Relative Price Effects on Export Performance: The Case of Nonelectrical Machinary

JACQUES R. ARTUS and SUSANA C. SOSA *

Abstract

JACQUES R. ARTUS and SUSANA C. SOSA *

JACQUES R. ARTUS and SUSANA C. SOSA *

This paper explores the effects of changes in export price competitiveness at a disaggregated level. A case study is considered—namely, the effects of relative price changes among the three main exporters of nonelectrical machinery–the Federal Republic of Germany, the United States, and the United Kingdom–on their relative export performances. All direct and cross-price elasticities of demand in a number of foreign markets are estimated in a consistent fashion. These markets are the European Economic Community (EEC), the rest of the Organization for Economic Cooperation and Development (OECD) countries (RO), and the developing countries (DC).1

The study has two major focuses: (i) it attempts to test the validity of the new “elasticity pessimism” that has arisen following the apparent failure of the large exchange rate changes and relative price changes of 1969–73 to affect trade balances significantly; 2 and (ii) it attempts to eliminate certain methodological flaws that reduce the usefulness of many existing empirical studies on foreign trade price elasticities. Concern with the methodological aspects has led to the present consideration of relative price effects in the world market for a specific group of products rather than to a broad study of countries’ total import and export flows. This choice, of course, limits the implications that can be drawn from the study as to the validity of the new elasticity pessimism; nevertheless, the results are relevant to this issue because of the importance of nonelectrical machinery in world trade 3 and because of the large relative price changes experienced by the three dominant suppliers of these products in recent years.4

On a methodological level, the present study incorporates several characteristics that are not often combined in empirical studies on foreign trade price elasticities:5

(1) The international demand functions are explicitly derived from a specific theory of demand, namely, the theory of demand for products distinguished by place of production developed by Armington (1969). This approach has the advantage of making clear what kinds of structural parameters are being estimated and what assumptions and constraints underlie the empirical model.

(2) Aggregation biases that plague most empirical studies have been limited through the choice of a specific group of products, the individual consideration of various foreign markets, and the recognition that, even at this level of disaggregation, countries’ exports are differentiated as far as product mixes are concerned. 6 This differentiation is taken into account in deriving the weighting schemes used in the calculation of the relative price terms.

(3) Lags between changes in relative prices and their effects on market shares are taken into account without the imposition of either uniform lag patterns on all of the variables in the equation, or of continuously declining weights. Account is also taken of the fact that unit values reflect prices negotiated at the time goods are contracted for, rather than current prices.

(4) Further, a country’s competitiveness depends not only on the prices it is able to offer but, to a large extent, also on nonprice factors such as delivery time, credit terms, reliability of the goods, and existence of technological dependence. In the present study, such factors are proxied by variables measuring relative delivery delays and by time trends. The delivery delay variable picks up both the effects of additional costs to buyers arising from delayed deliveries and the effects of other nonprice rationing variables that move together with delivery delays over the cycle.7

(5) Unit value indices are used only for the United Kingdom. Price series for the Federal Republic of Germany and for the United States are based on more reliable contract price series.8 There is no doubt, however, that the likelihood of significant observation errors in the price indices remains a serious potential source of bias.

(6) An attempt is made to take into account the elasticity of supply for exports and to reduce simultaneous-equation bias in the estimation of the price elasticities of demand.

The first section of the paper discusses the theoretical framework and the specification of the export equations. The second section presents and discusses the results of the estimation. Some conclusions suggested by the results follow.

I. Theoretical Approach

The Federal Republic of Germany, the United States, and the United Kingdom do not export exactly the same commodity mix of nonelectrical machinery. Germany tends to specialize in metalworking machinery, while the United States specializes to some extent in office machinery, and the United Kingdom specializes in power generating machinery. Even for a given kind of nonelectrical machinery, the products offered by the three countries may still differ as to technical specification, quality, reliability of after-sale service, complementarity with existing equipment, etc. A study of relative price effects on exports of the three countries must be based on a theoretical model that takes into account these differences in commodity mix and in product characteristics.

The approach followed here is based on the theoretical framework developed by Armington (1969) for products distinguished by place of production. Let us consider a given market and a specific kind of product, say lathes. Armington shows that, under certain assumptions, the demand Dj for lathes supplied by country j (in real terms) will depend only on the prices Pl of lathes supplied by all competing countries l (with l = l, …, j, …, n), and the total demand D for lathes (in real terms). 9 In mathematical notation, the demand equation can be expressed:

ln(Dj) = bj + ln(D) + Σlη¯j/lln(Pl)(1)

where ηj/l = income-compensated price elasticity (in real terms) between lathes supplied by country j and lathes supplied by country l.

equation (1) is based on two assumptions. First, the marginal rate of substitution between the lathes supplied by any two countries must be independent of the quantities demanded of other kinds of products, for example, other kinds of machinery, foodstuffs, or chemicals.10 Second, the utility functions governing choices among lathes supplied by the various countries must be linear and homogeneous. These two assumptions are necessary if the choice between lathes is to be isolated from consideration of the demand for products of other kinds.

Up to this point, only fairly mild assumptions have been made. The number of parameters remains large for econometric work, however, given the size of available observation samples. The next step in obtaining a model suitable for empirical work is to adopt the assumption that the elasticity of substitution is the same between any pair of suppliers. This latter assumption is likely to be more restrictive than the others; nevertheless, for a given market and a specific kind of product, the assumption still seems to be plausible. Under this latter assumption, the income-compensated price elasticities ηj/l can be expressed in terms of the substitution elasticity and the market shares, namely, 11

η¯j/l=slσ(2)

and

η¯j/j=(1sj)σ(3)

where the Hicks-Allen elasticity of substitution σ is positive or zero, and sl is the share of country l in the market considered.

In the present case of three competing countries, the constraints represented by equations (2) and (3) allow the nine price elasticities ηj/l to be derived from one substitution elasticity σ and the market shares sl.

Substituting equations (2) and (3) into equation (1) yields

ln(Dj/D)=bj+Σljσslln(Pl/Pj)(4)

The theoretical framework presented above relies on the assumption that the products considered—in this instance, lathes—all have the same end use; they are distinguishable only on the basis of the country of their production. For example, lathes made in the Federal Republic of Germany and the United States are assumed to be designed for the same tasks; however, they may still differ as to quality, reliability of after-sale service, etc. In practice, it is difficult to define commodity groups that are sufficiently homogeneous to satisfy this assumption. Even “lathes” may not constitute a sufficiently well-defined kind of product because of the need to distinguish lathes used for working metals from lathes used for working wood, and numerically-controlled lathes from other lathes. At the same time, the kind of disaggregation necessary to have “lathes” as a separate group is already too fine for empirical study, given the limitations on the availability of foreign trade prices.

The compromise solution retained in the present study is to use equation (4) for a larger commodity group—nonelectrical machinery, taken as a whole—but to redefine the trade weights sl to take into account different degrees of competitiveness between exporting countries owing to differences in the product composition of their exports of nonelectrical machinery to the market considered. More specifically, the nonelectrical machinery classification was disaggregated into 27 subgroups for purposes of calculating trade weights, 12 and the market share sl in equation (4) was replaced by a weighted share slj that is specific to the exporting country and is calculated as 13

slj=Σi(D¯ij/D¯j)(D¯il/ΣlD¯il)(5)

where the bar above a variable indicates that the base-period value of this variable is considered, and the subscript i refers to a particular subgroup of nonelectrical machinery.

The weighting scheme represented by equation (5) assumes that the price elasticity ηj/l of, say, German nonelectrical machinery with respect to the price of British nonelectrical machinery in a given market will be influenced by the British market shares for the various subgroups, weighted by the relative importance of these subgroups in the Federal Republic of Germany’s exports to the market.

In equation (4), prices are the only variables that influence the demand allocation between products supplied by the various countries. Various nonprice factors, such as delivery time, credit terms, reliability of the goods, and the existence of technological dependence, are viewed as fixed characteristics, which are reflected by the constant term, the trade shares, and the substitution elasticity. However, nonprice factors do change over time, and this always causes major difficulties for empirical work on price elasticities. The present analysis will follow the example of previous studies in this field by assuming that most of the effects of changes in the nonprice factors can be picked up by including two additional variables in the demand equations: (i) delivery delays for orders placed to supplier l, WTl, as a proxy for all “cyclical” elements; and (ii) a time trend t as a proxy for all “noncyclical” elements. No attempt will be made to identify effects of the nonprice factors on the values of the price elasticities.

Allowance will also be made for the existence of fixed adjustment and information costs by introducing lags in the price and waiting-time effects. Thus, the demand equation used in the empirical analysis is

ln(Dj/D)=bj+cjt+σ[Σljsljln(Pl/Pj)]L+δ[Σljsljln(WTl/WTj)]L+(6)

where l denotes a lag operator and e denotes a randomly distributed error term.

To avoid the simultaneous-equation bias that may result from not taking into account the supply response to price variations, the parameters of equation (6) are estimated in a simultaneous model composed of equation (6) and a price equation.14 The price function refers to the total exports of nonelectrical machinery by country j. It is postulated as a log-linear function of the total volume of exports, normal unit costs, and competitors’ prices, namely,

ln(pj)=dj+fjln(Xj)L+gjln(PWj)L+hjln(NUCj)L+u(7)

where

article image

The normal unit cost variable is a weighted average of the unit labor cost in manufacturing, adjusted for cyclical variations in output-perman-hour, and of the costs of raw materials, fuels, and services used as inputs by the manufacturing sector.

II. Empirical Findings

The model represented by equations (6) and (7) was employed to study relative price effects on exports of nonelectrical machinery by the Federal Republic of Germany, the United States, and the United Kingdom to three separate markets—the European Economic Community (EEC), the “rest of the OECD countries” (RO), and the developing countries (DC). 15 The three markets considered have been selected on the basis of an a priori belief that price elasticities in these markets may differ because of differences in economic conditions and import tariffs. Only three exporting countries are included in the study because of a lack of data for other countries; the fact that these three countries represent the bulk of world trade in nonelectrical machinery, however, limits the size of the biases that are thereby introduced in the analysis.16 This specification also abstracts from production of nonelectrical machinery in the markets considered. The sum of the exports of these three countries to a given market is, thus, assumed to represent the total demand for nonelectrical machinery—that is, the D variable.

data and functional form

Relative price variables in the demand equations were calculated on the basis of contract prices. Contract price indices were used for the Federal Republic of Germany and for the United States. A contract price index is not available for the United Kingdom, and it had to be proxied by a unit value index. This latter index was introduced with a lead of three quarters. U. K. delivery delays average about three quarters, so that contract prices should, on average, correspond to unit values observed three quarters later.17 Export prices of each country were assumed to be the same in all of its foreign markets. This assumption that countries do not practice price discrimination between their various foreign markets is a restrictive assumption; it is, however, unavoidable, given the lack of export price series disaggregated by foreign markets. 18

Two lagged price variables were used in the estimation of the demand equations. The variable used to measure the very-short-term adjustment of the volume of trade to changes in prices is the simple average of relative prices over the four preceding quarters—that is, t-1 through t-4. The second variable, which reflects a longer-term response of the volume of trade to price changes, is a simple average over periods t-5 through t-12.19 Thus, the model allows for the existence of delivery delays and for the possibility that buyers of the good may not react rapidly to changes in relative prices because of recognition and decision lags.

The delivery delays considered here are the actual delivery delays for foreign orders rather than quoted delivery delays. The assumption is either that quoted delivery delays reflect actual delivery delays, or that customers are good at predicting actual delivery delays. The actual delivery delays in a given quarter are taken to be equal to the delivery delay for the first order received during the quarter. The delivery delay WT in quarter t is defined as the number of quarters, 20 subsequent to quarter t-1, that are required to work off an amount of foreign orders equal to the stock of unfilled foreign orders at the end of quarter t-1. This definition rests on the assumption that new orders and deliveries relevant to each quarter are evenly distributed during that quarter. For the United States, series on total (home plus foreign) unfilled orders and shipments were used because separate series on foreign orders were not available for the period considered here. 21 Only one lagged delivery delay variable was used in the equation—a simple average of relative delivery delays over the four preceding quarters. Variables with longer lags were found to have a statistically insignificant coefficient with the “wrong” sign.

Various lag patterns were also introduced in the price equation (7). The dynamic adjustment of the export price to a change in the volume of exports may be influenced by two different effects—namely, (i) decision lags and the desire to minimize price variations may slow down the price increase that would be induced by an increase in demand and in exports, and (ii) the shift of resources from or to the export sector is likely to be only a slow process, so that an increase in demand could lead to a price increase that was large initially and then was reduced gradually when the shift of resources took place. The lag coefficients on the volume of exports could thus be positive and gradually increasing for the first few lag periods, and could then decrease and take negative values for a number of subsequent periods. The adjustment of the export price to variations in domestic costs and competitors’ prices may also follow a complicated path. For all three explanatory variables, the form of the distributed lag was, therefore, left unconstrained.

The coverage and particular statistical properties of the series employed in this study are examined in the Appendix.

empirical results

Parameters of equations (6) and (7) were estimated, for the three exporting countries and the three markets considered, from quarterly data by multiple regression methods using two-stage least squares. The observation period for the dependent variables Dj/D and Pj extends from the second quarter of 1962 to the third quarter of 1974. To obtain estimates of the substitution elasticity σ and the delivery delays parameter δ that would satisfy all three demand equations for a particular market simultaneously, stacked time series corresponding to the three exporting countries were used in the estimation.22

The results of the statistical estimation of the demand equations are presented in Table 1. For each of the three markets considered in this study, the structural equation (6) explains variations in market shares occurring over time with a reasonably high degree of precision; the standard errors of estimate range from 2.0 per cent in the EEC and the RO markets to 2.7 per cent in the DC market. All but one of the signs of the estimated coefficients are consistent with a priori expectations. The exception is the coefficient on the delivery delays variable for exports to the RO market, which is found to be negative, but small and not statistically significant. The waiting-time variable was subsequently deleted from the equation for this market. Increases in prices are always found to have a negative effect on the volume of exports.

Table 1.

Demand Equations: Constrained Regression Coefficients from Pooled Time Series1

ln(Dj/D) = bj + cjt + σ[Σljsljln(Pl/Pj)]L + δ[Σljsljln(WTl/WTj)]L + ϵ2(6)
article image

R2 denotes the coefficient of multiple correlation adjusted for degrees of freedom; S.E. denotes the standard error of estimate; and D-W denotes the Durbin-Watson statistic. The asterisk (*) indicates that an estimate is significant at the 5 per cent significance level. The notation σ1→4, for example, refers to the elasticity with respect to the period t-l through t-4.

Seasonal dummies have also been included in the regression equation.

Excluding the Federal Republic of Germany and the United Kingdom.

Excluding the Federal Republic of Germany, the United Kingdom, and the United States.

The Durbin-Watson statistic has no precise meaning in regression equations using pooled time series.

The estimated elasticities of substitution have not only the right sign but also have relatively small standard errors, at least as far as the long-run total price effect is concerned. The long-run estimated elasticities are as follows:23

article image

It is somewhat surprising that in the RO market the elasticity of substitution is found to be relatively small. The importance of tied foreign aid in the financing of imports in developing countries would have suggested that relative price effects were less important in that market. The results for the three markets are consistent, on the other hand, in indicating that price changes affect export volumes only with a considerable lag. In all three equations, the second relative price term, which corresponds to a lag of two to three years, was found to have a much larger coefficient than the first relative price term, the coefficient of which measures shorter-term effects.

The size of the estimated substitution elasticities should not mislead us. Once the substitution elasticities are translated into income-compensated direct and cross-price elasticities of demand by using equations (2) and (3), relative price effects are found to be small although not negligible. The complete matrices of derived direct and cross-price elasticities are presented in Table 2. The weighted average elasticities for the three markets are of particular interest because their diagonal elements indicate the effect of a 1 per cent increase in the export price of a country on the total volume of its exports to the three markets considered in this study—that is, they correspond to the traditional export price elasticity concept. For the first year following a price change, the export price elasticities are found to vary from –0.06 (0.04) for the United States to –0.09 (0.06) for the United Kingdom. 24 In the long run, the export price elasticity is found to be only –0.66 (0.04) for the Federal Republic of Germany, –1.043 (0.07) for the United Kingdom, and –0.64 (0.04) for the United States. The elasticities are either smaller than, or not much larger than, one, and indicate that the value of a country’s exports would not increase and might, in fact, decrease as a result of a unilateral fall in its export prices.

Table 2.

Federal Republic of Germany, United Kingdom, and United States: Derived Direct and Cross-Price Elasticities of Demand for Their Exports of Nonelectrical Machinery in Third Markets

article image

Excluding the Federal Republic of Germany and the United Kingdom.

Parentheses enclose standard errors of estimate.

Excluding the Federal Republic of Germany, the United Kingdom, and the United States.

The model represented by equation (6) is based on the assumption that the elasticity of substitution in a given market is the same for any pair of supplying countries. This is a restrictive assumption, the validity of which needs to be tested on available data. To test this assumption, the parameters of the more general equation

ln(Dj/D)=bj+cjt+η¯j/lln(Pl/Pj1)L+δj[Σljsljln(WTl/WTj)]L+(8)

were estimated for each exporting country with respect to the three separate markets considered. In equation (8), the only constraint imposed on the price elasticities is the homogeneity assumption.

The results are presented in Table 3. The small number of observations (50) and the large number of explanatory variables (9, including the 3 seasonal dummy variables) in each equation lead to inefficient estimates. The comparison between the two sets of estimates of the long-run elasticities for the three markets combined (presented in Table 4) is, however, interesting. In three cases, the unconstrained estimates are not consistent statistically with the constrained estimates presented in Table 4. The major difference between the two sets of estimates is that the long-run cross-price elasticities between, and the total export price elasticities for, U. K. and U. S. products are higher in the unconstrained estimation. The export price elasticities for U. K. and U. S. products are, respectively, –2.13 (0.39) rather than –1.04 (0.07), and –1.06 (0.20) rather than –0.64 (0.04). The implication of these results is that, even after some adjustments are made for differences in market and product specialization, the substitution elasticities between German (Fed. Rep.) and U. S. products, and between German (Fed. Rep.) and U. K. products, seem to be markedly less than the one between U. K. and U. S. products.

Table 3.

Demand Equations: Unconstrained Regression Coefficients1

ln(Dj/D)=bj+cjt+η¯j/lln(Pl/Pj)L+δj[Σljsljln(WTl/WTj)]L+2(8)
article image

The asterisk (*) indicates that an estimate is significant at the 5 per cent significance level. Parentheses enclose standard errors of estimate. The empty cells indicate parameters that were not estimated directly in the regression equation.

Seasonal dummies have been included in the regression equation; their estimated coefficients are not reported here.

R2 denotes the coefficient of multiple correlation adjusted for degrees of freedom; S.E. denotes the standard error of estimate; and D-W denotes the Durbin-Watson statistic.

Table 4.

Federal Republic of Germany, United Kingdom, and United States: Constrained and Unconstrained Estimates of the Long-Run Direct and Cross-Price Elasticities of Demand for Their Total Exports of Nonelectrical Machinery to the Three Markets Considered1

article image

Parentheses enclose standard errors of estimate. The check mark (✓) indicates that an unconstrained estimate is significantly different from the corresponding constrained estimate at the 5 per cent significance level.

The regression results for the supply equation are presented in Table 5. For the three exporting countries, equation (7) explains a high percentage of variations in export prices, ranging from 85 per cent for the United Kingdom to 99 per cent for the Federal Republic of Germany. Also, the standard errors of the estimates are low—0.4 per cent for the United States and the Federal Republic of Germany and 0.8 per cent for the United Kingdom.

Table 5.

Supply Equations: Regression Coefficients1

ln(Pj)=dj+fjln(Xj)L+gjln(PWj)L+hjln(NUCj)L+u2(7)
article image

R2 denotes the coefficient of multiple correlation, adjusted for degrees of freedom; S.E. denotes the standard error of estimate; and D-W denotes the Durbin-Watson statistic. The asterisk (*) indicates that an estimate is significant at the 5 per cent significance level. The notation σ1→4, for example, refers to the elasticity with respect to the period t-1 through t-4.

Seasonal dummies have also been included in the regression equation.

The volume-of-exports variable was not lagged for this exporting country.

The Durbin-Watson statistic has no precise meaning in regression equations using pooled time series.

Various lag patterns were tried for the volume of exports, competitors’ prices, and unit cost variables. No significant lagged effects, however, were found for the latter two variables. For the volume of exports, a simple lag of two periods was selected for the Federal Republic of Germany and the United States, and no lag was selected for the United Kingdom, since these lags gave the lowest standard errors of estimate. These results are, to some extent, disappointing; they point out the difficulty involved in modeling the supply side in foreign trade studies.

The coefficients on the volume of exports have the expected positive sign. They are also significantly different from zero at the 95 per cent level of significance for the Federal Republic of Germany and the United States. The implicit price elasticities of supply are shown in the following table:

article image

This is calculated as the inverse of the coefficient of the export volume variable in the price equation (7).

The estimated supply elasticity for the Federal Republic of Germany is similar to the one obtained by Goldstein and Khan (1976) for total exports. For the United Kingdom and the United States, however, the present estimates are quite different from the results obtained by those authors for total exports—1.4 for the United Kingdom and 6.6 for the United States.

Estimation of the demand equation (6) and the supply equation (7) by multiple regression methods using simple least squares yielded results rather similar to those reported previously. Thus the simultaneousequation bias that would be introduced by estimating the price elasticities of demand from equation (6) without taking into account the supply response would be small, at least in the present case.

III. Conclusion

Empirical observations, including observations for the period 1967–74, which was characterized by large changes in exchange rates and relative prices, support the view that, at least in one important market, the market for nonelectrical machinery, changes in price competitiveness have had a significant effect on the export volume of the various exporting countries. Price effects are, however, not extremely large and are felt rather slowly. When the substitution elasticity is assumed to be the same between any pair of supplying countries, the estimated long-run (3 years) substitution elasticities are 1.75 (0.17) in the EEC market, 0.90 (0.17) in the RO market, and 1.69 (0.24) in the DC market. The corresponding long-run export price elasticities are –0.66 (0.04) for the Federal Republic of Germany, –0.64 (0.04) for the United States, and –1.04 (0.07) for the United Kingdom. Short-run export price elasticities (1 year) are quite low; they range from –0.06 (0.04) for the United States to –0.09 (0.06) for the United Kingdom.

Some evidence suggests that the substitution elasticities between German and U.K. products, and between German and U.S. products, may be lower than the elasticity between U. K. and U. S. products. When the assumption that the substitution elasticity is the same between all pairs of supplying countries is dropped, the long-run export price elasticity estimates are –0.50 (0.12) for the Federal Republic of Germany, –2.13 (0.39) for the United Kingdom, and –1.06 (0.20) for the United States. The relatively low price elasticity obtained for German exports tends to substantiate the widespread belief that nonprice factors play an important role in determining demand for German products.25

The approach followed here is viewed as an improvement over that used in previous empirical work. It is not, however, without weaknesses, namely, that (i) the price data remain somewhat unreliable; 26(ii) important nonprice factors are not taken into account or are taken into account imperfectly by using proxy variables; and (iii) only the three major supplying countries are considered. It is felt that, on balance, these weaknesses may have led to a downward bias in the estimates of the price elasticities. There is also a need for studies on other industries before broader-based and more reliable conclusions can be derived as to the role of relative prices in international trade.

APPENDIX

Commodity and geographical coverage

The commodity group “nonelectrical machinery” consists of all goods included in Standard International Trade Classification (SITC) 71–that is, power generating machinery and implements, office machines, metal working machinery, textile and leather machinery, machines for special industries, and machinery and appliances (other than electrical) and machine parts not elsewhere specified.

The total volume of exports to each market is defined as the sum of the volume of exports of the Federal Republic of Germany, the United Kingdom, and the United States to the market considered. The markets are defined to exclude these three exporting countries. They represent three major economic regions: (1) the founding members of the European Economic Community (EEC): Belgium-Luxembourg, France, Italy, and the Netherlands; (2) the “rest of the OECD countries” (RO), consisting of the European Free Trade Area (EFTA) countries as of November 1972: Austria, Denmark, Finland 27, Iceland, Ireland, Norway, Portugal, Sweden, and Switzerland; all other OECD countries not included in group (1)—namely, Greece, Spain, Turkey, Canada, Japan, Australia, and New Zealand; plus South Africa (which is not an OECD member); (3) the developing countries (DC): Yugoslavia and the rest of Europe not included elsewhere, all Africa (except South Africa), all Latin America, all Asia (except Japan, the People’s Republic of China, North Korea, North Vietnam, and Mongolia).

Market shares

Values of the trade flows in current U. S. dollars were obtained from the OECD, Statistics of Foreign Trade, Series B, Trade by Commodities: Country Summaries. The deflators used to obtain series on the volume of trade flows were the unit value index for the United Kingdom, and the export contract price indices for the Federal Republic of Germany and the United States. The export contract price series were lagged by two quarters to obtain series more closely resembling unit value series. This lag is equal to the average estimated waiting time of orders placed in Germany and the United States. Sources of the price series are indicated below.

Series on the export volume of the United Kingdom and the United States were further adjusted for the effects of strikes and other disturbances. The adjustment to the value of British exports was estimated by using the adjustment factors for total exports published in the National Institute Economic Review by the National Institute of Economic and Social Research (U. K.). For the United States, the factor of adjustment is the ratio of actual to normal trade volumes, as calculated by Peter Isard (1975). For both countries, the adjustment factors were assumed to be the same in all three markets considered.

Prices

Export contract price indices form the basis for the relative price terms used as explanatory variables. The Federal Republic of Germany’s export contract price indices were obtained from the Federal Statistical Office in Wiesbaden in Fach serie G, Reihe 7. U. K. price series consist of export unit value series in national currency, adjusted for exchange rate changes and led two quarters to take into account average delivery delays. Annual data for 1958–60 and quarterly data since 1961 were provided directly by the U. K. Department of Trade and Industry, Statistical Division. Quarterly data for 1958–60 were obtained by linear interpolation.

U.S. export prices were obtained by splicing several series corresponding to different subperiods and subgroups. For the period 1953–64, annual price indices (June 1962 = 100) calculated by Kravis and Lipsey (1971) for seven three-digit SITCs (711, 712, 714, 715, 717, 718, and 719) were used. The weights used to aggregate these price series were the relative values of each of these subgroups in total exports for these groups in 1963. The values for the years 1958-60, for which Kravis and Lipsey give no data, were estimated by linear interpolation. Price series for the period 1964–74 were estimated on the basis of annual price indices (June 1967 = 100), based on June figures, for the period 1964–73, and on quarterly indices for 1974 (published in News by the U.S. Department of Labor, Bureau of Labor Statistics). The number of four- and five-digit SITCs for which data were available increased during the period. Therefore, the aggregation of the subgroup indices was performed for five subperiods covering the years 1964-74. During each subperiod, the number of SITC subgroups remained unchanged. The weights used were derived from the structure of the value of exports of the SITC 71 group in 1969. All the subseries were spliced. Then a quarterly series was derived by applying the interquarter rates of increase of the quarterly series of unit values for finished manufactures (1967=100), which was available for the whole 1964–74 period, to the values of the spliced annual series. The unit value index is published in Survey of Current Business, U. S. Department of Commerce, Bureau of Economic Analysis. Series on actual or proxy export price series are presented in Table 6.

Table 6.

Federal Republic of Germany, United Kingdom, and United States: Data on Contract Prices and Delivery Delays for Exports of Nonelectrical Machinery, Fourth Quarter 1960-Fourth Quarter 1974

article image

This series is derived by taking a series of export unit values in local currency, giving it a two-quarter lead, and then converting the series to U. S. dollars.

Delivery delays

The waiting-time series for orders placed with firms in the Federal Republic of Germany were estimated from monthly series on the value of export deliveries, and the value of new orders received from abroad. These data, produced by the Federal Statistical Office in Wiesbaden, were provided directly by the Deutsche Bundesbank. The series for orders on hand was derived from the previous two series and a benchmark estimate of the value of unfilled orders. This benchmark value was obtained by assuming that the value of export deliveries in the last two quarters of 1957 was equivalent to the value of unfilled orders at the end of the second quarter of 1957–postulating, therefore, a waiting time of two quarters when the sales trend was at a relative low level. The waiting-time series for the United Kingdom were derived from quarterly series on volume index numbers of export orders on hand and of export sales, which are published by the U. K. Central Statistical Office in its Monthly Digest of Statistics. For the United States, the quarterly series on the value of total (home plus abroad) new orders, unfilled orders, and shipments were obtained from the Survey of Current Business. It was assumed that the fluctuations of the delivery delay series for total orders reflected fluctuations in delays for foreign orders. Series on delivery delays are presented in Table 6.

Normal unit costs

The unit cost series are an aggregate of input price series and of normal unit labor cost indices for manufactures. The input price series are calculated from data on basic materials and fuel prices, and from yearly indices of implicit deflators of the gross national product originating in the following service sectors: transportation; communications; electricity, gas, and sanitary services; wholesale and retail trade; and banking, insurance and real estate. The source for these national accounts data is OECD, National Accounts of OECD Countries. The sources for basic materials and fuel prices are: Federal Republic of Germany, Wirtschaft und Statistik (published by the Federal Statistical Office in Wiesbaden); United Kingdom, Monthly Digest of Statistics; and United States, Survey of Current Business. The quarterly indices of normal unit labor cost in local currency (1970 = 100) were calculated by the Current Studies Division of the International Monetary Fund’s Research Department.

The weights used to calculate the aggregated input price series and the unit labor cost indices correspond to the shares of the basic materials, fuels, and services components of the value added in the manufacturing sector, as reported in national input-output tables.

REFERENCES

  • “Apel Answers Back,” The Banker, Vol. 127 (April 1977), pp. 3941.

  • Armington, Paul S., “A Theory of Demand for Products Distinguished by Place of Production,” Staff Papers, Vol. 16 (March 1969), pp. 15978.

  • Artus, Jacques R. (1970), “The Short-Run Effects of Domestic Demand Pressure on British Export Performance,” Staff Papers, Vol. 17 (July 1970), pp. 24776.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Artus, Jacques R. (1973), “The Short-Run Effects of Domestic Demand Pressure on Export Delivery Delays for Machinery,” Journal of International Economics, Vol. 3 (February 1973), pp. 2136.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Artus, Jacques R.(1974), “The Behavior of Export Prices for Manufactures,” Staff Papers, Vol. 21 (November 1974), pp. 583604.

  • Ball, R.J., and others, “The Relationship Between United Kingdom Export Performance in Manufactures and the Internal Pressure of Demand,” Economic Journal, Vol. 76 (September 1966), pp. 501518.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Barker, Terence S., “Aggregation Error and Estimates of the U. K. Import Demand Function,” in The Econometric Study of the United Kingdom, ed. by Kenneth, Hilton and David F. Heathfield(London, 1970), pp. 11545.

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, and Mohsin S. Khan, “The Supply and Demand for Exports: A Simultaneous Approach” (unpublished, International Monetary Fund, March 23, 1976), forthcoming in the May 1978 issue of the Review of Economics and Statistics.

    • Search Google Scholar
    • Export Citation
  • Hanoch, Giora, “CRESH Production Functions,” Econometrica, Vol. 39 (September 1971), pp. 695712.

  • Isard, Peter, “Dock Strike Adjustment Factors for Major Categories of U. S. Imports and Exports, 1958–1974,” International Finance Discussion Paper No. 60, Board of Governors of the Federal Reserve System (February 7, 1975).

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Kravis, Irving B., and Robert E. Lipsey, Price Competitiveness in World Trade, National Bureau of Economic Research (Columbia University Press, 1971).

    • Search Google Scholar
    • Export Citation
  • Leamer, Edward E., and Robert M. Stern, Quantitative International Economics (Boston, 1970).

  • Magee, Stephen P., “Prices, Incomes, and Foreign Trade,” in International Trade and Finance: Frontiers for Research, ed. by Peter B., Kenen, (Cambridge University Press, 1975), pp. 175252.

    • Search Google Scholar
    • Export Citation
  • Stern, Robert M., and others, Price Elasticities in International Trade: An Annotated Bibliography (London, 1976).

  • Steuer, M.D., and others, “The Effect of Waiting Times on Foreign Orders for Machine Tools,” Economica, Vol. 33 (November 1966), pp. 387403.

  • White, William H., “Bias in Export Substitution Elasticities Derived Through Use of Cross-Section (Sub-Market) Data” (unpublished, International Monetary Fund, March 4, 1970).

    • Search Google Scholar
    • Export Citation
  • Whitman, Marina v. N., “The Payments Adjustment Process and the Exchange Rate Regime: What Have We Learned?” American Economic Review—Papers and Proceedings, Vol. 65 (May 1975), pp. 13346.

    • Search Google Scholar
    • Export Citation
*

Mr. Artus, a staff member of the External Adjustment Division of the Research Department, holds degrees from the Faculty of Law and Economics in Paris and from the University of California at Berkeley.

Mrs. Sosa, a staff member in the External Adjustment Division, is a graduate of the Universidad de la República, Uruguay and received her masters degree from American University. She has earned certificates of graduate studies from the University of Chile and from the Latin American Institute for Economic and Social Planning in Santiago.

1

The study is concerned only with relative price effects in third markets, so that the EEC market is defined as excluding the Federal Republic of Germany and the United Kingdom, while the market formed by the “rest of the OECD countries” excludes the United States, the United Kingdom, and the Federal Republic of Germany, and includes South Africa (which is not an OECD member).

2

This elasticity pessimism was particularly widespread among economists and policymakers in the United States in 1972–73; see Whitman (1975).

3

Nonelectrical machinery accounted in 1974 for 21 per cent of the Federal Republic of Germany’s exports, 17 per cent of U. S. exports, and 19 per cent of U. K. exports. It accounted for 13 per cent of OECD countries’ exports to the OECD area, and for 17 per cent of their exports to the non-OECD area.

4

On the basis of the price indices used in the present study, the Federal Republic of Germany’s nonelectrical machinery export prices in dollar terms increased between 1969 and 1974 by close to 60 per cent relative to U. S. prices and by about 30 per cent with respect to British prices.

5

A number of excellent surveys of econometric work on foreign trade price elasticities and their weaknesses are available; for example, see Magee (1975), Leamer and Stern (1970), and Stern and others (1976).

6

The importance of allowing for the existence of differences in substitution elasticities between markets has been pointed out by White (1970), while the need to take into account the pattern of the various countries’ exports across markets has also been demonstrated by White (1970). Biases caused by improper aggregation of commodities have been analyzed by Leamer and Stern (1970, Appendix to Ch. 2, pp. 41–50) and Magee (1975). Barker (1970) has attempted to quantify aggregation biases involved in the estimation of the import price elasticities of U. K. imports.

7

The role played by delivery delays and other cyclical nonprice rationing variables in export equations has been explored by Ball and others (1966), Steuer and others (1966), and Artus (1970 and 1973).

8

The case against the use of unit value indices in econometric work on foreign trade price elasticities and the argument in favor of contract prices have been forcefully presented by Kravis and Lipsey (1971).

9

The deflator P is calculated from the formula

ln(P)=Σlslln(Pl)

where sl is the share of country l in total demand (in nominal terms).

10

This condition is often referred to as the independence assumption.

11

For a proof, see Hanoch (1971), pp. 698–99.

12

The list of subgroups is given in the Appendix.

13

This weighting scheme is to some extent equivalent to assuming that the substitution elasticity σ is the same for all subgroups of products in the market considered. However, the equations (4) for the subgroups cannot be easily aggregated because they are expressed in logs, so that the aggregate equation used here has to be considered as an approximation.

14

The price equation can be seen as an implicit supply equation. The parameter fj in equation (7) is the inverse of the supply price elasticity.

15

The EEC market used here includes the founding members except for the Federal Republic of Germany—that is, Belgium-Luxembourg, France, Italy, and the Netherlands. The countries included in the two other markets considered are listed in the Appendix.

An attempt at estimating relative price effects in the Sino-Soviet market failed because exports to that market are strongly influenced by governmental regulations and political factors.

16

These three major exporting countries account for 63 per cent of total exports of nonelectrical machinery by OECD countries to the rest of the world (excluding the three countries considered).

17

The empirical relation between U. K. export contract prices and unit values has been investigated by Artus (1974).

18

White (1970) has shown that the existence of such price discrimination may lead to biased estimates of the price elasticities.

19

Longer lags could not be obtained on the basis of the limited sample data used here. It is, in any case, difficult, if not impossible, to disentangle very long lags from the structural changes that are imperfectly proxied in time-series studies by simple time trends.

20

The fraction of the last quarter that is required to work off the amount of unfilled orders is calculated by dividing the remaining unfilled orders by the total amount of orders delivered during that quarter.

21

Thus, the calculation of WT for the United States rests on the further assumption that foreign orders keep their places in the total (home plus foreign) order queue.

22

Introducing all three countries into the equation may lead to some interdependence among the error terms, since the shares of the three countries must sum to one. The variables, however, are used in log form, so that the constraint that the sum of the shares must equal one does not introduce a strict linear dependence among the error terms.

23

The point estimates for the long-run elasticities are obtained by summing the values of σt→4 and σ5→12 presented in Table 1; their standard errors are calculated as the square root of the sum of the variances of the estimates of and σ5→12 plus twice the covariance between these estimates. The notation σ1→4, for example, refers to the elasticity with respect to the period t-1 through t-4.

24

Parentheses enclose standard errors of coefficients.

25

For example, this belief was recently expressed by Mr. Hans Apel, the [then] Minister of Finance of the Federal Republic of Germany. Asked to explain why German exports have not suffered much from the more than 30 per cent effective appreciation of the deutsche mark (DM) between early 1973 and March 1977, Mr. Apel answered:

The fact that this [appreciation of the DM] has been possible without endangering German exports, can be explained like this: obviously the variety and the quality of German goods fits almost exactly what the customer wants. Also, customers can rely on the dates of delivery promised by German suppliers being met. This is, to a large extent, due to good labour relations in our country. These points—for the buyer of German products—apparently are so important that the high price of the German currency unit has not had much influence on our exports. See “Apel Answers Back,” The Banker, Vol. 127 (April 1977), p. 40.

26

This is a recurring theme in empirical studies of foreign trade price elasticities, but one that cannot be overemphasized.

27

Finland is an associate member of the EFTA.