It is often taken for granted that an increase in risk will lead risk-averse individuals to reduce their efforts in the risky activity and to shift to less risky endeavors. This popular view has led many to conclude that exchange rate volatility, by increasing the risk of interna-tional trade activities, must in principle have a negative effect on trade. In this view exchange rate volatility leads economic agents to retrench into domestic activities. If one accepts this version of events, the only issue is an empirical one concerning the significance and magnitude of this effect.1

Abstract

It is often taken for granted that an increase in risk will lead risk-averse individuals to reduce their efforts in the risky activity and to shift to less risky endeavors. This popular view has led many to conclude that exchange rate volatility, by increasing the risk of interna-tional trade activities, must in principle have a negative effect on trade. In this view exchange rate volatility leads economic agents to retrench into domestic activities. If one accepts this version of events, the only issue is an empirical one concerning the significance and magnitude of this effect.1

It is often taken for granted that an increase in risk will lead risk-averse individuals to reduce their efforts in the risky activity and to shift to less risky endeavors. This popular view has led many to conclude that exchange rate volatility, by increasing the risk of interna-tional trade activities, must in principle have a negative effect on trade. In this view exchange rate volatility leads economic agents to retrench into domestic activities. If one accepts this version of events, the only issue is an empirical one concerning the significance and magnitude of this effect.1

I. The Theory

The modern theory of production and consumption under risk does not allow one to derive such clear-cut conclusions (see, for example, Newbery and Stiglitz (1981)), To show this ambiguity in the context of international trade decisions, a simple model is developed. The purpose of this theoretical exercise is not to analyze all aspects of risk in international trade, but rather to analyze the simplest possible problem of choice under risk. Even in this simple model, the theoretical effects of increased risk (from exchange rate volatility) on the supply of exports is ambiguous. In a more complicated model, this ambiguity will certainly persist.

Consider an individual producer who has the choice of producing for foreign and domestic markets (both perfectly competitive). He allocates a given amount of resources, x , to these two activities. The only source of risk is the price he obtains (in domestic currency) for the output he sells in the foreign market, and this risk is due to the uncertainty about the exchange rate.2 His total net revenue (profit) from these two activ-ities is defined as

Π¯=(p¯fqfwxf)+(pdqdwxd),(1)

where a tilde indicates that the variable is a random one; p¯f is the price of the output sold in the foreign market, measured in domestic currency and defined as p¯f=p*e¯,wheree¯ is the exchange rate (a random variable) and p* is the foreign currency price of the output sold in the foreign country; pd is the price of the output sold domestically; qf and qd are the quantities produced for the foreign and domestic markets, respectively; xf and xd are the amounts of resources (labor) used in producing for the foreign and the domestic markets, respectively (xf + xd = x, where x is assumed to be fixed);3 and w is the unit resource cost (wage rate), which is assumed to be the same in the two sectors (this assumption can easily be relaxed).

Output is defined by the production function q(x) so that

qf=q(xf)qd=q(xd).(2)

Note that the same technology is assumed lor the two sectors: this assumption can be relaxed without affecting the results.

Using equation (2) and the definitions xf+xd=xandp¯f=p*e¯, one can rewrite equation (1) as

Π¯=p*e¯q(xf)+pdq(xxf)wx.(3)

Because wx is a constant, it does not affect optimizing behavior. This term can therefore be dropped from the analysis, and the total revenue Y˜ can be defined as

Y˜=p*e˜q(xf)+pdq(xxf).(4)

The choice problem of the individual producer now amounts to select-ing an Xf that maximizes the expected utility of his total income Y˜; that

is,

maxEU(Y˜),

where U is a concave function of income Y˜ (that is, the producer is assumed to be risk averse). Using equation (4), one obtains

maxEU[p*e˜q(xf)+pdq(xxf)].(5)

To simplify the analysts, it is assumed that the utility function is sepa-rable in the two terms within the brackets. This amounts to assuming that the marginal utility of export revenue is independent of changes in domestic revenue:

max{EUf[p*e˜q(xf)]+Ud[pdq(xxf)]}.(6)

The first-order condition for an optimum then becomes

EUfp*e˜q(xf)=Udpdq(xxf),(7)

where Uf is the marginal utility of export revenue, Ud is the marginal utility of domestic revenue, and q’ is the marginal productivity of labor.

This optimum condition can also be rewritten as

EUfe¯=UdPdq(xxf)P*q(xf).(8)

The question of how an increase in the variability of e˜ affects the optimal amount of resources put into the export sector (xf) can now be addressed. To answer this question, one must find out how a “mean-preserving” spread in e˜ affects EUfe¯. If such an increase raises EUfe¯, then the right-hand side of equation (8) must also increase, which can only occur if xf increases. In other words, if an increase in the variability of the exchange rate increases the expected marginal utility of income from exports, then this increased variability will lead to more export activity. In contrast, if a higher spread of e˜ leads to a lower expected marginal utility of export income, then the producer will lower his export effort.

The question then boils down to whether the function Ufe¯ is convex or concave in e˜ If it is convex (concave), then every mean-preserving increase in the spread of e˜ will increase (decrease) the expected value of the function Ufe¯. This is made clear in Figure 1. To find out the conditions under which the function Ufe¯ is convex or concave, differentiate it twice with respect to e. This yields, after some manipulation,

d2Ufe¯de2=1e¯[R(1R)+RY˜f],(9)

where R=Uf"Yf/Uf is the coefficient of relative risk aversion, and Y˜f=P*e˜q is the export revenue.

If expression (9) is positive (negative), then the function Ufe¯ is convex (concave). It follows that convexity or concavity depends on the degree of risk aversion. Let us assume, first, that the coefficient of relative risk aversion (R) is constant (an assumption often made in the literature on risk). Then R’=0. One then has the simple result that convexity holds if R>1, and concavity holds if R<1.

Thus, if producers are sufficiently risk averse (R>1), an increase in exchange rate risk raises the expected marginal utility of export revenue and therefore induces them to increase their export activity. If, however, producers are not very risk averse (R<1), a higher exchange rate risk reduces the expected marginal utility of export revenues and therefore leads them to produce less for export.

Figure 1.
Figure 1.

Effects of an Increase in the Mean-Preserving Spread of e¯onEUfe¯

Citation: IMF Staff Papers 1988, 001; 10.5089/9781451956771.024.A003

This may seem a surprising result. The economic intuition underlying it is the following. Very risk-averse individuals worry about the worst possible outcome. As a result, when risk increases they will export more to avoid the possibility of a drastic decline in their revenues. Less risk-averse individuals are less concerned with extreme outcomes. They view the return on export activity now as less attractive given the increase in risk and decide to export less.

This result is more easily understood if one recognizes that an increase in risk has both a substitution and an income effect. The substitution effect is the one that comes to mind when an increase in risk is seen to lower the attractiveness of risky activities and to lead people to reduce these activities. There is, however, also an income effect that works in the opposite direction. When risk increases, the expected (total) utility of export revenue declines. This drop can be offset by increasing re-sources in the export sector. If the income effect dominates the substitution effect, higher exchange rate risk leads to greater export activity.

Another important distinction that aids understanding of these results is the effect that higher risk has on total and marginal utilities. The increased variability of the exchange rate increases the spread of export revenue around the mean. Because the utility function is concave, every mean-preserving spread of the export revenue lowers the expected total utility of these revenues. This effect is shown in Figure 2.

The same increase in risk, however, may increase the expected marginal utility of export revenue, thereby making it more attractive to increase exports. Thus, although exporters are universally made un-happy by the volatility of exchange rates (and may express their dis-pleasure by urging a change in the system), some may decide that they will be better off by exporting more.

The preceding results were derived under the assumption of constant relative risk aversion (R’ = 0). The results can be generalized by drop-ping this assumption. Higher exchange rate risk then leads to more exports if

R(1R)+RY˜f<0,

and to fewer exports if

R(1R)+RY˜f>0,

Qualitatively, the results are unchanged; that is, the effect on exports depends on the degree of risk aversion. If R’is negative (if risk aversion declines with the level of revenues), a lower degree of risk aversion is required for exports to be influenced positively. If R’ is positive, the degree of risk aversion that leads to more exports is increased.

Figure 2.
Figure 2.

Effect of a Mean-Preserving Spread of Export Revenue on Expected Utility

Citation: IMF Staff Papers 1988, 001; 10.5089/9781451956771.024.A003

It is important to stress here that the foregoing analysis has been derived from a utility function with very few restrictions. Only concavity and separability were assumed.4 In the literature on the effects of ex-change rate risk, models have been used in which very stringent restrictions were imposed on the utility function. In particular. Hooper and Kohlhagen (1978), who set the tone for much of the subsequent empirical research, used a utility function with constant absolute risk aversion. This stringency imposes on their model a very peculiar behavior of individuals toward risk. More precisely, it eliminates the income effect of changes in risk, so that an increase in risk always leads to a reduction of the risky activity (see Newbery and Stiglitz (1981, p. 88)). Thus, in their model the result that higher exchange rate variability leads to less exports follows immediately from the restriction imposed on the utility function.

II. The Political Economy of Exchange Rate Variability

The pure theory of behavior under risk does not allow derivation of the popular conclusion that an increase in exchange rate variability will in general lead to a decline in the supply of exports. One needs special assumptions to derive such a conclusion. There is another strand of the literature, however, that analyzes the effects of exchange rate variability on international trade and that appears to come to more clear-cut results with regard to this relationship. One can call the topic of this literature the “political economy of exchange rate variability.”

Although this literature is far removed from the level of formalization found in the pure risk theory, it is important to look at the problem from the perspective of political economy. The main ideas can be summarized as follows. Exchange rate changes that wander away from purchasing power parities (the so-called fundamentals) lead to adjustment problems and to “real” effects in the economy. This effect has also been called the “misalignment” problem (Williamson (1983)). These misalignments lead to a boom in the traded-goods sectors of countries whose currencies have become undervalued. In the countries whose currencies have be-come overvalued as a result of the swing in the real exchange rate, the traded-goods sectors are squeezed, leading to a loss of output and employment that is not easily absorbed in the short run by the other sectors in the economy.

The political economy part of this story is set in motion when, as a result of output and employment losses, individuals hurt by these developments organize themselves to pass protectionist legislation. As a result, markets become more protected, so that international trade is negatively affected.

This hypothesis only makes sense if there is some asymmetry in the protectionist tendencies—for example, when protectionist legislation passed when the currency tended to be overvalued is kept in place when the currency is in the undervaluation part of the cycle. If such asymmetries are present, then swings in the real exchange rates will lead to a trend like increase in protectionism and will negatively affect international trade. Thus, in general, this theory predicts that the volatility of real exchange rates over periods exceeding a few months or quarters are likely to lead to a reduction in the growth of international trade.

III. The Empirical Model

In this section a model is proposed that aims to explain the decline in the growth of international trade among the industrial countries since 1973, and to ascertain whether the increase in exchange rate variability contributed to this phenomenon.

To specify the empirical model, one may rely on what standard trade theory tells us about the determinants of international trade flows:

Xij=f(Yj,Pij,Tij,Sij,Sij,a).(10)

The growth of exports of country i to country j (X= ij) is a positive function of the growth rate of real income of country J(Yj) and a negative function of the rate of change of the price of country i’s goods relative to country j’s goods (Pij). In addition, the growth of trade between countries i and j will be influenced by the nature of the trade arrangements between the two countries (Tij). In particular, if countries i and j form a customs union, their trade flows should increase relative to the trade with other countries. This then also implies that, during the period of adjustment to the trade arrangement, the growth rates of trade among the members of the customs union will increased5

As argued in the previous section, trade theory is less clear about the sign of the effect of exchange rate variability (Sij) on trade flows. But if one takes the political economy theory seriously, one can expect that an increase in real exchange rate variability between currencies i and / will negatively affect trade flows between countries i and j. This theory then also predicts that the creeping protectionism caused by real exchange rate variability will reduce the growth rates of trade. Finally, other exogenous disturbances (a) may affect the growth rate of trade flows between countries i and j.

The following econometric equation was specified:

Xijt=btTijt+ctYjt+etSijt+gtPijt+at+uijt,(11)

i, j=1,..., n; t=1,3,

where Xijt is the average yearly growth rate of the export of country i to country j during period t; uijt is the error term. Two periods are distinguished: the fixed exchange rate period 1960-69, and the flexible exchange rate period 1973-84. The period 1970-72, which can best be considered as a transitional period between the two exchange rate regimes, is not considered. Note that, because Xijt is defined as an average growth rate over a ten-year and a twelve-year period, the model seeks to explain why the long-run growth rates of trade differ both between the two subperiods and between countries within each subperiod. The coefficients bt,ct, and et are allowed to be different in the two periods.

The variable Tijtrepresents the trade integration that has occurred among groups of countries. It is defined as a dummy variable in the following way. When a subgroup of countries (i = 1,...,k and k<n) forms a customs union during subperiod t, all the observations on the bilateral trade flows of these countries are given a value of unity (and of zero otherwise). Because there are several subgroups of countries that can be clustered together according to this criterion, there is more than one trade-integration variable.

The income variable is represented by Yjt, which is defined as the average yearly growth rate of gross domestic product (GDP) of country j(the importing country) during the subperiod f. Countries that have experienced a high growth rate of output during the subperiod t can also be expected to have experienced a high growth rate of their imports from the other countries in the sample. As with the other variables, the subperiods are 1960-69 (the fixed exchange rate period) and 1973-84 (the flexible exchange rate period).

Exchange rate variability is captured by the variable Sijt, which is defined as the variability of the yearly percentage changes of the bilateral exchange rate between currency i and currency j around the mean ob-served during subperiod t. This variable thus reflects long-run (yearly) movements of the bilateral exchange rates around the mean changes observed during, respectively, the fixed and the flexible exchange rate subperiods.6

Several alternative definitions of variability are considered here. First, the analysis distinguishes between nominal and real exchange rate variability. The former measures the variability of nominal exchange rates; the latter the variability of the nominal exchange rates corrected for inflation differentials (the real exchange rates). The theory developed in the previous section leads one to suspect that it is mainly through the variability of the real exchange rates that international trade flows are likely to be affected. Second, as a statistical measure of variability, the analysis uses the standard deviation of the yearly growth rates of the exchange rates around the mean (real and nominal). More detail about the precise definition of these variables is presented in the Appendix.

The relative price variable, Pijt, is defined as the average yearly change in the real exchange rate between currency i and currency j during period t. Thus, when during, say, 1973-84 the currency of country i has depreciated on average in real terms against currency j, Pijt is a positive number. In that case one can also expect that the export of country j to country j will have been stimulated, on average. Thus, one expects a positive coefficient for Pijt. (Note that the real exchange rate between currencies i and j is defined as the nominal exchange rate between currency i and j times the ratio of the wholesale price level of country; and country j.)

The final variable to be explained is at (t = 1,2): a1 is a constant term for the observations of the first period; a2 is a constant term for observations of the second period. Thus, at measures the extent to which the average growth rates of trade were different during the two subperiods, where these differences are unrelated to the other explanatory variables in the model. This variable can therefore be interpreted as measuring the effects of shocks (for example, the oil shock) that occurred since 1973 and that affected the trade flows of all the industrial countries in the sample in the same way.7

IV. Empirical Results

The model was estimated using bilateral trade flows among the ten major industrial countries (in terms of the size of their external trade): Belgium, Canada, France, the Federal Republic of Germany, Italy, Japan, the Netherlands, Switzerland, the United Kingdom, and the United States. Together, the external trade of these countries amounts to 87 percent of the international trade of all industrial countries and to 60 percent of world trade.

With ten countries in the sample, 90 bilateral trade flows were obtained during the fixed and the flexible exchange rate periods, for a total of 180 observations. The results of estimating equation (11) (with a SURE [seemingly unrelated regression estimation] procedure) are presented in Table 1. Note that the way in which the model is specified implies that the income, exchange rate, and relative price variables are not restricted to have identical coefficients during the fixed and the flexible exchange rate periods. Table 2 presents the results when such restrictions were imposed each time the difference between the coefficients of the two periods was not statistically significant.

Several trade-integration variables are distinguished. One. called EC-60, groups all the trade flows between the original members of the European Communities (EC) in the first subperiod. The second one.

Table 1.

Estimation Results for Equation (11), 1960-69 and 1973-84

article image
Note: Explanatory variables are defined in the text.

Numbers in parentheses are t-statistics.

EC-70, has the same grouping during the floating-rate period. The third one, EC-New, brings together the trade flows between the United Kingdom and the original EC countries during the second subperiod. This variable captures the effect of the admission of the United Kingdom to the EC. Finally, dummy variables have been added for the exports of Japan during the two subperiods (JAP-60 and JAP-70). These variables capture the exceptional phenomenon of Japanese trade creation during the two subperiods, which cannot be explained by income growth or by exchange rate variability.

The results in Table 1 can now be interpreted. First, the constant term for the second period tends to be somewhat larger than the one for the first period, but the difference is not statistically significant. Thus, the shocks (other than those specified in the model) that occurred after 1973 and that affected all industrial countries do not seem to have significantly affected trade flows among the industrial countries.

Table 2.

Estimation Results for Restricted Version of Equation (II), 1960-69 and 1973-84

article image
Note: Explanatory variables are defined in the text.

Numbers in parentheses are t-statistics.

Second, the trade-integration variables have a significant influence in the explanation of the bilateral trade flows. For example, during the fixed-rate period the existence of a customs union among the EC countries increased the yearly growth rate of the members’ trade (relative to the other countries in the sample) by approximately 6 percent (see the row for EC-60 in Table 1). During the flexible exchange rate period the opposite phenomenon is observed: the same group of countries now experience a small negative effect from their trade association (row EC-70 in Table 1). This negative trade-integration effect could be due to trade diversion after the admission of new members to the EC. During the same period, a positive and significant integration effect can be observed on the U.K.-EC trade of approximately 5 percent a year. This reflects the positive effects of the admission of the United Kingdom to the EC.

Third, 1 percent higher growth of GDP leads in the long run to approximately 1 percent higher growth of trade. This income elasticity of bilateral trade flows is not significantly different between the two subperiods.8 These results may seem surprising to those accustomed to time-series studies of the income elasticities of international trade flows. In these studies one usually finds long-run income elasticities that exceed unity, which implies that in the long run people tend to spend their whole incomes on imports. These results are typically obtained for periods when countries experienced strong trade integration. To the extent that one does not isolate the trade-integration effects (typically the case in time-series studies), these effects will show up in the income elasticity of the trade flows.9 As a result, one can say that time-series studies will tend to overestimate the long-run income elasticity of trade flows. In this study, where the trade-integration effects have been isolated, the hypothesis that the long-run income elasticity of trade is equal to unity cannot be rejected.

Fourth, the variability of the exchange rates as measured by the variable S has a negative effect on the growth rate of trade during the second (flexible exchange rate) period. As expected, this negative effect is significant only when real exchange rate variability is used. During the fixed exchange rate period, significant effects of exchange rate variability on trade are not evident. This is not surprising because exchange rate variability (nominal and real) tended to be small during this period.

Finally, the relative price variable, P, does not perform well as an explanatory variable during the first (fixed exchange rate) period. During the flexible exchange rate period, however, P has the expected sign and is statistically different from zero. This different performance of the relative price variable during the two subperiods is probably due to the fact that in the first period protracted real appreciations or depreciations of currencies were relatively rare, whereas such currency movements tended to occur more frequently during the flexible exchange rate period, thereby having significant effects on trade flows.

As mentioned above, Table 2 presents the results under the restriction that the coefficients for the fixed and flexible exchange rate periods be equal each time the difference between the coefficients of the two periods is not statistically significant. In addition, since the coefficient of the relative price variable P was not statistically different from zero during the fixed exchange rate period, we constrained the coefficient of P to be zero during that period. The results were very little affected by these restrictions. In particular, exchange rate variability was found to have significant negative effects on the growth rate of international trade. Nominal variability now becomes significant. The size of the coefficient for nominal variability, however, is about half as large as the coefficient for real variability.

One may thus conclude that the large variability of exchange rates (especially of real exchange rates) observed during the period of flexible rates had a significant negative effect on the growth rates of bilateral trade flows among the major industrial countries.

In contrast to many time-series studies, this study has been able to find such negative effects of exchange rate variability for two reasons. First, by using cross-section evidence, the effects of trade integration can be isolated among groups of countries. This is quite important, as the evidence on trade among the EC countries testifies. The original members of the EC experienced a substantial deceleration of their bilateral trade despite low exchange rate variability. That is, during the periods considered here, there has been a correlation between stable exchange rates and a slowdown in the integration process. Put differently, the countries that experienced relatively stable exchange rates were also those that experienced a strong decline in their trade-integration process since the 1970s. Thus, studies that do not take these trade effects into account will bias the effects of exchange rate variability. In particular, they will tend to underestimate the effects of exchange rate variability on international trade.

Second, the present analysis has concentrated on exchange rate variability of one year or more. This long-run exchange rate variability (misalignment) is a likely source of protectionist pressure and may ex-plain why international trade is negatively affected.10

The regression results contained in Tables 1 and 2 allow one to quantify the contribution of the independent variables to the explanation of the deceleration of international trade since 1973. Because the constant term and the coefficients of the income and exchange rate variables are not significantly different between the two periods, the regression results of Table 2 will be used for this purpose (these coefficients, therefore, are constrained to be equal in the two subperiods).

The procedure adopted is as follows. The coefficients of the independent variables are multiplied by the observed mean value of that variable in the first period and the second period, respectively. The difference between these two numbers measures the contribution of the change in that variable toward explaining the decline of the growth rate of trade.

This procedure is applied to evaluate the contribution of the decline of the growth rate of output and of the increase in exchange rate variability from 1960-69 to 1973-84. The contribution of the trade-integration variables is obtained by multiplying the mean value of these variables by the respective coefficients. This procedure yields average trade-integration effects in the fixed and flexible exchange rate periods. By subtracting the latter from the former, one obtains a number that tells by how much international trade flows have changed because of the changes that occurred in trade-integration patterns.

The results for the ten major industrial countries of the sample are presented in the first panel of Table 3. Estimates are presented for both definitions of exchange rate variability. The major part (about half) of the total decline in the growth rate of international trade among the major industrial countries (5.7 percent) is attributable to the deceleration of the growth rate of output. About 30 percent of the decline can be explained by trade-integration effects—that is, by the substantial slowdown of trade integration among the original EC countries during the second period (which was only partially offset by trade creation between these EC countries and the United Kingdom) and by the slow-down of the trade penetration of Japan in the industrial world. Finally, the increase in exchange rate variability observed since 1973 accounts for approximately 20 percent of the total decline in the growth rate of international trade.

Table 3.

Explanation of the Decline in the Growth Rate of International Trade Among Ten Major Industrial Countries from 1960-69 to 1973-84

(In percent)

article image
Source: Table 2.

One thus may conclude that, although the large variability of (real) exchange rates since 1973 is less important than the other two factors (the GDP growth rate and trade integration) as an explanation of the deceleration of international trade in the industrial world, it has contributed significantly to this phenomenon.

It may be surprising that the relative price variable does not appear in Table 3 as a contributing variable in the explanation of the slowdown in international trade since 1973. The reason is quite simple. The positive effect of a real depreciation of, say, currency i on country i’s exports to the other countries is offset by a negative effect of the other countries’ exports to country i. As a result, the relative price changes do not affect the average growth rate of the bilateral trade flows.

The results in the first panel of Table 3 explain the slowdown in international trade for the group of industrial countries as a whole. It is also useful to disaggregate these results, as in the remaining panels of Table 3. As stressed earlier, the decline of the bilateral trade flows among the original EC countries was substantially larger than that of other bilateral trade flows. Therefore, the sample of trade flows was divided in two parts—intra-EC flows (that is, among members of the original EC) and flows among all the others—-and the same computations as for the first panel were made. The results lead to the following interpretation. First, the trade-integration variables explain the largest part (two thirds) of the spectacular decline of the yearly growth rate of intra-EC trade flows, whereas the variability of exchange rates does not contribute much in the explanation of this decline. Second, the trade flows outside the original EC present a completely different picture. The substantially smaller decline of the growth rates of these trade flows is explained to a larger degree by the increase in exchange rate variability. Close to one third of the decline in the growth rate of trade can be accounted for by the exchange rate variable. Conversely, a smaller part of this decline is due to trade-integration effects. Thus, one can say that the slowdown of the growth rate of international trade flows outside the EC has been significantly affected by the increase in the degree of exchange rate variability.

V. Additional Empirical Evidence

In the preceding section the determinants of the long-run growth rates of international trade among the industrial countries were analyzed. The “long run” related to observations over a ten-year and a twelve-year period. It is useful to analyze shorter time horizons, especially when one wants to study the importance of relative price changes. Some important real exchange rate changes have followed a long cycle of about ten years. The dollar, for example, first depreciated during the 1973-79 period and appreciated during the 1980-84 period. By aggregating time observations over a twelve-year period, as in the previous section, much of this cycle in relative price changes is lost. Therefore, in this section the flexible exchange rate period is disaggregated into two subperiods, 1973-78 and 1979-84.

The model used is the same one specified in equation (11). The estimation results for the two subperiods of the flexible exchange rate period are presented in Table 4. When the standard deviations of the growth rates of the exchange rates are computed, one now has only six yearly observations. To increase the number of observations, the model was also estimated using quarterly observations of yearly growth rates.11 These results are shown in the last two columns of Table 4. The first two columns present the results using annual observations of yearly growth rates.

There are several noteworthy results in Table 4, First, the constant term declines significantly from the first to the second period, indicating a decline in the growth rate of trade since 1979 that cannot be explained by the other independent variables in the model. This exogenous decline could be due to supply shocks or to protectionist pressure that negatively affected trade flows. Second, some shifts in the trade-integration pattern seem to have occurred after 1979. The U.K.-EC process of trade creation slows down (although this variable remains statistically significant after 1979, indicating that the admission of the United Kingdom to the EC in the early 1970s was still a source of trade creation in the 1980s). Conversely, one observes a significant acceleration of the export drive by Japan after 1979.12

Third, the significance of the relative price variable has increased when compared with results in Tables 1 and 2. The size of the coefficients, however, is affected little by measuring changes over shorter (six-year) periods. Finally, as can be seen from the coefficients of S-73 and S-79, the variability of the exchange rates has, in general, a significant negative effect on the growth of international trade. (The exceptions are the coefficients of variability in the second period, S-79, when yearly observations of the growth rates are used.)

Table 4.

Estimation Results for Equation (11), 1973-78 and 1979-84

article image
Note: Explanatory variables are defined in the text.

Numbers in parentheses are t-statistics.

Summing up, one can state that the results in Table 4 do not change the conclusions of the previous section in a fundamental sense. The hypothesis that exchange rate volatility has negatively affected the growth of international trade in the industrial world during the period 1973-84 cannot be rejected.

VI. Conclusions

Growth rates of international trade among industrial countries have experienced a substantial decline since the early 1970s. This decline coincided with the start of the flexible exchange rate system. This paper has analyzed the question of the extent to which the increased variability of exchange rates can account for this phenomenon.

Contrary to popular thinking on the subject, the modern theory of production and consumption under risk does not allow one to conclude that an increase in exchange rate uncertainty has a negative effect on trade. In the context of a simple model of the supply of exports under risk, the analysis here has shown that, although an increase in exchange rate risk reduces welfare, it can also induce the exporter to increase his export activities. Whether he does so then depends on his degree of risk aversion.

In view of the ambiguity of risk effects (induced by exchange rate volatility) on export supply, it was then argued that if exchange rate variability has a negative effect on international trade, it must be through another mechanism. This mechanism was called the political economy effect of exchange rate variability. The flexible exchange rate regime has led to long swings in the real exchange rates of major currencies, “‘mis-alignments” that have led to adjustment problems in the tradable goods sectors and to political pressures for protectionism. This hypothesis then led, in the empirical part of the paper, to concentration on the effects of long-run movements of real exchange rates.

Cross-section evidence pooled with observations of two periods (1960-69 and 1973-84) showed that the long-run variability of real ex-change rates contributed in a significant way to the slowdown in the growth of international trade. By this analysis, close to 20 percent of the observed decline in the growth rate of international trade among the industrial countries can be attributed to the substantial increase in the long-run variability of real exchange rates.

Although significant, the increase in real exchange rate variability is not the most important factor in the explanation of the slowdown in international trade since 1973. The other factors that have had a more pronounced negative effect on trade are the decline in the growth rate of output (accounting for approximately 50 percent of the decline in the growth rate of trade) and the slowdown of the trade-integration process in the industrial world since the early 1970s (responsible for approximately 30 percent of the decline in the growth rate of trade). An issue that remains unresolved is whether the decline in the growth rate of out-put can be considered as an independent variable. Further work in this area should concentrate on the question of whether the slowdown in the growth rate of output could have been influenced by the decline in the growth of international trade.

APPENDIX Data Sources and Definitions

Bilateral exports were obtained from the OECD, Statistics of Foreign Trade, Overall Trade by Countries, Series A (Paris, various years). These are U.S. dollar figures. To compute the real growth rate of the export from country i to country /, the export figures were first converted into domestic currency by using the U.S. dollar exchange rate from the International Monetary Fund’s International Financial Statistics (IFS) (Washington, various years). This figure was then deflated by using the export price index (or export unit value) from the IFS, line 74.

Real bilateral exchange rates (Rijt) were computed as follows:

Rijt = EijtPjt/P/it,

where Eijt is the index of the market exchange rate of currency j in units of currency i (obtained from IFS by using the dollar exchange rates of currency i and j); and Pjt and Pit are the wholesale price indices of country j and i respectively (from IFS, line 63).

References

  • Akhtar, M.A., and R. Spence Hilton, “Exchange Rate Uncertainty and International Trade: Some Conceptual Issues and New Estimates for Germany and the United States,” Research Paper 8403 (New York: Federal Reserve Bank of New York, May 1984).

    • Search Google Scholar
    • Export Citation
  • Bailey, M. J., George S. Tavlas, and Michael Ulan, “Exchange Rate Variability and Trade Performance: Evidence for the Big Seven Industrial Countries,” Weltwirtschaftliches Archiv(Kiel) Vol. 122 (No. 3, 1986), pp. 466-77.

    • Search Google Scholar
    • Export Citation
  • Brodsky, David A., “Fixed Versus Flexible Exchange Rates and the Measurement of Exchange Rate Instability,” Journal of International Economics (Amsterdam), Vol. 16 (May 1984), pp. 295-310.

    • Search Google Scholar
    • Export Citation
  • Clifton, Eric V., “Real Exchange Rates, Import Penetration, and Protectionism in Industrial Countries,” Staff Papers, International Monetary Fund (Washington), Vol. 32 (September 1985), pp. 513-36.

    • Search Google Scholar
    • Export Citation
  • Coes, Donald, “The Crawling Peg and Exchange Rate Uncertainty,” in Exchange Rate Rules: The Theory, Performance, and Prospects of the Crawling Peg, ed. by John Williamson (New York: St. Martin’s, 1981), pp. 113-36.

    • Search Google Scholar
    • Export Citation
  • Cushman, David O., “The Effects of Real Exchange Rate Risk on International Trade,” Journal of International Economics (Amsterdam), Vol. 15 (August 1983), pp. 44-63.

    • Search Google Scholar
    • Export Citation
  • Frenkel, Jacob A., “Flexible Exchange Rates, Prices, and the Role of ‘News’: Lessons from the 1970s,” Journal of Political Economy (Chicago), Vol. 89 (August 1981), pp. 665-705.

    • Search Google Scholar
    • Export Citation
  • Gotur, Padma, “Effects of Exchange Rate Volatility on Trade: Some Further Evidence,” Staff Papers, International Monetary Fund (Washington), Vol. 32, No. 3 (September 1985), pp. 475-512.

    • Search Google Scholar
    • Export Citation
  • Grauer F.L.A., and R.H. Litzenberger, “The Pricing of Commodity Futures Contracts, Nominal Bonds and Other Risky Assets Under Commodity Price Uncertainty,” Journal of Finance (New York), Vol. 34 (March 1979), pp. 69-83.

    • Search Google Scholar
    • Export Citation
  • Hooper, Peter, and Steven W. Kohlhagen, “The Effect of Exchange Rate Uncertainty on the Prices and Volume of International Trade,” Journal of International Economics (Amsterdam), Vol. 8 (November 1978), pp. 483-511.

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, Exchange Rate Volatility and World Trade, Occasional Paper 28 (Washington, July 1984).

  • Kenen, Peter B., “Exchange Rate Instability: Measurement and Implications,” International Finance Section Research Memorandum (Princeton, New Jersey: Department of Economics, Princeton University, June 1979).

    • Search Google Scholar
    • Export Citation
  • Kenen, Peter B., and Dani Rodrik, “Measuring and Analyzing the Effects of Short-Term Volatility in Real Exchange Rates,” Review of Economics and Statistics (Cambridge, Massachusetts), Vol. 68 (May 1986), pp. 311-15.

    • Search Google Scholar
    • Export Citation
  • Krugman, Paul, “Is the Japan Problem Over?,” NBER Working Paper 1962 (Cambridge, Massachusetts: National Bureau of Economic Research, June 1986).

    • Search Google Scholar
    • Export Citation
  • Lanyi, Anthony, and Esther Suss, “Exchange Rate Variability: Alternative Measures and Interpretation,” Staff Papers, International Monetary Fund (Washington), Vol. 29 (December 1982), pp. 527-60.

    • Search Google Scholar
    • Export Citation
  • Newbery, David M.G., and Joseph E. Stiglitz, The Theory of Commodity Price Stabilization. A Study in the Economics of Risk (Oxford, England: Clarendon Press, 1981).

    • Search Google Scholar
    • Export Citation
  • Thursby, Jerry G., and Marie C. Thursby, “The Uncertainty Effects of Floating Exchange Rates: Empirical Evidence on International Trade Flows,” in Exchange Rates, Trade, and the U.S. Economy, ed. by Sven W. Arndt, Richard J. Sweeney, and Thomas D. Willett (Cambridge, Massachusetts: Ballinger, 1985), pp. 153-65.

    • Search Google Scholar
    • Export Citation
  • Williamson, John, The Exchange Rate System (Washington: Institute for International Economics, 1983).

*

Mr. De Grauwe is Professor of Economics at the Catholic University of Leuven, Belgium, and a Research Fellow at the Centre for European Policy Studies, Brussels. He is a graduate of the Catholic University of Leuven and The Johns Hopkins University. This paper was written while the author was a visiting scholar in the Research Department of the Fund.

1

This is also the view taken in the paper by Hooper and Kohlhagen (1978), which has been influential for further empirical studies. The latter include Akhtar and Hilton (1984), Coes (1981), Cushman (1983), Gotur (1985), Kenen and Rodrik (1986), and Thursby and Thursby (1985). For a survey of empirical studies, see International Monetary Fund (1984).

2

There is no capital market in this model. Therefore, it is also implicitly assumed that the exporter cannot diversify his exchange risk. Introducing a capital market in the model would enrich the analysis, but it would not change the basic conclusion that risk has an ambiguous effect on exports. See Grauer and Litzenberger (1979).

3

Note that this implies that the total productive effort is not affected by the change in risk. The focus of the analysis here is only on how the existing re-sources are allocated between the two sectors. For an analysis of the effects of risk on the productive effort, see Newbery and Stiglitz (1981, pp. 80-83).

4

If the separability assumption is dropped, one obtains a more complicated first-order condition than equation (8), and the simple relation between convexity or concavity of the function EUfe¯ and the degree of risk aversion R will no longer hold. The ambiguity of the effect of risk on export activity, however, remains a basic feature.

5

In the long run, when all adjustments have taken place, there is no reason for the growth rates of trade between countries forming a customs union to exceed the growth rates of their trade with other countries.

6

There are many methodological problems with measuring exchange rate variability. See Lanyi and Suss (1982), Kenen (1974), and Brodsky (1984).

7

For an analysis of the effects of oil price increases on trade flows, see Bailev. Tavlas, and Ulan (1986).

8

Regressions were also performed in which the income elasticity was different between EC countries ana the other countries in the sample. In general, no significant differences were found.

9

Technically, the trade-integration and income variables are typically correlated. Omitting the former biases the coefficient of the income variable upward.

10

A recent article by Clifton (1985) confirms the existence of a link between real exchange rate movements and protectionism.

11

It should be stressed that this procedure does not introduce serial correlation in the error terms, as it would in a time-series analysis. In the cross-section analysis performed here, the average yearly growth rates of exports, incomes, and exchange rates were computed first, on the basis of quarterly observations of these growth rates, during 1973-78 and 1979-84 respectively. Then the standard deviation was computed for the quarterly observations of the yearly changes in the exchange rates around the sample means of the two subperiods. Thus, one number was obtained for each bilateral trade flow during 1973-78 and 1979-84, respectively. These numbers were then used in the cross-section regression analysis.

12

See a recent paper by Krugman (1986), who observed the same phenomenon. Krugman interpreted this exceptional Japanese export drive as being caused by the strong deterioration of the Japanese terms of trade.

IMF Staff papers: Volume 35 No. 1
Author: International Monetary Fund. Research Dept.
  • View in gallery

    Effects of an Increase in the Mean-Preserving Spread of e¯onEUfe¯

  • View in gallery

    Effect of a Mean-Preserving Spread of Export Revenue on Expected Utility