Chapter

V Effects of Hedge Funds’ Strategies on Price Dynamics

Author(s):
Anne Jansen, Donald Mathieson, Barry Eichengreen, Laura Kodres, Bankim Chadha, and Sunil Sharma
Published Date:
May 1998
Share
  • ShareShare
Show Summary Details
Author(s)
Laura Kodres

Market price dynamics can be affected by the strategies implemented by hedge funds and other institutional investors in two ways. The strategies can act to stabilize prices or they can potentially destabilize prices. There are also a number of institutional practices that may inadvertently act to destabilize prices. A short description of their attributes are included since many market participants, including hedge funds, may be subject to these practices. While understanding how strategies can affect prices is important, the implications for financial stability hinge on whether such strategies do, in practice, make prices more volatile. Existing empirical evidence directly relating hedge fund activity to price volatility is slim to nonexistent. However, there are several studies that examine whether large market participants, including hedge funds, “herd” with other participants or with their own kind—that is, whether they take similar positions simultaneously or following one another. Circumstantial evidence can also be obtained from a study that examines the returns earned by hedge funds, their stated strategies, and the returns of standard asset classes. A third study, analyzing the actions of a set of large foreign currency market participants (which may include hedge funds), also provides insight into the connection between these large players’ activities and subsequent exchange rate volatility.

Stabilizing Strategies

Among the types of strategies employed by hedge funds and other market participants, two broad types are stabilizing—contrarian strategies and arbitrage strategies. Contrarian strategies, in which an investor buys when prices are deemed to be too low and sells when they are too high, contrary to current market movements, is an obvious case where prices would be naturally pushed back to their perceived fair value, thereby stabilizing prices.

By some accounts, arbitrage strategies may be viewed as neither stabilizing nor destabilizing in that the actions of arbitrageurs simply connect one market to another. For instance, arbitrage between a portfolio of stocks and a stock index futures contract limits the variability of the spread between the two prices but has no particular implication for the volatility of the price levels in the two markets. The two markets may move violently side by side, but the spread between them may be constant. However, several studies have shown that stock index arbitrage activity is in fact stabilizing, in the sense of reducing volatility of the underlying stocks.1

Because of their ability to both execute short positions and add leverage, in the 1960s and 1970s, hedge funds traded by holding long and short positions simultaneously, providing profits for investors in both rising and falling markets.2 In fact, their contrarian trading strategies won them the title “hedge” fund. In current markets, however, the incidence of contrarian strategies is difficult to gauge, although there is some circumstantial evidence, presented below, that a number of funds continue to use them. Moreover, there is ample anecdotal evidence that hedge funds attempt to search for markets in which prices have overshot their equilibrium values, based on fundamental or technical analysis. Anecdotal evidence, as well as the number of hedge funds identifying themselves with particular strategies, suggests that hedge funds are extensive users of arbitrage-based strategies.

Destabilizing Strategies

For the purposes of discussion, destabilizing strategies will be divided into those strategies that use existing prices to decide how to trade and those that use the positions of other market participants as the basis for trading decisions. Destabilizing trading that is based on prices is often referred to as positive feedback trading, a strategy in which participants buy after price increases and sell after price declines. If there are no offsetting forces, these participants can cause prices to “overshoot” their equilibrium value, adding volatility relative to that determined by fundamental information.3

Positive feedback trading can arise under a variety of circumstances—some of which can be related to various institutional features of markets. These include dynamic hedging, stop loss orders, and collateral or margin calls. On a less sophisticated level, positive feedback strategies also incorporate general trend-following behavior whereby investors use various technical rules to determine trends and reinforce them with their buying and selling behavior.

Among the strategies that induce positive feedback type behavior, dynamic hedging is the most complex. Dynamic hedging is the practice of altering the amount of the hedging instrument through time to make sure that losses on the underlying instrument are offset, as closely as possible, by gains on the hedging instrument (or vice versa). When hedging options, a hedge ratio, the amount of the hedging instrument needed, is provided by the option’s “delta.”4 The delta changes over time requiring the hedge to be adjusted—sometimes causing the hedging instrument to be sold at the same time as the underlying instrument’s price falls.

Using currency options as an example, the underlying intuition is as follows. Since options sellers know that, with some probability, they will incur losses as the currency depreciates, they sell, in advance, a certain proportion (measured by the delta) of the amount of currency underlying the options contract. Assuming the currency does, in fact, depreciate, options sellers have some profits on their short (sold) currency position with which to offset the losses incurred because of the buyers’ exercise of the put options. However, as the currency depreciates it becomes more probable that the put options will be exercised. The options seller will then sell increasing amounts of spot currency to cover these potential losses until the amount sold is almost equivalent to the amount underlying the contract. Thus, to hedge themselves, options sellers would be required to sell the currency in a falling market to maintain a hedged position, potentially exacerbating the original movement.

Hedge funds are typically buyers of options and do not need to hedge using dynamic hedging techniques. However, dealers (mostly large commercial and investment banks) that sell options to hedge funds either need to offset these positions with other interested parties or need to hedge themselves. Usually, options dealers hold portfolios in which they have a net short position in options, requiring them to manage the associated risk. Their ability to dynamically hedge depends on the liquidity underlying the option. The theory underlying dynamic hedging requires that price changes be continuous and that hedging will be most effective when adjustments to the amount of the underlying instrument are made on an ongoing basis.

The use of dynamic hedging techniques in Southeast Asian currencies purportedly caused a positive feedback response, adding volatility to some of these currencies. Apparently, put options on the Thai baht were priced relatively cheaply prior to the depreciation, encouraging hedge funds and other buyers to purchase them as another means of selling the baht. These trades left dealers with less premium with which to offset possible losses than if the puts had been appropriately priced. Moreover, without offsetting positions from another client base, dealers held net short options positions. However, the extent of dynamic hedging is not verifiable, and since the Southeast Asian currency markets were known to be somewhat illiquid compared with the major currency markets some dealers may have been reluctant to use the technique. Of course, leaving the positions unhedged may have had even larger profit/loss consequences for the dealers than inefficiently hedged positions.

Several other institutional features may also lead to positive feedback response. These are common to many markets and the participants that use them: they are not specific to hedge funds. Stop loss orders can give rise to positive feedback dynamics by having a previously submitted order to sell as losses are generated by price declines. The selling to limit losses continues to put downward pressure on prices. Similar dynamics exist for losses generated by price increases if short positions need to be closed out. Collateral calls or margin calls can sometimes lead to a positive feedback response as well. Collateral holders may require additional collateral or margin from their customers when prices fall and losses are incurred. Usually the collateral can be obtained by selling any number of instruments. However, a customer may have to sell the instrument whose price declined and caused the collateral call in the first place, possibly causing further price declines and losses. Although these market features are no different for hedge funds than other investors, some intermediaries that provide margin to hedge funds keep the funds on a very tight leash—margin calls are made daily and margin can be called intraday if necessary. To the extent that this is the case, hedge funds may sell into falling markets (or buy into rising markets) sooner than other of the intermediaries’ counterparties—and any leverage could potentially exacerbate the amount of margin or collateral that needs to be posted. On the other hand, since many hedge funds maintain short positions, positive feedback responses by hedge funds in market downswings may be less likely as these funds would have profits, not losses, in this environment.

Another class of destabilizing trading behavior is the result of participants taking similar positions to those of other market participants rather than basing their decisions explicitly on prices. Positions can be mimicked by directly observing other market participants’ positions or indirectly using the same set of information and analysis as other participants. When similar position-taking is undertaken strictly by following others’ positions, it is frequently termed herding.5 For example, herding may result from money managers rationally mimicking each other to make it more difficult to evaluate their competence, as in Scharfstein and Stein (1990), or they may herd if their performance is measured against a common peer group or benchmark as in Maug and Naik (1996). In particular, managers may rationally ignore their own private information when they perceive that their reputations in the labor market are based, in part, on whether—when they make unprofitable decisions—other managers, too, have low returns. Consequently, if one manager mimics the behavior of others this suggests to the labor market that he is more likely to be smart, whereas if he takes a contrarian position it is perceived that he is more likely to be “dumb,” all else equal. It is far from clear that hedge fund managers have the same incentives to mimic each other as, say, mutual fund or pension managers, since hedge fund managers have their own wealth invested in the fund and are not compensated relative to a benchmark but on total return. However, many hedge funds probably have access to the same information, and one could imagine, especially for macro hedge funds, that hedge funds could arrive at a similar assessment at approximately the same time, creating the illusion that they collude and purposely execute similar positions.

A second issue is not the question of whether hedge funds herd among themselves but whether other investors herd with them or follow their lead into various markets. Anecdotes abound that hedge funds are the “leaders” and other institutions, including proprietary trading desks and other investors, follow them closely. Some hedge funds refute that they are leaders, citing that proprietary trading desks of commercial and investment banks are looking at the same information as they are. Others suspect that other institutions are able to detect their positions and ride their coattails. Virtually all hedge funds adamantly deny the claim that they discuss their strategies among themselves.

Empirical Evidence

While there is little evidence linking hedge fund strategies to excess market volatility, there is some evidence regarding similar position-taking and the incidence of various investment styles among various institutional groups. Kodres and Pritsker (1996) use daily position data from the U.S. futures market to examine whether some classes of large market participants appear to herd or alter their positions similarly over time. The data not only provide a portrayal of large participants but also allow examination of several institutional subgroups, including hedge funds. The study was originally undertaken to assess whether specific institutional groups were more or less likely to move their positions in concert. The data, obtained from the Commodity Futures Trading Commission (CFTC), span August 3, 1992 to August 15, 1994, covering the two exchange rate crises in the exchange rate mechanism of the European Monetary System and the early 1994 period of turbulence in world bond markets.

The study estimates a necessary but not sufficient condition for herding: the propensity of large participants to buy (or sell) futures when other large participants buy (or sell). The institutional categories examined include broker-dealers, foreign banks, commercial banks (domestic), pension funds, mutual funds, and hedge funds.6 Statistically significant positive comovement of positions was detected for several contracts and for several institutional groups. It was found most consistently in the S&P 500 index contract for broker-dealers, pension funds, and hedge funds, meaning that these groups’ position changes were similar to the position changes of a randomly selected subset of their type of institution. Similar position-taking was also detected in the deutsche mark and Japanese yen contracts for broker-dealers and foreign banks. In addition, a sub-group consisting of the smallest one-third participants was separated in the data set to examine whether smaller participants herd with larger ones. The results showed little difference: positive co-movement was found for the same contract markets and participant types.

For hedge funds, the results suggest that similar position-taking occurs in the S&P 500 index contract and the three-month Eurodollar contract. When the hedge fund group is split into large and small hedge funds, similar position-taking is present in the S&P 500 index contract and the Japanese yen contract. Interestingly, however, when the lagged daily price change is added to the model, the results suggest that while hedge funds are mirroring the positions of other hedge funds they are also negative feedback trading—they are (together) trading against the price trend in a contrarian fashion. Interestingly, hedge funds appear to engage in negative feedback trading for six of the seven contracts in which they are active.

For purposes of this occasional paper, the analysis was extended to investigate whether other market participants’ position changes were positively correlated with those of hedge funds. The results, presented in table 5.1, show that the opposite is true. For the futures contracts and institutional groups that showed a statistically significant correlation between position changes of institutional groups and those of hedge funds, the correlation was negative, not positive, implying other institutional groups alter their positions in the opposite direction to those of hedge funds. One might suspect that these other institutional groups are slow to alter their positions and follow the hedge funds with a lag. Using the lagged position changes of hedge funds in the model results in statistically insignificant correlations with position changes of other institutions in most markets, with the lone exception of the Japanese yen where, again, broker-dealers change their position in the opposite direction.7 (See Table 5.2.)

Table 5.1. Contemporaneous Herding by Selected Institutional Groups with Hedge Funds
Model 1 PropbuysModel 2Model 3Number of Individual Participant Types
PropbuysNetbuysPropbuysΔPt-1
Five-year U.S. treasury note
Broker-dealer0.61871.1674-1.37620.68471.494335
(0.2680)(0.1215)(0.9156)(0.2468)(0.0675)
Foreign bank-0.20390.1835-0.8398-0.2894-1.227222
(0.5808)(0.4272)(0.7995)(0.6139)(0.8901)
Commercial bank-2.3210-1.2042-1.9517-1.80981.978113
(0.9899)(0.8857)(0.9745)(0.9648)(0.0240)
Ten-year U.S. treasury note
Broker-dealer-0.93900.8866-3.7083-0.80622.592038
(0.8261)(0.1876)(0.9999)(0.7899)(0.0048)
Foreign bank-1.07580.9725-0.1575-0.8928-0.262615
(0.8590)(0.8346)(0.5626)(0.8140)(0.6036)
Commercial bank-0.5063-0.1373-0.2055-0.35040.818511
(0.6937)(0.5546)(0.5814)(0.6370)(0.2065)
Thirty-year U.S. treasury bond
Broker-dealer-4.2063-1.3307-6.4781-3.65102.753946
(>0.9999)(0.9084)(>0.9999)(>0.9999)(0.0029)
Foreign bank0.65941.5504-2.29420.94300.665728
(0.2548)(0.0605)(0.9891)(0.1728)(0.2528)
Commercial bank0.12351.2239-1.80910.43901.973814
(0.4509)(0.1105)(0.9648)(0.3303)(0.0242)
Mutual fund-3.0921-2.3196-1.6062-2.18824.389318
(0.9990)(0.9898)(0.9459)(0.9857)(>0.9999)
Pension fund-0.11040.2885-0.6487-0.5796-1.994823
(0.5440)(0.3865)(0.7417)(0.7189)(0.9770)
S&P 500 index
Broker-dealer-6.0282-0.4638-8.2784-5.82701.163332
(>0.9999)(0.6786)(>0.9999)(>0.9999)(0.6122)
Mutual fund-1.77770.2245-3.1753-1.7877-1.721254
(0.9623)(0.4112)(0.9993)(0.9631)(0.9574)
Three-month Eurodollar
Broker-dealer-2.87600.0205-5.9120-2.9730-1.009858
(0.9980)(0.4918)(>0.9999)(0.9985)(0.8437)
Foreign bank-3.4412-1.7996-2.7306-3.4841-1.1316113
(0.9997)(0.9640)(0.9968)(0.9998)(0.8711)
Commercial bank-2.4899-1.4979-1.7395-2.52110.414724
(0.9936)(0.9329)(0.9590)(0.9941)(0.3392)
Deutsche mark
Broker-dealer-5.7500-1.9241-1.8076-6.6422-4.371034
(>0.9999)(0.9728)(>0.9999)(>0.9999)(>0.9999)
Foreign bank-4.9194-2.1794-4.0511-5.6114-4.499927
(>0.9999)(0.9853)(>0.9999)(>0.9999)(>0.9999)
Japanese yen
Broker-dealer-4.0407-1.4742-2.7890-4.6910-4.558331
(>0.9999)0.92980.9974(>0.9999)(>0.9999)
Foreign bank-2.34851.0571-4.4592-3.2414-7.032421
(0.9906)(0.1452)(>0.9999)(0.9994)(>0.9999)
Notes:The sample period is August 3, 1992 through August 15, 1994. Propbuys is the Z statistic associated with the variable measuring the proportion of buys relative to total trades executed by hedge funds representing the “herd.” Netbuys is the Z statistic associated with the variable measuring the net number of contracts purchased by hedge funds representing the herd. ΔPt-1 is the Z statistic associated with the lagged price change variable, measuring the degree of positive (or negative) feedback. The Z statistic is an aggregated statistic representing the sensitivity of the selected institutional group’s buy/sell decision to the explanatory variables in models 1 through 3. The number of individual participants included in the estimation is given in the final column. The number in parentheses is the probability (p-value) associated with a one-tailed test that the Z statistic is greater than zero. For a detailed description of the techniques used, see Kodres and Pritsker (1996).
Notes:The sample period is August 3, 1992 through August 15, 1994. Propbuys is the Z statistic associated with the variable measuring the proportion of buys relative to total trades executed by hedge funds representing the “herd.” Netbuys is the Z statistic associated with the variable measuring the net number of contracts purchased by hedge funds representing the herd. ΔPt-1 is the Z statistic associated with the lagged price change variable, measuring the degree of positive (or negative) feedback. The Z statistic is an aggregated statistic representing the sensitivity of the selected institutional group’s buy/sell decision to the explanatory variables in models 1 through 3. The number of individual participants included in the estimation is given in the final column. The number in parentheses is the probability (p-value) associated with a one-tailed test that the Z statistic is greater than zero. For a detailed description of the techniques used, see Kodres and Pritsker (1996).
Table 5.2.Herding by Selected Institutional Groups Following Hedge Funds by One Day
Model 1 Propbuyst-1Model 2Model 3Number of Individual Participant Types
Propbuyst-1Netbuyst-1Propbuyst-1ΔPt-1
Five-year U.S. treasury note
Broker-dealer1.50541.3933-0.13591.49131.626335
(0.0661)(0.0818)(0.5540)(0.0679)(0.0519)
Foreign bank-1.3564-1.51230.4214-1.4351-1.368022
(0.9125)(0.9348)(0.3367)(0.9244)(0.9143)
Commercial bank0.42681.2527-1.70870.77702.208413
(0.3348)(0.1052)(0.9562)(0.2186)(0.0136)
Ten-year U.S. treasury note
Broker-dealer-0.3356-0.80560.8053-0.26852.688438
(0.6314)(0.7898)(0.2103)(0.6059)(0.0036)
Foreign bank0.79451.3887-0.87770.58380.058515
(0.2135)(0.0825)(0.8099)(0.2797)(0.4767)
Commercial bank-0.1379-0.0853-0.1739-0.18690.933411
(0.5548)(0.5340)(0.5690)(0.5741)(0.1753)
Thirty-year U.S. treasury bond
Broker-dealer0.7541-0.26972.01320.40213.756746
(0.2254)(0.6063)(0.0220)(0.3438)(0.0001)
Foreign bank0.40090.48220.00960.36080.606228
(0.3442)(0.3148)(0.4962)(0.3591)(0.2722)
Commercial bank-0.9957-1.18640.84160.98252.036814
(0.8403)(0.8823)(0.2000)(0.8371)(0.0208)
Mutual fund1.16780.74420.24011.00634.701918
(0.1214)(0.2284)(0.4051)(0.1571)(0.0000)
Pension fund0.36810.9508-1.25160.7410-1.814023
(0.3564)(0.1708)(0.8946)(0.2293)(0.9652)
S&P 500 index
Broker-dealer1.38720.26901.50870.72511.517932
(0.0827)(0.3940)(0.0657)(0.2342)(0.0645)
Mutual fund-0.34990.0221-0.34340.3037-1.841754
(0.6368)(0.4912)(0.6343)(0.3807)(0.9672)
Three-month Eurodollar
Broker-dealer-1.4842-0.8635-1.0168-1.3717-0.623458
(0.9311)(0.8061)(0.8454)(0.9149)(0.7335)
Foreign bank-1.4853-0.4770-2.1722-1.3048-0.4781113
(0.9313)(0.6833)(0.9851)(0.9040)(0.6837)
Commercial bank1.09520.76910.42801.08110.828224
(0.1367)(0.2209)(0.3343)(0.1398)(0.2038)
Deutsche mark
Broker-dealer0.43430.7801-0.78161.2162-3.486934
(0.3320)(0.2177)(0.7828)(0.1120)(0.9998)
Foreign bank0.5007-0.53571.62241.2419-3.571327
(0.3083)(0.7039)(0.0524)(0.1071)(0.9998)
Japanese yen
Broker-dealer-2.1024-2.07420.6304-1.5493-3.581431
(0.9822)(0.9810)(0.2642)(0.9393)(0.9998)
Foreign bank-1.9971-0.8830-1.2416-1.2444-6.076921
(0.9771)(0.8114)(0.8928)(0.8933)(>0.9999)
Notes:The sample period is August 3, 1992 through August 15, 1994. Propbuys is the Z statistic associated with the variable measuring the proportion of buys relative to total trades executed by hedge funds lagged one period representing the “herd.” Netbuys is the Z statistic associated with the variable measuring the net number of contracts purchased by hedge funds lagged one period representing the herd. ΔPt-1 is the Z statistic associated with the lagged price change variable, measuring the degree of positive (or negative) feedback. The Z statistic is an aggregated statistic representing the sensitivity of the selected institutional group’s buy/sell decision to the explanatory variables in models 1 through 3. The number of individual participants included in the estimation is given in the final column. The number in parentheses is the probability (p-value) associated with a one-tailed test that the Z statistic is greater than zero. For a detailed description of the techniques used, see Kodres and Pritsker (1996).
Notes:The sample period is August 3, 1992 through August 15, 1994. Propbuys is the Z statistic associated with the variable measuring the proportion of buys relative to total trades executed by hedge funds lagged one period representing the “herd.” Netbuys is the Z statistic associated with the variable measuring the net number of contracts purchased by hedge funds lagged one period representing the herd. ΔPt-1 is the Z statistic associated with the lagged price change variable, measuring the degree of positive (or negative) feedback. The Z statistic is an aggregated statistic representing the sensitivity of the selected institutional group’s buy/sell decision to the explanatory variables in models 1 through 3. The number of individual participants included in the estimation is given in the final column. The number in parentheses is the probability (p-value) associated with a one-tailed test that the Z statistic is greater than zero. For a detailed description of the techniques used, see Kodres and Pritsker (1996).

The study is subject to several caveats. First, the number of large participants classified as hedge funds in some of the contracts is very small. For example, for the S&P 500 contract and the Japanese yen contract, where the results are relatively strong, the results depend on only seven and two hedge funds, respectively. Second, hedge funds are typically small users of futures markets. Among the large participants, the proportion of open interest held by hedge funds is highest for the five-year U.S. treasury note (about 10 percent) and is less than 2 percent in the currency contracts where, in general, the market in currency futures represents a minuscule part of the overall foreign exchange market. Moreover, the currency results only apply to five major currencies (the pound sterling, the Canadian dollar, the deutsche mark, the Japanese yen, and the Swiss franc) as there are few listed futures contracts on other currencies and the volumes for those that are listed are too small to be of interest to hedge funds. More generally, to attribute a positive correlation of positions to a potentially destabilizing situation, one would need to control for the arrival of market news and have a broader view of the institutions’ portfolios.

Using the model to predict the proportion of volume that could be attributed to correlated position-taking shows that the amounts never exceed 16 percent and are more often around 4 percent to 5 percent. This suggests that the extent of herding found in the study is unlikely to pose a systemic risk. More important, the correlated position-taking uncovered could be attributable to other trading strategies related to potentially offsetting cash positions held by the participants, such as index arbitrage or covered interest rate arbitrage.

Fung and Hsieh (1997) analyze investment styles of hedge funds and commodity trading advisors (CTAs) using data on monthly returns. Their purpose is to extend a model of mutual fund investment styles to the more dynamic, leveraged trading strategies of hedge funds, aiming to attribute hedge fund returns to a broader array of “styles.” They find that, unlike mutual funds, whose returns are highly correlated with standard asset classes (such as U.S. equity returns and U.S. bond returns), “hedge fund managers and CTAs generate returns that have low correlations to the returns of mutual funds and standard asset classes” (p. 277). To accommodate the sensitivity of the results to outliers, they divide the returns of each asset class into five “states” and find that three of the five styles they empirically identify, including one titled “global/macro,” are not sensitive to changes in chosen asset class returns during “normal states” but can be sensitive to selective markets during extreme states (when asset class returns are in the “tails” of their distribution). They conclude that funds classified as global/macro do not use buy-and-hold strategies in U.S. bonds, currencies, or emerging market equities. One of the identified strategies, termed “Systems/Opportunistic,” delivered positive performance in the states when extreme negative outcomes were recorded in equities and bonds, suggesting that a contrarian strategy may have been employed. Taken together, Fung and Hsieh show that hedge funds use a diverse set of strategies that are uncorrelated with the buy-and-hold strategies used by U.S. mutual funds. This leaves open the possibility that instead of contributing to excess volatility, hedge funds, acting as contrarians, lower volatility.

While not directly related to hedge funds, Wei and Kim (1997) examine the correlation of positions taken by large foreign exchange participants (as a whole) and subsequent movements in exchange rates. The data, collected and published by the U.S. Treasury, records weekly and monthly positions in the spot, forwards, futures, and options for the five major currencies.8 They find that these large participants, of which 29 were commercial banks and the remaining 7 were other financial institutions, were unable to forecast subsequent movements in the exchange rate—neither the direction nor the magnitude of future changes. They did, however, find a positive association between the absolute value of positions and a subsequent increase in exchange rate volatility, measured as the standard deviation of daily returns over various time horizons. The authors interpret these results as suggesting that the positions of large participants are taken, at least in part, to speculate on the level of exchange rate movements. Whether the seven remaining financial institutions in the study, or some subset of them, are hedge funds and how these institutions influence the results is not known, and thus it is difficult to assess the relevance of these results to hedge fund activities. Perhaps the best way to evaluate the results is to note that hedge funds are probably not alone in pursuing speculative positions in currency markets and that no one group of large participants seems to excel at making directional predictions.

See Neal (1993), for example.

See Jaffer(1996), p. 129.

See DeLong, Shleifer, Summers, and Waldmann (1990) for a model of destabilizing positive feedback trading.

The delta measures the change in the options’ price for a given change in the price of the underlying instrument: mathematically, the first derivative of the options’ price with respect to the underlying instrument. The change in the delta is measured by “gamma.” Gamma measures the sensitivity of the hedge ratio to changes in the price of the underlying instrument: mathematically, the second derivative of the option’s price with respect to the underlying instrument. A negative gamma means the hedging instrument needs to be reduced when the underlying price falls (or vice versa), leading to a positive feedback response.

Herding often connotes blindly (irrationally) following others’ movements. However, in most financial models of herding the participants are rationally following the behavior of others.

The categorization was done by the CFTC and, while the accounts are individually identifiable in the data, the names and some other characteristics were suppressed to preserve confidentiality.

Since every contract sold is offset by one purchased, position changes of all participants must be offsetting. Some may suspect, therefore, that a negative correlation of position changes would be forthcoming. However, the data consist of a subset of participants (large ones), and there is no mathematical or statistical reason why any offsetting position changes should be identifiable by institutional group.

See Section VI for more detail on the data.

References

    DeLongBradford J.AndreiShleiferLawrence H.Summers andRobert J.Waldmann1990Positive Feed-back Investment Strategies and Destabilizing Rational SpeculationJournal of Finance Vol. 45 (June) pp. 379–95.

    FungWilliam andDavid A.Hsieh1997Empirical Characteristics of Dynamic Trading Strategies: The Case of Hedge FundsReview of Financial Studies Vol. 10 (Summer) pp. 275–302.

    JafferSohail1996Evaluating Opportunities in Europe: The Institutional Investor’s Perspective,” inEvaluating and Implementing Hedge Fund Strategiesed. byRonald A.Lake (London: Euromoney Books).

    KodresLaura E. andMathewPritsker1996Directionally-Similar Position Taking and Herding By Large Futures Market Participants,” inRisk Measurement and Systemic Risk: Proceedings of a Joint Central Bank Research Conference (Washington: Board of Governors of the Federal Reserve System).

    MaugErnstNarayanNaik1996Herding and Delegated Portfolio Management: The Impact of Relative Performance Evaluation on Asset AllocationWorking Paper 223–1996 (London: London Business School).

    NealRobert1993Is Program Trading Destabilizing?Journal of Derivatives Vol. 1 pp. 64–77.

    ScharfsteinDavid S. andJeremy C.Stein1990Herd Behavior and InvestmentAmerican Economic Review Vol. 80 (June) pp. 465–79.

    WeiShang-jin andJungshikKim1997The Big Players in the Foreign Exchange Market: Do They Trade on Information or Noise?NBER Working Paper 6256 (Cambridge, Massachusetts: National Bureau of Economic Research).

    Other Resources Citing This Publication