Annex III Foreign Exchange Markets: Structure and Systemic Risks
- International Monetary Fund
- Published Date:
- September 1996
This annex begins with a summary of recent developments in the foreign exchange market. Drawing on the BIS’s most recent foreign exchange and derivatives turnover reports.1 current trends in the types of instruments and currencies, trading locations, and market participants are reviewed. The remarkable growth of electronic broking for spot market transactions over the last three years is documented and examined in light of growing concerns about its effects on exchange rate volatility. Market characteristics, such as liquidity—and its components, breadth, and depth—are also examined, providing a broad picture of the current workings of this important 24-hour global market. The second main section of this annex deals with the potential systemic risk posed by the large and numerous cross-border settlements accompanying foreign exchange trading. An analysis of the issues underlying foreign exchange settlement risk is presented along with a discussion of the private and public initiatives designed to reduce it. The section pays particular attention to the advantages and disadvantages of multilateral netting systems and global clearing banks as possible risk-reducing measures.
Global Market Developments
Average daily turnover in the global foreign exchange market, including spot, outright forward, and foreign exchange swap contracts, was estimated by the BIS in its fifth triennial survey at $1,190 billion in April 1995.2 The comparable figures from the previous triennial surveys of April 1992 and April 1989 were $820 billion and $590 billion, respectively. According to the survey, much of the 45 percent increase in turnover since the last survey is attributable to the depreciation of the U.S. dollar, the currency in which the survey reports results, against most major currencies; adjusted for these movements, turnover expanded by 20 percent over the three-year period. Including OTC foreign exchange options and currency swaps, and exchange-traded foreign exchange futures and options, adds $70 billion to average daily turnover.3
The geographic breakdown of the market turnover statistics reveals that the three major financial centers—the United Kingdom (London), the United States (New York), and Japan (Tokyo)—continued to maintain their dominance (Table 41).4 London increased its pre-eminence by recording 30 percent of the foreign exchange transactions in the 1995 survey, up from 27 percent in 1992. New York remained stable at 16 percent, and Tokyo’s share of turnover declined slightly to 10 percent (from 11 percent three years before). Two other East Asian centers. Singapore and Hong Kong, show signs of increasing influence. In the latest survey, turnover in Hong Kong increased by 49 percent and outstripped that of Switzerland, putting it fifth in the rankings. Singapore maintained the fourth-place ranking that it attained in 1992.5
After the foreign exchange market, the next biggest financial market is thought to be the market for U.S. government securities. Average turnover by dealers of U.S. government securities was $175 billion a day in April 1995, and it has grown at an annual average rate of only 5.7 percent over the last two years6 Average daily turnover in the world’s ten largest stock markets together amounted to about $42 billion in 1995, still only 3 percent of the volume recorded by the foreign exchange markets.
The foreign exchange market is universally regarded as the most liquid market in the world.7 In the foreign exchange market, a person can transact in any major currency at any lime during the day or night—it is the only truly global, 24-hour market. Although 24-hour “trading books” are maintained by many dealer banks, liquidity varies during the 24-hour period. Figure 1 provides a rough gauge of activity around the clock by examining tick changes, or changes in quotes, during the day from a foreign exchange quote vendor. The trading day starts in the Far East—Tokyo, Singapore, and Hong Kong—moves through Bahrain in the Middle East and continues into Europe—Frankfurt and London. It then moves to New York where the period of greatest liquidity occurs when both New York and London are open—mid-afternoon in London and early morning in New York. Between the closing of business in New York and its opening in the Far East, the market experiences a lull in activity, A striking feature of the market is that even during the business hours of a trading center, somewhere between one half and three quarters of the turnover is cross border, suggesting that many trades occur outside of business hours in the counterparty’s country.8
Using different measuring rods to examine liquidity, a typical dealing institution writes between 3.000 to 4,000 trading tickets for foreign exchange transactions during an average 24-hour day and about 50 percent more than that on busy days. Quoted prices can change 20 times a minute for major currencies, with prices on the dollar-deutsche mark rate changing up to 18.000 times during a single day. During periods of extreme stress, a single dealer may execute trades as quickly as one every two to four minutes.9
Although associated with liquidity, market depth refers to the number of orders placed at prices above and below the current market price, either explicitly held by an intermediary (in a “book” or an automated trading system) or implicitly held by market participants. A market may be liquid, in the sense that transactions can be executed rapidly with little price impact, but not necessarily deep. In a market that is not deep, news that moves price away from the current price results in a large price change since there are few orders surrounding that price. While it is difficult to directly measure depth in the predominantly OTC foreign exchange market, except for the most extreme market conditions, there generally are sufficient numbers of orders surrounding the current price that prices usually move in small increments, providing prima facie evidence that the foreign exchange market is deep.10
|April 1989||April 1992||April 1995|
|Amount||Percentage share||Amount||Percentage share||Percentage Change 1989-92||Amount||Percentage share||Percentage Change 1992-95|
|Total “net-gross” turnover3||717.9||100||1,076.2||100||38||1,571.8||100||46|
While estimated coverage of the foreign exchange market in most countries ranged between 90 percent and 100 percent, coverage in Austria and Italy was 86.5 percent and 77 percent, respectively.
Data for 1992 are not adjusted for local double counting.
Percentage change for 1989-92 is based on 21 reporting countries.
While estimated coverage of the foreign exchange market in most countries ranged between 90 percent and 100 percent, coverage in Austria and Italy was 86.5 percent and 77 percent, respectively.
Data for 1992 are not adjusted for local double counting.
Percentage change for 1989-92 is based on 21 reporting countries.
Figure 1.Intra-Day Tick Activity for U.S. Dollar/Deutsche Mark
Source: Guillaume and others (1994).
Market breadth refers to a situation in which orders surrounding the current price are considered “large” and come from widely divergent groups of participants. Transactions between $200 million and $500 million are not uncommon in the foreign exchange market and, at most limes, do not affect prices. Even in the less-frequently traded currencies, transaction sizes are becoming larger. In Latin America, for instance, a typical transaction may he $25 million to $50 million, roughly five times greater than a year ago. Thus, the breadth of the market in some lesser-traded currencies is increasing rapidly. As discussed in more detail below, the composition of market participants has also broadened to include institutions other than banks as major players. Hedge funds, mutual funds, pension funds, large corporations, and securities firms now all play a prominent role.
Types of Transactions
While the overall growth in the foreign exchange market has remained steady, after accounting for the dollar’s depreciation, the product composition of trading has changed dramatically. Growth in forward transactions has outstripped the growth in spot transactions by a margin of two to one; within the group of forward transactions, foreign exchange swaps continued to outpace outright forwards by a margin of six to one. In fact, the swap market accounted for nearly 85 percent of the increase in overall turnover. Turnover in outright forwards and foreign exchange-swaps accounted for 56 percent of the April 1995 turnover, and spot transactions accounted for the remaining 44 percent, compared with 51 percent and 49 percent, respectively, in April 1992.
Over two thirds of the growth in foreign exchange forward swaps is concentrated in maturities up to seven days. The reason for the increase in such short-term maturities is not well understood. However, many market participants have pointed to the growing sophistication of institutional investors and mid-sized or regional banks in managing their short-term cash needs. As these institutions become more globally diversified, they have started to manage their cash positions in the currencies That are to be used for their transactions, using short-dated forward swaps to do so instead of moving cash to the local currency and then back to the foreign currency with spot contracts. Another reason offered by market participants is that the increased use of bilateral netting agreements lowers credit exposures and permits more forward maturity positions to be undertaken.
Derivative contracts based on foreign exchange are gaining increasing importance in the foreign exchange market. OTC derivative contracts involving foreign exchange accounted for 37 percent of the estimated $47.5 trillion in outstanding notional principal of derivatives contracts at the end of March 1995, as reported by the first BIS survey of derivatives.11 Since notional principal only provides information about the outstanding face value of the contracts being held and not their economic value, gross market values were estimated as well.12 Of the $2.2 trillion representing the gross market value, which corresponds to about 5 percent of the reported notional amounts, 64 percent is accounted for by foreign exchange contracts.13 The slightly higher share in the aggregate market value figures mostly reflects the higher volatility of foreign exchange rates during the recording period (which raises estimates of market values.)14
Of the total OTC derivative contracts, 6 percent were foreign exchange options contracts. While still a relatively small proportion of derivative contracts, there is keen interest in foreign exchange options products. The hedging strategies of many “exotic” options products require highly liquid underlying instruments. Thus, “exotics” such as barrier options and quanto options are often written on the most liquid foreign currencies.15 Despite the interest in “high tech” options applications using an underlying foreign exchange instrument, the current estimates of exotic options by market practitioners range between 2 percent and 10 percent of foreign exchange options.16 Hence, so far, exotic options do not represent a “dangerously high” proportion of the outstanding notional amount of foreign exchange contracts or derivatives contracts.
OTC contracts dominate the foreign exchange derivatives market: exchange-traded futures and options contracts remain minuscule. The turnover of exchange-traded foreign exchange contracts is only 1 percent of foreign exchange derivative contracts. Futures and options exchanges are attempting to engineer a pickup in their foreign exchange volumes by designing contracts that are more easily tailored to customers’ needs and launching contracts on emerging market currencies such as the Mexican peso and Brazilian real.
Types of Currencies
The dollar continued to be the currency of choice, and was involved in 83 percent of all transactions in April 1995—a 1 percent increase over its share of turnover in April 1992—even after accounting for the dollar’s depreciation. The deutsche mark and the Japanese yen continued to be the second and third most actively traded currencies, accounting for 37 percent and 24 percent of transactions, respectively.17 The deutsche mark, along with the pound sterling and the Swiss franc, continued to lose ground relative to the French franc and other EMS currencies.18 This trend reflects the growing financial integration and trade among the European Union countries. In fact, in a number of European markets, local currency trades against the deutsche mark were, and probably still are, more common than against the dollar.19 Thus, currencies such as the Belgian franc, the Danish krone, the French franc, the Finnish markka, the Norwegian krone, the Portuguese escudo, the Spanish peseta, and the Swedish krona have all reported increases in turnover volumes.
While most trading centers are diversifying the number of currencies traded, the Tokyo market is becoming more concentrated in yen transactions with dollar-yen trades accounting for 76 percent of total turnover. Both Singapore and Hong Kong have become focal points for some of the other major currencies as well as many of the lesser-known Asian currencies, reducing the role Tokyo has traditionally played. For example, daily dollar-deutsche mark turnover in Tokyo is less than that in Hong Kong, Moreover, Tokyo accounts for just 5 percent of all “other currencies” transactions in the three major centers, and Hong Kong’s turnover in the “other currencies” is about double that of Tokyo’s.
Although data are available only for 1995 (from the Federal Reserve Bank of New York survey results), it appears that trading of lesser-known currencies has increased. Trading in the so-called emerging market currencies, such as the Mexican new peso, the New Zealand dollar, the Malaysian ringgit, the Thai baht,20 the Indian rupee, the South African rand, the Brazilian real, and the Slovak koruna has gained interest among dealers. This is evidenced by the number of firms that had trading volumes of $100 million or more in these currencies during April 1995 (Table 42).21 One reason for the growing activity is the increased importance of emerging markets in global trade and capital flows. There is also a slow, but steady, increase in the use of derivatives by corporations, who seek to cover exposures in the emerging markets in which they conduct business.
The supply side has likewise moved in favor of emerging market currencies. Profitability in many of the bread-and-butter products (spots and forwards in the major currencies) is declining and many bankers are attempting to gain a foothold in the more lucrative products, such as risk management tools in emerging markets. Since the risks are perceived to be higher, and the markets are less liquid, profitability is notably higher for emerging market products. Of course, as more deals are consummated in these markets, liquidity will be enhanced, eventually eroding profit margins here as well.
|Currency||Number of Firms1|
|Mexican new peso||36|
|New Zealand dollar||11|
|Hong Kong dollar||9|
|South African rand||3|
Number of foreign exchange dealers in New York, reporting trading volumes in excess of SKM) million during April 1995.
Number of foreign exchange dealers in New York, reporting trading volumes in excess of SKM) million during April 1995.
Types of Counterparties
About two thirds of foreign exchange transactions take place between bank dealers.22 Turnover attached to nonfinancial customer transactions accounts for about 16 percent, and another 20 percent is executed with other financial institutions. While the proportion of interdealer turnover appears to have dropped slightly from April 1992, mostly due to a rise in business with other financial institutions, there appear to be many more transactions than are necessary to transfer goods and services across borders. There is a technical reason, however, that contributes to the large number of transactions. For a bank to hedge an outright forward contract with a customer in a currency other than the U.S. dollar or deutsche mark (a cross-currency contract), the bank may need to undertake as many as four additional transactions.23 Hedging foreign exchange options transactions, too, requires additional spot transactions. Additionally, transactions used to hedge large customer transactions may need to be transferred through a number of dealers, each of whom are trying to maintain a specific position, before they optimally Fit into a dealer’s portfolio. Hence, even without counting outright position-taking, turnover is a multiple of the underlying transactions required to conduct trade of goods and services (Box 9). Many market commentators observe the high turnover and some view it as an indication that the market is “excessively” volatile. However, the high liquidity allows customers to enter the market, to obtain a hedge or execute another strategy, without delays and the accompanying exchange rate risk.
Among their nonfinancial customers, banks are seeing a broader range of counterparties. The dismantling of capital controls in many countries has widened the customer base, from those simply wishing to carry out transactions related to trade to those making investments. Thus, international investors, speculators, and other new players have entered the customer bases of most large banks. Furthermore, increased volatility and larger commercial exposures to exchange rate changes have made many corporations wary of leaving themselves unhedged. In addition, securities firms that are active in international debt and equity markets have also entered the foreign exchange market as intermediaries providing one-stop shopping to their increasingly sophisticated customers. These other financial institutions have gained market share, from 12 percent to 20 percent of total turnover, at the expense of both bank dealer transactions and transactions between dealers and nonfinancial customers.
Box 9.Microstructure of the Foreign Exchange Market
The variety of market participants and the protocols used in foreign exchange trading help determine the prices and efficiency of the market. While macroeconomic variables and relatively complicated models are frequently used to forecast broad trends in the prices of various currencies, prices can also be described using a microeconomic approach. Models of market microstruc-ture provided a more conducive framework for explaining the coexistence of the three most striking structural features of the foreign exchange market: (1) trading volume is enormous, (2) the share of interbank trading is very high, and (3) the transparency of order How is very low. In this regard, it is helpful to summarize the two main branches of microstructure: the inventory-control approach and the private-information approach.
At the core of the inventory-control approach is a recognition that risk-sharing among a market’s participants is incomplete in the short run (in contrast to the perfect pooling of risk in many macroeconomic models). A key implication is that purely allocational trades, arising entirely for hedging purposes, can influence price. The classic example is the risk-averse dealer1 whose position is buffeted by random incoming orders. If the dealer has a longer position than desired, then the dealer can lower the asking price to increase the chance the next customer will buy. which helps reduce the long position (an undesired short position having the opposite effect on price). In this way, order flow can affect prices, though for reasons unrelated to fundamental information.
The private-information approach, in contrast, shifts the focus away from risk aversion by recognizing that price determination does not rest solely on public information. If some participants are better informed than others, scrutinizing trading activity can provide clues about future market direction. One example of private information in the foreign exchange market is the orders and quotes that dealers observe but customers do not. (Put differently, all price-relevant information is not common knowledge.) This type of dealer-level private information is valuable in the short run for position-taking. Over lime, though, this private information is impounded in price as dealers collectively learn from the trading process.
The inventory-control approach links the first two of the three features cited above: enormous volume and predominant interbank trading. This link relies on the so-called hot-potato trading. Foreign exchange dealers use the hot-potato metaphor to refer to the repealed passing of an inventory imbalance from dealer to dealer until it reaches a dealer whose inventory discrepancy it neutralizes (the original imbalance being produced by a customer order). With very low spreads for direct interbank trades, unloading the position on another dealer is quite common. And there is no reason to presume that the position of the receiving dealer will be neutralized; dealers do adjust their two-way quotes to help control inventory, but a large adjustment would place both the bid and ask above (or below) those of another dealer, permitting arbitragers to buy low and sell high at the dealer’s expense; and quoting off-market on one side of the bid/ask spread is costly in terms of reputation. (While it is true that trading through voice brokers 2 is more likely to neutralize discrepancies, the cost of brokered trading is higher than dealing directly, at least for the largest banks.)
Thus, an important source of the high volume in this market is the multiplier effect that customer trades have on successive rounds of interbank trading. There are two interesting implications of this. First, much of the trading in foreign exchange is not speculative, but rather reflects the hedging—or inventory control—of dealers (an important fact to keep in mind when evaluating the case for transaction taxes in this market). Second, hot-potato trading may not be innocuous because the amount of private information impounded in the price depends on how much information can be gleaned from interbank trades. But if hot-potato trading increases the share of purely allocational trading relative to that which is based on information, then the efficiency of extracting information from order How is reduced, and the price reveals less information.
The efficiency with which private information is contained in the price is also related to the third distinctive feature above—the low transparency of order How. Transparency in foreign exchange is not regulated by law as in other multiple-dealer markets like the U.S. OTC equity market.3 Rather, transparency arises as a by-product of the method used for interbank trading. Specifically, customer-dealer trades and direct interbank trades are not disclosed to market participants; only brokered interbank trades generate marketwide signals of net demand in the market. Because dealers cannot condition their trades on marketwide net demand, order How remains relevant for price determination over multiple trading rounds. This provides the link between low transparency, high volume, and predominant interbank trading.
The overview above suggests that a core distinction between the macroeconomic and microeconomic approaches to exchange rate determination lies in the role of order flow. In the macroeconomic approach, order How is unimportant since information is common and risk-sharing is complete. Rather, structural equations based on public information imply that news is impounded in price without trading. In the microeconomic approach, in contrast, order flow is the vehicle through which price changes occur. Order flow affects prices either because risk-sharing is incomplete or because orders convey private information.1 A dealer bank in the foreign exchange market quotes both bid and ask prices and typically stands ready to trade orders of any size from a retail customer as well as the requests from smaller, regional banks.2 A voice broker constantly advertises anonymous dealer quotes of bid and ask prices through open lines (speakers) to many banks3 For example, on the U.S. NASDAQ trades must be disclosed within minutes.
Despite a growing diversity of “customers,” many central banks noted increased market concentration since 1992, as the proportion of trading taken up by the top banks continued to rise. This was most evident in the smaller markets where foreign banks have tended to close operations in order to consolidate the business in centers such as London, New York, Tokyo, Singapore, and Hong Kong. This has left fewer institutions in these smaller centers to share the market. In the major centers, too. concentration has increased. Between 1992 and 1995, the market share of the top ten dealers rose in Tokyo from 44 percent to51 percent, in New York from 41 percent to47 percent, and in London from 43 percent to 44 percent. Even among the top 20 banks, increased concentration is evident. They constitute 70 percent of the market in New York, up from 60 percent in 1992, and 68 percent in London, up from 63 percent in 1992.
This trend toward higher concentration obscures the considerable turnover among institutions at the top. Three of the top ten dealers in New York and in London were not present in that group in 1992. One interpretation is that transaction services are highly competitive, making it difficult to stay at the top. This is borne out in two related statistics. First, dealing spreads, the difference between the bid and ask prices for a currency, have narrowed on most highly liquid currencies to five “pips” from ten pips about ten years ago.24 Second, revenues from foreign exchange trading at most major banks have declined over the last several years.25
The increased turnover, along with lighter spreads and lower foreign exchange profits, has made the issue of the creditworthiness of individual counterparties ever more salient. Among the 20 largest dealers in New York, dealers who moved up in the rankings had higher credit ratings than those that did not. Some currency products, such as currency swaps, carry with them higher credit risks than their proportionate share in the overall market would suggest.26 leading dealers who trade them to become more cognizant of credit exposures. Customers, too. are becoming more credit conscious. While banks have in the past competed for customers’ business on the basis of their relationships and by offering timely quotes, corporate customers now rank credit rating as one of the most important variables they use when choosing a foreign exchange dealer.27
Electronic Broking Systems
A key development in recent years has been the greater use of electronic broking systems.28 Although most foreign exchange transactions are still made by traders talking to each other, electronic broking systems have made a formidable debut in the market. According to the BIS survey,29 in April 1995 about one in every seven transactions made through a broker was executed through an electronic system (Table 43).30 While voice brokers handle transactions for spot and a wide range of forward maturities, electronic broking systems currently only permit orders for spot transactions. In this area, however, use of electronic broking systems has skyrocketed: nonexis-lent three years ago, it now accounts for more than one third of all spot transactions and the trend is continuing. Two electronic broking systems31 have about 44 percent of the brokered business in London and in Tokyo; the yen-dollar spot transactions executed through electronic broking systems in Tokyo exceeded 50 percent of the brokered spot market in April 1996.32 In Singapore, 56 percent of the foreign exchange brokered business goes through electronic systems.
Proportion of principals’ total foreign exchange business.
Proportion of total market volume.
Proportion of interbank transactions.
Proportion of principals’ total foreign exchange business.
Proportion of total market volume.
Proportion of interbank transactions.
While the growth of electronic broking has been dramatic, its coverage is, as yet, narrow. Transactions have been concentrated in a relatively small number of currency pairs and only in the spot market. However, both of the electronic broking providers, Reuters (with the D2000-2 system) and Electronic Broking System (EBS), expect to extend their service to include forward contracts within the next year or so. The main stumbling block has been the implementation of a common methodology for calculating forward credit exposures. Every bank views spot exposures the same way—they addup the nominal amounts transacted. By contrast, every bank has a slightly different method for calculating the credit exposure of forward contracts. Hence, in order for the system to be able to discern whether a user is within its preset exposure limit with another counterparty, a common method for calculating the forward exposure is necessary.
While the number of currency pairs offered over the electronic broking systems appears quite broad, only a few of them could be considered liquid.33 For example, EBS lists 16 currency pairs, of which 6 or 7 have “good” liquidity and four have 1 “pip” spreads. Turnover on the most active currency pair, the dollar-deutsche mark, amounts to about $20-25 billion a day. In March 1996, the expected turnover in the dollar-yen currency pair for EBS, after linking up with M1NEX. the Tokyo-based electronic broking system, was about $8-9 billion a day. On the Reuters D2000-2 system, 26 currency pairs are listed and, according to Reuters, there is “reasonable” liquidity in about one half of them. Average trade size on EBS is about $3 million and greater: on Reuters, it is in the $12 million range. Thus, compared with Renters, EBS has larger trade sizes on fewer liquid currency pairs.
For the most part, electronic broking systems have taken away business that would otherwise have been conducted by voice brokers and interbank dealers. For example, in London, it is estimated that about half of the trades executed through the electronic broking system in the dollar-deutsche mark pair came from voice brokers and the other half from new brokered business, mostly low-value interbank transactions. This was also true for New York—about one half of the business was new to the broking community and the other half came from voice brokers. Mid-sized and regional banks, which have traditionally used interbank dealers for their foreign exchange transactions, have started to bypass these dealers in favor of the electronic broking systems. To some degree, their inclination to use the broking system is a result of the lack of transparency prevalent in the interbank dealer market, where quotes can be different for different customers.34 For the types of currencies and relatively small transactions that are typical of these institutions, the electronic broking systems offer a firm price and a very tight bid-ask spread. As with voice broking, electronic systems are used exclusively by banks and other financial institutions, although there has been some thought about extending membership to corporate end-users. An end-user is likely to want additional services beyond execution, such as strategic advice, trade recommendations, research, and risk appraisals—services that traditionally have been bundled with price quotes by interbank dealers. Unless these other services are provided separately from quotes, some end-users will continue to find it preferable to use interbank dealers for prices in order to obtain these services.
Reflecting the inroads made by electronic broking systems, there have been several recent mergers of foreign exchange brokerage businesses in London, and more are expected Some market observers believe that the future business will support only about 3 voice broking firms instead of the current 14, However, even managers at the electronic broking firms are not predicting the complete demise of voice broking. In addition to execution, a broker provides other information. For example, it is claimed that the tone of one’s voice, or an eager or reluctant affirmation, provides a clue to the market “tenor.” Thus, the information, flexibility, and discretion provided by voice brokers—which cannot be reproduced precisely in an automated system—affords voice brokers a valuable niche.
As noted above, electronic broking systems offer anonymity (since counterparties to a trade are only revealed after the trade has taken place), which can be considered either an advantage or a disadvantage. Anonymity might permit a counterparty executing a large trade to do so without moving the price. If the counterparty’s identity were revealed along with the large trade, then, knowing the counterparty, may reveal to market participants some knowledge about the future direction of the price. Participants will attempt to establish positions before the large trade, potentially moving the price prior to its execution. This possibility encourages these large transactors to trade when the market is “deep” so that their transaction will go unnoticed or to break up the trade and execute it on an anonymnous system. Thus, the anonymity of electronic broking systems may be helpful for these counterparties.
On the other side of the ledger, as greater attention is paid to the credit quality of the counterparty, anonymity can make some potential participants reluctant to join the system because they believe they might end up with counterparties whose credit rating is too low. In both Reuters 2000-2 system and EBS, most of this risk is mitigated because banks must set their own credit exposure limits to the other counterparties in the system, prior to placing orders on the system. Any trade that surpasses the limit will be rejected by the system when an execution is attempted.35 Despite this feature, however, potential participants may be reluctant to join the system if many of the potential counterparties are perceived to be of such low credit quality that the establishment of low limits would not permit effective use of the system.
A positive feature of electronic broking systems is the ability to lower the number of execution errors among participants. An electronic record of each trade is automatically contained in each transaction thereby making reference to voice recordings obsolete. In fact, one broking system features a “ticket output feed” which can be directly connected to a back office system that issues payment instructions for final settlement. Also, since credit limits are checked automatically, credit risk can be controlled prior to the execution of transactions. Thus, while not the impetus behind The development of electronic broking systems, settlement risk can be reduced by such systems.
An oft-cited concern about automated trading systems is that of price volatility. In particular, some market participants have voiced concern about the ability of the electronic broking systems to cope with the large number of transactions that may accompany a period of market stress. Some fear that the trades may overwhelm the system, as users lower order sizes and increase the number of orders attempting to ensure quick execution. These actions may backfire, slowing down executions and providing dated information to users.36 Even more serious, users might withdraw from the system entirely and either return to the interbank dealers and voice brokers or sit out the period of volatility. Either avenue would imply less-informative and less-transparent prices and, potentially, a loss of liquidity as the uncertainty hampers active trading.
The concerns are predicated on two suppositions. One is that the electronic systems are unable to process huge volumes. Managers at both Reuters and EBS argue that their electronic systems are currently operating far below capacity. When trading volumes consistently exceed a threshold level of around 60 percent of capacity, more computers are automatically accessed, providing additional capacity.37 The second supposition is that when users abandon the electronic broking system, liquidity will dry up and volatility will increase. However, if users simply revert back to trading with interbank dealers or voice brokers,38 then liquidity will just move from the electronic systems back to the OTC interdealer market. Total liquidity and volatility is likely to remain unaffected.39 If, on the other hand, prices are deemed less informative and less transparent due a slow or an underutilized electronic broking system, users may exit the market completely. In this situation, turnover and liquidity will fall and volatility is likely to rise.
At this time, electronic broking systems are still in the nascent stages of development and they suffer from a “chicken and egg” problem that routinely occurs in new markets. If liquidity is low, few participants want to trade. However, unless participants are willing to use the system, liquidity will remain low. Nevertheless, the most recent trends show that turnover on electronic broking systems is growing quickly. If, as expected, liquidity on these systems improves and users become more comfortable with the systems’ operations, users will be less likely to retract from the electronic broking systems even during a period of stress. In addition, technological advancements, such as the ability to enter contingent orders and an increase in processing speeds, are likely to be instituted, making order execution glitches more remote.
Management or Systemic Risk
The BIS survey estimates a global average foreign exchange volume of $1.2 trillion a day, implying that two-way contract payments of $2.4 trillion flowed through national payment and settlement systems in various countries in April 1995 (Box 10). In New York. 70 percent of this volume was executed by the top 20 banks. Many small transactions, as well as some very large ones, must be reconciled between counterparties and the funds exchanged. There are an estimated 125,000 to 150,000 transactions every day, necessitating 250,000 to 300,000 exchanges of currency. The Clearing House Interbank Payments System (CHIPS), through which most dollar foreign exchange transactions are settled, estimates that it alone facilitates $600 billion in foreign exchange transactions each day.
Large settlement sizes pose at least two problems. The first has been called Herstatt risk, named for the failure of Bankhaus Herstatt in 1974. Banks are exposed to large amounts of cross-border settlement risk because irrevocable settlement of the separate legs of a foreign exchange transaction may be made at different times. For example, delivery of yen to the Japanese correspondent bank of a New York bank in Tokyo occurs during Tokyo business hours. The corresponding delivery of dollars by a New York bank to the U.S. correspondent bank of a Japanese counterparty in New York would occur during New York business hours. Since the two national payment systems are never open at the same time, this gives rise to the risk that after the first counterparty has already delivered one side of the transaction, the other counterparty may go bankrupt and fail to deliver the off-setting currency.
More than 20 years after the collapse of Herstatt, there is no widely accepted method of quantifying settlement risk. Only recently have attempts been made to quantify and measure foreign exchange settlement risk. The Foreign Exchange Committee in its report entitled “’Reducing Foreign Exchange Settlement Risk” was the first to survey foreign exchange dealers and provide a methodology for examining settlement risk and a set of recommended best prac-tices.40 The more recent report from the Group of Ten Committee on Payment and Settlement Systems, the Allsopp Report,41 released in March 1996, builds on the earlier methodology, analyzes existing arrangements, and sets out a strategy for reducing settlement risk. The report finds that foreign exchange settlement is not just an intraday phenomenon and That interbank exposures can last, at a minimum, one to two business days. Further, it can take an additional one to two business days before a bank knows with certainty that it has received the requisite payments. The Allsopp Report finds that, given current practices, the amount at risk at a bank could exceed three day’s worth of trades, so that the exposure to even a single counterparty could exceed a bank’s capital. Indeed, the global estimates of daily turnover exceed the total equity of the world’s largest 300 banks. While the risk is only beginning to be recognized and quantified, recent foreign exchange payment defaults, including those of the Bank of Credit and Commerce International (BCCI) and Barings Plc,42 demonstrate that the risk cannot be ignored.
The second problem is a liquidity issue. Failures to pay could arise from operational or systems problems, as well as a counterparty bankruptcy. In most cases, operational failures can be resolved within a 24-to 48-hour period and overnight funding can be obtained to cover the failed delivery. It is not uncommon, however, to have more than $2 billion outstanding between banks overnight. A large operational failure could surpass the ability of even some of the best-capitalized institutions to access money markets, especially when notice of the failure is received during off-hours in the institution’s domestic market or when the undelivered currency is not one in which the exposed institution customarily borrows. This is an especially salient issue in emerging markets where transaction sizes are growing, but the physical infrastructure for payment and settlement systems may not yet be adequate to accommodate transactions in large numbers or sizes.
Box 10.Settlement Procedures
Foreign exchange transactions are not settled directly by the counterparties, but are routed through the national payment systems corresponding to the traded currencies. Typically, only banks licensed within the local jurisdiction have direct access to that country’s payment facilities. Most counterparties (other than local banks with direct access) will use a correspondent bank to carry out the settlement procedure. The typical process by which a dollar-pound sterling transaction is settled is shown in the figure.
In the United States, although dollar payment instructions can be routed directly through to Fedwire—the large-value, real-time gross settlement funds transfer system—settlement services for most of the dollar legs of foreign exchange transactions are provided by the Clearing House Interbank Payments System (CHIPS).1 CHIPS is a private, dollar-clearing system owned by the New York Clearing House Association, a group of major New York banks. CHIPS nets payment transactions multilaterally among its members and settles the net amounts at the end of the day. At the end of every business day, banks are notified regarding their net position. Banks with a net debit position within CHIPS transfer the owed funds via Fedwire to a CHIPS net settlement account at the Federal Reserve Bank of New York.2 Once all net debit positions have been paid, CHIPS releases payments to the accounts of banks with net credit positions. In the United Kingdom, a similarly structured payment system. Clearing House Association Payments System (CHAPS), settles pound sterling transactions through banks’ accounts held at the Bank of England, which is itself a member of CHAPS,3 CHAPS settles on a multilateral net basis at the end of each day across accounts held at the Bank of England.
Consider, for example, the common method to settle a sterling-dollar transaction between a U.S. bank that sells dollars for sterling to a U.K. bank, with the sterling payment being made first. Assuming the U.K. bank is a member of CHAPS, the U.K. bank pays sterling into its account at CHAPS.4 where the payments are netted with other payments to be received or paid by the U.K. bank. At the end of the business day. CHAPS makes the payment and credits the account of the CHAPS member bank acting on behalf of the U.S. bank.5 The U.S. bank’s account with the correspondent bank is, in turn, credited. The U.S. bank routes its dollar payment order to its U.K. counterparty through CHIPS.6 At the end of the U.S. business day, CHIPS requests payment from the U.S. bank for the amount of dollars to be paid to the U.K. bank and receives this sum in its account at Federal Reserve Bank of New York. After CHIPS receives the payment, it credits the U.S. based correspondent bank, a CHIPS member, that the U.K. bank uses for its dollar-denominated transactions. A simitar procedure is used for most major currencies, whereby banks typically utilize correspondent banks that have access to national payment systems.
Current Settlement Process
Either of these problems can lead to a situation in which a delivery failure causes a systemic problem. The most commonly articulated scenario is a “domino effect.” in which the failure of one large bank causes a second bank to fail, which causes a third, and so on. Another situation might arise in which, independently, a small number of institutions fail to deliver, causing other institutions to fail or to encounter liquidity problems. Using actual gross settlement numbers from a specific day in 1994 when the yen appreciated against the dollar by 5 percent. MULTINET, a proposed multilateral netting facility, was able to show that the failure of the participant with the largest position within their system could have caused the failure of a number of other participants. This example, constructed using only a few institutions, shows how a large settlement failure can be potentially disruptive to the smooth functioning of the international financial system.
Payment and Settlement Solutions
Settlement risks, both for individual banks and for the financial system as whole, can be ameliorated in a number of ways (see Box 10 for current settlement practices). Most observers believe that settlement risk would be eliminated with the use of a multicurrency “payment versus payment” (PVP) system. Using PVP, settlement in one currency is made if and only if payment in the other currency has been received.43 For the PVP system to eliminate Herstatt-type risk, a third party needs to assure that both sides to a transaction make payments before the amounts owed to the two parties are simultaneously released.44 The third party must also be able to guarantee the finality of payment, that is, the irrevocability of the payments and the ability of counterparties to use their payments when they are received. For this reason, it is often assumed that PVP requites the existence of a real-time gross settlement (RTGS) system which provides for immediate and final settlement of funds.45 As of May 1996. France. Japan, the Netherlands. Sweden. Switzerland, the United Kingdom, and the United Stales had operating RTGS systems. In Europe, the TARGET system will link together national RTGS systems.46 Members of the European Union are currently developing national RTGS systems to link to TARGET, if they have not done so already, to prepare for the implementation of Stage Three of the Economic and Monetary Union (EMU), scheduled for January 1, 1999.
The existence of RTGS systems is not a guarantee that PVP will be conducted and settlement risk eliminated; it only provides a technical platform to do so. Settlement risk cannot be eliminated if, for example, a dollar payment is made at the close of business in New York and a yen payment is made at the close of business in Tokyo, as is current practice. First, both parties to a foreign exchange transaction must choose to use a system whereby simultaneous finality can be assured. At the moment, banks are free to choose the payment system through which to make payments. For example, despite the fact that the U.S. Federal Reserve System operates an RTGS system, Fedwire, the bulk of dollar-denominated foreign exchange transactions go through the CHIPS for a number of reasons, not the least of which is that it is cheaper. Second, the PVP assurance would have to come from private sector procedures that would make use of the real-time settlement service. That is, some means of verification would be required to make sure payments have been received before others are released; this is not currently within the purview of the central banks that run, or plan to run, RTGS systems.47 The reason PVP eliminates Herstatt risk is that PVP obviates the settlement exposure taken on by the participants due the nonsynchronous guaranteed receipt of funds.48 Even without a PVP system, however, Herstatt risk could be eliminated if the timing of payments in the two currencies of a transaction were perfectly aligned so that payments acquire finality simultaneously. Under current arrangements, this can only happen if both countries have RTGS systems running at the same lime. Hence, what is necessary to eliminate Herstatt risk is that both legs of a foreign exchange transaction are guaranteed payment receipt finality at the same time.
A first step to facilitate PVP would be to lengthen the hours of operation of RTGS systems in the countries where they are operational. A major extension of hours in countries trading key currencies could open the way for various PVP systems to develop. Already there is some progress in this direction. In 1993, the Bank of Japan lengthened the hours of BOJ-NET by two hours to 5:00 p.m. local time. While the Bank of Japan’s extension did not aid cross-border settlements with the United States (since the extension did not produce overlapping periods between the two RTGS settlement systems), it did facilitate same-day settlement within Japan’s payment system. Recently, the Federal Reserve agreed to extend Fedwire’s hours of operation to a full 18 hours a day beginning in 1997, opening at 12:30 a.m. New York time. With the extended hours, Fedwire will be open during the entire European business day and will overlap with the final two and one half hours of trading in Japan during standard lime and three and one half hours during daylight savings time.
A second possible method for eliminating settlement risk would be to establish cross-border links within the RTGS systems run by the central banks so that verification assuring that, say, a yen payment had been received in Tokyo before the corresponding dollar payment would be released in New York, could be achieved automatically. No private sector third party verifying or aligning payment instructions to the respective RTGS systems would then be necessary. One potential side effect of linking national RTGS systems is that a disruption at one site, resulting from operational or liquidity problems, would affect other sites. This would be an especially difficult situation if the ability to access off-hour money markets was inhibited or the money markets were not deep enough to provide adequate liquidity for the duration of the disruption. Multiple central banks running such linked RTGS systems may be required to supply central bank credit and liquidity facilities until the site of the dislocation was able to adapt.
Another potential problem with linked RTGS systems is the impaired ability of an individual central bank to deal effectively with a local financial problem. Suppose, for example, that the local stock market fell and large securities firms were required to make margin payments using cash. Without enough available cash, securities firms would sell some of their stock of government securities and expect payment in central bank funds. If there were insufficient funds in the domestic system and payment failures occurred, this could delay payments to other domestic or foreign institutions and create a sort of payments “gridlock” (Box 11). While this situation could currently arise, with linked RTGS systems the present set of informal interdependencies would be institutionalized, perhaps requiring more information to be transferred among central banks about the involved institutions. Further, issues may arise regarding which central banks would be responsible for increasing liquidity and how the influx of liquidity was to be “mopped up” after the event.49 Coordinated decisions about the supervisory norms and rules for members or participating institutions to the RTGS systems would be required as well, since multiple central banks would be acting as liquidity providers or lenders of last resort.
Box 11.Payments Gridlock
A payments gridlock occurs when a disruption to a payments system prevents the timely settlement of some or all of the participants in the payments system. There are a number of ways in which such a gridlock can occur, although it is most frequently envisioned as stemming from a technical or operational problem. For example, a natural disaster in the location where payment system operations are housed could cause a system failure and payments gridlock. Generally, the situation is one in which the failure of one participant to make a payment into the system prevents complete payments to the other participants. When the other participants do not receive the funds owed to them, they might, in turn, be unable to make other payments. The failure of these participants to pay may imply that other participants are also unable to complete timely payments to others. A domino effect is created, potentially encompassing multiple participants in multiple payment systems.
The collapse of Barings Plc at the end of February 1995 illustrates how easily a failure of one institution can cause problems for others, potentially frustrating settlement within a payments system. On Friday, February 24, a clearing bank for ECU transactions sent an ECU payment to one of Barings correspondent banks for settlement on Monday, February 27. After it was established that Barings had collapsed and an administrator had been assigned, the sending bank attempted to reverse the transaction, realizing it was un-likely to be paid the other leg of the transaction on lime, but found that the rules of the ECU clearing system did not permit the payment to be canceled. According to the rules, the receiving correspondent bank was also unable to reverse the transaction. At the end of the day on Monday, the sending bank was in a net debil position in the clearing system. The bank agreed to borrow from another bank in the system that had a net credit position in order to cover the shortfall, al-lowing the settlement of all 45 ECU clearing banks to proceed. However, the sending bank could have decided not to borrow and ECU clearing would not have taken place, causing potential disruptions to other banks and customers in the ECU market, and perhaps, in other markets as well. It could have been the case that there was not enough liquidity in the system to permit the sending bank to borrow the requisite amount—again with a potential chain reaction.
An alternative to linking public RTGS systems is to form a global real-time PVP system run by private consortium, supranational organization, or group of central banks. Although in an early stage of development, one such global clearing bank has been put on the drawing board by the Group of Twenty, a group of 17 internationally active banks.50 In the private clearing bank, linked foreign exchange payment orders would be matched and immediately settled one at a time through the posting of debits and credits to member accounts at the clearing house bank, using the “continuous linked settlement” approach.51 The currencies eligible for the clearing house bank would be those in which RTGS systems are available in the home country and in which there are overlapping hours of operation for the national RTGS systems.
Although a private global clearing bank appears to be the most direct method for managing Herstatt risk, there are remaining challenges. First, the clearing bank’s ability to guarantee finality of payment in each country is uncertain. Finality would require a coordination of the legal status of settlements in each country and involve the location and corporate form of the clearing bank and its relation to the national settlement facilities (Box 12).
Second, the operation of a global clearing bank might have an impact on liquidity in short-term money markets and, thus, on central bank liquidity management, and. perhaps, on monetary policy objectives. If a global clearing bank required its members to pay large sums of money into their accounts to cover large volumes of settlements, it might drain liquidity from the domestic money markets. The drain of liquidity might offset the ability of a central bank to control short-term interest rates and their provision of intraday liquidity to domestic money markets, potentially increasing the volatility of demand for central bank credit. The liming, as well as the amounts, of funds involved in global clearing bank settlement procedures might also present a problem for short-term money market liquidity. Clearing bank procedures may require funds to be available to support settlement during the periods of overlapping hours between the various national RTGS systems. For instance, when Fedwire extends its operating hours in 1997, the expected overlap of the RTGS systems in the United States and Japan will he between 12:30 a.m. and 3:00 a.m., New York lime. It is unclear whether sufficient liquidity would develop in the U.S. market during these off-hours to support PVP settlement of dollar-yen transactions.52 Finally, a single system purporting to settle the majority of global foreign exchange payments would make it vulnerable to technological failures; several redundant systems would probably be required to safely minimize this risk. While the problems with this option are not insurmountable, harmonizing national laws and procedures for liquidity provision is a difficult and time-consuming process.
Box 12.Settlement Finality
Settlement is final when a settlement system’s rules deem that the payment is irrevocable and unconditional. The ability to use the funds for other purposes is not a guarantee of finality since there may be situations in which it may be permissible to request that the payment be returned after it has been disbursed. Thus, in some cases, even after payments have been made, they may subsequently be able to be revoked. When central banks operate a payments system and provide accounts through which payments are made, the central bank determines when (and if) settlement finality occurs. Funds that are received and deemed “final” through a central bank settlement system are sometimes referred to as “good funds.” Not all payments through a central bank have finality. For example, the non-Fedwire transactions through the Federal Reserve Banks in the United States do not guarantee immediate finality for checks as would be guaranteed through Fedwire.
It is generally accepted that only central banks can provide settlement finality since their role of lender-of-last-re sort permits them to supply unlimited funds (that is. print money) to compensate for any failures to deliver funds into the payments system by participants. However, other private payments systems often refer to their ability to assure settlement finality. In this case, settlement finality refers to the assurance provided by the private payment system that, in the event of one or multiple participant failures, settlement among the remaining participants will occur. The assurance of final settlement is based on the definitions for the irrevocability of payments and the procedures and protections built into the private payment system.
In theory, the establishment of a well-constructed global clearing bank could eliminate foreign currency settlement risk. However, because the development of such a system is still in its nascent stages, other approaches such as the development of multilateral and bilateral netting are receiving increased attention. Formal bilateral netting systems have been available since 1990 and informal bilateral arrangements can be privately negotiated between counterparties at any time. Two multilateral netting systems have been under development for some time and the first became available in August 1995. The notion underlying these approaches is that if settlement risk cannot yet be eliminated, at least it can be reduced and better managed. Netting systems (both bilateral and multilateral) can dramatically lower the size and number of payments transferred among counterparties, thereby eliminating a large proportion of the settlement risk Netting systems are not stand-alone methods for eliminating settlement risk, however. After payments are netted, banks must still utilize a payment system that guarantees finality of payments.
Bilateral netting systems periodically aggregate the amounts owed between counterparties and calculate one payment per currency for each pair of counterparties. Bilateral netting can reduce amounts at risk by an estimated 50 percent on average. Multilateral netting systems net the amounts owed among a group of counterparties through a clearing house arrangement resulting in one payment each day in a given currency to or from the clearing house by each counterparty. The reduction in settlement risk obtained through multilateral netting ranges from 73 percent, for a group of about 20 participants, to as much as 95 percent if there are more participants in the netting scheme.
There are currently four main netting systems operating for foreign exchange—two bilateral and two multilateral (Box 13). The bilateral systems allow pairs of banks to match and confirm their trades with one another and net their foreign exchange deals. The systems only calculate the netamounts the pairs of banks owe one another—there are no automatic payment facilities and neither of them assume foreign exchange exposures. The two multilateral netting systems are set up differently from each other, although there are some common elements.53 Both multilateral systems provide a clearing organization that operates as the counterparty to the trades within the system. They both require collateral and/or margin from their members under certain circumstances and both have established means of dealing with a defaulting member. Since a default by a member causes a potential immediate liquidity problem and losses to at least some of the other members, both multilateral netting systems have procedures for assuring the continuance of payments to other members.
One of the primary difficulties facing multilateral netting systems is to attain legal enforceability of the netted contracts. Compared with the creation of other types of clearing houses (for example, futures exchange clearing houses), a foreign exchange netting system cannot operate effectively without resolving the legal status of contracts in many different jurisdictions. First, the clearing house itself needs to be able to guarantee the contracts it enters into are legally binding. Also, the institutions from different legal jurisdictions need to guarantee their ability to net and enter contracts with the clearing house. In addition, the counterparties and clearing house need to assure themselves of access to collateral that may be held in yet a different legal jurisdiction.
Box 13.Bilateral and Multilateral Netting Systems
There are two widely used bilateral netting systems. In the SWIFT Accord system, the messaging service that SWIFT (Society for Worldwide Interbank Financial Telecommunication) provides its customers is combined with an optional bilateral netting service. In May 1996, 27 financial institutions used the netting service processing about 1,000 orders a day. FXNET, another bilateral netting service, provides pairs of banks with the ability to match, confirm, and net their transactions without the information going through a third party. In May 1996, the system was used by 70 banks. FXNET estimates that its $173 billion daily turnover is reduced to $80 billion by bilateral netting. The type of bilateral netting executed by FXNET is netting by novation. Novation implies that a new contract, representing the net position, replaces the old one as successive transactions are undertaken. It applies to payments in the same currency on the same settlement date.1 It is thought that this type of legal construct for netting is appealing to banks in the United States and Switzerland.
The two multilateral netting systems, the Exchange Clearing House (ECHO) and MULTINET. provide a set of members with the ability to net their payments within the group thereby lowering the amounts owed.2 Members make a single payment (breach currency to the central system, termed a clearing house. The clearing house reduces the settlement risk of the participants by lowering the number and amounts of the payments. However, it does not eliminate it, since all netted transactions still require settlement over a national payment system.
ECHO is based in London and supports 24-hour, global netting. As of May 1996. 13 banks use ECHO and another 7 banks are preparing to join the system. ECHO’S netting structure operates under English law and uses the contractual process of “open offer,” in which ECHO becomes the counterparty to a deal as soon as the eligible foreign exchange contracts are arranged by members, The contract is accepted by the clearing house without the need for immediate communication or prescreening. ECHO does not use the legal process of novation.
Although ECHO is a party to all transactions, the risk of loss resides with the participants who chose to deal with the clearing house. The participants who deal with a defaulting member within ECHO bear a pro rata share of any resulting loss: those that chose not to deal with that member suffer no loss. The advantage of this system, according to ECHO, is that the incentive to monitor credit risk of counterparties is preserved. However, to make sure the total possible loss to the clearing house, and its members, is kept within reasonable bounds, the clearing house has a series of limits that will restrict the amount of exposure a counterparty can present to the system with out also providing margin for the excess.3 The margin is called after a limit has been breached but must be supplied before settlement to assure time to withhold outgoing payments to the user it margin has not been posted.
To maintain sufficient reserves in the unlikely event of a default, ECHO holds collateral provided by the members and has a number of sources of liquidity. To conform to the Lamafalussy requirement that ECHO must be able to withstand the failure of its largest user, it holds collateral in excess of $1.1 billion.4 An additional 2.5 percent of a user’s exposure limit with ECHO is also held in reserve. ECHO has access to over $6 billion through committed swap and stand-by loan facilities with its nostro agents, the banks acting as correspondent banks in each country.
In MULTINET, the other proposed multilateral netting system, a slightly different scheme is employed. MULTINET would become the counterparty to a matched deal after determining that the deal is within the clearing house’s risk limits. A participant’s trade would be subjected to as many as five separate criteria before it is accepted. If rejected, the deal is to be settled using bilateral netting techniques. However, once the deal is accepted, a “novation and substitution” process would be used to establish a new contract between the respective counterparties and MULTINET. This process implies that any legal actions regarding the contracts would involve Multinet as the counterparty to such contracts.
The loss-sharing agreement for members of MULTI-NET is designed to be similar to ECHO’S. Like ECHO, it would divide up a failed payment among the members who dealt with the defaulting party and who would have had exposure on a bilateral basis. MULTINET would collect collateral from its members to be used for de-faults. The general collateral pool for settlement and liquidity risks for spot transactions will consist of $350 million at the start of the clearing house. Members submitting forward deals to MULTINET are expected to fully collateralize any forward replacement exposure as well as provide collateral to cover five days worth of market risk. If there is a failure to pay by one of the members, that member’s collateral would be used to satisfy the forward replacement exposure obligations.
MULTINET will attempt to implement PVP by not releasing funds to a member until that member has made the requisite payments into the clearing house. The payments will not be quite simultaneous since the clearing house sets its payment deadline one hour prior to the time it expects to release funds. To help efficiently process settlements, MULTINET intends to apply for direct access to Fedwire through its own account at the Federal Reserve rather than having to use an account at a correspondent bank in the United States.1 In the United States, novation of contracts can also occur across settlement dates.2 Since 1992. MULTINET has run a bilateral netting facility, which now consists of ten banks. MULTINET has not begun actual) settlements through its multilateral clearing system, but is expected lo begin operations in late 1996.3 Basically, the limits are based on the counterparty’s credit quality and its size, as measured by tier 1 capital.4 The $1.1 billion figure is equal to the largest user’s exposure limit to ECHO.
The legal uncertainty surrounding foreign exchange netting originates in differences in countries’ insolvency laws. The basic question is whether netting is valid when one of the counterparties goes bankrupt. When such a situation arises between counterparties in two jurisdictions, attempts to net will depend on the insolvency laws of the defaulting party.54 Another issue is which counterparty has control of the collateral and when. Multilateral netting systems have obtained legal opinions from all of the members’ jurisdictions, in addition to their own jurisdictions, hoping to gain legal certainty for their operations. However, since insolvencies involving multilateral, multicurrency foreign exchange netting systems have not yet occurred, there is no legal precedent on which to base a legal opinion. Thus far, then, the opinions represent only an informed interpretation of the relevant law.
Another design issue for a multilateral netting system is the risk management system and its collateral or margin requirements. To attract members and to satisfy regulators that the multilateral netting will lower settlement risks, netting systems need to assure that the clearing house does not take on settlement exposures that cannot be covered in the unlikely event of a failed payment or a bankruptcy of a user, While there are a number of methods to manage settlement exposures, the basic risk management tools of the multilateral netting system utilize a combination of real time exposure limits, the collection of collateral or margin, and a precise set of operating procedures to limit the amount of lime that settlement risks are present. Different combinations of these tools have different implications both for the costs to members and for benefits provided to the financial system as a whole.
There are numerous variations on the use of collateral. For example, members may be required to contribute, ex ante, to a collateral pool a proportion of their spot market volume and the entire balance of forward exposure in the system. Alternatively, margin may be requested, ex post, if a spot or forward exposure exceeds certain limits. The first option is costly for members, but provides a larger cushion in case of a failure. The second option is cheaper for members, but may provide less protection. The Lamfalussy minimum standards (standard IV) require that the “multilateral netting systems should, at a minimum, be capable of ensuring timely completion of daily settlements in the event of an inability to settle by the participant with the largest single nei-debit position.”55 While it may first appear that the more collateral the better, the use of collateral is not a panacea. The value and liquidity of collateral changes with market conditions. Valuation of collateral holdings may be especially problematic in turbulent market conditions, times in which a failure is already more likely. In fact, the problem may worsen because the value of the collateral might decline exactly when members are having difficulty making payments. Thus, the apportioned collateral may not he-enough to cover a loss when it occurs. Typically, careful attention is paid to requiring enough collateral for all but the most unlikely circumstances. However, it is the most unlikely circumstances with large costs that are of systemic concern.
Multilateral netting systems typically use a set of limits that prevent members from taking on “excessively” large exposures with other members and with the multilateral netting facility. These limits, however, are only as good as the risk management system that produced them. In fact, although a detailed risk analysis may be performed on the transactions within the netting system, the system may not have access to members’ full portfolios and thus cannot accurately assess the other risks attributable to each member. Additional data, such as the credit rating of a member, may be used as a proxy for other risks a member takes on, helping to more accurately set exposure limits. Although not all of the foreign exchange settlement risk can be controlled by a set of system exposure limits, the limits are the first line of defense to protect the netting system against excessive exposures taken on by one or more of its members.
The operating procedures of the multilateral netting system are critical to determining the extent to which settlement risk is reduced. Since by definition, netting takes place over a period of time, forms of PVP that require simultaneous intraday settlement of both legs of a foreign exchange transaction cannot be feasibly linked to netting systems. However, once the netting has been accomplished at a point in lime, the settlement of the remaining net payments can, in principle, be executed using a PVP mechanism. One of the two netting systems attempts an imprecise version of PVP in which the clearing house collects payments from participants before releasing its payments to the recipient participants. In most circumstances, the difference between the time when the multilateral netting system receives payments and makes its own payments is expected to be within a few hours. But. unless there is simultaneous finality of received payments, there remains some degree of Herstatt risk. Obviously, the closer in time the two legs of the transaction obtain finality, all else equal, the shorter the duration of Herstatt risk.
Still, whether a PVP mechanism is used or not, a disruption due to the failure of one counterparty affects other members of the system. Furthermore, there is no guarantee that the loss will always be small enough that the affected members will not have payment or liquidity problems of their own. That is, a domino effect could be created if the loss was large enough. To foreclose the possible transference of one failure to other members, multilateral netting systems need to be assured of an ability to acquire funding in cases where payments are withheld due to failures. Multilateral netting systems have broached the funding issue either by holding collateral or by assuring themselves of outside sources of liquidity, for example, lines of credit and foreign exchange swap facilities, mostly with member banks. As with collateral valuation in times of stress, it is unclear whether these lines of credit can be relied upon, since the member banks themselves may be affected by a liquidity problem. Ultimately, then, central banks serve as the backstop in a liquidity crisis, just as they would without multilateral netting systems. In theory, the central bank’s ability to monitor the multilateral netting system and the system’s capacity to reduce both the settlement amounts and the number of transactions imply that it is unlikely that the lendcr-of-last-resort role of central banks would be used.
Without considering other multilateral multicurrency clearing bank arrangements, the existence of two multilateral netting systems may not be self-sustaining. The degree of risk reduction is a function of the number of linked counterparties. The greatest degree of risk reduction is obtained when all the largest participants join the same system. It may not be cost-effective for a single bank to become a member unless the other banks with which the bank does most of its foreign exchange business join the same netting system.56 Furthermore, each bank waits to see what its counterparties do, delaying full-scale realization of the risk reduction possibilities until a sufficient minority join one netting system to make it cost-effective for the others.
With the recent Group of Twenty initiative to develop a global clearing bank, the bilateral and multilateral net-ling systems face further challenges. While, in principle, the two approaches to lowering Herstatt risk could be viewed as complementary, both require scarce funds from banks’ foreign exchange trading businesses.57 Further, as competing approaches to the reduction of Herstatt risk present themselves, banks may wait until one system emerges a clear “winner” before attempting to reduce their own settlement exposures. Banks perform both an evaluation of current cost as well as an evaluation of future costs, including the cost of employing specialized technology that may require augmentation or alteration if another system is subsequently used. Both netting systems and a global clearing bank become economically viable only when large payment volumes are transferred through the system.58 Hence, competition among the groups developing methods to lower Herstatt risk may have the effect of reducing the effectiveness of any one system.59 Moreover, competing systems have the unfortunate effect of slowing the adoption of strategies to reduce Herstatt risk as poiential users wail for a private sector “winner.”
Other Solutions to Settlement Risk
A natural extension of limits imposed on the members of a multilateral netting facility is the internal exposure limits in the institutions making up the foreign exchange market. Bilateral exposure limits to one’s counterparties help make an institution less vulnerable to a failure of one of its counterparties. An institution also lowers settlement risk to others by limiting its own overall exposure in the foreign exchange market. In most countries, voluntary limits are enhanced by rules regarding the degree to which exposures can be concentrated in one or several related institutions. In some countries—Belgium, Den-mark, and the United States—there are no explicit limits preventing individual banks from taking on excessive foreign exchange positions, and institutions set their own internal limits.60 In Australia and Portugal, foreign exchange exposure limits are imposed by the authorities, but are not public and are specific to each bank. In most other industrial countries (Austria, Finland. Germany, Japan, Netherlands, New Zealand, Norway, and Switzerland) there is a formal capital charge against foreign exchange positions implying an implicit exposure limit relative to the institutions’s capital.61
Recent market risk proposals from the Basle Committee on Banking Supervision, and their adoption over the next year and a half, should also serve to indirectly limit foreign exchange settlement risk to the extent it is correlated with the risk of market fluctuations in exchange rates. In the standardized approach for assessing market risk capital requirements, an 8 percent capital charge is applied to the higher of either the net long currency position or the net short currency position.62 The Basle Committee’s market risk proposal also permits banks to use their internal value-at-risk models as the basis for market risk capital requirements. In this approach, foreign exchange is one of the four major risk categories accounted for in the value-at-risk model.63 Currency exposures should be accounted for within the model for all currencies in which the bank has a significant exposure. Historic correlations among the currencies and between the currency category and other main risk factors are permitted in order to lower the measured risk within the model. However, the model’s result, a value-at-risk number, is multiplied by three, at a minimum, to arrive at the required capital. Both the standardized and the internal models methods thus provide an implicit limit on the amount of foreign exchange market risk, which may also indirectly limit the settlement exposures by potentially limiting the size of positions, that can be taken on by a bank since capital must be held against such positions.
The existing risk management techniques could also be augmented or altered to measure and manage settlement risks. For example, existing credit risk control processes used by banks could be adapted to identify and control the counterparty foreign exchange settlement exposures. Improved back office payment processing, correspondent banking arrangements, and bilateral netting capabilities may also reduce settlement risks. Altering the timing of payments and identifying final or failed receipts as soon as possible could also help banks shorten the duration of settlement risks. These solutions require no public sector involvement and are estimated to provide a substantial reduction in foreign exchange settlement exposure.
Large Exposure Monitoring
Private sector banks protect themselves from a defaulting counterparty by imposing internal exposure limits. However the aggregate exposures taken on by multiple institutions can still pose problems, especially when a similar group of institutions take similar positions. In this case, individual limits may not be sufficient and public sector surveillance may be warranted. This is mainly accomplished through large exposure monitoring.
Large exposure monitoring by supervisory and regulatory authorities exists in several Group of Ten countries. In the United States, for example, each foreign exchange market participant that had more than $50 billion worth of foreign exchange contracts at the end of any quarter within the past year is required to report.64 The reports are submitted once a week and cover positions in the Canadian dollar, deutsche mark, pound sterling, Swiss franc, and Japanese yen. The data are reviewed by the Federal Reserve banks and authorities at the U.S. Treasury Department for excessively large positions, but are not published or otherwise publicly disclosed. In principle, large position holders could be asked to reduce their positions. However, any action taken as a result of monitoring the large positions is not known.
In some countries, such as in France and Italy, monitoring is accomplished by having banks that execute transactions above a threshold report these transactions to the authorities. These types of reporting arrangements are rooted in recent attempts to limit money laundering in the home currency, not to specifically examine large exposures of foreign exchange market participants in multiple currencies. In fact, the thresholds for reporting typically refer to home currency-denominated transactions and not to transactions denominated in other currencies. Furthermore, in some cases, the reporting is aimed at transactions performed by “customers,” not transactions with other interbank participants.
The purpose of large exposure monitoring is to make sure that no one or group of participants is taking a position that could precipitate a systemic problem. Given the global nature of the market, to adequately monitor participants’ overall positions, authorities would need to collect data from the large participants in the foreign exchange trading centers, consolidate the data, and communicate their findings to one another. Authorities in some countries have Memoranda of Understanding (MOUs). enabling them to share such information with their counterparts in other countries. However, these MOUs are mainly established for the exchange of information after a crisis has emerged and enforcement proceedings are under way. Currently, no countries routinely share large exposure reports of OTC foreign exchange dealings.
Bank for International Settlements (1996d). This report aggregates and analyzes the surveys conducted by individual central banks.
These turnover estimates are adjusted for local and cross-border double counting and for gaps in reporting.
The $70 billion figure aggregates two types of transactions—OTC and on-exchange—that actually are not directly comparable, since it is not possible to eliminate double counting of on-exchange transactions, which was done for the OTC transactions.
A foreign exchange swap is a simultaneous exchange of a forward and a spot or forward contract. A currency swap is a contract under which periodic interest payments are made in foreign currency between counterparties for an agreed period of time and notional principal amounts are exchanged at maturity.
The geographic location for turnover was determined by where the transactions were entered into, not the location where the transaction was ultimately booked or managed.
The fastest growth rate among the major centers was recorded by France with a three-year growth rate of 74 percent.
These figures exclude repurchase and reverse-repurchase agreements by dealers. Including only financing through overnight and continuing repurchase agreements (since counting reverse-repurchases would result in some double counting among dealers), $470 billion average daily turnover would be added, resulting in $645 billion a day—about one half the turnover in the foreign exchange market.
A liquid market is one in which a position change can be initiated quickly with little disruption in the prevailing price.
Fifty-two percent of net turnover in London was with other banks outside of the United Kingdom. Salomon Brothers reports that about three quarters of interbank activity is carried out between domestic and international counterparties (Salomon Brothers (1995)).
This information is based on discussions with dealers, quote services, and foreign exchange messaging systems.
Given their close association, studies of foreign exchange electronic broking provide some evidence about the depth of the OTC market. These studies show that prices move in small increments and that quote revisions and transaction prices are closely tied. See, for example, Goodhart, Ito, and Payne (1996).
See Annex I for more discussion of the results of this survey.
Gross market value is calculated as the sum of the gross positive market value of contracts held by the reporting institutions plus the gross negative market value of their contracts with customers. The gross negative market value of their contracts with customers is equivalent to the positive market value of contracts held by the customers and represents their cost for replacing the contracts if the reporting institution defaulted. The gross market value reported in the BIS study represents the cost that would have to be incurred if all the contracts had to be replaced at market values existing on March 31, 1995 and roughly represents the value of claims on outstanding contracts.
This percentage, however, does not include the effects of netting or collateral agreements, both of which lower the exposure reflected in the market value. Nor does it include foreign exchange forwards or foreign exchange swaps held by survey participants in the United Kingdom because these items were not collected there.
In addition, currency swaps often have relatively high market values because, unlike interest rate swaps, the notional principal amounts are often exchanged at maturity between the counterparties.
A barrier option is a family of options whose pay-off pattern and survival to the expiration date depend not only on the final price of the underlying security but also on whether or not the underlying security sells at or goes through a pre-set barrier at any time during the life of the option. A quanto option is an option in which foreign exchange risks in an underlying security have been eliminated.
There are, as yet, no formal estimates of the outstanding amounts of exotic options.
Because each transaction involves two currencies by definition, the percentages add to 200 percent. Data are based on estimates made by reporting institutions that, by and large, do not connect the two currencies of a trade when they record the transaction and thus have no ready method of obtaining currency pair data. Hence, these data are measured with error and should only be viewed as providing information on broad trends.
The BIS has attributed the increase in the share of the other EMS currencies to an expansion in turnover and better reporting in certain centers rather than the increase in the number of currencies covered due to the addition of new members to the European Union.
Salomon Brothers (1995).
The estimated daily turnover in Thai baht is nearly $2 billion (“Forex is the key in Asia,” International Financing Review, Issue 1103, October 14, 1995, p. 121.).
There were 130 firms in the New York survey.
Some nonbank institutions may act like a “dealer,” quoting both sides of the market and holding an inventory of currencies. However, the largest actors in the foreign exchange market continue to be banks and, traditionally, the term “interbank market” refers to the active trading among banks acting as dealers. The term “dealers,” without further explanation, refers to bank dealers in the interbank market in what follows.
For instance, suppose a U.K. bank wrote a forward contract with a customer selling sterling for Netherlands guilders. To hedge this outright forward contract, the bank might first transform both the sterling and the Netherlands guilder currencies into deutsche mark (a more liquid vehicle currency) in the forward market, then execute two spot transactions (sterling-deutsche mark and Netherlands guilder-deutsche mark) to move the currency risk to the spot market. The forward and spot transactions can be executed together as a forward swap, constituting two transactions. Then, to hedge the spot exposure, offsetting spot transactions would be undertaken, producing a total of four transactions.
A pip refers to a one-unit change in the last quoted decimal place for a foreign exchange rate. For example, for a deutsche mark-dollar quote, 1 pip spread would be represented by quotes of 1.4810 and 1.4809. For the Japanese yen-dollar, quotes of 107.49 and 107.48 would also represent a 1 pip spread.
Such evidence, however, is subject to several interpretations since there are considerable differences with respect to what is considered revenue from foreign exchange trading by banks. For example, revenues from speculative position-taking and specialized derivative products may be included along with regular dealing revenues.
In the United Kingdom, about 36 percent of the aggregate market value in the derivatives market is composed of currency swaps. By contrast, this category amounts to only 7 percent of the total outstanding notional principle of derivatives. The higher market value, and thus credit risk, associated with currency swaps is due to the exchange of principle at maturity and their longer maturities relative to most other derivative products.
In the off-exchange foreign exchange market, there are two ways in which computer-aided transactions can take place: through an electronic dealing system, such as Reuters 2000-1, or an electronic broking system. In a “dealing” system, the counterparty is known to both transactors before the trade is executed, while in a “broking” system, the transactors know the counterparty only after the trade is consummated. The dealing system essentially provides a computer hookup with another counterparty. The only prices visible to either party are the ones associated with the connected counterparties. In the broking system, however, the prices provided represent the best bid and ask prices available from all participants in the system, enabling potential users to “hit” a particular price without making multiple phone calls to find a higher bid or a lower ask price.
Bank for International Settlements (1996d).
The Bank of England reports that in London 5 percent of total turnover goes through an electronic broker and that about 14 percent of broking business is conducted by an electronic broker. In the United States, the volume of turnover conducted through electronic broking systems was about 6 percent of total market turnover and 19 percent of volume in the brokers’ market. In Japan, the use of electronic broking systems is slightly less prevalent overall, taking up 4.5 percent of the total turnover among participants and 16 percent of brokered transactions.
Initially, there were three electronic broking systems. In March, 1996, MINEX, a Tokyo-based system, decided to become a partner of the Electronic Broking System (EBS) and has ceased independent operation.
Nihon Keizai Shimbun, May 30, 1996.
In London, about 77 percent of the deals conducted through electronic brokers are for the dollar-deutsche mark, the deutsche mark-French franc, and the deutsche mark-Swiss franc.
Quoted prices and sizes, and the market-makers offering such prices, are available from a number of information services, providing some transparency to the market.
Users see the best bid and offer prices available in the system as well as the best bid and offer prices of the counterparties with whom they are permitted to deal, subject to the bilateral credit limits within the system.
The information may be a poor representation of current conditions, either because a withdrawal of users means that new information is not reflected in the “order book,” or because the slower system response entails the display and execution of old, dated orders.
The slowdown of order submissions and executions on the automated trading systems offered by the New York Stock Exchange during the 1987 stock market crash have sensitized developers of automated trading systems to the problems of capacity constraints.
Some market participants have noted that, in periods of stress, the natural response of traders is to use systems of trading with which they are more comfortable. Since most foreign exchange traders have “grown up” using the interbank dealers and voice brokers, they are likely to revert to these avenues for executing trades.
In this scenario, the distribution of trading between dealers, voice brokers, and electronic brokers is not likely to affect liquidity. It may be true, however, that if the emergence of the electronic broking industry results in a decline in capacity of the voice brokers, liquidity may suffer if during periods of stress, traders want to rely more heavily on voice brokers.
See Foreign Exchange Committee (1994).
Bank for International Settlements (1996b).
The collapse of Barings Plc in February 1995 caused problems in ECU clearing that were resolved. However, the problems could easily have escalated into a situation where settlement could have been frustrated, causing the breakdown of payments among the 45 ECU clearing banks.
“Delivery versus payment,” the analogue to PVP in the securities industry, requires that payment be withheld until delivery of the security is made.
There are various mechanisms through which PVP can be effected, all of which assure participants that a final transfer in one currency will occur if and only if a final transfer of another currency also occurs. These mechanisms differ with respect to, first, whether recipients are guaranteed payment regardless of whether the related transfers take place or whether payments are refunded when the related transfers are not made, and, second, whether settlements take place simultaneously or sequentially. These different mechanisms have different implications for the liquidity needs and operational setup of the “PVP system.” For a more detailed discussion of these alternative mechanisms see pages 22 through 24 of the Bank for International Settlements (1996b).
An RTGS system is a gross settlement system in which processing and settlement take place in real time (continuously).
TARGET is the acronym for Trans-European Automated Real- Time Gross Settlement Express Transfer system.
Most central banks with RTGS systems currently take on some amount of credit exposure. In the United States, the RTGS system, Fedwire, imposes a credit limit (net debit cap) and an explicit price for the use of such credit; that is, a fee is charged for daylight overdrafts. Other central banks rely on collateral to offset credit exposures they take on and still others do not take on credit exposures but rely instead on queuing mechanisms to assure that adequate funds are in senders’ accounts before transfers are made to recipient accounts.
While some PVP mechanisms eliminate credit exposures within the settlement process, these mechanisms do not address liquidity risks. For example, if no payments are made out of the system until all other related payments are received, some counterparties receiving funds must wait to use these funds even though they have already paid corresponding funds into the system. This lowers the total available funds (liquidity) relative to an alternative system in which payments are made out of the system without the related payments having been received by the system. In this case, the credit exposure is borne by the system operator or intermediary.
The inflow of liquidity, if not removed in a concerted fashion, may contribute to inflation.
However, the outline of the global clearing bank presented here is based on information gained prior to May 1996. No formal documentation has yet been released.
This approach was first presented in New York Clearing House Association (1995).
Alternatively, a global clearing bank could be structured to use funds provided by a central bank to facilitate sequential settlements in which payments out of the clearing bank would be made prior to related payments into the clearing bank. A central bank may charge for the use of such central bank funds and, further, the use of funds within a global clearing bank may limit the remaining balances available for domestic payments. In fact, this would be a case in which a global clearing bank simply takes over the role each central bank currently plays in supplying credit for the functioning of its own RTGS system.
Although one of the systems, MULTINET, is not yet running, its system design is unlikely to change significantly prior to its expected introduction in late 1996.
The counterparties may specify in their agreements that the law of a sympathetic jurisdiction will apply. Whether this type of arrangement will be upheld varies from jurisdiction to jurisdiction.
Bank for International Settlements (1990), Part C, p. 26. The report, called "The Lamfalussy Report," provides minimum standards for the design and operation of cross-border and multicurrency netting and settlement schemes.
FXNET, one of the bilateral netting systems, and MULTINET, a multilateral netting system, are already discussing their potential integration.
Interestingly, some banks are members of both the netting systems and the Group of Twenty.
In netting systems, of course, the large number of payments is reduced within the system as a result of the netting. The proportionate reduction of exposures, however, is intimately tied to the number of participants and the dispersion of their transactions among the other participants.
On the other hand, if there were only one private organization offering services to reduce Herstatt risk, it could potentially engage in monopolistic pricing of its services.
In the United States, however, large exposure monitoring of foreign exchange positions is done weekly, as mentioned below.
See Hartmann (1995).
These positions are the sums of long and short positions, respectively, in various currencies valued in the reporting currency at the current spot rate.
The others are interest rates, equities, and commodities.
A market participant is broadly defined and includes not only banks, foreign banks, and broker dealers, but nonprofit institutions, sole proprietorships, mutual funds, and hedge funds as well. The report is filed on a fully consolidated basis, but data need not be collected from offices or subsidiaries that have “relatively” small position limits in terms of the reporter’s overall activity.