THE ANALYSIS OF THE ROLE OF RESERVES in the monetary system has generally focused on the demand for reserves. That is, the decisions of individual countries to hold various types of reserve assets and changes in these preferences are seen as affecting prices and economic activity in the world economy. One well-known conclusion of this type of research is that the demand for reserves has apparently not changed significantly in recent years despite very important changes in exchange rate regimes, international capital markets, and a number of other institutional developments.1 The apparent stability in demand for reserves might suggest that the role of reserves in the system has also not changed greatly.
The argument presented in this paper, however, is that, although the demand for reserves may appear to be little changed, the role of reserves in the current system is fundamentally different because the systemic constraints on the supply of reserves has changed. By concentrating on the demand for reserves, analysts have implicitly assumed that the supply of reserves is determined independently of economic policies of individual countries. As a result, governments’ reactions to an exogenous change in the supply of reserves in part determines the course of important economic variables such as employment and prices. This assumption may have been appropriate under past international monetary arrangements but continued reliance on this assumption can be seriously misleading.
An alternative framework is based on the assumption that each country faces a schedule that determines the terms under which reserves are supplied to that country. In most cases the relevant supply conditions are the terms on which the country can obtain credit in international financial markets. Thus, reserve holdings are, for most countries, seen as one aspect of an increasingly sophisticated strategy of financial management. This framework also suggests that reserves, per se, play a much more important role in the policies of countries that are effectively excluded from participating in private financial markets. For unconstrained countries, the cost of holding reserves can be compared to the spread between the marginal cost of obtaining credit and the rate of return earned on their reserve assets. For countries with limited or no access to credit markets the cost of obtaining reserves is related to the requirement that their net receipts from goods and services traded with the rest of the world be altered.
This way of looking at the problem introduces a number of interesting questions about the provision and composition of reserve holdings in the current system. A general review of recent developments in the pattern of reserve holdings reinforces the view that the apparent relationships between reserve holdings and other important economic variables have continued to conform to historical experience. However, there are alternative interpretations of these developments. While it may have been useful in the past to interpret reserve holdings as being demand determined, this is not a useful interpretation in the current system. The view that reserve holdings are, for most countries, an element in a more general financial strategy is developed further in the final section of this paper in the context of the management of the currency composition of reserve assets.
Changes in Holdings of Nongold Reserves
Although holdings of reserves—defined here as nongold reserves—grew at an average annual rate of 16 percent during the period 1970–81, they have exhibited a strong cyclical pattern in recent years (Table 1). After a 2 percent decline in 1982 (which was the first such decline since 1959) many countries attempted to rebuild their reserve holdings during 1983 and 1984. As a result, reserves for all countries rose from SDR 334 billion at the end of 1982 to SDR 404 billion at the end of November 1984 (an annual rate of increase of 10 percent). Between December 1982 and November 1984, holdings of industrial countries grew by 11 percent per annum and those of developing countries expanded at 8 percent per annum, with holdings of the oil exporting countries falling by 1 percent per annum and those of the non-oil developing countries increasing at an annual rate of 16 percent.
Non-Gold Reserves and SDR Allocations and Holdings, End of Years 1970–83 and End of November 1984
(In billions of SDRs)
The ratios for cumulative SDR allocations and holdings of SDRs are calculated by using imports and trade imbalances for all countries.
Non-Gold Reserves and SDR Allocations and Holdings, End of Years 1970–83 and End of November 1984
(In billions of SDRs)
Non-Gold Reserves | |||||||
---|---|---|---|---|---|---|---|
Year | All countries | Industrial countries | All developing countries | Oil exporting countries | Non-oil developing countries | Cumulative SDR Allocations | Holdings of SDRs1 |
1970 | 56.5 | 38.9 | 16.7 | 3.7 | 13.0 | 3.4 | 3.1 |
1971 | 88.1 | 65.7 | 21.3 | 6.5 | 14.8 | 6.4 | 5.9 |
1972 | 112.3 | 79.7 | 31.1 | 8.8 | 22.3 | 9.3 | 8.7 |
1973 | 118.4 | 77.8 | 39.2 | 10.8 | 28.4 | 9.3 | 8.8 |
1974 | 145.8 | 78.3 | 68.2 | 39.1 | 29.1 | 9.3 | 8.9 |
1975 | 160.7 | 83.6 | 75.6 | 47.0 | 28.6 | 9.3 | 8.8 |
1976 | 188.7 | 92.7 | 94.1 | 54.8 | 39.3 | 9.3 | 8.7 |
1977 | 231.1 | 118.9 | 109.9 | 60.8 | 49.1 | 9.3 | 8.1 |
1978 | 248.2 | 143.1 | 102.8 | 44.8 | 58.0 | 9.3 | 8.1 |
1979 | 274.8 | 153.2 | 119.7 | 54.9 | 64.8 | 13.3 | 12.5 |
1980 | 327.1 | 184.3 | 135.5 | 67.9 | 67.6 | 17.4 | 11.8 |
1981 | 341.4 | 185.1 | 145.1 | 73.8 | 71.3 | 21.4 | 16.4 |
1982 | 334.2 | 184.4 | 138.5 | 68.1 | 70.4 | 21.4 | 17.7 |
1983 | 369.4 | 204.9 | 146.8 | 67.1 | 79.1 | 21.4 | 14.4 |
Nov. 1984 | 404.3 | 224.0 | 160.3 | 67.3 | 93.0 | 21.4 | 16.2 |
The ratios for cumulative SDR allocations and holdings of SDRs are calculated by using imports and trade imbalances for all countries.
Non-Gold Reserves and SDR Allocations and Holdings, End of Years 1970–83 and End of November 1984
(In billions of SDRs)
Non-Gold Reserves | |||||||
---|---|---|---|---|---|---|---|
Year | All countries | Industrial countries | All developing countries | Oil exporting countries | Non-oil developing countries | Cumulative SDR Allocations | Holdings of SDRs1 |
1970 | 56.5 | 38.9 | 16.7 | 3.7 | 13.0 | 3.4 | 3.1 |
1971 | 88.1 | 65.7 | 21.3 | 6.5 | 14.8 | 6.4 | 5.9 |
1972 | 112.3 | 79.7 | 31.1 | 8.8 | 22.3 | 9.3 | 8.7 |
1973 | 118.4 | 77.8 | 39.2 | 10.8 | 28.4 | 9.3 | 8.8 |
1974 | 145.8 | 78.3 | 68.2 | 39.1 | 29.1 | 9.3 | 8.9 |
1975 | 160.7 | 83.6 | 75.6 | 47.0 | 28.6 | 9.3 | 8.8 |
1976 | 188.7 | 92.7 | 94.1 | 54.8 | 39.3 | 9.3 | 8.7 |
1977 | 231.1 | 118.9 | 109.9 | 60.8 | 49.1 | 9.3 | 8.1 |
1978 | 248.2 | 143.1 | 102.8 | 44.8 | 58.0 | 9.3 | 8.1 |
1979 | 274.8 | 153.2 | 119.7 | 54.9 | 64.8 | 13.3 | 12.5 |
1980 | 327.1 | 184.3 | 135.5 | 67.9 | 67.6 | 17.4 | 11.8 |
1981 | 341.4 | 185.1 | 145.1 | 73.8 | 71.3 | 21.4 | 16.4 |
1982 | 334.2 | 184.4 | 138.5 | 68.1 | 70.4 | 21.4 | 17.7 |
1983 | 369.4 | 204.9 | 146.8 | 67.1 | 79.1 | 21.4 | 14.4 |
Nov. 1984 | 404.3 | 224.0 | 160.3 | 67.3 | 93.0 | 21.4 | 16.2 |
The ratios for cumulative SDR allocations and holdings of SDRs are calculated by using imports and trade imbalances for all countries.
It is generally accepted that countries hold reserves to cushion changing needs for foreign exchange of their private and public sectors. Various ratios can show how these needs have shifted over the past few years. The accumulation of reserves has been reflected in only modest increases in the ratios of reserves to imports, for example (Chart 1). For all countries, this ratio rose from 20 percent at the end of 1981 to slightly over 21 percent at the end of 1983. Based on the latest estimates in the World Economic Outlook (WEO) of the International Monetary Fund for the growth in the SDR value of world imports, and assuming that reserve holdings grew at the same monthly rate during December 1984 as they did on average during the first 11 months of 1984, this ratio attained a value of 20.8 percent at the end of 1984. The stability of the overall ratio of reserves to imports in the presence of relatively rapid reserve accumulation implies that total reserve holdings have generally continued to expand in line with world imports.

Ratio of Non-Gold Reserves to Imports
(In percent)
Note: The annual rate of imports in the fourth quarter is the divisor of the stock of reserves at year’s end.1The ratios for 1984 are estimates based on the assumption that non-gold reserves grew during December 1984 at the same average monthly growth rate as observed during the first 11 months of 1984. Imports are taken as growing at the rates projected in the Fund’s World Economic Outlook.
Ratio of Non-Gold Reserves to Imports
(In percent)
Note: The annual rate of imports in the fourth quarter is the divisor of the stock of reserves at year’s end.1The ratios for 1984 are estimates based on the assumption that non-gold reserves grew during December 1984 at the same average monthly growth rate as observed during the first 11 months of 1984. Imports are taken as growing at the rates projected in the Fund’s World Economic Outlook.Ratio of Non-Gold Reserves to Imports
(In percent)
Note: The annual rate of imports in the fourth quarter is the divisor of the stock of reserves at year’s end.1The ratios for 1984 are estimates based on the assumption that non-gold reserves grew during December 1984 at the same average monthly growth rate as observed during the first 11 months of 1984. Imports are taken as growing at the rates projected in the Fund’s World Economic Outlook.These movements in the overall ratio of reserves to imports have encompassed diverse movements in the ratios for the major country groups. The ratio for the industrial countries has remained in the range of 16-17 percent throughout the period 1981–84. For the developing countries, the ratio first declined from 30 to 27 percent in 1980–82 but then recovered to almost 30 percent at the end of 1984, while the ratio for the oil exporting countries declined from 61 to 46 percent during the period 1980–82 before recovering to 51 percent in 1983 and 1984. The non-oil developing countries were able to increase the ratio of their aggregate reserves to imports from 19 percent in 1981 to nearly 23 percent at the end of 1984.
In attempting to evaluate the evolution of reserve holdings, the ratio of reserves to imports can be supplemented by other measures. One such measure compares reserve holdings with trade imbalances. If trade imbalances were to expand and be more variable, a country would typically want larger holdings of reserves to help maintain stable exchange market conditions. The ratio of reserves to aggregate trade imbalances for all countries did not increase in the period 1981–83; and, at-the end of 1983, this ratio was only marginally above the value observed in 1980, which was the lowest value attained in the period since 1974 (Chart 2).

Ratio of Non-Gold Reserves to Trade Imbalances
(In percent)
Note: Trade imbalances equal the sum of the absolute values of differences between exports and imports for individual countries in each country group.1 The ratio could not be calculated for 1984 due to a lack of data for the exports and imports of individual countries.
Ratio of Non-Gold Reserves to Trade Imbalances
(In percent)
Note: Trade imbalances equal the sum of the absolute values of differences between exports and imports for individual countries in each country group.1 The ratio could not be calculated for 1984 due to a lack of data for the exports and imports of individual countries.Ratio of Non-Gold Reserves to Trade Imbalances
(In percent)
Note: Trade imbalances equal the sum of the absolute values of differences between exports and imports for individual countries in each country group.1 The ratio could not be calculated for 1984 due to a lack of data for the exports and imports of individual countries.Although imports and trade imbalances provide measures of trade-related transactions, they do not indicate the size of international financial transactions. Since data on gross financial transactions are difficult to obtain for most countries, reserve holdings have been compared with selected measures of the stock of external debt on the assumption that gross flows are roughly proportionate to the existing stock of liabilities. One such measure is the ratio of reserves to external indebtedness to banks (Table 2). For all developing countries and the group of non-oil developing countries, this ratio has remained stable since the first half of 1982, although it rose for the non-oil developing countries in Asia, and (slightly) for those in Europe and the Western Hemisphere.
Ratio of Non-Gold Reserves to External Debt to Banks, First Half of 1981 to First Half of 1984
(In percent)
The following offshore centers are included in the first group but excluded from the second: Africa—Liberia; Asia—Hong Kong and Singapore; Middle East—Bahrain and Lebanon; and Western Hemisphere—Bahamas, Cayman Islands, Netherlands Antilles, and Panama.
Ratio of Non-Gold Reserves to External Debt to Banks, First Half of 1981 to First Half of 1984
(In percent)
1981 | 1982 | 1983 | 1984 | ||||||
---|---|---|---|---|---|---|---|---|---|
2nd half | 1st half | 2nd half | 1st half | 2nd half | 1st half | ||||
Including main offshore centers1 | |||||||||
All countries | 17 | 15 | 15 | 15 | 15 | 15 | |||
Developing countries | 21 | 19 | 18 | 18 | 18 | 18 | |||
Non-oil developing countries | 13 | 11 | 11 | 12 | 12 | 13 | |||
Of which: | |||||||||
Africa | 12 | 11 | 10 | 9 | 11 | 10 | |||
Asia | 22 | 21 | 23 | 25 | 25 | 26 | |||
Europe | 13 | 10 | 10 | 10 | 12 | 12 | |||
Middle East | 20 | 17 | 18 | 17 | 15 | 14 | |||
Western Hemisphere | 8 | 6 | 5 | 5 | 5 | 6 | |||
Argentina | 14 | 13 | 11 | 14 | 5 | 9 | |||
Brazil | 11 | 10 | 6 | 5 | 6 | 10 | |||
Mexico | 5 | 2 | 1 | 3 | 5 | 7 | |||
Excluding main offshore centers1 | |||||||||
All countries | 20 | 17 | 17 | 17 | 17 | 17 | |||
Developing countries | 37 | 32 | 30 | 29 | 29 | 30 | |||
Non-oil developing countries | 22 | 19 | 18 | 19 | 19 | 22 | |||
Of which: | |||||||||
Africa | 16 | 14 | 12 | 12 | 13 | 13 | |||
Asia | 49 | 49 | 52 | 58 | 60 | 65 | |||
Europe | 13 | 10 | 10 | 10 | 12 | 12 | |||
Middle East | 36 | 32 | 29 | 29 | 27 | 27 | |||
Western Hemisphere | 15 | 11 | 9 | 9 | 8 | 11 |
The following offshore centers are included in the first group but excluded from the second: Africa—Liberia; Asia—Hong Kong and Singapore; Middle East—Bahrain and Lebanon; and Western Hemisphere—Bahamas, Cayman Islands, Netherlands Antilles, and Panama.
Ratio of Non-Gold Reserves to External Debt to Banks, First Half of 1981 to First Half of 1984
(In percent)
1981 | 1982 | 1983 | 1984 | ||||||
---|---|---|---|---|---|---|---|---|---|
2nd half | 1st half | 2nd half | 1st half | 2nd half | 1st half | ||||
Including main offshore centers1 | |||||||||
All countries | 17 | 15 | 15 | 15 | 15 | 15 | |||
Developing countries | 21 | 19 | 18 | 18 | 18 | 18 | |||
Non-oil developing countries | 13 | 11 | 11 | 12 | 12 | 13 | |||
Of which: | |||||||||
Africa | 12 | 11 | 10 | 9 | 11 | 10 | |||
Asia | 22 | 21 | 23 | 25 | 25 | 26 | |||
Europe | 13 | 10 | 10 | 10 | 12 | 12 | |||
Middle East | 20 | 17 | 18 | 17 | 15 | 14 | |||
Western Hemisphere | 8 | 6 | 5 | 5 | 5 | 6 | |||
Argentina | 14 | 13 | 11 | 14 | 5 | 9 | |||
Brazil | 11 | 10 | 6 | 5 | 6 | 10 | |||
Mexico | 5 | 2 | 1 | 3 | 5 | 7 | |||
Excluding main offshore centers1 | |||||||||
All countries | 20 | 17 | 17 | 17 | 17 | 17 | |||
Developing countries | 37 | 32 | 30 | 29 | 29 | 30 | |||
Non-oil developing countries | 22 | 19 | 18 | 19 | 19 | 22 | |||
Of which: | |||||||||
Africa | 16 | 14 | 12 | 12 | 13 | 13 | |||
Asia | 49 | 49 | 52 | 58 | 60 | 65 | |||
Europe | 13 | 10 | 10 | 10 | 12 | 12 | |||
Middle East | 36 | 32 | 29 | 29 | 27 | 27 | |||
Western Hemisphere | 15 | 11 | 9 | 9 | 8 | 11 |
The following offshore centers are included in the first group but excluded from the second: Africa—Liberia; Asia—Hong Kong and Singapore; Middle East—Bahrain and Lebanon; and Western Hemisphere—Bahamas, Cayman Islands, Netherlands Antilles, and Panama.
These recent movements in reserve holdings are only one aspect of the evolution of international liquidity. An equally important factor has been the curtailment of the access of many countries to international financial markets during 1981 and 1982, which has sharply reduced the current and future availability of borrowed reserves. Moreover, it appears that access to markets may be restored only slowly for most of these countries. The reduced availability of borrowed reserves has meant that a number of countries have been able to increase their reserve holdings only through policies that generate current account surpluses.
The Historical Role of Reserves
The view that the evolution of reserve holdings is exogenous to the system and that the supply of reserves is an important determinant of economic policies is most appropriate in the context of a classical gold standard. Theoretically, in that system the physical stock of the reserve asset—gold, or some good substitute—is determined by mining technology while its nominal value is determined by fixing the price of gold in terms of national currencies. In issuing national currencies, governments had to take into account their obligation to convert currencies into gold at the fixed price. An excessive issue of currency, for example, led to a drain on reserves if market participants came to doubt that the convertibility obligation could be maintained. In such a system, the growth in the nominal value of the supply of reserves relative to the growth in demand for them is a central issue. If real economic activity grew more rapidly than the real gold stock or its substitutes, demand would have been greater than supply, leading to deflationary pressures.
In practice, the link between domestic monetary policies and the global supply of reserves was probably never as direct as suggested by this theoretical model. Indeed, the gold exchange standard as it existed in the Bretton Woods System can be seen as a gradual evolution away from the strict subservience of economic policies to changes in gold holdings. In particular, governments resorted to a variety of credit arrangements both among themselves and with private credit markets in order to soften the link between their economic policies and the constraint on these policies provided by the need to maintain convertibility.
However, the ability of governments to substitute credit, generally in the form of financial assets denominated in reserve currencies, for gold reserves and other assets that are close substitutes for gold, was limited. For example, a shift in preference of reserve holders toward gold and away from currency reserves required a general contraction of the monetary liabilities issued by authorities of individual countries. The preferences of reserve holders and the composition of reserve assets therefore had an important bearing on how well the system performed. In particular, the cost to the system of a sudden change in the composition of reserve holdings could be very large if the associated changes in domestic monetary policies were disruptive.
The gradual move away from convertibility was completed by the early 1970s. In this environment, authorities increasingly came to rely on other nominal anchors in conducting policy. Targets for monetary and credit aggregates as well as objectives for the growth of nominal spending have emerged as the dominant constraint on the policies of individual countries. A key feature of the current system is that there is no exogenous stock of reserves that provides a systemic coordinating constraint on setting these targets. In this regard, the surveillance of the Fund over the policies of individual countries has become the means by which international considerations are brought to bear on the conduct of policy in individual member countries.
If the nominal stock of reserves is no longer a constraint on monetary policies of countries, what role does it play in the current system? It can be analyzed in a framework that considers both demand and supply conditions. Turning briefly to the demand for reserves, this demand, usually related to so many months’ imports, is analogous to an individual’s demand for liquid assets. In addition, for some countries it might also be useful to identify changes in the demand for reserve assets that are associated with changes in the governments’ desired net foreign currency position. These changes need not involve shifts in reserve assets because a number of other financial positions can be manipulated in order to attain an equivalent position. As argued later, this is an important element in understanding the currency composition of reserve assets.
In its most streamlined version, the traditional approach would involve adding up the real demands for reserve assets of individual countries and comparing this to the exogenously determined global nominal stock of reserve assets. It is clear that only one world price level would be consistent with these two conditions. In the approach proposed here we can still imagine a determinate real demand for reserve assets. However, nominal prices in each country are determined by a decentralized set of monetary policy decisions of individual authorities. Moreover, the terms on which governments, as any other asset holder, can acquire liquid assets are also determined by the stance of policy in each country. For example, the terms on which a government can borrow dollars and hold liquid dollar balances are determined by conditions in dollar credit markets and by that country’s credit standing. For countries with strong credit ratings, the carrying costs of reserves can be measured by the loan deposit spread in international credit markets. Within wide ranges for most countries, reserve holdings can be adjusted without noticeably affecting this carrying cost.
Thus, we have an almost entirely different assumption concerning supply conditions for reserves. Rather than a fixed supply of nominal reserves, the more appropriate assumption is a very elastic supply of real reserve assets relative to carrying costs, with nominal values of both determined entirely outside the reserve creating and holding process. This change in the environment is bound to reduce and perhaps eliminate the value of analyses that rest on assumptions about causal links between broad measures of reserve holdings, either individually or globally, to other economic variables.
For countries with limited or no access to international credit markets, terms on which reserves can be acquired are different to those of creditworthy countries. For the less creditworthy, reserves can be accumulated only through current account surpluses or through increases in official credit. Clearly the types of change in economic policies required to alter the current account balance are in a different league compared with the decision to draw on a line of credit at a commercial bank. Moreover, a change in the terms on which credit can be obtained will have an important effect on these countries’ behavior.
The importance of the current account and official aid to less creditworthy countries underlines the significance of allocations to these countries. It cannot be doubted that for these countries an extension of official credit, for example, in the form of SDRs would be important to their adjustment efforts. However, in the event these SDRs were eventually accumulated by governments in countries with good credit ratings, it is difficult to predict how their behavior would be affected. The hypothesis that, even in the long run, the right to borrow represented by the SDR position would lead to more expansionary or inflationary policies seems unlikely to be useful. It could be argued, to the contrary, that given targets for monetary aggregates in the industrial countries, the increased demand for goods and services (if any) from debtor countries financed by SDR sales could, by increasing real output in industrial countries, generate downward pressure on prices. This illustrates that the reserve system remains important but in ways that are entirely unrelated to those suggested by the traditional analytic framework.
Currency Composition of Reserve Assets
The potential changes in the value of the dollar relative to other currencies that are also used as reserves has been suggested as a cause for concern in the current system. In particular, shifts in official preferences among reserves denominated in different currencies could be a source of problems comparable to shifts in preferences among gold and credit instruments in the Bretton Woods System. The analogy may be misleading. It is probably true that unstable policies in reserve currency countries could lead to defensive portfolio adjustments by both official asset holders and others. But it seems unlikely that these adjustments have had powerful effects on exchange rates. In the old system, a shift from foreign exchange reserves to gold was important since it constrained monetary policies in reserve currency countries. It is possible that concern about currency composition is an inappropriate extension of this argument.
Table 3 shows the shares of national currencies in total reserves in recent years. The decline in the share of dollar reserve assets, from about 77 percent in 1976 to 67 percent in 1980, may have reflected the desire of authorities to diversify their reserve holdings. However, this trend was partly reversed after 1980. There are several difficulties in interpreting these data. First, as noted in the Fund’s Annual Report for 1984, changes in these shares are, in part, due to valuation effects as the relative exchange values of currencies varied over a wide range during the period considered. Moreover, it may be appropriate to view the authorities’ preferences for reserve assets denominated in different currencies in the context of a more complex financial balance sheet.
Share of National Currencies in Total Identified Official Holdings of Foreign Exchange, End of Selected Years, 1976–83
(In percent)
Share of National Currencies in Total Identified Official Holdings of Foreign Exchange, End of Selected Years, 1976–83
(In percent)
1976 | 1977 | 1978 | 1979 | 1980 | 1981 | 1982 | 1983 | |
---|---|---|---|---|---|---|---|---|
U.S. dollars | 76.5 | 77.9 | 75.6 | 72.8 | 66.7 | 69.4 | 68.5 | 69.1 |
Pound sterling | 1.8 | 1.7 | 1.7 | 2.0 | 3.0 | 2.2 | 2.5 | 2.6 |
Deutsche mark | 9.0 | 9.2 | 11.0 | 12.6 | 15.1 | 13.2 | 12.5 | 11.9 |
French franc | 1.6 | 1.3 | 1.2 | 1.4 | 1.7 | 1.4 | 1.4 | 1.2 |
Swiss franc | 2.3 | 2.4 | 2.3 | 2.7 | 3.2 | 2.8 | 2.7 | 2.4 |
Netherlands guilder | 0.9 | 0.9 | 0.9 | 1.1 | 1.3 | 1.2 | 1.0 | 0.8 |
Japanese yen | 2.0 | 2.3 | 3.2 | 3.5 | 4.2 | 4.1 | 4.2 | 4.2 |
Unspecified | 5.9 | 4.3 | 4.2 | 4.0 | 4.8 | 5.7 | 7.2 | 7.8 |
100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 |
Share of National Currencies in Total Identified Official Holdings of Foreign Exchange, End of Selected Years, 1976–83
(In percent)
1976 | 1977 | 1978 | 1979 | 1980 | 1981 | 1982 | 1983 | |
---|---|---|---|---|---|---|---|---|
U.S. dollars | 76.5 | 77.9 | 75.6 | 72.8 | 66.7 | 69.4 | 68.5 | 69.1 |
Pound sterling | 1.8 | 1.7 | 1.7 | 2.0 | 3.0 | 2.2 | 2.5 | 2.6 |
Deutsche mark | 9.0 | 9.2 | 11.0 | 12.6 | 15.1 | 13.2 | 12.5 | 11.9 |
French franc | 1.6 | 1.3 | 1.2 | 1.4 | 1.7 | 1.4 | 1.4 | 1.2 |
Swiss franc | 2.3 | 2.4 | 2.3 | 2.7 | 3.2 | 2.8 | 2.7 | 2.4 |
Netherlands guilder | 0.9 | 0.9 | 0.9 | 1.1 | 1.3 | 1.2 | 1.0 | 0.8 |
Japanese yen | 2.0 | 2.3 | 3.2 | 3.5 | 4.2 | 4.1 | 4.2 | 4.2 |
Unspecified | 5.9 | 4.3 | 4.2 | 4.0 | 4.8 | 5.7 | 7.2 | 7.8 |
100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 | 100.0 |
For most countries the initial changes in exchange rates among the reserve currency countries are not limited to the effects on the value of their reserve holdings. The rapid growth in the use of international credit markets implies that the currency composition of government liabilities is also important. A recent paper on this subject provides an example of the importance of looking at both assets and liabilities in evaluating the currency preferences of 94 developing countries.2 Table 4 shows the results of this effort. The top panel in Table 4 shows the evolution of the currency composition of these countries’ reserve assets from 1974 to 1979. These data roughly conform to the experience of all countries for those years discussed above. A remarkable feature in the changes in the share of dollar-denominated reserves in the total is that it closely mirrors changes in the dollar’s exchange value as measured by the SDR-dollar rate shown in the last column of Table 4. That is, if we held exchange rates at their 1974 levels and recalculated the shares shown in Table 4, there would be little change in the dollar share of the portfolio. Such a result would seem to suggest that this large group of countries, on balance, passively accepted changes in the share of assets denominated in different currencies resulting from changes in exchange rates.
Currency Preferences of 93 Developing Countries
(End of period)
Currency Preferences of 93 Developing Countries
(End of period)
Total | Percent Denominated in: | SDR/$ | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Year | (In millions of dollars) | US$ | DM | STG | FF | JY | Other | (1974 = 100) | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Foreign Exchange Reserves | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1974 | 67,677 | 67 | 10 | 11 | 2 | 0 | 11 | 100.0 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1975 | 72,439 | 70 | 9 | 7 | 2 | 1 | 11 | 104.5 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1976 | 89,925 | 72 | 10 | 3 | 2 | 0 | 12 | 105.4 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1977 | 107,412 | 68 | 13 | 3 | 2 | 0 | 14 | 100.1 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1978 | 106,550 | 62 | 15 | 3 | 2 | 1 | 17 | 94.0 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1979 | 128,415 | 62 | 16 | 3 | 2 | 4 | 12 | 92.9 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
Currency Preferences of 93 Developing Countries
(End of period)
Total | Percent Denominated in: | SDR/$ | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
Year | (In millions of dollars) | US$ | DM | STG | FF | JY | Other | (1974 = 100) | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Foreign Exchange Reserves | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1974 | 67,677 | 67 | 10 | 11 | 2 | 0 | 11 | 100.0 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1975 | 72,439 | 70 | 9 | 7 | 2 | 1 | 11 | 104.5 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1976 | 89,925 | 72 | 10 | 3 | 2 | 0 | 12 | 105.4 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1977 | 107,412 | 68 | 13 | 3 | 2 | 0 | 14 | 100.1 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1978 | 106,550 | 62 | 15 | 3 | 2 | 1 | 17 | 94.0 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
1979 | 128,415 | 62 | 16 | 3 | 2 | 4 | 12 | 92.9 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
A possible explanation is that reserve assets are held for transactions and precautionary reasons, and the currency composition of such assets is dominated by the consideration that they can be easily liquidated and used to make payments. Adjusting the currency mix of transactions balances following an exchange rate change might not be immediately called for, particularly if nominal prices and payment patterns do not change along with exchange rates. Thus, a country with a given pattern of payments in, for example, U.S. dollars and deutsche mark might expect little change in the nominal level of those payments in each currency following a change in the dollar-mark exchange rate.
The lower panel of Table 4 shows estimates of size and currency composition of external debt for developing countries. The data show that external debt positions were larger and grew more rapidly as compared to reserve assets throughout the six-year period. Moreover, exchange rate changes did not dominate changes in the share of liabilities denominated in different currencies. In particular, the 10 percent fall in the dollar’s exchange value between 1976 and 1978 was associated with a 1 percent rise in the share of dollar-denominated liabilities for this group of countries.
The very different appearance of the information on the currency composition of reserve assets and external debt in Table 4 serves as a reminder that data on either assets or liabilities alone are likely to give a misleading picture of changes in overall foreign currency positions.
Conclusions
Under current international monetary arrangements, the supply of international reserves is best described as a schedule that relates the amount of reserves supplied by financial markets to different market conditions. Countries tend to adjust their reserve holdings until the benefits derived from such holdings are equal, at the margin, to the net cost of holding reserves. Under normal conditions, the marginal cost of obtaining and holding additional reserves in private financial markets is relatively constant. This implies that the demand for reserves by individual countries is generally accommodated with little or no change in overall economic conditions.
As with the level of reserve holdings, the currency composition of reserves can be viewed as the result of countries’ efforts to balance the usefulness of assets denominated in various currencies against the expected relative yields available on various assets. This implies that the currency composition of reserve assets might not closely reflect the overall currency preferences of individual countries.
Comment
Willem H. Buiter
Dooley has written a characteristically lucid paper on a subject matter for which clarity has a significant scarcity premium: the changing role and significance of international reserves.
The exchange rate systems he compares are an idealized gold standard (a system with an exogenously given quantity of international reserves) and a fixed exchange rate system without convertibility into an “outside” reserve asset. While the paper is not clear on the matter, I assume this means a system comprising a few reserve currency countries and a large number of small non-reserve currency countries. Some of the latter can borrow in the international financial markets to replenish their reserves while others are credit constrained.
The paper has very little to say about the “microfoundations” of the demand for and supply of international reserves and for the purposes of this discussion I also merely assume the existence of a reasonably stable demand function for international reserves as well as the usual well-behaved demand functions for domestic money.
Four main points are made by Dooley. First, for the authority of an individual country, international creditworthiness implies the ability to borrow international reserves in private financial markets. The supply schedule of international reserves to an individual creditworthy country is highly elastic; for simplicity it can be viewed as perfectly elastic, or horizontal. The opportunity cost of international reserves is measured by the spread between the marginal cost of obtaining external credit and the rate of return earned on the reserve assets.
The reason why Dooley’s assumption that reserve holdings can be varied without any effect on the opportunity cost of reserves represents such a departure from the ideal gold standard scenario is that, apparently, reserve holdings can be varied without altering domestic monetary policy. Domestic credit does not have to be reduced in order to attract more reserves or, equivalently, the discount rate does not have to be increased. Another way of stating this is that full sterilization is, for all practical purposes, possible up to the limit of the government’s ability to borrow externally to finance reserve losses. Furthermore such borrowing can proceed at a constant marginal cost over a very wide range. A net worth constraint has been substituted for a liquidity constraint; liquidity does not become an issue until solvency is threatened. The solvency limit is presumably given by the present value of the nation’s future expected trade balance surpluses or the fraction of these surpluses that the authority can appropriate and transfer to its creditors. Governments can borrow more today for any purpose (including defending their reserves) provided they can credibly commit themselves to a future trade balance surplus strategy that permits them to service and amortize the increased debt.
Dooley’s second point is that for a country that is not rationed in the international credit markets, the magnitude and currency composition of its reserve assets are a poor guide to that country’s demand for international transactions and precautionary balances, its currency preferences, and its overall foreign currency position. Its access to the international credit markets (and the rapid growth in the use of these markets) means that the currency composition of the authority’s foreign liabilities is as relevant as that of its assets in assessing the strength or weakness of the external financial position. Simply netting out liabilities in a given currency against assets may well yield a picture that is less misleading than the consideration of just the asset side of the balance sheet. To the extent that assets and liabilities denominated in a given currency are yet different in some important respects (such as maturity, whether exchange rates are fixed or floating, or their degree of sovereign risk) the “gross” balance sheet should be taken as the object of study. I have no disagreement with Dooley on this point.
Dooley’s next point is that an individual country without international creditworthiness will be rationed and unable to borrow in private international financial markets. Reserves can be acquired only by running current account surpluses. The cost of acquiring additional reserves today is then measured by the cost of import compression today or of shifting resources toward the production of exportables today. Note the difference with the net-worth-constrained authority discussed under the first point. This credit-constrained authority cannot borrow against the collateral or security of its planned or expected future trade balance surpluses. Resources only become available as they are earned currently. The returns earned from holding the reserves must of course still be credited against the cost of generating them. This cost, in the case of a rationed country, is not measured by the world interest rate (at which it cannot borrow) but by the shadow price of foreign resources, which exceeds the interest rate.
Finally, Dooley points out that under an international monetary standard with an exogenously given global quantity of the reserve asset (such as a pure gold standard with exogenous gold production and exogenous non-reserve demand for gold) and strict convertibility of national currencies into gold at a fixed price, the world economy has a clear nominal anchor. The current system without convertibility into any “outside” asset (whether commodity or fiduciary) has no such obvious nominal anchor. Dooley proceeds to say (with italics added) that “rather than a fixed supply of nominal reserves, the more appropriate assumption is a very elastic supply of real reserve assets relative to carrying costs, with nominal values of both determined entirely outside the reserve creating and holding process. This change in the environment is bound to reduce and perhaps eliminate the value of analyses that rest in assumptions about causal links between broad measures of reserve holdings, either individually or globally to other economic variables.”
Consider first the case of a single individual non-reserve currency country wishing to maintain a fixed exchange rate with a reserve currency. For simplicity, assume that strict purchasing power parity holds. With a fixed exchange rate, domestic and foreign price levels must move together. Without loss of generality, take the world price level to be constant. If domestic credit expansion exceeds the growth of the demand for money, the country will lose reserves. With a fixed exchange rate and constant price levels, nominal and real reserve losses are (given the appropriate choice of units) identical. A creditworthy country can afford to lose reserves for longer than a credit-constrained country, but there is a finite upper limit on the real value of the amount they can borrow, and any given real balance of payments deficit will exhaust this sooner or later, with or without a speculative attack. Monetary policy for a single small country is therefore still constrained by the need to avoid exceeding the credit limit defined by the net worth constraint. The link is less tight than for a country that cannot borrow reserves, but it is still there. Dooley apparently overstates his case here, but his insight is still valuable.
How about the system as a whole? The notion that a fixed exchange rate system without convertibility into some exogenously given asset has no nominal anchor, other than the monetary growth targets of the combined reserve currency countries, is a familiar one. The need for any real quantity of reserves by the non-reserve countries can be met with any rate of monetary growth in the reserve currency centers (ignoring fine points about absence of superneutrality). With monetary growth in the non-reserve currency countries adjusting to changes in the money growth rates of the reserve currency centers, and with the world rate of inflation following suit, the same real reserve equilibrium can be established at any world rate of inflation. The same will hold if the non-reserve currency countries are credit constrained. What drives the result is the absence of compulsory convertibility at fixed exchange rates into an “outside” asset whose quantity is exogenous. Dooley’s analysis doesn’t seem to add anything to the familiar picture as far as the operation of the system as a whole is concerned.
One area of disagreement with the author is on the effect of increased SDR allocations to developing countries on inflation in the industrial world. If the SDR issue were to take the form of a net increase in the global stock of outside financial assets, global demand would increase. Qualitatively the same result would follow if the SDR issue were to take the form of a transfer from countries with a low marginal propensity to spend (the richer countries) to those with a high marginal propensity to spend (the developing countries). How this would lead to downward pressure on industrial country prices, as Dooley argues it does, is not clear to me. Even if monetary targets are unaffected by the SDR issue, response to the issue could come through velocity, through prices, or through real output. On the familiar IS-LM model of aggregate supply and demand, at a given price level, an increase in financial wealth will shift the IS curve to the right and, if there is a financial wealth effect on money demand, the LM curve to the left. The interest rate would increase and the effect on the effective demand level of output (and therefore on velocity) is ambiguous. If the effective demand level of output increases, the price level will rise if the aggregate supply schedule is upward sloping or vertical. Unless there is a shift of the aggregate supply schedule to the right in response to an increase in SDR allocations, I cannot generate negative association between output and the price level. It would seem to make more sense to represent the SDR allocation as working solely through the demand side of the IS-LM model. If velocity is a behavioral constant, there will be full (100 percent) crowding out and neither output nor the price level will change. Note that any effect on prices from a once-off increase in SDR allocations will take the form of an increase in the price level—that is, the rate of inflation (or deflation) will change only temporarily.
Mr. Dooley works in the Research Department of the International Monetary Fund.
For example, see Jacob Frenkel, “International Liquidity and Monetary Control,” in International Money and Credit: The Policy Roles, edited by George M. von Furstenberg (Washington: International Monetary Fund, 1983).
“An Analysis of the Management of the Currency Composition of Reserve Assets and External Liabilities of Developing Countries,” Michael P. Dooley, 1983, in The Reconstruction of the International Financial Arrangements, edited by Robert Z. Aliber (London: McMillan, forthcoming).