Short-Run Effects of Exchange Rate Changes on Terms of Trade and Trade Balance

Erich Spitäller

doi:10.5089/9781451946864.024.A004

Jump to Content Jump to Main Navigation

Short-Run Effects of Exchange Rate Changes on Terms of Trade and Trade Balance

Author: Erich Spitäller¹

¹ 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

Publication Date:: 01 Jan 1980

eISBN:: 9781451946864

Language:: English

Keywords:: SP; mixed-citation publication-type; unit value; contract price; supplier price response; export demand; export unit; Exports; Exchange rate adjustments; Export prices; Valuation; origin and classification; Exchange rates; competitor price variable

Download PDF (1.5 MB)

The persistence of large payments imbalances in the face of considerable swings in exchange rates has imparted new urgency to questions about the functioning of the adjustment process under floating exchange rates. This paper addresses one particular aspect of adjustment, namely, the magnitude and short-run time path of the effects of exchange rate changes on export and import unit values. In the process, analysis turns to the associated terms of trade effect on the trade balance. Thus, some evidence is presented regarding the shape of the first segment of the J-curve—the initial deterioration in the trade balance following a devaluation—assuming the volumes of exports and imports to be constant during the twelve months following an exchange rate change.

Abstract

In Section I, a simple model of export demand and export supply is used to analyze the effects of exchange rate changes on export contract prices and export unit values. The transition from export contract prices to export unit values is made along the lines suggested in the recent literature. The section ends with the derivation of a distributed lag equation in which percentage changes in export unit values are related to corresponding changes in exchange rates, competitor prices, and production costs. Section II contains a behavioral relationship for changes in import prices and, making the same transition as in export contract prices, an import unit value equation. Accordingly, percentage changes in import unit values are expressed as a distributed lag function of changes in exchange rates and suppliers’ average export prices. The remainder of the section concerns the derivation of the size and distribution of the cumulative terms of trade effect on the trade balance, given the exchange rate effects on export and import unit values.

In Section III, the export and import unit value equations were estimated for a sample of ten countries using monthly data over the period January 1973 through April 1978. The countries in the sample are Belgium, Canada, France, the Federal Republic of Germany, Italy, Japan, the Netherlands, Sweden, the United Kingdom, and the United States. Estimation was carried out subject to certain assumptions regarding distribution and length of lag structures. The exchange rate effects estimated for the different countries are presented in the context of the cumulative monthly effects of a 10 per cent devaluation. Effects on export and import unit values are shown, together with corresponding terms of trade effects on the trade balance. A selective summary of results is presented in Section IV.

I. Effects of Exchange Rate Changes on Contract Prices and Unit Values of Exports

This section explores ways in which exchange rate movements influence export contract prices and export unit values. ¹ A simple structural model is used to trace the exchange rate effects. It is composed of (i) demand and supply equations for the volume of a country’s exports, as well as an equilibrium condition, and (ii) an expression of total factor costs in the production of exports in terms of labor and material costs. All relationships are expressed in per cent changes.

The demand for exports of a country depends on two factors, economic activity abroad and the export price of the country relative to the price of its competitors on export markets.

\begin{matrix} {\overset{*}{X}}_{D} = ϵ \overset{*}{G D} + γ [\overset{*}{X P} - (\overset{*}{W X P}) + \overset{*}{E X}] & ϵ > 0, γ < 0 & (1) \end{matrix}

where X_D denotes the demand for exports of the home country in volume terms. GD is an appropriately weighted average of real foreign economic activity, with weights corresponding to the importance of different export markets to the country in question. XP is the export contract price in local currency. WXP is a weighted index of competitor prices in competitor countries’ currencies, and EX is the correspondingly weighted “effective” exchange rate expressed as the value of these currencies in terms of local currency; the weights take into account both the geographic distribution of exports by the home country and the shares that different competitors have in different export markets. ² Specifically, the weighting schemes—shown here with reference to competitor prices—involve

\begin{matrix} {\overset{*}{W X P}}_{k} = \underset{j}{Σ} s_{k j} \underset{1 \neq k}{Σ} z_{l j} \overset{*}{X V_{l}} & (2) \end{matrix}

where the prices of the competitors of the home country, k—here approximated by the competitor country’s export unit value, XV_l—are weighted by the share of a particular market, j, in the total exports of country k, s_kj, and by the share that the competitor country, l, has in market j, z_lj. Accordingly,

\begin{matrix} s_{k j} = \frac{X_{k j}}{X_{k}} & (3) \end{matrix}

where X_kj represents the exports of the home country, k, to market j, and X_k represents the home country’s total exports; and

\begin{matrix} z_{l j} = \frac{X_{l j}}{\underset{1 \neq k}{Σ} X_{l j}} & (4) \end{matrix}

where X_lj represents the exports of the competitor country, l, to market j. All export figures are expressed in 1973 (base year) U. S. dollars.

The coefficients ε and γ represent the income and price elasticities of export demand. Asterisks above variables denote per cent changes from the immediately preceding period.

The supply of exports by the home country is written as a function of export prices and unit labor and material costs in the production of tradeables.

\begin{matrix} {\overset{*}{X}}_{S} = β (\overset{*}{X P} - \overset{*}{I N P C}) & β > 0 & (5) \end{matrix}

where X_s stands for the supply of exports by the home country in volume terms and β is the elasticity of supply. INPC denotes an index of unit variable input costs that is a weighted average of standard unit labor costs and material prices, with weights reflecting the shares of labor and materials in total variable factor inputs in the production of tradeables. In mathematical terms,

\begin{matrix} \overset{*}{I N P C} = α (\overset{*}{W} - \overset{*}{\bar{Q}}) + (1 - α) (\overset{*}{M A P} + \overset{*}{E $}) & α > 0 & (6) \end{matrix}

where W is the wage rate, $\bar{Q}$ is normal output per man-hour, MAP is the price of material inputs in U. S. dollars, and E$ is the local currency price of the U. S. dollar; α and (1—α) denote the. shares of labor and materials in total variable factor inputs, both final and intermediate, in production. The U. S. dollar exchange rate is used here to convert material prices into local currency equivalents because for a large number of important materials, including petroleum, world market prices are denominated in U. S. dollars.

Equilibrium of demand and supply of exports is written as

\begin{matrix} {\overset{*}{X}}_{D} = {\overset{*}{X}}_{S} = \overset{*}{X} & (7) \end{matrix}

The effects of exchange rate changes on export contract prices may be studied most conveniently if the export demand and supply equations and the equilibrium condition, equations (1), (5), and (7), are, on substitution, reduced to

\begin{matrix} \overset{*}{X P} = \frac{γ}{γ - β} (\overset{*}{W X P} + \overset{*}{E X}) - \frac{β}{γ - β} \overset{*}{I N P C} - \frac{ϵ}{γ - β} {\overset{*}{G D}}^{3} & (8) \end{matrix}

Export prices can be seen to adjust by the full amount of a change in competitiveness if either export supply is inelastic, β = 0, or export demand is infinitely elastic, γ = ∞. Conversely, they adjust by the full amount of a change in production costs if either export supply is infinitely elastic, β = ∞, or export demand is inelastic, γ = 0. In these two instances, export prices respond to exchange rate changes only to the extent of their effect on costs. In cases other than these four, the response of export prices to exchange rates will be a composite of their effects on competitiveness and on costs.

In view of the importance of the magnitudes of the elasticities of export demand and supply in this connection, it is well to recall their determinants. ⁴ The magnitude of the elasticity of export demand is inversely related to (i) the degree of specialization of a country’s exports and (ii) the share the country holds in world export markets. ⁵ In addition, it may be mentioned that, if the elasticity of exchange rate changes of competitor countries with respect to an exchange rate change in the home country is high, a given devaluation at home will induce a smaller response in export demand than if that elasticity were low. Accordingly, a tendency of competitor countries to follow suit when the home country devalues is tantamount to a situation of relatively inelastic export demand. The magnitude of the elasticity of supply depends on (i) the nature of goods exported, (ii) the proportion of domestically produced tradeables that is exported, and (iii) the degree of capacity utilization. The supply of industrial products is likely to be more elastic than that of agricultural products, at least in the very short run when the volume of agricultural output cannot be changed. Supply is more elastic when the proportion of tradeables is small rather than large, or when the economy operates below capacity rather than at full employment. Considering the different determinants of the demand and supply elasticities of exports, one could expect that in a large industrial country export prices in local currency would rise by much less in response to a devaluation than they would in a small country.

Substitution from equation (6) into equation (8) yields an export price equation in terms of unit labor costs, material prices, competitor prices, and exchange rates, assuming that in the very short run, which is the perspective here, economic activity abroad remains unchanged, that is, GD = 0.

\begin{matrix} \overset{*}{X P} = - δα \overset{*}{\bar{Q}} + δα \overset{*}{W} + δ (1 - α) [\overset{*}{M A P} + \overset{*}{E $}] + ω [\overset{*}{W X P} + \overset{*}{E X}] & (9) \end{matrix}

δ = - \frac{β}{γ - β} > 0, ω = \frac{β}{γ - β} > 0

Assuming that movements in the U. S. dollar exchange rate are proportional to movements in the multilateral effective exchange rate, EX, equation (9) may be written as

\begin{matrix} \overset{*}{X P} = - δ α \overset{*}{\bar{Q}} + δ α \overset{*}{W} + δ (1 - α) \overset{*}{M A P} + [δ u (1 - α) + ω] \overset{*}{E X} + ω \overset{*}{W X P} & 0 \leq u \leq 1 & (10) \end{matrix}

where u is a proportionality factor linking the two exchange rate variables, $\overset{*}{E $} and \overset{*}{E X}$ ; for the United States, where the variable $\overset{*}{E $}$ drops out, u is set to zero.

Keeping in mind that the purpose of this paper is to assess empirically the terms of trade effects of exchange rate changes and their implications for the trade balance, a transition has to be made from export contract prices to export unit values. How this transition is made is demonstrated in some of the recent literature. ⁶ Specifically, movements in export unit values in local currency are expressed as a function of a combination of current and past changes in export contract prices, on the one hand, and exchange rates, on the other. The maximum lag in this connection reflects the longest period of time that passes between conclusion of the export contract and export delivery. Due distinction is made between contracts invoiced in local currency and contracts invoiced in foreign currency. Denoting the export unit value in local currency by XV, the share of exports delivered in period t and contracted for in period t – i by c_i, $\underset{i}{Σ} c_{i} = 1$ , the share of contracts invoiced in foreign currency by b, the share of contracts invoiced in local currency by (1 – b)—with both shares assumed to be constant—and, finally, denoting the relevant exchange rate by EV—the local currency price of a weighted average of the foreign currencies used in invoicing—we may write

\begin{matrix} \overset{*}{X V} = Σ_{i = 0}^{n} c_{i} [b ({\overset{*}{X P}}_{- i} - {\overset{*}{E V}}_{- i})] + b \overset{*}{E V} + (1 - b) {\overset{*}{X P}}_{- i} & (11) \end{matrix}

\begin{matrix} \overset{*}{X V} = Σ_{i = 0}^{n} c_{i} [{\overset{*}{X P}}_{- i} + b (\overset{*}{E V} - {\overset{*}{E V}}_{- i})] & (12) \end{matrix}

Several aspects of these equations are worth noting. First, if all export contracts were invoiced in domestic currency, the exchange rate term would drop out and changes in export unit values would simply be a weighted average of current and past changes in export contract prices. Second, in all other instances, changes in export unit values would also reflect some valuation loss or gain attributable to exchange rate changes. Third, the exchange rate change $\overset{*}{E V}$ introduces yet another “effective” exchange rate, which may differ from $\overset{*}{E X}$ , as defined earlier. The reason for this difference is that the weights involved in computing EV reflect the shares of different foreign currencies used in invoicing, which do not necessarily correspond exactly to the geographic distribution of exports by the home country and the shares that its competitors have in different export markets. However, in what follows it is assumed for simplicity that the changes in the two effective exchange rates are the same, that is, $\overset{*}{E X} = \overset{*}{E V}$

On substitution of equation (10) into equation (12), the resulting export unit value equation reads

\begin{matrix} \overset{*}{X V} = Σ_{i = 0}^{n} c_{i} [- δ α {\overset{*}{\bar{Q}}}_{- i} + δ α {\overset{*}{W}}_{- i} + δ (1 - α) {\overset{*}{M A P}}_{- i} + ω {\overset{*}{W X P}}_{- i} + (δ u - δ u α + ω - b) {\overset{*}{E X}}_{- i} + b \overset{*}{E X}] & (13) \end{matrix}

From equation (13), it appears that for reasons of valuation alone, which may have nothing to do with the role of exchange rate changes in the determination of changes in export contract prices, the coefficients on the exchange rate and competitor price variables in an estimating equation are likely to differ to some extent. Equation (13) provides the basis for the empirical study of the response of export unit values to exchange rates in particular. One question of interest in this connection is the extent to which wage changes reflect exchange rate changes. While the way in which this may happen is pursued in Appendix I, it is here assumed as a first approximation that wage changes are effectively exogenous in the very short run.

II. Effects of Exchange Rate Changes on Import Prices, Import Unit Values, and Terms of Trade

The preceding section dealt with the determination of changes in export prices and export unit values and, in particular, with their response to exchange rate changes. This section first introduces import price and import unit value equations and then considers the short-run terms of trade effect of a devaluation on the trade balance. Changes in the price of total imports in local currency may be expressed as the weighted sum of changes in prices of manufactured and semimanufactured imports, on the one hand, and changes in the prices of material imports, on the other. In turn, price changes of the first category are a function of weighted averages of changes in the export prices of the home country’s suppliers in foreign currency and exchange rates, where the weights reflect the geographic distribution of the home country’s imports. It is conceivable that, in pricing their exports, supplier countries discriminate among their customers and that, in the particular instance of a devaluation in a country that figures prominently in their sales, suppliers would lower their selling prices to forestall some of the expected reduction in sales. As for the prices of material imports, the same weighted average of petroleum prices and other U. S. dollar denominated material prices used in the export price equation is used again, together with the U. S. dollar exchange rate. Accordingly, the import price equation may be written as

\begin{matrix} \overset{*}{M P} = ρ [σ (\overset{*}{S X P} + \overset{*}{E M})] + (1 - ρ) [\overset{*}{M A P} + \overset{*}{E $}] & ρ, σ > 0 & (14) \end{matrix}

where MP denotes the import price in local currency. SXP is the import weighted average of supplier export prices in foreign currency, ${\overset{*}{S X P}}_{k} = \underset{i}{Σ} V_{j k} {\overset{*}{X V}}_{j}$ , where V_jk is the share of country j in the imports of the home country, k, in the base year and XV is the export unit value, used here as a proxy for the export price. EM denotes the import weighted effective exchange rate, computed analogously; ρ is the share of manufacturers and semi-manufacturers in total imports; (1 – ρ) is the share of materials; and σ is the degree of price discrimination. Assuming movements in the U. S. dollar exchange rate to be proportional to movements in the multilateral effective exchange rate, EM, equation (14) may be rewritten as

\begin{matrix} \overset{*}{M P} = ρ σ \overset{*}{S X P} + (1 - ρ) \overset{*}{M A P} + [ρ σ + (1 - ρ) υ] \overset{*}{E M} & 0 \leq υ \leq 1 & (\begin{matrix} 15 \end{matrix}) \end{matrix}

where v is a proportjonality factor linking the two exchange rate variables, $\overset{*}{E $} and \overset{*}{E M}$ ; for the United States, where the variable $\overset{*}{E $}$ drops out, v is set to zero.

The transition from import prices to import unit values is made in a manner analogous to the transition from export prices to export unit values in the preceding section. Accordingly, account is taken of invoicing practices and of lags between the conclusion of purchase contracts and the delivery of imports. Again following specifications in recent works, ⁷ the import unit value equation reads

\begin{matrix} \overset{*}{M V} = Σ_{i = 0}^{n} e_{i} [g {\overset{*}{M P}}_{- i} + (1 - g) ρ σ {\overset{*}{S X P}}_{- i} + (1 - g) (1 - ρ) {\overset{*}{M A P}}_{- i} + (1 - g) \overset{*}{E M V}] & (16) \end{matrix}

where e_i is the share of imports in period t contracted for in period t - i, $\underset{i}{Σ} e_{i} = 1$ , g is the share of imports invoiced in local currency, and (1—g) is the share invoiced in foreign currency. EMV is the local currency equivalent of a weighted average of the foreign currencies used in invoicing. Although the weights involved need not correspond exactly to the geographic distribution of the home country’s imports, it is assumed in what follows that the two effective exchange rates, EM and EMV, are the same, that is, $\overset{*}{E M} = \overset{*}{E M V}$ . On this assumption, and on substitution from equation (15), the import unit value equation becomes

\begin{matrix} \overset{*}{M V} = Σ_{i = 0}^{n} e_{i} [ρ σ {\overset{*}{S X P}}_{- i} + (1 - g) {\overset{*}{M A P}}_{- i} + (g ρ σ + g υ - ρ υ) {\overset{*}{E M}}_{- i} + (1 - g) \overset{*}{E M}] & (17) \end{matrix}

Valuation gain or loss due to exchange rate changes would be reflected in the difference in direction and magnitude between the two exchange rate terms. From this it follows that, as in the export unit value equation, the coefficients of foreign supplier export prices and exchange rate changes in an estimating equation need not be the same.

On imposition of some constraints on the lag distributions involved in the export and import unit value equations, the effects of exchange rate changes on export and import unit values can be estimated. These estimates can then be used to simulate the cumulative effects of a given rate of devaluation on export and import unit values over a given period. From these, in turn, may be derived the corresponding terms of trade effects on the trade balance.

For example, the terms of trade effect of exchange rate changes on the value of the trade balance in U. S. dollars is calculated as

\begin{matrix} T B (i) = X [1 - \overset{*}{E X} / 100 + \overset{*}{X V} (i) \underset{E X}{*}] - M [1 - \overset{*}{E M} / 100 + \overset{*}{M V (i)} \underset{E M}{*}] & (18) \end{matrix}

where TB is the value of the trade balance in U. S. dollars, i is a time index, i = 1, …, n, X is the U. S. dollar value of exports in the base period, and M is the U. S. dollar value of imports in the base period. ${\overset{*}{X V} (i)}_{E X} and {\overset{*}{M V} (i)}_{\overset{*}{E M}}$ denote the cumulative proportionate change in export and import unit values in response to the exchange rate change. ⁸

III. Empirical Results, January 1973—April 1978

For purposes of estimation, the lag structures of the export and import unit value equations, equations (13) and (17), were assumed to be distributed along a third-degree polynomial without end constraints. Regarding the order of the polynomial, various alternatives ranging from four to twelve months were considered. Both equations were estimated in terms of monthly per cent changes for the period January 1973 through April 1978. In their general forms, the estimating equations are thus represented by

\begin{matrix} \overset{*}{X V} = \overset{*}{X V} [\overset{*}{E X} (L), \overset{*}{W X P} (L), \overset{*}{W} (L), \overset{*}{M A P} (L)] & (19) \end{matrix}

\begin{matrix} \overset{*}{M V} = \overset{*}{M V} [\overset{*}{E M} (L), \overset{*}{S X P} (L), \overset{*}{M A P} (L)] & (20) \end{matrix}

where (L) indicates that the lags on a particular variable are distributed along a third-degree polynomial. The empirical results are reported in the context of the cumulative effect of a 10 per cent devaluation on export unit values, import unit values, and the trade balance, where the trade balance effect is expressed relative to the U. S. dollar value of the trade balance in 1977.

Tables 1 and 2 show the effects of both a devaluation and a rise in competitor or supplier prices on export and import unit values in ten countries—Belgium, Canada, France, the Federal Republic of Germany, Italy, Japan, the Netherlands, Sweden, the United Kingdom, and the United States. These effects were obtained by multiplying by ten and cumulating the distributed lag coefficients of the exchange rate and foreign price variables in the estimated export and import unit value equations. Estimates of the coefficients of the wage cost variables in the export unit value equations are also reported in Table 1, but these variables enter only in concurrent form, or with a discreet lag, to avoid statistical difficulties that may arise from the inclusion of too large a number of lag distributions in the estimating equations. The effects shown in Tables 1 and 2 are presented graphically in Chart 1. Also shown in the chart are the terms of trade effects of the devaluation on the trade balance.

Table 1.

Export Unit Value Equations, Selected Countries, First Quarter 1973–Fourth Quarter 1978: Cumulative Effects of 10 Per Cent Devaluation in Export Weighted Exchange Rate (ER) and 10 Per Cent Rise in Competitor Prices (CP) on Export Unit Values in Local Currency¹

(Monthly per cent change)

^¹All effects are computed by multiplying by ten and cumulating the polynomial lag coefficients in the export unit value equations.

^²Canada and the United Kingdom are the only cases in which use of the wage variable in lagged form rather than concurrent form yielded more meaningful results. In the equation for Canada, wage changes are expressed in terms of a three-month moving average, lagged one period, that is, ${[1 / 3 (\overset{*}{W} + {\overset{*}{W}}_{- 1} + {\overset{*}{W}}_{- 2})]}_{- 1}$ . In the equation for the United Kingdom, wage changes are lagged one month, that is, ${\overset{*}{W}}_{- 1}$

^³Shift dummy variables of the zero/one type were used in the estimating equations to abstract from implausibly large monthly changes in unit values.

^⁴The standard error of the sum of coefficients is, like the coefficients themselves, multiplied by ten.

Table 1.

(Monthly per cent change)

		Belgium		Canada ²		France		Fed. Rep. of Germany		Italy
		ER	CP	ER	CP	ER	CP	ER	CP	ER	CP
Lags in Months
	0	1.1	2.3	1.5	6.0	1.8	11.0	0.4	4.9	3.9	7.8
	1	2.5	4.6	2.8	12.3	2.9	—	0.7	8.1	6.2	11.6
	2	3.9	6.8	3.9	—	3.6	—	0.9	10.0	7.4	—
	3	5.1	8.8	4.8	—	4.1	—	1.1	—	8.0	—
	4	6.0	10.9	5.6		4.5	—	1.3	—	8.2	—
	5	6.6	—	6.2	—	4.8	—	1.5	—	8.4	—
	6	6.8	—	6.7	—	5.3	—	1.7	—	8.7	—
	7	—	—	7.1	—	5.9	—	1.9	—	9.2	—
	8	—	—	7.5	—	—	—	2.1	—	9.8	—
	9	—	—	8.0	—	—	—	2.3	—	10.5	—
	10	—	—	8.6	—	—	—	2.5	—	—	—
	11	—	—	9.5	—	—	—	2.6	—	—	—
Standard error of sum of coefficients ⁴		7.2	4.5	4.8	5.7	2.8	3.9	3.2	3.3	3.6	4.1
Coefficients of cost variables
	$\overset{*}{M A P}$	0.30		0.12		0.00		0.01		-0.04
	t–statistic	(0.70)		(1.33)		(0.00)		(0.20)		(0.61)
	$\overset{*}{MAP}$	-0.30		2.44		-0.32		0.06		0.11
		(1.08)		(2.33)		(1.68)		(0.22)		(1.31)
Summary Statistics
	R²	0.37		0.29		0.37		0.32		0.56
	D-W	2.9		2.6		2.8		2.8		2.2
	SEE	1.3		2.4		1.4		1.0		1.5

		Japan		Netherlands		Sweden		United Kingdom ²		United States
		ER	CP	ER	CP	ER	CP	ER	CP	ER	CP ³
Lags in Months
	0	3.7	3.3	3.0	5.5	0.5	3.1	0.6	5.5	0.5	3.1
	1	5.3	6.2	3.1	9.3	-0.1	7.5	1.1	8.1	1.2	4.5
	2	5.9	8.5	5.8	13.5	-0.1	9.9	1.5	8.7	1.9	5.1
	3	—	10.0	6.8		0.0	10.4	2.0	8.2	2.5	5.5
	4	—	—	—	—	0.4	—	2.5	7.2	2.9	6.7
	5	—	—	—		1.5		3.0	6.2	3.1	7.1
	6	—	—	—	—	2.3		3.5	5.4	3.2	8.3
	7	—	—			3.3		4.0	5.2	—	9.3
	8					4.2	—	4.5	5.7	—	9.4
	9	—	—	—	—	4.8	—	5.0	6.7	—	—
	10	—	—	—	—	—	—	5.4	8.5	—
	11	—						5.6	10.5	—	—
Standard error of sum of coefficients ⁴		3.6	1.7	2.6	2.4	1.4	1.4	2.0	2.4	2.2	2.6
Coefficients of cost variables
	$\overset{*}{M A P}$	0.04		0.01		0.01		0.01		0.09
	t-statistic	(0.64)		(0.24)		(1.04)		(0.22)		(2.36)
	$\overset{*}{W}$	0.99		0.08		-0.18		0.28		-0.09
		(0.99)		(0.98)		(3.52)		(3.82)		(1.16)
Summary Statistics
	R²	0.35		0.50		0.83		0.49		0.61
	D-W	2.0		2.0		0.7		2.5		2.2
	SEE	1.9		1.1		0.4		0.8		1.0

^¹All effects are computed by multiplying by ten and cumulating the polynomial lag coefficients in the export unit value equations.

^³Shift dummy variables of the zero/one type were used in the estimating equations to abstract from implausibly large monthly changes in unit values.

^⁴The standard error of the sum of coefficients is, like the coefficients themselves, multiplied by ten.

Table 1.

(Monthly per cent change)

		Belgium		Canada ²		France		Fed. Rep. of Germany		Italy
		ER	CP	ER	CP	ER	CP	ER	CP	ER	CP
Lags in Months
	0	1.1	2.3	1.5	6.0	1.8	11.0	0.4	4.9	3.9	7.8
	1	2.5	4.6	2.8	12.3	2.9	—	0.7	8.1	6.2	11.6
	2	3.9	6.8	3.9	—	3.6	—	0.9	10.0	7.4	—
	3	5.1	8.8	4.8	—	4.1	—	1.1	—	8.0	—
	4	6.0	10.9	5.6		4.5	—	1.3	—	8.2	—
	5	6.6	—	6.2	—	4.8	—	1.5	—	8.4	—
	6	6.8	—	6.7	—	5.3	—	1.7	—	8.7	—
	7	—	—	7.1	—	5.9	—	1.9	—	9.2	—
	8	—	—	7.5	—	—	—	2.1	—	9.8	—
	9	—	—	8.0	—	—	—	2.3	—	10.5	—
	10	—	—	8.6	—	—	—	2.5	—	—	—
	11	—	—	9.5	—	—	—	2.6	—	—	—
Standard error of sum of coefficients ⁴		7.2	4.5	4.8	5.7	2.8	3.9	3.2	3.3	3.6	4.1
Coefficients of cost variables
	$\overset{*}{M A P}$	0.30		0.12		0.00		0.01		-0.04
	t–statistic	(0.70)		(1.33)		(0.00)		(0.20)		(0.61)
	$\overset{*}{MAP}$	-0.30		2.44		-0.32		0.06		0.11
		(1.08)		(2.33)		(1.68)		(0.22)		(1.31)
Summary Statistics
	R²	0.37		0.29		0.37		0.32		0.56
	D-W	2.9		2.6		2.8		2.8		2.2
	SEE	1.3		2.4		1.4		1.0		1.5

		Japan		Netherlands		Sweden		United Kingdom ²		United States
		ER	CP	ER	CP	ER	CP	ER	CP	ER	CP ³
Lags in Months
	0	3.7	3.3	3.0	5.5	0.5	3.1	0.6	5.5	0.5	3.1
	1	5.3	6.2	3.1	9.3	-0.1	7.5	1.1	8.1	1.2	4.5
	2	5.9	8.5	5.8	13.5	-0.1	9.9	1.5	8.7	1.9	5.1
	3	—	10.0	6.8		0.0	10.4	2.0	8.2	2.5	5.5
	4	—	—	—	—	0.4	—	2.5	7.2	2.9	6.7
	5	—	—	—		1.5		3.0	6.2	3.1	7.1
	6	—	—	—	—	2.3		3.5	5.4	3.2	8.3
	7	—	—			3.3		4.0	5.2	—	9.3
	8					4.2	—	4.5	5.7	—	9.4
	9	—	—	—	—	4.8	—	5.0	6.7	—	—
	10	—	—	—	—	—	—	5.4	8.5	—
	11	—						5.6	10.5	—	—
Standard error of sum of coefficients ⁴		3.6	1.7	2.6	2.4	1.4	1.4	2.0	2.4	2.2	2.6
Coefficients of cost variables
	$\overset{*}{M A P}$	0.04		0.01		0.01		0.01		0.09
	t-statistic	(0.64)		(0.24)		(1.04)		(0.22)		(2.36)
	$\overset{*}{W}$	0.99		0.08		-0.18		0.28		-0.09
		(0.99)		(0.98)		(3.52)		(3.82)		(1.16)
Summary Statistics
	R²	0.35		0.50		0.83		0.49		0.61
	D-W	2.0		2.0		0.7		2.5		2.2
	SEE	1.9		1.1		0.4		0.8		1.0

^¹All effects are computed by multiplying by ten and cumulating the polynomial lag coefficients in the export unit value equations.

^³Shift dummy variables of the zero/one type were used in the estimating equations to abstract from implausibly large monthly changes in unit values.

^⁴The standard error of the sum of coefficients is, like the coefficients themselves, multiplied by ten.

In interpreting results from a model that involves polynomial lag distributions, it is important to have some a priori notion about the shape of a particular distribution. In the present context, one could expect export unit values in local currency to rise more or less gradually in response to a given devaluation, possibly leveling off at some point. In any event, the rise should fall short of the amount of devaluation where the shortfall involved would depend on factors already noted, such as the strength of a country’s position on export markets and the reaction of factor costs to an exchange rate change. The response of import unit values is likely to reflect a more complete pass-through of the devaluation, with the speed of the pass-through depending on delivery lags, among other things. How consistent is the material presented in Tables 1 and 2 and in Chart 1 with these a priori notions?

The exchange rate effects on export unit values shown in Table 1 are appreciably smaller than the amount of devaluation for all countries except Canada and Italy, where they are equal or close to the amount of devaluation. For Belgium and the Netherlands, the effect amounts to about 70 per cent, for Sweden and the United Kingdom to about 50 and 55 per cent, and for the Federal Republic of Germany and the United States to about 25 and 30 per cent. It is apparent from the table that the length of the lag distribution involved in the exchange rate effect on export unit values differs among countries. In this connection, it should be noted that, whenever there was clear evidence of degenerating lag distribution, two criteria were applied in selecting results for tabulation. Either the length of the lag was shortened and the model was re-estimated or the cumulative effect was taken to be constant at the point where the distribution degenerated, omitting subsequent values not deemed meaningful. From Table 1 it appears that the cumulative effects of competitor price changes on export unit values is generally larger than the corresponding effects of an exchange rate change. While reasons were mentioned earlier in the paper for such a disparity there may also be statistical reasons. Specifically, a country’s export price would tend to move in line with the world export price, and the results reported in the table reflect this tendency. ⁹ While the tabulated effects of changes in exchange rates and competitor prices present a plausible picture, the performance of the cost variables is rather unsatisfactory and the explanatory power of the estimated equations is rather poor. A number of reasons may account for this: measurement errors are likely to be present in monthly data; the linkage between the prices and unit values of exports that has been posited is in reality not constant and is certain to be a source of instability, making it difficult to trace the impact of systematic influences; and the assumption of constant foreign economic activity may be too stringent. As for the cost variables, the limiting assumption that had to be made in their construction—involving interpolation of quarterly data for wage changes and the use of a composite of U. S. dollar denominated prices for petroleum and other materials to obtain a measure of material prices—may have adversely affected the results.

With regard to the cumulative effects of a 10 per cent devaluation on import unit values, shown in Table 2, the following applies. (1) The pass-through of devaluation is complete in nine of the ten countries; only in the Federal Republic of Germany is it less than complete. ¹⁰ (2) The time path of the pass-through differs among countries. Generally, however, the pass-through effect occurs more rapidly than the corresponding effect on export unit values. (3) As for differences in the lag distribution, the same points apply that have been made with respect to the results from the export unit value equations. (4) It appears that—much as in the export unit value equations—the response of import unit values to changes, in supplier prices generally tends to be faster than the response to exchange rate changes. Again, differential responses in this respect are quite plausible. A given change in conditions of general rate flexibility is not necessarily considered lasting, at least not to its full extent. ¹¹ By contrast, there is less reason to believe in price flexibility, so that price changes would appear to be more lasting. This disparate assessment of the nature of changes in exchange rates and foreign prices as either transient or lasting gives rise to differential impacts on costs. Even so, the reported response of import unit values to supplier prices is likely to reflect a common tendency in world price movements. In estimating the import unit value equations, three alternative specifications, differing in their treatment of raw material prices, were tried. The changes in these prices were alternatively included—in concurrent form or in distributed lag form—and excluded. None of this differential treatment had any marked effect on the estimated response of import unit values to exchange rates, although, in general, results from equations excluding material prices proved slightly preferable and were, accordingly, chosen for tabulation. ¹² Again, the explanatory power of the equation estimates tends to be as poor as in the estimates of the export unit value equations, and much for the same reasons.

Table 2.

Import Unit Value Equations, Selected Countries, January 1973–April 1978: Cumulative Effects of 10 Per Cent Devaluation in Import Weighted Exchange Rate (ER) and 10 Per Cent Rise in Supplier Prices (SP) on Import Unit Values in Local Currency¹

(Monthly per cent change)

^¹All effects are computed by multiplying by ten and cumulating the polynomial lag coefficients in the import unit value equations.

^²For Canada, the results shown are based on an equation including the local currency price of the U.S. dollar rather than the import weighted exchange rate, which yielded a somewhat lower cumulative price effect, namely 7.7 per cent.

^³A shift dummy variable of the zero/one type was used to take the 1973 oil price increase into account.

^⁴The standard error of the sum of coefficients is, like the coefficients themselves, multiplied by ten.

Table 2.

(Monthly per cent change)

		Belgium		Canada ²		France		Fed. Rep. of Germany		Italy
		ER	SP	ER	SP	ER	SP	ER	SP	ER	SP
Lags in Months
	0	4.0	2.1	3.6	1.9	3.7	11.9	4.6	4.6	5.5	14.4
	1	6.5	4.1	7.3	3.1	5.9	—	6.8	6.9	8.9	—
	2	8.0	5.7	9.2	3.8	7.1	—	7.3	7.7	10.9	—
	3	8.8	6.8	10.1	4.4	7.7	—	—	—	—	—
	4	9.3	7.4	—	4.9	7.9	—	—	—	—	—
	5	9.7	8.6	—	5.4	8.1	—	—	—	—	—
	6	10.2	—	—	6.0	8.3	—	—	—	—	—
	7	—	—	—	6.7	8.7	—	—	—	—	—
	8	—	—	—	7.3	9.2	—	—	—	—	—
	9	—	—	—	7.6	9.8	—	—	—	—	—
	10	—	—	—	—	10.3	—	—	—	—	—
	11	—	—	—	—	—	—	—	—	—	—
Standard error of sum of coefficients ⁴		3.2	2.4	10.2	3.5	4.2	2.4	5.8	2.4	6.3	4.2
Summary Statistics
	R²	0.31		0.11		0.57		0.48		0.63
	D-W	2.4		2.8		1.8		2.2		2.8
	SEE	1.7		2.7		1.9		1.5		2.5

		Japan ³		Netherlands		Sweden		United Kingdom		United States
		ER	SP	ER	SP	ER	SP	ER	SP	ER	SP
Lags in Months
	0	6.1	6.4	2.3	5.9	2.1	5.4	4.8	2.2	1.7	3.3
	1	11.5	10.2	3.9	9.7	4.4	9.1	7.4	3.9	2.7	5.5
	2	—	—	5.0	11.8	6.5	11.4	8.6	5.1	3.4	7.5
	3	—	—	5.7	—	8.3	—	9.0	5.9	4.1	9.0
	4	—	—	6.1	—	10.2	—	9.0	6.3	4.9	10.1
	5		—	6.4		—	—	9.0	6.4	5.9	—
	6	—	—	6.6	—	—	—	9.3	—	7.1	—
	7	—		6.9	—	—	—	9.8	—	8.6	—
	8	—	—	7.4	—	—	—	10.6	—	10.2	—
	9	—	—	8.3	—	—	—	—	—	—	—
	10	—	—	9.7	—	—	—		—	—	—
	11	—	—	11.6	—	—	—	—	—	—	—
Standard error of sum of coefficients ⁴		2.2	2.4	8.4	2.2	2.2	1.4	3.3	1.7	7.2	2.4
Summary Statistics
	R²	0.80		0.66		0.71		0.54		0.43
	D-W	1.9		2.3		1.0		1.8		2.5
	SEE	1.9		1.2		0.4		1.0		1.6

^¹All effects are computed by multiplying by ten and cumulating the polynomial lag coefficients in the import unit value equations.

^³A shift dummy variable of the zero/one type was used to take the 1973 oil price increase into account.

^⁴The standard error of the sum of coefficients is, like the coefficients themselves, multiplied by ten.

Table 2.

(Monthly per cent change)

		Belgium		Canada ²		France		Fed. Rep. of Germany		Italy
		ER	SP	ER	SP	ER	SP	ER	SP	ER	SP
Lags in Months
	0	4.0	2.1	3.6	1.9	3.7	11.9	4.6	4.6	5.5	14.4
	1	6.5	4.1	7.3	3.1	5.9	—	6.8	6.9	8.9	—
	2	8.0	5.7	9.2	3.8	7.1	—	7.3	7.7	10.9	—
	3	8.8	6.8	10.1	4.4	7.7	—	—	—	—	—
	4	9.3	7.4	—	4.9	7.9	—	—	—	—	—
	5	9.7	8.6	—	5.4	8.1	—	—	—	—	—
	6	10.2	—	—	6.0	8.3	—	—	—	—	—
	7	—	—	—	6.7	8.7	—	—	—	—	—
	8	—	—	—	7.3	9.2	—	—	—	—	—
	9	—	—	—	7.6	9.8	—	—	—	—	—
	10	—	—	—	—	10.3	—	—	—	—	—
	11	—	—	—	—	—	—	—	—	—	—
Standard error of sum of coefficients ⁴		3.2	2.4	10.2	3.5	4.2	2.4	5.8	2.4	6.3	4.2
Summary Statistics
	R²	0.31		0.11		0.57		0.48		0.63
	D-W	2.4		2.8		1.8		2.2		2.8
	SEE	1.7		2.7		1.9		1.5		2.5

		Japan ³		Netherlands		Sweden		United Kingdom		United States
		ER	SP	ER	SP	ER	SP	ER	SP	ER	SP
Lags in Months
	0	6.1	6.4	2.3	5.9	2.1	5.4	4.8	2.2	1.7	3.3
	1	11.5	10.2	3.9	9.7	4.4	9.1	7.4	3.9	2.7	5.5
	2	—	—	5.0	11.8	6.5	11.4	8.6	5.1	3.4	7.5
	3	—	—	5.7	—	8.3	—	9.0	5.9	4.1	9.0
	4	—	—	6.1	—	10.2	—	9.0	6.3	4.9	10.1
	5		—	6.4		—	—	9.0	6.4	5.9	—
	6	—	—	6.6	—	—	—	9.3	—	7.1	—
	7	—		6.9	—	—	—	9.8	—	8.6	—
	8	—	—	7.4	—	—	—	10.6	—	10.2	—
	9	—	—	8.3	—	—	—	—	—	—	—
	10	—	—	9.7	—	—	—		—	—	—
	11	—	—	11.6	—	—	—	—	—	—	—
Standard error of sum of coefficients ⁴		2.2	2.4	8.4	2.2	2.2	1.4	3.3	1.7	7.2	2.4
Summary Statistics
	R²	0.80		0.66		0.71		0.54		0.43
	D-W	1.9		2.3		1.0		1.8		2.5
	SEE	1.9		1.2		0.4		1.0		1.6

^¹All effects are computed by multiplying by ten and cumulating the polynomial lag coefficients in the import unit value equations.

^³A shift dummy variable of the zero/one type was used to take the 1973 oil price increase into account.

^⁴The standard error of the sum of coefficients is, like the coefficients themselves, multiplied by ten.

In simulating the effects of a devaluation on export and import unit values, on the one hand, and on the trade balance, on the other, one aspect has so far been disregarded. Specifically, it has been implicitly assumed that the devaluation would leave the prices of material suppliers, MAP, unchanged. The extent to which this assumption biases results needs to be examined. ¹³ If material exporters are sensitive to erosions in their purchasing power, a devaluation of the U. S. dollar in particular would elicit a price response of some importance. Very likely, the U. S. dollar denominated supplier prices of raw materials would rise to compensate for the valuation loss in the supplier countries’ export proceeds. ¹⁴ Assuming, for example, that these countries would want to keep the purchasing power of their export proceeds constant in terms of their import bill, they would adjust their supplier prices in the following manner:

\begin{matrix} \overset{*}{M A P} = {\overset{*}{E}}_{U S (\frac{u_{1} - u_{2}}{u_{3}})} & (21) \end{matrix}

where E_US is the rate of devaluation of the U. S. dollar in per cent, u₁ is the share of export proceeds denominated in the currency of the devaluing country, u₂ is the share of the raw material exporters’ import bill invoiced in the currency of the devaluing country, and u₃ is the share of U. S. dollars in export proceeds. Generally, it may be pointed out that whenever the share of export proceeds denominated in the currency of the devaluing country equals the share of the raw material exporters’ import bill, u₁ = u₂, the purchasing power of the export proceeds will remain unaffected and no price change will occur. Alternatively, whenever u₁ is greater than u₂, raw material exporters will incur a net valuation loss, so that they will raise their export prices, while, strictly speaking, the reverse applies in cases where u₁ is smaller than u₂, involving a valuation gain. Accordingly, where export proceeds are exclusively U. S. dollar denominated, the valuation loss is greatest and the consequent price increase the largest in a devaluation of the U. S. dollar, at least so long as the share of the import bill that is invoiced in currencies other than the U. S. dollar remains substantial.

So far, the discussion of the adjustment in raw material prices in response to a devaluation in an industrial country suggests that this adjustment would have but a small, if not negligible, effect on the import unit value of the devaluing country in all instances other than that of a U. S. dollar devaluation. Consequently, an alternative estimate of the effects of a 10 per cent devaluation on import unit values and on the trade balance, an estimate which in contrast to the tabulated and charted results takes explicit account of the exchange rate induced price adjustment of raw material exports, is provided here for the case of a U. S. dollar devaluation only. Chart 2 contrasts the effects that do allow for an adjustment in the prices of raw material exports with the effects that do not allow for such adjustment. For the purposes of the chart, it is arbitrarily assumed that the share of the raw material suppliers’ import bill corresponds to the weight of the U. S. dollar in the SDR basket as of July 1, 1978. It appears from the chart that, following a U. S. dollar devaluation, import unit values rise more rapidly when the response of supplier prices is taken into account than when this response is excluded, although after ten to eleven months the cumulative effect is the same in both instances. Similarly, while the trade balance deteriorates more rapidly when the supplier price response is considered, after ten to eleven months the cumulative deterioration is the same as when that response is disregarded. ¹⁵

IV. Conclusions

The estimated short-run exchange rate effects on export unit values suggest that the change in competitiveness attributable to a given movement in the exchange rate tends to be offset in amounts and over periods of time differing among countries. In most instances, the offset is smaller than the amount of the exchange rate change and occurs relatively slowly. ¹⁶ On the import side, it seems that the pass-through of an exchange rate change to import unit values is complete in all countries other than the Federal Republic of Germany, where some three fourths of the change is passed through, although the timing of the pass-through differs across countries. ¹⁷ Generally, the response of export or import unit values to competitor or supplier prices in industrial countries appears to be both stronger and more rapid than the response to exchange rate changes. While it has been demonstrated that the two kinds of response need not be the same, the apparent difference may also reflect common trends in price movements.

The estimates of the cumulative monthly exchange rate effects on export and import unit values are used, together with trade flow data in U. S. dollars for 1977, to simulate the short-run terms of trade effect of a 10 per cent devaluation on the trade balance for the ten countries in the sample. For the United States, the simulations take into account the devaluation induced adjustment of raw material prices denominated in U. S. dollars. The terms of trade effect provides an illustration of the first segment of the J-curve, the initial deterioration in the trade balance following a devaluation. The computations show that the low point of the deterioration may be reached some time during or at the end of the twelve-month period that is considered here, depending on the time paths of the exchange rate effects on export and import unit values in the different countries. ¹⁸ The U. S. dollar values of the deterioration at its low and end points may be put in per cent of the flow of imports or exports in 1977 U. S. dollars. This indicates, at current levels of trade, the per cent deterioration attributable to a 10 per cent devaluation of the effective exchange rate. In terms of imports, the maximum deterioration associated with the terms of trade effect of the devaluation at any time during the year following the devaluation ranges between 3 and 4 per cent in Belgium, France, Italy, and the Netherlands; between 4 and 6 per cent in Canada, Japan, and the United States; and between 8 and 10 per cent in the Federal Republic of Germany, Sweden, and the United Kingdom. By the end of the year, however, owing to the complete pass-through of the devaluation to import and export unit values in Canada and Italy, a terms of trade effect is no longer in evidence in these two countries; in other countries it ranges from 3 to 5 per cent, or 6 per cent for the Federal Republic of Germany. This evidence implies that if the reaction of trade volumes to the exchange rate is sluggish, as assumed here, the response of the trade account to an exchange rate change is in most instances likely to aggravate existing imbalances in international trade during the first year.

APPENDICES

I. Transmission of Exchange Rate Changes via Wage Movements to Export Prices and Unit Values

In the text of this paper, wage changes were assumed to be exogenous and no allowance was made for the possibility that they reflect, to some extent, changes in exchange rates. While this assumption is likely to be warranted in the very short run for most countries, it is nevertheless of interest how any transmission of exchange rate changes to wage movements, and in the process to movements in export prices and unit values, would occur. To illustrate the point, wage equations may be used, together with a scheme for inflationary expectations that distinguishes between the long run and the short run. The wage equation represents an expectations-augmented Phillips curve.

\begin{matrix} \overset{*}{W} = a_{11} \overset{*}{\bar{Q}} + a_{12} \overset{*}{P^{e}} (s) + a_{13} (Y / \bar{Y}) & 0 < a_{11}, a_{12} \leq 1, a_{13} > 0 & (22) \end{matrix}

where W is the wage rate, $\overset{*}{P^{e}} (s)$ is the rate of inflation expected in the short run, and $Y / \bar{Y}$ the deviation of real output from its trend value—measures the degree of capacity utilization. Long-run inflationary expectations, $\overset{*}{P^{e}} (l)$ , are assumed to be determined by corresponding expectations about exchange rate changes, $\overset{*}{E^{e}} (l)$ , and the world inflation rate, $\overset{*}{Π^{e}} (l)$ . ¹⁹

\begin{matrix} \overset{*}{P^{e}} (l) = \overset{*}{E^{e}} (l) + \overset{*}{Π^{e}} (l) & (23) \end{matrix}

Expected long-run changes are assumed to be proportional to corresponding changes expected for the short run:

\begin{matrix} \overset{*}{E^{e}} (l) + \overset{*}{{f E}^{e}} (s) & 0 < f < 1 & (24) \end{matrix}

\begin{matrix} \overset{*}{Π^{e}} (l) + \overset{*}{h^{e}} (s) & 0 < h < 1 & (25) \end{matrix}

and

\begin{matrix} \overset{*}{P^{e}} (l) + \overset{*}{{k P}^{e}} (s) & 0 < k < 1 & (26) \end{matrix}

Short-run expectations about exchange rate changes, $\overset{*}{E^{e}} (s)$ , and the world inflation rate, $\overset{*}{Π^{e}} (s)$ , are taken to be formed on the basis of observed current and past changes in these two variables.

\begin{matrix} \overset{*}{E^{e}} (s) = Σ_{j = 0}^{m} λ_{j} {\overset{*}{E}}_{- j} & λ_{j} > 0 & (27) \end{matrix}

and

\begin{matrix} \overset{*}{Π^{e}} (s) = Σ_{j = 0}^{m} θ_{j} {\overset{*}{Π}}_{- j} & θ_{j} > 0 & (28) \end{matrix}

Equations (23) through (28) may be combined to yield an expression for short-run inflationary expectations in terms of actual exchange rate changes and the world inflation rate.

\begin{matrix} \overset{*}{P^{e}} (s) = \frac{1}{k} ⌊ f Σ_{j = 0}^{m} λ_{j} {\overset{*}{E}}_{- j} + h Σ_{j = 0}^{m} θ_{j} {\overset{*}{Π}}_{- j} ⌋ & (29) \end{matrix}

On substitution of this expression and the wage equation into the export price equation (10)—where export price changes are expressed in terms of unit labor costs, material prices, competitor prices, and exchange rates—one obtains

\begin{matrix} \overset{*}{X P} = & (δ α a_{11} - δ α) \overset{*}{\bar{Q}} + δ α a_{12} υ ⌊ \frac{f}{k} Σ_{j = 0}^{m} λ_{j} \overset{*}{E X_{- j}} + \frac{h}{k} Σ_{j = 0}^{m} θ_{j} \overset{*}{W X P_{- j}} ⌋ \\ + δ α a_{13} (Y / \bar{Y}) + δ (1 - α) \overset{*}{M A P} + [δ u (1 - α) + ω] \overset{*}{E X} + ω \overset{*}{W X P} 0 < u, υ \leq 1 & (30) \end{matrix}

where u and υ are proportionality factors, assuming that changes in the different exchange rates are proportional to each other and that the same applies to changes in world prices and competitor prices,

\overset{*}{E $} = u \overset{*}{E X}, \overset{*}{E} = υ \overset{*}{E X}, \overset{*}{Π} = υ \overset{*}{W X P} .

Finally, on transforming changes in export contract prices into changes in export unit values, one may write

\begin{matrix} \overset{*}{X V} = & Σ_{i = 0}^{n} c_{i} [(δ α a_{11} - δ α) {\overset{*}{\bar{Q}}}_{- i} + δ α a_{13} {(Y / \bar{Y})}_{- i} + δ (1 - α) \overset{*}{M A P_{- i}} \\ + ω {\overset{*}{W X P}}_{- i} + (δ u - δ u α + ω - b) {\overset{*}{E X}}_{- i} + b \overset{*}{E X}] \\ + δ α a_{12} υ Σ_{i = 0}^{n} Σ_{j = 0}^{m} c_{i} {(\frac{f}{k} λ_{j} \overset{*}{E X} + \frac{h}{k} θ_{j} \overset{*}{W X P})}_{- i - j} & (31) \end{matrix}

recalling that changes in unit values are a function of current and past changes in contract prices and exchange rates.

II. Data Sources

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. 791-824.
- Search Google Scholar
- Export Citation
Artus, Jacques R., “The Behavior of Export Prices for Manufactures,” Staff Papers, Vol. 21 (November 1974), pp. 583-604.
- Search Google Scholar
- Export Citation
Artus, Jacques R., “The 1967 Devaluation of the Pound Sterling,” Staff Papers, Vol. 22 (November 1975), pp. 595-640.
- Search Google Scholar
- Export Citation
Bruno, Michael, “Exchange Rates, Import Costs, and Wage-Price Dynamics,” Journal of Political Economy, Vol. 86 (June 1978), pp. 379-403.
- Search Google Scholar
- Export Citation
Chiesa, C., and others, Un Modello di Analisi e Previsione del Settore Bilancia dei Pagamenti Correnti, Banca D’Italia, Modello Econometrico dell’Economia Italiana, II ed.—M2 BI (March 1978).
- Search Google Scholar
- Export Citation
Clark, Peter B., “The Effects of Recent Exchange Rate Changes on the U. S. Trade Balance,” in The Effects of Exchange Rate Adjustments: The Proceedings of a Conference Sponsored by OASIA Research, Department of the Treasury, April 4 and 5, 1974, ed. by Peter B. Clark, Dennis E. Logue, and Richard James Sweeney (Washington, 1974), pp. 201-36.
- 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. 147-206.
- Search Google Scholar
- Export Citation
Goldstein, Morris, “The Effect of Exchange Rate Changes on Wages and Prices in the United Kingdom: An Empirical Study,” Staff Papers, Vol. 21 (November 1974), pp. 694-739.
- Search Google Scholar
- Export Citation
Gylfason, Thorvaldur, “The Effect of Exchange Rate Changes on the Balance of Trade in Ten Industrial Countries” (unpublished, International Monetary Fund, October 5, 1978).
- Search Google Scholar
- Export Citation
Hooper, Peter, Forecasting U. S. Export and Import Prices and Volumes in a Changing World Economy, U. S. Board of Governors of the Federal Reserve System, International Finance Discussion Papers, No. 99 (December 1976).
- Search Google Scholar
- Export Citation
Hwa, E. C., “Price Determination in Several International Primary Commodity Markets: A Structural Analysis,” Staff Papers, Vol. 26 (March 1979), pp. 157-88.
- Search Google Scholar
- Export Citation
Junz, Helen B., and Rudolf R. Rhomberg, “Price Competitiveness in Export Trade Among Industrial Countries,” American Economic Review (May 1973), pp. 412-18.
- Search Google Scholar
- Export Citation
Kravis, Irving B., and Robert E. Lipsey, “Export Prices and the Transmission of Inflation,” American Economic Review, Papers and Proceedings, Vol. 67 (February 1977), pp. 155-63.
- Search Google Scholar
- Export Citation
Kreinin, Mordechai E., “The Effects of Exchange Rate Changes on the Prices and Volume of Foreign Trade,” Staff Papers, Vol. 24 (July 1977), pp. 297-329.
- Search Google Scholar
- Export Citation
Kwack, Sung Y., “Price Linkage in an Interdependent World Economy: Price Responses to Exchange Rate and Activity Changes,” in Analysis of Inflation: 1965-1974, ed. by Joel Popkin, Studies in Income and Wealth, Vol. 42 (National Bureau of Economic Research, 1977), pp. 447-77.
- Search Google Scholar
- Export Citation
Laidler, David E. W., “Expectations and the Behavior of Prices and Output Under Fixed and Flexible Exchange Rates,” Research Report No. 7620, University of Western Ontario (London, Canada, October 1976).
- Search Google Scholar
- Export Citation
Magee, Stephen P., “Currency Contracts, Pass-through, and Devaluation,” Brookings Papers on Economic Activity: 1 (1973), pp. 303-23.
- Search Google Scholar
- Export Citation
Miles, Marc A., “The Effects of Devaluation on the Trade Balance and the Balance of Payments: A New Perspective.” Federal Reserve Bank of New York, Research Paper No. 7605 (February 1976).
- Search Google Scholar
- Export Citation
Niehans, Jürg, “Some Doubts About the Efficacy of Monetary Policy Under Flexible Exchange Rates,” Journal of International Economics, Vol. 5 (August 1975), pp. 275-81.
- Search Google Scholar
- Export Citation
Odling-Smee, John, and Nicholas Hartley, “Some Effects of Exchange Rate Changes,” U. K. Treasury Working Papers, No. 2 (1978).
- Search Google Scholar
- Export Citation
Polak, J. J., and T. C. Chang, “Effect of Exchange Depreciation on a Country’s Export Price Level,” Staff Papers, Vol. 1 (February 1950), pp. 49-70.
- Search Google Scholar
- Export Citation
Rhomberg, Rudolf R., “Indices of Effective Exchange Rates,” Staff Papers, Vol. 23 (March 1976), pp. 88-112.
- Search Google Scholar
- Export Citation
Ripley, Duncan M., “The Movement of Export and Domestic Prices in the United States, Japan and Germany” (unpublished, International Monetary Fund, July 10, 1974).
- Search Google Scholar
- Export Citation
Xafa, Miranda, “The Effects of Exchange Rate Changes on Domestic Inflation and on Price Competitiveness: The Cases of Greece and Israel” (unpublished, International Monetary Fund, August 1978).
- Search Google Scholar
- Export Citation

Mr. Spitäller, currently an advisor at the National Bank of Lebanon and senior economist in the External Adjustments Division of the Research Department, is a graduate of the University of Graz, Austria, and of the School of Advanced International Studies of the Johns Hopkins University, Washington, D. C. He was formerly on the staff of the Organization for Economic Cooperation and Development.

For theoretical studies in this connection, see Polak and Chang (1950) and Bruno (1978). For empirical work on different countries, see Clark (1974), Kreinin (1977), Goldstein (1974), Odling-Smee and Hartley (1978), Chiesa, and others (1978), and Xafa (1978). For the relationship between export prices and export unit values, see Artus (1974) and Hooper (1976).

For different ways to calculate “effective” exchange rates, see, for example, Rhomberg (1976).

The weighting scheme used in this paper is one among several alternatives. For example, in computing the average competitor price, account could be taken not only of the prices of exporters competing with the home country in a given market but also of the domestic selling prices in that same market. The weights on these prices and on competing export prices alike would then be the appropriate shares in the total demand in the market—satisfied from both foreign and domestic sources—for the commodity category in question. Arguably, however, the scheme that is used here allows in an indirect manner for the domestic prices of the market receiving the exports of the home country and its competitors, since they all have to contend with those prices and presumably respond to them in the pricing of their own exports. One advantage of the present scheme in this connection is that the problem of summing export unit values and domestic prices for tradeables does not arise.

If supply is wholly inelastic, β = 0, the coefficient of the first term on the right-hand side of the equation equals unity, while the second term drops out. It is immaterial whether there is any effect of exchange rate changes on production costs. Alternatively, if supply is infinitely elastic, β = ∞ exchange rate changes affect export prices only via their impact on production costs but not through their effect on competitiveness. Assuming that there are no exchange rate effects on costs, a devaluation would leave the local currency price of exports unaffected, but it would lower their foreign currency price by the full amount of the devaluation. If export demand is price inelastic, γ = 0, export prices will change by the full amount of the change in production costs. It depends, therefore, on the response of costs to exchange rate changes whether there will be an exchange rate effect on export prices and how large it will be. Conversely, if export demand is infinitely price elastic, γ = ∞, it makes no difference what effect, if any, exchange rate changes have on costs; export prices would remain unchanged.

The determinants listed here are those given in Polak and Chang (1950), pp. 53–54.

Conceptually, a country whose exports are so highly specialized that no substitutes for them exist faces an inelastic demand curve, as does a country that enjoys a monopoly on export markets. In either instance, the country can have its export price differ from that of other exporters. By contrast, a country that is a marginal supplier of exports and that sells goods similar to those sold by others will be forced by competition to keep its prices more closely in line with the prices of its competitors.

See, especially, Artus (1975), Hooper (1976), and Magee (1973).

See Ahluwalia and Hernández-Catá (1975), Artus (1975), and Hooper (1976).

If one wanted to consider both the terms of trade effect and the volume effect of exchange rate changes on the value of the trade balance, the two bracketed terms on the right-hand side of equation (18) would be expanded by $+ \overset{*}{X Q} (i) \underset{E X}{*}$ and $- \overset{*}{M Q} (i) \underset{E M}{*}$ , respectively, where these expressions denote the cumulative proportionate changes in export and import volumes in response to the exchange rate change. On the assumption that the volume effect of exchange rate changes on the trade balance is weak or absent during the first six to twelve months following an exchange rate change, the terms of trade effect corresponds to the first portion of the J-curve effect of devaluation. Evidence of this J-curve effect may be found in Artus (1975, Table 10, p. 621), where the 1967 devaluation of the pound sterling is shown to have had an adverse effect during the first half of 1968; reference is also made to an estimate by the London Business School of an adverse effect during the whole of 1968—Artus (1975, Table 11, p. 622). Regarding more recent exchange rate changes, work at the U. K. Treasury points to an adverse effect during two quarters; see Odling-Smee and Hartley (1978). Clark (1974) reports a J-curve effect of the U. S. dollar depreciation between the second quarter of 1971 and the second quarter of 1972 (Table 6, pp. 230–31). Miles (1976) finds evidence of a deterioration in the trade balance during the year of a devaluation for a number of developed and developing countries. Chiesa, and others (1978) report findings that in Italy the initially adverse trade effects of devaluation are quickly offset within a year. The notion of an initial adverse effect of devaluation on trade is linked to the evidence of rather long lags—two to three years—in the volume effects of exchange rate changes; see Junz and Rhomberg (1973). By contrast, Gylfason (1978) has found lags to be much shorter.

An alternative to the export unit value equation reported in Table 1 was explored. This alternative was based on a price equation embodying the assumption that export prices and wholesale prices behave in such a way that their ratio would change only if the competitiveness of the country in question were to change because of either a change in exchange rates or a change in competitor prices. Accordingly, using the notation employed earlier,

X P = {{W P}^{δ} [(W X P) (E X)]}^{1 - δ}

where WP is the wholesale price index. In terms of per cent changes, the equation reads

\overset{*}{X P} = δ \overset{*}{W P} + ω [\overset{*}{W X P} + \overset{*}{E X}]

Under conditions where δ = 1, and ω = 0, that is, where either export demand is price inelastic, γ = 0, or export supply is infinitely elastic, β = ∞, changes in export prices are equal to changes in wholesale prices, XP = WP. On the empirical relationship between export prices and wholesale prices, see Ripley (1974) and Kravis and Lipsey (1977). Estimates of an export unit value equation that are based on a relationship between export prices, on the one hand, and wholesale prices, competitor prices, and exchange rates, on the other, reveal one general difference compared with the estimates shown in Table 1: the exchange rate effects on export unit values are consistently smaller for all countries, suggesting that the true effect is understated and that part of it is captured by the wholesale price variable. Accordingly, of the two kinds of estimates, those shown in Table 1 may be regarded as superior.

Similar pass-through effects were obtained by Kwack (1977) from annual equations.

For the notion that observed exchange rate changes will only in part be perceived as lasting, see Niehans (1975).

In instances where the concurrent changes in raw material prices proved statistically significant, or nearly so, they tended to reduce the effect of supplier prices shown in Table 2, supporting the presumption that they reflect world price movements in general rather than the movements in the export prices of the ten countries in the sample alone.

On a general level, a devaluation in any country implies a revaluation in the effective exchange rates of its trading partners of a magnitude that differs depending on the degree of interrelation in trade. In the countries registering such a revaluation, the local currency price of exports will tend to fall, or rise less rapidly than would otherwise be the case, in accordance with the response of export unit values to exchange rate changes indicated by the econometric estimates. On a more particular level, three types of effects from the devaluation may be distinguished: (i) effects on the prices of competitors in the world market for industrial goods, (ii) effects on the prices of suppliers of these same goods, and (iii) effects on the prices of suppliers of raw materials. The first two effects are demonstrably small enough to be disregarded without violence to reality.

Accordingly, this happens for the same reason that induces a rise in raw material prices in response to an acceleration of inflation in the export unit values of industrial countries exporting manufactures; see, for example,Hwa (1979).

It may be noted that the effects, inclusive rather than exclusive, of supplier response resemble more the results shown earlier in Table 2 and Chart 1, although they were based on an import unit value equation that did not explicitly contain the supplier prices of raw materials. Presumably, some of the response of import unit values to exchange rate changes in that equation reflects the response to supplier prices of raw materials induced by the exchange rate change.

For the countries in the sample, the ranges of offsets and time periods are between 40 and 68 per cent over three to seven months in Belgium, between 39 and 95 per cent over three to twelve months in Canada, between 36 and 59 per cent over three to nine months in France, between 39 and 100 per cent over one to ten months in Italy, 59 per cent over three months in Japan, 68 per cent over four months in the Netherlands, between 15 and 48 per cent over six to ten months in Sweden, between 30 and 56 per cent over six to twelve months in the United Kingdom, and between 20 and 32 per cent over three to seven months in the United States.

It takes up to seven months in Belgium, four months in Canada, eleven months in France, three months in the Federal Republic of Germany and Italy, two months in Japan, twelve months in the Netherlands, six months in Sweden, and nine to ten months in the United Kingdom and the United States.

The U. S. dollar value of the deterioration at its low point and, where it is different, its end point, amounts to in Belgium, 1.2 billion after three months and 970 million after seven months; Canada, 2.3 billion after three months and 220 million after twelve months; France, 2.8 billion after eleven months; the Federal Republic of Germany, 7.8 billion after three months and 5.9 billion after twelve months; Italy, 1.2 billion after two months while after ten months no deterioration can be observed; Japan, 3.7 billion after two months and 3.3 billion after three months; the Netherlands, 1.7 billion after twelve months; Sweden, 1.8 billion after five months and 980 million after ten months; the United Kingdom, 4.5 billion after four months and 2.5 billion after twelve months; and the United-States, 8.2 billion after nine to ten months.

Regarding the direction of causality that is implied here, a point made by Laidler (1976), may be recalled: “There is nothing a priori implausible about postulating that agents form their price expectations by looking at foreign prices and the exchange rate nor is there any reason to suppose that the average of such expectations about the rate of change of particular prices, which is what our expected rate of inflation for the economy as a whole amounts to, will in the long run be inconsistent with a domestic inflation rate determined solely by the domestic monetary expansion rate. Thus, we cannot rule out on a priori grounds the possibility that inflationary expectations in a flexible exchange rate economy are appropriately modelled as if directly influenced by the world inflation rate and the exchange rate.”

Other Publishers

Save
Cite
Email this content

Share Link

Copy this link, or click below to email it to a friend
Email this content
or copy the link directly:

https://www.elibrary.imf.org/view/journals/024/1980/002/article-A004-en.xml
The link was not copied. Your current browser may not support copying via this button.

Link copied successfully

IMF Staff papers: Volume 27 No. 2

Author: International Monetary Fund. Research Dept.

Volume/Issue:: Volume 1980: Issue 002

Publisher:: International Monetary Fund

ISBN:: 9781451946864

ISSN:: 1020-7635

Pages:: 250

DOI:: https://doi.org/10.5089/9781451946864.024

Keywords
I. Effects of Exchange Rate Changes on Contract Prices and Unit Values of Exports
II. Effects of Exchange Rate Changes on Import Prices, Import Unit Values, and Terms of Trade
III. Empirical Results, January 1973—April 1978
IV. Conclusions
APPENDICES
- I. Transmission of Exchange Rate Changes via Wage Movements to Export Prices and Unit Values
- II. Data Sources
REFERENCES

View raw image

Short–Run Terms of Trade Effects of 10 Per Cent Devaluation¹

View raw image

Short–Run Terms of Trade Effects of 10 Per Cent Devaluation¹

View raw image

Cumulative Effects of 10 Per Cent U.S. Dollar Devaluation on U.S. Terms of Trade and Trade Balance Under Alternative Assumptions Regarding Response of Supplier Prices of Raw Materials¹

Export unit values International Monetary Fund,	International Financial Statistics (IFS), line 74
Import unit values	IFS, line 75
Export and import weighted effective exchange rates are computed as weighted averages of bilateral exchange rates. In turn, these rates are computed as the ratio of a country’s local currency price per U. S. dollar to the partner country’s currency price per U. S. dollar;	IFS, line of
the weights are trade shares computed from trade flows in 1973 U. S. dollars	Organization for Economic Cooperation and Development, Statistics of Foreign Trade, Series B
Competitor prices and supplier prices are computed as a weighted average of partner countries’ export unit values; the weighting schemes are the same as those applying to the export and import weighted effective exchange rates
Wholesale prices	IFS, line 63
Wage rates	IFS, line 65, interpolated
Material prices are a weighted average of commodity prices other than petroleum prices and petroleum prices in U. S. dollars; the weights are the shares of raw materials other than fuel and fuel in the cost of total material inputs in semifinished and finished manufacturing output in 1970 (1967 for the United States)	IFS, lines 76 ax and 76 aa National sources
Merchandise imports and exports in 1977 …	IFS, lines 77 abd and 77 aad

Export unit values International Monetary Fund,	International Financial Statistics (IFS), line 74
Import unit values	IFS, line 75
Export and import weighted effective exchange rates are computed as weighted averages of bilateral exchange rates. In turn, these rates are computed as the ratio of a country’s local currency price per U. S. dollar to the partner country’s currency price per U. S. dollar;	IFS, line of
the weights are trade shares computed from trade flows in 1973 U. S. dollars	Organization for Economic Cooperation and Development, Statistics of Foreign Trade, Series B
Competitor prices and supplier prices are computed as a weighted average of partner countries’ export unit values; the weighting schemes are the same as those applying to the export and import weighted effective exchange rates
Wholesale prices	IFS, line 63
Wage rates	IFS, line 65, interpolated
Material prices are a weighted average of commodity prices other than petroleum prices and petroleum prices in U. S. dollars; the weights are the shares of raw materials other than fuel and fuel in the cost of total material inputs in semifinished and finished manufacturing output in 1970 (1967 for the United States)	IFS, lines 76 ax and 76 aa National sources
Merchandise imports and exports in 1977 …	IFS, lines 77 abd and 77 aad

Export unit values International Monetary Fund,	International Financial Statistics (IFS), line 74
Import unit values	IFS, line 75
Export and import weighted effective exchange rates are computed as weighted averages of bilateral exchange rates. In turn, these rates are computed as the ratio of a country’s local currency price per U. S. dollar to the partner country’s currency price per U. S. dollar;	IFS, line of
the weights are trade shares computed from trade flows in 1973 U. S. dollars	Organization for Economic Cooperation and Development, Statistics of Foreign Trade, Series B
Competitor prices and supplier prices are computed as a weighted average of partner countries’ export unit values; the weighting schemes are the same as those applying to the export and import weighted effective exchange rates
Wholesale prices	IFS, line 63
Wage rates	IFS, line 65, interpolated
Material prices are a weighted average of commodity prices other than petroleum prices and petroleum prices in U. S. dollars; the weights are the shares of raw materials other than fuel and fuel in the cost of total material inputs in semifinished and finished manufacturing output in 1970 (1967 for the United States)	IFS, lines 76 ax and 76 aa National sources
Merchandise imports and exports in 1977 …	IFS, lines 77 abd and 77 aad

Topics

Countries

Series

About

Resources

Abstract

I. Effects of Exchange Rate Changes on Contract Prices and Unit Values of Exports

II. Effects of Exchange Rate Changes on Import Prices, Import Unit Values, and Terms of Trade

III. Empirical Results, January 1973—April 1978

IV. Conclusions

APPENDICES

I. Transmission of Exchange Rate Changes via Wage Movements to Export Prices and Unit Values

II. Data Sources

REFERENCES

Same Series

Other IMF Content

Other Publishers

Asian Development Bank

Food and Agriculture Organization

Inter-American Development Bank

The World Bank

Share Link

Table of Contents

Headings

Figures

Metrics