When the Special Drawing Right (SDR) was created in 1969—the year when the First Amendment to the Fund’s Articles of Agreement went into effect—its value was set equal to 0.888671 gram of fine gold. That was also the gold value of the U. S. dollar, so that 1 SDR equaled 1 U. S. dollar. The SDR’s one-for-one relation to the U. S. dollar lasted for only about two and a half years. The devaluations of the dollar decided on in December 1971 at the Smithsonian Conference and in February 1973 led to increases in the dollar value of the SDR, first to $1.08571 and then to $1.20635. The latter value was maintained by the Fund for another year and a half, even though the new par value of the U. S. dollar ceased to be observed almost at once. Since July 1, 1974, the value of the SDR has been based on the market value of a basket of 16 currencies.


When the Special Drawing Right (SDR) was created in 1969—the year when the First Amendment to the Fund’s Articles of Agreement went into effect—its value was set equal to 0.888671 gram of fine gold. That was also the gold value of the U. S. dollar, so that 1 SDR equaled 1 U. S. dollar. The SDR’s one-for-one relation to the U. S. dollar lasted for only about two and a half years. The devaluations of the dollar decided on in December 1971 at the Smithsonian Conference and in February 1973 led to increases in the dollar value of the SDR, first to $1.08571 and then to $1.20635. The latter value was maintained by the Fund for another year and a half, even though the new par value of the U. S. dollar ceased to be observed almost at once. Since July 1, 1974, the value of the SDR has been based on the market value of a basket of 16 currencies.

When the Special Drawing Right (SDR) was created in 1969—the year when the First Amendment to the Fund’s Articles of Agreement went into effect—its value was set equal to 0.888671 gram of fine gold. That was also the gold value of the U. S. dollar, so that 1 SDR equaled 1 U. S. dollar. The SDR’s one-for-one relation to the U. S. dollar lasted for only about two and a half years. The devaluations of the dollar decided on in December 1971 at the Smithsonian Conference and in February 1973 led to increases in the dollar value of the SDR, first to $1.08571 and then to $1.20635. The latter value was maintained by the Fund for another year and a half, even though the new par value of the U. S. dollar ceased to be observed almost at once. Since July 1, 1974, the value of the SDR has been based on the market value of a basket of 16 currencies.

The considerations that lay behind the choice of the basket of currencies as the method of valuation of the SDR, the selection for this purpose of a particular basket (then called the “standard basket”), and the closely related question of the determination of the interest rate on the SDR have already been analyzed in a Fund pamphlet. 1 Since then, a number of important developments have occurred that have a bearing on these questions.

(1) The Second Amendment to the Articles of Agreement went into effect on April 1, 1978. The new Articles no longer define the “unit of value” of the SDR, nor do they specify its method of valuation. Article XV, Section 2 (Valuation of the special drawing right) gives no further precision than is contained in the statement that “the method of valuation of the special drawing right shall be determined by the Fund …,” but it does make any major change from the prevailing method difficult by providing that “an eighty-five percent majority of the total voting power shall be required for a change in the principle of valuation or a fundamental change in the application of the principle in effect.”

(2) The initial basket of currencies was replaced by a new basket on July 1, 1978; on that occasion, both the frequency of future changes and the formula that would be applied in making such changes were announced. 2

(3) The interest rate on the SDR was raised twice, in 1976 and as of the beginning of 1979.

(4) The Bank for International Settlements (BIS) and a number of commercial banks have begun to accept deposits denominated in SDRs. The Fund uses this new facility offered by the BIS to deposit in terms of SDRs part of the profits derived from the sale of the Fund’s gold, pending their disbursement as loans to low-income developing countries; some central banks also use it. These developments in the field of banking are of greater significance for the SDR as a reserve asset than the earlier issuance of some bonds denominated in SDRs, because deposits expressed in SDRs are close substitutes for holdings of SDRs themselves for those official institutions that are in a position to hold both.

These new developments, in conjunction with the Fund’s own experience with the SDR since 1974, shed additional light on the characteristics of the SDR as a basket of currencies. They call for further analysis, both of new problems that have arisen since 1974 and of certain problems that were recognized before then but have not yet been adequately discussed.

The subjects to be discussed in this paper are grouped under three headings. Section I deals with some unfinished business of the 1974 Report to the Board of Governors on Reform of the International Monetary System (hereinafter referred to as the Committee of Twenty (C-20) Report) on the question of the SDR as a basket of currencies. Section II discusses a variety of issues related to the composition, and changes in the composition, of the basket. Section III deals with the closely related question of the rate of interest on the SDR. At various points in Sections I and II, comparisons are made between the SDR and the European Currency Unit (ECU); in the design of the ECU, many problems similar to those encountered with the SDR no doubt presented themselves.

I. Nature of the Basket

The nature of the basket to be adopted for the calculation of the value of the SDR was actively discussed by the Executive Board of the Fund and by the Committee of Twenty and its Deputies from the spring of 1973 through the meeting of the C-20 in Rome in January of 1974. In that meeting, agreement was finally reached to base the valuation of the SDR on a standard basket of currencies. 3 It is clear from the text of the communique that support for that kind of basket was far from enthusiastic. 4 The paragraph begins with a bow to the asymmetrical basket: 5 “… further attention should be given to the question of protecting the SDR’s capital value against depreciation.” The agreement is then hedged around by a number of restrictive clauses—“in the present circumstances,” “for an interim period,” and “without prejudice to the method of valuation to be adopted in the reformed system” 6—indicating less than complete commitment. The fact is that the only consideration which brought about agreement was the proposition that none of the other three methods considered 7 could work under conditions of generalized floating. For a number of years, the method adopted was systematically referred to as the interim method of valuation. 8 More recently, this adjective has been dropped and, as was mentioned previously, the adoption of a different method would now only be possible by an 85 per cent majority of the total voting power in the Fund.

In part, the change in attitude toward the standard basket reflects the realization that the prospect of the arrival of the “reformed” system (i.e., a system “based on stable but adjustable par values and with floating rates recognized as providing a useful technique in particular situations”) 9 is even dimmer now than it was in 1974; under the amended Articles, “the introduction of a widespread system of exchange arrangements based on stable but adjustable par values” (Article IV, Section 4) also requires an 85 per cent majority. But it may also reflect a realization that even in a system that would make the introduction of some other approach possible, the standard basket would still, on balance, be the preferred method. It is of interest to note in this connection that, much like the SDR, the ECU is also valued by the standard basket technique, even though it is to function in the framework of a European Monetary System that is based on fixed but adjustable exchange rates.

There is, indeed, a broader case for the standard basket, which becomes evident as soon as the SDR is seen as, potentially or actually, a market asset. If the SDR is defined as a standard basket, it equals the sum of specified quantities of currencies. Each of these currencies has—at least in principle 10—an interest rate determined in a money market and spot and forward exchange rates against other currencies determined in foreign exchange markets. For each of the currencies, therefore, the interest arbitrage equation—interest rate differential vis-à-vis a reference currency equals the difference between the spot and the forward rate against that currency—must hold, again “in principle.” To the extent that markets function, positions in these currencies can be hedged.

Since the standard basket is a fixed combination of currencies, it has most of the properties that individual currencies have. It has not only a spot value in terms of any reference currency, which is widely quoted, but also—in principle—a forward value in terms of that currency, even though that forward value is not (or at least not commonly) quoted. It also has—in principle—a market interest rate, which equals the weighted sum of the interest rates in the 16 markets with the shares of these currencies in the SDR basket as weights. 11 Thus the standard basket SDR has the properties of a currency. The interest arbitrage equation must hold for it, and it can be hedged.

These properties are an essential precondition for the viability of a market SDR. By its nature, the SDR is the equivalent of a foreign currency for any bank, and banks generally want to avoid open positions in foreign currencies. They can readily acquire the major currencies—either spot or forward—as a hedge against any increases in deposits in these currencies that they may receive. If banks are to be able to accept deposits in SDRs—that is to say, if a market SDR is to exist—banks must be able to cover themselves when they accept SDR-denominated deposits.

As shown above, this should be possible in principle. One can assume that since a substantial number of commercial banks and the BIS are now willing to accept deposits in SDRs, it is also possible in practice. One of the ways in which banks could cover themselves when they accepted, say, U. S. dollars from a depositor who wanted to establish a deposit denominated in SDRs would be to invest the whole amount of dollars in some suitable interest-earning dollar asset of a maturity corresponding to that of the deposit and to enter into contracts to purchase forward the amount of each of the 15 other currencies that, according to the basket, was contained in the amount of SDRs created as a deposit by the bank. Alternatively, they could buy the component currencies spot and invest these. 12 Either technique gives banks a complete hedge for their SDR liabilities and net interest incomes, on the basis of which they can quote interest rates that they are prepared to pay on SDR deposits.

The SDR basket was revised after an initial period of four years and is subject to further revision at intervals of five years on the basis of an announced formula (from which, however, the Fund may decide to deviate). On the occasion of each change—assuming it were an important one—banks with SDR liabilities would have to adjust their forward contracts to remain fully hedged. The formula adopted makes it likely that changes in the composition of the SDR will not be very large and that they can be calculated by the market with increasing precision as the date for a change approaches.

While the standard basket SDR has the properties that permit it to become a market asset, this is not the case for an SDR valued on the basis of any of the alternative methods considered in the C-20. The basic reason is that these other SDRs are not combinations of existing currencies but assets that have no counterpart in the market.

Take, for example, the asymmetrical basket, which differs from the standard basket in that, on the occasion of the devaluation of a currency in the basket, the quantity of that currency would be increased in inverse proportion to the change in the par value. If any currency in that basket approaches a situation where it might be devalued, that component of the basket becomes entirely dissimilar to the currency in the market that carries the same name. The value of the real currency will be expected to fall sharply, and the currency will no doubt carry a large forward discount and a high interest rate. At the same time, the reasonable expectation with respect to the value of the corresponding component in an asymmetrical basket is that it will rise, since any decline in the par value or central rate of the currency is guaranteed to be offset by an increase in the number of units, and the currency is likely, after devaluation, to have a higher market value compared with its new par than it has in relation to its existing par (standing presumably at the lower end of the range around par).

In the adjustable basket, the divergence between “currencies” in the basket and currencies in the market occurs not only at the lower end of the range but at the upper end as well. Thus, neither of these two alternative baskets is a straightforward combination of currencies, and neither can, therefore, be considered as the equivalent of a currency. If the Fund had adopted either of these baskets for the SDR, banks could not have been expected to create market SDRs because they would not have been able to cover themselves.

The fourth method of valuing the SDR mentioned in the C-20 Report (put forward by the United States at one stage but never considered as a serious contender) does not involve a basket of currencies but makes the transactions value of the SDR in terms of any currency equal to its par value. Since currencies do not stay at their parities even in a par value system, this approach would produce a set of broken cross rates between exchange rates in the market and the SDR values of the same currencies. It would not just make it difficult to create market SDRs but, indeed, would make it impossible to envisage them.

There is a further reason why both the asymmetrical and the adjustable basket are incompatible with a market SDR, and that is their treatment of floating rates. The suggestion offered is that when a currency is floating, “the number of units of that currency in the basket would be increased in a similar way”; this advice is accompanied by a footnote indicating that the extent of the float, and hence the required change in the number of units, could be measured as some average of the change in the rate against the nonfloating currencies. 13 But consider the implications of this. If the necessary calculations for this purpose were made periodically (say, monthly), the value of the SDR toward the end of each period would acquire a high probability of a discrete change in a few days. In other words, while floating rates eliminate the market uncertainty of discrete changes in currency values, the treatment of such rates in these SDR baskets would almost certainly reintroduce the same drawback into the value of the SDR with a vengeance. If, for example, the value of a floating currency in terms of any particular yardstick (perhaps the SDR itself) had slipped by 10 per cent in the course of a month, any holder could be virtually certain that on the first day of the next month, when the adjustment in the number of units was made, the value of the SDR would be raised abruptly by 10 per cent multiplied by the weight of this currency in the basket. The possibility of such sudden changes in its value is about the last attribute that one would wish for in an SDR that was intended as a market asset. When the Fund changed the method of valuation of the SDR in 1974, and again when it changed the basket in 1978, it made sure that on the day of transition, the values of the SDR on the basis of the old formula and the new formula would be identical to the last decimal place; and the Fund made it clear that the same approach will be taken in any future changes. 14 The European Community, in launching the ECU, at once established the same principle for its asset. 15

The remedy to this difficulty would be to make any adjustment on a daily basis. To change the composition of the SDR every day would, to say the least, not be conducive to encouraging its use. But it would also be pointless. If one succeeded every day in making that small adjustment in the number of units of a floating currency in the basket that would just offset the change in the value of that currency from the day before, one would have effectively eliminated any impact of the exchange rate movements of that currency on the value of the basket. In other words, if there were any effective way of treating changes in a currency with a floating rate as a succession of devaluations or revaluations, it would have the same result as the elimination of the floating currency from the basket. The latter solution would, of course, be easily applied, but there is some question whether it would be desirable.

When the reform exercise blessed “floating in particular situations,” some may have entertained the hope that such exceptions to the system of stable but adjustable rates would not only be exceptional but would also affect only the less important currencies. After all, in the 25 years up to May 1971, floating had appeared necessary only for such currencies as the Canadian dollar, the Peruvian sol, and the Lebanese pound. To drop minor currencies from the basket for any period that their issuers might have recourse to floating would probably not be an unacceptable solution. But the experience of recent years shows that floating has been the exchange rate regime adopted primarily by countries with large economies, while countries with smaller (and more open) economies have tended to prefer various regimes of pegging. Thus, in the history of European monetary arrangements since 1973, the United Kingdom, France, and Italy have left the snake to be able to float, while the smaller participants have either stayed in the snake through considerable difficulties or (Sweden and Norway) have left it for alternative pegging arrangements. Any basket that excluded major currencies because they floated would risk becoming seriously unrepresentative of the membership (whether Fund-wide or regional) that it was supposed to reflect. In addition, there would be the crucial question whether it would be possible to determine that some large members of any group floated while the others did not. For example, while the deutsche mark has consistently remained in the European snake, it has also clearly been floating. When the European Community had to find a pragmatic answer to the question whether to include floating currencies in, or exclude them from, its ECU basket, the choice was made to include such currencies.

II. Composition of the Basket

Serious consideration of the composition of the SDR basket—the selection of the currencies to be included and the indicators from which the weights for each of the currencies could be derived—did not begin in the Fund until after the more fundamental issue of the nature of the basket had been decided.

The composition of the first (1974) basket reflected a fundamental approach: to give the SDR stability in purchasing power as a reserve asset. To this end, the selection of countries for inclusion in the basket and (with the exception of the additional weight for the U. S. dollar) the weights attributed to the currencies of these countries were made on the basis of their exports of goods and services. As a result of this approach, a country with an average composition of imports could expect stability in the purchasing power over imported goods and services of that part of its reserves held in SDRs. This stability would, admittedly, be subject to a number of qualifications:

(a) It would only apply to the “average country,” that is, the country whose composition of imports of goods and services by country of origin corresponded to the composition of world trade.

(b) It would protect such a country only from price changes in its imports owing to exchange rate changes, and not from price changes owing to inflation in the exporting countries.

(c) The stability to be expected would imply that export prices in domestic currency would not respond to changes in exchange rates. This is an assumption which could, at most, be valid for the exports of industrial countries, and the logic of the approach would therefore require that only such countries be included in the basket. The 16 countries whose currencies made up the 1974 basket, however, included at least two that would not have qualified on this ground (Australia and South Africa), and the inclusion of two oil exporting countries (Saudi Arabia and Iran) in the 1978 basket—replacing Denmark and South Africa—moved the basket even further away from its initial rationale.

(d) The additional weight for the U. S. dollar is not compatible with the logic of the approach, but it was considered attractive by many countries in the light of the special importance of the U. S. dollar in financial transactions.

In the review of the composition of the basket in 1978, the first issue to be decided was whether to make any change in the basket at all. Granted that the shares in trade flows which had guided the selection of the 1974 basket had changed, the link between these shares and the inclusion of currencies in the basket (as well as the weights with which they were included) was not so compelling in logic as to require a change after a limited number of years. Any change would hurt in some measure the “stability” of the SDR as a concept; it might retard the spread of the SDR as a measure of value; and it might cause some difficulties, both inside and outside the Fund, in connection with contracts in which the SDR had been adopted as such a measure. 16 There was some question whether the parties to such contracts would all be prepared to follow the change in the composition of the SDR, or whether the transition by the Fund to an SDR of a slightly different composition might have the effect of creating over time the confusion inherent in a family of SDRs—the “1974 SDR,” the “1978 SDR,” and so on. At the same time, however, it was realized that if the principle of “stability” were followed to the point of never changing the composition of the SDR, the unit could become increasingly remote from reality as reflected in current flows of trade. A time would eventually come when it would be felt that a change had to be made, and that change would be quite radical. These considerations led to the adoption of a procedure for periodic change, at stated intervals and on the basis of an announced formula, that would make the change predictable. The first change was made in 1978, and subsequent changes will be made at five-year intervals.

After this basic decision had been taken, a number of further issues needed resolution:

(i) The principle of selecting countries for inclusion in the basket on the basis of their ranking in the value of exports of goods and services was continued in the 1978 basket and will also apply to future baskets.

(ii) In 1974, the decision was taken to include countries whose share in world exports of goods and services was as small as 1 per cent. This yielded 16 currencies—a number that seemed politically more acceptable than, say, the 5 or 9 that would have resulted from higher cutoff percentages. The question for consideration in 1978 was whether to stick with a basket of approximately 16 currencies or to turn to a smaller number. The main advantage of a more restricted basket was the availability of corresponding market interest rates, which do not exist for some of the minor currencies making up the 16-currency basket. Here the conclusion was partly economic—to include a broad range of countries, so that movements in the value of the currency of any one country would not have too large an influence on the value of the SDR—and partly political—to avoid the appearance that the Fund was concentrating its interest on the major industrial countries. A subsidiary question was whether the countries to be included should be selected by maintaining the same cutoff point (1 per cent of exports of goods and services) or the same number of countries (16). The decision was taken in favor of the fixed number of countries; this would not only provide the appearance of less change but would also bring about less frequent entries and exits from the list than the fixed percentage of exports. There are many countries with trade shares in the neighborhood of 1 per cent of the world total, so that relatively small variations in trade could move such countries above or below the 1 per cent threshold. It would take a more substantial change in trade for an excluded country to pass not only a fixed percentage but also the last included country. 17

(iii) A third issue was whether to continue to base weights on the same criterion used to select countries—viz., exports of goods and services over the preceding five-year period, with a special adjustment for the United States—or to introduce certain other criteria as well. Here the decision was to continue the previous approach for the 1978 basket, but to use a double criterion in the selection of future weights, starting in 1983. At that time, the adjustment that has been made in the weight for the U. S. dollar (in a broad judgmental way, to reflect its financial importance) will be extended to all currencies on a formula basis. For this purpose, the amount of a country’s currency held in other members’ reserves will be used as the yardstick of its “financial importance.” This amount, taken as the average for the five-year period, will be added to the average annual values of the exports of goods and services to derive the weights for the 16 currencies. There is, of course, no compelling reason for a simple addition of average trade flows to average reserve stocks. The procedure described may, however, have recommended itself because when it is applied to the data for 1972 to 1976 it produces, by a formula, weights for all currencies (including the U. S. dollar) that are close to those actually incorporated in the 1978 basket. (See Table 1.) The total value of official holdings shown in Table 1 is about 20 per cent of the value of world exports of goods and services during the same period. The calculation described above thus amounts to an averaging of export shares and currency-holding shares, which have weights of approximately ⅚ and ⅙, respectively.

Table 1.

Weights for a Basket of 16 Currencies Based on Commercial and Financial Indicators

article image

Data are taken from International Monetary Fund, International Financial Statistics; average 1972 to 1976.

Average of end-of-year figures, 1972 to 1976. Data are partly estimated and based on International Monetary Fund, Annual Report 1978 (Washington, 1978), Table 16, p. 53 and on a 76-country sample published in the IMF Survey. (See H. Robert Heller and Malcolm Knight, “Recent Variations in Reserve Asset Holdings Are Result of Greater Resort to Floating Rates,” IMF Survey, Vol. 7 (May 22, 1978), pp. 154–57.) Available data are limited to four countries; however, the figures for the other 12 countries are known to be small.

While a financial variable will be assigned a role in the determination of the weights in future revisions of the basket, the construction of the 1978 basket still follows the approach set in 1974. Table 2 compares the two baskets in terms of weights and amounts of each currency. It will be noted that nearly half the currencies that appear in both baskets have the same weight in both. 18 Indeed, the major adjustment in the basket was not related to trade shares but to exchange rates. Some of the most important changes in the number of currency units in the basket are due to changes in the values of the currencies themselves: the appreciations of the deutsche mark and the yen explain the reductions in the number of units of these two currencies needed to produce an unchanged share of the SDR, and the depreciations of the Australian dollar and the Spanish peseta required substantial increases in the amounts of these currencies to restore the weights assigned to them, which were also unchanged. For most of the other currencies, the change in the number of units was small, mostly because the two underlying variables (trade shares and exchange rates) moved in opposite directions.

Table 2.

Comparison of Composition of 1978 and 1974 SDR Baskets

article image

Exchange rates measured in units of currency per SDR; a negative figure indicates appreciation, a positive figure depreciation.

The preceding paragraphs suggest two further comments on the SDR as a basket. The first relates to the issue of a “strong SDR,” the second to the issue of a principle for the selection of weights for the currencies included.

(1) In the discussions on the structure of the basket that took place in 1973, a good deal of emphasis was placed on designing an SDR with a strong capital value. In that context, considerable attention was given to the question whether it would be possible to enhance the weight of “strong currencies,” either in the original composition of the basket or by means of subsequent adjustments. It appeared, however, that there was no practical way of giving effect to any such desire, if for no other reason than that it is not possible to predict with any assurance the future strength of a currency on the basis of its past strength. But one observation could be offered in consolation: if a currency showed persistent weakness, its share in the SDR, on the basis of an initial number of units of that currency, would gradually decline; indeed, if the basket contained the currency of a country suffering from hyperinflation, the adjustment of the market value of that currency would, by itself, minimize its importance in the basket. In the same manner, the importance of strong currencies in the basket would grow as they appreciated.

It can now be noted, however, in the light of the 1978 adjustment of the SDR basket, that this “self-strengthening” tendency of the basket is effective only as long as the basket is not brought back into line with countries’ trade shares. Since, as noted above, the effect of changes in trade shares on the composition of the basket was on this occasion not very substantial, the main impact of at least the first periodic adjustment of the basket was to reduce the amounts of historically strong, and to increase the amounts of historically weak, currencies.

(2) In the formulas used in 1974 and 1978, a single guiding principle was applied to determine both the currencies to be included in the basket and, with only one exception, their weights. Currencies were selected on the basis of the value of the issuing countries’ exports of goods and services, with a certain cutoff point—countries with at least 1 per cent of the total of world exports were included in 1974, while countries with the 16 largest export shares were included in 1978; and the same value figures determined the weights for these currencies. The only exception was the weight for the U. S. dollar, which was put considerably higher than its trade share to reflect also that currency’s financial role.

There is no practical necessity to use the same indicators for selection and for the assignment of weights. In future SDR baskets, a separation between the two criteria will, in principle, be applied not only to the U. S. dollar but to all 16 currencies; selection will continue to be on the basis of trade shares, but weights will reflect both trade shares and a financial indicator. Such an approach, which divorces the criteria of weighting from the criterion of selection, contains the seeds of potential difficulties. A single criterion—even if it is based, as is the criterion of trade shares, on an underlying theory that is open to some question—has the advantage that it limits the issue of the composition of the basket to the question of the cutoff point. If selection and weighting are approached as separate issues, there is room for a proliferation of indicators to be reflected in the weights, without a compelling basis for the choice among them or for the relative importance to be attached to each.

Such a proliferation can be noted in the design of the European Unit of Account (EUA), to which the ECU has been set equal. The selection of the currencies for the unit did not require a statistical criterion; the unit includes the currencies of all nine members of the Community. In the absence of a statistical criterion for selection, there appeared to be no obvious choice for the weights. Four indicators were considered, viz.,

(a) shares in exports of goods and services to other Organization for Economic Cooperation and Development (OECD) countries in the base period 1969–73;

(b) shares in imports of goods and services from other OECD countries in the same period;

(c) shares in total gross national product in the same period; and

(d) shares in total quotas in the short-term monetary support system (these shares were, at that time, equal for the three largest countries—France, the Federal Republic of Germany, and the United Kingdom).

The weights adopted for individual countries are not, however, weighted averages of these four indicators alone; a number of other factors were taken into account in the formulation of country weights. 19

It should be noted that while the EUA had no provision for changes in weights, the weights of currencies in the ECU will be re-examined, and revised if necessary, at intervals of five years or, on request, if the weight of any currency has changed by 25 per cent. 20

III. Rate of Interest on the SDR

The approach selected for the determination of the rate of interest on the SDR can be seen in much clearer perspective now than at the time, in 1973–74, when this approach was a matter of lively discussion among the Fund staff, in the Executive Board, and in meetings of the Deputies of the Committee of Twenty.

What the staff had proposed, and what was in a considerably modified form adopted by the Executive Board, 21 can in retrospect be viewed as a proto-market approach to the determination of the SDR interest rate. The approach recognized that the SDR would have to coexist with the dollar and other currencies in the reserves of members, and that it should therefore carry an interest rate that made its total yield, as determined by exchange rate movements and market interest rates, compatible with, and thus comparable to, such other reserve assets. The competition, however, was not seen as a close one; accordingly, no great importance was attached to a construction of the interest rate formula for the SDR that would assure an interest return on the SDR precisely equal, or indeed (as it turned out) anywhere near equal, to the combined interest that could be earned on a package of 16 currencies assembled in the amounts in which these currencies appeared in the formula to determine the value of the SDR. The original interest rate formula adopted in 1974, while inspired by the principle of the equality of the SDR interest rate to a theoretical market-determined rate, nevertheless produced results that diverged widely from such a market-determined rate.

These divergences were due to a combination of three factors, which are discussed below. The discrepancies that might result from the operation of the first two of these three were recognized in principle, but they were expected to be relatively minor and probably random in nature; moreover, there were practical reasons for the choices that led to these two sets of discrepancies. The third factor was intended to set the SDR interest rate below a market-equivalent rate.

1. The Five-Currency Interest Rate Basket

It was expected that a weighted average of the market interest rates for the five major currencies in the basket, which together accounted for approximately 70 per cent of the total weight of the SDR, would provide a reasonable approximation of what would have been shown by a 16-currency interest rate basket if such a basket could have been constructed. As a practical matter, it could not have been, since reliable interest rates for about half of the currencies in the basket were not available.

The assumption that the five largest currencies presented a reasonable approximation to the whole basket for the purpose of the calculation of interest rates was not fully borne out by the experience during the four-year period of the first basket (mid-1974 to mid-1978). For the reasons already indicated, the evidence on this point that is based on interest rates is less than satisfactory, but there is strong indirect evidence based on the movements of exchange rates for the 5 currencies in the interest rate basket as against the 11 not in that basket. Table 3 shows a rather persistent increase in the SDR value of the (combined) five major currencies in the basket, from SDR 0.696 in mid-1974 to SDR 0.719 in mid-1978, an increase of 3.4 per cent over this four-year period. The corresponding decline in the combined value of the 11 other currencies was more than twice as large proportionately, down to 92.4 per cent of the original value by mid-1978. Over this short period, therefore, the 5 major currencies in the SDR—which, of course, showed wide fluctuations among themselves—jointly appreciated by no less than 12 per cent in comparison with the 11 minor currencies.

Table 3.

Value of Two Subgroups of SDR Basket, Second Quarter 1974–Second Quarter 1978

article image

End of second quarter of 1974 = 100.0.

One would expect that such different behavior in the value of two groups of currencies should also be reflected in a difference in interest rates. This expectation is confirmed by certain calculations made for an interest rate basket of 12 currencies. 22 These 12 currencies accounted for 93.5 per cent of the original weight of the valuation basket, and the combined value of the 7 additional currencies depreciated in comparison with the 5 major currencies by the same percentage, over the 4-year period, as that shown for the 11 other currencies in Table 3. Although the quality and representative character of some of the additional interest rates may be open to some question, the weighted average interest rate on the basis of this larger basket was about ⅔ per cent per annum higher over the same period than that calculated on the basis of a five-currency interest rate basket. This difference accords reasonably well with the 0.85 per cent average annual appreciation of the five-currency basket against the SDR that is shown in Column 2 of Table 3.

The remaining difference of about 0.2 per cent per annum is a pale reflection of the extent to which, even over a four-year period, exchange rate changes and interest rate differentials for individual currencies fail to cancel out. The return on interest rate and exchange rate changes combined exceeded that of the 12-currency basket by 11 per cent per annum for the yen and by 3 to 5 per cent for the deutsche mark, the French franc, the guilder, the Belgian franc, and the Austrian schilling, all of which had appreciated compared with the basket; the comparable return fell short of that of the basket by 2.5 per cent per annum for the U.S. dollar and by even higher percentages for four other currencies that had depreciated compared with the basket. Only for the Italian lira was the combined return close to that of the basket.

These findings indicate that the assumption, which was expressed in 1974, 23 that the interest rate on the five-currency basket could be considered as representative of the full basket was not confirmed by the experience of the subsequent four-year period and that during this period the interest rate on the SDR was subject to a downward bias on this account of the order of 0.4 percentage points (i.e., 60 per cent of the difference in interest rates between the five-currency basket and a more nearly complete basket). The wide divergences in combined returns also suggest that even when the movements in the value of the five-currency basket are in line with those of the SDR as a whole, the interest rate calculated on the basis of that basket could show a significant difference—in either direction—from the interest rate calculated on a broader basis.

These observations on a particular, relatively short, period are not presented to suggest that the smaller interest rate basket would systematically yield a lower interest rate than a more nearly complete basket. Indeed, it would be odd to find that the currencies of smaller countries had a systematic tendency to depreciate against those of larger countries.

2. Use of Fixed Weights in Calculation of Average Interest Rate

The weights employed in the SDR interest rate calculations for 1974 to 1978 were adopted when the basket method of valuation and interest rate determination was introduced in 1974. At that time, these weights corresponded to the SDR value of the contribution that each of the five currencies made to the total value of the SDR. As exchange rates changed, however, these contributions also changed. A formula that kept the weights in the interest rate calculation constant could, therefore, over time yield results that would diverge from the interest that would be earned on the amounts of the five currencies specified in the valuation basket.

During the four-year period analyzed, the largest difference would have been in the first quarter of 1977, when the combined market rate calculated on the basis of fixed weights was about 0.4 percentage points higher than it would have been if calculated on the basis of the amounts of the five currencies in the basket. (Because the calculation of the actual SDR interest rate involved both a 40 per cent discount (see below) and rounding to the nearest ¼, per cent, the difference between the two methods of calculation would have vanished in that rate.) The main reason for the difference was that the interest rate formula used a constant weight of 13 per cent for the sterling interest rate, even though by that time the share of sterling in the basket had declined to about 10 per cent; the same depreciation of sterling that was responsible for this decline in the share also caused a high interest rate on sterling.

This peculiarity of the method of calculation of the SDR interest rate thus produced a slight upward bias during some quarters of the period observed, and this bias may well be systematic.

3. The Security Discount

When it was first proposed that the SDR interest rate should be related to a weighted sum of the interest rates that could be earned on the currency components of the valuation basket, it was suggested that the actual interest rate on the SDR could appropriately be set somewhat lower than the result of this calculation, because holdings of SDRs conveyed a special benefit. This benefit was that the SDR was a less volatile asset than any individual currency. Countries could be expected to be prepared to pay something for this benefit by accepting a slightly lower interest rate than the combined interest rate of the components of the basket. The SDR could be considered to enjoy a negative risk premium, which became known as the “security discount.”

On further consideration, the theoretical case for a security discount does not stand up well. It may be true that countries would be willing to hold SDRs at a somewhat discounted interest rate if they had to make this sacrifice. But, with some effort, countries can duplicate the SDR by holding an appropriately distributed portfolio of currencies, and they can then earn the full interest rate on each of the 16 components. More practically, they can take advantage of the fact that there are banks prepared to do the assembly job so that holders can acquire ready-made deposits denominated in SDRs. On these deposits they can earn an interest rate quoted by these banks that equals the total interest earned on the package less the profits of the intermediary banks. In other words, in a competitive market, holders can acquire the benefit of the stability of the SDR without having to forgo any important interest income.

Whatever the theoretical case for a moderate security discount in the SDR interest rate, the rates on the SDR set by the Fund have so far involved a very substantial discount as compared with market rates. In mid-1974, when the SDR interest rate formula was adopted, short-term interest rates were very high, with a weighted average of about 10 per cent per annum. The SDR interest rate at that time was put at 5 per cent per annum. This wide discrepancy reflected in part a feeling that interest rates at the time were abnormally high, but it also was indicative of a desire not to make the SDR unduly competitive with other reserve assets, in particular the U. S. dollar.

Any implication of a precise discount of the SDR interest rate as against the weighted sum of market interest rates was, in any event, dispelled by the complicated formula by which the two rates were linked and which is shown as line A in Chart 1. Under this formula, the interest rate equaled 60 per cent of the combined market rate less a deduction of from 0.4 per cent at a combined market rate of 9 per cent or less to 1.6 per cent at a combined market rate of 11 per cent or more. 24

Chart 1.
Chart 1.

Relation Between SDR Interest Rate and Combined Market Interest Rate 1

Citation: IMF Staff Papers 1979, 004; 10.5089/9781451930474.024.A001

1 Line O covers the period January 1969–June 1974.Line A covers the period July 1974–June 1976.Line B covers the period July 1976–December 1978.Line C covers the period from January 1979 onward.

Since the initial formula was adopted in 1974, the Fund has taken two steps to bring the interest rate on the SDR closer to what might be considered a discounted average market rate. In the first step, taken in 1976, the zigzag line (A) was replaced by a straight line with a 60 per cent slope and without a deduction (B). In the second step, which became effective at the beginning of 1979 (at the same time allocation of SDRs was resumed), the ratio of the SDR rate to the calculated rate was raised from 60 per cent to 80 per cent (line C), leaving the SDR with a 20 per cent discount. The original interest rate of 1½ per cent per annum is shown by line O in Chart 1.

The second change occurred after the new (1978) basket had been adopted for the SDR and the weights for the five currencies had been adjusted to this. But the new interest rate basket was not widened so as to coincide with the valuation basket, nor was the formula adjusted so as to apply it to the fixed amounts of the currencies included in the interest rate basket. Identity of the two baskets could have been achieved if the valuation basket had been limited to about 9 currencies, but it would have been difficult, if not impossible, to find a reliable interest rate for each of the 16 currencies; expansion of the interest rate basket to somewhat more than 5, but still far less than 16, currencies did not appear particularly worthwhile, the less so because it would have meant the inclusion of some interest rates derived from less solid markets. Moreover, the 20 per cent discount would, in any event, swamp any refinement in coverage. The same line of reasoning also made it seem like a counsel of perfection to suggest a move from the simple fixed-weight formula to the theoretically more correct formula based on adding the interest that could be earned on the fixed amounts of each currency in the interest rate basket.

The preceding description of the evolution of the interest rate basket to date calls for a conclusion in terms of some more general considerations. The approach adopted in 1974 could be described as being guided—subject to quite substantial qualifications—by the principle of providing the SDR with an effective yield that was comparable with, but not equal to, the yield on reserve currencies, especially the U. S. dollar. The changes made in 1976 and 1979 brought the yield on the SDR closer to a competitive level without, however, closing the gap in favor of the dollar.

It seems plausible that countries’ views on the distribution of their reserve portfolios between reserve currencies and SDRs are affected by many considerations. The relative yield of the two assets plays a role in this, but by no means an exclusive role. Moreover, the elasticity of substitution between the SDR and reserve currencies in response to their relative interest rates cannot be very high. Factors limiting the substitution include constraints on the use of the SDR and limits on the ability of countries to obtain SDRs. In the light of considerations such as these, the approach selected in 1974 to provide the SDR with a yield that was market-related seemed adequate, in the sense that it did not place an undue burden on the rules for the holding and use of SDRs.

Two changes have taken place since that time. First, the rules on the use of SDRs have been considerably relaxed by the Second Amendment, which gives participants greater freedom to engage in transactions in SDRs and permits a wider range of operations in SDRs. Second, the importance of the interest rate that the Fund pays on the SDR has been greatly increased by the emergence—even if only on a small scale so far—of market SDRs. A number of large commercial banks now accept deposits denominated in SDRs with various maturities for which they quote interest rates. This makes it possible for central banks, and certain other official institutions that could be potential “other holders” of SDRs, to invest reserves in market SDRs. As a consequence, the interest rate that the Fund pays on SDRs, which was originally designed to be effective only in the field of weak competition between SDRs and dollars, will now increasingly have to perform the more demanding task of equilibrating holdings in a much more competitive field where SDRs are pitched against market SDRs (or should one call them Euro-SDRs?). As this competition becomes more important—as the Euro-SDR market itself becomes more competitive—it seems likely that the Fund will increasingly have to pay attention to the manner in which interest rates on SDR deposits in that market are determined.

There would seem to be two possible ways in which the Fund could respond to this situation. One approach would be to follow more closely in its own calculations the elements that enter into market calculations: the Fund could widen its interest rate basket, perhaps use some Euromarket interest rates rather than the domestic interest rates where these differed significantly, or derive implied interest rates from the margin between spot and forward quotations where the exchange market for a certain currency appeared to be more representative than the money market. In addition, the Fund might have to reduce the discount it applied below the present 20 per cent. This approach would no doubt involve considerable technical difficulties; indeed, it could bring about a situation in which the Fund would come close to replicating the calculations made by commercial banks in their determination of the interest rates that they were prepared to offer on SDR-denominated deposits. Alternatively, the Fund might, at that stage—once a really competitive market in SDR-denominated deposits had developed—decide to follow a different approach, viz., to derive the interest rate that it would pay on the SDR directly from the interest rates for SDR paper quoted in that market. This would eliminate all technical aspects and leave only the important policy decision as to how the Fund’s SDR rate should relate to the market SDR rate. In general, paper issued by national governments (e.g., treasury bills) carries a somewhat lower interest rate than comparable paper issued by commercial banks (e.g., certificates of deposit of the same maturity). Similarly, SDRs issued by the Fund could be expected to be fully competitive with market SDRs while paying a somewhat lower interest rate. It is too early to speculate on how wide a margin would be appropriate. The possible margin would not, in any event, be a single number. Much would depend on the quantity of SDRs allocated by the Fund and the quantity of SDR-denominated claims issued by a possible substitution account that would have to find a place in members’ portfolios and on the extent to which it were felt that, in the long run, reliance should be put on restraints on the holding and use of SDRs.

APPENDIX: The Special Drawing Right

Method of Valuation

1. Effective July 1, 1978, the value of one special drawing right shall be the sum of the values of specified amounts of the currencies listed in 2 below, the amounts of these currencies to be determined on June 30, 1978 in a manner that will ensure that, at the average exchange rates for the three-month period ending on that date, the shares of the currencies in the value of the special drawing right correspond to the weights specified for each currency in 2 below.

2. The currencies and the weights referred to in 1 above shall be as follows:

article image

3. The list of the currencies that determine the value of the special drawing right, and the amounts of these currencies, shall be revised with effect on July 1, 1983 and on the first day of each subsequent period of five years in accordance with the following formula, unless the Fund decides otherwise in connection with a revision:

  • (a) The currencies determining the value of the special drawing right shall be the currencies of the 16 members whose exports of goods and services during the five-year period ending 18 months before the effective date of the revision had the largest value, provided that a currency shall not replace another currency included in the list at the time of the determination unless the value of the exports of goods and services of the issuer of the former currency during the relevant period exceeds that of the issuer of the latter currency by at least 1 per cent.

  • (b) The amounts of the 16 currencies referred to in (a) above shall be determined on the last working day preceding the effective date of the relevant revision in a manner that will ensure that, at the average exchange rates for the three-month period ending on that date, the shares of these currencies in the value of the special drawing right correspond to percentage weights for these currencies, which shall be established for each currency in the proportion that the sum of the value of the balances of that currency held by the monetary authorities of other members and the value of the exports of goods and services of the issuer of the currency bears to the total sum of the same values for all 16 currencies during the relevant period, rounded to the nearest ½ of 1 per cent.

4. The determination of the amounts of the currencies in accordance with 1 and 3 above shall be made in a manner that will ensure that the value of the special drawing right in terms of currencies on the last working day preceding the five-year period for which the determination is made will be the same under the valuation in effect before and after revision.

Decision No. 5718-(78/46) G/S

March 31, 1978


Mr. Polak, the Economic Counsellor and Director of the Research Department, is a graduate of the University of Amsterdam. He was formerly a member of the League of Nations Secretariat, economist at the Netherlands Embassy in Washington, and Economic Adviser at UNRRA. He is the author of An International Economic System and of several other books and numerous articles in economic journals.


J. J. Polak, Valuation and Rate of Interest of the SDR, IMF Pamphlet Series, No. 18 (Washington, 1974). (Hereinafter referred to as Polak, Valuation and Rate of Interest of the SDR.)


The text of the Decision is reproduced in the Appendix.


Communiqué of January 18, 1974, paragraph 4. (Committee on Reform of the International Monetary System and Related Issues, International Monetary Reform: Documents of the Committee of Twenty (International Monetary Fund, Washington, 1974), p. 218.) (Hereinafter referred to as Documents of the Committee of Twenty.) The text refers to “a basket of currencies” without further specification; however, the intention was to adopt the “standard basket.”


One observer reports that all five of the major powers opposed its adoption through the greater part of the discussions. See John H. Williamson, The Failure of World Monetary Reform, 1971–74 (Sunbury-on-Thames, England, 1977), p. 180.


A description of that basket is given on page 632 of the present paper.


Documents of the Committee of Twenty, p. 218.


The “asymmetrical basket,” the “adjustable basket,” and the “par value” technique. For descriptions of these three methods, see Documents of the Committee of Twenty, pp. 44–45 or Polak, Valuation and Rate of Interest of the SDR, pp. 24–26.


See, for example, “Interim Valuation of the SDR: New Rule O-3 and Method of Determining and Collecting Exchange Rates,” International Monetary Fund, Annual Report 1974 (Washington, 1974), p. 116.


Documents of the Committee of Twenty, p. 8.


The qualification “in principle” is needed because, while all currencies in the basket are actively traded in some spot exchange market, not all have active forward markets and not all of the 16 issuing countries have well-developed money markets.


These shares vary over time as exchange rates change. For any currency, the share on a particular day equals the SDR value of the number of units of that currency in the basket—that is, the number of units multiplied by the exchange rate of that currency against the SDR on that day.


This is the technique generally used by the Bank for International Settlements. (See its Forty-Ninth Annual Report (Basle, 1979), p. 168.)


Cited in Polak, Valuation and Rate of Interest of the SDR, p. 25.


See paragraph 4 of Executive Board Decision No. 5718-(78/46) G/S of March 31, 1978, which is reproduced in the Appendix.


Resolution of the European Council of December 5, 1978, Part A, Paragraph 2.3 states that “revisions [in the composition of the basket] … will, by themselves, not modify the external value of the ECU.” This resolution is reproduced in the Fund’s Annual Report on Exchange Arrangements and Exchange Restrictions: 1979 (Washington, 1979), pp. 29–30.


In the loans contracted by the Fund under the oil facility, the lender obtained the right, in the event of a change in the way in which the value of the unit of the SDR was determined, to have the old (1974) method continue to apply (see International Monetary Fund, Annual Report 1974 (Washington, 1974), p. 126); when the change was made in 1978, however, no lender exercised this right. Creditors under the supplementary financing facility, by contrast, were given the option to require immediate repayment if the Fund decided to make a change in the method of valuation of the SDR, on the basis of the method of valuation in effect before the change (see International Monetary Fund, Annual Report 1978 (Washington, 1978), p. 117). Although the decision to establish this facility was taken in 1977, no claims under this facility were acquired before the 1978 decision on the valuation of the SDR had been taken. The quinquennial changes in the composition of the basket pursuant to this decision will not constitute changes in the method of valuation of the SDR.


Continuity was slightly enhanced by the provision that, to be included, any new country would not simply have to pass a country already in the basket, but would have to pass it by a margin of at least 1 per cent of the trade of the displaced country.


In order to avoid the impression of spurious precision, the weights are rounded to the nearest ½ per cent; this procedure can easily either hide or magnify certain changes which, especially for the countries with small weights, might not be unimportant in other contexts. It should also be noted that for the 13 currencies other than the dollar that appear in both baskets, the unrounded 1978 weights work out somewhat lower than the 1974 weights because the combined weight for the currencies of Saudi Arabia and Iran in the 1978 basket exceeded the combined weight for the currencies of Denmark and South Africa in the 1974 basket by 2½ per cent.


H. Joly Dixon, “The European Unit of Account,” Common Market Law Review, Vol. 14 (May 1977), pp. 191–208. The technical annex to this paper indicates that the European Community (EC) used the same technique to calculate the number of currency units as had been developed for the SDR. Cf. Polak, Valuation and Rate of Interest of the SDR, Appendix I, pp. 23–24. (One difference is that the number of units of each currency in the EUA is rounded to three significant digits, as against two digits in the SDR; perhaps the explanation is that, with nine currencies, the use of two digits does not give enough possible combinations to ensure the accuracy of the unit to the desired six significant digits.)


Resolution of the European Council of December 5, 1978 (cited in footnote 15), loc. cit.


The C-20 Deputies never came around to this approach. In their final report, they merely recommended that the interest rate should be set from time to time “in the light of changing market interest rates.” (See Polak, Valuation and Rate of Interest of the SDR, p. 19.)


The currencies added are those of Australia, Austria, Belgium, Canada, Italy, the Netherlands, and Spain.


Polak, Valuation and Rate of Interest of the SDR, p. 20.


The formula can be written as follows: i = 0.6c − 0.4 − 0.6(x–9)

i = 0.6 c – 0.4 – 0.6 (x–9)


i = interest rate on the SDR (before rounding)

c = combined market interest rate


The rate calculated by this formula is rounded to the nearest ¼ per cent; the resulting small steps are not shown in Chart 1