IX An Operational View of the Foreign Exchange Market

R. Johnston, and Mark Swinburne
Published Date:
September 1999
  • ShareShare
Show Summary Details

This section discusses the characteristics and typical practices of the foreign exchange market and the interactions with regulatory or market-based policy interventions. Written from an operational perspective, it explains something of the pressures that motivate a dealer when buying and selling currency and relates these to the conceptual issues that lie at the core of questions about the appropriate role of public policy in the foreign exchange market. The main focus is on reasonably advanced foreign currency markets, and especially on the role of banks as dealers and market-makers in the interbank market, as the center of the currency markets in general. Nevertheless, the principles not only apply to industrial countries but also to a fairly wide range of emerging market economies.

The second part of the section provides background on the structure of the market and its constituent parts and considers the environment in which dealing takes place and the effect this has on dealing decisions. The third part deals with the management of exchange transactions and markets. It discusses what is needed for an effective interbank market and what a central bank can do to help bring this about, examining, in particular, the nature of market making, pricing behavior, arbitrage and speculation, and profit seeking or loss avoidance. A concluding section links the preceding discussion to models of exchange market behavior.

Market Structure and Trends

The Agents

Although this section mainly concerns banks and the role they play in the wholesale foreign exchange markets, it is necessary to see how their role relates to that of other actors in the foreign currency markets. The major market participants may be conveniently grouped as follows:

First are the commercial and investment banks that make up the interbank market at the core of the currency exchange system. Some banks elect to make markets—that is, they quote firm bid and offer rates at which they stand ready to deal, in either direction, in all but the most severe market conditions. Other banks confine themselves to servicing customer needs. They account for the largest share of market turnover by far and perform a vital economic function of mediating currency flows. Most take positions at one time or another, and to one degree or another, on future exchange rates. Indeed, taking a view on likely future exchange rates is an inseparable aspect of the market-making role. As such, banks normally have to be willing to accept a significant amount of currency risk but must also be able to manage that risk or find ways of laying it off substantially. The net position banks take, at least beyond very short-term (within a day) time frames, are typically limited by either internal controls or supervisory requirements linked to their capital.

Second are the central banks. Aside from servicing the exchange requirements of government and sometimes other central banks, they generally do not provide market-making services, in more advanced markets at least. Charged with the broader responsibility to maintain reasonable market order, they may stand ready, as a residual supplier of domestic or local currency, to ensure that the foreign exchange market clears at a given exchange rate (depending on the exchange regime), but they are most conspicuous in the market when intervening to manage the exchange rate to absorb market pressures, when these are judged to be excessive. Central banks generally restrict dealing to the local market (reserves management transactions aside). Although at times they may deal overseas, or through nonbanks (e.g., when wishing to conceal their hand), active speculation by central banks on exchange rates is rare and frowned on by the central banking community, as it would involve taking a position against other central banks.

Third are large corporations, including transnational enterprises. They account for the largest net flow of funds across the exchanges and therefore have the greatest overall impact on currency values. Moreover, when they alter their hedging policy, accelerating or delaying cover for example, the market disturbance that can be caused is significant. Transnationals and other large corporations generally have little appetite for currency risk and creating exposures for the sole purpose of active speculation (as opposed to hedging) would be unusual for them.

Fourth are the derivative dealers, including the derivatives desks of banks. Although controlling relatively small business volumes, they can cause major, if short term, currency unrest when hedging positions in the cash markets. This is especially so when computer programs signal that an option book should be hedged in the cash markets, after the spot rate moves through a strike price. Most derivative dealing is based on arbitrage, where risk in one part of the market is hedged elsewhere. Although futures are used more widely now as a tool for speculation, a derivative dealer’s appetite for open risk is generally modest.

The last category of agents is the nonbank financial institutions, including fund managers and the currency hedge funds. To the extent that they invest in longer-term analysis and forecasting, they may be among the first to see upcoming currency pressures and realignments, which they can be quick to hedge against or exploit with large deals that may be temporarily destabilizing. This characterization is often seen to apply to the currency hedge funds, in particular, who can take exchange rate bets of a medium to longer-term duration, especially in the forward or derivative markets. They are seldom tempted to trade positions but rather retain them, once opened, until an objective is realized or until the decision is taken to cut them. Apart from the high-risk appetite of the currency hedge funds, the interest of other financial institutions tends to be tied to the protection of underlying investment portfolios. As the recent IMF study of hedge funds concluded, however, these organizations are not always “first,” nor are their expectations always accurate, let alone “self-fulfilling.”76 The study noted that, while fairly prominent, hedge funds were not conspicuously ahead of other important market players in Thailand’s recent currency crisis; that, if anything, they may have lagged behind other players in the speculation against other Asian currencies; and that while they earned their reputation to a large extent in the attacks on EMS currencies in 1992, they appear to have been caught largely unaware by the international bond market turbulence in 1994. Moreover, individual hedge fund deals may be large because of leverage possibilities and may cause market disturbance, as they can often obtain leverage of up to 20 times on an overnight basis, and even 50 times intraday, on their underlying investor resources. Also the cash they receive from investors can itself run in to hundreds of millions of dollars. Nevertheless, they are usually small in the overall scheme of things, when set against longer-term corporate and investment sector position taking or against the transaction sizes involved if banks switch their net positions, within established prudential limits.

Market Turnover and Products

According to the 1995 triennial BIS survey of foreign exchange activity in the 26 largest centers, turnover in all foreign exchange products amounted to some $ 1.26 trillion a day. Allowing for exchange rate adjustments, this was an increase of 30 percent from the previous triennial survey, more or less the same growth rate shown in the previous three-year period. Activity between banks was 64 percent of 1995 turnover, between banks and other financial institutions, 20 percent, while the balance of 16 percent represented business between banks and their nonfinancial customers. Daily turnover of traditional products averaged $1.19 trillion, of which 44 percent was spot, 7 percent outright forwards, and 49 percent swaps. Currency futures and options contributed an added $70 billion of turnover, which, although small by comparison, was still significant in absolute terms. Of the outright forwards and swaps, 53 and 71 percent, respectively, was concentrated in maturities up to one week (used by banks to manage liquidity), while business over one year was “rare.” The dominant form of trading in such derivative products has been the “over the-counter” (OTC) interbank market, rather than the exchange-traded alternative that requires more standardized products. Not large enough to be mentioned in the BIS survey, a recent and innovative product that is of growing importance in the trading of emerging market currencies is the “nondeliverable forward.” This product is settled for a cash consideration on a net basis—that is, only the net gains or losses are settled on maturity, with no exchange of principle sums. Turnover is estimated at $500 million to $1 billion a day, and it is understood that they have been actively used in recent trading of the Asian currencies, both in onshore and offshore currency markets.

Unfortunately, the BIS survey does not directly cover activity within emerging markets, (South Africa excepted), although $87.8 billion, or 7.7 percent of daily turnover, was a “residual” figure representing activity in the currencies of countries that did not participate in the survey. Nevertheless, there has been an unmistakable shift in dealing emphasis toward emerging markets in recent years. Deutsche Morgan Grenfell, for example, estimated in 1997 that 25–30 percent of its worldwide foreign exchange revenues came from exotic currencies, while Bank of America reported 25 percent of its activity was emerging-markets driven, up from only 5 percent in 1992. Exotic is market terminology for currencies that do not yet benefit from the large volumes, market liquidity, and well-developed infrastructure of most industrial country currencies and that may be prone to greater volatility. Meanwhile, even before the Asian crisis, HSBC Midland calculated its turnover in exotics had doubled in the preceding three years, and in the last few months Société Générale, Parisbas, and Standard Chartered have all announced the creation of new trading desks for emerging markets.

There are several reasons for this shift. The first is the underlying business need, with banks responding to their customers’ growing cross-border investment and trading needs, as local economies open to the outside world. The second is that banks are drawn by the fatter margins in markets where the arbitrage community is not yet present in great numbers—in so doing, of course, their own actions tend to drive down those margins as additional suppliers compete. A third reason is that EMU convergence has released dealing capacity in many banks, which has allowed them to bring forward their expansion plans for exotic currencies. It is estimated, for instance, that 20 percent of European foreign exchange business, and up to 40 percent in some countries, will be eroded by EMU. For example, the BIS survey showed almost a fourth of the daily $58 billion turnover in Paris was deutsche mark-French franc, all of which becomes redundant, post-EMU. Part of the capacity released will be absorbed by trading the new euros, but this still leaves a considerable surplus of manpower, capital, and credit resources for deployment elsewhere. Coinciding as it has done with the plans by banks to increase coverage of emerging markets, this EMU contraction has allowed those plans to be accelerated. The final reason for the increase in emerging market interest is the more general developmental phenomenon whereby markets grow vigorously once a certain critical threshold of liquidity is reached. As currency markets deepen and broaden for underlying business and other reasons, as above, this in itself encourages further entry. Liquidity creates confidence, which encourages more banks to do more business, which then creates more liquidity, and so on, at an accelerating pace. A not-too-distant historical example of this same phenomenon is the growth in the financial futures markets in London and Chicago. Doubtless, a number of the emerging markets are in the midst of this sort of process, notwithstanding periods of uncertainty like that associated with Asia recently.

A further noteworthy point here is that the rapid advances in global communications have rendered the concept of locally domiciled trading a thing of the past for many of the world’s currencies. Instead, they are today traded continuously and freely across national borders and time zones. At the same time, banks have established global networks of dealing offices to service their clients and to spearhead moves into new markets, but these offices also serve as havens to rehouse dealing books when regulatory constraints in any particular market become onerous. When economies begin opening to the outside world, there is an inevitable erosion of central control that goes with it, because governments have no direct jurisdiction over the free trading of their currencies in offshore centers. Nowadays, dealing quickly migrates when the business climate turns adverse, as a trading book requires little more than good communications and a good address to be operational, both of which can be readily found elsewhere in the global networks of dealing banks. The BIS survey underscored the fact that it is now commonplace for significant volumes of foreign exchange dealing to take place outside the country concerned. It reported that more dollar, deutsche mark, and French franc business was handled in London than in their respective countries, while only 30 percent of Swiss franc turnover actually occurred in Switzerland.

With the rapid growth in foreign exchange turnover, an increasingly recognized issue is the additional risks involved during settlement of both legs of a foreign currency transaction. Bank management and regulators have been concerned at the costs and credit risk issues in foreign exchange settlement for many years. Yet despite many high-level meetings, no one has come up with a solution that enjoys widespread support. A number of the main trading banks have acted on their own, however. For some time they have netted their deals bilaterally, combining all settlements due for the same value date into a single exchange of payments. To automate the bilateral netting process, several systems have been set up, including FXNET, which was established in 1986 by the dealing management of 14 active banks. It now nets the bilateral dealing of 61 participants, which account between them for 10–13 percent of global foreign exchange turnover. In so doing, around $100 billion of daily clearing is now eliminated. More recently, the Exchange Clearing House (ECHO), owned by 36 major banks, finally went live, offering the advantages of a full multilateral clearing facility. According to an internal Echo study, if multilateral netting were accepted universally, global settlements would be reduced by up to 95 percent. To date, the volumes handled by ECHO have been modest, around 1 percent of the market, and the take-up has been slow. Indeed, it has been far from clear that ECHO per se would be very actively supported, because the big dealing banks have already eliminated a major part of their settlement problem through FXNET at a charge of $1.50 a transaction, while ECHO, charging $5 a transaction, seems an expensive and rather marginal alternative. Partly reflecting such doubts, Echo, along with Multinet (another multilateral netting service), has just been merged into a new organization, Continuous Linked Settlement Services (CLSS). CLSS was established by the “G-20 banks,” and the previous shareholders of ECHO and Multinet are becoming shareholders in CLSS. CLSS will continue to provide netting facilities but is also planning to establish a bank to handle foreign exchange settlements among member banks. It will be linked with members’ own domestic, real-time gross settlement systems to provide simultaneous settlement of both legs of a foreign exchange transaction.

Mechanisms for Dealing

Currency dealers seldom now meet on a trading floor to transact their business, except in markets that have yet to progress beyond central bank auctions. Instead, they trade through terminals or voice brokers. The reason is that floors do not have the capacity to accommodate the often large numbers of banks that raise many deals with each other each day; nor can they accommodate the many screen-based services that banks need, to manage their portfolios. But in addition, the costs of a trading floor tend to be prohibitive—witness the current debate on the floor-based London International Financial Futures Exchange (LIFFE) exchange moving to cheaper screen-based dealing.

The major portion of interbank foreign exchange dealing is still negotiated bilaterally between banks following the traditional pattern of price inquiry followed by a deal, although the former use of telephone and telex has now been largely superseded by systems such as Reuter Dealing 2000–2001. The latter is a screen-based communication system, which is commonly used around the globe. The advantages of Reuter Dealing 2000–2001 in particular are its speed, integrity and reliability, and its facility for making multiple calls (up to four) simultaneously. In addition, its audit trail and electronic data feed to banks’ mainframes greatly reduce the scope for costly dealer and processing error, and for fraud. Nevertheless, voice brokers continue to negotiate important volumes of business for the banks, especially in foreign exchange swaps.

Beyond the regular dealing systems like Reuter Dealing 2000–2001, newer electronic systems are changing the nature of market trading in a deeper way. In particular, within the last few years, an electronic alternative to the price inquiry and deal systems has finally overcome dealer prejudice and been successfully introduced to the market. The automated deal matching systems of Electronic Brokering System (EBS)—owned by 14 major banks—and Reuter Dealing 2000–2002, are now estimated to handle around 60 percent of all the brokered business in London, while they have virtually replaced the voice brokers in Asia.77 These deal-matching systems function through desktop terminals, with the amounts and rates to be dealt keyed in by the dealer. It is not possible to offer different terms for counterparties of different credit standing, but each dealer’s terminal is preprogrammed with the names and limits for acceptable counterparties. Only when there is a match with an acceptable counterparty are the names of the counterparts exchanged, but revealed to no one else.

The systems particularly suit the market’s middle order transactions of a few million dollars each, which they match smoothly and uneventfully—although larger deals still need to be negotiated directly between banks (the EBS average deal is just under $3 million). But the systems trade on such fine dealing spreads—typically between 1 to 3 basis points compared with the 5 points previously common—that market makers find the systems are making serious inroads into their traditional business. It is a trend that seems likely to continue. Indeed, there is a clear prospect that electronic order matching could become the dominant dealing medium in future. With the loss of a significant part of their business, voice brokering is going through a significant retrenchment. Yet, the voice broker’s skill at securing prices under even difficult market conditions means they will probably remain important in many less liquid markets that do not have the turnover to support electronic matching.

Both EBS and Reuter see significant growth ahead. EBS already publishes its worldwide turnover figures. A recent report showed over 32,000 EBS transactions a day, totaling $94.3 billion. With 760–770 clients, compared with Reuter’s 1,200 or so but with smaller turnover, EBS claims 46 percent of all brokered business in London, and as much as 80–90 percent in Asia. Reuter has around 15 percent of the London market, according to EBS. Reuter Dealing 2000–2002, meanwhile, covers 35 currency pairs in 33 countries, and is in the course of extending its coverage to emerging markets. Their current move to cover the Mexican and Russian markets is the forerunner of this. Until recently, the two providers handled only spot dealing, but Reuter has just launched a service in forwards for 16 currency pairs in 21 countries—Reuter Dealing 2000–2002 Forwards. EBS is expected to follow shortly, while the brokers are marketing their own automated forwards systems. Conscious of the fundamental change in the way the market does it business, the BIS is to collate data on automated brokering for inclusion in its next triennial survey.

The virtue of electronic matching for regulators and administrators is its relative ease of surveillance and price discovery, and the heightened efficiency it brings from reduced spreads. The operating costs for the banks are not particularly large either. For example, the initiator of a transaction on Reuter Dealing 2000–2002 pays $25, which is insignificant when set against the $326 benefit from improving, say the dollar—deutsche mark rate by 2 basis points on a $3 million deal (e.g., from 1.8400 to 1.8402). The monthly charge for a Reuter Dealing 2000–2002 communication terminal varies between $2,000 and $4,000. A disadvantage that some see, however, is the possibility that such automated systems may not do as good a job in preserving liquidity as voice brokers do, at times when markets are especially turbulent (see below).

Profitability and Competition Trends

Foreign exchange business is an important source of income for the banking community. Even with the EMU-related stability in Europe, the competition of electronic dealing, and the growing relative importance of nonbank financial institutions, revenues for the major banks listed in Table 39, as a group, were only slightly lower in 1996 (a relatively quiet year) than in the previous years. Meanwhile, early results for 1997 suggest a strong resurgence of revenues. Citibank posted record foreign exchange income of $1,225 billion for 1997, while Chase Manhattan Bank nearly doubled its revenues. In the United Kingdom, foreign exchange earnings at HSBC Holdings were up 72 percent to $1,004 billion and at Standard Chartered up 84 percent to $498 million. One important reason for this is that banks have been shifting the focus of their business away from traditional market making and proprietary trading, where margins are relatively thin, to client-related business (selling more profitable complex derivatives, for example) and to emerging markets (where profit margins are better).

Table 39.Foreign Exchange Trading Revenues(In millions of U.S. dollars)
19961995Percent Change
HSBC Midland597.0609.0-2
Swiss Bk Warburg594.0422.6+41
Chase Manhattan444.0584.0-24
JP Morgan320.0253.0+26
Bank of America316.0303.0+4
Standard Chartered261.9303.4-14
State Street126.0141.0-11
Barclays BZW107.0145.3-14
Republic Nat Bk98.0113.0-13
American Express72.079.0-9
Bank of New York67.060.0+ 12
Sources: FX Week and Deutsche Morgan Grenfell from Financial Times Survey, April 18, 1997.
Sources: FX Week and Deutsche Morgan Grenfell from Financial Times Survey, April 18, 1997.

Another related factor has been a continuing trend toward bank consolidation, affecting foreign exchange business and other areas of banking. The BIS survey notes the continuing trend for foreign exchange business to be concentrated in the hands of fewer banks, with the top 10 banks controlling 44 percent of 1995 turnover in London, 47 percent in the United States, and 51 percent in Tokyo. Seen another way, 75 percent of 1995 turnover in the six largest centers was accounted for by 11 percent or less of banks. Yet this trend does not appear to represent a diminution of competition in major foreign exchange markets—more the opposite, in fact. It reflects the intense competitive pressures in the financial sector more broadly, which are encouraging rationalizations, mergers, and acquisitions, as previously distinct—if not protected—market segments are breaking down internationally.

Market behavior and Policy Interactions

This section examines major foreign exchange dealing considerations and how dealing behavior can affect the movements in the exchange rate.

Nature of Market Making

Whether or not banks operating in the foreign exchange market engage in market making depends on the level of their in-house trading skills and their appetite for risk. Those with limited ambitions confine themselves to servicing customer needs alone and immediately lay off positions with market makers. But banks that elect to be market makers additionally offer two-way dealing services to the interbank market—which places them in a strong position to bid for larger customer deals—while they also establish reciprocal facilities with each other to lay off their own excess risk. In all but exceptional circumstances, market makers stand ready to make firm dealing prices, where the spread and dealable sum are “understood” and consistent—for example, in major markets not more than a 5 point spread and good for a minimum $3 million. There are different tiers of these understandings allowing, for instance, for more ambitious banks to trade with each other for bigger amounts and on closer spreads. The bid-offer spread quoted by a market maker is determined by the competition both of other market makers and the electronic matching systems. Thus, for a large trade, a bank might expect to be quoted a 5-point spread (say 1.8400–1.8405 on U.S. dollar-deutsche mark, equivalent to 0.027 percent); as noted earlier, the electronic systems usually deal off lower spreads, possibly as little as 1 point, albeit for smaller transactions.

Box 10.A Day of Spot Market Making

The first concern of a dealer approaching a new business day is to discover the price. For the major currencies traded continuously around the globe this is easy: a London dealer, for example, checks the sterling price in Hong Kong SAR that may then be used as the first dealing quotation or the dealer may shade it up or down according to his preference. Alternatively, if the Hong Kong SAR price does not seem to reflect fully the latest news, the dealer could deal on it and open his first speculative position of the day. In the case of a local currency that floats but is not traded internationally, the opening price is more difficult to assess as there is no open market to check. So the dealer has to calculate its theoretical value and then adjust it according to his instincts. The basis is always the previous evening’s closing level, but this normally needs adjusting to reflect factors such as international market developments (e.g., a generally firmer dollar), local financial and economic developments, and less tangibly, political developments. Making the first price of the day is then something of a step into the unknown—so it is usually quoted with a wider spread to give some protection.

There is something of a race to be the first to make a call to another dealer because the estimation of the rate can then be checked. The dealer called, however, has had no opportunity to confirm his own assessment of the price but is obligated, nonetheless, to make a two-way dealing price. A dealer is only interested in firm prices—ideas or indications are of little value to him. So if the price quoted broadly agrees with the caller’s view there is unlikely to be a deal. But if this is not the case, two things can happen: first the caller could deal and open a position if confident his view was correct and the price quoted was wrong; or second, the dealer could amend his own view of the rate. By this process of calling and dealing, a broad consensus on the rate emerges.

An example of dealing could be the following. When a caller sells dollars to a market maker, the market maker encourages an offsetting deal to close the position and take his profit. The market maker does this by lowering his price (while maintaining the spread) to make the best dollar offer in the market. The next wholesale buyer of dollars ought then to deal with the dealer. But if having dropped the rate to encourage the offsetting deal, the next few callers did the opposite and sold dollars again—possibly the dealer had failed to realize the rate was even lower—the dealer might decide the position had become too large and that proactive action was called for to deal with it. Although dealers have a strong preference for dealing on their own price to conserve the spread, which is often the difference between profit and loss, they would in these circumstances call other banks and sell them dollars to reduce or close the position completely. And if they believed the price would fall further they could even sell more dollars than needed and take an opposite short position. Doubtless, at this point dealers would have made a loss, far from uncommon for market makers; but they would hope to make it up from the short dollar position they had taken and from the day’s later market making. As market makers can only control the prices they quote and not the deals that others do with them, it is impossible for a dealer to guarantee a square position at all times.

Yet a dealer must have broad control of the dealing book, the objective being that it is no larger than is consistent with the depth of the market—for example, that the position can always be closed with not more than four calls to other market makers. So it follows that market unrest that reduces the tradable amounts between market makers, dictates smaller positions. Equally, a spell of poor dealing normally calls for a lower risk profile. As the close of the day’s business approaches, a dealer will begin fine-tuning his book to ensure he can meet overnight limits. As all banks are doing the same, liquidity begins to contract. The dealer is then less accommodating to callers and will make wider spreads and deal for smaller amounts. Finally, when the position is as the dealer wishes it to be, the dealer will announce to callers that he is closed for business for the day.

Since a market maker is under a virtual obligation to quote dealing prices on demand, it is difficult for him to regulate his residual position with any precision, given that a counterpart’s deal may not always suit the book. But with markets unpredictable and potentially dangerous, it is most important that the broad shape of the book is controlled. There are two techniques dealers use for this. First, they vary the rates they quote, moving them higher or lower and possibly widening the spread, to encourage or discourage counterparties. Second, there are times when they must aggressively lay off risk on the prices of other market makers, whether at a profit or a loss. In fact, knowing when to deal on the prices of others, and having the courage to do so, is generally held to be the key to a dealer’s survival (see Box 10 for more specifics on market makers’ typical activities).

An active market maker trading one of the major currencies could easily handle many hundreds, or in some cases thousands, of deals each day with turnover running to several billion dollars. With such magnitudes transacted, even quite small price movements can make a significant difference to profits and losses. This leads to a culture of instant decisions. Prices require an immediate response and, even then, may be changed several times on the same call. Indeed, to allow for this, the electronic systems have been designed so that dealing prices may be placed under immediate reference by means of a single key stroke. Against a background of continual exchange rate uncertainty there is an inherent tension in dealing rooms that is fanned by loudspeakers disseminating constant news and rumors from around the world. Hence, perhaps, the common image of frenetic and very short-term oriented decision making. Yet this popular image misses important parts of the story, as discussed further below.

Importance of Liquidity

Like other asset markets, foreign exchange markets need high levels of liquidity to function most efficiently. Liquidity may be described as a market’s capacity and readiness to accommodate deals at fine prices without undue disturbance—that is, its ability to absorb even large transactions, without moving the market price unduly against those transactions. Beyond the benefits for the ultimate end users, liquidity is the lifeline that allows market makers to lay off risk and control exposures; allows an exit when dealing decisions go wrong; and exerts a constant downward pressure on transaction costs. As already noted, liquidity emerges only slowly in a market at first, but then, once it reaches some critical mass, tends to become self-perpetuating, growing more rapidly as confidence in the market begins to takes hold. In conceptual terms, market liquidity can be seen as a form of positive externality or public good, providing benefits to all actual or potential users of the market.

Liquidity is produced in several ways. First, liquidity is produced when continuous and natural two-way business permits deals to be absorbed uneventfully, as with electronic deal matching. Second, it is the product of banks agreeing to make markets to each other, for even without an immediate deal to offset an inward transaction, a market maker has to be prepared to accept uncovered risk on his book, at least temporarily. Third, the act of speculation generates liquidity by allowing imbalances in supply and demand to be carried forward to a later time, in anticipation of a reversal of the market. In this regard, the intraday and overnight speculation of the banks play a key role. Indeed, no financial asset market could operate efficiently without speculators to provide liquidity.

To a large extent then, liquidity is the result of banks being ready to accept risk on their books; but before this can happen, it is first essential that management is confident that their dealers possess the necessary skills to do so. However, the skill of dealing is less a question of mathematical competence than of learning to understand market “psychology”—how others are likely to react to developments in a changing and uncertain environment. But these are not skills that are acquired from books; they must be learned from experience, through sitting at the dealing desk and actually dealing, and this may entail some interim costs in the form of dealing losses. Exchange rate quotations are a sensitive and revealing barometer of currency value that is constantly adjusted to reflect current and expected order flows and prevailing sentiment. As such, it is the key indicator to help market makers manage risk and it is therefore imperative that they be able to interpret its everchanging message. They must, for instance, understand the significance of competing bids and offers, how banks respond when prices are quoted to them and how they react when a deal is made on their price, whether dealing spreads change, and so on. The result is that management is most unlikely to invest in training and expose its bank to the prospect of losses, unless it is clearly in its interest to do so—and the abstract cause of market liquidity on its own will not be a sufficient motive. In short, without a profit motivation, no skills will be developed; and with no skills, there will be no market making and no meaningful liquidity.

The large currency markets have high liquidity and routinely accommodate large deals without stress. However, some commentators feel that the growth of anonymous electronic matching systems may erode this, especially, in future episodes of serious currency unrest. Before automated deal matching became a force, market making was a well-oiled practice that allowed risk to be laid off through direct contacts between dealers and with minimal market disturbance. Voice brokers, too, played a part in this through generating a regular stream of competitive bids and offers in all but the most difficult conditions. The direct human challenge, dealer to dealer and broker to dealer, was an important element in the process. But although the electronic systems have worked admirably so far, confidently handling a growing share of middle and smaller market business, the fear of some is that these orders could be withheld at a time of future turbulence. And since electronic matching provides no facility to challenge a market maker directly for a price, they can, as it were, hide anonymously behind their terminals. Meanwhile, on this view, the old market-making skills are being lost as a new generation of dealers is brought up on electronic dealing, and the “understandings” between banks are being allowed to fall into disuse. Still, even if the above concern is valid in principle, it remains to be seen whether electronic matching systems will in fact largely replace the price inquiry and deal process—especially if markets recognize and value the advantages of more direct contact. As noted earlier, larger transactions are still usually negotiated directly between banks, while smaller markets may not have the turnover to support electronic matching.

Position Taking, Risk Management, and Supervision

Foreign exchange is one of the few areas of banking where large sums of money can be lost very quickly—which is why both management and supervisors impose strict controls on the operations of dealers. The precise requirements of the two are different, however, since management seeks to prevent any erosion of shareholder value, while supervisors concern themselves with the sort of catastrophic loss that places depositors’ funds at risk and possibly threatens systemic failure. Well-managed banks scrutinize every aspect of dealing that could give rise to losses and usually impose a comprehensive range of mandatory internal controls that tightly circumscribe the exercise of any dealer discretion. These will include limits on market and counterparty risk and would normally require that positions are regularly marked to market (revalued), with both realized and unrealized profits and losses accounted for. Supervisors, on the other hand, are generally more concerned with overnight and longer-lasting currency positions. Beyond exposure limitations perse, they typically require bank management to demonstrate that they have a proper understanding of the hazards in foreign exchange and that they have installed effective monitoring, measurement, and control systems so that the risks of significant loss are reduced to the barest minimum. Intraday currency exposures, though larger than those held overnight, are not generally subject to supervisory limits (although they may well be curtailed by management). Supervisors consider them to be less dangerous in general, in that they can still be closed in a working market if things go wrong, which is not the case overnight when the dealing room is closed. Yet this perhaps gives a misleading notion of overnight risk, because banks frequently protect their overnight exposures by leaving stop-loss orders with centers that are open.

In the course of a day’s dealing, banks may accumulate or initiate substantial open currency positions—especially when acting as market makers. These actions may have a noticeable impact effect on the rate, but, because they must be largely squared before the end of the day, their long-term consequences are seldom significant. Positions carried overnight or for a longer term, on the other hand, can be of greater consequence since they cause a net movement of funds across the exchanges until closed. If banks use their limits to the full, and simultaneously change from an overbought to an oversold position, or vice versa, the impact on the rate could be clearly felt. Yet, this power is frequently exaggerated, because in practice banks never act in unison and to the full extent of their limits. Moreover, supervisory limits are usually highly restrictive. For example, in the past, at least, industrial country supervisors typically set such limits at around 20 percent of a bank’s capital, which would only expose a bank to a direct loss of about 5 percent of shareholders’ funds, even for an exchange rate movement of 25 percent. Levels of 15–20 percent have been common in industrial countries, though comparability is made more difficult by different calculation methods. A number of industrial countries have already moved away from direct open position limits or are in the process of doing so, on the view that a more holistic view of banks’ risk exposure and risk management is desirable. The apparent latitude now afforded some banks to decide their own levels of overnight risk under broader market risk-based capital requirements, in theory, opens the possibility of greater influence on currency values. But it seems unlikely that this will lead to any pronounced change in banks’ behavior because, by aggregating the limits for their branches and subsidiaries, the banks concerned have probably not been short of facilities anyway, while U.S. banks have been free of a direct regulatory constraint on open positions for many years.

A dealer will open and close many different and perhaps alternating long and short positions during the day to exploit anomalies. Thus, a currency that has risen suddenly and sharply could be due for a correction, and a short position would be opened. But because markets are unpredictable, a position could be rapidly closed or even reversed if an expected move did not soon occur, even if it meant taking a loss. Indeed, the final tally of the day’s performance for an active dealer would be the product of many individual profits and losses during the session. An overnight position taker may also exploit market inconsistencies. Thus an exposure could be taken against the closing trend in the expectation that an end-of-day imbalance was temporary and likely to reverse the following morning. Yet, at other times a position in line with the closing trend could be justified on the basis that trends frequently perpetuate themselves.

Dealing Behavior and Fundamentals

It is often assumed that a dealer’s overwhelming preoccupation is with matters of profit. Although that is important, the greater preoccupation is likely to be the fear of loss, since losses arise all too easily in a tense trading environment where time horizons are very short and positions must be closed each day to satisfy internal and supervisory limits. This is compounded by the uncomfortable reality of dealing that the worst losses invariably exceed the best profits—the reason being, first, that human nature being what it is, profits tend to be taken too soon and losses run for too long; and second, when the consensus expectation of dealers is wrong, competition among them to cut positions may drive the rate further away from their previous expectation and increases losses. Conversely, when the market has correctly anticipated a trend, the action of taking profits arrests the trend and reduces additional profit opportunities.

Spot dealers are not primarily concerned whether their decisions are rational or not, in the sense of being consistent with economic fundamentals. In the end, the only logic that makes sense, in this context, is to be long of a currency appreciating and short of one depreciating. Observed price and volume movements, in other words, are the dealers’ main short-term indicators of market sentiment, and most of the time no amount of economic analysis and forecasting is likely to provide much additional help in picking movements in the very short term. The result is that, up to a point at least, a dealer may follow a trend on the simple premise that trends tend to perpetuate themselves. In fact the momentum of trends is often so powerful that it may take a brave dealer to be the first to oppose one, even though many may believe a trend has gone too far.

Short-term dealers are, therefore, often influenced by the forecasts of technical analysts (chartists). Extrapolating from trends, technical analysis is widely used to endorse or even initiate trading decisions, so much so in fact that it can cause bouts of short-term price instability when signaling price thresholds that, when breached, trigger additional self-fulfilling movements. Charts may also be used to support longer-term dealing decisions although, naturally, with the passage of time the study of past trends is less relevant.

The value of broader “fundamental” analysis is more important in dealing, however, if the time scale is long enough to allow for short-term market wrinkles to be ironed out, or when there is a need to interpret significant news releases. Even here, however, the dealer’s own knowledge of how markets react to news may be just as helpful in the very short term. In support of longer-term or proprietary trading, banks make extensive use of the services of external as well as in-house economic analysts. Thus, it can be quite misleading to think of foreign exchange markets as excessively driven by “short termism” based on an extrapolation of the behavior of spot interbank dealers. This point is discussed in the conclusion of this section and linked to some conceptual analyses of exchange market behavior.

A different facet of exchange market behavior is the view that major banks may be able to manipulate markets for their own gain. The view has some validity, up to a point, but again it is important not to overstate the importance of, or understate the constraints on, such behavior. Specifically, a prominent bank can leverage a strong position and make exchange rates move profitably by inducing the market to draw the wrong conclusions about its business. It could do this in the following way. The customer orders that banks handle are confidential and when large enough, they can move the exchange rate. As a result there is continual guessing about what a bank is doing when it deals aggressively: is it operating for a customer or is it dealing on its own account? The distinction is important. Customer orders can have a lasting effect on the rate, whereas own-account dealing is more neutral because it must generally be unwound at the end of the day. A bank can exploit this uncertainty. It might, for instance, secure a large commercial deal at a price that, in isolation, produces a loss for the dealer. The attraction of the deal, however, is the opportunity it provides for leverage. If the dealer covers twice the amount of the customer deal in the market, and the market is left with a larger imbalance, the rate is more likely to change. Fearing the customer order was larger than it actually was, and that the price would move even more, other banks may then decide to cut their positions and take their losses. This may then cause the rate to fall sufficiently for the original bank to take a profit on its excess position, which more than covers the loss on the underlying commercial deal. From beginning to end, the exercise would typically take not more than 10 minutes. Once successfully accomplished a few times, the instigating bank might even establish a reputation for itself, not for its market leverage, which the banks would remain unaware of, but for its dealing skills—that is, the fact that the market always seemed to follow the direction of its dealing. At that point, the other banks would be inclined to follow whatever the bank did. This would increase the initiating bank’s short-term powers even more, for without an underlying order, it could make the rate move just by dealing with several banks. As long as it is not too ambitious, a bank operating in the market this way can fairly easily cover its tracks by channeling deals through different market makers, so that no one knows for sure who is behind the activity.

For these tactics to be successful, however, it requires a background of currency instability and uncertainty, with the exchange rate fluctuating relatively sharply in both directions. In such conditions, a market might offer little resistance to a new trend started by a leveraging bank that concealed its hand. And the sums needed to start a market moving depend very much on circumstances. If, in one of the large markets, sentiment were to be fragile, $100 million might be enough; whereas in a more confident and deeper market, $500 million could be readily absorbed with little effect on the rate. As a percentage of market turnover these may not be large sums, but the danger for a leveraging bank is that larger positions carry higher risks if the stratagem fails, which can easily happen. Meanwhile, although exotic currencies can be moved with smaller volumes, positions must still remain consistent with market liquidity to allow for a ready exit. Of course, the added consideration in exotic currency dealing is that smaller liquidity permits a defense to be mounted more easily by the central bank.

If a bank tried leveraging its position in an orderly market, however, it is much more likely to encounter stiff resistance as confident banks willingly absorbed the pressure without disturbance to the rate. They would look upon the attempted manipulation simply as an opportunity to obtain cheap currency. Even in quiet and relatively stable markets, a single bank could impose its will if it has a particularly dominant competitive position, perhaps with privileged access to major clients. Nevertheless, leveraging is seen by most market practitioners as a legitimate tactic in a market of professionals, one they would like to engage in themselves given the opportunity. Few in the market consider it to be improper or unethical, as a bank that leverages its book does not deliberately deceive the market in the sense of spreading misleading or false information, which would be quite a different issue. Rather, it realizes its objectives by doing the opposite and saying nothing, instead allowing other banks to draw their own conclusion about its business. In fact, it would be most unusual that a bank attempted to deceive the market willfully. Were it to do so, it would be dealt with severely by the market itself (through ostracism) and quite likely by official sanction too (possibly withdrawal of a foreign exchange license).

Policy Interventions in the Foreign Exchange Markets

Policy interventions in currency markets take several forms, ranging from market-based operations to regulatory measures that aim to restrict market behavior. The most basic policy intervention, however, is the choice of exchange rate regime, which can have a great deal to do with the incentives for foreign exchange market development, and more particularly for the active management of currency risk. When central bank currency transactions are conducted at rates that are predetermined, and excessively rigid—under fixed or pegged regimes within very narrow bands for example, or when a “floating” rate is never actually allowed to move—there is little or no uncertainty and the banks have no risk to concern themselves with, at least not unless or until the regime collapses. Customer deals are covered directly with the central bank (or matched in-house), which is little more than a clerical exercise that develops no risk awareness. But if instead banks are given no guarantee of direct access to the central bank at “known” prices, and if intervention is channeled at varying rates through broader market operations, rather than made available to individual banks, on demand, at an open window, then the element of uncertainty produced would compel the banks to acquire risk management expertise and train their dealers in such aspects. The point is valid even under a fixed or pegged regime, provided it allows at least some movement in market rates around a central rate. Individual banks would not know for sure where, when, and at what price the central bank would next appear. The better incentives for risk management by banks would tend to stimulate currency market trading and growing liquidity, while also encouraging nonbanks to pay more attention to exchange rate risk. The Asian crisis amply demonstrated the costs that can be involved if excessively rigid exchange rates encourage either banks or nonbanks, or both, to ignore exchange rate risk.

The recent events in Asia and elsewhere have also reconfirmed old lessons about how to defend and how not to defend specific exchange rate levels. In a nutshell, exchange rates that are clearly unrealistic will probably not be defensible and the many examples of central banks attempting to defend unrealistic exchange rates over the past few decades have provided currency markets with a rich source of income. Yet, there is a significant gray area here, as to what is “unrealistic,” and examples of fixed or targeted rates that have been successfully defended against substantial speculative attacks abound. The most important factors appear ro be twofold.

First, the authorities need to display a strong commitment. Specifically, they need to be clearly willing to bear the costs of a defense, particularly in terms of higher interest rates, but also in terms of fiscal retrenchment and determined financial sector and other structural policy changes, as required. For several reasons, interest rate defenses are often seen as controversial (and certainly they can be painful). Interest rate defenses may not always work, and even if they do work, they may involve high interest rates, at least for a time. For example, Sweden’s interest rate defense in 1992 was unsuccessful, even though short-term interest rates rose to over 500 percent. (Short rates, on an annual basis, are required if a depreciation is widely expected to be imminent: as a simple illustration, if a 10 percent depreciation is expected within a month, an annual interest rate of some 214 percent would be required on one-month domestic currency investments, just to compensate for the expected exchange rate loss.) Moreover, more sophisticated speculators will have already locked in domestic currency liquidity and credit at lower interest rates, as they opened speculative positions against the local currency. But neither of these concerns is an argument against increasing interest rates if the policy objective is to defend the existing exchange rate. The point of interest rate responses is primarily to affect the behavior of those who have not yet committed themselves to speculative positions against the local currency, so that the market is not one-sided. If those who have already taken speculative positions are thereby forced to close out their positions at a loss, so much the better. But that is a side benefit. Overall, interest rate increases may not be sufficient to avert a currency collapse, but they are typically necessary.

Second, the authorities need to act early. Often, authorities have allowed speculative pressures to build up, absorbing these through continued direct intervention in the currency markets but not allowing these to spill over into higher domestic interest rates (let alone contemplating preemptive interest rate increases). The temptation to delay a strong defense, and simply absorb speculative pressures initially, helps explain the often observed phenomenon of an apparently sudden eruption of a currency crisis, even if an underlying fragility had been clear for some time. In many instances, authorities may have also delayed the buildup of full-blown speculation by successfully massaging opinion, convincing the markets for a time that the exchange rate is credible and sustainable. But when the official will and capacity to continue a defense, and broaden it beyond sterilized foreign currency intervention, begins to be questioned, then speculation begins in earnest. Central banks need to be willing to act quickly to restrict the domestic currency credit they themselves are supplying, through one means or another, that is, fueling a speculative attack and holding down short-term interest rates. The need for policies that, to the extent possible, are preemptive is valid for both interest rate and other defensive measures. If extremely high short-term interest rates are needed because speculation has been allowed to build up too far, the market has to take account of the political, if not economic, sustainability of the rates. In other words, the interest rate defense may not be credible. The same general point applies to other defense measures as well, such as fiscal and structural policy changes. Markets look for determined, credible, and timely decision making in these areas, and undue delays or half measures raise doubts about the authorities’ commitment. In short then, “too little, too late” is the recipe for an unsuccessful defense.

Sometimes, central banks have conducted their foreign currency interventions in nonspot markets—forwards, futures, and derivatives, for example. Intervention in these forms has often turned out to be quite problematic, however, because there can be a significant temptation to overuse the intervention in the hope the speculative pressures will abate. These points do not relate to the use of such transactions for purely reserves management ends, rather than currency intervention; nor to the use of foreign exchange swaps as an instrument of domestic monetary management. This temptation arises from two factors that are often perceived, somewhat ironically, as the main advantages of such operations as compared with simple spot market or money market intervention. First, intervention in forwards, futures, or derivatives markets appears to offer the ability to defend an exchange rate while economizing on the use of official foreign currency reserves. The spot exchange rate can be influenced quite strongly by such operations, but without the immediate use of the same volume of official reserves and, correspondingly, without the same need to sterilize the foreign exchange intervention to leave domestic monetary conditions unchanged (if that is the policy). Thus, there may be no obvious limit to the intervention, or at least the limit is less obvious.78 If the pressures continue, however, the day of reckoning for the spot rate, and the central bank’s balance sheet, is not avoided.

Second, the use of nonspot interventions can be less transparent, especially when central banks’ publicly reported holdings of foreign exchange reserves do not take account of the actual or potential commitments under forwards, futures, or derivatives contracts. This was a substantive issue in both Thailand and Korea where, even though there were market suspicions that the central banks had a substantial amount of such commitments, their actual volume proved to be an additional unpleasant surprise when it became known. Authorities sometimes see nontransparency as a means of leading market participants to believe that there is underlying market (rather than official) support, presumably because, once there is a perceived lack of market support, it is inevitable that market participants will question how seriously the authorities are willing to defend a given exchange rate. Therefore, if there is a reluctance on the authorities’ part to take more fundamental actions to defend an exchange rate (such as allowing or initiating a monetary tightening, and making determined adjustments to fiscal and structural policies), the temptation to overuse less transparent intervention methods may be strong.

In any event, it is doubtful that nontransparent intervention methods are in fact superior to a clear official policy reaction function, as part of a transparent monetary and exchange management framework, with a single, clear, medium-term objective. A clearly oriented and relatively independent central bank may well help in this regard. In the case of a currency board arrangement, a credibility benefit arises instead from the fact that domestic interest rates adjust automatically to foreign currency flows, without the need for specific monetary policy decisions that might delay the interest rate response. Certainly, markets will start to suspect the existence of nontransparent interventions as soon as they begin to become sizable. At that point, non transparent interventions are more likely to be counterproductive, since they send a negative signal about the willingness of the authorities to bear the costs of a serious defense of the exchange rate.

The other type of policy intervention is to endeavor to reassert direct regulatory controls on markets. Apart from direct capital controls, examples sometimes advocated include measures like circuit breakers and market closure, aimed at the operation of the market; credit controls; a turnover tax to increase costs for speculators or taxes on short-term inflows; the reduction or withdrawal of supervisory limits and prohibitions on banks accepting “speculative” deals; and so on. Yet for countries integrating into the global economy, experience has shown that controls are seldom effective for very long, unless perhaps they are very repressive indeed. Such controls seek to combat economic incentives for capital flight, but in general they do not remove those incentives: rather they serve to convert them into incentives for circumvention of regulatory or tax-like constraints. Credit controls, for example, are complex to implement and difficult to reimpose once lifted. They are often the instrument of their own undoing, since they entice domestic liquidity to seek the higher returns that then become available in the offshore black market. Thus, even if effective in the local market, circuit breakers, market closure, turnover taxes, and constraints on the freedom of local banks all tend to drive business to free trading offshore markets. Corporate treasurers and their advisers are adept at finding ways to actually or in effect get money out of the country through such techniques as leading and lagging, creative intra-group invoicing, and—increasingly important these days—specially structured derivative products such as those with the payoff tied to an exchange rate. When the rewards are high enough, offshore demand will always be satisfied. Foreign exchange knows few boundaries, and in the absence of a strict regulatory body with a global reach, it will migrate and leak overseas to avoid the imposition of controls. It is, meanwhile, a moot point whether the imposition of controls might actually hasten capital flight by undermining investor confidence even further.

Yet it would be misleading to suggest that controls never work, because they certainly catch some agents. But the sophisticated operators are seldom among them, because they anticipate and cover themselves ahead of controls being introduced. Rather, those affected tend to be the domestic enterprises carrying unhedged currency exposures, with neither a proper appreciation of the risks involved nor the financial connections to do much about them, once they realize the predicament they are in.

Meanwhile, formal institutional controls on the structure and organization of the interbank market in the main dealing centers are mostly minimal, if indeed they exist at all. It is widely accepted, among industrial country regulators, at least, that markets perform best if they are allowed to evolve naturally within an appropriate broad incentive structure, free of burdensome overregulation. Freedom from overregulation safeguards the interests of the market’s end users by encouraging competition between banks, and this is reflected in finer prices and better service. Equally, when a market can adapt organically to changing external circumstances, the informal understandings governing relations between banks—bid-offer spread, deal size, and so on—are able to evolve smoothly. On the other hand, the general standards of behavior expected of practitioners are normally promulgated in an officially approved (but not usually officially designed) code of conduct, the object of which is to create a professional and ethical framework to foster good market practice and confidence. Most dealing centers have produced their own codes, all of which follow the same broad lines, and the international association of dealers has also issued a document with universal application, adopted and modified as appropriate in many less developed currency markets.


From time to time questions are raised about the “efficiency” or “rationality” of foreign exchange markets, particularly at times of currency crisis, and more generally whenever policymakers perceive that exchange rates are substantially “misaligned.” Sometimes, these questions seem to mainly reflect discomfort on the part of policymakers about the judgment on economic policies and fundamentals that markets are revealing through their actions. At other times, there may be deeper underlying concerns that markets are “failing” in some important sense. A critical issue in such cases, which however is often not very well-defined, is exactly in what sense markets are thought to be failing—that is, relative to what standards, and why? The “why” is particularly important because any public policy response to a perceived market failure is not likely to resolve the problem—and tends to create additional, unintended distortions instead—if the reason for the failure is not clearly identified and addressed as directly as possible. It should also be remembered that political and other pressures on policymakers can often lead to a strong temptation to find “quick fixes” that address symptoms rather than causes. In turn, both the “why” of the failure, and the definition of the standards against which failure is defined need to be based on an appropriate analytical framework that adequately takes into account how exchange markets actually behave. Defining “failure” relative to a model that bears insufficient resemblance to how exchange markets actually work would not be a sound basis for public policy formulation. With such a model, what appears to a “market failure” could very well be a case of “model failure” instead. For a general discussion of this issue, see, for example, Taumanoff (1984). In this regard, it seems fair to say that there is still a good way to go before generally accepted, sufficiently rich economic models are developed to adequately explain important aspects of exchange market behavior and the exchange rate formulation process, especially for shorter time frames. This section has sought to contribute to richer analysis by providing a more operational perspective on these issues.

The need for a richer analysis of foreign exchange market behavior can be illustrated in two specific respects highlighted in this section. One is the fact that interbank market dealers in particular often take short-term actions in the foreign exchange markets with scant apparent regard to analysis of “fundamentals.” In doing so, they often seem to rely on their own intuition about the behavior of clients and other dealers, and to varying extents also on “technical analysis” methods, such as charting. Such “noise trading” behavior is often at the core of models of speculative bubbles or attacks in foreign exchange markets, and of analyses taking the view that the exchange markets “fail.” Evidence for this view is often based on a survey of market practitioners presented in Taylor and Allen (1992). This point often seems to be extrapolated, however, to serve as a more general description of both interbank dealer behavior beyond the day-to-day time frame and the behavior of market participants other than the dealers themselves. More formally, the efficiency of the foreign exchange market as a whole may be seen to be compromised by the interreaction of “chartist” and “fundamentalist” behavior.79 But such an extrapolation beyond short-term dealer behavior is problematic for a number of reasons. First, since economic models of exchange rate behavior are well recognized as particularly weak over shorter-term horizons, dealers can hardly be accused of irrationality and inefficiency if they do not rely much on analysis based on them. In the absence of strong guidance from that direction, the most efficient response for dealers over very short horizons is to focus on the movements in market prices and volumes as summary indicators of market trends. Second, short-term dealer behavior is constrained by internal dealing strategies, position limits, and the like that are set on a somewhat less frequent basis but that are more likely to be influenced by “fundamentals analysis” (even if often not in terms of fully specified, quantitative exchange rate models). Indeed, the Taylor and Allen survey results are quite consistent with this, showing the relative importance of technical analysis falling off substantially as the time frame lengthens, and showing too that a substantial majority of survey respondents perceived that fundamentals and technical analysis were seen as complementary rather than mutually exclusive. Third, the balance between the use of fundamentals and technical analysis may well change according to circumstances, and in a way that, though relatively complex, may be quite consistent with rationality.80

The second aspect relates to nonbank end users in the foreign exchange market. While the dealers can be influential and may indulge in various strategies aimed at securing very short-term gains, in the final analysis, the positions they can take on their own behalf are relatively limited at day’s end, compared with the volumes that large clients can move. Given the quasi-public good nature of services provided by currency market makers, it is perhaps reasonable to view the scope to make such short-term gains as an indirect return to production of that public good. And furthermore, the signals and sentiments to which they are responding are for the most part derived from the actions of those large nonbank customers. The recent questions some have raised about the role of hedge funds in leading exchange markets demonstrate the point—such organizations are sometimes thought to be market movers who precipitate “herd behavior” on the part of other market participants. As pointed out in Eichengreen, Mathieson, and others (1998), it is not at all clear that this is actually the case for those institutions, more than others. But even if it were, the more important question is whether such behavior is irrational or inefficient given the environment in which markets operate, and in particular whether it is as divorced from fundamentals as the day-to-day activity of dealers appears (on the surface) to be.

In general, uncertainty and limited, costly, or heterogenous information are all inherent in the real world, and aversion to loss is a major driving force in the short term. Models that assume these factors away are fundamentally incomplete, while models that attempt to explicitly take this into account produce complex interactions. But these more complex models demonstrate that there can be a number of reasons why short-term dealer behavior may, to a large extent, reflect a quite reasonable approach to acquiring information on more fundamental market trends, in a decentralized market where different views on the future are held. They also demonstrate why apparently observed behavior such as herding, speculative bubbles, and the like may be quite rational.81 These considerations also seem to provide some shorter-term, and more micro, foundation for the view in some recent empirical work that exchange rates do seem to revert to fundamental (e.g., purchasing power parity) equilibria over time, and that they do not seem to be “excessively volatile.”82

Once one moves toward an understanding of why certain types of market behavior occur, it tends to become more difficult to draw simple conclusions about market failure. If key aspects of market behavior are in fact rational and efficient given the constraints under which markets inevitably operate, the issue for public policy then becomes more one of whether and how policy measures can improve the environment in which markets work, rather than directly intervening in market behavior. Central in this regard is the promotion of greater certainty about economic and financial conditions, and especially about the nature of current and future public policy, both at the macroeconomic and the microeconomic/structural levels.

Note: This section was prepared mainly by David Mitchem and Mark Swinburne.


For a more detailed description of the operation of the Reuters Dealing 2000–2002 system, see Chapter 4 of Frankel and others (1996).


Freedman (1991) makes the same point.


See, among others, De Long and others (1990).


An interesting recent attempt to empirically model the simultaneous existence of chartists and fundamentalists is found in Vigfusson (1997). This paper finds that, though chartists seem to dominate foreign exchange market trading for much of the time, this is typically in periods that are relatively tranquil or smoothly trending. In contrast, fundamentalists dominate in the less frequent periods of greater turbulence, but this turbulence is associated with reequilibration rather than the opposite. The result has some intuitive appeal—fundamentalists dominate when views of the fundamentals have changed, or where the exchange rate has drifted sufficiently far from what seems justified by the fundamentals that arbitrage and speculation to correct the discrepancy will be profitable.


Some recent attempts to model behavior of exchange market participants along these lines include Lyons (1991); and Peiers (1997). Also, Eichengreen, Mathieson, and others (1998) note several reasons why herd behavior, for example, may be quite rational.


On these points, see respectively, for example, MacDonald (1995); and Bartolini and Bodnar (1996).

    Other Resources Citing This Publication