Import Discipline in the U.S. Manufacturing Sector
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
  • | 2 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

The proposition that imports provide a competitive constraint or discipline on the price-raising ability of domestic producers has long been part of the case for a liberal trade policy. Even if output in the domestic industry is concentrated among few producers and even if there is little countervailing power from either consumers or organized labor, actual and potential competition from imports are said to be sufficient to discourage monopolistic price behavior. For if domestic producers consistently maintain a price above the landed price of imports (for similar goods),1 they face the same prospective loss of output, profits, and employment as would ensue if there were effective internal price competition. As such, the expectation is that, other things being equal, the rate of change of domestic producers’ prices will be smaller, the greater the increase in import competition.2

Abstract

The proposition that imports provide a competitive constraint or discipline on the price-raising ability of domestic producers has long been part of the case for a liberal trade policy. Even if output in the domestic industry is concentrated among few producers and even if there is little countervailing power from either consumers or organized labor, actual and potential competition from imports are said to be sufficient to discourage monopolistic price behavior. For if domestic producers consistently maintain a price above the landed price of imports (for similar goods),1 they face the same prospective loss of output, profits, and employment as would ensue if there were effective internal price competition. As such, the expectation is that, other things being equal, the rate of change of domestic producers’ prices will be smaller, the greater the increase in import competition.2

The proposition that imports provide a competitive constraint or discipline on the price-raising ability of domestic producers has long been part of the case for a liberal trade policy. Even if output in the domestic industry is concentrated among few producers and even if there is little countervailing power from either consumers or organized labor, actual and potential competition from imports are said to be sufficient to discourage monopolistic price behavior. For if domestic producers consistently maintain a price above the landed price of imports (for similar goods),1 they face the same prospective loss of output, profits, and employment as would ensue if there were effective internal price competition. As such, the expectation is that, other things being equal, the rate of change of domestic producers’ prices will be smaller, the greater the increase in import competition.2

This paper presents estimates of the effect of changes in import competition on the domestic pricing behavior of U. S. manufacturing industries during the 1972–76 period. Changes in import competition are measured by changes in import penetration, where import penetration is defined as the ratio of the value of imports to the value of domestic shipments. The analysis is conducted (using annual data) at the disaggregated four-digit Standard Industrial Classification (SIC) industry level so that the domestic price effects of import competition can be studied across industries as well as over time. Indeed, two strong advantages of the cross-sectional approach are that variations in import competition are typically much larger across industries than for aggregate sectors over time,3 and that the industry disaggregation yields relatively large samples for estimation.

This study differs from earlier empirical work on import discipline in three major respects.4 First, our method of holding “other things equal” is different from that used in some studies. Specifically, we include the change in the industry’s factor input costs and the change in demand for the industry’s output in the price-change equations as control variables, so that the independent effect of changes in import competition on prices can be identified. This is in contrast to the popular approaches of either comparing price behavior in periods before and after large changes in import competition have taken place for a given group of industries or of comparing price behavior in industries subject to unusually weak or strong import competition with price behavior for all manufacturing industries during an identical time period.5 In our view, neither the “before-after approach” nor the “other industry control group approach” is satisfactory for inferring how prices would have behaved in a given industry if import competition had been different from what was actually observed. The before-after approach is likely to falter because the other major determinants of price changes (i.e., cost changes, demand conditions, incomes policy) are not likely to remain constant across the two time periods. Similarly, the control group approach will yield misleading answers if either the other determinants of price changes differ between the control industries and the subject industries or prices in the control group industries are not set independently of those in the subject industries.6 By controlling for cost and demand changes within the industry itself over a given time period, our approach avoids the worst of these pitfalls.

A second difference is that this study treats the pricing discipline of imports as dependent on the competitive market structure of the industry. Specifically, we argue that import competition should be expected to exert greater restraint on prices in highly concentrated industries.7 The presumption here is that a competitive domestic market will have already kept price close to marginal cost, thereby eliminating much of the potential disciplinary force of imports. In contrast, the greater deviation of price from marginal cost in highly concentrated industries provides considerable scope for import discipline. We find that this theoretical precept is supported in large measure by our empirical results.

The third major difference is that we pay particular attention to an important econometric problem that has plagued most earlier studies of import discipline (e.g., Krause (1962), Mancke (1968)).8 We refer to the simultaneous relationship between changes in domestic prices, on the one hand, and changes in import penetration and domestic demand, on the other. Because of this simultaneity, ordinary-least-squares (OLS) estimates of the effect of changes in import penetration on the change in domestic prices will be biased; further, under plausible assumptions, this bias can be large enough to seriously distort the “true” restraining price influence of import competition. In this paper, we employ two-stage-least-squares (TSLS) estimation methods to overcome the simultaneity problem, and we find, as expected, notable differences between the OLS and TSLS estimates.

The rest of the paper is organized into three parts. Section I outlines several theoretical models, including the conventional price leadership model, that suggest how import competition affects domestic price behavior. In the end, we draw on these models to support a three-equation empirical model in which the change in domestic prices is a function, inter alia, of the change in import penetration and the change in market demand and in which the change in import penetration and the change in market demand are each functions, inter alia, of the change in domestic prices. Our estimation results are presented in Section II. Section III states the study’s conclusions about the role of import competition in determining the behavior of domestic manufacturing prices. Finally, our data base is described in the Appendix, which also includes some supplementary estimation results.

Previewing what follows, our main empirical finding is that increases in import penetration do have a significant restraining effect on the inflation rate in U. S. manufacturing industries, and the effect is greater, the higher the level of concentration in the industry. Our estimates suggest that for industries with an “average” level of concentration (as denoted by an eight-firm shipments concentration ratio of about 50 per cent), a 10 per cent increase in the ratio of imports to shipments would, ceteris paribus, be accompanied by a fall in domestic prices of about 1 per cent. While small relative to the influence of cost and demand factors in price determination, such estimates suggest that policymakers should not disregard the domestic price implications of various policies designed to protect U. S. manufacturing industries from import competition.

I. Domestic Price Determination and Import Competition

The theory of import discipline is founded on the notion that, in analyzing the market structure of an industry, foreign as well as domestic firms must be counted as actual or potential competitors. The theory contends that oligopolistic behavior among firms in an industry, if such behavior truly exists, is more likely to exist among domestic firms and that, by contrast, foreign firms serving the domestic market are likely to constitute the competitive fringe of a market, continuously threatening the price-setting power of domestic firms. Thus, import competition is viewed as a countervailing force against whatever power domestic firms may have to raise prices above the competitive and socially efficient level of actual production costs.

This description of the role of import competition is consistent with the assumptions and general outline of the well-known price leadership model found in the literature on monopolistic competition.9 The basic price leadership model assumes that by virtue of its large size, a dominant firm in an industry may have enough market power to establish and maintain an industry-wide price sufficient to earn the dominant firm abnormally large profits. In setting a profit-maximizing price, the dominant firm is assumed to have knowledge of demand conditions in the market served by the industry, just as a pure monopolist does, but also to have knowledge of the supply capabilities of the competitive fringe of remaining firms in the industry.

In applying this framework to study the notion of import discipline, we divide producers serving the domestic market into the group of domestic producers and the group of foreign producers. Following the popular notion of import discipline, we assume foreign producers act as perfect competitors in the domestic market and that domestic firms attempt to coordinate their activities as if they were a single producer with perfect knowledge of overall market demand and of foreign supply capabilities. For now, we assume domestic and foreign firms produce identical goods, but, as we demonstrate later, the implications are the same when domestic goods and imports are treated as imperfect substitutes.

Under the foregoing assumptions, domestic producers will attempt to maximize their joint profit function π,

π=P·(DM)C((DM),Z,K)(1)

where P denotes price; (D · M) denotes the level of demand facing domestic producers (i.e., total domestic demand, D, less the supply of imports, M); and C(−) denotes domestic producers’ combined total costs of production, which are assumed to be a function of demand facing domestic producers (DM), variable factor prices (Z), and the fixed stock of productive capital (K). Differentiating equation (1) with respect to P yields the (maximum) profit condition

P=MC[D·(1MD),Z,K]+D·(1MD)MD(2)

where MC[·] denotes the marginal cost of domestic producers’ combined output, and M′ and D′ denote the incremental responses of import supply and total domestic demand, respectively, to changes of price.10 Marginal cost, the partial derivative of marginal cost with respect to its first two arguments, and M′ are all expected to be positive; D′ and the partial derivative of marginal cost with respect to capital are expected to be negative.

Two key implications of equation (2) are that effective price leadership by domestic firms will result in the market price being set above the marginal cost of domestic firms and that the size of this divergence will vary inversely with the price elasticity of import supply (M′), the price elasticity of total domestic demand (D′), and the market share of imports (M/D). Note also that in the limiting cases where the import supply elasticity is infinite, or where the price elasticity of demand is infinite (so that domestic firms cannot collude effectively), the second term in equation (2) vanishes, and the socially desirable equality of price and marginal cost is restored. Even in this case, however, it is interesting that an increase in the market share of imports still reduces the market price via its effects on firms’ marginal cost—that is, with total domestic demand given, an increase in imports lowers the output of domestic firms and pushes them to a lower point on their upward-sloping short-run marginal cost curves.

Next, consider the case where imports and domestic goods are imperfect substitutes. If, following Pugel (1980), we can assume that demands for similar domestic and imported goods are functions solely of their relative prices and of total spending on these goods, then the profit condition analogous to equation (2) can be written as

P=MC[EF(1PM·ME),Z,K]+EF(1PM·ME)(η1+η2Γ)(2)

where P, MC[·], and M are defined as before; E denotes the value of total expenditures on similar domestic and imported goods; PM denotes the price of imports; η1 and η2 are partial derivatives of demand for domestic output with respect to P and PM, respectively; and Γ is the (reduced-form) incremental response of import prices to changes in domestic prices.

Equation (2′) is, in fact, a simple generalization of the price leadership model in equation (2) and carries with it, by and large, the same implications. Because we anticipate that the quantity (η1 + η2 Γ), which measures the total change in demand for the domestically produced good in response to an incremental change of P, will be negative, price will again be greater than marginal cost so long as domestic firms can effectively coordinate their actions. Further, the power of domestic firms to set prices above marginal costs is once again seen to be a decreasing function of the price responsiveness of domestic demand and the market share of imports. The price-setting power of domestic firms also remains inversely related to the ability of foreign producers to supply the domestic market, for the greater the supply elasticity of imports, the smaller will be the value of Γ, and hence the smaller will be the difference between price and marginal cost.

It follows then that, regardless of whether imports are regarded as perfect or imperfect substitutes for the output of domestic firms, the price leadership model indicates that import competition (as represented by the market penetration of imports and by the import supply elasticity) serves to restrain prices. We can next ask whether alternative models of imperfect competition yield similar implications for import competition. One such model is the so-called limit pricing model.11 In this model, the dominant firm or firms in an industry are assumed to set the market price at a level low enough to discourage entry by potential competitors. Clearly this theory does not deny the hypothesis of import discipline; it simply suggests that threatened, rather than the actual, entry of imports to the domestic market constrains the price-setting behavior of domestic firms. The weakness of this theory, though, would appear to be its failure to explain the reason for observed increases of import penetration. Undoubtedly, the theory of limit pricing could be modified to include an adjustment process by which errors in domestic firms’ limit pricing strategies (i.e., observed changes in import penetration) provoke price adjustments. But in such an event, the limit pricing model would begin to approximate the price leadership model and thus would offer little advantage over the latter model.

A second, less well-developed alternative model is based on the often-heard assertion that domestic firms with price-setting power prefer to allow their market shares to be eroded by increasing imports rather than allow prices to fall. This is because domestic firms believe supply capabilities of foreign producers are limited and, hence, that inroads made by imports will likely be small, and possibly only temporary, given the pressures on import supply of cyclical foreign demand conditions. From either a short-run or long-run perspective, the denial of the import-discipline hypothesis advanced by this theory would seem deficient with respect to the traditional assumption of profit-maximizing behavior by firms. If foreign capacity to supply the domestic market is limited or temporary, the optimal response of domestic firms to increasing import competition is not to maintain a given price. The appropriate profit-maximizing response is still to adjust prices downward, in the first case, or, in the second case, to allow prices to fall only temporarily. Failure to adjust prices, by contrast, implies an undue erosion of profits in the short run and incentives for expanded production capacity abroad and further erosion of profits in the long run.

Identifying the theoretical effect of import competition on domestic price behavior via the price leadership model or otherwise is one thing. Measuring or estimating this effect is quite another.

The first significant problem for empirical work is that the parameters that capture the effects of import competition on domestic prices in equations (2) and (2′) are, for the most part, unobservable at either a disaggregated or aggregate level. Specifically, there are simply no good estimates of import supply elasticities, of domestic demand price elasticities (M′ and D′, respectively, in equation (2)), or of the elasticity of import prices with respect to domestic prices (Γ in equation (2′)). What is observable at the industry level is the import market share (MD). We therefore have to assume in our empirical work that these other import-competition parameters vary with the import market share—that is, we assume that the import supply elasticity and the domestic demand price elasticity are higher, the greater is the import share.12 One defense of this assumption of convenience is supplied by Caves ((1974), p. 6)

  • What should constrain the profits of import-competing producers is not the market share held by imports but the responsiveness of their supply to an increase in the domestic price above the competitive level. A small share ex post could be associated with an elastic supply of imports, and hence a “limit price” allowing little permanent excess profit. Considering that the industries for which we have data are at least somewhat heterogeneous, however, the successful performance of the import-share variable grows more plausible, according to the following argument: Some well-defined products (hereafter subproducts) classified to an industry have no close importable substitutes, while others are subject to high cross-elasticities with respect to the prices of imports and face varying amounts of actual import competition. The import share, as a weighted average of these situations, probably reflects the prevalence of subproducts with close importable substitutes.

A second practical problem is how to give concrete representation to the notion that the restraining effect of imports ought to depend on the market structure of the domestic industry. In other words, as implied earlier, import discipline is conditional in the sense that if the industry is purely competitive, price will already equal marginal cost and, hence, there will be no monopolistic price premium to be eliminated by import competition. Following the lead of Pugel (1980), we have assumed that the effect of import competition varies linearly with the industry’s concentration ratio (CR), where CR serves as a proxy for the ability of dominant firms in the industry to set prices above costs. Using a1 to represent the elasticity of the domestic price with respect to changes in the import market share (a1 < 0), we assume13

a1=α0+α1CR(3)

We are aware of the shortcomings of the concentration ratio as a measure of competitive forces in an industry (see, for instance, Comanor and Wilson (1967)), but it is still probably the best available proxy for the purpose at hand.

The third major problem is how to deal with the obvious simultaneous relationship between domestic price changes and import penetration changes and also with the (less obvious) possible simultaneity between domestic price changes and changes in total market demand for both the imported good and the domestic good. This problem is worthy of attention because, under plausible assumptions, it can be shown that such simultaneity will lead to a serious understatement of the effects of both import-penetration changes and changes in market demand on domestic price changes. In other words, ordinary-least-squares estimates of the effect of these variables on domestic price changes will be biased. To illustrate the nature of the problem, consider the following simple three-equation model:

ΔP=a0+a1ΔIP+a2ΔMD+a3ΔZ+1(4)
ΔIP=b0+b1ΔP++2(5)
ΔMD=c0+c1ΔP++3(6)

where, again, P denotes the domestic price, IP denotes the import share variable, MD denotes total market demand for the industry’s output (total expenditure on similar domestic and imported goods deflated by the price of domestic goods), Z denotes other factors affecting domestic prices, ε1 … ε3 are error terms, and Δ denotes the log difference operator.14 The skeletal forms of equations (5) and (6) serve only to indicate that both ΔIP and ΔMD are believed to be functions of ΔP. Specification of the remaining determinants of these two variables is left to later discussion.

Given this model, the sign of the expected bias in OLS estimates of the coefficients on ΔIP and ΔMD in equation (4)—namely, a1 and a2, respectively—will depend on the signs of the following two covariances:15

Cov(ΔIP,1)b1σ12+σ121c1a2b1a1(7)
Cov(ΔMD,1)c1σ12+σ131c1a2b1a1(8)

where σ12 is the variance of ε1 σ12 is the covariance between ε1 and ε2, and σ13 is the covariance between ε1 and ε3.

To ascertain the direction of the bias in OLS estimates requires determination of the signs of the parameters appearing in expressions (7) and (8). Of those parameters, the two for which we can offer no firm a priori expectation are the error covariances σ12 and σ13; for this reason, we have to resort to the usual assumption (e.g., Maddala (1977)) that these two covariances are zero.

From the import discipline model discussed earlier comes the expectation that a1 and a2 in the price change equation (4) will carry negative and positive signs, respectively. That is, an increase in the import share will decrease the rate of increase of domestic prices, whereas an increase in market demand for domestic and imported goods will increase the rate of price change. Because the change in import share (ΔIP) is defined in value, rather than volume, terms, we know from demand theory that b1 will be positive or negative, according to whether the elasticity of substitution between imported and domestic manufactures is greater than or less than unity. At the four-digit SIC level of disaggregation used in this paper, we expect c1 to be negative, on the grounds that the total real (deflated) market demand for a class of domestic and imported goods will fall as their domestic price increases and consumers consequently switch their demand to other classes of goods.

To sum up, if the covariance of errors across equations is assumed to be zero (σ12, σ13 = 0), if a2,b1 > 0, and if a1,c1 < 0, then the OLS estimates of the effects of changes in import penetration (a1^) and of changes in market demand (a2^) on domestic prices will both be understated.

Our response to this potentially serious simultaneity problem is to estimate the parameters of the domestic price change equation by two-stage least squares. While TSLS estimates are less efficient (i.e., have larger asymptotic variances) than various full-information estimation techniques, they do provide consistent estimates and do not require a thorough specification and estimation of all the structural equations of a simultaneous model.

In order to carry out the TSLS estimation, it is necessary to specify two or more independent variables that influence either the change in the import share (ΔIP) or the change in total market demand (ΔMD) but are not among the structural determinants of the change in domestic prices (ΔP). In selecting these independent determinants of ΔIP and ΔMD, we have relied on a standard “utility tree” notion of demand theory.16 More specifically, we posit that expenditures on goods in each industry class can be determined on the basis of total nominal expenditure and simple indices of prices of goods in each industry, whereas the market share of imports depends only on the relative price of similar domestic and foreign goods. In other words, consumers are viewed as implicitly engaging in a two-step procedure: in the first step, they decide how much to spend on the goods (both domestic and imported) of industry class i on the basis of their total income or expenditure and of the prices of these goods relative to those of other broad industry classes; in the second step, they allocate their expenditure on goods of industry class i (given from step 1) as between domestic and imported goods, solely on the basis of the relative price of imports. In symbols, this can be written as

MDi=fi(E,Φ1,Φ2,,Φn)(9)
Φi=wiPi+(1wi)PMi(10)
IPi=hi(PMi/Pi)(11)

where the subscript denotes the industry category; Pi, and PMi denote prices of similar domestic and imported manufactures, respectively, in the ith industry category; Φi denotes the overall price index for the ith industry class (with wi the weight of domestic goods in the index); and E denotes nominal expenditures on all manufactures.

To summarize the main points of this section, we have argued: (i) that the price leadership model predicts a restraining effect of an increase in import penetration on the change in domestic prices; (ii) that this restraining effect of import competition should be stronger in more concentrated domestic industries; and (iii) that an unbiased estimate of the effect of changes in import penetration on domestic prices can be obtained only if the simultaneous relationship among import penetration changes, domestic price changes, and market demand changes is accounted for and if other determinants of domestic price changes are held constant. In combination, these arguments suggest an estimating equation for domestic price changes (ΔP) of the following general form:

ΔPi=γ0+γ1ΔUCi+γ2ΔMDi^+γ3ΔIPi^+γ4CRiΔIPi^+γ5CRiΔMDi^(12)

γ1, γ2, γ5 > 0; γ3, γ4 < 0

where ΔP denotes the change in the domestic price, ΔUC denotes a proxy for the change in unit variable costs, AMD denotes the change in market demand, ΔIP denotes the change in import penetration, CR denotes the concentration ratio, the subscript i denotes the ith industry, and the superscript ^ denotes the predicted value of a variable from a first-stage regression of that variable on only the exogenous variables in the model. As in most price studies, we expect to find a positive effect of marginal cost changes and of demand changes on industry price changes—that is, γ1, γ2 >; 0. Our central concern here, however, will be to determine: (i) whether an increase in import penetration has a significant negative effect on the change in domestic prices—that is, whether (γ3 + γ4 CR) <; 0; (ii) whether this import discipline effect works more strongly, or perhaps only works at all, in more concentrated industries—that is, whether γ4 <; 0 or |γ4| >; |γ3|; and (iii) whether OLS estimates of γ2, γ3, γ4, and γ5 show less effect of ΔIP and ΔMD on ΔP than do TSLS estimates, as suggested by a two-way causation among ΔP, on the one hand, and ΔIP and ΔMD, on the other—that is, whether

TSLS OLS

i|i = 2−5 > |γi|i = 2−5

II. Empirical Results

In the preceding section, we have argued that it is not possible to identify the independent effect of changes in import competition on domestic price behavior unless “other factors” affecting industry price changes are held constant. In Table 1, we set out a series of price change regressions, all estimated by OLS, where the only explanatory variables are industry cost changes and changes in industry demand.17 In regressions (la)−(lc), the control variables are changes in full unit variable costs, AFUVC(that is, the change in unit material cost plus the change in unit labor cost), and one of three alternative proxies for the change in industry demand—namely, the change in total real market demand (ΔMD), the change in the ratio of inventories to shipments (ΔINVSM), and the change in the real value of shipments (ΔRS).18 The expectation is that ΔFUVC, ΔMD, and ΔRS should all carry positive signs, while ΔINVSM should carry a negative one. In regressions (ld)–(lf), the change in industry unit variable cost is disaggregated into the change in unit material cost (ΔUMC), the change in the wage rate of production workers (ΔWRP), the change in real output per production worker (ΔPR), and the change in the gross book value of assets in the industry (ΔTA), so as to permit the individual coefficients to vary. In all cases, the equations are estimated by OLS, and the sample size is restricted (344–356 observations) to those industries and time periods for which data are available on all relevant variables (including import prices). As shown in Table 7 in the Appendix, however, the estimated coefficients are generally quite stable as one moves from the largest possible sample (1,512–1,584 observations) to the much smaller one used here.19

Table 1.

Price Change Regressions Using Pooled Cross-Section Data for 1973–76, Ordinary-Least-Squares Estimation, and No Foreign Variablesl

article image

Figures in parentheses under regression coefficient estimates are f-ratios. In the last column, n denotes the total number of observations, while the numbers in parentheses give the number of four-digit Standard Industrial Classification (SIC) categories included in the regression. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

Without going into great detail, the main features of Table 1 are: (i) the “control variables” explain a large proportion (R2 = 0.77 to 0.82) of the interindustry variation of domestic price changes; (ii) the industry cost-change variables, whether disaggregated into primary components or not, are rather consistently significant at the 1 per cent level with the expected signs;20 and (iii) the performance of the industry demand-change variables is poor, with four of the six equations yielding demand variables that are either insignificant or wrongly signed. All in all, the equations reported in Table 1 are reasonably representative of the results obtained by other researchers (see Nordhaus (1972) for a survey) for the U. S. manufacturing sector, although we should point out that most previous empirical studies used more aggregate data.

In Table 2, we add the change in import penetration (ΔIP) to the set of explanatory variables employed in Table 1. In line with the arguments made earlier, we allow the effect of import discipline to depend on the industry’s market structure by having the change in import penetration interact with the eight-firm concentration ratio in the industry (CR · ΔIP). If the effect of import competition on domestic prices is greater in more concentrated industries, the estimated coefficient on CR · ΔIP should be negative and significant. The final amendment we have made for Table 2 (although it has little effect on the results of main interest) is to permit the effect of industry demand changes on prices to vary also with the industry concentration ratio (i.e., to allow a given demand change to have a larger effect on prices in concentrated industries where there is no presumption that price changes equal changes in marginal costs). As before, the equations are estimated by OLS and therefore ignore any simultaneity between price changes and the explanatory variables.

Table 2.

Price Change Regressions Using Pooled Cross-Section Data for 1973–76, Ordinary-Least-Squares Estimation, and Import Penetration Changes1

article image

Regressions (2a)–(2c) are based on a sample of 356 observations (89 Standard Industrial Classifications (SICs)), while regressions (2d)–(2f) are based on 344 observations (86 SICs). Figures in parentheses under regression coefficient estimates are t-ratios. A single asterisk denotes statistical significance at the 5 percent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

Three results in Table 2 are of particular importance for the purpose of this paper. First, the change in import penetration itself (ΔIP) either is insignificant (regressions (2d)–(2f)) or is significant with the wrong (positive) sign (regressions (2a)–(2c)), thereby providing no support for the traditional version of the import discipline hypothesis. Second, there is evidence of a restraining effect of import penetration changes on the rate of inflation that increases with the degree of concentration in the domestic industry, a result that supports the modified version of the import discipline hypothesis. Note that the variable CR · ΔIP always carries the hypothesized negative sign, and it is significant at the 5 per cent level in all those regressions (2a)–(2c) where costs are represented by ΔFUVC. Third, even where significant, the size of the estimated coefficients on ΔIP and CR · ΔIP are quite small, suggesting only a small influence of changes in import competition on domestic price behavior (once any effects of import competition on industry costs and industry demand are held constant). Although less relevant for the subject of this paper, the regressions in Table 2 also imply that industry demand changes have a greater effect on prices in more concentrated industries. Observe that the demand-CR variables all carry the expected signs.

The next question to be addressed is how the estimates of the effect of import competition on domestic prices would be altered if the simultaneity between domestic price changes, on the one hand, and import penetration changes and demand changes for the industry, on the other, were accounted for. Recall from Section I that under plausible assumptions, we showed that OLS estimates of the effects of both ΔIP and the demand variables on ΔP would be understated.

Table 3 provides the TSLS estimates of price change equations where the estimated values of ΔIP and the demand variables replace their actual values. Two conclusions stand out. First, although the estimated coefficient on ΔIP remains positive (i.e., wrongly signed), it is now always insignificant, never reaching a f-value greater than 1.06. Second, and more important, the estimated coefficient on CR·ΔIP increases in size (anywhere from three to five times), with five of six estimates falling in the −0.22 to −0.43 range; in all six equations in Table 3, the coefficient on CR·ΔIP carries the expected negative sign, and it is statistically significant at the 10 per cent level in four of the six regressions—(3a), (3b), (3c), and (3e). Similarly, it is interesting to note that the estimated coefficients of the interactive demand variables also increase as predicted in the TSLS estimates, although the increase is not as dramatic as that of the import penetration variable.

Table 3.

Price-Change Regressions Using Pooled Cross-Section Data for 1973–76, Two-Stage-Least-Squares Estimation, and Import Penetration Changesl

article image

Regressions (3a)–(3c) are based on a sample of 356 observations (89 Standard Industrial Classifications(SICs)), while regressions (3d)–(3f) are based on 344 observations (86 SICs). Figures in parentheses under regression coefficient estimates are r-ratios. All r-statistics are adjusted to their asymptotic values. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

The same general pattern of results emerges when the import penetration variable is expressed in volume, rather than in value, terms. As indicated in Tables 4 and 5, the TSLS estimates of CR·ΔIP are anywhere from one-and-a-half to three times larger than the OLS estimates, the estimated coefficients of CR·ΔIP always carry the expected negative sign, and they are statistically significant at the 10 per cent level or better in all six equations. The main difference is that the estimated elasticity of the domestic price change with respect to the import penetration change is much smaller in the volume version than in the value version (−0.04 to −0.07 compared with −0.14 to −0.43). In addition, this difference in elasticities is only very partially offset by the slightly larger size of the mean change in import penetration in volume terms compared with the mean change in import penetration in value terms (10.15 per cent compared with 7.89 per cent), at least for our sample industries and time period.

Table 4.

Price Change Regressions Using Pooled Cross-Section Data for 1973–76, Ordinary-Least-Squares Estimation, and Quantity Measure of Import Penetration1

article image

Regressions (4a)–(4c) are based on a sample of 356 observations (89 Standard Industrial Classifications (SICs)), while regressions (4d)–(4f) are based on 344 observations (86 SICs). Figures in parentheses under regression coefficients are f-ratios. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

Table 5.

Price Change Regressions Using Pooled Cross-Section Data for 1973–76, Two-Stage-Least-Squares Estimation, and Quantity Measure of Import Penetration1

article image

Regressions (5a)–(5c) are based on a sample of 356 observations (89 Standard Industrial Classifications (SICs)), while regressions (5d)–(5f) are based on 344 observations (86 SICs). Figures in parentheses under regression coefficient estimates are t-ratios. All r-statistics are adjusted to their asymptotic values. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

Taken together, the TSLS regression results suggest strongly that increases in import penetration do restrain inflationary price behavior in the U. S. manufacturing sector; and the restraint is greater, the higher is the degree of concentration in the domestic industry. Yet the results also suggest that this restraint or discipline is not very great for the manufacturing sector as a whole. For example, if we assume an elasticity of domestic price change with respect to CR·ΔIP of about −0.25, and if we allow CR and ΔIP to take on their mean values over the sample of 0.50 and 7.89 per cent, respectively, then we arrive at the estimate that an average increase in import penetration lowers the rate of inflation in U. S. manufacturing by about 1.0 per cent [−1.0 = (−0.25)(0.50)(7.89)], relative to what it would be in the absence of any change in import penetration. This is not a negligible contribution, but it is certainly small relative to the contributions of domestic costs and excess demand factors.

Three qualifications to this conclusion should, however, be noted. First, the contribution of import discipline to price restraint in individual manufacturing industries could be much higher than that indicated in the preceding paragraph. This is because annual changes in import penetration on the order of 10–25 per cent would be unusual for the manufacturing sector as a whole but not for individual industries.21 Second, it is important to recall that the equations presented here measure only the direct effect, or so-called sympathetic effect, of import competition on domestic prices, given the behavior of unit costs in the industry. No attempt is made to estimate the effect of, say, import price changes on unit cost changes in the industry. In other words, the full effects of import competition on domestic prices are likely to be larger than the effects estimated here.22 Third and finally, the estimates of import discipline presented in this paper—while based, we think, on a superior methodology relative to those used in earlier studies—still suffer from several problems that could mask the “true” effects of import discipline. Foremost among these problems is the relatively low explanatory power of the first-stage regressions for the change in import penetration. As shown in Table 9 in the Appendix, we were able to explain only about 5 to 7 per cent of the variance in import penetration changes across industries when the value measure was used. Fortunately, the first-stage results for the volume measure of import penetration changes (ΔRIP) were much better, carrying R2s in the 0.36 to 0.44 range.23 Also, the estimated elasticity of ΔRIP with respect to import price changes of −0.92 to −0.97 is consistent, both with earlier estimates of the elasticity of substitution between U. S. manufactures and imported manufactures and with the near-zero elasticity of import penetration changes in value terms with respect to import price changes.24 Attempts to find simple (i.e., lagged import penetration changes) instrumental variables for ΔIP and ΔRIP that had better explanatory power yielded disappointing results; for example, the correlation between ΔIPt and ΔIPt − 1 was only 0.024. Other potential problems that could mask the true discipline effect include possible nonlinearities (large versus small changes) or asymmetries (positive versus negative changes) in the effects of ΔIP on domestic prices, weaknesses of the concentration ratio as a measure of the competitive market structure in the industry, and inadequacies of the unit-value-type price indices used for import prices.

III. Conclusions

Much of the recent debate in the United States and elsewhere about policies that shelter domestic industries from import competition has focused on the actual or potential employment losses in industries facing stiff and increasing import competition. Less attention has been paid to another important objective of public policy—namely, control of inflation. Clearly, to the extent that empirical evidence is lacking on the effects of import competition on domestic price behavior, it makes it easier for policymakers to discount the importance of price effects in weighing such policies.

In this paper, we have used a pooled cross-section of data on U. S. domestic manufacturing industries and on U. S. imports to test the disciplinary effect of changes in import competition on domestic prices. We have used changes in import penetration to represent changes in import competition under constant domestic factor costs and constant market demand conditions,25 and have taken explicit account of the two-way causation between domestic price changes, on the one hand, and changes in both import penetration and market demand, on the other. Our principal conclusions can be summarized as follows:

(1) Increases in import penetration, ceteris paribus, do seem to have a negative (i.e., restraining) effect on the domestic inflation rate in the U.S. manufacturing sector. It thus appears that import competition not only affects profit rates in U. S. industries (see Pugel (1980) and Marvel (1980)) but price behavior as well—and, in both cases, in the expected direction.

(2) The disciplinary effects of import competition are, however, conditional upon the market structure of the domestic industry. More specifically, import discipline is stronger in highly concentrated industries than in less concentrated (presumably more competitive) ones.

(3) Relative to influences of unit cost changes and excess demand conditions on domestic price changes, the effect of changes in import penetration is small Nonetheless, our estimates strongly support the case for a liberal trade policy. To pick an example of immediate policy interest, consider the case of the highly concentrated U. S. auto industry and the associated recent pleas to protect it by introducing quantitative restrictions that would roll back the market share of imports to the pre 1978 level of about 20 per cent (versus the 1980 level of roughly 25 per cent). Given an eight-firm concentration ratio of unity, if we assume that the elasticity of domestic prices with respect to changes in import penetration is the same for autos as for all manufacturing (i.e., −0.25) and that the estimated decrease in the import market share is 20 per cent, we obtain an estimated increase of approximately 5 per cent in domestic auto prices [5.0 = (−0.25)(1.0)(−20.0)]. This is by no means a small number, given the fact that the U. S. producer price index for passenger autos increased by 7.6 per cent in 1979.

(4) The estimated effect of changes in import penetration on domestic prices is reasonably robust with respect to alternative specifications of the control variables (unit cost changes and demand factors) in the equation, but is quite sensitive to both the method of estimation (i.e., OLS versus TSLS) and the definition of import penetration itself (i.e., volume versus value terms). Specifically, the estimated effects are considerably larger when the TSLS method is used and when import penetration is measured in value terms. Also, the statistical significance of the import discipline coefficient tends to vary considerably across different specifications of the basic equation, but it usually is significant at the 10 per cent level in the TSLS estimates.

APPENDIX: Data Base and Definition of Variables

Our data consist of annual observations on U. S. imports and on prices, production, and costs of U. S. manufacturing industries by four-digit Standard Industrial Classification (SIC) categories over the period 1972–76. These data are taken from three primary sources. From the Annual Survey of Manufactures 1976 (United States, Department of Commerce (1978)) and the Census of Manufactures 1972 (United States, Department of Commerce (1975)), we the value of shipments, inventories, capital stock, and wages and man-hours of production workers. From the U.S. Bureau of Labor Statistics, we obtained on computer tape the price indices for output by domestic firms that the Bureau uses to compute its published producer price series. Finally, from the U. S. Bureau of the Census, we obtained (also on computer tape) computations of the value and volume of U. S. imports by SIC category. A descriptive summary of the variables finally employed to estimate our model is presented in Table 6.

Though for many variables we have observations across all 450 four-digit industry categories (as seen in Table 6), the coverage of several variables is quite limited by comparison. In particular, the necessity of using data on the volume of imports (to measure import prices) severely reduces the effective size of the data base. Indeed, there are at most only 95 categories for which we have import quantity data.

Table 6.

Data Base

article image

SIC is an abbreviation for Standard Industrial Classification.

ASM = Annual Survey of Manufactures 1976 (United States, Department of Commerce (1978)); CM = Census of Manufactures, 1972 (United States, Department of Commerce (1975)); BLS = Bureau of Labor Statistics (computer tape); CB = United States, Bureau of the Census (computer tape); and IFS = International Monetary Fund, International Financial Statistics.

wj=VSj/jiVSj

Table 7.

Price Change Regressions Using Pooled Cross-Section Data for 1973–76, Ordinary-Least-Squares Estimation with Alternative Sample Sizes1

Figures in parentheses under regression coefficient estimates are t-ratios. In the last column, n denotes the total number of observations, while the numbers in parentheses give the number of four-digit Standard Industrial Classification (SIC) categories included in the regression. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

Attributable to rounding.

The import quantity data are limited for two reasons. First, the Census Bureau statistics often combine more than one four-digit SIC category into a single group. This is because import statistics are originally collected on a goods basis, whereas firms in more than one SIC category often produce many of the same products. We have not restricted our data base to those categories of imports for which only a single four-digit SIC code is identified, however. Instead, we have retained the composite import categories by simply assuming that the composite categories are represented by the primary four-digit SIC code identified by the Census Bureau in each case.

Second, the import quantity data are limited, because even when classified by four-digit SIC groups, imports are still quite heterogeneous; and therefore, the Census Bureau does not attempt to measure physical quantities for most groups.

The final number of four-digit SIC categories in our TSLS sample is less than 95. Only 86 categories are found by the intersection across SIC groups of the data used in our regression analysis. Despite this small number, it must be remembered that our data base on changes of variables spans four years. Thus, our pooled cross-section constitutes a sample of 344 observations (86 categories × 4 years). Further, it should be noted that the 86 categories span the complete range of 2-digit categories of manufactures.

Several of the individual variables posed particular specification problems. To begin with, measuring import penetration with precision requires matching not only import data but also export data to four-digit SIC manufacturing shipments data in order to adjust shipments to reflect domestic consumption of domestic manufactures. To reduce the problem of reconciling trade and shipments data, we have chosen to represent import penetration as simply the ratio of imports to shipments. Our assumption is that the price behavior of U. S. firms is affected equally by the scale of import competition to U. S. firms’ total shipments and by the scale of import competition to their shipments solely to the domestic market. A similar problem was that consistent data on import duties paid were not available for the whole 1972–76 period; nor was it possible to obtain data on the cost of transporting domestic goods from the point of shipment to the final user. Ideally, the numerator of the import penetration variable should include import duties paid, while the denominator should include domestic transportation charges. We were able to take some comfort from the fact that for the two years (1975 and 1976) for which data on import duties by four-digit SIC industries were available, the correlation between ΔIP inclusive and exclusive of duties was quite high (in excess of 0.98).

Obtaining good instrumental variables to represent “other prices” in the first-stage regression equations for ΔIP and ΔMD also presented problems. Obviously, given the large number (86) of four-digit industries in our sample, they could not all be used as instruments without critically reducing the degrees of freedom of the estimating equation. Our solution was to employ indices of domestic producers’ prices by two-digit manufacturing industries to define two price series.26 The first variable, P2, is the two-digit price index of which the given four-digit industry is a part—for example, the two-digit category, primary metal products (SIC 33), would be the relevant P2 for both of the four-digit subcategories smelter and refined copper (SIC 3331) and steel foundries (SIC 3323). This variable is assumed to represent the prices of closely related four-digit industries. The second variable, P0, is an index of prices in other two-digit categories (i.e., nonrelated four-digit industries).

In some specifications of the basic domestic price-change equation (12), we represented the change in the industry’s costs by the changes in the prices of variable factor inputs and by the stock of productive physical capital (i.e., see equation (2)).27 For the latter, we used the gross book value of assets in the industry. This proxy is deficient because it is based on historical costs, which understate current value, and also because it includes accumulated depreciation of assets, which overstates current value. Book value remains, however, the only convenient representation of industry stocks of capital at the four-digit level of disaggregation. As regards factor input prices, good wage-rate data are available at the four-digit level, but this is not so for material input prices. Data are available on unit material costs, so that changes in material input prices can be obtained only under the assumption that material requirements per unit of manufacturing output remained constant over the 1972–76 period.

Finally, we have used U.S. gross national product rather than expenditures on all manufactures as the “scale” or “income” variable in the first-stage regression equation for changes in total market demand (ΔMD). This reduces the collinearity between manufacturing prices and expenditure, and is still consistent with the two-step demand model outlined earlier.

Table 8.

Ordinary-Least-Squares Regressions for Changes in Demand Variables Using Pooled Cross-Section Data for 1973–761

article image

Figures in parentheses under regression coefficient estimates are t-ratios. In the last column, n denotes the total number of observations, while the numbers in parentheses give the number of four-digit Standard Industrial Classification (SIC) categories included in the regression. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

Table 9.

Ordinary-Least-Squares Regressions for Changes in Import Penetration Using Pooled Cross-Section Data for 1973–76: Comparison of Value and Volume Measures of Import Penetration1

article image

Figures in parentheses under regression coefficient estimates are t-ratios. In the last column, n denotes the total number of observations, while the numbers in parentheses give the number of four-digit Standard Industrial Classification (SIC) categories included in the regression. A single asterisk denotes statistical significance at the 5 per cent level. A double asterisk denotes statistical significance at the 1 per cent level. SEE denotes standard error of the estimate.

Estimate is multiplied by 100.

REFERENCES

  • Ahluwalia, Isher J., and Ernesto Hernández-Catá, “An Econometric Model of U. S. Merchandise Imports Under Fixed and Fluctuating Exchange Rates, 1959–73,” Staff Papers, Vol. 22 (November 1975), pp. 791824.

    • Search Google Scholar
    • Export Citation
  • Caves, Richard E., International Trade, International Investment, and Imperfect Markets, Special Papers in International Economics, No. 10, International Finance Section, Department of Economics, Princeton University (November 1974).

    • Search Google Scholar
    • Export Citation
  • Clayman, Jacob, “Exploding the ‘Inflationary’ Myth,” Viewpoint, Vol. 9 (Spring 1979), pp. 15.

  • Comanor, William S., and Thomas A. Wilson, “Advertising Market Structure and Performance,” Review of Economics and Statistics, Vol. 49 (November 1967), pp. 42340.

    • Search Google Scholar
    • Export Citation
  • Deppler, Michael C., and Duncan M. Ripley, “The World Trade Model: Merchandise Trade,” Staff Papers, Vol. 25 (March 1978), pp. 147206.

    • Search Google Scholar
    • Export Citation
  • Esposito, Louis, and Frances F. Esposito, “Foreign Competition and Domestic Industry Profitability,” Review of Economics and Statistics, Vol. 53 (November 1971), pp. 34353.

    • Search Google Scholar
    • Export Citation
  • Goldstein, Morris, and Mohsin S. Khan, “Large Versus Small Price Changes and the Demand for Imports,” Staff Papers, Vol. 23 (March 1976), pp. 200225.

    • Search Google Scholar
    • Export Citation
  • Gregory, R. G., “United States Imports and Internal Pressure of Demand: 1948–68,” American Economic Review, Vol. 61 (March 1971), pp. 2847.

    • Search Google Scholar
    • Export Citation
  • Hadar, Joseph, Mathematical Theory of Economic Behavior (Reading, Massachusetts, 1971).

  • Hooper, Peter, “The Stability of Income and Price Elasticities in U. S. Trade, 1957–77,” International Finance Discussion Paper, No. 119, Board of Governors of the Federal Reserve System (June 1978).

    • Search Google Scholar
    • Export Citation
  • Hooper, Peter, and Barbara Lowrey, “Impact of the Dollar Depreciation on the U. S. Price Level: An Analytical Survey of Empirical Estimates,” International Finance Discussion Paper, No. 128, Board of Governors of the Federal Reserve System (January 1979).

    • Search Google Scholar
    • Export Citation
  • Jones, J. C. H., L. Laudadio, and M. Percy, “Market Structure and Profitability in Canadian Manufacturing Industry: Some Cross-Section Results,” Canadian Journal of Economics, Vol. 6 (August 1973), pp. 35668.

    • Search Google Scholar
    • Export Citation
  • Krause, Lawrence B., “Import Discipline: The Case of the United States Steel Industry,” Journal of Industrial Economics, Vol. 11 (November 1962), pp. 3347.

    • Search Google Scholar
    • Export Citation
  • Maddala, G. S., Econometrics (New York, 1977).

  • Mancke, Richard, “The Determinants of Steel Prices in the U. S.: 1947–65,” Journal of Industrial Economics, Vol. 16 (April 1968), pp. 14761.

    • Search Google Scholar
    • Export Citation
  • Marvel, H. P., “Foreign Trade and Domestic Competition,” Economic Inquiry, Vol. 18 (January 1980), pp. 10322.

  • Murray, T., and Ginman P., “An Empirical Examination of the Traditional Aggregate Import Demand Model,” Review of Economics and Statistics, Vol. 58 (February 1976), pp. 7580.

    • Search Google Scholar
    • Export Citation
  • Nordhaus, William D., “Recent Developments in Price Dynamics,” in The Econometrics of Price Determination, Conference October 30–31, 1970, Washington, D.C., ed. by Otto Eckstein, Board of Governors of the Federal Reserve System and the Social Science Research Council (June 1972), pp. 1649.

    • Search Google Scholar
    • Export Citation
  • Pugel, Thomas A., “Foreign Trade and US Market Performance,” Journal of Industrial Economics, Vol. 29 (December 1980), pp. 11929.

    • Search Google Scholar
    • Export Citation
  • Scherer, F. M., Industrial Market Structure and Economic Performance (Chicago, 1970).

  • United States, Department of Commerce, Bureau of the Census, Annual Survey of Manufactures 1976, Industry Profiles (1972–76), Series M76(AS)-7 (Washington, 1978).

    • Search Google Scholar
    • Export Citation
  • United States, Department of Commerce, Bureau of the Census, Census of Manufactures, 1972: Concentration Ratios in Manufacturing (Washington, 1975).

    • Search Google Scholar
    • Export Citation

SUMMARIES

Effects of Central Bank Intervention in the Foreign Exchange Markethans genberg (pages 451–76)

The article surveys theoretical issues and empirical evidence on the effects of central bank intervention in foreign exchange markets on exchange rate movements. The focus is on the ability of fully sterilized intervention to influence exchange rates in a predictable manner.

Theoretical considerations suggest that the exchange rate may be affected by intervention if assets denominated in different currencies are imperfect substitutes, thus creating opportunities for intervention (undertaken for portfolio balance reasons) to modify interest differentials and, hence, the exchange rate. Alternatively, intervention may have a systematic influence on expectations of future monetary policy and, therefore (as a result of speculation), on the current exchange rate. The empirical evidence on either of these factors is flawed because of inherent difficulties in measuring expectations, on the one hand, and in obtaining good data on the currency composition of intervention and asset stocks, on the other. Evidence drawn from estimates of exchange rate equations does not, in general, confirm the existence of significant and stable portfolio balance effects. The same conclusion emerges when one looks at the effects of intervention on risk premiums in the foreign exchange markets, on differences between offshore and domestic interest rates, and on covered international interest differentials.

The lack of any firm evidence that portfolio balance effects resulting from intervention in foreign exchange markets affect exchange rates predictably suggests that intervention policy cannot be distinguished from general monetary policy for practical purposes. This, in turn, implies that attempts to use these two policy instruments independently may lead to lesser, rather than to greater, exchange rate stability.

Incentives for International Currency Diversification by U. S. Financial Investorsgeorge m. von furstenberg (pages 477–94)

In recent years, official holdings of foreign exchange increasingly have been diversified by currency. If the desire for such diversification were also to extend to nationals without significant foreign payments obligations, traditional monetary relationships might be disturbed and the manageability of national economies by their own measures might be reduced. For this reason, it is important to ascertain whether U. S. investors in liquid assets might have reason to expect to realize higher rates of return upon reconversion into dollars if they invested in currencies other than the dollar.

Using a mean-variance framework of analysis, the study finds that while higher rates of return would, in fact, have been realized if U. S. investors had chosen to diversify their liquid asset portfolios to some extent by certain rules, they would not have obtained sufficient compensation for the risk they would have run by doing so. The high variability in real rates of return that U.S. residents are bound to experience when seeking to increase portfolio returns by going through other currencies thus provides a protective barrier around the dollar; similarly, such variability provides a protective barrier around other national currencies that prevents home residents without significant foreign payments obligations from exploiting expectations based on the superior performance (unanticipated appreciation) of other currencies in the past. Speculations based on the extrapolation of past differences in the real rates of return realized on different currencies, with rates expressed in dollars of constant purchasing power in the United States, thus would not have proved sufficiently profitable to warrant the additional risk of holding foreign monies or liquid assets.

Monetary Stabilization With and Without Government Credibilityjacques r. artus (pages 495–533)

This paper presents an analytical and empirical study of the effects of monetary stabilization programs on prices, output, and exchange rates in major industrial countries. The Federal Republic of Germany is taken as a case study. The focus is on an investigation of: (i) the rational-expectations view that the monetary authorities can curb inflation rapidly at little cost in terms of lost output and unemployment by reducing money growth and simultaneously convincing private market participants that the reduction in money growth will be long-lived; and (ii) the asset-market view that a monetary stabilization program tends to cause an appreciation of the real exchange rate that speeds up price adjustment in the labor and goods markets, by lowering prices of imported goods, but also worsens the foreign trade performance of the country.

The analysis uses a simple monetarist model in which the long-run inflationary expectations of private market participants are assumed to reflect the monetary policy of the authorities. Contrary to existing empirical work in this field, the model makes allowance for discretionary changes in the policy stance of the monetary authorities. It also allows for the existence of contract prices in the goods and labor markets and views the interest rate and the exchange rate as variables that are determined in the financial asset markets in the long run. Following econometric estimation of its parameters for the Federal Republic of Germany, the model is used to simulate the effects of a monetary stabilization program under two alternative assumptions—in one case the program affects money growth expectations, and in the other it does not. The usual caveats concerning econometric results apply to the results obtained.

The study shows that even if the sudden adoption of a new restrictive monetary policy is successful in changing the money growth and inflationary expectations of private market participants, the policy is still likely to have a major and persistent negative effect on the level of economic activity. In fact, in the present study, the successful impact of the new policy on expectations is seen not only as the necessary condition for the policy’s ultimate success in restoring the noninflationary conditions necessary for sustained economic growth in the long run but also as the major source of recessionary tendencies during the first few years. The marked initial fall in output in the Federal Republic of Germany occurs because of the adjustment in expectations, not in spite of it. A major reason for this effect is the existence of long-term contractual wage arrangements. Even when such contracts tend to cover only a one-year period, as in the Federal Republic of Germany, they may delay wage-price adjustment sufficiently when there is a sudden reduction in the money growth rate to cause a major recession.

As to the role of the exchange rate, the hypothesis that monetary stabilization causes exchange-rate overshooting in comparison with the inflation rate differential is confirmed by the present study in the case in which money growth expectations adjust. Quantitatively, however, the implications of the exchange rate overshooting for both output and inflation seem to be minor, at least for the Federal Republic of Germany. This is mainly because import prices per se do not seem to have a major effect on the rate of inflation, in terms of the gross domestic product (GDP) deflator, unless the variation in import prices is extremely large, as it was in 1973–75. Further, the short-run foreign trade price elasticity is too low to allow exchange rate movement to have a marked effect on output through its effects on foreign trade. Exchange rate overshooting is nevertheless large and persistent and may, therefore, be a significant source of concern to national and foreign authorities, if only because it results for a number of years in improved terms of trade for the country that adopts a restrictive monetary policy, while it also leads to a reduction in that country’s market share in volume terms.

The Case For and Against “Disequilibrium” Moneywilliam h. white (pages 534–72)

The customary demand-for-money functions, such as those used in setting money growth targets, are criticized for providing lags in the adjustment of the demand for money to changes in variables like income while overlooking lags in the nonbank private sector’s attempt to reverse undesired exogenous or unexpected changes in the money supply. The econometric evidence for temporary passive buffer stock reactions to undesired money is, however, found to be unacceptable. In addition, scope for significant undemanded amounts of money is restricted or eliminated when the normal forms of money change—which directly change interest rates, real income, and prices—displace the typical illustration of “windfall” money. (Of course, the combination of small immediate effects of money growth on income plus a low unlagged income-demand elasticity can, on average, account for little of the current money changes observed; but the lagged effects of earlier income changes on current demand for money are available to account, on average, for most of the rest.) Moreover, the initial efforts to eliminate any surviving undemanded change, which are only partial, tend to restore positions on the demand curve immediately by inducing changes in interest and prices; for these latter, the lag in portfolio readjustment owing to transactions costs is bypassed, and the given portfolio is converted into the desired one via the high long-run demand elasticity. Furthermore, businesses in the countries where overdraft credit is unavailable can eliminate surplus cash almost as quickly and cheaply as businesses having overdraft credit available to them. And active money markets; certificates of deposit; and easily repaid, high-cost consumer credits have in the last decade yielded—even in some developing countries—fairly quick adjustments for nonborrowing businesses and households.

Additional behavioral relationships that are sometimes proposed to justify buffer-stock behavior (i.e., relevance of the slowly responding expected interest rate instead of the actual rate, liquidity preference shifts when interest changes unexpectedly) seem illogical or quantitatively unimportant. Nevertheless, corrected models and data purged of biases might still demonstrate a modest amount of buffer-stock behavior.

International Trade of Multinational Corporations and Its Responsiveness to Changes in Aggregate Demand and Relative Pricesdavid j. goldsbrough (pages 573–99)

A substantial proportion of international trade now consists of direct intra-firm transactions between various affiliates of the same multinational firm. The size and direction of such trade flows are determined by choice-of-location decisions within the firm. The growing share of such internalized transactions in total trade may have important implications for the response of a country’s trade flows to changes in economic activity and relative competitiveness. This paper tests whether the response of such internal trade flows to shifts in aggregate demand and relative prices differs from the response of more conventional trade. A number of a priori reasons for expecting a lower underlying price elasticity for intrafirm trade are considered. The econometric implications of using artificial transfer prices are also discussed, and it is shown that they generally cause a downward bias in estimated price elasticities.

The empirical section of the paper discusses the fitting of a simplified version of the Fund’s World Trade Model to data on the transactions of U. S. multinational companies and their overseas affiliates and comparing the results with those obtained for conventional trade flows. Given the nature of the available data, it is only possible to estimate the model for U. S. manufacturing imports from overseas affiliates and for exports from U.S. affiliates in each of the major industrial countries. No adequate time series exist on exports from U. S. parents to their overseas affiliates. The results show that while there is no significant difference in the response of intrafirm and conventional trade to shifts in aggregate demand, the estimated price responsiveness of intrafirm trade is significantly lower. In general, the normal trade model fits poorly for intrafirm trade flows, and a somewhat different model, based on a choice-of-location framework, may be called for.

Import Discipline in the U. S. Manufacturing Sectordean a. derosa and morris goldstein (pages 600–634)

This paper investigates empirically the proposition that imports provide a competitive constraint, or discipline, on the price-raising ability of domestic producers. Estimates of the effect of changes in import competition (measured as changes in the ratio of the value of imports to the value of domestic shipments) on the domestic pricing behavior of U. S. manufacturing industries are presented for the 1972–76 period using data disaggregated by four-digit Standard Industrial Classification (SIC) industry categories, so that the domestic price effects of import competition can be studied across industries as well as over time.

The study differs from earlier empirical investigations of import discipline in three major respects. First, changes in industry factor input costs and changes in demand for the industry output are included in the price-change equations as control variables, so that the independent effect of changes in import competition on prices can be identified. Second, the study treats the pricing discipline of imports as dependent on the competitive market structure of the industry. And finally, the study employs two-stage-least-squares estimation methods to overcome the simultaneous relationship between changes in domestic prices, on the one hand, and changes in import penetration and domestic demand, on the other.

The main empirical finding of the study is that increases in import competition do have a significant restraining effect on the inflation rate in U. S. manufacturing industries; and the effect is greater, the higher the level of concentration in the industry. Although the influence of import discipline is found to be small relative to the influence of cost and demand factors in price determination, the estimates still suggest that policymakers should not disregard the domestic price implications of various policies designed to protect U. S. manufacturing industries from import competition.

RESUMES

Effets de l’intervention de la banque centrale dans les marches des changeshans genberg (pages 451–76)

Cet article examine les questions théoriques et les preuves empiriques relatives aux effets qu’exerce sur les mouvements du taux de change l’intervention de la banque centrale dans les marchés des changes. Il fait le point sur le pouvoir d’une intervention totalement stérilisée d’influencer sur les taux de change de manière prévisible.

Des considérations théoriques donnent à penser que le taux de change peutêtre influencé par l’intervention si les actifs libellés en diverses monnaies sont des substituts imparfaits, créant ainsi les occasions d’intervention visant à modifier les différentiels d’intérêt et, par conséquent, le taux de change par suite du “rééquilibrage” du portefeuille par le secteur privé. Autrement, l’intervention peut exercer une influence systématique sur les anticipations de politique monétaire future et, done, (par suite de la spéculation) sur le taux de change courant. La preuve empirique dégagée de l’un ou l’autre de ces facteurs est boiteuse en raison des difficultés inhérentes, d’une part, à la mesure des anticipations et, d’autre part, à l’obtention de données péremptoires sur la composition monétaire de l’intervention et les stocks d’actifs. En général, la preuve à partir des estimations d’équations de taux de change ne confirme pas l’existence d’effets significatifs et durables sur l’équilibre du portefeuille. On arrive à la même conclusion quand on observe les effets de l’intervention sur les primes de risque dans les marchés des changes, sur les différences entre taux d’intérêt offshore et taux d’intérêt intérieurs, et sur les différentiels d’intérêt internationaux couverts.

L’absence de toute indication solide que les effets d’équilibre du portefeuille par suite de l’intervention dans les marchés des changes ont une incidence prévisible sur les taux de change méne à penser que la politique d’intervention ne saurait être distincte de la politique monétaire générate aux fins d’application. Il s’ensuit, par ricochet, que les tentatives d’utilisation indépendante de ces deux instruments peuvent se solder par une diminution plutôt qu’une augmentation de la stabilité du taux de change.

Peut-on inciter les investisseurs financiers des Etats-Unis à diversifier la composition de leurs placements en devises?george m. von furstenberg (pages 477–94)

Depuis quelques années, les autorités nationales diversifient de plus en plus les avoirs officiels en devises. Si les ressortissants d’un pays donné, qui n’ont pas d’importantes obligations de paiements extérieurs, suivaient leur exemple, les relations monétaires traditionnelles pourraient s’en trouver perturbées et l’efficacité des mesures économiques que prendrait le pays en être diminuée. Il importe done de déterminer si les investisseurs des Etats-Unis qui placent des avoirs liquides ne seraient pas fondés à espérer obtenir des taux de rendement plus élevés lors de la reconversion en dollars de leurs placements en devises.

Sur la base d’un cadre d’analyse fondée sur la moyenne et sur la variance des taux de change, l’étude établit que, s’il est vrai que les investisseurs auraient en fait bénéficié de taux de rendement plus élevés s’ils avaient choisi de diversifier leurs avoirs liquides dans une certaine mesure, conformément à certaines règies, ils n’auraient pas été suffisamment dédommagés pour le risque ainsi couru. La forte volatilité des taux de rentabilité réels à laquelle les résidents des Etats-Unis s’exposeront nécessairement lorsqu’ils chercheront à accroître la rentabilité de leurs portefeuilles en opérant en autres devises est done, pour le dollar, une barrière de protection; de même, cette variabilité protège les monnaies des autres pays en empêchant les ressortissants de ceux-ci, qui n’ont pas d’obligations de paiements extérieurs importantes, d’exploiter les anticipations fondées sur le meilleur comportement d’autres monnaies (appréciation non anticipée), dans le passé. Les spéculations fondées sur l’extrapolation de différences enregistrées dans le passé entre les taux réels de rendement obtenus avec diverses devises (taux exprimés en dollars à pouvoir d’achat constant aux Etats-Unis) ne se seraient done pas révélées suffisamment rentables pour justifier le risque additionnel que présente la détention de monnaies étrangères ou d’avoirs liquides extérieurs.

Stabilisation monétaire avec ou sans crédibilité gouvernementalejacques r. artus (pages 495–533)

L’auteur procède à une étude analytique et empirique des effets des programmes de stabilisation monétaire sur les prix, la production et les taux de change dans les grands pays industrialisés. Il utilise comme exemple le cas de la République fédérate d’Allemagne. La recherche examine en particulier la validité des deux hypothèses suivantes: i) dans l’optique de la formation rationnelle des anticipations, les autorités monétaires peuvent juguler rapidement l’inflation sans réduire sensiblement la production et l’emploi en abaissant le taux de croissance de la masse monétaire et en donnant aux agents économiques privés de bonnes raisons de penser que cette baisse sera durable; ii) dans l’optique de l’importance que revêtent les marchés financiers pour la détermination à court terme des taux de change, le programme de stabilisation monétaire tend, d’une part, à provoquer une appréciation du taux de change réel qui accélère l’ajustement des prix sur les marchés du travail et des biens, en abaissant le prix des biens importés, mais il entraîne, d’autre part, une détérioration des résultats du commerce extérieur du pays considéré.

L’analyse utilise un modèle monétariste simple dans lequel, par hypothèse, les anticipations inflationnistes à long terme des agents économiques privés reflètent la politique monétaire des autorités. Contrairement aux études empiriques déjà effectuées dans ce domaine, le modèle prend en considération les modifications discrétionnaires que les autorités monétaires apportent à leur politique. Il prend également en compte le fait qu’il existe des prix fixés par accord sur les marchés des biens et du travail et il considère le taux d’intérêt et le taux de change comme des variables déterminées sur le marché des actifs financiers à long terme. Une fois que l’estimation économétrique des paramètres a été effectuée pour la République fédérate d’Allemagne, le modèle permet de simuler l’incidence d’un programme de stabilisation monétaire dans deux hypothèses différentes: dans la première, le programme a une incidence sur les anticipations en matière de croissance de la masse monétaire; dans la seconde, il n’en a pas. Les réserves que l’on formule normalement pour les études économétriques s’appliquent à ces simulations.

L’étude fait ressortir que, même si elle permet de modifier la croissance de la masse monétaire et les anticipations inflationnistes des agents économiques privés, l’adoption soudaine d’une nouvelle politique monétaire restrictive aura probablement une incidence négative considérable et durable sur le niveau de l’activité économique. En fait, dans la présente étude, l’incidence positive de la nouvelle politique sur les anticipations est considérée non seulement comme une condition sans laquelle la nouvelle politique ne parviendra pas, en fin de compte, à rétablir le climat non inflationniste indispensable à une croissance économique soutenue en longue période, mais aussi comme la principale source de tendances récessionnistes pendant les toutes premières années de son application. S’il se produit, au départ, une chute sensible de la production en République fédérate d’Allemagne, c’est précisément en raison de l’ajustement des anticipations et non malgré lui. L’une des principales causes de ce phénomène est l’existence de conventions collectives à long terme. Même lorsqu’elles sont d’une duree d’un an, comme c’est le cas en République fédérate d’Allemagne, ces conventions risquent—si le taux de croissance de la monnaie vient à baisser brusquement—de retarder suffisamment l’ajustement salaires-prix pour provoquer une profonde récession.

Pour ce qui est du rôle du taux de change, la présente étude corrobore l’hypothèse selon laquelle la stabilisation monétaire provoque une surréaction du taux de change par rapport au différentiel de taux d’inflation lorsque les anticipations relatives à la croissance monétaire s’ajustent. Quantitativement, toutefois, les conséquences que cette surréaction du taux de change entraîne à la fois pour la production et pour l’inflation semblent mineures, du moins dans le cas de la République fédérate d’Allemagne. Cela tient principalement à ce que les prix à l’importation ne semblent pas, par eux-mêmes, avoir une incidence considérable sur le taux d’inflation, mesuré par le déflateur du produit intérieur brut (PIB), sauf si leurs variations sont très fortes, comme ce fut le cas en 1973–75. En outre, à court terme, l’élasticité-prix des échanges extérieurs est trop faible pour que les fluctuations du taux de change aient de profondes répercussions sur la production, du fait de leur incidence sur le commerce extérieur. La surréaction du taux de change est néanmoins considérable et persistante; elle risque done d’être un grave sujet d’inquiétude à la fois pour les autorités nationales et pour celles des autres pays, du seul fait qu’elle provoque pendant un certain nombre d’années une amélioration sensible des termes de l’échange du pays qui adopte une politique monétaire restrictive, tout en ay ant pour effet de réduire sa part du marché en termes de volume.

Pour et contre un stock de monnaie en déséquilibrewilliam h. white (pages 534–72)

On reproche aux fonctions traditionnelles de la demande de monnaie, telles que celles qui servent à fixer les objectifs de croissance monétaire, de prendre en compte les retards avec lesquels la demande de monnaie s’ajuste aux variables changeantes, comme le revenu, tout en négligeant les retards qui caractérisent les efforts que déploie le secteur privé non bancaire pour inverser des variations exogènes non souhaitées ou des variations inattendues de la masse monétaire. Toutefois, l’étude économétrique des réactions passives du type stock-tampon à la monnaie non souhaitée donne des résultats jugés inacceptables. En outre, les possibilités d’apparition de montants de monnaie non souhaités mais significatifs sont limitées ou éliminées lorsque les variations habituelles de la masse monétaire—qui agissent directement sur les taux d’intérêt, le revenu réel et les prix—faussent l’exemple classique de monnaie “tombée du del”. (Il va de soi que la combinaison d’une faible incidence immédiate de la croissance monétaire sur le revenu et d’une faible élasticité instantanée revenu-demande ne peut, dans l’ensemble, expliquer qu’une petite partie des variations observées de la demande courante de monnaie; mais la plupart des autres variations inexpliquées peuvent généralement être imputées aux effets décalés des variations antérieures du revenu sur la demande courante de monnaie.) En outre, les efforts initiaux visant à éliminer toute variation restante non recherchée, efforts qui ne sont que partiels, tendent à rétablir les positions sur la courbe de la demande en provoquant immédiatement des variations des taux d’intérêt et des prix; pour ces derniers, on évite le retard avec lequel s’opère le réajustement de portefeuille en raison du coût des transactions, et le portefeuille effectif devient le portefeuille recherché par le jeu de la forte élasticité de la demande à long terme. En outre, dans les pays où le découvert bancaire n’existe pas, les entreprises peuvent se débarrasser de leur excédent de liquidité presque aussi rapidement, et presque à aussi bon compte, que les entreprises qui disposent d’un tel découvert. Les marchés monétaires actifs, les certificats de dépôt et les crédits à la consommation, facilement remboursés, ont, au cours de la décennie passée, produit—même dans certains pays en développement—des ajustements assez rapides dans le cas d’entreprises et de ménages ne recourant pas à l’emprunt.

Les autres relations de comportement que l’on avance parfois à l’appui d’un comportement du type stock-tampon (par exemple la pertinence de taux d’intérêt attendus qui réagissent lentement, au lieu de taux effectifs, le retour à la préférence pour la liquidité quand les taux d’intérêt varient de façon inattendue) apparaissent illogiques ou quantitativement négligeables. Néanmoins, des modèles corrigés et des données non biaisées pourraient encore démontrer l’existence d’une légère incidence du comportement du type stock-tampon.

Les échanges internationaux des sociétés multinationals et leur sensibilité aux variations de la demande globale et des prix relatifsdavid j. golds-brough (pages 573–99)

Les échanges internationaux qui s’effectuent de nos jours prennent en grande partie la forme d’opérations directes intra-société, c’est-à-dire d’opérations entre différentes filiales d’une même société multinationale. L’ampleur et l’orientation de ces courants d’échange sont déterminées par les décisions prises au sein de la société sur le choix du lieu de production. Etant donné la place de plus en plus grande qu’elles occupent dans le total des échanges, ces transactions intra-société peuvent avoir d’importantes conséquences pour la façon dont les flux commerciaux d’un pays réagissent aux variations de l’activité économique et de la compétitivité relative. La présente étude a pour objet de déterminer si la réaction de ces flux commerciaux intra-société aux variations de la demande globale et des prix relatifs diffère de celle des échanges de type traditionnel. Après avoir examiné un certain nombre de raisons “a priori” incitant à croire que l’élasticité-prix sous-jacente sera normalement inférieure dans le cas des échanges intra-société, elle donne un aperçu des conséquences économétriques qu’entraîne l’utilisation de prix de transfert artificiels en précisant que ces prix ont tendance à fausser en baisse les estimations de l’élasticité-prix.

Dans la section empirique de l’étude, on adapte une version simplifiée du modèle du commerce mondial du Fonds aux données sur les transactions de sociétés multinationales américaines et de leurs filiales à l’étranger et on compare les résultats que donne le modèle pour les opérations intra-société à ceux qu’on obtient pour les courants d’échange traditionnels. Etant donné la nature des données disponibles, le modèle ne peut être estimé que pour les produits manufacturés que des sociétés multinationales américaines importent de leurs filiales à l’étranger et pour les produits exportés vers les Etats-Unis par des filiales américaines situées dans chacun des principaux pays industrialisés. Il n’existe aucune série chronologique adéquate sur les exportations des sociétés américaines vers leurs filiales à l’étranger. Les résultats indiquent que, s’il n’y a pas de différence significative entre la réaction des échanges intra-société aux variations de la demande globale et celle des échanges de type traditionnel aux mêmes variations, la sensibilité estimée des échanges intra-société aux variations des prix est, en revanche, nettement inférieure à celle des autres échanges. En général, le modèle commercial normal est peu adapté aux échanges commerciaux intra-société et il serait peut-être préférable d’utiliser un modèle légèrement différent, fonde sur une approche qui tient compte du choix du lieu de production.

Discipline imposée par les importations au secteur manufacturier des Etats-Unisdean a. derosa et morris goldstein (pages 600–634)

Dans cette étude, les auteurs procèdent à un test empirique de la thèse selon laquelle les importations, en faisant concurrence aux produits nationaux, limitent les possibilités pour les producteurs de majorer leurs prix et, de ce fait, imposent une discipline. Ils présentent des estimations de l’incidence que les variations du degré de la concurrence exercée par les produits importés (mesurées par les variations du rapport entre la valeur des importations et la valeur des livraisons de produits nationaux) ont eue sur le processus de fixation des prix intérieurs dans l’industrie manufacturière aux Etats-Unis pendant la période 1972–76; ces estimations, qui se fondent sur des données désagrégées, ventilées par catégories de branches d’activité—telles qu’elles figurent dans la Classification internationale type, par industrie, de toutes les branches d’activité économique, où les positions sont désignées par un nombre à 4 chiffres—, permettent d’analyser, d’une branche d’activité à l’autre et au cours des années, l’incidence que la concurrence des importations exerce sur les prix intérieurs.

L’étude diffère des recherches empiriques qui ont été précédemment effectuées dans ce domaine à un triple point de vue. Premièrement, les variations que subit la demande portant sur les produits industriels sont incorporées dans les équations de variation de prix comme variables de contrôle; de ce fait, il est possible d’identifier l’effet propre que les variations du degré de la concurrence exercée par les importations ont sur les prix. Deuxièmement, l’étude traite la discipline que les importations imposent en matière de fixation des prix comme une variable dépendante de la structure du marché concurrentiel dans l’industrie. Enfin, elle utilise la méthode d’estimation des doubles moindres carrés pour faire ressortir la relation simultanée entre les variations de prix intérieurs, d’une part, et les variations de la demande intérieure et de la pénétration des importations, d’autre part.

La principale conclusion empirique qui se dégage de cette étude est que l’augmentation du degré de la concurrence exercée par les importations a certainement un effet restrictif sur le taux d’inflation dans l’industrie manufacturière américaine; et l’effet est d’autant plus grand que le niveau de concentration dans l’industrie est plus élevé. L’incidence que la discipline imposée par les importations exerce sur la fixation des prix est certes faible par rapport à celle des coûts et de la demande, comme l’étude permet de le constater, mais les estimations indiquent néanmoins que les responsables de la politique ne devraient pas ignorer les implications pour les prix intérieurs de certaines mesures visant à protéger les industries manufacturières américaines contre la concurrence des produits importés.

RESUMENES

Efectos de la intervención de los bancos centrales en los mercados de divisashans genberg (páginas 451–76)

En este artículo se examinan algunas consideraciones teóricas y demostraciones empíricas referentes a los efectos que la intervención de los bancos centrales en los mercados de divisas produce en las fluctuaciones de los tipos de cambio. El aspecto principal consiste en determinar hasta qué punto se puede prever la influencia que ejercerá en el tipo de cambio una intervención cuyo impacto en la oferta monetaria interna hay a sido totalmente neutralizado.

Las consideraciones teóricas parecen indicar que la intervención puede influir en el tipo de cambio si los activos denominados en distintas monedas no son del todo intercambiables, con lo cual se crean oportunidades de intervención para modificar las diferencias de tipo de interés y, por consiguiente, el tipo de cambio, pues el sector privado modifica la composición de la cartera para reequilibrarla. Por otra parte, la intervención puede ejercer una influencia sistemática en las expectativas del público en cuanto al curso que seguirá la política monetaria y, por consiguiente (como resultado de la especulación), en el tipo de cambio en vigor. Los resultados empíricos sobre la base de uno u otro de estos factores adolecen de deficiencias debido, por una parte, a la dificultad inherente a la medición de las expectativas y, por otra, a la obtención de datos exactos sobre monedas utilizadas en la intervención y tenencias de activos. Los resultados basados en la estimatión de ecuaciones del tipo de cambio no confirman, en general, que se produzcan efectos significativos y estables en lo que respecta al equilibrio de la cartera. Se llega a la misma conclusión al observar los efectos de la intervención sobre las primas de riesgo en los mercados de divisas, sobre las diferencias entre los tipos de interés externos y los internos y sobre las diferencias cubiertas de los tipos de interés internacionales.

La falta de pruebas concluyentes de que los efectos de la intervención en los mercados de divisas sobre el equilibrio de la cartera influyen de manera previsible en los tipos de cambio parece indicar que desde el punto de vista práctico no cabe separar la política de intervención de la política monetaria general. Esto signiflca que si ambos instrumentos de política se utilizan independientemente puede que la estabilidad de los tipos de cambio disminuya en vez de aumentar.

Incentivos a la diversificación monetaria internacional por parte de los inversionistas financieros privados de Estados Unidosgeorge m. von furstenberg (páginas 477–94)

En los últimos años se ha diversificado crecientemente la composición monetaria de las tenencias oficiales de divisas. El inversionista privado que no tenga grandes obligaciones de pagos en el exterior y esté interesado en diversificar sus activos líquidos podría dar lugar a perturbaciones de las relaciones monetarias tradicionales que redujesen el grado de acierto del control de las economías nacionales. Por este motivo, conviene determinar si el inversionista estadounidense podría en realidad contar con un mayor rendimiento al reconvertir en dólares la invesión que hubiera realizado en otras monedas.

Usando un cuadro de análisis de promedio y varianza, en el estudio se llega a la conclusión de que, si bien el inversionista habría obtenido de hecho—hasta cierto punto y según algunos criterios—tasas superiores de rendimiento de haber optado por diversificar su cartera de activos líquidos, ese inversionista no hubiera obtenido suficiente compensatión por el riesgo que supone el hacerlo. La elevada variabilidad de las tasas reales del rendimiento que los residentes de Estados Unidos habrían obtenido al tratar de aumentar la rentabilidad de sus carteras invirtiendo en varias monedas constituye en cierta manera una protección para el dólar; de la misma forma, dicha variabilidad protege a las demás monedas, pues el residente del país emisor, que carezca de importantes obligaciones de pagos en el exterior, no podrá aprovechar las expectativas fundadas en la evolución pasada más favorable (apreciación imprevista) de otras monedas. Por consiguiente, la especulación basada en la extrapolación de pasadas diferencias en las tasas reales de rendimiento obtenidas con distintas monedas, si dichas tasas se expresan en dólares de poder adquisitivo constante en Estados Unidos, no habría resultado lo bastante lucrativa para justificar el riesgo adicional que representa el mantener monedas o activos líquidos extranjeros.

Estabilización monetaria y credibilidad del gobiernojacques r. artus (páginas 495–533)

Se presenta en este trabajo un estudio analítico y empírico de los efectos que producen los programas de estabilización monetaria en los precios, el producto y el tipo de cambio de los principales países industriales. Se utiliza para el estudio el caso de la República Federal de Alemania. El trabajo se centra en la investigatión de: i) el punto de vista basado en expectativas racionales de que las autoridades monetarias pueden frenar rápidamente la inflatión sin tener que pagar un elevado precio por razón del producto perdido y el desempleo si reducen la expansión del dinero y convencen simultáneamente a los participantes en el mercado privado de que la reducción del crecimiento monetario será perdurable, y ii) el punto de vista basado en el mercado de activos de que el programa de estabilización monetaria tiende a producir una apreciación del tipo de cambio real que acelera el ajuste de los precios en los mercados de trabajo y de bienes—al reducir el precio de los bienes importados—aunque empeora la situación del comercio exterior del país.

Se emplea para el análisis un modelo monetarista sencillo en el que se supone que las expectativas inflacionarias a largo plazo de los participantes en el mercado privado son reflejo de la política monetaria de las autoridades. A diferencia de otros trabajos empíricos sobre este tema, el modelo tiene en cuenta los cambios discrecionales de la orientación de la política seguida por las autoridades monetarias. También tiene en cuenta la existencia de precios contractuales en los mercados de bienes y de trabajo, y considera que el tipo de interés y el tipo de cambio son variables determinadas a largo plazo por los mercados de activos financieros. Tras estimar econométricamente los parámetros en el caso de la República Federal de Alemania, se utiliza el modelo para simular los efectos de un programa de estabilización monetaria en función de dos criterios opuestos: en un caso el programa influye en las expectativas de expansión del dinero mientras que en el otro no influye. Deben tenerse presentes las consabidas advertencias sobre el carácter de los resultados econométricos.

Del estudio se desprende que, incluso si la repentina adopción de una nueva politica monetaria restrictiva logra modificar la expansión del dinero y las expectativas inflacionarias de los participantes en el mercado privado, lo probable es que la política tenga un considerable y persistente efecto negativo en el nivel de la actividad económica. De hecho, en el actual estudio, el efecto de la nueva política en las expectativas se considera no sólo como condición necesaria para el éxito definitivo de la política de restablecer las condiciones no inflacionarias que exige el crecimiento económico sostenido a largo plazo, sino también como fuente principal de las tendencias recesivas durante los primeros años. El acentuado descenso inicial del producto en la República Federal de Alemania ocurre debido al reajuste de las expectativas, no a pesar de éste. Una causa importante de este efecto es la existencia de contratos salariales a largo plazo. Incluso en la situación en que dichos contratos tienden a abarcar únicamente períodos de un año—como ocurre en la República Federal de Alemania—puede que el ajuste de los salarios y precios ante una reducción repentina de la tasa de crecimiento del dinero se retrase tanto que llegue a causar una importante recesión.

En cuanto a la función que el tipo de cambio desempeña, la hipótesis de que la estabilización monetaria ajusta el tipo de cambio más aliá del diferencial de la tasa de inflación se ve confirmada en el actual estudio cuando se produce un reajuste de las expectativas de expansión del dinero. No obstante, cuantitativamente, las repercusiones de ese exceso en el producto y la inflación parecen ser de poca importancia, al menos en el caso de la República Federal de Alemania. Ello se debe principalmente a que los precios de las importaciones, en cuanto tales, no parecen repercutir mucho en la tasa de inflación, en función del deflactor del producto interno bruto (PIB), a menos que las variaciones de dichos precios sean muy considerables, como ocurrió en el período de 1973–75. Es más, la elasticidad-precio del comercio exterior a corto plazo es demasiado baja para permitir que las variaciones del tipo de cambio tengan un efecto señalado en el producto por el cauce de sus efectos en el comercio exterior. No obstante, los efectos derivados del excesivo ajuste del tipo de cambio son considerables y persistentes y pueden, por consiguiente, ser motivo de preocupación para las autoridades nacionales y extranjeras, aunque sólo sea porque inicia un período de varios años de mejor relación de intercambio para el país que adopta la política monetaria restrictiva, aunque también da origen a una menor participación del país en el mercado en lo que se refiere al volumen.

Los argumentos en pro y en contra del dinero de “desequilibrio”william h. white (páginas 534–72)

Las funciones de demanda de dinero habituales, como por ejemplo las que se utilizan al fijar los objetivos de expansión del dinero, han sido objeto de críticas porque dan cabida a desfases en el ajuste de la demanda de dinero a las variaciones que experimentan variables como el ingreso, mientras que prescinden de los desfases que se producen en el intento del sector privado no bancario de imprimir sentido inverso a las variaciones exógenas o inesperadas de la oferta monetaria que no se deseen. No obstante, se llega a la conclusión de que son inaceptables las pruebas econométricas de que se dispone relativas a las reacciones pasivas temporales “de amortiguación” ante una cantidad de dinero no deseada. Además, la posibilidad de que haya cantidades significativas de dinero no demandado se ve restringida o eliminada cuando las formas normales de variación del dinero—que a su vez modifican directamente los tipos de interés, el ingreso real y los precios—desplazan el ejemplo típico de dinero recibido de forma inesperada. (Por supuesto, la combinación de los pequeños efectos inmediatos que produce la expansión del dinero en el ingreso y de una elasticidad baja de la demanda sin desfase con respecto al ingreso sólo puede explicar, en promedio, una pequeña parte de las variaciones corrientes del dinero observadas; pero los efectos desfasados de variaciones anteriores del ingreso en la demanda corriente de dinero pueden explicar, en promedio, la mayor parte del resto.) Además, los esfuerzos iniciales para eliminar toda variación no demandada que todavía persista, y que son tan solo parciales, tienden a restablecer inmediatamente posiciones en la curva de demanda al inducir variaciones en los tipos de interés y en los precios; en lo que respecta a estas últimas, se pasa por alto el desfase del reajuste de cartera ocasionado por los costos de las transacciones, y la cartera dada se convierte en la deseada a través de la elevada elasticidad de la demanda a largo plazo. Además, las empresas de los países en los que no se dispone de crédito para girar en descubierto pueden eliminar el excedente de efectivo casi con la misma rapidez y bajo costo que las empresas que pueden girar en descubierto. Y los mercados activos del dinero, los certificados de depósito y el crédito al consumidor de elevado costo y fácilmente rembolsado han facilitado en el último decenio—incluso en algunos países en desarrollo—ajustes bastante rápidos de las empresas y las unidades familiares que no recurren al crédito.

Otras relaciones de comportamiento que a veces se proponen para justificar el comportamiento amortiguador (es decir, pertinencia del tipo de interés previsto que reacciona con lentitud en vez del tipo efectivo, cambios en la preferencia de liquidez cuando varía inesperadamente el tipo de interés) parecen ilógicas o cuantitativamente carentes de importancia. No obstante, los modelos corregidos y los datos de los que se hayan eliminado los errores sistemáticos podrían quizá seguir indicando un cierto comportamiento amortiguador.

El comercio internacional de las empresas multinacionales y su sensibilidad ante las variaciones de la demanda global y los precios relativosdavid j. golds-brough (páginas 573–99)

Una proporción considerable del comercio internacional consiste actualmente en transacciones directas entre diversas filiales de una misma empresa multinacional. La magnitud y dirección de estas corrientes comerciales se determinan mediante decisiones, adoptadas dentro de la empresa, sobre la elección de lugares. La proporción creciente de estas transacciones internas con respecto al comercio total puede tener consecuencias importantes con respecto a la sensibilidad de las corrientes comerciales de un país ante las variaciones de la actividad económica y la competitividad relativa. En este trabajo se trata de determinar si la sensibilidad de estas corrientes de comercio interno ante las variaciones de la demanda global y los precios relativos difiere de la sensibilidad del comercio más convencional. Se consideran ciertas razones a priori que permitirían suponer una menor elasticidad-precio en el caso del comercio que se efectúa dentro de una misma empresa. También se analizan las repercusiones econométricas de la utilización de precios de transferencia artificiales, y se demuestra que, en general, estos precios originan un sesgo a la baja en las elasticidades-precio estimadas.

En la parte empírica de este estudio se examina la aplicación de una versión simplificada del Modelo de Comercio Mundial, del Fondo, a los datos sobre las transacciones de las empresas multinacionales estadounidenses y sus Males en el exterior, y la comparación de los resultados con los obtenidos en el caso de las corrientes comerciales convencionales. Dado el carácter de los datos disponibles, sólo es posible estimar el modelo en el caso de las importaciones estadounidenses de manufacturas provenientes de filiales en el exterior y las exportaciones de las filiales de empresas estadounidenses en cada uno de los principales países industriales. No existen series cronológicas adecuadas relativas a la exportación de las casas matrices estadounidenses a sus filiales en el exterior. Los resultados indican que si bien no hay una diferencia significativa en cuanto a la sensibilidad del comercio efectuado dentro de una misma empresa y del comercio convencional ante las variaciones de la demanda global, la sensibilidad estimada de los precios es muy inferior en el caso del comercio que tiene lugar dentro de una misma empresa. En general, el modelo normal de comercio se adapta mal a las corrientes comerciales que existen dentro de una misma empresa y puede ser necesario aplicar un modelo algo diferente, basado en la elección de lugares.

Disciplina impuesta por la importación en el sector manufacturero de Estados Unidosdean a. derosa y morris goldstein (páginas 600–634)

En este trabajo se estudia empíricamente la tesis de que la importación impone una limitación de la competitividad, o disciplina, a la capacidad de los productores nacionales para subir los precios. Se presentan estimaciones del efecto causado por las variaciones de la competitividad de las importaciones (medida como variación de la razón entre el valor de las importaciones y el valor de los envíos nacionales) en el comportamiento de los precios internos de las industrias manufactureras de Estados Unidos en relación con el período 1972–76 utilizando datos desagregados, clasificados según las categorías industriales de cuatro dígitos de la Clasificación Industrial Uniforme, de modo que los efectos de la competencia de las importaciones en los precios internos puede estudiarse en relación con las diversas industrias y a lo largo del tiempo.

Este estudio difiere de investigaciones empíricas anteriores de la disciplina impuesta por la importación en tres aspectos importantes. En primer lugar, las variaciones de los costos de los factores que son insumos de la industria y las variaciones de la demanda de la producción de la industria se incluyen en las ecuaciones de variación de los precios como variables de control, de modo que se puede identificar el efecto independiente causado por las variaciones de la competitividad de las importaciones en los precios. En segundo lugar, el estudio considera la disciplina que en los precios imponen las importaciones como un elemento dependiente de la estructura del mercado competitivo de la industria. Finalmente, en el estudio se utilizan métodos de estimación de mínimos cuadrados de dos etapas para salvar la dificultad que plantea la relación simultánea entre la variación de los precios internos, por un lado, y la variación de la penetración de las importaciones y la demanda interna, por el otro.

La principal conclusión empírica del estudio es que los incrementos de la competitividad de las importaciones tienen un considerable efecto restrictivo sobre la tasa de inflación en el sector manufacturero de Estados Unidos, y este efecto es más pronunciado cuanto mayor es la concentración industrial. Si bien la influencia de la disciplina impuesta por la importación resulta pequeña en relación con la influencia de los factores de costo y demanda en la determinación de los precios, las estimaciones indican que los responsables de la política económica no deberían dejar de lado las repercusiones que tienen en los precios internos las diversas medidas encaminadas a proteger las industrias manufactureras estadounidenses contra la competencia de las importaciones.

In statistical matter (except in the résumés and resúmenes) throughout this issue,

Dots (…) indicate that data are not available;

A dash (—) indicates that the figure is zero or less than half the final digit shown, or that the item does not exist;

A single dot (.) indicates decimals;

A comma (,) separates thousands and millions;

“Billion” means a thousand million;

A short dash (–) is used between years or months (e.g., 1977–79 or January–October) to indicate a total of the years or months inclusive of the beginning and ending years or months;

A stroke (/) is used between years (e.g., 1978/79) to indicate a fiscal year or a crop year;

Components of tables may not add to totals shown because of rounding.

FINANCIAL POLICY WORKSHOPS: The Case of Kenya

This book offers a series of workshops on Kenya that are used as a case study in the IMF Institute’s course on Financial Analysis and Policy for officials of the Fund’s member countries. The workshops are designed to involve participants in all phases of work leading to the formulation of a consistent financial program. It is hoped that participants will be able to apply what they learn to their own countries.

The topics of the workshops are as follows:

  1. Monetary and Financial Survey

  2. Government Finance Statistics

  3. Balance of Payments Statistics

  4. Flow of Funds

  5. The Polak Model: An Application

  6. Projection of Monetary Aggregates

  7. Revenue Forecasting

  8. Balance of Payments Forecasting

  9. Financial Programming

Each workshop deals with a particular sector of the economy, provides relevant background information, and contains exercises and questions (with answers) relating to the statistical data. They draw heavily on the Fund’s experience with programs of balance of payments adjustment. For each workshop, actual data from Kenya permit practical application of analytical and forecasting techniques. Numerous tables present data for analysis, forecasting, and financial programming on a national level. Charts and a map illustrate the book.

Clothbound, US$12.50 (postpaid by surface mail; extra for air mail) Pp. xx+315

Advice on payment in currencies other than the U.S. dollar will be given on request. For information and to place orders, please write to:

External Relations Department

Attention: Publications

International Monetary Fund

Washington, D.C. 20431 U.S.A.

*

Mr. DeRosa, an economist with the U. S. Department of the Treasury, is a graduate of the University of Washington and the University of Oregon.

Mr. Goldstein, Advisor in the Research Department, was on leave at the U. S. Department of the Treasury when this paper was prepared. He is a graduate of Rutgers University and New York University, and was formerly a Research Fellow in Economics at the Brookings Institution.

The views expressed in this paper are solely those of the authors and do not necessarily represent the views of the U. S. Treasury. In addition to colleagues in the Fund, the authors would like to thank Jacob Dreyer, Mike Finger, Angelo Mascaro, and Fred Sterbenz for helpful comments on an earlier draft. Keith Hall and Bill Murden provided particularly efficient research assistance.

1

The landed price of imports is the world price of the good plus transportation costs, tariffs, and any other artificial impediments to trade.

2

It is not clear that the rate of inflation of domestic producer prices will also be lower, the greater is the level of import competition. On the one hand, as with unionization or other restrictions on competition, there is a general presumption that a lower level of import competition will raise the price of protected products/industries relative to unprotected ones. This suggests that import competition can affect the rate of inflation only if the extent of import competition (the share of protected products in total consumption) is changing or if the protected/unprotected price differential is growing. On the other hand, and as we show later on, the level of import competition may be a reliable indicator of the supply capabilities of foreign producers to meet domestic demand and, as such, may contribute independently to restraining inflation of domestic output prices.

3

To take a specific example, for aggregate manufactures, the standard deviation of changes in the ratio of imports to shipments over the period 1972–76 was only about 3 percentage points. In contrast, the standard deviation of this variable across four-digit components was about 15 percentage points.

4

In what follows, we discuss only the effects of import competition on domestic pricing. Much of the literature on import discipline, however, relates to the effect of import competition on profitability of domestic producers; see, for example, Esposito and Esposito (1971), Jones and others (1973), and, most recently, Pugel (1980) and Marvel (1980).

5

See, for instance, the summary by Clayman (1979) of a recent U. S. Labor Department study of the price behavior of products under import relief (via orderly marketing arrangements).

6

The independence issue is perhaps most easily understood by considering the problem of how to estimate the effect of unionization on wages. The basic point is that one cannot take the observed difference between wages in union and nonunion firms as representing the influence of unionization because wages in the nonunion firms will be set partly with the purpose of deterring unionization in those firms. Hence, unionization affects wages in both the control group and the subject group, and a large effect of unionization on wages is perfectly compatible with no difference in observed wages as between union and nonunion firms. The parallel here is that if prices in protected (from imports) industries are set in part so as not to lose sales to nonprotected industries, the observed difference in prices will not represent the effect of import competition on prices. In more general terms, the control group approach will not work unless the control group is truly unaffected by the policy or variable under study.

7

Pugel (1980) makes the same argument for profits.

8

More recently, the simultaneity issue is being tackled directly (e.g., see Marvel (1980)) but not, to our knowledge, for price determination, as opposed to profit determination.

9

See, for instance, Hadar (1971), pp. 115–17.

10

Generally, the price leadership model is presented diagramatically as is done, for instance, in Marvel (1980), p. 105. It is not difficult, however, to derive equation (2) by construction from the demand and supply schedules depicted in the usual diagrams illustrating the price leadership model.

11

A discussion of the limit pricing model can be found in Scherer (1970), Chapter 8, pp. 219–34.

12

It should be noted, of course, that if the elasticities of import supply and domestic demand are quite stable over a particular estimation period, then the influence of these import discipline factors will be captured in a regression of domestic producers’ prices on costs, demand conditions, and import market shares as a part of the estimated coefficient of the import-market-share variable.

13

One encounters the problem that the error term may become heteroscedastic (e.g., see Goldstein and Khan (1976)); nevertheless, a common practice is to estimate the equation by OLS, since the estimates will still be unbiased, and the significance of the coefficients α0 and α1 will still provide a test of the import discipline hypothesis.

14

Specification of the log difference of variables is dictated by our intention to use cross-sectional industry data to test the import-discipline model. Taking the log difference of variables is very nearly equivalent to measuring the proportional change of variables, especially when year-to-year changes are very small. The advantage of either transformation of the data is that it tends to standardize the scale of variables across industry categories.

15

Equations (7) and (8) are approximations because they neglect possible second-order relationships among the disturbances, ε1, ε2, and ε3. The omission of these secondary relationships serves solely to simplify, not alter, our discussion of simultaneous-equation bias.

16

As a practical matter, it makes little difference whether the utility tree notion of demand is assumed, because by and large a more general model of demand would still suggest many of the same TSLS instruments. Our only purpose in assuming separability of consumers’ utility functions is to bring these variables into focus as directly and simply as possible.

17

As was noted earlier, Δ denotes the year-to-year change in the logarithm of the variable. The estimated coefficients can therefore be interpreted as elasticities.

18

We used these three alternative proxies for industry demand changes because each of them has certain weaknesses not shared by the others for the purpose at hand. For example, the inventory variable can be expected to be collinear with the import penetration variable when inventories are replenished by imports. The change in real output covers only the output of domestic firms when we really want to measure demand changes for domestic firms plus imports. Finally, the market demand variable, while covering both imports and domestic output, is difficult to express in real terms because there is no obviously superior deflator when import prices do not equal domestic prices.

19

The only exceptions to this statement are the coefficients on the industry demand variable, which sometimes change appreciably in sign and significance.

20

The capital-stock variable is the one cost component variable that is insignificant in each of the regressions (Id)—(If).

21

To take two specific examples, in 1976 the ratio of imports to shipments increased about 50 per cent in the specialty tool and die industry (3544) and increased about 35 per cent in the television components industry (3672).

22

Since most earlier aggregate studies of the effect of import price changes on domestic price changes for the United States do not hold unit materials cost constant (see Hooper and Lowrey (1979) for a survey), it is not surprising that they obtain larger effects than those presented here.

23

The first-stage regressions for the alternative demand variables appear in Table 8 in the Appendix.

24

See Hooper (1978), Gregory (1971), and Deppler and Ripley (1978) for estimates of the U. S. price elasticity of demand for imports. Though price elasticities of import demand and import substitution are different parameters, they are related. In particular, if one is unity, so is the other. In some other regression results not presented here, we found that the elasticity of substitution of the volume of import penetration changes with respect to domestic price changes was statistically greater than unity at the 5 per cent level, thus providing support for our earlier supposition that OLS estimates of the value measure of import penetration would be biased toward understatement of the true effect. This finding of an apparent difference between values of substitution elasticities with respect to import and domestic prices is not uncommon. In particular, both Ahluwalia and Hernández-Catá (1975) and Murray and Ginman (1976) report finding a significantly larger elasticity of demand for U.S. imports with respect to domestic prices than with respect to foreign prices.

25

One strong advantage of using import penetration to represent import competition is that it permits one to estimate the domestic price effects of policies, such as voluntary export restraints and import quotas, that may not have an immediate or easily determined effect on alternative measures of import competition (such as, for instance, the price of imports).

26

Our data on import prices are too limited to make it possible to define two-digit manufacturing price indices having both domestic and foreign components, such as are suggested in equation (10).

27

One reason for using such a specification is apparent from the first term in either equation (2) or (2′). Both these equations suggest that one effect of import competition is to restrain prices by discouraging domestic firms from producing at high-cost rates of output. When unit costs are used in the price change equation rather than factor prices, this effect of import competition cannot be identified. In other words, with demand pressures held constant, average variable cost itself will be a function of import penetration; the factor price specification of the price-change equation permits this effect to be identified, whereas the unit cost specification does not.

IMF Staff papers: Volume 28 No. 3