An Analysis of Factors Influencing the Level of SDR Holdings in Non-Oil Developing Countries
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

During the period 1973-78, no special drawing right (SDR) allocations were made. This period may therefore afford us an opportunity to study how countries adjust their SDR holdings in the absence of supply shifts caused by new allocations. Non-oil developing countries are particularly useful to study, because most of them have been consistent net users of SDRs and have been able to initiate transfers of SDRs, while net acquirers or the designated recipients of SDRs have been more passive. After we show that non-oil developing countries generally did not reduce their SDR holdings to the lowest possible levels permitted under reconstitution limits, we identify economic factors that may influence the level of SDR holdings in relation to cumulative allocations.

Abstract

During the period 1973-78, no special drawing right (SDR) allocations were made. This period may therefore afford us an opportunity to study how countries adjust their SDR holdings in the absence of supply shifts caused by new allocations. Non-oil developing countries are particularly useful to study, because most of them have been consistent net users of SDRs and have been able to initiate transfers of SDRs, while net acquirers or the designated recipients of SDRs have been more passive. After we show that non-oil developing countries generally did not reduce their SDR holdings to the lowest possible levels permitted under reconstitution limits, we identify economic factors that may influence the level of SDR holdings in relation to cumulative allocations.

I. Summary

During the period 1973-78, no special drawing right (SDR) allocations were made. This period may therefore afford us an opportunity to study how countries adjust their SDR holdings in the absence of supply shifts caused by new allocations. Non-oil developing countries are particularly useful to study, because most of them have been consistent net users of SDRs and have been able to initiate transfers of SDRs, while net acquirers or the designated recipients of SDRs have been more passive. After we show that non-oil developing countries generally did not reduce their SDR holdings to the lowest possible levels permitted under reconstitution limits, we identify economic factors that may influence the level of SDR holdings in relation to cumulative allocations.

We found possible indication that a rise in the SDR interest rate relative to the combined market rate used in the SDR interest rate basket encouraged a greater degree of retention of SDRs by non-oil developing countries as a group toward the end of the past decade. Since the SDR interest rate was fixed at approximately 60 per cent of the combined market rate from the middle of 1974 through 1978, the influence of other factors could be examined separately over this period through the pooling of annual data for 27 major non-oil developing countries. This analysis showed that, on the one hand, higher debt burdens, adjusted for differences in the normal rate of economic growth of countries, do not have a statistically significant effect on SDR holdings expressed as percentages of allocations. On the other hand, the positive effect of the ratio of non-gold reserves to imports is statistically significant in most regressions, with SDR holdings tending to rise somewhat less than in proportion to reserves.

The upshot is that to the extent SDR holdings expressed as percentages of allocations were influenced by economic, and not just legal or administrative, factors, portfolio-balance and cost considerations were the only systematic influences on such holdings that could be identified in this paper for the period 1974-78. Since the uses of SDRs have continued to be deregulated since that time and since the interest rate on the SDR has been raised to the full combined market rate, the economic determinants of how many SDRs non-oil developing countries choose to hold in relation to their cumulative allocations are likely to become more distinct and important than we have shown in this first effort to quantify their role.

II. Introduction

Since the start of the fourth basic period in January 1982 is only six months away, decisions will soon be made concerning the amount of special drawing rights (SDRs) to be allocated during the period. In deciding upon the allocation of SDRs, the International Monetary Fund is directed by its charter to “seek to meet the long-term global need … to supplement existing reserve assets in such manner as will promote the attainment of its purposes and will avoid economic stagnation and deflation as well as excess demand and inflation in the world” (Article XVIII, Section 1(a)). It is helpful to explore the pattern of the net use that has been made of past allocations to identify some of the economic factors that may influence the use of allocations increasingly in the future. In this paper, we examine the factors that have played a role in SDR transactions initiated by non-oil developing countries during the years 1974-78. Because no SDRs were allocated after January 1972 and until the beginning of 1979, the latter part of this period affords us a unique opportunity to explore the behavioral and administrative factors that have influenced the desired distribution of SDR holdings in relation to past cumulative allocations and to other reserve assets.

All members of the Fund are eligible to participate in the SDR Department and to receive allocations; and as of August 1980, all members were participants. SDR allocations are organized around successive periods of five years, or of differing length if the Fund so decides, that are known as basic periods. There have been three basic periods to date, with the fourth scheduled to begin on January 1, 1982. 1 A country whose holdings of SDRs differ from the level of its cumulative allocation 2 is said to have either made net use, or experienced net acquisition, of SDRs, depending on whether its holdings are, respectively, less than or greater than its cumulative allocation. The Fund assesses charges on cumulative allocations and pays interest at the same rate on SDR holdings. Hence, net charges are assessed when net use has been made, and net interest is paid on net acquisition. 3

During the 11 years since the first allocation of SDRs, the rules and requirements of the Fund for use and acquisition have undergone numerous and significant changes. Particularly since the Second Amendment to the Articles of Agreement (which became effective on April 1, 1978), many transactions and operations are no longer specifically regulated by the Fund, although they must conform to its general purposes as required in Article XIX, Section 2 (d). Yet the options of net users and net acquirers of SDRs have remained somewhat asymmetric, as participating countries in strong reserve positions may be designated to receive SDRs up to an acceptance limit on holdings equal to 200 per cent in excess of their cumulative allocations, unless a higher limit is agreed between a participant and the Fund. Countries with balance of payments financing needs or weak reserve positions can make net use of SDRs at their own initiative, while countries with holdings at or above cumulative allocations generally acquire additional SDRs when designated by the Fund to do so. Countries that made net use of SDRs could replenish their holdings by purchasing SDRs from the Fund’s General Resources Account for the purpose of reconstitution 4 or via transactions by agreement with other participants, whereas surplus countries with holdings above cumulative allocations and continuing strength in their balance of payments and gross reserve position could only lower their holdings through transactions by agreement. 5 Thus, countries making net use of SDRs had a greater degree of control over the level of their holdings than countries making net acquisitions. 6 Consequently, our study focuses primarily on that group of members, namely, the non-oil developing countries, that have been consistent net users of SDRs.

To investigate the behavioral factors that determine the use made of given cumulative allocations by particular countries, it is convenient to choose a period during which no SDR allocations were made. Although no allocations were made after the start of 1972 until the beginning of 1979, our analysis is confined to the period extending from the fourth quarter of 1974 to the third quarter of 1978 for several reasons:

  • (1) Some time was required, after the last allocation in the first basic period, for the ratio of SDR holdings to allocations to approach the changing (“moving stock equilibrium”) level ultimately desired in the absence of new allocations. Later on, that ratio began to be influenced, in the fourth quarter of 1978, by anticipations of the 1979 allocation. As shown in Chart 1, the SDR holdings of non-oil developing countries tend to be reduced in the months before scheduled allocations are made, as well as in the months immediately following allocations. Such reductions continue until stock equilibrium is restored.

  • (2) Export proceeds of non-oil developing countries were extremely variable during the commodity price boom-and-bust cycle of 1972-74,7 so that omitting this period is advisable to ensure that the period analyzed is more stable.

  • (3) No change was made in reconstitution requirements, which first applied to the five-year period that ended at the start of 1975, until the start of 1979.

  • (4) The interest rate on SDR holdings was equal to about 60 per cent of the combined rate on a basket of five money-market instruments 8 beginning on July 1, 1974, and was much lower earlier, so that a consistent method of determining the relative rates of return and cost prevailed during the period 1975-78. 9

Chart 1.
Chart 1.

SDR Holdings of Non-Oil Developing Countries, January 1970-May 1980 1

Citation: IMF Staff Papers 1981, 002; 10.5089/9781451946871.024.A003

1 Allocations of SDRs were made in January of the following years: 1970, 1971, 1972, 1979, and 1980.

The paper is organized as follows: Section III analyzes general trends in SDR transactions, concentrating on the influence of administrative factors in determining the timing and types of transactions engaged in. This is of interest primarily as background material for the more formal analysis of behavioral factors in the next two sections. Section III shows that the SDR holdings of non-oil developing countries have, in general, not been reduced to the maximum extent permitted under Fund regulations; financial management, and not just administrative limits, determines the size of these holdings. In Section IV, the decisions of non-oil developing countries to either hold or use SDRs are investigated in this light by considering their debt burdens and the interest that could have been earned on alternative reserve assets. In Section V, these debt and portfolio considerations are incorporated into a single-equation model that attempts to explain SDR holdings expressed as percentages of countries’ allocations. The conclusions are presented in Section VI.

III. General Trends in SDR Transactions

Over the period 1974-78, distinct patterns of transactions in SDRs emerged after the initial adjustment to the first round of allocations was completed. The SDR holdings of non-oil developing countries could generally be increased or decreased freely in response to economic, and not just to legal, factors if such holdings were well above reconstitution limits. While this was not the case for all individual countries, 10 it appears to have in been true for the group of non-oil developing countries as a whole. As Chart 1 shows, SDR holdings of non-oil developing countries expressed as percentages of their allocations were substantially above the 30 per cent level throughout the past decade. Thus, both economic and institutional factors have influenced the different uses of SDRs that are detailed in subsequent paragraphs.

The two categories of transactions undertaken through the SDR Department are transactions among participants and transactions with the General Resources Account (GRA) of the Fund. Only SDR transactions with the GRA can change the holdings of all members and the importance of SDRs in international reserves. Table 1 presents a breakdown of the two types of transactions for the financial years (FY) 1975-78 (the Fund’s financial year extends from May 1-April 30). Perhaps the most striking aspect of this period is the shift from negligible changes in the SDR holdings of the GRA in FY 1975 and FY 1976 to considerable net acquisition of SDRs by the GRA in FY 1977 and FY 1978. This shift was due to the upsurge in charges on Fund holdings of currencies associated with expanded use of the Fund’s resources over the years 1975-78, particularly those borrowed under the oil facilities established in 1974 and 1975, and to the increase, from approximately 2.85 per cent in April 1974 to an average of 5.94 per cent in FY 1976, in the average GRA charge on currency holdings in excess of quota that arose from countries’ use of Fund facilities. 11 Even though participants could acquire SDRs from the GRA for currencies acceptable to the Fund if they needed SDRs to pay any charge or assessment, such payments often involved a decline in national holdings of SDRs, particularly for the non-oil developing countries. Consequently, transfers of SDRs to the GRA should, in general, rise as indebtedness to the GRA increases.

Table 1.

All Participants: Transfers of Special Drawing Rights, 1975-78

(In millions of SDRs)

article image
Source: Annual Report 1980, Table 28, p. 92, and Table 1.16, p. 124.

The Fund’s financial year extends from May 1-April 30.

SDR 396 million was purchased and repurchased on the same day by the United Kingdom.

SDR 832 million was used in connection with the postponement of the discharge of repurchase obligations under Article V, Section 1(b), which was in effect prior to the Second Amendment, that had accrued and were (formerly) payable in gold, and in connection with the settlement of amounts accrued and payable in SDRs. See Annual Report 1978, pp. 62, 68.

In FY 1978, up to 25 per cent of quota increases determined in the Sixth General Review of Quotas could, at the member’s option, be paid in SDRs.

Includes assessments and interest received on General Resources Account holdings.

Including amounts acquired as part of purchases.

Includes remuneration, interest, and transfer charges on Fund borrowings from members.

Transactions Between Participants

The two forms of transactions between participants are transactions with designation 12 and transactions by agreement. Transactions with designation, which at first were the main SDR transaction used for balance of payments financing, fell steadily, as a percentage of cumulative allocations, from FY 1975 to FY 1977. Transactions with designation declined in part because the SDR holdings of some of the user countries were approaching the lower limits set by reconstitution obligations and other less formal constraints. 13 Chart 2 shows SDR holdings expressed as percentages of cumulative allocations for two groups of countries making net use of SDRs in transactions with designation during the period FY 1975-FY 1977. 14 The holdings for non-oil developing countries stabilize at about 39 per cent of allocations, considerably above the level of about 20 per cent maintained by the group of industrial and oil exporting countries. This suggests that while both groups of countries were approaching zones of increasing resistance to further reductions in holdings, a constraint was felt at a much higher level of holdings by the non-oil developing countries than by other net user countries. Countries with a history of comparatively easy access to credit (industrial countries and oil exporters) and high turnaround potential appear to have been willing to reduce their SDR holdings further than non-oil developing countries to finance what they perceive as temporary balance of payments deficits. 15 In both groups of countries, administratively constrained outcomes, with SDR holdings remaining at their statutory minimums for extended periods, are rare.

Chart 2.
Chart 2.

SDR Holdings of Countries Making Net Use of SDRs in Transactions with Designation, FY 1975-FY 1977 1

Citation: IMF Staff Papers 1981, 002; 10.5089/9781451946871.024.A003

1 The Fund’s financial year extends from May 1-April 30.2 Argentina, Chile, Costa Rica, Dominican Republic, The Gambia, Israel, Jamaica, Kenya, Korea, Mauritius, Mexico, Peru, Sri Lanka, Sudan, Tanzania, Western Samoa, the People’s Democratic Republic of Yemen, Zaïre, Zambia.3 Australia, Indonesia, Italy, New Zealand.

The second form of transaction between participants is known as the transaction by agreement. Prior to August 1976, this type of transaction was allowed only if the user of SDRs had a balance of payments need. An Executive Board decision adopted in August 1976 16 broadened this form of transaction by waiving the requirement of balance of payments need in cases where the transaction brought holdings closer to net cumulative allocations for both parties to the agreement or promoted reconstitution. This decision led to a fivefold increase in transactions by agreement between FY 1976 and FY 1978. Because purchases of SDRs from the GRA were subject to specific limits and requirements but transactions by agreement were not, participants who had made net use of SDRs could, following this decision, obtain larger amounts of SDRs. For the most part, non-oil developing countries acquired SDRs in these transactions, while a few industrial countries supplied SDRs (see Table 2). Transactions by agreement thus helped move non-oil developing countries further into the range in which their SDR holdings could increase or decrease symmetrically in response to economic and financial considerations.

Table 2.

Transactions by Agreement, by Country Group, 1975-78

(In millions of SDRs)

article image
Source: Annual Report 1975, 1976, 1977, and 1978.

The Fund’s financial year extends from May 1-April 30.

The only transactions by agreement in FY 1975 and FY 1976 involved settlement of obligations between European countries that had arisen owing to intervention in foreign exchange markets under the European Monetary Cooperation Agreement.

Countries with years of transactions: Belgium (75, 76, 77), Canada (77, 78), Denmark (75, 76, 77, 78), Federal Republic of Germany (75, 76, 77, 78), Netherlands (75, 76, 77, 78).

SDR 155.4 million was received by the Federal Republic of Germany.

SDR 467.0 million was received by Italy.

Countries with years of transactions: Argentina (78), Australia (78), Burma (77,78), Central African Empire (78), Chad (78), Chile (77, 78), Costa Rica (77), El Salvador (77), Ghana (77, 78), Greece (77, 78), Haiti (78), Iceland (78), Indonesia (78), Israel (77, 78), Ivory Coast (77), Jamaica (77, 78), Kenya (78), Korea (77, 78), Lao People’s Democratic Republic (78), Madagascar (77), Mexico (77, 78), Morocco (78), Panama (78), Papua New Guinea (78), Peru (77, 78), Philippines (78), Senegal (77), South Africa (78), Sudan (77, 78), Tanzania (77, 78), Turkey (77, 78), Uganda (78), Yugoslavia (77, 78), Zaïre (78), Zambia (77).

Transactions with General Resources Account

As discussed previously, the use of SDRs to pay GRA charges and to make repurchases in the GRA bears a direct relationship to the volume of borrowing from the GRA. This implies that changes in quotas and in the rates of charges will have consequences for SDR transactions. In addition, up to 25 per cent of increases in quotas under the Sixth General Review could, at each member’s option, be paid in SDRs. 17 As noted in Table 1, SDR 201 million was used during FY 1978 to pay increases in subscriptions under the Sixth General Review of Quotas. 18

For non-oil developing countries, regression analysis of the relationship between borrowing from the GRA and the use of SDRs shows that when the use of Fund credit, measured as a percentage of quota, rises, SDR holdings, measured as percentages of net cumulative allocations, tend to decline. 19 For the quarterly periods between the fourth quarter of 1974 and the third quarter of 1978, the individual country results showed a significant negative relationship in 13 out of the 27 countries examined. Thus, the aggregate relationship between GRA use and SDR use noted earlier is also evident on an individual country basis. Since the use of Fund credit tends to increase when countries’ reserve positions worsen and their capital-market borrowing opportunities decrease, it would not be surprising for SDR holdings to fall along with other reserve assets; but the link between SDR use and the use of Fund credit appears to be particularly close.

IV. Behavioral Factors: Non-Oil Developing Countries

Unlike most of the industrial countries that have had relatively strong payments and reserve positions, the non-oil developing countries have generally met the need criterion required by the Fund for the use of SDRs in transactions with designation. Hence, these countries have had the option to use their allocations of SDRs and should, therefore, provide the best opportunity to view the behavioral factors influencing SDR transactions.

We begin by looking at the cost considerations of using SDRs versus other reserve assets. Next, a model is developed that attempts to explain holdings of SDRs measured as percentages of net cumulative allocations. Relationships between SDR holdings, other reserve assets, and official borrowings are considered. Finally, we specify and estimate the model using pooled time-series data for non-oil developing countries over the period 1974-77.

Cost Considerations

To appreciate the conflicting pressures on non-oil developing countries to both use and hold SDRs, it is instructive to consider the costs associated with changing the composition of reserves. The most important alternative to increased SDR use is a reduction in the foreign exchange component of international reserves. While non-oil developing countries’ foreign exchange holdings are generally borrowed, rather than earned through payments surpluses, 20 a change in their foreign exchange holdings need not affect the size of external borrowing. There is clearly a relation between the size of reserves and the supply of loans to non-oil developing countries. However, changes in the composition of reserves may not affect supply conditions. For instance, special banking relationships aside, a purchase of SDRs with foreign exchange that raises SDR holdings and reduces foreign exchange holdings so as to leave total reserves unchanged is unlikely to affect the external credit supply available to, or the demand for external credit on the part of, non-oil developing countries. Thus, foreign borrowing would not change as the composition of reserves was adjusted.

This example also shows that the distinction between borrowed and unborrowed reserves is immaterial in the absence of new allocations. In the previous example, SDR holdings, which are frequently identified as “unborrowed” reserves, increase while “borrowed” foreign exchange decreases, yet the size of external borrowing remains unchanged. With the cumulative amount of SDR allocations fixed, the rate of return earned by raising SDR holdings is given by the dollar rate equivalent to the Fund’s interest rate on the SDR while the cost of doing so is given by the dollar rate forgone on the official deposits of foreign exchange withdrawn from foreign banks. 21

These matters need to be explained in greater detail. The interest cost of holding borrowed foreign exchange is equal to the difference between the rate at which it is borrowed and the rate at which it can be deposited. Borrowing rates for non-oil developing countries are typically quoted at a margin (spread) above the London interbank offer rate (LIBOR). The margin varies from country to country—for example, in 1979, loans were made at margins as low as 56 basis points (Malaysia) and as high as 215 basis points (Pakistan). A foreign deposit rate that is indicative of the investment returns available to non-oil developing countries is the three-month London Eurodollar deposit rate. The act of holding any amount of SDRs allocated is essentially costless, because interest is earned on that amount at the same rate as that paid on cumulative allocations. 22 Hence, the cost advantage of holding SDRs allocated rather than foreign exchange in international reserves is equal to the difference between the borrowing and deposit rates at the margin at which the size of this spread is determined.

While this correctly describes the benefits of substituting SDR holdings obtained through allocations for borrowed foreign exchange when total reserves are fixed, there were no new SDR allocations during the period examined in this paper. Thus, any increase in SDR holdings had to be financed either by reducing foreign exchange holdings23 or by increasing the size of external borrowing, ceteris paribus. If the latter course were taken, the cost of raising SDR holdings through increased foreign borrowing should be compared with the cost of raising official holdings of foreign exchange through a like increase in borrowing. The net cost differential of choosing the former over the latter course of action would then be precisely the same as in the case where SDR holdings are substituted directly for foreign exchange holdings. Because the borrowing costs would be the same in either event, holding more borrowed SDRs (SDRs whose acquisition is financed through foreign borrowing) rather than borrowed foreign exchange has an expected cost equal to the difference between the expected dollar rate of interest on foreign deposits and that on the SDR. 24 Thus this last differential is the one relevant for determining the appropriate composition of non-gold reserves during periods when there are no allocations.

While the dollar rate of interest on official deposits in foreign banks is known for the term of the deposit, the dollar equivalent of any rate set by the Fund on the SDR is uncertain. This rate can be determined only upon realization unless SDRs are bought or sold forward for dollars, depending on whether an SDR payment or receipt, respectively, is involved. During the period examined here, such forward transactions were not permitted, so that an expected dollar rate on the SDR must be compared with a known dollar rate on foreign deposits. Reserve holders are thus forced to form expectations about the future U. S. dollar price of the SDR, and the degree of risk they attach to any such forecast, in deciding upon the composition of their reserves between SDR holdings and foreign exchange holdings.

Table 3 shows that it would have been unprofitable ex post to increase SDR holdings in 1975 and 1976, because the fall in the dollar price of the SDR made the dollar rate of interest on the SDR far lower than the dollar rate on official deposits of foreign exchange. The opposite occurred in 1977 and 1978, when the dollar price of the SDR rose appreciably. Perhaps in lagged response to these developments, the SDR holdings of non-oil developing countries, as shown in Chart 1, fell until 1977 and then rose. 25 Whether or not changes in the expected profitability of SDR holdings were the cause in this instance. 26 it is reasonable to assume that the extent of SDR use is not entirely independent of its relative cost. Rather, an increase in the interest rate on the SDR in relation to the combined rate used in the SDR interest rate basket or a pattern of increases in the dollar price of SDRs (that are not fully reflected in correspondingly higher U. S. dollar deposit rates) will eventually raise the SDR holdings of non-oil developing countries and lower SDR use in relation to a fixed total of cumulative allocations.

Table 3.

Non-Oil Developing Countries: Comparative Use and Holding Costs for Foreign Exchange and SDRs, 1975-78 1

(In per cent per annum)

article image

The country group classification system of the International Monetary Fund’s publication, International Financial Statistics, is used. The annual SDR rate is the average of the SDR rates prevailing in each quarter of the year. Each quarterly rate is determined on the basis of money-market rates prevailing in the middle 60 days of the previous quarter. Hence, this rate is matched with the average quarterly three-month Eurodollar rate in the previous quarter. The latter rate is compounded to obtain the annual rate in the second row of figures.

The following example illustrates the derivation. To calculate the equivalent dollar rate of return for 1978, convert $1.21471 into 1 SDR at the beginning of the year, let it earn 3.75 per cent interest in SDRs, and then convert SDR 1.0375 into $1.35164 at the end of the year to get (1.35164/1.21471) - 1 = 11.27 per cent.

Average quarterly rates for fourth-quarter-through-third-quarter periods were used with quarterly compounding. The average quarterly rates for three-month Eurodollar deposits are computed from daily quotations of broker bid rates at noon in London. The average quarterly series is from International Financial Statistics.

Debt Considerations

In the short run, many other factors can impinge on the SDR holdings of developing countries. For instance, countries with unusually high debt service burdens are likely to face the highest spreads on, or outright refusals to let them continue past rates of growth of, foreign borrowing. With foreign exchange reserves tied up as compensating balances or on deposit to safeguard banking relationships, SDR holdings may be reduced disproportionately to finance a balance of payments disequilibrium. The opposite expectation, that SDR holdings will rise to leverage borrowing capacity, is far less persuasive. Leverage is derived from SDR allocations to the extent they increase a country’s net worth. 27 An increase in SDR holdings in relation to a fixed total cumulative allocation confers no additional leverage opportunities, since net worth is unchanged. We therefore expect an inverse relation between SDR holdings, measured as percentages of allocations, and the debt service burden. Although debt service “burdens” are often identified with the size of foreign debt payments (amortization plus interest) in relation to exports, the ratio of interest payments to income or gross national product (GNP) provides a clearer indication of the relative size of the potential transfers out of income or product that may be deemed burdensome. 28 Before such burdens can be compared between countries, however, they must be calibrated by reference to differences in their rates of economic growth.

A country facing an initial debt-to-income ratio of D/Y, with an annual real interest rate r and a real GNP growth rate n, can maintain that ratio by having its debts outstanding grow by n per cent a year. If rn this means that 100 (n/r) per cent of the interest on the foreign debt can be paid by simply issuing more debt. If n > r, then not only can interest on the debt be paid by simply issuing more debt but additional debt must be issued if the debt-to-income ratio is not to fall. 29 If n and r were converted to nominal rates by adding the (expected) inflation rate to both, the same qualitative conclusion would hold. Obviously, therefore, a reduction in r relative to n reduces the degree to which the net expansion of debt must be used to cover interest charges before it is available for other purposes. Furthermore, if r/n declines because the economic growth rate n increases, the net expansion of international debt that is warranted to maintain a given debt-to-income ratio itself increases.

Alternatively, if countries wish to maintain the same cost of interest net of the growth dividend, (r-n)D, in relation to income, then, for any given r > n, the debt-to-income ratio is higher, the greater is n. This last result indicates that the sustainable debt-to-income ratio is higher, ceteris paribus, the greater the rate of economic growth. Consequently, any empirical assessment of the debt position of various countries requires consideration of differences in economic growth prospects.

V. The Model and Results

The model developed from the foregoing considerations focuses on portfolio composition and debt factors in attempting to explain SDR holdings, expressed as percentages of net cumulative allocations, by non-oil developing countries. Because it was shown that the yield on additional SDR holdings has at times been greater, and at other times smaller, than the yield on foreign exchange holdings, SDR holdings would be expected to be a normal part of any diversified reserve portfolio, with or without reconstitution limits. Such holdings are therefore expected to be positively related to the size of non-gold reserves. To allow pooling across countries, SDR holdings expressed as percentages of allocations, SDRP, are regressed on non-gold reserves excluding SDR holdings expressed as percentages of imports, R/M. 30

Additional regressors are derived from the interest cost of debt relative to income net of the growth dividend, (r-n)D/Y. The hypothesis is that the higher the adjusted debt burden of a country, the lower its SDR holdings are likely to be, because countries may attempt to cope with a financing crisis by drawing down their SDR holdings as a last resort, while holding an ample cushion of SDRs in reserve at other times. The real interest rate on external borrowing, r, is approximately equal to the size of the spread over LIBOR, since the real rate of interest represented by LIBOR was approximately zero on average over the period 1974-1978. 31 However, since representative spreads are not continuously available for many countries, r is left implicit in the regression coefficient of D/Y. Furthermore, the rate of economic growth, n, is entered separately rather than multiplicatively in our model to allow it to indicate that, given D/Y, the debt burden is lower and SDR holdings are higher, the higher is n. Although countries with higher n can sustain a higher debt income ratio, D/Y, than countries with lower n and still have the same debt burden adjusted for the growth dividend, the relation between n and D/Y is not consistently positive in the data. Rather, as Table 4 shows, the debt-to-gross domestic product (GDP) ratio first increases and then decreases as n increases across country groups. While this would make it feasible to enter D/Y and n(D/Y) jointly in the regression, leaving D/Y implicit in the regression coefficient of n allows the structural coefficient on n(D/Y) to vary to the extent D/Y varies among countries. Specifically, the higher is D/Y, the lower is the structural coefficient on n(D/Y) that is implied by a fixed regression coefficient on n. The hypothesis supporting coefficient variability is that the higher is D/Y, the less effect a given rate of n has in reducing the debt burden, since the feasibility of maintaining that growth rate is reduced by a higher D/Y. The final form of the estimating equation is therefore

Table 4.

Non-Oil Developing Countries: Average Annual Rate of Growth of Real GDP Compared with Average Debt-to-GDP Ratio, Averaged Across Countries

article image
Sources: International Monetary Fund, International Financial Statistics, various issues; and International Bank for Reconstruction and Development, World Debt Tables: External Public Debt of Developing Countries, various issues.

The four country groups are as follows:

  • (A) Brazil, Dominican Republic, Korea

  • (B) Colombia, Costa Rica, El Salvador, Kenya, Mexico, Morocco, Philippines, Tanzania, Turkey

  • (C) Greece, Honduras, India, Israel, Pakistan, Panama, Peru, Sri Lanka, Zambia

  • (D) Chile, Cyprus, Jamaica, Uganda, Uruguay, Zaïre.

Computed as (GDP77/GDP71)1/6 - 1.

Debt outstanding including undisbursed amounts, total all lenders, as measured by the International Bank for Reconstruction and Development. This data includes only debt of public entities and private debt guaranteed by official institutions in the borrowing country.

SDRP=a0+a1(D/Y)+a2n+a3(R/M)+u(1)

RESULTS

Table 5 presents regression results for equation (1) using data from 1974-77 pooled over 27 non-oil developing countries. Column (1) reports an ordinary least-squares (OLS) regression on the entire sample. 32 The coefficient on the reserve variable is significant and of the expected positive sign. Increasing non-gold reserves excluding SDRs by 1 per cent of imports yields an increase in SDR holdings equal to 0.6 per cent of cumulative allocations. 33 This relationship is the strongest and most consistent one observed. Apparently, the countries that we analyze attempt to maintain a measure of balance among the components of their reserve holdings. 34 The performances of the debt-to-income ratio and of the growth rate are somewhat disappointing, and only the negative sign on the former conforms to expectations.

Table 5.

Regression Results Using Data Pooled Over 27 Non-Oil Developing Countries1

article image

See the Appendix for a list of these countries and a description of data sources. The dependent variable is SDRP. Annual series are calculated by averaging three quarterly observations on end-of-period holdings of SDRs expressed as percentages of net cumulative allocations from the third quarter of one year to the first quarter of the next.

The observations were weighted by the square root of the reciprocal of cumulative allocations.

Standard errors of the coefficients are in parentheses. An asterisk denotes significance at the 5 per cent level.

End-of-year debt outstanding, including amounts undisbursed, expressed as a percentage of the flow of gross domestic product over the year. We use nominal gross domestic product instead of nominal national income because data for the latter are unavailable. Initially, we tried using the ratio of total debt service payments (amortization plus interest) to gross domestic product in place of the debt-to-GDP ratio and growth rate, but we were unable to obtain any significant effect on SDR holdings.

For the rate of economic growth, we use the geometric average annual rate of growth of real GDP for each country over the period 1971-77. Instead of calculating growth rates from data for the years in our sample, we use an average growth rate calculated over a longer period to represent long-term growth prospects.

Reserves expressed as a percentage of imports (R/M) is calculated from a quarterly series by averaging third-quarter-to-first-quarter observations so as to center the annual series at the end of the year. Each of the quarterly observations is defined as [end-of-quarter t reserve position in the Fund plus foreign exchange holdings]/[imports in quarter t plus imports in quarter (t + 1)] times 50.

For the weighted regression, these statistics are calculated using the untransformed data.

SEE denotes standard error of estimate.

Although neither the coefficient on the debt-to-income ratio nor the coefficient on the growth rate in regression (1) in Table 5 is significant, the former borders on significance at the 90 per cent level whereas the latter is very insignificant. Thus, the negative sign on the debt variable suggests some weak support for the hypothesis that when borrowing limits are approached, SDR holdings tend to be drawn down disproportionately, compared with other components of a country’s reserves, as a last resort.

In a cross-section/time-series model, error terms may well have variances that differ across countries. Hence, to test the efficiency of our results, we re-estimated the model using the generalized least-squares technique. We assumed experimentally that the error variances were positively related to the level of cumulative SDR allocations. Column (2) in Table 5 presents the results for this regression, in which all variables were weighted by the square root of the reciprocal of cumulative allocations. The same pattern of significance and signs emerges as in regression (1) with unweighted data. This tends to suggest that the error variances in the unweighted model using ratio variables are not systematically linked to country size.

To examine the stability of the model, we re-estimated equation (1) using only cross-sectional data for each year in our sample. These four regressions are also reported in Table 5. Inspection of the magnitude and signs of the coefficients shows what appear to be highly unstable estimates. For instance, the regression coefficient on RIM falls from a high of about 1 in 1974 and 1975 to a low of 0.3 in 1977. The earlier years conform most closely to the pooled regression over the entire period, whereas the later years have no explanatory power. An F-test revealed that these differences in the regression coefficients, though large, are not significant statistically.

A possible explanation for the failure of D/Y to have a statistically significant effect on SDR holdings is that the countries analyzed are too diverse in their borrowing abilities. To achieve greater homogeneity in this regard, we separated those countries that borrowed heavily in the Eurocurrency markets during the period 1974-77 from those that did not. 35 The regression reported in Table 6 for the former group of countries yields a significant negative sign on the debt variable, although again the growth rate is not significant and has the wrong sign. This indicates that countries borrowing heavily on international markets exhibit a more systematic relationship between borrowing limits and SDR holdings than our group of non-oil developing countries as a whole. In addition, the reserve variable is not significant. This suggests that the strength of the portfolio composition effect may vary greatly between countries as well as over time, although the direction of the effect of R/M on SDRP is the same in all regressions.

Table 6.

Ordinary Least-Squares Regression Using Data for 1974-77 Pooled Across Non-Oil Developing Countries With Access to the Euroloan Market1

article image

Dependent variable is SDRP.

Brazil, Chile, Colombia, Greece, Korea, Mexico, Morocco, Peru, Philippines, Turkey

Standard errors of the coefficients are in parentheses. An asterisk denotes significance at the 5 per cent level.

VI. Conclusions

In this paper, we have considered various behavioral influences on the SDR holdings of non-oil developing countries. This group of countries offers the best opportunity for analyzing SDR holdings because they exercise substantial discretion in initiating transactions to both use and acquire SDRs. By contrast, industrial countries and oil exporting countries usually have had less freedom of choice, since their generally strong reserve and payments positions preclude use of SDRs in transactions with designation and require passive acceptance of SDRs when these countries are designated by the Fund. 36 In addition, because a major purpose of the SDR is to supplement reserves and to smooth payments financing, it is of considerable interest to analyze countries with consistently weak payments positions.

Our study indicates that there may be an aggregate time-series relationship between cost factors and the use of SDR holdings. As the relative interest cost of using SDRs instead of foreign exchange has risen, the net use of SDRs has fallen. Should a causal interpretation of this association be warranted, it implies that, for non-oil developing countries, administrative increases in the SDR interest rate reduce net use of SDRs and result in increased holdings. Hence, calculation of the potential benefit accruing to net users under future allocations should include not only consideration of the expected relative interest charge on SDR use, but also consideration of the effect of changes in the relative interest charge on the extent of SDR use. 37

To test for the influence of portfolio and debt factors, we constructed a single-equation model of SDR holdings as percentages of allocations. While the regression results are hardly conclusive, they do suggest two points. First, under given cumulative allocations, determining how many SDRs to hold in non-gold reserves appears to be principally a problem of maintaining balance between reserve assets, which, while free of default risk, have risky valuations and returns. Thus, the SDR holdings of non-oil developing countries rise consistently with their other non-gold reserve assets, given the expected return on each. 38 Second, the hypothesis that countries with high debt burdens are running down SDR holdings until their external positions and borrowing opportunities improve is not well supported, and differences in rates of economic growth appear to have no statistically significant effect on SDR holdings. Hence, it may be useful to concentrate on the size of non-gold reserves in relation to imports and on the expected interest rate on SDR holdings compared with that on other reserve assets to gauge the demand for SDR holdings by particular countries or groups of countries that tend to be net users of SDRs. These economic factors are likely to be more important, compared with administrative factors, the less restricted the extent and purpose of net use.

APPENDIX

The regressions use data for all non-oil developing countries with net cumulative allocations of at least SDR 15 million during the period 1974-77 for which data are available.

The countries used in the regression analysis are:

article image

All variables are entered in percentage form. Calculation of the variables and their sources are as follows:

R/M = Annual series centered at the end of the year by averaging three annualized quarterly percentages calculated as 25 times (end-of-period reserve position in Fund plus foreign exchange)/(imports), where imports are the average of quarterly flows in the current and the following quarter. The annual series is an average of the percentages so derived for the (end of the) third quarter, the fourth quarter, and the following first quarter. Merchandise imports are expressed on a c. i. f. basis and converted from U. S. dollars to SDRs by use of the average quarterly SDR price of the dollar. All data are from various issues of the International Monetary Fund’s monthly publication, International Financial Statistics (IFS).

D/Y = End-of-year debt outstanding including undisbursed amounts, total all lenders (as measured by the International Bank for Reconstruction and Development), as a percentage of the flow of gross domestic product over the year. These data include only debt of public entities and private debt guaranteed by official institutions in the borrowing country. Debt data are taken from International Bank for Reconstruction and Development, World Debt Tables: External Public Debt of Developing Countries, Vols. 1, 2 (Washington, 1979).

n = Average annual percentage rate of growth of real GDP from 1971 to 1977, calculated as 100[GDP77/GDP71)1/6 - 1], with real GDP taken from IFS.

SDRP = Annual series calculated, using data from IFS, by averaging quarterly end-of-period holdings of SDRs, expressed as percentages of cumulative allocations, over the third and fourth quarters and the following first quarter.

*

Mr. Murphy, a summer intern in the Financial Studies Division of the Research Department when this paper was prepared, is a graduate of Williams College. He is currently pursuing graduate studies in economics at the Massachusetts Institute of Technology.

Mr. von Furstenberg, Chief of the Financial Studies Division of the Research Department, was formerly Professor of Economics at Indiana University and a Senior Staff Economist of the U. S. President’s Council of Economic Advisers.

1

The first basic period, January 1, 1970 to December 31, 1972, witnessed allocations in all three years, whereas there were no allocations during the second basic period, January 1, 1973 to December 31, 1977. The current (third) basic period, which extends from January 1, 1978 through December 31, 1981, witnessed no allocation in 1978, followed by allocations in 1979, 1980, and 1981.

2

The term “net cumulative allocation” is used in the Articles of Agreement of the IMF because allocations may be followed by cancellations. However, no cancellations have been undertaken or proposed to date.

3

Net charges and net interest are settled at the end of each financial year on April 30.

4

The reconstitution requirement states that average daily holdings, as a percentage of average daily cumulative allocations over five-year periods ending with successive quarters, not fall below a set percentage. Prior to 1979, the reconstitution limit was 30 per cent. From that time on, it was 15 per cent until it was eliminated in May 1981. Each quarter, the Fund calculated for every country the amount of future SDR holdings it needed to meet the reconstitution requirement for all current five-year reconstitution periods. When a country needed to obtain SDRs in order to meet the reconstitution requirement, it was permitted to purchase SDRs from the General Resources Account, either by means of a tranche drawing, subject to conditionality, or to obtain SDRs in exchange for another member’s currency specified by the Fund.

5

Prior to August 1976, transactions by agreement between participants normally required that the user in the transaction have a balance of payments need to obtain SDRs. After that date, the requirement of need was waived provided the transaction brought both parties’ holdings of SDRs closer to their cumulative allocations. Hence, before August 1976, it was even more difficult for net acquirers to alter the level of their holdings. Since the Second Amendment became effective in 1978, net acquirers have had substantially more discretion with respect to their SDR holdings because the requirement that transactions by agreement bring both parties’ holdings of SDRs closer to their cumulative allocations has been abolished.

Countries can also lower their SDR holdings by using them to pay charges on drawings and to make repurchases in the General Resources Account. However, this possibility is less likely to be available to net acquirers than net users, since the former tend to have strong balance of payments and reserve positions and are thus unlikely to use Fund resources.

6

Detailed information on the rules and requirements is contained in International Monetary Fund, Annual Report of the Executive Directors for the Financial Year Ended April 30, 1974 through 1980 (hereinafter referred to as Annual Report 19—); Special Drawing Account: Manual of Procedures (Washington, 1970); and Special Drawing Account: Manual of Procedures, Supplement No. 1: Principles and Procedures for Reconstitution (Washington, 1971).

7

From the third quarter of 1972 to the third quarter of 1974, the export unit value index increased 90 per cent for the group of non-oil developing countries versus only 54 per cent for the group of industrial countries. See International Monetary Fund, International Financial Statistics (IFS).

8

The SDR interest rate is a weighted average of the rates of interest on five money-market instruments in the five major industrial countries. The instruments and weights were

article image
The SDR interest rate was adjusted quarterly to reflect interest rates during the 60-day period ending on the fifteenth day of the month preceding the end of each quarter. It was rounded to the nearest one-quarter of 1 per cent of the rate determined by the formula.

9

The rate was 60 per cent of the combined rate (rounded to the nearest one-quarter of 1 per cent) from July 1, 1976 through December 31, 1978. The complex rate determination formula in effect from July 1, 1974 through June 30, 1976 yielded a rate equal, on average, to 56 per cent of the combined rate for the six quarters ending with the second quarter of 1976.

10

To investigate whether the SDR holdings of particular countries were close to the reconstitution limit of 30 per cent, end-of-month holdings, averaged over 60 months, were expressed as percentages of the cumulative allocations outstanding averaged over the same 60 months. If the resulting ratio of moving averages was below 32 per cent, current holdings were judged to be effectively constrained by the reconstitution limits, in that they could not be reduced substantially below 30 per cent of cumulative allocations without violating this limit. Specifically, the SDR holdings of 2 of the 27 countries identified in the Appendix, Pakistan and Uruguay, were never substantially above reconstitution limits during the period 1975-78; and the same was true for Chile in 1976-77, Costa Rica in 1975-77, Panama in 1975-76, Tanzania 1977-78, and Zaïre in 1976-78. Six other countries had SDR holdings close to reconstitution limits in 1 out of the 4 years, 1975-78. This still leaves a majority of the 27 countries with SDR holdings generally well above reconstitution limits throughout the entire period.

11

See Annual Report 1977, pp. 61-72. Since March 1977, SDRs can be included in the currency budget, making them available for sale to members drawing on the Fund, in order to reduce the Fund’s own holdings. See Annual Report 1977, pp. 56-57.

12

A transaction with designation is initiated by a participant in need of balance of payments financing. The Fund then designates one or more participants with strong gross reserve and balance of payments positions to accept the initiating member’s SDRs in exchange for a freely usable currency of the designee’s choosing. Prior to the Second Amendment, “currency convertible in fact” had to be supplied—meaning the U.S. dollar, the pound sterling, or the French franc—or currencies cross-convertible therewith.

13

Non-oil developing countries as a whole were net users of SDRs in transactions with designation during the period FY 1975 to FY 1978, while the group consisting of industrial and oil exporting countries combined were net acquirers. Individual countries in the latter group did, however, make net use of SDRs in transactions with designation.

14

Only countries making net use of SDRs in transactions with designation over the period FY 1975 through FY 1977 were chosen, since Executive Board Decision No. 5355-(77/36) G/S, adopted March 15, 1977 (see Annual Report 1977, pp. 105-106), greatly improved participants’ access to SDRs by weakening constraints on transactions with designation when holdings are at low levels. This decision permitted the Managing Director to propose that participants in the Special Drawing Account acquire SDRs, rather than currencies, as part of purchases from the GRA. Even if participants’ holdings of SDRs were low initially, they could thus obtain additional SDRs to spend.

15

Although Chart 2 shows SDR holdings, expressed as percentages of cumulative allocations, dipping well below the reconstitution requirement of 30 per cent, this does not necessarily mean that these countries’ holdings fell below (or even came close to) reconstitution limits. Because the requirement stated that average daily holdings, expressed as percentages of average daily net cumulative allocations over five-year periods ending with successive quarters, not fall below 30 per cent, time periods during which holdings were below 30 per cent could be averaged with periods during which holdings were above 30 per cent. The main point to be gained from Chart 2 is that non-oil developing countries making net use of SDRs in designated transactions held significantly more SDRs, expressed as percentages of their allocations, than those industrial and oil exporting countries also making net use through transactions with designation.

16

Executive Board Decision No. 5167-(76/120) G/S, adopted August 2, 1976. See Annual Report 1977, pp. 102-103.

17

Members of the Fund that are participants in the Special Drawing Rights Department (currently all members) must pay 25 per cent of the increase in quotas under the Seventh General Review in SDRs. These increases went into effect for most members in December 1980.

18

See Annual Report 1976, pp. 103-11.

19

The equation estimated had SDR holdings measured as percentages of allocations in period t as the dependent variable, with a constant term and Use of Fund Credit, measured as a percentage of quota in period t, as the explanatory variables. Use of Fund Credit—which is identified in the Fund’s publication, International Financial Statistics—measures the net use by a member of its conditional drawing rights in the Fund.

20

The official or officially-guaranteed debt owed by non-oil developing countries to international lenders is generally several times as large as their total holdings of non-gold reserves. Because repayment of external debt is thus always a possible use of foreign exchange, such reserves may be described as borrowed, although proceeds of individual foreign loans are fungible and are unlikely to be devoted to a single purpose, such as the acquisition of reserves.

21

Naturally, rates of return or cost could be expressed in any common currency other than the dollar, or in the SDR itself, to make them comparable. The dollar rate of interest on the SDR is the rate payable in SDRs by the Fund on SDR holdings where this rate is adjusted to reflect changes in the dollar price of the SDR. For instance, if interest is paid in SDRs at the rate of 5 per cent per annum on the SDR, but the dollar value of the SDR rises by 2 per cent during the year on account of dollar depreciation, the implied dollar-denominated rate of return is approximately 7 per cent per annum, as the exchange gain is added to the SDR interest rate.

22

If SDR holdings rise above cumulative allocations, interest is earned on the amount exceeding cumulative allocations. If SDR holdings are below cumulative allocations, charges are incurred at the same rate on the amount of net use.

23

A country that purchases SDRs with its own currency would experience no reduction in gross foreign exchange holdings, but an increase in liquid liabilities to official foreign agencies would worsen its net foreign exchange position.

24

This remains true in both cases if the additional borrowing takes the form of credit tranche drawings from the Fund rather than of increased borrowing in private capital markets. However, according to Article XIX, Section 3 (a), participants are not expected to use their special drawing rights “for the sole purpose of changing the composition of their reserves.

25

Since the share of U. S. dollars in the foreign exchange reserves of even the non-oil developing countries has been at least twice as high as the dollar’s share in the SDR basket, and since investment in the 16 currencies that constituted the SDR has been impractical, increasing investment in those currencies has not been a relevant alternative to increasing SDR holdings.

26

Among the countries included in the pooled cross-section study (see the Appendix), India received SDR 52.5 million by designation in the financial year ended April 30, 1978, yet its SDR holdings were lower at the end of April 1978 than one year earlier, though they were still far above reconstitution limits. Countries may therefore be designated to receive SDRs and still not raise their SDR holdings during a given accounting period. Conversely, any increase in SDR holdings achieved by designation would not necessarily be involuntary from the viewpoint of the country concerned.

27

The net worth obtained from SDR allocations derives from the entitlement to borrow at below-market rates through net use of SDRs that is conferred by these allocations. The capitalized value of this entitlement is bound to be only a small fraction of the amount allocated. The size of this fraction differs with alternative external borrowing costs and the needs of particular countries.

28

Amortization is a function of the maturity structure of the debt, and it is unclear why short debt is necessarily more burdensome than long debt. It is also unclear why the international debt burden of a country should rise when debt payments and GNP are fixed, as the production of import substitutes is increased and that of exports declines. In other words, the debt burden should not be measured in such a way as to be a function of the degree of autarky.

29

For a more detailed development of this argument in the context of steady growth, see George M. von Furstenberg, “The Long-Term Effects of Government Deficits on the U. S. Output Potential,” Journal of Finance, Vol. 33 (June 1978), pp. 989-1001.

30

In all instances, using total non-gold reserves yielded results similar to those subsequently reported.

31

Taking the six-month LIBOR rate, compounding it semiannually to obtain an annual series, and deflating the sum of principal and interest by the U. S. consumer price index yields a real interest rate close to zero for 1974 through 1978.

32

Because holdings of SDRs are part of total reserves, it is likely that SDR holdings and other reserves are determined simultaneously. To correct for possible bias and inconsistency in our estimates, we used the two-stage least-squares technique to estimate equation (1). Instruments for the reserve variable (RIM) were a variability measure of export earnings and the exogenous variable in equation (1). The results were highly unstable and exhibited almost no explanatory power, so we did not pursue this approach further.

33

To the extent SDR holdings fall short of cumulative allocations, the percentage increase in SDR holdings would be higher.

34

This is in spite of the fact that a participant is not permitted to engage in transactions with designation “for the sole purpose of changing the composition of its reserves” (Article XIX, Section 3(a)).

35

Eurocurrency credits are credits granted by banks out of Eurocurrency funds either on deposit with them or borrowed by them in the Eurocurrency market. See International Bank for Reconstruction and Development, Borrowing in International Capital Markets (May 1980).

36

As discussed in the introductory paragraphs and in Section III of this paper, these countries have had the freedom to enter into transactions by agreement without the requirement of need since August 1976, and have had even more freedom since the Second Amendment to the Articles of Agreement became effective in April 1978. Limits on a participant’s obligation to provide currencies in exchange for SDRs are set in Article XIX, Section 4 for transactions with designation. Since the Second Amendment, there have been no specific restrictions on the conditions under which transactions by agreement can take place.

37

Even without actual net use, SDR allocations will be beneficial to non-oil developing countries to the extent that the dollar rate of interest on their foreign borrowing is above the dollar (or expected dollar equivalent) rate of interest on their foreign exchange holdings deposited abroad. However, whether SDR holdings and foreign borrowing, or foreign exchange holdings and foreign borrowing, are reduced after new allocations have been received depends on portfolio-balance considerations and on the difference between the dollar rate on foreign exchange deposited abroad and the expected dollar equivalent of the net rate of charge on the country’s net use of SDRs. The smaller this difference, the less the incentive to reduce SDR holdings below allocations and the lower the benefits per SDR allocated.

38

While the actual rate of return is known for the length of each contract in its currency of denomination, exchange rate movements must be forecast before the rates of return expected to be earned on investment in different currencies or in the SDR can be compared. Although holding SDRs bearing interest at less than 100 per cent of the combined market rate used in the SDR interest rate basket assured a lower return on SDR holdings than on holdings of the 16 currencies in the former SDR basket, acquiring such direct holdings may have been neither feasible nor economical. (The same would not apply to the 5 currencies used to define the SDR since the start of 1981.) Hence, SDR holdings may have provided benefits by facilitating efficient reserve diversification that may have compensated holders for the below-market rate they earned.

IMF Staff papers: Volume 28 No. 2
Author: International Monetary Fund. Research Dept.