Separate Exchange Markets for Capital and Current Transactions

During the past few years there has been considerable discussion about how to control international capital movements, particularly those occurring during periods of exchange rate uncertainty. While such movements have sometimes anticipated necessary exchange rate changes, this has not always been the case, and they have frequently merely hampered official efforts to restore order to exchange markets and to conduct effective national monetary policies. One possible method for dealing with these disequilibrating capital flows is a system of separate exchange markets for capital and current transactions. It has been argued that, in such a system, a fluctuating exchange rate for capital transactions would remove pressure from official reserves caused by large shifts in capital flows, while at the same time insulating foreign trade from exchange rate fluctuations and eliminating the need for inefficient discretionary restrictions on capital transactions.


During the past few years there has been considerable discussion about how to control international capital movements, particularly those occurring during periods of exchange rate uncertainty. While such movements have sometimes anticipated necessary exchange rate changes, this has not always been the case, and they have frequently merely hampered official efforts to restore order to exchange markets and to conduct effective national monetary policies. One possible method for dealing with these disequilibrating capital flows is a system of separate exchange markets for capital and current transactions. It has been argued that, in such a system, a fluctuating exchange rate for capital transactions would remove pressure from official reserves caused by large shifts in capital flows, while at the same time insulating foreign trade from exchange rate fluctuations and eliminating the need for inefficient discretionary restrictions on capital transactions.

During the past few years there has been considerable discussion about how to control international capital movements, particularly those occurring during periods of exchange rate uncertainty. While such movements have sometimes anticipated necessary exchange rate changes, this has not always been the case, and they have frequently merely hampered official efforts to restore order to exchange markets and to conduct effective national monetary policies. One possible method for dealing with these disequilibrating capital flows is a system of separate exchange markets for capital and current transactions. It has been argued that, in such a system, a fluctuating exchange rate for capital transactions would remove pressure from official reserves caused by large shifts in capital flows, while at the same time insulating foreign trade from exchange rate fluctuations and eliminating the need for inefficient discretionary restrictions on capital transactions.

Several writers have described the principles of a system of separate exchange markets,1 and the theoretical comparison of this system with alternative policy mixes has been developed in considerable detail. Meanwhile, a system of dual exchange markets, similar to that which had already been in effect for many years in the Belgian-Luxembourg Economic Union (BLEU) was adopted and maintained in France from 1971 to 1974 and in Italy from 1973 to 1974.2 The modus operandi of the dual exchange markets in these countries has been markedly different from that suggested by the theoretical models. The purpose of this paper is to bridge the gap between theory and practice, and thus to focus on the key problems inherent in operating a dual exchange market system. In order to permit concentration on the principal issues, some aspects of the dual exchange markets, such as the role of forward exchange markets, have been given less attention than they perhaps deserve. The paper also does not analyze the operations of some variants of the dual exchange market, such as markets through which only securities purchases are channeled.

The discussion is presented as follows. The first section briefly describes the model of a dual exchange market, as developed chiefly by Fleming (1974). The second section discusses the major ways in which dual markets are likely to differ—and have in practice differed—from the model and the implications of these differences. Succeeding sections discuss the performance of dual exchange markets with respect to overall balance of payments adjustment and in comparison with quantitative restrictions on capital movements. A concluding section summarizes the argument and attempts to specify the available policy choices with regard to operating a separate exchange market.

I. A Dual Exchange Market System: The Model

The fundamental characteristic of a dual exchange market system is the channeling of current and capital transactions into separate exchange markets—an official exchange market for current account transactions and a financial exchange market for capital account transactions.3 The domestic currency may stand at a discount or a premium in the financial exchange market, as compared with the official exchange rate. It is essential to the present model that the financial exchange rate be floating, although official intervention in the financial market is not precluded (see below). The discussion in this section and the next assumes a pegged official exchange rate; the consequences of a floating rate are discussed later in the paper.

In a dual exchange market system, international capital movements respond to changes in the financial exchange rate as the result of two separate motives on the part of economic units involved. The first of these is a rate of return motive, which arises from the fact that earnings on foreign investment are (as current account items) transacted at the official exchange rate, and therefore the effective rate of return on foreign investment is a function of the proportionate spread between the official and financial exchange rates. For instance, an increase in the premium on the domestic currency—let us, for brevity, call it the franc—in the financial exchange market lowers the effective rate of return on investment by nonresidents in the home country;4 if that premium is 5 per cent, a domestic interest rate of 8 per cent implies an effective rate of return of 7.6 per cent on investment in franc assets by nonresidents.5

The second way in which a change in the financial exchange rate influences capital flows is through a capital gains motive, which arises from the expectation that a change in the rate will be followed either by a reversal toward, or by a continued movement away from, the previous rate. In the first of these cases, funds tend to move out of the currency which has just appreciated, whereas in the opposite case funds continue to move into the appreciating currency.

In addition to these two mechanisms, by means of which private capital flows are determined by the level of the financial exchange rate, a crucial element in the operation of a dual market system is the policy of the authorities with regard to their intervention in the financial exchange market. At one extreme would be a nonintervention policy. Another alternative is a neutral intervention policy, according to which the monetary authority sells (buys) foreign exchange in the financial exchange market equal to the net increase (loss) in official reserves arising from a current account surplus (deficit). Thus, a neutral intervention policy ensures overall balance of payments equilibrium, with the imbalance on current account exactly offset by an equal imbalance of opposite sign on capital account.6 Any intervention policy not following either of these rules can be designated a variable intervention policy.

The standard argument for dual exchange markets assumes that a broadly neutral intervention policy is being pursued and that effective implementation of such a policy is feasible.

(1) An intervention policy can be described as broadly neutral even if it includes occasional or seasonal smoothing operations. Such a policy may also have to be varied over short periods in order to prevent speculation in the financial exchange market because of expected official intervention. A neutral intervention policy might also be modified to take into account desired changes in the exchange rate or in official reserves. Taking these qualifications into account, only a neutral intervention policy is defensible as a long-run policy; any other is inconsistent with an objective of overall external balance and cannot be compared with alternative means of achieving international payments adjustment (see Fleming, 1971, p. 303 and 1974, pp. 3–4).

(2) Effective implementation of a neutral intervention policy means essentially two things: that a spread between the official and financial exchange rates can be maintained which is large enough to influence the volume of capital flows and that official operations in the financial exchange market can offset speculative attacks on the official rate (for example, through leads and lags in trade payments). To the extent that a spread between the two rates is not maintained, the dual exchange market loses its raison d’être of dampening speculative capital movements; and whenever the authorities are unable to intervene in the financial exchange market to the extent of changes in official reserves, they will be unable to protect the official exchange rate without undesired reserve changes.

If, however, both of these conditions are sustained in practice, it follows directly that the dual exchange market is a superior instrument for the control of disequilibrating capital flows compared to several major alternatives. While performing the same task as quantitative controls over capital movements, it has the advantages over the latter system of being automatic and of allocating funds efficiently among different types of capital flow (see Fleming, 1974, pp. 6–11). It is a more flexible instrument than the fiscal-monetary policy mix and is less likely than the latter to interfere with growth objectives (see Fleming, 1974, pp. 13–15 and Baratierri and Ragazzi, pp. 364–69). In comparison with a floating unified exchange rate, it ensures an environment of greater exchange rate stability for foreign trade.7 In contrast to any system of fixed unified rates, it relieves official reserves from the pressures caused by volatile capital movements. Moreover, it ensures—provided a neutral intervention policy is pursued—a sustained overall payments balance.

In addition to these advantages, which follow from the assumptions listed above, it has been claimed that there are two additional advantages of a dual market system: that such a system is inherently more likely to produce rational and internationally cooperative exchange rate action on the part of national authorities than is either a system of fixed unified rates or one of floating unified rates, and that separation of the two exchange markets is more efficient for allocating resources and less costly to administer than is a system of quantitative controls. These points will be discussed later in this paper.

First, however, it will be necessary to examine in greater detail the conditions upon which the major potential benefits of a dual exchange market system depend—the willingness and ability of the authorities to maintain both separate exchange markets and a neutral intervention policy.

II. Incomplete Market Separation and its Consequences

The prerequisite for a successful dual exchange market is that the two exchange markets are in fact kept effectively separate. To be sure, perfect separation of the markets is virtually impossible; in practice, significant links between dual exchange markets have been present whenever such a system has been tried. Whether these links have impeded achieving the ultimate objectives of separate exchange markets is a matter of degree and, in the absence of empirical evidence, of judgment. However, it is clear that to the extent that links are established between an official and a financial exchange market, the system is undermined in two ways: first, the sustainable spread between the two exchange rates is reduced, thereby diminishing the incentives to capital flows to move in a counter-speculative direction; and, second, capital movements initially passing through the financial exchange market tend to spill over into the official market, thereby making it more difficult for the monetary authorities to insulate the official rate from speculative disturbances. It is thus useful to examine in some detail the difficulties encountered in trying to separate exchange markets and the types of links between the market that can arise even in the presence of controls.

The separation of the financial exchange market from the official exchange market requires extensive controls over both the foreign exchange transactions of residents and the domestic currency (franc) transactions of nonresidents. Generally, controlling the former requires a system of exchange controls, while for the latter it necessitates regulation of nonresidents’ franc bank accounts. In recent European experience, such regulation has been based on a separation between two types of nonresident accounts, one pertaining to current and the other to capital transactions. There have ordinarily been no direct limitations on the increase or reduction of balances in nonresident accounts;8 the control system affects solely transfers from these accounts to residents.

The freedom to purchase or sell foreign currencies in exchange for francs in nonresident “commercial” accounts permits one type of capital movement, in the form of changes in nonresidents’ balances in domestic currency, to take place in the official exchange market. Changes in nonresident balances have in recent years been an important means of currency speculation, yet outright prohibition of such changes would interfere fundamentally with the mechanism for effecting trade payments and has therefore not been undertaken except for short periods in extraordinary circumstances.

Another capital item which has been included in the official market is trade financing. “Leads and lags” in the settlement of commercial claims, or variations in the initial terms of commercial financing, are believed to have been responsible for a large proportion of the short-term capital movements occurring during the foreign exchange crises of recent years. Thus, so long as trade financing9 remains in the official market, the dual exchange market cannot effectively act as a buffer for official reserves against speculative forces. However, a separation of trade financing from the exchange market for current transactions would not only be extremely difficult to administer but would also eliminate an important advantage of separate exchange markets—namely, that they shield current account transactions from the exchange rate fluctuations caused by capital movements.

There are, moreover, certain types of international transfers which, while technically current account items, can easily be used as vehicles for the acquisition of foreign assets. These include remittances of savings by foreign workers, family remittances, and tourist expenditures. Including these items in the financial exchange market does away with the danger that they serve as a channel for capital movements coming through the official market. However, conducting certain current account transactions (such as tourism) at a different exchange rate from others (such as commodity trade) does involve a palpable distortion of relative prices.10 Obviously, this objection is of substantive importance only if the spread between the official and financial exchange rates is significant.

Another possible channel for disguised capital movements deserves separate mention: payments related to capital service or royalties. Such items have in fact been included in the financial franc market in France. While this eliminates the possibility of capital movements occurring under this guise in the official market, it also eliminates the rate of return motive for moving capital between countries in response to changes in the spread between the official and financial exchange rates.

Another method for either concealed capital movements through the official market or concealed current account payments in the financial exchange market is the false invoicing of trade transactions. For instance, if the domestic currency is relatively depreciated in the financial exchange market, there will be an incentive for underinvoicing exports, with the difference between the actual and the invoiced export proceeds entering the country in the guise of a capital inflow at the more favorable financial exchange rate.11 Alternatively, the noninvoiced export proceeds may be retained abroad, which effectively eliminates for the exporter the exchange loss he would have suffered if he had surrendered those proceeds in the official market and then purchased foreign exchange in the financial exchange market in order to effect the corresponding capital outflow.

In the cases just mentioned, the effect of evasion is to place additional pressure on official reserves and to narrow the spread between the exchange rates in the two markets: for example, when the financial franc is relatively depreciated, the consequent underinvoicing of exports tends to reduce the inflow of foreign exchange on the official market and to appreciate the financial franc. The same effect results from the above-mentioned cases of capital movements occurring through the official market. For instance, to the extent that a net capital inflow moves through the official market, there is a tendency for official reserves to rise and for the relative appreciation of the financial exchange rate to have been less than it would have been if separation between the two exchange markets had been more complete.

Another link between the two exchange markets may arise from the operation of forward markets. If the authorities do not take steps to assign particular types of forward exchange transactions to one or the other exchange market—for example, trade-related forward exchange dealings to the official market—there will be widespread arbitrage between the two forward markets and, as a consequence, strong transmission of speculative pressures between the two spot exchange markets. Trade-related forward dealings have in fact been assigned to the official market in the BLEU, France, and Italy. Even in these cases, however, forward exchange dealings can create indirect links between the two spot markets, so long as a forward market in financial francs is permitted to operate. This occurs in the first instance because the forward premium or discount on the financial franc gives a clear signal of market expectations, thereby affecting the supply and demand for both forward and spot exchange in the official market. Moreover, banks generally make forward transactions on their own account to offset the effect on their foreign currency positions of their clients’ trade-related forward dealings; since the banks are likely to be balancing their overall positions—rather than their positions within each market—their own account dealings would take place in the financial exchange market, and the forward rates that banks offer their official-market clients would be related to the forward rate in the financial market. The two forward markets would, therefore, be closely linked and the spot markets thereby indirectly so.12

In summary, it is virtually inevitable that links will exist between the two exchange markets, in the form either of capital transactions taking place legally or illegally in the official exchange market or of current transactions taking place in the financial exchange market. These links tend to lead to a breakdown of the separation between the two markets. This breakdown occurs in one or both of two ways: (1) the financial exchange rate tends to stay close to the official rate, and (2) official reserves are affected directly by speculative flows. The first implies the second, since it comes about by a shift of certain transactions from one market to the other. However, (2) may occur without (1) occurring—for example, through shifts in the leads and lags in trade payments, which do not represent any shift of transactions from the financial to the official exchange market.

The extent to which links between the two exchange markets lead to a breakdown of the separation between the two markets in either of the above senses depends on several factors. One is the size of the links relative to the magnitude of transactions in the financial market. Another is the cost of evading the market separation. This cost depends on the stringency with which the system of controls designed to separate the exchange markets is administered and on the severity of the penalties for evading those controls. It is an important factor only to the extent that capital flows cannot take place legally through the official market, nor current transactions, in the financial market. If these possibilities were nonexistent—if evasion were the only link between the exchange markets—and if at the same time possibilities for evasion were so large that transactions involving evasion could dominate the financial exchange rate, the spread between the two exchange rates would not exceed the costs of undertaking evasion.13 Whenever there is a substantial mixing of capital and current transactions in at least one of the markets, the cost of evasion does not ensure that rates in the two markets are kept apart, although it remains a factor tending to keep them apart.

Finally, however, the intervention policy of the authorities determines whether or not links between the markets actually bring about unification of the exchange rate or changes in official reserves. A nonintervention policy permits one or both of these to occur. On the other hand, a neutral intervention policy would generally offset the effects of a shift in transactions from one market to the other—for example, a capital flow channeled through the official market. For this reason, a neutral intervention policy is necessary if a dual exchange market is to defend the official exchange rate against heavy speculative capital movements. The chief technical drawback of such a policy—that it involves increases or decreases in the monetary authority’s liabilities to the public (see Section I)—would in most circumstances be rather minor compared to the advantage of providing the authorities with the means of preventing an undesired change in the official exchange rate.

In some circumstances, however, a neutral intervention policy may not entirely offset the effect on the financial exchange rate of illegal evasions of regulations or of illegal arbitrage between the two markets. This is because such illegal operations may establish an implicit financial exchange rate which changes the net demand for foreign exchange on the financial market at the prevailing market rate. An analysis of such operations is given in Appendix II.

Taking the preceding caveat into account, it is nevertheless reasonable to expect that a neutral intervention policy would offset much of the effect of evasion and arbitrage. However, despite the apparent superiority of a neutral intervention policy over a nonintervention policy when large speculative capital movements are tending to break down the separation between the two exchange markets, recent European experience reveals a widespread reluctance to intervene in the financial exchange market. What intervention has taken place seems to have been only in response to minor flurries or seasonal factors, and it appears that neither systematic intervention over the longer term nor massive short-run intervention to meet major speculative disturbances has been undertaken. This may be due in part to fear on the part of the authorities that the financial exchange market would become heavily influenced by expectations with regard to official intervention, and that any regular pattern of official intervention, such as a fixed intervention point, would encourage formation of such expectations. In addition, governments may feel that the combination of heavy official intervention in the financial exchange market and a wide or increased spread between the two exchange rates14 would be taken by the market as signs of pressure on the official rate and would consequently provoke additional speculation through the official market; simultaneously, the higher spread between rates would lead to widespread evasion. While in theory it is conceivable that a spiral of ever-increasing pressure on official reserves could be offset by ever-increasing official intervention, this is apparently not considered by monetary authorities to be a manageable policy.15

As a consequence of avoiding large-scale or protracted official intervention in the financial exchange market, the authorities in the countries concerned have found it necessary to “balance” each market, if necessary by permitting or requiring specific categories of current transactions to be carried out in the financial market, or capital transactions in the official market. This mixing of transactions is also influenced by the need to avoid evasion.16

For example, in Belgium, inward and outward payments on account of capital service and certain other current account items may be settled in either the official or the “free” market; moreover, private travel expenses, when settled with banknotes, are transacted in the free market. Under the system used in France from August 23, 1971 to March 21, 1974, current payments (by the private sector) passing through the financial exchange market included such large items as nonresident travel expenditures in France, residents’ travel expenditures abroad, capital service (except on commercial credits), and workers’ remittances. Under the Italian scheme in effect from January 22, 1973 to March 22, 1974, current and capital transactions were as nearly as possible separated into the corresponding exchange markets; however, as in the other countries with a dual exchange market, commercial credits remained in the official market and all purchases and sales of foreign banknotes—which accounted for a substantial portion of tourist expenditures and workers’ remittances—were effected in the financial exchange market. Details on the operation of these markets are given in Appendix III.

The actual structure of the major European dual exchange markets and the policies followed by the authorities suggest that this type of exchange system will operate differently from the model described in the previous section. Because there are two major types of capital flows which can take place legally through the official market—namely, variations in the terms of payment and changes in nonresident domestic currency balances—speculative pressures have had a direct effect on official reserves. These pressures have on occasion been massive and have prompted the authorities to take measures of direct control over capital movements (see Appendix III). Moreover, the considerable links existing between the two markets have meant that a large spread encourages considerable evasion, making it eventually impossible to sustain such a spread over a long period of time.17 This in turn means that the rate of return motive has had little role to play in such a system and that any heavy or sustained speculative pressures operating in the financial market have tended to spill over into the official market. Under these circumstances, the financial exchange market has at best mitigated speculative capital movements; it has certainly not prevented the countries involved from being swept along in the major exchange crises that have occurred during this period.

The relatively poor data on capital movements for countries employing dual exchange markets, as well as the often dominating importance of various other factors, makes it difficult to substantiate the above assertions on the basis of quantitative data. The spreads between the official and financial exchange rates of the BLEU, France, and Italy (see Table 1) have tended to be under 5 per cent.18 In the BLEU, the only period of substantial divergence between the two rates during the period under consideration was April–July 1974, and again in April–June 1975, during which time the financial exchange rate was usually at a discount of 3 to 5 per cent to the official rate; in all other months during the period covered the differential between the two rates was under 2 per cent. In France, the spread exceeded 4 per cent from mid-June through August 1972, for a few days in December 1971, and also for a week in November–December 1973. However, in Italy, spreads of between 4 and 6 per cent prevailed during 8 of the 15 months during which the financial exchange market was in existence.19

Table 1.

Percentage Spread Between Financial and Official Exchange Rates

(Monthly averages; + = premium of financial over official rate)

article image
Source: New York quotations at 12:00 noon, as reported by the IMF Treasurer’s Department.

Financial exchange market introduced effective August 23, 1971.

Financial exchange market introduced effective January 22, 1973.

Financial exchange market abolished effective March 21, 1974.

Financial exchange market abolished effective March 22, 1974.

The significance of this data is somewhat lessened by ignorance as to the extent of direct or indirect official intervention in the financial exchange market. By “indirect intervention” is meant the timing of capital movements under government control or influence (or the choice of market for such movements): for instance, the “compensatory borrowing” arranged by the Bank of Italy but actually carried out by private or semipublic firms. It is perhaps significant, however, that whenever heavy speculative pressures were occurring, the authorities found it necessary to introduce or intensify quantitative controls on capital movements, or, alternatively, to move the official exchange rate. The actions of the authorities themselves, as well as the large reserve movements which occasionally took place, suggest the inadequacy of dual exchange markets to stem severe speculative pressures.20

Official actions in early 1974 also suggest a desire for a balanced set of transactions in each market. The French and Italian financial exchange markets were abolished in March, at a time when the probable impact of the changes in oil prices was to produce greater imbalances in both current and capital accounts. The BLEU has retained its dual exchange market; although this is presumably due to the fact that the marché libre is a long-established feature of the BLEU banking and exchange system, it is also significant that, unlike France and Italy, the BLEU did not face a large current account deficit in 1974. However, in January 1974, in the face of possible current account pressures, the BLEU authorities permitted outward payments of investment earnings to be made through either exchange market, instead of only through the official market; this again indicates a tendency to balance exchange markets.

To summarize the above discussion, a neutral intervention policy has in fact not been followed by industrial countries with dual exchange markets, and, for this reason, the dual exchange market has not played as major a role in dampening speculative capital flows as one would expect on the basis of a theoretical model. It is, of course, possible that a dual exchange market would have been a more effective instrument for controlling capital flows if a neutral intervention policy had been followed. Nonetheless, as shown above, such a policy alone might not prevent a tendency for the exchange rates to move together; the evasion possibly producing this tendency could be prevented only by extensive administrative controls. A more essential point, however, is that large-scale official intervention in the financial exchange market could lead to destabilizing speculative capital flows, both in the financial and in the official market. The latter will necessarily be prone to such flows through two channels—leads and lags in trade payments and changes in foreign “current account” balances in the domestic currency. These two items cannot be separated from current payments; at the same time, quantitative restrictions with regard to either are unlikely to be effective without hampering seriously both trading relationships and trade-financing procedures.21

There are two alternative ways of managing a dual exchange market system. First, there is a nonintervention or variable intervention policy, which, given the links between the two exchange markets, can only be sustained if markets are balanced; this in turn may require a mixing of transactions, which to some degree runs counter to the rationale of such a system—namely, that it insulates current account transactions from exchange rate disturbances caused by capital movements. Such a policy tends to break down in the face of heavy speculative pressures, because such pressures tend either to be channeled through or to spill over into the official market, and the only adequate official response—large-scale intervention—is by definition ruled out as an alternative. Second, there is a neutral intervention policy, which runs the serious risk of provoking destabilizing capital movements precisely when the need for dampening such movements is most needed—when there is nascent speculation, leading to changes in leads and lags and in nonresident balances which are offset by abnormally large official intervention in the financial exchange market. But a growing spread between the two exchange rates will eventually become impossible to sustain in the face of growing incentives for evasion; and growing official intervention in the financial exchange market may at some stage become both impractical and politically distasteful. These are apparently the reasons why, in practice, the first alternative has been chosen.

III. Dual Exchange Markets and Balance of Payments Adjustment

In the previous section, the efficacy of the dual exchange market was discussed in relation to disequilibrating capital movements of a temporary nature. However, an exchange system cannot be judged solely by the scope that it affords the monetary authorities to defend the exchange rate against speculative attacks. A criterion of at least equal importance is whether such a system tends to promote achievement of general balance of payments objectives. As Williamson (1973) has recently shown, these objectives are more complicated than the traditionally cited zero balance. A country may wish to readjust its level of official reserves at a particular rate of change, and at the same time maintain a particular structure of the balance of payments as reflected in the corresponding balances on current and “underlying” capital account.22

When this approach is taken, the number of balance of payments targets is two instead of the traditional one, and thus two instruments are required to reach them. For instance, while the exchange rate may be used as the instrument to bring about the desired rate of reserve accumulation or decumulation, given the net flow of long-term capital, some other instrument (such as direct controls) is generally necessary to bring that net flow to the desired level. This is a fortiori the case if the exchange rate is a determining factor for “underlying” capital movements, since, in general, the exchange rate that produces the desired change in reserves will not be the same as that which brings about the desired underlying capital flow.

From this standpoint, a dual exchange market appears at first glance to have important advantages. With such a market, the monetary authorities can vary both the official exchange rate and their intervention policy in the financial exchange market so as to produce just the combination of overall reserve accumulation or decumulation that they desire, together with their target for either capital or current balance.23 However, the capital account balance in this case includes short-term reversible movements. To bring about a desired level of long-term capital flow would require additional measures (a tax-subsidy mix or controls). Thus, if it is the long-term capital account which is of particular interest to the policymakers as regards the structure of the balance of payments, the dual exchange market cannot replace traditional capital control techniques. If, on the other hand, it is the current account on which interest is focused24—i.e., the authorities are indifferent as to the composition of the capital account—then the dual exchange market would accomplish the twin balance of payments objectives subject, however, to the qualification that there can be, as discussed above, substantial short-term capital movements channeled through the official market. While such movements are often of a temporary nature, a longer-term current account objective might have to be corrected for secular trends in the levels of nonresident domestic currency balances and of trade financing.

From an international viewpoint, it is important to judge how well a dual exchange market system would perform, relative to alternative systems, with regard to the environment for overall balance of payments adjustment. One might phrase this criterion as follows: under such a system, how quickly do the pressures created by payments imbalance lead to exchange rate changes, either directly (as in floating rates) or indirectly through decisions of the authorities?

It has come to be recognized, for example, that, under a system of fixed exchange rates, changes in official reserves exert a stronger pressure on the authorities of deficit countries than on those of surplus countries and that in either case exchange rate action has often been very sluggish in responding even to persistent reserve changes. By contrast, under a system of unified floating rates, the exchange rate adjusts to every change—no matter how temporary—in the balance of payments (except to the extent that the rate is managed through official intervention). Would a system of dual exchange markets with a neutral intervention policy offer a reasonable compromise between these extremes of underadjustment and overadjustment of the exchange rate?25

In attempting to answer this question, two subsidiary questions must be dealt with: (1) under a dual exchange market system is it likely to be easier than under alternative systems for the national authorities to perceive and correctly diagnose a persistent imbalance in international payments; and (2) under a dual exchange market system, are the pressures created by such an imbalance likely to result in a more or less appropriate adjustment than under alternative exchange rate systems?

With respect to the first question, let us begin by comparing a system of dual exchange rates (with a fixed official rate) with a system of fixed unified rates. Under the latter system, it has sometimes been noted that a basic disequilibrium in the balance of payments is easier to recognize in practice than to define in theory. The opposite might well be the case under a dual exchange rate system, particularly if such a system were employed by several major countries. A simple definition of equilibrium can be derived from the model set out in Section I, but the system as actually operated would produce more ambiguous results than the traditional criterion of reserve changes.

From the model of a single country with a dual exchange rate system and a neutral intervention policy, it follows that if the financial exchange rate persistently diverges in the same direction from the official rate, there must be a basic disequilibrium in the balance of payments, i.e., there would be a persistent loss of reserves if the exchange rate were unified.26 This must be the case if the net inflow of capital responds positively to a depreciation in the financial exchange rate. For example, suppose the domestic currency is at a persistent discount on the financial exchange market. At that discount, under a neutral intervention policy, the net capital inflow is just equal to the current account balance. If exchange markets were unified at the official exchange rate, the net capital inflow would decline,27 the current account would remain unchanged, and there would, therefore, be a loss of official reserves. By similar reasoning, if the domestic currency is at a persistent premium on the financial market, an underlying surplus in the balance of payments is indicated.28

It follows from this line of reasoning that, if a neutral intervention policy is being followed, a persistent discount or premium on the home currency in the financial exchange market is as clear a signal for the need to change the official exchange rate as a persistent decrease or increase in reserves would be. However, this is true only in the case of a neutral intervention policy. For example, with a nonintervention policy, there may be a net gain or loss of official reserves equal to the balance on current account, and a zero balance on capital account: under these circumstances, a persistently appreciated or depreciated financial exchange rate does not indicate unambiguously whether an underlying disequilibrium exists.29 This is a fortiori true for a variable intervention policy. Since, in practice, either nonintervention or variable intervention policies have been followed and there has been a substantial mixing of current and capital transactions in the two exchange markets, any persistent undirectional spread between the two exchange rates observed in practice will give only an uncertain indication of the need to change the official rate.

The second question to be considered is whether a persistent spread between the official and financial exchange rates is likely to exert greater pressure for adjustment than a persistent change in reserves. Let us first consider the case of a dual exchange market system with a fixed official exchange rate and a neutral intervention policy. It is clear that under such a system a discount on the franc in the financial exchange market could, at least in principle, be sustained indefinitely; the same is not true of a loss of reserves. The contrast between a premium on the franc in the financial exchange market and an accumulation of reserves is less sharp, since a reserve increase puts less pressure on the monetary authorities to take exchange rate action than does a reserve decrease. However, events of recent years have shown that the effect of a large cumulative increase in reserves can be considerable; in contrast, there is no cumulative effect in the case of a sustained premium on the domestic currency in the financial exchange market. For these reasons, reserve changes would appear in both cases to be more effective than a spread between official and financial exchange rates for inducing monetary authorities to carry out appropriate exchange rate adjustments.

When one considers the dual exchange market system as it has actually operated in recent European experience, the contrast with fixed unified exchange rates becomes less marked. Since, as explained in the previous section, a dual exchange market system with a nonintervention or variable intervention policy does not in practice eliminate large-scale changes in official reserves, the pressures on the authorities under such a system are different only in degree, and most likely not in very large degree, from those under fixed exchange rates. For analogous reasons, a dual exchange market system with a floating official exchange rate—what Fleming (1974, pp. 24–25) has called a “hybrid system”—will probably in practice differ little from a unified floating rate; in either case, the exchange rate relevant to foreign trade will be subject to considerable variation, unless there is a large element of official management.30

At the present time it would be most relevant to compare a dual exchange market (with a managed floating official exchange rate) with the highly managed floating unified exchange rates now prevailing. As implied above, the differences between these two alternatives from the standpoint of an individual country are not likely to be very great. There might, however, be a significant difference from an international standpoint if a number of major countries established dual exchange markets. The double exchange rates for each country could, as Fleming (1974, p. 7) has emphasized, create large spreads between financial exchange rates: this would imply possibly substantial underlying disequilibria. However, it would no longer be possible in a multilateral context to gauge the existence of a disequilibrium for an individual country by whether the financial exchange rate was appreciated or depreciated relative to the official rate. It would thus become even more difficult than it is at present to find a generally accepted and credible basis for international agreement on exchange rates31 and thus to achieve some measure of international collaboration in this field.

IV. Dual Exchange Markets Compared with Quantitative Restrictions on Capital Transactions

There can be no doubt that allocative efficiency is greater under a system of separate exchange markets than when capital movements are controlled by direct quantitative restrictions (QRs) (Fleming, 1974, pp. 8–13 and 25–27). However, it is arguable that, as has also been claimed, a dual exchange market entails lower administrative costs than quantitative restrictions on capital transactions. This section is chiefly concerned with the latter issue, with brief mention of certain other considerations which might tend to favor the use of QRs on capital movements rather than the use of a dual exchange market.

Fleming argues for the lower administrative costs of a dual exchange market system on the grounds that QRs are “inevitably uneven in their incidence,” unlike the dual exchange market, which creates the equivalent of a uniform tax or subsidy on all capital movements. The uneven incidence of QRs implies that the capital flows being inhibited by QRs must be of higher average profitability than those flows being discouraged by a dual exchange market; thus, the incentive to evade controls must on average be higher under QRs than under a dual exchange market system. Fleming (1974, pp. 10–11) concludes that “… if a set of administrative restrictions is compared with a particular level of the capital [financial] exchange rate (relative to the exchange rate on current transactions) both of which, in the absence of evasion, would have an equal effect on the balance of payments, the capital [financial] rate would be likely to give a smaller incentive to evasion than would the quantitative measures…. Since evasion would be less profitable under a dual market than under a regime of direct administrative restrictions over capital flows, it may be assumed that there would be less of it for a given degree of severity in the policing of the controls or that the policing itself would be less severe.”

To adopt Fleming’s conclusion in unqualified form, it would be necessary to assume that the administrative cost of a dual exchange market system “in the absence of evasion” is no more than that of a system of QRs.32 However, this is not necessarily the case. A dual exchange market (in the form in which it has been set up in the BLEU, France, and Italy) requires the additional bookkeeping involved in two sets of accounts for nonresidents’ domestic currency balances and residents’ foreign currency balances, whereas the QRs may require little administrative cost at all if they take the form of banning outright certain forms of capital flow but making no discriminatory judgments about other forms.

An additional premise necessary to support Fleming’s conclusion is that the administrative cost of preventing evasion (per unit of balance of payments effect) is at least as great under QRs as under the dual exchange market. This is not necessarily true, particularly when a system of separate exchange markets is seen as a whole and not just with respect to the direct task of controlling capital transactions. It is argued below that not only does a dual exchange market system share much, if not most, of the costs of a system relying on QRs on capital transactions, but in some respects it may be more costly to administer effectively.

While it has been plausibly argued that capital transactions could be directed through the financial exchange market solely by means of complete control of current transactions,33 the opposite is not the case. To achieve reasonably effective control over the entire range of capital transactions, it is necessary to have concurrent controls over current transactions. This would not be the case if it were possible for the authorities to obtain reliable documentation about each capital transaction and noncurrent account transfer; but this would be exceedingly difficult for domestic assets being transferred and virtually impossible for foreign assets. The bulk of current transactions by contrast are largely related to more readily measurable items such as the value of goods passing through customs, the domestic earnings of foreign workers, tourist expenditures, and so on. Thus, strict controls can reduce the incidence of such practices as the underinvoicing or overinvoicing of exports and imports, but it is difficult to conceive of a way to control directly the proper invoicing of all capital transactions, so as to avoid either a current account component in such transactions or hidden capital movements in the official exchange market, without simultaneous controls over current transactions.

This conclusion applies, of course, to QRs as well as to the dual exchange market. However, it follows from what has been said and from the far greater number of individual current account than capital account transactions, that the bulk of administrative costs under either system of capital controls must be spent on controlling current account transactions. This is a fortiori the case if we accept as part of an effective dual market system the supplementary controls suggested by Fleming (1974, pp. 13–14 and 23) to avoid speculative capital movements through variations in nonresident balances (for current account transactions) and in the terms of payment in foreign trade. Nothing said thus far contradicts the conclusion that controls over capital transactions per se are less costly under the dual market, but that conclusion has been put in a necessary perspective. In view of the costs of controlling current transactions, the saving of costs of capital controls entailed by a dual exchange market would be a relatively minor factor. Moreover, the system of double accounts for nonresidents’ domestic currency and residents’ foreign currency accounts increases the public and private costs of conducting and controlling both current and capital transactions. It is thus an open question whether quantitative capital controls entail a higher net cost at all for the economy as a whole, even if the allocative inefficiencies created by QRs are taken into account.

Aside from cost considerations, which do not clearly favor the dual exchange market, there are other reasons why, at least in some cases, QRs may be preferred as a means of controlling capital movements. One such reason is that changes in QRs are likely to have a quicker and more predictable effect than relative exchange rate movements in the dual market. This is particularly the case when an exchange crisis arises. In recent years, in countries with stringent direct controls over capital movements, there have been only two significant loopholes—nonresident bank balances in domestic currency and leads and lags. As already mentioned, the same loopholes are no less a problem with a dual exchange market. However, when such a market is in operation, an exchange crisis brings simultaneous pressures in the official market (nonresident balances and leads and lags) and in the financial market; the pressures in the latter only exacerbate, directly and indirectly, the pressures in the former, thus introducing an additional element of potential instability to the system that is impossible to predict in advance or control when it occurs. By comparison, the system of QRs, while possessing the same loopholes, seems at least the more predictable of the two alternatives.

Quantitative restrictions on capital flows may also be preferred when there is a separate policy objective relating to the size, composition, and direction of long-term capital movements (see Section III). Since the dual exchange market provides no way of discriminating between short-term and long-term capital flows, such an objective can be achieved only by means of direct restrictions on certain types of capital movements, notably direct investment, securities purchases, and real estate transactions. Moreover, once quantitative or discretionary controls are in effect for these types of transactions, the marginal cost of extending these controls to other types of capital transactions, if and when this is required for overall balance of payments reasons, may be relatively small compared to the cost of installing a dual exchange market system.34

Nothing that has so far been said should be construed as a general argument in favor of QRs. This method of capital controls, together with the controls over current transactions required to make the controls over capital transactions effective, entails substantial administrative costs and misallocation of resources. To a large extent, however, this also is true of a dual exchange market. Unless there is a separate balance of payments objective with regard to the long-term capital account, it is by no means clear that either of the alternatives being considered here is superior to floating unified exchange rates or a system under which exchange rates are relatively stable and imbalances are dealt with by some combination of financing and an adjustment of monetary policies. However, when the latter methods are inadequate and the monetary authorities nevertheless desire to hold an exchange rate against speculative pressure, either QRs or a dual exchange market—or, conceivably, some combination of the two—will be required. The dual exchange market is superior on grounds of allocative efficiency but not necessarily on any others.

V. Conclusions

In Section II it was concluded that in practice the device of separate exchange markets for capital and current transactions has been of limited usefulness, inadequate to halt or to protect official reserves from heavy or protracted speculative pressures. However, in the absence of comparisons between dual exchange markets and alternative methods of controlling capital movements, this conclusion, although meaningful, is somewhat empty. Such comparisons have been developed by Fleming (1974) and by Barattieri and Ragazzi (1971, pp. 363–64 and 368–72), and some comparisons between a dual exchange market and alternative arrangements have been made earlier in this paper. Rather than repeat these discussions, it is preferable to examine whether the effectiveness of dual exchange markets might be increased by direct controls over certain capital movements, by allowing the official exchange rate to float, or by a more active intervention policy. In addition to focusing on the policy choices involved in managing dual exchange markets, this discussion may also make it possible to delimit more precisely than before the specific contribution of a dual exchange market system to the task of controlling disruptive capital movements.

First of all, it has been suggested (Fleming, 1974, pp. 11 and 19) that a system of dual exchange markets would provide a more effective barrier against speculative capital movements if it was supplemented by direct restrictions on changes in nonresident domestic currency deposits and on variations in the terms of commercial payments. However, this conclusion is greatly weakened by the consideration that in no industrial country have the authorities done more than reduce somewhat the possible scope for leads and lags. This is not to say that a much stricter control of payments terms is inconceivable; but as a practical matter severe controls of this sort would in many countries be considered as an unacceptably great impediment to normal commerce except in wartime circumstances and they would involve a considerable administrative burden. Similar considerations apply to severe restrictions on nonresident domestic currency deposits, although here the scope for official action is greater. However, while some improvement may be possible along these lines, it is unlikely that this would suffice to prevent heavy pressures on official reserves during exchange crises.

Another means of possibly increasing the effectiveness of a dual exchange market would be a “hybrid system” in which the official exchange rate is allowed to float.35 To some extent, such a system would no longer fulfill one of the chief objectives of maintaining separate exchange markets—namely, insulating foreign trade from exchange rate fluctuations caused by volatile capital flows. However, such a system might be appropriate in circumstances where there exists a basic disequilibrium in the balance of payments or where demand management is inadequate.

At first glance, it might seem that this hybrid system, like successful hybrids in agriculture, would combine the advantages of both its genetic forebears: the stabler exchange rate for foreign trade provided by dual exchange markets, and the more flexible response to irresistible speculative pressures provided by a unified floating exchange rate.36 This would presumably result if the authorities followed a neutral intervention policy, modified by allowing the official rate to move whenever this is considered appropriate, and perhaps also by permitting a fairly wide margin of variation around the official rate (Fleming, 1974, pp. 19–20). Such a policy might in practice be little different from one in which an official central rate is fixed within wide margins but changed fairly frequently (e.g., once or twice a year). In fact, in times of exchange rate uncertainty, the difference between a dual exchange market system with a “fixed” official rate and one with a “floating” official rate diminishes. In either case, links between the two exchange markets tend to put pressures on official reserves, and the authorities will either have to bend to these pressures or be willing to engage in massive intervention in the financial exchange market. As a matter of fact, this description of the problem faced by the authorities would nearly fit the situation when a unified exchange rate is floating during a period of uncertainty. However, to the extent that a dual exchange market succeeds in insulating the official market from speculative pressures, it may also increase the temptation to the authorities to avoid taking justified exchange rate action.

Once again, the logic of the discussion leads to the fundamental point that the effectiveness of a dual exchange market depends crucially on the type of intervention policy followed by the authorities. If the latter are prepared to follow a policy of neutral intervention, or at least a policy of variable intervention with heavy intervention in the financial market when necessary, it is possible that dual exchange markets, whether with a “fixed” or a “floating” official rate, might be effective in stemming speculative capital flows.37 If the authorities are not prepared to follow such a policy, dual exchange markets can provide only weak and temporary protection. They will be effective only in discouraging speculative flows through certain channels—such as securities and real estate transactions—which cannot be easily substituted or disguised. However, this function could also be performed by a more limited secondary market such as the investment currency market in the United Kingdom and the O-Guilder market in the Netherlands; in fact, the effectiveness of such markets with respect to the particular types of capital transactions involved (but not with respect to the capital account as a whole) might be greater than that of an expanded dual market because of the greater premia or discounts which are possible to maintain in the markets of more limited scope.

Official reluctance to intervene on a large scale, or even at all, in the financial exchange market has been discussed earlier. This reluctance is apparently related to a fear that such intervention might in some circumstances trigger increased speculative capital flows, and it may also be felt that capital flows tend to be more equilibrating in a free market than in one in which visible official intervention occurs. To be sure, a similar argument can be made against official intervention in a unified exchange market. However, without judging the validity of these arguments, it should be noted that they ought to be balanced against the obvious advantage of an active intervention policy, namely, that it makes it possible to defend the official rate while at the same time shielding official reserves from undesired changes. In the absence of past experience, it is impossible to tell whether such a policy could defeat a heavy speculative movement. Until such a policy is actually tried under such circumstances, the usefulness of separate exchange markets for capital and current transactions must remain an open question.


I. Algebraic Formulation of Exchange Rate and Interest Rate Relationships

If both capital gains and the effective rate of return are taken into account, the expected percentage gain (over a given period t) of a nonresident investing in the home country can be expressed as


where Ith is the relevant domestic interest rate over time period t; ERt0 is the average expected spot official exchange rate applicable to interest payments during period t, weighted by the size of interest payments due; ERtf is the expected spot financial exchange rate at the end of the period t; and SRf is the spot financial exchange rate at present. The exchange rates are defined as “dollars” (foreign currency) per unit of “francs” (domestic currency). For simplicity, the following expressions are defined:




(Here ft is simply the expected percentage rate of change of the financial exchange rate; et is a more complicated concept, because ERt0 is a weighted average of different expected official exchange rates over the period.) If interest arbitrage operates perfectly, equilibrium in the financial exchange market is achieved when


where Itw is the interest rate in the rest of the world (on some basis comparable to Ith) and p is the premium of the financial franc over the official franc—that is,


Since (Ithet) is of relatively negligible magnitude, equation (1) can be reduced to


If there is a forward market for financial francs, it is possible to derive the following version of the well-known interest equalization formula:


where FPtf is the forward premium on the financial franc. However, it should be mentioned that the validity of equation (3) depends on such assumptions as a frictionless and riskless international capital market, the absence of transaction costs, and an unlimited volume of funds available for interest arbitrage.

Finally, it can be noted that the equilibrium conditions—equations (2) and (3)—are reduced in the absence of expected exchange rate changes to the following:


The right-hand expression of this simplified condition is the basis of the numerical example given in Section I.

II. Effects of Evasion and Arbitrage Under a Neutral Intervention Policy

A neutral intervention policy may not entirely offset the effect on the financial exchange rate of shifts of transactions between the two markets when such shifts take place as a result of outright evasion of regulations. For example, if the domestic currency (franc) is at a premium on the financial exchange market, nonresidents will be induced to seek ways of bringing money into the home country at the official exchange rate, while residents will be induced to make import payments at the financial rate. These inducements are all the greater if the official rate is expected to be revised upward. To accomplish these aims, nonresidents might enter into arrangements with home-country importers whereby the latter underinvoice imports, the difference being made up by a separate transaction through the financial exchange market. Such a transaction could take either of two forms: (1) the acquisition of foreign exchange by the home country importer, which is then used to pay the foreign exporter, or (2) the acquisition of financial franc assets by the foreign exporter. Let us consider each of these cases in turn.

(1) The home-country importer could “repay” a “loan received abroad for foreign expenses”; this could only be done in a country without strict registration of all foreign loans.38 Alternatively, he could purchase a foreign asset, resell it abroad, and use the proceeds to pay the foreign exporter. In either instance the result is to lower the effective price of imports and at the same time increase the private demand for foreign exchange on the financial exchange market. The home country’s demand for imports will also rise, and, under a neutral intervention policy, increase the monetary authorities’ demand for foreign exchange in the financial exchange market. Whether the latter occurs depends on the price elasticity of demand for imports and the proportion of import payments being channeled through the financial exchange market. There will in any case be a tendency for the premium on the financial franc to be reduced.

(2) When the quid pro quo to the underinvoiced import payment takes the form of the foreign exporter acquiring financial franc balances or franc assets, the result is less determinate. Suppose, for instance, that the home country importer sells the foreign exporter a franc bond for less than the market price, or, alternatively, “loans” him financial franc balances.39 If the foreign exporter repatriates the entirety of these new franc acquisitions, the result is the same as in case (1). However, the availability of these cheap franc assets may increase net foreign demand for franc assets. If so, the increased demand for foreign exchange on the financial market caused by the rise in the home demand for imports will to some extent be offset by an increase in foreign demand for francs. The extent of this offset will depend on the relative elasticities of demand involved and on the implicit effective exchange rate which the importer and exporter have determined between themselves.

A similar line of reasoning applies when there is an initial discount on the financial franc. In this case, home countrymen will engage in the underinvoicing of exports and with the difference will acquire foreign assets either for holding abroad or bringing back through the financial exchange market. (Underinvoicing of exports is a well-known practice, often engaged in to evade restrictions on capital outflows; where authorities have tried to counter this practice by more stringent enforcement of exchange surrender requirements, including the specification of minimum prices in export licenses, this has sometimes led to keeping export prices secret.) In this case, there will be a net increase in exports and, possibly, a net increase in home demand for foreign assets. Again, the net effect on the financial exchange rate depends on the relevant elasticities.

Another type of evasion is illegal arbitrage between the two markets. For example, suppose there is a premium on the financial franc; a multinational corporation might then send funds out of the country through the financial market and bring them back through the official market under the guise of additional royalty payments to the home country branch. Under a neutral intervention policy, such operations have no effect on the financial franc (domestic currency) when the aim of the operation is to make a franc profit, since the amount of foreign exchange bought on the financial market is equal to that sold on the official market and thus resold by the authorities on the financial market. The only net effect on the financial exchange rate is in the very short run. On the other hand, if the arbitrage is conducted by foreigners aiming at a dollar (foreign currency) profit, fewer dollars will be sold on the official market than will be bought on the financial market, with the effect that there will tend to be a decline in the premium on the financial franc.

III. Institutional Features of Recent European Dual Exchange Market Systems

This Appendix gives a brief description of the dual exchange market systems operating in recent years in Belgium-Luxembourg, France, and Italy, and of the additional measures to control capital movements which the authorities of these countries took during the periods in which dual exchange markets were in effect. A summary of the structures of the three dual exchange systems, showing also any important changes in these structures, is given in Table 2.

Table 2.

Features of Dual Exchange Markets in Belgium-Luxembourg, France, and Italy

article image
Key: O: Official market.F: Financial market.Option: Transaction can be settled in either market.O, F: Certain items are settled in one market and the remainder in the other.

For France, the dual exchange market regulations shown by this table did not apply to Operations Account countries. For Belgium, they have not applied to “bilateral countries” (Burundi, Rwanda, and Zaïre) with which special payments arrangements have been in effect.

Previous to the establishment of the financial exchange market, there had been a more limited secondary exchange market, the security currency (devises-titres) market. This market, which was formally merged with the financial franc market on October 20, 1971, was used for settlement of transactions related to residents’ purchases and sales of French or foreign securities abroad. For a brief period in December 1971, a separate type of nonresident account, the “Patrimonial Assets Account,” was established to accommodate nonresident purchases and sales of French securities and real estate.

Important features of the system were in effect before April 1, 1959, on which date most of Belgium’s trading partners became included in Convertible Account arrangements. The presentation here ignores minor changes between 1959 and 1971.

Includes trade-related payments such as freight, insurance and commercial expenses, and transit trade (but see footnote 5). Forward exchange transactions directly related to trade generally had to take place in the official market and were in some cases restricted.

During this period, transit trade transactions had to take place in the financial exchange market.

Settlements by private travel agencies took place in the official market; all others took place in the financial market.

Amortization of loans in lire by nonresidents was channeled through the financial market.

These refer to private sector business expenses. Public sector administration expenses took place in the official market in all cases.

Includes workers’ remittances, remittances by emigrants and immigrants, family maintenance payments, and other personal transfers (except for gifts and inheritances).

From January 22, 1973 through January 31, 1973, banknote transactions took place in the official market.

These items, depending on their nature, were assigned to the same markets as private sector transactions.

Includes transportation, commissions and brokerage fees, professional fees, payments relating to intellectual property, rentals, payments for technical services, and transactions on foreign commodity markets. In the case of France, this category also included surrender of export proceeds in advance of the contracted due date and mixed import-export transactions (with more than 25 per cent in unspecified services).

Repatriation of certain foreign long-term investments was channeled through the official market (optional before May 11, 1971).


The Belgian-Luxembourg Economic Union (BLEU) has had a system of dual exchange markets since 1951; since the end of 1957, the system has, broadly speaking, been maintained in its present form and coverage. The experience with this system through 1972 has been described by De Looper and Mountford (1973). The structure of the two exchange markets is shown in this Appendix, Table 2. The setting in which the BLEU dual market has operated differs sharply from that in France (see below), where most types of capital movements have been tightly controlled; by contrast, the BLEU has since the mid-1950s allowed both current and capital outward payments to be substantially free from restrictions.

On May 11, 1971 certain important changes were made in the system—in particular, the almost complete separation of the official market and the “free” (financial) market. Previously, these markets had been in effect linked to some extent, owing to regulations which permitted all payments to be made on the free market and all receipts to be channeled through the official market, while outward payments eligible for the official market continued to be strictly controlled. As a result of these regulations, market forces had prevented the free market franc from rising to more than a small premium over the official franc but not from falling to a substantial discount. After May 11, only a small number of current transactions could be effected through either exchange market; in addition, individual licenses could be granted to allow certain capital transactions through the official market. However, domestic and foreign banknotes could be bought and sold on the free exchange market. On January 28, 1974, outward payments with respect to investment earnings were again permitted to be channeled through either market, thereby re-establishing a major link between the markets.

Administration of the dual exchange system is through a dual system of nonresident domestic currency accounts (Convertible Accounts, balances on which are equivalent to currencies negotiated on the official market, and Financial Accounts, which are related to free market transactions), supplemented by an equally supervised dual system of resident foreign currency accounts, related to the official market and the financial market, respectively. Residents receiving foreign currency proceeds from current account transactions which are to be used later for current payments may deposit such proceeds in Controlled Resident Accounts in Foreign Currency. In practice, transactions of amounts less than specified ceilings do not get reported to the exchange control authority, the Belgo-Luxembourg Exchange Institute (Institut Belgo-Luxembourgeois du Change or IBLC). Thus, correct classification of smaller transactions is the sole responsibility of the commercial banks.

During the various exchange crises of the past several years, the major loophole through which short-term capital could enter at the official market rate during periods of expected appreciation of the Belgian franc was the Convertible Accounts: notably, nonresidents were free to create balances in such accounts by the sale of eligible currencies in the official market. In order to prevent this, the BLEU authorities on several occasions took steps to penalize or block such inflows. From October 1969 to September 1971, banks in Belgium and Luxembourg were requested to maintain a ceiling on their net external spot debtor positions; between June and September, this ceiling was reinforced by banks’ being advised to deposit any excess over this ceiling in special noninterest-bearing blocked accounts in the National Bank of Beligum. The ceiling was re-established in March 1972 and has remained in force, with modification, since that time. In July 1972 a “gentleman’s agreement” between the National Bank of Belgium and the Belgian banks re-established a reserve requirement against the banks’ Convertible Account liabilities; a similar system was established in Luxembourg in January 1973 and was continued in both countries until February 1974. More direct measures to limit inflows into Convertible Accounts were taken during a two-week period in March 1973; essentially, the IBLC froze any new inflows into Convertible Accounts except for the purpose of making permitted payments to the debit of Convertible Accounts. Moreover, in March 1973 banks in Belgium and Luxembourg agreed with their respective monetary authorities to charge a negative interest rate in the form of a “special commission” of 0.25 per cent per week on holdings of Belgian or Luxembourg francs in Convertible Accounts in excess of the daily average of holdings in the account concerned during the last quarter of 1972. This measure was in effect from April 1, 1973 until it was suspended, effective January 1, 1974.

Although leads and lags do not seem to have played as important a role for Belgium during the past several years as has been the case for some other European countries, the BLEU authorities have required exchange control permission for import payments taking place three months before or after customs clearance and for export proceeds collected more than six months after the date of exportation. There are also regulations pertaining to the permissible length of time between receipt of export proceeds in convertible currencies and their sale to an authorized bank or their deposit in a special resident foreign currency account for making current payments at a later date.


A system of dual exchange markets was in operation in France between August 23, 1971 and March 21, 1974. Prior to the establishment of a separate “financial franc market,” there had been in effect (since August 11, 1969) a more limited “security currency” (devises-titres) market; through this market were channeled purchases and sales of foreign exchange relating to transactions by residents with respect to certain French and foreign securities held abroad. The security currency market was merged with the financial franc market on October 20, 1971; thus, between August 23 and October 20, there was effectively a system of triple exchange markets. It should be noted (by way of contrast with the BLEU’s dual exchange market system) that the introducion of the financial market in France was not accompanied by the lifting of existing restrictions on outward transfers.

Separation of the two exchange markets was enforced by exchange control and by establishment of two types of nonresident franc accounts—Foreign Accounts and Financial Accounts, for operations relating to the official market and the financial franc market, respectively. Transfers between the two types of accounts could be effected only by conversion through spot exchange transactions in the two exchange markets.

The dual exchange market regulations were not applicable to Operations Account countries.40 Each of these countries had a dual exchange market system parallel with the French one, applying to all countries other than France and its Overseas Departments and Territories41 and other Operations Account countries.42 The official and financial exchange rates of the Operations Account countries were based on (1) the fixed rates between their currencies and the French franc, and (2) the foreign currency rates in the Paris exchange market. Thus, the same percentage spread between the official and financial rates vis-à-vis the rest of the world was maintained by all French Franc Area countries.

At the outset, a number of important categories of current account transactions were assigned to the financial franc market; in fact, the only transactions to be settled through the official market were merchandise trade (including incidental expenses but excluding transit trade) and public sector current transactions (see Table 2). While some current transaction items were shifted to the official market in May 1972 (see Table 2), important categories such as travel/tourism, investment income, workers’ remittances, and banknote transactions remained in the financial franc market until the latter’s abolition. This was done in order to put into the financial market those current account items that were most difficult to control and to balance the deficit in those items against the net capital inflow expected.43 There were no options between the two markets, but in rare cases the determination as to market eligibility was made by administrative decision.

Despite the assignment to the financial franc market of many categories of current account items that were potential vehicles for exchange speculation, heavy speculation in favor of the franc through the official market took place on several occasions, and the French authorities took various measures to restrict inflows into Foreign Accounts.44 Prior to establishing the financial franc market, the authorities had raised reserve requirements on nonresident accounts (May 2 and August 5, 1971), instructed banks to prevent any further deterioration in their overall net position in foreign currencies and francs in respect of foreign currencies (effective August 3, 1971), and had obtained the agreement of commercial banks (later backed by a mandatory provision) to pay no interest on nonresidents’ term deposits in francs where these were for periods of less than 90 days. On December 3, 1971 the Ministry of Economy and Finance in effect threatened the inconvertibility of all nonresident-held French francs by announcing that from December 10, nonresidents holding francs on Foreign Accounts or Financial Accounts could only use the balances for payments in francs to residents and no longer for conversion into foreign currency. This action, which amounted to a threat of a partial freeze on nonresident accounts, was only partially alleviated by continuing permission (December 8) to shift balances from one type of account to the other through spot transactions in the two exchange markets.

Measures to prevent the inflow of foreign funds into Foreign Accounts were taken again in March 1973. Banks and other financial institutions were prevented from remunerating directly or indirectly any nonresident franc deposits which were either sight deposits or time deposits with a maturity of up to 180 days;45 a 100 per cent reserve requirement was applied against any increases after January 4, 1973 in nonresidents’ demand and time deposits in francs;45 and the net long forward position in foreign currencies of resident banks vis-à-vis their foreign correspondents was limited to that of February 28.

In addition to measures limiting the inflow of funds into the Foreign Accounts, the French authorities took various steps during this period to limit possible variations in the terms of payment. In addition, they gradually liberalized regulations with respect to residents’ purchases of foreign securities and real estate, as well as travel allowances and personal transfers, and on one occasion (March 1973) restricted nonresident purchases of French securities. These measures were intended to reduce pressures on the financial exchange market.

However, despite the occasional severe pressures on the official market, which prompted the steps described above, the authorities pursued a declared policy of nonintervention in the financial market.46 Throughout most of the life of the dual market, speculative pressures were in favor of the tranc, as indicated by a premium on the financial franc; the situation was reversed in the latter half of 1974, when a discount on the financial franc developed (see Table 1 of the paper).47 On January 19, 1975, the authorities announced that their intervention in the official market would no longer be aimed at maintaining specified margins between the French franc. The departure of the franc from the “snake” was followed by a reduction in the discount on the financial franc. On March 21, 1974 the financial franc market was abolished.


In Italy, a dual exchange market was introduced on January 22, 1973; it continued to operate until March 22, 1974. At the time it was introduced, the official exchange rate was confined within announced limits, but from February 13, 1974 on, the lira was floating independently. In general, current account transactions were channeled through the official exchange market, which was named the ordinary exchange market, and capital account transactions through the financial exchange market (see this Appendix, Table 2). The system was administered by the banks by means of two separate types of nonresident lire accounts (Foreign Accounts and Capital Accounts) and of resident foreign currency accounts (Ordinary Foreign Currency Accounts and Financial Foreign Currency Accounts).48 However, certain major current account items were settled in the financial exchange market. By far the most important of such items were those settled by foreign banknotes whose purchase and sale took place on the financial exchange market as of February 1, 1973.49 A substantial part of tourist receipts and workers’ remittances were thereby assigned to the financial exchange market.50 In practice, some of these transactions probably continued to take place on the extra-legal “parallel” market (see below). As in the case of France, it is not known how the foreign currency proceeds of public sector external borrowing were allocated between the two exchange markets.

As in BLEU and France, the Italian authorities imposed certain controls on the terms of import and export payments, especially with respect to advance payments for imports and the surrender of export receipts, 90 per cent of which had to be collected within 90 days of shipment, unless special approval was obtained. After export proceeds were collected, they could be held in domestic banks for up to one month before being converted to lire.

The purpose of introducing a dual exchange market was to dampen the outflow of capital, which had accelerated in the latter part of 1972 and early 1973. At the end of July 1973, the authorities took the further step of requiring residents exporting capital to deposit, in a noninterest-bearing account, a sum equivalent to 50 per cent of the amount exported.51 The authorities believe that this measure was to some extent evaded by expansion of the “parallel” market. The Bank of Italy estimates that the outflow of capital associated with banknote transactions on the parallel market was between Lit 370 billion and Lit 680 billion. Aside from this channel of evasion, it was also felt by the authorities that effective functioning of the system was hindered by cross-operations between the financial and the official exchange markets and by fraudulent imports of gold (allegedly bought for industrial purposes). For this reason, and because a steep rise in Italian interest rates had reduced the arbitrage incentive for capital outflow, it was decided in March 1974 to abolish the financial market, “which for some while had lost all practical significance.”52


  • Banca d’Italia, Report for the Year 1973 (Rome, abridged version, 1974).

  • Barattieri, Vittorio, and Giorgio Ragazzi, An Analysis of the Two-Tier Foreign Exchange Market,Banca Nazionale del Lavoro, Quarterly Review, No. 99 (December 1971), pp. 35472.

    • Search Google Scholar
    • Export Citation
  • De Looper, J. H. C., and Alexander Mountford, Belgo-Luxembourg Economic Union Evolved Dual Exchange Market During Two Decades,IMF Survey, January 8, 1973, pp. 24.

    • Search Google Scholar
    • Export Citation
  • Decaluwe, B., Two-Tier Exchange Markets and Other Systems: A Comparison,Tijdschrift voor Economie, Vol. 19, No. 1 (1974), pp. 5579.

    • Search Google Scholar
    • Export Citation
  • Deutsche Bundesbank, Report of the Deutsche Bundesbank for the Year 1971 (Frankfurt am Main, 1972).

  • Dickie, Paul M., and David B. Noursi, Dual Markets: The Case of the Syrian Arab Republic,Staff Papers, Vol. 22 (July 1975), pp. 45668.

    • Search Google Scholar
    • Export Citation
  • Fleming, J. Marcus, “Dual Exchange Rates for Current and Capital Transactions: A Theoretical Examination,” in his Essays in International Economics (Harvard University Press, 1971), pp. 296325.

    • Search Google Scholar
    • Export Citation
  • Fleming, J. Marcus, Dual Exchange Markets and Other Remedies for Disruptive Capital Flows,Staff Papers, Vol. 21 (March 1974), pp. 127.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, Annual Report on Exchange Restrictions (Washington, various years).

  • Meade, J. E., The Balance of Payments (London, 1951).

  • Salin, Pascal, Un Double Marché des Changes, Est-il Justifié?Revue d’Economie Politique, Vol. 81, No. 6 (November-December 1971), pp. 95974.

    • Search Google Scholar
    • Export Citation
  • Schmitt, Hans O., International Monetary System: Three Options for Reform,International Affairs, Vol. 50 (April 1974), pp. 193210.

    • Search Google Scholar
    • Export Citation
  • Triffin, Robert, “The Dollar Currency Rate in Switzerland” (unpublished, International Monetary Fund, November 17, 1947).

  • Wahl, Jacques Henri, Le Dispositif Français de Lutte Contre les Afflux de Capitaux,Aussenwirtschaft, Nos. 1 and 2 (March/June 1973), pp. 6173.

    • Search Google Scholar
    • Export Citation
  • Williamson, John, Payments Objectives and Economic Welfare,Staff Papers, Vol. 20 (November 1973), pp. 57390.


Mr. Lanyi, Assistant to the Director, Exchange and Trade Relations Department, is a graduate of Harvard University and received his doctorate from the University of California at Berkeley. He formerly was on the faculty of Princeton University.


See especially, Fleming (1971 and 1974). Other papers on this subject are Barattieri and Ragazzi (1971), Salin (1971), and Decaluwe (1974). The first study on this type of exchange market may well have been Triffin’s (1947); other Fund studies of dual exchange markets in specific countries include those by the European Department (1970) and by Dickie and Noursi (1975).


In several European countries there have been separate exchange markets (“closed circuit markets”) limited to certain types of capital transactions, such as the investment currency market in the United Kingdom, the O-Guilder market in the Netherlands, and, until 1971, the security currency market in France. The discussion in this paper concentrates on the broader type of separate exchange market employed in the BLEU, France, and Italy. “Free” exchange markets for capital transactions have also been employed in a number of other countries, notably in the Middle East and Latin America, but the dual exchange markets employed in these regions have usually included various current account transactions, since they have been used for purposes other than moderating capital flows—for instance, for gradual devaluation or for allowing invisibles and capital to be uncontrolled where it was felt that effective controls could be applied only to trade transactions.


Discussion of the methods employed to separate the two markets, and the inherent limitations involved, is taken up in the following section.


This also increases the effective rate of return on investment abroad by residents.


An exposition of the algebraic relationships among the official exchange rate, financial exchange rate, and interest rates is given in Appendix I.


A neutral intervention policy, as defined here, also has the effect of increasing or decreasing the monetary authority’s liabilities to the public, depending on whether foreign exchange is being bought or sold in the financial exchange market at a premium or a discount. On this point see Barattieri and Ragazzi, pp. 361–62. Fleming (1974, p. 4) has defined—although not advocated—an alternate type of neutral intervention policy, which would equalize the domestic currency bought (sold) in exchange for foreign exchange in the financial market with that sold (bought) in the official market. While this type of intervention policy would have no domestic monetary effects, it would not ensure balance of payments equilibrium.


This is not necessarily the case if the official exchange rate is also floating.


Both the French and the Belgian authorities have on occasion imposed ceilings on nonresident franc holdings.


Trade financing means any lagged payment terms in international trade, whether directly between buyer and seller or through commercial banks.


For example, under the French system (when it was in existence) the U. K. tourist in France bought a French good at a different cost in terms of sterling than did the U. K. importer.


In the opposite case of a relatively appreciated financial exchange rate, nonresidents have an incentive to underinvoice their exports, which appear as underinvoiced imports from the viewpoint of the home country.


This is an admittedly brief discussion of a complex topic which, it can be hoped, will someday receive fuller and more systematic study.


These costs include the official penalties (adjusted for the risk of being caught) and administrative costs for the evader.


There are conceivable circumstances in which unusually heavy pressures on official reserves under a neutral intervention policy would result, at least for a time, in a diminishing spread between two exchange rates—that is, when there is initially a surplus (deficit) on current account, a financial exchange rate more depreciated (appreciated) than the official rate, and an increase in speculative flows through the official market in favor of (against) the domestic currency. However, it is more likely that the state of market expectations is already reflected in the relative depreciatedness or appreciatedness of the financial exchange rate.


As long as the market’s elasticity of expectations with respect to the financial exchange rate is less than unity, an equilibrium rate must finally be reached at which no speculation on a further change will occur. However, if the new financial exchange rate is at a wide spread vis-à-vis the official rate, speculative pressure on the official market may continue to be intense.


The 1971 Report of the Deutsche Bundesbank (1972), in arguing that a two-tier foreign exchange market would not be suitable for the Federal Republic of Germany, makes this point as follows: “… the German current account is more or less balanced only because the very large trade surplus is matched by deficits on foreign travel and unilateral transfers of approximately equal size. It would hardly be possible, however, adequately to subject foreign travel and remittances by foreign workers to foreign exchange controls; and if these items were assigned to a ‘free’ foreign exchange market, both the controlled market for commercial transactions and the free market for ‘financial’ transactions would be highly distorted from the outset.”


Unless the objective of official intervention in the exchange markets was to sustain such a spread.


For each of these countries there were a few days during this period when the spread rose to 6–9 per cent.


The larger spreads in the Italian case may have reflected the fact that the Italian financial market (unlike its French and Belgian counterparts) was based on a significantly unbalanced capital account. The only major balancing element (apart from the possible conversion of foreign loans) was the sale of foreign banknotes, resulting, for instance, from workers’ earnings abroad. It might be hypothesized that the spread in the Italian financial market reflected the net opportunity cost of buying lire in the well-established markets for banknotes in other European countries; however, the data for testing such a hypothesis do not exist.


To the extent that the latter took the form of leads and lags in commercial payments, or changes in nonresident balances for current account payments, this was an inadequacy shared by other methods of controlling capital movements. A description of the dual exchange market systems in the BLEU, France, and Italy during 1971–74, as well as of the chief capital control measures taken during the exchange crises over this period, is given in Appendix III.


Controls on leads and lags—such as fixing the terms of payment by law—interfere with the freedom of exporters and importers to determine mutually convenient terms; they are also cumbersome and costly to enforce. However, the harmful effects of severe controls on leads and lags are mitigated if there is freedom for capital to move through the financial market. Controls on commercial balances involve inequities such as how one fairly determines the ceiling on balances without, for example, penalizing newcomers—and added transaction costs for trade payments. It may still be thought that added costs to trade are a price worth paying to control capital movements.


For the purposes of Williamson’s analysis, reversible short-term capital movements are included in the concept of reserve changes. Williamson’s concept of “underlying capital movements” is similar to Meade’s “autonomous capital movements.” See Meade (1951), chapter 1.


Given the change in reserves, only one of these can be independently determined.


Schmitt has argued that in some cases the current account is the focus of this structural objective (Schmitt, 1974, especially pp. 200–205).


Aspects of this issue are discussed in Fleming (1971, pp. 311–12, and 1974, pp. 20–25).


The period of time to which “persistent loss” refers is subject to differences of view. Some would choose the entire length of a business cycle; others would take a shorter period. The optimum frequency of exchange rate adjustment may differ among countries, which makes it impossible to give a single generally applicable criterion.


This is true by the assumption that the net inflow of capital responds positively to a depreciation of the financial exchange rate.


These conclusions would not necessarily be correct if one employed a definition of external payments imbalance according to which short-term capital movements are included “below the line.”


In the case of a current account surplus combined with a depreciated financial exchange rate or in that of a current account deficit combined with an appreciated financial exchange rate, there may be either a deficit, surplus, or zero underlying balance (if the exchange markets were unified), depending on the elasticity of capital flows with respect to changes in the exchange rate and on the size of the spread between the two rates. However, a zero or positive current account balance, combined with an appreciated financial exchange rate, unambiguously indicates an underlying surplus, while a zero or negative current account balance combined with a depreciated financial exchange rate shows that there is an underlying deficit.


In fact, under a hybrid system the official rate may be even more variable than under unified (unmanaged) floating if (a) the trade account responds perversely (in “J-curve” fashion) to changes in the rate and (b) speculation in general tends to be stabilizing with regard to medium-term trends in the rate.


To cite just one difficulty in this regard, calculations of an equilibrium set of rates would have to be based on elasticities of capital movements as well as on those of trade flows.


This only underlines the fact that it is very hard to understand what is the concrete meaning of a set of controls existing “in the absence of evasion”—for example, what kind of behavior on the part of private individuals does it assume?


Fleming (1971, pp. 297–99) gives this as an example of the general proposition that thorough control over one class of transactions, ensuring that all and only transactions in that class pass through the corresponding exchange market, implies control over the other class of transactions as well. However, as argued here, the only possible application of Fleming’s proposition is the one he gives.


To be sure, the cost of installing a system of QRs where none had previously existed might not be less than that of introducing a dual exchange market.


See Fleming (1974, pp. 19–20). A system of this type was in effect in Italy from February 13, 1973 to March 22, 1974; to a lesser degree, it can be said to have been in effect in France from March 19, 1973 to March 21, 1974, although for most of this period France was maintaining margins against other “snake” currencies.


There is another advantage to a unified floating exchange rate over dual exchange markets, which is that it eliminates the distortions arising from different exchange rates applying to different types of capital flow (trade financing in the official market and other capital in the financial market), as well as the resource cost due to the efforts necessary to evade controls and to the administration of controls against evasion. See Fleming (1974), p. 18.


However, this would not be the case if the effect of official intervention is itself disequilibrating (see next paragraph).


Since those advocating a dual exchange market system stress its superiority to a system of quantitative restrictions, one can assume for the sake of this discussion that no such restrictions exist in the case being described here.


Here again, the lack of restrictions (or at least looseness of controls) would make possible this type of evasion. Controls preventing such evasion would need to be quite thorough and complex, with considerable checking into details of individual transactions.


Countries whose bank of issue was linked with the French Treasury by an Operations Account on August 23, 1971 were: Cameroon, the Central African Republic, Chad, the People’s Republic of the Congo, Dahomey, Gabon, Ivory Coast, the Malagasy Republic, Mali, Mauritania, Niger, Senegal, Togo, and Upper Volta. In 1973, the Malagasy Republic (July 1) and Mauritania (July 9) ceased to maintain an Operations Account with the French Treasury and were henceforth considered foreign countries for exchange control purposes.


With the exception of the Territory of the Afars and the Issas.


All current and capital payments within the French Franc Area continued in general to be unrestricted and uncontrolled and were effected at the existing fixed exchange rates.


A senior French Treasury official (Wahl, 1973, p. 66) has explained this as follows: “… la répartition des transactions courantes entre les deux marchés des changes n’était pas commandée seulement par des considérations rationnelles, mais par des raisons matérielles de contrôle et aussi par une considération supplémentaire, la volonté de laisser l’essentiel du déficit de la balance des transactions unilatérales et des services susciter une demande de devises sur le marché du franc financier.”


As in the BLEU, the dual market regulations did not prevent banks abroad from creating official market balances in domestic currency by converting foreign currency in the official market.


Exempt from this measure were: (1) persons physically and juridically resident in Operations Account countries; and (2) international organizations, central banks, and foreign public financial institutions (for their balances on September 30, 1972).


However, it is not known how the proceeds of public sector external borrowing were allocated between the two markets.


This may have reflected the pressures to which the official market franc was subjected as a result of France’s participation in the “snake” arrangement.


Both types of foreign currency accounts remained subject to the rule previously in effect that funds entered on such accounts had to be sold by the end of the calendar month following that in which the credit entry concerned had taken place.


However, exporters were permitted to receive payment in foreign banknotes that were accepted in the country of issue for credit to bank accounts, balances in which were convertible at official market rates.


Other items included amortization of loans, depreciation of direct investments, accrued interest on bonds, engagement bonuses (received by artists, technicians, and sportsmen) and interest on balances in Capital Accounts.


Certain exemptions were granted to the deposit requirement, chiefly relating to direct investments and to bond issues by international organizations and bodies.


The material in this paragraph is based entirely on Banca d’Italia (1974, especially pp. 64–67).