An Analysis of Exchange Market Intervention of Industrial and Developing Countries

There is a wide range of views among governments concerning the usefulness and limitations of exchange market intervention policy. The objective of this paper is to provide a framework that can help identify and, to a limited extent, evaluate the basis for this range of views.

Abstract

There is a wide range of views among governments concerning the usefulness and limitations of exchange market intervention policy. The objective of this paper is to provide a framework that can help identify and, to a limited extent, evaluate the basis for this range of views.

There is a wide range of views among governments concerning the usefulness and limitations of exchange market intervention policy. The objective of this paper is to provide a framework that can help identify and, to a limited extent, evaluate the basis for this range of views.

The first step in developing this framework is to propose a definition of exchange market intervention. It is argued in the next section that intervention policy can only be usefully defined and evaluated in the context of a model of exchange rate determination. Since no widely accepted model is yet available, this paper sets out a simple model designed to highlight the possible independent contribution of exchange market intervention to economic policy. The model suggests that intervention policy can be usefully defined as management of the currency composition of interest-bearing public debt.

Following this definition, exchange market intervention includes any government transaction, or set of transactions, that changes the relative supplies of official nonmonetary debt denominated in different currencies held by the private sector. Changes in the supply of domestic or foreign money held by the private sector, which are often associated with the mechanics of official transactions in exchange markets, are assumed to be offset by appropriate purchases or sales of nonmonetary debt. An advantage of this definition of intervention is that it permits a clear distinction to be drawn between monetary policy and exchange market intervention policy.

Throughout the analysis, it should be remembered that this paper is not questioning the widely held view that monetary policy is probably the single most important determinant of exchange rates. Nor does it address the important question of the role that the exchange rate or exchange market conditions might play in formulating monetary policy.1 What it does address is the fact that a government, inclusive of the monetary authority, can be thought of as making two decisions concerning its financial balance sheet. First, it must decide what part of its debt will take the form of base money; this will be referred to here as “monetary policy.” Second, the government must decide in what currency the balance of its nonmonetary debt will be denominated; we call this “intervention policy.” This framework is based on the observation that governments are not free to alter the currency denomination of their base money. No country allows other governments (or private counterfeiters) to issue monetary liabilities denominated in its currency, nor would a government fail to offset any substantial holdings of its base money by another government. Governments, however, do hold and issue interest-bearing debt denominated in foreign currencies.

Although the mechanics and forms of exchange market intervention transactions vary from country to country and over time, this simple framework for evaluating a government’s financial balance sheet provides a complete summary of its exchange market intervention policy.2 Note that our definitions of monetary policy and exchange market intervention policy do not depend upon the intent of the governments involved, the exchange rate regime (i.e., fixed or floating), or the particular mechanisms employed to alter the government’s balance sheet.

The independence of our definitions of policy tools from the intent of the government is particularly important. A more conventional definition of exchange market intervention might include transactions in which changes in the monetary base are generated by sales or purchases of foreign exchange reserves. This definition is often associated with models in which assets denominated in different currencies are assumed to be perfect substitutes. In this framework, changes in the monetary base that result from purchases or sales of domestic securities are identical to exchange market intervention in terms of their effects on interest rates and exchange rates. The only distinction between intervention policy and “domestic” open market operations is that the former is a response to exchange rate objectives while the latter might be directed toward other policy objectives. An important insight gained from this framework is that if securities denominated in different currencies are perfect substitutes, the government has only one policy tool at its disposal and cannot attain both an exchange rate objective and an independent objective for another endogenous variable, such as an interest rate or the monetary base.

A shortcoming of this framework is that it draws attention away from the fact that if securities denominated in different currencies are not perfect substitutes, changes in the mix of government securities denominated in different currencies will affect interest rates and exchange rates even if the monetary base is unchanged. Moreover, by focusing on the impact of exchange market intervention on the monetary base, analysts have concluded that exchange market intervention policies of developed countries have continued to play an important role in these countries’ economic policies.3 However, as shown in Section II, the framework developed in this paper reveals a clear change in the policies of several industrial countries after the demise of the fixed exchange rate system in early 1973. In fact, several major industrial countries have virtually suspended their use of exchange market intervention policy, as defined in this paper, since the introduction of managed floating exchange rates.

Although interest in management of the maturity of public debt (what might be called closed economy debt management policy) has faded in recent years, it is possible that the currency denomination of public debt is an important policy tool. A necessary, but not a sufficient, condition for the effectiveness of both closed and open economy debt management policy is that the financial assets involved be imperfect substitutes in the portfolios of private wealth holders. If the nonmonetary assets bought and sold by governments are perfect or nearly perfect substitutes, debt management policies will not measurably affect market prices and therefore cannot contribute to policy objectives. In the context of domestic debt management policy, empirical studies have largely failed to support the view that changes in the relative supplies of long-maturity and short-maturity nonmonetary debt influenced the spread between long and short interest rates.4 Given this result, there was no point in pursuing the impact that “twisting” the yield curve would have on policy objectives.

The analogous problem for exchange market intervention policy is to test the hypothesis that nonmonetary financial assets denominated in different currencies are imperfect substitutes. If assets denominated in different currencies are imperfect substitutes, changes in relative supplies generated by exchange market intervention would require changes in the relative expected yields in order to restore equilibrium. In general, a change in the equilibrium level of relative expected yields would be accompanied by changes in both current exchange rates and expected paths for future exchange rates. While the nature of the exchange rate adjustments depends upon a number of further considerations, it is well known that any policy that alters real expected yields can have a powerful impact on the level of spot exchange rates.5 The possibility that changes in expected yields generate changes in exchange rates that exceed, or “overshoot,” changes in the long-run expected value of the exchange rate sharpens one’s interest in the imperfect substitutes hypothesis. The model developed in Section I is designed to explore the possibility that outside debts of governments are imperfect substitutes owing to either political or exchange risk. If assets issued by different governments are imperfect substitutes owing to political risk, changes in external indebtedness that accompany current account imbalances may have powerful effects on exchange rates, even if there is no change in the currency denomination of outside assets. It follows that empirical tests of the effectiveness of intervention policy must carefully distinguish between changes in the external indebtedness of different countries and changes in the stock of outside debt denominated in a given currency.

A review of existing empirical work in Section II suggests that the expected rates of return on assets that are identical except for their currency denomination do appear to vary over time. These differences in expected yields, often referred to as exchange risk premiums, are consistent with the hypothesis that government debts denominated in different currencies are imperfect substitutes. There are, however, other plausible explanations for the observed yield differentials. Moreover, studies that have attempted to link changes in risk premiums to official foreign exchange market intervention have met with limited success. Thus, one cannot rule out the possibility that intervention can have a predictable impact on policy objectives, but the case is far from established.

In Section III, this paper presents estimates for exchange market intervention activity for a large number of industrial and developing countries. A surprising implication of these estimates is that exchange market intervention policies of several major industrial countries since 1973 have generated only minor changes in the relative supplies of bonds denominated in these countries’ currencies. However, developing countries and smaller industrial countries have employed this policy tool more extensively, in that an increasing share of their total net debt has been denominated in foreign currencies. Moreover, since this foreign currency debt is largely denominated in major currencies, the intervention policies of developing countries have generated relatively large changes in the relative supplies of bonds denominated in the currencies of major industrial countries. The failure of existing empirical work to establish a link between intervention policies of industrial countries and yield differentials or exchange rates among their currencies might be due to the failure to model the effects of third country intervention policies. While a multi-country framework may be necessary in order to test the importance of exchange market intervention policy, data for the intervention activities of developing countries are currently available only on an annual basis for a few years; such limitations mean that formal hypothesis testing is of little value.

In the concluding section, two objectives are suggested that seem to be consistent with observed exchange market intervention policies. First, intervention policy might contribute to the objective of minimizing the expected real costs of servicing a country’s public debt. This might be an important factor in understanding the intervention policies of developing countries. Second, exchange market intervention might be useful at times when the government’s forecasts for exchange rates are different from the private sector’s forecasts that are embodied in market exchange rates. Since a difference of opinion between the government and the private sector would presumably narrow as events unfolded, there is a strong presumption that such intervention would be reversed so as to return to the long-term debt management strategy described earlier. This second objective of intervention policy may be important to industrial countries. Intervention policies based on either of these objectives might be important determinants of exchange rates.

I. A Portfolio-Balance Model with Political Risk and Exchange Rate Risk

In order to address the policy of exchange market intervention, a model of the financial sector of an open economy is developed in this section.6 The model can be simplified by assuming that various financial assets are perfect substitutes or that the country considered is small enough so that it is a price taker in both financial and goods markets. As is discussed later on, such assumptions have proven useful in focusing on several important aspects of exchange rate determination. However, the ineffectiveness of intervention policy, as defined in this paper, follows directly from these simplifying assumptions.

A financial sector that provides testable hypotheses concerning the effectiveness of exchange market intervention policy involves two countries—the United States and the Federal Republic of Germany—and three outside financial assets issued by the government of each country. The assets are the following:

M = dollar money issued only by the U.S. Government

BUS$ = dollar-denominated securities issued by the U.S. Government

BUSDM = deutsche mark-denominated securities issued by the U.S. Government

M* = deutsche mark money issued only by the German Government

BGDM = deutsche mark-denominated securities issued by the German Government

BG$ = dollar-denominated securities issued by the German Government

The demands for these assets are assumed to be different for residents of the two countries. Two types of risk are associated with a given financial asset. First, exchange risk is associated with uncertain outcomes for future exchange rates.7 Second, political risk is associated with the possibility that a government will default on, or penalize, net nonresident holdings of financial claims on domestic residents.

In order to deal with a six-asset financial sector, it is necessary to impose some simplifying assumptions. A potentially useful assumption is that private wealth holders make separable decisions concerning the share of their bond portfolios denominated in each currency and the share of net claims on each country. This is a strong assumption, in that it requires, among other conditions, that exchange risk be independent of political risk.8 Separability implies, for example, that sales of deutsche mark-denominated bonds against dollar-denominated bonds issued by the United States Government that are offset by sales of dollar-denominated bonds against deutsche mark-denominated bonds issued by the German Government would not generate changes in market yields. This is because political risk would not change, since the net indebtedness of both governments would not change, and exchange risk would not change since the global supply of deutsche mark-denominated and dollar-denominated outside bonds would not change. Our assumption also suggests the following restrictions on yield differentials:

RUSDMRUS$=E%X+(1)
RGDMRG$=E%X+(2)
RGDMRUSDM=E%D+α(3)
RG$RUS$=E%D+α(4)

Equations (1) and (2) state that the interest rate differentials between dollar-denominated and deutsche mark-denominated securities that carry the same political risk—that is, those issued by the same government—will be the same and will equal the expected change in the spot exchange rate plus the exchange risk premium. These two equations can be interpreted as a covered-interest-rate-parity condition that is known to hold for Eurocurrency deposits that carry the same political risk.9 It follows that the interest rate differential defined in equations (1) and (2) will always be equal to the market forward exchange rate. In this model, forward exchange market intervention by a government is exactly equivalent to a sale of its dollar-denominated security for its deutsche mark-denominated security, or the reverse. Thus, forward exchange market intervention is equivalent to a transaction that changes the relative supplies of two of the four bonds already defined in the model. It is important to note that private sector transactions in the forward exchange market that do not involve governments merely transfer the existing stock of outside exchange risk associated with the existing stock of outside government bonds.10 There is no independent market-clearing condition for the forward exchange rate.

Equations (3) and (4) state that the interest rate differentials between bonds issued by the U.S. and German governments that carry the same exchange risk—that is, those denominated in the same currency—are the same and are equal to the differential expected rate of default plus the differential political risk premium α. Political risk is associated with the probability that a government will force complete or partial default on domestic debt held by nonresidents. It is also assumed that the risk of default is the same, regardless of the currency denomination of the debt. For simplicity, it is assumed that only net indebtedness is subject to political risk. That is, in the event of partial or total default, the holders of claims on the defaulting country will be compensated to the extent that other domestic residents have liabilities to the defaulting country.

Conditions (1)-(4) allow a considerable simplification of the financial sector of the model. U.S. residents’ demand for claims on German residents is viewed as depending on the relative rates of return on bonds denominated in the same currency issued by the German and U.S. governments and upon U.S. wealth (measured, for consistency, in deutsche mark):

BUSG=1/XBG$US+BGDMUS=f(RG$RUS$,RGDMRUSDM,WUS)(5)

Substituting equations (3) and (4) into (5) yields

BUSG=f(E%D+α,WUS)(6)

where E%D denotes the mean probability of default, which is viewed as depending upon net indebtedness of German residents, and α denotes the political risk premium associated with the net indebtedness of German residents. U.S. residents’ demand for deutsche mark-denominated bonds is viewed as depending upon the yield differential between deutsche mark-denominated and dollar-denominated issues of both governments.

BUSDM=g(RGDMRG$,RUSDMRUS$,WUS)(7)

Substituting equations (1) and (2) into (7) yields

BUSDM=g(E%X+,WUS)(8)

where E%X denotes the expected annual rate appreciation of the deutsche mark against the dollar and ℓ denotes the exchange risk premium necessary to induce the private sector to hold the existing stock of dollar-denominated and deutsche mark-denominated bonds.

A similar derivation for German residents yields

BGG=h(E%D+α,WG)(9)
BGDM=i(E%X+,WG)(10)

There are now two market-clearing conditions. For bonds issued by the German government

BG=BUSG+BGG=f(E%D+α,WUS)+h(E%D+α,WG)(11)

And for deutsche mark-denominated bonds issued by both governments11

BDM=BUSDM+BUDM=g(E%X+,WUS)+i(E%X+,WG)(12)

The market-clearing conditions (11) and (12) tell an interesting story about asset demand in a world where political risk is independent of exchange risk. Equation (11) states that the demand for bonds issued by the German government depends upon the difference between market interest rates on deutsche mark- (or dollar-) denominated bonds issued by the German Government and deutsche mark- (or dollar-) denominated bonds issued by the U.S. Government, adjusted for the expected loss owing to default and the risk premium, if any, necessary to induce wealth holders to hold the existing stock of securities subject to political risk. Condition (12) states that the demand for deutsche mark-denominated bonds issued by both the U.S. and German governments depends upon the difference between interest rates on deutsche mark-denominated bonds issued by the Federal Republic of Germany (United States) and dollar-denominated bonds issued by the Federal Republic of Germany (United States), adjusted for the expected change in the exchange rate and the exchange risk premium, if any, necessary to induce wealth holders to hold the existing supply of deutsche mark-denominated outside debt. For both market-clearing conditions, the distribution of world wealth between U.S. and German residents is important if, as we would expect, their demand functions are different.

Exchange market intervention policy does not directly affect condition (11) since this is the market-clearing condition for each government’s net debt, and intervention changes only the currency composition of the debt but not its level. Intervention does directly affect condition (12), since a change in the relative supply of deutsche mark-denominated and dollar-denominated bonds requires a change in the relative expected yields on these assets.

If the model were extended to include three or more countries, political risk premiums would be determined by each country’s net external debt, but exchange risk premiums would be determined by the global supply of net government debt denominated in each currency. It follows that the measure of exchange market intervention policy that was relevant to a given pair of currencies would not be restricted to the balance sheets of the two governments that issued the currencies.

POSSIBLE SIMPLIFICATIONS OF FINANCIAL SECTOR

The financial sector that has been developed here is sufficiently general to encompass a variety of models developed in recent years, which can be viewed as special cases. Our primary objective in reviewing other models is to see what, if anything, can be said about exchange market intervention policy in more streamlined versions of the model.

A large number of papers associated with the monetary approach to the balance of payments are based on the assumption that securities denominated in different currencies and issued by different governments are perfect substitutes.12 The simplest of such models deals with a small, open economy and assumes fixed exchange rates. In such a model, there is only one bond market, and since the country considered is “small,” the interest rate necessary to clear the bond market is unaffected by disturbances originating in the small country. In terms of the model developed here, all of the right-hand-side variables in equations (1)-(4) are set at zero, so that there is one world interest rate that is exogenous to the small country. In such a model, exchange market intervention as defined in this paper is, by assumption, a trivial policy.

In flexible exchange rate versions of the monetary model, wealth holders are assumed to be insensitive to exchange risk and political risk, so that equations (1)-(4) can be simplified to

RUS$RGDM=E%X(13)

where the large country interest rate, RUS$, is exogenous and the small country rate, RGDM, is sensitive to changes in the small country money supply. If prices in the small country are assumed to adjust slowly, an unexpected change in the money supply can change the real interest rate and, in turn, generate exchange rate changes that overshoot their long-run equilibrium level. While the flexible exchange rate model is somewhat more complex than the fixed exchange rate model, there is still, by assumption, no scope for exchange market intervention in this model.13

Another class of models assumes that assets denominated in different currencies are imperfect substitutes but that political risk is not a factor in portfolio decisions. Conditions (1)-(4) then can be summarized as

RUS$RGDM=E%+(14)

where , which denotes the exchange risk premium, depends in part upon the stocks of dollar and deutsche mark bonds outstanding and the distribution of wealth between U.S. and German residents.14

This model has quite striking implications for intervention policy. It is generally assumed in such models that residents of the different countries have different demands for assets denominated in various currencies, so that equation (12) takes the form given earlier. In such a model, intervention changes the stock of deutsche mark-denominated and dollar-denominated bonds that the private sector is forced (at market-clearing prices) to hold. A change in BDM relative to B$, holding the total world bond supply and world wealth constant, requires a change in ℓ, so that the expected yield on the reduced supply of BDM falls.

A current account imbalance shifts wealth from U.S. to German residents. For example, a U.S. current account deficit implies a transfer of wealth from U.S. residents to German residents. If German residents have a relatively stronger preference for deutsche mark-denominated bonds, the world demand for deutsche mark bonds will rise and, other things being equal, the expected yield on deutsche mark bonds will fall. The interesting point is that there is always some intervention policy that will exactly offset the change in expected yields generated by a current account imbalance. It follows that a current account deficit will not affect exchange rates as long as the deficit government is expected to supply the foreign-currency-denominated bonds necessary to satisfy the portfolio preferences of foreign wealth holders. Models assuming that all current account imbalances necessarily result in increased holdings of foreign-currency-denominated financial assets by residents of the surplus country avoid this problem, but they are implicitly based on the assumption that governments never choose to intervene in the foreign exchange market.15

It is concluded from this brief overview of existing models that an appraisal of intervention policy requires a model that deals with both political risk and exchange risk as determinants of private portfolio behavior.

One way to simplify such a model is to rule out expected paths for the level of real exchange rates that generate levels of net international indebtedness that are expected to increase without limit. Expected paths for current accounts that imply unlimited growth in a country’s net international indebtedness can be ruled out if, at some future date, the associated rise in debt service costs owing to an increasing political risk premium will make default optimal for the debtor government. Since the present value of any nonresident holdings of debt on the day of default is zero, such a path for exchange rates cannot be optimal. If it is assumed that only one exchange rate level is consistent with a sustainable current account position in a steady state, a long-run equilibrium level of the exchange rate s in some future time period T can be determined. The path that exchange rates are expected to follow toward sT is a component of the expected yields on all the financial assets in the model. Foreign exchange market intervention policy plays a role in determining the steady-state exchange rate, since the debt service burden the government is expected to incur depends, in part, upon the currency denomination of its debt. A country that never issues or holds foreign-currency-denominated debt (never intervenes in the foreign exchange market) might be expected to pay an exchange risk premium on its securities that would depend upon the parameters in condition (12). If this is the case, the sustainable net indebtedness of a country over time will depend in part on its exchange market intervention policy, which is one of the many factors that will determine the level of sT. Thus, if exchange risk premiums are not zero, intervention policy does have “real” effects and can influence the equilibrium path for exchange rates.

II. Empirical Tests of Model

Two approaches are used to evaluate empirically the effectiveness of exchange market intervention policy. The more direct approach is to solve condition (12) for the exchange rate and to attempt to estimate parameters that measure the effect on exchange rates of changes in the supplies of bonds denominated in different currencies.

Comprehensive reviews of such efforts by Genberg (1981) and Meese and Rogoff (1981) indicate that in practice this approach has not provided convincing evidence for a stable relationship between intervention policies and exchange rates. As pointed out in Mussa (1979) and Dooley and Isard (1981), changes in exchange rates seem to be dominated by revisions in expectations concerning future values of money supplies, bond supplies, current account balances, and other important variables rather than the outcomes for such variables during the period in question. Perhaps because of the difficulty in specifying expectations for these variables, existing models have succeeded in explaining only a small share of the observed variability in exchange rates since 1973.

An alternative approach to evaluating the possible effectiveness of exchange market intervention is to note that a necessary, but not a sufficient, condition for intervention to be effective is that bonds denominated in different currencies be imperfect substitutes. A contribution of the model developed in this paper is to provide a basis for identifying which of the many possible interest rate differentials available can be exploited to test the imperfect substitutes hypothesis.

In order to test this hypothesis, it is necessary to identify securities that are pure counterparts to the theoretical bonds in the model developed previously. In particular, we must be able to measure supplies and rates of return on securities that carry equivalent political risks but are denominated in different currencies. Although industrial countries do, from time to time, issue foreign currency securities (for example, the Carter bonds issued by the United States), there are often special marketing or other restrictions on such issues that make comparison with domestic currency securities difficult. In the case of developing countries, comparison is difficult because their domestic currency securities may not be widely traded and may be subject to administered interest rates or other market imperfections. These problems have motivated several efforts to disentangle information available in Eurocurrency market interest rates.

One can view the Eurocurrency market as the combined balance sheets of Eurobanks. In order to avoid needless complexity, it is assumed that Eurobanks are relatively averse to exchange risk and always match assets and liabilities by currency denomination. Each of the assets of Eurobanks are equivalent to one of the four bonds in the financial sector described earlier. Since Eurobanks hold a wide variety of financial assets, we are assuming that claims on the private sector of a given country are perfect substitutes for the outside bonds of that country’s government denominated in the same currency. Since a Eurobank is by definition not a resident of either country, it faces political risk on all of its assets.

For example, in offering a Euromark deposit, the Eurobank must consider its investment opportunities, which consist of deutsche mark-denominated bonds issued by U.S. or German residents. The expected yields on these two bonds would differ because of the different political risk associated with each. The Eurobank deposit, in contrast, can be viewed as carrying no political risk for the depositor, since it can be expected that the bank’s capital will act as a guarantee in the event of a decrease in the value of the Eurobank’s assets. It follows that the difference between the Euromark deposit interest rate and the interest rate on U.S. and German government deutsche mark-denominated bonds is a measure of the market’s valuation of political risks associated with claims on German and U.S. residents.

Following the assumption made in this paper that only the net debtor country carries positive political risk, the Eurocurrency rate will be equal to the interest rate (plus a normal deposit-loan spread) at which the net creditor country can issue bonds denominated in various currencies. For empirical work, this paper will assume that the net creditor or zero political risk country is the United States. Thus, the Eurodollar loan rate, EDOL, is assumed to be identical to the yield, RUS$, that the U.S. Government would pay on a similar security. More important for the purposes of this paper is the implication that the Euromark loan rate, EDM, is identical to the normally unobservable rate, RUSDM, at which the U.S. Treasury could sell deutsche mark-denominated securities. Moreover, since the difference between the Euromark deposit rate and the German Government bond rate, RGDM, must be the same as the difference between the Eurodollar deposit rate and the interest rate on dollar-denominated German Government bonds, a value can also be derived for the normally unobservable rate, RG$, at which the German Government can sell dollar-denominated bonds. To summarize, we can obtain the following information from Eurocurrency interest rates:

RUS$=EDOL(13)
RUSDM=EDM(14)
RGDM=EDM+E%D+α(15)
RG$=EDOL+E%D+α(16)=EDOL+(RGDMEDM)

Perhaps the most interesting hypothesis suggested by these relationships is that the market forward exchange rate, which is known to equal the Eurocurrency interest rate differential, reflects the differential yield that both governments face in marketing debt denominated in the two currencies. The observed forward premium is independent of political risk and depends only on the expected value of the change in the exchange rate, E%X, and the risk premium, , associated with the existing stocks of all governments’ bonds denominated in different currencies. Moreover, exchange market intervention, which does not alter political risk, will have no direct effect on the differences between onshore and offshore interest rates on assets denominated in the same currency.

One can now substitute observable interest rates into equations (11) and (12), assume a linear relationship, and solve for the interest rate differential.

RGDMEDM=a0+a1BG+a2WUS+a3WG+a4E%D+a5α(17)
EDMEDOL=b0+b1BDM+b2WUS+b3WG+b4E%X+b5(18)

Relationships similar to equations (17) and (18) have been tested in a number of papers using a wide variety of techniques. In general, the models tested have not carefully distinguished between political and exchange risk, but the regression forms are similar to those developed here. The differential between domestic and offshore interest rates paid on bonds denominated in deutsche mark was tested in Dooley and Isard (1980), with the additional specification that capital controls might also affect the interest differential. In that study, it was concluded that while capital controls in place were the most important determinant of the very large differentials that marked the 1971–74 time period, political risk also seemed to explain a significant part of these interest rate differentials. This finding has an important bearing on the model developed here, because it suggests that a country’s prospects for future increases in external indebtedness can have an important impact on the terms on which it can borrow.

Although political risk has had measurable effects on interest rates, the second part of the story—the importance of the currency denomination of government debt in determining the interest rate differentials—has proven more difficult to support. A number of approaches to testing this relationship have been employed. The most common is to impose the condition that b4 = 1, so that

EDMEDOLE%X=b0+b1BDM+b2WUS+b3WG+b5(19)

Under the null hypothesis that securities are perfect substitutes, all the coefficients on the right-hand side are zero and equation (19) becomes the well-known open-interest-parity condition. If it is further assumed that the forecast of E%X conditional on today’s information is formed rationally, then the ex post observed differences between the interest differential EDM - EDOL (often measured as the forward exchange premium) and the changes in the exchange rate are measures of forecast errors for the change in the exchange rate. Under the joint hypotheses of rational expectations and perfect substitutes, such forecast errors should be serially uncorrelated and have a zero mean.

A number of studies have attempted to test this joint hypothesis. While the results of such studies are mixed, a number of careful studies, notably a recent paper by Hansen and Hodrick (1980), have concluded that the perfect substitutes-rational expectations hypothesis is not consistent with the data. Thus, it appears that exchange market intervention might be an effective policy tool. There are, however, only three studies known to the author that have attempted to test a regression hypothesis similar to equation (18). Dooley and Isard (1981) tested a variant of equation (19) for quarterly data for 1973–78. That study was primarily directed toward determining the likely maximum size of risk premiums under some plausible assumptions concerning desired portfolio shares for residents of four regions: the United States, the Federal Republic of Germany, Organization of Petroleum Exporting Countries (OPEC) members, and the rest of the world. In that paper, it was found that point estimates of risk premiums appear to be quite small. Using similar data, Frankel (1979, 1982 b) tested a more general form of equation (19) and concluded that there was no support for the imperfect substitutes hypothesis.

In the next section, data is presented on the intervention activities or a large number of countries in order to assess the importance of a prominent feature of the theoretical model developed in this paper, which is that the relevant bond supplies in equation (18) are not restricted to the dollar and deutsche mark bonds issued by the Federal Republic of Germany and the United States but also include all outside assets denominated in these two currencies. These data suggest that the two-country models that lie behind existing empirical work lead to substantial mismeasurements of the regressors in equation (18).

III. Exchange Market Intervention of Industrial and Developing Countries: 1973-80

In this section, the paper presents estimates for exchange market intervention for both industrial and developing countries. The data indicate that the intervention policies of industrial countries since 1973 have generated changes in the supplies of bonds denominated in various currencies that are quite small relative to the growth of total debt. For many industrial countries, foreign currency debt positions were large relative to the total stock of debt outstanding in March 1973, but these positions remained nearly unchanged through the end of 1980, while the stock of total net debt tripled.

The apparent unwillingness of industrial countries to denominate a share of their debt in foreign currencies may have led private market participants to regard intervention transactions of industrial countries as representing only transitory changes in the relative supplies of securities denominated in different currencies. In contrast, estimates for intervention policies of developing countries indicate that this group of countries increased the share of its total net debt denominated in the currencies of industrial countries between 1974 and 1978. Recalling that political risk depends upon the net claims on a given country but that exchange risk is a function of the net positions that all governments take in a given currency, it follows from the model developed in this paper that the intervention policies of developing countries may have been important in determining an exchange risk premium between currencies such as the U.S. dollar and the deutsche mark.

In evaluating the importance of developing countries’ intervention policies, it is probably necessary to take into account the fact that these policies depend, in part, upon conditions in exchange markets and international credit markets. The possible endogenous nature of developing country intervention policies does not alter the fact that private wealth holders must be willing to hold the total supply of outside debt denominated in a given currency.16 Given the rapid changes in the financial positions of both oil producing and other developing countries in recent years, the collective intervention decisions of these governments has played an important role in determining the currency denomination of financial assets held by the private sector.

Official transactions included in the estimates of intervention presented in this section follow from the model developed in Section I. In particular, the definition of intervention is neither limited to transactions initiated with the intention of affecting exchange rates or exchange market conditions nor is it limited to transactions initiated by the part of the government that is normally concerned with exchange markets. As was discussed in Section I, any transaction that is equivalent to an exchange with the private sector of nonmonetary debt denominated in different currencies is identified as intervention. It is important to recall that the theoretical model developed in Section I was based on the assumption that only net positions in various currencies are relevant. If a country were to issue a dollar-denominated bond, for example by borrowing dollars in the Eurodollar market, and invest the proceeds in a dollar bond, for example in a U.S. Treasury security, there would be no change in the private sector’s net holdings of dollar-denominated nonmonetary debt and therefore no net intervention. It follows that the management of the size and currency composition of a country’s official reserve assets is a component of its intervention policy that cannot be evaluated without taking into account that country’s management of the currency denomination of its nonmonetary debt.

Another important feature of the estimates of intervention reported in this paper is that to the extent possible, all transactions that alter the net currency position of the government are included. In addition to the changes in the currency denomination of official assets and liabilities, changes in the private sector’s position that are mandated or subsidized by the government should be included. For example, if a private or quasi-private firm is induced by the government to alter its net holdings of foreign currency debt, this should be included in that country’s measured intervention activity.17 Finally, all official forward exchange transactions should be included in an estimate of intervention.

Although this definition of intervention policy is in many ways a more comprehensive one than is usually considered, there is an important sense in which the definition of intervention is limited in this discussion. The most important limitation is that other policy tools that might have a profound impact on exchange rates are not considered a part of intervention policy. In particular, the management of the government’s monetary liabilities might be directed partly toward an exchange rate objective and might be more important than intervention policy in explaining movements in exchange rates.

INDUSTRIAL COUNTRIES

Our estimates of exchange market intervention of industrial countries are summarized in Charts 1-12 in the Appendix. The solid line in each chart shows the value (in domestic currency units) of the central government’s net nonmonetary debt. The broken line represents an estimate of the stock of industrial country governments’ debt that is denominated in domestic currency and held by the public. If there had never been any intervention in foreign exchange markets by industrial countries, these two aggregates would always equal one another and would measure the cumulative budget deficit of the government less sales of debt to the monetary authority. Exchange market intervention policy of all industrial countries in this country’s currency is measured by the divergence between these two debt aggregates. For example, in the first quarter of 1973, the German Government’s net nonmonetary debt totaled about DM 45 billion. But the value of deutsche mark-denominated nonmonetary debt of industrial country governments held by the public was more than two and a half times this amount, or about DM 116 billion. Thus, industrial countries’ cumulative exchange market intervention policies under the fixed exchange rate system had generated a supply of deutsche mark-denominated bonds in the hands of the public that was more than twice what would have been required to finance the Federal Republic of Germany’s budget deficits. For Japan, the supply of yen-denominated debt in the hands of the public in early 1973 was a bit less than twice what would have been necessary to finance Japan’s budget deficits; while for France, the value of French franc-denominated debt was about one and a quarter times the value of total French net debt outstanding. As shown in Table 1, similar ratios for other industrial countries, with the exception of the United States, stood between unity for the United Kingdom and 1.2 for Sweden. The United States was in the opposite position from the rest of the industrial countries at the close of the fixed rate system. The United States’ net nonmonetary debt totaled about $266 billion in early 1973, while U.S. dollar-denominated debt in the hands of the public was only about 70 per cent of this amount.

Chart 1.
Chart 1.

United States: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of U.S. dollars

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 2.
Chart 2.

Canada: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of Canadian dollars

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 3.
Chart 3.

Australia: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of Australian dollars

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 4.
Chart 4.

Japan: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1980

Trillions of yen

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 5.
Chart 5.

Belgium: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of Belgian francs

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 6.
Chart 6.

Finland: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of markka

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 7.
Chart 7.

France: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of French francs

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 8.
Chart 8.

Federal Republic of Germany: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of deutsche mark

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 9.
Chart 9.

Italy: Estimated Exchange Market Intervention, First Quarter 1975-First Quarter 1981

Trillions of lira

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 10.
Chart 10.

Netherlands: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of guilders

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 11.
Chart 11.

Sweden: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of krona

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Chart 12.
Chart 12.

United Kingdom: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

Billions of pounds

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

Table 1.

Twelve Industrial Countries: Ratio of Net Nonmonetary Debt of Central Governments, Denominated in Domestic Currency, to Total Net Nonmonetary Debt, First Quarter 1973 and First Quarter 1981

article image

Second quarter 1973.

First quarter 1975.

First quarter 1980.

Third quarter 1980.

It seems clear from this data that the exchange market intervention policies of industrial countries played a major role in determining the currency denomination of assets in the hands of the public before 1973. As shown in Charts 112 in the Appendix, however, the currency denomination of nonmonetary debt issued by governments of these industrial countries has been determined since 1973 almost exclusively by the size of the fiscal deficits of various countries. This is most clearly the case for the United States, the Federal Republic of Germany, and Japan—countries that have received the bulk of the attention in studies of exchange market intervention policy. As shown in Table 1, the ratio of total net debt to domestic-currency-denominated debt of these countries moved rapidly toward unity from 1973 to 1981. Exceptions to this tendency included the United Kingdom and Italy, whose policies appear to have remained essentially unchanged. Finally, there was a clear tendency for the ratio of domestic currency to total net debt to decrease to less than unity for France and several of the smaller industrial countries including Belgium, Canada, Australia and Finland. As is shown in the charts, this tendency most often became apparent after 1977.

DEVELOPING COUNTRIES

Data for the exchange market intervention activities of 95 developing countries for 1974-78 are shown in Chart 13. The chart is similar to those presented for the industrial countries. The solid line shows an estimate of the dollar value of total net nonmonetary debt of the central governments of these countries, while the broken line shows the dollar value of domestic-currency-denominated nonmonetary debt of these countries. The difference is, by definition, the net value of these governments’ foreign currency liabilities—that is, their foreign exchange reserve assets less their foreign-currency-denominated nonmonetary debt.

Chart 13.
Chart 13.

Ninety-Five Developing Countries: Estimated Exchange Market Intervention, 1974-78

Citation: IMF Staff Papers 1982, 002; 10.5089/9781451946888.024.A004

The application of this paper’s definition of exchange market intervention to developing countries generates difficulties in both interpretation and measurement. A comprehensive analysis of the financial policies of developing countries would take into account the fact that, for many of them, such policies are severely limited by poorly developed markets for domestic currency debt. Moreover, the prevalence in developing countries of exchange controls that limit private exchanges of financial assets tends to isolate the market for domestic currency debt from international markets and thus undoubtedly alters the effects of intervention on exchange rates and other variables. An analysis of the role played in developing countries by the debt management policy defined in this paper as intervention is a subject for further research. This paper has a narrower focus, in that it attempts to evaluate the net effect these policy decisions have had on the supplies of debt denominated in major currencies.

Difficult measurement problems include the treatment of foreign currency positions of parastatal industries. In addition, a substantial share—about half in 1977—of the foreign currency debt of developing countries is held by governments of developed countries and by multilateral organizations. We assume that these governments and multilateral institutions pass this debt through, in the same currency, to the private sector. A consistent accounting for both developed and developing countries’ intervention would require inclusion of official claims of developed countries on developing countries in the measure of developed countries’ official assets. To the extent that such credits to developing countries have a substantial grant component, a full accounting of the pass-through would include implicit claims on future tax revenues of developed countries. If such a procedure were followed, it would not materially alter this paper’s measure of developed country intervention policy.

In a group of countries this large, there are, of course, a wide range of financial policies. If, for example, the value of foreign currency reserve assets of oil exporting countries had been always exactly equal to the net foreign currency liabilities of non-oil developing countries, there would be no intervention for the developing countries, and the two lines in Chart 13 would coincide. In fact, between 1974 and 1978, the $40 billion increase in foreign currency assets held by oil exporting and non-oil developing countries included in the sample was more than matched by the $108 billion increase in their foreign-currency-denominated liabilities.

The important difference in the data for these countries as compared with the industrial countries considered previously is the continued importance of exchange market intervention policy after 1974. At the end of 1974, data for these countries are similar to that for the industrial countries, in that a substantial share (about 26 per cent) of their total nonmonetary debt was denominated in foreign currencies. However, since 1974, while the industrial countries have not denominated a significant share of their increasing debt in foreign currencies, the developing countries included in this sample have denominated an increasing share of their net debt in foreign currencies. By 1978, the last year for which there was reasonably complete data, the share of developing countries’ foreign currency debt had increased to about 32 per cent of total debt. Partial data for 1979 and 1980 suggests that this trend probably continued.

In this environment, a change in the intervention policy of developing countries would not be interpreted as a transitory change in the relative supplies of bonds denominated in different currencies. Instead, such a change in policy could reasonably be interpreted as changing the expected relative supplies of, for example, dollar-denominated and deutsche mark-denominated bonds over an extended time period. This may be the reason that so-called reserve diversification has received considerable attention in recent years. The model developed in this paper, though, would suggest that only the net reserve position is important.

Table 2 shows a breakdown by currency denomination of the net debt position of 95 developing countries. As is shown in the table, the increase in the share of net foreign-currency-denominated debt for these countries was more than accounted for by an increase in the share of net dollar-denominated debt from -1 per cent to 16 per cent.18 The intervention activities of these developing countries added about $50 billion in dollar-denominated assets to private portfolios. In contrast, the share of developing country net debt denominated in deutsche mark was never large and, in fact, fell during the period. Even taking into account the difficulties in accurately measuring and interpreting the intervention policies of developing countries that were discussed earlier, it would appear that the intervention policies of developing countries and smaller industrial countries may have been more important in altering the relative supplies of dollar-denominated and deutsche mark-denominated securities than the intervention policies of the United States, the Federal Republic of Germany, and other major industrial countries. A very tentative conclusion is that, on balance, the intervention policies of developing countries may have contributed to pressure on dollar exchange rates. Unfortunately, data are available on an annual basis for only a few years and therefore are not useful for formal hypothesis testing. If the ideas advanced here prove to be of sufficient interest, it might be worthwhile to undertake the formidable task of gathering and refining quarterly data.

Table 2.

Ninety-Five Developing Countries: Net Nonmonetary Debt by Currency Denomination, 1974-78

article image

A natural extension of this research is a reappraisal of the existing literature on the importance of countries’ preferences for reserve assets denominated in different currencies. The model developed in this paper suggests that changes in holdings of reserve assets must be considered together with changes in official liabilities. Given the rapid increases in foreign currency liabilities outstanding, changes in the shares of liabilities denominated in various currencies are likely to dominate observed changes in the net position of that country in a given currency. For example, the fall in the share of reserve assets denominated in dollars during the period is overshadowed by the rise in the net share of dollar-denominated liabilities issued by developing countries. Studies that attempt to define an optimal policy for the management of the currency composition of reserve assets address only a part, and not the most important part, of a country’s exchange market intervention policy.

IV. Objectives for Exchange Market Intervention Policy

In this final section, policy objectives are suggested that are consistent with the observed patterns of recent exchange market intervention policies of developing and industrial countries. The primary conclusion offered is that exchange rates among major industrial countries are more likely to be systematically influenced by the intervention policies of developing countries than by the intervention policies of the industrial countries themselves.

Although there are undoubtedly many different objectives for exchange market intervention policies among developing countries, the dominant objective seems to be to minimize the expected cost of servicing their central government’s debt. One important implication of this imputed objective is that developing countries will continue to denominate a substantial share of their debt in foreign currencies. If assets denominated in different currencies are not perfect substitutes, a developing country faces the problem of choosing an optimal mix of net positions in domestic and foreign currencies. In most cases, the exchange risk premium necessary to induce nonresidents to hold the government’s domestic currency securities is likely to be greater than that on similar securities denominated in the currencies of industrial countries. For this reason, one should expect that a substantial share of the flow of developing country debt will continue to be denominated in the currencies of industrial countries. The collective decisions of developing countries as to the denomination of their net debt is likely to continue to be an important determinant of the supply of government securities denominated in these currencies.

The industrial countries, though, can be thought of as attempting to influence exchange rates over limited time horizons, where the size and duration of intervention initiatives are constrained by the fundamental decision not to accumulate foreign currency debt over time.

It seems likely that the effect of intervention policies under these conditions would be quite weak unless private expectations are influenced by the information conveyed by the government’s intervention policy. As was argued previously, the current exchange rate depends upon market participants’ expectations about future money supplies, bond supplies, current account imbalances, and a number of other variables. Given the behavior of industrial countries since 1973, an intervention initiative is not likely to lead to a revision of expectations concerning the currency composition of the flow of official debt over time. It is likely, however, that the exchange rate expectations of the government and the private sector will differ at times. This could be due to a different set of expectations for some important variable, such as the rate of relative growth of money supplies, or a different understanding of the parameters of the model that relate such variables to the exchange rate.

The market exchange rate depends only upon private expectations unless the government changes the relative supplies of securities or money in response to a difference in opinion concerning the equilibrium exchange rate. Suppose the government and private expectations for the growth of the domestic money stock are not the same. If the government adheres to its long-run intervention policy, the market exchange rate will follow a path that is incorrect according to the government view. The government could announce that the market is wrong, but this might have little impact. Or the government could engage in “supplementary intervention” in the exchange market and continue to do so to move the exchange rate toward its “correct” anticipated trajectory. This might be called zero accumulation intervention, since, when either the private sector or the government realizes its mistake, the intervention must be reversed to keep the exchange rate on the path consistent with the desired long-run debt management policy. Supplementary intervention can affect exchange rates by temporarily changing the relative supplies of bonds denominated in different currencies. Perhaps more important is the possibility that such intervention will be a signal to private market participants of what the government’s outlook is for the exchange rate and perhaps for monetary policy. Almost by definition, such intervention will not generate stable changes in interest rates or exchange rates, since such changes depend on a learning process on the part of both the government and private sector.

APPENDICES: I. Data Sources

Industrial countries

Data are from various issues of the Fund’s monthly publication International Financial Statistics (IFS) and national sources. Net nonmonetary debt is total central government debt minus holdings of the monetary authority. Domestic-currency-denominated nonmonetary debt is this total plus foreign exchange reserves adjusted for European Currency Unit (ECU) gold valuation changes, minus foreign-currency-denominated liabilities. Estimates of foreign-currency-denominated debt are drawn from IFS and national sources. Estimates for publicly guaranteed foreign-currency-denominated debt are included for Australia, Canada, France, Italy, and the United Kingdom. Forward exchange positions are included for Canada.

Developing countries

Data on foreign-currency-denominated debt are from the World Bank Debtor Reporting System and include public debt, publicly guaranteed debt, and a small amount of private external debt, all of which have original maturities of more than one year. Data on foreign currency reserve assets are from national sources. Data on domestic currency debt are estimated from data available in IFS for 37 countries. These 37 countries, which account for 30 per cent of total external debt, are assumed to issue domestic-currency-denominated debt and foreign-currency-denominated debt in the same ratio as the full sample of 95 countries.

II. Exchange Market Intervention of Selected Industrial and Developing Countries

BIBLIOGRAPHY

  • Aliber, Robert Z.,The Interest Rate Parity Theorem: A Reinterpretation,Journal of Political Economy, Vol. 81 (November-December 1973), pp. 145159.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Baker, Charles C.,The Basis and Practice of Treasury Debt Management,in The Political Economy of Policy-Making: Essays in Honor of Will E. Mason, ed. by Michael P. Dooley, Herbert M. Kaufman, and Raymond E. Lombra (Beverly Hills, California, 1979), pp. 175218.

    • Search Google Scholar
    • Export Citation
  • Balbach, Anatol B.,The Mechanics of Intervention in Exchange Markets,Federal Reserve Bank of St. Louis, Review, Vol. 60 (February 1978), pp. 27.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Bilson, John F. O.,Recent Developments in Monetary Models of Exchange Rate Determination,Staff Papers, Vol. 26 (June 1979), pp. 20123.

  • Boyer, Russell S.,Optimal Foreign Exchange Market Intervention,Journal of Political Economy, Vol. 86 (December 1978), pp. 104555.

  • Cass, David, and Joseph E. Stiglitz,The Structure of Investor Preferences and Asset Returns, and Separability in Portfolio Allocation: A Contribution to the Pure Theory of Mutual Funds,Journal of Economic Theory, Vol. 2 (June 1970), pp. 12260.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dooley, Michael P.,Foreign Exchange Market Intervention,in The Political Economy of Policy-Making: Essays in Honor of Will E. Mason, ed. by Michael P. Dooley, Herbert M. Kaufman, and Raymond E. Lombra (Beverly Hills, California, 1979), pp. 22131.

    • Search Google Scholar
    • Export Citation
  • Dooley, Michael P., and Peter Isard,Capital Controls, Political Risk, and Deviations from Interest-Rate Parity,Journal of Political Economy, Vol. 88 (April 1980), pp. 37084.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dooley, Michael P., and Peter Isard,The Portfolio-Balance Model of Exchange Rates and Some Evidence that Risk Premiums are Small” (unpublished, Board of Governors of the Federal Reserve System, May 1981).

    • Search Google Scholar
    • Export Citation
  • Dooley, Michael P., and Peter Isard, (1982 a), “A Portfolio-Balance, Rational-Expectations Model of the Dollar-Mark Rate,Journal of International Economics, Vol. 12 (May 1982), pp. 25776.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dooley, Michael P., and Peter Isard, (1982 b), “The Role of the Current Account in Exchange Rate Determination,Journal of Political Economy, Vol. 90 (forthcoming in 1982).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger,Expectations and Exchange Rate Dynamics,Journal of Political Economy, Vol. 84 (December 1976), pp. 116176.

  • Dornbusch, Rudiger,Exchange Rate Risk and the Macroeconomics of Exchange Rate Determination” (unpublished, National Bureau of Economic Research, January 1980).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Fischer, Stanley,Stability and Exchange Rate System in a Monetarist Model of the Balance of Payments,in The Political Economy of Monetary Reform, ed. by Robert Z. Aliber (Montclair, New Jersey, 1977), pp. 5973.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frankel, Jeffrey,A Test of the Existence of the Risk Premium in the Foreign Exchange Market vs. the Hypothesis of Perfect Substitutability,International Finance Discussion Papers, No. 149 (Board of Governors of the Federal Reserve System, August 1979).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frankel, Jeffrey, (1982 a), “Monetary and Portfolio-Balance Models of Exchange Rate Determination,in Economic Interdependence and Flexible Exchange Rates, ed. by J. Bhandari (MIT Press, forthcoming in 1982).

    • Search Google Scholar
    • Export Citation
  • Frankel, Jeffrey, (1982 b), “A Test of Perfect Substitutability in the Foreign Exchange Market,Southern Economic Journal (forthcoming in 1982).

    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A.,A Monetary Approach to the Exchange Rate: Doctrinal Aspects and Empirical Evidence,Scandinavian Journal of Economics, Vol. 78 (No. 2, 1976), pp. 200224.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A., and Richard M. Levich,Covered Interest Arbitrage: Unexploited Profits?Journal of Political Economy, Vol. 83 (April 1975), pp. 32538.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A., and Joshua Aizenman,Aspects of Optimal Management of Exchange Rates” (unpublished, National Bureau of Economic Research, September 1981).

    • Search Google Scholar
    • Export Citation
  • Genberg, Hans,Effects of Central Bank Intervention in the Foreign Exchange Market,Staff Papers, Vol. 28 (September 1981), pp. 45176.

  • Girton, Lance, and Dale W. Henderson,A Two Country Model of Financial Capital Movements as Stock Adjustments with Emphasis on the Effects of Central Bank Policy,International Finance Discussion Papers, No. 24 (Board of Governors of the Federal Reserve System, 1973).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Hansen, Lars P., and Robert J. Hodrick,Forward Exchange Rates as Optimal Predictors of Future Spot Rates: An Econometric Analysis,Journal of Political Economy, Vol. 88 (October 1980), pp. 82953.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Henderson, Dale W.,Exchange Market Intervention Operations: Their Effects and Their Role in Financial Policy” (unpublished, Board of Governors of the Federal Reserve System, January 1982). This paper will appear in Exchange Rates: Theory and Practice, ed. by John F. O. Bilson and Richard C. Marston (University of Chicago Press, forthcoming in 1983).

    • Search Google Scholar
    • Export Citation
  • Henderson, Dale W., and Kenneth Rogoff,Negative Net Foreign Asset Positions and Stability in a World Portfolio Balance Model,Journal of International Economics (forthcoming in 1982).

    • Search Google Scholar
    • Export Citation
  • Hodrick, Robert J.,An Empirical Analysis of the Monetary Approach to the Determination of the Exchange Rate,in The Economics of Exchange Rates: Selected Studies, ed. by Jacob A. Frenkel and Harry G. Johnson (Reading, Massachusetts, 1978), pp. 97116.

    • Search Google Scholar
    • Export Citation
  • Isard, Peter,An Accounting Framework and Some Issues for Modelling How Exchange Rates Respond to the News,International Finance Discussion Papers, No. 200 (Board of Governors of the Federal Reserve System, January 1982).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kouri, Pentti J. K.,The Determinants of the Forward Premium,Institute for International Economic Studies, Seminar Paper No. 62 (Stockholm, August 1976).

    • Search Google Scholar
    • Export Citation
  • McKinnon, Ronald I.,Portfolio Balance and International Payments Adjustment,in Monetary Problems of the International Economy, ed. by Robert A. Mundell and Alexander K. Swoboda (University of Chicago Press, 1969), pp. 199234.

    • Search Google Scholar
    • Export Citation
  • Meese, Richard A., and Kenneth Rogoff,Empirical Exchange Rate Models of the Seventies: Do They Fit Out of Sample?” (unpublished, Board of Governors of the Federal Reserve System, January 1981).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mussa, Michael,Empirical Regularities in the Behavior of Exchange Rates and Theories of the Foreign Exchange Market,in Policies for Employment, Prices, and Exchange Rates, ed. by Karl Brunner and Allan H. Meltzer, Vol. 11, Carnegie-Rochester Conference Series on Public Policy (Amsterdam, 1979), pp. 957.

    • Search Google Scholar
    • Export Citation
  • Mussa, Michael, The Role of Official Intervention, Occasional Paper No. 6, Group of Thirty (New York, 1981).

  • Rodriguez, Carlos A.,The Role of Trade Flows in Exchange Rate Determination: A Rational Expectations Approach,Journal of Political Economy, Vol. 88 (December 1980), pp. 114858.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Roper, Don E., and Stephen J. Turnovsky,Optimal Exchange Market Intervention in a Simple Stochastic Macro Model,Canadian Journal of Economics, Vol. 13 (May 1980), pp. 296309.

    • Crossref
    • Search Google Scholar
    • Export Citation
*

Mr. Dooley, Assistant Chief of the Developing Country Studies Division of the Research Department, is a graduate of Duquesne University, the University of Delaware, and Pennsylvania State University.

The author would like to thank colleagues in the Fund as well as Arturo Brillembourg and Peter Isard for helpful comments on an earlier draft of this paper.

1

See Fischer (1977), Boyer (1978), Roper and Turnovsky (1980), and Frenkel and Aizenman (1981) for studies that address this question. In these studies, foreign exchange market intervention is defined as changes in the monetary base motivated by exchange rate changes. The framework developed in this paper would define such transactions as monetary policy regardless of the motivation behind the transactions.

2

See Balbach (1978) for a summary of several types of intervention transactions.

3

See, for example, Dooley (1979) and Mussa (1981).

4

See Baker (1979) for a review of this literature.

5

See Dornbusch (1976) and Isard (1981) for models in which changes in expected yields on assets denominated in different currencies generate changes in spot exchange rates.

6

The model is an extension of portfolio-balance models developed by McKinnon (1969), Girton and Henderson (1973), and others. See Henderson (1982) for an analysis of intervention policy in which a similar financial sector is a component of a more complete model.

7

See Kouri (1976) and Dornbusch (1980) for models of utility-maximizing portfolio selection in the presence of exchange risk.

8

In addition to independence, further strong assumptions are necessary concerning the behavior of wealth holders. See Cass and Stiglitz (1970) for a discussion of the conditions under which separability can be assumed.

10

More generally, this paper follows the conventional assumption that private financial positions in various currencies net to zero, so that market-clearing conditions include only government-issued outside financial assets.

11

Since data are not available on the net positions of U.S. and German residents denominated in dollars and marks, this paper will ignore the effect of changes in the spot exchange rate on wealth in both equations (11) and (12). For discussion of the role of revaluation of wealth resulting from changes in exchange rates, see Henderson and Rogoff (1982).

15

This point is made in Dooley and Isard (1982 b) in reference to Rodriguez (1980).

16

Simply put, an outside asset is an asset that the private sector does not incorporate into its own balance sheet. Changes in the relative supplies of outside assets can be assumed to be exogenous or endogenous in a model of exchange rate determination. See Henderson (1982).

17

As described in the Appendix, it is assumed that for developing countries, all foreign currency debt reported by the World Bank Debtor Reporting System is ultimately the responsibility of the government.

18

The negative share for 1974 means that the group of countries had a net asset position in dollars.

IMF Staff papers: Volume 29 No. 2
Author: International Monetary Fund. Research Dept.
  • View in gallery

    United States: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of U.S. dollars

  • View in gallery

    Canada: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of Canadian dollars

  • View in gallery

    Australia: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of Australian dollars

  • View in gallery

    Japan: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1980

    Trillions of yen

  • View in gallery

    Belgium: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of Belgian francs

  • View in gallery

    Finland: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of markka

  • View in gallery

    France: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of French francs

  • View in gallery

    Federal Republic of Germany: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of deutsche mark

  • View in gallery

    Italy: Estimated Exchange Market Intervention, First Quarter 1975-First Quarter 1981

    Trillions of lira

  • View in gallery

    Netherlands: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of guilders

  • View in gallery

    Sweden: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of krona

  • View in gallery

    United Kingdom: Estimated Exchange Market Intervention, First Quarter 1973-First Quarter 1981

    Billions of pounds

  • View in gallery

    Ninety-Five Developing Countries: Estimated Exchange Market Intervention, 1974-78