Discipline Hypothesis

The discipline hypothesis is based on the assumption that the balance of payments constraint under fixed exchange rates acts as a check on the pursuit of inflationary policies. Because a devaluation would be regarded by the public as an admission of the failure of government policies, a high-inflation country will sooner or later be obliged to alter its policies in such a way as to bring the inflation rate into line with that of its neighbors. The perception by the public that a given parity must be defended prompts the adoption of otherwise unpopular policies of demand restraint. Surplus countries under fixed rates are said to be subject to a weaker discipline, either because there is no similar constraint on the accumulation of reserves or because the revaluation that is necessary to avoid such accumulation carries no political liability. Under floating rates, this anti-inflationary discipline is absent because (so it is argued) the only consequence of a relatively high inflation rate is a depreciating currency.

The discipline hypothesis prompts the following observations.

(1) Whatever the differences in anti-inflationary discipline between truly fixed rates and purely floating rates, these distinctions become blurred in a comparison of adjustable par values and managed floating.³⁶

Yet the latter two regimes are the relevant ones for the observable macroeconomic policy behavior of the postwar period.

(2) The political cost of devaluation under fixed rates should not be exaggerated. From 1955 to 1971 the longest period without an exchange rate change by any of the 16 OECD countries was five years (1962–66). Likewise, the number of exchange rate adjustments by high-inflation participants since the inception of the European Monetary System in 1979, as well as the number of departures by high-inflation countries from the “snake” in an earlier period, suggest that devaluation is not viewed as a political catastrophe.³⁷ All of this is also consistent with the emerging literature on public choice that relates voting behavior and government popularity to economic variables. It is found in the literature that only the traditional domestic macroeconomic variables (real income growth, inflation, and unemployment) count and that only recent performance is important (i.e., voters have short memories) (Fair (1978); Frey and Scheneider (1978)).

(3) In those cases where there have been conflicts between internal and external balance under floating rates, these conflicts have by no means always been resolved in favor of the internal target. Black (1978), for example, after studying such conflicts over the 1973–76 period, reports (p. 626):

• In most cases, some influence of the external target on monetary or fiscal policy is evident, except for Germany in 1973, the United Kingdom in 1974, and Canada in 1974. Furthermore, the influence of external targets appears to have been rising, as the 1976 conflict cases (France, Italy, Canada, the United Kingdom, and Sweden) have all been resolved in favor of the external target over the internal target.

Such revealed preference for the external target suggests that the balance of perceived political costs is more complex than is sometimes alleged.

(4) The 1979–83 policy experience in industrial countries is evidence that anti-inflationary discipline can be restored without fixed exchange rates. Indeed, the deceleration in growth rates of narrow and broad money that took place in most of the major industrial countries in 1979–82 coincided with relatively high variability of both nominal and real exchange rates.³⁸ This episode of monetary restraint reinforces the point that, while fixed exchange rates are one way of making a nonaccommodation strategy more credible, they are not the only way—and perhaps not even the most effective way. Alternative policy instruments that can be used for anti-inflationary discipline include tax-based incomes policies, preannounced money supply targets, constitutional limits on budget imbalances, stabilization programs of the International Monetary Fund, and, more generally, the aversion of public opinion to the continuation of price inflation and its attendant distortions.

To summarize, the inflation performance of industrial countries over the past decade has been so poor that it is difficult to reject a call for more discipline. The near quadrupling (from 1.2 to 4.2 percent) of the ratio of industrial countries’ government fiscal deficits to gross national product (GNP) between 1963–72 and 1973–82 is just one indicator of how times have changed. But it does not follow from this that fixed exchange rates are either a necessary or sufficient mechanism for establishing such discipline. The restoration of anti-inflationary discipline in 1979–82 in the face of both high unemployment and strong pressure for monetary expansion demonstrates that fixed rates are not always “necessary.” Likewise, during the past two decades there are too many examples of exchange rate targets giving way to employment targets when push came to shove to believe that fixed rates are “sufficient.” In fact, if it is necessary to generalize from the experience after World War II, it would be more accurate to say that both the degree and inter-country dispersion of macroeconomic discipline determine the exchange rate regime rather than the reverse.

Ratchet Hypothesis

Turning to the ratchet hypothesis, the basic argument is that floating rates have an inflationary bias because, in a world of downward price inflexibility, devaluations lead to price increases in the devaluing country but to no (or smaller) offsetting price decreases in the revaluing country. For example, if Country A devalued its exchange rate by 10 percent vis-à-vis Country B and then a year later reversed the process by revaluing by 10 percent—so as to cause no net change in the exchange rate over the whole period—the ratchet hypothesis indicates that domestic prices would be higher in both countries and so, too, would the world price level.

The key to the inflation predictions of the ratchet hypothesis is the proposition that prices do not fall in the revaluing (appreciating) country (or at least do not fall by as much as they rise in the depreciating country). Two explanations have been put forward for this proposition. One is that export prices (expressed in domestic currency) in the devaluing country will rise by the full extent of the devaluation, so that import prices (again, in domestic currency terms) in the revaluing country will not fall after an exchange rate change. The second is that import prices will fall in the revaluing country but, as such a decline will be viewed as temporary and as producers only respond to those cost or demand changes they deem to be permanent, these declines in import prices will not provoke any decline in domestic prices.

While the ratchet hypothesis has a certain intuitive appeal that derives from the apparent stickiness of money wages and prices in industrial countries, empirical tests of this hypothesis have almost unanimously been unkind to it. To begin with, most estimates of the export price response to exchange rate changes find that export prices do not rise by the full extent of a devaluation except in the smallest, most open, industrial economies (Robinson, Webb, and Townsend (1979); Goldstein and Khan (1982)). The same conclusion also applies to export price behavior following tariff changes (e.g., Kreinin (1961)).

Moreover, the implication of the ratchet hypothesis that declines in the domestic currency price of imports should occur very infrequently is not supported by the facts. For example, Pigott, Sweeney, and Willett (1975) found that such declines occurred in about 35 percent of the quarters from 1957 to 1974, at least for six major industrial countries taken together.³⁹ Further, the fall in the domestic currency price of imports following specific revaluations or appreciations (e.g., the 1978 appreciations of the yen and the deutsche mark and the 1961 and 1969 revaluations of the deutsche mark, etc.) also points in the same direction.

Negative changes in import prices also do not appear to have a different effect on domestic prices than positive changes do. Goldstein (1977), in a series of a pooled cross-section time series regressions for five large industrial countries, found no evidence of an asymmetry in the domestic price effects of decreases versus increases in import prices—and these results held for both the gross domestic product deflator and the price of manufactures. Similarly, Finger and DeRosa (1978) examined the relationship between final product prices and raw commodity input prices for 20 product groups during the 1950–75 period and found no evidence of a ratchet effect. Further, a recent Fund staff study (IMF (1984 a)) on exchange rate volatility and world trade could not find any significant independent effect of exchange rate variability on inflation rates for the seven largest industrial countries over the 1958–81 period.

Finally, a recent study for the Organization for Economic Cooperation and Development (OECD) by Encaoua, Geroski, and Miller (1983) indicates that the trend toward greater downward price inflexibility that was observed in major industrial countries since the early 1960s was finally halted during the recession that began in 1979. It is thus dangerous to assume that all cost or demand decreases (whether induced by exchange rates or otherwise) will leave prices unchanged, while all cost and demand increases are quickly passed forward into price increases.

In sum, while few would disagree that the downward inflexibility of finished goods prices and money wages during recessions has been a formidable obstacle to more successful macroeconomic policy in industrial countries over the past two decades, there is little evidence that exchange rate fluctuations have compounded that problem to any important degree. Further, the roots of the downward price inflexibility itself probably lie in the accommodating macroeconomic policies followed by most governments over this period. As long as workers and producers believe that governments can sustain contractionary demand policies for only short periods, they will be reluctant to reduce prices and wages during recessions. In this respect, one compensation of the 1979 recession is that such beliefs are now likely to be held with much less confidence; consistent with that supposition, the trend toward greater downward price inflexibility that had apparently been going on since the early 1960s in major industrial countries seems to have been halted in the 1979–82 period.

Vicious Circle Hypothesis

The charge that floating rates give rise to “vicious and virtuous circles” is heard much less often now than it was seven or eight years ago (1975–76) when disparities in economic performance between strong industrial countries (e.g., Japan and the Federal Republic of Germany) and weak ones (the United Kingdom and Italy) were particularly large. This may partly reflect the growing awareness that inflation differentials are not the overriding determinants of exchange rate changes. Even more so, it reflects the change in fortunes of some of the formerly strong and weak currencies.

Nevertheless, some important issues in the vicious circle debate warrant an examination.⁴⁰ The basic proposition is that a depreciation immediately raises the local currency price of imports. These import price increases then feed quickly through to domestic and export prices, which in turn induce higher money wages, higher domestic prices, more exchange rate depreciation, etc. If the trade balance displays significant J-curve effects and if the exchange rate depreciates in response to expectations of future current account deficits and of higher inflation, the inflation-depreciation spiral can be even quicker and more adverse. The virtuous circle is just the opposite, with the appreciating exchange rate lowering import prices, which in turn lower domestic prices, etc. Fixed rates are said to break these circles because the relatively stable export prices of low-inflation countries restrain the increase in domestic prices in high-inflation countries (and conversely).

A fair appraisal of the vicious circle hypothesis would seem to require acknowledging the validity of three aspects of it.

(1) There can be no doubt that a depreciating exchange rate does have a significant inflationary effect on the depreciating country’s import, domestic, and export prices. Tables 7 and 8 in the Appendix, which are adapted from Goldstein and Khan (1982), give some representative estimates of those effects for a sample of industrial countries (including the seven largest ones). For an average industrial country, a 10 percent depreciation would induce: (a) an 8–10 percent increase in import prices within six months; (b) a 1.5–4.0 percent increase in consumer prices within a year or so; and (c) a 5–8 percent increase in local currency export prices within two years. Also, there is a definite country pattern to these effects. The smaller, relatively open, and more highly indexed countries obtain less relative price advantage and more domestic inflation from a given depreciation than do the larger, less open ones.⁴¹ From this perspective, it is perhaps not surprising that the countries in the former group have in the main rejected independent floating in favor of less flexible exchange arrangements.

(2) A second legitimate point of the vicious circle hypothesis is that floating rates probably shorten the time lag between money supply changes and domestic price changes because money supply changes are often transmitted rapidly into exchange rate depreciation.⁴² As acknowledged by Wallich (1977), this problem limits the scope for antirecessionary action under floating rates, especially in an environment where most industrial countries have had to deal simultaneously with high inflation and high unemployment. The problem is apt to be particularly troublesome when a country with weak currency and high unemployment has to engineer a recovery in the face of an unfavorable foreign interest rate differential.

(3) Yet a third point of merit in the vicious circle hypothesis is the claim that the exchange rate movements that begin the vicious circle—and even some that extend it—need not necessarily be the fault of the policymaking authorities in the country with the depreciating currency. This follows from the widely accepted premise that the current exchange rate depends heavily on expectations about the future exchange rate and from the observation that the list of factors affecting these expectations is long and varied—including not only monetary and fiscal policies in the home country but also unexpected policy changes in foreign countries, new political developments abroad, current account “news” in other countries, changes in intervention practices, etc. In short, anything that affects the supply of (or the demand for) assets denominated in the depreciating currency or close substitute currencies could initiate an exchange rate change. Further, even after a decade of floating rates, the possibility of “inefficiencies” in the foreign exchange market that magnify and prolong departures of actual rates from equilibrium rates cannot be ruled out.⁴³ These inefficiencies may arise from risk aversion combined with legal and regulatory constraints on open foreign exchange positions (i.e., too little stabilizing speculation, as suggested by McKinnon (1976) and Artus and Crockett (1978)), or from a mistaken appraisal of fundamentals by market participants (Dornbusch (1983)), or from the possibility of changes in regime that overwhelm the authorities’ good intentions regarding policymaking (the so-called peso problem), or from speculative “bubbles” that continue to grow larger even in the face of perceived disequilibria because nobody knows when the crash will occur (Blanchard (1979)). Whatever the cause, the point is that weak currencies can be subjected to excessive downward pressure relative to longer-term equilibrium levels.⁴⁴ Tobin (1980, pp. 157–58) has provided a good summary of the problem:

• . . . foreign exchange markets are necessarily adrift without anchors. . . . In these markets, as in other markets for financial instruments, speculation on future prices is the dominating preoccupation of the participants. In the ideal world of rational expectations, the anthropomorphic personified “market” would base its expectations on informed estimates of equilibrium exchange rates. Speculation would be the engine that moves actual rates to the equilibrium set. In fact no one has any good basis for estimating the equilibrium dollar-mark parity for 1980 or 1985, to which current rates might be related. That parity depends on a host of incalculables—not just the future paths of the two economies and of the rest of the world, but the future portfolio preferences of the world’s wealth owners. … In the absence of any consensus on fundamentals, the markets are dominated—like those for gold, rare paintings, and—yes, often equities—by traders in the game of guessing what other traders are going to think.

From another point of view, it is not hard to identify two shortcomings of the vicious circle hypothesis.

(1) A first shortcoming is that the vicious circle hypothesis fails to recognize that an excessive rate of domestic monetary expansion will often be the driving force behind both exchange rate depreciation and high domestic inflation. In this sense, the fact that exchange rates typically respond faster to money supply changes than do domestic prices can create the optical illusion that exchange rate movements are causing domestic price increases, when in reality it is domestic monetary policy that is the real culprit. This point becomes particularly significant whenever one moves beyond the initial stages of the vicious circle because the domestic price increases induced by depreciation also reduce the real value of money balances. As asset holders seek to restore these balances to the desired level, they will spend less on all goods, including imports, creating an incipient current (or capital) account surplus and an exchange rate appreciation (and this will be created even if no relative price advantage is obtained from the depreciation). That is why it is unusual for a vicious circle to be sustained without accommodating money supply behavior. Most studies of policy behavior prior to vicious circles find evidence of just such monetary accommodation (see Bank for International Settlements (1976) and Gordon (1977)). It is also the basis for Haberler’s (1980, p.31) conclusion that “. . . countries are not by chance on one side or the other [of the vicious/virtuous circle].”

(2) The second major deficiency of the vicious circle argument is that it puts forward too simplistic a view of exchange rate determination. Empirical studies show that month-to-month changes in exchange rates are not well correlated with month-to-month inflation differentials. Indeed, whenever an exchange rate changes substantially over a short time span, this change is almost always accompanied by a significant divergence from purchasing power parity (Mussa (1983)). Often, the key variable is the market’s evaluation of the prospects for monetary and fiscal policies in the weak country. The turnaround in the U.S. dollar after the November 1, 1978 package of measures in the United States is perhaps the classic case in point, but there are other notable escapes from the vicious circle as well (e.g., the pound sterling after the acceptance of the Fund’s stand-by arrangement in late 1976).

To summarize, with the benefit of hindsight, it is clear that each side in the original vicious circle debate undervalued the arguments of the other. Based on their own experience, the larger and less open industrial economies underestimated the domestic price effects of depreciation for others, and they oversold the rationality of the market in correctly valuing weak currencies—a rationality that they are increasingly questioning now that some of their own currencies are weak. At the same time, the smaller and more open economies underestimated the role of domestic monetary and fiscal policies in sustaining the vicious circle as well as their own capacity to escape from it by altering these policies. In short, floating rates can exacerbate intercountry inflation differentials, but they need not.

Control of the Money Supply

Perhaps the main reason why floating rates looked so appealing in the latter years under the Bretton Woods system was that the incompatibility of a fixed exchange rate with a relatively independent monetary policy had become painfully obvious. Nowhere was this more apparent than in the Federal Republic of Germany and Switzerland, where restrictive monetary measures (taken in large part to avoid imported inflation) induced capital inflows, official intervention to support the U.S. dollar, more restrictive monetary measures, more capital inflows, etc.—creating, in effect, a monetary vicious circle. In February and March of 1973 alone, the Deutsche Bundesbank purchased $8.5 billion—only to succumb to floating rates the next month. The motive for that decision was clear, as later confirmed by Emminger (1977, p. 4):

• For countries like Germany and Switzerland, the main—or even only—reason why they went over to floating in the spring of 1973 was the necessity to regain control over their own money supply. . . .”

The message that a country could not simultaneously maintain a fixed exchange rate, allow freedom for international capital movements, and have an independently determined money supply was not new (e.g., Kouri and Porter (1974); Frenkel and Johnson (1976); and IMF (1977)). It had long been recognized in the monetary approach to the balance of payments that, under fixed rates, decreases (increases) in domestic credit would be offset by increases (decreases) in international reserves. Further, for a small country,⁵³ this offset would be complete so that the authorities would be able to control the composition of the money supply (i.e., the mix between the domestic and foreign components of the money supply) but not its level; if the country was large or if there was less than full employment, the offset would be only partial, because domestic credit would affect domestic prices and output as well as the balance of payments (e.g., Aghevli and Rodriguez (1979)).

Considerable empirical literature now exists on the behavior of international capital flows and on the sterilization attempts of monetary authorities during the period of fixed rates.⁵⁴ While estimates vary considerably across studies, as a group they suggest that industrial countries found it possible, but at times very difficult, to control their money supplies under fixed rates,⁵⁵ with perhaps the monetary authorities in the Federal Republic of Germany, Switzerland, Belgium, Austria, and France having more trouble than those in Japan, the United States, the United Kingdom, and Italy (Hickman and Schleicher (1978) and Laney (1979)).

Then, what about the control of the money supply under floating rates? The presumption of greater control is derived from the absence of any obligation to use exchange market intervention to peg the exchange rate. Thus, exchange market pressures take the form of price changes (exchange rate changes) rather than volume changes (reserve movements) and the foreign component of the monetary base ceases to be a source of changes in the money supply.

The difficulty, however, is that the authorities must regard the exchange rate exclusively as a policy instrument and not as a target. The more they manage the exchange rate, the more they relinquish the added degree of freedom. As noted earlier, the heavy amount of exchange market intervention conducted by industrial countries during the past decade stands as testimony that policymaking authorities do not regard the either/or choice between control of the money supply and control of exchange rates as acceptable. Instead, their behavior reveals a preference for an intermediate solution, where the authorities keep an eye on both targets.⁵⁶ This suggests that, under floating rates, countries have had more control over the money supply than under fixed rates but that the difference is some-what less marked than early supporters of floating rates had anticipated.⁵⁷ Two observers sum up recent experience:

• From the experience of the past seven years, it is also apparent that the behavior of exchange rates influences the conduct of monetary policy, but usually only after exchange rates have moved substantially away from what the authorities regard as appropriate or desirable values. (Mussa (1981, p. 24))

• … in many countries the exchange rate has achieved a comeback in the minds of policy-makers as one of the most important prices in the economy. Not that a fixed (or a fixed but adjustable) rate has again become a policy goal. . . . Thus, we have seen cases where monetary policy in general, and interest rate policy in particular, became largely geared to the exchange rate; not only in smaller countries like Belgium or Austria, but also in countries like Britain and Germany. (Emminger (1982, p. 2))

Effectiveness of Monetary Policy

The second aspect of the case for floating rates as a boon to monetary policy stems from two early theoretical results of Fleming (1962) and Mundell (1968). The first result is that, under conditions of high capital mobility, a given increase in the money supply produces a larger increase in income under floating rates than under fixed rates. The argument unfolds as follows. Under floating rates, the money supply increase yields a temporary fall in domestic interest rates relative to foreign rates. This induces an incipient capital outflow, a depreciation of the exchange rate, an improvement in competitiveness, and an expansion in net exports. In contrast, under fixed rates, the same interest rate differential induces a realized capital outflow that restores the original money supply and the domestic interest rate, thereby preventing any effect on the domestic level of income. The second result is that, under flexible rates and high capital mobility, expansionary monetary policy has a comparative advantage in raising the level of domestic income over expansionary fiscal policy. The difference is that, whereas the former is accompanied by a fall in domestic interest rates, the latter is accompanied by a rise. As a result, the initial income stimulus under fiscal policy is choked off, or at least blunted, by currency appreciation, but with monetary policy it is reinforced by currency depreciation.

Whatever its merits as a representation of how monetary policy might have worked under flexible rates in an environment similar to that of the 1960s, the Mundell-Fleming model has at least four weaknesses when applied to the 1970s or 1980s.

(1) The first is the assumption that exchange rate changes translate quickly into changes in competitiveness. This would be true if feedbacks from the exchange rate to domestic factor costs and prices were insignificant. As indicated earlier, however, the empirical evidence shows that such feedbacks do exist and that they can be sizable, especially in the smaller, more open, and more highly indexed industrial countries. The greater are these feedbacks, the smaller is the competitive price advantage achieved by depreciation and hence the smaller is the expansion in net exports. In fact, when real wages are rigid and unaffected by exchange rate changes, expansionary monetary policy will affect only prices and the exchange rate, but will have no effect on real output, employment, or the trade balance.⁵⁸ This problem is particularly relevant because there are indications that real wage rigidity is much greater in Europe than in the United States, with Japan being in the middle (Sachs (1983); Grubb, Jackman, and Layard (1982)). This implies that the United States gets more advantage from exchange rate movements than do the Europeans. Branson (1983, p. 58) summarizes this argument:

• These results suggest a pattern of differences in adjustment to exchange rate changes between Europe and the United States . . . an exchange rate change will move relative prices and the balance on current account in the United States, and also influence output, all in the expected “stabilizing” direction. In Europe, however, the movement in the exchange rate will mainly move the overall price level, with minimal effects on the trade balance or output. So the exchange rate is reasonably viewed as an effective instrument for stabilizing the current account in the United States. In Europe, however, exchange rate fluctuations are equally reasonably viewed as essentially destabilizing the price level. The result is policy conflict based on different implicit assumptions about the underlying structure of labor markets and wage behavior.

(2) The second weakness of the Mundell-Fleming model is the assumption that changes in competitiveness will yield rapid improvements in the depreciating country’s trade balance. Because in the short run (less than a year) import prices rise more rapidly than export prices in response to depreciation and there has not been enough time for the volume of trade to adjust very much, it is quite common for the response of trade balance to depreciation to follow a J-curve (Spitäller (1980)). During this short run, the stimulating effects of monetary expansion will thus be reduced, not strengthened, by depreciation (Niehans (1975)). Over time the initial perverse trade balance effects will be checked and then reversed as export price increases catch up with import price increases and as the responses of imports and exports grow large in volume. Nevertheless, this means that, if anything, the comparative advantage of expansionary monetary policy lies in the medium to long run, not in the short run.

(3) A third limitation relates to the size (and unpredictability) of the exchange rate change induced by domestic monetary expansion when domestic and foreign assets (including currencies) are close substitutes. The point here is that exchange rate changes can go much further than the authorities would like. For example, depreciations that were looked on with favor because they would diminish current account deficits can become cause for concern, as asset holders switch out of the weak currency, and as the weak currency’s store of value—and perhaps even its unit of account and medium of exchange functions—is replaced by stronger currencies. Thus, as recognized in the so-called currency substitution literature, high asset substitutability can limit the scope for expansionary monetary policy (Calvo and Rodriguez (1977); Kareken and Wallace (1978); Brillembourg and Schadler (1979)).

(4) Yet a fourth necessary amendment to the Mundell-Fleming model is the incorporation of exchange rate expectations into the choice between domestic and foreign assets. Specifically, under floating rates, asset holders will not necessarily select the asset with a higher nominal interest rate, unless the interest rate differential exceeds the expected depreciation of that currency relative to the one with a lower interest rate. The important implication of this familiar interest-rate-parity condition is that monetary policy operates on exchange rates not only via its direct effect on interest rates but also via its indirect effect on expected future exchange rates. For example, success in halting a depreciation with restrictive monetary policy is likely to hinge as much on convincing the market that this policy is relatively permanent (thus affecting the future exchange rate) as in maneuvering a favorable interest rate differential. The shortcoming of looking only at nominal interest rates is perhaps best illustrated by noting that the U.S. dollar was depreciating relative to the deutsche mark from mid–1976 through most of 1978 despite a rise in U.S. interest rates relative to those abroad (Dornbusch (1979)). A related point of interest is that the movement of the exchange rate itself can be a useful indicator of the appropriate stance of monetary policy under floating rates. Because the nominal interest rate reflects the sum of the real rate of interest and the expected rate of inflation, a rise in the nominal interest rate could reflect an increase in either of the two unobservable factors, each with different implications for the stance of monetary policy (Mussa (1981) and Frenkel (1983 c)). The exchange rate may help to disentangle that puzzle. If the exchange rate is appreciating in the face of a favorable interest rate differential, it signifies a rise in the real rate of interest that may call for easing of domestic monetary policy relative to that abroad; on the other hand, a joint indication of a currency depreciation and a favorable interest rate differential implies that inflation expectations are the culprit, thus signaling, ceteris paribus, monetary restraint.

It is thus clear in retrospect that the case for monetary policy under floating rates was oversold. Many of the perceived constraints on monetary policy during the period of fixed rates turned out not to be constraints imposed by the exchange rate regime but rather constraints imposed by the openness of national economies. As succinctly put by Frenkel (1983 c, p. 49):

• … These constraints are reflected in either a reduced ability to influence the instruments of monetary policy (like the nominal money supply under fixed exchange rates), or in a reduced ability to influence the targets of monetary policy (like the level of real output), or in an increased prudence in the use of monetary policy because of the potentially undesirable effects on expectations.

In view of that, those who counsel that monetary policy be directed more toward stabilization of the exchange rate need to consider what policy instruments will then be directed toward domestic objectives.⁵⁹ In this respect, the past record of fiscal policy hardly makes it an attractive, especially in view of the seemingly structural nature of some present-day budget deficits. Thus, while the comparative advantage of monetary policy under flexible rates is undoubtedly smaller than originally thought, it is still evident, especially in the case of countries that conduct monetary policy in a stable and responsible way.

Capital Mobility and Insulation

The notion that floating rates can insulate a country from foreign disturbances is not a bad working assumption if the international mobility of capital is low, but it is serious misrepresentation when applied to the world of high capital mobility in the 1970s and 1980s. Indeed, one of the key messages of the asset market view of exchange rates is that anything that affects asset supply or asset demand can alter exchange rates and thus affect real variables (i.e., real output and employment) in both the home and the foreign country. In this sense, the relevant question is not whether floating rates can transmit foreign disturbances but rather how they do so vis-à-vis fixed rates.

Two rather well-known results from the theoretical literature are worth repeating:⁶¹ (1) a foreign monetary disturbance will have opposite effects on foreign and domestic output under floating rates but will move output in the same direction under fixed rates; and (2) a foreign expenditure disturbance, whether induced by fiscal policy or otherwise, will be transmitted to domestic output with greater strength under floating rates than under fixed rates. In brief, these results are based on the assumption that foreign monetary expansion lowers the foreign interest rate while a foreign expenditure disturbance raises it; hence, the two types of disturbances produce opposite exchange rate movements which, in turn, imply opposite net trade balance and real output effects for the home country.

Perhaps the key policy implication of these theoretical results is that countries cannot rely on floating rates to protect them from foreign policy changes. Instead, if they want such insulation, either they or the foreign country must take some countervailing action. For example, if the home country wants to prevent foreign monetary expansion from reducing home output or foreign fiscal expansion from increasing home inflation, it will have to prevent an interest rate differential from appearing so as to stabilize the exchange rate and thereby choke off the main channel of transmission. Similarly, if an autonomous shift in asset demand is not to impinge upon domestic policy, either the home country or the foreign country will have to alter the relative supply of assets of home and foreign currency so as to offset these demand shifts. In principle such countervailing actions can always be devised, but there may be formidable constraints on their practical application (e.g., fiscal policy may not be flexible enough, especially in the direction of restraint, to make rapid changes in the policy mix; or the substitutability of domestic and foreign assets may be so high as to preclude small changes in relative asset supplies from having much of an effect on exchange rates; or the home country might not want to follow the foreign country’s policy lead because its domestic unemployment/inflation picture is different). In any case, the essential point is that, with high international mobility of capital, countries will be obliged to work hard to obtain a reasonable degree of insulation from foreign disturbances and, even then, many constraints will preclude complete insulation.

Monetary Versus Real Shocks

A second reason why the insulating properties of floating rates may have been overestimated is that floating rates seem to have a comparative advantage against monetary or overall price level shocks—the types of shocks that probably predominated in the 1950s and 1960s. For example, a floating rate provides good potential for insulation against a rise in the world price level because an appreciation of the domestic currency proportionate to the increase in foreign prices prevents wealth or relative price effects from taking place. Floating rates cannot, however, provide effective insulation against relative price changes among different classes of traded goods (e.g., between oil or food and other tradables) because they cannot alter relative prices at that level of disaggregation. In fact, floating rates have a comparative disadvantage in protecting against real shocks because, unlike fixed rates, they effectively prevent the balance of payments from serving as a cushioning device to smooth domestic consumption (Frenkel and Aizenman (1982); Flood and Marion (1982)). These distinctions are not academic because, as is well known, the most important shocks of the 1970s were real shocks involving large changes in the relative price of tradable commodities.⁶² In this connection, it is worth reporting that theoretical models that simulate the effects of a relative price increase for an important intermediate input (e.g., energy products) in an economy with floating rates and sticky real wages typically find that the authorities are powerless to protect the economy from some increase in unemployment, except under some restrictive conditions (Buiter (1978); Argy and Salop (1979)).

In the real world, countries will be faced with both monetary and real shocks, and they will not know in advance which types of shock will predominate. In this situation, an intermediate degree of exchange rate flexibility can be optimal, but exchange rate policy should not be expected to produce complete insulation from either real or monetary shocks.

From an empirical viewpoint, there is no evidence to suggest that the period of floating rates has been characterized by a weaker international transmission of disturbances than was the period of adjustable par values. Studies by Ripley (1979), Hickman and Schleicher (1978), and Swoboda (1983) all find that synchronization of movements in real economic activity and of monetary variables among industrial countries has typically been somewhat higher under floating rates.⁶³ Such evidence, however, is consistent not only with greater transmission of disturbances under floating rates but with other hypotheses as well, including the greater incidence of common external shocks and the usual policy responses to them in the period of floating rates. This is, for example, the explanation favored by Artus (1983) in interpreting Swoboda’s (1983, p. 103) evidence of increased synchronization:

• In my view it is largely because of these common tendencies [the increase in the size of social transfers, the policy of monetary accommodation of wage and price increases, the growth of rigidities in labor markets, and the decrease in the share of income going to capital, mainly in Europe], and because of the two waves of oil price increases, that industrial countries have all jointly moved into a period of stagflation since 1973. All being in the same situation, it is not surprising either that they have tended to adopt fairly similar policies.

To summarize, the past ten years make it clear that the old view of floating rates as premier insulators against a whole range of foreign disturbances is inaccurate. Instead, floating rates should be considered as having a comparative advantage against some types of disturbance and a comparative disadvantage against others. Also, because floating rates cannot provide complete insulation against the representative bundle of foreign disturbances, the case for policy activism to combat such disturbances (including in some instances greater exchange market intervention) is thereby strengthen ed. by the same token, the case for coordination of policies is also strengthened so as to minimize conflicts between countries’ policy actions.

Current Account Imbalances Alone

As suggested earlier, serious problems are associated with measuring external payments adjustment by reference to the current account alone. Nevertheless, in view of (a) the elusive nature of normal capital flows, (b) the prominence given to the current account as a barometer of the need for adjustment in some earlier periods,⁶⁶ and (c) the availability of relatively long time series data, there is merit in reviewing the historical record.

Table 2 presents three characteristics of current account imbalances (in relation to gross national product) for seven larger and eight smaller industrial countries during the period of floating rates (1973–82) and during the last decade of adjustable par values (1963–72).⁶⁷ The mean ratio of the current account to gross national product is employed as a rough indicator of the average degree of current account adjustment over the period,⁶⁸ while the standard deviation and first-order serial correlation statistics proxy the variation and persistence of these current account imbalances over the period. Ceteris paribus, external adjustment is assumed to be less satisfactory, the larger are these three summary statistics.⁶⁹

Table 2.

Current Account Imbalances as a Percentage of GNP: Selected Industrial Countries, 1963–72 and 1973–82¹

¹Current account includes goods, services, and all current transfers, both private and official.

²Country means take into account the sign of current account imbalances. In contrast, group means are based on absolute values of country means.

³Statistic reported is the estimated coefficient on the lagged dependent variable in the first-order autoregressive equation; * indicates statistical significance at the 95 percent level.

⁴1967–72 only. Prior to 1967 balance of payments data for France included data for overseas territories.

⁵Excludes France.

Table 2.

Current Account Imbalances as a Percentage of GNP: Selected Industrial Countries, 1963–72 and 1973–82¹

	Mean2		Standard Deviation		Serial Correlation3
Country	1963–72	1973–82	1963–72	1973–82	1963–72	1973–82
Larger countries
United States	0.30	0.05	0.46	0.57	0.86*	0.19
Canada	–0.85	–1.31	1.17	1.16	0.60*	0.74*
Japan	0.79	0.20	1.15	1.02	0.83*	0.22
Germany, Fed. Rep.	0.53	0.20	1.07	1.63	0.61*	0.74*
Italy	1.90	–0.96	1.47	2.19	0.81*	0.34
United Kingdom	0.21	–0.19	1.11	1.85	0.46	0.62*
France	–0.184	–0.25	0.574	1.18	…	…
Unweighted average	0.68	0.46	1.00	1.37	0.705	0.485
Smaller countries
Australia	–2.62	–3.04	1.71	1.92	0.83*	1.01*
Austria	–0.48	–1.65	0.71	1.33	0.45	0.68*
Denmark	–1.95	–3.27	1.15	1.13	0.68*	0.96*
Finland	–1.44	–2.35	1.24	2.78	0.66*	0.73*
Netherlands	0.04	1.37	1.14	2.07	–0.12*	0.75*
Norway	–1.88	–4.19	1.78	5.95	0.03	0.84*
Spain	–0.50	–1.90	1.39	2.05	0.63*	0.73*
Sweden	–0.24	–1.64	0.85	2.01	0.61	0.80*
Unweighted average	1.14	2.42	1.26	2.39	0.50	0.81

¹Current account includes goods, services, and all current transfers, both private and official.

²Country means take into account the sign of current account imbalances. In contrast, group means are based on absolute values of country means.

³Statistic reported is the estimated coefficient on the lagged dependent variable in the first-order autoregressive equation; * indicates statistical significance at the 95 percent level.

⁴1967–72 only. Prior to 1967 balance of payments data for France included data for overseas territories.

⁵Excludes France.

View Table

Table 2.

Current Account Imbalances as a Percentage of GNP: Selected Industrial Countries, 1963–72 and 1973–82¹

	Mean2		Standard Deviation		Serial Correlation3
Country	1963–72	1973–82	1963–72	1973–82	1963–72	1973–82
Larger countries
United States	0.30	0.05	0.46	0.57	0.86*	0.19
Canada	–0.85	–1.31	1.17	1.16	0.60*	0.74*
Japan	0.79	0.20	1.15	1.02	0.83*	0.22
Germany, Fed. Rep.	0.53	0.20	1.07	1.63	0.61*	0.74*
Italy	1.90	–0.96	1.47	2.19	0.81*	0.34
United Kingdom	0.21	–0.19	1.11	1.85	0.46	0.62*
France	–0.184	–0.25	0.574	1.18	…	…
Unweighted average	0.68	0.46	1.00	1.37	0.705	0.485
Smaller countries
Australia	–2.62	–3.04	1.71	1.92	0.83*	1.01*
Austria	–0.48	–1.65	0.71	1.33	0.45	0.68*
Denmark	–1.95	–3.27	1.15	1.13	0.68*	0.96*
Finland	–1.44	–2.35	1.24	2.78	0.66*	0.73*
Netherlands	0.04	1.37	1.14	2.07	–0.12*	0.75*
Norway	–1.88	–4.19	1.78	5.95	0.03	0.84*
Spain	–0.50	–1.90	1.39	2.05	0.63*	0.73*
Sweden	–0.24	–1.64	0.85	2.01	0.61	0.80*
Unweighted average	1.14	2.42	1.26	2.39	0.50	0.81

¹Current account includes goods, services, and all current transfers, both private and official.

²Country means take into account the sign of current account imbalances. In contrast, group means are based on absolute values of country means.

³Statistic reported is the estimated coefficient on the lagged dependent variable in the first-order autoregressive equation; * indicates statistical significance at the 95 percent level.

⁴1967–72 only. Prior to 1967 balance of payments data for France included data for overseas territories.

⁵Excludes France.

The story told by Table 2 is straightforward. For the larger industrial countries, there is no indication that current account adjustment has been less satisfactory under floating rates than under the Bretton Woods system. In fact, the figures in Table 2 suggest that average current account imbalances for the larger countries have been noticeably smaller and less persistent (although with larger variation) under floating rates than during the 1963–72 period. At the same time, the current account performance of the smaller industrial countries does seem to have deteriorated during the period of floating rates, with larger average current account imbalances, greater amplitude in current account swings, and more year-to-year persistence in current account imbalances.⁷⁰ In interpreting Table 2, it is also worth reporting that the main qualitative conclusions for both country groups are unaffected if the 1953–62 period is substituted for 1963–72,⁷¹ or if official transfers are excluded from the current account, or if a rough adjustment for the global current account asymmetry is made for the larger industrial countries,⁷² or if the partial data for France are excluded from the calculations.

Chart 3 provides some complementary information on current account adjustment in 1973–82 versus 1963–72 by showing the cumulative current account imbalance for each of the sample countries. This time, the imbalances are not scaled by gross national product (GNP) and the calculation starts fresh in each period. Most interesting, and not previously hinted at in Table 2, is how long it takes for initial current account imbalances to be reversed—under both floating rates and adjustable par values. For whatever reason, when such a complete reversal does take place, it typically requires from three to seven years—hardly a rapid adjustment.

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

View Full Size

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Cumulative Current Account Balances for Selected Industrial Countries, 1963–72 and 1973–82 ¹

(In billions of domestic currency)

¹ Current account includes goods, services, and all current transfers, both private and official.² In trillions of yen.

• Download Figure
• Download Figure
• Download Figure
• Download figure as PowerPoint slide

Current Account Positions and Normal Capital Flows

One important drawback of measuring external adjustment by considering only the current account is that it ignores the possibility of continuing intercountry differences in savings behavior and in real rates of return on investment. Such intercountry differences make it possible for a country with a relatively low domestic savings rate but with relatively attractive domestic investment opportunities to run a persistent current account deficit by drawing on foreign savings. Further, so long as the host country invests those foreign savings wisely (i.e., obtains a rate of return in excess of the cost of borrowing), there is no reason why it cannot sustain a current account deficit for a prolonged period and, just as important, there is no presumption that such a continuing current account imbalance would be suboptimal from a global welfare viewpoint.⁷³

For these reasons, the equilibrium balance of payments is usually defined not in terms of the current account alone but rather in terms of a current account position that can be financed by sustainable or “normal” capital flows. As is well-known, there are a host of serious problems in defining and measuring normal capital flows.⁷⁴ For the purposes of the present study, normal capital flows were approximated by a four-year moving average of the ratio of actual net private capital flows (inclusive of errors and omissions) to GNP.⁷⁵ Given such a measure, it is then possible to use the sum of the current account and normal capital flows as an approximate indicator of the extent of external disequilibrium—that is, the closer this sum is to zero, the more satisfactory external adjustment is presumed to be.

Table 3 presents the relevant comparison between the periods of floating rates and the adjustable peg for this second crude indicator of external adjustment.⁷⁶ Because of considerations of data availability for capital flows, the number of smaller industrial countries in the sample falls from eight to four. Also, the use of a four-year moving average for normal capital flows involves the loss of two observations from the beginning of the data period and one from the end; hence, the period of adjustable pegs is now 1965–72 and the period of floating rates is 1973–81.

Table 3.

Current Account Imbalances Plus Normal Private Net Capital Flows as a Percentage of GNP: Selected Industrial Countries, 1965–72 and 1973–81¹

¹Current account includes goods, services, and all current transfers, both private and official. Private net capital flows include net errors and omissions.

²Country means take into account the sign of imbalances. In contrast, group means are based on absolute values of country means.

³Statistic reported is the estimated coefficient on the lagged dependent variable in the first-order autoregressive equation; * indicates statistical significance at the 95 percent level.

⁴1967–72 only. Prior to 1967 balance of payments data for France included data for overseas territories.

Table 3.

Current Account Imbalances Plus Normal Private Net Capital Flows as a Percentage of GNP: Selected Industrial Countries, 1965–72 and 1973–81¹

	Mean2		Standard Deviation		Serial Correlation3
Country	1965–72	1973–81	1965–72	1973–81	1966–72	1973–81
Larger countries
United States	–0.62	–0.60	0.60	0.66	1.51*	0.57*
Canada	0.25	–0.26	1.15	0.75	0.17	–0.29
Japan	0.94	0.04	0.93	1.00	0.86*	0.19*
Germany, Fed. Rep.	0.75	0.35	0.97	1.31	0.85*	0.54*
Italy	0.64	0.00	1.00	2.51	0.29	0.27
United Kingdom	0.07	–0.13	1.57	1.51	0.62*	0.25
France	0.294	0.18	1.104	0.90	…	…
Unweighted average	0.51	0.22	1.05	1.23	0.72	0.35
Smaller countries
Denmark	0.21	0.31	1.03	1.16	0.19	–0.11
Netherlands	0.76	0.41	1.11	1.53	0.33	0.42
Norway	0.95	1.64	1.73	3.11	0.02	0.69
Sweden	–0.01	0.18	0.72	1.16	0.47	–0.04
Unweighted average	0.48	0.64	1.16	1.74	0.25	0.31

¹Current account includes goods, services, and all current transfers, both private and official. Private net capital flows include net errors and omissions.

²Country means take into account the sign of imbalances. In contrast, group means are based on absolute values of country means.

³Statistic reported is the estimated coefficient on the lagged dependent variable in the first-order autoregressive equation; * indicates statistical significance at the 95 percent level.

⁴1967–72 only. Prior to 1967 balance of payments data for France included data for overseas territories.

View Table

Table 3.

Current Account Imbalances Plus Normal Private Net Capital Flows as a Percentage of GNP: Selected Industrial Countries, 1965–72 and 1973–81¹

	Mean2		Standard Deviation		Serial Correlation3
Country	1965–72	1973–81	1965–72	1973–81	1966–72	1973–81
Larger countries
United States	–0.62	–0.60	0.60	0.66	1.51*	0.57*
Canada	0.25	–0.26	1.15	0.75	0.17	–0.29
Japan	0.94	0.04	0.93	1.00	0.86*	0.19*
Germany, Fed. Rep.	0.75	0.35	0.97	1.31	0.85*	0.54*
Italy	0.64	0.00	1.00	2.51	0.29	0.27
United Kingdom	0.07	–0.13	1.57	1.51	0.62*	0.25
France	0.294	0.18	1.104	0.90	…	…
Unweighted average	0.51	0.22	1.05	1.23	0.72	0.35
Smaller countries
Denmark	0.21	0.31	1.03	1.16	0.19	–0.11
Netherlands	0.76	0.41	1.11	1.53	0.33	0.42
Norway	0.95	1.64	1.73	3.11	0.02	0.69
Sweden	–0.01	0.18	0.72	1.16	0.47	–0.04
Unweighted average	0.48	0.64	1.16	1.74	0.25	0.31

¹Current account includes goods, services, and all current transfers, both private and official. Private net capital flows include net errors and omissions.

²Country means take into account the sign of imbalances. In contrast, group means are based on absolute values of country means.

³Statistic reported is the estimated coefficient on the lagged dependent variable in the first-order autoregressive equation; * indicates statistical significance at the 95 percent level.

⁴1967–72 only. Prior to 1967 balance of payments data for France included data for overseas territories.

Two conclusions stand out in Table 3. The first important conclusion, is that, as in Table 2, external adjustment for the larger industrial countries shows up as considerably better during the period of floating rates than during the last decade of adjustable par values. Both the average size of the imbalance and its persistence (i.e., the serial correlation) are noticeably smaller in 1973–81 than in 1965–72. On the minus side, the average dispersion of these imbalances is larger in 1973–81 than in 1965–72. Perhaps deserving of special notice is the dramatic improvement in external adjustment recorded in 1973–81 by the two countries (Japan and the Federal Republic of Germany), with the largest estimated average external imbalance in 1965–72.⁷⁷ As before, however, it is apparent that the smaller industrial countries have not shared in this improved external adjustment. When account is taken of normal capital flows, external adjustment is still worse (in terms of average size, amplitude, and persistence) in the period of floating rates, although the period differences are much reduced vis-à-vis the current account measures shown in Table 2.⁷⁸

The second important conclusion emanating from Table 3 is that the size of external imbalances is much reduced during both periods when external adjustment is defined to include the capital account as well as the current account. In other words, on average, private net capital flows have acted as an offset to current account imbalances—although certainly not a perfect offset. This point is brought out more clearly in Chart 4, which shows the annual ratios to GNP of the current account and of net private capital flows for each of the subject countries over the whole 1963–83 period. For the larger industrial countries, the typical pattern is for both the current account and the private capital account to fluctuate around their long-term trend values (of zero to roughly 1 percent of GNP), and for positive (negative) deviations in the current account to be paired with negative (positive) deviations in the private capital account.⁷⁹ Thus, the rule still holds that private capital flows help to finance current account imbalances rather than add to them. This does not mean that there have not been episodes when capital flows exacerbated current account problems. Indeed, Chart 4 shows that the last few years of the Bretton Woods system (1969–72), as well as the 1977–79 period, fit the characterization that “when the current account gets bad the capital account gets worse” (Dornbusch (1980, p. 173)). These periods of strong disequilibrium and of great pressure on exchange rates also provide one explanation for the heavy official intervention that took place then, as authorities “leaned against the wind” to dampen exchange rate movements.

Ratio of the Current Account and Private Net Capital Flows to GNP in Selected Industrial Countries, 1963–82 ¹

Sources: IMF, International Financial Statistics and the data file of the External Adjustment Division of the Fund’s Research Department.¹ Current account includes goods, services, and all current transfers, both private and official. Net private capital flows include net errors and omissions.

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Ratio of the Current Account and Private Net Capital Flows to GNP in Selected Industrial Countries, 1963–82 ¹

Sources: IMF, International Financial Statistics and the data file of the External Adjustment Division of the Fund’s Research Department.¹ Current account includes goods, services, and all current transfers, both private and official. Net private capital flows include net errors and omissions.

• Download Figure
• Download figure as PowerPoint slide

Ratio of the Current Account and Private Net Capital Flows to GNP in Selected Industrial Countries, 1963–82 ¹

Sources: IMF, International Financial Statistics and the data file of the External Adjustment Division of the Fund’s Research Department.¹ Current account includes goods, services, and all current transfers, both private and official. Net private capital flows include net errors and omissions.

• Download Figure
• Download figure as PowerPoint slide

As regards the smaller industrial countries, there is a more pronounced trend in net private capital flows associated either with the financing of large-scale energy projects (e.g., Norway) or with a substantial increase in official or quasi-official borrowing (e.g., Denmark and Sweden).⁸⁰ Note, however, that for these smaller industrial countries the negative covariance between the current account and private net capital flows still holds and that, as a result, external imbalance inclusive of capital flows is typically much smaller than without them.

The Equilibrium Balance of Payments

Even though estimates of a more comprehensive construct of the equilibrium balance of payments do not extend back beyond 1975 or 1976, these estimates are still worth examining for what they imply about the adequacy of external payments adjustment during the last seven or eight years of managed floating. Two such estimates or studies merit separate mention.

The first is that recently done by Williamson (1983). Although he estimates discrepancies between actual and so-called fundamental equilibrium exchange rates rather than differences between actual and equilibrium payments imbalances, it is possible to infer the latter from the former because Williamson (1983, p. 14) defines his fundamental equilibrium exchange rate as essentially the exchange rate that would make the actual balance of payments equal to the equilibrium one. Also, the types of adjustment that are made to actual current account and capital account balances (e.g., adjustments for cyclical demand effects, permanent exogenous changes in the terms of trade, continuing intercountry differences in labor productivity, and the switch in the status of the United Kingdom since 1977 from a major oil importer to a significant oil exporter) are at least close relatives of the types of adjustment suggested here. For these reasons, it is probably legitimate to use Williamson’s estimates of real exchange rate disequilibria (i.e., differences between actual and fundamental real equilibrium exchange rates) as a rough index of external payments maladjustments.

Williamson’s (1983) estimates cover the five major currency countries for the 1976–83 period. Two of his conclusions are relevant for this paper. First, there is no indication that the size of external adjustment imbalances has been declining over time during the past seven years. On the contrary, for three of the five major industrial countries (each of which has adopted an exchange rate arrangement of independently floating), the maladjustments in the 1981–83 period have been as large or larger than at any other time during the past seven years. Second, whatever the mean degree of external payments adjustment for the period of floating rates, this mean conceals substantial mal-adjustments, often in the opposite direction, during individual one-year or two-year periods (e.g., the United Kingdom in 1976 versus 1980–83, Japan in 1978 versus 1982–83, and France in 1980 versus 1982–83).

The second study, or series of studies, on equilibrium payments balances was done by the Fund staff over the 1971–83 period. A brief history of the development of that methodology in the Fund (usually referred to as the underlying payments balance method) is contained in a recent Fund staff study (IMF (1984 b)) dealing with issues in the assessment of exchange rates of industrial countries.⁸¹ While the approach taken by the staff to estimate equilibrium payments positions in those exercises falls within the general framework outlined here, there is one important difference. Whereas the earlier calculations in this paper have been static or backward looking, the exercises in the Fund are explicitly forward looking. Specifically, they seek to estimate the likely paths of the current account (adjusted for temporary factors) and of normal capital flows during the next two or three years, given the continuation of present exchange rates, anticipated macroeconomic policies in the subject countries, the delayed effects of past exchange rate changes, and a number of other expected future developments.⁸² In short, the aim of the staff exercises has really been to spell out the medium-term balance of payments implications of past and present exchange rates and of expected future macroeconomic policies. If those balance of payments developments are deemed unsustainable or undesirable in the sense that “underlying” current accounts are quite different from normal capital flows, then the normative implication is that either planned policies or present exchange rates need to change to prevent those undesirable balance of payments scenarios from taking place.

Given the focus of this paper, perhaps the key conclusion arising from these staff exercises is that the period of floating rates has been marked by many instances in which anticipated policies, together with prevailing exchange rates, did imply undesirable or unsustainable external payments outcomes. Table 4, taken from a recent Fund staff study (IMF (1984 b)), provides just one representative example of such a staff calculation applied to the November 1983 situation. The key point to note in that table is that the estimates of normal capital flows given in column (11) are quite different from the estimates of underlying current accounts given in columns (9) and (10).⁸³ In short, without changes in macroeconomic policies and/or changes in real exchange rates, the outlook for effective external payments adjustment was not encouraging from the vantage point of November 1983.

Table 4.

Underlying Payments Balances on Current Account and Normal Capital Flows¹

(In billions of U.S. dollars)

Source: IMF, Issues in the Assessment of Exchange Rates of Industrial Countries, Occasional Paper No. 29 (July 1984).

¹All accounts are in constant 1983 prices.

²In Japan, recent import liberalization measures increased import volumes; adjustments are required to give appropriate balances for the whole of 1983, both for Japan and for its trading partners.

³Estimates of the effects of changes in exchange rates and domestic price levels that occurred from 1980 through November 1983 but are not reflected in 1983 trade flows.

⁴The effect of assumed changes in relative cyclical positions to 1986 on trade and services balances in terms of 1983 dollars.

⁵Effects over three years of estimated trend factors, including changes in fuels balances.

⁶The 1983 balances adjusted by the factors in columns (2) through (5).

⁷A further adjustment is made in column (7) for changes in relative cyclical positions which arise if prospective U.S. demand growth is assumed to be ½ percent per annum higher, and prospective demand growth in the Federal Republic of Germany and Japan is assumed to be ½ percent lower than in the central scenario.

⁸Includes effects over three years of estimated trend in official transfers.

⁹The sum of columns (6) and (8).

¹⁰The sum of columns (7) and (8).

¹¹Reflects errors, omissions, and asymmetries in reported statistics, plus balance with other countries including the U.S.S.R. and other countries of Eastern Europe that are not Fund members. In the table, this asymmetry is assumed to remain constant at the projected 1983 level.

Table 4.

Underlying Payments Balances on Current Account and Normal Capital Flows¹

(In billions of U.S. dollars)

		Adjustment for:
	Current Balance Projection 1983 (1)	Temporary disturbances2 (2)	Recent relative price changes3 (3)	Changes in relative cyclical positions4 (4)	Medium-term tendencies5 (5)	Scenario I: Underlying Current Balance6 (6)	Scenario II: Underlying Current Balance7 (7)	Official Transfers8 (8)	Scenario I: Underlying Current Balance Including Official Transfers9 (9)	Scenario II: Underlying Current Balance Including Official Transfers10 (10)	Normal Capital Flows (11)
United States	–34.2	0.2	–28.2	–10.2	–7.8	–80.3	–85.5	–5.6	–85.9	–91.1	– 5 to –10
Germany, Fed. Rep.	9.0	—	10.9	0.8	–3.1	17.9	20.0	–6.5	11.4	13.5	0
Japan	22.3	–0.6	0.8	–1.6	6.4	27.3	30.0	–2.0	25.3	28.0	– 5 to –10
Other industrial countries	1.8	0.1	16.5	–0.2	3.1	21.0	20.7	–8.0	13.0	12.7	+10 to –20
Total industrial countries	–1.2	–0.3	—	–11.2	–1.4	–14.1	–14.8	–22.1	–36.2	–36.9	—
Other countries	–72.6	0.3	—	11.2	1.4	–59.7	–59.0	22.1	–37.6	–36.9	—
Total11	–73.8	—	—	—	—	–73.8	–73.8	—	–73.8	–73.8	—

Source: IMF, Issues in the Assessment of Exchange Rates of Industrial Countries, Occasional Paper No. 29 (July 1984).

¹All accounts are in constant 1983 prices.

²In Japan, recent import liberalization measures increased import volumes; adjustments are required to give appropriate balances for the whole of 1983, both for Japan and for its trading partners.

³Estimates of the effects of changes in exchange rates and domestic price levels that occurred from 1980 through November 1983 but are not reflected in 1983 trade flows.

⁴The effect of assumed changes in relative cyclical positions to 1986 on trade and services balances in terms of 1983 dollars.

⁵Effects over three years of estimated trend factors, including changes in fuels balances.

⁶The 1983 balances adjusted by the factors in columns (2) through (5).

⁷A further adjustment is made in column (7) for changes in relative cyclical positions which arise if prospective U.S. demand growth is assumed to be ½ percent per annum higher, and prospective demand growth in the Federal Republic of Germany and Japan is assumed to be ½ percent lower than in the central scenario.

⁸Includes effects over three years of estimated trend in official transfers.

⁹The sum of columns (6) and (8).

¹⁰The sum of columns (7) and (8).

¹¹Reflects errors, omissions, and asymmetries in reported statistics, plus balance with other countries including the U.S.S.R. and other countries of Eastern Europe that are not Fund members. In the table, this asymmetry is assumed to remain constant at the projected 1983 level.

View Table

Table 4.

Underlying Payments Balances on Current Account and Normal Capital Flows¹

(In billions of U.S. dollars)

		Adjustment for:
	Current Balance Projection 1983 (1)	Temporary disturbances2 (2)	Recent relative price changes3 (3)	Changes in relative cyclical positions4 (4)	Medium-term tendencies5 (5)	Scenario I: Underlying Current Balance6 (6)	Scenario II: Underlying Current Balance7 (7)	Official Transfers8 (8)	Scenario I: Underlying Current Balance Including Official Transfers9 (9)	Scenario II: Underlying Current Balance Including Official Transfers10 (10)	Normal Capital Flows (11)
United States	–34.2	0.2	–28.2	–10.2	–7.8	–80.3	–85.5	–5.6	–85.9	–91.1	– 5 to –10
Germany, Fed. Rep.	9.0	—	10.9	0.8	–3.1	17.9	20.0	–6.5	11.4	13.5	0
Japan	22.3	–0.6	0.8	–1.6	6.4	27.3	30.0	–2.0	25.3	28.0	– 5 to –10
Other industrial countries	1.8	0.1	16.5	–0.2	3.1	21.0	20.7	–8.0	13.0	12.7	+10 to –20
Total industrial countries	–1.2	–0.3	—	–11.2	–1.4	–14.1	–14.8	–22.1	–36.2	–36.9	—
Other countries	–72.6	0.3	—	11.2	1.4	–59.7	–59.0	22.1	–37.6	–36.9	—
Total11	–73.8	—	—	—	—	–73.8	–73.8	—	–73.8	–73.8	—

Source: IMF, Issues in the Assessment of Exchange Rates of Industrial Countries, Occasional Paper No. 29 (July 1984).

¹All accounts are in constant 1983 prices.

²In Japan, recent import liberalization measures increased import volumes; adjustments are required to give appropriate balances for the whole of 1983, both for Japan and for its trading partners.

³Estimates of the effects of changes in exchange rates and domestic price levels that occurred from 1980 through November 1983 but are not reflected in 1983 trade flows.

⁴The effect of assumed changes in relative cyclical positions to 1986 on trade and services balances in terms of 1983 dollars.

⁵Effects over three years of estimated trend factors, including changes in fuels balances.

⁶The 1983 balances adjusted by the factors in columns (2) through (5).

⁷A further adjustment is made in column (7) for changes in relative cyclical positions which arise if prospective U.S. demand growth is assumed to be ½ percent per annum higher, and prospective demand growth in the Federal Republic of Germany and Japan is assumed to be ½ percent lower than in the central scenario.

⁸Includes effects over three years of estimated trend in official transfers.

⁹The sum of columns (6) and (8).

¹⁰The sum of columns (7) and (8).

¹¹Reflects errors, omissions, and asymmetries in reported statistics, plus balance with other countries including the U.S.S.R. and other countries of Eastern Europe that are not Fund members. In the table, this asymmetry is assumed to remain constant at the projected 1983 level.

To summarize, the record of external payments adjustment during the period of floating rates has been a mixed one. From the available summary indicators, there is little evidence that external adjustment has been less complete or slower under floating rates than under the adjustable peg system. In fact, the performance of the larger industrial countries appears on average to have been better in the period of floating rates while that of the smaller industrial countries looks worse. At the same time, any notion that floating rates produce prompt equilibration of external payments balances can be safely discarded. When effective external adjustment does take place, it takes a number of years and there have been many instances of large discrepancies between actual and equilibrium payments balances at prevailing exchange rates.

Relative Price Changes Versus Relative Income Movements⁸⁵

One of the cornerstones of the case for floating rates is that greater exchange rate flexibility permits the substitution of expenditure-switching policies for expenditure-reducing ones, with the result that the cost of external adjustment (in terms of output and employment) is reduced, especially in a context of downward price inflexibility (Friedman (1953), Meade (1955), and Johnson (1973)). Implicit in this conclusion is the assumption that, once a country’s effective exchange rate moves in the direction required for, say, current account adjustment, the resulting relative price effects will act as a powerful aid in securing that adjustment.

While the foregoing argument is unassailable in a theoretical sense, critics of floating rates have expressed doubts about the practical efficacy of exchange rate changes as an adjustment mechanism. Those doubts have two origins—one conjectural and the other historical. The conjectural one is that the uncertainty associated with floating rates could significantly reduce the size of price elasticities for traded goods. The historical one is based on the observation that there have been several occasions during the period of floating rates when sizable current account imbalances have persisted in the face of sizable exchange rate changes (in the right direction).

The price-elasticity-pessimism thesis has perhaps been put foward most clearly by Niehans (1975, p. 276).

• . . . consider the probable effect of flexible rates on foreign trade elasticities. It is convenient to introduce the distinction between the actual exchange rate and what, in analogy to permanent income, may be called the permanent exchange rate. . . .

• What matters for trade flows in physical units is mostly the permanent rate. Major changes in the international division of labor require new production facilities, new distribution networks, new sources of supply, and the development of new markets. Most firms will try to avoid making such long-term decisions on the basis of exchange rates which turn out to be only temporary. . . . With flexible rates, in view of the slow adjustment of permanent rates to actual rates, this process will be even slower, and many fluctuations in actual rates will have hardly any effect on permanent rates, and thus on trade flows.

• and by McKinnon (1978, p. 4):

• . . . with the advent of floating, the future direction of exchange rate movements has proved highly uncertain. . . . And it may not be in the interest of merchants to engage in active arbitrage in industrial commodities if tomorrow’s exchange rate is unknown. Hence, the quantitites of goods traded respond sluggishly to exchange rate fluctuations giving rise to a modern version of elasticity pessimism. . . .

Whatever the plausibility of the Niehans-McKinnon argument, there is as yet no firm empirical support for it. A recent survey by Goldstein and Khan (1982), for example, does not find any consistent tendency for estimated price elasticities to be lower in equations estimated with 1970s data than in those based on data from earlier periods. It needs to be acknowledged, however, that the number of studies using data for the period of floating rates is still relatively small, and that composition effects (i.e., changes in the weights of low and high elasticity goods in total exports or imports) can distort the independent effect of the exchange rate regime on these elasticities.

Turning to the historical argument, Table 5 documents perhaps the most striking of these alleged perverse episodes for the major industrial countries: namely, the 1976–78 period. In particular, worthy of note is the seemingly perverse association between nominal exchange rate changes (row 7) and current account developments (rows 4–6) in the three largest industrial countries (the Federal Republic of Germany, Japan, and the United States).

Table 5.

Trade Balances, Current Accounts, Exchange Rates, Relative Prices, and Cyclical Income Movements for the Seven Largest Industrial Countries, 1976–78

Sources: IMF, World Economic Outlook (various annual issues); IMF, International Financial Statistics, various issues; and Artus and Turner (1978).

¹In billions of U.S. dollars.

²Includes goods, services, and all current transfers, both private and official.

³The Fund’s multilateral exchange rate model (MERM) index. A positive figure denotes an appreciation, a negative one a depreciation.

⁴Percentage change in ratio of own to competitors’ wholesale prices for manufactures. A positive figure denotes deterioration in country’s position, a negative one improvement.

⁵Percentage change in ratio of own to competitors’ wholesale prices for manufactures, adjusted for effective exchange rates. A positive figure denotes deterioration in country’s position, whereas a negative one denotes improvement. Since the relationship is multiplicative rather than additive, numbers in the two rows above will not sum to those in this row.

⁶Gross domestic product at market prices for France, Italy, and the United Kingdom.

⁷Defined as potential output less actual output, as percentage of actual output. A negative figure indicates manufacturing sector is operating at less than normal capacity.

Table 5.

Trade Balances, Current Accounts, Exchange Rates, Relative Prices, and Cyclical Income Movements for the Seven Largest Industrial Countries, 1976–78

	Canada	France	Fed. Rep. of Germany	Italy	Japan	United Kingdom	United States
Trade balance1
1976	1.8	–4.9	16.2	–4.2	9.8	–7.0	–9.5
1977	3.1	–3.3	19.6	–0.1	17.3	–3.9	–31.1
1978	4.1	0.1	24.7	2.9	25.6	–3.0	–34.0
Current account1, 2
1976	–4.2	–3.4	3.9	–2.9	3.7	–1.5	4.2
1977	–4.1	–0.4	4.1	2.4	10.9	0.1	–14.5
1978	–4.3	7.1	9.2	6.2	17.5	2.2	–15.5
Percentage change, 1976–78
Nominal effective exchange rate3	–16.7	–7.8	13.7	–16.2	32.0	–6.0	–10.4
Unadjusted relative wholesale prices4	3.8	4.6	–9.4	18.7	–12.5	18.5	1.5
Adjusted relative wholesal e prices (real exchange rate)5	–13.6	–3.5	3.1	–0.4	15.3	11.7	–9.0
Percentage change in real GNP/GDP6
1975	1.2	0.2	–1.6	–3.6	2.4	–1.1	–1.2
1976	5.5	5.2	5.6	5.9	5.3	3.4	5.4
1977	2.1	3.0	2.8	1.9	5.3	1.6	5.5
1978	3.6	3.7	3.5	2.7	5.1	3.9	5.0
Output gap in manufacturing7
1975	–8.5	–6.9	–11.1	–13.1	–22.3	–9.4	–12.3
1976	–6.2	–3.9	–5.3	–5.9	–17.2	–7.9	–6.4
1977	–7.2	–3.0	–4.2	–7. 0	–15.4	–6. 5	–2.7
1978	–5. 0	–2.6	–4.1	–8.1	–13.1	–6.1	0.0

Sources: IMF, World Economic Outlook (various annual issues); IMF, International Financial Statistics, various issues; and Artus and Turner (1978).

¹In billions of U.S. dollars.

²Includes goods, services, and all current transfers, both private and official.

³The Fund’s multilateral exchange rate model (MERM) index. A positive figure denotes an appreciation, a negative one a depreciation.

⁴Percentage change in ratio of own to competitors’ wholesale prices for manufactures. A positive figure denotes deterioration in country’s position, a negative one improvement.

⁵Percentage change in ratio of own to competitors’ wholesale prices for manufactures, adjusted for effective exchange rates. A positive figure denotes deterioration in country’s position, whereas a negative one denotes improvement. Since the relationship is multiplicative rather than additive, numbers in the two rows above will not sum to those in this row.

⁶Gross domestic product at market prices for France, Italy, and the United Kingdom.

⁷Defined as potential output less actual output, as percentage of actual output. A negative figure indicates manufacturing sector is operating at less than normal capacity.

View Table

Table 5.

Trade Balances, Current Accounts, Exchange Rates, Relative Prices, and Cyclical Income Movements for the Seven Largest Industrial Countries, 1976–78

	Canada	France	Fed. Rep. of Germany	Italy	Japan	United Kingdom	United States
Trade balance1
1976	1.8	–4.9	16.2	–4.2	9.8	–7.0	–9.5
1977	3.1	–3.3	19.6	–0.1	17.3	–3.9	–31.1
1978	4.1	0.1	24.7	2.9	25.6	–3.0	–34.0
Current account1, 2
1976	–4.2	–3.4	3.9	–2.9	3.7	–1.5	4.2
1977	–4.1	–0.4	4.1	2.4	10.9	0.1	–14.5
1978	–4.3	7.1	9.2	6.2	17.5	2.2	–15.5
Percentage change, 1976–78
Nominal effective exchange rate3	–16.7	–7.8	13.7	–16.2	32.0	–6.0	–10.4
Unadjusted relative wholesale prices4	3.8	4.6	–9.4	18.7	–12.5	18.5	1.5
Adjusted relative wholesal e prices (real exchange rate)5	–13.6	–3.5	3.1	–0.4	15.3	11.7	–9.0
Percentage change in real GNP/GDP6
1975	1.2	0.2	–1.6	–3.6	2.4	–1.1	–1.2
1976	5.5	5.2	5.6	5.9	5.3	3.4	5.4
1977	2.1	3.0	2.8	1.9	5.3	1.6	5.5
1978	3.6	3.7	3.5	2.7	5.1	3.9	5.0
Output gap in manufacturing7
1975	–8.5	–6.9	–11.1	–13.1	–22.3	–9.4	–12.3
1976	–6.2	–3.9	–5.3	–5.9	–17.2	–7.9	–6.4
1977	–7.2	–3.0	–4.2	–7. 0	–15.4	–6. 5	–2.7
1978	–5. 0	–2.6	–4.1	–8.1	–13.1	–6.1	0.0

Sources: IMF, World Economic Outlook (various annual issues); IMF, International Financial Statistics, various issues; and Artus and Turner (1978).

¹In billions of U.S. dollars.

²Includes goods, services, and all current transfers, both private and official.

³The Fund’s multilateral exchange rate model (MERM) index. A positive figure denotes an appreciation, a negative one a depreciation.

⁴Percentage change in ratio of own to competitors’ wholesale prices for manufactures. A positive figure denotes deterioration in country’s position, a negative one improvement.

⁵Percentage change in ratio of own to competitors’ wholesale prices for manufactures, adjusted for effective exchange rates. A positive figure denotes deterioration in country’s position, whereas a negative one denotes improvement. Since the relationship is multiplicative rather than additive, numbers in the two rows above will not sum to those in this row.

⁶Gross domestic product at market prices for France, Italy, and the United Kingdom.

⁷Defined as potential output less actual output, as percentage of actual output. A negative figure indicates manufacturing sector is operating at less than normal capacity.

There are at least three reasons why this evidence should be interpreted cautiously, however.

(1) Exchange rates are only one component of a country’s competitive position. The other component is the behavior of the prices or costs of traded goods in that country relative to those of its competitors—that is, real rather than nominal exchange rate changes count for trade flows. Divergences from purchasing power parity have been much larger during the period of floating rates than under the adjustable peg (Genberg (1978)), but there have nevertheless been episodes Floating Rates and External Adjustment when relative inflation rates have offset much of the effect of nominal exchange rate changes on countries’ competitive price positions. The 1976–78 period was one of these. As shown in rows 8 and 9 of Table 5, this relative inflation offset was most pronounced for the countries with the most abnormal inflation performance—that is, Italy and the United Kingdom on the high-inflation side, and Japan and the Federal Republic of Germany on the low-inflation side. In cases where the size of real exchange rate changes is modest, a large relative price impact on current account outcomes should not be expected.

(2) A second factor that works against a clear association between exchange rate changes and current accounts is that relative price changes affect the volumes of imports and exports in the short run differently from in the long run. By now, there is a considerable empirical literature on relative price elasticities in international trade (Goldstein and Khan (1982)). This literature suggests that those elasticities are significant and reasonably large over the long run (two to three years) in most industrial countries but also that they are much smaller over the short run (up to one year). A consensus estimate would be that short-run price elasticities are only about half as large as the long-run ones. Given that the long-run elasticities themselves probably lie in the range of –0.5 to –1.0 for total imports and –1.25 to –2.5 for total exports in a representative industrial country (Goldstein and Khan (1982)), this means that the short-run response of the trade balance to an exchange rate change can be perverse. For example, Spitäller (1980) shows that the short-run deterioration or improvement in the trade balance following a depreciation or appreciation, respectively, will probably last about four or five quarters and that, at its worst point, it could amount to about 8 to 10 percent of the local currency value of imports.

(3) The third, and probably most compelling, reason why exchange rate changes and current account out-comes do not always move together is that relative price changes are not the only, or even the most important, determinant of current account movements over the short to medium run. Over a one-year period the combined income elasticities of demand for imports and exports will generally be two to four times larger than the sum of relative price elasticities (Deppler and Ripley (1978); Hooper (1978); Goldstein and Khan (1982)), and there are some indications that income movements would still be more powerful even after three years.⁸⁶ Thus, if relative income movements work on trade flows in a direction opposite to that of relative price factors, there need be no close correspondence between the latter and current accounts. By all reports, such relative cyclical developments were responsible for much of both the deterioration in the U.S. current account and the improvements in the current accounts of the Federal Republic of Germany and Japan between 1975 and 1978 that are reflected in Table 5 (Lawrence (1978) and Wallich (1978)). This is quite a common phenomenon; whenever a country’s growth rate (relative to its potential) exceeds that abroad, its current account usually deteriorates.

In sum, the contribution made by exchange rate changes to current account adjustment is at times hidden from the naked eye, especially over the short run, but this contribution is still important.

Adjustment in Goods Versus Asset Markets

In this paper equilibrium payments balances, and by analogy the equilibrium exchange rate, have been defined in terms of a current account equal to normal capital flows (after correction for temporary factors, cyclical effects, etc.). Thus, both normal flows of goods and normal transactions in assets are included in the concept of “fundamentals.” Other observers, however, often choose a different definition of “fundamentals,” ranging from the current account alone, to all transactions in financial assets, to the behavior of the major determinants of such trade flows or changes in asset supplies or demand (e.g., real incomes, relative inflation rates, money supplies, interest rates, etc.). It is thus easy to understand why the adequacy of external adjustment or exchange rate behavior under floating rates—measured as the difference between actual outturns and that corresponding to the fundamentals—elicits so many divergent views.

Regardless of the precise definition of “fundamentals,” however, there is the indisputable basic economic reality that adjustments to disturbances in goods and labor markets typically take much longer than those in financial markets. This means that prices in those financial markets (namely, interest rates and exchange rates) can be expected to bear the brunt of the short-run adjustment to unanticipated economic developments. And because these financial prices have to compensate for the stickiness of goods and labor prices, their short-run response is often much larger than the long-run one (after other prices have also moved). In now popular terminology, these financial prices “overshoot.”⁸⁷

This overshooting of exchange rates and other financial prices would not perhaps be cause for much concern if the overshooting did not last long, or if the disturbances that initiated it were solely of domestic origin, or if we could always identify the nature of those disturbances, or if the induced exchange rate changes did not have a powerful effect on the real side of the economy. Unfortunately, experience with floating rates suggests that none of these conditions is true. As noted earlier, it would seem that there have been periods of two to three years’ duration under floating rates when the structure of payments balances and of exchange rates was out of line with fundamentals and when this overshooting was most closely connected with developments in financial markets. The international aspect of the problem arises because financial markets today are so well diversified and integrated that assets denominated in different currencies are close substitutes for one another. Hence, anything that affects the current rate of return, risk factors, or the expected future rate of return in a partner country can have swift and strong repercussions on the home country’s asset prices, including its exchange rate. Further, while in some cases it is easy to locate the source of the overshooting disturbance in terms of large shifts in the stance of monetary or fiscal policy, in other cases it is not, with exchange rates continuing to move out of line after real interest rate differentials have stabilized or after the current account has begun to move in the opposite direction. Last, it has to be recalled that although the exchange rate is often viewed as an asset price (i.e., the relative price of two monies), it is also a key component of the relative price of national outputs; for this reason, it will have strong effects on the demand and supply for traded goods, and hence on employment in traded goods industries. This fact is often brought home most vividly after the implementation of a program of monetary restraint. In the initial stages, the effects of the resultant increase in real interest rates on economic activity outweigh the relative price effects of exchange rate appreciation and produce an improvement in the current account. Later on, however, while the capital account continues to be aided by the high real interest rate, the current account deteriorates as the adverse relative price effects of a higher real exchange rate take their toll on the country’s trade flows.

Having noted the seriousness of the overshooting problem, it is equally relevant to point out that the solution may not lie in pegging of exchange rates for at least two reasons. One is that such an action may just transfer disturbances from the exchange market to goods and labor markets at even greater social cost (Frenkel and Mussa (1980)). In this respect, at least one advantage of the exchange market is that insurance against unforeseen (short-term) fluctuations can be purchased through forward contracts. And because prices are sticky in goods and labor markets, a consequence of shifting more of the adjustment to those markets is that real output and employment may suffer more. The second reason is that pegged rates can give rise to their own brand of overshooting. Machlup (1979, p. 76) makes this point as follows:

• The economists who blame particular appreciations and depreciations of currencies on “overshooting in a regime of floating rates” forget that the discrete and deliberate adjustments of fixed or pegged exchange rates, especially the official devaluations, usually involve much wider overshooting. If a currency had long been overvalued by its fixed parity, and the authorities at last decided on a devaluation, they regularly chose a new parity which undervalued the currency at current levels of prices. Among the justifications for such “excessive” devaluations was usually the argument that the monetary reserves, depleted in the period of overvaluation, had to be replenished in the subsequent period of undervaluation. No one was ever ashamed of such overshooting in the official adjustment of the official par value; yet the same observers blame the system of managed floating for allowing the market to overshoot the “right” exchange rate. What used to be the rule for devaluations is now regarded as disorderly in depreciations.

Symmetry of Adjustment

As is well-known, the period under the Bretton Woods system was marked by frequently and strongly expressed concerns on the part of some countries over the alleged lack of symmetry in external adjustment. Two types of asymmetry were most discussed.⁸⁸ First, there was the charge that surplus countries were subject to a much weaker discipline than deficit countries. Second, there were complaints about the special role of reserve centers, especially the United States. To the Europeans, this special role constituted an unwarranted privilege because the United States alone could finance payments deficits by liability as opposed to asset settlement.⁸⁹ On the other hand, to the United States the special role of the dollar as numeraire of the system came to be seen as a burden because it precluded the initiation of exchange rate action as an adjustment mechanism. What can be said about the validity of these allegations and about their subsequent existence under managed floating?

The proposition that balance of payments deficits prompted stronger adjustment measures than comparable surpluses under the adjustable peg system is supported in what is probably the most thorough study of the issue. Specifically, after studying balance of payments developments in nine industrial countries over the 1950–66 period, Michaely (1971, pp. 63–64) concludes:

• Countries whose monetary policy generally responds to changes in the balance of payments tend to make exceptions to this pattern of behavior mainly when they are in surplus. Similarly, compliance of monetary policy with balance-of-payments requirements in generally noncomplying countries tends to be found at times of deficits. . . . The loss of reserves is viewed with concern; but their accumulation . . .is viewed, in fact, with satisfaction or indifference.

The 1970 report by the Executive Directors on the role of exchange rates in the adjustment of international payments (IMF (1970, p. 38)), similarly suggests that adjustment pressures on deficit countries were stronger than those on surplus countries during the 1950s and 1960s because “. . . reserve accumulation is not subject to limit in the same way as exhaustion of reserves or of borrowing facilities.” Likewise, the figures on mean external imbalances for the larger industrial countries, shown in Tables 2 and 3, also point in the same direction. If the United States is excluded because of its special position, the countries with the largest mean imbalances and with the most persistence in those imbalances under the Bretton Woods system seemed to be the surplus countries.

The fact that payments imbalances and real exchange movements have both been more variable under floating rates than before, and the probability that nominal exchange rate appreciations are harder to sterilize under floating rates than reserve increases were under the adjustable peg, have seemingly combined to reduce such surplus/deficit asymmetries. Again, Tables 2 and 3 suggest that there has been no obvious pattern to mean payments imbalances along surplus/deficit lines.

In the case of asymmetries associated with the special role of the United States, the extent of asymmetry appears to have been reduced during the period of floating rates. The diversification of official reserve holdings away from the dollar toward other currencies (principally the deutsche mark and the yen) and, most important, the spread of liability settlement of external imbalances to even non-reserve-currency countries, have reduced the special privileges of the dollar.⁹⁰ But by the same token, any special burdens would also seem to have dissipated with the abandonment of dollar convertibility into gold in 1971 and with the sharp increase in the variability of the real effective exchange rate of the dollar over the 1973–83 period.

To summarize, the degree of external payments adjustment under managed floating may not be as swift or as predictable as would be desirable, but it is hard to attribute any shortcomings to systematic asymmetries across classes or types of countries.

Rules Versus Discretion in Adjustment

Yet another frequently heard criticism of the present exchange rate system is that it relies too much on discretionary policy actions by country authorities to secure external adjustment. The implication is that a more automatic system that relied on more specific rules would yield a better result. Further, there is a related criticism that both external adjustment and exchange rate stability would be enhanced if there was a more formal and more explicit mechanism for co-ordination of economic policies across countries. What can be said about these two criticisms?

To begin with, it is sensible to acknowledge that if one were to align alternative exchange rate systems along a spectrum according to either the degree of automatism of the adjustment process or the mix betweeen rules and discretion in initiating adjustment, the results would suggest that the present system is closer to the complete discretion pole than to the rules-only pole. In this sense, the pure gold standard with its automatic specie flow mechanism, the adjustable peg system with its clear implications for the subordination of domestic monetary policy to the exchange rate except during fundamental disequilibrium, the objective indicator system (based, say, on reserve levels) with its automatic trigger for the initiation of adjustment actions, mechanistic crawling peg schemes with their automatic adjustment of the exchange rate, or even a pure floating system with its complete prohibition of all official intervention in the exchange market—all could be considered less discretionary than the present system. In much the same way, it can readily be agreed that efforts at coordination of economic policies during the period of floating rates represent at most a middle ground along a hypothetical spectrum between completely activist and completely passive coordination strategies.⁹¹ In this context, efforts have gone beyond the exchange of forecasts and policy intentions to encompass occasional common actions (e.g., the U.S. dollar support package of November 1, 1978 or the setting of explicit targets for energy conservation efforts within the International Energy Agency); agreements on short-term exchange rate management policies (e.g., intermittent joint countering of disorderly market conditions); sometime agreement on medium-term expenditure and/or energy policies (e.g., the final agreements of the Bonn economic summit of 1978); and most recently, the extension of Fund surveillance to a multilateral Group of Five format.⁹² At the same time, coordination efforts have stopped well short of binding agreements on either exchange rate targets or rates of monetary expansion. For want of a better label, the present system might therefore be characterized as a discretionary and decentralized system, with loose coordination among the main players but with tighter coordination and disaster relief during crises.

Although the issues of rules versus discretion and of the optimal coordination of policies are still ones of little agreement, past experience would seem to point to the following four conclusions.

(1) Whatever the combination of rules and discretion, a prerequisite for successful external adjustment is the pursuit of stable, credible, and balanced macroeconomic policies at the national level. Without such policy behavior, the greater autonomy of policy instruments in more discretion-based systems will not produce greater policy effectiveness, and similarly without it, the policy rules in more rules-based or automatic systems will not be observed. This is undoubtedly what led Frenkel (1983 b, p. 112) to conclude:

• If governments were willing to follow policies that are consistent with the maintenance of a gold standard, then the gold standard itself would not be necessary; if, however, governments are not willing to follow such policies, then the introduction of the gold standard per se will not restore stability, as before long the standard will have to be abandoned.

That is, the need for good policy is not diminished by the presence or absence of automatic adjustment rules. This lesson is especially pertinent to countries that have adopted independently floating exchange arrangements. Two clear implications of the asset market view of exchange rates are: (1) that the current exchange rate will be much influenced by the expected future exchange rate; and (2) that the expected future rate will be much influenced by expected future macroeconomic policies. Since instability in present policies generates uncertainty about future policies, it is easy to see why stable and credible policies are a sine qua non for greater stability in exchange rates. Indeed, one of the main reasons why the adjustment process has not worked better under floating rates is because market participants have undoubtedly had difficulty in finding a good anchor for longer-term exchange rate expectations. The harder it is to make an informed judgment about the future course of policies, the more one can expect the erroneous extrapolation of short-term events. By the same token, a country that has established a credible long-term policy posture can count on the market to be more forgiving of short-term deviations because such deviations do not materially affect the long-term anchor. All of this is consistent with the conclusion of the Versailles economic summit meeting of 1982 that the maintenance of the internal and external values of currencies “rests primarily on convergence of policies designed to achieve lower inflation, higher employment, and renewed economic growth.”⁹³

(2) A second conclusion is that, whatever the mix between rules and discretion, effective external adjustment will not be forthcoming unless countries take into account the external repercussions of their own macroeconomic policy actions on their trading partners.⁹⁴ The notion that, under floating rates, each country can decide independently its own policy stance and mix and let the exchange rate settle all conflicts in the market place is neither realistic nor helpful. It is not a realistic notion because, as argued earlier, floating rates are not capable of providing enough insulation from other countries’ policy actions to make independent targeting work. It is also not helpful because failure to take into account other countries’ policies is likely over time to induce retaliatory actions by those who are unhappy with the verdict of the market. In the end, therefore, the path to external adjustment will be slower and the eventual equilibrium less satisfactory than they would be if some coordination of policies took place.

But taking external repercussions “into account” in setting domestic policies is not the same thing as being dominated by external considerations. It is unlikely, at least for the larger industrial countries, that a fully centralized and coordinated decision-making system would be either realistic or helpful. It would not be realistic because, even with the recent narrowing of inflation differentials among the largest industrial countries, it is unlikely that these countries would be willing to fully subordinate domestic policy, particularly monetary policy, to the goal of a fixed exchange rate. It would not be helpful because such an assignment rule would leave internal balance to be handled by market forces, or fiscal policy, or incomes policies. It seems wiser to continue to focus monetary policy on the still formidable task of achieving long-run price stability and sustainable growth.

What all this suggests for external adjustment during the period of floating rates is that such adjustment would have been smoother and exchange rate behavior would have been less variable, if there had been a more intensive and more regular effort to appraise the consistency of countries’ intended economic policies at the then prevailing exchange rates and, if, when such inconsistencies arose, there had been more compromise at the margin by all parties to reduce them.

The difficulties of achieving such improved coordination should not, however, be underestimated. In this respect, Polak (1981) has pointed out the natural limits to coordination and even the benefits of decentralized decision making (via the market) in those cases when negotiation is unsuccessful. He notes that coordination can be difficult because, inter alia: (1) exchange rates (and often interest rates) are by their nature competitive in the sense that one country’s gain is frequently the other’s loss; (2) the compromise of growth and inflation objectives at the national level often leaves little room for further compromise on demand policies at the international level; and (3) international decision making on floating rates is inherently more difficult than for infrequent par value changes. In a similar vein, Solomon (1982), commenting on the problems of securing monetary coordination in past economic summits, mentions that: (1) the perception of independent monetary policy may be necessary in some countries for sustaining confidence that monetary policy will not be inflationary in the long run; (2) the approach to formulating and conducting monetary policy differs greatly across countries; and (3) there is a logically prior need to coordinate domestic monetary policy with domestic fiscal policy.

But all these barriers to greater coordination of policies do not mean that efforts should not be made to improve the situation. Volcker (1977, p. 36) lends support to this position:

• Questions of the policy mix and timing are not rigid and preordained in any of our countries. I would like to think the time has come to be a little more open in putting those kinds of issues on this and other tables as part of the processes of comprehension and consultation. It seems to me a necessary part of running the floating rate system as effectively as I think it can be—and should be—run.

(3) The third conclusion regarding rules versus discretion in adjustment is that automatic adjustment systems usually turn out to be less automatic in practice than in theory and that very specific adjustment or exchange rate rules run the risk of becoming liabilities when the global environment changes in unexpected ways. The first point about automatic adjustment is best illustrated by the historical gold standard. Its specie flow mechanism is often held up as the ideal in automatic adjustment. But serious students of the historical gold standard seem to agree that internal balance considerations sometimes caused country authorities to offset or sterilize the effects of gold flows. Cooper (1982, p. 28), for example, notes that: “offsetting actions by central banks, in periods of contraction as well as periods of expansion, even took place often in the heyday of the historical gold standard.” Similarly, in describing the 1914–21 period in the United States, the report of the Gold Commission (U.S. Congress (1982, p. 67)) states that “the postwar increase in the quantity of money occurred because the Federal Reserve System did not observe the rules of the gold standard but exercised discretion. The subsequent collapse occurred because the power to manage money was not limited by the requirement to maintain gold reserve requirements.”

The perils of very specific adjustment rules can be seen by reference to a number of past systems or proposals. Perhaps the best example is the Bretton Woods system. As noted by Polak (1981), the Bretton Woods rules of the game were written so as to place all exchange rate changes under strict international supervision. By so doing, the danger of renewed competitive exchange alterations was minimized. This was just what was suitable in the early years of Bretton Woods; but during the late 1960s and early 1970s, these same rules became a liability when the need arose for a greater degree of exchange flexibility. Another example is that in the 1960s the idea of a “crawling peg,” based on inflation differentials, seemed like the right antidote for sticky exchange rates and for inefficiencies in central management of exchange rates. But after the plethora of real economic disturbances in the 1970s, that proposal’s neglect of the need for real exchange rate adjustment now seems like a serious drawback. To some extent, even the objective indicator proposal (based on reserves) considered by the Committee of Twenty appears outmoded in a world in which reserves are determined more by demand than by supply. In short, the moral would seem to be that limitations on the ability to forecast the future global economic environment put constraints on the usefulness of very specific adjustment rules—even if countries could be convinced to obey those rules strictly.

In summation, although the present exchange rate system relies more on discretion in initiating adjustment measures than do some alternative systems, there is little to suggest that performance would have been better over the past ten years if adjustment rules had been used instead. A more promising avenue would be to try to achieve a better anchor for longer-term exchange rate expectations by inducing countries both to pursue more stable and credible macroeconomic policies and, in framing these policies, to take more explicit account of their repercussions on other countries.