Two questions arise when large changes occur in exchange rates among the currencies of the major industrial countries. First, what are the effects of those changes on other important economic variables both domestically and abroad, including current account balances, output, prices, and interest rates? Second, what policy reactions, if any, should they entail? These questions are particularly timely in view of the large swings in exchange rates that have characterized the floating-rate era and especially the past five or six years. The 23 percent decline in the nominal effective exchange rate of the U.S. dollar and the 47 percent appreciation of the Japanese yen against the dollar from March 1985 through April 1986, and the decision by the five largest industrial countries in September 1985 to encourage changes in their exchange rates, have added to the urgency of developing a clear understanding of the processes that are involved and of their impact.

Two questions arise when large changes occur in exchange rates among the currencies of the major industrial countries. First, what are the effects of those changes on other important economic variables both domestically and abroad, including current account balances, output, prices, and interest rates? Second, what policy reactions, if any, should they entail? These questions are particularly timely in view of the large swings in exchange rates that have characterized the floating-rate era and especially the past five or six years. The 23 percent decline in the nominal effective exchange rate of the U.S. dollar and the 47 percent appreciation of the Japanese yen against the dollar from March 1985 through April 1986, and the decision by the five largest industrial countries in September 1985 to encourage changes in their exchange rates, have added to the urgency of developing a clear understanding of the processes that are involved and of their impact.

The scope of this paper is limited in three important ways. First, it deals only with the large industrial countries that have floating exchange rates. To examine the issues that arise for smaller countries with floating rates, as well as for countries with less flexible exchange rate arrangements or fixed rates, would require a different approach from that taken in this paper; available empirical studies pertain almost exclusively to the largest countries. Second, the paper is not concerned with very short-run fluctuations in exchange rates that are reversed within a few months, nor is it concerned with the problems that might arise from any persistent and long-run misalignment of exchange rates. Although the study addresses certain long-run issues at least peripherally, its main emphasis is on the effects that exchange rate movements have over the first few years following a policy change or other disturbance. This limitation is consistent with its empirical and policy-oriented approach. Third, no attempt is made to assess the extent to which exchange rates might in practice have been misaligned. Although it is important to reach a judgment—even if such a judgment must be to some extent subjective—on whether a given currency is undervalued or overvalued in order to determine the appropriate policy responses to currency movements, such judgments are beyond the scope of this paper.

Although the questions that are raised in this paper are central to many important policy issues, any attempt to answer them raises serious empirical and methodological problems. Foremost among these is that exchange rates respond to changes in many variables, including various policy instruments as well as private market behavior. The effects of exchange rate movements on the world economy depend critically on the origins of these movements, and it is usually difficult in practice to disentangle the precise reasons for movements in exchange rates. Second, the effects of exchange rate movements on the policy options and constraints facing national authorities depend on a broad range of factors, including the origins of the exchange rate changes, the nature of the trade and financial linkages among countries, how expectations are formed, and whether exchange rates were initially thought to be out of line. The difficulties raised by these and other issues are discussed in the paper. It must, however, be emphasized that the paper does not attempt to discuss systematically the determinants of exchange rate changes. It briefly describes the processes by which exchange rates are thought to respond to monetary or fiscal policies or to shifts in portfolio preferences of private asset holders (among other factors), but it does not assess the empirical evidence on the relative importance or the stability of those factors.1

Main Issues

The past several years have been characterized by a number of very large swings in nominal exchange rates among the major currencies. Furthermore, real exchange rates have also undergone very large changes, reflecting the fact that nominal rates have not been closely correlated with relative price movements. In many instances, yearly real effective exchange rate changes have been even larger than nominal changes (Chart 13).2 Italy and, to a lesser extent, France, have been exceptions; in those two countries, both of which have participated in the various European exchange rate arrangements in place during most of this period, nominal effective exchange rate changes have predominantly reflected inflation differentials.

Chart 13.
Chart 13.
Chart 13.

Major Industrial Countries: Nominal and Real Effective Exchange Rates, 1973–86

(Indices, 1975–84=100)

1 Rates based on the Fund’s Multilateral Exchange Rate Model.2 Relative normalized unit labor costs adjusted for exchange rate changes. Costs for the first quarter of 1986 have been estimated.

The importance of real exchange rate movements in recent years may be illustrated by reference to two of the most prominent cases: the large swings in the effective values of the pound sterling and the U.S. dollar. The pound sterling appreciated in nominal effective terms by 27 percent from the second quarter of 1978 to the first quarter of 1981 and then depreciated by 29 percent through the first quarter of 1985. During most of the six-year period, costs in the United Kingdom were rising somewhat faster than those of its trading partners. The real appreciation of the pound sterling to the first quarter of 1981 amounted to 59 percent, and the subsequent real depreciation was 18 percent. Similarly, the U.S. dollar appreciated in nominal effective terms by 63 percent from the third quarter of 1980 to the first quarter of 1985 and then depreciated by 19 percent during the following four quarters. The corresponding real changes amounted to 58 percent and 21 percent, respectively. Thus in both cases, exchange rate movements were essentially unrelated to movements in relative costs.

In assessing the effects of exchange rate changes on the economy, the fact that such changes have frequently been at least as large in real terms as in nominal terms must be borne firmly in mind. Nominal exchange rate changes that merely offset differences in price performance among countries will be benign so long as the origins of the change are monetary; indeed, in these cases they are even beneficial in helping economies insulate themselves from disturbances that affect the price level differently in different countries. In contrast, although changes in real exchange rates may be essential to the process of adjustment to certain nonmonetary disturbances, they may be disruptive when the disturbances arise primarily from monetary factors or from speculation that is unrelated to economic conditions. This distinction is examined more closely below, notably in the next section.

A second issue that is discussed at various points throughout this paper is that the effects of exchange rate changes may vary substantially depending on the source of the change. Suppose, for example, that the U.S. dollar were to depreciate against all other major currencies because of an exogenous shift in private market preferences away from interest-bearing dollar-denominated assets, with no change in monetary or fiscal policies. This type of disturbance is discussed in the next section. In general, such a shift would be expected to lead not only to a depreciation of the dollar, but also to a rise in (nominal and real) interest rates on dollar-denominated assets, reflecting the reduced demand for them, and to a similar (though not necessarily equal) decline in interest rates on assets denominated in other currencies.

In response to this kind of portfolio disturbance, one would normally expect an eventual strengthening of the current account balance of the country with the depreciating currency (the United States, in this example) and a corresponding temporary increase in the growth of output. However, a decline in the growth of domestic demand or aggregate supply could offset the higher output arising from the strengthening of the external balance. In the countries whose currencies are appreciating, growth in output would temporarily decline as a result of the weakening of the current account balance but could rise through a stimulus to domestic demand or through improved supply conditions. So in all the countries involved, the direction of the effect on output would be theoretically ambiguous, while the direction of the shifts in current account balances would be more straightforward.

Now suppose, alternatively, that the depreciation were to result from an expansionary monetary policy rather than from an exogenous private portfolio shift. In this case, real interest rates would be at least temporarily reduced in the depreciating country and—probably by a smaller amount—in the world as a whole. In this case, in contrast to the preceding example, the risk premium of the two currencies is assumed not to change. Thus the interest rate differential between two countries should change only if the expected rate of depreciation were to change. An expansionary monetary policy is likely to lead to depreciation, and perhaps to “overshooting” of the nominal exchange rate; that is, the nominal exchange rate may initially depreciate by more than the ultimate effect of the policy shift on prices. Therefore, following the initial depreciation, market participants will come to expect an appreciation. In this case, interest rates in the expanding country will have to fall by more than those in other countries.

After the monetary expansion one would expect to see a temporary boost to domestic demand and to output in the depreciating country, but the direction of the effect on the current account balance would be ambiguous: the depreciation would contribute to a strengthening, while the rise in domestic demand would work in the opposite direction. As a related example, if the initial policy action were a monetary contraction abroad instead of a monetary expansion at home, the implications for exchange rates might be similar, but the temporary effects on each country’s demand and output would obviously be quite different.

Monetary policy changes in one country may lead either to higher or to lower economic activity in other countries, depending on the strength of various linkages. Monetary expansion, by raising aggregate demand at home, would tend to increase demand for foreign goods, other things being equal. This effect would tend to raise both economic activity and interest rates abroad, assuming unchanged real money balances. On the other hand, the appreciation of foreign currencies would also tend to reduce prices in those countries and thereby to raise real money balances. This effect would allow a fall in interest rates abroad. The net effect of these forces is difficult to establish a priori.

Fiscal policy may also have an important influence on exchange rates, with still different macroeconomic implications. A contractionary fiscal policy, whether implemented through reductions in government expenditure or through tax increases, could be expected to reduce real domestic interest rates and thus could contribute to a depreciation of the exchange rate. In contrast to the preceding example, this policy shift leads at least temporarily to a reduction in domestic demand and thus to an unambiguous strengthening of the current account balance. However, the depreciating effect on the exchange rate may be applicable primarily to the largest industrial countries, owing to the limited financing resources in smaller countries. For example, shifts in fiscal policy in many countries may be accommodated by monetary shifts, if monetary growth is determined in large part by the need to finance the government borrowing requirement, with little or no effect on real interest rates. In these circumstances, the dominant effect of fiscal contraction could be to strengthen the current account balance through the contraction of domestic demand, with the exchange rate appreciating rather than the reverse.

As this brief discussion suggests, it should not be surprising to find that there is no clear pattern to the relationship between exchange rate changes and other variables, such as interest rates, output growth, and current account balances. That finding would not imply that the causal relationships are weak, but rather that they are complex and depend crucially on the origin of the exchange rate changes. In order to provide some order to the discussion, the next section examines in some detail the empirical evidence on the effects of exchange rate changes that result from private portfolio shifts. The evidence relating to policy-induced exchange rate movements is then described in the subsequent section.

The third broad issue in this area relates to the policy implications of changes induced by movements in exchange rates. It is very difficult to derive general conclusions on this issue, because the implications depend on the initial circumstances facing national authorities as well as on the changes themselves. The importance of initial conditions may be illustrated by the appreciation of the U.S. dollar that persisted through the early part of 1985. A major factor that is thought to have contributed to this appreciation was the sharp rise in real interest rates in the United States from 1981 to 1983, which in turn resulted in large measure from the combination of an expansionary fiscal policy and a restrictive monetary policy during a good part of that period. The dollar’s appreciation was largely responsible for the deterioration of the U.S. current account position from 1982 to 1985, but it also contributed to the reduction in inflation in the United States.

The appreciation of the dollar—or, more accurately, the combination of policies that helped to produce it—substantially altered the policy constraints facing other countries, in that it contributed to a rise in world interest rates, substantial changes in relative prices, and a strengthening of the current account balances of other countries. Whether output growth in any particular country would have been higher or lower under different U.S. policies would depend on the relative importance of these various effects, which are examined in detail below.

Effects of Shifts in Private Portfolios with Unchanged Policies

This section examines the effects of exchange rate changes that result from an exogenous shift in the preferences of private market participants regarding the desired currency composition of their portfolios or the geographic location of the borrower.3 Such shifts might arise, for example, because of political uncertainties in one or more countries. Monetary and fiscal policies, as well as private spending propensities, are assumed not to change; in particular the exogenous components of the fiscal position are fixed, as are the growth rates of the relevant monetary aggregates. These assumptions do not mean that all changes in the economy that result from the portfolio shift can be attributed directly to the initial change in the exchange rate; other variables, in particular interest rates, will also be immediately affected and will have major effects on the economy. Nonetheless, these assumptions should isolate the effects of exchange rate changes as cleanly as possible in a world of floating (and therefore endogenous) exchange rates.

The analysis in this section makes several additional assumptions. First, costs and prices are assumed not to respond immediately to exchange rate changes. Thus a change in the nominal exchange rate is equivalent, initially, to a change in the real exchange rate; that is, to a change in relative costs or prices adjusted for exchange rate changes. Over time, as prices and costs respond to a portfolio disturbance, the real exchange rate will move back toward its initial value.

Second, it is assumed that the long-run equilibrium real exchange rate may change in response to any nonmonetary disturbance that alters the short-run exchange rate.4 The reasoning is as follows. Any change in the real exchange rate will alter the flow of current account balances and thus the cumulated stock of external net assets or debt. This change, in turn, will alter the flow of debt service payments over time and thereby generate a different long-run real exchange rate from the one that existed before the portfolio shift. Thus if a portfolio shift were to occur in an economy in long-run equilibrium and cause an appreciation on impact, it would ultimately lead the exchange rate to depreciate below the initial level. The degree to which the subsequent depreciation will exceed the initial appreciation will be largely determined by the size and persistence of the current account effects.5 This is a long-run concern and is not a significant factor in much of the following analysis, which concentrates on the short to medium run.

Finally, the analysis assumes that exchange markets are in equilibrium; the current level of the exchange rate may or may not be sustainable over the long run, but it is assumed to be market clearing, given the currency composition of the existing stock of assets. It is further assumed that the portfolio shift is not disruptive to the market; that is, it does not generate explosive or unstable shifts in market expectations, and it does not give rise to speculative bubbles. These two assumptions are at least implicit in all of the empirical models that are examined below; without them, one could only speculate as to the possible effects of any disturbance.

Effects on Prices

Economic theory suggests that a nominal exchange rate depreciation will lead to a rise in domestic prices, but the extent of that increase—that is, the question of whether prices and costs rise by enough to offset the effect of the depreciation on the real exchange rate—is an empirical issue. The measurement of this effect has received a great deal of attention in the literature, prompted in large measure by the need to understand the effects of the extraordinary swings in the external value of the U.S. dollar since 1973. Unfortunately, this task involves studying the effect of one endogenous variable on another; specifically, the problem is that it is logically impossible to derive a reduced-form equation relating prices to the exchange rate. Both variables are simultaneously determined; exchange rate changes affect prices and vice versa. Neither unambiguously “causes” the other; changes in both ultimately stem from a change in some exogenous factor. Consequently, empirical studies should and often do use simultaneous equation estimation techniques in an effort to derive consistent estimates of the linkages between prices and exchange rates.

In the analysis of the causal relationship running from exchange rates to prices, it is useful to begin by describing the principal structural linkages that relate these two variables. First, there is a fairly straightforward direct link. A depreciation of the exchange rate exerts upward pressure on the domestic currency price of imports and, consequently, on consumer prices.6 Evidence on the magnitude of this direct pass-through effect to import prices is provided by Spitäller (1980), in a study of nine major industrial countries. Spitäller finds that, although the effect is not immediate, a fairly complete pass-through of exchange rate changes to import prices occurs after two years. In the short run, firms normally absorb part of the rise in import costs, limiting the rise in domestic prices by allowing profit margins to fall. This delay in the price response appears to be attributable in part to uncertainty about the permanence of a given exchange rate change, and in part to the preference of firms to maintain market shares rather than profits.

While domestic consumer price and gross output indices incorporate the direct effects of an exchange rate change, value-added deflators do not. Conceptually, a change in the price of imports should have no direct effect on the GNP deflator, since it would affect the value of consumption and imports by the same amount.7 Only as the indirect effects of an exchange rate change come into force will costs, and hence output prices, be affected. In the case of a value-added deflator, such as a GNP deflator, any exchange rate effect must work entirely by changing the cost of primary factors.8

There are several indirect channels by which exchange rates can affect prices. The first of these runs from exchange rate changes to the price of competitively traded commodities and thus to production costs. Other things equal, a depreciation would be expected to lead to an increase in the local-currency price of primary commodities. The size of this effect will depend on several factors, the importance of which may be seen by assuming for the moment that the supply of the commodity is fixed. In this case, the size of the price increase will depend essentially on the elasticity of demand in each market as well as on the relative share of exports destined for each market. The more price-elastic is the demand in the depreciating country relative to that of the appreciating country, the smaller will be the price increase measured in the depreciating currency and the larger will be the price decrease measured in the appreciating currency. In addition, the larger is the market of the country with the depreciating currency—and the larger are the markets of countries whose currencies are pegged to the depreciating currency—the smaller would be the rise in the price of the primary commodity expressed in that currency (see Appendix I).

A second important indirect channel involves the dynamics of the wage-price determination process. In the standard explanation of this process, a depreciation directly raises import prices, which in turn raise the consumption price index. That increase raises, with a lag, nominal wages.9 Higher wages raise the output price as producers strive to restore profit margins; although the increase in wages is less than proportional, it causes the consumption price index to rise further. This process—prices onto wages and wages back onto prices—will continue, with each successive round being less than the last until the long-run price level is eventually reached (see Appendix II).

It has sometimes been argued that the process that has just been described is not self-damping—that exchange rate depreciation induces a rise in prices and thus further depreciation, in an endless “vicious circle.” Thus flexible exchange rates would constitute an independent source of inflationary pressure that is inherently unstable. Bilson (1979) has examined the empirical relevance of the vicious circle hypothesis by developing a model in which exchange rate movements and price movements are both determined ultimately by monetary policy actions. Exchange rate changes can magnify price movements, but they cannot lead to an explosive pattern unless the authorities lose control of monetary growth. Bilson’s model shows how, when asset markets adjust faster than goods markets, a monetary shock will cause exchange rates to change before prices, creating the appearance of a vicious circle but not in a manner in which exchange rates can be said to cause prices to become unstable. The vicious-circle argument thus does not appear to be of significant practical importance.

A number of recent single-equation and small-model studies of the effect of exchange rate changes on prices are summarized in Table 36; the large-scale simulation results will be examined later. In the studies referenced in Table 36, exchange rates are treated as exogenous; although monetary and fiscal policies are not always explicitly controlled for in these studies, it is assumed here that these changes may be viewed as arising from portfolio shifts. A cursory study of this table reveals a rather wide range of estimates of the effect of a 10 percent depreciation in the U.S. dollar on U.S. prices, ranging from nearly zero in the study by Glassman to 2.4 percent in the study by Sachs. Evidence on the effect of exchange rate changes on prices in the other industrial countries, though relatively meager, suggests a stronger effect, a finding that is consistent with the more open nature of the other industrial economies. Using pooled data, Bruno estimates the effect of a 10 percent depreciation in the non-U.S. OECD countries to be 3.7 percent. This number is consistent with the wide range reported by Artus and McGuirk, as discussed below. In general, the lower estimates of price effects tend to be found in studies based on input-output tables and single equations. Indirect competitive effects and wage-price dynamics are excluded in the input-output studies. Single-equation studies can, in principle, overcome these problems with instrumental-variable estimation techniques and a sufficiently general lag structure; in practice they seem to have only partially captured these effects when compared with the results of small models that explicitly model these channels.

Table 36.

Long-Run Price Effects of a 10 Percent Exchange Rate Change: Single Equation and Small Model Results

article image

CPI = consumer price index.

The two-equation Dornbusch-Fischer (1984) model and the seven-equation Sachs (1985) model suggest that a 10 percent depreciation of the nominal effective value of the U.S. dollar will cause the U.S. consumer price index to rise by 2.1–2.4 percent over a two- to four-year period. These results are particularly interesting for two reasons. First, they make use of more than one equation, so the wage-price dynamics may be explicitly modeled.10 Second, the parameter estimates are based on a longer data base than the earlier studies, and one in which the dollar has undergone substantial movements in both directions.11 The Sachs and Dornbusch-Fischer results can be compared with the consensus estimate of Hooper and Lowrey (1979), often cited in the literature, of 1.5–1.75 percent. An updated consensus would appear to be somewhat higher, perhaps on the order of 2 percent.

The Fund’s Multilateral Exchange Rate Model (MERM), although large, does not contain a comprehensive list of macroeconomic variables but rather focuses on the demand for and supply of 5 goods in 18 industrial countries and 2 regions. Consequently the results reported by Artus and McGuirk (1981) are discussed here, rather than later with the macro models. The MERM is a complex set of microeconomic relationships designed to answer questions about the long-run effect of exchange rate changes on prices. The parameters are not estimated directly but rather are based on work done elsewhere or on theoretical judgments.

Artus and McGuirk report on two MERM simulations in which the home currency was depreciated by 10 percent. In both simulations they assume that policies keep real output constant; the first simulation assumes a low (50 percent) feedback of prices onto wages, while the second assumes a high (70 percent) feedback. The 10 percent depreciation in the low feedback scenario leads eventually to increases in GDP deflators that range from 1.4 percent in the United States to 3 percent in the Benelux countries. The same depreciation in the high feedback scenario leads to increases in the GDP deflator ranging from about 4.3 percent in the United States, Canada, Spain, and Denmark to 6.8 percent in the Netherlands. In all instances the effect on the domestic demand deflator is somewhat higher; as noted earlier, the domestic demand deflator measures prices of imported as well as domestically produced goods and services, and it therefore reflects both the direct and indirect effects of a depreciation on prices. The GNP deflator, in contrast, reflects only indirect effects. Because a depreciation is less than completely passed through to wages in the MERM (real wages fall in response to a depreciation), output prices rise by less than consumption prices.

Effects on International Trade

Exchange rate changes that arise from a shift in private portfolio preferences with no change in national policies affect the flow of international trade in a number of ways. First, there is a direct relative price effect, since a change in the exchange rate alters import and export prices. Normally, this process is expected to involve an initial perverse (“J-curve”) effect; over time, however, a depreciation is expected to lead to a strengthening of the trade balance. Second, there are indirect effects to the extent that domestic demand responds to the exchange rate change or to the associated change in domestic prices or the change in domestic interest rates.

This subsection uses a partial equilibrium framework in which exchange rates, domestic demand, and domestic price deflators are assumed to be fixed, but in which export and import prices and volumes are determined endogenously. A useful and comprehensive survey of price effects in international trade can be found in Goldstein and Khan (1985); Table 37 is based partly on their work.

Table 37.

Long-Run Import and Export Demand Price Elasticities1

article image

Based in part on Table 4.2 in Goldstein and Khan (1985), p. 1081.

Several empirical regularities with respect to the larger industrial countries emerge from a study of Table 37. First, price elasticities of export demand are uniformly greater in absolute value than price elasticities of import demand, probably because of the relatively greater concentration of exports of the surveyed countries in manufactured goods. While the point estimates of these elasticities vary somewhat, export elasticities exceed import elasticities by nearly the same amount in four of the five studies that estimate both elasticities. Second, the cross-country models have very similar properties. The Goldstein-Khan (1978) results correspond very closely with Deppler-Ripley (1978), as do the Heien (1968) findings. Recent work by Spencer (1984), which updates the Deppler-Ripley study, finds elasticities that are about 10 percent smaller than those in the Deppler-Ripley report but about the same relationship between import and export price elasticities. Beenstock and Minford (1976) found somewhat higher elasticities than the others, but the relationship is again the same. These regularities imply that any given relative price changes would be expected to yield very similar trade balance effects in all of these studies.

In order to gauge the overall effect of a change in exchange rates on merchandise trade, it is useful to examine simulations with the World Trade Model (WTM), a model based directly on the Spencer and Deppler-Ripley studies. For this purpose, the model simulated the effects of a 20 percent depreciation of the U.S. dollar against all industrial country currencies except for the Canadian dollar; this depreciation occurs in the second semester of 1985 and in broad terms approximates the actual behavior of exchange rates at that time. Such a depreciation would be expected to have several effects. Consequently, the WTM is simulated with the following assumptions. First, for the reasons outlined above, the dollar price of primary commodities would be expected to rise. Research on the magnitude of this direct pass-through effect to commodity prices is quite limited; the WTM simulation uses a rule of thumb, based on staff estimates, that three fourths of the nominal depreciation feeds through to non-oil commodity prices measured in U.S. dollars.12 Forty percent of this effect is assumed to come in the first six months after the depreciation and the remaining 60 percent in the following six months. This assumption is arbitrary, but it serves to capture the lags introduced by the existence of contracts. In addition, for purposes of the analysis, it has also been assumed that oil prices increase by a like amount. This assumption is imposed in order to maintain relative prices constant, not because it is necessarily believed to be realistic.

Second, a depreciation of the dollar would have temporary upward effects on the rates of price inflation in the United States and Canada, as well as downward effects elsewhere. Higher import prices in the United States and Canada, as well as the increased foreign demand for domestic output that would result from a depreciation, would cause factor prices in these two countries to increase. The more recent of the studies cited in Table 37 (such as Sachs (1985) and Dornbusch and Fischer (1984)) suggest that perhaps one fifth to one fourth of the depreciation would feed through to U.S. prices within two and a half years. Therefore, a 20 percent depreciation of the U.S. and Canadian dollars against the other industrial country currencies is assumed to cause the U.S. price level to be 4 percent higher at the end of 1987 than it otherwise would have been. Wages are assumed to follow prices with a six-month lag.

Given the relative openness of the Canadian economy, and the importance of its economic ties with the United States, it is assumed that price and wage reactions in Canada would be very similar to those in the United States. In the other industrial countries, whose nominal exchange rates would appreciate, price and wage movements would be in the opposite direction. These effects would vary from country to country depending on the closeness of their trading links with the United States and Canada. On average, prices and wages outside of the United States and Canada are estimated to fall by about 4 percent relative to what they would have been had exchange rates been unchanged.

Table 38 presents the results of simulating the WTM for five semesters beginning in the second half of 1985, using the exchange rate, price, and wage assumptions set out above. Rates of growth of domestic demand are held fixed in each country, while rates of growth of GNP are permitted to vary in response to changes in the trade balance. It should be pointed out that the real effective exchange rate change for each country—which is what matters for trade flows in the model—is substantially less than the 20 percent nominal change postulated in the simulation. This is partly because each country conducts a significant part of its trade with countries whose currency has not changed relative to its own (so that nominal effective rates change by less than 20 percent), and partly because the induced changes in costs and prices offset part of the shift in competitiveness resulting from a nominal exchange rate change.13

Table 38.

Estimated Changes in Merchandise Trade Balances in Response to a 20 Percent Nominal Depreciation of the U.S. and Canadian Dollars1

(In billions of U.S. dollars, annual rates)

article image

Based on World Trade Model simulations, as described in the text. Changes in effective exchange rates are smaller than 20 percent. A positive value indicates a reduction in a deficit or an increase in a surplus relative to the baseline projection.

France, the Federal Republic of Germany, Italy, Japan, and the United Kingdom.

The change in relative prices causes U.S. and Canadian export volumes to rise and import volumes to fall, and the converse in the other industrial countries. In nominal terms, the U.S. trade deficit is estimated to fall relative to the baseline path only after a J-curve effect in the first half year; the decline reaches $19 billion at an annual rate in the third semester after the depreciation and then stabilizes close to that level. By the end of two and a half years of simulation, relative price movements have offset about 40 percent of the nominal depreciation; the dollar has depreciated, in real effective terms, by about 9 percent. Thus every 1 percent change in the real effective exchange rate has been associated with about a $2 billion strengthening of the U.S. trade balance. The other industrial countries also show a strengthening of their trade balances, owing mainly to J-curve effects but also because changes in trade balances are valued in U.S. dollars, not in local currency. In the simulation this effect persists into the second six months following the exchange rate change, but by the third semester the volume effects are strong enough to lead to a reduction in the aggregate trade balance of Japan and the major European countries.

Price effects on the imports of agricultural and manufactured goods are estimated in the WTM to be essentially complete after a year and a half, but the effects on the imports of raw materials, and especially fuels, are estimated to take longer. Similarly, on the export side, a portion of the response, which varies greatly from country to country, is estimated to occur after the end of the simulation period examined here.

A fall in the external value of the U.S. dollar would also exert significant effects on service account balances in industrial countries. Travel and tourism deficits would decline in the United States and Canada, while in the other industrial countries they would move in the opposite direction, at least in local currency terms; the exchange rate sensitivity of these types of service flows is particularly difficult to judge. In the short run, income flows from portfolio investments, which are largely dollar denominated, will be sensitive to a change in the interest rate but not very responsive to a change in the exchange rate. On the other hand, income from direct investment flows will be sensitive to exchange rate changes even in the short run. Over the longer run, cumulated changes in the current account brought on by a dollar depreciation will lead to further shifts in investment income that will reinforce the changes in trade balances. Other service items, such as shipping and insurance, would be expected to move in line with merchandise trade. Consequently, the total effect on current account balances would be substantially larger than the merchandise trade figures presented in Table 38.

A model of U.S. current account transactions recently developed by Dunaway (1985) suggests that a real depreciation of the magnitude discussed above would lead to a strengthening of the U.S. current account balance by roughly $34 billion after two years. Of this, about $8 billion would come from a rise in net receipts from service transactions, largely reflecting the impact of the depreciation of the dollar on U.S. direct investment income receipts.

Effects on Output Growth

To the extent that a dollar depreciation changes current account positions, it will also have a direct effect on GNP growth rates, even under the assumption of unchanged domestic demand in the industrial countries. That is, with this assumption the change in GNP growth rates is attributable entirely to changes in trade volumes.14 An increase in the growth of export volumes would raise a country’s real GNP growth rate, while faster growth in import volumes would decrease it.

The results of the WTM simulation discussed above can be used to calculate the magnitude of this effect. Because this is a partial-equilibrium calculation, it should be regarded as indicative of the minimum range of these effects. During the six-month period when the 20 percent dollar depreciation is assumed to occur, the net positive effect on the real trade balance of the United States is estimated to be quite small: about ⅕ of 1 percentage point of GNP. In the other major industrial countries, the real trade balance is weaker, on average, by ⅓ of 1 percent of GNP. In one year’s time, volume adjustments would cause U.S. real output to be about 1 percent greater than it would have been in the absence of a depreciation. Output in the other major industrial countries would be, on average, 0.9 percent lower. Subsequently, these growth rate effects would taper off; trend growth rates would be little affected, although the level of output would be higher in North America and lower elsewhere.

Effects Measured with Large Models

The empirical studies examined in the preceding subsection focused on single-equation and other partial-equilibrium models. These studies are useful for measuring and evaluating specific relationships that are important in the transmission of exchange rate effects. However, they do not enable one to measure overall effects very clearly; a number of indirect channels have been ignored that might interact in complex ways.

A full general equilibrium model is required to quantify the role of these various channels in the context of the overall effect of exchange rate movements. For example, exchange rate changes will alter relative prices and thus net exports; this change in aggregate demand will exert pressure on the general price level. Similarly, a general equilibrium model is needed in order to analyze the effects of an exchange rate change on aggregate supply, and hence price changes, as production costs respond to an exchange rate change.

Simulation evidence on the effects of exchange rate changes that arise from exogenous portfolio shifts is rather sparse. There are, however, at least three published studies dealing with three different models.15 Kling (1985) has simulated and analyzed such a shock using two models maintained by the U.S. Board of Governors of the Federal Reserve Sytem: the MIT-Penn-SSRC (MPS) model and the Multicountry Model (MCM). The MPS model is a large-scale U.S. macro model with an abbreviated international sector, and the MCM is a large-scale international model that focuses on international linkages.16

Kling’s simulations imply that a depreciation of the U.S. dollar will lead to a rise in the interest rate on U.S. dollar-denominated assets relative to those on foreign-currency denominated assets. Dollar prices in the United States also rise relative to local-currency prices in the other industrial countries. Conceptually, the output effect under these conditions is indeterminate. Output will rise as a result of the effects of the depreciation on the current account balance, but this effect could be offset by liquidity and aggregate supply effects, as discussed in the next section. As it turns out, the MCM conforms to the “standard” case—output rises—while the MPS shows a decline in output.17 Kling concludes, on the basis of a simplified nine-equation model, that the truth lies somewhere between no change and the MCM results; that is, the depreciation of the dollar would result in an increase in U.S. output, but by a smaller amount than indicated by the MCM simulation. Table 39 presents a summary of Kling’s findings.

Table 39.

Effects of a Portfolio Shift in the MCM and MPS Models Yielding a 10 Percent Nominal Depreciation of the U.S. Dollar over Three Years

(Deviations from control)

article image
Source: Based on Kling (1985), p. 20.

Haas and Symansky (1983) analyze sterilized intervention shocks with a version of the MCM that is similar, but not identical, to the version used by Kling.18 Sterilized intervention by the authorities is analytically equivalent to a portfolio shift by private wealth holders; in both cases the prices of financial assets must adjust to balance the altered composition of demand for home- and foreign-currency bonds. Haas and Symansky report on four simulations; a $10 billion purchase of U.S. dollar-denominated assets in Canada, in the Federal Republic of Germany, in Japan, and in the United Kingdom.19 The simulations suggest that the flow magnitudes required to achieve a significant change in the exchange rate are substantial. The $10 billion shock gives rise to impact changes in exchange rates ranging from 2 percent in Germany to 5 percent in the United Kingdom. By the fourth quarter of the simulations, the nominal exchange rate effects were essentially zero. The price effects, owing to the relatively small movement in the exchange rate, were also small—less than 0.3 percent—as were the output and interest rate effects.

Sachs (1985) has also simulated a portfolio shift, using a relatively small but fully articulated multi-country model. This model, in contrast to the MPS and MCM models, employs forward-looking expectations on the exchange rate. Sachs’ simulations, based on a 27.5 percent depreciation of the nominal effective value of the U.S. dollar brought on by a portfolio shift, show domestic output initially expanding (as the relative price effect dominates the interest rate effect) and then contracting as the reverse happens. Table 40 summarizes these results.

Table 40.

Effects of a Portfolio Shift in Sachs’ Model that Yields a 27.5 Percent Impact Nominal Depreciation

article image
Source: Based on Sachs (1985), Table 11, p. 182.

Deviations from the long-run equilibrium real exchange rate in percent.

A negative number indicates an increase in the gap.

The conclusions of this section, in which an exchange rate change is assumed to result from an exogenous shift in private market preferences regarding the desired currency composition of portfolios, can be summarized as follows. With respect to prices, recent work with small models suggests that, ceteris paribus, a 20 percent change in a country’s nominal effective exchange rate would cause the GNP deflator to rise by 4–5 percent over a two- to four-year period. Virtually all of these estimates, however, pertain solely to the U.S. dollar; in view of the relatively greater openness of most other economies, the figure would be expected to be significantly higher in other countries. With respect to trade balances, available empirical work suggests that a 20 percent depreciation of the U.S. and Canadian dollars against the currencies of the other industrial countries, again in a partial equilibrium context, would result in a strengthening of the balance on merchandise trade of about $20 billion, at annual rates, in the United States and Canada combined and a decrease of about $14 billion in the other major industrial countries after two and a half years. During the first year, there would be a temporary increase in the growth rate of GNP in the United States of about 1 percent and a corresponding decrease in the GNP growth rate in Japan and the major European economies of nearly the same amount. The simulations of a portfolio shift in several general equilibrium models show that, because interest rate and exchange rate effects work in opposite directions, the output effects are indeterminate; the results depend critically on the parameters used and the expectations process postulated.

Effects of Monetary and Fiscal Policies and Supply Shocks

Factors other than shifts in the preferences of investors have also had a systematic influence over the pattern of exchange rates. Some of the more important of these factors—shifts in monetary or fiscal policies and changes in the relationships determining the supply of goods and services—are discussed in this section. It will be seen that the relationship between exchange rate movements and other variables such as output and inflation rates depends critically on the source of the change.

One problem that makes the analysis of this section rather conjectural is that most empirical models are quite limited in what they reveal about the effects of monetary and fiscal policies on exchange rates. The basic reason for this limitation is that the effect of any disturbance on exchange rates depends critically on how it affects market participants’ expectations about the course of relevant variables such as interest rates and inflation rates. In practice, these expectational effects are probably not very stable, so that the observed relationship between policy actions and exchange rates will depend on the particular circumstances under which each policy is enacted. Empirical models, moreover, must rely on highly imperfect representations of the processes by which expectations are formed.

Since 1980, industrial countries have been greatly affected by shifts in saving and investment behavior—especially changes in government dissaving (see Knight and Masson (1986)). These effects are illustrated in the next subsection, which examines the observed relationships between measures of the fiscal stance and real exchange rates. It is more difficult to calculate a single measure of the monetary policy stance, and conventional measures have been distorted in recent years by financial innovations that have shifted the demands for monetary aggregates. However, this subsection does discuss the evidence on the international effects of monetary policies and surveys from global models.

In order to illustrate the empirical importance of specifying the cause of exchange rate movements, Tables 41 and 42 compare the effects of four different disturbances (scaled to generate equal changes in exchange rates): a portfolio shift, a U.S. monetary expansion, a U.S. fiscal contraction, and a combination of fiscal contraction in the United States and fiscal expansion in other industrial countries. The model that is used here (MINIMOD) is a small, two-region global model developed in the Research Department of the Fund.20 At present, there are two versions of this model, which differ in the way market participants are assumed to form expectations regarding the future paths of inflation rates, long-term interest rates, and exchange rates. In the version of the model simulated for Table 41, expectations of these variables are formed adaptively, so that the value expected for next period is a weighted average of the values in the present and the previous periods. The portfolio shift is assumed to involve a permanent increase—at the initial exchange rate—in the required rate of return on U.S. dollar assets by 5 percentage points, for instance, because of an increase in the perceived risk of those assets relative to assets denominated in foreign currencies. Because of the resulting depreciation of the dollar, the ex post rise in U.S. interest rates relative to foreign rates is estimated to be much smaller—2.4 percent after five years—than the ex ante shift.

Table 41.

Comparison of Cumulative Effects of Different Shocks, Assuming Adaptive Expectations

(Percentage deviations from baseline)

article image
Source: Based on the quarterly MINIMOD model; see Haas and Masson (1986). Figures reported are averages of the four quarters. Variables are: the effective exchange rate (E), narrow money (M1), real government spending (GOV), real gross national product (GNP), GNP price deflator (PGNP), current account balance (CA), short-term interest rate (RS). For the exchange rate, a positive number indicates appreciation.

Deviations from baseline, as ratio to baseline GNP.

Deviations from baseline, in billions of U.S. dollars.

Deviations from baseline, in percentage points.

A permanent increase in the required rate of return on dollar assets by 5 percentage points.

Scaled to produce the same exchange rate change as the portfolio shift.

Table 42.

Comparison of Cumulative Effects of Different Shocks, Assuming Rational Expectations

(Percentage deviations from baseline)

article image
Note: For notes, see Table 41. The simulations use the same exogenous values as in that table. Rational expectations are defined here as perfect foresight of the relevant variables.

The results presented in the first panel of Table 41 suggest that such a shift would produce a gradual depreciation of the dollar, amounting to about 8 percent after five years, a gradual rise in short-term U.S. interest rates, and a small fall in interest rates abroad. The output and price effects are relatively modest: a small increase in activity and the GNP price deflator in the United States, and small declines in activity and prices in other countries. The U.S. current account weakens at first but then improves as the dollar’s depreciation stimulates the volume of U.S. net exports.

For the second panel of Table 41, it is supposed that the portfolio shift out of U.S. dollar assets does not take place, but that instead the U.S. monetary authorities ease monetary policy so as to increase the money supply by enough to cause a dollar depreciation of the same magnitude. This monetary policy shift—an 8.3 percent rise in the stock of money, spread over five years—amounts to an average increase in the rate of monetary growth of 1.6 percent a year, although the initial shock is much larger. Compared with the first panel of Table 41, there are a number of differences in the outcome for other variables. In the case of monetary expansion, interest rates in the U.S. fall sharply instead of rising. The boost to U.S. economic activity is much larger here than in the portfolio-shift case, and so is the increase in U.S. prices. Consistent with this, the negative current account effects for the U.S. are larger and last longer than in panel 1. The rest of the world’s activity, prices, and interest rates are affected relatively little by this monetary shock. Activity initially rises and then falls, reflecting corresponding movements in the current account balance, while prices and interest rates decline throughout the simulation period.

Panel 3 considers the same exchange rate effect brought about by a decrease in U.S. real federal government expenditure, holding the money supply unchanged both in the United States and abroad. The simulated decline in government spending, equivalent to roughly 4 percent of U.S. GNP, is commensurate with the increase that took place in the United States from 1979 to 1983, when federal government expenditure rose from 21.9 percent of GNP to 26.4 percent. It is, however, unrealistic to suppose either that such a large decrease could take place in one year and be sustained thereafter, rather than be spread over several years, or that such a sudden decrease would be required in order to generate the depreciation shown in the table. As will be seen below, in the discussion of Table 42, different assumptions about expectations would generate the same exchange rate movement with a much smaller decline in government spending.

In response to this hypothetical decline, GNP in the United States is sharply lower in the first two years—rather than higher as in panels 1 and 2. Similarly, the U.S. GNP deflator falls rather than rises, and the current account improves by a substantial amount—more than $50 billion by year five. Effects on the rest of the world are also larger in this case: the fall in aggregate demand in the United States, adding to the contractionary effect abroad of the appreciation of other currencies against the U.S. dollar, produces a fall in rest-of-world activity relative to baseline of 1.5 percent after five years and a similar decline in the level of the GNP deflator.

Panel 4 considers a sustained decline in government spending in the United States—arbitrarily chosen to be 3 percent of U.S. GNP—and a fiscal expansion in the rest of the world by an amount sufficient to bring about the same dollar depreciation as in the portfolio shift simulation. Government spending abroad rises by roughly 2.5 percent of GNP in the first year, but subsequently declines to a level 1 percent of GNP higher than in the baseline. (However, the point made above concerning sudden changes in fiscal stance also applies here.) In this scenario, output abroad rises in the first year by roughly the amount of the U.S. decline, rather than falling. The U.S. current account improves but by somewhat less than in panel 3.

The effects of policies will depend on the way expectations are formed, as is illustrated by comparison of Tables 41 and 42. In Table 42, the same changes were simulated as in Table 41: a portfolio shift out of U.S. dollars, a U.S. monetary expansion, a U.S. fiscal contraction, and a combination of U.S. fiscal contraction and fiscal expansion abroad—all equal in magnitude to those in Table 41—but here it is assumed that private market participants correctly anticipate the future paths of the relevant variables and take into account the structure of the model in forming their expectations of endogenous variables. Reflecting common usage, this is referred to here as “rational expectations,” even though it is rational only on the assumption that the model is correct, and in the absence of uncertainty concerning future policies.

The qualitative conclusions concerning the relationship between the cause of the exchange rate movement and the effects on other variables remain unchanged; however, the size of exchange rate movements is much larger on impact in Table 42 than in Table 41, while the initial output effects are smaller. In all four panels of Table 42, the exchange rate overshoots in response to an exogenous change, and then gradually moves back in the direction of the baseline path. The reason for this can be seen when considering the portfolio shift. When adjusted for exchange rate movements, returns on dollar assets must be 5 percentage points higher than in the baseline; since the nominal interest differential does not move in favor of U.S. assets by this full amount, some of the expected return must come from an expected appreciation of the dollar. Under rational expectations with perfect foresight, this can occur only if the dollar depreciates enough initially that it subsequently is expected to appreciate—and actually does so. Under adaptive expectations, because the exchange rate expected for next period reflects lagged exchange rate movements, it is continually below its current value, so the dollar is continually expected to appreciate even though, in Table 41, it in fact depreciates. As a result of the larger exchange rate movements under rational expectations (and also larger declines in long-term interest rates, which are not reported in the table), U.S. output decreases less on impact in response to fiscal contraction, and output effects are no longer negative after five years.

In reality, individuals do not have perfect foresight, and they must form their expectations in an environment of uncertainty. Nevertheless, they are likely to use what incomplete knowledge they may have of the likelihood of future events. The world is thus not likely to operate in accordance with either adaptive or rational expectations. In addition, it may be difficult to distinguish between the effects of policy shocks and the portfolio preferences of investors, because the certainty with which expectations of future policy actions are held is affected by the policies themselves. In particular, announcements of future policies may not be credible unless the authorities have adhered to past policy commitments. Hence, policy actions may induce a change in the attractiveness of the liabilities of the government concerned and thus may affect the portfolio preferences of investors. For example, a monetary expansion, even though temporary, may lead to doubt as to the commitment of the authorities to fight inflation and hence lead to a shift away from financial assets denominated in that currency.

In addition, confidence effects that are not captured by the simulations reported above may be important when fiscal contraction occurs in the context of an initial fiscal stance that is clearly unsustainable. In such a situation, a fiscal contraction may not bring about as large negative output effects as those reported for the United States in panel 3 of Tables 41 and 42. Cuts in government expenditure in these circumstances may stimulate real investment and increase foreign demand for the country’s assets, or at least not decrease them to as great an extent as implied in those tables. Consequently, economic activity may not be as adversely affected, nor the exchange rate depreciate as much as in Tables 41 and 42.

Effects of Monetary Policies

The standard theory of the effects of monetary expansion on exchange rates can be stated quite simply. A sustained increase in the rate of growth of the money supply is expected to lead to a sustained rate of depreciation in the nominal exchange rate and to at least a temporary depreciation in the country’s real exchange rate. In the short run, the direct effect of monetary expansion will be a reduction in interest rates (reflecting the increase in liquidity in the economy). This reduction will make domestic securities less attractive relative to those denominated in other country’s currencies; investors will shift assets in favor of foreign securities, inducing a depreciation in the exchange rate.

Because goods prices and wage rates are slower to adjust to shocks such as a change in monetary growth than are financial variables, the country’s price level will temporarily rise by less than the depreciation of the nominal exchange rate. Thus the real exchange rate will tend to depreciate as well. Correspondingly, to the extent that there are unemployed resources in the economy, monetary expansion will lead to some temporary increase in output. Over time, however, domestic inflationary pressures will drive the real exchange rate back toward its initial level.21 A temporary monetary expansion (that is, an increase in the level but not in the sustained rate of growth of money) will have somewhat different effects on the exchange rate. If a monetary expansion is followed by a contraction, perhaps because the central bank initially accommodates a shock to the economy but then returns to its targeted path for the money supply, then there is no reason to expect a long-run change in the nominal exchange rate after the economy has reached a new equilibrium. In this case, the exchange rate may depreciate initially and then appreciate, as interest rates temporarily fall, then rise above their initial levels, to return eventually to those levels.

Given that the ultimate effects on exchange rates and other variables depend in important ways on the subsequent path of the money supply, individuals’ expectations of the future stance of policy will clearly be important for the impact effects of monetary policy changes. These expectations are necessarily formed in a context of uncertainty concerning the intentions of the authorities, even if there are public announcements of intended policy actions. In many circumstances, there is likely to be some skepticism concerning the ability and willingness of the authorities to stick with their announced policy. In such circumstances, a monetary expansion may have quite different effects from those in a situation where the authorities have a high degree of credibility; in the latter case, a temporary acceleration of monetary growth, accompanied by an announcement that the underlying stance of policy had not changed, might not lead to a large depreciation of the exchange rate.

In light of the importance of expectations in the transmission of the effects of monetary policy, it is not surprising that simple inspection of data on monetary growth and exchange rates for the large industrial countries does not generally reveal a stable relationship between these variables.22 Furthermore, most single-equation econometric studies have found little or no relationship. In contrast, simulations with large-scale models, which take account of a complete system of simultaneous macroeconomic relationships, offer more hope for explaining the behavior of exchange rates. In general, large models find that changes in monetary growth generate movements in exchange rates that accord reasonably well with the theoretical description outlined above. It should be noted, however, that even these models have made large errors in predicting recent exchange rate movements.

Table 43 presents the results of simulations of several national and linked multicountry models, taken from various published sources. In all cases, the policy experiment was a monetary expansion sufficient to bring about a decrease in a domestic policy-controlled short-term interest rate by 1 percentage point relative to a baseline solution to the model, while other countries kept their interest rates unchanged.23 The simulated effects on real output, a domestic price deflator, and the foreign currency value of the domestic currency, as percentage deviations from their baseline values, are reported.

Table 43.

Simulated Effects in Home Country of a 1 Percentage Point Decrease in a Policy-Controlled Domestic Interest Rate

(Percentage deviations from baseline)

article image

Source: Chan-Lee and Kato (1984), Table 10. Variables are real GNP, the GNP deflator (PGNP), and the exchange rate (with a positive number indicating appreciation).

Obtained from the Federal Reserve Board early in 1986; simulations were performed using the then-standard version of the MCM. The four- to six-month commercial paper rate was decreased in the case of the United States, the call money rate for Japan, and the three-month Treasury bill rate for Germany. PA is the absorption deflator.

Source: Helliwell and Padmore (1985), Table 4.1. Exchange rates were recalculated so that an increase indicates appreciation.

Obtained by communication with OECD early in 1986; simulations were performed using the then-standard version of Interlink. PGDP is the gross domestic product deflator.

The drop in interest rates produces a depreciation of the exchange rate and a rise in both real output and the price level in each of the models, but the magnitudes differ considerably. The Japanese Economic Planning Agency (EPA) model predicts a depreciation of the deutsche mark against the dollar of about 8 percent after four years, while the other linked world models produce a depreciation of less than half that amount. The simulated increases in output also differ greatly, ranging from 2.7 percent (after four years) in Japan with the MCM down to 0.1 percent in the United States with the EPA model. However, output generally rises by a little less than 1 percent in response to a 1 percentage point decrease in the country’s short-term interest rate.

Over at least the first three or four years of all the simulations, the nominal exchange rate depreciation exceeds the rise in prices, leading to a real exchange rate depreciation as well. The size of movements in output, prices, and exchange rates differs widely across models and across countries, however. To some extent, this may reflect differences in the definitions of the variables and in how the simulations are performed. More important are major differences in the structure and parameters of models; indeed, different versions of the same model, estimated over different data periods, can give very different results, especially for the magnitude of exchange rate changes.

The models represented in Table 43 explicitly or implicitly embody the assumptions that expectations of exchange rates and other financial variables adapt gradually to observed values. If the nature of the monetary policy change is recognized by private market participants, for instance because it has been announced by a credible policy authority, and if its effects on the economy are properly understood, then the results of the policy change can be quite different, as comparison of Tables 41 and 42 has shown. With perfect foresight (“rational expectations”), the exchange rate tends to overshoot its long-run level on impact, and then gradually appreciates, while under adaptive expectations, exchange rate depreciation is modest initially, but the dollar depreciates steadily thereafter.

An implication of the sensitivity of the effects of monetary policy to different expectational assumptions is that shifts in expectations themselves can have large effects on exchange markets and indirectly on other variables. Furthermore, since there may be considerable uncertainty concerning the intentions of the monetary authorities, changes in confidence, statements by public officials, and the degree of credibility of announced policy may make a great difference to the way monetary policy affects the economy.

The direction of transmission of monetary policy changes onto economic activity in other countries is in principle ambiguous under floating exchange rates. A monetary expansion increases income at home, which in turn raises the demand for foreign goods; however, the resulting currency depreciation will tend to limit this rise in demand. The exchange rate movement will tend to weaken current account balances abroad, which would reduce output growth in those countries; on the other hand, it would also reduce foreign prices in local-currency terms, which would raise real balances abroad and thereby tend to raise output. There also are linkages through aggregate supply that arise through changes in the prices of intermediate inputs that are utilized in production and through the sensitivity of real wages to exchange rate changes.

Originating with the work of Mundell (1962) and Fleming (1962), a conventional wisdom was established on the relative importance of some of these different effects and the exchange rate implications of various policy changes. In the Mundell-Fleming view, an expansionary monetary policy in a large country is viewed as giving rise to a depreciation of that country’s currency and a fall in output abroad. This conclusion depends on particular assumptions about the form of policy changes, the behavior of expectations, and the role of competitiveness effects in the international transmission process. The conclusion that an expansionary monetary policy in a large country will contract output abroad depends crucially on an implied deterioration in competitiveness abroad outweighing any positive effects that might arise either from increased liquidity abroad or from higher activity in the large country spilling over into higher demand for other countries’ exports.

Negative transmission effects (that is, monetary expansion at home promoting contraction abroad) depend also on a relatively small role being assigned to supply effects. As noted by Argy and Salop (1979) and by Bruno and Sachs (1985), in a situation in which wage earners attempt to hold constant the real wage measured in terms of a bundle of consumption goods (the consumption wage), while firms base their hiring decisions on the wage measured in terms of their output price (the product wage), monetary expansion in a large country can raise output abroad by reducing the product wage and hence increasing employment. The mechanism by which this occurs is an appreciation of foreign currencies, which increases the purchasing power of a given nominal wage in foreign countries, thus reducing any upward pressure on nominal wages. As a result, with the price of output in these countries not directly affected by the appreciation, increasing employment becomes profitable for firms there.

Empirical macroeconomic models give differing conclusions concerning the transmission of monetary policy changes. MINIMOD implies negligible transmission effects for U.S. monetary expansion under adaptive expectations, at least for the first five years (Table 41); under rational expectations, the transmission effects are negative, and increasingly so (Table 42). The large multicountry models surveyed here themselves give conflicting answers: the MCM implies negative transmission of U.S. monetary policy changes, but INTERLINK implies positive transmission effects, while the EPA model suggests that the transmission effects of U.S. policy changes are positive on Japan and negative on Germany (Table 44). Simulations of monetary contraction taken by other countries also give mixed results. Thus, it is difficult to draw firm conclusions from empirical studies as to whether a country experiencing appreciation, as a result of monetary expansion in another country, should expect that development to push its output higher or lower.

Table 44.

Simulated Effects on Real GNP in Other Countries of a 1 Percentage Point Decrease in a Policy-Controlled Domestic Interest Rate

(Percentage deviations from baseline)

article image
Note: For sources of simulation results, see notes to Table 43. Home-country effects are reproduced from Table 43.

Effects of Fiscal Policies

Fiscal policies have a less straightforward effect on the exchange rate than do monetary policies, but it is clear that in some circumstances they have had a large and persistent influence on the levels of both nominal and real exchange rates and on other macroeconomic variables. It seems likely that decreases in government spending can in some cases be associated with exchange rate appreciation, and in others with exchange rate depreciation, depending on accompanying monetary policies, the degree of international financial integration, and expectations that the public is led to form about future fiscal and monetary policies. It is thus important to understand more fully the nature of the effects of fiscal policies on exchange rates before examining the effects of the resulting exchange rate changes on other variables; obviously, the effects of a depreciation will depend heavily on whether it results from fiscal contraction or from fiscal expansion.

The ambiguity of the fiscal effect on exchange rates has several causes. First, fiscal policies have been accompanied by different monetary policy stances in different countries; it is important to specify how the monetary authorities react to fiscal contraction and to decreases in the financing needs of the government. Second, because a fiscal surplus lowers the stock of government debt it is important to know whether the decrease in debt will lead to lower interest rates on home-country bonds relative to the expected returns on other interest-bearing assets, or whether instead those assets are viewed as perfect substitutes. If the demand by foreigners for interest-bearing claims on the country is not perfectly elastic, then the improvement in the current account may not be offset by an ex ante capital account deterioration. Consequently, a fiscal contraction could lead to appreciation. Finally, because the prospect of large deficits may cause investors to doubt the sustainability of a certain course of tax and expenditure policies, a fiscal contraction may bring about favorable confidence effects that tend to offset the negative impact on activity. The implications of these conflicting forces for the exchange rate are discussed in more detail below.

Accompanying monetary policies are important to fiscal policy effects because interest rates respond to the fiscal change in a manner that depends on the reactions of the monetary authorities. If the central bank does not respond to the decline in nominal demand by decreasing the supply of money, then a decline in government spending will tend to be accompanied by a fall in nominal and real interest rates. As a result, some of the fiscal contraction will be offset by increased private sector demand. The fall in interest rates will also tend to depreciate the exchange rate, further offsetting the demand effects. Another possible reaction on the part of the monetary authorities is to keep short-term interest rates unchanged and therefore to contract the money supply endogenously. In this case the ultimate consequences for prices might be more deflationary, and the exchange rate would also be more likely to appreciate than in the first case.

Chart 14 shows that the relationship between the real exchange rate and a measure of the stance of fiscal policy—a fiscal impulse variable that measures discretionary changes in budget deficits relative to GNP and is calculated relative to other countries’ fiscal impulses—can be diverse. The United States and the Federal Republic of Germany—among others—have generally adhered more closely in recent years to monetary targets, while some smaller European countries have been closer to a policy of maintaining unchanged interest rates. A generally nonaccommodating monetary policy has helped explain why fiscal expansion in the United States since 1981 has been associated with exchange rate appreciation, and fiscal contraction in Germany and Japan generally has been associated with real depreciations. In contrast, fiscal expansion in France and Italy during 1981 was accompanied by weakness of the franc and (to a lesser extent) the lira. Conversely, fiscal contraction in the United Kingdom in the 1979–81 period was accompanied by a tight monetary policy, and the pound sterling appreciated in real terms by over 30 percent (Chart 14).24 Nonetheless, in each of these cases, it is doubtless true that other factors were important in determining actual exchange rate movements; for the United Kingdom and Japan, in particular, oil market developments would have been a major part of the story, as discussed in the next section.

Chart 14.
Chart 14.
Chart 14.

Major Industrial Countries: Relative Fiscal Impulses and Percentage Changes in Real Effective Exchange Rates, 1976–85

Note: The fiscal impulse is a cyclically adjusted indicator of stimulative (if positive) or restrictive (if negative) shifts in government fiscal operations.

The extent to which domestic bonds are good substitutes for foreign bonds is also important (see Sachs and Wyplosz (1986)). In the extreme case of perfect substitutability (that is, perfectly elastic capital flows) the return expected on domestic bonds would have to equal the return expected on foreign bonds, as any positive differential in favor of one of the bonds would be offset by an incipient unlimited increase in the demand for that bond. Therefore, if fiscal contraction were associated with unchanged monetary targets, leading to a decrease in domestic interest rates relative to foreign rates, as discussed above, then the exchange rate would have to depreciate in both nominal and real terms—and perhaps overshoot its eventual equilibrium level—in order to generate the offsetting expected appreciation necessary to equalize expected returns. If, on the other hand, domestic and foreign bonds were not perfect substitutes, then the fact that fiscal contraction decreased the stock of domestic bonds would of itself require an expected return differential in favor of foreign bonds to open up in order to induce investors to hold both bonds. If the fall in domestic interest rates were not sufficient to accomplish this, then the exchange rate would have to appreciate, in order to generate a sufficiently large expected depreciation.

In practice, available empirical evidence for most major industrial countries suggests that bonds that are denominated in different currencies but that otherwise have similar characteristics with respect to risk, maturity, and tax treatment are generally good (though not perfect) substitutes.25 Therefore, it is unlikely that in ordinary circumstances nonaccommodated fiscal contraction would lead to exchange rate appreciation. However, the greater are capital controls and the costs of taking a position in foreign-currency assets, the less substitutable domestic and foreign assets are likely to be. France and Italy have in recent years exercised control over residents’ purchases of foreign financial assets; these controls are relevant to the lack of association between fiscal expansion and real appreciation in those countries (Chart 14), in addition to considerations of monetary accommodation, discussed above.

In some circumstances another factor would come into play, namely confidence concerning the sustainability of fiscal policy (see Masson (1985)). An expansionary fiscal policy might have to be reversed eventually because it led to an accumulation of indebtedness that fed upon itself if real interest rates exceeded the real rate of growth of government revenues. At some point, the government would no longer be able to continue to finance itself through bond issues, and it would have to turn either to a contractionary fiscal policy or to a switch to financing deficits through issuing money. The anticipation of either of these occurrences might lead to a depreciation, not an appreciation, in response even to a nonaccommodated expansionary fiscal policy. In the former case, investors might expect negative effects on future economic activity and the possibility of increased taxation on interest income; in the latter, debt denominated in the domestic currency would be more risky, as an eventual resort to inflationary finance would amount to a partial default on the real value of bonds by the government. Conversely, a contractionary fiscal policy in these circumstances might allay fears of inflationary finance and lead to an exchange rate appreciation.

For a number of reasons, fiscal contraction is more likely to be associated with depreciation in countries with well-developed domestic financial markets and international demand for claims denominated in their own currencies. The extent to which central banks can resist pressures to monetize deficits, thus forcing the government to finance itself through sales of bonds to the private sector or to residents, may be limited in countries where the financial markets and banking system are not well developed. International demand for claims denominated in a country’s currency may be a function of the country’s size; in practice it is mainly the currencies of larger industrial countries—principally the U.S. dollar, the deutsche mark, the Japanese yen, and the pound sterling; and to a lesser extent the French franc, the Dutch guilder, the Canadian dollar, and the Swiss franc—that serve to denominate a significant amount of international borrowing. In consequence, bonds denominated in other currencies are less likely to be close substitutes either for one another or for bonds denominated in the currencies of countries with well-developed international capital markets, and it is in the latter countries that fiscal contraction is most likely to be associated with depreciation.

The effects of a decrease in government spending on output, prices, and interest rates in the home country are likely to be negative, as the decreased stimulus to aggregate demand causes a contraction of economic activity. Lower economic activity puts downward pressure on prices; money demand decreases, so that if the money supply is not decreased interest rates must fall. However, the transmission of fiscal stimulus at home onto foreign activity is less clear. Lower domestic activity decreases the demand for foreign goods as well as domestic goods, but in principle this effect could be offset by liquidity and supply-side considerations, as discussed earlier in the context of monetary policy. Conventional wisdom, originating with Mundell (1962) and Fleming (1962), regards fiscal contraction as leading to a contraction of output abroad—that is, fiscal policy is transmitted positively. However, in a two-country version of the Mundell-Fleming model, as well as in more general models, the sign of the effect is in principle ambiguous.

Simulations of empirical macroeconomic models help to pin down the relative magnitudes of the factors discussed above, except perhaps the question of confidence in the sustainability of policies and the significance of forward-looking expectations. Table 45 presents the simulated responses of several large national and linked multicountry models to an autonomous government expenditure decrease. As expected, output effects are negative and in most cases imply an initial multiplier greater than unity. Only for the United States does fiscal contraction lead to a substantial exchange rate depreciation in all three models.

Table 45.

Simulated Domestic Effects of a Sustained Decline in Real Government Expenditure Equal to 1 Percent of GNP

(Percentage deviations from baseline)

article image

Source: Chan-Lee and Kato (1984), Tables 5 and 6. Results for Germany are “with accommodating monetary policy,” but induced money supply changes are quite small.

Obtained from the Federal Reserve Board early in 1986; simulations were performed with the then-standard version of the MCM, holding money supplies constant (M1 in the United States, M2 in Japan, and central bank money in Germany).

Helliwell and Padmore (1985), Tables 3.1, 3.3, and 3.4.

Obtained by communication with OECD early in 1986; simulations were performed using the then-standard version of Interlink.

For several reasons, only limited credence should be given to these quantitative results. The elasticity of capital flows facing these three countries has probably increased in recent years, to an extent not captured by estimates based on historical data. Furthermore, exchange rate expectations in these models are assumed to follow a simple adaptive scheme; as illustrated in Tables 41 and 42, forward-looking behavior is likely to strengthen pressures for depreciation in response to fiscal contraction. The magnitude of the exchange rate change in fact differs considerably from model to model, with short-run and long-run effects also being quite different. In fact, for Germany and Japan, simulations of the EPA model give one-year and four-year effects with different signs: fiscal contraction depresses output, lowers interest rates, and depreciates the currency in the short run, but in the long run leads to a smaller stock of indebtedness to nonresidents and to an improvement of the services account, which requires a worsening of the trade account. The latter is achieved by a fall of competitiveness via an appreciation of the real exchange rate.

Table 46 presents cross-country effects of fiscal contraction. For these three multicountry models, the Mundell-Fleming predictions are borne out and fiscal policy is transmitted positively. Cross-country effects are of course largest for a contraction that occurs in the United States; however, a fiscal policy change in the Federal Republic of Germany would certainly have important repercussions for other industrial countries in Europe.

Table 46.

Simulated Effects on Real GNP in Other Countries of a Sustained Decline in Real Government Expenditure Equal to 1 Percent of GNP

(Percentage deviations from baseline)

article image
Note: For sources of simulation results, see notes to Table 45. Home-country effects are reproduced from Table 45.

The MINIMOD simulations reported in panel 3 of Tables 41 and 42 also imply positive transmission effects. It appears that the direct transmission of lower demand through decreased imports more than offsets the positive effect on foreign activity of lower interest rates. However, the timing of the effects on activity at home and abroad of fiscal contraction may be different. In particular, foreign contraction may lag that in the country withdrawing the fiscal stimulus, as decreased imports are due to the indirect response of the domestic private sector rather than directly to lower government spending abroad (see Table 42).

A caveat to the simulation results presented here—particularly those that relate to the effects of changes in policies—is that none of these simulations makes adequate allowance for credibility effects. A credible and sustained change in fiscal policy can, in practice, have quite different implications from those of a policy change that is not believed likely to be sustained. Hence, for example, a credible reduction in a country’s fiscal deficit that is widely expected to be maintained over a number of years might be expected to have a larger effect in the short run in lowering long-term interest rates, and negative output effects might thereby be attenuated. Effects on the exchange rate may be lessened, and conceivably fiscal contraction could lead to appreciation. Notwithstanding these points, it is generally difficult to take account of credibility effects in model simulations. Instead, care and judgment must be used in assessing results; where it is thought necessary, some adjustment can be made.

Effects of Supply Shocks

Shocks to the supply of goods and services may also have substantial and persistent effects on the pattern of exchange rates. These shocks may take a variety of forms such as changes in the technology of production that lead to changes in productivity, for given inputs of factors; exogenous changes in the real prices of inputs to production; or changes in endowments of fixed or quasi-fixed factors of production. A decrease in trend productivity seems to have occurred in many industrial countries in the first part of the decade of the 1970s, but the reasons for this shift are still unclear (see Bruno and Sachs (1985) for a discussion of some of the issues). Factor prices may change for a number of reasons; for instance, wage bargaining may succeed in raising the real wage, at least temporarily, and the actions of producing countries may raise the prices of primary commodities. Finally, since factors of production are not fully mobile either within a country or internationally, an event that causes a major change in the supply of those factors available for domestic production or a major change in the demand for domestic goods would have substantial real effects.

This discussion focuses on a shock that has been very important in the last 15 years, namely changes in the world price of energy. In 1973–74 and in 1979–80 there were major increases in the price of petroleum, and more recently real oil prices have declined considerably. These changes have meant that the real price of energy products has changed dramatically in all industrial countries. Because of the importance of energy use by industries and households, considerable changes in economic activity have resulted, both sectorally and for the whole economy, as well as changes in overall price levels. Because of differences in oil dependence, an oil price change can be also expected to affect the exchange rates among industrial countries.

The analysis of oil price effects is difficult in most macroeconomic models because until recently those models did not have full-fledged oil sectors, nor did they take account of energy as a separate factor of production. The impact of an energy price increase will depend on the importance of energy in production as well as in consumption and on the possibilities of substituting other factors for it. In addition, effects of a price increase on the domestic supply of energy will be important in gauging the different movements in industrial countries’ trade balances and exchange rates, given quite different endowments of energy resources. Both demand and supply effects should take into account inter-fuel substitution, which itself will depend on a country’s natural gas resources and electric generating capacity since there are limits to the tradability of these forms of energy. Finally, the investment preferences of oil exporters and their eventual spending propensities will influence the exchange rates among the currencies of industrial countries.

Given these complications, it is natural that an assessment of effects of oil price increases should supplement existing models with outside sources of information. This was the procedure used by McGuirk (1983); for that study, estimates of the current account effects of the oil price rises that occurred from 1972 to 1980 on current account balances projected for 1985 were first obtained from other sources, and the MERM was then solved to get an estimate of the real exchange rate levels that would be necessary to offset these effects. The results there suggested that the real exchange rate movements consistent with the real oil price increase that occurred from 1972 to 1980 might be quite large, with net energy exporters needing large real appreciations (23 percent for the United Kingdom and 12 percent for Canada) and large energy importers with no domestic energy resources needing large real depreciations (28 percent for Japan).

Helliwell and Padmore (1985) survey some results from linked multicountry models. Both the Japanese EPA model and the MCM predict a large depreciation of the yen and a modest appreciation of the U.S. dollar in response to an oil price increase, but they disagree on impacts on other currencies, and it is not clear what features of the models lead to these divergent results. These simulations are subject to the caveats mentioned above concerning the inadequate treatment in most large models of energy linkages.

Though effects of oil price changes on exchange rates are subject to dispute, and clearly are country specific, there seems to be a consensus concerning their effects on output and inflation in industrial countries. Oil price increases will raise costs of production and hence prices, and also tend to depress aggregate supply. To the extent that they transfer income to countries with higher propensities to save (which was the case at the time of the first oil shock, but is less true now), the oil price increases will tend to reduce world aggregate demand as well. Though standardized model simulations are not readily available, Helliwell and Padmore (1985, p. 1135) conclude that “a 10 percent increase in world oil prices lowers real GNP in a typical industrial country by between 0.5 and 1 percent in the second year and increases the consumer price by somewhat more.” These stagflationary effects pose a policy dilemma for the industrial countries concerned: on the one hand, it may be desirable to limit output losses by expansionary monetary or fiscal policies, but on the other hand, a concern for inflationary pressures suggests that macroeconomic policies should resist the increase in domestic prices in order to avoid a wage-price spiral.

Implications of Exchange Rate Movements for Policy Implementation

The evaluation of policy reactions to the economic changes associated with exchange rate movements cannot be considered in a vacuum; it depends critically on the conditions under which an exchange rate change occurs. In particular, a central concern is to take account of the medium-term framework in which policies are now implemented. How to modify current policies—perhaps by temporarily raising or lowering targeted growth rates for monetary aggregates or by raising or lowering tax rates—without losing the discipline and the credibility that are necessary to keep the economy on a desired path over the medium term involves political questions that are beyond the scope of this review. Those issues should nonetheless be kept in mind in evaluating the economic issues that are being discussed.

The question of appropriate policy reactions hinges on a number of specific issues about which it is very difficult to generalize. First, what were the conditions prevailing before exchange rates changed? If exchange rates were initially thought to be appropriate and sustainable over the medium term, the response to a movement away from those levels would be quite different from the response to a movement toward sustainable levels. Nonetheless, it must be recognized that even a movement toward sustainable levels may pose difficulties for the formulation of policies in at least some of the countries involved.

A second issue is whether the exchange rate movement itself resulted from a policy shift or from a shift in private market behavior. As described above, the economic consequences of the exchange rate movement depend critically on its source. And third, the timing of the change may be quite important. Short-term fluctuations in exchange rates may be disruptive, but they do not in general call for changes in economic policies. The postwar experience with short-term adjustments to monetary and fiscal policies has not led to favorable conclusions; on the contrary, the importance of maintaining a medium-term perspective on policy settings is now widely accepted. Nonetheless, large sudden movements in exchange rates may be more disruptive than slower adjustments and may require a relatively large, though possibly temporary, policy response. Each of these issues is discussed further below.

Another point to be kept in mind is that a policy reaction to a change in conditions brought about by (or associated with) exchange rate movements is not the same as using policy to stabilize exchange rates. In a sense, one could regard a fixed exchange rate system—such as the European Monetary System—or a system of “target zones” for exchange rates as a means of systematizing policy responses to exchange rate changes, or rather to changes in conditions that give rise to exchange rate movements. Nonetheless, even if the authorities have no exchange rate objective at all, they are unlikely to be indifferent to shifts in output, prices, or current account balances that are associated with these changes. For example, the rate of monetary growth that is appropriate under initial conditions may no longer be appropriate after an exchange rate depreciation alters the path of the current account balance and (at least temporarily) both output and inflation.

A final general point to be made is that the policy reactions considered in this section are limited to shifts in monetary or fiscal policies. In some circumstances, sterilized intervention might be considered as an additional policy tool, especially if the exchange rate movements were thought to be temporary and unrelated to shifts in underlying economic and financial conditions. However, the focus of this paper is on the consequences of large sustained movements in exchange rates, and it is unlikely in this context that sterilized intervention could play more than a very limited subsidiary role (see Jurgenson (1983)). Other policy actions, including those designed to improve labor productivity or otherwise improve the functioning of the supply side of the economy, would not generally be appropriate responses to the kinds of shocks being discussed here. Such policies may well be warranted, but their desirability would probably not be affected one way or the other by movements in exchange rates.

In order to analyze the main relationships that are of interest here, this section—like the section on the effects of portfolio shifts—focuses primarily on a single example, rather than trying to cover a wide variety of possible situations. For the central example in this section, it is assumed that the U.S. dollar is initially viewed as being at an unsustainably high level and then depreciates rapidly by 20 percent against all other major currencies except the Canadian dollar.

This example is intended in part to reflect events that occurred during 1985 and the first quarter of 1986. By March 1985, the real effective value of the U.S. dollar was almost 40 percent above its average value for the 1975–84 decade (see Chart 13). Then, over the next 12 months, the dollar depreciated in real terms by close to 30 percent against a weighted average of other industrial-country currencies excluding the Canadian dollar, while appreciating by about 1 percent against the latter. But the example is also intended to reflect the concerns that have been expressed regarding the outlook for 1986 and 1987; because of the large deficit in the U.S. current account, questions have continued to be raised about the sustainability of the dollar exchange rate, even after the depreciation that has already occurred.26

Reactions to a Portfolio Shift

Based on the results described earlier, it is clear that a 20 percent depreciation of the U.S. and Canadian dollars (resulting from a private portfolio shift) would have sizable effects, not only on those two countries but on most other countries as well. It is difficult to summarize those results in terms of precise magnitudes, but the following may be taken as representative conclusions from the simulation studies described above. Output in the North American countries would rise relative to its baseline path, because the direct effect of the increase in the current account balance would be larger than the offsetting effect on domestic demand of the rise in domestic interest rates. The net rise in output might be on the order of 1 percent per annum over a period of one to two years.

On average, output in other industrial countries would decline (relative to baseline) by an amount similar to the rise in North America, but this effect would not be uniform. As a general rule, the largest declines would be expected to occur in countries with the closest trade linkages with the United States and Canada. However, the expected declines in interest rates and in prices (relative to baseline) in these countries could possibly modify this conclusion. Interest rate declines in particular might not be uniform across countries; those countries where financial markets are closely integrated with those in the United States and whose securities are close substitutes for those denominated in dollars would be less likely to experience a sharp decline in interest rates relative to those in North America. In addition, the conduct of monetary policy will differ among countries, with implications for interest rates.

The empirical evidence on the effects of a portfolio shift is not sufficiently detailed to enable one to draw firm conclusions about which countries would experience the largest declines in output corresponding to the increases in North America after this kind of exchange rate shock. Nonetheless, some tentative observations may be made on the basis of the considerations just outlined. Specifically, in view of the high degree of intra-European trade, European countries could be adversely affected to a rather smaller extent than would Japan, which conducts more substantial trade with both the United States and Canada. For example, 32 percent of Japan’s exports in 1983 went to North America, whereas the comparable figure for German exports was only 8 percent. In any event, however, it is clear that for at least some countries, the changes would be large enough to warrant consideration of possible policy responses.

The policy implications of these various effects would clearly depend on whether the impacts on the economy were desired or not. For example, in the North American countries, output growth and interest rates would both be rising, with the strengthening of the current account balance being only partially offset by weakening domestic demand. The authorities might of course be willing to accept this outcome, especially under circumstances where the current account balance was initially quite weak. If, however, it were deemed desirable to attempt to restore the status quo ante with respect to both the level of output and its composition between domestic demand (notably investment) and the external balance, the clearest method would be through sterilized intervention or through monetary restraint. A reduction in the rate of growth of the money supply would tend to reverse the incipient depreciation of the exchange rate. Consequently, the strengthening of the current account balance would also be partially reversed, and the stimulus to output could be completely reversed. In that event, the composition of output would still be different from what it was initially, because domestic interest rates would need to rise in order to induce people to hold the existing stock of dollar-denominated securities. Nonetheless, the domestic/external mix might be less altered than with the portfolio shock alone.27

An alternative course would be to adopt a more restrictive fiscal policy. A reduction in the rate of growth in government spending or an increase in taxes would tend to reduce interest rates and to reduce domestic demand further, but with less crowding out of investment. Even though these changes, on balance, would be likely to accentuate the depreciation of the exchange rate, they also would be likely partially to offset the inflationary consequences of the rise in the current account. The composition of output would inevitably be altered more than it would have been with monetary contraction, but the total could, at least in theory, be brought back to its previous path. This course of action would therefore be preferable in situations where the current account was initially weak but where output growth was sufficiently strong that the principal policy objective was to restrain inflationary pressures.

Fiscal contraction in North America would be a particularly appropriate response in conditions such as those prevailing in 1985. The sharp rise in the fiscal deficits of the United States and Canada in 1982 and 1983, and the continuation of large deficits in 1984 and 1985, had not only contributed to the real appreciation of their currencies but also had led to concerns about the sustainability of the policies producing those deficits. A portfolio-induced depreciation of the exchange rate, to the extent that it helped to reduce the external deficit and thereby to raise output growth at a time when growth was otherwise projected to be relatively slow, would facilitate a tightening of the fiscal stance in these circumstances.

In this example, other countries would have declines in output and in domestic interest rates, so the policy options would in general be symmetric to those for the North American countries. That is, expansion of monetary growth or of fiscal policies could potentially restore the original path of output, with monetary expansion coming closer to restoring the original composition of output (if desired). However, the costs of an easing of policy may be higher than those of a contraction.

Fiscal easing could force the economy onto an undesirable medium-term course. Monetary easing could lead to a loss of confidence regarding the authorities’ commitment to an anti-inflationary policy, which in turn could set in motion inflationary expectations that would again make it more difficult to regain the appropriate policy path once the need for adjustment had passed.28 Therefore, as a practical matter in situations where the credibility of anti-inflationary policies was in doubt, the countries facing an appreciation of their currencies as a result of a portfolio shift might be constrained to accept the consequences and to leave any major policy response to the countries whose currencies were depreciating.

Reactions to Monetary Expansion

It was noted earlier that the macroeconomic effects of a depreciation arising from monetary expansion would be very different from those of a private portfolio shift. However, the differences would be much more apparent in the country undertaking the monetary expansion than they would be abroad. This subsection continues with the example of a 20 percent depreciation in the U.S. and Canadian dollars relative to other industrial-country currencies, but it is now assumed that the depreciation results from an increase in the stock of money in the two North American countries. Those countries would expect to see a much larger temporary stimulus to demand and output than in the previous case, because now domestic demand would be rising and thus reinforcing the stimulus coming from the strengthening of the current account balance. Interest rates would decline, rather than rise, until the inflationary consequences of the monetary expansion fully took hold.

Output in other countries could either rise or fall (relative to baseline), and the empirical studies discussed above do not suggest that this effect would be substantial or that it would persist. Interest rates would decline; in the absence of a portfolio shift, interest rate declines abroad would probably be smaller than those in North America, but this effect depends critically on the nature of market expectations.29 In most countries, it appears that the major effects would be on the composition of output—with an increase in domestic demand (especially interest-sensitive components such as investment) offsetting a decline in the current account—rather than a change in the total.

The appropriate policy responses abroad in the face of a North American monetary expansion are even more complex to summarize than in the previous case, given the ambiguity of the sign of the effect of the currency movement on output. If, as seems likely, the net induced changes in output are small in most countries, then the best policy response in this case might well be to maintain the original stance of both monetary and fiscal policy. After all, if the initial exchange rate configuration was characterized by an “overvalued” dollar, then the shifts in current account positions brought about by a devaluation would have to be accepted as a necessary part of the adjustment process. The stimulus to domestic demand abroad brought about by the monetary expansion in North America would at the same time serve to offset temporarily the deleterious effects on output.

Some countries might nonetheless choose to use such an occasion to ease policies, because the exchange rate movement itself might relieve an important constraint on domestic policy formation. Since the counterpart of the “overvalued” dollar in this example would have been an undervaluation of other countries’ currencies, those countries may have felt the need to maintain domestic interest rates at levels higher than would otherwise have been appropriate in order to limit the exchange market imbalance. Although in principle that objective could have been achieved through fiscal expansion as well as by monetary contraction, the desire to consolidate fiscal positions steadily over the medium term may have prevented the authorities from taking that option. Domestic considerations would thus make a temporary easing of monetary policy desirable in the face of the postulated exchange market developments.

The difficulty with this scenario is that the monetary expansion itself—either in North America or abroad—could lead to an unsustainable outcome. This difficulty would certainly arise if the monetary authorities lacked credibility—that is, if even a temporary monetary expansion were to lead quickly to a rise in prices that offset the real gains. In addition, it would arise in the North American countries if the expansion merely masked the effects of an unsustainable course for fiscal policy. If the overvalued currency arose primarily because of an excessively loose fiscal stance, then monetary policy could not be expected to do more than temporarily relieve the symptoms of the underlying imbalance.

Reactions to Fiscal Contraction

The final example to be examined here concerns the effects of a fiscal contraction in North America that is large enough to bring about a 20 percent depreciation.30 As with a monetary expansion, this policy action would be expected to reduce interest rates, both at home and abroad, to strengthen current account balances in North America, and to weaken those balances in other countries. The consequences for output, however, would be very different, as noted earlier. Output would temporarily decline in the countries implementing fiscal contraction, while the effect on other countries would in principle be ambiguous. The empirical studies discussed above suggested that output would temporarily decline in both the Federal Republic of Germany and Japan, but evidence relating to the effects in other countries is less clear. Most countries would normally experience a decline in output, reflecting the weakening of their current account balances. Some countries, however, might experience a rise in output as a result of the favorable effects of reduced local-currency prices on liquidity and on wage pressures.

The policy options facing these countries are also potentially quite different, depending on whether output rises or falls. For countries whose output falls, the qualitative description of the situation is quite similar to that arising from a portfolio shift: output and interest rates decline, and the current account balance weakens. Hence, the policy options are the same as in that case. Either fiscal easing or monetary easing could help to offset the decline in output; monetary easing might come closer to restoring the original composition of output between domestic demand and the external balance, but fiscal expansion might be preferable in some circumstances. Specifically, if the shift in current account balances were viewed as a desirable outcome, or if it were felt that monetary expansion would bring a greater inflationary risk in spite of the deflationary impact of the exchange rate movement, then fiscal easing would clearly be the preferred policy.

If some countries were to have a rise in output, their policy options would be similar to those produced by monetary expansion in North America. It appears paradoxical that monetary expansion and fiscal contraction could have the same policy implications abroad, but it should be recalled that this conclusion applies only to countries that are affected primarily by the disturbance to financial, rather than trade, linkages. In any event, because these circumstances would probably imply only a very small effect on output, the only reason to modify monetary or fiscal policies in such cases would be if it were felt that inappropriate exchange rates had previously constrained the authorities to follow an undesirably tight monetary policy.

In summary, the policy options created or modified by exchange rate movements would depend on a variety of factors, including the causes of the exchange rate change, the initial conditions, and the nature of the linkages between countries. Even though it is impossible to generalize about policy implications, a number of conclusions may be drawn from this discussion.

First, even if the exchange rate movement is toward more sustainable levels, the implications will not necessarily be favorable for all countries. Notably, countries whose exchange rates had been undervalued may experience lower output growth as the appreciation of their currencies puts downward pressure on their current account positions. Second, exchange rate movements brought about by monetary expansion in one country would be unlikely to have sizable effects on output growth in other countries; other types of shocks, such as private portfolio shifts or fiscal actions, have potentially larger effects. The reason for this dichotomy appears to be that the transmission of monetary policy to other countries involves offsetting effects to a relatively greater degree than with these other shocks. Hence, shifts in monetary policy in one country would not in general appear to justify policy reactions in other countries. Third, although shifts in fiscal policy are likely to affect output in the same direction in the home country and in foreign countries, the degree to which policy constraints are affected in the latter countries may vary greatly, depending on a number of factors. In some cases, the direction of the effect could even be reversed.


This examination of the consequences of large exchange rate shifts among industrial countries has taken account of the fact that these consequences vary substantially depending on the origin of the shift. This topic is, of course, very broad and complex, and it has proved necessary to limit the analysis in a number of ways. The paper has focused primarily on the effects that would normally be expected to occur during the first two or three years following a major exchange rate movement, and it has emphasized the empirical difficulties involved in analyzing the role of exchange rate movements in transmitting the effects of disturbances among countries.

The analysis has drawn on previous empirical studies and some additional simulation exercises performed specifically for this study. This empirical work has made it possible to derive some tentative conclusions about the direction of some effects that are theoretically ambiguous. Nonetheless, it is important to keep in mind that even the most careful empirical work in this area must rely on a number of simplifying assumptions, notably about the formation of expectations. In addition, it is often very difficult in practice to identify the precise causes of a particular exchange rate movement.

The following are some of the major conclusions that may be drawn from this study. Because of the limitations just described, they should be viewed only as a general guide.

1. One way that a major shift in exchange rates could come about would be through a shift in private portfolio preferences. It appears from the evidence that it would take a very large shift to bring about a large (say 20 percent) change in industrial-country exchange rates in the absence of changes in monetary or fiscal policies, and it is not clear whether actual shifts have been large enough to produce changes of that size. Nonetheless, the possibility certainly cannot be ruled out, especially in view of the very large shifts in exchange rates that have taken place during the past several years.

2. A major effect of a shift in portfolio preferences out of a currency would be a rise in prices measured in that currency, and a decline in prices measured in other currencies. However, although one would normally expect import prices to rise (with a lag of up to two years) by the full amount of the depreciation, the rise in the overall price level would be much smaller. Estimates of the size of this pass-through effect cluster around one fifth for the United States (for example, a 4 percent increase in the price level spread over two to three years, following a 20 percent depreciation), but the effect is no doubt rather higher for other countries.

3. A portfolio-induced nominal effective depreciation would also tend to raise interest rates in the home country. The size of this increase and the direction of the effect on interest rates in other countries would depend critically on what happened to the expected rate of depreciation (rather than just the size of the actual depreciation). The evidence suggests that short-term interest rates in the home country could rise by perhaps 2 percentage points in response to a 20 percent depreciation and that rates would decline slightly in other countries.

4. A depreciation induced by a portfolio shift would also tend to raise output growth temporarily in the country experiencing depreciation, but this effect might not be very large, owing to offsetting effects. The current account balance would generally strengthen in real terms, thus raising domestic output; however, the rise in interest rates and in domestic prices would have the opposite effect. The empirical evidence suggests that the positive effect would dominate, but it is less clear on the likely size of the net effect. Output growth in other countries would be expected to decline, probably on average by a magnitude similar to the increase in the home country.

5. An expansionary monetary policy, or a restrictive fiscal policy with unchanged monetary growth, could also lead to a depreciation of that country’s exchange rate. In the case of a fiscal contraction, the direction of the effect on exchange rates is theoretically ambiguous; the empirical evidence is quite clear that a depreciation would result for the United States, and the weight of the evidence for Germany and Japan also leans in that direction.

6. A depreciation arising from monetary expansion would have effects that differ from those of a private portfolio shift in a number of ways. First, prices in the home country (the country undertaking the monetary expansion) would rise to a much greater extent. Over a period of several years, domestic prices should rise by as much as the nominal depreciation of the exchange rate. In the short to medium run, the real exchange rate would depreciate, though by less than in the portfolio-shift case. Output growth would also temporarily rise to a greater extent, but output growth in other countries would not be greatly affected.

7. If the depreciation arose from fiscal contraction, output growth would temporarily decline in the home country, in contrast to the other two examples. Prices would also decline relative to their baseline path, as the effects of the slowdown in activity would outweigh the inflationary effects of the depreciation. Prices would probably decline in other countries as well. As for output, evidence has been presented here showing that a fiscal contraction in the United States would tend to reduce output growth temporarily in other large countries; however, it is possible that some smaller countries could experience increased output growth as a result of the associated rise in liquidity abroad.

8. Exchange rate movements could also arise from real shocks such as innovations affecting productivity growth or major changes in energy prices. For example, a decline in oil prices is expected to contribute to appreciations of the currencies of countries that are heavily dependent on oil imports (such as Japan) and to depreciations of the currencies of countries with substantial indigenous oil supplies (notably Canada and the United Kingdom). Although it is not possible to generalize about the implications of these country-specific results, it is clear both that these exchange rate movements can be large and disruptive but also that they are a necessary part of the adjustment to real disturbances.

9. Large exchange rate movements can have important implications for the room to maneuver in implementing macroeconomic policies. Whether these movements are toward or away from fundamental equilibrium values, they may have adverse effects for some countries and beneficial implications for others. For example, countries whose currencies had been undervalued and are now appreciating as a result of a portfolio shift might face temporary declines in output growth and in inflation that would alter the optimal paths for monetary growth and fiscal variables in these countries.

10. The policy implications of a monetary shift in one country for other countries whose exchange rates changed as a result would probably not be very great, because—as noted above—the output consequences would be quite small in most instances. The exchange rate movement could, however, lead to important shifts in the mix of output between domestic demand and the external balance, possibly implying a need for policy adjustment in some countries.

11. Exchange rate movements arising from a shift in fiscal policy would be likely to have more pervasive real effects and hence greater policy implications for other countries than movements arising from a monetary shock. In that case, adjustments to monetary or fiscal policy might be warranted in some circumstances, taking account of the need for a stable policy over the longer run. In particular, to the extent that fiscal contraction in one country generated temporary declines in output in other countries, it might be necessary for some of those countries to relax somewhat their own restraint on fiscal policy in order to avoid a decline in output.

Appendix I: Effect of Exchange Rate Changes on Prices of Internationally Traded Commodities

This appendix describes how a change in exchange rates among the industrial countries can affect the price of competitively traded commodities as discussed in the section on portfolio shifts. The basic point can be stated at the outset; the price of a commodity is determined by supply and demand elasticities and market shares, not by the currency in which it is denominated.

These relationships can be illustrated by considering a case in which developing countries produce a fixed amount of a primary commodity that is sold in the United States as well as in other industrial countries. Begin by assuming a dollar depreciation against the currencies of the other countries. Market equilibrium requires that, with a fixed supply of the commodity, the change in total industrial country demand be zero. In the limiting case where there is no demand in the other industrial countries, the change in U.S. demand must be zero. Consequently the dollar price of the commodity cannot have changed. If, on the other hand, the only market for the commodity is in the other countries, the price in their currencies will be unchanged.31 In the intermediate case, where the commodity is demanded in both markets, demand elasticities become a factor.

The model in this appendix divides the world economy into three areas: the United States, other industrial countries, and developing countries.32 The model permits all three economies to produce and consume a single commodity; it also assumes that the various elasticities are common to all three economies. This latter assumption can be relaxed, but the increase in the complexity of the algebra is great relative to any additional insight. The analysis also assumes that developing countries fix their real exchange rate. This is done in order to “solve out” their exchange rate. Other assumptions—such as pegging the developing countries’ currency to the U.S. dollar or to the currency of other industrial countries—are possible. Their adoption would modify the following analysis in a predictable way.

The Model

article image

The supply functions may be written as:


and the demand functions as:


The relative changes in world supply and demand, dsw and dqw, can be written as weighted sums of the changes in the three economies:


The effect of a change of the exchange rate, e, on the dollar price of commodity prices, p, can be solved for by making an explicit assumption about ed, differentiating equations (1) and (2), substituting the results into equations (3) and (4), and equating equations (3) and (4), the equilibrium conditions.

The derivation that follows assumes that the real effective exchange rate of the developing countries is kept constant; thus:


where α is the U.S. weight in the developing countries’ basket of foreign goods. Equations (1) to (5) yield:




Equation (6) shows that the change in the dollar price of a commodity is a weighted sum of changes in general price levels, world real income, dyw, and the exchange rate. Specifically, the effect of a change in the exchange rate on the dollar price of a commodity (1–β) increases as U.S. imports fall relative to total imports, as U.S. exports fall relative to total exports, and as supply elasticities fall.

Appendix II: Wage-Price Dynamics

The following three conventional wage-price equations demonstrate how wage-price dynamics can greatly magnify the initial direct effect of an exchange rate change on prices.



article image