Perhaps the element that is most commonly taken into account in attempting to assess whether a country’s exchange rate is at an appropriate level is its price competitiveness. Such an assessment normally takes the form of a straightforward comparison of past movements in prices at home and abroad, after adjustment for exchange rate changes. If the comparison indicates that the country considered has experienced significant losses or gains in its price competitiveness in relation to a base period when its position was considered adequate, this finding is viewed as establishing a presumption that the current exchange rate is unsustainable.8 Sustainability can, of course, be restored through a change in the nominal exchange rate, a change in the level of domestic currency prices relative to those abroad, or some combination of the two. In many cases, a comparison of movements in production costs is made as a supplement, or an alternative, to a comparison of movements in prices. Indicators of relative prices or costs adjusted for exchange rate changes are here referred to as “real exchange rates.”
The Fund’s regular Article IV consultation reports on member countries normally include a discussion of the relevant country’s international competitiveness, and this discussion is one of the bases for the section of the staff appraisal that deals with the exchange rate. In addition, following the review of surveillance procedures in March 1983, the Fund decided to initiate, on an experimental basis, a system whereby the Executive Board would be notified regularly of all large changes in real effective exchange rates.
Outside the Fund, national authorities in virtually all industrial countries also pay close attention to various indicators of their cost and price competitiveness. For example, the recent report on intervention states that:
While a comparison of inflation rates—even in the longer run—requires considerable judgment regarding the choice of appropriate price indices and the base period, members agreed that it was a useful concept for assessing the exchange rate.9
In general, conclusions obtained from the application of this method are only considered as tentative. They are useful to sound an alert but, because such indicators can be quite distorted, they should not per se create alarm. Given the dearth of other suitable indicators, however, there are times when considerable practical importance is attached to the results of analyses of international competitiveness.
This section outlines the major assumptions that underlie the use of indicators of international competitiveness for the assessment of exchange rates and then describes the main statistical measures of competitiveness that are employed by the staff. The reliability of the preferred indicator—that based on relative normalized unit labor costs in manufacturing—is then considered in detail. After a comment on competitiveness indicators based on consumer prices, the main conclusions of the section are briefly summarized.
Basic Assumptions of the Competitiveness Analysis
The use of indicators of international price competitiveness to assess the exchange rate is an application of the relative purchasing power parity (PPP) theory of exchange rates. Without entering into the intricacies of the various versions of this theory, it is important to note that it deals with currency relationships where: (1) countries produce similar goods and there are no major impediments to free trade, (2) adjustment lags in the goods markets are considerable because of information lags and other market inefficiencies, and (3) countries’ overall price levels in local currency tend to be determined primarily by domestic factors.
The first assumption is necessary in order for exchange rates to converge toward their PPP level in the longer run. If countries produced completely different goods, or if there were major impediments to free trade, there would be little or no tendency—even in the longer run—for countries’ price levels to converge when converted into the same numeraire currency. The second assumption is required for the view that PPP holds in the long run to be consistent with short-run divergences between national price levels; if goods market prices adjusted immediately there would be no need for the analysis of trends in competitiveness, since nominal exchange rates would continuously move to offset inflation rate differentials. The third assumption is necessary for the analysis to be relevant to the assessment of the exchange rate, rather than relative overall price levels in local currencies.
The empirical relevance of the second assumption concerning the existence of adjustment lags is widely accepted, since there is ample evidence that buyers in the international goods market do not immediately change their sources of supply in response to month-to-month (or even year-to-year) changes in relative prices. More basic issues arise in judging the empirical relevance of the first and third assumptions. On the whole, even though there is room for concern about restrictions that impede free trade in goods and services, industrial countries tend to maintain a fairly liberal trade system with relatively low tariffs, at least as far as nonagricultural products are concerned.
There is much evidence, however, that the assumption that countries produce similar goods—that is, goods that are close substitutes in demand—does not fit with conditions in the real world. In the first place, the degree of specialization in various categories of tradable goods varies significantly across countries. For example, Australia and Canada export relatively more raw and semi-processed materials than most other industrial countries.
More generally, specialization along the lines of comparative advantage is the basic characteristic of free international trade. To illustrate this point, the structure of production of the manufacturing sector—the dominant segment of the tradable-goods producing sector in industrial countries—is presented in Table 1 for the three largest industrial countries and for three smaller countries. The table indicates significant cross-country differences in the proportion of total value added that is contributed by the various industries. It is also relevant to note that evidence at the industry level, such as that given in Table 1, does not reflect the full extent of structural differences because much specialization occurs at the level of individual products, rather than across broad industrial categories. Furthermore, even in cases where the degree of specialization at the individual product level is similar in different countries, the specific types of commodities produced may still be greatly differentiated on account of differences in quality, delivery dates, after-sales service, and many other characteristics.
Selected Countries: Shares of Value Added in Manufacturing1
Because the level of disaggregation of the input-output tables varies across countries, not all categories of manufactures were available on a consistent basis for all countries. For some countries, the disaggregation of metals into ferrous and nonferrous, of machinery into electrical, nonelectrical, and precision instruments, and of transport equipment into road vehicles, aircraft, and ships was estimated by the staff.
Selected Countries: Shares of Value Added in Manufacturing1
| United States | Federal Republic of Germany | Japan | Italy | Belgium | Sweden | |
|---|---|---|---|---|---|---|
| Food, beverages, and tobacco | 13.1 | 9.5 | 10.5 | 9.3 | 15.4 | 10.6 |
| Textiles, weaving, apparel, and leather | 7.4 | 6.7 | 6.7 | 15.1 | 11.4 | 7.0 |
| Wood and wood products | 12.7 | 10.5 | 10.1 | 11.0 | 9.6 | 22.2 |
| Chemicals and rubber products | 11.1 | 13.0 | 9.7 | 11.6 | 9.8 | 7.8 |
| Petroleum refining | 2.5 | 0.5 | 4.7 | 1.2 | 6.9 | 0.8 |
| Nonmetallic minerals | 3.7 | 5.0 | 4.2 | 6.7 | 7.1 | 4.8 |
| Iron and steel | 5.3 | 4.5 | 3.9 | 4.9 | 9.3 | 5.2 |
| Nonferrous metals | 2.1 | 0.8 | 1.6 | 1.1 | 2.6 | 1.0 |
| Metal products | 6.6 | 11.6 | 6.4 | 7.2 | 7.6 | 3.6 |
| Nonelectrical machinery | 8.6 | 12.3 | 11.7 | 8.3 | 6.6 | 17.8 |
| Electrical machinery | 11.1 | 11.2 | 10.5 | 6.6 | 5.4 | 6.8 |
| Precision instruments | 2.5 | 3.4 | 1.8 | 2.4 | 1.6 | 1.2 |
| Road vehicles | 7.4 | 5.7 | 6.0 | 4.8 | 3.9 | 7.2 |
| Aircraft | 2.9 | 0.3 | — | 1.2 | 0.9 | 0.3 |
| Ships | 1.6 | 1.2 | 4.1 | 1.4 | 0.9 | 2.9 |
| Miscellaneous | 1.6 | 3.8 | 3.1 | 1.2 | 0.9 | 0.6 |
Because the level of disaggregation of the input-output tables varies across countries, not all categories of manufactures were available on a consistent basis for all countries. For some countries, the disaggregation of metals into ferrous and nonferrous, of machinery into electrical, nonelectrical, and precision instruments, and of transport equipment into road vehicles, aircraft, and ships was estimated by the staff.
Selected Countries: Shares of Value Added in Manufacturing1
| United States | Federal Republic of Germany | Japan | Italy | Belgium | Sweden | |
|---|---|---|---|---|---|---|
| Food, beverages, and tobacco | 13.1 | 9.5 | 10.5 | 9.3 | 15.4 | 10.6 |
| Textiles, weaving, apparel, and leather | 7.4 | 6.7 | 6.7 | 15.1 | 11.4 | 7.0 |
| Wood and wood products | 12.7 | 10.5 | 10.1 | 11.0 | 9.6 | 22.2 |
| Chemicals and rubber products | 11.1 | 13.0 | 9.7 | 11.6 | 9.8 | 7.8 |
| Petroleum refining | 2.5 | 0.5 | 4.7 | 1.2 | 6.9 | 0.8 |
| Nonmetallic minerals | 3.7 | 5.0 | 4.2 | 6.7 | 7.1 | 4.8 |
| Iron and steel | 5.3 | 4.5 | 3.9 | 4.9 | 9.3 | 5.2 |
| Nonferrous metals | 2.1 | 0.8 | 1.6 | 1.1 | 2.6 | 1.0 |
| Metal products | 6.6 | 11.6 | 6.4 | 7.2 | 7.6 | 3.6 |
| Nonelectrical machinery | 8.6 | 12.3 | 11.7 | 8.3 | 6.6 | 17.8 |
| Electrical machinery | 11.1 | 11.2 | 10.5 | 6.6 | 5.4 | 6.8 |
| Precision instruments | 2.5 | 3.4 | 1.8 | 2.4 | 1.6 | 1.2 |
| Road vehicles | 7.4 | 5.7 | 6.0 | 4.8 | 3.9 | 7.2 |
| Aircraft | 2.9 | 0.3 | — | 1.2 | 0.9 | 0.3 |
| Ships | 1.6 | 1.2 | 4.1 | 1.4 | 0.9 | 2.9 |
| Miscellaneous | 1.6 | 3.8 | 3.1 | 1.2 | 0.9 | 0.6 |
Because the level of disaggregation of the input-output tables varies across countries, not all categories of manufactures were available on a consistent basis for all countries. For some countries, the disaggregation of metals into ferrous and nonferrous, of machinery into electrical, nonelectrical, and precision instruments, and of transport equipment into road vehicles, aircraft, and ships was estimated by the staff.
A second reason for differences in the composition of output across countries is that there are many goods and services whose use is country specific. For example, there is obviously little arbitrage between prices of newly produced houses in the United States and in France. Many types of services, including both government services and those provided by the private sector, are also country specific. It is possible to exclude most of these “nontradable” goods from comparisons of international price levels, but this only reduces the problem. For example, if a country’s tradable goods prices remain in line with those of similar goods in the world market, it will not experience a decline in domestic and foreign demands for these goods. However, if the ratio of nontradable goods prices to tradable goods prices increases, it may still experience a shift in domestic production from the tradable sector to the nontradable sector. Thus, an analysis that focuses on the relative prices of tradable goods among countries—without taking into account relative prices between tradable and nontradable goods within each country—may at times be misleading because it ignores a major determinant of the sustainable exchange rate.
These observations are not intended to detract from the role played by relative prices in a country’s foreign trade performance. Nevertheless, they do suggest that the PPP theory has to be qualified in two respects. First, no matter how much care is taken in the choice of the price indices, the resulting indicator of competitiveness may still fail to measure the true level of competitiveness because of composition effects; that is, because the weights given to the various items in the price indices differ significantly across countries. This problem may be quite significant because rates of growth of productivity and, therefore, rates of change of prices, differ substantially among sectors. For example, a country may appear to be gaining competitiveness on the basis of aggregate price indices when in fact the relative change in the indices may be caused by the country’s heavy specialization in an industry that has a relatively high rate of productivity growth and a declining price trend when compared with other industries. Thus, whenever possible, a study of competitiveness should include an examination of whether the results are likely to be influenced by composition effects.
Second, the PPP theory also has to be qualified to take into account the fact that product differentiation among countries significantly limits the elasticities of substitution among the goods that they supply, even in the longer run. This has the major implication that any sustained changes in a country’s endowment of economic resources, the pattern of demand for its products, or the determinants of its net capital flows may have to be offset by a change in price competitiveness if the country’s overall payments balance is to remain in equilibrium in the longer run. Thus, any assessment of an exchange rate requires both a consideration of changes in competitiveness that have actually taken place and an analysis of the change that would be called for to restore the overall balance of payments to a sustainable level.10
There are also cases, particularly among the smaller industrial countries, where the validity of the third assumption of the PPP approach—namely, that the overall price level in local currency tends to be determined mainly by domestic factors—is less tenable. As long as rigidities in the goods and labor markets are not too extreme, overall price levels will normally remain consistent with domestic monetary developments in the longer run. In this sense, domestic monetary policy can be viewed as the major determinant of a country’s overall price level. Nevertheless, large changes in nominal exchange rates can create pressures on the monetary authorities that cannot always be fully resisted. Although the strength of these pressures differs from country to country depending in part on the degree of openness of the economy and the pervasiveness of wage indexation, some allowance must normally be made for the feedback effect of exchange rate changes on costs and prices in local currency. For example, a change in the nominal exchange rate of 10 percent might induce a more or less automatic rise in domestic currency costs and prices of, say, 3 percent. Thus, a study that concluded that the nominal exchange rate was 10 percent too high on the basis of current prices at home and abroad might imply that the nominal rate should be changed by 13 percent in order for it to return to a sustainable level in real terms.
Some Measures of Competitiveness Used by the Fund Staff
Most of the work done in the Fund on the measurement of competitiveness among industrial countries has focused on the five quarterly indicators now published in International Financial Statistics (IFS). All five indicators measure aspects of the international competitiveness of a country’s manufacturing sector. The first three indices—those for relative export unit values, wholesale prices, and value added deflators—reflect price competitiveness. The last two indices—those for relative unit labor costs and normalized unit labor costs—are intended to reflect cost competitiveness. The weighting scheme for all of the indicators has been designed to take account not only of bilateral trade, but also of the degree to which various countries effectively compete because they tend to export the same types of products to the same geographic markets.11
For the purpose of assessing the sustainability of exchange rates, the first two indicators have important limitations that tend to cause them to underestimate changes in international price competitiveness. For the indicator based on export unit values this problem arises because, even though goods may be differentiated, international competition places limits on how far an exporter’s prices can diverge from those charged by competitors. The indicator also fails to reflect the prices of goods that are not exported anymore because their domestic production costs exceed the prices at which they would be competitive in world markets. Similarly, prices for the relatively homogeneous products that have a heavy weight in the wholesale price index tend to be similar across countries after adjustment for exchange rate changes, but this is hardly a sign that competitiveness among countries is constant.12 In addition, both wholesale and export price indicators are often strongly biased as a result of composition effects because the structure of exports and the weighting structure of wholesale price indices differ substantially across countries.
While the next indicator of price competitiveness—the index based on value added deflators in manufacturing—is normally more relevant, it also has important disadvantages. The main drawback of this indicator is that actual data are published with a long lag and are available quarterly for only a few countries. It is generally necessary for the staff to extrapolate for five or six quarters beyond the most recent benchmark year on the basis of wholesale prices for manufactures adjusted to exclude the influence of changes in raw materials prices. This is a severe limitation since, as explained above, movements in wholesale price indices can often be a misleading indicator of changes in competitiveness.
Given the problems with measures of price competitiveness, more reliance is generally placed on indices of labor cost developments in the manufacturing sector. However, the fourth indicator—that based on current labor costs—is very sensitive to cyclical changes in productivity over periods of several years. For this reason, the indicator based on normalized unit labor costs (i.e., adjusted for cyclical variations in productivity) has generally been regarded as the most useful in assessing exchange rates, and it was recently selected to implement the experimental system of information notices for the 13 member countries for which the indicator is available.13 Thus, it is particularly useful to discuss the main characteristics of this indicator at some length and to analyze its past behavior.
The series on normalized unit labor costs is calculated by dividing an index of actual hourly compensation per worker by an index of output per manhour in the manufacturing sector adjusted so as to eliminate the estimated effect of the cyclical swings in economic activity on productivity.14 The indicator for each country represents the ratio of that country’s index of normalized unit labor costs to a weighted geometric average of corresponding indices for the other 13 industrial countries included in the analysis, after all of the national indices are expressed in terms of a common currency.
To the extent that the price structure in a country’s factor markets tends to fluctuate less over time than the price structure of its goods markets, the normalized unit labor cost indicator benefits from its focus on factor prices rather than goods prices; in particular, there is less risk of a composition bias. Moreover, factor prices in the tradable goods sector—particularly money wage rates—are less subject than goods prices to direct foreign influences. Price indices for factors of production in the tradable goods sector can, therefore, be used directly. There is less need to estimate the relative long-run evolution of the tradable versus the nontradable sectors, or to make allowance for the distorting effects of competitors’ prices on domestic prices.
However, these conceptual advantages are relative rather than absolute; that is, they are significant mainly in relation to the major disadvantages of alternative indicators. In particular, the risk of serious composition bias, while smaller than for alternative indicators, is not negligible. Rates of growth of productivity may differ widely among industries and, as indicated above, such cross-country differences may reflect primarily their differing compositions of total manufacturing output. Moreover, since the only factor of production included in this indicator is labor, it may be biased because of crosscountry differences in longer-run rates of change in capital/labor ratios. To eliminate this source of bias it would be necessary to adjust the changes in productivity (in terms of output per manhour) for the effects of changes in capital/labor ratios; this is a difficult task that so far has not been attempted by the staff.
Reliability of the Normalized Unit Labor Cost Indicator
The practical reliability of the normalized unit labor cost indicator can be assessed by (i) considering its historical performance, (ii) analyzing whether merchandise trade flows are, in fact, responsive to movements in this particular indicator of international competitiveness, and (iii) determining whether developments in normalized unit labor costs in local currencies are largely independent of nominal exchange rate movements.
Historical Performance of the Normalized Unit Labor Cost Indicator
As an aid to analyzing the historical behavior of the indicator of relative normalized unit labor costs, Chart 2 presents quarterly data for 12 member countries.15 The period covered in the chart runs from the beginning of 1963 to the first quarter of 1983. On the strict assumptions of the PPP theory, one would expect the data for the real effective exchange rate based on normalized unit labor costs to exhibit relatively short-lived fluctuations around a more or less stable long-run trend. Furthermore, one would also expect fluctuations around the trend to be closely associated with overshooting of the nominal exchange rate resulting from short-run developments in financial markets. Even if these assumptions were not fully satisfied, however, international price comparisons could still be useful so long as the timing and magnitude of shifts in cost and price competitiveness were clearly associated with recognized major structural changes in the goods and capital markets.
Selected Industrial Countries: Relative Normalized Unit Labor Costs in Manufacturing, Adjusted for Exchange Rate Changes, First Quarter 1963–Third Quarter 1983
(Period average 1963–82 = 100)
The information in Chart 2 suggests that, on the above criteria, the record of the normalized unit labor cost indicator of the real exchange rate during the past 20 years has been mixed. First, there are five countries—the United States, Japan, Canada, Italy, and the Netherlands—for which the level of the normalized unit labor cost indicator in the first quarter of 1983 differed by less than 15 percent from the level that had prevailed in the first quarter of 1963. However, for these countries the difference between the maximum and minimum levels of the real effective exchange rate during the 20-year period was more than 30 percent, and in the case of the United States the average level of the real effective exchange rate during the period 1973–82 was 35 percent lower than it had been during the preceding decade. In addition, while marked movements in the real effective exchange rate were associated with similar movements in nominal effective exchange rates, such movements tended to persist rather than being reversed in a few years. Examples are the appreciation of the Dutch guilder between 1963 and 1974, and the depreciation of the U.S. dollar between 1969 and 1973.
Second, in two cases—Norway between 1969 and 1977 and the United Kingdom between 1976 and 1981—the sharp appreciation in the real effective exchange rate was associated with a major structural change, namely, the discovery and exploitation of North Sea oil reserves. Even in these cases, however, it is noteworthy that the changes in real effective exchange rates associated with the structural shifts were very large, raising the question of whether the movements that actually occurred were an appropriate response to the change in comparative advantage, or whether they partly represented overshooting of the nominal effective exchange rate. The competitiveness indicators, taken by themselves, provide only limited information on this issue. Furthermore, while the shifts that occurred in Norway and the United Kingdom were particularly large ones, most other industrial economies also experienced significant structural changes during this period. The magnitudes of the changes in the real effective exchange rates of Norway and the United Kingdom suggest that even the effects of smaller structural changes on the sustainable exchange rate cannot necessarily be neglected in making assessments of international price competitiveness.
Third, there are several cases where the indicator depicted in Chart 2 did not suggest a worsening of international competitiveness, but the countries concerned experienced growing external imbalances. For example, the current account and overall payments positions of Belgium and Sweden began to weaken from the mid-1970s, even though their international competitiveness as measured by the relative unit labor cost indicator showed little deterioration during this period. Furthermore, despite the substantial apparent improvement in its competitiveness according to the conventional measures during the period 1979–81, Belgium failed to show a significant recovery of export volumes, manufacturing sector employment, or current account balance. In Sweden, the combined effects of the 1981 and 1982 devaluations led in 1983 to improved international competitiveness, a gain in market shares for manufactured exports, and an improvement in the profitability of Swedish industry. Nevertheless, the experience of these two countries suggests the need for caution in interpreting statistical indicators of competitiveness.
The country cases cited above suggest that there is a variety of situations where standard indicators such as relative normalized unit labor costs appear to provide only limited information about changes in a country’s international competitiveness or in the sustainable level of its exchange rate. For this reason, the Fund, in analyzing exchange rate sustainability on the basis of competitiveness indicators, has on occasion extended its analysis to incorporate other types of evidence including trends in manufacturing sector profitability, export market shares, and capital formation in the sectors exposed to foreign competition. Such supplementary indicators suggest the possibility that during the later years of the 1970s these countries experienced a significant deterioration in their overall international competitiveness, even though their indices of relative normalized unit labor costs (Chart 2) were showing an improvement. Two likely causes of such a development are a loss of comparative advantage in traditional export industries (such as shipbuilding, machine tools, and autos (Sweden), and textiles and steel (Belgium)); and/or a process of “marginalization” in which the squeeze on profits caused by domestic cost increases in excess of the rise in world market prices forced the least viable firms in the tradable goods sector out of business.
For example, during the mid-1970s Belgium experienced strong wage-push pressures that sharply reduced the rate of growth of output in the manufacturing sector and led to the fastest rate of decline in manufacturing sector employment among the industrial countries. But, because marginal firms were being eliminated at a relatively rapid rate, the measured growth of labor productivity in the Belgian manufacturing sector accelerated, and standard indicators of the real effective exchange rate actually showed a gain in Belgium’s international competitiveness after 1977. Similar developments may have affected the indices of competitiveness for such countries as Denmark, the Netherlands, and Sweden. Such explanations provide an ex post rationale for the lack of correlation between the apparent improvement in relative unit labor costs and current account developments, but they give little indication of the size of the change in the real exchange rate index that would have produced a sustainable overall payments position.
Fourth, in some cases the timing of the change in the indicator depicted in Chart 2 differed substantially from the timing of major structural changes in the goods and capital markets. For example, between 1969 and 1974 the real effective exchange rate of the U.S. dollar depreciated by 29 percent, that for the Federal Republic of Germany appreciated by 32 percent, and that for Japan appreciated by 15 percent. These large changes in the indicators of international competitiveness occurred during the transition from the Bretton Woods regime of fixed exchange rates to one of managed flexibility, but they were not associated with any specific developments in the structure of goods or capital markets. It is widely assumed that during the period prior to 1970–73 there was a steady deterioration in the competitiveness of the United States relative to Germany and Japan which accounted for the large change that occurred subsequently. However, the empirical indicator of cost and price competitiveness, which remained relatively stable for all three of these countries during the period from 1963 to 1969 (Chart 2). gave little prior warning about either the size of this disequilibrium or the need for a change in real exchange rates. Even after the fact, the timing of exchange rate movements can only be attributed to the change that occurred in the exchange rate regime, suggesting that during this period national authorities were able, through their policy actions, to maintain exchange rates at unsustainable levels for substantial periods of time. While such leeway has doubtless been reduced to the extent that capital has become increasingly mobile since the early 1970s, the possibility remains that competitiveness indices may give only limited warning of such misalignments.
Fifth, since standard indicators of a country’s international competitiveness are based on weighted averages of bilateral real exchange rates, the issue arises as to whether the weights used are an accurate reflection of the relative importance of partner countries. If this were not the case, changes in bilateral real exchange rates between countries that are close trading partners could lead to substantial longer-term changes in the structure of trade, even if—because of offsetting movements in bilateral rates—standard measures of overall competitiveness appeared stable. This issue is important because actual movements in bilateral rates have been considerable. Chart 3 presents data on the real bilateral exchange rates between the deutsche mark and the currencies of five of its partner countries (France, Italy, Netherlands, Belgium, and Denmark) in the European Monetary System (EMS). It is apparent from these data that even in the case of closely integrated economies that belong to a regional exchange arrangement, movements in real bilateral exchange rates were rather large, both prior to the setting up of the EMS and subsequently.
Selected Members of EMS: Bilateral Relative Normalized Unit Labor Costs in Manufacturing, Adjusted for Exchange Rate Changes, First Quarter 1963-Third Quarter 1983
(Period average 1963–82 = 100)
Responsiveness of Trade Flows to Changes in the Normalized Unit Labor Cost Indicator
An alternative way of assessing the reliability of the normalized unit labor cost indicator is to see how well it performs as an explanatory variable in an econometric model of foreign trade flows. To be a highly reliable guide for assessing the sustainability of the exchange rate, the indicator should consistently show a strong effect on foreign trade flows, especially in the long run. This linkage should be reflected in relatively large and statistically robust estimates of the long-run elasticities of import and export volumes with respect to the indicator.
As an experiment, the equations for the volumes of exports and imports of manufactures in the Fund’s world trade mode16 were reestimated using the normalized unit labor cost indicator as the variable reflecting each country’s international competitiveness. The results are presented in Table 2. The estimated long-run elasticities vary considerably across countries, but, on the whole, they cannot be described as large. For example, considering the case of a country where imports and exports are initially equal and where the estimated elasticities are 0.5 in absolute terms for both imports and exports, a 10 percent increase in relative normalized unit labor costs would lead to a trade balance deficit in constant prices amounting to 10 percent of the initial flow of exports or imports. Allowing for a rise in the terms of trade of say 5 percent,17 the deficit in nominal terms would be only 5 percent of the initial flow of exports or imports. Moreover, the standard errors of the estimated elasticities are rather high, indicating that the magnitude of the response of foreign trade flows to the standard measure of cost competitiveness remains uncertain.
Fourteen Industrial Countries: Elasticities of Imports and Exports of Manufactures With Respect to Relative Normalized Unit Labor Costs1
Elasticities are defined such that their expected signs are negative. The impact elasticity gives the response of import or export volume in the first six months; the long-run elasticity gives the total response, which is generally assumed to occur within 2–3 years. Standard errors are in parentheses.
Fourteen Industrial Countries: Elasticities of Imports and Exports of Manufactures With Respect to Relative Normalized Unit Labor Costs1
| Import Volume |
Export Volume |
|||
|---|---|---|---|---|
| Impact Elasticity | Long-Run Elasticity | Impact Elasticity | Long-Run Elasticity | |
| Austria | –0.09 | –0.72 | –0.33 | –0.56 |
| (0.22) | (0.33) | (0.16) | (0.24) | |
| Belgium | — | — | –0.06 | –0.52 |
| (0.17) | (0.29) | |||
| Canada | –0.12 | –0.12 | –0.33 | – 1.14 |
| (0.19) | (0.19) | (0.08) | (0.26) | |
| Denmark | –0.17 | –0.17 | –0.12 | –0.12 |
| (0.09) | (0.09) | (0.17) | (0.17) | |
| France | –0.23 | –0.60 | –0.13 | –0.48 |
| (0.12) | (0.21) | (0.07) | (0.14) | |
| Germany, Federal | –0.08 | –0.49 | — | –0.36 |
| Republic of | (0.18) | (0.18) | (0.13) | |
| Italy | –0.58 | –0.58 | — | –0.51 |
| (0.30) | (0.30) | (0.40) | ||
| Japan | –0.06 | – 1.27 | –0.35 | –1.30 |
| (0.14) | (0.32) | (0.10) | (0.28) | |
| Netherlands | –0.18 | –0.20 | — | –0.07 |
| (0.44) | (0.55) | (0.40) | ||
| Norway | –0.11 | –0.47 | –0.23 | – 1.25 |
| (0.30) | (0.59) | (0.21) | (0.70) | |
| Sweden | –0.33 | –0.45 | –0.42 | –0.82 |
| (0.24) | (0.28) | (0.11) | (0.17) | |
| Switzerland | — | – 1.10 | — | –0.18 |
| (0.25) | (0.18) | |||
| United Kingdom | –0.26 | –0.78 | — | –0.16 |
| (0.12) | (0.22) | (0.18) | ||
| United States | — | –1.03 | –0.26 | –0.93 |
| (0.23) | (0.17) | (0.19) | ||
Elasticities are defined such that their expected signs are negative. The impact elasticity gives the response of import or export volume in the first six months; the long-run elasticity gives the total response, which is generally assumed to occur within 2–3 years. Standard errors are in parentheses.
Fourteen Industrial Countries: Elasticities of Imports and Exports of Manufactures With Respect to Relative Normalized Unit Labor Costs1
| Import Volume |
Export Volume |
|||
|---|---|---|---|---|
| Impact Elasticity | Long-Run Elasticity | Impact Elasticity | Long-Run Elasticity | |
| Austria | –0.09 | –0.72 | –0.33 | –0.56 |
| (0.22) | (0.33) | (0.16) | (0.24) | |
| Belgium | — | — | –0.06 | –0.52 |
| (0.17) | (0.29) | |||
| Canada | –0.12 | –0.12 | –0.33 | – 1.14 |
| (0.19) | (0.19) | (0.08) | (0.26) | |
| Denmark | –0.17 | –0.17 | –0.12 | –0.12 |
| (0.09) | (0.09) | (0.17) | (0.17) | |
| France | –0.23 | –0.60 | –0.13 | –0.48 |
| (0.12) | (0.21) | (0.07) | (0.14) | |
| Germany, Federal | –0.08 | –0.49 | — | –0.36 |
| Republic of | (0.18) | (0.18) | (0.13) | |
| Italy | –0.58 | –0.58 | — | –0.51 |
| (0.30) | (0.30) | (0.40) | ||
| Japan | –0.06 | – 1.27 | –0.35 | –1.30 |
| (0.14) | (0.32) | (0.10) | (0.28) | |
| Netherlands | –0.18 | –0.20 | — | –0.07 |
| (0.44) | (0.55) | (0.40) | ||
| Norway | –0.11 | –0.47 | –0.23 | – 1.25 |
| (0.30) | (0.59) | (0.21) | (0.70) | |
| Sweden | –0.33 | –0.45 | –0.42 | –0.82 |
| (0.24) | (0.28) | (0.11) | (0.17) | |
| Switzerland | — | – 1.10 | — | –0.18 |
| (0.25) | (0.18) | |||
| United Kingdom | –0.26 | –0.78 | — | –0.16 |
| (0.12) | (0.22) | (0.18) | ||
| United States | — | –1.03 | –0.26 | –0.93 |
| (0.23) | (0.17) | (0.19) | ||
Elasticities are defined such that their expected signs are negative. The impact elasticity gives the response of import or export volume in the first six months; the long-run elasticity gives the total response, which is generally assumed to occur within 2–3 years. Standard errors are in parentheses.
The evidence in Table 2 must be viewed as quite tentative. In particular, it is extremely difficult to distinguish among trend developments, structural shifts, and the longer-run effects of lagged competitiveness changes in the econometric determination of foreign trade flows. Thus, the relatively low estimated elasticities with respect to the chosen competitiveness indicator may well be the result of the necessarily simplistic specification of the trade equations.18 Nevertheless, the evidence does suggest that the responsiveness of trade flows to this indicator, while sizable, is not so large and so precisely known that the indicator, taken by itself, can always be relied upon as a fully satisfactory measure of the sustainable exchange rate.
Feedback Effects of Exchange Rate Changes on Normalized Unit Labor Costs
A further issue that arises in using indicators of international competitiveness is that of the extent to which a major change in the nominal effective exchange rate will create feedback effects that alter the levels of prices or unit labor costs in a direction that tends to restore purchasing power parity. Chart 4 provides evidence on this issue for periods corresponding to relatively large changes in the nominal effective exchange rates of three major industrial countries (the United States, Japan, and the Federal Republic of Germany) and three of the smaller countries (Belgium, Norway, and Sweden). In the charts, the dashed line is the nominal effective exchange rate (foreign currency per unit of domestic currency), and the solid line is the ratio of partner country unit labor costs expressed in local currency to those in the country under consideration. If there is a significant feedback effect from exchange rates on domestic labor costs, then one would expect to see an appreciation of the nominal effective exchange rate (a rise in the dashed line) followed by a rise in the solid line, reflecting a slower rate of increase in domestic-currency unit labor costs relative to those prevailing abroad. The six panels in Chart 4, however, provide little evidence that the feedback effects of exchange rate changes on domestic unit labor costs are significant for either the larger or the smaller countries considered.
Selected Industrial Countries: Nominal Effective Exchange Rates and Relative Normalized Unit Labor Costs, in Local Currencies
(First quarter of sample period = 100)
1Foreign currency per unit of domestic currency.2Ratio of partners’ to own labor costs.Note on Consumer Price Indicators
Even though, as noted above, most of the work done in the Fund on the measurement of competitiveness among industrial countries has focused on the five indicators for manufacturing published in IFS—particularly on the indicator based on normalized unit labor costs—there are a number of countries where considerations of data availability have so far precluded the calculation of such indicators. For these countries, as well as for developing countries, the measurement of competitiveness is often based on consumer price indices, as for example in the system of information notices for real exchange rate movements that is being undertaken on an experimental basis by the Fund. For this reason, it is useful to compare the indicator based on consumer prices with that based on normalized unit labor costs in cases where the two indicators can be calculated. To illustrate the results of such a comparison, the left-hand panels of Chart 5 display the real bilateral exchange rates between Japan and the United States, the Federal Republic of Germany, and the United Kingdom, respectively, based on both consumer price indices (dotted line) and indices of normalized unit labor costs (solid line). For all three cases, the two indices essentially show a similar pattern of variation throughout the period chosen. Nevertheless, owing to their differing longer-term trend movements, the levels of the two indicators diverged markedly by the end of the period of the two decades shown in the chart. For example, compared with the first quarter 1963 base period, the cumulative net appreciation of Japan’s real exchange rate with the United States in the first quarter of 1983 was 71 percent, when measured in terms of the indicator based on consumer prices, but only 29 percent when measured in terms of normalized unit labor costs. Japan’s real exchange rate with the Federal Republic of Germany experienced a net appreciation of nearly 64 percent in terms of consumer prices over the same period, whereas the normalized unit labor cost indicator, by contrast, indicated a net depreciation of 6 percent. Similarly, using the former indicator, Japan’s real exchange rate with the United Kingdom experienced a net appreciation of 66 percent, while the latter measure showed a depreciation of 8 percent against sterling.
Selected Industrial Countries: Indicators of Competitiveness Based on Consumer Prices
1Foreign currency per unit of domestic currency.2Ratio of partners’ to own labor costs or consumer prices.Given Japan’s trade performance as compared with the other three countries cited in Chart 5, the indications of cumulative changes in Japan’s international competitiveness provided by the normalized unit labor cost indicator seem much more plausible than those based on consumer prices. This conclusion is not surprising, since long-run increases in the relative prices of nontraded goods were a major source of trend movements in Japan’s consumer price index during the period considered. These observations suggest that although relative normalized unit labor costs and consumer prices may give similar indications of short run developments, the former indicator is superior for the analysis of trends in international competitiveness over periods spanning several years.
A further problem that arises in using competitiveness indicators based on consumer prices, illustrated in the right-hand panel of Chart 5, is the fact that consumer prices appear to be more sensitive than unit labor cost indicators to the feedback effects of nominal exchange rate changes, particularly because a significant portion of the consumption basket consists of imports. The right-hand panels of Chart 5 are the same as those presented in Chart 4, except that a dotted line has been added to represent the ratio of partner-country to domestic consumer prices, with each measured in local currency. In all three cases, there appears to be some evidence of a feedback from exchange rate changes to domestic consumer prices. For example, the February 1982 devaluation in Belgium appears to have had a stronger feedback effect on domestic consumer prices than on unit labor costs in local currency, owing to the large weight of imported goods in the consumer price index and the implementation of a strict incomes policy.
Summary
The discussion of this section suggests the need for caution in using standard indices of international competitiveness to assess the longer-run sustainability of a country’s exchange rate and external position, particularly at times when major changes are occurring in underlying real sector conditions. There can often be cases where a given exchange rate is no longer sustainable, despite the apparent stability of conventional measures of the real effective exchange rate, and the opposite case is no less common. In such circumstances, evidence based on indices of competitiveness must be supplemented by other information on the economy, such as data on the return on capital in the tradable goods sector, or on the growth of the nontradables sector (particularly public services) relative to the traded sector over time.
Instead of analyzing changes in aggregate price indices relative to a base period, a more direct method of assessment would be to compare the prices of specific goods and services in the home market with those prevailing abroad, after expressing both in a common currency. How-ever, the necessary data are not normally available for a sufficiently wide range of commodities.
“Report of Working Group on Exchange Market Intervention,” p. 31.
This issue will be considered in more detail in Section IV, below.
The weighting scheme is built up from disaggregated trade data for manufactures and reflects both the relative importance of a country’s trading partners in its bilateral trade and the relative importance of its competitors in third markets.
The wholesale price index is also subject to problems of double counting, since the prices of certain intermediate commodities may be counted both separately and as part of final output.
The data are published for the following member countries: the United States, Canada, Japan, France, Federal Republic of Germany, Italy, Norway, Sweden, Denmark, Austria, Belgium, the Netherlands, and the United Kingdom; they are also available for Switzerland.
A technical note on the indicator is provided in each issue of IFS, listed in the Contents as “Cost and Price Comparisons for Manufacturing.”
As noted above, the indicators of the cost and price competitiveness for the manufacturing sector are available in IFS for 14 countries. For purposes of exposition, however, only 12 of these countries are included in the panels of Chart 2. The omitted countries are Austria and Switzerland.
See Michael C. Deppler and Duncan M. Ripley, “The World Trade Model: Merchandise Trade,”Staff Papers, International Monetary Fund (Washington), Vol. 25 (March 1978), pp. 147–206.
See Jacques R. Artus, “The Behavior of Export Prices for Manufactures,”Staff Papers, International Monetary Fund (Washington), Vol. 21 (November 1974), pp. 583–604.
In addition to specification problems related to the trade equations, the low estimated elasticities are likely to reflect the existence of measurement errors in the normalized unit labor cost indicator, as well as the fact that this indicator is far from being a perfect proxy for the competitiveness of exportables.
