The goal of this fifth session of the seminar was to examine the potential for the SDR in the provision of conditional liquidity to member countries, rather than unconditional liquidity, as at present. The speakers—Jacques Polak, Marcello de Ceceo and Francesco Giavazzi, ZHU Xiaohua, and György Surányi—discussed a number of questions related to this general theme. Does the increasing integration of international financial markets, and the growing stock of internationally mobile private capital, imply a greater need for official safety-net mechanisms? If so, should such a mechanism involve the Fund and the SDR? Is there a case for voluntarily redistributing SDRs after allocation into a Fund-administered account that could be used to supplement the conditional financing available to Fund members? Should the SDR-financing technique be extended to all of the financial operations of the Fund?

The goal of this fifth session of the seminar was to examine the potential for the SDR in the provision of conditional liquidity to member countries, rather than unconditional liquidity, as at present. The speakers—Jacques Polak, Marcello de Ceceo and Francesco Giavazzi, ZHU Xiaohua, and György Surányi—discussed a number of questions related to this general theme. Does the increasing integration of international financial markets, and the growing stock of internationally mobile private capital, imply a greater need for official safety-net mechanisms? If so, should such a mechanism involve the Fund and the SDR? Is there a case for voluntarily redistributing SDRs after allocation into a Fund-administered account that could be used to supplement the conditional financing available to Fund members? Should the SDR-financing technique be extended to all of the financial operations of the Fund?

Should the SDR Become the Sole Financing Technique for the IMF?

Jacques J. Polak

Summary of Findings

Analysis of the Fund’s simultaneous experience with two financing techniques for its credit operations—currencies for conditional, and SDRs for unconditional, credit—suggests that it would be beneficial to adopt the SDR technique for both types of operations, because

  • the SDR technique gives creditors a preferred asset—in terms of remuneration and easy usability—that could be further improved by allowing commercial banks to become “other holders” of SDRs;

  • that technique eliminates the need for the current practice of seeking excessively large quota increases, only 30 percent of which can be used to raise Fund lending;

  • the consolidation of the General Resources and SDR Accounts should facilitate finding an economically efficient formula for the allocation of the Fund’s general expenditure, including possibly a higher rate of charge on the net use of SDRs than on SDR creditor positions;

  • the elimination of members’ contributions in SDRs and currencies when their quotas are increased, and hence of the appearance of budgetary commitments, should facilitate the periodic quota rounds, without weakening the role of quotas as the governor of the Fund’s credit extension;

  • the totality of the changes envisaged would make a much needed contribution to the transparency of the Fund’s financial structure, not least by giving the institution a meaningful balance sheet.

An Essential Distinction: SDR as Asset Versus Allocation of SDRs

The introduction of SDRs into the Articles of Agreement of the Fund in 1969 at the end of some five years of negotiation brought two new features to the institution: (1) a new way for the Fund to extend credit, by the allocation of unconditional liquidity to the membership as a whole, in addition to the existing method of providing conditional credit on a case-by-case basis to particular members that encountered payments difficulties; and (2) the financing of this new form of Fund credit by the simple method of making entries on the Fund’s books, in contrast to the prevailing method based on resources contributed by the membership through quotas or, more exceptionally, through lending to the Fund. The new acronym, SDR, became emblematic of the entire novation: its liquidity objective as well as the financing technique supporting it. But in considering the future of the SDR, it is essential to consider the two separately. Other papers at this seminar discuss the merits and demerits of the provision of liquidity by means of allocation to the membership as a whole or to a large subsection of it. This paper will concentrate on the SDR as a financial technique. It will conclude that this technique is superior to that based on contributed currencies and that the Fund would benefit from applying it not only to its unconditional lending (whether maintained at its present level or raised by new allocations) but also to its conditional lending.

Failure to distinguish between the technique and the purpose for which it is used—between the instrument and the transactions conducted with it—gave rise to a debate on the question, as it was phrased, “whether the SDR was money or credit.” But since that is not a meaningful question, it has no answer. The question probably arose from a confusion of two legitimate questions: (1) does the SDR fit under one’s definition of “money,” or should it be considered as quasi-money or as a nonmonetary financial asset? And (2) is the allocation of SDRs an extension of credit, a grant, or perhaps something in between? I have dealt with these questions before (Polak, 1971); in this paper, I have found it convenient to treat the SDR as money, and SDR allocations as the extension of unconditional Fund credit.

The Three Financing Techniques of the IMF

In the course of its history, including its prehistory, the Fund has engaged in two major debates on the technique to be used to provide members with financial resources and it has gained experience with three different techniques for this purpose. This section will touch briefly on these debates and appraise the three techniques in terms of their merits in meeting the objectives of the institution.

The “Mixed Bag of Currencies” of the White Plan1

The plans for postwar monetary cooperation prepared by the United States and the United Kingdom in 1942 both contained provisions to make additional foreign exchange resources available to countries in payments deficit. Under the U.S. (“White”) plan, the resources for this purpose would be contributed to the Fund by each member in gold and its own currency. The Fund would then use its “mixed bag of currencies” to (in effect) extend credit to the countries that had a payments deficit “in that currency.” (The term “credit” was studiously avoided: the Fund “sold” the currency required against an equivalent amount of the currency of the country in deficit, which the latter country was expected to “repurchase” as it overcame its difficulties.) By contrast, the approach in the British (Keynes) plan was based on British banking practice: countries would be given overdraft facilities that, if used, gave the creditor a balance in international currency, called bancor, on the books of the Fund. This, to British thinking, seemed the natural way to proceed: it avoided notions of bilateral payments needs and the “scarce currency” problem and it finessed the question of the source of the capital for the new institution.

Although many other differences between the U.S. and the British plans were rather quickly removed by compromises, it took more than two years, until the Joint Statement in April 1944, before the British acquiesced in the currencies approach. The Americans insisted on this approach with the argument that the U.S. Congress would have difficulties with the introduction of an international currency and with a financial technique that was unfamiliar in the United States, where the term “overdraft” has a connotation of financial impropriety. The more weighty economic argument was that the potential U.S. commitment to give balance of payments assistance to the rest of the world through the Fund would be more than twice as large under the Keynes Plan as under the White Plan: under the latter, the maximum contribution that could be claimed from the United States through the Fund equaled the U.S. quota ($2.75 billion); under the former, the sum of the quotas of all other members (about $6 billion).

The Fund thus began its operations by the technique of “selling” contributed currencies, and for the first ten years or so this proved satisfactory all around—but for two reasons that were specific to the early postwar years. (1) The payments problems for which members approached the Fund were almost exclusively dollar problems, so that there was little question that the U.S. dollar was the currency “needed” under the provisions of the Articles; up to 1958, the Fund used almost exclusively dollars in its transactions. (2) The dollars sold by the Fund were mostly spent on imports from the United States or willingly held if acquired by other countries; thus the United States did not lose gold as a result of the Fund’s use of dollars, and in any event the size of U.S. gold reserves was not a matter of concern at that time.

Halfway House Between the White Plan and the Keynes Plan

All of this changed as a wider range of currencies became de facto convertible in the late 1950s. Convertibility made the specification of a “currency … needed for making in that currency payments…” economically meaningless and thus cleared the way, the Fund’s lawyers cooperating, for a use of a wider range of currencies. But countries were not generally prepared to hold currencies other than the dollar in their reserves; hence the issuers of these currencies could expect to lose reserves as amounts drawn on the Fund were presented for conversion. Having only recently rebuilt their reserves to reasonably comfortable levels, these countries proved reluctant to have their currencies drawn from the Fund if this meant a loss of reserves. Thus, for the Fund to use the wide range of the currencies it held, it was necessary to establish the fact—by a Fund decision on gold tranche drawings, by appropriate publicity, and ultimately (in 1969) by amendment—that reserve positions held in the Fund, created by reductions in the Fund’s holdings of a members currency below quota, fully deserved to be considered reserves.2

These positions, whose value was at the time maintained in gold, made their entry in the statistics on international reserves in about 1963, under the name “gold tranche positions.” Prudence prevented the IMF staff from suggesting that they be called “bancor,” but it was well realized that the Fund’s financing technique had quietly moved at least halfway from the White Plan to the Keynes Plan: although the “acceptance limits” of the White Plan remained in force, the new approach to transactions meant that the extension of credit by the Fund created reserves on the books of the Fund in a way very similar to that envisaged by the Keynes Plan.3

As long as the value of reserve positions in the Fund was expected to be maintained in terms of gold, the issue of the payment of interest on these positions remained subdued, although creditor countries insisted on the inclusion in the First Amendment of a rate of remuneration on these positions (in excess of 25 percent of quota), which was set at 1½ percent, equal to the initial interest rate on the SDR. Over time, as the reference value of these positions moved from gold to the SDR as a basket of currencies, it became accepted that they should normally be remunerated at a competitive market rate; in the Second Amendment this rate was defined as 80 percent of the SDR rate as the minimum and the full SDR rate as the maximum.

The SDR Technique

Thus, by the mid-1960s, when the international discussions on the design of a supplementary liquidity mechanism were about to start, the Fund had developed a system to finance its credit operations, which, although it had not yet been fully sanctioned by new legal provisions, could be seen as a functioning halfway house between the Keynes and the White plans. In the view of the staff, the financing of any new system of unconditional access could either be grafted onto the existing mechanism by means of a set of unconditional drawing rights (as in the gold tranche), or be designed as a separate operation in terms of “units” that would in fact be like Keynes’s bancor. As these alternatives were considered, it became clear that any policy desiderata (magnitude of access, decision-making process, repayability, and so on) could be attached to either of the two technical approaches. As in the 1940s, the real controversy was not about financial techniques but about the degree of control under which any new mechanism would dispense liquidity: under what conditions, to which countries, in which amounts, and subject to what reimbursement. As each of these issues was finally settled in a manner that mostly reflected the views of the majority of the Fund membership, a concession was made to the views of the minority in the choice of the name of the new instrument; it became “special drawing right” rather than “reserve unit” or any other set of words that would suggest the creation of a new asset with recognized reserve characteristics (Gold, 1971).

Experience with Alternative Techniques

The choice made late in the liquidity discussions of the 1960s to choose a different financing technique for the new liquidity mechanism than the one in existence for the Fund’s conditional credit implied that the institution would be running two financial systems side by side. There have now been 25 years of experience with this dual arrangement. What does it tell us about the relative merits of the two approaches?

Experience with the SDR Technique

In many respects the SDR system has worked extremely well in the technical sense as a means of extending credit across the board to the Fund membership and of allowing members to use SDRs as reserve assets. (This, of course, is not a judgment on the success of the SDR scheme in the broader sense, including the lofty objectives that were added to the Articles of Agreement in the Second Amendment: “… promoting better international surveillance of international liquidity and making the special drawing right the principal reserve asset in the international monetary system” (Article V, Section 7).) Although a substantial number of members joined the Fund too late to have benefited from allocations, all members are now “participants” and can hold SDRs. The somewhat novel mechanism of designation, which had its forerunner in the practices with respect to the currencies selected for drawings that the Fund had developed since the late 1950s, has operated without difficulties. Even more significant, this mechanism has been found to be redundant since late 1987: in practice it has been replaced by arrangements that the Fund has established with about a dozen members that are willing to acquire or sell SDRs, within agreed limits, against usable currencies (International Monetary Fund, Annual Report, 1995, pp. 148-50). With these arrangements in place for the conversion of SDRs, members wanting to use their SDRs to acquire currencies do not have to claim a balance of payments need. Another indicator of reduced concern among the membership about acceptance limits was implicit in the proposal made by the countries of the Group of Seven for a special allocation of SDRs to members that had not, or not fully, participated in earlier allocations; with the exception of Switzerland, these were all potential users of SDRs, and 25 years ago any allocation that failed to generate matching increases in acceptance limits of prospective creditor countries would have been considered unthinkable.

Nevertheless, some notable weaknesses in the mechanism have surfaced, which are mainly related to the fact that commercial banks cannot under the Articles be “other holders” of SDRs.4 As a consequence, SDRs cannot be used directly in intervention in exchange markets; nor can they be used as compensating balances for credit granted by banks. These may be at least some of the reasons why countries with precarious reserves—primarily developing countries—typically hold only a small proportion of the SDRs that have been allocated to them.

This experience with the SDR mechanism suggests the desirability of an amendment to the Articles that would permit commercial banks to become “other holders.”5 The objective of this change would be to induce a limited number of commercial banks to make a market in SDRs, at small spreads, essentially as a service to their central bank customers. There are a number of hopeful signs in recent experience with the SDR that suggest that this objective could be achieved. Most industrial countries, including the United States, Germany, and Japan, have been willing to hold SDRs in excess of their allocations; indeed, although the Fund has been able to find willing buyers for SDRs, it has sometimes been unable to find willing sellers. Designation has, as noted, proved to be redundant even without the additional cushion of demand that would be created by bringing commercial banks into the market; the relative decline of SDRs outstanding as a proportion of total reserves (see Box 1) may well have been a factor in this development. The interest rate for SDR deposits in private markets, as reported in International Financial Statistics, is very close to the SDR interest rate paid by the Fund. That interest rate, and the fact that members would have a frequent demand for SDRs for payments to the Fund, would provide a strong support level for the SDR in a free market. If judged necessary, this support could be reinforced by maintaining designation (not necessarily in the same manner as under the present Articles) on a stand-by basis.

Reserves in the Fund Compared with Other Reserves

Countries holding part of their reserves in reserve positions in the Fund or in SDRs have a natural concern about the proportions of their total reserves held in these forms. The table below indicates the development of these proportions from 1970 to 1995 for the industrial countries as a group, which account for the bulk of the credit to the Fund. (Total reserves are defined including gold for 1970 and excluding it for subsequent years.) As is well known, the SDR proportion has declined since the last allocation in the early 1980s, but the proportion of reserves held in reserve positions in the Fund has also been lower in recent years than in most of the 1970s and 1980s.

Reserves of Industrial Countries

article image
Source: International Monetary Fund, International Financial Statistics.

Experience with the Modified Currencies Technique

The introduction of the SDR system made it possible for the General Resources Account to hold SDRs, and that account received large amounts of them in charges and repayments and, since the Second Amendment, in payments for quotas when these were increased. Thus a substantial part of the business of that account in the last 25 years has been conducted in SDRs and has benefited from the smooth operation of the SDR mechanism. For example, when Mexico drew from the Fund an amount equal to SDR 5.3 billion in February 1995, the Fund used SDR 3.5 billion in SDRs in a single transaction and Mexico sold that entire amount to other members for currencies. But the Fund cannot conduct all its transactions in SDRs (at the end of fiscal 1995, its SDR holdings had been reduced to about SDR 1 billion), and it must therefore rely on being able to draw on its holdings of currencies. Here, it has encountered two difficulties.

First, although the Fund has accepted in principle to pay the full SDR interest rate on reserve tranche positions in excess of a certain minimum holding for each member,6 it has in fact never done so: in the last ten years it has actually reduced the rate paid to creditors from about 90 percent to the statutory minimum 80 percent of the SDR rate, in order to offset uncollectible charges of some members and to build up special reserves against possible future defaults. The below-market remuneration of creditor positions has tended to make agreement on the selection of currencies for use in Fund transactions more difficult; in 1990 the Fund found it necessary, after a long and difficult debate, to modify its policy (which had been in effect for some thirty years) under which the guiding criterion for the choice of creditor currencies had been proportionality to members’ gold and foreign exchange reserves, by introducing an overriding criterion of broad proportionality to quotas (Annual Report, 1992, p. 72; Annual Report, 1995, p. 138).

The difficulties encountered in the use of its holdings of usable currencies, at the same time that the Fund could dispense with the corresponding rules for designation in the SDR mechanism, is one indication of the weakness of the currency mechanism in the execution of the Fund’s standard credit transactions. A more serious weakness results from the attitude the Fund and its creditors have gradually developed toward measures of the Fund’s liquidity. The Fund needs of course a measure—even if it cannot be very precise—of its “liquidity”: the assets it has at its disposal for future transactions. It defines that liquidity as its holdings of “usable” currencies and SDRs, adjusted for commitments under arrangements, less an allowance for needed working balances (reminiscent of a prescribed minimum number of taxicabs to be kept at the station?) and an estimate of amounts of usable currencies that may become unusable.

There might be some advantage in expressing the Fund’s liquidity as a ratio to an indicator of potential demand for drawings, but a reference value for this purpose would be hard to find—although one can note an interesting recent attempt to appraise the liquidity remaining after the large arrangement with Mexico in terms of the number of future “Mexicos” that the Fund might be able to handle. Faced with the difficulty of finding a good general reference value, the Fund has opted for one that looks plausible but is in fact wrongly focused—namely, the total of its “liquid liabilities,” defined as the sum of members’ reserve tranche positions and lending to the Fund.7 Because of the importance that the liquidity ratio has acquired in recent policy discussions in the Fund, it deserves a critical look.

Calls on the Fund’s liquid assets arise overwhelmingly from drawings in the credit tranches, not from reserve tranche drawings by creditor countries. When these countries need to dip into their reserves, they tend to use their holdings of foreign exchange, which, for the major creditors, are many times larger then their reserve tranche positions (see Box 1), rather than go through the procedure of reserve tranche drawings on the Fund. In fact, not a single industrial country has made a reserve tranche drawing in the last eight years; in the preceding seven years such drawings from this group of countries totaled about SDR 700 million, or less than 2 percent of the SDR 48.5 billion credit tranche purchases in the same period.

A further curiosity of the liquidity ratio as defined is that while it appears to focus on demands from the Fund’s creditors only, it does not make any allowance for the resources potentially available to meet these demands under the General Arrangements to Borrow (GAB). Yet the GAB were specifically designed to strengthen the Fund in crisis situations affecting the Group of Ten countries—the main creditors of the Fund—and were in fact drawn upon to finance part of the reserve tranche drawing of the United States in 1978. This omission is more than a statistical quibble. Even before the planned doubling in size of the GAB, its resources—SDR 18.5 billion, almost all in usable currencies—account for nearly 60 percent of the SDR 31.6 billion reserve tranche positions of all (GAB and non-GAB) industrial countries (see Box 1) and are more than 25 times as large as the total of the reserve tranche drawings made by these countries in the last 15 years.

Over the years, the view has gained ground in the Fund that the liquidity ratio as defined should not, except perhaps for brief periods, fall below 70 percent. There appears to be no record to substantiate the choice of this number—so widely in excess of both the practice in national banking systems and the Fund’s experience with respect to reserve tranche drawings. Yet the 70 percent figure has become well established. In a December 1989 statement to the Executive Directors in connection with the Ninth General Review of Quotas, the Managing Director noted the forecast of a ratio of 56 percent a year ahead, “which is far below the 70 percent ratio which provides the necessary safeguard of the monetary character of the Fund.” The calculations attached to his statement derived the required size of the ninth quota increase on the assumption that the liquidity ratio five years ahead, after the estimated increase in the Fund’s credit over that period, should not fall below 70 percent at the end of the period. And most Executive Directors were reported to share the Managing Director’s views on the required degree of liquidity.8

Finally, the minimum ratio described as 70 percent is in fact an even higher number because of the staff’s practice (mentioned above) of paring some 20 percent (a figure that can be inferred from the data presented in recent Annual Reports) from the actual figure of usable assets to eliminate the part it considers not really usable. Thus the 70 percent ratio stands in fact for a liquidity ratio of (70/80=) 87.5 percent.

The gradual elevation of the 70 percent ratio from an innocuous historical average to a norm does not mean that transactions stop when this figure is reached. Instead, its major influence is on the institution’s approach to the quinquennial quota cycle. The fixation on a high liquidity ratio, together with the unpredictability of the demand for Fund drawings and the traditional slowness of quota reviews, has led to the practice in the Fund to begin to seek large quota increases while it still has large usable resources. But at the same time the approach followed makes quota increases very inefficient in terms of raising the ability of the Fund to engage in net additional transactions. As shown in Box 2, the logical outcome of the liquidity ratio approach is that any quota increase raises the lending capacity of the Fund by only about 30 percent of that increase. That same approach, moreover, renders the one instrument specifically designed to strengthen Fund liquidity—quota contributions in SDRs—almost wholly dysfunctional as these SDRs add almost nothing to the Fund’s liquidity ratio.

The case may perhaps be made that quota increases in the Fund are costless to contributing members, in the sense that they receive alterntive assets for the SDRs contributed or the amounts of currency used by the Fund. But this argument has not proved convincing to the parliaments of all member countries; and in any event the political exercise of wringing quota increases out of 180 parliaments is not a costless process, as has become increasingly clear in recent quota exercises.

Conclusions on Alternative Techniques

The preceding appraisal of the Fund’s experience with the two financing techniques it has used side by side for the last 25 years has brought to light a fundamental weakness of the currencies technique:

Effects of a Quota Increase on Fund Lending Capacity

We consider separately the effects of (1) contributions in the member’s own currency (for 75 percent of the quota increase) and (2) contributions in SDRs (for 25 percent of the increase).

1 Currency contributions. In order to leave an incremental liquidity ratio of 87.5 percent (7/8), almost one-half (7/15) of any currency contribution will have to be added to reserves, while only slightly over one-half (8/15) can be used to extend new credit and thus create new reserve tranche positions of the same amount. Assuming that countries with usable currencies account for 2/3 of the total quota increase, the 75 percent currency contributions of these countries will thus allow a credit expansion of 2/3 × 3/4 × 8/15 = 26.7 percent of the total quota increase.

2. SDR contributions. Assume all members contribute 25 percent of their quota increases in SDRs (or in the usable currencies of other members). As these contributions create reserve tranche positions of the same amount, they allow for a credit expansion of only 1/8 × 1/4 = 3.125 percent of the total quota increase.

On the assumptions made (which if anything are optimistic), there will be room for a credit expansion of 26.7 + 3.125 = just under 30 percent.

What is particularly striking is the smallness of the effect of the SDR payments, be-cause the liquidity ratio approach subjects these payments to an 87.5 percent “reserve requirement.”

it provides creditor countries with an automatic drawing right that is a less satisfactory “asset” than the directly usable SDRs that creditor countries earn when they make their currencies available under the alternative technique. It is time to acknowledge that the view held by many, including the IMF staff, at the time of the SDR exercise, that automatic drawing rights and “units” could be designed to work to equal satisfaction has not been confirmed by the Fund’s experience. The introduction of a market in SDRs among commercial banks, as recommended, would enhance further the comparative advantage of SDRs over currencies as the financing technique for conditional Fund credit.

There is thus every reason for the Fund, after 50 years of applying the currencies technique to its conditional credit activities, to give serious consideration to switching these activities to what has proved to be the superior SDR technique and to unify all its operations under a single financing technique.9 This would give members the same asset in exchange for amounts lent through the conditional and unconditional windows of the Fund. It would, like the dollars and other currencies held by central banks, be a marketable asset. Since 1971, U.S. dollars have not been convertible into gold; central banks have nevertheless willingly held them in ever increasing amounts because they can freely be sold for other assets. In the same way, marketability should suffice for central bank holdings of reserve assets in the Fund, provided these assets have a competitive interest rate and controls are in place to ensure that their issuance is not pushed beyond the absorption capacity of the market (the question of control is discussed below).

As the Managing Director said in his statement referred to above, the “monetary character of the Fund” requires “the liquidity and immediate usability of members’ claims on the Fund.” But these two attributes of the claims need not be ensured by their encashability, backed by large balances of idle resources; they can be ensured equally well—probably better—by the assets being exchangeable against currencies—as under the SDR designation system, in the proto-market currently organized by the staff, or in a true market run by the leading commercial banks.

The essentially bookkeeping steps required to transform the Fund to using the SDR as its single financing technique raise no questions of principle; being too technical for the text and too integral to the argument to be banned to an annex, these steps, and some of their implications, are described in Box 3.

Transposition to an SDR Basis of the Fund’s Conditional Credit Operations

This subject will be discussed in two steps: (1) the operation of future Fund credit transactions on an SDR basis, and (2) the transposition of the transactions outstanding at the time of the change in the system.

(1)The conduct in SDRs of future transactions. The technique envisaged is both simple and familiar: it is the same as that of a commercial bank. When the debtor takes up a certain amount of credit, he receives a balance on the books of the bank in the agreed currency. In the case of the Fund, this is the SDR. Credit creation by the Fund thus raises the amount of SDRs in circulation, just as credit creation by a German bank raises the amount of deutsche mark in circulation. Repayment of Fund credit, or the payment of charges, automatically reduces the SDR circulation. The Fund does not “cancel” the SDRs paid to it; the repayment itself reduces the amount outstanding. The debt of the member will no longer be recorded as additional Fund holdings of the currency of the debtor—the SDR value of which would have to be maintained—but as a Fund claim expressed in SDRs, (The Fund already holds such claims for structural adjustment facility (SAF) loans.) In short, the transition would be from an artificial to a natural structure.

(2)The transposition of outstanding transactions into SDRs. Once it had been decided to finance future Fund transactions by SDR entries, it would be desirable to do away with the existing stock of currencies by extending the same technique retroactively to past transactions. Technically, this would be achieved by the Fund “repaying” all quota subscriptions by returning to each member 100 percent of its quota in its own currency; if the Fund’s holdings of its currency were smaller than that amount, it would pay the balance in SDRs—newly created except for the SDRs held by the Fund on the day of transition. In this way, quotas and the corresponding currency holdings would be eliminated from the balance sheet and creditor countries would acquire an amount

equal to their reserve tranche in SDRs. The Fund’s remaining holdings of the currencies of debtor members, which would equal these countries’ use of Fund credit, would appear on the balance sheet as “Fund credit,” the term already in use in the Fund’s statistics. Comparison of Panels A and B of Table 1 below shows how this process of transformation of the Fund’s balance sheet as of April 30, 1995, would work out. Quotas, now shown as a P.M. entry at the bottom of the balance sheet, would continue to determine a wide range of members’ rights and obligations, but they would no longer reflect financial contributions to the institution. Thus, the role of quotas would have become exactly parallel to that in the SDR Account.

Table 1.

Balance Sheet of IMF General Department, April 30, 1995

(In billions of SDRs)

article image
Source: International Monetary Fund, Annual Report, 1995, pp. 138, 231, and 236.Note: Data for the General Resources Account and the Special Disbursement Account have been combined. Also, a number of smaller items have been combined. Totals may not add because of rounding.

The balance sheet presented in Panel B of Table 1 is more informative about the Fund’s finances than that in Panel A. Indeed, the Fund must be the only financial institution in the world whose published balance sheet reveals neither the amount of its outstanding credits nor that of its sight liabilities. To find the former, one has to look in the notes to the published balance sheet; the latter are found only in the text of the Annual Report as part of the discussion of the liquidity ratio.

The transformation from Panel A to Panel B, which is essentially a bookkeeping operation, reduces the balance sheet total by approximately three fourths. It also leads to a presentation of the financial size of the Fund that makes it at least somewhat comparable to that of other financial institutions in the world, although it fails to show among the Fund’s assets its large holdings (about SDR 60 billion) of usable currencies.

To bring the consolidation of the Fund’s two departments to its logical conclusion, one would like to present a single balance sheet for the institution as a whole. Perhaps somewhat surprising, however, the Fund does not publish a balance sheet of the SDR Department (which means that there is also no balance sheet of the Fund as a whole!), but it provides a table called “Statements of Allocations and Holdings” that can be converted into an extremely simple balance sheet format (see Table 2).

Table 2.

Balance Sheet of IMF General Department, April 30, 1995

(In billions of SDRs)

article image

Combining Panel B of Table 1 and Table 2 leads to a balance sheet of an integrated Fund, as in Table 3, in which the SDR 1.0 billion holdings of the General Resources Account shown in Table 2—a claim of the Fund on itself—are netted out against an equal reduction in allocations, from SDR 21.5 billion to SDR 20.5 billion.

Table 3.

Balance Sheet of IMF General Department, April 30, 1995

(In billions of SDRs)

article image

Further Improvements in an SDR-Based Fund

The merger of the two accounts in the Fund, in addition to bringing about much-needed clarity to its finances and considerable simplification to the Articles of Agreement, would also make possible two important improvements in the structure of the Fund.

More Rational Distribution of Costs of the Fund

There is a striking contrast between the ways in which the costs of running the Fund, including the necessary provisioning, are covered in the General and SDR Departments. The expenses of the SDR Department (without any provisioning) are covered by an “assessment” on all participants in proportion to the allocations they have received.10 The costs of the General Department have been covered by a multiplicity of devices, including—in very different proportions over the years—various specific charges on debtors; a premium on the SDR interest rate as part of the periodic charge to debtors; the payment of remuneration to creditors at a rate below the SDR interest rate; and (the most important source in recent years) the nonremuneration of a basic amount, equal to 25 percent of its 1978 quota, for each creditor. Dissatisfied with the inequities and disincentives of the resulting patchwork, Executive Directors searched for a more rational system and reported having reached “a wide consensus… to pursue further the possibility of an amendment of the Articles that would… allow for a quota-based system of sharing the cost of financing the Fund....” (Annual Report, 1995, p. 144). Under such an amendment, members would contribute to the general expenditures of the Fund in proportion to their current quotas, much as they contribute to the expenses of the SDR Department in proportion to the total allocations they have received.

Progress on this matter appears to be stalled on a central political issue: what part of the Fund’s expenses should be covered by a charge proportional to quotas, and what part by a spread between the debit and credit interest rates. In other words, how much of these expenses would be paid by the Fund’s debtors, and how much by its creditors? The new approach also raises a more technical issue—the unfamiliarity with assessment in the General Department and the need for parliamentary approval that its introduction could provoke. Both of these issues appear in a new light once the Fund’s two departments are considered together.

There is an anomaly in the different rates of charge applied to debtor positions in the two departments: in spite of a widely held view in the Fund that conditionality of credit should be associated with some degree of concessionality in its cost, the Fund’s rate of charge for conditional credit (in the General Department) is higher than the rate applying to the use of unconditional credit in the SDR Department. The main historical reason for this lies in the original approach of a rigid separation between the two accounts, which required an exact balance between the interest payments and receipts of the SDR Department. Merging the two departments would remove the need for this anomaly. It would open the possibility of a higher charge on net users of SDRs, which would be expected to encourage early reconstitution and a more balanced distribution of SDR holdings among members. More generally, the introduction into the income structure of the Fund as a whole of additional variables that affect different members in different ways would enlarge the scope for compromises on the distribution of the cost burden.

The novelty that assessment would introduce in the General Department might also appear less formidable if it were noted that in the SDR Department, members are subject not only to an annual assessment of costs—although at a very low rate—but also to an annual interest charge on their net cumulative allocation that is well in excess of anything the Fund might want to collect by way of an assessment to cover administrative expenditures.11

The Changed Function of Quotas in an SDR-Based Fund

Perhaps the principal lesson to be learned from the two sets of negotiations discussed at the beginning of this paper—those that led to the Bretton Woods agreement and those that brought forth the Rio agreement on the SDR—is that the acceptability of a particular technique to finance the Fund will depend crucially on its effects on the membership’s control over the volume of credit extension by the institution.

At present, the Fund’s ability to extend credit is governed by its stock of usable currencies, which in turn is primarily determined by periodic quota reviews. How, then, would the membership exercise control over the total of Fund credit if quota increases no longer served to replenish the Fund’s holdings of currencies? Would quotas still have a function in this context?

Not necessarily: it would be possible to rely simply on the decisions by the Executive Directors to approve individual credits. But this would represent a major loosening of the control that members and their parliaments now have over the financial activities of the institution. There is no reason to expect that a change in financing technique should result in such loss of control. That control could be maintained, in an SDR-based Fund, by instituting a statutory maximum for outstanding SDRs, equal to a certain percentage—presumably 100 percent—of the sum of quotas, plus 100 percent of the sum of net cumulative allocations. Since the present financing technique has given the Fund excessively large quotas for the amount of credit that it wants to extend, this formula would eliminate the need for a general quota increase for one or two decades. Indeed, it might well be considered prudent to include a transitional provision that would, pending a new decision by the Board of Governors, put a lower figure than 100 percent of present quotas on the amount of SDRs allowed to be outstanding as a result of Fund credit operations. Note that a statutory ceiling on outstanding SDRs would preserve the flexibility now existing for the Fund to grant credit by nonmonetary means: the sale of gold (for SDRs) and, more practically, borrowing (again in SDRs).

After the transitional period referred to, the Fund would from time to time require quota increases as world trade, and perhaps especially world capital movements, grew. These quota increases would also provide the opportunity, as they do now, to adjust quotas to changes in the relative economic importance of countries. Parliaments that had to approve quota increases would be aware of what these increases would involve in terms of annual contributions, by way of assessments, to the costs of running the Fund. But quota increases would no longer have to be paid for in SDRs or currencies and thus carry the appearance, though not the reality, of major budgetary commitments.

The Use of SDRs in the Financing of Official Safety Net Mechanisms

Marcello de Cecco and Francesco Giavazzi1

In the aftermath of the Mexican crisis, the IMF decided—at the Interim Committee meeting of October 1995—to enhance its crisis management role by establishing “exceptional procedures (an emergency financing mechanism) that would enable the Fund to respond promptly—with sizable, front-loaded financing, when necessary—to deal with potential Mexico-style crises.” The IMF Managing Director added that the new facility would not necessarily imply exceptional financing: there will be circumstances “when there is tremendously pressing need for action and ample IMF liquidity. This was the case with Mexico. When the IMF faces a liquidity squeeze at the same time as an emergency, then we will have to activate the emergency financing mechanism of the General Arrangements to Borrow (GAB).”

The question addressed in this paper is whether the IMF should be involved in Mexico-style crises and, if the answer is positive, whether the existing IMF financing mechanisms are adequate—in terms of both the volume of funds that can be mobilized and the speed at which they can be mobilized—to support the crisis management role that the Fund would play.

Short History of SDRs

An international monetary system should facilitate, by its working, the exchange of goods and services between different countries, as well as favoring capital transactions. It is generally agreed that in order to do so the system must be based on an internationally accepted money, which can be either issued by a national central bank and used internationally or issued by an international monetary agency like the IMF. For all practical purposes, the dollar, and to a lesser extent the pound sterling and later the deutsche mark and the yen, have been used as international monies. Using the dollar, however—which was and remains the pivot of the international monetary system—by its very success created the so-called Triffin dilemma: since dollars would enter the international payments system as a result of U.S. external deficits, eventually the ratio of gold reserves to dollar liabilities would become so small that confidence in the dollar would be undermined. Alternatively, with a U.S. external surplus, there would be a dollar scarcity. Sufficient liquidity for the international economy would be ensured only by a lucky accident. An international reserve currency not issued by a national central bank, it was believed, would solve the Triffin dilemma. Thus, in the late 1960s, when confidence in the dollar was at its lowest, but the international monetary system was still in its Bretton Woods first edition, with most currencies pegged to the dollar and the dollar pegged to gold, the SDR was created to reduce dependence on the dollar and on other nationally issued reserve currencies. At the Annual Meeting of the IMF in Rio de Janeiro in 1967 it was agreed to establish a new reserve asset administered by the Fund; in 1969, the First Amendment to the Articles of agreement expanded the IMF’s responsibilities in matters concerning international liquidity by authorizing the Fund to create SDRs.

The main purpose of the SDR was to enable the IMF to increase the supply of reserve assets in a timely manner whenever need arose. The SDR was to serve a number of objectives, which the IMF’s Managing Director summarized as follows: (1) to expand international trade, economic activity, and development; (2) to promote a multilateral payments system and the elimination of exchange restrictions; (3) to promote exchange rate stability and orderly exchange rate adjustments; (4) to correct balance of payments disequilibria without resorting to measures destructive of national and international prosperity; and (5) to prevent world economic stagnation and deflation, as well as excess world demand. The new reserve units would be allocated to each member country in proportion to its quota in the IMF. Issues would continue over time in line with the anticipated growth of demand for international reserves. Eventually the role of the dollar and of other national currencies as reserve assets would be replaced by the SDR. SDRs were regulated so as to be transferable only between central banks: they were not made to be used directly. Central banks would use them to acquire from other central banks an equivalent amount of a convertible currency. Recourse to SDRs by individual central banks was moderated by ruling that the use of SDRs beyond 70 percent of cumulative allocations required eventual reconstitution such that, over any period of five consecutive years, the country’s average daily holdings of SDRs would amount to at least 30 percent of its cumulative allocation.

Soon after SDRs were created, however, as worldwide inflation increased in the 1970s the idea of further expanding the stock of international reserves came to be considered the opposite of what was needed—even if such reserves were issued by the IMF and did not represent a debt for any one country. After the events of the second half of 1971, when the dollar’s link with gold was cut, the system of fixed but adjustable exchange rate parities was abandoned. The Fund’s Articles were subsequently amended, and a system of more flexible exchange rates emerged. The new system, however, was soon found to require stabilization by central banks to an extent that was no less than the previous one had required.

A World Money?

In a paper he wrote about fifteen years ago, after the Latin American debt crisis of the early 1980s, Stanley Fischer (1983) examined the behavior of the IMF in the crisis and concluded that, by using its resources and its influence to prevent an international financial crisis from becoming a financial collapse, the IMF had performed the functions of a central bank. He added that, by the First Amendment to its Articles of Agreement and by the creation of the SDR, it had also taken upon itself the task of performing the second function of a modern central bank—that of using its portfolio to stabilize the trade cycle—but that it had not succeeded in performing that function except during the 1982-83 crisis. It had not, however, in the 1982-83 crisis, been the ultimate source of liquidity for the world financial system, Fischer noted. The smallness of its portfolio and the lengthy negotiation required for an increase in the stock of SDRs had prevented it from becoming the lender of last resort. The SDR, Fischer went on to say, remained far from being a world money: it was used as a unit of account by some organizations; official SDRs served as a store of value to governments that held them; and there were some privately issued SDRs—SDR-denominated bonds and bank liabilities. But the SDR was not used as a medium of exchange. It was more comparable to a treasury bill with desirable yield characteristics. Furthermore, there was no link between official holdings of SDRs and private transactions in SDRs. Changes in the stock of official SDRs were changes in the quantities of credit and reserves available to central banks but were not directly changes in the stock of money.

Little has changed since Fischer wrote his paper. In December 1994 a great speculative attack affected the Mexican peso and reverberated throughout Latin America, spreading to currencies whose sale had little fundamental justification. The IMF did play a role in the Mexican crisis, but, as the regulations concerning the issue and use of SDRs have remained unchanged since the 1982-83 crisis, we could come to the same conclusions as Fischer. Nor have the questions he raised about the Fund’s role as a world central bank become less important since he expressed them.

How could SDRs become more like an international money? Quite correctly, Fischer indicated their most important present shortcoming: the fact that no link exists between the circuit where official SDRs circulate—that of central banks—and the private financial circuit—where a small amount of privately created SDR-denominated securities exist. What beleaguers the present system of SDR issue and use is that they never touch the hands of private operators. They were never allowed to become a parallel currency, and they have been valued, since their link to gold was severed, according to the value of a basket of national currencies. The present SDR system is part and parcel of the IMF-centered system, a system of official reserve pooling and recycling meant to replace, for most purposes, the international private market for short-term loans to monetary authorities. Fischer concluded that, at the time of his writing, the difficulties experienced with reaching an international agreement on the further creation of SDRs indicated that the world was not ready for a world money—and this quite apart from the theoretical desirability of a world money, or even of a world exchange standard.

Emergency Finance and Development of a Foreign Exchange Market

Fifteen years later there is little reason to change Fischer’s conclusion, as the world has scarcely become more ready for a world money or a world exchange standard. A more limited question is, however, worth asking: whether it is justified today to advocate a reform mechanism of SDR issue and management that would allow the IMF to contain speculative attacks against individual currencies and to prevent them from spreading to other currencies and inducing an international financial crisis. Is it conceivable to introduce a system of SDR creation and management that would allow the Fund to muster sufficient resources to defend a currency under speculative attack, without having to transform the Fund into a world central bank? Could the Fund, in other words, become an issuer of credit without becoming an issuer of money? Could SDRs be effectively used to allay speculative attacks just by changing the way in which they are issued? To answer these questions we have to examine how the international monetary system has been transformed since the Bretton Woods agreements. When the IMF was created in the 1940s, the dominant concern—shaped by the experience of the collapse of the international trading system in the 1930s—was the reconstruction of an orderly flow of trade among nations. International capital movements were seen as potentially disturbing the orderly flow of trade, thus calling for a system of managed payments among countries, centered around the IMF and the World Bank. At the time, the alternative view of the world, based on free markets for goods and capital, was defeated. But in the last twenty-five years we have witnessed the re-emergence of a market-based international financial system, where the hierarchy of government and central-bank-organized payments has become ever more uncertain, and authorities have to resort more and more to private markets to pursue their aims.

The public circuit of balance of payments adjustment through a centralized clearing of reserves between surplus and deficit countries managed by the IMF has gradually given way to a day-to-day involvement by central banks as participants in the private foreign exchange markets. Central banks borrow from and lend reserves to these markets, just like normal operators. In so doing, they pursue mostly macroeconomic objectives: they use markets to try to keep their own national currencies at parities that they, and their governments, consider “right.” But they know that they cannot impose parities that do not convince at least one half of the market. More than anything else, the duty of a central bank in the foreign exchange market has become that of avoiding that its actions might convince traders that a one-way bet against their currency is possible. In a world of free capital movements, central banks have lost the power to swing a market by brute force—by the impact of their reserves on market transactions. They can only be effective when market sentiment is oscillating and can be swung in one direction by a little nudging. But similar swings can also be produced by strong market participants, who can place bets as large as those of central banks. This was the case with the likes of Mr. Sindona in the 1970s and is the case with hedge funds today. If exchange rates were determined only by fundamentals, the task of the central bank would be clear: it would have to concentrate on the purely domestic task of ensuring that its country had good fundamentals all the time, and speculation would play itself out in the markets, with the ruin of the speculator who bet against the country’s fundamentals. As we know, however, parities and fundamentals only tally in the long run. In the short run, divergence is common. The question then becomes whether the central bank ought to play the markets in order to bet in favor of its own currency, if its fundamentals are good but markets do not seem to appreciate that, and short the currency.

In the major industrial countries, where the national currency market is wide, deep, and resilient, it is reasonable to expect that there will be market makers to keep it orderly at all times except the worst, when excess turbulence may be experienced because of exceptional external circumstances. The central bank will borrow reserves from the market, and, provided it is willing to engage in forward contracts to an unlimited extent, it can hold a parity, against the will of the market, for a limited time if, for instance, it believes that an unfavorable trade balance, which has prompted a speculative attack against the currency, will turn around shortly. It is reasonable to expect large national commercial banks to be natural market makers for their currency. The task of the central bank is to give them a hand when such exceptional circumstances arise, and it may need, in the worst cases, the help of other central banks to do so. In the international financial system that has evolved in the last quarter century, currencies that enjoy well-organized and resilient markets have mostly been defended in this way, with central banks getting together to mount support operations through the Group of Ten at the Bank for International Settlements (BIS), more often than through the IMF.

Most countries, however, cannot rely on a deep market for their currency and are therefore compelled to borrow in other, and more widely marketable, currencies. Their borrowing capacity, however, is not unlimited, and the market knows approximately when the upper limit will be reached. The problem is then to avoid a free fall of the exchange rate, which might stop at totally disruptive levels for that country—and for other countries as well—as the speculative attack can spread to similar currencies. Over the years, however, many more currencies have become convertible and have started being freely negotiated on the markets. This has radically changed the notion of official reserves. Central banks have given up their monopoly of foreign exchange. Foreign exchange is now negotiated directly by banks and nonbanks, including nonfinancial companies and individuals. Anybody can sell the national currency to obtain foreign exchange. If central banks want to establish and maintain a peg, they must sell the national currency to obtain foreign exchange. With the huge arbitrage and speculation flows that characterize today’s financial markets, a peg can only be maintained by inducing capital inflows. The stock of gold and accumulated foreign exchange is a relic of the old system, of very limited use when a currency is subject to a speculative attack. Capital inflows are motivated by interest rate arbitrage and by the expectation of a capital gain on national bonds. As a result, countries have resorted to convertibility, with currency pegging, as an instrument to effect—by inducing capital inflows and thereby reinforcing the credibility of national monetary authorities—internal stabilization, rather than being a measure taken when stabilization has been achieved.

As liberalization of exchange rates spreads wider, following the rapid development of the countries of Asia and Latin America and, more recently, the return to the fold of the countries of Central and Eastern Europe, the problem has arisen of how to regulate the international exchanges in the transition period between the convertibility of a currency and its negotiation in a well-organized, deep, and resilient market. Only when confronting a foreign exchange market for its currency that is recently established, lacks proper market makers, is not backed by a well-structured forward market, and therefore shows dangerously high volatility will a central bank feel the inadequacy of its reserves for the task of stabilizing the market, even if they are ample compared with those of developed countries’ central banks. The central bank will also fear that a speculative attack might spread contagion to financial assets denominated in the national currency, so that the currency loses its function as a store of value domestically and dollarization ensues. It has often been argued that exchange controls ought not to be abandoned before the fundamentals of a country are properly stabilized. But if exchange controls persist, it is very difficult to build up a market for the country’s own currency that will exhibit the necessary breadth, depth, and resilience, and local banks will not learn to work in free financial markets. Authorities have to take the plunge at some stage and face the consequences. Otherwise, they are in a vicious circle, where the fear of an undeveloped foreign exchange market prevents full liberalization and perpetuates the situation. This is where international assistance is needed. While a rapidly expanding country is developing the free external market for its currency, speculation might force it to abandon the effort and to return to controls. Alternatively, speculation might force its exchange rate to levels that are inappropriate for its own domestic stability, or for the stability of its trading partners, or for the stability of the international financial system. Is there a role for the IMF in helping quickly developing countries to create large and resilient markets for their currencies? Would SDRs issued and administered according to rules different from those presently applying be capable of alleviating the difficulties arising from such situations? Or should the IMF think of other ways to achieve the same end?

Present SDRs

SDRs, in their present form, are hardly suited to look after these problems, although, as we saw above, at the time of the First Amendment to the Articles of Agreement, the then Managing Director of the IMF considered a regular new supply of them necessary to bring orderliness to foreign exchange markets. The fact is that the mechanism to issue SDRs has no flexibility. They can be created by a decision of the IMF Board of Governors, but once created they remain in the system and are a once-and-for-all addition to its reserves, although there is a procedure for cancellation. SDRs represent unconditional finance by the IMF. They are, by statute, issued to every member according to its quota and, after many changes, have been given a valuation and interest rates that make them desirable to holders. The Fund has also organized a sort of interbank market for them. However, the crucial problem concerning their potential role in stabilizing exchange markets has not been solved. SDRs remain to this day enclosed in the “public circuit” of international reserves. They are not usable on normal private currency markets, nor are they used in IMF conditional financing. They are created as a result of a cumbersome procedure, and once they are there, they remain there. If they were to be used to stabilize exchange markets, they would have to be created by a much more elastic procedure and be used as a part of the Fund’s conditional financing. Once they had served their purpose in one particular instance, a way would have to be found to bring them back to the Fund so that the fears of permanently inflating the world’s liquidity by their addition would be allayed.

To be approved, an SDR allocation needs an 85 percent majority of Fund quotas. Special allocations are not possible under the Articles of Agreement and would require an amendment approved and ratified by countries holding at least 85 percent of the voting power in the Fund. The debate in favor or against new SDR allocations has in fact concentrated on their characteristic of being a net addition to, rather than a change in the composition of, total international reserves. The question of whether SDRs ought to be used to foster international economic development has also been discussed. The debate has often turned into one about inflationary finance, and developed countries have argued that a new allocation of SDRs is unnecessary because there is no long-term global need to supplement international reserves. Hans Tietmeyer, President of the Deutsche Bundesbank, has observed, however, that there are good reasons for being in favor of enabling the Fund’s new members to participate fully in the system through a special allocation of SDRs that would ensure the equitable participation of all members in the mechanism. But even this concession by the staunchest opponent of new general allocations of SDRs is catering to a special need: that of fairness to new members of the IMF, that is, to the former socialist countries of Central and Eastern Europe.

The use of SDRs to fight speculative attacks against individual currencies, preventing such attacks from spreading to other currencies, would require facing the issue of whether a centralized strategy of containment of speculative attacks is preferable to one that relies on voluntary cooperation among central banks in the foreign exchange market to achieve the same end. If we can satisfy ourselves about the superiority of the former, we have to provide convincing arguments about the superiority of the IMF as the protagonist of the centralized strategy over, for instance, the Federal Reserve, or a joint effort by the U.S. Treasury and the Federal Reserve. If we manage to surmount this first hurdle, we have to examine ways of reforming the system of SDR creation and allocation, so that it may serve efficiently as an instrument to fight speculative attacks.

Currency Crises and Need for IMF Financing

There are typically two types of currency crisis: the crisis that is unjustified by fundamentals and is thus very similar to a “bank run,” and the crisis that results from the buildup of an unsustainable macroeconomic position. The two are, of course, difficult to distinguish: the timing of an attack often coincides with the moment when speculators change opinion on the sustainability of a country’s exchange rate. It is only the government and the central bank that call the attack “unjustified,” denying that current policies are unsustainable. Moreover, as we shall discuss, an attack may become self-fulfilling if speculators anticipate that the authorities will change their policies as a result of the attack. Thus, the difference is between devaluations that are unavoidable because the external position of the country cannot be sustained, and devaluations that become unavoidable the moment that a speculative attack occurs.

Self-Fulfilling Attacks: Sunspots and Other Attacks Unjustified by Fundamentals

As with a bank, a country can be subject to a speculative attack on its currency even if its exchange rate policy—a fixed parity vis-à-vis another currency, or a basket of currencies, or a predetermined exchange rate path, as in a crawling peg system—could be sustained indefinitely in the absence of an attack. However, if an attack occurs, it could be self-fulfilling, that is, it could lead to the country being forced to abandon its exchange rate policy. Obstfeld (1996) describes a simple and illuminating example of a self-fulfilling attack. Two traders and a central bank are engaged in defending a parity that, without an attack, could be sustained indefinitely. When does an attack result in a devaluation? There are two polar cases: in one, the reserves of the central bank are large enough to resist an attack staged simultaneously by the two traders; in the other, the volume of reserves is so small that the central bank gives in even if only one trader attacks it. These situations correspond to the traditional model of speculative attacks (see, for example, Krugman, 1979): fundamentals are either consistent with the peg, or they are not. But there is a third possibility: the reserves of the central bank are large enough to resist an attack staged by one trader, but not one by the two simultaneously. In this case reserves are neither so large as to make a successful attack impossible, nor so small as to make it inevitable. An attack succeeds if the two speculators coordinate their moves; otherwise the exchange rate parity survives.2

Another case is that of an attack that becomes self-fulfilling because fundamentals change as a result of the attack; that is, speculators anticipate that when the attack materializes, the authorities will change domestic policies, and that the new policies will be inconsistent with the pre-attack parity (on this point, see Obstfeld, 1994). Often, policies change because the attack causes a sharp increase in nominal interest rates that the authorities are not prepared to bear, because of the level and maturity structure of the public debt, or, as in the U.K. experience of September 1992, because of the effect of the increase in interest rates on the cost of households’ mortgages. It thus becomes difficult to distinguish attacks that are “unjustified by fundamentals” from those that are not. In the end all attacks are justified ex post. The distinction should be between devaluations that are unavoidable given current policies, and devaluations that become unavoidable if the country is subject to an attack.

There are obvious arguments for avoiding a devaluation that is not justified by current policies. The strongest is the risk of contagion through an international spillover of the devaluation (for a model of contagion, see Gerlach and Smets, 1994). The country whose currency has been successfully attacked experiences a sudden improvement in competitiveness at the expense of its trading partners, which have an incentive to restore competitiveness through a matching devaluation (alternatively, the anticipation of such an incentive may be sufficient to trigger another attack).3

How can self-fulfilling attacks be ruled out? The central bank must convince speculators that its ability to defend the parity is beyond doubt, either because it directly holds a sufficient amount of reserves, or because other central banks are prepared to line up sufficient ammunition. Since the central bank itself is unlikely to have enough reserves to resist an attack,4 the ability to borrow from the market or from other central banks through bilateral swap arrangements is crucial: in case of need a country under attack must be able to draw upon its credit lines up to the point at which speculators will withdraw from the market. If the commitment to extend unlimited credit lines to the country under attack is credible, such credit lines are unlikely ever to be used.5

Can individual central banks arrange such credit lines directly, or should an international agency intervene and provide a coordinating mechanism? In particular, is there a role for the Fund in avoiding self fulfilling attacks? The European Monetary System (EMS) is one example of an exchange rate arrangement in which the response of central banks to an attack on a member currency is coordinated. The instrument through which the response is coordinated is the very short-term financing facility (VSTFF) administered by the European Monetary Cooperation Fund. The facility is a mechanism through which two central banks whose bilateral exchange rate has reached the limit of the fluctuation band automatically extend unlimited credit one to the other—the relatively stronger to the relatively weaker currency. The credit line is denominated in ECUs and is unbounded: in the EMS jargon it is called “marginal intervention,” since it only applies at the outer limits of the fluctuation bands.6 The automaticity of the VSTFF is the feature that eventually brought the exchange rate mechanism (ERM) to an end. In September 1992 the Bundesbank felt that it was unable to fully sterilize the effect of domestic liquidity of its compulsory exchange market interventions—particularly in defense of the lira and of the pound sterling. Thus it stopped complying with the formal rules of the EMS. The VSTFF suffers, however, from another weakness that would be particularly relevant if one was to consider this facility a model for the design of new safety net mechanisms. The VSTFF is not a mechanism to redistribute risk across different lending countries. The credit lines extended through the VSTFF are bilateral. It is true that if a currency was to reach its fluctuation margins vis-à-vis more currencies simultaneously, all the corresponding central banks would be asked to extend credit lines through the VSTFF; but the ERM parity grid is such that this rarely happens. Normally a member currency hits its margin vis-à-vis a single currency—typically the deutsche mark—and it is thus a single central bank—typically the Bundesbank—that is called to extend credit via the VSTFF. As we shall see in the next section, risk diversification should instead be an important feature of any international safety net mechanism designed to defend currencies from speculative attack.

In countries in which financial markets are well developed and which belong to integrated financial areas, central banks can use the markets to diversify the risk associated with intervention. In Europe, in September-October 1992, weak-currency countries borrowed a total of $30 billion from the markets (see International Monetary Fund, 1993). But this is often not an option for countries in which the financial market is thin or which are not part of an integrated financial area—for example, in Latin America or in Eastern and Central Europe. When an exchange rate crisis develops, central banks in these countries would typically call up the Federal Reserve, or the Bundesbank, depending on their location in the world. Risk would not be diversified, and a single central bank may be unwilling to commit enough liquidity, or—what is enough to stage an attack—speculators may anticipate that it is unwilling to provide enough liquidity. In such situations an international agency could serve two functions: it could arrange the credit lines by coordinating other central banks on a good equilibrium; and it could do so by spreading the risk across them. A timely response is essential. To avoid the risk that a delay may result in the attack fulfilling itself, the agency should be able to provide the needed liquidity fast, and in an amount that is large enough to corner the speculators. The agency must line up true ammunition, that is, reserves that in case of need could be used in the market. Otherwise its commitment would not be credible. If the attack is truly unjustified by fundamentals, such reserves are unlikely ever to be used.

Financing Unsustainable Policies

Speculative attacks unjustified by the pre-attack fundamentals are rare. As mentioned above, the timing of an attack often coincides with a change in market sentiment about a country’s exchange rate; it is the government and the central bank that call the attack “unjustified,” denying that current policies are unsustainable. Often this happens when the real exchange rate has become overvalued, for instance, because the government has used the exchange rate as the nominal anchor for bringing down inflation, but prices have kept rising, albeit at a slower pace. At this time it is common to listen to government officials pointing to “structural change” that will eventually make the current parity consistent with a sustainable current account. In Mexico in 1994 it was the anticipation of the effects of NAFTA; in Italy in 1992 it was the structural changes expected to take place in the service sector that should have produced more competition and lower absolute prices. Meanwhile the current account deficit grows larger, easily financed by capital inflows betting on the “convergence game.” Then, all at once, market sentiment changes, and the central bank wakes up to the reality that most capital inflows have come in the form of short-term deposits that can be withdrawn overnight.

In such cases the devaluation is welcome and often leaves more winners than losers. The losers are the central bank and those domestic residents who had borrowed in foreign currency without asking why it was so much cheaper compared with domestic currency loans. Wealth is reallocated from these borrowers to those savers—often the very same individuals—who had shifted their wealth abroad anticipating the devaluation. Real wages fall, but the sharp increase in international competitiveness is good news for industry and for employment as well. The European experience following the September 1992 devaluations—but the Mexican experience as well—indicates that the real effects of the devaluation can be long lasting. In Italy 42 months after the devaluation competitiveness is 30 percent above its level in August 1992, the month before the devaluation; in Mexico 15 months have eroded only 20 percent of the competitiveness gain produced by the devaluation of December 1994.

In such a situation any amount of money spent to defend the parity is wasted, as the European crisis of September 1992 demonstrates. International agencies should stay clear of intervention directed at defending unsustainable parities. Instead they should concentrate on “surveillance,” that is, on convincing governments that the longer they stick to the unsustainable exchange rate parity, the smaller are the chances that a soft adjustment can be engineered. For example, in a program designed to bring down inflation from very high levels, the exchange rate is a very useful anchor in the first stages of the plan: the question is how to avoid the buildup of a large overvaluation. Bruno (1995) convincingly, given his direct experience, argues that “the main question is not whether stabilization by means of the exchange rate is good or bad policy per se, but what the optimal time of exit is,” And, we would add, what the accompanying policies should be at the time when the country abandons the peg.

Creating the Conditions for an “Orderly” Devaluation

Not all devaluations are mostly good news—and Mexico stands in sharp contrast to the European experience of 1992 and to that of Italy in particular. Why did the Mexican devaluation leave mostly losers, and why was the IMF right to intervene? The main difference between Italy and Mexico was the currency denomination of the public debt. In Italy, where the debt is almost entirely denominated in lire, the devaluation had no effect on the liquidity position of the Government. In Mexico, on the contrary, by December 1994, over 75 percent of the debt was denominated in dollars and was mostly of very short maturity (Tesobonos). The devaluation created a short-run liquidity crisis, and debt default became a serious possibility: the financial and real disruptions associated with a default would have vastly offset the direct expansionary effects of the devaluation.

In such a situation, providing the country with enough foreign exchange is the only way to avoid the liquidity crisis turning into a default. Debt default would mean further depreciation, well beyond that needed to restore competitiveness, and would thus open up the channel for international contagion. Contrary to what happens during a speculative attack, the amount of resources needed to rule out a default is not “unbounded”: it is equal to the liquidity needs of the treasury—and of the banking system to the extent that its short-run liabilities are also denominated in foreign currency. Since intervention is directed at overcoming a liquidity crisis, it should be made clear that the foreign exchange provided is temporary and must be reimbursed as soon as the current account turns around.

The concern for an orderly devaluation—one that avoids an overshooting of the exchange rate—is more general than in the example described above. Contagion can occur not only through an international spillover of the devaluation but also within the domestic financial market, if the devaluation induces a shift into foreign assets of portions of domestic wealth that normally would not be traded internationally. Typically only a subset of the stock of a country’s wealth is traded internationally. As shown by French and Poterba (1991), among others, individual portfolios are heavily specialized in domestic securities, even more so if one considers the value of houses that in many countries represent the most important component of private wealth. The wealth allocation decision by many individual investors, however, may be characterized by a nonlinearity. As long as the expected devaluation remains “reasonable”—taking, for example, some measure of purchasing power parity as the benchmark—domestic residents will not diversify their wealth—their housing wealth in particular—internationally, even if the expected return from holding domestic assets does not increase one-to-one with the expected devaluation. Thus, within this range, they behave differently from international financial arbitrageurs. (Transaction costs may be enough to explain this behavior.) However, if the exchange rate embarks on a “free fall,” even those domestic residents who so far have not behaved as the arbitrageurs will start shifting their wealth from domestic to foreign assets—they may even start to sell their houses or to mortgage them. When they do so the amount of money that comes to the foreign exchange market can be of some orders of magnitude larger than that mobilized by speculators. The central bank could in principle resist by raising the home interest rate sufficiently, but this would amount to the disappearance of the home currency and to dollarization of the economy.

The concern for domestic contagion explains why central banks often spend large amounts of reserves to “accompany” a devaluation to ensure that it happens in an orderly manner and that the exchange rate does not overshoot. Moreover, since domestic contagion, by accelerating the devaluation, can produce international contagion, it provides one more argument for external intervention aimed at helping central banks carry out orderly devaluations.

Proposal for Reforming IMF Financing Mechanisms

We argued above that making emergency credit lines available to a central bank is justified in two situations: to help it resist a devaluation that is unwarranted by current policies and to allow it to carry out an orderly devaluation when one is unavoidable. Who should provide such credit lines? An individual central bank, the Federal Reserve, for instance; a club of central banks, such as the BIS; or the IMF? Extending an emergency credit line always involves some risk: in the end the devaluation may not be avoided, for instance, because it was not really unjustified by fundamentals; domestic contagion and international spillovers may not be avoided. Emergency credit lines should be designed to maximize the diversification of such risk.

In the Mexican experience, the substantial role played by the Federal Reserve in extending credit lines to the Bank of Mexico weakened the dollar, as markets assessed the risk that the U.S. central bank was taking and guessed that it would refrain from raising interest rates for some time. In Europe, one of the arguments—perhaps the only economically valid one—for the Maastricht fiscal criteria is the risk that the consequences on the financial market—and on the payment system in particular—of debt default in one country could force the European central bank to intervene, thus deviating from its monetary targets.

An institution already exists that helps diversify such risks: the BIS, which was also active during the Mexican crisis. The Fund, however, could provide a superior risk-sharing mechanism by spreading the risk across all its members that are creditor countries—including, for instance, Korea, which is not a member of the BIS.

But the existing IMF financing mechanisms are ill-suited to play such a role. The Articles of Agreement authorize the Fund to decide the sources, timing, magnitude, terms, maturity, and techniques of its borrowing. The Fund is permitted to borrow currencies from any source, including private ones. But so far it has borrowed only from official sources. Until the mid-1970s, it borrowed only from industrial countries. After the first oil crisis it resorted to borrowing from oil producers. When Japan became the main surplus country, in 1986 the Fund concluded a borrowing arrangement with that country. The problem with all borrowing agreements is not that they require lengthy negotiations with lenders. That is only to be expected. The problem is that once the agreement is reached, the resources thus obtained are pooled and managed in the General Resources Account and used for stand-by arrangements, that is, for conditional lending. Procedures for stand-by arrangements take time, and can hardly be accelerated. This contrasts with the number one requirement of crisis intervention—speed.

Thus the problem with current procedures is that the Fund’s General Resources Account can be replenished independently of the outbreak of a crisis but cannot be used in the very short time necessary to allay a speculative run on a currency. It is therefore not particularly useful to ask lenders to increase the existing credit lines they have with the Fund if the resources thus obtained cannot be used for the purpose mentioned to justify the increase. The same applies to quota increases or to any other call for adding to the resources available to the Fund, motivated by the need to fight speculative attacks.

How could the IMF financing mechanisms be changed? Any plan should satisfy three characteristics: (1) speed, because crisis intervention cannot wait; (2) risk redistribution, because speed will always come at the cost of higher risk; this needs to be redistributed across lending countries, and the justification for IMF intervention lies in the ability of this institution, compared with other international agencies, to redistribute risk across a large number of potential lenders; and (3) temporary creation of liquidity, because a mechanism designed for crisis management should not produce a permanent increase in international liquidity. A possible new mechanism based on the SDR could thus have the following features:

  • The Fund should be authorized to issue SDRs in unlimited quantity when it decides that intervention is necessary and justified, according to one of the criteria described above. In principle, a resolution to this effect should be taken by the IMF Executive Board by simple majority. But the Board is unlikely to be able to come to a decision fast enough for intervention to be effective. One option is to give a mandate to the Managing Director, who should take the responsibility for deciding when intervention is justified. Shifting the responsibility to the Managing Director of the Fund is a way to solve the trade-off between speed and risk diversification that lies at the core of any safety net mechanism.

  • The borrowing country would obtain intervention currencies by exchanging the SDRs received by the Fund, for instance, with the Federal Reserve. The Federal Reserve would thus be in a better position relative to the situation it would find itself in if it had unilaterally extended a credit line to the country under attack: its credit is vis-à-vis the IMF, not that individual country.

  • The Federal Reserve—to continue with the example—must be authorized to exchange the SDRs thus received with the central banks of other creditor countries that are members of the IMF. These transactions should take place pro quota, that is, it should not be allowed to exchange the SDRs only with the Bundesbank or the Bank of Japan. The quotas could be either the existing Fund quotas or some more representative allocation.

  • This mechanism has now spread the risk associated with intervention across all creditor countries, rather than concentrating it in a single country—the United States in our example—or a subset of creditors, as would normally be the case if diversification had been arranged through the BIS. This mechanism, moreover, by spreading the risk across a larger number of countries, reduces the need to support the credit lines with real guarantees because, in case of default, the effects would not be as damaging as if the risk was concentrated in a single country.

  • As soon as the borrowing country has overcome the crisis, and devaluation either has been avoided or has occurred in an orderly fashion, its central bank starts repaying the loan to the Fund—in dollars or another reserve currency—and such installments are immediately used by the Fund to cancel the SDRs that it had originally created and that have been spread across all its members that are creditor countries.

Concluding Remarks

The mechanism suggested above would satisfy the three requirements that we have indicated: speed, risk redistribution, and temporary creation of liquidity.7 However, it cannot avoid the intrinsic risk associated with the creation of a safety net mechanism: the availability at the IMF of an emergency financing mechanism that can be rapidly mobilized increases the pool of accessible reserves—even more so if such reserves can be mobilized in a way that spreads the risk across many countries. This creates a moral hazard problem that could result in needed policy adjustments being delayed. As indicated in the IMF (1993) study of the ERM crisis, there is no easy solution to this tradeoff. The experience of currency crises and collapses—as surveyed for instance in Dornbusch, Goldfajn, and Valdés (1995)—suggests that while a safety net mechanism that allows a country to carry out an orderly devaluation is always desirable, activating such a mechanism to help a country postpone a devaluation that the authorities consider “unjustified by fundamentals” is much more risky and should thus be considered by the authorities with much caution.

The Potential Role of the SDR in Improving the Reserve System

ZHU Xiaohua1

There seems to be a puzzle in making institutional arrangements: we set up rules to regulate, promote, and assist in the progress and growth of the international economy, in the hope of maintaining orderly developments. Yet we have found it difficult to do so. Changes often take place in such a drastic and unpredictable manner that we either have to admit exceptions to our rules or end up adapting our rules to these developments. The past fifty years have witnessed sufficient evidence of this paradox. The establishment of the Bretton Woods system was followed by a chain of events that only weakened it—the dollar glut, the closing of the U.S. gold window, and the floating of major exchange rates, all of which conspired to bring about the amendment of the Articles of Agreement of the Fund. Similarly, countries made a commitment in the Second Amendment of the Articles of Agreement to make the SDR the “principal reserve asset in the international monetary system” (Article XVIII), but later developments in the world economy seemed to have only drawn us further away from this pledge.

It is hard, if not impossible, to explain this puzzle, even with the benefit of hindsight. But if we can learn anything from the past, it is that our otherwise deeper insights are sometimes blurred by such factors as national interests and political will. It seems to me that a prudent start, firm commitment to our agreed goals, and a spirit of compromise are essential in making institutional efforts.

Critical Evaluation of Current Reserve System

If we dwell on the future, an assessment of the current system seems in order. The current international monetary system, characterized by floating exchange rates, national autonomy in adjustment of external balances, and a mixed international reserve mechanism, is described as a “nonsystem” (de Vries, 1976, and Williamson, 1977), Within this nonsystem, the component that governs reserve creation and distribution is by no means smoothly functioning and well organized. Three features are distinctive in this context, namely, the predominance of holdings of national currencies in liquid international reserves, the heavy reliance on cross-border credits in creating these reserves, and the absence of a mechanism to control the supply.

Experience and history show that whether the world operates under a system of flexible or fixed exchange rates, the international community seeks to establish a reserve system that has some inherently desirable characteristics. These include the ability of the system to inspire confidence, to foster financial stability, and to provide adequate and inexpensive reserves in an equitable manner. Based on these criteria, I find the functioning of the current reserve system unsatisfactory. I give my reasoning below, under the traditional headings of asymmetry, adequacy, confidence, and stability, to which I shall try to bring some new sense.

Asymmetry in Reserve Creation

Asymmetry in reserve creation and balance of payment adjustment was found to be a major flaw in the gold exchange system. The reserve currency country was able to finance its payment deficits through the issuance of liabilities, while nonreserve currency countries had to settle any such deficits by asset transfers. This aspect of asymmetry has not vanished despite changes that have taken place in the international monetary system. What is different is that the number of reserve currencies has now increased and the advantage of reserve creation is enjoyed by a few reserve currency countries, not the United States alone.

A more fundamental defect of the current reserve system with respect to asymmetry in reserve creation, however, lies in the very nature of the prevailing reserve creation mechanism that the bulk of reserves are now provided by the market. Some countries can now increase their foreign exchange reserves easily, either by borrowing or by purchasing reserve currencies with national currency. Other countries, however, find their ability to obtain international reserves extremely limited. They cannot conduct currency exchanges because of the absence of an international market for their national currencies; they also find it difficult to borrow reserves because their access to the credit market is blocked by their perceived lack of creditworthiness.

Asymmetry therefore exists largely between creditworthy and non-creditworthy countries. Creditworthy countries can borrow at a normal market interest rate, which can be very low for countries with good credit standings. Those without access to the market, however, can only acquire reserve currencies either by obtaining conditional credits from the Fund or by generating a surplus in their external accounts, usually through costly adjustment efforts. This adds to their already straitened economic conditions and high costs. Creditworthy countries, on the other hand, do not induce identical costs in acquiring reserves. Borrowing rates differ because they reflect the different credit ratings of countries. Generally the major industrial countries have the lowest borrowing rates, whereas developing countries have to pay higher risk premiums or in some cases penalty rates to obtain the necessary reserves.

So creditworthiness makes the difference. It is argued that this is an economic law and there is no reason for an exception to be made when the borrowers are governments or central banks (see, for example, Schroder, 1990). Creditworthiness acts as a constraint for countries to bring their houses into good economic order.

Profit-driven private lenders would therefore have to evaluate the creditworthiness of countries case by case and in an objective and equitable manner so that countries’ credit ratings can fully reflect both their conduct of policy and improvements in payment positions. But this may well not be so. Market decisions are influenced by a number of factors, which include the conditions, expectations, and sentiment of the market itself, a country’s stage of development, location, and even political and cultural elements. Private lenders are also strongly influenced by the policies and regulations of their governments, which themselves are borrowers of reserves.

The effectiveness of the credit constraint is also questionable. First, this constraint is not automatic. Although a country may not want to follow policies that could jeopardize its credit standing, there is no guarantee that it will pursue the necessary adjustment policies. The constraint is especially loosened when a country can manage to service its debt by means other than policy conduct, such as by issuing more debt. On the other hand, if a non-creditworthy country has pursued proper policies to improve its balance of payments position, there is no guarantee that its credit ranking will go up. As an example, after the debt crisis in the 1980s, developing countries often found it difficult to restore their credit ratings despite all the improvements they had made.2 Second, this constraint is also not symmetric. While it may impose pressures on non-creditworthy countries to adjust their policies by keeping them out of the market, it does less to affect the policies of creditworthy countries, especially the major industrial countries.

Inadequate Reserve Supply

Can the private market that produces the bulk of reserves function smoothly enough to provide a stable and ample source of liquidity for the world? Some features of the private credit market have evoked extensive study and debate. These include unwise lending and borrowing, inadequate insurance (see, for example, Corden, 1983), volatility and uncertainty, and overshooting. These features imply that the terms and conditions of the market could undergo sizable and dramatic changes that will affect the global availability of reserves. In hindsight, we can see that the market can go so far as to expand its lending excessively (the late 1970s and early 1980s) and also can lag substantially behind the growth of the world demand for liquidity (mid-1980s). The recent crisis in Mexico again reminds us that borrowed reserves are too vulnerable to market sentiment.

The issue of reserve adequacy also reflects the problem of asymmetry. Some countries have a virtually unlimited source of borrowed reserves, but at the same time others are faced with a critical shortage of liquidity. It is true that pure efficiency considerations disregard distribution effects, but it is also true that efficiency can be achieved by changes in distribution. The relationship between reserve adequacy and distribution was well summarized in the IMF’s Annual Report, 1966:

The economic significance both of reserve stocks and of reserve growth depends on their distribution among countries. If the distribution of the stock is very uneven in relation to need, as it was at the end of World War II, even a relatively high level of global reserves may be insufficient to serve the world’s needs. In general, any improvement in the distribution of reserves is likely to have much the same effect on the world economy as an increase in total reserves.

Fluctuations in market conditions and the uneven distribution of reserves combine to highlight the inadequacy issue. Inadequate reserve supply could jeopardize growth and structural adjustments of both developed and developing countries and adversely affect the global economy.

Lack of Confidence in Reserve Mediums

Lack of confidence in a reserve medium is indicated by shifting to other assets. The only way in which confidence can be guaranteed is if there is only one reserve asset, which, for example, could be gold, or a single currency, or a fiduciary asset like the SDR. A less powerful way to keep confidence, in particular in a multireserve system, is to maintain relatively stable exchange rates between different reserve assets.

One of the prominent features of the current monetary system is the floating of major exchange rates. Although countries frequently intervene in the foreign exchange market, the exchange rates between major reserve currencies are far from being stable. This exposes countries holding reserves to constant exchange risks, which add to the costs of maintaining reserves. Countries are compelled to exercise portfolio choices to spread risks and to reduce costs, which in turn raises transaction costs.

The confidence issue is made more complex because the exchange market itself is influenced by a number of factors, especially the macroeconomic policies of reserve currency countries and the changing market perceptions to these policies. The value of international reserves is therefore largely affected by the policies of major currency countries. The inconsistency of these policies leads to fluctuations in exchange rates, interest rates, and prices, which combine to impose burdens on countries holding reserves and to impair their adjustment efforts.

Systematic Instability

A desirable reserve system should help promote sustainable growth with price stability. As a minimum this requires that the reserve asset be produced without causing a monetary shock. And it would be ideal if the creation of reserves were countercyclical, so that creation increased in recession and restricted in boom times. The current reserve system falls short of these desirabilities. First, there is a lack of effective international control over global liquidity, and the private market can experience periods of excessive expansion and contraction. Second, the inconsistency of policies of reserve currency countries causes frequent fluctuations in various economic variables. Third, the private market is more likely to be cyclical than countercyclical, given the intensive interaction between the capital market and real sectors of the economy.

Another weakness concerns new developments in the capital market. Two distinct developments occurred during the 1990s: the dramatic increase in the size and volatility of cross-border capital movements and the continuing innovation and deregulation in the international financial markets. These made individual countries extremely vulnerable to shifts in market sentiment and surges in capital flows and to subsequent spillover effects. The possibility of spillover of crisis indicates a systemic weakness. Unfortunately, no international mechanism is yet in existence to confront these dangers.

Potential Role of the SDR

Can an SDR mechanism help overcome the weakness of the multicurrency reserve system? History seems to provide negative answers. It was in the company of the SDR that the multicurrency system weighed its anchor and started possessing all the merits and demerits mentioned. However, it would be arbitrary to conclude that the SDR is useless for tackling the problems of this system. It is undeniable that we did too little to try to achieve such a goal. We have failed to reach agreement on further allocations since 1981, we have failed to put the substitution account into effect, and we have even failed to reach agreement on enhancing the SDR as a reserve asset. It is hard to believe that an international reserve, with a share of less than 3 percent in global reserves and a quality that evokes constant criticism, can fulfill the mission of becoming “the principal reserve asset” and of eliminating the several drawbacks of the multicurrency system.

The SDR is not a stereotype whose usefulness has been ruled out by developments in the international monetary system. The floating of exchange rates did not remove the role of reserves in external adjustment, especially in the short term. Few countries allow clean floating to take care of their balance of payments adjustments, and evidence shows that the accumulation of nongold reserves has grown over time. If there is a case for reserves, I see no reason why there is not a case for an international reserve asset like the SDR. The booms in the financial market did not eliminate the justification for official asset settlements, and hence for settlements through the SDR. The recent crisis in Mexico and its spillover effects in other emerging markets have undoubtedly highlighted the qualitative differences between borrowed reserves and owned reserves. Creditworthy countries that generate their reserves through the credit market are highly vulnerable to market surges. Expanding the holdings of owned reserves like the SDR will improve the quality of global reserves and enhance systemic stability.

A long-term scenario for the reserve role of the SDR is also being discussed (see Schroder, 1990). If a national currency is to act as the principal international currency, this country must be less open so that it is immune to world market instabilities and can maintain stable monetary and economic policies. On the other hand, the foreign economic relations and nontradable sector of this country must be large enough to satisfy other countries’ need for liquidity, which requires that it has a large share of world GNP but a relatively small foreign trade sector. The United States best fulfilled these conditions in the postwar period, but no country is up to these criteria today and in the foreseeable future. Therefore, a fiduciary asset like the SDR will become a natural solution.

The question now is what changes should be brought to the operation and character of the SDR, so that this reserve asset can better adapt to its environment.

Weaknesses of the Present SDR Mechanism

The present SDR mechanism does not function as the international community had expected. Its shortcomings have given rise to considerable debate. One weakness of the SDR concerns valuation. In the present method of valuation, the SDR’s value is determined by the value of a basket of currencies. The purchasing power of the SDR fluctuates with the weighted average purchasing power of the component currencies of the basket. Thus the value of the SDR is relatively stable compared with any of the five currencies in the basket. It has been hoped that for this reason countries would readily accept the SDR as a major unit of account and store of value, so that if the holding of SDRs grows over time, the overall quality of global reserves will be gradually improved.

Evidence shows that the SDR is not in effect being widely used as either a unit of account or a store of value. A straightforward reason is that every country can now devise its own basket of currencies just as the IMF devises its SDR. A self-made basket not only has the same value as the SDR, but it is also more flexible. It can be used directly as a medium of exchange and for carrying out arbitrage transactions. It is also argued that the value of the SDR is not stable in real terms because the five component currencies are exposed to erosion by inflation, which is affected by diversified economic policies in these five countries (see Coats, 1990).

The second weakness of the SDR is its inability to be used in market transactions. Almost all SDR transactions are carried out through transactions with the Fund, and the scope for private use is extremely limited. The SDR therefore cannot, like other assets, be used directly by countries to intervene in the foreign exchange markets—a major defect for a reserve asset.

The third weakness has to do with its interest rate. The interest rate on the SDR is a weighted average of interest rates on selected short-term monetary instruments in the money market. Although it is in fact a short-term rate, there is no time limit on the use of SDRs. A criticism is that countries receive long-term loans and pay short-term interest rates when using their allocated SDRs (see Corden, 1983, and Schroder, 1990), and buyers of SDRs have to invest in a long-term asset for which they receive short-term interest. As a result, there is a substantial disequilibrium between the demand for and supply of SDRs, that is, a disparity between the need to use allocated SDRs and the willingness to buy more SDRs. The perception is that to avoid piling up too many SDRs, some net buyers of SDRs choose to protect themselves by opposing any further allocations (Schroder, 1990).

The fourth shortcoming of the SDR mechanism lies in its procedures for allocation and cancellation. Almost every proposal for allocation evoked futile debates over the definition of “long-term global need for reserves” and reached an impasse. Not enough attention has been paid to the uneven distribution of global reserves and the asymmetry in reserve creation. The allocation and cancellation procedures also lack enough flexibility to reflect changes in the terms and conditions of the capital market that makes the bulk of reserves available.

To establish a strong and efficient SDR mechanism, improvements should be made both quantitatively and qualitatively in its operation and character. Quantitative improvements concern allocation and the proposed postallocation redistribution schemes. Only if the SDR is of substantial quantity and even distribution can it help tackle the inadequacy and asymmetry problems. A well-controlled SDR mechanism can also be used to maintain stability in the international monetary system. Qualitative improvements have to do with the reserve character of the SDR. An attractive SDR can inspire confidence and help improve the overall quality of global reserves.

Enhance the Reserve Character of the SDR

Qualitative improvements in the SDR is not a new topic. Exploring the qualitative issue becomes more important under the present impasse on SDR allocation, because weakness in its reserve character is one of the obstacles impeding quantitative progress, and proper qualitative improvements can increase incentives for allocations and contribute to the quantitative issue.

To enhance the reserve character of the SDR, it seems desirable to improve its method of valuation and to endow it with a stable purchasing power. Only when it has a solid value and is widely used as a unit of account can its further development be expected. Until now, several different proposals for “hardening” the SDR have been distinguished.3 One of these suggests a change in the weight of basket currencies in line with their inflation rates so that the value of the SDR is protected from erosion by inflation. A second proposal goes further. It tries to free the SDR from the influence of national currencies and define the SDR in terms of a specific commodity basket. A third proposal envisages a linkage of the value of the SDR with gold (see, for example, Legarda, 1987). Under the present conditions of generalized floating of major currencies and a multireserve system, the first method seems more pragmatic and feasible. But in the long run, an international money that is not affected by any national currency and policy is obviously a more ideal arrangement. Therefore, the second and the third methods deserve further exploration.

The promotion of the private use of the SDR is equally important. In a floating rate system, a very important function of reserves is for holding to intervene in the foreign exchange market. It is hard to say that the SDR is a reserve money if it is virtually useless in market transactions. A number of approaches have been suggested to create “private SDRs” and to use SDRs to finance intervention. These include issuing certificates denominated in SDRs to be held by private users (Coats, 1990) and setting up a “clearinghouse” to ease transactions (Kenen, 1983). There has been so much concentration on this discussion that it now seems more like a matter for consensus and action than a technical issue.

I am doubtful about the hypothesis that there are reluctant holders of SDRs just because the interest rate is low. Lack of usability could be a far more important reason, and the diminishing enthusiasm is largely due to the changes in the world economy. Even the mere fact of political unwillingness could be a bigger obstacle than a low interest rate. Raising the interest rate of the SDR is therefore unlikely to clear the SDR market.

However, a more market-oriented interest rate will no doubt increase the attractiveness of the SDR as a store of value. Hence, some prudent adjustments in the interest rate structure of SDRs seem warranted. A starting point could be to vest the SDR with different interest rates in line with different terms and holdings of SDRs. A user that repurchases its SDRs in one year pays a one-year interest rate. If it fails to do so, it will have to pay a two-year interest rate from the beginning of the second year. By this method the SDR becomes a more attractive reserve asset, and users are encouraged to carry out more efficient adjustment policies.

Principles on Allocation

If allocations of SDRs are to be increased, improvements in the allocation procedures seem necessary. With regard to the allocation method, I do not believe that any better approach can be found to replace general allocations. The key issue is when and how to make allocation decisions. Although setting up some definite rules proves hard, it is practical that we reach consensus on at least one or two principles to govern allocation decisions, or our debates will too easily end in deadlock.

One of these principles could be to put more emphasis on the effects of an allocation on the global economy rather than on a judgment of the existence of a liquidity shortage. By comparing the benefits, such as the promotion of adjustment and economic development, with the possible adverse impact of an allocation, some judgments can be made to serve as a guide for decision making. In making these judgments, quantitative analysis would be more convincing than literal arguments. In view of the low levels of reserves in a large number of countries—which has constrained their economic development and adjustment efforts—I do believe that a new round of general allocation is now justified.

A second principle could be to counteract the trends in the international credit market. That is, an allocation should be implemented when there is market inactivity, and cancellation should be involved when there is overlending in the market. In this way, a sufficient and stable source of liquidity is available and reserve creation could be made countercyclical.

A third principle could be to stress the stability element. The recent financial turbulence illustrates that even creditworthy countries that can meet their reserve needs by borrowing from the credit markets face substantial risk in maintaining reserves. By keeping a considerable percentage of owned reserves in the international monetary system through gradual SDR allocations, the overall quality of global reserves would be improved and risks reduced. By setting up such a facility as a safety net—possibly through the SDR mechanism—the Fund’s resources would be enlarged, and its capacity to deal with systemic instabilities enhanced. Having a lender-of-last-resort in the monetary system will no doubt help restore confidence in the market and benefit countries’ adjustment policies.

Proposal for Redistribution

The idea of postallocation redistribution first appeared in the September 1986 communiqué of the Interim Committee as a subject for further consideration by the Executive Board and was intensively debated on several occasions thereafter. The basic idea is to redistribute allocated SDRs in the direction of developing countries, so that the unevenness in reserve holdings can be eased, and allocated SDRs can be put into use by those countries that have more urgent need for reserves. A benefit of this scheme is that it will help the adjustment and growth of developing countries and improve the “overall social welfare” of the world economy. This is also in accordance with the aim of “attainment of a better balance of payments equilibrium, as well as the likelihood of a better working of the adjustment process in the future” (Article XVIII).

Difficulties encountered in establishing the facility are essentially a matter of risk taking. Under some proposals, creditors would lend to the Fund’s General Resources Account and the lending risk is to be assumed by the Fund. In contrast, other proposals envisaged country-to-country bilateral lending in which the Fund plays an inactive role. In April 1993, the Managing Director proposed a redistribution based on an administered account in the Fund, which would involve transfers of SDRs to selected countries with Fund-supported reform programs and under which lending countries would associate themselves with the risk undertaken by the Fund.4

Impressed by the idea of a special account for SDR redistribution, I would like to present some initial views in favor of the setting up of such an account and some suggestions.

  • The aim of this account is to provide an appropriate channel through which net holders of SDRs can voluntarily transfer their SDR holdings to countries that are in need of additional reserves. This temporary arrangement does not change the operations of the SDR Department and does not crowd out regular Fund resources.

  • A general allocation of SDRs would be made in accordance with the Articles of Agreement. Members of the SDR Department would agree in advance of the allocation to set up an SDR administered account in the Fund, under an arrangement along the lines indicated below.

  • The SDR administered account would be established by the issuance of SDR-denominated shares to be held by voluntary participants. By exchanging SDRs for these shares, participants would entrust the Fund to manage a given amount of their allocated SDRs. The Fund would act as the account manager or trustee. The Fund then would use these SDRs to provide assistance to members to supplement the financing available for regular Fund programs.

  • Considering its reserve character, the eligibility and terms of redistributing SDRs could be made more flexible than under the General Resources Account. Countries that pursue sound economic policies but are facing difficulties or emergencies resulting from institutional, technical, or other exogenous factors could also be considered for help, even if they are at the time not under a Fund program. Terms of lending could also be made longer than regular Fund programs.

  • Shareholders of this account would retain their ownership over those channeled SDRs. Members with holdings in excess of their allocations would continue to earn net interest and those with holdings below their allocations would still pay a net charge. What is different is that SDR users of this account would have to pay a risk premium in addition to regular SDR charges, which is to be shared by the shareholders of this account.

  • The issuance of shares would bring about flexibility and liquidity. The par value of each share would be small so that not only large holders but also countries that hold only moderate amounts of SDRs and prescribed holders of SDRs can participate. By making these shares negotiable, participants could exchange their shares for SDRs when they themselves are in need of reserves.

  • Prior to the termination of the account, there could be a review of the situation to determine whether to extend it.

I perceive this redistribution account as nothing more than a temporary arrangement to ease the current unevenness in SDR distribution. To improve the current reserve system, more fundamental measures will have to be taken to better equip the SDR as a reserve asset, to reform the reserve creation process, and to strengthen the role of the IMF in orchestrating the international adjustment process.

Concluding Remarks

As I have pointed out, the extent to which the SDR could contribute to the monetary system depends on the extent to which we improve the SDR mechanism itself—both qualitatively and quantitatively.

By putting the SDR under international control, an adequate supply of world liquidity could be ensured. The SDR is used not only to supplement existing reserve assets but also to counteract any overreaction of the private credit market, and hence to maintain a stable source of liquidity.

Tackling the asymmetry in reserve creation does not mean creating an unlimited amount of SDRs at low cost for every country, because this would remove the constraints on countries to adjust their policies and would produce adverse effects. On the contrary, symmetry is achieved by putting countries under the impartial surveillance of the Fund and by general allocation and postallocation distribution of fiduciary reserves. Higher interest rates or Fund surveillance will provide the necessary protection against undue expenditures of resources.

The stable purchasing power of the SDR combined with usability in the private market will significantly strengthen its role as a unit of account, store of value, and medium of exchange. This will facilitate trade and commerce and reduce the shifts between assets, so that the confidence problem will be largely alleviated.

A stable supply of reserves will help foster financial stability. A highquality reserve money—the SDR—that substitutes for borrowed reserves will reduce lending risks and enhance the stability of the system. By creating a safety net facility, possibly using the SDR, the Fund will act as the lender of last resort to forestall any potential crisis in the financial system and maintain the stability of the world economy.

Does the World Need a New Official Safety Net to Deal with Potential Financial Crisis?

György Surányi

Need for Official International Safety Net

The history of finance keeps extensive records of events of extraordinary importance that are not related to periods of war or revolution, but that derive from panic discussed at length by the science of mass psychology. Pioneering the creation of panic was the English-turned-French banker John Law who, depending on taste, is called either notorious or famous and who “discovered” the distinct ability of banks to create money by granting credit. Although the development of the “central bank” role in the banking system was basically tied to the need to satisfy the borrowing requirement of the treasury and to the monopoly of banknote issue, the lender of last resort function—one of the central bank’s tasks with its importance stressed tentatively but increasingly loudly from the beginning—cannot be neglected.

Kindleberger (1978, chap. 10) and, with reference to him, Eichengreen (1995, pp. 247-50), use the notion of “international lender of last resort.” According to Kindleberger, the international lender of last resort should secure two distinct things for countries struggling with payments difficulties: cheap long-term loans and open markets. In the international monetary system created at Bretton Woods after the Second World War, the IMF is considered best able to fulfill this function.

The International Monetary Fund could well have accomplished this function if the objectives of securing an undisturbed operation of the monetary system created after the Second World War had fully materialized, that is: the race for devaluations among countries is eliminated; and persistent payment imbalances do not develop and, consequently, member states are not compelled to conduct deflationary policies, which might lead to a drastic rise in unemployment on the one hand, and to the introduction of import exchange and capital restrictions on the other.

Had these goals materialized, there would have been only temporary balance of payments deficits, which could have been addressed by the newly created monetary system under the umbrella of the International Monetary Fund. Instead, temporary payments deficits proved rather persistent, and the tools designed to treat them had (or would have) to be used to maintain the original objectives. The discord about means and ends inevitably leads to distortion of the system and to deficiencies in its operations.

While international capital markets continue to integrate, the instruments available to influence the smooth functioning of the system have remained broadly unchanged. In countries newly integrated into the international monetary system, domestic markets have developed at an accelerated pace, often implying the risk of collapse. It is necessary therefore at least to question whether there is a need to create a new official safety net.

In light of the risks inherent in the international monetary system, it is unlikely that a central banker responsible for maintaining the financial system in his own country would give a negative answer. However, it must not be a simple positive answer either, since, in my view, both the solution of the problem and the determination of an acceptable scale for such a safety net are extremely complex tasks.

In my view, the starting point is to determine an adequate size for the safety net. The easiest way to understand the essence of my idea is through the following analogy. Engineers engaged in flood prevention can almost precisely define the height of the embankment that can protect against floods. It is well known that the impact of floods that occur a couple of times a year can be anticipated with reasonable precision; and higher-than-normal levels of water occur only in intervals of 20, 30, or 50 years. The question is whether it is proper to adjust the height of the embankment for floods that recur every 20, 30, or 50 years, or for floods that are expected every year? The answer will inevitably be a mix of economic as well as aesthetic aspects; the costs of building and maintaining an embankment suitable to protect against floods expected at wide intervals will exceed the potential damages caused by those rare floods (which, of course, use of “short-term” instruments attempts to reduce); further, the view of a city such as Budapest would be spoiled by a colossal concrete structure ten meters high that protects both banks of the River Danube.

Perhaps the analogy will obviate any further explanation of my answer.

Although lack of agreement about the original purposes and the tools at hand and the large amount of integration of national financial markets make it important to restructure, supplement, or enlarge the existing official safety net, the creation of a safety net offering guarantees against risks of the potential collapse of any national financial market cannot be targeted.

IMF or Central Banks?

The international lender of last resort function of the IMF and the swap network of the central banks developed simultaneously, side by side, from the 1950s. The most developed countries have no official standpoint on which of the two should be the ultimate lender of last resort.


In 1962 the resources of the IMF were supplemented by the GAB (General Arrangements to Borrow). Within this framework the Group of Ten increased the $15 billion quota by $6 billion, with the aim of allowing the IMF to make use of it as a means of last resort lending (if funds of the member state and of the IMF should prove to be insufficient) to withstand the effects of certain unfavorable capital movements (Kindleberger, 1978, p. 202). The use of the GAB at that time was restricted to the financing of loans to GAB creditors, reflecting a situation in which capital movements threatening the stability of the international monetary system concentrated mainly on industrial countries. This amount not only proved to be inadequate in a number of cases, but, in addition, the decision-making mechanism of the Fund was overly complicated. Decisions took weeks to make, whereas the efficient intervention could have been secured only by operative decision making and action.

Meanwhile, the existence of hot monies became more and more pronounced on the money markets by the 1960s. All this was only partly due to inflation; a role was also played in this kind of asset holding by the increasing weight of financial instruments (the rise in the ratio of aggregate money to income—financial deepening). By the same token, and also as a result, investor behavior also changed: the foreign exchange started to become a liquid asset and an object of speculation (Kindleberger, 1978, p. 202).

Central Bank Swap Network

The Basle Agreement can be considered the new international lender of last resort function created by the central banks jointly, in response to the appearance of hot monies. In 1961, when the pound sterling came under pressure, several central banks, under the umbrella of the Bank for International Settlements (BIS), extended credit to the Bank of England. The $1 billion at the command of the Bank proved sufficient against the speculative run. This was the turnaround, and the IMF began gradually to withdraw into the background as lender of last resort: from this time the club of central banks became the primary lender in emergency. The club was summoned several times during the decade. The United States assumed a leading role in negotiating the syndicated loans of central banks. The swap line technique was applied in lending: central banks opened liability accounts for the other partner banks, while the liability item was balanced by another item on the asset side in the partners currency. In exchange, the partner did the same, thus the liability of the one became the asset of the other and vice versa (Kindleberger, 1978, pp. 204-5). In principle, this “current account” had to show a semiannual closing balance.

The Federal Reserve and the Bank of France were the first to use this technique in 1962, which was followed by similar arrangements between the Federal Reserve and the Belgian and Dutch central banks. In 1962 the swap network included 11 central banks, with $2 billion capital. By 1973, the credit line of the swap network had been raised to $18 billion.

It was expected by some—mostly in academic circles—that, owing to the freely floating exchange rates, the role of the official international lender of last resort would soon become redundant. It was assumed that, as no government would take explicit responsibility for the exchange rate, there would be no need for official intervention to defend it, and the market mechanisms would adjust rates according to fundamentals. Speculation does not destabilize but—on the contrary—stabilizes, because it eliminates quickly and effectively the chances for arbitrage. In this way speculators will ruin themselves: hot monies will decrease drastically, and perhaps subside. The reason for stability according to the optimists was that governments’ commitment to fixed exchange rates hindered the market in making the necessary corrections quickly. This may only have delayed the adjustments—but could not terminate their causes—which inevitably occurred later in a much more severe form.

In this scenario the lender of last resort could only have had a domestic role. However, the failure of the Herstatt Bank in 1974, caused by exchange rate speculation, demonstrated the need for international coordination of national monetary control and supervision, because the nonregulated gray zone was continuing to grow. At that time, international organizations started to make efforts aimed at regulating international banking groups. Demand for this kind of coordination has been increasing ever since the start of the liberalization and external deregulation wave of the 1980s. National banks or their clubs have always had a major role in the process.

In 1974 the supervisory bodies of the United States warned U.S. banks against extending loans to Italy again, because the country’s solvency was doubtful. After the European Economic Community had rejected the Italian request, Germany finally granted credit to Italy against gold as collateral. As a result of the oil crisis of the 1970s, Italy was facing severe payment difficulties; it was far from being alone, however; a number of other developed and developing importers of oil found themselves in a similar situation. The United States, with a view to reducing financial uncertainties that were mounting as a result of the oil crisis, was considering introducing a financial safety net, which was later scrapped by Congress. Today the situation of industrial countries is completely different. The sophisticated network of large central banks, as well as the easy access of rich countries to private capital markets, appears to offer both sufficient frontline last resort lending and a safety net.

Once More the IMF?

The growing indebtedness of the developing and, later, of the former socialist countries raises much more different problems. In this context the role of the IMF is crucial: the focus of its activities has more and more decidedly shifted in this direction. Central banks of developing countries typically are not members of the swap network of central banks of rich countries. The IMF can scarcely be replaced in the future. If it is not primarily the lender of last resort—as its own resources are scarce—it is probably the organizer of ultimate loans, as a catalyst and provider of surveillance for the use of credits. At the time of the outbreak of the debt crisis in the 1980s, the IMF was the only organization that, by virtue of its position, original purpose, and experience, could come to the rescue of countries in trouble and of their commercial banks. This proved to be a historical step that has determined the development of the activities of the IMF since then.

Should the SDR Retain the Role of Reserve Currency?

Preparation for Creation of the SDR

The creation of the SDR was originally decided to allow member states to have sufficient reserve assets in order to intervene in defense of fixed exchange rates, provided that the most important goal of the IMF materialized, and members could achieve longer-term financial equilibrium. This expectation was included in the founding documents of the IMF as the scarce currency clause.

The proposal put forward by Keynes at the time the Articles of the IMF were elaborated, which was aimed at eliminating persistent international financial imbalances, had been abandoned by 1943. This would have been the so-called surplus clause, which said that pent-up excess liquidity in countries with balance of payments surpluses should be recycled into countries with financial disequilibrium through the Bretton Woods institutions. At the same time, Keynes intended a lasting role for restrictions on capital movements—including administrative policy instruments. Under the political pressure of the United States, this proposal was not included in the Articles in its original form. However, restrictions on convertibility were allowed in the area of certain, mainly short-term, speculative, movements (capital controls), and, further, commercial restrictions on a country issuing currency and declared scarce by the IMF were tolerated (scarce money clause). This meant that, instead of the bilateral (debtor and creditor) accommodation proposed by Keynes, unilateral accommodation of debtors became a requirement (Grieve Smith, 1995, p. 291). Never was there a need, however, to apply the scarce currency clause, because this was made unnecessary following the war by Marshall aid and later by the emerging balance of payments deficit of the United States.1

The role of the dollar in the Bretton Woods system meant that the countries that were willing to hold dollars as a reserve currency had to tolerate the permanent balance of payments deficit of the United States. The strain inherent in this solution was pointed out by Triffin for the first time.2

Introduction of the SDR

The introduction of the SDR had continuously been pondered by IMF member states at length throughout the decade, before it was finally decided in 1969.3 The first allocation took place in 1970. The time of the introduction makes it obvious that there were profound changes in the circumstances of decision and action that were originally used to justify the creation of the international reserve currency.

The unfolding developments quelled worries: there was an oversupply of U.S. dollars. By now, there are different arguments supporting the introduction of the SDR than those accepted at the time of its elaboration.

The European countries—mainly France—accused the United States of deriving seigniorage income from the central role of its currency at the expense of the other states. In 1972, immediately before the presidential elections, the United States was more eager than ever to continue with its monetary expansion, whereas Germany was doing just the opposite: from fear of inflation they raised interest rates. This resulted in an outflow of capital from the United States. Eventually, the Germans were forced to cut interest rates. However, solution of the structural conflict could not be delayed for too long. It was expected by some experts that this type of conflict could be easily eliminated if and when an international reserve currency was introduced. The benefits of reserve creation (seigniorage income) would then spread more widely among the world economic community. Some proposals, like the “link” proposal, even wanted to use this income for supporting the development of the Third World. The link proposal, however, was later dismissed, partly because industrial countries disagreed with its basic principles and arguments; and partly because it had lost its attractiveness for Third World countries by 1972-73 because inflation emerged as a major problem and international liquidity rapidly increased during those years (Crockett, 1994, pp. 86-7). Moreover, the conflict between the roles of national and international reserve holdings can be reduced if divergent movements are equilibrated in the exchange rate of the international reserve currency.

The oversupply of dollars flooded the Eurodollar market, which—in the form of petrodollars following the oil price shock—made it possible for individual countries to acquire dollar balances from non-U.S. residents so that its adverse impact on the balance of payments of the United States became avoidable. The gradually increasing international liquidity from the 1970s, which in the beginning meant negative real interest rates for developing countries in cases of borrowing on the international capital markets, was paradoxically accompanied in the 1980s by high international real rates of interest, and its lasting survival and effects are still being felt today. Struggling with problems of competitiveness, the oil price hike, and the deteriorating balance of payments of developing countries that became indebted at lower interest rates had led to a debt crisis by the 1980s (Mexico, Poland, etc.). While rich countries had obtained increasingly easier access to international reserves on the private markets, countries affected by payment problems could only obtain funds at increasingly higher premiums, and several Third World debtors could not even raise reserves from private sources. Demand for official international reserves multiplied, and therefore mainly the Third World countries urged repetitions of allocations and/or increases in IMF quotas. Their ranks were joined by some former socialist countries that also did not have access to international capital markets.

From a practical point of view, the existence of an international reserve currency can be supported by the notion that debts in such a currency could be managed and repaid in a simpler way than by managing simultaneously a portfolio of individual currencies exposed to larger cross-rate volatility. This is worth consideration by developing countries that have little experience in tackling such problems.

Effect of Development of Macroeconomic Theory

In addition to practical considerations supporting the international reserve currency and, as such, the SDR, attention must be called to the change that occurred in the theory of economic policies after the Second World War. This theoretical and economic policy change took place together with a new stage in the globalization of the world economy. As a result, (Keynesian) economic policies conducted by the state, which operated with aggregate demand management, were replaced by neoclassical policies based on the self-efforts of economic agents.

By the 1980s it had become obvious that an economic policy that, on the basis of a choice between inflation and unemployment (Phillips trade-off), maintained that (nearly) full employment could be feasible at the price of a modest rate of inflation was not credible any longer. As we saw earlier, the documents creating the Bretton Woods institutions emphasize the achievement of full employment as one of the most important objectives. In the 1980s, however, governments’ major goal became to bring down inflation, and the elimination of unemployment was subordinated to this ambition. More specifically, a durable solution of unemployment could only be expected from sustainable growth, which was considered viable by securing a reasonably low long-term inflation rate.

From the point of view of international reserves, this change in the approach carried critical importance. Macroeconomic management of demand following the war built on the exogenous conception of money (Crockett, 1994, pp. 86-7). This theory, which has generally been connected with monetarism, was an essential pillar of economics before Keynes. It remained an organic part of the synthesis economics in the postwar period and, incorporated in the Polak model that provided a basis for monetary programming applied by the IMF, is still present in the thinking of those formulating practical economic policies. It was not the appearance of monetarism that brought about the theory of exogenous money supply.

The essence of the approach is the assumption that there is a relatively stable demand for money by economic agents, against which money supply is determined by the monetary authorities. In 1965, at the time of the Ossola report that paved the way for the SDR, this concept in the context of an open economy sounded as follows: the starting point is fixed exchange rates with every country endeavoring to maintain them. Adhering to this condition they are making efforts to achieve external and internal equilibrium. A precondition for maintaining durable external equilibrium was to secure smaller fluctuations by manipulating aggregate demand in the opposite direction. A full employment of factors is assured by stable growth. There was only need to adjust rates for the sake of fundamental corrections.

The role of international reserves has been reduced to provide opportunity to maintain the exchange rate parity at times of temporary external imbalances. Reserves in excess of demand would have caused inflation, while reserves below that threatened to lead to balance of payments restrictions and deflationary monetary policy. In this system the institution providing international reserves had to allocate exactly the volumes that were needed for the achievement of the optimal rate of growth/inflation. The IMF has made genuine efforts to estimate future developments in demand for international reserves.

Crockett (1994, pp. 83-5) calls attention to an important difference between reserves and liquidity: to quantify international liquidity is much more difficult than to quantify reserves. Reserves can be measured more or less exactly under the internationally accepted definition. Ex post they are absolutely recognizable. The case is different, however, with liquidity, because it also includes reserves that have not been created but that are at the disposal under certain conditions of some sort of arrangement or commitment. If they are not used or run down, complete knowledge about them is not possible. Nevertheless, they exist and exert certain influences. It is these difficulties of quantification that prompt us to speak of reserves, instead of talking about liquidity, and this is why we use the two terms as synonyms.

Experience has shown that it is still difficult to estimate short-term needs for reserves; and the long-term trend increases with the help of a rule of thumb, more or less parallel with the expansion in international trade, and this rough empirical estimation cannot be improved even with the most sophisticated procedures (Crockett, 1994).

Two things are worth singling out: the predictability of (international) demand for money and the problem of the endogenous or exogenous nature of (international) money.

First, an attempt at predicting demand for money can fail even when it is exogenous, according to its traditional conception: if the system is affected by an external shock, structural variables become unstable and unpredictable. In a broad sense, the appearance of financial innovations can be placed here, which have facilitated the portfolio adjustment between the various monetary aggregates and blurred the differences between the aggregates themselves. By the 1980s most of the industrial countries had given up earlier monetary policies based on the prediction of target aggregates, in which they were inspired by monetarism.

Second, demand for money can be predicted well also with stable structural characteristics (short of shocks), even if money enters into circulation through endogenous mechanisms (as credit money). In this case nominal national income and money supply grow hand in hand—in the same way as the theory of exogenous money supply predicts. However, the direction of causality is from price level toward money, instead of the opposite. Further, the theory of endogenous money supply does not deny that overspending by governments exerts inflationary pressures, and that this component of money in circulation can be treated as exogenous.

Without going into details, the analogy between national and international issuance of money must not be overemphasized. While there are issuances of money as well as output at the national level, there is also an international mechanism for issuing money, but there is naturally no such mechanism for output.

The assumptions concerning the theory of exogenous money from the supply side did not materialize. This started from the assumption that the expansion in liquidity was the result of the deliberate creation of money. In reality, liquidity created in such a way accounted for only a marginal part of international reserves: as we saw it, it was the dollar that played a major part in the increase of international reserves. The gradual decline in the role of gold directly stimulated the creation of the SDR: gold reserves reached their peak in 1965, followed by a gradual fall.

When in the first allocation of SDRs, between 1970 and 1972, SDR 9.3 billion of reserves were created, that amount augmented the stock of reserves by approximately 10 percent. However, what the experts anticipated did not occur: the reserve function of national currencies did not decline, but on the contrary, it increased. The immediate reason for this was a dramatic deterioration in the balance of payments of the United States, not least because of the escalation of the Vietnam War. Speculation against the dollar strengthened, which increased the weight of other national currencies among reserves.

At that time, the official opinion of the Group of Ten countries, however, was that, in addition to the system of flexible exchange rates, there was a need to predict the demand for international reserves, as they did not share the view that foreign exchange markets automatically supply the necessary volumes of reserves (exogenous approach to international reserves); the recycling of petrodollars made it necessary to issue international reserves, since oil exporting countries were inclined to hold their revenues in money balances instead of making direct foreign investments in countries in deficit; and, as a reflection, the external imbalance of oil importing countries enhanced the risk of speculative runs on their currencies, while the stock of their international reserves was falling quite rapidly.

In reality, however, 1973 marked the beginning of a battle against inflation in industrial countries, and the above considerations became overshadowed. The development of the commitment to monetary restriction characterized this period, which was fully achieved in the 1980s and became part of official policy. That is why later the issuance of SDRs failed to pick up considerably. The Annual Reports of the IMF devoted just a few pages to the subject, which appeared to be no more than a fulfillment of a compulsory task to satisfy the expectations of shareholders rather than playing a major role in the actual operations of the IMF, and which could have been discussed in detail. We saw that this had practical and economic policy reasons; however, it had become theoretically clear by now that it was necessary to issue international reserves to promote the growth of world economy and trade.

Pros and Cons of New Allocations

First, the demand for a new allocation of SDRs raises doubts about whether such a demand can be harmonized with the original purpose of its creation. Second, the question is also whether the SDR is the most appropriate instrument for supplementing the international reserve holdings of developing countries.

The last allocation took place in 1981. The rise in the use of SDRs stemmed mostly from quota increases and interest payments made to the IMF. Its use is not tied to conditionality. The majority of proposals today refer only to SDR allocations with conditionality and/or post-allocation redistribution. Williamson’s proposal (1994, pp. 61-3) appears to be exceptional, as he does not speak of conditionality, but of equity, in connection with a particular kind of allocation: a partial remedy of an existing injustice.

The following arguments can be brought forward for international allocations based on conditionality, link, and equity:

  • Developing and transition countries have access to reserves on the private capital markets only under conditions that require enormous efforts from them. This may have two different forms: substantial export surpluses, which require severe restriction on domestic consumption, or borrowing at high interest rates, which again demands exorbitant surrender of real goods.

  • The volume of allocation that is vital for developing countries and countries in transition represents just a minor portion of world trade and world GDP. That is why its effects resulting in the redistribution of real resources and/or in the increase in international interest rate levels would be negligible.

None of the above arguments are considered to be particular calls for the use of the SDR. There are some international financial institutions that could fill the needed function much more explicitly: some of them were created to arrange the international transfer of resources. Nor do these and many other arguments justify why transfers of this kind should be arranged by creating additional liquidity, especially if the uses of these resources are otherwise tied to IMF conditionality. In this case the SDR would not provide greater freedom of use, compared with normal IMF or other types of resources.

  • According to Williamson, when the developing countries, in order to acquire reserves, are forced to achieve surplus exports and/or obtain foreign borrowing at prohibitive interest rates, they actually provide rich countries with a form of subsidy. This is why allocation could only partially compensate for, or reduce, this unjust practice. Some might say that it is difficult to bring the argument of justice into a controversy, as it naturally has different meanings for the partners. And even if the developed world shares Williamson’s sense of justice, which is far from obvious, why should justice be restored through the creation of money? (Masera, 1994, and Mundell, 1994).

All three approaches above employ symptomatic treatment to redress problems that appear to be inherently structural. Confidence in countries whose macroeconomic policies are accepted by creditors/ investors cannot be restored by a one-time increase in reserves. Continuous issuance of liquidity can only delay necessary structural arrangements, while confidence in the system itself can be shaken.

Proposals in general would tie new allocations to IMF conditionality. However, the requirements of the IMF practically harmonize with capital market criteria. Therefore, if a country is appropriate for one of the two (IMF), it will likely be appropriate for the other (capital markets).

Finally, there is a version of proposals that would try to avoid the negative repercussions of liquidity increases by prescribing the holding of resources on reserve accounts in beneficiary countries. This would not cause interest rate increases or exert inflationary effects, as there would be no other unwarranted consequence. It would demonstratively boost confidence in the currency of the country. This argument can be challenged also: if the macroeconomic policy of a country is considered reliable, the level of reserves is only one element helping to sustain market confidence. However, when macroeconomic policy appears unsustainable in the long run, even high levels of reserves will not offer any kind of guarantee.

One lesson from the emerging markets’ crises is that unsustainable macroeconomic policies cannot be bettered by the creation of international liquidity. Another lesson is that capital markets never respond immediately: sometimes they react too late, when social costs are too high. (In Mexico, foreign investors only moved when residential investors had already fled from the peso.)

In this respect the Mexican crisis did not deliver any novel feature that would prompt a revision of the above conclusions.

General Discussion

Morris Goldstein, reacting to de Cecco and Giavazzi’s argument that two of the desirable characteristics for a financial safety net were risk diversification and speed, questioned to what extent these two were in conflict and how any such conflict could be resolved. For risk diversification, a large number of creditors was desirable; for speedy decision making, a small number. If the United States had been unable to contribute to the Mexico rescue package in January 1995, and if the Fund had had to draw on the General Arrangements to Borrow, he wondered if agreement could have been secured quickly enough to have avoided default. He assumed that the authors’ solution was to give the Managing Director the authority to decide quickly in such an emergency, but he doubted whether the idea of giving such a blank check would appeal to the major creditors. Therefore, how could the circle of speed and risk sharing be addressed?

Peter Kenen remarked that he did not regard credit risk as the primary question. Experience with the functioning of the very short-term financing facility (VSTFF) of the European Monetary System had shown that the issue had not been the credit risk borne by the strong currency countries, particularly Germany, but the risk of monetization resulting from an open-ended credit creation arrangement. He had examined the key passage in the report of the EC central bank governors on this question, which conveyed emphatically that there could be no automatic or mechanistic response involving symmetrical actions by surplus and deficit countries and no interference with control over monetary conditions in the country issuing the intervention currency. Opposition was likely therefore to be based not so much on credit risk and moral hazard considerations as on the monetary control issue. He interpreted the EC report as saying that the automaticity of the EMS arrangements was dead. If de Cecco and Giavazzi wanted to revive it in a somewhat different context, Kenen said he could envisage a revival of those objections.

Francesco Giavazzi, responding to Goldstein, agreed that there was a trade-off. Risk diversification was a desirable characteristic that came at the cost of the difficulty of coordinating the roles of the different countries involved. Giving the Fund this coordination role, and the Managing Director the initiative in these decisions, was a point in this tradeoff. Giving such a blank check was comparable to the president of an individual central bank having the mandate to make decisions when its board was deciding whether to intervene in a particular crisis. Their suggestion had been to replicate that example.

In response to Kenen, Giavazzi commented on the significant difference between the VSTFF and what he and his co-author had envisaged. The VSTFF had been designed to prevent realignment: whenever currencies hit the boundaries, it was used to provide liquidity and prevent realignment. Unlimited intervention through such a facility had now been ruled out, because trying to prevent realignment might be extremely costly in terms of creating automatic liquidity. He and de Cecco had envisaged using this mechanism in a devaluation to prevent a temporary liquidity crisis—which caused contagion either internationally or in the domestic market—not to create a permanent increase in liquidity.

Barry Eichengreen asked de Cecco and Giavazzi to comment on alternative means of financing the emergency financing facility they had proposed so that it could be contrasted with a method suggested by Kenen. Kenen had proposed having countries experiencing large capital inflows deposit some portion of those inflows into a pool at the Fund and some multiple of those deposits could then be disbursed either conditionally or unconditionally to countries experiencing currency crises.

Marcello de Cecco responded by recalling a comment on Mexico from György Surányi’s paper: intervention had been fast, and everyone had known that they were going all the way. He questioned, however, whether Mexico was like every other country. He contrasted the philosophy behind Surányi’s comments on Mexico with cases such as Italy, where no “big daddy” or “sugar daddy” had been available. He recognized the danger of monetization to which Kenen had referred and as it was perceived by the EC central bank governors, but there had to be a way of ensuring that automaticity was implied even when “big daddy” was not there. He could support the proposal mentioned by Eichengreen, because it was important to find some way of providing a safety net. Alluding again to Surányi’s paper and the height of the dam needed in a currency crisis, he observed that if a very large mobilization of resources was needed and the potential was not there, the problem would be who was going to mobilize it and how much would be needed. Rather than provide for an uncertain future, might it not be better to plan as if a flood—not too heavy—occurred every year and provide accordingly?

J. de Beaufort Wijnholds agreed with Kenen that the real issue was not so much risk diversification as the modality of applying credit, and whether credit was conditional or unconditional. His understanding of de Cecco and Giavazzi’s proposal was that, by using the SDR mechanism, it was unconditional with unlimited support. That kind of short-term facility had not received much support when discussed in the Fund. But agreement had been reached on a procedure for an emergency financing mechanism—rather than a separate facility—to address problems of the type experienced in Mexico, in an accelerated approach and with appropriate conditionality.

John Williamson welcomed Zhu’s suggestion that interest rates should be raised on the net use of SDRs for an extended period. It would provide an alternative way of generating an income for the Fund and also complemented the ideas raised by Iqbal, Mohammed, and Zaidi (chap. 6), It would act as an incentive for countries that lacked access to other forms of reserve assets to use SDRs only as reserve assets and not to make extended net use of them.

No market mechanism presently existed to ensure that the outstanding stock of SDRs was willingly held, Williamson continued. Although there had been no need to resort to designation in recent years, it was clearly a potential problem. The interest rate paid to holders of SDRs should be determined in some sort of auction mechanism. Once an average interest rate above the interest rate in the Group of Five countries—whose currencies compose the SDR basket—had been obtained, there would be a margin to play with that would allow a market-related interest rate for SDR holders to be determined and thus ensure that the outstanding stock of SDRs was held willingly.

Responding to a question from Jacques Polak about why he advocated market determination of the interest rate and not the price, and why he was not considering a market for SDRs such as existed for ECUs, Williamson observed that while that was an alternative, he had always been attracted to the basket valuation of the SDR. The deviation between the official and the market price of the ECU made it a relatively unattractive alternative. He would prefer to let the interest rate be determined in the market.

Rudolf Rhomberg shared Williamson’s attraction to the basket valuation, which defined what was ultimately obtainable if an SDR was sold. But that should not determine what interest rate had to be paid when SDR balances were drawn down for a specified period such as 3, 6, or 12 months. These rates could differ for different maturities, and he agreed that longer-term use should normally carry a higher interest rate. Rhomberg favored having a country decide ex ante for how long it wanted to make use of a specified amount of its SDRs; the country would then pay an interest rate based on the length of that period, rather than follow the practice, which had previously applied to the use of Fund credit, of paying interest at a certain rate for an initial period, with increased rates for subsequent periods during which balances remained outstanding.

Ariel Buira was troubled by the idea of charging a higher interest rate for the longer-term use of the SDR. If SDRs were used for longer, it would simply be because of a shortage of other sources of liquidity. It would be the Fund’s role to supplement these sources, and the need would be permanent. If the SDR was considered a line of credit, it should perhaps have a market-related rate of interest; but if it was being used as the result of a lack of liquidity in the system, it would be unfair to charge a higher rate for a systemic fault.

Francis Woehrling objected that if countries held their SDRs as reserves, they would not be charged any interest on the balance. The high interest rate would be charged only on prolonged use of SDRs.

In response, Rhomberg indicated that he had imagined two very different situations. One was iike holding a checking account that carried a low interest rate that stayed the same regardless of how much was in the account. An element of the SDR mechanism would always be like that. The other was the situation in which additional amounts of SDRs could be made available under crisis conditions, as had been discussed. There could be shorter- or longer-term loans in SDRs, and they could carry different interest rates.

As a comment on Jacques Polak’s proposal, Kenen remarked that the existing method of quota financing, as Polak had indicated, gave the Fund currencies that it could use, whereas an SDR allocation gave certain countries SDRs they did not even want to hold, making the present arrangement extraordinarily inefficient. In both cases the amounts to be raised or mobilized were far larger than certain participating countries would like, and one of the advantages of Polak’s proposal was that it did not have that redundancy.


  • Coats, Warren L, Jr., “The SDR as a Means of Payment,” Staff Papers, International Monetary Fund, Vol. 29 (September 1982), pp. 42236.

    • Search Google Scholar
    • Export Citation
  • Duisenberg, W.F., and A. Szász, “The Monetary Character of the IMF,” in international Financial Policy: Essays in Honor of Jacques j. Polak, ed. by Jacob A. Frenkel and Morris Goldstein (Washington: International Monetary Fund, 1991), pp. 25466.

    • Search Google Scholar
    • Export Citation
  • Gold, Joseph, Special Drawing Rights—The Role of Language, IMF Pamphlet Series, No. 15 (Washington: International Monetary Fund, 1971).

    • Search Google Scholar
    • Export Citation
  • Horsefield, J. Keith, The International Monetary Fund, 1945-1965 (Washington: International Monetary Fund, 1969).

  • International Monetary Fund, Annual Report, various issues.

  • Kenen, Peter B., “Use of the SDR to Supplement or Substitute for Other Means of Finance,” in International Money and Credit: The Policy Roles, ed. by George M. von Furstenberg (Washington: International Monetary Fund, 1983), pp. 32760.

    • Search Google Scholar
    • Export Citation
  • Polak, Jacques J., Some Reflections on the Nature of Special Drawing Rights, IMF Pamphlet Series, No. 16 (Washington: International Monetary Fund, 1971).

    • Search Google Scholar
    • Export Citation
  • Polak, Jacques J., Thoughts on an International Monetary Fund Based Fully on the SDR, IMF Pamphlet Series, No. 28 (Washington: International Monetary Fund, 1979).

    • Search Google Scholar
    • Export Citation
  • Roosa, Robert V., The Dollar and World Liquidity (New York: Random House, 1967).


  • Bruno, Michael, “Comments and Discussion on Dornbusch, Goldfajn and Valdés,” Brookings Papers on Economic Activity: 2 (Washington: Brookings Institution, 1995), pp. 27865.

    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger, Dan Goldfajn, and Rodrigo O. Valdés, “Currency Crises and Collapses,” Brookings Papers on Economic Activity: 2 (Washington: Brookings Institution, 1995), pp. 21970.

    • Search Google Scholar
    • Export Citation
  • Fischer, Stanley, “The SDR and the IMF: Toward a World Central Bank?” in International Money and Credit: The Policy Roles, ed. by George M. von Furstenberg (Washington: International Monetary Fund, 1983), pp. 17999.

    • Search Google Scholar
    • Export Citation
  • French, Kenneth R., and James M. Poterba, “Investor Diversification and International Equity Markets,” NBER Working Paper No. 3609 (Cambridge, Massachusetts: National Bureau of Economic Research, January 1991).

    • Search Google Scholar
    • Export Citation
  • Gerlach, Stephan, and Frank Smets, “Contagious Speculative Attacks,” CEPR Discussion Paper No. 1055 (London: Centre for Economic Policy Research, November 1994).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, International Capital Markets. Part I. Exchange Rate Management and International Capital Flows, World Economic and Financial Surveys (Washington: International Monetary Fund, April 1993).

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

    • Search Google Scholar
    • Export Citation
  • Kindleberger, Charles P., Manias, Panics and Crashes: A History of Financial Crises (London: Basic Books, 1978).

  • Krugman, Paul, “A Model of Balance-of-Payments Crises,” journal of Money, Credit and Banking, Vol. 11 (August 1979), pp. 31125.

  • Obstfeld, Maurice, “The Logic of Currency Crises,” Cahiers Economiques et Monétaires, Banque de France, Vol. 43 (1994), pp. 189213.

    • Search Google Scholar
    • Export Citation
  • Obstfeld, Maurice, “Models of Currency Crises with Self-Fulfilling Features,” European Economic Review, Vol. 40 (April 1996), pp. 103747.

    • Search Google Scholar
    • Export Citation
  • Obstfeld, Maurice, and Kenneth Rogoff, “The Mirage of Fixed Exchange Rates,” journal of Economic Perspectives, Vol. 9 (Fall 1995), pp. 7396.

    • Search Google Scholar
    • Export Citation


  • Coats, Warren L., Jr., “Enhancing the Attractiveness of the SDR,” World Development, Vol. 18 (July 1990), pp. 97588.

  • Corden, W.M., “Is There an Important Role for an International Reserve Asset Such as the SDR?” in International Money and Credit: The Policy Roles, ed. by George M. von Furstenberg (Washington: International Monetary Fund, 1983), pp. 21347.

    • Search Google Scholar
    • Export Citation
  • de Vries, Margaret Garritsen, The International Monetary Fund 1966–1971: The System Under Stress, Vol. I (Washington: International Monetary Fund, 1976).

    • Search Google Scholar
    • Export Citation
  • Kenen, Peter B., “Use of the SDR to Supplement or Substitute for Other Means of Finance,” in International Money and Credit: The Policy Roles, ed. by George M. von Furstenberg (Washington: International Monetary Fund, 1983), pp. 32760.

    • Search Google Scholar
    • Export Citation
  • Legarda, Benito, “Issues Arising in Making the SDR the Principal International Reserve Asset,” in The International Monetary System and its Reform, Part 2, ed. by Sidney Dell (Amsterdam: North-Holland, 1987), pp. 38791.

    • Search Google Scholar
    • Export Citation
  • Schroder, Jurgen, “The Role of the SDRs in the International Monetary System,” in international and European Monetary Systems, ed. by Emil-Maria Claason (London: Praeger, 1990), pp. 6576.

    • Search Google Scholar
    • Export Citation
  • Thakur, Subhash M., “The Hard SDR: An Exploratory Analysis,” Staff Papers, International Monetary Fund, Vol. 41 (September 1994), pp. 46087.

    • Search Google Scholar
    • Export Citation
  • Williamson, John, The Failure of World Monetary Reform, 1971–74 (Sunbury-on-Thames, England: Nelson, 1977).


  • Crockett, Andrew, “The Role of Market and Official Channels in the Supply of International Liquidity,” in The International Monetary System, ed. by Peter B. Kenen, Francesco Papadia, and Fabrizio Saccomanni (Cambridge: Cambridge University Press, 1994), pp. 82100.

    • Search Google Scholar
    • Export Citation
  • Eichengreen, Barry, “Hegemonic Stability Theories of the International Monetary System,” in International Political Economy: Perspectives on Global Power and Wealth, ed. by Jeffry A. Frieden and David A. Lake (London; New York: Routledge, 1995), pp. 23054.

    • Search Google Scholar
    • Export Citation
  • Grieve Smith, John, “A Programme for Reform,” in Managing the Global Economy, ed. by Jonathan Michie and John Grieve Smith (Oxford: Oxford University Press, 1995), pp. 287304.

    • Search Google Scholar
    • Export Citation
  • Kindleberger, Charles P., Manias, Panics and Crashes: A History of Financial Crises (New York: Basic Books, 1978).

  • Masera, Rainer S., “Discussion,” in The International Monetary System, ed. by Peter B. Kenen, Francesco Papadia, and Fabrizio Saccomanni (Cambridge: Cambridge University Press, 1994), pp. 6573.

    • Search Google Scholar
    • Export Citation
  • Mundell, Robert A., “Discussion,” in The International Monetary System, ed. by Peter B. Kenen, Francesco Papadia, and Fabrizio Saccomanni (Cambridge: Cambridge University Press, 1994), pp. 7481.

    • Search Google Scholar
    • Export Citation
  • Triffin, Robert, Gold and the Dollar Crisis: The Future of Convertibility (New Haven, Connecticut: Yale University Press, 1960).

  • Williamson, John, “The Rise and Fall of the Concept of International Liquidity,” in The International Monetary System, ed. by Peter B. Kenen, Francesco Papadia, and Fabrizio Saccomanni (Cambridge: Cambridge University Press, 1994), pp. 5364.

    • Search Google Scholar
    • Export Citation

Horsefield chaps. 1 –2, passim.


Robert Roosa has recorded his discovery of the new asset: “It did not occur to many of us in the United States that, as dollars were paid out by the International Monetary Fund over the early postwar years, we were gaining a valuable asset in the parallel increase in our ‘super-gold tranche’ position, or, more properly, our ‘net creditor position in the Fund’.” (From a 1964 speech published in Roosa (1967), pp. 156–57.)


I recall a visit to the Fund in 1958 by Lucius Thompson-McCausIand, an advisor to the Bank of England, who raised the question of when it would become possible to revive the approach to liquidity creation of the Keynes Plan. 1 could tell him that in a quiet way we were doing exactly that by the approach to the use of nondollar currencies that the staff was developing and indeed putting into practice.


Nor, apparently, have the commercial banks been sufficiently keen on finding a way around this restriction to hold “positions in SDRs” without actually holding SDRs. The Fund itself did discover the needed loophole, as explained in its Annual Report, 1994 (p. 145): “As the Fund cannot hold SDRs in the SDA [Special Disbursement Account] or in its capacity as trustee of accounts administered for the benefit of members [the largest of which is the ESAF trust fund], SDRs for these accounts are held by prescribed holders [in fact, the Bank for International Settlements] on behalf of the Fund.”


Versions of this idea, without the need for amendment, have been proposed by Coats (1982) and Kenen (1983).


Since the Second Amendment of the Articles, this minimum is an amount equal to 25 percent of a member’s quota on April 1,1978, and a comparable amount for members that joined the Fund after that date.


The first published reference to the “liquidity ratio” so defined is found in the Fund’s Annual Report, 1990, p. 51, but the concept was used in internal documents some years earlier. A chart showing the ratio retroactively to 1979 was first published in the Annual Report, 1993, p. 82.


Duisenberg and Szász (1991), p. 260, Their statement indicating a “consensus” on this issue goes a little too far: there were important exceptions among directors.


For an earlier suggestion along these lines, see Polak (1979).


That assessment is very low as the Fund uses a narrow definition of the expenses allocated to the SDR Department. Thus, none of the costs of the 1994 Madrid Annual Meeting, where the issue of SDR allocation was one of the most prominent subjects, entered into the SDR assessment.


For members holding SDRs equal to their allocation, this charge is offset by the credit interest they receive. But for the many members that hold only few of the SDRs that were originally allocated to them, there is no such offset.


The authors thank Fabrizio Saccomanni, Pierpalo Benigno, various seminar participants, and James M. Bough ton for useful comments.


The example points to the role of “large” speculators and of “herd behavior” by other traders. For a discussion of these issues, see International Monetary Fund (1995).


Country-specific “sunspots” could also be internationally correlated, reflecting regional contagion effects on capital flows.


As shown by Obstfeld and Rogoff (1995), even within the OECD few central banks own reserves as large as the monetary base: in September 1994, out of 14 OECD central banks only 4 had nongold reserves at least as large as their monetary base, among them Mexico; if we count gold reserves as well, the number rises to 6.


Kindleberger (1978), pp. 186–7, describes a silver crisis in Hamburg in December 1857. A train came from Frankfurt full of silver. It pulled into the Hamburg railway station and left. The sight of the silver was enough to kill the attack.


The credit lines activated through the VSTFF mature 75 days after the end of the month in which intervention has taken place and can be renewed for a further period of three months at the request of the debtor central bank. However, while the initial credit line is unlimited, renewal is not: it cannot exceed a limit equal to twice the country’s quota in the short-term monetary support facility (STMS)—a mechanism designed to provide short-term financing for balance of payments needs. These quotas are remarkably smaller than the resources European countries can draw from the IMF, which are limited to 90 percent of a country’s quota (except in “special circumstances”). The ratios of IMF quotas to the quotas in the STMS, computed for 1987 using an ECU/SDR exchange rate of 1.1196, were: Benelux, 4.431; Denmark, 3.29; France, 3.08; Germany, 3.71; Ireland, 4.10; Italy, 2,998; Netherlands, 4.669; and United Kingdom, 4.526.

Central banks can also use the VSTFF for intramarginal intervention, but in this case access to the VSTFF is not automatic: it is subject to authorization of the central bank whose currency is being drawn.


Creating a secondary market for SDRs would help to redistribute the risk further. For such a market to function, however, the price of SDRs should no longer be pegged: SDRs would thus acquire characteristics that are similar to those of officially issued ECUs.


The views expressed in this paper are those of the author and do not necessarily reflect those of the authorities of the People’s Bank of China.


In fact some proposals were designed to reverse the underlending of the credit market: for example, at the 1985 Annual Meeting of the IMF, proposals by U.S. Secretary of the Treasury James A. Baker.


Thakur (1994) discusses the first two.


For a summary statement of the Managing Director’s proposal, see IMF Survey, May 3, 1993, p. 134. For a summary of the discussion of the issue by the Executive Board, see Annual Report, 1993, pp. 55–7.


The U.S. balance of payments recorded a deficit in 1950, then swung temporarily into surplus in 1951–52, just to register persistent shortfalls again. Eichengreen (1995), p. 246.


The generally cited passage is Triffin (1960). Eichengreen, Williamson, and others pointed out that Triffin had already called attention to this dynamic instability in 1947.


The team led by Rinaldo Ossola, president of the Group of Ten, made the milestone report in connection with the creation of the SDR.

  • de Vries, Margaret Garritsen, The International Monetary Fund 1966–71: The System Under Stress (Washington: International Monetary Fund, 1976).

    • Search Google Scholar
    • Export Citation
  • de Vries, Margaret Garritsen, The International Monetary Fund 1972–78: Cooperation on Trial (Washington: International Monetary Fund, 1985).

    • Search Google Scholar
    • Export Citation
  • de Vries, Margaret Garritsen, The IMF in a Changing World 1945–85 (Washington: International Monetary Fund, 1986).

  • Horsefield, Keith, The International Monetary Fund 1945–65: Twenty Years of International Monetary Cooperation (Washington: International Monetary Fund, 1969).

    • Search Google Scholar
    • Export Citation