Recent events have cast doubt on simple traditional views as to the relationship between the exchange rate regime and policy interdependence or, more generally, economic interdependence. Fixed exchange rates enforce a great deal of economic and policy interdependence; flexible rates allow a lessening of the transmission of disturbances through markets and for increased policy autonomy. Or so the story went. That story now appears incorrect, or at least incomplete, on both factual and analytical grounds. Theory has suggested new—or rediscovered old—channels of transmission of disturbances under floating exchange rates.1 National business cycles have been as, or more, closely in phase across major industrial nations under managed floating as in the 1960s.

Abstract

Recent events have cast doubt on simple traditional views as to the relationship between the exchange rate regime and policy interdependence or, more generally, economic interdependence. Fixed exchange rates enforce a great deal of economic and policy interdependence; flexible rates allow a lessening of the transmission of disturbances through markets and for increased policy autonomy. Or so the story went. That story now appears incorrect, or at least incomplete, on both factual and analytical grounds. Theory has suggested new—or rediscovered old—channels of transmission of disturbances under floating exchange rates.1 National business cycles have been as, or more, closely in phase across major industrial nations under managed floating as in the 1960s.

Recent events have cast doubt on simple traditional views as to the relationship between the exchange rate regime and policy interdependence or, more generally, economic interdependence. Fixed exchange rates enforce a great deal of economic and policy interdependence; flexible rates allow a lessening of the transmission of disturbances through markets and for increased policy autonomy. Or so the story went. That story now appears incorrect, or at least incomplete, on both factual and analytical grounds. Theory has suggested new—or rediscovered old—channels of transmission of disturbances under floating exchange rates.1 National business cycles have been as, or more, closely in phase across major industrial nations under managed floating as in the 1960s.

The aim of this paper is threefold. It broadly discusses the relationship between the exchange rate regime and policy or economic interdependence. This is done in section II, which, after a general discussion of targets and instruments of policy under alternative international monetary regimes, discusses briefly transmission under flexible rates and the notion of a world business cycle. The section concludes that for many relationships interdependence is likely to be similar under fixed and flexible exchange rates but that the channels of transmission tend to differ. Section III speculates on the implications for the conduct of policy. It begins with a discussion of rules versus discretion, and of consistency versus credibility or commitment. It concludes with a brief discussion of the relevance of the argument to two issues raised by the organizers of this conference—the prospects for a regional arrangement such as the European Monetary System and for the return to fixed rates between Europe and the United States.

Before proceeding to such broad generalities, however, the next section offers some statistical background evidence on the correlation between main macroeconomic variables in six countries: the United States; the Federal Republic of Germany, France, and the United Kingdom; Japan, and Canada. The choice of these countries was dictated by a number of considerations. They are the six major industrial countries, and our analysis is confined to the “Atlantic” world. Data were readily available for most series of interest for these countries. And the division of the sample between the United States, the three European countries, and the two non-European ones should help detect patterns of association specific to these three groups. This should also help to shed some light on the issue of U.S. dominance over the world economy. Further evidence from a few recent studies of the transmission mechanism is cited to round out the section’s selective survey of factual evidence.

I. Some Empirical Evidence

Economic interdependence at a macroeconomic level can manifest itself in a variety of ways that range from co-movements (contemporaneously and at various lags) of national target or “endogenous” variables, such as price levels, inflation rates, real output movements, or interest rates, to co-movements (positive or negative) of instrument variables (not necessarily exogenous), such as money stocks or fiscal variables. Such associations across countries are only an indication that some interdependence exists. They reveal little as to the nature and source of the interdependence. In particular, much further analysis would be needed to determine whether causality runs from, say, the United States to the rest of the world; whether, instead, these associations are responses to common outside shocks with differences in national time profiles of response due to differences in national economic structure; whether the similarities in the behavior of output, prices, and interest rates are due to common policy responses rather than to transmission through private markets; or whether there exists a world business cycle, the origins of which may be monetary or real, of which national cycles are a part.2 Nevertheless, simple measures of association provide a useful start, especially if they help us find similarities and dissimilarities in behavior under fixed and flexible exchange rates.

This section presents, for the six countries mentioned in the introduction, correlations and a principal component analysis of changes in the following variables: the price level (consumer price index—CPI), real output (gross national product—GNP), interest rates (long-term government bond yields), money supplies (Mi), and government spending in real terms. All growth rates are calculated as the percentage change of the value of the variable in the current quarter relative to its value four quarters ago—except for interest rate changes, which are first differences over the previous quarter. The characteristics of the data are shown in Table 1.

Table 1.

Characteristics of the Data1

(Mean growth rates)

article image
Sources: International Monetary Fund, International Financial Statistics (IFS). Data on the five variables were obtained using the following lines on IFS country pages: CPI (line 64); money (line 34); GNP (line 99 ar); interest rates (line 61); government expenditure (line 82); exceptions noted below; Organization for Economic Cooperation and Development (OECD), French industrial production (for GNP); and U.S. Department of Commerce, government expenditure, 1960-67.

The first figure in each cell refers to the 1960, first quarter to 1971, first quarter period. The second figure refers to the period that began in the second quarter of 1974 and ended in the second quarter of 1982 for the CPI, the first quarter of 1981 for GNP, the fourth quarter of 1981 for M1 and interest rates, the third quarter of 1981 for government expenditure. For interest rate levels, the second figure refers to 1973, third quarter to 1981, fourth quarter.

Real government expenditure was obtained by deflating through the CPI. All growth rates, except interest rates, calculated as log X1 – log Xt-4. Interest rate changes are first differences. The series that were not available (indicated by …) are either incomplete over the sample period or too unreliable to be included.

All estimates were calculated for three periods: first quarter 1960-first quarter 1981 (or up to second quarter 1982 where available) and two subperiods, first quarter 1960-first quarter 1971 and second quarter 1974-first quarter 1981 (or up to second quarter 1982). The tables only report results for the two subperiods for clarity and to focus on the contrast between the fixed and the flexible exchange rate periods. The flexible rate period starts in the second quarter of 1974, as that observation represents the change of the variable relative to the second quarter of 1973.3 Quarterly observations of rates of change were calculated relative to the previous year’s corresponding quarter in order to smooth the series somewhat and to eliminate part of seasonality.

Table 1 contains few surprises and confirms the well-known broad characteristics of macroeconomic behavior in the 1960s and 1970s. The average rate of inflation almost triples when the flexible rate period is compared with the fixed rate period. That flexible rates allow for greater divergence of national inflation rates, as predicted by elementary analysis, is confirmed by the tripling of the standard deviation of mean national inflation rates around the six-country average between the first and second periods. Turning to the growth of output, there is a generalized drop for the second period, as witnessed by a decline in the average growth rate by more than half, from 5.3 percent to 2.2 percent. The decline in output growth is at least as generalized as the preceding high growth performance (compare also the standard deviations). On the surface at least, and contrary to expectations, there does appear to be “real” interdependence under flexible, as well as under fixed, exchange rates.

Averages of first differences in interest rates (row 4a) are not of much interest and are recorded here only for the sake of completeness. As for interest rate level figures (row 4b), they tell a familiar story. There is a significant rise in average interest rates in the second period and the dispersion in interest rate levels across countries rose substantially in the 1970s. As was true of inflation rates, interest rates appear to be more independent across countries during the flexible rate period than the fixed rate period.

As for the remaining two (instrument) variables, no clear pattern emerges at first sight. Nor does theory suggest a priori strong differences in their behavior under alternative exchange rate regimes.4 Mean money growth rises significantly for the United States and the United Kingdom and falls sharply for Japan. Excluding the latter country, average money growth does rise from 6.6 percent to 8.4 percent for the average of the remaining five countries (while the standard deviation of national money growth rates remains constant). Turning to government expenditure growth in real terms, the sample is limited to four countries only. Mean growth in that variable declines for every country in the sample during the flexible rate period. On average, however, it declines by significantly less than the real growth of output, indicating a rising share of government expenditure in total expenditure and, therefore, some common tendency in fiscal policy.

These broad tendencies are of little help in answering a number of relevant questions, among which five may be listed:

  • (1) Can one say anything about short- or medium-term interdependence of inflation rates, output growth rates, etc., across countries?

  • (2) Can one attribute some of the variation in national series to a common international factor, the remainder to individual country variation?

  • (3) Are answers to these questions the same for innovations in the series as well as for the series themselves?

  • (4) Are there systematic differences in the degree of association (interdependence?) across the two subperiods of our sample? Do these hold for all variables?

  • (5) Can one detect regional groupings in the degree of interdependence? Specifically, do European countries behave differently from, say, the United States?

The correlations and principal component analysis contained in Tables 2 to 5 may help to begin to answer these questions. A word of caution, however: correlations and principal components are essentially descriptive statistics, ways of organizing the data, that cannot in themselves provide any answer to a number of questions that are as interesting as, or more interesting than, those asked above. In particular, they cannot tell us anything about the source of a change in behavior between our two subsample periods, however much we may suspect that the change in the exchange rate regime or raw material shocks may have played a dominant role. Furthermore, they do not allow any inference as to causality or the nature and time profile of the “transmission mechanism.”

Table 2.

Inflation Rates

article image

CPVE = cumulative percentage of variance explained.

Table 3.

Real GNP Growth

article image

CPVE = cumulative percentage of variance explained.

Table 4.

Money Growth Rates

article image

CPVE = cumulative percentage of variance explained.

Table 5.

Interest Rate Changes

article image

CPVE = cumulative percentage of variance explained.

Tables 2-5 are each divided into two parts. Panel A reports simple correlations of the relevant variable (see Table 1 for definitions). Each cell contains two correlation coefficients, the first referring to the first subsample period, the second to the second subperiod. Panel B contains the results of the analysis of the principal components for both the original series and innovations in the series. Innovations were obtained by regressing, for each sample period separately, the variable on a constant, a time trend, three seasonal dummies, and four lags of the variable. The residuals are the innovations.

Principal component analysis is convenient for searching for a common element (or a number of common elements) that could explain the variance of a number of individual time series.5 Principal components are linear combinations of the original variables that explain increasingly higher proportions of the total variance of the variables. The first principal component is the linear combination that, by itself, explains the greatest portion of the variance (explainable by a single component), and so on for the second and other components. When the number of components is equal to the number of variables, all the variance has been accounted for. Obviously, series whose variance has a strong common element will require few principal components to explain a large fraction of the total variance. Tables 2-5 list the cumulative percentage of variance explained by the first three principal components (line CPVE at the bottom of the tables). Each column (P1, P2, P3) reports the factor loadings of each individual country series on, respectively, the first, second, and third principal component. These loadings are a measure of the association of each country’s series with the appropriate principal component.6 Provided that (as is true here) the original series have been appropriately standardized, the square of the factor loading is a measure of the variance of each country’s series explained by the corresponding principal component. Moreover, some inference as to regional groupings can be made by an examination of the pattern of factor loadings. Finally, note that in what follows the analysis is limited to contemporaneous correlations and no formal statistical test is used to support statements as to differences in various coefficients.7

Table 2 presents results for CPI inflation rates. The first clear implication is that contemporaneous correlations of quarterly observations of changes in national CPIs are much less strong than those among CPI levels, at least under fixed exchange rates, as is well known.8 Second, these correlations appear, if anything, stronger under flexible than under fixed exchange rates, contrary to what mean rates of inflation over longer time spans indicate. Third, no very clear regional pattern emerges. These findings are confirmed by the results of the principal component analysis of the original series. The first and second principal components account for, respectively, 14 percent and 9 percent more of the cumulative variance under flexible than under fixed exchange rates. Under flexible rates, the factor loadings of the first principal component turn out to be fairly high except for Japan and the United Kingdom, which also have fairly high negative loadings on the second principal component. This is the only weak regional pattern of note. Somewhat the same story is told by the results for innovations, although both the CPVEs and the edge of flexible over fixed rate explanatory power are substantially reduced. No clear regional pattern emerges. In brief, the main finding is the increased correlation (”commonality”) of price level changes during the second subsample period.

Results for the contemporaneous association of real output changes are more clear-cut. They are presented in Table 3. In terms of correlation coefficients, there is a notable increase of their value from fixed to flexible rates. Moreover, under flexible rates, all correlation coefficients are positive. In that second period, the highest coefficient is for France-Federal Republic of Germany, the lowest for Canada-Japan. For the second period, correlation coefficients are usually higher for output than for CPI changes. Turning to principal components, the explanatory power of the first principal component under flexible rates is quite high and equal to 69 percent of the original series. The edge of the CPVE under flexible rates over that under fixed rates is 37 percent and 21 percent, respectively, for the first and second principal components. In terms of factor loadings, consider first fixed exchange rates. The loadings on the first principal component (P1) are not high, except for Canada and the United States. Loadings on P2 are negative, except for these two countries. This indicates a regional pattern of variation of real output with one grouping composed of the North American countries, another of the other countries. For the second subsample period, factor loadings on P1 are higher for all countries except Canada. In terms of the fraction of variance in each country’s output changes that is accounted for by the first principal component (the square of factor loadings), the lowest is 51 percent for Japan, the second highest is the United States with 81 percent, and the highest is Germany with 90 percent. The regional pattern is less clear-cut than for the fixed rate period, although the United States and Canada may again constitute one group.

Turning to innovations in the real output series, the pattern resembles that obtained for the original series. A “reasonable” portion of variance is still explained by the first and second factors under floating rates. A United States-Canada versus other country groupings appears again for both the fixed and the flexible rate periods. In the first period, the factor loadings on P1 are positive only for the North American countries and are higher than those of other countries (except Japan) for P2. For the second period, the loadings for Canada and the United States on P1 are the lowest and are the only positive ones for P1. In brief, output changes seem more strongly associated under flexible than under fixed rates, with a strong contemporaneous common element (the first principal component) under flexible rates. Furthermore, a distinction between a European versus North American cycle does seem to make sense under both exchange rate regimes.

The results on money growth rates presented in Table 4 are, in contrast, much less clear. Some correlation coefficients are positive, some are negative; they are, on the whole, fairly low although they may tend to be slightly higher in the second than in the first period. The highest coefficients appear in the second period and are those that relate German money growth to that of France, the United Kingdom, and Canada. Some confirmation of these tendencies appears in the figures of Panel B of Table 4. CPVEs for the original series are higher in the second period than in the first period. Under flexible rates, all factor loadings on P1 are positive but the factor loading for the United States is very small, the one for Germany highest. This is in contrast to the fixed rate period where the factor loading on P1 is highest for the United States and lowest for the Federal Republic of Germany (in addition to a varied sign pattern for P1).

These patterns are not confirmed by the principal components analysis of innovations in money growth. This is the only time in all tables when the cumulative percentage of variance explained under flexible rates is not higher than under fixed rates. Moreover, the CPVEs are rather low. Also, the factor loadings on P1 for Canada and the United States are similar for the two periods and are high relative to the others, although fairly low in absolute value. In fact, there is reduced association for the Federal Republic of Germany in the second period. To summarize, no strong pattern of association among money growth appears across countries, nor are regional groupings suggested except in tenuous fashion. It remains true that the association appears somewhat higher under flexible than under fixed rates for the original series.

Finally,9 consider interest rate changes as in Table 5. Here correlation coefficients are all positive. Most (8 out of 10) are larger in the second than in the first period. The Canadian-U.S. interest rate correlation is highest for both periods. The cumulative percentage of variance explained by the first two principal components is quite high for the second period. Again, more association appears for the second period than for the first, both for the original series and for innovations. A regional grouping of North America versus Europe also finds some support in both periods. The factor loadings on P1 are highest for the United States and Canada and negative factor loadings on P2 appear for two of the other three countries in both periods. Furthermore, note that the results for innovations are quite similar to those for the original series. This is not surprising to the extent that interest rates follow a stochastic process that is close to a random walk. The major portion of changes in interest rates will then actually be innovations.10 Thus, factor loadings are very similar for the original series and for innovations. The latter have the highest CPVEs of the innovation series in the tables; and the proportion of the variance in interest rate innovations accounted for by P1 for each country is quite respectable, ranging from 30 percent for France to 77 percent for the United States under flexible rates.

Three general conclusions can be drawn from Tables 2 to 5. First, the degree of association among all four types of variables examined rises from the first to the second period. This appears clearly from comparisons of the cumulative proportion of variance explained by successive principal components. It remains true for both the original series and for innovations, with the single exception of money innovations where the CPVEs are basically the same under fixed and flexible exchange rates. Furthermore, factor loadings on P1 are frequently reasonably high, especially under flexible exchange rates, indicating that a substantial portion of changes in countries’ macroeconomic performance can be accounted for by a common factor. This is particularly true of changes in real output under flexible exchange rates. The strength of the connection in changes in real economic activity, especially in the second period, constitutes our second general conclusion and lends some support to the notion of a world business cycle.

Third, some evidence of regional factors can be garnered from the tables. In particular, a North America versus “Europe” (or “Europe” and Japan) distinction seems relevant for the behavior of changes in interest rates and real output under both fixed and flexible rates. A similar pattern may be sustained, albeit weakly, for money growth under fixed but not under flexible rates. This pattern does not seem to appear in the behavior of inflation rates. Finally, looking at the Federal Republic of Germany’s factor loadings on P1 for the various variables under flexible rates suggests a rising influence of the common factor on that country’s economy.

When assessing these findings it is useful to remember that for two reasons principal components will tend to underestimate the common component in the movement of various series. First, as they are based on contemporaneous correlations only, they tend to underestimate the common factor to the extent that it influences national variables with differing lags. Second, to the extent that national variables are subject to measurement error, the latter may lead to the unwarranted identification of a “country” factor and to a decrease in the measured influence of the first principal component. In assessing the findings, it is also useful to remember, again, that they do not allow any inference about causality. Nor do they allow us to attribute a clear cause to the increase in “interdependence” in the second subsample period.

One possibility is that the world economy has been subjected to larger common shocks during the second period, that there has been much more variance to be explained, rather than that “transmission” should have been higher in that period because of a change in the international monetary regime. To get some insight into this issue, the ratios of the standard deviations of innovations in the flexible to the fixed rate period are reported in Table 6. Reading across the lines of the table, it appears that a systematic increase in variance for all countries is present only for interest rate changes. Reading down, it appears that only for the United States (and except for M1 for the United Kingdom) does the variance of innovations systematically increase in the second in comparison with the first period. The question of why “interdependence” has increased recently thus remains to be explained. Both the change in exchange rate regime and the increased strength of common (e.g., supply) shocks remain candidates.

Table 6.

Ratios of Standard Deviations of Innovations

(Second period/first period)

article image

Two recent studies—one by Genberg, Saidi, and Swoboda (1982), the other by Huber and Saidi (1982)—help shed some light on these issues. The first study analyses in some detail the interdependence between U.S., Canadian, French, and German interest rates (and exchange rates). The data set is different and consists of monthly observations on short-term and long-term end-of-period interest rates. The results, however, are by and large consistent with those presented above. The Genberg, Saidi, and Swoboda paper investigates causality and the dynamics of response by means of causal ordering (Sims) tests and multivariate vector autoregressions (VARs). The results most relevant here are as follows:

  • (1) U.S. interest rate innovations appear to cause German and, especially, Canadian innovations. U.K. innovations are more exogenous.

  • (2) The fraction of the variance in foreign interest rate innovations explained by U.S. innovations is significant but not huge.

  • (3) As in this paper, the variance of innovations (from the VARs) does show an increase for the United States and the United Kingdom from the first to the second subperiod. In contrast to this paper, there is a decrease for Canada and the Federal Republic of Germany.11

  • (4) As indicated by a reduction of the portion of national interest rate variance accounted for by its own innovations, interdependence appears larger in the flexible than in the fixed rate period.

  • (5) Lags in transmission between countries are small, the largest effect always taking place within two months. Patterns also differ between the two subperiods.

In brief, as far as interest rate changes are concerned, these findings confirm higher interdependence in the flexible than in the fixed rate period. Part of that interdependence does seem to arise from transmission from the United States to other countries, but a substantial (probably more important) part from common response across countries to common underlying disturbances. Finally, that response tends to differ across exchange rate regimes indicating a different role for interest rate changes in the transmission mechanism under fixed and flexible exchange rates.

This last conjecture finds some confirmation in the study by Huber and Saidi (1982). Their results are derived from estimates of vector autoregressions for the United States, Canada, France, and the Federal Republic of Germany. For each country, four variables are typically included at first: money, prices, industrial production, and a short-term interest rate. From our vantage point, the most relevant result of the first part of their analysis is that responses across exchange rate regimes appear fairly stable for the United States, but that the dynamic structure of responses seems to differ across regimes for the remaining three countries. This suggests differences in the transmission mechanism and a role for foreign and domestic variables in explaining domestic business cycles. This hypothesis is examined in the next sections of the Huber and Saidi paper, where two “foreign” variables are included in the VAR systems for Canada, France, and the Federal Republic of Germany. The two variables used to capture foreign shocks are deviations from purchasing power parity (the real exchange rate) and movements in the U.S. interest rate. Only a few of the many interesting results obtained by Huber and Saidi can be mentioned here. First, a substantial part of innovations that appeared at first as domestic can, in fact, be attributed to foreign variables. Second, in the authors’ own words:12

… in the flexible rate period unanticipated real exchange rate movements account for a substantial fraction of the variance of domestic variables. The same role is played by innovations in the U.S. T-bill rate in the fixed exchange rate period. The results support the notion that there is a common underlying world business cycle, with an overlying domestic cycle. Differences in exchange rate regimes imply different channels for the effects of the international cycle. More exchange rate flexibility does not imply an insulation of domestic variables from foreign disturbances.

These conclusions are consistent with the evidence presented in this section. It remains for the next two sections to discuss briefly possible reasons for increased interdependence since 1973 and to draw some implications for policy.

II. Why So Much Interdependence Under Flexible Rates?

The preceding section suggests that there has been more interdependence under flexible than under fixed exchange rates. This is perhaps most surprising for the interdependence in real output in view of the main elements of the traditional literature on adjustment under fixed versus flexible rates. This section begins with a brief restatement of the traditional reasons adduced for interdependence under fixed and insulation under flexible exchange rates. It continues with more recent explanations for some degree of transmission under flexible rates. However, as is argued next, neither traditional nor more recent theorizing is entirely adequate to analyze the world business cycles, the existence of which is suggested by the data. The section concludes by sketching briefly the type of a priori reason that might support the world business cycle hypothesis.

The integrating characteristics of fixed exchange rates rest in the fact that, if they are strictly fixed and expected to remain so fixed, the world is effectively on a common currency standard. Changes in the currency composition of the public’s demand for money are met by adjustments in the currency composition of the supply of “world money” at the going fixed exchange rate. With goods arbitrage and in the absence of changes in barriers to trade, the rate of inflation of prices of traded goods will be uniform across countries; so, eventually, will be the rate of change of prices of nontraded goods in the absence of changes in technology and tastes. Any long-run differences in national inflation rates would thus reflect either measurement error or changes in relative prices due to changes in technology, tastes, or impediments to trade. To the extent that capital markets are integrated and that assets denominated in various currencies are otherwise close substitutes, nominal and real interest rates will tend to be equalized. We are in the realm of interregional economics with all the usual channels of transmission operating. There is some scope for national (state, regional) fiscal policy autonomy but monetary policy must be devoted to the maintenance of fixed exchange rates. Its scope is limited to determining the ratio of domestic to international reserve assets of the central bank within the bounds compatible with maintenance of the parity—that is, within the bounds imposed by the necessity of maintaining credibility of the central bank’s commitment to fixed exchange rates.

In such a world, if it is composed of either small or medium-sized countries of roughly equal size, and if international reserves are of the outside variety (gold, SDRs), we should observe close correlation among national price indices and among interest rate movements. For national money stocks the direction and extent of correlation is less clear. It depends on the source of the underlying disturbance and on the time profile of response. Thus, an exogenous increase in the domestic assets of one central bank may result in a negative correlation over some time span, as at first the domestic money supply expands with unchanged money supplies abroad and, then, as the payments adjustment mechanism works itself out, the domestic money supply contracts somewhat while money supplies abroad expand.13 The pattern of response and correlation would be different if the initial shock had come from a change in the demand for money (general or currency specific) or from a gold discovery. As for output changes, most models predict that a boom at home will spill over into a boom abroad through the trade balance.14 One may, however, witness different cycles at home and abroad if the cycles are industry specific (or if different industries react with different lags) and if industries are not spread evenly through the fixed exchange rate world. In brief, “target variables” (output, employment, inflation) are closely interdependent and positively correlated; policy instruments are effectively interdependent in the sense that monetary policy loses all but nominal autonomy, but their movements are not necessarily positively correlated across countries.

For large countries and/or reserve issuing countries, monetary policy does have some autonomy. A large country’s monetary policy can have a significant effect on the world’s money stock and on other macroeconomic variables by dint of its sheer size. A reserve currency issuing country can set the course of world monetary aggregates and of world prices, as its money stock is not constrained by the availability of outside international assets. This is one main juncture at which policy interdependence under fixed exchange rates can lead to policy conflict—i.e., if there is disagreement among the reserve issuing country and others as to the proper stance of macroeconomic (here mainly monetary) policy.

The “real world” differs, of course, from this idealized picture and offers some scope for independent movements in national macroeconomic variables and some autonomy for domestic macroeconomic policy. Exchange rates are not strictly fixed but fluctuate within margins, divergences in national price levels are possible through short-run relative price changes, countries are not all small, substitutability between assets is less than perfect, and some sterilization of reserve flows is possible, especially over the business cycle. The extent of policy autonomy is, however, likely to remain limited to the fairly short run, as long as exchange rates remain basically fixed.15

In one sense, the breakdown of the Bretton Woods system was brought about by the policy conflict among the reserve currency issuing country (the United States) and several other countries, as mentioned above. The expectation was that the freeing of exchange rates would lead to greater autonomy in national monetary policy and to less interdependence in both inflation rates and output changes. That floating rates afford more autonomy for monetary policy and more scope for divergence in national price level movements is consistent with common sense, most theoretical models, and facts (no country has experienced an isolated hyperinflation and at the same time kept its exchange rate fixed). That flexible rates should afford more insulation with respect to national output movements is consistent with primitive models of the open economy under floating.

The central feature of these (comparative statics) models that ensure insulation is that net exports are independent, in equilibrium, of the exchange rate. The latter changes until the foreign trade balance is equilibrated (equal to autonomous capital flows). Should domestic expenditure be inadequate to sustain the target of output, domestic monetary and fiscal policies can be adjusted appropriately. Nominal exchange rate changes, however, will affect the terms of trade especially if domestic prices are rigid, and hence real income. Nevertheless output need not be affected unless, as the classic Laursen-Metzler argument puts its, saving depends on the terms of trade.

The Laursen-Metzler article is the first in a long series of theoretical arguments that have questioned the output-insulating properties of flexible rates.16 It predicts that a boom abroad will create a depression at home; it also emphasizes the possibility for domestic expenditure policy of counteracting such inverse transmission. Another channel for transmission under flexible rates is through the capital account. If the latter is no longer exogenous, neither is the trade account or the net foreign demand for the home economy’s output. This channel of transmission is, of course, at the center of the Mundell-Fleming model. Assuming perfect capital mobility (here equalization of nominal interest rates across countries), Mundell shows that a fiscal (investment) boom at home is transmitted abroad, whereas a monetary expansion at home increases income there but reduces it abroad.17 The reason is that a fiscal expansion that raises the domestic interest rate creates a capital inflow, a depreciation of the foreign currency, and an increase in net foreign exports which raises output abroad. In contrast, a domestic monetary expansion induces a depreciation of the domestic currency, a capital outflow, and an increase in domestic net exports (and output), the counterpart to which is a fall in foreign exports and output.

One major problem with the Mundell-Fleming model is its static nature, in particular with respect to expectations. Its assumption that exchange rate expectations are static justifies the assumed equality of nominal interest rates across countries but rules out any exchange rate dynamics related to movements in expectations. The asset approach models of the 1970s, notably those proposed by Dornbusch, remedy this deficiency by requiring the equality of the domestic nominal interest rate with the foreign rate plus the expected rate of depreciation of the domestic currency. Combined with a model of expectations formation and the assumption that output and/or prices adjust more slowly than exchange rates, such models produce interactions between domestic monetary policy, interest rates, exchange rates, and output and/or prices that appear suggestive of recent experience. In particular, the exchange rate is seen to overshoot, in the short run, its long-run equilibrium value in response to unanticipated monetary disturbances.

Although these models have usually been used to study the effects of domestic disturbances given foreign variables (the small country assumption), their extension to a two-country framework and to the analysis of transmission readily suggests itself. Monetary contraction at home by appreciating the real exchange rate (depreciating the foreign currency in real terms) creates inflation and an expansion of output abroad. This is precisely the example used by Dornbusch (1982) in a recent paper to analyze the so-called dollar problem. Recent monetary tightness in the United States has been cited as a major source of policy conflict between the United States and Europe. High U.S. interest rates have prevented European countries from lowering their interest rates to stimulate output in Europe, as such a lowering of interest rates would result in currency depreciation and inflation. Worse, the tightening of U.S. monetary policy, while it reduces inflation there, increases inflation in Europe without much of an increase in European growth because Europe suffers a terms of trade deterioration that offsets in part the benefits to output from increased competitiveness. Europe is in the awkward position of having to tighten its monetary policy to fight the imported inflation (due to U.S. deflation), thus losing the benefits from the added growth.18

Traditional and more recent models do play an important role in delineating a number of possible channels of transmission of shocks from country to country and in drawing attention to potential domestic and foreign effects of policy actions. In most cases, however, they predict negative transmission of foreign disturbances, either for output or for price changes. They also point to greater scope for domestic policy to counteract the effects of foreign shocks under flexible rather than under fixed rates.19 The increased interdependence of both output changes and short-run price changes under floating rates that has been documented above suggests that models endowed with a richer structure and stochastic specification are required to provide an explanation of international and domestic business cycles.20 Such an explanation is, of course, an exceedingly difficult task; it would, at an empirical level, require the specification and estimation both of the dynamic relationships among macroeconomic variables within and across countries and of the stochastic structure of the shocks to which the international economy is subjected.

Nevertheless, one might suggest that the facts of interdependence under flexible rates support the notion of a world business cycle on which national cycles are superimposed. Transmission is, of course, part of the story but there are several reasons why current models are not entirely adequate to deal with the notion. Among them the assumed structure of output and production figures prominently. The standard models deal with final outputs and with a few homogeneous categories of goods. The aggregation criterion for these composite goods is whether they are exports or imports, traded or nontraded, rather than the structure of their production or the industrial category to which they belong.

The point is that the common element in cyclical behavior across countries may be the result of shocks that affect specific stages in the production process and specific industries. Alternatively, these shocks may affect demand across the board but the dynamic response of specific outputs will differ depending on the stage in the production process and across industries. The result of either shock will be the appearance of a common transnational cycle and of national deviations from the common trend, independently of the exchange rate regime. The extent of common cyclical response and of national differences will depend mainly on the distribution of industries across countries; the more even the distribution the greater the coherence with the world cycle—and vice versa. It is still true that the nature of the exchange rate regime is likely to affect the response pattern and that “transmission” will play a role. But, for some purposes, it may be more informative to divide the world into industrial sectors rather than along national borders.

There are a number of obvious examples. The oil price rise is the standard prototype of a shock that has affected most industrial countries simultaneously but differently, depending on the structure of their production. To study the nature of the response requires a model that admits of imported inputs and differing structures of production across countries.21 Similarly, a number of industrial sectors may be affected in much the same manner, in whichever country they are located. The automobile sector is one example. To the extent that the typical response to changes in aggregate demand is different for automobiles than for other goods, the cycles for countries with a heavy concentration in automobile production will tend to be correlated more closely with each other than with others. Technological change also tends to have industrywide transnational effects rather than country specific effects. In an increasingly integrated industrialized world, common business cycles should arise whether exchange rates are fixed or not.

The increasing degree of economic integration in the last few decades (whether in terms of trade in goods, of trade in assets, or of socioeconomic attitudes and institutions) may account partly for increased interdependence irrespective of the exchange rate regime. In such an integrated world economy, expectations will also tend to show common movements. Industrial gloom (or bullishness) does tend to spread. In addition, common policy responses to common shocks may reinforce the tendency toward a world business cycle. The policy autonomy afforded by flexible rates in fighting shocks originating abroad is of little avail if the shocks are not transmitted along traditional lines but affect several countries simultaneously and directly.

In brief, the close interdependence in business cycles that was witnessed in the 1970s may well find its origin in a world business cycle in a closely integrated international economy rather than in the transmission of disturbances from one country, or group of countries, to the rest of the world. This is not to deny that traditional channels play a role; but not necessarily a dominant one. World business cycles can occur under both fixed and flexible rates, although the channels of transmission will most likely differ. There is, however, one sense in which the changeover to flexible rates may have altered the dynamics of output changes within and across countries. It has been associated with a basic change in domestic monetary regimes toward fiat money that may have altered the nature and extent of the shocks impinging on the world economy.

III. Some Policy Implications

Strong interdependence and the existence of a world business cycle under floating exchange rates do not imply very novel messages for policy. These findings do, however, provide some background against which to discuss briefly two questions addressed to me by the organizers of this conference: the extent to which a currency union arrangement like the European Monetary System influences the response of countries within the union to external shocks; and the arguments for or against a return to a system of fixed exchange rates between Europe and the United States. These very broad questions cannot be answered satisfactorily here, and the following brief remarks will only touch on them and raise a general question concerning the nature of policy commitments.

As to the first question, the theoretical argument would be superficially that the European Monetary System should help insulate member countries from shocks external to the monetary union, while increasing interdependence and susceptibility to shocks arising within the union. This answer is, of course, predicated on the assumption that flexible rates toward the rest of the world insulate the union as a whole, while fixed rates within the union increase interdependence among the members. The assumption does not receive strong support from the empirical evidence we have reviewed. Of course, it may be that the increased interdependence since 1973 is not due to the adoption of floating rates; interdependence might have been as strong, or even stronger, had fixed exchange rates been maintained. However, for shocks to all industrial countries, such as raw material price increases, there is little that can be achieved by floating against nonmembers.

There is one respect in which a currency union might help reduce the impact of outside shocks. And that is by reducing inside shocks. More specifically, a currency union may reduce uncertainty with respect to the future course of individual members’ monetary policy, thus reducing monetary shocks (innovations). The contribution that lowering the monetary shocks occurring inside the union may make to insulate the members stems partly from the role of real exchange rate variations in transmitting shocks under flexible rates (see Section I). With lower unexpected monetary variability there should be less real exchange rate variability. There are two prerequisites for this to occur. First, uncertainty as to individual members’ monetary policy should not be replaced by uncertainty as to the union’s monetary policy, be it the result of some ill-defined averaging process or that of a deliberate “dollar policy” of the union. The second is that the commitment to fixed rates within the union by individual members be credible.22 It is left to the reader to decide whether these prerequisites are met in the European Monetary System.

One further aspect of the empirical findings may have a bearing on this issue. The principal component analysis indicated the existence of a regional factor, especially with respect to co-movements in interest rates, on the one hand, and in real output, on the other. The factor suggested that our sample divides into a North American versus European (or rest of the world) group. This might suggest that Europe constitutes an adequate grouping for a currency union. However, note that the grouping holds for the fixed as well as for the flexible rate period and thus does not constitute, prima facie, a case for a currency union for the three European countries.

Turning to a return to fixity of exchange rates between the United States and Europe, the argument developed in this paper implies that interdependence in real output does not establish a case against such a return. The reason is that output interdependence appears as strong or stronger under flexible as under fixed rates. Moreover, the advent of floating has not diminished the extent to which U.S. variables “cause” European ones (in the VARs mentioned above), even if the channels of transmission are different. And popular views as to recent transmission of U.S. monetary policy indicate that many observers believe U.S. dominance to be even stronger now than it was in the 1960s. The reluctance to return to fixed exchange rates has other grounds.

In the first place, it remains true that price level dependence is much stronger under fixed than under flexible rates and that adoption of the former would enforce much greater convergence of mean inflation rates over the medium term than was true in the 1970s. It is also true that fixed exchange rates would deny countries monetary policy autonomy, however illusory the exercise of that autonomy may be in controlling national business cycles. In the second place, fixed exchange rates are part of a package that governments are loath to discuss let alone agree on, because to work properly, fixed exchange rates require adhesion to a set of rules and agreement on acceptable international reserves and on the process generating their growth. Once these rules are accepted fixed exchange rates force consistency of national monetary policies, at the cost (which may be a benefit) of loss of national monetary autonomy.

The case for not returning to fixed rates between Europe and the United States is, in this perspective, the lack of agreement on such rules. Floating rates have the advantage of letting monetary policy conflicts be resolved by market forces. But inconsistent or unstable policies will also cause an unsatisfactory working of flexible rates—even if that system remains viable during policy conflicts. Hans Genberg and the author have argued elsewhere that floating rates may require as much commitment as fixed rates to work properly.23 That is, a properly functioning system of flexible rates does not remove the need for rules. The need for policy responses that represent medium- to long-term commitments does not change fundamentally with a change in the exchange rate regime, although the form of the commitment does change. This conjecture is based on recent work on rules versus discretion and on the inconsistency of optimal plans, notably by Kydland and Prescott (1977) and by Barro (1982). This is not the place to restate the argument. Suffice it to say that the absence of precommitted monetary policies (as would be implied by the pursuit of optimal discretionary policies) might well lead to either increased variability of real exchange rates or suboptimal outcomes from a welfare point of view, or both. Note that precommitted policy does not have to be a set of binding laws but can be the outcome of consistent behavior and the absence of optimal control policies.

The basic issue, in this context, is not so much whether it is preferable to have fixed or floating rates between Europe and the United States. It is, rather, that precommitted and credible policies be adopted under either alternative. Such credibility may be simpler to achieve for nonreserve currency issuing countries under fixed rates. “All” they need to do is to peg and follow essentially gold standard rules. For the reserve issuing country there is no simple precommitted rule that can be readily incorporated into the country’s “monetary constitution.” To that extent, credible precommitment may be more difficult to achieve for such a country. Be that as it may, the adoption of implicit or explicit rules that are adhered to and are credible can make an important contribution to eliminating unexpected policy shocks as a major source of cyclical fluctuations in world output.

REFERENCES

  • Barro, Robert J., High and Volatile Inflation—the Role of Discretionary Monetary Policy,paper written for the Conference on Monetary Policy, Financial Markets, and the Real Economy (International Center for Monetary and Banking Studies, Geneva, June 1982).

    • Search Google Scholar
    • Export Citation
  • Bruno, Michael and Jeffrey Sachs, Supply versus Demand Approaches to the Problem of Stagflation,” in Herberg Giersch, ed., Macroeconomic Policies for Growth and Stability (Tubingen, 1981), pp. 1560.

    • Search Google Scholar
    • Export Citation
  • Dhrymes, Phoebus J., Econometrics: Statistical Foundations and Applications (New York, 1970).

  • Dornbusch, Rudiger, Equilibrium and Disequilibrium Exchange Rates,paper prepared for the Conference on Monetary Policy, Financial Markets, and the Real Economy (International Center for Monetary and Banking Studies, Geneva, June 1982).

    • Search Google Scholar
    • Export Citation
  • Dornbusch, Rudiger, and Jacob A. Frenkel, eds., International Economic Policy: Theory and Evidence (Baltimore, 1979).

  • Genberg, Hans (1977a), “The Concept and Measurement of the World Price Level and Rate of Inflation,Journal of Monetary Economics, Vol. 3 (April 1977), pp. 23152.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Genberg, Hans (1977b), “Policy Autonomy of Small Countries” in Inflation Theory and Anti-Inflation Policy, Proceedings of a conference held by the International Economic Association at Saltsjōbaden, Sweden, ed. by Erik Lundberg (New York), pp. 183208.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Genberg, Hans, Nasser Saidi, and Alexander K. Swoboda, American and European Interest Rates and Exchange Rates: U.S. Hegemony or Interdependence?paper prepared for the Conference on Monetary Policy, Financial Markets, and the Real Economy (International Center for Monetary and Banking Studies, Geneva, June 1982).

    • Search Google Scholar
    • Export Citation
  • Genberg, Hans, and Alexander K. Swoboda, Fixed Exchange Rates, Flexible Exchange Rates, or the Middle of the Road: A Reexamination of the Arguments in View of Recent Experience,paper presented at the 1982 Wingspread Conference on the Evolving International Financial System.

    • Search Google Scholar
    • Export Citation
  • Huber, Gerard, and Nasser Saidi, Business Cycles under Alternative Exchange-Rate Regimes: 1960-1981,preliminary draft (Geneva, August 1982).

    • Search Google Scholar
    • Export Citation
  • Jones, Ronald W., and Douglas D. Purvis, International Differences in Response to Common External Shocks: The Role of Purchasing Power Parity” in Emil Claassen and Pascal Salin, eds., Recent Issues in the Theory of Flexible Exchange Rates, Vol. 8 (Amsterdam, 1983).

    • Search Google Scholar
    • Export Citation
  • Kydland, Finn E., and Edward C. Prescott, Rules Rather than Discretion: The Inconsistency of Optimal Plans,Journal of Political Economy, Vol. 85 (June 1977), pp. 47391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Mundell, Robert A., International Economics (New York, 1968).

  • Theil, Henri, Principles of Econometrics (New York, 1971).

Comments

Jacques R. Artus

One of the interesting arguments advanced by Swoboda is that the interdependence among industrial countries, in part, reflects the integration of their production systems through trade in intermediate products. Another and related argument is that the interdependence among industrial countries, in part, reflects their common dependence on imported fuels and raw materials. When there is a change in the prices of these products, as a result of a supply shock or a change in demand in one of the major industrial countries, this constitutes a real shock rather than a monetary shock. Such a real shock has a similar and simultaneous impact on all industrial countries importing the products affected. Those arguments are not original, but Swoboda is quite right to point out that they have been largely neglected in both traditional and more recent theoretical models.

Swoboda is also quite right to point out that, in an integrated world, it is not only the expectational factors related to economic phenomena that tend to be common but also the political and social factors. He fails, however, to provide us with much explanation or illustration of this important point. I would have liked to see him mention some of the common tendencies that have dominated the evolution of most industrial countries during the 1960s and 1970s. I have in mind 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. In my view it is largely because of these common tendencies, 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.

Where I do not follow Swoboda completely is in his argument that there is now a “world business cycle.” No doubt activity has tended to go up and down concurrently in the main industrial countries during the 1970s and early 1980s. However, much of this correlation results also from the two waves of oil price increases that followed the Middle East conflict in 1973 and the Iranian revolution in 1979. The increased integration of the production systems of the industrial countries and the common nature of many expectational, political, and social factors are important elements of interdependence, but this does not mean that the amount of interdependence is such as to lead to a world cycle. There, the burden of proof is on Swoboda. In addition, the notion of a cycle implies a certain regularity as to frequency and amplitude, and it is unclear from where this regularity would emanate. As for myself, I would not be surprised if 1983 and 1984 were to be characterized by a marked recovery in the United States and a continuation of stagnation in many other industrial countries, thus bringing about an early death to the notion of a world cycle.

I have even more problems following Swoboda in his empirical work. Both his correlation analysis and his principal-component analysis confirm a well-known fact, namely, that the cross-country correlation of price level, output, and interest rate changes was higher during the 1970s and early 1980s than during the 1960s. What can we conclude from this as to the implications of flexible exchange rates for economic interdependence? We can conclude that the move toward greater exchange rate flexibility did not protect countries from the common real external shocks experienced during the 1970s and early 1980s, or from the common problems inherited from the past. This, however, is not too useful; there have never been any serious beliefs that flexible rates would do this. In particular, as pointed out by Swoboda, it has been recognized for many years that flexible rates would not fully insulate countries from real shocks originating abroad. The meaningful question is whether flexible rates reduce or enhance interdependence, in particular, in the presence of monetary shocks. Swoboda’s empirical analysis fails to cast any light on this point.

Swoboda is careful to point out the limits of his empirical analysis. At the same time, he does suggest that there is no evidence that common shocks have been larger during the flexible rate period than during the 1960s, and thus he leaves the reader with the impression that the change in exchange rate regimes is likely to be behind the increase in interdependence. In my view, the argument on the size of common shocks is quite weak. It is based on the observation that the standard deviations of “innovations” in prices and output changes are not larger during the flexible rate period than during the 1960s. But such an observation does not in any way contradict the hypothesis that the shocks are more “common” during the flexible rate period. In addition, because of the way the innovations were obtained,1 the standard deviations of innovations are more likely to reflect the erratic nature of quarterly series than the size of major shocks. On the whole, it is hard to believe that the simultaneity across countries of the increases in prices, followed by decreases in output, in 1974-75 and 1980-81 was related mainly to the nature of the exchange rate regime rather than to the two waves of oil price increases.

All this is not to say that flexible rates have provided the insulation effects that most economists expected from them, in particular, for external monetary shocks. In fact, as I have argued elsewhere,2 there are reasons to believe that they did not, and that, in some cases, they even increased the interdependence among countries. To show this convincingly, however, the empirical study would have to separate the effects of exchange rate flexibility from the effects of other factors, including the effects of common real shocks. In particular, the study would have to focus on the problem of exchange rate overshooting, its repercussion on domestic prices, and the tendency for national authorities to adjust their monetary policies to reduce the impact of external monetary shocks on their exchange rates.

Swoboda does refer in his theoretical section to the work of Dornbusch on the overshooting mechanism and on the implications for European countries of the shift to a policy of monetary restraint in the United States. He also refers to the tendency for European countries to fight the imported inflation resulting from the appreciation of the U.S. dollar by tightening their own monetary policies,3 and explains that this can ultimately lead to a positive transmission of foreign disturbances. But he places little weight on this explanation because he finds no evidence in his empirical analysis of a strong pattern of association among money growth across countries. Here again, part of the problem may be that Swoboda is focusing narrowly on the pattern of association on a quarterly basis, while what is really important is the pattern of association among broad swings in money growth. Chart 1 shows the evolution of the most closely watched monetary aggregates for the United States (M2), Japan (M2 + CDs), and the Federal Republic of Germany (central bank money) from 1976 to 1982, after deflating them by the domestic demand deflators and the levels of potential gross domestic product (GDP) in real terms, and after adjusting them for long-run changes in velocity. It is rather difficult to argue that there has been no pattern of association among these variables over the past seven years.

Chart 1.
Chart 1.

Adjusted Real Money Stocks, 1976-821

Citation: IMF Staff Papers 1983, 001; 10.5089/9781451956658.024.A004

1 In each country, the nominal stock of money was deflated by the aggregate domestic demand deflator and the level of potential output, then adjusted for the long-run change in velocity. The following annual percentage changes in velocity were used: +¼ percent in the United States, –1½ percent in Japan, and –½ percent in the Federal Republic of Germany. The growth rates of potential output were assumed to be 5 percent from 1976 to 1982 and 4½ percent in 1983 in Japan; 3 percent from 1976 to 1978 and 2¾ percent from 1979 to 1983 in the United States; and about 2½ percent from 1976 to 1980 and 2 percent from 1981 to 1983 in the Federal Republic of Germany.

To conclude, let me look briefly at the European Monetary System (EMS). The theoretical argument is that the System will help to isolate member countries from an external shock, in particular a monetary shock originating in the United States. Swoboda comments briefly that the empirical evidence does not provide strong support for this argument. In a recent paper,4 I estimated and simulated a small model in order to study the effects of a shift to monetary restraint in the United States on the deutsche mark-dollar exchange rate and the German economy. I concluded that when Germany does not change its interest rates and other EMS countries keep their real exchange rates constant vis-á-vis the deutsche mark, a shift to monetary restraint in the United States of a magnitude similar to the 1979 shift leads to an extremely sharp depreciation of the deutsche mark vis-á-vis the U.S. dollar, a moderate increase in the German domestic deflator, and a moderate increase in the German GDP gap (ratio of actual over potential GDP). If other EMS countries appreciate in line with the U.S. dollar, the increases in the German domestic demand deflator and the GDP gap become much larger, about twice as large as in the previous simulation. There is, therefore, no doubt that, in this case, Germany benefits from constant real exchange rates with other EMS countries.

But the results are quite different when simulation of the model is made under the assumption that Germany chooses to keep its interest rates in line with U.S. interest rates, thus duplicating the U.S. shift to monetary restraint. In this case, if all the other members of the EMS follow suit, then practically all the major industrial countries move together toward a policy of monetary restraint and the effects tend to cumulate into a large world recession. To conclude, the EMS helps to isolate activity in member countries when the EMS is floating vis-á-vis the rest of the world. When the EMS implies that most countries adjust their interest rates together, it aggravates the effects of the external shock.

Jacob A. Frenkel

Swoboda’s paper on “Exchange Rate Regimes and U.S. European Policy Interdependence” deals with key issues. It contains useful evidence and stimulating analysis. My remarks are divided into two parts. The first deals with the empirical findings and their interpretation and the second concerns the relevance of the analysis for the choice among alternative exchange rate regimes.

empirical findings and interpretation

In comparing the empirical regularities that have characterized the fixed exchange rate period of the 1960s with those of the flexible exchange rate period of the 1970s, Swoboda notes that, generally, the move to the flexible exchange rate regime has been associated with a rise in the mean and in the variance of the rates of inflation while it has been associated with a decline in the mean and in the variance of the rates of growth of real GNP. It is relevant to point out, however, that these changes in the means and the variances have resulted in almost no change in the coefficients of variation of the two series. Therefore, prior to drawing an inference from these findings it would be useful to establish which of the measures of variability are more relevant for such comparisons.

The second set of findings relates to correlation and to the analysis of principal components. Swoboda’s computations show that, on the whole, the cross-country correlations among inflation rates, real GNP growth, money growth rates, and interest rate changes have not been altered dramatically by the move to flexible exchange rates. In fact, based on the principal component analysis of these variables, Swoboda concludes that the degree of association among countries has even slightly increased during the flexible exchange rate period. Based on these findings, Swoboda advances the interesting hypothesis that the economic system may be governed by a world business cycle rather than by the traditional mechanism of an international transmission of local business cycles.

Prior to discussing this interpretation of the evidence, it might be useful to ask what are the alternative circumstances that could have caused such patterns of association. In principle, patterns of association that are similar over periods might indicate that, first, the effective exchange rate regime has not changed much over time. That is, the decade of the 1960s was not in practice a completely fixed rate period, while the 1970s was not in practice a completely flexible rate period. Second, these patterns could indicate that, even though the periods were characterized by sharply different exchange rate regimes, the similarities in the associations among countries indicate that the prevailing theory of the transmission mechanism needs revision. Third, it may indicate that the shocks confronting the system during the 1960s were very different from those of the 1970s. Finally, the observed correlations reflect the combination of differences in exchange rate regimes, differences in the nature and origin of the shocks and the patterns of government policies that have been adopted in face of the shocks. As the observed data reflect the outcome of the joint influence of the exchange rate regime, the nature of the shocks, and the response of policies, it is evident that one may not ascribe the entire difference between the periods to the differences between the prevailing exchange rate regimes. To avoid any misunderstanding in this respect I would even have suggested modifying the title of the paper from “Exchange Rate Regimes and U.S. European Policy Interdependence” to “U.S.-European Policy Interdependence during the 1960s and 1970s.”

Swoboda’s hypothesis that the co-movements of the various series may represent a world business cycle is interesting. It could be explored further by changing the sample of countries in the analysis. At present the countries under consideration represent mainly oil consuming countries. As all have been confronted by a similar external oil shock, it is conceivable that the positive correlation among them reflects the outcome of a common policy response to a common external shock rather than indicating a pattern which is specific to the prevailing exchange rate regime. It would be useful to extend the sample so as to include an oil producing country. If the extended sample also produces a similar pattern of correlations, then the world business cycle hypothesis would receive further support.

Finally, it is relevant to note that if the observed record reflects the reaction of the various countries in the sample to a common set of shocks, then the received theory of the transmission mechanism may still be perfectly valid for the analysis of the mechanisms by which a specific policy shock is transmitted from one country to the rest of the world.

choice of exchange rate regime

One of the interesting findings that are reported by Swoboda concerns the regional pattern of variations of the various variables. He finds that, based on the variations of at least some of the variables, one could classify countries into groups, of which one would consist of the North American countries and the other of the European countries in the sample.

There remains, however, the question what does this country grouping indicate? Does it indicate that the countries within each group are sufficiently similar so as to justify a monetary union (or a variant thereof like the European Monetary System)? Or should they consider joining a customs union? Or should they even consider a political union? Since each one of the various forms of integration depends on different considerations, it might be useful to incorporate into the analysis some of the factors that determine whether a specific union is being called for. For example, it is reasonable to argue that prospective members of a monetary union should possess common tastes concerning the common course of inflation. The desired rate of inflation in turn depends, among other things, on the efficiency of the fiscal system. Similarly, one could predict monetary mergers for countries among which there exists a positive correlation of exogenous disturbances, so that expansionary or contractionary policies are mutually consistent with the needs of all members. These and similar considerations might be usefully added before concluding from the patterns of the country grouping whether the European Monetary System does or does not include the appropriate countries.

One of the criteria employed in Swoboda’s analysis of the choice of an exchange rate regime is the degree of policy autonomy that is afforded by alternative regimes. In evaluating this criterion, however, three issues need to be clarified. First, a distinction should be made between the autonomy in the use of a policy instrument and the efficacy of this instrument in influencing a policy target. For example, in principle, a flexible exchange rate regime affords full autonomy in the use of monetary policy—in the same sense that, in contrast to a fixed exchange rate regime, the monetary authority controls the nominal money stock—but one may envisage a situation whereby the control over the money stock does not provide the authorities with the power to alter the course of real economic activity. How should we evaluate such autonomy? Second, is it possible to ask what is so good about autonomy? It is obvious that one needs a criterion with which to evaluate the cost and benefits of policy autonomy. Here it is important to know whether the policymaker is likely to abuse or properly use the autonomy granted to him and, as is evident, no single general answer can be given to this question. Third, the optimal exchange rate regime depends on the nature and the origin of the shocks that affect the economy. It is clear, therefore, that the question of the optimal degree of autonomy that should be granted in the use of various policy instruments cannot be answered properly without reference to the characteristics of the shocks that affect the economy.

Swoboda concludes his paper with a discussion of the issue of rules versus discretion. He emphasizes, correctly, that the commitment to fixed rates must be credible, and that the adoption of credible rules that are being adhered to can make an important contribution by eliminating policy shocks. Granted all these benefits, the intriguing questions that remain are why do these beneficial rules always get broken? and will the world monetary system eventually turn into a regime with a credible stable rule?

In order to gain perspective, it may be instructive to recall that over a century ago, in his Principles of Political Economy, John Stuart Mill regretfully concluded that

So much barbarism, however, still remains in the transactions of most civilized nations, that almost all independent countries choose to assert their nationality by having, to their own inconvenience and their neighbours’, a peculiar currency of their own.

In predicting the future Mill believed that eventually the international monetary system would evolve into a unified currency area, a process that would be brought about by “the progress of political improvement.”

Mill’s prediction has obviously been refuted. National governments have generally preferred to have their own monetary unit and international monetary rules have generally been broken. The apparent inconsistency between what seems to be in the general interest and what is actually being practiced may be rationalized by recalling that, as a general rule, governments tend to discount the future heavily since their time horizons are relatively short. Consequently, faced with a conflict between internal and external targets, elected officials (who wish to be re-elected) will typically sacrifice external obligations to domestic goals by renouncing previous commitments to the international rules of game. In view of this, it is important to choose an exchange rate regime that is capable of accommodating this general tendency of policymakers. Swoboda’s reasoning seems to be consistent with this principle. He thus argues that “Floating rates have the advantage of letting monetary policy conflicts be resolved by market forces,” and he goes on emphasizing correctly that “inconsistent or unstable policies will also cause an unsatisfactory working of flexible rates—even if that system remains viable during policy conflicts.” I fully agree with this argument as it indicates that in comparing exchange rate regimes, the mere survivability of the system under a specific exchange rate regime need not imply the superiority of this regime. Rather, the comparison should be based on the cost of sustaining the system under alternative exchange rate regimes. And here one needs to evaluate the cost of the various crises under an adjustable peg regime with the corresponding costs under a regime of flexible rates.

My final remark concerns discipline. It has been argued that the obligation to peg the rate would alter the conduct of policy by introducing discipline. If true, then this argument suggests a strong case for fixed exchange rates. Experience seems to suggest, however, that national governments are unlikely to adjust the conduct of domestic policies so as to be disciplined by the exchange rate regime. Rather, it is probable that the exchange rate regime is more likely to adjust to whatever discipline national governments choose. It may be noted in passing that this is one of the more potent arguments against the restoration of the gold standard. 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.

*

Alexander K. Swoboda is Professor of Economics at the Graduate Institute of International Studies and Director of the International Center for Monetary Studies in Geneva, Switzerland. A Yale graduate, Mr. Swoboda has taught at the University of Chicago, Johns Hopkins University Bologna Center, the London School of Economics and Political Science, and Harvard University. His main writings and chief research interests are in international monetary economics, macroeconomics, and international banking and finance.

1

This paper will touch only briefly on channels of transmission under alternative exchange rate regimes as this is the topic of Dornbusch’s paper for this conference.

2

A number of statistical methods that go well beyond the scope of this paper begin to deal with such questions. These include tests of causal ordering, vector autoregression methods, and index models.

3

Except, of course, for interest rates, where the second quarter of 1974 is the change relative to the first quarter of 1974.

4

See also the next section.

5

For an exposition of principal components analysis, see, for instance, Theil (1971) or Dhrymes (1970). For a relevant application in the present context, see Genberg (1977 a).

6

In fact, the factor loading of a variable is the regression coefficient of the variable on the appropriate principal component.

7

As a check, correlations and principal components were also calculated on levels of prices, output, money, and interest rates for the countries in the sample, except Japan. Briefly stated, the results are as follows. Price levels are highly correlated in both subperiods and the CPVE by the first principal component is 0.98 and 0.99, respectively. Similiar results obtain for money levels. For real output levels, correlations are higher in the first subperiod, Pi accounting for 0.97 of total variance, whereas, in the second subperiod, P1 accounts for 0.87, and P1 and P2 for 0.96, of total variance. For interest rates, correlations are again higher in the first subperiod; the first two principal components account for, respectively, 0.97 and 0.89 of total variance.

8

See, for instance, Genberg (1977 a).

9

We do not consider government expenditure in the text, as there are only four series. Suffice it to say here that principal components show the association to be substantially higher under flexible rates for the original series but not for innovations.

10

This is confirmed by an examination of the regression equations used to estimate the innovations in interest rates.

11

The difference probably arises from differences in the data (smoothed quarterly observations on quarterly long-term rates here and point-in-time monthly observations on short-term rates there) and in estimation procedure.

13

These movements in money supplies may also lead to transitory negative correlations (or lessened positive correlations) in national inflation rates.

14

The Mundellian special case of perfect capital mobility and fiscal expansion is a possible exception.

15

For empirical evidence supporting this statement and a survey of some relevant literature, see Genberg (1977 b).

16

For references, a survey, and some extensions of the standard literature on macroeconomic interdependence under floating rates see the chapter by Michael Mussa in Dornbusch and Frenkel (1979). See also the comment by A. Swoboda.

17

Mundell’s 1964 argument is reproduced in Mundell (1968), chapter 18.

18

For the full argument, see Dornbusch (1982), pp. 20-24.

19

Of course, the autonomy may not have been used and policies may have been so much more similar across countries after 1973 as to produce the greater association in output, inflation, and interest rates. The evidence on money reviewed in Section I, however, does not support this hypothesis.

20

We may also note in passing that the VAR estimates of Huber and Saidi (1982) do not support the overshooting hypothesis. They find that for France and the Federal Republic of Germany an unexpected increase in domestic money leads to an initial real exchange appreciation of the domestic currency in contradiction to the hypothesis.

21

For recent analyses and intermediate input shocks (or more generally supply shocks) see Bruno and Sachs (1981) and Jones and Purvis (1982).

22

The credibility issue is discussed briefly below.

1

The reader will remember that the innovations were obtained by regressing, for each sample period separately, quarterly observations of the variable on a constant, time trend, three seasonal dummies, and four lags of the variable. The residuals were assumed to be the innovations.

2

See Jacques R. Artus and John H. Young, “Fixed and Flexible Exchange Rates: A Renewal of the Debate,” Staff Papers, Vol. 26 (December 1979).

3

It should be observed that during 1979-82 European countries had to fight not only the imported inflation resulting from the appreciation of the U.S. dollar but also, and mainly, the imported inflation resulting from the increase in oil prices in 1979-80 and their own domestically generated inflation.

4

Jacques R. Artus, “Effects of U.S. Monetary Restraint on the DM-$ Exchange Rate and the German Economy,” Working Paper No. 926, National Bureau of Economic Research, Inc. (July 1982), scheduled to be published in Exchange Rate: Theory and Practice, eds. R. Marston and J. Bilson, (University of Chicago).