10 Trends in Central Bank Reserves Management

D. Folkerts-Landau, and Marcel Cassard
Published Date:
July 2000
  • ShareShare
Show Summary Details
Jean-François Rigaudy

This paper reviews trends in central bank reserves management over recent years.1 It draws upon theoretical and statistical work prepared in various areas of the Bank for International Settlements (Monetary and Economic Department, Banking Department) and upon our own experience from contacts with central bank reserves managers.

Admittedly, an exercise that intends to identify the main trends and key factors runs the risk of oversimplification—since the world of central banks is made up of a wide spectrum of organizations, functions, and responsibilities. As a result of these differences, there is a great variety of ways in which central banks approach the question of reserves management. The matter is further complicated by the fact that confidentiality is often an important factor for central banks. Nevertheless, we have made every effort to guard against excessive generalizations. Some common features, however, can be identified:

  • Despite theoretical literature on the optimal level of reserves, their size is, most of the time and in most of the countries, a consequence of economic policy and particularly monetary policy or foreign exchange policy at the central bank. Few central banks have a reserve objective per se, although the size of reserves could in some cases represent an additional (and acute) constraint to economic policy.

  • Central banks hold reserves usually for three main reasons, namely: transaction needs, intervention purposes, and wealth accumulation. Each central bank has its own combination of these three features and no particular one appears to have been dominant over recent years. In addition, Goldstein and Turner have shown that a healthy cushion of international reserves appears to have taken on added importance for preventing crises in domestic banking systems in emerging market economies.2

This paper investigates the main statistical changes in foreign exchange reserves within the past few years. It then describes some significant changes that have taken place with regard to the objectives and investment policies of central banks. Finally, it points out some trends that could noticeably slow or even reverse these recent changes.

Evolution of Foreign Exchange Reserves

Foreign exchange reserves have grown at an increasing rate in recent years, while the currency composition of reserves has remained relatively constant. This has coincided with the trend toward a more integrated global economy. In the years up to 1996, the nominal size of reserves in dollar terms increased at an average annual rate of 14.3 percent (11.3 percent in 1993, 14.8 percent in 1994, and 16.8 percent in 1995). The growth in reserves in 1994 and 1995 was at least two percentage points higher than the 1986-92 average annual growth rate of 12.9 percent. This growth in reserves has not been symmetric between industrialized countries and emerging market economies (see Figures 10.1 and 10.2, and Table 10.1). From 1993 to 1995, industrialized countries experienced a slower growth in reserves than during the period from 1986 to 1992 (10.1 percent average annual growth rate against 11.2 percent), while over the same period, the foreign exchange reserves of emerging market economies grew at a rate of 18.7 percent—compared with 15.2 percent during the period from 1986 to 1992. As a consequence of this faster rate of growth, by the end of 1995, the total foreign exchange reserves of all central banks had reached a combined amount of $1,382 billion, of which $655 billion (47.4 percent) were held by central banks of industrialized countries and $727 billion (52.6 percent) were held by central banks of emerging market economies. Thus, the sharp growth in the reserves of developing countries has led to a major shift in the distribution of world reserves. Industrialized countries no longer hold the major portion of world foreign exchange reserves.3

Figure 10.1.World Foreign Exchange Reserves

(In billions of U.S. dollars)

Source: IMF, International Financial Statistics.

Figure 10.2.Shares of Total Foreign Exchange Reserves

(In percent)

Source: IMF, International Financial Statistics.

Table 10.1.Annual Growth in Foreign Exchange Reserves1
Country Type1986-922199119921993199419951993-95
Industrial countries11.2-3.7-4.84.612.013.710.1
United States14.9-11.9-12.93.8-0.819.17.0
Emerging markets15.222.36.318.917.519.618.7
Latin America20.639.239.423.5-5.716.313.2
Eastern Europe13.61.2-7.566.826.595.460.3
Total foreign exchange reserves3
Sources: IMF, national data, and BIS.

All numbers are presented in percentage terms.

Annualized growth over the entire period.

These numbers are expressed in billions of U.S. dollars.

Sources: IMF, national data, and BIS.

All numbers are presented in percentage terms.

Annualized growth over the entire period.

These numbers are expressed in billions of U.S. dollars.

Within the industrialized group of countries, Japan has been the one exception to the trend toward slower growth in reserves in recent years. In Japan, foreign exchange reserves increased 40.7 percent on average per year between 1993 and 1995, whereas the level of reserves in other developed countries remained relatively stable. Among industrialized countries, at the end of 1995, the bulk of the reserves were held by Japan ($172.4 billion), Germany ($60.2 billion), and the United States ($49.1 billion).

During the same period, the reserves of emerging market economies swelled and caused their share of the total to reach more than 50 percent. However, the growth was different between regions. In Latin America, foreign exchange reserves grew much slower over those three years (13.2 percent average annual growth rate) than they did in 1986-92 (20.6 percent). The level of Latin American reserves was $125.2 billion at the end of 1995. Also, although Eastern European countries experienced an impressive 60.3 percent average annual growth rate from 1993 to 1995 compared with the 13.6 percent average annual growth from 1986 to 1992, the total level of reserves held by central banks of that region were still small in relative terms ($82.8 billion). In March, the largest contribution to the rapid growth of reserves was taking place in Asian emerging market economies. Foreign exchange reserves of this region, which amounted to $423.6 billion at the end of 1995, had grown steadily at an average of 18 percent a year since 1986.

The size ($1,382 billion) and growth of foreign exchange reserves (14.3 percent from 1992-95) can be put into perspective by comparing it to world foreign exchange activity. As of April 1995, the average daily turnover of global foreign exchange markets (composed of spot, outright forward, and foreign exchange swap contracts) was estimated to be nearly $1.2 trillion ($1,190 billion).4 Furthermore, between 1992 and 1995, foreign exchange transactions increased by 45 percent or 12 percent annually. The size of total reserves ($1,382 billion) grew at an annual rate of 14.3 percent over 1993-95. Although there might be no logical link between these figures, it is nevertheless interesting to note the similarity between the growth in reserves and the growth in global foreign exchange turnover (see Figure 10.3).

Figure 10.3.World Global Foreign Exchange Turnover and Foreign Exchange Reserves

(In billions of U.S. dollars)

Source: Bank for International Settlements (BIS).

Part of the enormous increase in world reserves in recent years is attributable to factors such as the accumulation of interest, central bank foreign exchange activity, and increased trade and capital flows. Some idea of how much the growth in reserves arises from pure interest accumulation may be gained by calculating since 1986 the proceeds of investing the reserve total for each currency in the respective three-month LIBID (London Interbank Bid Rate) index, and converting back into U.S. dollars at each year-end. This approximates to separating out the accumulation of reserves purely attributable to leaving the funds in liquid assets. In fact, this rough calculation shows that as of year-end 1995, 47.2 percent of reserve growth from 1985 to 1995 in dollar terms would have been obtained from merely investing in three-month LIBID instruments and converting the returns into dollars (see Figure 10.4).

Figure 10.4.Foreign Exchange Reserves Over Time

(In billions of U.S. dollars)

Source: Bank for International Settlements (BIS).

Finally, it is interesting to compare the growth in reserves in different geographical regions with standard criteria for determining the adequate level of reserve holdings. The optimal level of reserves will certainly differ from country to country. One well-known rule of thumb is to have a level of reserves that covers three months of imports. Another approach is to determine a reserve level upon the contribution of trade to GDP, more commonly known as the average propensity to trade. The average propensity to trade is usually the ratio of exports to imports or an average of the two to GDP. Although there has been no exact preferred ratio established, the general idea is that a country might want relatively more reserves when its trade-to-GDP ratio is higher. Yet another method to assess an appropriate level of reserves, suggested by Scott Roger, is to examine current account variability by measuring the standard deviation of current account changes each year.5

Most of these measures are included in Table 10.2, which compares the evolution between two periods—1985 to 1991, and 1991 to 1995. Column 2 of the table identifies a country or synthesizes a “representative” country by taking an average of the reserves of each country in the region (weighted according to their share of the total). To examine the evolution of reserves, three methodologies for determining reserve levels, summarized in the previous Paragraph, are compared for different regions over the two periods. Column 4 of the table represents the average month cover of foreign exchange reserves for various countries and “representative” countries of different regions. It indicates that for industrial countries, the G-10, the United States, Japan, and Germany, foreign exchange reserves have adjusted to the increase or decrease in trade experienced by them in recent years. In other words, the ratios have remained relatively constant between the two periods. For Asian countries, these ratios have decreased. This indicates that for the “representative” country in Asia, growth in trade has outpaced the growth in reserves—a point of some significance in the light of the recent enormous growth in Asian reserves. Despite the decrease in this ratio for Asia, ratios there were still three times as high as the ratio for the representative G-10 country.

Table 10.2.Nominal Foreign Exchange Reserves and Various Metrics
Trade1Current AccountVolatility

(In billions of U.S. dollars)
Foreign exchange/trade

Month cover

Trade GDP


(In billions of U.S. dollars)
Foreign exchange


(In billions of U.S. dollars)
1991 to 19951991 to 19951995
United States43.46.20.710.557.30.649.1
Latin America17.961.87.417.76.32.8125.2
Eastern Europe5.930.83.630.33.41.782.8
1985 to 19911995 to 19911991
United States24.
Latin America6.238.14.516.33.61.763.4
Eastern Europe2.216.31.927.
Source: IMF, International Financial Statistics; national data; and BIS.Note: Asia = China, India, Hong Kong SAR, Korea, Singapore, Taiwan Province of China, Indonesia, Malaysia, Philippines, and Thailand; Latin America = Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela; and Eastern Europe = Czech Republic, Hungary, Poland, Romania, and Turkey.

Trade is defined as the sum of exports and imports divided by two.

Average stands for average of foreign exchange reserves and is computed by taking a geometric mean for each of the countries over the given period. For groups of countries, such as the G-10, a weighted average of all the countries within the group is taken to get the final figure. The weights are based upon percentage holding of foreign exchange reserves. This methodology attempts to capture a representative figure of that region based upon its share of foreign exchange holdings. This is in no way a “representative” country in terms of economic weight (as indicated by GDP); however, it may be a more interesting weight to analyze world foreign exchange reserves.

S.D. represents the standard deviation of annual current account changes.

This figure is the ratio of the average foreign exchange reserves over the period to the current account variability as defined by its standard deviation.

The total figures represent the total foreign exchange reserves of each region as of the ending period. In the upper panel, this is as of end-1995 and in the lower panel, this is as of end-1991.

Source: IMF, International Financial Statistics; national data; and BIS.Note: Asia = China, India, Hong Kong SAR, Korea, Singapore, Taiwan Province of China, Indonesia, Malaysia, Philippines, and Thailand; Latin America = Argentina, Brazil, Chile, Colombia, Mexico, Peru, and Venezuela; and Eastern Europe = Czech Republic, Hungary, Poland, Romania, and Turkey.

Trade is defined as the sum of exports and imports divided by two.

Average stands for average of foreign exchange reserves and is computed by taking a geometric mean for each of the countries over the given period. For groups of countries, such as the G-10, a weighted average of all the countries within the group is taken to get the final figure. The weights are based upon percentage holding of foreign exchange reserves. This methodology attempts to capture a representative figure of that region based upon its share of foreign exchange holdings. This is in no way a “representative” country in terms of economic weight (as indicated by GDP); however, it may be a more interesting weight to analyze world foreign exchange reserves.

S.D. represents the standard deviation of annual current account changes.

This figure is the ratio of the average foreign exchange reserves over the period to the current account variability as defined by its standard deviation.

The total figures represent the total foreign exchange reserves of each region as of the ending period. In the upper panel, this is as of end-1995 and in the lower panel, this is as of end-1991.

It is at this point that column 5 is interesting. One will notice that the Asian trade-to-GDP ratio is three times that of the G-10 representative. It might be quite logical for Asian countries to have more reserves given the importance of trade in their economies. For Eastern Europe and Latin America, the month cover actually increased. Finally, one may observe the evolution of the current account criterion in columns 7 and 8. Overall, current account variability has not changed much from the earlier period, except for the United States. The ratio of foreign exchange reserves to current account variability (column 8) has remained relatively stable for most countries, except Japan, Asia, and Latin America. Japan’s figure increased by approximately 30 percent, Asia’s by about the same amount, and that of Latin America by roughly 60 percent. Thus, if one believes that current account variability is a reasonable criterion to judge the adequacy of foreign exchange reserve holdings, then the Asian reserves had increased by too much—or conditions had changed so as to require a higher ratio.

Perhaps the most telling story from these figures is that the growth in Asian and Japanese foreign exchange reserves was so substantial. While some measures indicate that the growth was either sufficient or too little as a consequence of economic events, other measures indicated that this growth had exceeded what was warranted.

Over the same period, the composition of reserves, excluding gold, has remained relatively constant. Since 1970, there has been a decline in the percentage of U.S. dollar reserves within the overall total, though this decline has stabilized in recent years (see Table 10.3). Indeed, the currency distribution of reserves since 1992 has been relatively constant, with the exception of holdings of Japanese yen (which have declined by 12.8 percent since 1994). It is not easy to draw clear conclusions about a change in the composition of aggregate reserve holdings. Horii (1986) has shown that roughly only 1 percent of the 12.8 percent drop in the U.S. dollar share of reserves between 1972 and 1984 was attributable to a change in the preferred holdings of currencies by central banks.6 A much more significant factor was the greater accumulation of non-U.S. dollar currencies by countries with different reserve baskets. Horii also found no evidence that the currency composition of foreign exchange reserves could be explained by international trade patterns or traditional risk and return arguments. It seems that a large part of currency management focuses upon liquidity concerns—hence, unsurprisingly, the large holdings of U.S. dollars.

Table 10.3.Currency Composition of Foreign Exchange Reserves1
U.S. dollar77.267.268.356.260.961.460.761.4
Japanese yenn.a.
Deutsche mark1.914.813.616.915.015.615.916.1
Pound sterling10.
French franc1.
Dutch guildern.a.n.a.
Swiss franc0.
Private ECUn.a.n.a.
Source: BIS.

The compositions are computed for year-end figures of foreign exchange reserves. All numbers are expressed in percentages.

Source: BIS.

The compositions are computed for year-end figures of foreign exchange reserves. All numbers are expressed in percentages.

Finally, a word or two should be said about gold. The estimated 35,000 tons of gold held by central banks (equivalent to about $430 billion) still represent a significant proportion of global reserves, despite a steady decline from about 90 percent in 1970 to 20 percent today (see Figure 10.5). The effect on market sentiment of central bank gold sales should not be underestimated, but such sales as have occurred have been modest in comparison with the level of total central bank holdings, and the proportional decline is almost entirely due to the rise in foreign currency reserves.

Figure 10.5.Gold Share of Central Bank Reserves

(In percent)

Source: IMF, International Financial Statistics.

Significant Changes in Objectives and Investment Policies

The significant developments over recent years in central bank reserves management have been largely prompted by dramatic changes in the external environment, including the internationalization of money and capital markets as well as the impressive advances made in technology and information systems. It is, however, also the result of profound changes that have taken place within central banks themselves, whether arising from increased accountability for their actions demanded by the public, or by the trend toward greater independence, or the significant improvement in communication between central bankers themselves, fostered by international organizations (such as the BIS) and private financial institutions. In many cases, of course, it will be a combination of all these factors.

This section reviews some of the more important changes currently being experienced by central bank reserves managers.

Progressive Increase in Transparency

Reserves management has for a long time been considered one of the most confidential activities of central banks, partly because of its links to foreign exchange intervention, but also because of the central banks’ own traditions of secrecy. This confidentiality with regard to reserves management is still predominant among central banks. Nevertheless, some changes are emerging. Regulators have pushed commercial institutions toward more transparency in the presentation of their published accounts. Following the same path, finance ministries and governments increasingly require detailed and sophisticated information from central banks on the management and performance of national reserves. As a price for greater independence, central banks could in the future be subject to demands for more transparency as their accountabilities become more precisely identified. In emerging market economies, such transparency may to a certain extent be imposed by the market—which will require more information before committing substantial funding to a country.

At present, a minority of central banks make their investment policies and precise performance figures available to the public (such as Australia and New Zealand) or through a mark-to-market end-of-year balance sheet (such as the Netherlands and Sweden). This trend may be expected to accelerate in the future.

Significant Changes in Definition, Objectives, and Constraints of Reserves Management Policy

The security and safety objective remains paramount to reserves management in central banks, leading toward a tightening of credit constraints and a better assessment of the risks incurred. Central banks are averse to credit risk, such as exposure to losses through defaults. Even if the losses are eventually recouped through litigation, most central banks do not want the risk of having their names even mentioned in this connection (given the reputation risk and cost involved).

To some extent, as a result of this credit aversion, central banks seem to price credit risk differently from other participants. Over recent years, therefore, few central banks have loosened their credit risk constraints. On the contrary, most of them have moved toward tighter rules in response to the perceived weakening of credit ratings among banks and a general tendency toward concentration. More specifically, the trend toward tighter rules will only have been encouraged by various examples of central banks being unwittingly caught up in bank failures. The growth of marketable debt in most developed countries has resulted in a marked increase in the availability of liquid and highly rated government securities, providing central banks with attractive opportunities to diversify and replace bank risk.

The sensitivity of central banks to credit risk is such that there is a clear trend to develop credit risk groups charged with the independent assessment of banking counterparties and even, sometimes, of sovereign entities—with a view to setting limits for transactions. For this credit risk function, central banks are drawing on their own analytical capacity but also, like the market, are using the services of major rating agencies. It is also worth mentioning that, with regard to a specific failure of a financial institution, conflict may arise between the role of a central bank as an investor, on the one hand, and as a regulator on the other.

Liquidity has also remained a major requirement for central banks, and they have noticeably improved and refined their liquidity management. Of course, liquidity remains difficult to grasp and to quantify as there is no precise economic model to analyze it. Liquidity is relative and often dependent on market conditions. The liquidity premium cannot easily be isolated from the market or credit risk premiums. For central banks, liquidity has another dimension, which is the necessity not to unsettle markets and create financial disturbances that could conflict with other policy objectives.

As measured by the relative share of securities in foreign exchange reserves (BIS computations show that securities represented 75 percent of central bank foreign exchange reserves in 1995), liquidity has continued to increase in recent years, mainly in relation to the development of liquid securities markets in most of the currencies that are held as reserves.7

In addition, it is widely recognized that the dominance of the U.S. dollar in foreign exchange reserves (particularly reserves kept for intervention purposes) can also be partially explained by the role of the dollar as a transaction currency and the liquidity of the U.S. money and capital markets. There have, however, been several changes to the approach of central banks to liquidity management:

  • Many central banks have split their reserves between liquidity and investment portfolios in order to deal with different sets of objectives and constraints. Liquidity portfolios usually have a short-term management horizon and duration. The low, but effectively riskless, return achieved on this portfolio is then viewed as the price to pay for liquidity. The key question when managing such a portfolio is more to determine its size than its asset composition (which is usually short-term money market instruments). Some central banks have also recognized that through derivatives they might even, to some extent, split the management of market risk and liquidity risk. For example, through the use of futures, it is possible to hold securities across the curve (which are sometimes more liquid than “current” issues).

  • Some progress has been achieved in the qualitative analysis of liquidity by central banks. Most now assess carefully their potential liquidity needs through historical and statistical analysis of past interventions and, for some, worst-case scenarios or stress tests. Nevertheless, when evaluating the potential need for liquidity, central bank managers also need to adjust their portfolios and take full account of recent developments in the economic, and particularly in the foreign exchange, policies of their countries.

  • There has also been some evolution in the approach to liquidity by central banks. This has traditionally consisted of defining liquidity risk as the risk that money cannot be recovered quickly, or recovered only at considerable cost. The ability to sell or to liquidate positions is obviously relative and linked to the depth of the market and its capacity to absorb large trades. Apparently, globalization of financial markets and the extension of hedging possibilities have enhanced the liquidity of most markets. However, there is a growing concern that, due to the concentration of the industry, the liquidity of some markets would be quite vulnerable to the potential default of a major player.

  • Another approach is to view liquidity risk as arising from funding mismatches in cash flows. In other words, to bear liquidity risk is to run the risk of not being able to meet cash needs or to run the risk of not having funds readily available when required. As compared to the traditional approach (ease of liquidation), this alternative considers that cash needs can be met not only by liquidating assets but also through borrowing outright or against collateral.

  • Indeed, the rapid development of repurchase (repo) and swap markets has made it easier for a central bank to borrow foreign currency for a short period of time against its domestic currency, as well as against other kinds of collateral: securities, other foreign currencies, gold, and so on. Some central banks have in addition developed agreements on repo lines with other central banks: this type of agreement has sometimes been concluded between institutions in the same geographical area (such as the recent bilateral U.S. dollar repurchase agreements concluded among a number of Asian countries).

  • Nevertheless, liquidity in the sense of financing capacities still remains a relative concept—as the conditions under which funds can be borrowed vary significantly from one debtor to another. Each central bank has to assess its own trade-off between holding securities, which can be easily liquidated, and its borrowing costs. Globally, it seems that liquidity management has become much easier for countries that enjoy a high-quality credit rating and are well regarded by financial markets. Others have also benefited from the overall increase in the liquidity of markets, but to a much lesser extent.

  • The most noticeable change as regards central bank reserve management objectives is the increasing weight placed on return. Return maximization is still a “tertiary” objective but is playing a steadily increasing role in reserves management. The main reasons for this evolution include the following.

    • In many countries, there is increasing pressure from governments to maximize income from reserves to benefit government finances or to service external debt.

    • There is a more widespread perception that new financial management techniques provide cost-effective ways of implementing a conservative management policy and therefore a return-oriented management is no longer contradictory to a conservative management.

    • High returns are required as evidence of the professionalism of the staff (as good as the private sector) and of the reputation for efficiency of the central bank.

    • High returns are necessary to justify and to finance the evergrowing investments needed (staff, IT, back office, and so on).

Although it is dangerous to generalize, it seems that the emphasis on return is often linked to the size of the reserves; the higher the level of reserves, the more weight attached to achieving a higher return. This is consistent with the idea that central banks seek first to meet their transaction and intervention needs, only then turning to the goal of maximizing returns. It is also consistent with general economic and financial theory, which assumes that the richer an investor is, the less risk averse he or she is.

The increasing appetite for return puts more emphasis on the formulation of a precise return objective. Most central banks claim that their objective is to achieve a “reasonable,” “satisfactory,” or “optimal” return. However, central banks have different attitudes toward this question. Given the uncertainty of their investment horizon and of their currency allocation, most of them set relative return objectives versus a benchmark. A few of them define a minimum total return objective at the start of each new financial year, specifying the reference currency (which could be the domestic currency, a specific foreign currency, or a currency basket) and a reference currency allocation.

Increased Appetite for Return Has Led to More Active Management

The tremendous improvements in communication and technology have helped markets to become much more efficient: mispricings, arbitrage opportunities, and “anomaly trading” have more or less disappeared or at least have become so limited in time and in size that they require leveraged positions to take advantage of them. Figure 10.6, prepared by J.P. Morgan, shows indeed a significant reduction in the magnitude of term-structure “errors” in the main government bond markets (in the example, over 10 years, the average difference between the 7-year government bond yield and its theoretical value as calculated by a term-structure model based on 2-year, 5-year, and 10-year bonds declined substantially). In other words, central banks can no longer rely mainly on traditional and relatively riskless arbitrage, an activity on which most of them originally built their reserves management capacities.

Figure 10.6.Tighter Term Structures

(In basis points)

Source: National sources.

Note: Average standard error in regressing daily changes seven-year government yield on 2-, 5-, and 10-year benchmarks in the United States, United Kingdom, Japan, Germany, and France during the first six months of the year.

Many central banks have since moved toward a more active management of their interest rate risk and, less frequently, of their foreign exchange risk. Most of them actively manage the duration of their portfolios in tight limits to increase relative return versus a benchmark that represents passive management: over recent years, central banks have become major players in short- and medium-term securities of the principal G-10 government and money markets. In response to this development, a large number of private financial institutions have over the years created specialized central bank desks, and they devote significant efforts to providing information and training to central bank staff.8

Only a few central banks actively manage their currency risk. The size of their foreign exchange positions is, however, often limited.

Finally, to increase their return, most central banks have partly eased the constraints on the maximum authorized maturity of investment products and on the type of instruments permitted. However, both diversification and maturity-lengthening remain limited for the present.

Performance Measurement and Risk Assessment

Benchmarks have become the norm as regards internal performance assessment of reserves management. Indeed, they present many advantages: as a means of communication and control across the whole organization; as an expression of risk preferences; as a means of the long-term investment policy of the management and the board; as a market-neutral position; and as an objective for portfolio managers. Over recent years, many central banks have found it useful to use such techniques, if only to revisit and increase the consistency of their reserves management policies.

However, many central banks still have a traditional cost-accounting approach for publishing their results. The inconsistency of this with the benchmark “total rate of return” approach remains limited—provided the average duration of the portfolio is kept rather short (up to one year). There is much more of a problem when funds are invested in the longer part of the yield curve, where capital gains and losses can often be significant. Both solutions have drawbacks: mark-to-market accounting can create a higher volatility in the yearly profit-and-loss accounts owing to market fluctuations (this sort of volatility is usually regarded as counterproductive for the central bank); book cost-accounting, on the other hand, can create profits and losses that are not directly related to market movements. It can even sometimes interfere with the management of the portfolio when it contains large potential losses.

Benchmarks are usually rather short (from three months to two years) for two main reasons. One reason is that, owing to the specific purposes of reserve assets (they could be unexpectedly required for interventions), central banks cannot always assume they will keep the same amount of funds over a full interest cycle: the usual risk/return analysis, which in efficient markets would argue for a long duration, cannot simply be applied to central bank reserve management. Another reason is that being conservative investors, central banks keep their investments short to avoid a negative total return even on short periods—though, over recent years, there has been a modest lengthening due to an increased appetite for return. As mentioned earlier, only a few central banks actively manage foreign exchange risk, so global benchmarks are the exception, and most central banks have different benchmarks for each currency.

Considerable efforts have been made by central banks in defining appropriate benchmarks. Many central banks have developed internal, sometimes sophisticated, benchmarks, which they maintain and improve constantly. But there has also been a recent trend toward using external benchmarks, which can be viewed as providing more credibility to the investment and performance measurement process. More important, the availability, the range of countries, markets, and market segments covered by external indices and subindices have considerably increased—allowing central banks to combine them easily in a way that best suits their needs.

There is nevertheless a risk in focusing excessively on the benchmark as shown by the 1994 bond markets sell-off. A sharp increase in market volatility and adverse movement in interest rates could indeed lead to a significant and unexpected negative return. Many central banks have come to realize the shortcomings of this approach and also of their traditional risk measurement, which generally was based on duration analysis. Hence, the welcome given by central bank reserve managers to the rapid development of new risk management techniques.

Indeed, central banks have shown great interest in these new techniques and in the concept of value-at-risk (VAR), in particular. Among factors of note here has been a progressive recognition by central banks that financial risks are an integral part of their core business activities (not only for reserve management, but in other areas such as the application of monetary policy as well). Some of them are therefore keen to use the most recent and sophisticated techniques to measure their financial risks. There has also been substantial progress in information technology and the availability to a large investor base of pre-computed standard risk data (particularly through J.P. Morgan’s RiskMetrics set of correlations and volatilities), which has made the task of measuring these financial risks much easier. In addition, central banks are usually thoroughly involved or are responsible for the supervision of their domestic banking systems. A deep, theoretical, and practical understanding of financial risks is regarded today as essential in carrying out this task.9

A few central banks are already using VAR or similar concepts at an operational level to define their risks and objectives more accurately. Many others are considering it. This process is only in its early phases. It remains to be seen if these techniques, which could lead to substantial changes in the management of funds, will be accepted and adopted at higher management levels within central banks.

Increased Use of External Portfolio Managers10

Over recent years, many central banks have entrusted one or more external asset management companies with a modest part of their reserves. They do so for a variety of motives, including the fact that the use of external portfolio managers allows them to invest indirectly in certain markets where it would be difficult to be active directly, either for time zone reasons or because they lack the required expertise. External portfolio managers are often used as yardsticks for comparisons with risks and returns achieved internally, even if the guidelines established are often quite different from internal ones. The major reasons for using external managers, however, lies in the desire of central banks to benefit from more specialized and different skills in portfolio management. An external portfolio manager usually offers comprehensive systems for risk and performance measurement, training facilities, market research, and, more importantly, the opportunity to profit from an ongoing dialogue about fund management with a firm using private sector investment knowledge and style.

A minority of central banks have pushed this reasoning further: whether for lack of internal resources, the conviction that the private sector can perform as well as internal managers at a competitive price, or else the willingness to foster a local fund management industry, some institutions (such as the Hong Kong Monetary Authority and the Central Bank of Argentina) have decided to entrust a larger part of their funds for management by the private sector.

Striking evidence of the importance attached to external portfolio management by certain central banks is the ever-more elaborate process for selecting external managers. If public tenders for external portfolio managers are still the exception in the world of central banks, the information required has become comprehensive.11 In some cases, use is made of specialized external consultants to assist the central bank in its selection process.

Despite some disappointment about actual experience on performance, particularly when markets are performing badly (for example in 1994), this trend is expected to continue. It should be stressed in conclusion, however, that many central banks have found that “managing” an external portfolio manager can sometimes prove as demanding and time-consuming as managing the funds for themselves directly.

Evolution of the Use of Derivatives

While many institutions remain cautious, other central banks have increased their use of derivatives for reserves management purposes, considering these instruments as an additional tool to manage their portfolio rather than a specific asset class. Studies on derivatives conducted in the early 1990s by individual central banks or by specific groups have certainly contributed to this progressive change of attitude.12

Two factors in particular have encouraged this development:

  • On the asset side, the futures markets have rapidly come to be used as a cheaper way than cash securities trading to manage the duration of portfolios (this is particularly evident for the Bund market).

  • From early on, those central banks responsible for managing liabilities have, like other borrowers, made extensive use of interest rate and currency swaps for issuing debt in the most appropriate currencies or markets, and adjusting the risk characteristics of their portfolio.

Further development in the use of derivatives has gone hand in hand with progress achieved in the level of expertise of both staff and management, the quality of risk controls, and the development of IT systems to monitor and process derivatives transactions just as it has with the majority of institutional investors.

At present, central banks are mostly using straightforward, linear products such as futures, forwards, short, and medium-term currency and interest rate swaps (often referred to as “first generation” derivatives). Sophisticated structured products, which were extensively marketed to central banks in the early 1990s, have seen a declining interest owing to their lack of liquidity and transparency—and perhaps also to the increased capacity of central banks to adjust their risk profiles themselves by using common (plain vanilla) derivatives. This said, the major leap forward in the use of derivatives has occurred in the options market: the features specific to options (asymmetry and nonlinearity) make them a genuinely different class of product that would allow the implementation of new investment strategies and perhaps further lead to a more radical overhaul of investment policies. On the other hand, due to these characteristics, options are more difficult to price, process, and monitor, and the risk assessment is more complex.

At present, few central banks are using options for intervention purposes.13 However, there is a growing number using them for portfolio management purposes, such as foreign exchange options, interest rate, or bond options and even in some cases “exotic options” (such as barrier options) but in limited amounts. This development will certainly continue although the ever-increasing complexity of derivatives markets clearly acts as a brake to their use in central bank reserves management.

Factors Liable to Slow These Trends

The recent developments in reserves management outlined above, and in particular the increased focus on the wealth-accumulation function, may slow down in the coming years. Three significant elements could indeed stand against or in some limited cases even reverse this evolution: the impossibility for a large number of central banks to manage gold as a day-to-day asset; the increase in the mobility of international capital flows; and the difficulties of operating market activities in public sector institutions.

Central Banks Will Have Little Opportunity to Manage Gold Reserves Actively With a View to Maximizing Return

Gold over recent years has not represented a profitable investment for central banks. Figure 10.7 shows the excess returns obtainable by investing in one-year treasury bills rather than gold.14 The annualized return over the entire period from 1978 to 1995 is 8.6 percent for the one-year bills, 4.84 percent for gold (price returns), and 5.87 percent for gold (including price and lease rate terms with a 1 percent annual lease rate). Not only has the gold price deteriorated significantly, from its heady $800 per ounce highs to below $400, but the yields available to those institutions who lend their gold into the limited deposit market have been relatively poor in comparison to other high-grade investment outlets. The more recent decline in the price of gold, traditionally held by investors as an inflation hedge, can be partly explained by the relatively high level of real interest rates and indeed the general fall in inflation expectations. The unresolved debate about possible IMF sales, the continuing forward sales by mining companies, and the uncertain ramifications of EMU have also recently been weighing on the market.

Figure 10.7.Excess Returns of U.S. Twelve-Month LIBID Over Gold

(In percent)

Source: Author’s calculations.

Nevertheless, many major central banks still see gold as a guarantee of central bank credibility and a symbol of stability. It is also regarded as the ultimate prudential asset, a source of liquidity of last resort, and as such there has been a certain reluctance to mobilize gold holdings. A rise in gold deposit yields in the mid-1990s, a byproduct of forward sales by gold producers, has pushed a substantial number of central banks to place some of their gold assets on the lease market (whose size is estimated at between 2,500 tons to 3,000 tons). However, the possibility of any part of the remaining 32,500 tons of official gold coming onto the market could be enough to put a cap on gold yields.

Increase in the Mobility of International Capital Flows

The world economy is becoming increasingly more integrated as convertibility gains ground in emerging market economies. This has been evidenced by an increase in capital flows between countries, especially from industrialized to emerging market economies. For example, between 1991 and 1995, the average annual capital inflows to Asian and Latin American countries was estimated at $69.88 billion. This was much higher than averages for previous years, which were in the order of about $12.4 billion a year.15

As a result, any country, whether industrialized or developing, could experience large and sometimes abrupt capital inflows and outflows. In such a global market, every objective or policy regarding reserves management is superseded by the pressing demand for liquidity.

As for emerging market economies facing large inflows of capital, particularly portfolio inflows, extra care is needed in managing reserves and maintaining substantial buffer stocks to react quickly. The Mexican crisis of 1994 offers a stern warning in this respect: most of the capital inflows, which had represented on average 6.8 percent of GDP from 1990 to 1994, left the country in less than a month (December 1994), prompting a dramatic depreciation. This had a spillover effect in some Latin American and Asian countries (the so-called “tequila effect”) and triggered substantial capital outflows that required some of these countries to intervene heavily.16

Countries maintaining a fixed or semi-fixed exchange rate target should be aware that the size of international markets and the mobility of capital is such that massive interventions could be needed over a short period to resist speculative attacks. The 1992 ERM crisis in this respect is a particularly telling example. According to a G-10 report prepared in 1993, some DM 284 billion (nearly 20 percent of worldwide foreign exchange reserves at that time) were sold in the markets by European central banks (including those of the Nordic countries) between June and December without preventing a partial collapse of the system.17

The case for countries maintaining a floating exchange rate target is not much better. While it is true that such countries may not face a speculative attack as traditionally defined, there is still the possibility that an exchange rate may suffer large swings—unwelcome to the central bank. The latter may want to dampen this volatility, particularly to prevent the exchange rate from reaching certain “trigger” points. This may require substantial foreign exchange interventions as well. The magnitude of international capital flows imposes on central banks the need to maintain, and even in some cases to tighten, their liquidity criteria.

Difficulty of Operating Market Activities in Public Sector Institutions

Operating market activities in public sector institutions such as central banks poses specific problems. Some of these are related to the nature of the decision-making process and human resources management within central banks. Other problems arise owing to the difficulties faced by such institutions in making the necessary investments required to keep up with the rapid development of financial markets.

As mentioned earlier, the management of foreign exchange reserves is a sensitive issue for central banks, since market participants scrutinize and attach great importance to any central bank market activity and might therefore misinterpret reserves management transactions.18 However, foreign exchange reserves management may sometimes seem a subsidiary activity for central banks in comparison with monetary policy issues, which are its primary responsibility. Moreover, the prevailing culture in central banks may be far less profit-oriented and money-driven than in the private sector.

In many central banks, there may be a conflict between their specific needs for recruiting specialized and experienced portfolio managers and their general hiring policy aimed at recruiting generalists. Moreover, as public institutions, central banks cannot compete effectively with the salaries offered by private institutions for premium talent. Central banks can end up losing staff to the private sector, which can be disruptive to their small portfolio management teams. It needs to be emphasized that central banks need professional staff who can manage reserves well but their role is different from that of a private sector fund manager. The qualities needed by the central bank reserve manager are not necessarily the same as those in the private sector.

The cost of technical infrastructures and human resources that would be incurred in keeping pace with the market has increased tremendously in recent times. Indeed, markets are more efficient, technology has been changing rapidly, and the markets are dominated by large players that can make substantial investments in technology. It is difficult for central banks to make similar investments in technology owing to institutional constraints, a lack of customers to whom they can pass on the costs, less clear risk/return goals, and the instability of foreign exchange reserves. For this reason, central banks run the risk that many midsized institutions face—of falling behind in terms of technology.


This paper has discussed many of the issues regarding foreign exchange reserves and their relationship to a changing world economy. In particular, it has been noted that the recent growth in world reserves has been more rapid than in the past, though the currency composition of reserves has remained relatively stable. Some of the recent changes in central bank management of reserves have also been documented. Although liquidity is still the central issue for foreign exchange management, return maximization is becoming a more important concern. The future evolution of foreign exchange reserves may prove to be much more complicated. In the final part of this paper, a few of the potential complications were touched upon, such as the management of gold reserves, the increased mobility of capital, and the difficulties of managing market activities in public institutions. However, there may be yet another: the creation of the euro. What will be the evolution of a new European currency with respect to other world currencies? Will it bring a halt to the strong growth in foreign exchange reserves—taking into consideration the fact that a European central bank might need less foreign exchange reserves for the purpose of trade or current account stabilization? To what extent will the euro challenge the dollar as the dominant world currency?

Despite these questions, it is hard to imagine that the considerable development of the management of foreign exchange reserves observed in recent years could substantially reverse its course. However, managing reserves will never be identical to private fund management and will continue to reflect for a Gavin objectives and constraints of individual countries.


    Bank for International Settlements1996“Central Bank Survey of Foreign Exchange and Derivatives Market Activity 1995” (Basle: Bank for International Settlements, Monetary and Economic Dept.).

    Bank for International Settlements1996BIS Annual Report (Basle: Bank for International Settlements).

    GavinMichael and others1996“The Macroeconomics of Capital Flows to Latin America: Experience and Policy Issues,” in Banking Crises in Latin Americaedited byRicardoHausmann and LilianaRojas-Suárez (Washington: Inter-American Development Bank; distributed by Johns Hopkins University Press).

    GoldsteinMorris and PhilipTurner1996“Banking Crisis in Emerging Economies: Origins and Policy Options,”BIS Economic Paper No. 46 (Basle: Bank for International Settlements, Monetary and Economics Dept.).

    Group of Ten Group of Deputies1993“International Capital Movements and Foreign Exchange Markets: A Report to the Ministers and Governors” (Rome).

    Hong Kong Monetary Authorityvarious issuesQuarterly Bulletin (Hong Kong: Hong Kong Monetary Authority).

    HoriiAkinari1986“The Evolution of Reserve Currency Diversification,”BIS Economic Paper No. 18 (Basle: Bank for International Settlements, Monetary and Economic Dept.).

    NugéeJohn“Use of External Managers in Central Bank Reserves Management,”Quarterly Bulletin (Hong Kong: Hong Kong Monetary Authority).

    RogerScott1993“The Management of Foreign Exchange Reserves,”BIS Economic Paper No. 38 (Basle: Bank for International Settlements, Monetary and Economic Dept.).

See Scott Roger, “The Management of Foreign Exchange Reserves,” and “Overview of Reserve Management by Central Banks,” a paper presented in Florence in 1994 by Jean-François Rigaudy and in Cartagena by Pierre Cardon.

See Goldstein and Turner, “Banking Crises in Emerging Economies: Origins and Policy Options.”

The average percentage of reserves held by industrialized countries between 1985 and 1994 was 57 percent.

See BIS, “Central Bank Survey of Foreign Exchange and Derivatives Market Activity 1995.”

See Scone Roger, “The Management of Foreign Exchange Reserves.”

See Akinari Horii, “The Evolution of Reserve Currency Diversification.”

There is, of course, wide disparity between the liquidity of different securities.

Some of the central hank desks have recruited former central hank reserves managers.

In the proposals of the Basle Committee on Banking Supervision to amend the 1988 Capital Accord to incorporate market risk, regulators have accepted that hanks could use their internal models of VAR to measure and report their market risk, subject to a number of carefully defined conditions.

On this topic, see John Nugée, “Use of External Managers in Central Bank Reserves Management.”

Usually, a potential candidate for managing a portfolio on behalf of a central bank has to provide a complete file detailing the track record, the customer base, the size of fund under management, the management philosophy, the fee structure, the qualifications of the staff, the turnover in the staff, the staff compensation system, the projected lime spent on the central bank account, the projected time spent on research, the training provided, the performance objectives, the decision-making process, and so on.

In this respect, the report of the G-10 working group chaired by Mr. Hannoun, Deputy Governor of the Banque de France, which addressed macroeconomic and monetary policy issues raised by derivatives markets, played a significant role.

In practice, one of the most powerful arguments against this idea is that the signal sent to the market by a central bank intervening in options is less clear to market participants than the one sent through a direct intervention in the spot market (connected to a clear exchange rate level).

The returns for gold are based on yearly price changes and do not include revenues from leasing gold.

See BIS Annual Report, June 1996.

On this point, some economists have suggested that temporary controls or taxes on capital inflows may in some circumstances provide helpful protection against international financial turmoil (based on the experience of Chile and Colombia). See Michael Gavin and others, “The Macroeconomics of Capital Flows to Latin America: Experience and Policy Issues,” in Hausmann and Rojas-Suárez.

See “International Capital Movements and Foreign Exchange Markets.”

This is even more important since in market economies, credibility and reputation have become the main instruments of central bank monetary policy.

    Other Resources Citing This Publication