The Prudential Regulation and Management of Foreign Exchange Risk
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Contributor Notes

Author’s E-Mail Address: rabrams@imf.org

This paper examines issues in the prudential management and regulation of foreign exchange risk. It begins with measurement issues, notably converting foreign currency items into domestic currency terms, and calculating foreign exchange positions. The focus then shifts to managing foreign exchange risks. Although the key to effective management lies in the bank’s reporting and internal control systems, regulators frequently seek to limit such risks directly. This usually involves limiting the overall open position in terms of bank capital or requiring that capital be set aside against such risks.

Abstract

This paper examines issues in the prudential management and regulation of foreign exchange risk. It begins with measurement issues, notably converting foreign currency items into domestic currency terms, and calculating foreign exchange positions. The focus then shifts to managing foreign exchange risks. Although the key to effective management lies in the bank’s reporting and internal control systems, regulators frequently seek to limit such risks directly. This usually involves limiting the overall open position in terms of bank capital or requiring that capital be set aside against such risks.

I. Introduction

This paper examines issues in the prudential management and regulation of foreign exchange risk. Foreign exchange risk is a form of market risk.2 It arises when a bank holds foreign currency denominated assets or liabilities, with the risk being that exposures from open foreign currency positions may give rise to losses (or gains) in domestic currency terms as a result of exchange rate movements.3

The management of foreign exchange risk, as with other forms of credit or market risk, is based on five building blocks.4 First, accounting procedures and rules should result in the values of the assets and liabilities being accurately reflected in financial statements. Second, formulae need to be designed to allow the accurate measurement of the various forms of risk. Third, steps should be taken so that unexpected shocks associated with those risks do not unduly undermine the individual institution. Fourth, timely and accurate financial information on each specific risk should be prepared and promptly delivered to the interested parties, including the supervisor. Fifth, each institution should have a system of internal controls that both protect the safety of it assets, and ensure that informed decisions are taken at the appropriate level.

In recent years, probably the most contentious aspect of foreign exchange risk management has been the choice of method to limit such risks. Traditionally, this has been done by placing limits on the institution’s foreign currency exposure. Such limits are either absolute or expressed as a percentage of the bank’s capital base, and take no consideration of capital adequacy or the other risks in the bank’s portfolio. More recently, however, an increasing number of national supervisory authorities have started to require specific capital allocations against open foreign exchange positions and other market risks (in addition to or instead of the traditional approach). This practice compels a bank to designate capital holdings against its open foreign exchange positions, over and above the capital it holds for other reasons, for example, “Basle” capital requirements for credit risk. This makes it more difficult for weakly capitalized banks to take on new risks (perhaps preventing the weakest from having open foreign exchange positions). However, it gives a bank greater flexibility in choosing the risks it will accept by allowing managers to allocate a bank’s capital between credit and market risks, including foreign exchange risk. The two approaches, thus, achieve rather different prudential objectives, with absolute limits restricting the fluctuations in income that can be caused by exchange rate fluctuations, independent of the level of bank capital, while capital requirements seek to use capital as a buffer against potential losses from the bank’s foreign exchange exposure.

The importance of managing foreign exchange risk has increased with the trend toward global economic and financial integration, and the associated increase in trade, liberalization of financial markets, and dismantling of capital controls. These factors have combined to increase the risk that banks, consciously or inadvertently, may hold large open positions, and that adverse exchange rate movements may seriously damage their capital base.5

This paper explores issues relating to the prudential regulation of foreign exchange risk, by examining both actual practices and alternatives that are under consideration. Section II explores alternative methods of measuring foreign exchange risks. Section III explores methods of managing foreign exchange risks, and Section IV concludes. The appendix contains tables presenting the findings of a survey of the prudential supervision of foreign exchange operations in 19 countries.6

II. Measuring Foreign Exchange Risk

If foreign exchange risks are to be managed, they must first be quantified. This is a two-part process. First, accounting issues must be addressed, notably recording foreign exchange transactions and assets and liabilities in domestic currency terms to correspond to all other information in the financial reports. Second, one must determine how to measure the size of the bank’s foreign currency positions, in individual currencies, groups of currencies, and in all currencies taken together.

A. Procedures for Recording and Converting Foreign Currency Assets and Liabilities

While accounting standards are not central to this paper, they do have implications for measuring the values of foreign currency denominated assets and liabilities in domestic currency terms. In this regard, particular attention must be paid to recording procedures, conversion methods, and the treatment of revaluation gains and losses.

Three aspects of the recording of foreign assets and liabilities must be addressed. First, while banks keep their primary accounts in the domestic currency, they should also have parallel ledger accounts for each currency, generally with separate accounts for each type of transaction in each currency. Second, rules are needed regarding the currency in which foreign assets and liabilities are recorded. In general, assets and liabilities should be recorded in the currency in which the risk is denominated, regardless of the currency in which they were initially paid or received.7 Third, foreign currency accounts must be converted into domestic currency equivalent for inclusion in financial statements and prudential reports. This requires that revaluation accounts be used for recording translation gains and losses.

Accounting standards generally prescribe particular conversion methods to govern the conversion of foreign currency transactions and balances into their domestic currency equivalent. In most countries, banks should convert and record foreign currency transactions (such as a spot sale or a loan) by applying the exchange rate at the date of the transaction to the foreign currency account. Foreign currency balances should be converted or translated using either the closing rate method or the nonmonetary method, depending on the type of asset or liability concerned. Monetary items should be reported using the closing rate method, where items are converted at closing exchange rate for the period (for example, the spot exchange rate for balance sheet data). In some countries, certain items, such as real estate or other fixed assets, are treated as nonmonetary items, and their value, for balance sheet purposes, is generally based on an historic rate, such as the exchange rate that existed at the date of the transaction, or the time that the value of a particular item was determined.

Most countries require use of the closing rate method for monetary assets.8 Of the 19 countries surveyed, 16 require the use of the closing rate for monetary assets. However, six of the surveyed countries also allow some fixed assets to be treated as nonmonetary assets and converted at an historic exchange rate (Appendix I, Table 2).

The impact of the conversion method on valuation gains or losses can vary markedly depending on the composition of a bank’s balance sheet. In general, the greater the share of a bank’s portfolio that is converted using the closing rate, the greater the impact of exchange rate variations on its balance sheet and income. However, such exposures may be hedged. On the other hand, if a banks chooses to hedge nonmonetary items, its balance sheet may be more vulnerable to exchange variations than if all items were converted using the closing rate.

The treatment of valuation gains and losses is important because it can affect a bank’s capital and its taxable income. Virtually all countries surveyed pass valuation gains and losses on monetary assets and liabilities to the income and expenditure account, so they are immediately reflected in bank capital and are included in, or deducted from, taxable income. However, many countries exclude unrealized valuation gains on certain fixed assets, and foreign exchange-denominated capital, from taxable income. Four sample countries—Czech Republic, Korea, Singapore, and the United States—record all fixed assets at the closing rate, and revaluation gains and losses are recorded directly in the capital accounts. Since these accounts are not part of the income and expenditure statement, any unrealized valuation gains are not subject to taxation. Eight other countries, including France, Malaysia, and Spain, record certain fixed assets at historical cost, so that exchange rate changes do not affect the value of these assets until they are disposed of. The down side of this simpler alternative is that it may result in unrealized gains or losses being hidden on the balance sheet; thus, for the sake of transparency, it is useful to add memorandum items showing the value of these items at current exchange rates.

B. Measuring the Size of Foreign Currency Positions

For purposes of regulation and internal control of foreign currency positions, attention is normally paid to (1) single currency positions or position in an individual currency or group of currencies with a single—or nearly single—exchange rate; and (2) the overall foreign position, which measures a bank’s foreign exchange position in all currencies taken together.

Determining the size of a single currency position

Regulations on a foreign currency position must enumerate the items to include in the calculation. Until recently, there was no broadly accepted formula for the calculation but in 1996, the Basle Committee took the position that the calculation should be made by summing up the following items:

  • “the net spot position (that is, all asset items less all liability items, including accrued interest, denominated in the currency in question);

  • the net forward position (that is, all amounts to be received less all amounts to be paid under forward foreign exchange transactions, including currency futures and the principal on currency swaps not included in the spot position);

  • guarantees (and similar instruments) that are certain to be called and likely to be irrecoverable;

  • net future income/expense not yet accrued but already fully hedged (at the discretion of the reporting bank);

Positions in composite currencies, such as the ECU, need to be separately reported but, for measuring banks’ open positions, may be either treated as a currency in their own right or split into their component parts on a consistent basis.”9

A single currency position is used to measure the potential losses or gains to the bank for each unit of change in the given exchange rate. However, it should only measure the gains and losses that result from changes in the spot exchange rate (with any interest rate risk managed separately). Thus, the Basle Committee has taken the view that forward currency positions should be valued at current spot exchange rates, since forward rates also reflect current interest rate differentials. Similarly, the calculation should exclude items that are carried at historical cost.

Measuring the overall foreign currency position

Traditional measures

Supervisors and bank managers alike find it convenient to use a single summary figure to measure a bank’s overall foreign exchange exposure. Several definitions are commonly used, which vary primarily with regard to the degree of netting of long and short positions across currencies. The three most common are: (1) the gross aggregate position, which is the sum of all net short and all net long positions; (2) the net aggregate position, which is the absolute value of all short positions less all long positions, in other words, the position in the domestic currency, abstracting from all other positions; and (3) the shorthand position, which is the greater of the sum of the short positions and the sum of the long positions (See Box 1 for the formulae and an example).

“Traditional” Measures of the Overall Foreign Exchange Position

A. The Formulae

Let Fi denote the single currency, i, position. If Fi>0 (Fi<0), then a long (short) position exists. For purposes of calculation, let the absolute value of Fi>0 be Pi, and the absolute value of Fi< 0 be Ni, and let P be the sum of all Pi, and N be the sum of all Ni, then the three measures are defined as follows:

  • Gross aggregate position (GAP), GAP= P + N

  • Net aggregate position (NAP), NAP= absolute value of (P - N).

  • Shorthand position (SHP), SHP= maximum (P, N).

From the definitions, it follows that: GAP > SHP > NAP, and SHP = (GAP + NAP)/2

B. An Example

As an example, assume a bank has the following foreign exchange positions (measured in domestic currency equivalent):

article image

Then P = 170, and N = 175, and GAP = 345; NAP = 5; and SHP = 175

The appropriate summary measure depends on the correlation of exchange rate changes between the currencies in which a bank holds open positions. If exchange rate movements are perfectly correlated, as with currencies within a well-functioning monetary union, then cross-currency exchange risk can be theoretically ignored (in effect the exposure to one currency in the union can be treated the same as the exposure to any other currency in the union). In this case, the appropriate measure of exposure would be the net aggregate position, since this particular formula offsets long positions against short positions. On the other hand, if exchange movements are uncorrelated, then the losses (or gains) on an open long position would not be expected to be offset by gains (or losses) on an open short position. In this case, the appropriate measure of exposure is the gross aggregate position, since it sums short and long positions. The shorthand position, on the other hand, is a compromise between the other two measures, and is therefore appropriate when there is some, but not perfect, correlation among the currency exchange rates.

The Basle Committee has come out in favor of the use of the shorthand position for calculating capital requirements.10 However, it has made no explicit recommendation as to which measure to use when setting limits on a bank’s overall open foreign exchange position. These decisions would presumably be based on the expected correlations in movements of currencies in which domestic banks hold open positions. However, the shorthand position appears to be the best compromise if the extreme assumptions clearly do not apply.11

While the value of using the net aggregate position for prudential purposes is open to dispute, since it places no limits on the positions a bank can take in third currencies, it does measure the bank’s overall own position in the reporting currency, so this formula is generally used in reports which are produced for monetary or exchange control purposes.

Other methods of managing foreign exchange risk

The traditional measures of foreign exchange risk, however, have three drawbacks. First, as noted, they can take only partial account of the correlation in movements between the various currencies. Second, they take no account of the relative size of variations between currencies with the result that open positions in currencies, whose exchange rate relationships with the home currency are quite stable, are given the same implicit risk-weight as those which have a far more unstable relationship. For example in the Netherlands, the traditional measures would give open positions in deutsche mark the same treatment as open positions in U.S. dollars or Russian rubles. Third, these systems treat foreign exchange risk as being distinct and independent from the other risks faced by the bank, preventing the substitution of foreign exchange risk for other forms of risk.

One way to address the above noted problems is through the use of Value-at-Risk models (VAR). VAR models are used to measure the market or other portfolio risks for financial assets by estimating the greatest loss in the value that can be expected in a given trading book, with a given probability over a specified holding period. Thus, for example, the problem could involve estimating the maximum loss or value-at-risk at the 99 percent confidence interval, for a given foreign exchange portfolio over a ten-day holding period.12 There are numerous ways of estimating VAR models, however, the three most common are the variance/covariance, historical simulation, and Monte Carlo approaches.

As a related issue, supervisors should only set the broad operational guidelines regarding the estimation of the banks’ VAR models. It should not be heavily involved in deciding how frequently the parameters of a bank’s model must be revised, and especially in determining whether special adjustments must be made to incorporate changes in the expected behavior of particular currencies. Furthermore, letting banks make their own estimates allows them to use various weighting schemes to estimate parameters and even to include information on their expectations regarding currency behavior.

The variance/covariance (VC) approach

The VC approach uses summary statistics on exchange rate variability and correlations between currencies to estimate potential losses. The coefficients of the model are generally estimated using historic data. However, users of this model must address a number of statistical, informational, and policy problems. The first problem is that it assumes the predicted foreign exchange gain or loss will have a multi-variate normal distribution, which seems not to be the case, since exchange rates appear to have density functions that are flatter than a normal distribution (Booth and Glassman, 1987). Thus, if normality is assumed, the capital requirements for a given confidence level will be underestimated.13 Techniques have been developed to avoid the problems associated with the assumption of normality, but they have additional informational requirements (Leahy, 1991).

A second problem is that the VC model has large informational requirements. For example, even assuming normality, estimating the distribution of outcomes for a portfolio with six currencies, requires 27 parameters (6 means, 6 variances and 15 covariances). But bank portfolios are often far more diverse, with correspondingly greater informational demands. Furthermore, basing statistics strictly on historical data may exclude important available information, while incorporating new information requires either arbitrary adjustments in the coefficients or the use of ad hoc weighting schemes for estimating those coefficients.

A third problem is determining who should estimate the model. Clearly, this is a job for the banks, not the supervisor.14 However, even so, the supervisors must be able to test the coefficients of the model to ensure that it is sound, and does not allow the bank to pursue unacceptably risky policies, which in turn may require the supervisor to maintain its own VAR model. Furthermore, in many countries the banks would be unwilling to share a single model, implying not only that each bank may need to bear the cost of creating and updating its own model (or buying it from a vendor), but that the supervisor be able to test each bank’s model.

Historical simulation approach

The historical simulation model uses historic data to calculate changes in a trading book that would have been experienced during that simulation period. This approach offers several distinct advantages over the VC approach. First, it requires no assumptions about the distribution of the variables; instead, the estimated value at risk at each confidence level is the actual loss that would have occurred during that period, with that portfolio, given the events that actually transpired. Second, the variances and covariances of each asset need not be estimated. Third, once the details of the simulation are determined, ad hoc adjustments are not normally considered.15 The major downside of the problem is that historical simulations do not make full use of the current information available, and do not incorporate information on changes in underlying conditions, such as changes in exchange rate arrangements or in the outlook for particular currencies.

Monte Carlo approach

The Monte Carlo approach is used to estimate the value at risk based on a large sample of simulated price paths for the assets in the portfolio. Generally the price movements are simulated using a dynamic model, which is given a large number of shocks produced by randomly generated data (assumptions must also be made about the distribution of the random data). Then, if the assumed model of price dynamics is correct, the simulation may produce highly reliable estimates of the value at risk. An advantage of this approach is that the simulations can be designed to take maximum advantage of the latest available information on exchange rate movements. In addition, if the underlying price model is thought not to have changed, historic data can also be sampled randomly as the input data for the simulations. A disadvantage is that the underlying model will be arbitrary, and if the dynamic model is not correct, neither will the estimate of the value at risk.

Stress tests

One shortcoming of VAR models is that they are based on normal day-to-day developments, and only incorporate exceptional shocks insofar as they occurred during the estimation period for the VC model or in the simulation period for the historical simulation model. Even then, no additional weight is given to these observations. However, large shocks can and do occur, and it is in these situations that banks are most likely to have the capital bases severely damaged. It is for this reason that the Basle Committee has come out in favor of supporting VAR models only with the use of stress tests.16

Stress tests are used to estimate potential losses under extreme assumptions. This may start by estimating the peak losses that would have been experienced using a historical simulation. This may be supported by estimating the impact of several notable historical shocks, such as the fall in equities prices in 1987, or the Exchange Rate Mechanism (ERM) crises in 1992 and 1993. Such testing can also be supported by tests based on guesses regarding the types of shocks that could occur in existing operating environment.

Structural positions

Banks in countries with weak currencies often wish to hold structural or ongoing long positions in foreign exchange. The problem arises because a depreciating domestic currency will cause the value of bank capital, in foreign currency terms, to decline if the bank’s foreign currency position is balanced (unless the hedge is achieved with a put option).17 Thus, if a bank holds significant foreign-denominated assets, it may be forced to increase its capital or reduce its assets to meet regulatory capital requirements, unless it can generate additional capital (in local currency terms).

If banks are allowed to hold such structural positions, then such positions should be excluded from the calculation of the open position in the corresponding currencies. The Basle Committee’s proposal is that national supervisors should be free to allow their banks to hold structural positions, and for all these positions to be excluded from standard prudential limits, if certain conditions are met.18 In particular: (1) the positions must be nondealing in nature;19 (2) the national authority must be satisfied that it does no more than protect the bank’s capital adequacy ratio; and (3) the exclusion is applied consistently. Support for this approach appears broad-based.

Holding structural positions raises several regulatory issues. Perhaps the most sensitive issue relates to the treatment of revaluation gains and losses on the structural position, particularly their distribution and taxation. Since the profits derived from structural positions are unrealized and are intended to support the bank’s ongoing operations, there is a strong case that these profits should be neither taxed nor distributed. One way to protect them from such actions would be to record such gains in a special revaluation account rather than moving them through the profit and loss account.

Consolidated measure of foreign exchange risk

There is broad international support for the consolidated supervision of risk.20 This is because problems in one affiliate can affect a financial or corporate group as a whole, and consolidation reduces the likelihood that risks taken by one institution can be “hidden” in other unsupervised institutions. However, measuring risk on a consolidated basis also permits intra-group netting, and may be thus less demanding. Furthermore, supervision on a consolidated basis should not be used as a reason to preclude the measurement of risk at an individual institution.

Difficulties may arise when applying this principle. One problem is that internationally active banks may adjust their end-of-day positions by passing assets or liabilities to an affiliate in a later time zone; a legitimate way to continuously manage positions. However, as some prudential reports are prepared by consolidating the individual reports of the affiliates at the end of the business day in each time zone, they may understate the bank’s true open position. Capital controls may also cause problems. For example, a bank may have a closed position on a consolidated basis, but if some of its revaluation gains are booked in a country which limits the distribution of such profits, these gains cannot be used to offset the bank’s revaluation losses in other countries.

III. The Management of Foreign Exchange Risk

The prudential control of foreign exchange risk should begin by ensuring that the bank has effective systems of internal control to allow it to be able to adequately manage the risks that it faces.21 Bank supervisors also normally require that foreign exchange reporting systems be in place to enable them to monitor banks’ open positions. Supervisors also often impose other prudential controls, including direct limits on the range of foreign exchange operations a bank may carry out, limits on banks’ open foreign exchange positions, and capital requirements against foreign exchange risk.

A. Internal Controls

A strong system of internal controls is a prerequisite for effectively managing market risks, particularly foreign exchange risks where large positions may be quickly created. While bankers are responsible for devising these systems, supervisors are responsible for ensuring that the framework is adequate and enforced by bank management. A part of these controls often involves the imposition of limits on foreign exchange positions, capital requirements, or both; however, supervisors in many countries also issue regulations or guidelines regarding the design of banks’ internal control systems.22 The guidance offered varies from quite general to highly specific, and in some cases it includes rules specifically relating to foreign exchange-related operations. The areas covered usually include accounting and information standards, allocation of responsibilities, and internal compliance.

Accounting and information standards23

The internal control system must be supported by a system for monitoring foreign exchange risk. This system must: (1) immediately record all foreign exchange operations and maintain a permanent record of all transactions; (2) immediately and continuously update positions in each traded currency; and (3) continuously measure the results, for each and all currencies. It must also ensure that foreign currency claims and obligations originating from all of the bank’s operations are quickly entered to allow accurate monitoring of the bank’s overall foreign exchange position. The full evaluation of such systems generally requires an on-site inspection to assess both the design of the monitoring system, the computer systems on which the system operates, and how the bank actually uses these systems.

Allocation of responsibilities

Supervisors should also ensure that a bank has properly allocated internal oversight responsibilities. These responsibilities should be assigned by the board of directors. In small banks, the responsibility for foreign exchange operations may be delegated to a senior operating officer, while in larger banks, it may be appropriate to assign such responsibilities to a committee of senior managers. The committee generally includes the managers responsible for the major lines of business, including the lending function and foreign exchange transactions.

An increasing number of banks now use an independent risk control unit, which operates outside of the bank’s trading units to design and control the overall risk management system.24 This group is charged with developing and testing the bank’s risk management system and for allocating risk limits, as defined by the bank’s internal model and/or regulatory standards, to the operational units. Risk control units normally report directly to the bank’s senior management.

The internal control system should include a written policy for foreign exchange operations, approved by the board of directors. The policy need not be lengthy and detailed, but it should provide guidance on the tolerance level for foreign exchange risk. This includes requiring the establishment of internal limits on foreign currency positions, and naming the persons responsible for proposing and approving the internal limits. The board should also be required to approve these limits.

Key aspects of this policy include the establishment of limits on overnight and intra-day open positions in each currency, and for all currencies combined. In addition, any domestic branches that are not on-line with the head office should have sub-limits delegated to them to ensure that the bank’s overall limits are not violated. A wholly owned subsidiary is normally treated as a branch for this purpose, while joint venture banks are treated as independent institutions. However, the parent bank should know the limits of its affiliates in order to avoid an undesired level of risk on a consolidated basis.

Of the 19 countries surveyed, eleven had regulations or guidelines governing general requirements for an internal control function, while two (Colombia and the United States) implicitly established these requirements (Appendix I, Table 3). Nine of the survey countries have regulations requiring written policies on internal controls, while three (Indonesia, Singapore, and the United States) have no general requirement, but a written policy was required by the supervisors.

Internal compliance

A bank must also have systems to monitor compliance with its internal rules. As a minimum, this requires that: (1) the front and back offices be fully separated—the back office must never report to the front office; (2) a qualified officer to be responsible for monitoring compliance; (3) controllers to have specialized knowledge of foreign exchange; (4) controllers be independent of the foreign exchange department, and to report directly to the chief executive officer or the board of directors;25 and (5) a minimal periodicity be established for assessing compliance. Supervisors also often insist on receiving a regular report describing the steps taken to implement foreign exchange policy guidelines and noting any major change in their internal organization of the foreign exchange function.

B. Reporting Practices

The supervision of banks’ foreign exchange exposures requires a formal reporting system. Regulations must set: (1) the format of the reporting forms; (2) the frequency with which the various reports are prepared; and (3) the frequency with which the various reports must be sent to the supervisory authorities. While all of the countries surveyed require some form of reporting on banks’ currency position, and most require the banks’ to calculate their single and the overall positions at least daily, most supervisors surveyed only require that reports be sent on a weekly or a monthly basis.

C. Direct Limits on Banks’ Foreign Exchange Operations

One way to limit the foreign exchange risks a bank may take on to directly limit its foreign exchange operations. This may be done with licensing requirements or by supervisory order. While such licensing may be explicit, the need for a formal arrangement is not altogether clear; in some cases these requirements may be a carryover from the era of exchange controls. Only seven of the nineteen countries surveyed have specific licensing requirements on general foreign exchange operations, and none of the industrial countries surveyed have this requirement. Some, however, do require that a bank receive special permission from the supervisor prior to engaging in certain riskier foreign exchange operations, such as swaps, derivatives, or writing options.

D. Limits on the Size of Open Positions

As noted, most countries impose prudential limitations on banks’ open foreign exchange positions. However, these limits are defined differently in different countries. In most, the limits are set as a percentage of a bank’s capital, but in a few countries other considerations also apply. For example, in India the overall limit must be formally approved, while in Malaysia the quality of bank management is factored in by the authorities in setting the overall limit.

Limits on single currency positions

In theory, limit’s on a bank’s single currency position should be based on the level of risk of that currency, and the income volatility that the bank or the supervisory authority will accept as a result of that exposure. However, supervisors are reticent to explicitly commit to the size of acceptable fluctuations in income derived from changes in exchange rates, and relatively few countries’ regulations differentiate between currencies.26

The use of explicit single currency limits is not very common. Of the nineteen countries in the survey, only four, Czech Republic, Hong Kong Special Administrative Region (Hong Kong SAR), Malawi, and Oman set such limits (see Appendix I, Table 4). Three of the four also have differential limits depending on the currency. Hong Kong SAR and Malawi allow banks to hold larger positions in U.S. dollars, while Czech Republic allows larger positions in “convertible” currencies, which include those of Australia, Canada, the EU, Japan and the United States. Regulators in Hong Kong SAR may also explicitly reduce the single currency limits for a given currency.

Single currency limits vary considerably. Oman allows positions of up to 40 percent of capital, while the Czech Republic sets limit at 15 percent of capital for “convertible” currencies, and 2 percent of capital for other currencies. Malawi sets a limit of 15 percent of capital for U.S. dollar positions and 5 percent for other currencies, and Hong Kong SAR uses 10 percent of capital as a norm, but supervisors may allow larger positions in U.S. dollars and smaller limits for some other currencies.

None of the countries surveyed currently imposes a sub-ceiling on open positions in groups of currencies, although several have done so in the past.

Limits on the overall position

While single currency limits help manage the risk of income fluctuations resulting from bilateral exchange rate movements, limits on the overall foreign currency position are designed to reduce the fluctuations in a bank’s income from either a general movement in the domestic currency or from exchange rate volatility. From a bank’s perspective, limits on the overall position are generally the most important constraint on their foreign currency portfolio.

Most countries, including fifteen of the nineteen countries surveyed, limit their banks’ overall foreign currency positions. The limits are normally set as a percentage of bank capital. However, the choice of calculation method varies considerably, with four using the shorthand position, three the net aggregate position, three the gross aggregate position, and five some other calculation. Among the countries using the shorthand method, the limits vary from 4 percent in the Czech Republic, for all “other” currencies, to 35 percent in Malawi. On the other hand, survey countries using the net aggregate method set limits in the range of 20-25 percent of capital, while those using the gross aggregate method set limits ranging from 21 percent of capital in Germany to 40 percent in Oman.27

E. Capital Requirements

In recent years, there has been a growing trend toward the use of capital requirements in the prudential control of banks’ market risks, including foreign exchange risks. While open position limits remain the instrument of choice in most countries, an increasing number of countries are implementing capital requirements. Five survey countries, France, India, the Netherlands, Singapore, and Spain now use capital requirements, while Germany and Hong Kong have announced plans to introduce these requirements by end-1997. In most cases, the changeover to capital requirements was closely related to the issuance of either the EU directive in 1993 or the Basle Committee proposal on market risks in early 1996.

The Basle Committee proposal, which was to be implemented by Committee member countries by end-1997, gives a framework for setting minimum capital requirements on all forms of market risks, including foreign exchange risk. It specifies that capital requirements against each form of market risk should be based on the potential losses associated with that exposure, and that the capital charges on all market risks should be calculated so as not to create artificial incentives for favoring one class of instruments over another. By treating all market risks in a similar manner, the plan allows a bank to substitute one type of risk for another.

The proposal allows capital charges against a bank’s foreign exchange position to be calculated in two ways.28 The first, which is expected to be the more common, estimates the bank’s overall foreign exchange position using the shorthand method, and requires a minimum capital charge of 8 percent against that position.29 The second allows the use of a bank’s own internal risk model to estimate capital requirements. It gives considerable flexibility regarding the choice of model, but the use of this approach is subject to stringent conditions, which are discussed below.

The EU directive, which was implemented by end-1995, bears many similarities to the Basle proposal in its rules regarding the allocation of capital for market risks. But there are also a number of differences; some of the more important are:30

  • the directive applies to a broader group of financial institutions, including securities firms;

  • while both approaches allow the use of the shorthand model, with an 8 percent capital requirement, the only alternative model permitted in the directive is a simulation model, based on historic data, that cannot take account of covariances between markets. A bank may use its internal model if it can be shown to have higher capital requirements than the standard model, that is, the bank’s model is more conservative than the regulator’s;

  • the directive does not permit similar assets, for example, equities, in different markets to have differing volatilities; and

  • the directive allows lower capital requirements for currencies whose movements are closely correlated with the home currency or are bound together by governmental agreement, such as the EMS.

Prerequisites for using VAR models

While the Basle Committee proposal gives bank’s great flexibility in their choice of VAR model, it also urges caution in permitting the use of such systems. Accordingly, it sets out detailed standards on when such models should be used, as well as qualitative and quantitative standards on their use and stress testing.31

Since model-based systems are computer intensive, complex, and have large informational requirements, the Committee recommends that the supervisory authority only give approval to use the model if, inter alia:

  • “it is satisfied that the bank’s risk management system is conceptually sound and is implemented with integrity;

  • the bank has, in the supervisory authority’s view, sufficient numbers of skilled staff in the use of sophisticated models not only in the trading area but also in the risk control, audit, and, if necessary, back-office areas;

  • the bank’s models have, in the supervisory authority’s judgement, a proven track record of reasonable accuracy in measuring risk;

  • the bank regularly conducts stress tests”32

Thus, it would appear that this approach requires a degree of sophistication on the part of both the banks and their supervisors that is not currently present in numerous countries, so supervisors in these countries should not consider allowing the use of such models until the necessary criteria are met.

Qualitative standards

The Committee proposes a range of qualitative standards to help assure the supervisory authority that the models are “conceptually sound and implemented with integrity.” These standards cover the bank’s overall risk management systems, which are a part of the bank’s overall internal control system, and include the need for an independent risk control unit; active involvement of management and the bank’s board in risk control; a fully documented system of risk controls, and a routine to ensure compliance with those controls; independent reviews of the risk measurement system by the bank’s internal audit system; regular ex post testing of the risk measurement model’s results; and integration of the model into the day-to-day risk management process, including its use in setting internal trading and exposure limits. In short, the bank’s risk management systems need to be comprehensive, well developed, and fully integrated with the bank’s operations.

Quantitative standards

The Basle Committee makes no proposal regarding the type of VAR model banks must use, noting that VC, historical simulation, and Monte Carlo models would all be acceptable. However, the proposal does establish minimum quantitative standards for the use of the model. The basic requirements are that the model be run using daily data, with a ten-day holding period used to calculate price shocks, and that the historic observation period for calculating the value at risk be at least one year, and estimated using the 99th percentile of a one-tailed confidence interval. Data should be updated at least quarterly, more frequently if the supervisor believes that there has been a rise in exchange rate volatility. The bank is also free to allow for correlations across financial instruments, if the supervisory authority is satisfied with the soundness and integrity of those estimates. In addition, capital requirements must be met daily, based on the higher of the requirements estimated on the previous day or the average daily value at risk over the previous sixty business days multiplied by a multiplication factor of not less than three.33 The multiplication factor may also be further increased if the performance of a bank’s model is deemed less than fully satisfactory.

The Committee also proposes that banks using internal models engage in rigorous stress testing of their portfolios to estimate peak potential losses, and that they report data on the bank’s recent peak losses compared to the estimates of those peaks made by the bank’s own model. Basing the required capital allocation on peak losses in a series of well-designed stress tests could be one way of replacing the minimum multiplication factor.

The definition of capital

The Basle Committee recommends that the definition of capital for covering market risk be the same as that in the Basle Capital Accord.34 However, national supervisors may also permit banks to use an additional form of subordinated debt, Tier 3 capital, solely to cover market risks.35 The amount of Tier 3 capital used to cover capital requirements on market risks may not exceed 250 percent of the primary (Tier 1) capital allotted to cover capital requirements from market risk.36

Capital requirements with a weak or developing banking system

From the above, there are clear advantages to using capital requirements to manage banks’ foreign exchange exposure versus the use of fixed percentage limits. The recognition of this led the members of the Basle Committee to support the adoption of capital requirements for its member countries. However, replacing fixed limits with the Basle Committee proposals on capital requirements may not be appropriate in all cases. To begin with the use of VAR models places stringent demands on both the banks and the banking supervisors in a country, and the use of such models is only appropriate in the most sophisticated countries. However, even basing capital requirements on the shorthand model may result in banks being permitted to take on levels of risk that may be higher than the domestic supervisors would favor, particularly if the banking system is not well developed or there are concerns about the accuracy of the data on bank capital.

The differences between the use of absolute limits and the Basle Committee system are highlighted in Table 1. The table shows that fixed percentage limits more tightly limit the potential loss a bank might face from an adverse exchange rate movement if its capital level considerably exceed the minimum capital ratio that is required against credit risk (and other market risks). Furthermore, capital requirements allow a rapid increase in the permissible maximum exposure—and hence maximum permissible losses—as the capital ratio rises. This, by itself, may raise serious concerns on the part of supervisory authorities.

Table 1.

Risks Using Capital Requirements versus Percentage Limits1/

article image

Assumes that capital allocated for counterparty risk is 8 percent of total assets, and total assets are 100.

Assumes that capital for market risks is 8 percent of the bank’s shorthand foreign exchange position, and does not necessarily need to allocate additional capital for other market risks.

Capital less capital set aside for counterparty risk, multiplied by the capital requirement.

Assumes the limit on the bank’s shorthand position is 40 percent of total capital.

Capital multiplied by the percentage foreign exchange limit.

Far more serious problems may arise if there are concerns about the accuracy of the data on bank capital. For example, if banks are not fully provisioning against potential loan losses, then capital will be overstated. In these circumstances, the reported capital ratio may be high, even though the true level of capital may be inadequate or even negative. Thus, the bank may find it to its advantage to pursue a risky so-far-broke strategy in the foreign exchange market, in the belief that inaction may result in the bank’s closure. Furthermore, if the bank is considering pursuing such a strategy, the benefits of providing inaccurate data on bank capital increase sharply because of the greater leverage permitted with capital requirements. Hence, the supervisory authority should be convinced of the accuracy of banking data before fixed percentage exposure limits are abandoned altogether. In the interim, a combination of the two may be the most desirable approach.

IV. Conclusion

This paper examined issues in the management and prudential regulation of foreign exchange risk. In the course of the discussions, a number of issues came to the fore, but this section will highlight only a few of those. A central, but frequently forgotten, point is that the key to the effective management of foreign exchange risk lies in the bank’s own risk management systems. This should begin with an effective system of internal controls, based on stringent accounting and informational standards, and a clear allocation of responsibilities. While the institution is responsible for developing its internal control system, the supervisory authority should be expected to ensure that the system is adequate and used.

Banks’ overall foreign exchange exposures have generally been measured using one of three formulae, based on unweighted summations of their long and short positions in individual foreign currencies. These include: the gross aggregate position, which implicitly assumes that there is no correlation in currency movements (uncommon); the net aggregate, which assumes that the correlations in currency movements are perfect (unlikely); and the shorthand position, which assumes that the correlations are between the two extremes. While each formula has its advantages and disadvantages, the approach itself has three serious drawbacks: (1) it does not take full account of correlations in currency movements; (2) it implicitly assumes that the risk in every currency is the same; and (3) it treats foreign exchange risk as independent from the other risks faced by the bank. While no overall measure dominates the others, the shorthand position would appear to be the most appropriate for most banks in most countries. On the other hand, the use of the net aggregate measure should generally not be used, since it takes no account of the risks posed by positions in third currencies.37

One way to address these shortcomings is with VAR models, which measure portfolio risks by estimating the maximum loss that can be expected with a given likelihood over a specified holding period. There are three main VAR models, each of which has its advantages and disadvantages, and there is no consensus as to which approach constitutes the best practice. The variance/covariance model is estimated using summary statistics on exchange rate variability and correlations between currencies. It is potentially quite accurate, but has large informational requirements, and estimating the coefficients of the model is not straightforward. The historical simulation model, on the other hand, uses a simulation of actual exchange rate developments over a specified period. While this avoids the main drawbacks of the first approach, it takes no account of changes in underlying relationships. Finally, the Monte Carlo model uses a model of exchange rate changes developed by the user to estimate the value at risk by repeated simulations. Although this allows for the effective incorporation of assumed changes in behavioral relationships, such assumptions also make the underlying model arbitrary.

One shortcoming of VAR models is that they focus on normal exchange variations, and not an infrequent but large shocks that may take place.38 A widely accepted way of addressing this problem is by also subjecting the bank’s portfolio to stress tests, which simulate major shocks based either on actual experience or on estimates of potential shocks that could take place.

Supervisors also generally seek to control banks’ foreign exchange positions, either directly as a percentage of bank capital or through capital requirements. While limits on individual currency positions are on the wane, direct limits on overall positions remain common. Since the same overall position limits must apply to all banks in a system, shorthand limits are probably the most appropriate, although consideration could be given to allowing closely tied currencies to be aggregated.

An increasing number of countries are using capital requirements, instead of or in addition to direct limits. Much of the impetus for this change has come from the Basle Committee’s proposal on their use. The proposal requires banks to allocate capital, over and above that allocated for credit risk, specifically for all market risks, including foreign exchange risk. The capital requirement may be calculated in two ways. The most common is to set it at 8 percent of the bank’s shorthand position (plus the absolute value of its position in gold). A bank may also use its internal VAR model to estimate its capital requirement, but the use of such models must be subject to stringent conditions, including, inter alia, a qualitative assessment of the bank’s risk management system, and an agreement on the parameters of the model. The proposal also requires that capital requirements be at least three times larger than the value-at-risk estimated by the model at the 99 percent confidence level, which reflects a limited experience in the use of VAR models as a supervisory tool.

There are distinct advantages to the use of capital requirements, even if the use of VAR models is only appropriate for the most financially sophisticated bankers and supervisors. Still, some national authorities may not find it appropriate to rely solely on capital requirements. First, some supervisors argue that overall limits on banks’ foreign exchange positions are needed because they play a role similar to limits on credit exposure to single borrowers, namely that they avoid concentration risk. Second, capital requirements allow open limits to rise sharply as the bank’s capital/asset ratio rises above the minimum required for credit risks, which may permit a bank with limited skills in foreign exchange risk management to take excessively risky positions. Third, and potentially most serious, is that data on bank capital are frequently not accurately reported, and capital requirements may allow a misreporting bank to pursue far riskier foreign exchange strategies than it could with percentage limits. Thus, some supervisors may wish to either set higher capital requirements against foreign exchange positions than the minimum standards laid out in the Basle proposals, and/or retain maximum percentage exposure limits, particularly if they are concerned about the accuracy of the banks’ reported capital data.

APPENDIX I

Table 2.

Licensing, Accounting, Conversion and Revaluation1/

article image
article image
Source: Data provided by national authorities.

The detailed survey results are available upon request. The survey was conducted in 1996.

First column refers to whether or not commercial banks are required to obtain a specific license for engaging in foreign currency activities.

Second column refers to whether or not commercial banks have to follow specific accounting rules for foreign exchange transactions.

Third column refers to the exchange rate used for converting into domestic currency assets and liabilities denominated in foreign currency.

Table 3.

Internal Control and Reporting Practices

article image
article image
Source: Data provided by national authorities.

“No” in the first column means that the implementation of an internal monitoring and control system for foreign currency operations is not specifically required in the country regulations. However, the supervisors can establish general requirements.

The second and third columns report whether specific requirements are included in the regulations are issued by the supervisors.

In column four, (S) means sent to supervisors and (P) means that banks are required to prepare the report but they need not send it to the supervisors.

Table 4.

Limits and Capital Requirements

article image
article image
Source: Data provided by national authorities

References

  • Basle Committee on Banking Supervision, 1996, Amendment to the Capital Accord to Incorporate Market Risks (Basle: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation
  • Basle Committee on Banking Supervision, 1992, Minimum Standards for the Supervision of International Banking Groups and their Cross-Border Establishments (Basle: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation
  • Basle Committee on Banking Supervision, 1988, International Convergence of Capital Measurement and Capital Standards (Basle: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation
  • Basle Committee on Banking Supervision, 1983, Principles for the Supervision of Banks′ Foreign Establishments (Basle: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation
  • Basle Committee on Banking Supervision, 1980, Supervision of Banks’ Foreign Exchange Positions (Basle: Bank for International Settlements).

    • Search Google Scholar
    • Export Citation
  • BIS, 1996, Committee on Payment and Settlement Systems, prepared by the Settlement Risk in Foreign Exchange Transactions of the central banks of the Group of Ten Countries.

    • Search Google Scholar
    • Export Citation
  • Boothe, Paul, and Debra Glassman, The Statistical Distribution of Exchange Rates: Empirical Evidence and Economic Implications,” Journal of International Economics. (22), May 1997.

    • Search Google Scholar
    • Export Citation
  • European Union, Council Directive 93/6/EEC of 15 March 1993 onThe Capital Adequacy of Investment Firms and Credit Institutions,” Luxembourg: Office of Official Publications of the European Communities.

    • Search Google Scholar
    • Export Citation
  • Folkerts-Landau, David, and Carl-Johan Lindgren, 1998, Toward a Framework for Financial Stability, World Economic and Financial Surveys (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation
  • Hartmann, Philipp, 1995, “Capital Adequacy and Foreign Exchange Risk Regulation: Recent Developments in Industrial Countries, Special Paper No. LXXVII, December 1995, (London: London School of Economics Financial Markets Group).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, 1995, International Capital Markets: Developments, Prospects, and Issues (Washington).

  • Jackson, Patricia, 1995, “Risk Measurement and Capital Requirements for Banks,” Quarterly Bulletin, 35 (2), pp. 177-184 (London: Bank of England)

    • Search Google Scholar
    • Export Citation
  • Johnson, Norman Lloyd and Samuel Kotz, 1971, Distribution in Statistics: Continuous Univariate Distribution (New York: John Wiley and Sons).

    • Search Google Scholar
    • Export Citation
  • Leahy, Michael P., 1991, “Determining Foreign Exchange Risk and Bank Capital Requirements.” International Finance Discussion Paper, No 400, (Washington: Board of Governors of the Federal Reserve System).

    • Search Google Scholar
    • Export Citation
1

The authors would like to thank Olivier Frécaut, Peter Hayward, John Leimone, Carl-Johan Lindgren, Tom Nordman, and Inci Otker for helpful discussion, comments, and suggestions, and Natalie Baumer for her editorial assistance. Mr. Abrams is Deputy Chief in the Banking Supervision and Regulation Division. Ms. Beato was a consultant in the Banking Supervision and Regulation Division when the paper was originally drafted, and she is now Principal Economist in the Infrastructure and Financial Markets Division of the Inter-American Development Bank.

2

Market risk may be defined as “the risk of losses in on- and off-balance sheet positions arising from movements in market prices.” Basle Committee on Banking Supervision (Basle Committee, 1996), p.l. Exchange risk may be defined as “the risk that a bank may suffer losses as a result of adverse movement during a period in which it has an open position, either spot or forward, or a combination of the two in an individual foreign currency.” Basle Committee (1980), p. 1.

3

Certain risks related to foreign exchange operations are not covered in this paper, including inter alia, risk of losses when only one leg of a foreign exchange transaction is settled (called Herstatt risk). For a discussion of foreign exchange settlement risks associated with foreign exchange transactions, see Bank for International Settlement (1996).

4

Although this paper focuses on more supervisors’ efforts to regulate banks’ foreign exchange risk, best practices of bankers and supervisors should be essentially the same. The difference is that while a prudent banker seeks to maximize the value of the bank, a supervisor seeks to maintain an efficient banking system, and avoid the systemic difficulties associated with multiple banks failures.

5

Increasingly deep and efficient markets also allow a trader to take a position more quickly, so conventional reporting requirements and examination procedures that rely on limits and capital charges are less effective than they once were.

6

As used in this paper, the term “country” does not in all cases refer to a territorial entity that is a state as understood by international law and practice. In particular, as used here, the term “country” also may cover Hong Kong Special Administrative Region.

7

A variety of rules are used regarding the recording of fixed assets located abroad, but they are usually recorded in the currency of the country in which these assets are located.

8

However, the current spot rate is normally used for internal risk management operations if the institution trades on an ongoing basis.

9

Basle Committee (1996), p. 23. The proposal also includes a discussion of managing the risks associated with foreign exchange options, but this is beyond the scope of this paper.

11

If a bank has positions in groups of currencies whose movements are highly correlated, for example, in EMS currencies for a non-EMS country, it may be preferable to aggregate positions in those currencies and use the gross aggregate or shorthand method, rather than use those holdings as a reason for using the net aggregate position.

12

For a useful discussion of VAR models, as well as Basle Committee proposals regarding the use of these models, see IMF (1995), pp. 141—147.

13

Problems may also arise because both of these distributions allow a non-zero probability of the exchange rate turning negative unless, as Johnson and Kotz (1971) suggest, natural logarithms are used.

14

Imposing the supervisor’s model on the banks would also require that the supervisor accept the moral—and perhaps legal—responsibility for any extraordinary losses arising, should market developments diverge significantly from the parameters of the model. Such an arrangement might also encourage a bank to take risks which, while excessive based on its own model, are within the limits of the supervisor’s model, on the belief that should serious losses arise, the bank may seek restitution from the supervisor on the grounds that the supervisor’s model encouraged it to engage in excessive risk taking.

15

The model also does not have to face problems with nonlinearities, such as those associated with options contracts, which can undermine VC models.

17

In high inflation countries with low bank profitability, banks may also wish to hold a structural foreign currency position to hedge their capital against inflation.

19

Thus, structural positions would also include foreign assets that the regulator allows to be recorded at historical cost. Hence a bank may have a notional structural position rather than holding excess assets explicitly in foreign currency as reflected in its balance sheet.

20

The principle of consolidated supervision of banks was laid out in the Basle Committee’s Concordat in 1983, which has subsequently been amended several times, see Basle Committee (1983 and 1992, respectively).

21

For example, see Folkerts-Landau and Lindgren (1998), p. 39.

22

Basle Committee (1980) contains a succinct discussion of the role of bank management in ensuring the safety of a bank’s foreign exchange operations.

23

Procedures for recording and converting foreign currency assets and liabilities are discussed in II.A.

24

For further discussion, see III.E and Basle Committee (1996), pp. 39-41.

25

Some modern internal control systems assign primary responsibility for designing and implementing internal controls to line officers, while internal audit tests that proper controls are in place, and makes spot checks to ensure that they are implemented effectively. In the case of foreign exchange trading, compliance is enforced through the back office, which reports to either a senior manager or committee charged with compliance; not unlike the traditional audit function.

26

A parallel example is with credit risk, where most supervisors place a single limit on large exposures, independent of the borrower (other than government).

27

Some countries also set differential limits set on their banks’ overall long and overall short foreign exchange positions. Asymmetric limits would only seem appropriate when the risks are asymmetric. Otherwise, one must question whether the limit is being imposed solely for prudential purposes, or whether it is also being used as a capital control. Similar questions may arise when a country frequently adjusts its foreign exchange exposure limits, particularly if the changes are not in response to changes in either the capital positions of the banks or perceived improvements in banks’ ability to manage foreign exchange risk.

28

National supervisors may also exempt a bank from capital requirements on its foreign exchange positions, if its gross aggregate position does not exceed 100 percent of capital, and its net aggregate position is less than 2 percent of capital. Basle Committee (1996), p. 26.

29

The position also includes the bank’s open position in gold.

30

European Union (1993). For a further discussion of the differences, see Hartmann (1995), pp. 11-12, and Jackson (1995), pp. 183-4.

31

These standards are laid out in Basle Committee (1996), pp. 38-49.

33

The use of a multiplication factor greater than one may reflect uncertainties associated with the limited experience in the use of VAR models as a supervisory tool.

35

Tier 3 capital consists of unsecured, subordinated debt, with an original maturity of at least two years, and may not be repaid before the agreed repayment date without the approval of the supervisor. It must also have a lock-in clause stipulating that neither interest nor principal may be paid (even at maturity), if such payments would cause the institution to fall below, or remain below, its minimum capital requirements.

36

Tier 2 capital may also be substituted for Tier 3 capital, up to the same 250 percent limit. However, some Committee member countries will continue to require that at least half of all bank capital be primary capital, that is, Tier 1. Basle Committee (1996), p. 7.

37

Unless, of course, all or virtually all exposures are in a single currency.

38

Such as that experienced in several Asian countries in 1997.