It has become a commonplace observation that one of the principal factors undermining the international monetary order that was set up at Bretton Woods has been the ever-growing magnitude of payments disequilibria caused by temporary and often reversible capital flows. Doubtless the regime of fixed though adjustable par values which was the centerpiece of the Bretton Woods system has been subjected to considerable stress by other factors also, notably the increasing importance of the cost-push element in inflation, with its differential impact on national competitive positions; but it is the extreme volatility of international capital flows, in response to interest differences or speculative incentives, that has compounded these difficulties to such a point that the feasibility of restoring the par value system itself can be called in question.


It has become a commonplace observation that one of the principal factors undermining the international monetary order that was set up at Bretton Woods has been the ever-growing magnitude of payments disequilibria caused by temporary and often reversible capital flows. Doubtless the regime of fixed though adjustable par values which was the centerpiece of the Bretton Woods system has been subjected to considerable stress by other factors also, notably the increasing importance of the cost-push element in inflation, with its differential impact on national competitive positions; but it is the extreme volatility of international capital flows, in response to interest differences or speculative incentives, that has compounded these difficulties to such a point that the feasibility of restoring the par value system itself can be called in question.

I. Background of the Study

It has become a commonplace observation that one of the principal factors undermining the international monetary order that was set up at Bretton Woods has been the ever-growing magnitude of payments disequilibria caused by temporary and often reversible capital flows. Doubtless the regime of fixed though adjustable par values which was the centerpiece of the Bretton Woods system has been subjected to considerable stress by other factors also, notably the increasing importance of the cost-push element in inflation, with its differential impact on national competitive positions; but it is the extreme volatility of international capital flows, in response to interest differences or speculative incentives, that has compounded these difficulties to such a point that the feasibility of restoring the par value system itself can be called in question.

These disruptive capital flows have the effect of reducing not only the stability of exchange rates but also the independence of monetary policies and the internal financial stability of countries. Indeed, in the case of the countries receiving such flows, the desire to avoid losing control of the domesti

article image
money situation has been the primary motive for allowing exchange rates to rise.

Countries that have been exposed to disruptive capital flows, and that do not wish their exchange rates to be determined entirely by market forces, have generally sought to cope with such flows, at least in part, either by controlling or by financing them. “Financing,” which includes not only official borrowing and lending abroad but also the rise and fall of reserves, is generally accompanied by policies to offset the disturbing effects of the flows on domestic monetary conditions and on interest rates. “Controlling” is used in a very wide sense to cover not only restrictions on individual capital transactions and quantitative regulations governing the external assets or liabilities of banks and other residents but also more indirect methods of influencing capital flows, such as interest rate policies, fiscal measures, official intervention on the forward exchange market, and various arrangements affecting spot exchange rates.

This paper is particularly concerned with a technique for influencing capital flows that has attracted increasing attention in recent years both in official circles and among economists. This method involves the setting up of separate exchange markets, with separate exchange rates for current and capital transactions, respectively. A dual exchange market of a fairly full-fledged character has been established for many years in Belgium, although it is only since 1971 that the Belgian financial franc has been free to rise to a premium as well as fall to a discount. Similar—though not identical—dual markets were set up in France in 1971 and in Italy in 1973. Partial exchange markets applicable to particular types of capital transactions have existed for many years in the United Kingdom, France, and the Netherlands, and segregated exchange markets covering various types of both capital and current transactions have been common in several developing countries, particularly in Latin America.

The aim of the present paper is to assess the relative merits of dual markets as a means of dealing with disequilibrating capital flows (or more precisely with the temporary and reversible payments imbalances that are frequently associated with such flows), as compared with each of the other main techniques available for this purpose. Through such a serial comparison it is hoped to achieve an overall evaluation of the dual market device and, by using it as a standard of comparison, to bring out the special characteristics of the other approaches.

For these evaluations, though taking into account the lessons of experience, this paper considers the various techniques as they might be applied rather than as they have been applied. This is particularly necessary in the case of dual markets, since this technique is a relatively recent one and its potentialities have not yet been fully explored.

In general the paper deals with arrangements that provide for only two exchange markets—one comprising (as far as possible) all capital transactions and the other (with the same qualification) all current transactions. And for the most part it is assumed that the exchange rate for current transactions is pegged by official intervention in the vicinity of parity, as has indeed generally been true of the Belgian and French (though not of the Italian) dual market.

II. Main Features of the Dual Market

The operation of a dual exchange market has been described in some detail in a previous article.1 It calls for the segregation of exchange transactions into two categories, connected respectively with current and with capital payments and receipts. As explained in that article,2 such segregation necessitates a close control over at least one of the categories of underlying transactions (usually the current items), as well as over actual payments and receipts, and it involves the establishment of two kinds of nonresident balances in domestic currency and, at least on the part of the foreign exchange banks, two kinds of resident balances in foreign currency.

It is sometimes thought to be of the essence of the dual exchange market that the rate for capital transactions is allowed to float freely without official intervention. This is a misunderstanding of the possibilities of the system. There is no reason why the authorities should not buy or sell foreign currency for domestic currency on the capital exchange market. Indeed, if they wish that market to make its maximum contribution to the equilibrium of the balance of payments as a whole, they must so intervene, selling in the capital transactions market the foreign exchange they are acquiring in the current transactions market and buying in the former the foreign exchange they are selling in the latter.3 This will give them a profit (or a loss) according as they bring the “capital” rate closer to (or pry it apart from) the “current” rate.

There are two mechanisms whereby movements in the “capital” exchange rate affect capital flows, both inward and outward: 4 (1) A change in the actual rate associated with an equal change in expected future rates results in a proportionate change in the yield of domestic capital invested abroad or foreign capital invested in the country, provided that interest in both directions is transferred as a current item through the official market; and (2) A change in the actual rate unaccompanied by a corresponding change in the expected future rate—for example, a change in the actual rate that is expected to be reversed—in addition creates the prospect of a corresponding capital gain on any temporary investment in one direction and of a capital loss on any investment in the other direction.

Of the two mechanisms, the second offers a much more powerful incentive for short-term placements. For example, a 10 per cent rise in a country’s “capital” rate that was expected to be permanent might reduce the yield on foreign investment in that country only from 5 per cent to about 4½ per cent, and increase the yield of investment abroad from 5 per cent to 5½ per cent. But a 10 per cent rise in the “capital” rate that was expected to be reversed in six months would give, in addition, a 10 per cent capital gain over that period. Owing to this speculative element, a temporary change in the relationship of the “capital” rate to the “current” rate might be strong enough to counteract expectations of a change in the latter rate, if the “capital” rate was expected to revert to a normal relationship to the “current” rate.

When the “capital” exchange rate is at a discount from the “current” rate, the operation of the first mechanism described above is likely to be weakened by the difficulty of ensuring that interest and dividend receipts from the foreign investments of residents are in fact transferred through the market for current transactions. It may be partly for this reason that, under the Belgian system, such receipts can be transferred through whichever channel yields the most profit. By contrast, the purchase of foreign investments by British residents under the investment currency scheme is controlled, and earnings have to be repatriated through the regular exchange market.

The first mechanism, however, is more reliable than the second, which depends on the existence of a negative elasticity of expectations. This, in turn, is probably dependent on a variety of factors—such as (1) an expectation that a change in the “current” rate or other adjustment measures will remove the speculative motive for capital flows and allow a narrowing of the spread between the two rates, (2) an expectation that evasion of controls will occur whereby current transactions pass for capital ones (or vice versa), or (3) an expectation of a shift in official intervention on the exchange market for capital transactions designed to draw the rates closer together. The last two expectations assume that the authorities will tolerate a widening of the overall imbalance in their external transactions.

To the extent that the second mechanism is operative, movements in the “capital” rate, as has been seen, operate with special force on short-term capital flows—which may be what the situation requires, if the expectations of temporariness on which this mechanism depends are warranted.

In any realistic appraisal of the dual exchange market, it must be noted that it is not administratively practicable to achieve a full segregation of capital and current transactions. As long as some flexibility is allowed in the timing of current payments, variations cannot be prevented in the amount of trade credit given and received that passes through the official (that is, the current) market. Again, there may be variations in the amount of working balances arising out of, or to be used for, current transactions held in domestic currency by nonresidents or in foreign currencies by residents, including the banks and dealers in foreign exchange. In addition to working balances proper, the banks may be allowed some variability in their spot positions to enable them to provide a reasonably stable market in forward exchange for their commercial customers. Some types of current transactions, notably those that are financed by travelers checks or banknotes, are usually put through the financial market, because if they were admitted to the official market they would serve as a cover for illicit capital transactions. On the other hand, it is difficult to prevent evasions under which certain types of current transactions—for example, border trade, trade in diamonds, emigrants’ or immigrants’ remittances, interest and dividend payments, and other current invisibles—are paid for through the “capital” exchange market, when the rate there is more profitable than the rate in the “current” market. It would be particularly difficult to ensure that foreign firms paid the appropriate tax on retained and reinvested earnings when the “capital” rate was at a premium, or that national firms paid this tax on their retained earnings of their foreign branches when the “capital” rate was at a discount. There may also be inhibitions about applying an adverse rate to the repatriation of capital, both resident and especially nonresident. All of these difficulties also apply to capital control systems of all kinds, not merely to the dual market.

The general effect of these loopholes and imperfections in the segregation of capital and current transactions for the operation of the dual market depends on the intervention policy of the authorities.5 If in the absence of evasion the existence of the dual market would be a factor making for equilibrium in the balance of payments, any increase in evasion by reducing the discrepancy between the “current” and the “capital” exchange rates would tend to promote a payments imbalance. If, however, the authorities altered their intervention on the capital market in such a way as to restore the original rate discrepancy, they would—despite the incentive given to further evasion—reduce the overall payments imbalance to approximately the original, evasion-free level.6

Insofar as property income transfers are allowed to pass through the capital exchange market or insofar as they do so illicitly, one of the basic mechanisms of the dual market—the effect of the rate discrepancy on the yield of capital—will be crippled. It will operate only where it continues to be profitable for investment and investment income to pass through separate exchange markets—that is, only for foreign capital where the domestic currency is at a discount in the capital exchange market or only for domestic capital where the “capital” rate is at a premium. If, as in France, the foreign investment income of nationals has to be repatriated at the “capital” rate, even the latter effect may be precluded. However, as has been seen, the speculative or “capital gain” effect of the dual market is probably more important than the “yield” effect.

As was shown in the author’s previously cited article and as will be seen later in this paper, some of the most interesting characteristics of the dual exchange market system arise when it is applied simultaneously by a number of countries—on a national basis, not collectively. In this case capital flows between these countries will be influenced in a direction and to an extent determined by the relative premiums or discounts on their “capital,” as compared with their “current” rates. Thus, flows from a country where the “capital” rate is at a premium of 10 per cent to a country where it is at a premium of 5 per cent will benefit from a subsidy of approximately 5 per cent.

III. Evaluation of Dual Exchange Market

The alternative techniques for dealing with temporary and reversible payments imbalances, with which the dual exchange market is compared, may be listed as follows:

(1) Quantitative or administrative restrictions on capital transactions or balance sheet positions.

(2) Taxes or subsidies affecting capital transactions or income from capital.

(3) Official intervention in forward exchange markets.

(4) Monetary or interest rate policies.

(5) Official financing or use of reserves.

(6) Floating exchange rates or wider margins.

No consideration has been given in this paper to systems of par value adjustment or nonadjustment—such as absolutely fixed parities or parities which cannot alter by more than a small percentage a year. Although proponents of these systems have claimed great merits for the avoidance or mitigation of disequilibrating capital flows, examination of them would be long and the author is skeptical both of their feasibility and of their power to fulfill the primary function of a par value system—namely, the correction of fundamental disequilibria.7

dual exchange markets versus capital restrictions

The dual exchange market, when compared with quantitative or administrative capital restrictions as a means of influencing net capital flows, has two major advantages and a single disadvantage. To the extent that it is effective, it is likely to be preferable from the standpoint of economic welfare to any type of quantitative or administrative restrictions; it is likely to achieve a higher degree of effectiveness against evasion for a given cost of administration than any form of restriction which relies on the screening of individual transactions. It may, however, have a stronger disequilibrating effect on exchange rate speculation.

The superiority of the dual exchange market from the welfare standpoint arises from its use of the coordinating mechanism of the market. Over the wide range of transactions which it covers, it is equivalent to a uniform, though not constant, percentage tax (subsidy) on purchases of securities by domestic residents from foreign residents, combined with an equal subsidy (tax) on sales of securities by domestic to foreign residents (all taxes or subsidies being measured as a percentage of the price in the domestic market). Quantitative or administrative restrictions, on the other hand, are inevitably uneven in their incidence because of their piecemeal character and because they operate by determining quantities of assets or liabilities traded (or held vis-à-vis the outside world) rather than by price differences. In most cases they operate in the same manner as taxes to restrict transactions in one direction without operating (as subsidies do) to encourage transactions in the opposite sense.

Let us assume that the price of a security in any country represents its true marginal utility to the holders in that country. Then the following can be shown. If country A has placed on the importation of two types of securities from country B restrictions that are unequal (in the sense that they are equivalent to unequal ad valorem import taxes)—then if the import of the lower taxed security were increased and that of the higher taxed security were reduced—as would tend to happen if the two rates of tax were brought closer together—in such a way as to leave the balance of payments between the two countries unaffected, while any repercussions on the prices of the securities traded between the two countries were compensated by unilateral transfers, any such adjustment would be beneficial to some of the residents of both countries and harmful to none. Such adjustments could continue with advantage until the rate of tax became equal on all imports of securities from B to A.

It can also be shown that, on the same assumption with respect to the payment of compensatory transfers from B to A, country A could with advantage promote the exportation of securities to country B by means of a uniform subsidy on all securities and that the subsidy on exportation should equal the tax on importation of securities when measured in the percentage of the prices of the taxed or subsidized securities in country B (see Appendix, Section B). Now, this is precisely the effect that would be achieved if country A applied a uniform exchange premium on its currency in capital transactions with country B. Since price repercussions would not, in reality, be compensated by unilateral transfers, the optimal relationship between taxes and subsidies on capital flows would be somewhat more complicated than that described above.8 The latter situation, however, probably represents the approximation to the optimal situation most suitable for adoption as a convention to govern international relationships.9

By analogous reasoning it can be shown that restrictions and incentives affecting capital transactions among any group of countries may be assumed to be optimal if they result from application by each country of a uniform exchange premium or discount on capital transactions with all other countries. This implies that capital transactions between any two countries would be affected by the premium or discount, as the case may be, in both countries, so that, for example, a capital transaction between two countries which had equal premiums on their capital exchange rates would in effect be free from taxation or subsidization.

The discussion here involves the choice between one method of capital control and another, and it is assumed that “current” exchange rates, price levels, and the balance of payments on current account remain approximately unaffected by this choice. Therefore, the welfare superiority of the dual exchange rate technique relates not to its effect on the flow of real resources nor on the geographical distribution of physical capital among countries but mainly to its effect on changes in the composition of portfolios and on the distribution of real investment among industries and types of firms within countries. The general effect of substituting dual exchange rates for quantitative controls would be to foster (1) the international diversification of portfolios and (2) multinational business. There may be some social disadvantages in doing this, but if it is desired, for example, to keep domestic industry in domestic hands, this should be promoted by suitably adapted measures of a permanent nature rather than by methods that vary in their effectiveness with the balance of payments situation.

A second advantage of the dual exchange market over direct administrative restrictions on capital flows is that, with an equal degree of severity in penalizing evasion, it is likely to achieve either a given balance of payments effect with less administrative effort than the direct restrictions on capital flows or a greater balance of payments effect with the same administrative effort. The reason is that, if a set of administrative restrictions is compared with a particular level of the capital 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 rate would be likely to give a smaller incentive to evasion than would the quantitative measures.

Administrative interference with capital flows generally operates by inhibiting or restricting rather than by promoting such flows. Moreover, as has been seen, it restricts very unevenly.10 Some transactions that are barely profitable may get through the controls, while others that would be very profitable are inhibited. A uniform tax on such flows (with an effect, in the absence of evasion, equal to that of the restrictions) would be likely to promote less evasion, since the transactions inhibited by the tax would all be of a less profitable sort; while some of those inhibited by administrative interference would be of a more profitable sort where the incentive to evade would be greater. A dual exchange rate that has an equivalent evasion-free effect offers an even smaller incentive to evasion than the uniform tax. Since part of its effect is achieved through promoting transactions by subsidization, the inhibiting exchange premium or discount is lower than a uniform tax of equivalent effect, and the transactions inhibited by it are of lower average profitability. It is true that the effect of the dual market in subsidizing capital flows in one direction may enhance the profitability of illicit capital flows through the current exchange market in the contrary direction, but this is likely to be a second-order effect.

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. There is reason to think that some of the countries that have applied dual rates, notably Belgium, have taken advantage of this to apply less onerous surveillance and enforcement measures than they would have done had they relied instead on quantitative restrictions. It is sometimes asserted that “experience” has shown that the dual market system is undermined by evasion as soon as the premium or discount on the capital exchange market widens beyond a moderate level—at least for any considerable period of time. It seems possible, however, that this is true only where the authorities have been unwilling to police and enforce the segregation of capital and current transactions with the same energy that they would have had to use to enforce capital restrictions of a more familiar type.

An indication of this is seen in the very substantial discounts prevailing over periods of years in markets such as the market for security sterling, in which nonresidents bought and sold currency usable for transactions in domestic securities—or in markets such as the U.K. market for investment currency, in which residents buy and sell foreign currency derived from and usable for transactions in foreign securities. Facilities for arbitrage between two such markets are all that is necessary to create an exchange market for capital transactions, and it is difficult to believe that the addition of this element of arbitrage need fatally weaken the barriers to evasion.

The incentive to evade restrictions depends not only on the profitability of such evasion but also on the ease with which it can be carried out. There is no reason why controls over capital flows such as those operating through the licensing of actual transactions or payments should have any advantage over the controls required for implementation of the dual market in regard to evasion or to the ease of preventing it. The situation may be otherwise, however, when capital flows are controlled indirectly through quantitative regulation of the external positions or through foreign exchange parities of banks or other enterprises. Such regulations, although they have the other defects of quantitative restrictions described above, probably have the merit of being easier to enforce with a given degree of effectiveness than the regulations underlying dual exchange markets. However, they can provide an alternative to direct controls over capital transactions or to dual exchange markets over only a part of the field.

Evasion is undesirable in that: (1) it produces allocative inefficiencies among different types of capital flows and different types of current transactions; (2) it absorbs productive resources in the actual process of evasion and in the prevention thereof; (3) it involves a loss of income to the authorities; and (4) it either weakens the desired equilibrating effect in the balance of payments or, if this is countered by intensifying restrictions, intensifies the other evils mentioned. The superiority of dual markets over direct administrative restrictions in the matter of evasion therefore yields advantages in all of these respects.

As regards the cost of administering the controls, a distinction has to be drawn between direct controls over capital transactions and controls over external positions. The former are probably more costly to operate than dual markets for equal effectiveness, not only because of the greater incentive to evasion but for other reasons as well. Not only do they require government decisions as to the quantities of the different categories of external assets or liabilities to be held or transacted by residents; they also require that the different types of transactions be not only identified, as in the case of dual markets, but also regulated as to quantity. In the case of dual markets the first requirement does not exist; the second stops at identification. Where quantitative controls are exercised with respect to amounts held rather than transacted, however, the costs of ensuring that external positions are what they should be are probably quite small for large holders such as banks and financial institutions.

The main disadvantage of dual markets as compared to quantitative capital restrictions lies in the visibility of the exchange premia and discounts on the capital exchange market and in the speculative repercussions which may arise therefrom on the current exchange market. By creating an expectation that the official rate will move in the direction of the “capital” rate, such repercussions may reduce the extent to which the “capital” rate is expected to revert to its original norm, may increase the spread between the two rates that is required to keep capital flows under control, and may strengthen the incentive to evasion.

The question arises whether dual exchange markets, by their very effectiveness in maintaining temporary payments balance when it is inadvisable to adjust the current account balance of payments, might become an obstacle to adjustment of the current balance when appropriate. In this context, however, the greater visibility of market premia and discounts as compared with quantitative barriers, which was a disadvantage in dealing with temporary disequilibria, may be advantageous in that such premia and discounts, if continued over a long period, can be taken as an indicator of the need for fundamental adjustment.

What has been said thus far about administrative and quantitative restrictions relates to restrictions on capital transactions not directly linked to current transactions. However, some of these restrictions have to do with types of capital flows directly related to current transactions, such as changes in foreign trade credit outstanding, in the temporary holding by residents of working balances in foreign currency derived from export receipts or to be used for purchase of imports, or in analogous working balances held by nonresidents in domestic currency. Variations in trade credit and in such working balances can give rise to significant net capital flows (“leads and lags”) in response to changes in relative national interest rates and to very large net flows in response to changes in expectations about exchange rates.

To prevent such flows from taking place through the “current” exchange market, as would be required for a complete segregation of current and capital transactions, it would be necessary to synchronize current payments with the corresponding current transactions and also with the associated exchange transactions, so that any trade credit extended would have to be carried out as a separate transaction on the capital exchange market. To enact a real synchronization would be inconvenient and possibly seriously discouraging to trade. It may therefore be preferable to tolerate such flows through the “current” exchange market or to limit them by a looser type of regulation governing the timing of payments ii relation to the underlying transactions, the amounts of residents’ and nonresidents’ working balances, or the time elapsing between credits and debits to such balances. Such regulations should not be regarded as alternatives to the use of a capital exchange market but as “imperfect” complements to such use.

dual exchange markets versus fiscal interventions

Fiscal interventions affecting capital flows may take the form of taxes or subsidies on capital transactions, on the external assets or liabilities of residents, or on income receipts and payments derived from such external assets or liabilities. They may also take the form of taxes or subsidies on domestic capital outlays that will permit domestic interest rates to be higher or lower than they would otherwise be, thus attracting or repelling capital flows.

Taxes and subsidies on capital transactions, if enforced by exchange controls, would resemble dual markets in potential effectiveness and susceptibility of evasion. They would, however, be far less flexible than dual markets and would therefore be less effective in arresting short-term fluctuations in payments imbalances. Also, since such interventions are generally applied without the enforcement aid of exchange control, they are likely to be far more open to evasion and far less effective than dual markets. Certain types of taxes or subsidies on external positions may be enforceable without the aid of controls over transactions, but they are likely to cover only a limited range of potential capital flows. Taxes and subsidies on domestic capital outlays, with the associated variations in interest rates, might well be the most effective and comprehensive fiscal instrument for influencing capital flows, but they suffer from the same lack of flexibility as other fiscal instruments.

dual exchange markets versus forward exchange market intervention

If a country is in temporary payments surplus, its authorities may seek to check the inflow by acquiring foreign exchange on the forward exchange market or by “swapping out” reserves to its commercial banks; if it is in temporary deficit, they may seek to check the outflow by supporting the national currency on the forward exchange market. Such forward market intervention is likely to evoke private counter-transactions of a speculative or hedging character, since it will make such transactions cheaper; occasionally, however, if the movement in forward rates affects expectations about future spot rates, it may have the opposite effect. In any event, the greater part (if not the whole) of the amount of official intervention is likely to be matched by a shift in private balances covered forward, and it is this shift that produces the effect on reserves desired by the authorities.

Forward market intervention, like the dual market, is a very flexible technique, affecting both inward and outward movements of funds and both resident and nonresident capital. Moreover, it has the great merit of being able to influence to some extent just those types of capital flow—leads and lags connected with current transactions, and shifts in working balances—which the dual market cannot reach. On the other hand, it is more limited in scope, being confined to short-term funds of those (banks, traders, and others) who wish to avoid exchange risk and therefore cover their foreign exchange positions on the forward market or of those who can be induced to shift their speculative positions from the spot market to the future market.

Again, forward market intervention may be expensive, if, for example, it is applied to resist an interest-motivated flow of funds or a speculative flow that is followed by a change in the spot exchange rate in the expected direction. In the first case the authorities are likely to have to buy forward exchange at a premium or sell it at a discount; in the second case, although the opposite is true, they may have to take a loss on outstanding contracts when the spot rate is adjusted. With dual markets, on the other hand, the authorities can usually obtain some relief from the undesired influx or efflux of reserves without intervening in the capital exchange market, and thus avoid losses; only if they have to intervene to push the forward and official rates still further apart will they begin to create losses.

Because of the difference in their range of influence, dual exchange markets and intervention on the forward market for current transactions may be used with advantage as complementary instruments of policy.

dual markets versus fiscal-monetary policy

When countries are affected by balance of payments deficits or surpluses that are believed to be temporary, the response traditionally recommended by conservative economists and bankers is for the deficit countries to raise interest rates and contract credit and for the surplus countries to lower interest rates and expand credit in the attempt to evoke equilibrating capital flows or to stem disequilibrating ones. Today it is generally recognized that such a policy carries with it dangers to domestic stabilization policy, and it is customary to recommend its application in conjunction with fiscal policies designed to counteract any undesired inflationary or deflationary effects of the monetary policies. This fiscal-monetary mix, with its mélange of Keynesian and pre-Keynesian elements, is popular among those who are reassured by the familiar and market-respecting character of the instruments employed. In reality, however, of all the policies for coping with disruptive capital flows, it is one that is most likely to generate diseconomies and distortions.

In comparing reliance on the fiscal-monetary mix with the dual exchange market system, it is necessary to draw a distinction between a mix in which the fiscal element is confined to measures affecting capital outlay and a mix in which it is confined to measures affecting consumption. As an example of the former case, a country in temporary deficit might raise its interest rate by a restrictive monetary policy and offset the effect on domestic capital outlay or investment expenditure by subsidizing the latter. Such a policy would in principle operate in a manner similar to a dual exchange market in which the financial rate was allowed—or made—to fall. There are, however, certain differences. The fiscal-monetary mix would affect all sorts of capital flows, including those (such as leads and lags) which normally bypass the financial exchange market; to this extent it is superior to the dual market system. On the other hand, certain (relatively unimportant) diseconomies would result from the separation of the interest rates applicable to domestic savings and investment, but these would not occur under the dual market system. The real inferiority of the fiscal-monetary mix, however, lies in the area of practical application. It would not be feasible in practice to apply taxes or subsidies to all forms of capital outlay in a country, and it would be still less feasible to vary them flexibly in harmony with interest rates so as to bring about the desired level of net capital flows—a result that the market can bring about automatically through variations in the financial exchange rate.

By extending fiscal manipulation beyond the area of capital expenditure to forms of public expenditures and taxation that presumably affect consumption, it would become technically somewhat more feasible to offset the aggregate demand effects of internal rate variations designed to bring about the appropriate capital flows. Even in this case, however, it would hardly be feasible to vary budgetary surpluses and deficits to the extent that would be required to offset the very large variations in interest rates needed to restrain disequilibrating capital flows of the speculative sort (as distinct from interest-motivated flows resulting from differential phasing of cyclical situations in different countries). In principle it should be technically possible to deal with exogenous demand variations associated with the business cycle while keeping interest rates at levels that, in the absence of exchange speculation, would evoke the required inflows or outflows of capital. Such a result, however, is one which governments are understandably reluctant to attempt, since it involves considerable diseconomies and sectoral maladjustments.

The diseconomies in question arise from the fact that if budgetary surpluses or deficits are used to stabilize aggregate demand by expanding or contracting consumption, they are likely to cause the level of national savings to diverge from the level corresponding to the true preferences of individuals and public authorities.11 If the circumstances that give rise to this necessity are temporary and reversible, no permanent harm will have been done. Considerable harm may be done, however, if undersaving or oversaving persists for a long period of years and is not recognized for what it is—a symptom of basic disequilibrium in the balance of payments.

Sectoral maladjustments are likely to arise from the fiscal-monetary mix where this is used as a way of dealing with payments disequilibrium evoked by speculative flows or by interest arbitrage associated with cyclical discrepancies in the incentive to invest. Variations in demand induced by interest rate variations affect primarily the construction industries and secondarily (partly through the effect on the stock market) investment goods industries. Compensating variations in budgets primarily affect consumer goods industries. Such compensation is therefore imperfect and is likely to result in a patchy situation combining unemployment and scarcity in different sectors. Admittedly, where the payments disequilibria are due to temporary export variations, the inevitable sectoral maladjustments may be more easily mitigated by using fiscal policy for internal stabilization and by using monetary policy to equilibrate external payments and receipts than they could be by attempting to use monetary policy for stabilization purposes. Even in the last-mentioned case, however, a still better result might be attained by using intervention on a capital exchange market to equilibrate the balance of payments without disturbing the domestic interest rate and by using selective budgetary policy to offset the effects of export variation on sectoral and total demand.

In summary, the dual exchange market offers a far more flexible technique of equilibration than the fiscal-monetary mix, and it is free from the deleterious effects of the latter on savings and sectoral adjustment. Despite the gaps in its coverage of capital flows (for example, leads and lags), it is far superior to the mix in dealing with violent speculative movements of funds; its advantage may be somewhat less marked in dealing with more moderate flows of a cyclical nature, though in this case the gaps in coverage may be less important. In this case a combination of the two techniques may be advantageous, especially if the dual market is applied in such a way as to permit substantial evasion.

dual exchange market versus official financing

There are two methods whereby governments or monetary authorities can finance balance of payments deficits (surpluses): (1) by using (accumulating) reserves; and (2) by borrowing abroad (repaying foreign loans).

There are two great disadvantages of official financing, particularly in the form of reserve movements, from which dual exchange markets (like other devices for preserving payments balance) are free and two others which they share.

(1) Reserve losses tend to bring about a contraction of money supply and bank credit along with a decline in security prices and expenditures, while reserve increases tend to have the opposite effects. These effects, which are likely to be unwelcome to the authorities from the standpoint of demand management, in principle can be prevented by a combination of high marginal bank reserve requirements, variable reserve requirements, official purchases and sales of securities, quantitative credit controls, and so on. However, particularly in the case of reserve inflows, some authorities are not equipped to carry these measures as far as would be necessary for complete offsetting and others are unwilling to do so, either because they impair the profits or liquidity of particular banks or because open market sales are expensive for the authorities themselves. From a national standpoint, also, it is costly for a country to permit foreigners to acquire assets carrying a substantial yield in exchange for reserves yielding a low return or none at all.

The financing of payments imbalances through official lending or borrowing gives rise to much the same difficulties for monetary management as do reserve movements. Increases or reductions in official lending act like reserve changes and indeed usually are reserve changes; increases or reductions in official liabilities do not help to offset the domestic effects of payments surpluses or deficits. Insofar as the sterilization of payments imbalances is concerned, one country can help another only by borrowing or lending in the other’s private credit or capital markets.

Given the technical and psychological limitations of official offsetting policies, reliance on reserve movements or on liability financing to meet external disequilibria carries the danger that the countries concerned will be exposed to inflationary pressures when in payments surplus and to deflationary pressures and unemployment when in payments deficit.

(2) Access to certain forms of official financing, if available to countries generally on a scale that would suffice to meet possible deficits caused by temporary capital flows, would be likely to impart an inflationary bias to the world economy. This would not apply to Fund drawing facilities in the credit tranches, because these can be utilized by countries only on condition that they adopt prudent noninflationary policies. However, it would constitute an objection to the provision of reserves, including currency holdings and SDRs on the scale that would be required for this purpose, since the yield on such reserves (in the form of interest and possible capital appreciation) is generally considerably lower than that of assets which the holding country could acquire at home or abroad if its authorities chose to adopt a more expansionary policy that would use up reserves. Although a separate exchange market for capital transactions, on the other hand, could also be manipulated to finance the payments deficits entailed by an inflationary policy, it would not offer the same temptation to the authorities, since the funds it would attract would have to be remunerated at the current rates on international markets. Moreover, if the authorities were to attract such funds by depreciating the “capital” rate, they would themselves incur increasing financial losses eventually.

On the other hand, official financing is likely to have certain advantages over the dual exchange market in that (1) it may involve less distortion and disturbance between different sectors of national capital markets, and (2) by reason of its very limitations in dealing with short-term disequilibria, it may offer less resistance to fundamental adjustment in the face of persistent disequilibria.

(1) The inevitable evasion and bypassing of the obstacles to capital flow set up by a “capital” exchange rate that diverges from the “current” rate might lead to certain distortions in the relative prices of different types of assets in countries applying the dual market, which a skillful policy of stabilization and offsetting of reserve movements would have avoided. This, however, is not likely to be of great quantitative importance.

(2) It seems evident that, however clearly a persistent premium or discount on the financial exchange market would signal the existence of a fundamental disequilibrium, a persistent reserve flow or even a persistent resort to borrowing by monetary authorities would signal it even more clearly. Moreover, owing to the necessarily limited availability of official financing, persistent losses of reserves are almost bound to enforce adjustment on the part of deficit countries, although the same is not equally true of reserve gains by surplus countries, until such gains grow to massive dimensions.

dual exchange markets versus flexible unitary rates

Flexible rates here mean rates that float, subject only to intervention sufficient to smooth out minor short-term fluctuations. Such floating may have no definite limits or may occur within rather wide margins on either side of a central rate fixed against an intervention currency. It is convenient to begin by comparing dual exchange rates to flexible unified rates that are unhampered by formal or informal margins.

Flexible unified spot exchange rates tend to make capital flows more equilibrating or less disequilibrating to the extent that, as the rate falls, the expectation that it will fall (rise) further is reduced and the expectation that it will rise (fall) again is increased. Given such a negative elasticity of exchange rate expectations, disequilibrating elements affecting the balance of payments will tend to produce rate movements that bring equilibrating capital flows into play and thus limit the movement in the rate. The probability that, after half a year or so, changes in the exchange rate are likely to begin to produce an equilibrating effect on the current account items in the balance of payments—an effect that will grow more powerful as time passes—may be a good reason why rate expectations beyond the short run should behave in the manner described above. Many of those who move funds in response to exchange rate changes, however, are influenced by short-run considerations; therefore, it is unfortunate that the early effects of rate changes on the current account are likely to be perverse. In these circumstances, even if rate changes are at first checked by expectations of early reversal, these expectations may well be disappointed and hence revised so that rates in a “clean” float may move far from the level toward which they might reasonably be expected to move in the longer run. This is a weakness in uniform floating exchange rates that is not shared to the same extent by the floating “capital” rates in a dual rate system.

Experience with floating rates has been very diverse. When they have been in operation over a considerable period with reasonably stable internal conditions and no large structural disequilibria in the balance of payments, flexible rates have shown considerable stability in the face of random shocks. This is especially true where the authorities have intervened to smooth out day-to-day variations and have striven, if only indirectly through financial policies, to keep longer-term variations within a normal range. However, where floating has occurred in circumstances of basic disequilibrium or where demand management has been inadequate, wide fluctuations in rates have taken place.

The disadvantages of flexible unified spot rates, as compared with dual markets, are that they are disturbing to trade and to the foreign trade industries.12 While the exchange risks of particular trade transactions can usually be covered chiefly in the forward exchange market, the risks of production and especially of investment in the foreign trade industries cannot be so covered. Such investment may thus be misdirected or unduly curtailed.

Again, a system of flexible rates may be more open to the abuse of competitive exchange depreciation than would a dual market, because such depreciation could in the former case be achieved through market intervention whereas in the latter it would require a par value change.

On the other hand, since flexible rates exercise their equilibrating effects on all forms of capital flow (including leads and lags and variations in convertible balances) and since they are not subject to evasion, they are free from distortions between different types of capital flow and from the waste of resources in the evasion and enforcement of controls to which dual rates are subject. For the same reason they are far more certain of achieving the required equilibrating effect on the balance of payments. This consideration may in practice make floating rates inevitable where really massive speculative capital flows are concerned, even when in principle other solutions might be preferred.

Finally, flexible rates have the important advantage that if (as very often happens) an apparently temporary payments disequilibrium masks a basic disequilibrium or is accompanied by one, the consequent movement in the spot rate will initiate the basic adjustment that is required, while a dual market would simply bottle up the disequilibrium until a persistent—or persistently growing—premium or discount in the “capital” rate showed it to be basic in character.

There are two ways in which the variations in floating rates may be limited by official intervention. The first is the adoption of rigid, publicly announced, but widely spaced margins of exchange rate variation around an announced though adjustable central rate. The second is a much more informal arrangement, under which there is no announced margin or central rate but the authorities in fact take action, either through direct intervention on the exchange market or through the adoption of the sort of equilibrating short-term balance of payments policies that have been discussed in earlier sections of this paper.13 Thus, they prevent rates from diverging too far from whatever rate would seem at the time to be mostly likely, if maintained, to bring about payments equilibrium over the next several years. This second type of limitation would seem to be a sine qua non if the flexible rate system is to operate with a reasonable degree of stability. Such limits would greatly enhance the equilibrating effects of exchange rate changes, since the market is more likely to anticipate a return to a normal rate if it feels that the authorities themselves have such a norm in mind and are prepared to enforce it. A readiness on the part of the authorities to intervene in this way would thus strengthen the case for using floating rates (within the limits) as an instrument to deal with temporary and reversible disequilibria, in preference to other instruments, such as dual rates.

A system of announced central rates with wide but rigid margins within which exchange rates could float freely might exercise an even stronger equilibrating effect on capital flows than the system of informally limited floating just discussed—not to mention a system of unlimited floating—if there is confidence that the central rate and the margins are likely to be held. If, however, there is a fear of devaluation or revaluation of the central rates, the opposite may be true, since under the informal intervention system there is no clear target to speculate against and those speculating in such a way as to exaggerate exchange rate fluctuations can never be sure when they may encounter aggressive official opposition on the exchange market. Moreover, if there is a basic payments disequilibrium, the fixed central rates and margins may inhibit or delay the necessary adjustments in the same way as the fixed “current” exchange rates under the dual market system.

IV. Conclusions

From what has been said, it would seem that dual exchange markets, supplemented as necessary by controls over the timing of current payments and by controls or incentives affecting changes in nonresident holdings of convertible balances, have considerable advantages over most of the other methods of dealing with temporary and reversible payments disequilibria—with the possible exception of unified rates which (within reasonable limits) are allowed to float freely, subject to day-to-day smoothing operations. They are, at least potentially, more effective and more economic in their incidence than quantitative restrictions on capital transactions, more flexible than fiscal measures, more flexible and less disturbing to production and saving than the fiscal-monetary mix, of wider scope than forward market intervention, less inflationary than some forms of compensatory official financing, and less disturbing to trade—though also less effective—than floating unitary rates. Where persistent disequilibria are present, however, adjustment in the spot exchange rates is clearly indicated, and this can be most smoothly and easily attained by the use of floating rates.

possibility of a hybrid system

This suggests the possibility of a hybrid system under which (1) capital and current transactions would be segregated in separate exchange markets; (2) both the “current” and the “capital” rates would be allowed to float; (3) official sales of foreign exchange on one market would be balanced by official sales on the other, so that there would be no net reserve change or at least any net reserve change would be confined to amounts deemed internationally appropriate; and (4) the gross amount of intervention would be such as to keep rates on the two markets together, except when the rate thus unified tended to diverge too far from its presumed longer-term norm, in which event the “current” rate would be prevented by intervention from diverging further from its normal level, while the “capital” rate would find its level subject to the operation of rule (3).14

For example, if the two rates were equal but were tending to fall too far below the “norm,” the authorities could intervene on both markets in such a way as to produce a discount of the “capital” rate compared with the “current” rate. This would raise the “current” rate or prevent it from falling as far as it otherwise would. However, if the discount on the “capital” rate persisted too long, it would be a sign that the “norm” for the “current” rate might have been set too high.15

Under a system of this kind it would not be necessary for large long-lasting divergences between the two rates to emerge. It might therefore be unnecessary to take any very drastic steps to interfere with capital flows through the “current” market in the form of leads and lags, and so on.

principles for operation of dual market system

It may be well to conclude by reiterating two principles that should govern the operation of a dual market system, whether of the simple or of the “hybrid” type:

(1) The capital exchange market, and in particular official intervention on that market, should be conducted so as to promote short-run equilibrium in the overall balance of payments of the country concerned. This should be interpreted as compatible with aiming to achieve a rate of reserve growth that is internationally appropriate in relation to world reserve growth and to the need to achieve a gradual improvement in the international distribution of reserves.

(2) The persistence of chronic premia (discounts) in the “capital” exchange rate as compared with the “current” exchange rate should be recognized as one indication of the need to raise (lower) the par value of the country concerned or (in a floating exchange rate system) the range within which the exchange rate for current transactions is expected to move in the medium term.

APPENDIX Optimal Relationship Between Taxes and Subsidies on Capital Flows

Let there be only two countries, A and B.

U, UA, and UB ≡ utility in general, in A, and in B

qx ≡ amount of securities (capital account items) exported from A to B

qm ≡ amount of securities imported from B to A

PxA and PxB ≡ price of A’s securities in A and B

PmA and PmB ≡ price of S’s securities in A and B

t ≡ proportionate tax in A on import of securities from B


s ≡ proportionate subsidy in A on export of securities to B


ϵ ≡ elasticity of net supply of B-securities to A


μ ≡ (negative) elasticity of net demand for A-securities in B


UA and UB ≡ marginal utility of money in A and in B.

Then U = UA (qm, qx, …) + UB (qm, qx, …)



A. Constant Payments Imbalance

Let t and s vary in such a way as to leave unchanged the balance of payments between A and B. Since the exchange rate and level of money income in A and B are kept constant, it is assumed that the current account items in the balance of payments remain constant also, with only the flow of securities being affected. The balance of payments constraint is, therefore, as follows:


where δ signifies a total differential.

As t and s vary, subject to this constraint, the effects on UA, UB and U are :


Since competitive conditions are assumed to prevail in security markets in both A and B, and since the marginal utility of money is assumed to be the same for all residents of each country, though different as between countries, therefore:






From equation (1),


Substituting in equation (2),


When t and s are at their optimal levels, subject to the balance of payments constraint, δU = 0, that is:


The foregoing analysis assumes independent demand curves for A’s securities and B’s securities in B and ignores cross-elasticities linking changes in demand for A’s securities to changes in the price of B’s securities and changes in demand for B’s securities to changes in the price of A’s securities. In the case of securities, one would expect such cross-elasticities to be high. The broad effect of taking cross-elasticities into account, when s and t are brought closer together, would be similar to that of a higher level of μ and ϵ.

B. Compensatory Payments Between Countries

If the variations in t and s had been accompanied by compensatory payments (positive or negative) from A to B, sufficient to prevent any changes in UB, then, by hypothesis:




This equation (1’) now takes the place of the balance of payments constraint (1). Then


which, from equation (1),


At the optimum point,




Mr. Fleming, Deputy Director in the Research Department, is a graduate of Edinburgh University. He was formerly a member of the League of Nations Secretariat, Deputy Director of the Economic Section of the U.K. Cabinet Offices, U.K. representative on the Economic and Employment Commission of the United Nations, and visiting Professor of Economics at Columbia University. He is the author of Essays in International Economics and numerous articles in economic journals.


J. Marcus Fleming, “Dual Exchange Rates for Current and Capital Trans actions: A Theoretical Examination,” in Essays in International Economics (Har vard University Press, 1971), pp. 296–325. See also Vittorio Barattieri and Giorgie Ragazzi, “An Analysis of the Two-Tier Foreign Exchange Market,” Banca Nazion ale del Lavoro, Quarterly Review, No. 99 (December 1971); Pascal Salin, “Ui double marché des changes est-il justifié?” Revue d’Economie Politique, No. (November-December 1971), pp. 959-74; and Anthony M. Lanyi, “Separat Exchange Markets for Capital and Current Transactions” (unpublished, Intei national Monetary Fund, 1973).


Fleming, op. cit., pp. 298 ff.


If the system is to make its maximum contribution to buffering external influences on the domestic monetary system, a slightly different criterion must be followed—that of buying or selling as much domestic currency on the capital market as is sold or bought on the official market. This involves some reserve changes, corresponding to net profits or losses.


Compare Barattieri and Ragazzi, op. cit.

Herman C. Verwilst (doctoral candidate, The Johns Hopkins University) has drawn my attention to a third possible mechanism whereby movements in the “capital” rate might affect capital flows, namely, through a change in the degree of uncertainty of investors regarding the future level of that rate. A rise in uncertainty such as might occur when the “capital” rate diverged from its normal relation to the “current” rate would presumably discourage capital flows, both inward and outward, and also reduce the net capital flow corresponding to any given level of the capital rate.


Lanyi, op. cit.


The reason why maintenance of the original rate discrepancy does not maintain precisely the original payments balance is that the evaders make profits—in reserves or in domestic currencies—at the expense of the authorities. Maintenance of the original payments balance in the face of evasion thus calls for some widening of the original rate discrepancy when the domestic currency is at a discount in the “capital” exchange market, but is compatible with some narrowing of the discrepancy when the domestic currency is at a premium in that market.


See, however, John H. Williamson, The Crawling Peg, Essays in International Finance, No. 50 (Princeton University Press, December 1965).


More precisely, as can be seen from the Appendix, Section A, optimization requires that



  • ϵ = elasticity of net supply of B securities to A

  • μ = elasticity of net demand for A securities in B

  • UA, UB = marginal utility of money in A, B

  • t = the tax in A on the import of securities from B as a proportion of the price in B

  • s = the subsidy in A on the export of securities to B as a proportion of the price in B.

This formula will permit equality between t and s, that is, a uniform capital exchange rate only if either

(a) the elasticity of net demand for A’s securities in B and the elasticity of net supply of B’s securities to A are both infinite


(b) the marginal utility of money in A relative to that in B happens to be precisely equal to the exchange rate for B’s currency in terms of A’s currency for capital transactions. In this case both numerator and denominator on the right-hand side of the equation would be zero.


The foregoing argument calls for some qualification to take account of the possibility that, given the uncertainty of the future “capital” rates in any dual exchange market system, such a system may create greater uncertainty regarding the profitability of capital flows both inward and outward than would a system of administrative restraints on such flows (compare footnote 4). To the extent that this is the case, there is a possibility that the dual exchange market system may unduly restrict gross capital flows in a way that the system of administrative restraints would not, though the effect on net flows might be welcome.


Apart from the unevenness, inseparable quantitative controls and licensing arrangements, countries are often loath to apply such restrictions to the outflow of nonresident funds or the inflow of national funds held abroad. Inhibitions about applying the “capital” rate to such transactions may be less marked.


This is true whether the variation in investment demand that has to be compensated for is due to a change in interest rates undertaken for balance of payments reasons or to a change in the incentive to invest to which monetary policy, for balance of payments reasons, fails to respond.


Both flexible unitary rates and dual markets are open to objection because of the unduly discouraging effect which their uncertainty may exercise on gross capital flows, especially in the medium term, and because of their tendency to discourage net capital flows whether or not this is called for in the interests of short-term equilibrium. Neither appears to have a differential advantage over the other in this respect.


The use of the dual exchange market as an instrument to limit the range of variation of a unified floating rate is discussed below.


Some of the features of this arrangement are approximated in Italy’s present dual float.


It is vital to the operation of such a system that official intervention on the two exchange markets follow rules (3) and (4) above. Any attempt to operate the system without official intervention could be saved from disaster only by leakages from one market to the other, since a completely segregated current exchange market would be highly unstable as the impact effect of exchange rate changes on the current account would naturally be perverse.