THE TRANSFORMATION of the international monetary system since 1971 has provided governments of member countries of the International Monetary Fund with a far wider range of choice with regard to their exchange rate regimes than had previously existed. Before 1971 most member countries of the Fund had a declared par value for their currencies, with margins of 1 percent and the obligation to maintain the par value unless a “fundamental disequilibrium” could be shown to have arisen. There were, however, a small number of countries that did not maintain stable par values. Some member countries in Latin America experienced rapid rates of domestic inflation that necessitated gradual depreciation of their exchange rates in relation to their intervention currencies, and both Lebanon and Canada had extensive experience with floating exchange rates. Since 1971, and especially since 1973, several different types of exchange arrangements emerged, and they were formally legalized as possible choices in the Second Amendment of the Fund’s Articles of Agreement on April 1, 1978. Countries could adopt a link to an external standard by pegging to another currency, to the SDR, or to a self-selected basket of currencies. Those countries wishing for greater flexibility could choose “clean” floating, “managed” floating with various degrees of intervention, or a rule for officially controlling the movement in the exchange rate according to objective indicators. Variants of those schemes have also appeared—such as pegging to an undisclosed basket of currencies, but with the authorities’ reserving the right to make occasional shifts in the rate or to operate the scheme within fairly wide margins.

With the emergence of a variety of exchange rate regimes, increasing attention has been given to the rationale for choosing one type of regime over another. Because the major industrial countries have adopted a system of floating either singly or jointly, several studies have focused on the problem that developing countries have in choosing their regimes. It is the purpose of this paper to survey these studies and to suggest directions in which future research might most usefully be pursued.

I. Overview of Principal Issues

The selection of an exchange rate regime by a developing country in a world of floating among the major convertible currencies is typically viewed as involving a sequence of choices or considerations. In the first instance, the question addressed is whether a developing country can and should let its currency float independently. Floating in this context is in general considered as a “clean” or at least moderately clean float; that is, the exchange rate is primarily the outcome of a market-clearing process for foreign exchange among domestic and foreign residents. With the emphasis on the role of the market in the determination of the exchange rate, the question then posed by a number of authors is whether conditions exist in developing countries that would allow an independent float. Black (1976) and Branson and Katseli- Papaefstratiou (1981; hereafter cited as B-K), for example, have argued that, whereas the potential benefits of a floating exchange rate are not necessarily limited to industrialized countries, certain characteristics found in many developing countries rule out floating as a feasible or realistic option. Other authors proceed directly to consider whether a fixed or flexible exchange rate regime is preferable in light of such characteristics as the susceptibility of the economy to internal and external disturbances, the “openness” of the economy, and domestic attitudes toward inflation. Much of this discussion of feasibility and optimality springs directly from the extensive literature on the choice between fixed and flexible exchange rate systems for maintaining internal and external balance, and from the classic studies on optimum currency areas and monetary integration by Mundell (1961a), McKinnon (1963), Kenen (1969), and Corden (1972). It is perhaps significant that flexible exchange rates are considered as synonymous with floating rates in most of this literature, and that the studies thus are oriented toward countries with financial markets that are well developed internally and integrated internationally. Flexibility, however, does not necessarily imply that the exchange rate is determined by the interaction of market forces without official intervention (that is, a “free” or “clean” float). The authorities may control or directly determine the exchange rate and have to decide on the form and substance of their policies for exchange rate management. It is primarily from this standpoint that this part of the literature is relevant to the choice of exchange rate regime by developing countries.

The next stage in the choice of regime proceeds if and when a determination is made, on the grounds of feasibility or optimality, that a floating exchange arrangement is ruled out. The monetary authorities of the developing country must then decide on the link between the national currency and some external standard, and the problem to be analyzed is whether a single-currency peg or some form of basket peg should be chosen. One approach is to choose a peg so as to reduce or offset as far as possible the effects on or cost to the economy of the developing country of exchange rate variability among foreign currencies, whereas the level of the exchange rate may be chosen or changed in light of other targets or policy objectives. Another approach is to consider that the choice of peg involves broader policy considerations than those related to short-term currency fluctuations—in particular, that the choice of peg can be used as an instrument to help maintain an appropriate level of the exchange rate, thus obviating or reducing the need for the monetary authorities to make discretionary changes in the peg.

The sequential approach generally adopted in the literature on the choice of exchange rate regime provides a convenient way in which to organize a more detailed consideration of the principal contributions and issues. The following section discusses the arguments on the feasibility of floating and the state of foreign exchange and financial markets in developing countries. Section III examines how some of the ideas and criteria advanced in the literature on optimum currency areas have been considered applicable to the problem of how a developing country should choose its exchange rate regime. Particular attention is paid to the issue of monetary independence and to the fact that countries with significant differences in inflation rates cannot maintain fixed exchange rates but must instead adopt some form of flexible exchange rate regime. Section IV considers additional arguments concerning exchange rate flexibility and the link to policies for exchange rate management. Section V concerns the choice of peg, and empirical studies on the choice of regime are the subject of Section VI. The paper concludes with some suggestions on the most promising areas for future research.

II. Floating and Domestic Exchange and Financial Markets

One important factor stressed as preventing or militating against a policy of floating for a developing country is the inadequate development of domestic financial markets and their lack of integration with world markets. Black (1976) and B-K (1981) for example, have cited as a condition of feasibility for floating the existence of asset markets integrated into the international system. In B-K’s analysis, there are in fact two conditions: first, that domestic financial markets of some minimum depth exist; second, that domestic and foreign currency assets are substitutes in the private portfolios of wealth holders. B-K argue that if these conditions are satisfied, then in the short run the exchange rate will be determined by equilibrium conditions in financial markets, and the stability of the exchange rate will depend on the overall stability of these markets. Countries with integrated financial asset markets can expect a floating rate to be stable in the short run. This is an application of the asset-market approach to exchange rate determination, in which the exchange rate in the short run is explained primarily in terms of the relative demands for and supplies of domestic and foreign financial assets.

With respect to the flow market for foreign exchange, the asset-market approach emphasizes the importance of capital account transactions that are the result of asset holders’ adjusting, or attempting to adjust, their portfolios in response to factors affecting the desired stocks of domestic and foreign currency assets. B-K argue that if asset markets in a developing country are not integrated with international markets, then the exchange rate will be determined by current account flows; that is, by demand and supply of foreign exchange emanating from the goods market. The short-run stability of the foreign exchange market will, therefore, depend on the satisfaction of the Marshall-Lerner conditions on trade elasticities. The feasibility problem, as put forward by B-K, is that countries having any degree of market power would be unlikely on theoretical grounds to satisfy these conditions in the near term. They also point to empirical evidence from trade models that suggests the absence or low values of contemporaneous price terms. On more general grounds, it is argued that prices for traded goods may change gradually and that lags operate in the adjustment of trade and service account flows to relative price changes. In the period immediately following an exchange rate change, the terms of trade may move against a country whose currency has depreciated and may offset any short-run effect on trade volumes of the exchange rate change; that is, the so-called J-curve effect may be evident.

The issue of exchange market stability and speculative capital flows in the presence of lags in price and quantity adjustment has been examined more closely by Driskill and McCafferty (1980).¹ The assumption made in their model is that the volume of trade in any given market period depends on prior expectations held about the exchange rate, and that lags therefore occur in the adjustment of trade volumes to their desired levels. An unanticipated change in the exchange rate that arises from a random disturbance in the trade balance may have no effect on current trade volumes and a possibly perverse influence on the trade balance, owing to price effects (that is, through the terms of trade). Speculators are assumed to have a stock demand for net foreign assets that is sensitive to expected capital gains, and thus desired changes in the stock arise from changes in the expected appreciation or depreciation of the exchange rate. Walrasian or short-run instability of the foreign exchange market may arise when there is insufficient speculative demand. Using a rational expectations framework, Driskill and McCafferty show that stability is more likely when the degree of risk aversion by speculators is low relative to the variance of the stochastic disturbance and when the response of trade volumes to anticipated exchange rate changes is high. The conclusion reached is that some minimum level of responsiveness of short-term capital flows to expected relative asset yields is needed for Walrasian and dynamic stability in a foreign exchange market otherwise rendered unstable by J-curve effects.² This model can be used to illustrate the central point made by B-K: if financial market separation prevents the possibility of speculative capital flows, then in the short run the floating rate will be unstable. The central bank will have to intervene and “make the market” for foreign exchange, thereby eliminating free floating as a feasible regime.

A potential weakness of the feasibility argument considered by B-K is that developing countries that are strictly price takers for their imports and exports could satisfy the Marshall-Lerner conditions.³ B-K suggest, other things being equal, that these countries could float their currencies even without integrated asset markets. A network of foreign exchange traders or dealers could conceivably hold transactions or working balances in foreign exchange,⁴ perform the important function of clearing international payments for goods and services, and effectively make a market for foreign exchange that would display intraperiod and dynamic stability. In response to disturbances over time, the exchange rate would move so as to keep the trade balance close to zero, with only transitory changes in the private sector’s (that is, foreign exchange dealers’) holdings of transactions balances. The problem for such countries, B-K argue, is that floating may not be feasible on other grounds—in particular, that these economies are typically highly “open” in the structure of their trade and production. A floating exchange rate would be volatile and would undermine the domestic currency—an argument from the literature on optimum currency areas that will be considered further in Section III.

Another important limitation to the main thrust of B-K’s argument as it affects policy choice is the implicit conclusion that a fairly clear-cut distinction can be drawn between countries whose financial markets are internationally integrated and those whose financial markets are not. If on the one hand it is determined that most developing countries fall into the second category, then it is fairly clear by this criterion that the policy choice is what to peg to and how to adjust the peg. If on the other hand attention is focused on the degree of financial integration, then the policy choice in many cases becomes less of a dichotomy, and more one of the appropriate amount and kind of exchange rate management.

Black (1976) has examined more extensively the characteristics of developing countries’ exchange and financial markets in comparison with those present in industrialized countries. The features of an advanced financial system emphasized by Black are the presence of institutions that efficiently and competitively intermediate the demands and supplies of various financial assets on behalf of the residents of the countries concerned. The intermediaries and other agents deal in short-term government or private paper (or both), bank deposits, currency, bank loans, and spot and forward foreign exchange. There is, in sum, a highly developed market for short-term financial capital, the efficiency and depth of which permit rapid portfolio adjustment and arbitrage in financial assets so that a portfolio-balance model again provides an appropriate framework for analyzing the short-run determination of interest and exchange rates.

The exchange and financial markets of developing countries, Black argues, typically do not have similar breadth and depth or such a range of institutional development. Forward exchange facilities are often absent, and markets for common stocks, securities, and bills are often poorly developed, thin, or nonexistent. The predominant source of financial intermediation is the banking system, which offers a small range of financial instruments and whose size and structure is limited relative to that found in developed financial systems.

In some developing countries there may indeed be insufficient participation to ensure the effective functioning of competitive markets for foreign exchange and for domestic financial assets and liabilities. Import and export transactions may be concentrated in relatively few hands, the bulk of foreign capital flows are often public, and economies of scale may limit the number of domestic financial intermediaries that can also act as dealers in the foreign exchange market. This type of situation is often found in small island states and in certain African countries, where the market structure of financial intermediation tends to be concentrated and would not provide for effective competition in determining interest rates or the exchange rate. Under such circumstances, allowing the exchange rate to be determined by market forces is not a realistic option, and it is appropriate that the authorities directly manage the exchange rate by setting the price (often including margins for intermediaries) at which foreign transactions are to take place and around which policies of intervention can be organized.

Relatively low levels of financial and associated institutional development are, however, often found in conjunction with restrictions on current and capital account transactions and controls on the yields from available financial instruments. Black argues that the successful introduction of floating rates would require in many developing countries a substantial commitment of real resources to the development of adequate exchange and financial markets, as well as a willingness by the authorities to eschew exchange and payments restrictions and devices used to intervene in domestic financial markets. The development, for example, of an active forward market in foreign exchange depends heavily on allowing free movement of short-term capital (McKinnon (1979)). As Crockett and Nsouli (1976) have also pointed out, there is likely to be an important relation between policies adopted by a particular country and the ability and willingness of the non-government sector to commit resources for the development of well-functioning foreign exchange and financial markets. Lack of institutional structure and efficient intermediation may be less the result of particular characteristics (for example, a low level of overall economic development, small scale of markets, high transaction and information costs) than of government controls and restrictions.

An attempt, therefore, can be made to distinguish between economic factors that may favor a particular type of exchange arrangement and institutional features that may be the results of the policies followed by the authorities. It is sufficient to remark here that, in several developing countries, attempts to control directly prices and quantities in the economy are prevalent and are particularly manifest in foreign exchange and financial transactions. Freedom to make and receive current payments is often curtailed by import quotas and restrictions on access to foreign exchange; inward and outward capital movements are controlled; and governments attempt to centralize foreign exchange transactions in the central bank or use the commercial banks as heavily regulated agents to enforce the exchange control and payments regime. The monetary authorities determine the exchange rate or rates at which various transactions are officially to take place and at which foreign exchange will be made available to transactors deemed eligible. In the domestic financial markets, deposit and loan rates of banks are often fixed and subject to ceilings that are low relative to the inflation rate and are combined with selective credit controls. Government budget deficits are frequently financed through direct borrowing from the central bank, by the sale of bills and securities at pegged interest rates to captive intermediaries and institutions, or by a combination of these means. The extent to which these features are found in developing countries clearly varies widely, but in such a policy environment it is not surprising that the development of organized foreign exchange and financial markets is retarded and that such markets remain rudimentary in comparison with those found in industrialized countries.

It cannot be inferred that in any particular developing country well-organized markets for foreign exchange and various financial instruments would emerge rapidly if only governments would lift restrictions and controls. The point to be made is that government policies often can influence over time the development of foreign exchange and other financial markets and the extent to which such markets are separated from international money and capital markets because of barriers to international trade in financial assets. Whether, therefore, underdevelopment of exchange markets and separation of financial markets per se are legitimate arguments against floating would seem to depend not only on the existing state of such markets but also on their potential to become broader, more developed, and better integrated within a favorable policy environment.

McKinnon (1979), for example, considers that, for a competitive and unified foreign exchange market to emerge, substantial freedom must be afforded nonbank residents to make and receive payments on current account, and the financial institutions must be in a position to intermediate these demands and supplies efficiently. In some developing countries, particularly in Latin America and Asia, restrictions and controls on current transactions have been much reduced in scope or effectively eliminated. Financial reforms have been undertaken that have stimulated competition between, and development of, financial intermediaries. At the same time, domestic financial markets have been opened up to international influences through the partial relaxation of controls and other restrictions on trade in financial assets. These types of policy changes can be expected to yield significant benefits in the form of improved resource allocation and are, in most instances, complementary.⁵ Elimination of quantitative controls and exchange restrictions on current account transactions (including associated trade financing), for example, will usually require for the efficiency of the payments-clearing process that greater freedom be given to the financial intermediaries to engage in domestic interbank foreign exchange transactions, to hold foreign exchange assets, and to negotiate lines of credit and other such facilities with foreign banks. Moreover, if financial institutions are to be permitted increased freedom to undertake such activity and to broaden the scope of both domestic and international intermediation, liberalization of the domestic interest rate structure and regulatory system is required.

With interest rate reforms and current account liberalization, nonbank residents may also gain more direct access to international short-term capital markets through open account financing and other trade-related instruments. In most cases, some degree of market segmentation is likely to continue. Some parts of the domestic financial markets may remain controlled or subject to different reserve or liquidity requirements, and the link between foreign and domestic financial markets may be circumscribed by the regulations and restrictions relating to the acquisition of foreign assets and the incurring of foreign debt.⁶ Nevertheless, there is likely to be sufficient freedom in the trade and payments regime and sufficient depth of participation and institutional development that the monetary authorities no longer perform the key intermediary role in the payments-clearing process. Floating may well constitute, therefore, a feasible exchange rate regime for some developing countries; several developing countries with relatively advanced financial systems have indeed experienced periods in which the exchange rate has floated.

One further point that links the issue of controls and the feasibility of floating should be considered. Some developing countries have found that if the trading and domestic financial system remains repressed, continuing incentives are provided to residents to evade controls through illegal transactions at official exchange rates, through the operations of an unofficial parallel market for foreign exchange, or through both. By such means, domestic residents can often acquire outside the official financial system significant stocks of foreign exchange as substitutes for domestic currency assets. In parallel or black markets, the exchange rate of course is floating, and the existence of such markets indicates that some form of official floating rate may well be possible, even if restricted to certain transactions (as in a dual market system). Such a step would help draw foreign exchange resources held outside the official financial system into the open and pave the way for the development of a unified market for foreign exchange.

III. Choice of Exchange Rate Regime and Optimum Currency Areas

Systematic attempts to define the characteristics of areas for which it is optimal to have a single currency regime began with the classic article by Mundell (1961a), who suggested that the extent of factor mobility be the principal criterion determining the domain of the optimum currency area. McKinnon (1963) advanced the criterion of the openness of the economy, whereas Kenen (1969) argued that product diversification in trade should be considered a major determinant of whether a country should opt to form an independent currency area or not. Other contributions to the literature have suggested additional criteria, such as the similarity of inflation rates and the degree of policy coordination among potential members of a currency area. Surveys of the literature on optimum currency areas by Ishiyama (1975) and Tower and Willet (1976) present the various issues and considerations that have been advanced to determine whether a particular country should join with others to form a common currency area.

Optimum currency area theory sets out to address the question of which areas or countries should adopt genuinely fixed exchange rates among themselves, allowing variability of their exchange rate in relation to other currency blocs. When exchange rates between major currencies are floating, a peg by a developing country to one of the foreign currencies gives rise to what can be considered a currency area; the literature on optimum currency areas would appear, therefore, to be of immediate relevance to the choice of regime. By and large, however, the theoretical

analysis surveyed by Ishiyama and by Tower and Willet confines itself to the major factors determining whether it is desirable for two countries or areas to join together in a currency area. The focus is almost exclusively on relations within the area, with stability (including exchange rate stability) in the rest of the world assumed or with the interaction between members of the potential currency area and other currency blocs left outside the scope of the analysis.

Heller (1978) and Bird (1979), for example, have invoked optimum currency area considerations as relevant to the decision of a developing country to adopt a fixed link between the domestic currency and an external standard, with the link in the form of either a single-currency peg or a peg to a basket of currencies. For optimum currency areas, however, it clearly makes a difference what form the peg takes. Whereas a peg to a single foreign currency gives rise to a form of currency area in which the exchange rates of the currencies concerned float jointly against other foreign currencies, the adoption of a basket peg in general will not have this effect. With a basket peg, the value of the domestic currency floats against the major currencies, and, rather than reflecting acceptance of optimum currency area arguments in favor of a link to a single currency, choice of a basket peg may be interpreted as a rejection of such arguments (Ishiyama (1975)).

Perhaps the most frequent reference to the literature on optimum currency areas in discussions of exchange regimes for developing countries is the argument that such countries, characterized by small open economies, must or should peg in order to secure the monetary value of their currencies. Mundell (1961a), for example, has stressed the benefits that formation of currency areas has for the usefulness of money. The essence of the case for a full currency union is the extension, to economic relations between two countries, of the real resource savings yielded by the use of a single currency as the medium of exchange, unit of account, and store of value. McKinnon (1963) has argued that, if a small open economy issues its own currency and allows the currency to float against that of its larger trading partner, the likely amplitude of fluctuations in the exchange rate would tend to undermine the domestic currency in performing its monetary functions and would encourage agents in the economy to substitute foreign currency for the domestic currency. In such an economy, which is highly open in terms of the ratio of tradables to nontradables in its output, the variability of a floating exchange rate would cause corresponding fluctuations in the domestic currency price of tradables. With monetary policy perhaps capable only of affecting the price of nontradable output, the result would be considerable variability in both the price level and relative prices. This volatility would not only make the real rate of return on domestic currency holdings uncertain compared with that of the foreign currency but would also make the latter preferable as the numeraire in domestic accounts and contractual obligations.

Through a broad appeal to the arguments of openness and currency substitution, B-K (1981) and Connolly (1982) consider that floating may not be a feasible exchange rate regime for many developing countries. A floating rate would undermine the local currency, and domestic residents would want contracts effectively denominated in a foreign currency; thus, apart from legal requirements, there would be no basis for the demand for local currency. As openness increased, the more likely would a floating rate lead to the erosion of the demand for local currency.

In assessing the importance of these arguments, there are several considerations to take into account. The original argument of McKinnon (1963) largely abstracts from differences in inflation rates and concentrates on the instability of the domestic price of traded goods caused by exchange rate fluctuations. Corden (1972) has noted that, to sustain McKinnon’s argument, it is necessary to assume that the foreign price index for tradable goods is stable and that a floating rate would display a high degree of variability even if other currencies in the system were supplied in a stable fashion. Such variability may, however, be associated with other characteristics found in conjunction with openness in developing countries, such as the dependence on particular primary product exports that are subject to price cycles or the incidence of real domestic supply shocks. Nevertheless, the argument for a peg is that the domestic price level and relative prices will be more stable than they would be under a floating exchange rate. The variance in the real rate of return on domestic currency holdings would be similar to that on foreign currencies under a peg, and the stability of relative prices and the price level would secure the continued use of the domestic currency.

On a theoretical level, however, there are unresolved issues about the conditions under which direct displacement of one national currency by another is likely to occur (Girton and Roper (1981)). Most models of currency substitution, for example, do not explicity treat the service benefits of the national currency in internal transactions, and thus it is difficult to explain satisfactorily why the domestic currency remains in general the accepted medium of exchange and unit of account within a country’s borders (despite often significant variability in the real rate of return on domestic currency compared with that on some foreign currencies). Nevertheless, it is not uncommon to find that residents in some developing countries do hold significant stocks of foreign currency, that the prices of some goods are quoted in foreign currency, and that foreign currencies are also used in internal transactions. The problem appears most serious in developing countries where inflation rates are high and variable, the exchange rate changes frequently, and there is uncertainty about future government policy and exchange rate movements. A compounding factor comes into play when nominal yields on less liquid assets are repressed, in which case foreign currency and other assets denominated in foreign currency can substitute for interest-bearing domestic stores of value as well as for transaction balances (Tanzi and Blejer (1981)).

Although there may be reservations about whether the arguments of openness and currency substitution will make floating an infeasible exchange rate regime, the considerations just cited do raise important questions about the difficulties that authorities in a developing country may face in managing domestic currency and monetary policies under a flexible exchange rate. The issue is related to the constraints on macroeconomic policy, cited in the literature on optimum currency areas, that are implicit in the choice between floating and pegging. The maintenance of an unchanged peg over time between the currency of a developing country and a foreign currency or basket of currencies implies that domestic authorities import on a secular basis the monetary policy or policies of the standard to which the domestic currency is pegged. In addition, in the absence of recourse to restrictions, control of the domestic money supply may be largely subordinated to the balance of payments constraint.

In the literature on fixed versus flexible exchange rates and related issues that antedates the Bretton Woods breakdown, it was argued that the chief long-run benefit of a floating rate was the ability to achieve monetary independence; that is, the ability to choose the inflation rate independently (see, for example, Johnson (1972)). The pursuit of a domestic monetary policy, for example, that yielded a lower secular rate of inflation than that prevailing in the rest of the world would allow the country in question to have a more efficient medium of exchange and unit of account, thereby achieving a better intertemporal allocation of resources. More important, however, is the argument that a country would have the freedom to choose an optimum point on its Phillips curve. It has been argued, however, that this freedom might be more limited the smaller and more open the economy is (Tower and Willet (1976)). This question is related to the more general issue of whether persistent money illusion is likely to exist, and thus whether it is possible in the long-run to lower the path of real wages by adopting a higher rate of inflation. The prevailing view now seems to be that if policymakers cannot be sure of the nature and extent of the long-run trade-offs, then they are not in a good position to exploit the putative benefits of monetary autonomy.

One reason that a developing country might still seek to inflate at a faster rate than that ruling in the rest of the world would be to increase the yield of the tax on money balances (that is, a different rate of seignorage than that chosen by trading partners or the rest of the world). Nevertheless, the literature on currency substitution suggests the potential difficulties of attempts to over-utilize this source of taxation (see Khan and Ramirez-Rojas (1984)) and of the welfare costs to the economy of high inflation rates, even if inflation is fully anticipated. Expectations about future inflation rates, however, are not likely to be held with certainty; this factor, in turn, may cause undesirable shifts in resource allocation and income distribution, may create difficulties in establishing indexing schemes and procedures, and may impose other costs on the economy. The potential resource costs are therefore significant, whereas the long-term benefits of a high inflation rate may be considered uncertain or illusory.

Finally, and most important, there are the not inconsiderable risks of inflation getting out of control, given the difficulties of bringing down wage and price increases once inflationary expectations have become embedded in the economy. For some developing countries the constraints on internal policies implicit in the maintenance of a fixed link to an external standard may not be unwelcome because they provide a framework within which to organize consistent monetary, fiscal, and demand-management policies. Thus it may well be easier to define policies with a view to keeping wages, prices, and the level of economic activity consistent with the fixed link to an external standard than to set up independent policy objectives. It can, however, be argued that if the average externally given inflation rate is significantly positive and variable, a domestic target for a lower and more stable inflation rate is likely to be regarded as a legitimate objective of policy.

The incidence of and attitude toward the benefits and costs of different inflation rates does vary among countries, not least of all in the developing countries. For those developing countries that wish to pursue or, in the case of certain countries (for example, in Latin America), that seem forced to accept inflation rates substantially different from the world norm, an exchange rate regime offering nominal exchange rate flexibility should and must be chosen. Although in much of the literature on fixed versus flexible rates it was implicitly assumed that the required degree of nominal flexibility would be achieved by letting the exchange rate float, some high-inflation developing countries have introduced various forms of exchange rate, interest rate, and wage indexation to cope with the external and internal consequences of high domestic inflation rates. Crawling pegs or gliding parities, which are forms of exchange rate regime qualitatively different from a fixed (or occasionally adjustable) peg or a float, could be used for inflation targeting or to offset the external consequences of a high domestic inflation rate. Discussion of such regimes forms a distinct subset in the literature on the choice of regime that is of particular relevance to developing countries. To this and other aspects of the flexibility of exchange rates, particularly those related to stabilization and adjustment policies, the discussion now turns.

IV. Exchange Rate Flexibility and Developing Countries

Although for many developing countries, there are generally well-recognized difficulties associated with considering floating exchange rates, floating is not the only form of exchange rate flexibility that can be envisaged. Two questions therefore arise: first, whether in response to various shocks and disturbances flexibility in the exchange rate (as opposed to a fixed link to an external standard) provides some degree of insulation, assists in economic stabilization, and facilitates adjustment; second, if it does have these beneficial effects, how the appropriate degree and timing of flexibility are to be achieved. In particular, the second question concerns the role that might be played by a crawling peg regime.

There is, of course, an extensive literature on the source of disturbances and insulation from them under fixed and flexible exchange rates, and on the factors affecting the process and ease of adjustment under different exchange rate regimes (a review of such issues is presented in Artus and Young (1979)). It has also been widely recognized that the size and nature of the shocks that an open economy faces may be important determinants of the optimal degree of exchange rate flexibility.⁷ Attention in recent years has, however, tended to focus on the developed, industrialized economies, extending the work of Mundell (1961b) and Fleming (1962) and using models characterized by well-developed and integrated financial capital markets. For many developing countries the appropriateness of this framework comes into question. The departure point for several contributions considered here concerning the issue of exchange rate flexibility is to abstract from arbitrage in financial assets and from net private capital flows.

A strong restatement of arguments in favor of exchange rate flexibility in response to certain disturbances has been made by Flanders and Helpman (1978; hereafter cited as F-H). They consider exchange rate policy for an economy that is a price taker for tradable goods but that also produces nontradable goods. Commodity arbitrage is assumed to be perfect, thereby permitting price and exchange rate aggregation among foreign countries, and the only financial asset explicitly treated in the model is the local money, which is not traded internationally. The country faces external disturbances, in the form of shifts or fluctuations in the world price of tradables, and internal relative demand shocks caused by a switch or fluctuation in preference between tradable and nontradable goods. With the exchange rate floating, the authorities determine the money supply, and the exchange rate moves perfectly to maintain external balance; under a fixed exchange rate, external balance is maintained by the authorities’ implementing a well-organized policy that equates the money supply to money demand at each point in time.⁸ The latter assumption implies that there is an adjustment to a real domestic supply shock (for example), either under a flexible exchange rate or a fixed exchange rate.⁹ If in the F-H model, wages and the price of nontradable goods are flexible; full employment is maintained; and there is nothing to choose between the two regimes in response to shocks—except that in the case of fluctuations in world prices a flexible exchange rate will yield greater price stability, thereby enhancing the “moneyness” of the domestic currency.¹⁰

The case for exchange rate flexibility rests on the familiar argument of the downward inflexibility in domestic wages and the price of nontradables. F-H show that, for a foreign price shock, flexibility in the exchange rate ensures full employment in their model. For a fall in the foreign currency price of tradables, the effect on the domestic price of tradables and relative prices (and hence price-cost relations in the two sectors) is offset by a change in the exchange rate. Exchange rate flexibility counters the rigidity in the domestic price of nontradables, thereby permitting the equilibrium relative price ratio and full employment to be maintained. Under a fixed exchange rate, however, a downward foreign price shock causes a fall in the domestic price of tradables and conflict between internal and external balance. If monetary policy is consistent with external balance, unemployment will result; full employment may be attainable only at the cost of a balance of payments deficit (that is not sustainable) or of implementing particular types of government expenditure and taxation policies. If domestic relative demand disturbances take place, the results depend on whether preferences shift in favor of or against nontradable goods. The general proposition advanced, however, is that domestic monetary policy can be adjusted in the model under a flexible exchange rate so as to attain full employment; a flexible exchange rate is always as good as, or better than, a fixed exchange rate in the presence of downward price and wage rigidities.¹¹

Although F-H concentrate on the effects of particular types of shocks on the levels of output and employment, Black (1976) has emphasized the stability of domestic relative prices and the structure of production and consumption. The economy remains on its production possibility frontier, although possibly with underemployment and inefficient use of resources. At the initial relative price ratio, the economy is in internal and external balance; that is, the markets for nontradables and tradables are both in equilibrium. Optimal exchange rate policy depends on the type and nature of the shocks that the developing country faces. The impact of a fall in the world price of tradables decreases the domestic price of tradables with a fixed exchange rate and leads to a departure of relative prices from the initial equilibrium ratio and to money market disequilibrium. These effects and the resultant process of internal and external adjustment can be avoided if, as foreign prices fall, the exchange rate changes to offset the fall. For transitory domestic supply shocks, Black considers that the optimal policy is to peg the exchange rate through reserve use, foreign official borrowing, or both, thus providing some insulation to the domestic economy. This degree of insulation would reduce the incidence of domestic relative price movements and resource shifts compared with those required if the exchange rate were flexible.¹²

Whereas F-H and Black are interested in the effects of certain shocks on domestic relative price stability and internal and external balance, Lipschitz (1978) has emphasized internal balance in terms of the stabilization of real absorption in a one-good model. For the economy in question, output is exogenously determined but is subject to domestic supply shocks, considered of particular importance for primary-product-producing developing countries; the economy also faces domestic money demand shocks. The authorities’ objective is to stabilize real absorption, a function of the level of real output and the excess supply of real money balances. There is no international trade in financial assets (except by the authorities), and, under a fixed exchange rate, changes in the foreign component of the monetary base reflect the output-absorption gap. Lipschitz finds that, for a domestic supply (output) shock, fixing the exchange rate is superior for stabilizing absorption, since with a flexible rate absorption will be constrained to output. The reason for this superiority is clear: with a fall in real output, absorption can only be prevented from falling similarly if real resources are forthcoming from abroad, and in the model this is done by running a balance of payments deficit financed by the monetary authorities’ dishoarding or borrowing at the fixed exchange rate.¹³ A money demand shock, however, will have no effect on absorption under a flexible exchange rate, but under a fixed rate it will cause a departure from internal and external balance unless the authorities have perfect foresight and can appropriately manipulate the domestic component of the monetary base. In a dynamic extension of the model—in which, for example, reserve movements in one period must subsequently be reversed—the prescriptions hold with respect to reducing the variance of absorption over time. The conclusion reached is that optimal exchange rate policy should depend on the type of transitory disturbances likely to be encountered by the particular developing country.

In principle, strong conclusions about exchange rate flexibility emerge from these types of simple abstract models, but considerable caution must be exercised in drawing similarly strong policy implications about the choice of exchange rate regime. In F-H, for example, the bases for comparison are the polar cases of a “pure” floating exchange rate and a permanently pegged rate. The objection to the latter is more that, with price and wage rigidities, the policies required to achieve external balance in the face of certain types of shocks may be too costly in domestic output and employment.¹⁴ The model also restricts itself to potential conflicts between internal and external balance under a fixed exchange rate if policies are directed toward the latter target. External balance (that is, the current account balance) is in general regarded as a medium-run rather than a short-run objective, so that the analytical framework does not consider the circumstances under which it may be optimal to depart from short-run external balance in response to certain shocks or cyclical disturbances through the use of reserves or official borrowing.

It is these latter considerations that are central to the conclusions reached by Lipschitz (1978). In his model optimal exchange rate policy, when shocks are readily identifiable, is to allow the exchange rate to correct for domestic money demand shocks but not for temporary supply disturbances. The effects of transitory domestic supply shocks on the economy should be cushioned at a pegged exchange rate by reserve use or official borrowing (or both) to finance the temporary current account imbalances. In addition, there may be the possibility, if sufficient resources are available, of reducing the effect of reserve movements on the monetary base through sterilized intervention, thus further cushioning the real impact of the shock.

Domestic supply shocks are of particular concern to some developing countries that produce exportable or importable primary agricultural products subject to variable harvest conditions, disease, and so forth. In addition, many primary product markets are subject to periodic price shocks and to cyclical movements in export prices caused in large measure by output fluctuations in the rest of the world. Because of a lack of diversification in the composition of exports and in production patterns, such disturbances significantly affect the amount of real absorption that given levels of real output can finance. Assistance to such economies to finance temporary balance of payments deficits and to help the authorities in their efforts to stabilize real expenditure relative to real output over time is, after all, the rationale for the Stabex scheme of the European Community, the Fund’s compensatory financing facility for export shortfalls and cereal import excesses, and other forms of compensatory financing extended to developing countries with limited access to international capital markets.

The conclusions relating to domestic monetary disturbances are subject to greater qualification in the Lipschitz model. All goods are tradable, and the focus of attention is only on real absorption; exchange rate flexibility costlessly eliminates disequilibrium in the money market and constrains absorption to output. There is no relative price structure to be altered, output is exogenously determined, and there is no particular importance attached to the price level or the exchange rate. The necessary changes in the price level required to eliminate monetary disequilibria are achieved immediately, and without any real effects, by way of the exchange rate. As pointed out by Lipschitz, there is no room in his simple model to analyze the interaction between various types of shocks, relative prices, and exchange rates. This qualification, he suggests, might weaken the case put forward for exchange rate flexibility in response to domestic monetary disturbances. In Black’s model (1976), in contrast, monetary or price disturbances at home and abroad manifest themselves in domestic relative price changes at a fixed exchange rate, and it is the undesirability of these relative price changes in terms of internal and external balance that recommends exchange rate flexibility.

In general the conclusion that has been drawn from these and other simple theoretical models is that the optimal exchange rate regime in the face of various types of disturbances may be one intermediate between fixed and fully flexible exchange rates (see, for example, Frenkel and Aizenman (1981)). The policy problem is to determine the appropriate degree and type of exchange rate management consistent with the amount and kind of information available and with the financial and administrative abilities of the authorities in the developing country concerned. It is important in this respect to consider the institutional features and development of the foreign exchange market and that these elements also influence demands made on the skills of the country’s monetary authorities.

In developing countries where foreign exchange markets are not well developed, the monetary authorities set and announce the exchange rate at which transactions are to take place and at which they will buy and sell foreign exchange to the financial intermediaries. The monetary authorities directly determine the value of the domestic currency in relation to the intervention currency by adopting a price-setting rule or convention. The issue of flexibility for policymakers in many developing countries concerns the content and form of this price-setting rule. With an adjustable peg, the price-setting rule applied is relatively straightforward: the exchange rate of the domestic currency is kept constant against a single foreign currency or average of foreign currencies. Departures from the rule (that is, a change in the level of the rate against the chosen standard) may be made occasionally, in a discrete fashion or in steps, in response to signals or cumulative evidence that adjustment in the underlying current account position is required. An alternative is to include in the price-setting rule more continuous reference to some set of variables or indicators, as in, for example, an exchange rate regime using a form of crawling peg or gliding parity.

As Williamson (1981) has indicated, most of the original proposals for the operation of a crawling peg concern the use of such an exchange rate regime to facilitate payments adjustment among the industrialized countries. In practice, however, such regimes have primarily been used by a relatively small number of developing countries with atypically high inflation rates to neutralize the effects of domestic inflation on their external accounts. Although there have seldom been explicitly stated rules for the crawling pegs, the most accurate description is that the authorities in these countries have principally managed the domestic price of foreign exchange on the basis of a calculation of relative purchasing power parity. Both foreign and domestic price performance are taken into account in making frequent and small adjustments in the exchange rate.¹⁵

The management of exchange rate policy, as noted earlier in this paper, should also be related in part to the inflation objective or “target” of the authorities. If this target is in excess of inflation rates prevailing elsewhere, the exchange rate should be depreciated to offset the inflation differential. For countries with such preferences or propensities for inflation, large devaluations or jumps in the exchange rate at infrequent intervals (as in an adjustable peg arrangement) do not prevent the exchange rate from moving out of alignment in the interim (with attendant repercussions for the competitiveness and performance of the traded-goods sectors), may add to the uncertainty surrounding prospective economic developments, and in general lead to highly disruptive speculative fluctuations in the supply of and demand for foreign currency. Even in countries where there are strict exchange controls, such speculative disturbances can be sizable and are likely to increase in importance as the domestic financial system becomes more developed.¹⁶ In contrast, a crawling peg arrangement, by adjusting the exchange rate frequently and in small steps, can avoid many of the disadvantages inherent in large and abrupt devaluations.

The adoption of some form of crawling peg need not necessarily be restricted to those developing countries where economic policies yield inflation rates that are exceptionally high by international standards. Developing countries following conservative monetary policies could affect a gradual appreciation of the domestic currency through a crawling peg that is based on an assessment of underlying price and cost trends. In any event, the authorities would be applying some form of rule or convention for relative purchasing power parity in the management of exchange rate policy, thereby avoiding trend movements away from an initial value of some real exchange rate indicator. As in the case of countries with high inflation rates, however, it is the differences in longer-term monetary policies and inflationary trends that argue for a crawling peg regime, rather than short-term considerations of macroeconomic stabilization that arise from the incidence of various random shocks or disturbances.

The extent to which it may be desirable for more developing countries to adopt some form of convention or rule for purchasing power parity remains an open question. Williamson (1982) has strongly recommended an overall exchange rate policy that is dedicated to preserving the constancy of the real exchange rate, except when there is a perceived need to change the real exchange rate to promote external payments adjustment. He argues that the most efficient way to preserve a constant real exchange rate is to use a crawling peg based on a relative inflation rate indicator. An inflation rate differential also forms part of the policy rule for exchange rate management articulated by Black (1976). The rule, which Black refers to as “government-managed floating,” implicitly focuses on the more or less continuous stabilization of some real effective exchange rate, or REER, index (for more detail on use of this index, see the next section). A more explicit theoretical treatment of the rule with respect to REER indices is presented in B-K (1982).¹⁷

This policy rule has given rise to some controversy in the literature on exchange arrangements and “optimal” currency baskets. The rule is sometimes referred to, somewhat anomalously, as “pegging” the REER index or as “pegging” to a “real” currency basket. Lipschitz and Sundararajan (1980; hereafter cited as L-S) have argued that, whereas under certain conditions stabilization of some REER index can be considered an appropriate policy objective, a policy rule of the Black-B-K type, which in the limit implies continuous stabilization of a real exchange rate index, cannot be implemented in practice because the pertinent price data are only available discretely and usually with a considerable time lag. Whether the absence of contemporaneous price data is an insurmountable obstacle to the implementation of this type of policy rule has been questioned by Williamson (1982). His conclusion is that the high degree of serial correlation in inflation rates can be exploited, together with other available information, to make reasonable estimates of current inflation rates, and hence of price performance at home compared with that in competitor or partner countries.¹⁸ The implementation of a reasonable approximation to this kind of policy rule would thus be possible.¹⁹

A related issue is whether more developing countries should attempt to implement a policy rule of this type using the kind of price indices that usually are the most readily available. Consumer price indices, for example, suffer from several limitations in determining changes in competitiveness in the traded-goods sectors. Over time, differential productivity growth in traded- and nontraded-goods sectors is often a principal source of trend movements in the consumer price index; in the short-run, the indices are often affected by temporary factors and by feedback effects from exchange rate changes. In many instances, consumer price and wholesale price indices suffer from coverage limitations and from the presence of price controls and subsidies that make evaluation of the underlying rate of inflation and cost trends more difficult. A more general problem is that the assessment of price and cost trends in many developing countries is likely to be surrounded by a measure of uncertainty and to be made in circumstances that may not lend themselves to an unambiguous or precise estimate. For these reasons, the practical wisdom of uncritically adopting a policy rule based on a simple objective indicator, such as relative consumer price or wholesale price index performance, is questionable. A more prudent approach recognizes that important elements of judgment are required of the monetary authorities in using available price and cost data to operate this kind of policy for managing the exchange rate.

Apart from these important practical and technical issues, there is general agreement on a theoretical level that an exchange rate management policy oriented toward the stabilization of some real exchange rate index will be insufficient to maintain external balance. Changes in economic conditions or circumstances that a developing country confronts, such as a terms of trade shock that is not expected to be transitory, may call for a change in exchange rate and other policies to affect an alteration in the real exchange rate for external balance adjustment. A group of possible external balance indicators that could be used in exchange rate management were examined by Kenen (1975), and theoretical aspects of Kenen’s analysis have been followed up in an optimal control framework by Branson and Macedo (1980). Williamson (1981) has discussed a proposal along similar lines for managing the real exchange rate to cure or prevent the emergence of underlying current account disequilibrium.

The area of exchange rate management and the use of indicators continues to be an important one for further theoretical and applied research of particular importance to developing countries. Reflecting some of the considerations outlined, there has been a movement toward the adoption of more active exchange rate management in developing countries (see International Monetary Fund (1982, 1983)). At the same time, there has been growth in the number of countries whose exchange arrangements are based on a link between the domestic currency and a basket of foreign currencies. Where the precise composition and margins of the basket are not publicly announced, there can be elements of flexibility or management in these types of “pegged” exchange arrangements (Aghevli (1981)).

V. The Choice of Peg

The previous section considered arguments that have been or can be invoked in favor of a flexible exchange arrangement for a developing country. Notwithstanding those arguments, many developing countries opt to determine the value of their currencies by adopting a price-setting rule of the adjustable peg type. The substantive issue is what objective(s) are to be realized through the choice of peg and, correspondingly, what problems and circumstances should be viewed as relevant for the specification of these objective(s).

A characteristic of the period since the demise of the Bretton Woods system has been the variability of exchange rates between major currencies. As well as showing considerable short-run volatility (daily, weekly, monthly), the bilateral exchange rates between at least some of the major currencies have also exhibited fluctuations or swings over longer periods—variations that frequently have not demonstrated a close relation with corresponding relative price and cost performance. There have been, therefore, significant departures in both the short and medium term from indicators of relative purchasing power parity among the currencies of major industrialized countries. This behavior cannot be adequately explained in terms of changes in “underlying” conditions or “fundamental” determinants, such as the apparent need for current account adjustment and emerging differences in the relative rates of expansion of national money supplies.

Concern about the repercussions of these kinds of exchange rate fluctuation has been central to much of the discussion on how a developing country should choose its peg. The choice of peg is thus seen as the appropriate policy instrument to minimize or to avoid at least some of the harmful consequences of exchange rate variability among foreign currencies (see the literature survey in Williamson (1982)). In several studies, the undesirable consequences have been seen as the induced instability or variability of particular target variables, so that the choice of peg concerns insulating (or stabilizing) these variables to the extent possible from the effects of exchange rate fluctuations among foreign currencies. A common element is to utilize an effective exchange rate (EER) index to consider the effects of exchange rate movements between the domestic currency and foreign currencies (Black (1976), F-H (1979)). Although originally an outgrowth of exercises designed to isolate the effect of a vector of exchange rate changes on a country’s trade balance over a medium-term period, the index can in principle focus on any variable affected over various time horizons by exchange rates. Ideally, such an index attaches weights to a country’s bilateral exchange rates in relation to foreign currencies to provide in summary and convenient form an indicator of the net effect of exchange rate changes on a selected variable in the chosen time period. When exchange rate movements are considered in isolation, the measurement is often referred to as a nominal EER index; if allowance is made for different rates of price or cost inflation at home and abroad, the index is usually referred to as a real EER (or REER) index. In principle the solution to choosing a peg with the objective of stabilizing or insulating a variable in the event of exchange rate fluctuations between foreign currencies appears straightforward: the country should peg to a basket of currencies, whereby the weights used in constructing the basket are the same as those used in the EER index.²⁰

The weights used in calculating an EER index relevant to one particular variable or time period may well be different from those appropriate for another; hence, as B-K (1981) have pointed out, choosing a peg following this type of methodology may involve potential trade-offs. In practice there are likely to be considerable difficulties to be faced in determining weights for particular variables and in evaluating the trade-offs. Nevertheless, that several variables have been suggested in the literature is indicative of different views about what the consequences of currency fluctuations are that should or can be minimized by pegging to a currency basket. It is useful, therefore, to review the objectives that have been considered to guide the choice of peg.

VI. Empirical Studies

Because over a decade has passed since the breakdown of the Bretton Woods system and because significant changes have been monitored in exchange arrangements adopted by member countries of the Fund, it is not surpising that efforts have been made in the interim to examine the factors or considerations that have actually influenced the choice of exchange arrangement. To try to answer this question, several attempts have been made to see whether the kind of advice proffered has been taken into account by decision makers in choosing their exchange arrangements or regimes.

Heller (1978), Dreyer (1978), and Holden and Holden (1976) have used a variety of statistical techniques with cross-section data to test hypotheses that the choice of regime has been determined by the considerations advanced in theoretical analyses. The explanatory variables used in general have been based on ideas drawn from the literature on optimum currency areas, such as indicators of relative openness and country size. An important problem faced in these studies is the difficulty in giving numerical values to different exchange regimes; as a result, the statistical techniques used must rely on assigning countries in the sample to one of a small number of groupings. The groupings used follow the classification published in the Fund’s Annual Report (International Monetary Fund (1982)), which until 1982 was based on members’ declared exchange arrangements. The classification was subject to a number of shortcomings. For example, some countries, declared as being pegged to the SDR, in practice kept the value of their currency closely linked to the U.S. dollar, whereas other countries not declared as peggers included countries with market-determined floating rates as well as countries with nonconvertible currencies that had adopted forms of direct exchange rate management.²³

Despite these problems, attempts were made to test the association between types of exchange regimes and economic characteristics found in the sample group of countries. Heller found that the choice between peg or float is largely explained by the openness of the economy (the ratio of imports to gross national product, GNP), size (GNP), and a measure of geographical trade concentration. For peggers, trade concentration was found to be the principal factor affecting the choice between pegging to the U.S. dollar or the French franc and pegging to a currency basket, although the ability of the function to predict the choice between a dollar peg and a basket peg was considerably weaker than functions concerning choices between the franc and a basket and between the dollar and the franc. Dreyer also found that pegging is associated with the small size and openness of the economy, but not necessarily with trade concentration. Holden and Holden, like Heller, used discriminant analysis and similar explanatory variables to predict choice between regimes, but they considered a more differentiated classification of regime, including under separate regimes those countries with cooperative arrangements and those officially classified as adjusting their exchange rates according to a set of indicators. The results of this study were not clear-cut, and like the other statistical analyses suggest either that there are weaknesses in the theoretical literature in specifying how structural criteria should be used in making the choice of regime or that the assumption that the actual choice is made on the basis of full economic information is incorrect.

An alternative to the wide-ranging statistical approach is to examine the experience of individual countries or groups of countries. Aghevli (1981) examined the exchange regimes and exchange rate policies of eight Asian countries during the 1973-78 period. Some of the countries maintained a de facto peg to the U.S. dollar during most of that period, whereas countries that had pegged to the pound sterling before 1971 changed their arrangement and instead linked their currency to a composite of foreign currencies. Aghevli found evidence that all the countries were reluctant to make adjustments in their pegged exchange rates to take into account differences in inflation performance at home and abroad, policy decisions that tended to give rise to real effective appreciation in high-inflation countries and to real effective depreciation of low-inflation countries.

The choice of peg did, however, influence how the real and nominal effective exchange rates of the countries evolved over the period. Those pegged to the dollar all had inflation rates higher than those of their trading partners, and the trend depreciation of the dollar over the period helped to offset the inflation differential. Where the inflation differential was relatively small, the choice of a dollar peg helped to keep the real effective exchange rate relatively unchanged, but for countries with higher inflation the nominal effective depreciation associated with the dollar peg was not sufficiently large to avoid relatively large discrete adjustments in the value of these currencies in relation to the dollar. Real effective depreciation was recorded by those countries pegging to a basket of currencies because, although their nominal effective rates were stable, their inflation rates were lower than those of their trading partners or competitors. After the 1973-78 period covered by Aghevli’s study, the secular depreciation of the dollar began to be reversed so that, other things being equal, the maintenance of dollar pegs would have exacerbated the external balance difficulties of the group of higher-inflation countries—and thus would have reinforced Aghevli’s argument that exchange regimes affording more flexibility should be chosen by these countries. In the post sample period, at least one of the high-inflation countries (the Republic of Korea) did alter its exchange arrangement from a dollar peg to a government-managed flexible rate. Such analysis sheds further doubt on whether empirical work on the actual choice of regime based on cross-section data should be used in any prescriptive way by an individual country in determining its exchange regime.

VII. Concluding Remarks

This survey has reviewed what have been seen as the principal issues to be considered by a developing country in choosing its exchange rate regime. One concluding remark that can be made with surety is that a developing country’s choice of exchange regime cannot be discussed adequately in terms of a simple float-peg dichotomy. Independent floating is often considered infeasible for many developing countries because of the underdevelopment of domestic financial markets. Although inadequate development of such markets may be viewed as a structural characteristic from an analytical viewpoint, it may not or need not be so considered for purposes of policy. Moreover, ruling out some form of “clean” or “managed” floating does not remove the possibility of adopting other types of flexible exchange rate arrangement, particularly as a means of dealing with inflation differentials. In addition, despite earlier somewhat disappointing attempts to link the choice of regime to structural characteristics discussed in the literature on optimum currency areas, some of the more recent theoretical research has been attempting to pursue the links between macroeconomic stability, economic structure, and exchange rate management. Hence the optimal degree of exchange rate flexibility would seem to be an important area for further research. Such research can be expected to provide valuable insights into the extent to which the authorities of a developing country can or should modify simple rules or assignments for determining exchange rates in the present international monetary environment.