Prices and Export Performance of Industrial Countries, 1953-63
Author:
Ms. Helen B. Junz https://isni.org/isni/0000000404811396 International Monetary Fund

Search for other papers by Ms. Helen B. Junz in
Current site
Google Scholar Close
and
Rudolf Rhomberg
Search for other papers by Rudolf Rhomberg in
Current site
Google Scholar Close
Publication Date:
01 Jan 1965
eISBN:
9781451947212
Language:
English
Keywords:
SP; currency; deposit-money bank; price; price change; unit value; price effect; price sensitivity; price relative; Export prices; Competition; Price indexes
Download PDF (2.7 MB)

This paper is a revised version of an interim report presented to a joint session of the American Statistical Association and the American Economic Association at the Annual Meetings of the Allied Social Science Associations in Chicago, December 28, 1964.

Abstract

This paper is a revised version of an interim report presented to a joint session of the American Statistical Association and the American Economic Association at the Annual Meetings of the Allied Social Science Associations in Chicago, December 28, 1964.

The responsiveness of trade flows to relative price changes is an important factor to be considered when deciding upon policy measures aimed at correcting imbalances in international payments. For example, the imposition of temporary charges on imports and the decision to rebate certain indirect taxes on exports in the United Kingdom late in 1964, the devaluation of the Canadian currency in 1962, and the revaluation of the deutsche mark and the Netherlands guilder in 1961 were all policy decisions in which account had to be taken of the effect of price changes on exports and imports.

More generally, for many countries the choice among domestic economic policy measures is influenced by the effects they are thought to have, either directly or through their impact on incomes and price levels, on the external payments position. It becomes important, therefore, to ascertain the nature and magnitude of these effects. But changes in economic activity, shifts in demand, and price movements are to some extent interrelated, and it is exceedingly difficult to assess the separate effect of any one factor.1 However, statistical experiments can be designed in such a way as to minimize the difficulties of separating the effects of price changes from the influences of changes in economic activity on international trade. The present paper is an attempt in that direction.

Despite a number of technical and theoretical limitations, the results indicate fairly clearly that price competitiveness plays an identifiable, though not a dominant, role in the export performance of industrial countries in markets for manufactured products. When cross-section and time-series approaches were combined, export market shares were found to be significantly affected by relative price changes.2 However, the proportion of the observed variation in market shares which is explained by price factors is not very high. The results support the hypothesis that price changes show their full effect only in the course of several years. Price effects were found to be stronger, and price elasticities higher, when changes in prices and exports over 4-year periods were compared than when year-to-year changes were analyzed.

Measurement of Price Competitiveness

The problem

The study is concerned with exports of manufactures by 11 major industrial countries3 to selected market areas over the period 1953–63. The markets are those of the 11 countries themselves and Switzerland.4 In some computations, 5 composite markets are also considered, namely (1) the world market, (2) all major industrial countries, i.e., the 11 exporting countries plus Switzerland, (3) the European Economic Community (EEC),5 (4) the industrial countries excluding the EEC, and (5) the “rest of the world,” i.e., all countries excluding the 11 countries and Switzerland. Exports of manufactures consist of Standard International Trade Classification (SITC) sections 5 (chemicals), 7 (machinery and transport equipment), and 6 plus 8 (basic manufactures and miscellaneous manufactured goods) of the revised United Nations Code.6 Annual data are used.

The principal objective is to identify competitive price effects upon exports of the 11 countries to the various markets. For this purpose, a country's “price competitiveness” is measured in terms of several alternative ratios of one of its own price or cost indices to a weighted average of the corresponding indices of the other exporting countries. Changes in these price or cost ratios are put in relation to the relative export performance of the 11 countries. Demand for foreign manufactured products in each of the markets in the sample is taken as given. The study is not concerned with the manner in which this demand is influenced by economic activity in a particular market, by the prices of imports relative to the prices of domestic import-competing products, or by a number of other factors which affect imports. Attention is focused, instead, upon changes in the distribution of a given total amount of exports to a particular market among competing supplier countries, e.g., upon changes in the competing countries' market shares.7

In order to gauge the influence of price factors on export performance, three sets of statistical estimates were made: (1) elasticities of substitution, in each separate market, of the exports of each supplying country for the exports of all other suppliers of the sample;8 (2) coefficients of responsiveness of export shares of all exporters in all markets to changes in relative prices; and (3) measures of the relation between relative prices and the deviations of actual exports from those which would have occurred if exporting countries had maintained their previous shares in every market.

The extent to which the isolation of price influences can be expected to be successful depends upon the appropriateness of the assumption that influences other than relative prices of exports are not systematically related to changes in these relative prices but exercise their influence on market shares in a random manner. In fact, there is little doubt that this assumption is not fully warranted. One must expect, for instance, that an economy operating at or near full capacity will experience difficulties in increasing its exports in line with the growth of world trade. In a system of perfectly flexible prices, limitations of capacity should be fully reflected in rising export prices. But if prices are not perfectly flexible, their changes may be smaller than those which would clear the market, with the result that order books and delivery periods will be lengthened when over-all demand presses against capacity. If the entire joint effect of somewhat higher prices and a lessened ability or willingness to supply commodities on customary delivery terms is ascribed to the price increases alone, the elasticity of substitution of the country's exports for those of its competitors will tend to be overestimated.

Shifts in the commodity composition of demand in particular markets create a further difficulty. If imports of various commodities are differently affected by changes in economic activity, expansion of demand in a particular market will favor those exporters supplying the commodities for which demand increases the most. An exporting country whose major exports are favored by a shift in the pattern of demand in a certain market will, if anything, be enabled to increase its prices while at the same time it gains a larger share in that market. One would thus find an observation with the “wrong sign.” This difficulty will be the less important, the more uniform are the income elasticities of demand for the goods supplied by competing exporting countries to each market.9 However, the average income elasticity of demand for the exports of one country may be different from that for the exports of all other supplying countries in an individual market if the export bundle of that country differs substantially from the average, or typical, commodity pattern of trade among all industrial countries. As a result, the price sensitivity of its exports may be underestimated or even appear to be “perverse.”

Influence of structural factors

Some indication of the importance of changes in commodity composition and in the geographical distribution of demand is given in Table 1. Changes in exports may be thought of as having occurred because of four main factors: (1) growth in over-all demand, (2) changes in the geographical distribution of demand, i.e., differential growth of import demand in individual markets, (3) changes in the commodity composition of demand, i.e., differential growth in demand for different commodities in each market, and (4) changes in relative prices and other factors.

Table 1.

Selected Industrial Countries: Effect of Geographic and Commodity Composition on the Growth of Exports to Industrial Markets, 1953–55 to 1956–59 and 1956–59 to 1960–631

(In per cent)

article image

Changes in the total value of exports of each country to the other 10 countries and Switzerland (column 1) are compared with changes computed on three alternative assumptions, namely, first, that countries maintained their market shares of the previous period in the industrial market as whole, i.e., in the combined market of the other 10 countries and Switzerland (column 2); second, that they maintained their market shares in the market of each of the industrial countries; and third, that they maintained their market shares in each of three commodity groups in each of the industrial countries. The difference between the changes in the value of exports derived on the second assumption and column 2 is a measure of the effect of changes in geographic distribution of demand on the growth in countries' exports (column 3); that between changes in the value of exports derived under the second and third assumptions is an indication of the effect of changes in commodity composition on the growth of countries' exports (column 4). The three commodity groups are SITC 5 (chemicals), 7 (machinery and transport equipment), and 6 plus 8 (basic manufactures and miscellaneous manufactured products) of the revised United Nations Code. Market shares and percentage changes are computed from averages for three periods (1953–55, 1956–59, and 1960–63).

For most countries, differences between actual export performance (column 1) and that which could have been expected on the basis of over-all growth in demand (column 2) were fairly consistent for the two time periods shown (1953–55 to 1956–59 and 1956–59 to 1960–63). Exports of Belgium, Canada, the United Kingdom, and the United States grew more slowly, and those of France, Germany, Italy, Japan, and Sweden grew faster, than over-all demand in both periods.

Changes in geographical distribution, as shown in column 3 of Table 1, were much more important in explaining export growth in the later period than they had been in the earlier one. For the EEC countries, for example, favorable shifts in the geographical pattern of demand (to a large extent associated with the economic integration of the EEC) explained about one fifth of the changes in their combined exports during the period 1956–59 to 1960–63. For the earlier period, 1953–55 to 1956–59, only a fractional portion—just over one twentieth—of the increase was explained by the geographical factor.

By contrast, the United States, Canada, and Japan, whose principal markets among the countries considered lie outside the EEC, had to overcome a very considerable adverse geographical effect in the later period, because their major markets grew at a slower rate than the industrial market as a whole. Thus Japan's rapid growth in exports is even more remarkable when the geographical factor is taken into account, and the relatively small expansion of U. S. exports appears to have resulted to a large extent from geographical shifts and only to a minor extent from relative price changes and other factors.

In contrast to geographical changes in the pattern of demand, shifts in commodity composition, shown in column 4 of Table 1, affected growth in individual countries' exports to only a moderate extent. During both periods shown, changes in the distribution of demand between the three broad commodity groups adversely affected the exports of Austria, Belgium, Canada, France, and Japan, and in the last period also those of the Netherlands. For these countries, except Canada and—in recent years—Japan, this adverse movement was more than compensated, or at least mitigated, by favorable changes in the geographical distribution of demand. In the later period, in all but one instance, the effects of the geographical factor dominated those resulting from changes in commodity distribution of demand.

Adjustment for changes in commodity composition of demand therefore appears less pressing than elimination of the influence of geographic differences in market growth. For this reason, and also because price data by commodity class are not available, changes in commodity composition were generally left out of account in the estimation of price effects presented in this paper. On the other hand, allowance was made throughout the study for the effect of shifts in the distribution of demand for all manufactured goods among markets.

Price and cost variables

Three alternative price or cost variables were considered: (1) Unit values of exports of manufactures were chosen since constant-weight export price indices are not available for most countries. (2) The manufacturing component of the wholesale price index for the exporting country was taken as an alternative measure of price competitiveness. (3) Wage costs per unit of output in the manufacturing sector were used in order to test the hypothesis that, whereas prices of internationally traded goods must always move in close correspondence, costs (and therefore profits) may diverge and thus provide differential incentives to export or to sell in the home market. All three indicators were adjusted for changes in exchange rates so as to reflect prices or costs in U.S. dollars.

A comparison of movements in these price and cost indices and in the index of productivity per man-hour for 4 major industrial countries is given in Chart 1.10 The relation between changes in the three price or cost indices varied considerably from country to country. For instance, between 1953 and 1963, Germany experienced increases in labor costs far in excess of the rise in wholesale prices, which in turn advanced more than export unit values. A notable divergence of price and cost trends occurred in Japan, where the index of unit labor costs rose by 14 per cent over the 11-year period, while wholesale prices remained roughly constant and export unit values declined by 13 per cent. In the United States, by contrast, unit labor costs increased less between 1953 and 1963 than those of any other industrial country covered by the study, yet the rise in U.S. export unit values exceeded that in any of the other countries.

Chart 1.
Chart 1.

Selected Industrial Countries: Indices (1953 = 100) of Prices, Costs, and Productivity in Manufacturing, 1953–641

Citation: IMF Staff Papers 1965, 002; 10.5089/9781451947212.024.A003

1 From Table 8 (pp. 266–67).

Some of the divergence in movements between the various price indicators may be explained by the fact that in some countries pricing policies for export markets differ from those for the domestic market. Moreover, increases in investment and technological advances may be greater in export industries than in the manufacturing sector in general. Increases in productivity in export industries may therefore be greater than those for all industries, so that increases in unit labor costs or wholesale prices are not necessarily associated with corresponding increases in export prices.11

As a result of the divergent movements in these indices, the size, and indeed the direction, of computed changes in a country's price competitiveness depend upon the choice of the price or cost index on which the measure of price competitiveness is based. Since there appears to be a closer correspondence between shorter-run movements in at least two of the indices (export unit values and wholesale prices) than between their long-run trends, measures of short-run changes in countries' competitive positions, such as deviations from trend, are somewhat less affected by that choice.

Limitations of price and cost data

Proper measurement of price competitiveness would require indices of “delivered” prices by market. Such indices would measure—in addition to changes in basic export prices—changes in transportation and distribution costs and in tariffs. However, no such measure is currently available. The fact that existing price indicators do not reflect discriminatory changes in tariffs has become particularly important in recent years.

Unit value indices, the closest approximation to an export price series generally available, are based on a country's f.o.b. exports to all regions and do not, therefore, take into account differences in prices to ultimate purchasers in different markets. This may be a stronger limitation for the United States, the United Kingdom, and Japan, whose exports are concentrated heavily in nonindustrial areas, than for the other exporters which trade mainly with industrial countries. For these other exporters, the unit value index represents at least an average indicator for the industrial markets, while for the United States, the United Kingdom, and Japan the index reflects to a large extent unit values for goods going to nonindustrial markets, for which commodity composition and pricing policies may differ from those applying to industrial markets.

The usefulness of unit value indices is particularly limited by the fact that they reflect only the prices of goods which are actually exported. If goods cease to be exported, for example because they are overpriced, they will be excluded from the unit value index. Thus a country could “price itself out of world markets,” suffer a loss in market shares, and at the same time appear to have had stable, or even declining, export prices as measured by the unit value index. In these circumstances, a loss of exports would erroneously be ascribed to factors other than price changes.

Wholesale prices are thought to give, at least in this respect, a better indication than do unit values of changes in general price competitiveness of a country, since they measure price changes not only for goods which are actually exported but also for goods which are potential exports. However, in contrast to indices of export unit values, wholesale price indices do not reflect price changes caused by changes in export subsidies and tax rebates. Moreover, they do reflect changes in prices of imported goods. To the extent that these goods re-enter international trade either as component parts or as re-exports, the inclusion of their prices in a measure of export price competitiveness is quite appropriate; but to the extent that they are destined for domestic use, wholesale prices are less useful as indicators of external price competitiveness, particularly for countries where imports account for a large portion of domestic expenditure.12 Nevertheless, since home-produced goods compete, in many instances, directly with imports, movements of their prices cannot for long diverge to any large extent from those of the imported substitutes. Therefore, the inclusion of prices of imported goods does not invalidate the use of wholesale price indices, at least as supporting evidence, in the explanation of changes in export performance resulting from price factors.

A major objection to both wholesale prices and unit labor costs as measures of export price competitiveness relates to differences in coverage between these two series, on the one hand, and the export data, on the other: the manufacturing sector as defined by the Standard International Trade Classification, on which the export series and export unit value indices are based, differs considerably from that defined by the International Standard Industrial Classification published by the United Nations, on which the wholesale price indices and the series underlying the indices of unit labor costs and productivity are generally based.

The measurement of movements of unit labor costs presents technical difficulties because of differences in coverage and method of construction of the basic data series. Remaining inadequacies in the indices that were constructed are, however, likely to be outweighed by other considerations which would tend to make changes in labor costs a questionable explanatory factor of export performance. The relative importance of labor costs in over-all unit costs varies from country to country and over time. For example, European producers often have higher unit costs of raw materials than do U.S. enterprises, giving rise, on this ground alone, to an initial difference in the ratio of labor costs to total costs as well as to greater variation in this ratio over time.13 Thus changes in labor costs do not necessarily indicate changes in total costs.

For reasons given in the preceding paragraphs, none of the three indicators is an ideal measure of price changes for the purpose of this study. Yet each of these indices reflects at least part of the total variation in the price and cost elements which determine a country's price competitiveness. Furthermore, when evidence of price sensitivity can be based on computations using each of the indicators in turn, reservation about the use of any one of them is reduced.

Indices of price competitiveness

To measure changes in the degree of price competitiveness, a country's prices must be brought into relation to the prices of other exporting countries. Three sets of indices of price competitiveness, corresponding to the three price or cost variables used, were constructed for each exporting country. Each set consists of 1 index for each market, obtained by dividing the exporter's price index by the weighted average of the corresponding price indices of the other exporting countries; the weights are their exports of the preceding year (in 1953 U.S. dollars) to the particular market. These indices were computed not only for the 12 single-country markets but also for the 5 composite markets. For each exporter there are thus 16 indices of price competitiveness, 1 index for each of the markets.14 An improvement in competitiveness is indicated by a decrease in a country's index. A particular country's index of price competitiveness varies from market to market in accordance with the shares of its competitors.15 The variations may be quite large. For example, as shown in Table 2, Germany's unit value index rose by 7 per cent between 1953 and 1963, but its index of price competitiveness based on unit values deteriorated (i.e., rose) by more than 11 per cent in the U.S. market and improved (i.e., declined) by almost 8 per cent in the Canadian market. The deterioration in the U.S. market is explained by the fact that Germany's chief competitors there, measured by the volume of their exports to that market, are Canada and Japan, whose unit value indices declined over the period 1953–63. In the Canadian market, on the other hand, Germany faces competition primarily from the United States and the United Kingdom, whose unit value indices rose appreciably more than did that of Germany.

Table 2.

Selected Industrial Countries: Changes in Export Unit Value Indices and in Indices of Price Competitiveness Based on Export Unit Values in Selected Markets, 1953–63 and 1959–63 1

(In per cent)

article image

Export unit value indices are based on data in U.S. dollars. The index of price competitiveness of an exporter in a particular market is defined as the ratio of that exporter's unit value index to the average of the unit value indices of all other exporting countries of the sample, weighted by their export shares in that market.

As may be seen from Table 2 and Chart 2, relative price trends in a number of major countries changed substantially after 1959.16 Most continental European countries improved their relative price positions, in terms of export unit values, fairly steadily between 1953 and 1959, but between 1959 and 1963 their indices of price competitiveness either deteriorated, as in France, Germany, and the Netherlands, or improved at a slower rate, as in Italy. This shift in relative price trends is explained in part by differential trends in prices in individual countries, but also to a large extent by changes in exchange rates (devaluation of the French franc in 1957 and 1958, revaluation of the deutsche mark and the Netherlands guilder in 1961, and depreciation of the Canadian dollar between 1959 and 1962). During the period 1959–63, price stability in the United States led to an improvement in U.S. price competitiveness in some of the markets and to a slowing in its rate of decline in others. These price developments in the United States, which has a large share in many markets, also contributed substantially to the rise (i.e., worsening) of the competitiveness indices of most European countries. Preliminary data for 1964, shown in Table 9 and Chart 2, indicate that this trend continued.

Table 3.

Selected Industrial Countries: Estimated Elasticities of Substitution of Exports of Manufactures to Various markets for Competitors' Exports With Respect to Changes in Indices of Price Competitiveness based on Export Unit values1

article image

This table lists the regression coefficients, ηij, equation (1), Appendix I (p. 261). They represent elasticities of substitution of the exports of countries i (listed in the heading) to markets j (in the stub) for the exports of the other exporting countries of the sample to that market. Coefficients which are significant at the 5 per cent level are shown in boldface type; coefficients which had a positive sign are indicated by a plus sign.

Table 4.

Selected Industrial Countries: Size Distribution of Estimated Elasticities of Substitution by Exporting Country and Market from Annual Data 1953–63

(Number of cases)

article image
Source: Based on data in Table 3.
Table 5.

Selected Industrial Countries: Short-run Price Effects—Regression Equations with Percentage Changes from Preceding year in Market Shares (Dependent Variable) and in Alternative Price or Cost Relatives (Independent Variable)1

article image

Dependent variable: percentage change from preceding year in exporting countries' market shares (A) for all exporters, all markets, and all years, (B) for each exporter in all markets and all years, (C) in each market for all exporters and all years, and (D) in each year for all exporters and all markets. Independent variable: percentage change from preceding year in the ratio of the price or cost index of the exporting country to the average of the corresponding price or cost indices of all exporting countries of the sample weighted by their exports to individual markets. Regression coefficients that are statistically significant at the 5 per cent level are shown in boldface type.

Table 6.

Selected Industrial Countries: Short-Run Price Effects—Regression Equations with Percentage Changes from Preceding year in Market Shares (Dependent Variable) and in Alternative Price or Cost Relatives (Independent Variable)1

article image

The sample period (1953–63) was divided into three subperiods, (1) 1953–55,(2) 1956–59, and (3) 1960–63; for each variable used, percentage changes from the annual average of the first to that of the second, and from that of the second to that of the third, of these subperiods were computed. Dependent variable: percentage change from preceding subperiod in exporting countries' market shares (A) for all exporters, all markets, and both (second and third) subperiods, (B) for each exporter in all markets and both subperiods, (C) in each market for all exporters and both subperiods, and (D) in each subperiod for all exporters and all markets. Independent variable: percentage change from preceding subperiod in the ratio of the price or cost index of the exporting country to the average of the corresponding price or cost indices of all exporting countries of the sample weighted by their exports to individual markets. Regression coefficients that are statistically significant at the 5 per cent level are shown in boldface type.

Table 7.

Selected Industrial Countries: Price Effects on Total Exports of Manufactures—Regression Equations of Deviations of Actual From Constant-Shares Exports on Changes in Alternative Price or Cost Relatives1

article image

The variables in these regression equations are percentage changes (A) of each of the eight years 1956–63 from the annual average of the three preceding years or (B) of the annual average of each of two subperiods, 1956–59 and 1960–63, from the annual average of the preceding subperiod (1953–55 and 1956–59, respectively). Price effects computed with variables under (A) and (B) are referred to as “short-run” price effects and “longer-run” price effects, respectively. In each set of computations, the variables are defined as follows: Dependent variable—change in actual exports of manufactures minus change in exports computed on the assumption that exporters had maintained their previous shares in the market of each of the other industrial countries, expressed as a percentage of previous actual exports. “Previous shares” are those observed on average (A) during the three preceding years or (B) during the preceding subperiod. In the equations labeled “export volume,” the computation relating to the dependent variables uses export data in 1953 prices, and the constant-shares assumption relates to shares of total exports of manufactures in each market. In the equations labeled “total export value,” the same computations are carried out with export data in current prices. In the equations labeled “export value by SITC group,” the computations relating to the dependent variables are also carried out with export data in current prices, but constant-shares exports are derived on the assumption that exporters maintained their previous shares in each of three commodity groups (SITC 5, 7, and 6 plus 8) in each market. Independent variable—Percentage change in the average of each exporter's price relatives in all markets considered, weighted by the exporter's sales of manufactures to these markets.

Table 8.

Selected Industrial Countries: Manufactures−Indices of Prices and Costs in U.S. Dollars, and Index of Productivity, 1953–64

(1953 = 100)

article image
article image
Sources: See Appendix II (pp. 263–65).

Partly estimated.

Table 9.

Selected Industrial Countries: Indices of Price Competitiveness for Exports of Manufactures to the Industrial Market, 1953–641

(1953 = 100)

article image
article image

The industrial market consists of the following countries: Austria, Belgium, Canada, France, Federal Republic of Germany, Italy, Japan, Netherlands, Sweden, Switzerland, United Kingdom, and United States. Indices of price competitiveness are ratios of prices (or costs) of the exporting country to the weighted average of prices (or costs) of its competitors. In computing a country's index of price competitiveness, exports to that country have not been included in the weights applied to prices of competing countries. Data for 1964 are partly estimated.

Chart 2.
Chart 2.
Chart 2.

Selected Industrial Countries: Indices (1953 = 100) of Price Competitiveness and Percentage Market Shares for Exports of Manufactures to the Industrial Market, 1953–641

Citation: IMF Staff Papers 1965, 002; 10.5089/9781451947212.024.A003

Only 2 of the 11 exporting countries showed a fairly consistent price behavior throughout the entire period: Japan's index of price competitiveness in the world market improved almost consistently at an annual rate of nearly 2 per cent, while the price position of the United Kingdom deteriorated at an annual rate of almost 1.5 per cent. Between 1953 and 1959, only Italy showed a greater relative price improvement than did Japan, and only the United States experienced a greater deterioration in its relative price position than did the United Kingdom. Between 1959 and 1963, Japan's improvement in price competitiveness was exceeded only by that of Canada, whose exchange rate depreciated during this period; and the decline in the United Kingdom's price competitiveness was exceeded only by that of Germany, mainly as a consequence of the revaluation of the deutsche mark.

Although Chart 2 seems to indicate a closer correspondence between movements in market shares and relative export unit values than between the former and either wholesale prices or unit labor costs, the question which of the three price or cost series constitutes the most appropriate base for a measure of price competitiveness was not prejudged on this evidence alone. All three indicators were therefore used in turn in the statistical computations relating variations in export performance to changes in price competitiveness.

Statistical Results

Elasticities of substitution by exporter and market

In the first set of computations, intended to be exploratory, time-series data were used to derive, separately for each market, elasticities of substitution of the exports of each supplier for those of the other exporting countries of the sample. For each of the 11 countries, 16 markets were considered, namely those of the other 10 exporting countries, Switzerland, and the 5 composite markets. Thus 176 elasticities of substitution were computed for each of the 3 competitiveness indices. For each market, the relationship was established between proportionate changes in the ratio of one country's exports to the exports of the other countries of the sample (“export ratios”) and proportionate changes in the country's index of price competitiveness as defined in the preceding section. The export data used were “real exports” in 1953 prices, i.e., exports in current prices deflated by the index (1953 = 100) of export unit values. Since the export ratios may exhibit trends over the period studied which result from factors other than relative price movements, a linear trend term was included in the estimating equations; consequently, the calculated elasticities of substitution reflect the association of deviations from trend in the two variables. It is unavoidable in this procedure that a longer-run movement in the export ratio that in fact reflects the longer-run drift in the price competitiveness index will also be absorbed in the trend term. The resulting elasticities should, therefore, be regarded as short-run elasticities. The exact form of the estimating equation is shown in Appendix I (pp. 260–61).

The 176 elasticities estimated on the basis of relative export unit values are presented in Table 3, and summarized in Table 4. The majority, 126 (or 72 per cent), show the expected negative sign; of these, 44 (or 25 per cent of all estimated coefficients) are significantly different from zero at the 5 per cent level.17

Despite the unevenness in the results of this set of computations, it is possible to draw some tentative conclusions. There appears to be considerable variation in the sensitivity of exports to price changes among exporting countries and markets. These differences may in part be the result of varying influences of nonprice factors for which no allowance is made in the computations. The estimated elasticities of substitution in individual markets are, on the whole, somewhat higher in absolute value than those relating to the 5 composite market areas. The 27 significant negative coefficients relating to individual markets have an average value of −5.1, while the 17 significant coefficients relating to composite markets have an average value of −2.1. Geographic aggregation of markets appears to introduce a significant bias toward lower elasticity estimates. This type of aggregation bias is also evident in most of the individual columns of Table 3, where the estimated elasticities of substitution for composite markets are lower than the estimates for the component markets. For instance, for France, Germany, Italy, the United Kingdom, and the United States, the separately estimated elasticities for the EEC and non-EEC industrial markets are larger (in absolute value) than those for the combined market of industrial countries, which comprises the 2 submarkets. The elasticity of substitution for Germany's exports in the composite market of the non-EEC industrial countries is −2.4, yet the weighted average of the individually estimated elasticities in these markets is −3.3. A similar downward bias must be expected to result from aggregation of different commodities. The global elasticity estimates generally available from econometric studies must, therefore, be interpreted with caution; they may seriously underestimate the true effects of relative price changes.

Corresponding computations with the alternative indices of competitiveness, based on wholesale prices and unit labor costs, tend to corroborate the results obtained with export unit values but are, on the whole, less reliable. Of 113 negative coefficients obtained in the computations with wholesale prices, 30 are significant at the 5 per cent level; of these, 22 relate to exporting countries and markets for which significant negative elasticities were also obtained with export unit values. Using labor costs, 123 negative coefficients were computed; of 42 that are significant, 27 are for exporting countries and markets for which the computations based on export unit values also give significant negative cofficients. The averages of the significant negative coefficients for individual and composite markets are, respectively, −4.2 and −2.4 for the equations with wholesale prices and −2.6 and −1.4 for those with unit labor costs.

The facts that more than one fourth of the elasticity coefficients have the theoretically unexpected sign, and that a high proportion of estimates are statistically not significant, indicate that in many cases imperfections of the data and the influence of nonprice factors prevent reliable estimation of substitution elasticities, country by country and market by market.18 The influence of these factors, which disturb or modify the normal relation between export volumes and prices, is particularly evident in a statistical analysis based on a very small number of observations. In an attempt to overcome this difficulty, an approach combining cross-section and time-series analysis was chosen.

Market shares and relative prices: combined cross-section and time-series results

The second set of computations relates percentage changes in the shares of the 11 exporting countries in each of the single-country markets, i.e., those of the 11 countries and Switzerland, to percentage changes in the corresponding price relatives. The estimated coefficients can be interpreted as elasticities of market shares with respect to changes in price relatives. Unit values of exports of manufactures, wholesale prices, and unit labor costs were again used as alternative price variables. The use of percentage changes from a preceding period permits comparison of variations in market shares and price relatives among exporting countries and markets. It is therefore possible to cover in one statistical computation the experience of all exporting countries in all markets during the sample period.

The price variable used in these computations is somewhat different from the indicator of price competitiveness defined earlier in this paper. The equations used to estimate the elasticities of substitution by exporting country and market described in the preceding section relate the ratio of a country's exports to those of its competitors, as the dependent variable, to the ratio of its price index to the weighted average of the price indices of the competing countries. By contrast, the equations described in this section take as the dependent variable the percentage changes in market shares, i.e., the ratios of countries' exports to those of all countries (including itself) of the sample. The appropriate price variable is, therefore, the ratio of the price index of the specified exporting country to the weighted average of the corresponding price indices of all 11 countries; the weights are again the exports of the 11 countries to the specified market. This price variable is referred to here as a price relative, to distinguish it from the index of price competitiveness used elsewhere in this paper. Increases or decreases in price relatives show the deterioration or improvement, respectively, of a country's price competitiveness in a particular market in relation to the average price behavior of the whole group of 11 countries. Variations in price relatives are somewhat smaller, particularly for countries with large market shares, than the variations of the corresponding indices of price competitiveness, just as changes in export shares are smaller than the corresponding changes in the ratio of a country's exports to those of the other countries of the sample.

The elasticities of substitution presented in the preceding section were computed from annual deviations from trend of the export and price ratios. They are thus estimates of short-run elasticities. The increased number of observations in the present computation permitted, in addition to estimates of similar short-run coefficients of responsiveness of market shares to price relatives, an assessment of longer-run coefficients of responsiveness. The equations were therefore alternatively fitted to (1) year-to-year percentage changes in the variables, and (2) two percentage changes between the annual averages of three subperiods. The three subperiods are 1953–55, 1956–59, and 1960–63. This subdivision was dictated essentially by the desire to have, as far as possible, nonoverlapping subperiods of equal length; no attempt was made to divide the sample period into segments representing complete cycles, a task which would, at any rate, have been frustrated by the differences in timing of the movements of business activity in the various countries. Nevertheless, the chosen subperiods exhibit similar cyclical characteristics, at least in the sense that each includes one business recession affecting a number of industrial countries.

The short-run sample consists of 1,210 observations (ten year-to-year percentage changes for 11 countries, each exporting to 11 single-country markets). The longer-run sample consists of 242 observations (two period-to-period, percentage changes covering the same exporting countries and markets). In addition to the global computations with all observations, separate elasticity estimates were also computed from various subsets of observations, namely, for each exporting country, for each market, and for each year or subperiod.19

The estimated short-run and longer-run coefficients of responsiveness of market shares to changes in price relatives are shown in Tables 5 and 6, respectively. The extent to which percentage changes in the price relatives explain short-run variations in market shares is rather small. The values of the coefficients of determination (r¯2) range from zero to around 0.25; in the majority of cases, they are less than 0.10. Nevertheless, the t ratios indicate that many of the regression coefficients, even where r¯2 is low, are significantly different from zero, so that the effect of changes in relative prices, though in many cases dwarfed by the influence of other factors not accounted for in the regression equations, is clearly evident.20

The elasticity of market shares with respect to changes in unit value relatives for the whole sample of 1,210 observations is −1.85; this coefficient is 7½ times greater than its standard error. Of the 33 coefficients for individual exporting countries, markets, and years, all except 2 have a negative sign. The 16 coefficients that are significantly different from zero at the 5 per cent level range from −1.49 to −3.92.

The results of the regression equations with wholesale price relatives indicate that in many cases changes in this measure of price competitiveness are less closely associated with changes in market shares than are changes in relative unit values. Among the 33 estimated coefficients, 12 have a positive sign and only 8 are both negative and significantly different from zero. Export unit values prove clearly superior also to unit labor costs as an explanatory factor of changes in market shares. However, in many instances where the elasticity coefficients computed with unit value relatives appear, by statistical criteria, fairly reliable (as for the exporting countries, France, Germany, Italy, and Sweden, and the markets of Canada and the United States), the results are corroborated by those obtained with wholesale price relatives, and sometimes also by those with labor cost relatives.

Those elasticity coefficients computed with wholesale price relatives that are statistically significant are often somewhat higher than the corresponding coefficients computed with unit value relatives. This may be due to the fact that the inclusion of prices of imported goods in the wholesale price index may result in smaller variations, among countries, in wholesale prices than in export unit values and thus in larger elasticity estimates.

The longer-run price elasticities shown in Table 6 indicate the responsiveness of market shares to changes in price relatives when a period of 3½ or 4 years (between the midpoints of the three subperiods) is allowed for adjustments of export shares to relative price changes. The extent to which price factors explain variations in export performance over such a time span is considerably greater, at least for the equations with unit value relatives, than that observed in the computations of short-run price effects. Although some coefficients of determination are still zero, many range from over 10 per cent to almost 50 per cent. In the equation with all 242 observations, changes in relative unit values explain 23 per cent of the observed variation in market shares. As would be expected, these longer-run elasticities are appreciably higher than the year-to-year elasticities shown in Table 5. For the entire sample of all exporters, markets, and subperiods, a longer-run elasticity of almost −3 is found, a value which is over 8 times greater than its standard error. Of the 25 elasticity coefficients for individual exporting countries, markets, and subperiods, only 1 shows a positive sign;21 the 14 statistically significant coefficients range in value from −2.29 to −6.80.

As an explanatory factor of variations in market shares, the superiority, by statistical criteria, of unit values over wholesale prices and, particularly, over unit labor costs is again evident in the tabulation of the longer-run price effects. The elasticity of market shares with respect to wholesale price relatives is −2.3, but the equation explains only 6 per cent of the observed variation in market shares. Among the 25 coefficients for exporting countries, markets, and subperiods, 5 fail to have a negative sign and only 7 are both negative and statistically significant; these do, however, correspond in magnitude reasonably closely to the corresponding elasticity coefficients obtained with unit value relatives. None of the coefficients relating to labor cost relatives is both negative and significant.

A comparison of Tables 5 and 6 shows that the estimated longer-run elasticities of market shares with respect to unit value relatives are generally higher than the corresponding short-run elasticities, though there are some exceptions among the coefficients that are not significant. These instances of reversal of the expected pattern could, in part, be the result of shifts in commodity composition, which are likely to play a larger role over a longer period than from one year to another.22 The next (and last) set of computations, described below, permits at least a partial assessment of the probable influence of changes in commodity composition on the price effects estimated from aggregate data on exports of manufactures.

Relative prices and over-all export performance

This set of calculations explored the relationship between price competitiveness and deviations of countries' actual exports to all industrial markets from a “constant-shares norm,” i.e., from exports computed on the assumption that each country had maintained its previous market shares in each of the industrial countries.23 The use of percentage changes again permitted a combination of cross-section and time-series approaches. Export performance was compared among countries by expressing the deviations of actual from constant-shares exports as percentages of actual exports in the period on which the norm was based. These percentage deviations thus measure the percentage gains or losses from a preceding period in a country's exports, which are due to factors other than market growth in its customer countries.

Two sets of percentage changes were again used in order to identify short-run and longer-run price effects on countries' export performance. In the first set of 8 annual observations (1956–63), the constant-shares norm for each year was based on the assumption of maintenance of a country's shares in each market at the average of the 3 preceding years.24 In the second set of computations, leading to the estimation of longer-run price effects, the constant-shares norms for two subperiods (1956–59 and 1960–63) were based on the assumption that shares in each market were maintained at the average of the preceding subperiod (1953–55 and 1956–59, respectively). The price variables to which the percentage deviations of actual from constant-shares exports were related were the percentage changes from the preceding period in the weighted average of the exporting countries' price or cost relatives; the weights applied to a country's price relatives in various markets were its exports to those markets, i.e., the importance of each market in its exports to all industrial countries. The equations in which percentage changes from the 3 preceding years were used estimate relatively short-run price effects, namely those that make themselves felt over an average adjustment period of 2 years (from the center year of the 3 preceding years to the current year); the average adjustment period of exports to price changes in the equations using percentage changes from the preceding subperiod is 3½ years for the first observation (1953–55 to 1956–59) and 4 years for the second observation (1956–59 to 1960–63).

The constant-shares norms were computed in three ways. First, they were based, as in all preceding computations, on total exports of manufactures in 1953 prices. In order to ascertain the extent to which changes in commodity composition affect the estimated price responsiveness of total manufacturing exports, constant-shares norms were alternatively computed for each of the three commodity groups (SITC 5, 7, and 6 plus 8). The sum of these commodity-group norms indicates the value of exports that would have been achieved if the country had maintained its previous shares in each commodity class in every market. Since export unit values are not available by commodity class, it was not possible to deflate these alternative constant-shares norms by an appropriate index; for this reason, percentage deviations of actual from constant-shares exports in current prices were used as the dependent variable in the second set of equations. The third set of constant-shares norms, computed to allow comparison of results with and without correction for changes in commodity distribution, was based on total exports in current prices.25

The results, shown in Table 7, indicate a significant responsiveness of deviations of actual from constant-shares exports to changes in unit value relatives for all tested equations. In the computations where constant-shares norms are based on the total volume of exports (“export volume” equations), the estimated short-run price elasticity of exports with respect to changes in unit value relatives is −3.3 for the entire 8-year period 1956–63. The data for this equation and the regression line are shown in Chart 3 (p. 254). The corresponding value for the first half of this period is somewhat higher and that for the latter half somewhat lower, though these variations are not statistically significant. According to these equations, relative price changes explain about one half of the observed variation in the percentage gains or losses of actual, compared with constant-shares, exports; for the second half of the period, the percentage of the total variation explained is somewhat smaller.26

Chart 3.
Chart 3.

Short-Run Price effect: Percentage Deviations of Actual from Constant-Shares Exports and Percentage Changes in Relative Unit Values of Exports, 1956–63 1

Citation: IMF Staff Papers 1965, 002; 10.5089/9781451947212.024.A003

1 Exports of manufactures of each of the 11 countries to the other 10 countries and Switzerland. Percentage changes are taken from the average of the 3 preceding years. Numbers attached to country symbols indicate years (1 = change to 1956, 2 = change to 1957, etc.).

The longer-run elasticity of exports with respect to unit value relatives for the equation covering both subperiods is −5.5; the elasticities estimated separately for each of the two subperiods are almost exactly the same, namely about −5.5. These equations, too, explain about one half of the total observed variation in the dependent variable, except for the second subperiod, where this proportion is only two fifths. The data for the equation covering both subperiods and the regression line are shown in Chart 4 (p. 255).

Chart 4.
Chart 4.

Longer-Run Price Effect: Percentage Deviations of Actual from Constant-Shares Export and Percentage Changes in Relative Unit Values of Exports, Two Subperiods 1

Citation: IMF Staff Papers 1965, 002; 10.5089/9781451947212.024.A003

1 Exports of manufactures of each of the 11 countries to the other 10 countries and Switzerland. Percentage changes are taken from the preceding subperiod. Numbers attached to country symbols indicate subperiods (1 = change from average 1953–55 to average 1956–59; 2 = change from average 1956–59 to average 1960–63).

The computations with wholesale price relatives again support the results obtained with unit value relatives, but the values of r¯2 are considerably smaller and the regression coefficients are in most cases not significant. As in previous computations, the equations with unit labor costs are markedly inferior to those computed with export unit values. But in contrast to the results of the second set of equations, all regression coefficients for the volume equations, including those based on unit labor costs, show the expected negative sign.

Comparison of the equations in which deviations of actual from constant-shares exports are computed by commodity group as well as by market (“export value by SITC groups” equations), and those in which such deviations are derived from total exports by market in current prices (“total export value” equations), indicates the effect of shifts in commodity composition of demand on the elasticity estimates based on unit value relatives. For the three aggregative commodity groups considered, this effect is very small. The regression coefficients of the equations by SITC groups are uniformly somewhat larger in absolute value, and the coefficients of determination (r¯2) are somewhat higher, than in the total value equations. The bias toward smaller unit value elasticities resulting from shifts in commodity composition in the total value equations amounts, however, to no more than 3–5 per cent of the estimated elasticity coefficients.

The relation between the volume of exports, the value of exports, and the unit value index is such that the unit value elasticity of the value of exports should, theoretically, equal that of real exports plus unity, i.e., it should be smaller in absolute value by 1. The elasticities for the total value equations in Table 7 generally fall short of those derived from volume equations by somewhat more than unity. However, this does not necessarily imply the existence of a statistical bias owing to deflation of export data by unit values. It can be explained to a large degree by the fact that a country's unit value relatives vary less than its unit values. The reason is that, in obtaining these relatives, a country's unit values are divided by the average of the unit values of all countries, including the country in question.27

Chart 3 shows a considerable dispersion of the plotted points around the regression line of the volume equation for the years 1956–63. Evidently, nonprice influences account for a large part of the observed variation in export performance. The importance of price effects becomes somewhat clearer if only observations representing relatively large changes in price competitiveness (2 per cent or more) and relatively large deviations from the export norm (5 per cent or more) are considered. Among the 30 observations satisfying both criteria, there is only one instance of a deterioration in price competitiveness accompanied by a rise in exports relative to the norm, and there are two instances in which an improvement in price competitiveness is associated with a decline in exports relative to the norm. For the remaining 27 observations (90 per cent of the total), improvement in competitiveness is associated with export gains when compared with the constant-shares norm (12 observations), and deterioration in competitiveness with export losses (15 observations).

Inspection of Chart 3 also shows that the dispersion of the plotted points around the regression line is not random among countries. The data for some countries, like Germany and Canada, lie generally along a line whose slope is similar to that of the regression line, but some distance to the right or to the left of it. The data for other countries, such as Italy or the United Kingdom, are clustered in one segment of the chart but do not themselves exhibit a clear linear pattern. The slope of the regression line itself, that is to say the estimated elasticity of exports with respect to changes in unit value relatives, is influenced both by the pattern over time of observations pertaining to each country and by the position in the chart of all observations pertaining to one country. Some countries, like Italy and Japan, consistently gained in price competitiveness from year to year throughout almost the entire period and had large increases in exports in excess of gains explained by the growth of demand in the markets which they supply. This influences the estimated elasticity as much as the fact that, for instance, Germany's exports generally exceeded—by an amount which varied closely with its gains or losses in price competitiveness—the exports which it would have had in accordance with the growth of its markets.

In order to separate time-series elements from the cross-section elements discussed in the preceding paragraph, the equation shown in Chart 3 was recomputed in such a way as to remove the influence of the average of those gains and losses in each country's exports which were not due to changes in price competitiveness over time.28 This computation gave, as it were, an “average” time-series price elasticity of exports with respect to changes in unit value relatives for all 11 countries. This elasticity is −2.5.29

This alternative computation amounts to an important confirmation of the results discussed earlier. It shows that the estimated price effects do not depend to any large extent on a coincidence of favorable or unfavorable price movements with other factors which have persistently favored or harmed the export performance of individual countries. The major part of the apparent over-all price effect is due to changes in individual countries' price competitiveness over time.

The data used for the computation of longer-run elasticities of real exports with respect to changes in unit-value relatives, and the regression line of the equation for both subperiods, are plotted in Chart 4. During both periods (1953–55 to 1956–59 and 1956–59 to 1960–63), the increase in exports of Italy and Japan was considerably in excess of the growth of the industrial markets; and the rise in exports of most other countries—particularly Canada, the United Kingdom, and Austria—fell short of the amount which would have resulted from maintenance of shares in the industrial markets.

The regression lines fitted separately to the observations of the first and second subperiods (not shown in the chart) have nearly the same slope (indicating an elasticity of −5.5) as that computed from all observations. These cross-section elasticities for individual subperiods may be compared with results from corresponding computations reported by Fleming and Tsiang;30 for a group of exporting countries similar to the group used in the present study but covering their total exports to all destinations, Fleming and Tsiang obtained, for the period 1948–53, a substitution elasticity of −1.7 for real exports of manufactures with respect to unit value relatives. It is not surprising that price responsiveness appears to have played a less important role during this earlier period, when discriminatory import restrictions, supply limitations, and the postwar readjustment of the economies of the industrial countries determined trade flows to a much greater extent than they did during the sample period of this study. Moreover, the elasticity computed by Fleming and Tsiang may be lower than those found in this study because it was derived from data on exports to composite markets rather than to single-country markets (see the discussion of the effect of geographic aggregation, p. 242).

Conclusion

While there is a clear association between movements in relative prices and export performance, it is apparent that nonprice factors play an important role in the determination of a country's exports. The most prominent of these, the rate of growth of over-all demand in the exporting country's markets, was taken as given in this study; all computations were arranged in such a way as to show relationships between price movements and changes in relative export performance, measured by changes in market shares or in the ratios of a country's exports to those of its competitors. Other nonprice determinants of exports were not taken into account. The degree to which changes in the price indices used in the study fall short of “explaining” the observed variation in relative export performance is an indication both of imperfections in these indices and of the influence of nonprice factors.

Some nonprice influences, as well as price factors, make themselves felt gradually and over a long period of time. As a result, market shares often exhibit rising or falling trends. The computations in this study were designed to avoid uncritical ascription of trends in market shares to trends in relative prices. The influence of trends was either formally eliminated (in the first set of computations) or reduced by the use of percentage changes from the preceding year or from the average of a recent period (in the second and third sets).

Some evidence of the effect of relative prices on export performance emerges from the elasticities of substitution estimated for each exporter in each market (Tables 3 and 4). These estimates were, however, based on a small number of observations, and the results are somewhat erratic.

The estimated relations between price changes and changes in market shares for all exporters, markets, and time periods (Tables 5 and 6) are not subject to the limitations imposed by a small number of observations, but they are limited in another way: they are based on the assumption that the sensitivity of market shares to price changes is uniform among exporting countries or markets, or both. Such estimates can indicate an average range for the measure of the importance of price competitiveness in the explanation of changes in export-market shares of all countries and in all markets; but they cannot be taken to establish the price sensitivity of exports of a particular country or of those of several countries in a particular market.

The association between changes in export unit values and year-to-year changes in market shares found in this second set of computations is not very strong, though the influence of price competitiveness is evident even there. When a somewhat longer adjustment time is allowed by relating changes in annual averages of market shares to those of export unit values over periods about 4 years apart, a considerably stronger price effect is found. The longer-run elasticity of market shares with respect to unit value relatives, computed as an average for all exporting countries and markets, is estimated at −3, implying—subject to the reservations expressed in the preceding paragraph—that a 1 per cent rise (or fall) in export unit values, relative to the average of unit values of all exporting countries, would be expected to result in a 3 per cent reduction (or increase) of a “typical” country's market shares. On the assumption of uniform growth in all markets considered, this would be equivalent to a loss (or gain) of 3 per cent in the volume, and of 2 per cent in the value, of exports compared with the volume or value of exports to these markets which would have been achieved in the absence of relative changes in unit values.

The results of the third set of equations, shown in Table 7 and illustrated in Charts 3 and 4, are subject to the same limitations as the second set with regard to the assumption of uniform price sensitivity among exporting countries and markets. Nevertheless, they constitute the most persuasive confirmation of the importance of price changes in the export performance of industrial countries. About one half of the variation in industrial countries' over-all export performance, measured against a norm implying maintenance of previous market shares, can be attributed to changes in relative export unit values. From these equations it appears that, on average, a 1 per cent improvement (or deterioration) in price competitiveness may, other things being equal, be expected to result in a gain (or loss) in the volume of exports of more than 3 per cent in the shorter run (i.e., over a 2-year adjustment period) and of about 5 per cent in the longer run (i.e., over a 4-year adjustment period); the corresponding proportionate changes in the value of exports would be 2 per cent and 4 per cent, respectively.31

These findings are also supported by the computations of effects on export performance of relative changes in wholesale prices and, to a lesser extent, unit labor costs in the manufacturing sector. But the association of these indicators with export performance was found to be much more tenuous than that between relative changes in exports and in export unit values. It appears that, on the whole, unit value indices are the most useful indicators currently available for the measurement of price competitiveness of manufactures in international trade.

The statistical results presented in this paper lead to the conclusion that the price elasticities of demand for exports of manufactures of individual supplying countries may be rather higher than was previously supposed. An elasticity of −2 is often used as a rule of thumb when it is necessary to make quantitative assumptions about the effects of relative price changes on individual countries' exports. With regard to exports of manufactures by industrial countries to industrial markets, the findings of this study suggest that, in a longer-run context, a value for the price elasticity in the range of −3 to −5 may be a more appropriate assumption.

Broader aspects of competitiveness

This study has been confined to the investigation of one specific aspect of competitiveness in world markets, namely, the influence of prices on countries' shares in export markets.

Maintenance of specific shares is not, of course, by itself of intrinsic merit. If equilibrium in external payments can be achieved in some other way—for example, through a reduction in import requirements or through improvements in the terms of trade or in the balance on capital account—declining shares in export markets can be sustained without necessarily indicating a loss of what may be broadly termed competitiveness.

There are many other aspects of competitiveness which were deliberately excluded from consideration in this study. Since domestic suppliers compete in their home markets with foreign suppliers, changes in the proportion of the home market that is supplied from abroad should be considered in assessing the competitiveness of countries' industrial output in international markets. If competitiveness is defined as forcefulness of domestic producers in seeking out profit opportunities at home and abroad, export losses replaced by direct production in foreign markets, which raises investment income from abroad, may not represent a net loss of competitiveness. On the other hand, lack of investment opportunities at home, which results in stimulating investment abroad, could be considered a loss of competitiveness in a wider sense. In its broadest aspect, consideration of competitiveness would include all factors that determine the utilization and allocation of a country's resources and shape its cost and price structure.

APPENDICES

I. The Equations

The elasticity of substitution, η, between two goods of which quantities x1 and x2 are purchased at prices p1 and p2 is defined as

n = d ( x 1 x 2 ) u ( p 1 p 2 ) . p 1 p 2 x 1 x 2 = d log ( x 1 x 2 ) d log ( p 1 p 2 ) .

In the first set of equations described in the text, exports of manufactures (in constant prices) of country i to market j represent one of the two goods, and exports of all other countries of the sample to the same market represent the second good. The elasticity of substitution, nij, of the exports of country i to market j for the exports of all other countries to that market is computed from an equation of the form

log x ¯ ijt = n ij log p ¯ ijt + T ij t + y ij + u ijt ( 1 ) ( t = 1 , 2 , ….. , 11 ) .

This equation is computed from 11 annual observations for each country, i, and each market, j, provided i≠j so that there are i(j−1) equations; x¯ijt is the ratio of exports, deflated by the unit value index, of country i to market j in year (t+1952) to the deflated exports of all exporting countries of the study—excluding country i (and country j, if it is one of the exporting countries)—to market j in the same year; p¯ijt is one of the three indices of price competitiveness in year (t+ 1952), for instance, the ratio of the export unit value of country i to the weighted average of its competitors' export unit values, the weights being their deflated exports of the preceding year to market j; ηijt, τijt and Yijt are constants estimated by ordinary least-squares regression; uijt is the unexplained residual. The inclusion of the trend term τijt makes the estimated coefficient ηij identical to that which would have been obtained if deviations from linear trend of the variables x¯ijt and p¯ijt had been correlated.

In equation (1), the regression coefficient ηij is an estimate of

∂ log x ¯ ij ∂ log p ¯ ij

over the sample period and thus of the elasticity of substitution of exports of country i to market j for those of all other countries to that market.

Deflation of export values by the unit value index raises the question of statistical bias in equations in which export unit values are also used as independent variables. Random errors in the unit value index would lead to errors of the opposite sign in the deflated export figures. The inverse correspondence of these two sets of errors would result in negative correlation between the two variables even if they were otherwise uncorrelated, and thus in an overstatement of the (absolute value of the) estimated elasticities of substitution. To test for the presence of such bias, the equations with competitiveness indices based on export unit values are also computed with undeflated export ratios as dependent variables. In the absence of bias, the elasticities of substitution should equal those computed with deflated export ratios plus unity, since

d ( x 1 p 1 x 2 p 2 ) d ( p 1 p 2 ) . p 1 p 2 x 1 p 1 x 2 p 2 = d ( x 1 x 2 ) d ( p 1 p 2 ) . p 1 p 2 x 1 x 2 + 1 = n + 1.

In fact, the average difference between the two regression coefficients for the 176 elasticities computed from equation (1) and the alternative equation with undeflated export ratios is 0.97. In most instances (151 coefficients), the difference falls in the range from 0.9 to 1.1. It is therefore concluded that deflation by unit values does not result in significant statistical bias.

In the second set of equations, the annual observations for all 11 exporting countries and the 12 single-country markets are pooled. The equation with all annual observations is of the form

x ijt * − x ij , t − 1 * x ij , t − 1 * = α 1 p ijt * − p ij , t − 1 * p ij , t − 1 * + α 0 + v ijt ( i = 1 , … , 11 ; j = 1 , … , i − 1 , i + 1 , … , 12 ; t = 1953 , … , 1963 ) ( 2 )

where xijt* is the market share of country i in market j (computed in terms of deflated exports); pijt* is one of the three “price relatives,” i.e., the ratio of the price or cost index of country i to the average of those of all countries (including country i) weighted by their deflated exports to market j; α1 and α0 are constants estimated by least-squares regression; and vijt is the unexplained residual. In this version, the coefficient α1 is an estimate of the “short-run” elasticity of market shares with respect to changes in the specified price relative.

The coefficients of equation (2) are also estimated with (xijt*−xij,t−1*)/xij,t−1* and (pijt*−pij,t−1*)/pij,t−1* defined as proportional changes from the preceding subperiod, i.e., from the annual average of 1953–55 to that of 1956–59 and from the annual average of 1956–59 to that of 1960–63. In this version, the coefficient α1 is an estimate of the “longer-run” price elasticity of market shares.

The coefficients in equation (2) are also estimated, in turn, with data for single exporting countries (j = 1, j = 2, etc.), for single markets (j = 1, j = 2, etc.), and for single years or subperiods (t = 1953,t = 1954, etc.; or t = 1956–59 and t = 1960–63).

The typical equation for the third set of computations with 8 annual observations for each of the 11 exporting countries is of the form

( x it − x i , t − θ ) − Σ j [ x ij , t − θ * ( M ij − M j , t − θ ) ] X i , t − θ = β 1 p it * * − p i , t − θ * * p i , t − θ * * + β 0 + w it ( i = 1 , … , 11 ; t = 1 , … , 8 ) ( 3 )

where Xi is total exports of manufactures of country i to all other industrial countries, Mj is total exports of manufactures to country j from all other industrial countries, xij* is the market share of country i in country j, pi** is the weighted average of the price relatives (pij*) of country i in the markets of the industrial countries j (weighted by the exports of country i to each market j), (β1 and (β0 are constants estimated by least-squares regression, and wi is the unexplained residual; θ represents an average lag of 2 years; the respective values are averages of the 3 preceding years, e.g., Xi, t-θ = ⅓ (Xi,t-1 + Xi,t-2 + Xi,t-3). The term (Xit − Xi,t-θ) represents the actual change in the exports of country i, while the term Σj[xij,t−θ*(Mjt−Mj,t−θ)] is the change that would have occurred if country i had maintained its shares of the period referred to by the subscript (t−θ) in all individual markets.

Equation (3) is also computed with t referring to one of the subperiods, 1956–59 (t = 1) or 1960–63 (t = 2), and (t-θ) referring to the preceding subperiod, 1953–55 and 1956–59, respectively. The coefficient β1 is in this case interpreted as a “longer-run” price elasticity.

In addition, the dependent variable in equation (3) is computed from export data in current prices with the “constant-shares” change in exports, Σj[xij,t−θ*(Mjt−Mj,t−θ)], alternatively (a) derived from exports of total manufactures and (b) obtained as the sum of corresponding terms computed for each of three commodity groups (SITC 5, 7, and 6 plus 8).

In the text, it is noted that, in equation (3), the difference between the coefficients β1 alternatively computed with export data in constant and current prices generally exceeds unity. This is not necessarily an indication of statistical bias owing to deflation of export data by export unit values. The expected difference exceeds unity by an amount which depends on the distribution of market shares of the exporting countries included in the calculation. When a country's export unit value changes by 1 per cent, the induced change in its exports in current prices is less, by approximately 1 per cent, than that in exports in constant prices. But, at the same time, the price relative used as the independent variable in the equation changes by less than 1 per cent; this is so because a country's unit value index is expressed as a ratio to the weighted average of unit value indices of all countries including itself. Assume, for instance, that the price elasticity of real exports is −5, that a particular country's unit value declines by 1 per cent, and that the unit values of other countries remain unchanged; if the country's average share in the markets considered is one fifth, its unit value relative will decline by about 0.8 per cent and its exports in constant prices will therefore increase by 4 per cent. Exports in current prices will increase by 3 per cent, giving an apparent elasticity of the value of exports with respect to price relatives of −3.75 (3 divided by −0.8). This apparent elasticity of the value of exports falls short of the assumed elasticity of real exports by 1.25 rather than by 1.00.

Equation (3), in the version with 8 annual observations, is also re-estimated with a set of 10 dummy variables, 1 for each of the 11 countries except the United States. These dummy variables take the value of unity if the observation refers to the country in question, and the value of zero otherwise. The regression coefficients of these dummy variables reflect that part of the average proportionate deviation of a country's actual from constant-shares exports that is not explained by changes in price relatives over time; for the eleventh country, the United States, this factor is expressed in the constant term of the equation. The regression coefficient of the price variable in this equation gives an average elasticity (for all countries) of exports with respect to changes over time in the price relatives.

II. Definitions and Sources of Data

All the data used were defined to refer to manufactured goods or to the manufacturing sector of the economies of the countries included in the study. The basic data are (1) value of exports, (2) unit value of exports, (3) wholesale prices, (4) hourly wage costs, (5) exchange rates, (6) industrial production, (7) employment, and (8) average hours worked.

Value of exports

Exports of manufactures were defined as exports comprised under the UN Standard International Trade Classification (SITC) 5–8, revised. The SITC classification was revised in 1960; data prior to that year were adjusted to conform as much as possible to the new definition. For the United States, “special category” exports were excluded.

Sources: United Nations, Commodity Trade Statistics, Series D; Organization for European Economic Cooperation (OEEC), Statistical Bulletins, Foreign Trade Series IV; Organization for Economic Cooperation and Development (OECD), Statistical Bulletins, Foreign Trade Series C.

Unit value of exports

Unit value indices of exports of manufactures (SITC 5–8) are published by the United Nations for all countries included in this study, except Austria. Unpublished material to carry the available series back to 1953 was made available by the Statistical Office of the United Nations. For Austria, a unit value index of exports of manufactures was constructed on the basis of quantity and value indices for each of the SITC sections for the periods 1953–60 and 1961–63; the two series were linked at the breakpoint (1960–61) on the basis of the Austrian wholesale price index.

Sources: United Nations, Monthly Bulletin of Statistics; OEEC, Statistical Bulletins, Foreign Trade Series IV; OECD, Statistical Bulletins, Foreign Trade Series B; national sources.

Wholesale prices

The wholesale price indices were chosen to conform as closely as possible to the coverage of the trade data. They relate generally to manufactured goods.

Sources: OECD, General Statistics; national sources.

Hourly wage costs

Hourly wage costs were defined as total expenditure on labor per hour in the manufacturing sector, where labor expenditure consists of wage and other direct payments plus legally required and voluntary supplements paid either to the employee or into special employee benefit funds. However, full labor payments could not be obtained in all cases, nor was the coverage the same for all countries. Adjustments were made to make the various national series as consistent as possible.

Sources: OECD, General Statistics; International Labor Office (ILO), International Labor Review; John H. Chandler and Patrick C. Jackman, Unit Labor Costs in Manufacturing: Trends in Eight Countries (U.S. Department of Labor, Bureau of Labor Statistics, unpublished manuscript); national sources.

Exchange rates

Exchange rates were defined as market rates.

Sources: Board of Governors of the Federal Reserve System, Federal Reserve Bulletin and unpublished data; International Monetary Fund, International Financial Statistics.

Industrial production

Industrial production was defined as the output of the manufacturing sector according to the International Standard Industrial Classification published by the United Nations.

Sources: OECD, General Statistics; national sources.

Employment

Employment was defined as employment of wage earners in the manufacturing sector in order to have the series as consistent as possible with the wage cost series. However, an employment series so defined could be obtained for only one country, Italy; data for all other countries refer to employment of wage earners and salaried employees combined.

Sources: OECD, General Statistics; ILO, International Labor Review; national sources.

Average hours worked

Hours worked were defined as hours actually worked or paid per week in manufacturing industries and refer to wage earners only.

Sources: OECD, General Statistics; ILO, International Labor Review; national sources.

Prix et exportations des pays industriels, 1953–63

Résumé

Cette étude essaie de déterminer I’effet de changements de prix ou de coûts, mesurés par trois différents indicateurs, sur les exportations, vers divers marchés, de 11 pays exportateurs importants de produits manufacturés. Les indices de compétitivité des prix sont dérivés pour chaque pays exportateur, dans chaque marché, et on présente le calcul des élasticités de substitution des exportations de chaque pays sur tel ou tel marché particulier par rapport aux exportations de tous les autres pays industriels sur ces marchés. En outre, la réaction de la participation de chaque pays dans les divers marches aux changements de prix ou de coûts relatifs est calculée tant pour l'ensemble des exportateurs et des marchés qui constituent I’échantillon, que pour chaque pays exportateur, chaque marché et chaque année (ou sous-période).

Selon les auteurs, la compétitivité des prix joue un rôle important, mais non prédominant, dans les résultats d'exportation. Mais l'importance relative attribuée aux mouvements des coûts ou des prix comme facteur explicatif des changements des exportations varie considérablement selon les pays fournisseurs et les marchés. Les prix relatifs calculés sur la base des valeurs unitaires des exportations donnent de meilleurs résultats statistiques que ceux qui sont dérivés des prix de gros ou des coûts de la main d'œuvre. Les élasticités de substitution qui ont été calculées, et les élasticités “cross-section” des proportions du marché par rapport aux changements des prix relatifs sont généralement plus élevées en valeur absolue que les élasticités de prix figurant dans les fonctions de demande d'importation dérivées de l'analyse de series chronologiques. Les élasticités à court terme (pour les changements d'une annee a l'autre) se situeraient typiquement dans les limites de -2 à -3, tandis que les élasticités à plus long terme (pour les changements entre périodes de quatre ans) peuvent atteindre prés du double. L'article examine les limitations de la méthode d'analyse et les imperfections des indicateurs de prix utilisés, et donne en outre une indication de I%#x2019;influence qu'ont sur le développement des exportations les différences entre les taux de croissance des divers marchés et des demandes pour diverses catégories de produits.

Los precios y las exportaciones de los países industriales, 1953–63

Resumen

Este artículo procura determinar el efecto que las variaciones de precio o de costo, según medidas por tres indicadores alternatives, ejercen sobre la evolución de las exportaciones en diversos mercados de 11 de los principales países exportadores de productos manufacturados. Deriva índices del poder competitivo en materia de precios para cada pais exportador en cada uno de dichos mercados, y ofrece estimaciones de las elasticidades de substitución que indican el grado en que las exportaciones de todos los demás países industriales son substituibles, en los mismos mercados, por las exportaciones de cada país. Además, se computa para el conjunto de los exportadores y mercados incluidos en la muestra, así como también para cada exportador, mercado y año (o subperíodo), la reacción de las respectivas participaciones de los exportadores en cada mercado a los cambios operados en los precios o costos relativos.

Hállase que el poder competitivo de los precios desempeña un papel importante, mas no preponderante, en la evolución de las exportaciones. Sin embargo, la importancia relativa de los movimientos de los precios o costos, como explicación de los cambios operados en las exportaciones, varía considerablemente según sea cada país abastecedor y cada mercado. Los precios relativos basados en los valores unitarios de las exportaciones rinden mejores resultados estadísticos que los que se basan en los precios al por mayor o en el costo de la mano de obra. Las elasticidades de substitución computadas y las elasticidades “cross-section” de las participaciones respectivas de los exportadores en los mercados frente a los cambios experimentados en los precios relativos, por lo general son más elevadas en su valor absoluto que la elasticidad-precio encontrada en las funciones de la demanda de importaciones basadas en los análisis de series cronológicas. Las elasticidades a corto plazo (de cambios operados de un año a otro) pueden situarse típicamente dentro de los límites de -2 a -3, en tanto que las elasticidades a largo plazo (de cambios operados de un cuatrienio a otro) pueden ser casi el doble. El estudio examina las limitaciones del procedimiento y las deficiencias de los indicadores de precios que han sido utilizados, y proporciona cierta noción de la influencia que el crecimiento diferencial de los mercados y de la demanda por diversas clases de productos ejerce sobre la evolución de las exportaciones.
*

Mrs. Junz is an economist in the Division of International Finance, Board of Governors of the Federal Reserve System. She is a graduate of the University of Amsterdam and of the New School for Social Research, New York. Mr. Rhomberg, Chief of the Special Studies Division of the International Monetary Fund, is a graduate of the University of Vienna and of Yale University, and has been a member of the faculty of the University of Connecticut.

1

For a review of the literature on the technical problems encountered in the statistical estimation of these price and income effects, see S. J. Prais, “Econometric Research in International Trade: A Review,” Kyklos, Vol. XV (1962), pp. 560–79, and D. J. Morgan and W. J. Corlett, “The Influence of Price in International Trade: A Study in Method,” Journal of the Royal Statistical Society (A), Vol. 114 (1951), pp. 307–52.

2

Prominent among a number of studies employing a cross-section approach, or a combination of time-series and cross-section methods, are the following: G. D. A. MacDougall, “British and American Exports: A Study Suggested by the Theory of Comparative Costs,” Economic Journal, Vol. LXI (1951), pp. 697–724, and Vol. LXII (1952), pp. 487–521; J. M. Fleming and S. C. Tsiang, “Changes in Competitive Strength and Export Shares of Major Industrial Countries,” Staff Papers, Vol. V (1956–57), pp. 218–48; R. M. Stern, “British and American Productivity and Comparative Costs in International Trade,” Oxford Economic Papers, Vol. 14, No. 3 (October 1962), pp 275–96; and A. L. Ginsburg and R. M. Stern, “The Determinants of the Factors Affecting American and British Exports in the Interwar and Postwar Periods,” a paper presented at the Annual Meeting of the Econometric Society in Chicago, December 1964.

3

The 11 countries are Austria, Belgium-Luxembourg (hereinafter referred to as Belgium), Canada, France, Federal Republic of Germany, Italy, Japan, Netherlands, Sweden, United Kingdom, and United States. These countries accounted for 72 per cent of world exports of manufactures in 1962.

4

Switzerland was excluded from the group of exporting countries because of lack of certain data.

5

The members of the European Economic Community (Common Market) are Belgium, France, Federal Republic of Germany, Italy, Luxembourg, and Netherlands.

6

The data and data sources are further described in Appendix II (pp. 263–65).

7

Recent studies that make use of market-share analysis as a measure of changes in particular countries' trade positions include Bela Balassa, “Recent Developments in the Competitiveness of American Industry and Prospects for the Future,” Factors Affecting the United States Balance of Payments (U.S. Congress, Joint Economic Committee, Washington, 1962), pp. 29–54; Richard N. Cooper, “The Competitive Position of the United States,” The Dollar in Crisis, Seymour E. Harris, ed. (New York, 1961), pp. 137–64; R. L. Major, “World Trade in Manufactures,” National Institute Economic Review (London), July 1960, pp. 18–27; National Economic Development Council, Export Trends (London, 1963); Anne Romanis, “Relative Growth of Exports of Manufactures of United States and Other Industrial Countries,” Staff Papers, Vol. VIII (1960–61), pp. 241–73; and U.S. Department of Commerce, Bureau of International Commerce, U.S. Share of World Markets for Manufactured Products, Analysis of Changes from 1954 Through 1961 (Washington, 1964).

8

The elasticity of substitution is defined as the proportionate change in the ratio of the quantities of two goods purchased, divided by the proportionate change in the ratio of their prices.

9

See J. J. Polak, “Note on the Measurement of Elasticity of Substitution in International Trade,” Review of Economics and Statistics, Vol. XXXII (1950), pp. 16–20, where it is shown that the standard formula for the elasticity of substitution, on the assumption of unchanged income, does not yield the true elasticity of substitution if income is in fact changing, unless the income elasticities of demand for the two substitutes are equal.

10

The data underlying Chart 1 and corresponding data for the remaining 7 countries of the sample are shown in Table 8 (pp. 266–67).

11

Further evidence supporting this explanation of a divergence of movements of export prices and domestic prices or costs is contained in the following studies: Balassa, op. cit.; E. Benoit, Europe at Sixes and Sevens (New York, 1961), pp. 137 ff.; H. S. Cheng, “Relative Movements in the Prices of Exports of Manufactures,” Staff Papers, Vol. IX (1962), pp. 80–106; Cooper, op. cit.; and Sir Donald MacDougall, The Dollar Problem: A Reappraisal (Essays in International Finance, No. 35, Princeton, N.J., 1960).

12

Inclusion of imported manufactured goods in the commodity samples on which wholesale price indices are based will tend to reduce variations in these indices among industrial countries.

13

Theodore R. Gates, “Production Costs Here and Abroad,” Studies in Business Economics, No. 61 (National Industrial Conference Board, New York, 1958), pp. 32 ff.

14

Since the markets include those of the exporting countries themselves, each exporter sells to only 16 markets.

15

In the absence of separate export unit values by area of destination, the variation in rivals' market shares provides the only differentiation of a country's competitiveness among markets.

16

The data underlying Chart 2 and corresponding data for the remaining 3 countries of the sample are shown in Table 9 (pp. 268–69).

17

Deflation of the export data by export unit values could conceivably result in a negative bias in, and thus an overstatement of, the estimated elasticities when the index of price competitiveness is also based on export unit values. However, a test for the existence of such bias, which is described in Appendix I (pp. 261–62), led to the conclusion that there is in fact no significant bias of this sort in the elasticity estimates presented in the text.

18

The accuracy of the export and unit value data for Canada may have been affected by certain errors of reporting by exporters. The basic data are supposed to be reported in terms of Canadian dollars; it appears, however, that some companies have at times reported their exports in terms of U.S. dollars. The value of such exports will be understated in the official statistics if the Canadian dollar exchanges for less than US$1.00, and overstated if it exchanges for more than US$1.00. Since annual averages of the exchange value of the Canadian dollar varied between US$1,043 and US$0,927 during the period of the study, these data difficulties may have contributed materially to the finding that elasticities of substitution in the single-market computations for Canada were consistently perverse. However, since the extent of the difficulties was not known, no correction could be made, nor was Canada excluded from the study on these grounds.

19

The exact form of these equations is shown in Appendix I (p. 262).

20

In interpreting the coefficients of determination, it must be remembered that the variables are in the form of percentage changes of ratios. Relatively small values for r¯2, compared with values of r¯2 usually found in time-series analysis, were thus to be expected. However, values of 0.10 for r¯2 do not indicate a high degree of correlation by any standard.

21

The positive value relates to Canada as an exporting country; it is conceivable that incorrect data are, in part, responsible for the positive coefficient. As noted in footnote 18, the accuracy of the Canadian export and unit value data may have been affected by certain reporting errors.

22

The other structural factor, geographic differences in market growth, is allowed for in these estimates, since the observations refer to changes in shares in individual markets.

23

Aggregating deviations from constant-share exports to single markets eliminates some of the problems that arise in using market-share analysis in composite markets. For example, it is theoretically possible for a country to have declining market shares in every single market, but at the same time to increase its share in the composite market. This could occur if demand for imports increased more rapidly than the composite market average in those markets in which the exporting country has large shares, and increased more slowly (or declined) in those markets in which its shares are small.

24

Similar computations using shares of the preceding years as norm were carried out in an unpublished analysis by Miss A. Romanis; see also International Monetary Fund, Annual Report of the Executive Directors for the Fiscal Year Ended April 30, 1964 (Washington, 1964), pp. 123–30. Since the present study deals with series covering periods longer than those for the series used in the Annual Report, it was possible to introduce a norm based on average shares in the 3 preceding years. This norm was introduced in order to overcome the difficulty which arises from the fact that an unusual increase in a country's exports in a particular year will produce a large positive deviation from the norm based on shares of the preceding year, while in the following year, although exports return merely to a more normal level, a correspondingly large negative deviation is registered.

25

Equations were computed for all 88 observations (8 years for each of 11 countries), using annual data for the period 1956–63 as well as data for the first and second halves of the period (1956–59 and 1960–63). Similarly, equations based on changes between subperiods were processed for all 22 observations (2 subperiods for each of 11 countries) and for 11 observations for each of the 2 subperiods separately; the last are pure cross-section equations in which all observations refer to the same time period. For a more detailed description of the variables used in the equations reported in this section, see Appendix I (pp. 262–63).

26

Some evidence of a less close association between price relatives and changes in export performance in the last few years of the sample period, compared with the earlier years, was also found in the second set of equations (see Tables 5 and 6). However, the indication of a change in the basic relationship between relative prices and export performance in recent years may appear stronger than actually warranted. Discriminatory tariff changes and a growing flexibility in differential pricing by market—two factors which are not accounted for in the price indicators—probably operated much more strongly in recent years than in the earlier period. In addition, the steady losses in market shares of the United States and the United Kingdom in the period 1953–59, which were paralleled by a steady deterioration in their relative price positions, are often, at least in part, explained by the re-emergence of the ability of war-damaged industrial nations to supply world and home markets with industrial products which immediately after the war could be obtained from only a few countries, chiefly the United States and the United Kingdom. Thus, elasticity estimates for the early period may actually indicate a somewhat stronger sensitivity to price changes than those for more recent years because of factors affecting both ultimate costs to purchasers and market shares, for which no allowance is made in the equations.

27

See Appendix I (p. 263).

28

This was achieved by the use of dummy variables that absorbed special influences, whatever their nature, which made some countries show a persistently better or worse export performance than could be explained by changes in each country's price competitiveness over time. The technique is explained in Appendix I (p. 263).

29

This value is more than 7 times as large as its standard error, and the coefficient of determination is 0.78. This coefficient is substantially higher than that reported for the equation shown in Chart 3, since the average of positive or negative deviations of the observations for each country from the regression line is “explained” by the regression coefficient of the respective dummy variable. The interpretation of the coefficient of determination is that 78 per cent of the total observed variation in the dependent variable is explained by changes in the unit value relatives and the special circumstances of each country which resulted, on average during the period studied, in larger gains or smaller losses, or the reverse, in exports, compared with those which can be explained on the basis of market growth and price competitiveness. The partial coefficient of determination of unit value relatives in the equation with dummy variables is 0.32.

30

J. M. Fleming and S. C. Tsiang, op. cit.

31

The longer-run price elasticities computed from the third set of equations are thus somewhat higher than those implied by the second set. The difference in the value of the elasticities (−5 versus −3), which results largely from the somewhat different orientation of the two sets of computations, is hardly statistically significant. These estimates are, therefore, not mutually inconsistent.

Same Series

Other IMF Content

Other Publishers

  • Collapse
  • Expand
Cover IMF Staff Papers
IMF Staff papers: Volume 12 No. 2
Author:
International Monetary Fund. Research Dept.
Volume/Issue:
Volume 1965: Issue 002
Publisher:
International Monetary Fund
ISBN:
9781451947212
ISSN:
1020-7635
Pages:
70
DOI:
https://doi.org/10.5089/9781451947212.024
  • Chart 1.
    View in gallery
    Chart 1.

    Selected Industrial Countries: Indices (1953 = 100) of Prices, Costs, and Productivity in Manufacturing, 1953–641

  • Chart 2. Chart 2.
    View in gallery
    Chart 2.

    Selected Industrial Countries: Indices (1953 = 100) of Price Competitiveness and Percentage Market Shares for Exports of Manufactures to the Industrial Market, 1953–641

  • Chart 3.
    View in gallery
    Chart 3.

    Short-Run Price effect: Percentage Deviations of Actual from Constant-Shares Exports and Percentage Changes in Relative Unit Values of Exports, 1956–63 1

  • Chart 4.
    View in gallery
    Chart 4.

    Longer-Run Price Effect: Percentage Deviations of Actual from Constant-Shares Export and Percentage Changes in Relative Unit Values of Exports, Two Subperiods 1

Chart 1.

Selected Industrial Countries: Indices (1953 = 100) of Prices, Costs, and Productivity in Manufacturing, 1953–641
Chart 2.

Selected Industrial Countries: Indices (1953 = 100) of Price Competitiveness and Percentage Market Shares for Exports of Manufactures to the Industrial Market, 1953–641
Chart 2.

Selected Industrial Countries: Indices (1953 = 100) of Price Competitiveness and Percentage Market Shares for Exports of Manufactures to the Industrial Market, 1953–641
Chart 3.

Short-Run Price effect: Percentage Deviations of Actual from Constant-Shares Exports and Percentage Changes in Relative Unit Values of Exports, 1956–63 1
Chart 4.

Longer-Run Price Effect: Percentage Deviations of Actual from Constant-Shares Export and Percentage Changes in Relative Unit Values of Exports, Two Subperiods 1