The Short-Run Effects of Domestic Demand Pressure on British Export Performance

This paper is based on the author’s unpublished doctoral dissertation, “A Theoretical and Empirical Investigation of the Short-Run Effects of Domestic Demand Pressure on British Export Performance, 1954–1967,” presented to the University of California at Berkeley in 1969. The research project was financed by the Institute of International Studies at the University.

Abstract

This paper is based on the author’s unpublished doctoral dissertation, “A Theoretical and Empirical Investigation of the Short-Run Effects of Domestic Demand Pressure on British Export Performance, 1954–1967,” presented to the University of California at Berkeley in 1969. The research project was financed by the Institute of International Studies at the University.

This paper is based on the author’s unpublished doctoral dissertation, “A Theoretical and Empirical Investigation of the Short-Run Effects of Domestic Demand Pressure on British Export Performance, 1954–1967,” presented to the University of California at Berkeley in 1969. The research project was financed by the Institute of International Studies at the University.

ALTHOUGH there have been several recent theoretical and empirical studies of the short-run response of British exports to changes in the pressure of domestic demand, there is still much uncertainty about the direction, magnitude, and time pattern of this response. The theoretical discussion of this problem has been confused. Economists and businessmen with expansionist views 1 have maintained that domestic expansion reduces the average cost per unit and induces technological progress, making it easier for exporters to compete with foreign producers. Those with more traditional views have argued that when domestic demand increases the quantity of goods available for export is decreased, wage rates and prices rise, and waiting times are lengthened in the industries working on export orders. Very often, no clear distinction has been made between export values and volumes, short-run and long-run effects, and marginal and average costs. Furthermore, the debate has been influenced by political preferences.2

Empirical studies have not provided convincing answers. The recent study of international trade by the Organization for Economic Cooperation and Development (OECD) concluded that British exports had not been significantly affected by demand pressure during the period 1955–65.3 However, the econometric analysis by Ball, Eaton, and Steuer suggests that a high internal pressure of demand tended to decrease the volume and value of the U.K. exports of manufactures during the period 1954–64.4 A similar conclusion was reached by Renton in his study of the behavior of British manufactured exports to industrial countries from the first quarter of 1956 through the third quarter of 1966.5 On the other hand, Smyth found that it was the rate of change of demand pressure, rather than its absolute level, that had influenced the U.K. export performance during the period 1953–65.6 These results are difficult to reconcile.

This paper is a critical review of present-day theory and empirical knowledge relevant to this problem. A more systematic approach to the empirical study of the export response to a change in domestic demand has been designed; this new approach was then used to study the short-run response of British exports of motor vehicles (Standard International Trade Classification—SITC-732) and chemicals (SITC-5) to a change in home demand.

I. Theoretical Analysis

Short-run export behavior and balance of payments theories

The theoretical study of the short-run response of manufactured exports to changes in the pressure of domestic demand under a system of fixed exchange rates has long been neglected, mainly because balance of payments theories were not suitable to deal with the problem. The classical theory has often been adduced to argue that, owing to unfavorable relative-price effects, an increase in domestic demand would reduce the supply of export goods. However, this theory dealt mainly with long-run equilibrium analysis. Assuming that prices (including money wage rates) were more flexible than employment and real national income, it focused on the effects of changes in relative prices rather than on the role played by the variations in real national income that accompany changes in domestic or foreign demand.7 This theory is, therefore, not directly relevant to the study of export behavior during business cycles, because the main characteristics of these cycles are fluctuations in real and money incomes rather than in prices.

The Keynesian approach to balance of payments theory has focused on the effects of changes in real income, but, curiously, it does not provide a useful framework for analyzing the export response to changes in domestic demand during business cycles. Its implicit assumption that the structure of costs and prices is not modified by changes in real national income leads to the conclusion that exports are not directly influenced by changes in domestic demand. However, this assumption is correct only if the prices of the factors of production are not influenced by a change in economic activity and if firms are operating in a zone of constant marginal costs. It is not even certain that these conditions would be met in the trough of a severe depression. If we consider the economic situation of the leading industrial countries since World War II, it is clear that these conditions have never been met.8

In view of the gap between present-day economic conditions and Keynes’ assumptions, it is not surprising to find that the Keynesian analysis of short-run export determinants is very misleading when applied to the behavior of exports of the main industrial countries during the past 20 years. The exports of country i are assumed to depend only on the level of real income in the rest of the world. An increase (decrease) in domestic demand in country i will produce a rise (fall) in its level of imports and, through the foreign-trade multiplier mechanism, a related rise (fall) in the level of real income of the rest of the world. Thus, country i’s exports will rise (fall) as a result of an increase (decrease) in its domestic demand because of the changes in income in the rest of the world that are thereby induced. This analysis is certainly important in the explanation of the situation of a dominant economy, such as that of the United States in relation to the rest of the world, or that of France in relation to the rest of the French franc area. However, if we consider a typical economy, it is hard to believe that this indirect effect will be important, particularly in view of the possibility that foreign countries may use monetary or fiscal policy to counterbalance the effects of the foreign-trade multiplier. In fact, the alterations in the structure of costs and prices that accompany a change in the level of domestic demand may have a much more important effect on exports than do the induced income changes abroad.

The Keynesian approach, like the classical analysis, deals only with an extreme example; during the past 20 years, economists have been attempting to generalize these approaches to make them more useful. However, even modern balance of payments analysis has not yet fully integrated income effects and relative-price effects in a “general theory,” although economists are now quite conscious that in the short run the two mechanisms are operating together and that the relative weights of these two effects depend on the amount of unused resources existing in the economy.9 Furthermore, this analysis applies, strictly speaking, only to the particular instance of perfect competition in the world market and in the domestic market. It does not tell us what happens if the main export sectors have an oligopolistic or monopolistic structure, or if firms are maximizing their long-run profits rather than their short-run profits, or if selling expenses or waiting times are more important than prices in determining the competitiveness of a country in the world market.

There is clearly a need for a more general balance of payments theory that would give us a better understanding of the short-run behavior of exports during business cycles. However, before such a general theory can be built, much partial analysis of a theoretical and empirical nature is necessary, because little is known of the influence of various market structures on the short-run behavior of firms in export markets.

Short-run export behavior and the theory of the firm

An analysis of the effects of a change in the level of domestic demand on exports at the level of the individual firm is useful for the identification of the main factors that affect the relationship between domestic demand and exports at the macroeconomic level. Such a microeconomic approach makes it possible to account for factors not usually considered in macroeconomic analysis, such as market and cost structures, objectives of the entrepreneurs, and nonprice competitive factors. So far, however, few attempts have been made to derive meaningful propositions on a firm’s short-run export behavior from the modern theory of the firm. Ball is one of the few economists to have made such an attempt.10 The following analysis is essentially a critical review and extension of his work. First, we consider a simple monopolistic discriminatory-pricing model that seems to depict the main characteristics of the situation faced by a large number of British exporting firms. Then we consider how the conclusions derived from such a simple model have to be modified when we take into account such important elements as the short-run rigidity of domestic prices and the influence of selling services.

Monopolistic discriminatory pricing

As a framework for his analysis, Ball uses a simple model in which a short-run profit-maximizing firm is confronted by sloped demand curves on two relatively isolated markets, the domestic and the world market.11 If there is an increase in domestic demand (i.e., an upward shift of the domestic demand curve), the following two propositions can be derived.12

(1) The direction of the change in export volume is uniquely determined by the slope of the firm’s marginal cost curve in the neighborhood of the initial equilibrium output. If the marginal cost rises (falls) as output increases, there will be a decrease (increase) in the volume of exports.

(2) The magnitude of this change in export volume is determined by the slope of the marginal cost curve and by the demand elasticities in the domestic and foreign markets. The steeper is the slope of the marginal cost curve and the higher the foreign demand elasticity, the larger will be the change in export volume; the higher is the domestic demand elasticity, the smaller will be the change in export volume.

Ball suggests that for most of the British exporting firms the slope of the marginal cost curve in the neighborhood of the initial equilibrium output and the demand elasticities in the domestic and foreign markets are such that, under the assumptions of the model, a reduction (rise) in domestic demand should result in a significant increase (decrease) in exports. Empirical evidence tends to support his position. In many instances the domestic market for a particular group of commodities is well protected by tariffs and the number of national suppliers is very small; thus, the elasticity of the domestic demand curve facing a particular firm may be very low. A typical example is the British motor-vehicle industry, where five large manufacturers are supplying a domestic market protected by the McKenna duty of 30 per cent ad valorem on the landed cost of imported cars. On the other hand, owing to the existence of a large number of suppliers and buyers in the world market, the elasticity of foreign demand for manufactured goods supplied by a particular country is probably relatively high. Most of the recent econometric studies have found high substitution elasticities between industrial countries’ manufactured exports.13

As to the slope of the marginal cost curve, it is not very likely, and to the author’s knowledge has never been claimed, that in a major part of British export industries the marginal costs fall significantly as output increases. The hypothesis that short-run marginal costs are constant over wide ranges of output is not convincing either. This notion arises from empirical studies of individual business firms.14 However, most of these studies were conducted for a limited number of U.S. industries over a period characterized by a low level of activity. It is not surprising, therefore, that they did not detect the existence of increasing marginal costs. A recent survey of these studies by Walters 15 concludes that the hypothesis of increasing marginal costs as output rises has not been refuted by direct empirical evidence. Even though direct empirical verification is lacking, the assumption that in the short run marginal costs increase with rising outputs seems to be the most plausible one.

These reflections lead to the conclusion that a rise (decline) in domestic demand should be unfavorable (favorable) to exports. However, this conclusion rests on the validity of the assumptions of this simple model. We shall now show how the analysis can be modified if some of these assumptions do not hold.

Rigidity of the domestic price

A plausible alternative model of discriminatory pricing suggested by Ball16 is one in which prices are not influenced by short-run changes in demand. Entrepreneurs are assumed to sell as much as possible on the two markets under the constraint that the price on each of these markets remains constant. Under these conditions, a decrease in domestic demand would result in an increase in the volume of exports only if there was excess foreign demand but no excess domestic demand before the change in domestic demand occurred.

Ball does not consider the case where only export prices are adapted to short-run changes in demand-and-supply conditions. However, this case may be more relevant. It seems rational to assume that export prices are adapted to short-run changes in demand-and-cost conditions because the world market is to a large extent competitive. By contrast, domestic prices of exports may not be influenced by short-run changes in the level of domestic demand if the domestic market has an oligopolistic structure or if maximization of long-run profit precludes price changes in response to short-run changes in market conditions. Indeed, empirical investigation has shown that in the United Kingdom domestic prices of manufactures were insensitive to cyclical variations in the pressure of demand.17 Furthermore, studies of particular British export industries, such as motor vehicles 18 and chemicals, 19 have also concluded that domestic prices are not substantially affected by short-run fluctuations in demand. In these circumstances, a change in domestic demand would have a more powerful impact on the export volume than would a change in the domestic price. This is so because the entire adjustment to a change in domestic demand has to be made by shifting the firm’s supply curve for the foreign market.

Selling services

Another objection made by Ball to the use of the simple short-run discriminatory-pricing model is that the volume of exports may be more sensitive to the variations in selling services than to price changes. If we assume, as Ball does, that an increase in selling services requires a fairly substantial investment that can be completed only after a considerable period of time, we are led to the conclusion that such investment will be made only on long-run considerations and certainly not in response to a short-run decrease in the level of domestic demand. However, it seems doubtful that this is equally important in every export sector. In many industries, e.g., in chemicals, steel, and machinery, export sales are made to a relatively small number of large foreign firms and may not require a large fixed investment in advertising and marketing. In these industries, other factors—such as delivery delays, credit facilities, and personal contact with foreign entrepreneurs—may be important in the determination of the quantity of exports. Some of these factors can be modified relatively quickly. Even in the automobile industry, the volume of sales in a foreign market may be capable of being expanded rapidly as a result of changes in the amount of advertising for new models, once the initial sales structure has been established. Although an expansion of foreign sales may often require a large and lumpy initial investment in selling services, the possibility of a response of exports to short-run sales efforts motivated by a cyclical decline in domestic demand need not, in general, be precluded.

Conclusion

As a result of his reflections on the role of selling services and on the absence of short-run profit maximization, Ball concludes that “the argument that restricting credit [which would result in a decrease of domestic demand] will have a direct effect on the expansion of exports in the short run is highly dubious.”20 The preceding reconsideration of this analysis would seem to cast some doubt on this conclusion. Moreover, during an expansion (recession) of the whole economy the firm’s cost curves may be shifted upward (downward) relative to those of foreign competitors as a result of changes in wages or in the prices of materials. It is not implausible, therefore, that an expansion (recession) of domestic demand may be significantly unfavorable (favorable) to exports. However, these effects will depend in large measure on the cost and market structures of the main export sectors; and furthermore, their impact on exports may be very much reduced whenever exports are differentiated in quality or style from the corresponding products sold on the domestic market, since shifting production from domestic to foreign models or vice versa may in the short run involve high technical costs.21

Although theoretical microeconomic analysis of this type is suggestive, it is not conclusive. At any rate, it cannot provide a basis for a quantitative assessment of the effect of variations in demand pressure on export performance in general or in major industries. Such an assessment can only be made through empirical investigation either of the behavior of individual firms or of major export sectors.

II. Empirical Analysis

Although it would in principle be possible, as the preceding analysis has shown, to derive meaningful propositions about the short-run export response of British industries to a change in domestic demand from an empirical study of their cost and market structures 22 and behavior patterns, this approach is not very promising for the time being, because sufficiently precise information on these structures and patterns is not available. A more promising approach to an empirical study of this short-run export response lies in an analysis of the observed past export behavior of the British economy as a whole or of certain broad sectors. This is the approach that was used in Ball, Eaton, and Steuer’s pioneering study,23 and subsequently by Renton,24 Smyth,25 and Adams, Eguchi, and Meyer-zu-Schlochtern.26

Econometric analysis of U.K. export behavior

To measure the impact of a change in domestic demand on British export performance in manufactures, Ball, Eaton, and Steuer have estimated the parameters (A, α, β, r, γ1, γ2, and γ3) of the following export function: 27

(1)Xuk,t=AXw,tαCuk,tβer.tD1γ1D2γ2D3γ3eεt
article image

The parameter estimates were derived by means of ordinary least-squares multiple regression technique from quarterly observations from 1954 to 1964. To avoid problems resulting from high multicollinearity between Xw,t and t (time), α was constrained to 1. Under this constraint, the estimated value of β was negative and significantly different from zero at a 5 per cent level. Similar results were obtained when Xuk and Xw were measured in constant prices, or when the relationship was expressed in simple linear form. The conclusion seemed straightforward: a high (low) level of domestic demand had been unfavorable (favorable) to British export performance in manufactures, at least in the short run.

However, several criticisms can be made of this analysis:

1. The proxy variable used to represent the development of import demand in foreign markets is not adequate. It has long been recognized that, owing to sectoral and geographical specialization of exporting countries, the effect of a change in world demand on each country’s exports depends on the composition of this increase in demand by geographic area and commodity class. A change in world exports does not necessarily reflect a similar change of demand in the traditional British export markets.

2. There is no reason why the elasticity of British exports in manufactures with respect to world trade in manufactures should be equal to unity, since the United Kingdom may be specialized in cyclically sensitive or insensitive products. In fact, several recent studies tend to show that this elasticity may be significantly lower than unity. Houthakker and Magee have estimated the income elasticity of U.K. exports at 0.86.29 The National Institute of Economic and Social Research (NIESR) has shown that the more rapidly world trade is expanding, the more rapidly the U.K. share in world trade is declining.30 The misspecification involved in setting α equal to 1 may have serious consequences for the estimation of β, since Cuk and Xw may also be highly correlated. As McGeehan notes, this misspecification may have led Ball, Eaton, and Steuer to ascribe to the influence of domestic demand what are really world-demand effects.31 These authors should have investigated the sensitivity of their parameter estimates to changes in the dubious assumption that α equals 1.

3. Equation (1) is also unrealistic because it assumes an instantaneous impact of a change in capacity utilization on exports. If we use quarterly data, as Ball, Eaton, and Steuer have done, time lags must be considered, because reactions to changes are usually slow. In fact, there is probably a considerable lag between the time when domestic demand changes and the time when entrepreneurs become cognizant of this change and consider it sufficiently important to justify a change in their export policies. It is also likely that there is a lag between a change in companies’ export policies and the impact of this change on exports.

Similar criticisms can be made of the studies by Renton, Smith, and Adams, Eguchi, and Meyer-zu-Schlochtern.32 The study by Adams, Eguchi, and Meyer-zu-Schlochtern is the only one that identifies the influence of the British sectoral and geographical export specialization. Neither Smith nor Adams, Eguchi, and Meyer-zu-Schlochtern consider the possibility of lags. Renton assumes a lag of one quarter but does not test this assumption. Furthermore, all these studies have been made at a highly aggregate level.33 In Section I (pp. 250–55) we show that the magnitude and time pattern of an industry’s export response to a change in domestic demand depends in large measure on its cost and market structure. Thus, different industries may have quite different export behavior, which a study based on aggregate data may be unable to detect.34

A more systematic approach is needed. Each export industry should be considered separately, and due allowance must be made for the influence of the British geographical export specialization. The time pattern of the relationship between domestic demand pressure and exports should be more fully explored. Finally, if it is impossible to estimate simultaneously the influence of foreign demand and of domestic demand pressure on British exports owing to high multicollinearity between these two variables, the coefficient of the foreign demand variable should not be constrained to only one particular value but rather to a set of plausible values. In what follows, this approach will be used to study the short-run response of British exports of motor vehicles and chemicals to a change in domestic demand.

Short-run export response of the British chemical and motor-vehicle industries to a change in domestic demand

These two industries were selected for this study for two reasons: (1) they are particularly interesting and (2) the relevant data are available. The motor-vehicle industry is interesting because it has experienced wide fluctuations in domestic demand, partly induced by changes in hire-purchase controls used by the Government for the specific purpose of influencing exports.35 The chemical industry is interesting because, owing to the relative importance of overhead costs in this industry, marginal costs remain low until full capacity utilization has been reached.36 Both industries are oligopolistic in structure; as was argued in Section I, this should strengthen their export response to a change in domestic demand.

Hypothesis and approach

To estimate the magnitude and time pattern of the export response of these industries to a change in domestic demand, the parameters of the following equations were estimated from quarterly observations by employing simple least-squares multiple regression technique: 37

(2)Xc,t=AcWc,tαc(Uc,tnKc,tn)βc,nD1γc,1D2γc,2D3γc,3ercteut
(3)Xv,t=AvWv,tαv(Uv,tnKv,tn)βv,nD1γv,1D2γv,2D3γv,3ervteut
article image

The precise definitions of these variables and the data are given in the Appendix.

These equations are similar to those used by Ball, Eaton, and Steuer. However, the effects of the British geographical specialization are now fully identified by the variable W, which is a weighted average of the indices of imports in the British foreign markets, the weight for each foreign market being the share of British exports that it absorbed in the base year, 1958. Furthermore, we do not assume an instantaneous impact of the change in capacity utilization on exports; and, at least as a first step, α is not constrained but estimated from the data, and a relative measure of capacity utilization is used.

Instead of tracing separately the short-run effects that a change in domestic demand pressure has on the volume of exports and on export prices as a result of the shift of the export supply curves that it induces, the current value index of exports was used as the dependent variable. This method is not wholly satisfactory, but the export price data that could be used as a deflator to obtain volume indices are either not available or of dubious quality.39

The dependent variable has been cleaned thoroughly. British exports to non-OECD members of the sterling area were not included because of the special economic relations existing between the United Kingdom and these countries. The periods under study, the first quarter of 1960 through the third quarter of 1967 for chemicals and the first quarter of 1960 through the third quarter of 1966 for motor vehicles,40 were chosen because they were relatively free from large trade disturbances.41 Quarterly data on British exports were adjusted for dockworkers’ and seamen’s strikes (see Appendix).

Empirical findings

Estimation of the parameters of equations (2) and (3). As a first step, the parameters of equations (2) and (3) have been estimated without imposing any smoothness conditions on the shape of the lag distribution. Because there is no satisfactory method for determining the minimum and maximum time lags from the data,42 it was assumed a priori that these lags were not less than one quarter or longer than six quarters. Even after making this assumption, we found that the multicollinearity among successive values of the ratio U/K were so high that the individual coefficients β could not be estimated with any precision. However, the sums of these coefficients, hereafter referred to as Bc and Bv, were found to be highly significant; and, as the work progressed, most of our attention came to be paid to these sums, which express the “total” elasticities of exports with respect to the relative capacity utilization. The three variables (U/K)t-1, (U/K)t-2, and (U/K)t-6 were discarded at a later stage because their coefficients were far from being significant and their introduction in the equation increased the standard error of B^candB^v.43 As can be seen from the results presented in Tables 1 and 2, regression equations (2.1) and (3.1), these total elasticities were found to be negative and significantly different from zero at the 95 per cent level of confidence in both industries. However, this analysis suffers from several weaknesses. These are discussed in the remainder of Section II.

Table 1.

Regression Coefficients: U.K. Exports of Chemicals, First Quarter 1960—Third Quarter 19671

article image

Numbers in parentheses are standard errors.

s = Standard error of the regression.

d = Durbin-Watson statistics.

Table 2.

Regression Coefficients: U.K. Exports of Motor, Vehicles First Quarter 1960—Third Quarter 19661

article image

Numbers in parentheses are standard errors.

s = Standard error of the regression.

d = Durbin-Watson statistics.

First, it must be noted that the introduction of the relative measure of demand pressure (U/K) rests on the assumption that other countries’ exports respond to changes in demand pressure in the same way as do British exports. This assumption may not be justified, because market structures and entrepreneurial behavior may vary from one country to another. In fact, it was found that only the coefficients of U were significantly different from zero when the variables U and K were introduced separately in the regression equations. Furthermore, the standard errors of B^candB^v and the standard error of the regression for motor vehicles are smaller when the variables K are omitted from the equations—see Tables 1 and 2, regression equations (2.2) and (3.2).44 This should not be surprising, because the variables Kc and Kv have not experienced large fluctuations during this period; therefore, their effects cannot be identified. If the variables K are omitted, the estimated value of B is −0.388 (0.108)45 for exports of motor vehicles and −0.341 (0.141) for exports of chemicals. The fit is rather good, and the Durbin-Watson statistics indicate that there is no serious autocorrelation.

Estimation of the lag pattern by Almon’s method. The second weakness of the approach described above is that it does not yield much information on the time pattern of the relation between exports and demand pressure. First, the elimination, on statistical criteria, of (U/K)t-1, (U/K)t-2, and (U/K)t-6 from the regression equations is hardly convincing on theoretical grounds. Furthermore, the standard deviation of each of the remaining β^’s is very high, owing to fairly strict multicollinearities among the successive values of Ut and Kt. To overcome this problem of multicollinearity, we have imposed a constraint on the shape of the lag function, according to the method proposed by Almon.46 A polynomial of the third degree was used as lag function, and the value of the coefficients of (Ut-0) and (Ut-7) were constrained to zero.

The results of the estimation of this distributed lag model are presented in Tables 3 and 4, regression equations (2.3) and (3.3). The value of B^v confirms the results obtained previously, although the new results tend to prove that—by considering only the lagged values (Ut-3) to (Ut-5)—there had been a slight underestimation of the absolute value of the total elasticity of exports of motor vehicles with respect to the level of capacity utilization. These results show also that two thirds of the export response of the motor-vehicle industry to a change in demand pressure occurs between the third and the fifth quarters after this change. For chemical exports, the results are less conclusive; the estimated value of Bc is reduced to −0.266 (0.175), and it is no more significantly different from zero at a 95 per cent level. Furthermore, the concentration of the export response in periods t—5 and t—6 should tend to show that the hypothesis that βc,t-1 is equal to zero may not be verified. However, if we make the same estimation but use a polynomial of degree 2 without constraining βt-7 to zero,47 it can be seen in Tables 3 and 4, regression equations (2.4) and (3.4), that neither β^c,t7norβ^v,t7 is significantly different from zero at the 95 per cent confidence level. Nevertheless, the continuous increase of the absolute value of the coefficients seems hardly reasonable.

Table 3.

Polynomial Distributed Lag Results: Exports of Chemicals

article image

s = Standard error of the regression.

d = Durbin-Watson statistics.

Table 4.

Polynomial Distributed Lag Results: Exports of Motor Vehicles

article image

s = Standard error of the regression.

d = Durbin-Watson statistics.

Robustness of the parameter estimates. A further problem arises from the multicollinearity between the trend and the market variable.48 This is particularly true for chemicals, where this multicollinearity may be the reason for the low estimate of αc and the existence of an increasing trend. To see how this has affected the estimates of Bc and Bv and their standard errors, we have estimated these parameters after constraining the value of a successively to 1.0, 0.8, and 0.6. Since it is most unlikely that the real value of α could be larger than 1.0 or smaller than 0.6, this analysis should give us a good idea of the robustness of our estimates of Bc and Bv.

At first the estimation was made without imposing any smoothness conditions on the lag distribution. The results presented in Table 2, regression equations (3.5)-(3.7), show that the total elasticity of British exports of motor vehicles with respect to the level of capacity utilization in this industry does not depend very much on the assumption on the value of αv. However, for the chemical industry, it can be seen in Table 1, regression equations (2.5)–(2.7), that this total elasticity changes from −0.599 (0.167) when αc is assumed equal to 1 to −0.386 (0.126) when αc is assumed equal to 0.6.

If we consider the results for polynomial lags that are given in Tables 5 and 6, regression equations (2.8) and (3.8), it is also clear that it is only for the chemical industry that the estimate of the time pattern of the export response is affected by the value that we give to α. When αc is assumed equal to 1, it is found that three fourths of the export response of the chemical industry occurs between the third and the fifth quarters after the change in the pressure of demand. This estimate of the time pattern of the relation seems reasonable; however, the difference between this estimate and the one obtained when αc is unconstrained—i.e., regression equations (2.3) and (2.4)—casts a serious doubt on the confidence that we can have in it.

Table 5.

Polynomial Distributed Lag Results: Export Share of Chemicals

(Regression Equation 2.8)1

article image

Constraints: βc,0 = 0 and polynomial of degree 2.

Table 6.

Polynomial Distributed Lag Results: Export Share of Motor Vehicles

(Regression Equation 3.8)1

article image

Constraints: βv,0 = 0 and polynomial of degree 2.

III. Conclusion

A brief summary of the main results of our empirical study is presented in Table 7.

Table 7.

Regression Results: Summary1

article image

Numbers in parentheses are standard errors of the coefficients.

These results indicate that from the first quarter of 1960 through the third quarter of 1967 a 10 per cent increase (decrease) in the level of capacity utilization in the motor-vehicle industry resulted in a decrease (increase) in its exports by about 3–4 per cent. For the chemical industry the results are less clear, because the estimate of the effect of a change in the level of capacity utilization is affected significantly by a change in the assumed value of the export elasticity with respect to world demand. However, if the value of this elasticity is close to 0.6, the results for the chemical industry are similar to those obtained for the motor-vehicle industry. In both sectors, there was a lag of approximately three or four quarters between the change in the level of capacity utilization and its main impact on exports.

It seems hardly necessary to emphasize the tentative nature of this empirical study, particularly in view of the short time period used for analysis and of the weakness of a single-equation model. However, if these empirical results 49 are added to those obtained by Ball, Eaton, and Steuer 50 and by Renton, 51 and if the a priori analysis of this question is borne in mind, the conclusion that an increase (decrease) in the level of domestic demand has a short-run, unfavorable (favorable) effect on the value of British exports seems unavoidable. In fact, the direct impact of a change in domestic demand on exports seems highly significant from a statistical and economic point of view. These results show that to ignore completely this direct effect as in the Keynesian analysis, or to neglect it as in the absorption approach to balance of payments theory, is a serious mistake.

APPENDICES

A. Definition of Variables and Source of Data

Export value indices, X

These are quarterly value indices of exports to all foreign countries except non-OECD members of the sterling area.52 Data on quarterly export values were obtained from the OECD publications, Series B: Commodity Trade—Analysis by Main Regions and Series C: Commodity Trade: Exports—Detailed Analysis by Products. Allowance for major seamen’s and dockworkers’ strikes was made by taking the deviations of the monthly export values corresponding to a strike period from the export trend computed by the Board of Trade 53 and by correcting the previous data accordingly. Corrections for strike periods were as follows:

Motor Vehicles (SITC-732)

article image

Chemicals (SITC-5)

article image

“Market” indices, W

These are weighted aggregates of import indices in the British foreign export markets. More precisely,

Wi=ΣjQj,iXj,iX¯j,iwithQj,i=XUKj,i,1963XUK,i,1963
article image

The world has been divided into 11 areas: the United States; Canada; Germany; the United Kingdom; France; Italy, Belgium, Luxembourg, the Netherlands; Rest of OECD member countries; Overseas Sterling Area;55 Latin America; Overseas Franc Area;56 and Rest of the World. For computing W, the weights of the Overseas Sterling Area, Overseas Franc Area, and the United Kingdom were set equal to zero. Data on OECD’s exports to these areas were taken from the OECD Series B and C commodity trade publications mentioned earlier.

Levels of capacity utilization, U

These variables are derived from quarterly indices of physical output by the “trend-through-peaks” method, often referred to as the Wharton School method (from the Wharton School of Finance and Commerce, Philadelphia, Pennsylvania). The quarterly index of physical output of the motor-vehicle industry was derived from series on the number of automobiles and commercial vehicles produced by the British industry. These series were taken from the Monthly Digest of Statistics, published by the U.K. Central Statistical Office. They were seasonally adjusted by using seasonal coefficients estimated by multiple correlation analysis. For the chemical industry, we used the seasonally adjusted index of physical output of the “Chemicals and allied industries,” also published in the Monthly Digest of Statistics. The definition of the group “Chemicals and allied industries” is different from the definition for SITC-5 because it includes coke ovens and oil refinery products. However, in the index to “Chemicals and allied industries,” coke ovens and oil refineries have been given a weight of only 9 out of a total weight of 68. The peaks were found in the first quarter of 1955, 1960, and 1964 for the motor-vehicle industry, and in the first three quarters of 1960 and the first quarter of 1965 for the chemical industry.

Competitors’ levels of capacity utilization, K

Levels of capacity utilization were computed for France and Germany using the quarterly indices of physical output published, respectively, by the Institut National de la Statistique et des Etudes Economiques in Bulletin Mensuel de Statistique and by Statistisches Bundesamt (Wiesbaden) in Industrie und Handwerk—Reihe 2. For the United States, we used the indices of capacity utilization computed by the econometric unit at the Wharton School. The variables Kv and Kc are weighted averages of these three levels of capacity utilization, the weights being, respectively, the relative shares of these three countries in world trade in motor vehicles and in chemicals in the first quarter of 1963.

B. Data on British Motor-Vehicle Industry, First Quarter 1958—Third Quarter 1966

(For indices, 1958 (I) = 100)
article image

In the regression, (Uv) and (Kv) are introduced in index form with the first quarter of 1958 equal to 100.

C. Data on British Chemical Industry, First Quarter 1958—Third Quarter 1967

(For indices, 1958 (I) = 100)
article image

In the regression, (Uc) and (Kc) are introduced in index form with the first quarter of 1958 equal to 100.

Les effets à court terme exercés par les pressions de la demande intérieure sur le comportement des exportations britanniques

Résumé

Ce document présente une analyse des théories actuelles et des connaissances empiriques relatives aux effets à court terme exercés par la pression de la demande intérieure sur le comportement des exportations britanniques. L’auteur a fait appel aux théories de la balance des paiements et à des éléments de la théorie de l’entreprise pour montrer qu’il n’est pas exclu qu’en période courte un accroissement (affaiblissement) de la demande intérieure exerce une influence défavorable (favorable) sur les exportations.

Après avoir rappelé l’hypothèse traditionnelle selon laquelle toute augmentation de la demande intérieure a pour résultat de diminuer la capacité de concurrence du pays en matière d’exportation, du fait qu’elle augmente les coûts marginaux et les prix de ses exportations, l’auteur montre que cette hypothèse est encore renforcée par l’existence d’éléments monopolistiques dans le secteur d’exportation et par la rigidité des prix intérieurs. Il souligne également que si d’importants capitaux doivent souvent être investis initialement et en une seule fois dans les services de ventes pour pouvoir augmenter le chiffre d’affaires avec l’étranger, il ne faut cependant pas en général écarter la possibilité d’une réaction des exportations en réponse à une promotion à court terme des ventes suscitée par un affaiblissement cyclique de la demande intérieure.

Après un examen critique de l’analyse économétrique actuellement utilisée pour étudier cette réaction des exportations, une approche plus systématique est proposée pour mesurer les effets exercés en période courte par une variation de la demande intérieure sur les exportations britanniques de véhicules automobiles routiers (Classifications type pour le commerce international—CTCI—732) et de produits chimiques (CTCI—5); le profil chronologique de ces effets est en particulier exploré plus à fond que dans les études antérieures. Les résultats de cette étude empirique indiquent qu’en période courte, une augmentation (diminution) de la demande intérieure de véhicules automobiles routiers et de produits chimiques exerce une influence défavorable (favorable)—significative du point de vue statistique et d’une grandeur non négligeable—sur la valeur des exportations britanniques de ces produits. Dans les deux secteurs, on a enregistré un décalage d’environ trois à quatre trimestres entre le moment où le niveau de la demande intérieure a varié et celui où son effet s’est fait pleinement sentir sur les exportations.

Los efectos que ejerce a corto plazo sobre la exportación británica la presión de la demanda interna

Resumen

Este trabajo es un análisis de las teorías actuales y conocimiento empírico pertinentes a los efectos de corto plazo que ejerce en la exportación británica la presión de la demanda interna. Se hace uso de las teorías de la balanza de pagos y de algunos elementos de la teoría de la empresa, para demostrar que no es inverosímil que una expansión (recesión) a corto plazo de la demanda interna tenga un efecto desfavorable (favorable) sobre la exportación.

Se demuestra que la hipótesis tradicional de que un aumento de demanda interna disminuye la competitividad de las exportaciones de un país, al elevar los costos y precios marginales de las mismas, queda reforzada por la existencia de elementos monopolistas en el sector de exportación y por la rigidez de los precios internos. También se mantiene que, aunque una expansión de las ventas al exterior suele necesitar una inversión inicial cuantiosa y concentrada en los servicios de ventas, en general, no hay por qué excluir la posibilidad de que reaccionen las exportaciones ante la promoción a corto plazo de las ventas, motivada por una disminución coyuntural en la demanda interna.

Tras un examen crítico del análisis econométrico existente sobre dicha reacción de las exportaciones, se sigue un enfoque más sistemático con la finalidad de medir los efectos que una variación en la demanda interna ejerce a corto plazo sobre la exportación británica de vehículos automotores (CUCI 732) y de productos químicos (CUCI 5); es decir, que en comparación con los estudios anteriores, en especial se ha explorado más a fondo la configuración cronológica de dichos efectos. Los resultados de este estudio empírico indican que, a corto plazo, un aumento (disminución) de la demanda interna de vehículos automotores y de productos químicos ejerció un desfavorable (favorable) efecto, estadísticamente significativo, y de cierta magnitud, sobre el valor de la exportación británica de dichos productos. En ambos sectores se registró un desfase de unos 3 ó 4 trimestres entre la variación en el nivel de la demanda interna y su impacto principal en la exportación.

*

Mr. Artus, economist in the Special Studies Division of the Research Department, has degrees from the Faculty of Law and Economics in Paris and from the University of California at Berkeley.

1

A presentation of the expansionist and traditional views on the impact of domestic business cycles on exports can be found in Use Mintz, Cyclical Fluctuations in the Exports of the United States Since 1879 (National Bureau of Economic Research, Columbia University Press, 1967). The term “expansionist view” refers to the theory that an economic expansion (recession) has a favorable (unfavorable) effect on exports.

2

This situation was well described by T.W. Hutchison (Economics and Economic Policy in Britain, 1946-1966, London, 1968, p. 218), “In the absence of convincing empirical evidence, the precise relationships between the level of domestic demand and exports were inevitably a question of a priori speculation and guesswork, and, of course, there was a strong temptation to make whatever assumption fitted best with one’s general ideas and inclinations about policy.”

3

F.G. Adams, H. Eguchi, and F. Meyer-zu-Schlochtern, An Econometric Analysis of International Trade (Organization for Economic Cooperation and Development, January 1969).

4

R.J. Ball, J.R. Eaton, and M.D. Steuer, “The Relationship Between United Kingdom Export Performance in Manufactures and the Internal Pressure of Demand,” The Economic Journal, Vol. LXXVI (1966), pp. 501-18.

5

G.A. Renton, “Forecasting British Exports of Manufactures to Industrial Countries,” National Institute of Economic and Social Research, National Institute Economic Review (November 1967), pp. 35-51.

6

D.J. Smyth, “Stop-Go and United Kingdom Exports of Manufactures,” Bulletin of the Oxford University Institute of Economics and Statistics, Vol. 30 (1968), pp. 25-36.

7

John M. Letiche, Balance of Payments and Economic Growth (New York, 1967), pp. 51-57. As Letiche has noted, some of the more important economists of the nineteenth century, such as Thornton and Cairnes, also had occasion to speak of the mechanisms of adjustment of the balance of payments in the short run; and, in this instance, they considered the effects of changes in income and expenditure. However, this aspect of their analysis was neither fully developed nor satisfactorily integrated with their analysis in terms of relative prices.

8

At the end of World War II, Keynes became quite conscious of the role of costs and prices in the mechanisms of adjustment. See Lord Keynes, “The Balance of Payments of the United States,” The Economic Journal, Vol. LVI (1946), pp. 172-87, and Letiche’s analysis of Keynesian evolution on internal and external adjustments, op. cit., pp. 100-103. It is unfortunate that most of Keynes’ early disciples overemphasized the simple case of constant prices at a time when this assumption was becoming less and less relevant.

9

See, for example, Letiche, op. cit., pp. 86-91, and M. O. Clement, Richard L. Pfister, and Kenneth J. Rothwell, Theoretical Issues in International Economics (Boston, 1967).

10

R. J. Ball, “Credit Restriction and the Supply of Exports,” The Manchester School of Economic and Social Studies, Vol. XXIX (1961), pp. 161-72.

11

This model is nothing more than the simple monopolistic discriminatory-pricing model presented by Joan Robinson in Chapter 15, “Price Discrimination,” The Economics of Imperfect Competition (London, 1933), pp. 179-202.

12

For a mathematical proof of these statements, see Ball, op. cit., pp. 169-72.

13

See Helen B. Junz and Rudolf R. Rhomberg, “Prices and Export Performance of Industrial Countries, 1953-63,” Staff Papers, Vol. XII (1965), pp. 224-71, and Mordechai E. Kreinin, “Price Elasticities in International Trade,” The Review of Economics and Statistics, Vol. XLIX (1967), pp. 510-16.

However, it should be noted that most empirical studies have failed to estimate a significant, negative relationship between the volume of U.K. manufactured exports and relative prices. See Adams, Eguchi, and Meyer-zu-Schlochtern, op. cit., and Ball, Eaton, and Steuer, op. cit.

14

See Joel Dean, Managerial Economics (New York, 1951), and J. Johnston, Statistical Cost Analysis (New York, 1960).

15

A. A. Walters, “Production and Cost Functions: An Econometric Survey,” Econometrica, Vol. 31 (1963), pp. 1-66.

16

Ball, op. cit., pp. 163-64.

17

R. R. Neild, Pricing and Employment in the Trade Cycle: A Study of British Manufacturing Industry, 1950-61 (National Institute of Economic and Social Research, Occasional Papers, No. XXI, Cambridge University Press, 1963). As Eckstein and Fromm have noted, Neild’s results may be due to a certain extent to his use of equations in the price-level form, which gives a heavier weight to cost elements because of their common trend values (Otto Eckstein and Gary Fromm, “The Price Equation,” The American Economic Review, Vol. LVIII, December 1968, pp. 1159-83). Using the same data as Neild but in quarterly-change form, Rushdy and Lund have found that the level of demand was a statistically significant factor in explaining price changes (F. Rushdy and P. J. Lund, “The Effect of Demand on Prices in British Manufacturing Industry,” The Review of Economic Studies, Vol. XXXIV, 1967, pp. 361-73). However, this is true only for some of a large number of equations they used; and even in these “successful equations” the magnitude of the coefficient of the pressure-of-demand variable is small.

18

G. Maxcy and Aubrey Silberston, The Motor Industry (London, 1959), Chapters 7 and 8.

19

W. B. Reddaway, “The Chemical Industry,” Chapter 6 in The Structure of the British Industry: A Symposium, ed. by Duncan Burn (National Institute of Economic and Social Research, Economic and Social Studies, No. XV, Cambridge University Press, 1958), Vol. 1, pp. 218-59.

20

See Ball, op. cit., pp. 168-69.

21

Wells gives many examples of goods for which export models are different from domestic ones. (See Sidney John Wells, British Export Performance: A Comparative Study, Cambridge University Press, 1964.) However, it is not difficult to find examples where export and domestic models are similar. Even when they are not similar, the opportunity costs of shifting between export and domestic models may not always be very high.

22

A tentative analysis of the cost and market structures of the British chemical, motor-vehicle, and nonelectrical machinery industries was made in Artus, op. cit.

23

See Ball, Eaton, and Steuer, op. cit.

24

See Renton, op. cit.

25

See Smyth, op. cit.

26

See Adams, Eguchi, and Meyer-zu-Schlochtern, op. cit.

27

The parameter β represents the elasticity of exports with respect to capacity utilization; it can be assumed that the elasticity with respect to domestic demand is equal to (β·σ) where σ is the elasticity of capacity utilization with respect to domestic demand.

These authors also did a considerable amount of work using slightly different specifications. In particular, they tried rather unsuccessfully to identify the long-run factors.

28

This variable is measured by the ratio of British industrial output (seasonally adjusted) to its trend value.

29

H. S. Houthakker and Stephen P. Magee, “Income and Price Elasticities in World Trade,” The Review of Economics and Statistics, Vol. LI (1969), pp. 111-25.

30

NIESR, National Institute Economic Review (February 1967), pp. 16-19.

31

Joy M. McGeehan, “Competitiveness: A Survey of Recent Literature,” The Economic Journal, Vol. LXXVIII (1968), pp. 243-62.

A misspecification of α is, in fact, an omission of the variable Xw with coefficient α—α from the right side of the equation, α being the misspecified value of α. In this instance, the bias of the estimates of the coefficients of the other variables Z present in the equation is equal to (α—α) MXwz Mzz1 where MXwz Mzz1 is defined by the regression coefficient of Xw on the variables Z. (See E. Malinvaud, Statistical Methods of Econometrics, Chicago, 1966, p. 264.) If we remember that Mzz1 is always positive, and that it is likely that Mxwcuk is positive (i.e., Xw and Cuk are positively correlated), we see that, assuming that α=1, when it is in fact smaller than 1 (i.e., α<α), biases downward the estimate of the coefficient of Cuk.

32

A further difficulty with this study is that the meaning of the coefficient of the variable that measures the relative pressure of demand is not clear, since these authors have also introduced export prices in their export function.

33

An exception is the study made by M. D. Steuer, R. J. Ball, and J. R. Eaton, “The Effect of Waiting Times on Foreign Orders for Machine Tools,” Economica, New Series, Vol. XXXIII (1966), pp. 387–03.

34

This point has rightly been emphasized by Ball, Eaton, and Steuer, op. cit., pp. 501-502.

35

See Artus, op. cit., pp. 69-74.

36

Ibid., pp. 109-110.

37

The assumptions on which this specification of an export equation rests have been discussed in Ball, Eaton, and Steuer, op. cit.; Adams, Eguchi, and Meyer-zu-Schlochtern, op. cit., Appendix to Chapter IV; and more fully in Artus, op. cit., Chapter III.

38

The non-OECD members of the sterling area include all the sterling area except Ireland, Iceland, and the United Kingdom.

39

The export price indices published recently by Kravis and Lipsey are available only up to 1964. (See Irving B. Kravis and Robert E. Lipsey, “New Measures of U.S. International Price Competitiveness, 1953-64,” in Toward Improved Social Economic Measurement, National Bureau of Economic Research, Forty-Eighth Annual Report, June 1968, pp. 21-28.)

40

The last quarter of 1966 and the first three quarters of 1967 were not considered because, during this period, exports of motor vehicles were affected not only by the dockworkers’ strike (September-October 1967) and the expectation of a devaluation but also by widespread stoppages in the motor-vehicle industry (autumn 1966). A more detailed regression analysis covering the period 1954-66 can be found in Artus, op. cit.

41

The changes in country’s market shares in world trade in chemicals and motor vehicles that occurred during the 1950’s reflects mainly the post-World War II reconstruction of these industries in continental Europe. World trade in motor vehicles was also very much affected late in the 1950’s by the enormous increase in the demand for small automobiles in North America, which obliged the General Motors Corporation and the Ford Motor Company to import and market “compacts” produced by their British subsidiaries. See Artus, op. cit., pp. 96-98.

42

For a technical discussion of this problem, see Malinvaud, op. cit., p. 482.

43

The sign ^ indicates the estimated value of the parameter. We know that var(B^)=var(c’β^)=σ2c’Vc, where σ2V is the variance-covariance matrix of the estimates of the individual coefficients β, c is a column vector of units, and c’ is the transpose of c. See J. Johnston, Econometric Methods (New York, 1963), pp. 131-32.

44

Regression equations (2.5)—(2.7) and (3.5)—(3.7) are similar to equations (2.2) and (3.2), but a constant has been imposed on the value of α. They will be explained under Robustness of the parameter estimates (pp. 264-69); they are in Tables 1 and 2 to facilitate the comparison of these various results.

45

Numbers in parentheses are standard errors.

46

This method has a strong advantage over the use of a geometric distribution because it allows for a more general specification of the shape of the lag structure, and because it does not assume that the disturbance terms and all explanatory variables have the same lag structure. See Shirley Almon, “The Distributed Lag Between Capital Appropriations and Expenditures,” Econometrica, Vol. 33 (1965), pp. 178-96.

47

We decreased by 1 the degree of the polynomial to keep constant the number of parameters used to estimate the lag distribution.

48
If we consider the variance-covariance matrix of the estimated coefficients in the regression equation for chemicals (2.2), we have
var(r^c)=9.845×106,var(α^c)=1.346×102,cov(α^c,r^c)=3.616×104,R¯2α^c,r^c=1[[var(r^c)var(α^c)cov(α^c,rc)2]/var(r^c)var(α^c)]=0.9868.
In the regression equation for motor vehicles (3.2), the multicollinearity is slightly less strict. We have
var(r^v)=3.661×105,var(α^v)=4.314×102,cov(α^v,r^v)=1.216×103,R¯2α^v,r^v=1[[var(r^v)var(α^v)cov(α^v,rv)2]/var(r^v)var(α^v)]=0.9363.
49

Similar results were found for the British nonelectrical machinery industry. An increase (decrease) in the level of domestic orders had an unfavorable (favorable) effect on the level of British exports. (See Artus, op. cit., Chapter VII.)

50

It should be noted that the estimated values of Bc, and Bv, are very similar to the estimated value of B for all exports in manufactures that Ball, Eaton, and Steuer obtained by using equation (1) after constraining o to 1. Using quarterly data on the value of British exports in manufactures from the first quarter of 1954 through the fourth quarter of 1964, they found B^ equal to −0.395 (0.114). See Ball, Eaton, and Steuer, op. cit., p. 511, regression equation (7).

51

See Renton, op. cit.

52

Libyan Arab Republic included; Iraq excluded.

53

See U.K. Central Statistical Office, Economic Trends, various issues.

54

United States and Canada included; Japan excluded.

55

Libyan Arab Republic included; Iraq excluded.

56

Viet-Nam, Laos, and Cambodia excluded; Morocco, Tunisia, and Algeria included; Guinea excluded after 1958.

IMF Staff papers: Volume 17 No. 2
Author: International Monetary Fund. Research Dept.