with the behavior underlying the disposition effect (see Shefrin and Statman, 1985). These longer holding times coincide with lower performance. While prior studies document that professional traders have a tendency to hold their losing trades longer than their winning trades (see, for example, Locke and Mann (2005) and Garvey and Murphy (2004)), the professional traders observed in prior studies were not required to close out of all of their positions by a predetermined time.5 In our setting, traders are forced to exit their positions by a predetermined time (i.e. the end of the trading day) and our main focus is on how traders holding times and performance vary across the day leading up to the inventory liquidation deadline.
We find that the difference between losing and winning round-trip holding times systematically rises throughout the day and that it rises to its highest level just prior to the inventory liquidation deadline. Moreover, trading performance significantly declines as the liquidation deadline draws near. Our sample traders often have difficulty realizing certain losses and they have a tendency to hold onto them until the very last moment. While the firm's efforts do not statistically eliminate a trader's tendency to hold losing trades longer than winning trades, they clearly do have an influence on trader behavior.
Inventory liquidation requirements ensure losses get realized, but firms (traders) also rely on price control mechanisms to do the job. While trade price is often heavily controlled in institutional trading settings, trade size usually is not. Institutional market participants trade in large trade sizes and their ability to execute these large trade sizes in their entirety is often driven by market conditions. Thus, professional traders need flexibility with respect to trade size. While we find that traders adhere to a highly disciplined approach with respect to their exit prices, trade size considerably varies and traders let their losses run longer on larger size trades. Consequently, they are more unprofitable when they trade in larger trade sizes.
These traders were constantly being drilled on the dangers of holding losses too long, but they tended to hold onto their losses despite the warning. Their tendency to hold losses for longer periods of time resulted in lower performance. If the firm did not require traders to close out of their positions by the end of the day, or if they did not implement any control measures, presumably losses would be held for even longer periods of time and performance would be a lot worse. This is why it is so important for financial firms to implement control mechanisms.
Our results highlight the complexities involved with implementing optimal risk control mechanisms to circumvent traders' aversion to realizing losses. If firms implement control mechanisms that are too stringent, they are likely to conflict with traders' overall strategies and objectives because, as our research shows, trader behavior is heavily influenced by control mechanisms. On the other hand, if firms do nothing they open themselves up for considerable risks. Thus, our findings imply that firms do need to implement control mechanisms, but they need to be very careful with how they enforce this because traders align their strategies with control mechanisms. Because psychological biases are so intertwined in many people's decision-making processes, firms are unlikely to eliminate these biases, but they can lessen the damage caused by them.
I. Related Research
Kahneman and Tversky's (1979) prospect theory provides a descriptive framework for decision-making under risk. A central theme in their research is the role of loss. Much of the behavioral finance literature focuses around how market participants make decisions when they are confronted with the prospect of a loss. Shefrin and Statman (1985) were the first to apply prospect theory to a financial market setting. They also placed prospect theory in a wider theoretical framework that includes mental accounting, regret aversion, and self-control. These factors together help explain theoretically why traders have a tendency to hold their losing trades much longer than their winning trades, a behavior that is commonly known as the "disposition effect".
Traders, who exhibit behavior that is consistent with the disposition effect, think about stock purchases within separate mental accounts (see Thaler, 1 985) then apply prospect theory decision rules to each mental account. The disposition effect has proven to be quite pervasive in US markets.6 For example, research shows that individual investors (e.g., Barber and Odean (1999) and Odean (1998)), mutual fund managers (e.g., Frazzini (2006) and Scherbina and Jin (2005)), and professional traders (e.g., Garvey and Murphy (2004) and Heisler (1994)) all exhibit signs of this behavior. Some of the more recent studies identify individual trader characteristics that are correlated with the disposition effect. For example, Dhar and Zhu (2007) find that individuals' income level, occupational status, etc. are important factors in indicating who is more susceptible to the disposition effect.
While most researchers examine traders' unwillingness to take a loss through holding times and focus on decisionmaking in a single period setting, some other studies have looked elsewhere. They examine a trader's reluctance to realize losses using other risk measures and focus on decisionmaking in a multi-period setting. For example, Covai and Shumway (2005) and Garvey et al. (2007) find traders who have experienced prior morning losses engage in subsequent afternoon risk-taking as measured through increased trading activity, larger size trading, etc. These findings are consistent with the same behavioral tendency that leads traders to hold their losing trades too long.
Like much of the prior research, we examine trader resistance to loss realization through holding-time decisions, and we examine trader holding-time decisions in a singleperiod setting. Our motivation is not to examine if traders suffer from the get even behavior that underlies the disposition effect, but rather our motivation is to examine how (if) traders respond to institutional control mechanisms to prevent this behavior.
Our data originates from a National Securities Dealer. The firm had several trading operations and our focus is on the firm's proprietary trading operation for US equities. The data covers June 3, 2002 through May 30, 2003. During this oneyear period, the US stock markets were open for 25 1 days. In total, the 150 traders combined to execute 2.5 billion shares through 1 .3 million transactions on 693 securities. For every transaction, the data reveals the identity of the trader, the execution time, the type of trade (marketable versus limit order), the action taken (buy, sell, short, and cover), the volume, the price, the market where the order was sent, the contra party on the trade (if given), the location of the trader (the traders were located in five branch offices), and various other trade execution information.
Each trader's sole objective was to generate intraday trading profits utilizing firm capital. Consequently, the traders trade often and they also trade in large trade sizes. The average trader executes 75 per day and the average executed trade size is for 1,925 shares. This average trade size is more than three times the average trade size in US equity markets.7 Trading activity is concentrated in certain stocks (mostly Nasdaq) on certain days, often accounting for a sizeable portion of a particular stock's trading volume.8 While the traders trade often, they also set market prices often. Approximately 65% of their trades provide liquidity, while 35% of their trades take liquidity. The traders set market prices in the various trading venues which US equities trade in.9
To get an idea of the trading intensity difference between institutional and retail market participants, consider a sample of retail brokerage accounts studied by Barber and Odean (2000).10 They analyze the trading behavior and performance of retail market participants, who trade through a US discount brokerage firm. In total, 66,465 households execute 1,969,701 stock trades over a sixyear period ending in December, 1996. Our 150 traders execute 1,316,334 stock trades over just a one-year period. Institutional market participants dominate the trading landscape in US equity markets, yet much of the academic literature examining trader decision-making in equity markets is focused on retail market participants. While some studies do examine institutional trading (see, more recently, Conrad, Johnson, and Wahal, 2002), this literature largely focuses on transaction costs, and prior studies do not examine how institutional traders respond to risk control mechanisms.
Our sample traders received continual training on various trading strategies from the firms' management, yet they had considerable freedom with selecting stocks to trade. The compensation of the traders is solely tied to their trading performance. Thus, the traders had a clear incentive to maximize their trading performance. For our purposes, the most intriguing aspect of this particular setting is the firm's efforts to get traders to realize their trading losses. The firm is not alone with respect to their efforts in this regard. Getting traders to take losses is a common problem that securities firms and their risk managers constantly grapple with.
III. Empirical Results
The objective of our study is to directly examine how traders react to institutional control mechanisms that are primarily intended to get them to accept their losses. Specifically, we measure the effects of institutional control mechanisms on traders' holding-time decisions. Round-trip performance and holding-time measures are not included in the raw transaction data. In order to determine the gains and losses for each round-trip, we use an intraday round-trip matching procedure similar to the one used in Garvey and Murphy (2005). We pair off opening trades with the subsequent closing trade(s) in the same day. The traders did not always open and close positions with two trades. A trader could combine a closing transaction with an opening transaction, or they could lay off part of an open position. Regardless of whether trades opened, closed, or open and closed a position simultaneously, we searched forward in time each day until the opening position was closed out, and we kept track of execution times, accumulated inventory, and corresponding prices paid or received. The matching procedure creates 730,417 roundtrip trades from the 1 .3 million trades. We then calculate out the corresponding holding time for each round-trip. In order to do this, we calculate the holding time between the intraday opening and closing transaction(s). If a trader accumulates inventory before they eventually close out of their position, we use a weighted average between the various opening positions.11
From the matching procedure, there are 290,248 winning round-trips (gross round-trip trading profits above $0), 209,271 losing round-trips (gross round-trip trading profits below $0), and 230,898 break-even round-trips (gross roundtrip trading profits equal to zero). The frequency of breakeven round-trips highlights how focused mese traders are on their trade purchase price (the reference point). The traders do not hold their open positions for long. And when they enter into a position on one side of me market, they generally seek to quickly offset their position by trading on me opposite side of the market. For example, the mean holding time per round-trip is 780 seconds and the median holding time per round-trip is 205 seconds.
The sample period we observe was a difficult time to trade US equities. Many securities firms were reporting steep losses on their equity trading desks.12 Our firm was not immune to these difficult trading conditions, yet the traders did experience many profitable (and unprofitable) round-trips which enable us to examine meir holding time decisions in both the domain of gain and loss.
B. Do Institutional Risk Control Mechanisms Eliminate Behavioral Biases?
Table I provides information on the overall holding time difference between winning and losing round-trip trades. Despite the firm's efforts to get traders to realize their losses in a timely fashion, traders hold their losing round-trip trades considerably longer man meir winning round-trip trades (note that me firm's control mechanisms were in place over our entire sample period). On average, traders hold meir losing round-trip trades for 1 ,274 seconds and their winning roundtrip trades for 568 seconds. The difference of 706 seconds is statistically different from zero at the 1% level. The magnitude of the overall holding time difference is rather surprising given the firm's continual efforts to get traders to realize meir trading losses. It would be interesting to see how this result would change if the firm did not engage in any efforts to get traders to realize meir losses. Presumably, traders would hold meir losses for even longer periods of time.
In order to check the robustness of our initial result, we examine the holding time differences for each individual trader. This allows us to see if certain traders are driving our overall result. On a mean holding time basis, 145 out of 1 50 traders hold meir losing round-trip trades longer man their winning round-trip trades (135 differences are statistically significant). On a median holding time basis, 146 out of 1 50 traders hold their losing round-trip trades longer than their winning round-trip trades (139 differences are statistically significant). The individual trader results coincide with the aggregate trader results.13
In our setting, holding positions for longer periods of time is generally undesirable. The objective of the traders is to rapidly enter and exit positions in order to profit from small price changes. The trader's information is short-lived, so when traders' open positions are held for extended periods of time, it is a good indicator that the position has moved against the trader and they are primarily holding it in order to recover from the loss. In Figure 1, we segregate me roundtrip profits into five holding time categories. Round-trip trading profits systematically decline with longer holding times. When traders hold their trades under (over) five minutes they are profitable (unprofitable). Trader profits are highest when mey hold meir trades for under one minute, and trader profits are lowest when mey hold meir trades for more than 15 minutes.14 These results highlight the trader's short-term strategies. They also show how important holding times are for a trader's success. A professional trader's decision to hold trades for slightly longer periods of time could make the difference between being profitable or unprofitable on an overall basis.
C. How Does Inventory Control Influence Trader Behavior?
While the firm's efforts do not appear to eliminate biases from traders' decisions, meir risk control measures do have an influence on trader behavior. This influence is quite strong in the moments just prior to the inventory liquidation deadline, which highlights how deadline effects coincide witii a trader's resistance to realize losses. Table II and Figure 2 provide information on round-trip holding times across the day. Traders hold losses longer man gains, but these holding time patterns do not remain constant throughout the day. The difference in holding times between losing and winning roundtrips systematically rises throughout the day and it dramatically increases in me moments just prior to the firm's mandatory close-out period. The results indicate that traders often hold losses up until the very last moment before mey are forced to realize mem. The trader's behavior is not a desirable reaction to what the firm is trying to accomplish, but me liquidation requirement appears necessary. Without this control mechanism in place, these traders' reluctance to take losses would have most likely resulted in larger trading losses.
In Table III, we report me gross round-trip trading profits for each half hour of the trading day. Trading profits are highest in me initial opening period, or the period which is furthest away from the liquidation deadline, and statistically different from zero at the 1% level. Trading profits steadily decline until noon. Trading profits are positive after 12:00 p.m., but they steadily decline again until me close of trading. The sharp decline in traders' profits in me moments just before the close, along with the holding time patterns, indicates that trading is significantly driven by the firm's control mechanism. The average round-trip profit in the last 30 minutes is -$2.56, whereas me average round-trip profit at otiier intraday times is $0. 1 8. The average round-trip trading profit in the closing 30 minutes is 1 ,522% less man the average round-trip trading profit at other intraday times. The end-ofday trading losses can be broken down further. While the traders lost $136,934 in the final 30-minute period, approximately 63% of this occurred in me final 5 minutes of the trading day (note mat they lost money in each 5 minute interval in the final 30 minute period). The magnitude of the losses mat occurred in each five minute interval is highlighted in Figure 3.
Trading profits most likely continue to decline until noon because traders often break for lunch. When traders leave their trading terminals, they usually close out of their positions. For many traders, the lunch period serves as another period for realizing losses. However, the midday close-out period is not binding like the end-of-the-day period is, and losses are generally held for shorter periods of time leading up to the midday period. This is why losses are far more pronounced at the close of the day man they are at the middle of the day.
While me end of day closeout period induces trade and forces traders to realize their losses, it is interesting to theorize about what would occur if the firm did not have mis control mechanism in place. If me firm allowed traders to hold meir positions overnight, would mis give traders greater flexibility with implementing meir trading strategies and subsequently improve performance? Or, would removing me liquidation deadline result in traders holding meir losses for significantly longer periods of time resulting in catastrophic losses? While it is not possible to definitively answer these questions from our available data, we compute a hypothetical performance measure assuming some trades were held overnight. Because trading in the very last moments of the day appears highly driven by me firm's control mechanism, we recalculate trading profits for positions closed out in the final 15 minutes of me trading day. Contrary to using the round-trip trade price at the end of me day to determine profits, we assume traders closed their positions at me opening price on the following (trading) day. In order to do mis, we obtained opening price data on sample stocks traded from the Center for Research in Security Prices (CRSP) database and recalculated trading profits for positions closed out in the final minutes of the day. There are 24,332 round-trip observations, and the average closing position is for 1,674 shares. Under the adjusted closing price, the average round-trip trading profit declines from -$3.95 to -$4.17 (note mat a few anomalous observations are dropped from me analysis). While me firm's liquidation requirement imposes a constraint on traders, most of the losing positions traders were forced to realize at the end of me day continued to decline in value into the next day of trading.
Our firm's inventory liquidation requirement is not unique. Many securities firms require their market making, proprietary, arbitrage, and other types of traders to end the day flat, or they significantly restrict traders ability to accumulate inventory from day to day.15 If traders at other firms exhibit behavioral tendencies similar to these traders, than our results provide insight into some factors that drive intraday order flow patterns. Researchers have long known that intraday trading activity in US equity markets exhibits a U-shaped pattern across the main trading hours (9:30 a.m. to 4:00 p.m.). The Admati and Pfleiderer (1988) theory has served as a prominent explanation for these intraday volume patterns. While our traders' intraday trading activity closely resembles a U-shape pattern, there are competing factors inducing this pattern across the day. Trading at the open, on average, appears to be motivated by short-term information (i.e. gross round-trip trading profits are statistically different from zero). Trading activity rises in the second half of the day, but this time the rise in trading activity corresponds with a decrease in performance. The rise in trading activity in the second half of me day, on average, seems driven more by the firm's risk control mechanism rather than their traders' normal trading practices.
While our findings might be useful for providing insight into intraday order flow patterns, they are also potentially useful in understanding why trading activity levels may predictably rise or decline on certain days, at certain times of the year, or in response to certain situations. Our results provide direct evidence on how trader behavior varies within a set trading time horizon and when a deadline exists. We would expect similar loss-averse behavioral patterns to occur over various time horizon settings in which there is some type of deadline being imposed or self-imposed on a decision-maker (e.g., a fund manager, trader, retail investor, etc.), such as with a performance evaluation period, compliance or audit period, a government tax period, a maintenance margin level, etc. Economics literature has found evidence of mis "wait until the very last moment" approach in other settings (e.g., with bargaining negotiations), and our results provide some evidence on how psychological trading biases and deadline effects interact in a financial market setting. There has been very little direct research on this front because researchers studying trader behavior in financial market settings often lack data on traders' time horizons or the time horizon set by the employee's institution.16 Thus, it is difficult to measure how trader behavior varies over a trading time horizon if the trading time horizon is not truly known.
D. How Does Price Control Influence Trader Behavior?
While the end-of-day inventory liquidation requirement is the most binding control mechanism the firm has in place, the firm trained the traders to adhere to a disciplined exit strategy (e.g., use of stop loss mechanisms) to ensure loss realization. Stop loss mechanisms can be employed either explicitly (attached with the opening order) or through a self- imposed rule. The firm attempted to monitor trader exit prices through the trading manager and the traders uniformly exited most, but not all, of their positions within a very tight pricing range. In Table IV, we report the distribution of round-trip price changes for both winning and losing round-trips. The median price change for both winning and losing round-trips is one cent.17
The traders prefer trading in larger trade sizes in order to maximize their trading profits, but they are often forced to trade in smaller trade sizes due to factors beyond their control. For example, suppose a trader submits a 5,000 share limit order at the underlying best bid price. If an incoming 1,000 share order executes against the trader's order, but then the market price moves sharply away from the trader's bid price, the trader is left with 80% of the original order unfilled and will be forced to reassess execution strategy.
While trade sizes vary with underlying market conditions, traders (firms) do not usually reset price control mechanisms in accordance with trade size (i.e. on a percentage basis). In institutional trading settings such as ours, the emphasis is on disciplined trading and adhering to a specific and well defined trading strategy. Our traders are trained to enter and exit their positions within a very tight price range and they typically offset their positions within one or two cents. If these traders were to constantly reset their exit prices on a trade size percentage basis, this would create a much less disciplined approach to trading, and it would give them incentives to deviate from their normal trading practices. The traders are not trained to capture large price changes. Instead, they are trained to capture small price changes while trading frequently on both sides of the market.
When price is heavily controlled and traders are given greater leeway with respect to trade size, this leaves firms vulnerable to heightened risk-taking with larger size trades. The existing stock price in relation to the opening stock price determines whether a trade is for a capital gain or loss, but trade size, along with trade price, determines the magnitude of a trading gain or loss. When traders enter into larger trades and the price moves against them, the magnitude of their trading losses will increase and according to decision-making theory, they will have an increasing desire to get even.
We segregate round-trip trade sizes into five trade size categories: 1) trade sizes less than 250 shares, 2) trade sizes greater than 249 shares and less than 1,000 shares, 3) trade sizes greater than 999 shares and less than 2,000 shares, 4) trade sizes greater than 1,999 shares and less than 3,000 shares, and 5) trade sizes greater than 2,999 shares. The overall holding time results for each trade size category, and for gains and losses, are reported in Figure 4A. While losing round-trips are held considerably longer than winning roundtrips for each trade size category, the difference systematically widens with trade size. We suspect this pattern is the result of traders moving deeper into the red with their larger trade sizes. As losses surmount and traders move further away from the break point, they will have an increasing desire to gamble (hold trades longer) in order to get back to the break even point. Yet, control mechanisms are not in place or are much weaker to stop this undesirable behavior because the emphasis is on disciplined trading with respect to uniform price control.
In Panel A of Table V, we examine overall performance and trade size. The absolute difference between the average trading gain and loss correspondingly widens with trade size. In general, we expect holding times to rise with larger trade sizes because it is more challenging to execute larger trades than smaller trades. However, this does not explain the widening gap between losing and winning round-trips with respect to trade size. Most of the trading losses can be attributed to trading in larger trade sizes (3,000 or more shares), where the loss-gain holding time difference is most pronounced. This suggests that traders' decision to ride their losses longer with larger trade sizes is costly.
We check the robustness of our trade size results by controlling differences in liquidity across the stocks traded. How liquid a stock is can affect both holding times and trading profits. We expect, on average, holding times to be lower on more liquid stocks. And, on average, we expect the price impact (if any) incurred executing a trade to be smaller on more liquid stocks. Variations in price impacts will be reflected in trading profits. In order to control liquidity differences across stocks, we sort the stocks traded into two groups (liquid vs. illiquid stocks) based on their average daily turnover ratios (volume / shares outstanding) over our one year sample period. Volume and share outstanding data is obtained from the CRSP database. We are able to use matching trade data from CRSP for more than 97% of the trading activity in our data.
We compute holding time differences and performance differences for both liquid stocks and illiquid stocks according to our trade size classifications. The holding time results are reported in Figures 4B and 4C and the performance results are reported in Panels B and C of Table V. As expected, most trading activity occurs on liquid stocks and holding times are much lower on liquid stocks. For both liquid stocks and illiquid stocks, losing round-trips are held considerably longer than winning round-trips for each trade size category and the difference systematically widens with trade size. In general, the absolute difference between the average trading gain and loss correspondingly widens with trade size too. Although there is a sharp drop in the performance difference for illiquid stocks under the largest trade size category, there are very few large trade size observations on illiquid stocks.
The trade size results highlight the need to assess institutional market participants' resistance to loss realization (and its associate costs) at the individual trade level. Our results also highlight the need to devise control mechanisms, among other things, on a situational trade basis. For example, a firm may analyze the trading decisions of its fund managers and find that overall, their fund managers do not exhibit a tendency to avoid realizing their losses. Consequently, the firm might feel less of a need to implement control mechanisms to prevent traders from taking excessive risks when they are confronted with the prospect of a loss. While the fund managers may not exhibit a tendency to avoid realizing their losses on an overall basis, they may exhibit a tendency to do so with their larger holdings, which would pose a significant (preventable) risk that is not easily detectable though casual analyses of the overall trading data.
One of the more well known psychological tendencies that permeates Wall Street trading desks is the traders' aversion to realizing losses. Traders have a tendency to hold their losing trades too long because they are predisposed to get even with their losses. By all accounts, this behavior is undesirable and can be quite costly. Securities firms are well aware of the costs that arise with this behavior and they often implement risk control mechanisms to prevent (limit) it from occurring. In our paper, we examine whether such measures actually work and how they influence proprietary stock trader behavior.
Despite our sample firm's efforts to get traders more comfortable with realizing their trading losses through training, managerial oversight, trader access to a licensed psychologist, discipline price control, and inventory liquidation, we find that professional traders still have difficulties accepting their losses. For example, traders hold losing trades more than twice as long as winning trades and these longer holding times coincide with lower trading profits.
While our results highlight how difficult it is for institutions to rid psychological biases from the traders' decisions, our results also highlight the complexities involved with implementing efficient control mechanisms to get traders to realize their losses. When firms force traders to liquidate their inventory and realize their losses, professional traders respond by holding their losses up until the very last moment. When firms heavily focus on disciplined trading and uniform price control to ensure loss realization, professional traders respond by holding their losses longer on larger size trades. Clearly, these are not desirable behavioral responses to the firms underlying objective. However, if financial institutions impose stricter control mechanisms to get their traders to realize their losses sooner, the additional trading constraints will likely begin to start conflicting with the traders' overall strategies and trading practices. On the other hand, if control mechanisms put in place are too lax, losing trades will be held for longer periods of time and losses will surmount.
Securities firms implement control mechanisms to improve performance and reduce risk, but their efforts to get employees to accept their losses has much broader implications. Our results show that institutional risk control can have a strong influence on trader behavior. Future studies, which provide institution detail on the design of control mechanisms being used at other financial institutions and how employees respond to them, would be insightful for both creating optimal risk control mechanisms and also for determining their overall effects in the marketplace.
We analyze how proprietary stock traders, who work on behalf of a National Securities Dealer, react to institutional control mechanisms that are primarily intended to get them to realize their trading losses.
Traders hold losses longer than gains, but these holding time patterns do not remain constant throughout the day. The difference in holding times between losing and winning roundtrips systematically rises throughout the day and it dramatically increases in the moments just prior to the firm's mandatory close-out period.
While losing round-trips are held considerably longer than winning round-trips for each trade size category, the difference systematically widens with trade size.
1 See, for example, Odean (1998), Grinblatt and Keloharju (2001), and Covai and Shumway (2005).
2 Statman and Caldwell (1987) discuss risk control and behavioral biases in the context of capital budgeting decisions.
3 One of the more notable cases, or largest losses resulting from this behavior, occurred with Nicholas Leeson. Mr. Leeson incurred over $1.4 billion in trading losses in 1995, which led to the demise of his employer, 232 yearold Barings PLC.
4 We sat in on the firms training sessions for traders, reviewed trader manuals, had several discussions with management and traders, and observed traders trade so that we could better prepare this paper.
5 For example, Garvey and Murphy (2004) examine proprietary stock traders who mainly offset their positions intraday, but the traders can and do hold positions overnight. Locke and Mann (2005) assume commodity traders end each day flat when determining trader gains and losses. However, the authors do not report, or are not aware of, any mandatory time period in which traders are required to liquidate their inventory.
6 Researchers have also found strong evidence of the disposition effect in Finland (e.g. Grinblatt and Keloharju, 2001), Israel (e.g. Shapira and Venezia, 2001), China (e.g. Feng and Seasholes, 2005), Taiwan (e.g. Barber et al., 2007) and other countries.
7 The average trade size for NYSE (Nasdaq) stocks was 488 (579) shares in 2003 (Source: NYSE and Nasdaq data).
8 The only stocks traded every day were Sun Microsystems (SUNW) and JDS Uniphase (JDSU). The traders accounted for 1.5% and 3.3% of the annual share volume of SUNW and JDSU respectively. SUNW and JDSU are two of the more actively traded US stocks.
9 The traders often traded on The Island ECN, which reported its trades through the Cincinnati Stock Exchange. See Nguyen, Van Ness, and Van Ness (2004) for a discussion on the reporting of Island trades on the Cincinnati Stock Exchange.
10 The data has been used in several studies in finance literature.
11 For example, suppose a trader opens up a 2,000 share position of Yahoo at 10:30:00 a.m. and then purchases another 4,000 shares of Yahoo at 10:30:10 a.m. If the next trade were a sell of 6,000 shares of Yahoo at 10:30:20 a.m., the holding time on the round trip trade is 13.33 seconds (1/3 * 20 seconds + 2/3 * 10 seconds).
12 For example, the revenues of broker dealers with a Nasdaq market making operation fell over 70% during 2000-2004 (GAO, 2005). Our firm had a Nasdaq market making operation.
13 The average round-trip holding time among traders ranges from 122 to 3,176 seconds.
14 The average round-trip gain is $13.70 and the average round-trip loss is $19.23. The absolute trading profit difference is statistically significant from zero at the 1% level.
15 There are many self-employed (retail) traders who adhere to this rule on a self-imposed basis. In the US, these retail day traders typically trade through direct access brokers. Bear Stearns finds that the more active traders (25+ trades per day) who trade through direct access firms account for approximately 40% of Nasdaq/NYSE trading volume (Goldberg and Lupercio, 2004). Thus, retail and institutional traders, who often end the trading day flat, account for a very large percentage of overall daily trading volume in US equity markets.
16 For example, Benartzi and Thaler (1995) assume that the average investors holding time period is one year. This assumption has been applied in other research settings (e.g. Odean, 1998).
17 Recently adopted Regulation National Market System (Reg. NMS) eliminated sub-penny pricing for securities priced above $0.99.
Admati, A.R. and P. Pfleiderer, 1988, "A Theory of Intraday Patterns: Volume and Price Variability," Review of Financial Studies 1 (No. 1), 3-40.
Barber, B.M. and T. Odean, 1999, "The Courage of Misguided Convictions: The Trading Behavior of Individual Investors," Financial Analysts Journal 56 (No. 6), 41-55.
Barber, B.M. and T. Odean, 2000, "Trading is Hazardous to Your Wealth: The Common Stock Investment Performance of Individual Investors," Journal of Finance 55 (No. 2), 773-806.
Barber, B., Y. Lee, Y. Lu, and T. Odean, 2007, "Is the Aggregate Investor Reluctant to Realize Losses: Evidence from Taiwan," European Financial Management, 13 (No. 3), 371-388.
Benartzi, S. and R. Thaler, 1995, "Myopic Loss Aversion and the Equity Premium Puzzle," Quarterly Journal of Economics 110 (No. 1), 73-92.
Conrad, J., K.M. Johnson, and S. Wahal, 2002, "The Trading of Institutional Investors: Theory and Evidence," Journal of Applied Finance 12 (No. 1), 7-14.
Coval, J. and T. Shumway, 2005, "Do Behavioral Biases Affect Prices?" Journal of Finance 60 (No. 1), 1-34.
Dhar, R. and N. Zhu, 2007, "Up Close and Personal: An Individual Level Analysis of the Disposition Effect," Management Science, forthcoming.
Feng, L. and M. Seasholes, 2005, Do Investor Sophistication and Trading Experience Eliminate Behavioral Biases in Financial Markets?" Review of Finance 9 (No. 3), 305-351.
Frazzini, A., 2006, "The Disposition Effect and Underreaction to News," Journal of Finance 61 (No. 4), 2017-2046.
Garvey, R. and A. Murphy, 2004, "Are Professional Traders Too Slow to Realize Their Losses?" Financial Analysts Journal 60 (No. 4), 35-43.
Garvey, R. and A. Murphy, 2005, "The Profitability of Active Stock Traders," Journal of Applied Finance 15 (No. 2), 93-100.
Garvey, R., A. Murphy, and F. Wu, 2007, "Do Losses Linger? Evidence from Proprietary Stock Traders," Journal of Portfolio Management 33 (No. 4), 75-83.
Goldberg, D.C. and A. Lupercio, 2004, "Cruising at 30,000: SemiPro Numbers Level Off, but Trading Volumes Rise," Bear Steams Company Report, August.
Government Accountability Office (GAO), 2005, "Decimal Pricing has Contributed to Lower Trading Costs and a More Challenging Trading Environment," GAO Report, May. Full report available online at: http://www.gao.gov/new.items/d05535.pdf.
Grinblatt, M. and M. Keloharju, 2001, "What Makes Investors Trade?" Journal of Finance 2 (No. 2), 589-616.
Heisler, J., 1994, "Loss Aversion in a Futures Market: An Empirical Test," Review of Futures Markets 13 (No. 3), 793-822.
Locke, RR. and S.C. Mann, 2005, "Professional Trader Discipline and Trade Disposition," Journal of Financial Economics, 76 (No. 2), 401-44.
Nguyen, V.T., B.F. Van Ness, and R.A. Van Ness, 2004, "The Reporting of Island Trades on the Cincinnati Stock Exchange," Journal of Applied Finance, 14 (No. 2), 30-39.
Odean, T., 1998, "Are Investors Reluctant to Realize Their Losses?" Journal of Finance 53 (No. 5), 1775-1798.
Scherbina, A. and L. Jin, 2005, "Change is Good or The Disposition Effect among Mutual Fund Managers," Working Paper, Harvard Business School.
Shapira, Z. and I. Venezia, 2001, "Patterns of Behavior of Professionally Managed and Independent Investors," Journal of Banking and Finance 25 (No. 8), 1573-1587.
Shefrin, H. and M. Statman, 1985, "The Disposition to Sell Winners Too Early and Ride Losers Too Long: Theory and Evidence," Journal of Finance 40 (No. 3), 777-790.
Statman, M. and D. Caldwell, 1987, "Applying Behavioral Finance to Capital Budgeting: Project Terminations," Financial Management 16 (No. 4), 7-15.
Thaler, R., 1985, "Mental Accounting and Consumer Choice," Marketing Science 4 (No. 3), 199-214.
Ryan Garvey is an Associate Professor of Finance at Duquesne University in Pittsburgh, PA. Fei Wu is a Senior lecturer in Finance at Massey University in Palmerston North, New Zealand.
We would like to thank the Editor, Ramesh Rao, and an anonymous referee for helpful comments and suggestions on a prior draft. We are also grateful to executives at the US Securities Firm for providing proprietary data.…
Questia, a part of Gale, Cengage Learning. www.questia.com
Publication information: Article title: The Effects of Institutional Risk Control on Trader Behavior. Contributors: Garvey, Ryan - Author, Wu, Fei - Author. Journal title: Journal of Applied Finance. Volume: 18. Issue: 2 Publication date: Fall 2008. Page number: 22+. © Financial Management Association Spring 2008. Provided by ProQuest LLC. All Rights Reserved.
This material is protected by copyright and, with the exception of fair use, may not be further copied, distributed or transmitted in any form or by any means.