The Effects of Recent Exchange Rate Changes on the U.S. Trade Balance

K.7 ‘ (#725 in RFD Series)

INTERNATIONAL FINANCE DISCUSSION PAPERS

THE EFFECTS OF RECENT EXCHANGE RATE CHANGES ON THE U.S. TRADE BALANCE

by

Peter B. Clark

Discussion Paper No. 52, September 9, 1974

Division of International Finance

Board of Governors of the Federal Reserve System

The analysis and conclusions of this paper represent the views of the author and should not be interpreted as reflecting the views of the Board of Governors of the Federal Reserve System or its staff. Discussion papers in many cases are circulated in preliminary form to stimulate discussion and comment and are not to be cited or quoted without the permission of the author.

wn et ET Oa NT aerate

The Effects of Recent Exchange Rate Changes on the U.S. Trade Balance*

Peter B. Clark

As a result of the multilaterai exchange rate changes that took place between the first quarter of 1971 and the second quarter of 1973, the dollar depreciated on a trade-weighted basis by about fourteen percent.? This paper presents some estimates of the change in the value of U.S. merchandise imports and exports over this nine-quarter period that may be ascribed to these adjustments in exchange rates. In the first section of the paper an attempt is made to calculate the rise in the dollar prices of U.S. imports and exports brought about by the depreciation of the dollar. The second section then gives the results of estimating import and export price elasticities of demand. Finally, in the third section the calculated price increases and import and export demand elasticities are combined to yield estimates of the impact of the dollar depreciation on the U.S. trade balance, quarter by quarter, between 19711 - 197311.

This paper is frankly in the partial equilibrium mold.” The one step in the direction of a general equilibrium analysis is the calculation of the impact of the dollar depreciation on the prices of U.S. imports and exports. However, the induced effects of exchange rate changes on domestic prices, both in the United States and abroad, are not measured and thus are implicitly assumed to be zero.” Similarly, the real income effects of the dollar depreciation are also ignored. Since the induced responses in the levels of domestic prices and real

income may partially or fully offset the international price advantage

deme ne eT eal

-2e

for U.S. imports and exports prought about by the exchange rate changes, the estimates set forth here provide an upper pound for the full mutatis mutandis effects of the depreciation of the dollar during 1971-1973." 1. Effects of Exchange Rate Changes on the Prices of U.S. Lmports

and Exports

For a country that looms 4s large in world trade as the United States, it is unlikely that the domestic-currencys i.e., U.S. dollar, prices of its imports and exports would rise by the full amount of a cepreciation of the dollar. A knowledge of supply elasticities would, together with demand elasticities, enable one to compute the percentage change in U.S. import and export prices associated with any given dollar devaluation. The estimation of such supply elasticities is, however, notoriously difficult. First of all, in the case of trade in finished manufactures, the assumption of perfectly competitive conditions re quired to derive a supply curve is itself questionable. And second, for those classes of goods, @-B+> agricultural commodities, whose prices are determined in highly competitive markets, the proper specification . - of supply conditions requires extensive institutional knowledge that the author jacks.

For these reasons the following expedient has been adopted in this paper: U.S. imports and exports have been dichotomized on the basis of the degree of competition in the market in which they are bought and sold. On the one hand, there are commodities that have standard char~

acteristics and are traded on organized international commodity markets.

-3-

Arbitrage tends to make the prices of these goods in different countries differ by no more than the exchange rate and freight, insurance and interest charges. Such arbitrage is not perfect, however , because of tariffs, quotas, and special arrangements such as the Common Market's variable levies on agricultural imports. Nevertheless, for these commodities the perrectly competitive model is a good approximation,

so that the impact of the multilateral exchange rate changes during 1971- 1973 on the numeraire (in this case, the dollar) price of these commo~ dities can be calculated from knowledge of the magnitude of the exchange rate changes and on the basis of assumptions regarding supply .and demand elasticities. On the other hand, there are commodities that are not homogeneous, being characterized by substantial product differentiation. These goods are not traded on organized markets, and sellers have a substantial degree of discretion in setting prices. Since for these commodities the assumptions of perfect competition do not hold, the impact of the depreciation of the dollar on their prices is estimated

by using an alternative pricing assumption.

A. Multilateral Exchange Rate Effects on Dollar Prices of Traded Goods Under Perfectly Competitive Conditions

The approach adopted here is an application of the methodology suggested in a recent paper by Ridler and Yandle [1972] and described py Yandle in his contribution to this volume.° They show that under conditions of perfect competition the effect of a multilateral exchange rate realignment on the numeraire price of an internationally-traded commodity depends on the shifts in each country's export supply and

import demand in response to the change in its exchange rate vis-a-vis

-h-

the numeraire currency. This result is derived here in a somewhat more general context. .

Consider the total demand (D) and total supply (S) for a particular commodity where there are n countries in the world. Abstracting from all other factors affecting demand and supply except the world market

price, and assuming constant demand and supply elasticities, we have:

n n b; (1) Ss = 2154 = 2194 BaP? as > 0; b; > 0. 2 n n dq. (2) D= pera = 42404 BaP? c, ? 0, d. < 0. . where Ry = price of the numeraire currency in terms of the it country's currency

p = price of the commodity in terms of the numeraire currency. To determine the percentage change in the numeraire price in response to multilateral exchange rate changes, set D = § and take total differentials, thereby obtaining:

n dk, n dR,

n = - ,2,b )/ (E> 7 Ed ,D,)-

dP CE 14yPy R, i=l Pa R, i i

(3) > *

Thus the change in the numeraire price of the commodity is seen to depend on supply and demand responses induced by the exchange rate adjustments.

Alternatively, the change in the numeraire price can be expressed in terms of export supply and import demand elasticities. In this context of homogeneous goods, exports &,) represent Noxcess" domestic supply, i-e-> the difference between total domestic supply (s,) and

total domestic demand (D;)+

-5-

(4) x, = s, - Ds where the index j refers to exporting countries. Similarly, imports (M,) represent "excess domestic demand: (5) M, = Ds ~ 5, where the index i refers to importing countries.

Since by definition:

(6) mn, = wig _ =i b i M, i My i (7) x, = "iy - "i d j %; j %, j where mn, = import elasticity of demand x, = export elasticity of supply, *

it follows that the percentage change in the numeraire price can also be expressed in terms of these two parameters and the quantities im- ‘ported and exported:

h dk, . Ry k h

a = Gama RD 7 52144 R RONGE M;)

(8) ger%y%y 7 42a" where there are h importers and k exporters, and h + k < n. For empirical applications equation (8) is more useful than equation (3) because data are more readily available for the exports and imports of commodities

than for the total amount produced and consumed domestically. The only drawback to using equation (8) is that, strictly speaking, it is valid only at a point, since m, and x, are themselves functions of the exchange rate.

The expression given by equation (8) can be used directly to compute

the change in the numeraire price of an individual commodity brought

-6~

about by the exchange rate adjustments of many countries. However, for the purpose of analyzing the effects on the entire trade account of an exchange rate realignment, it is necessary to consider broad aggregates of commodities and measure the weighted average of the changes in the prices of the individual commodities making up the aggregate. Using

the share of the individual commodity in the aggregate as the weight

for that commodity greatly simplifies the computation. Thus the weighted average, or index, of price changes in a class of K commodities, repre-

sented by ¢ , is given by:

=> &«K dP dP = —t (9) P~ 521"4 P, J where M xX -t.—1 Ws 7 x? and _th , M, = total world imports of j commodity X. = total world exports of ha commodity J K K M = ,f,.M, = X= .2,X,, i1.€e, total world imports of the K commoj=l j jrl j

dities equals total world exports of the same commodities.

If one had estimates of the import demand and export supply elas-

“ ticities for each of the K commodities, then one could use this information, together with data on imports, exports and exchange rate changes, to compute the weighted average price change. However, for the broad aggregates considered here, which include thousands of individual commodities, such information is not available. Therefore the simplifying assumption was made that the import demand elasticity (m) is the same

for all commodities in the aggregate and for all countries, and similarly

ce coe ang MR om rar re gm smn RE Om Pe Ce I a ae eR EE LLP EN AS I ee

ay Pa

for the export supply elasticity (x).! ‘On the basis of these assumptions 4t can be shown that the weighted average of the percentage changes in

the individual commodities is given by:

2 8 omy xX, dR, qo) 5 hi Gan Gem,” where M, = total imports of country i of the K commodities included in the aggregate x, = total exports of country i of the K commodities included in

the aggregate.

Thus the expression given by equation (10) is in fact a weighted average of exchange rate changes, where the weights depend on the import demand and export supply elasticities as well as the country's share in world imports and exports of the class of commodities in question.

Equation (10) was used to compute the impact of the 1971-1973 exchange rate changes on the dollar prices of three broad classes of commodities. These classes were dictated by the type of disaggregation used for the estimation of import and export price elasticities of demand

described in the next section. There are two classes on the side of

-U.S. imports: foods, feeds and beverages, and industrial materials and

supplies (less fuels and lubricants), and one class of U.S. exports, namely agricultural exports. It is clearly the case that not all of the items in these classes are homogeneous goods traded on international commodity markets linked by arbitrage. Nevertheless, the model of competitive markets should provide a fairly good approximation of the manner in which the prices of the commodities respond to exchange rate

changes.

ne eee eee cn MEE RAIN ON IR atte TROT IE OTE ERE en eR IRE TTR STR TT TENE PRED ERTIES FE RS RE TERE yee

-3-

In using equation (10) the dollar was taken to be the numeraire currency and therefore exchange rates were measured in local currency units per dollar, with the first quarter of 1971 taken as the base period. Ideally, one would like to have total world imports (or exports) of the classes of commodities in question and include all countries in the sample. Since such data are not readily available, the sub-sample provided by OECD countries was used instead. Thus the larger of total OECD imports oT exports for each class of commodities was used as a proxy for the world total.® In the absence of precise estimates for the import demand and export supply elasticities, a range of values for these parameters was used.” The results of using equation (10), together with the elasticities assumed to obtain these results, are given in Table l.

The results in Table 1 depend heavily on the assumption that the commodities in each category are traded under highly competitive con~ ditions. Only under this assumption will countries’ supply and demand for the commodities respond automatically to exchange rate changes. Since for some commodities there are few sellers, and there are in fact barriers between markets in individual countries, such as quotas, @x~ change control and special marketing arrangements such as the Common Market's variable levy on agricultural imports, the results in Table 1 may overstate the increase in the dollar prices of these commodities resulting from the exchange rate realignments that took place between 19711 ~ 197311. Another reason why these price rises may be overstated

is that the sample of countries is limited to members of the OECD, and

~9-

sadueyo 93f1 asuvyoxs 0} sasuodsai queserdaa sadueyo aozaid ayq snyL Ajddns a42 pue Jaqienb youve ut 7/w sem AQToOFAseTSa pu asucdsea ATddns ay 3eyR pure avek auo UTYIFM aqzetduoo sea aATesay [erepsqy sy jo s1

‘sadeq aperq dogo - B2ep epral ATaaqaenb ‘satajzunod AnoZ assay 10J SuTIeT [NG eAAessy Telepeq worjy soqes

oursAo0gj JO pieog ‘assTion

a

“SoTtastqeIS TePoueuTy PeuofFJeuszaquy u

*zaqaenb yuezano ay} se [Tem se siaqaenb snofAeid uF ‘zajaenb ysee uf g/x 07 Tenba asuodsal qeyq pue ‘sieak omy uTYITM ajeTdwod sem

Bwap 34I asuodsai puewep ay} 4vyy peuwnsse sem 3] *uwaqshS ‘y udayyey fq saftiodajzed esn—pue OUT pasueiieed J uaafZ3 saqer yRuoW-jo-pus jo aBeravAe

RE

‘eqTaeTsosn, pue pueyTeoy ‘Aexany ‘aoee19 wey. 13420 SafTiqunod [Te oJ — saqzeA adueyoxy

Ayyep jo a8eraae ATAaqaenb

:s90ano0g

mmm

047°0 08°0-

ord 91°9 zS°¢ €0°S L6°€ LS°Z oz'T 07°0 -40°0

(€)

Sseytddng [eTaysnpuy Teanypnoyasy sntd seS8eiaAeg pue spseaq ‘spoo4

9¢°0 79°0-

40°L €8°s oT°S 09°47 6S°¢ o£? 70°T cse°0 90°0

(Z)

90°9 76°7 62°97 TB°e L6°¢ T6°T 88°0 62°0 s0°0

(1)

s°0 O°T O°I- $L°0- L6°8 0z°8 Led z9°9 a) 09°S GL°S L8°9 69° 9L°€ T6°2 Zu°2 Let 4T°T 97°0 6£°0 80°0 L0°0 (€) (Z)

a ne

TL°L ct’9 cO°S 90°47 9o°€ 80°? 86°0 £e°0 90°0

(T)

(squeosfaqnt pue stenj ssot) sefttddng pue STeyjzsqe_ TeFaAysnpuy

o£ °0 0L°0-

70°9 c0°S 93°49 0c*? Sere LT’? cO°T 9€°0 L0°0

(€)

Saseloaeg

07°0

9S°0-

L9°S 49°97 Lv o8°€ TO’ 76°T 16°0 ze°0 90°0

(Z)

0s*0

07° 0-

Oz°sS

€c°"

69°¢.

oe°'e 6S°% 99°T L£L°0 92°0 ¢0°0

(1)

pue speay ‘spooj

(ITL6T worly eseeroUT adequeoied) sdnoig A3Fpoumoy pazoaTes JO S8dF41d AeTTOd ods au} uo sasgueyg aqey odueyor|

6I-TL6T 243 JO QOedWE *T PTIFL

x uw

SOFIPOFISeTY uny-3u0] TI€Z6T

1el6t

AIZL6T IIIZL6T TIzL6T

1cL6Tt

AITL6T ILITL6T

TITZ6T

za qaeno

Leer me

1 ae EERE EE Oe RRR INTE © Nera

ap ceoeppe eteee so narap tapey eames 1 <tiggiges —reaoeaagatyes gutta atin p as AMIgrEnting ty *.

-10-

thus it leaves out a large number of countries that devalued relative to the United States during this period. While the results in Table 1 may therefore err on the high side, they are nevertheless considerably lower than the fourteen percent trade-weighted devaluation of the dollar between 19711 and 197311.°

_ ‘The figures reported in Table 1 are for the percentage changes in spot prices. Because of delivery delays, forward exchange commitments and long-term contracts, however, the value of current-quarter imports and exports will be less a reflection of current-quarter spot prices than spot prices in previous periods. As a rough correction for these lags it was assumed that current-quarter U.S. imports and exports (for the commodity class in question) reflect the dollar price that occurs in the previous quarter. Thus for the results reported in section III, which gives the combined price and quantity effects of the devaluation, the figures in Table 1 were lagged one quarter.

B. Exchange Rate Effects on the Dollar Prices of U.S. Imports and Exports Under Imperfectly Competitive Conditions

Not all commodities that enter the U.S. trade account have their

_ prices determined under highly competitive market conditions. Many finished manufactures are differentiated products and are not sold on organized markets. If there are few sellers of the commodities exported and imported by the United States, then tthe producers of these goods will have considerable discretion in setting prices. In fact, there is some direct evidence (see Dunn {1970]) that exporting firms are able

to administer the prices of traded goods. Furthermore, the occurrence

of dumping implies that firms have the ability to discriminate between

re eee emer mpm vee NE TT TE open eer epee TENT

-11l-

domestic and foreign markets. Finally, such discrimination is consistent with the large discrepancy that is observed between the change in export prices and in wholesale prices in those countries that revalued by significant amounts against the dollar between 1971 and 1973.

One approach to this price-setting behavior is to formulate a profitmaximizing model of a discriminating monopolist that takes account of the ability of the seller to segment the domestic from the foreign, i.e€-; export market. One such model was developed in a previous version of this paper." It unfortunately involves the difficulty that it does not lead to an explicit expression for the export price in terms of all the ex~ ogenous variables in the model. In addition, it implies that if demand elasticities are constant, the relationship between the domestic price and the price charged to foreign buyers for the same commodity will also be constant, which, as noted above, does not appear to have been the case during the recent period of exchange rate adjustments.

Therefore instead of pursuing this explicit profit-maximizing avenue, a more heuristic approach based on the notion of a markup equation has been adopted here. The basic idea is that price, P, represents some

‘markup over standard unit input costs:

(11) p = ycuic’ + uc’) where . y = markup factor, where y ? 1 ULC. = standard unit labor costs N . 12 UMC = standard unit materials costs.

The price is set by the firm so as to cover all costs of production,

cee we meee men net ELA EE MAMAN PRP EARE THOS

cee me nee Rh RM ELE OT LIPOIC ES eee TURES NT Ta

-12-

including capital costs, which are included in y. It is also assumed that the firm does not adjust its price to every change in actual costs, but rather responds only to changes in “ctandard," i.e., long-run or expected unit input costs.

Even though the firm can set its price, it need not therefore be impervious to demand conditions. In particular, for a firm that sells the same or similar goods both to domestic residents and to foreigners, the markup on goods sold abroad would tend to reflect both foreign and domestic demand conditions. By raising the price charged to buyers abroad in the face of increased foreign demand, the firm can directly increase its profits. The firm may also vary its markup on exports, and hence its export price, in response to an increase in domestic demand for the good. The markup on the exported commodity can therefore be

represented by:

(12) YF £(K, CUys CU;s PC,) where LK. = capital costs cu; = domestic capacity utilization rate CU, = foreign capacity utilization rate PC, = the price of foreign goods that compete with the export good.

In equation (12) the markup factor varies positively in response to changes in the arguments in the function. The capacity utilization variables are proxies for domestic and foreign demand conditions. The competitive price variable, PC es is designed to take account of shifts

in foreign demand caused by changes in the prices of goods that compete

-13-

with the export commodity. In particular, it is assumed that the exportcd commodity is an imperfect substitute for, and therefore competes with, the exports of other countries. This competitive price variable has two components: an exchange rate term, R, and the price of competitive exports expressed in foreign currency, PXes since PC, is defined as PC, =R°* PX ee A depreciation of the exporter's currency, for example, will raise R and thereby shift demand to the exporter because of the rise in competitors’ prices (expressed in the exporter's currency). The exporter can therefore increase his profits by raising his markup and therefore his price. The competitive price variable thus provides the link between exchange rate changes and export prices.t>

Combining equations (11) and (12) and assuming that capital costs, K, are constant, P can be expressed as a log-linear function of the following explanatory variables: (13) InP =a,+ a, In CU, + Gy in CU, +O, In PC,

0 1 ea nme’ +a, nM. 4 5 The log-linear form has two justifications. The theoretical rationale is that in the absence of constant returns to scale, explicit profitmaximizing models indicate that the price depends on both cost and demand factors in a non-linear manner." From the pragmatic viewpoint the log-linear form makes it possible to decompose the PC. term into its

two components and estimate the sepdrate influence of each on the export

price.

- °

eer rr tee eon NIT OIL EO ERR ER eee ts

were mm eT

~14- 1. Impact of the Dollar Depreciation on the Price of U.S. Exports of Manufactured Goods

A modified version of equation (13) was used to estimate the extent to which the dollar prices of U.S. exports of manufactured goods rose as a result of the effective depreciation of the dollar during 1971-1973. First, the vic’ variable was separated into compensation per man-hour (W) and output per man-hour (OMH) in order to take account of the possibility that firms may respond differently to changes in the two components in unit labor costs. The second modification involves the specification of standard unit input costs. Four-quarter moving averages were used as proxies for standard W, OMH and UMC in initial estimates of equation (13), but the statistical results were somewhat inferior to those obtained using the actual values of the regressors. Therefore, these actual values were used in the results reported here.

In estimating equation (13) P was represented by a weighted average of the unit value indexes for U.S. exports of semi-manufactures and finished manufactures, the weights being the relative proportion of these two exports in 1965-1966. Since P is expressed in dollars, the competitive price term, PCE» was also expressed in dollars. This varjable was constructed as a weighted average of seven countries’ (Belgium, Canada, France, Germany, Japan, Netherlands, and United Kingdom) unit value or price indexes for exports of manufactured products, with shares of exports of manufactures in 1965-1966 as weights.?> These same weights were used to construct a weighted average of the aggregate Wharton capacity utilization indexes of the same seven countries, CU,- A similar

f£ index, CU,» was used for the United States. W and OMH were for the

-15-

total U.S. manufacturing sector. Finally, UMC was approximated by the wholesale price index for crude materials less food. All variables are seasonally-adjusted index numbers. Data sources are given in the appendix.

The foreign capacity utilization variable, CU; was consistently insignificant and therefore does not appear in the results, reported in Table 2, which are based on quarterly observations over the period 19631 - 197311. A second degree polynomial distributed lag over six quarters and constrained to zero at both ends was used for the wage rate, W. A second degree polynomial distributed lag over four quarters and constrained to zero only at the far end was used for output per man—-hour, OMH. Both variables are significant and have reasonable coefficients. The U.S. capacity utilization variable lagged one quarter, Uy has a significant coefficient, indicating that U.S. exporters of manufactured goods vary their prices to foreign buyers in response to overall demand pressures in the United States. Finally, the proxy for. unit materials cost lagged one period, UMC Ly also has a significant influence on the export price.

The results in Table 2 indicate that prices of U.S. manufactured . exports do not respond positively to increases in the prices of competitive exports, PC, In regression 1 the current value of PC. has a negative sign, and in regression 2 the sum of the coefficients of a second degree, four-quarter polynomial distributed lag constrained at the far end also has a negative sign. In regressions 3 and 4 the PC.

variable was detomposed into a weighted average exchange rate term, R,

ane mame UE ret Toners cemepeammaraccmes ene te eel penn al Soe eee 8 wr mee

-16-

“squapoTgye0e Bet eur 5° uns ay} 10JF 21e

(ot z-) (97° 1) HNO pue M 193 gopast ze3s—-37 pue sjusToOFZJe09 ea :390N €t°- =o . & (gz°Z-) (09° 1) gi°- ST’ Z (61° €-) (¥2") 1L0°- 120° is (12° 1) (€8°-) a 1 a 0 ssey (91° €-) (92°) (11° 9) (67°9) (67°9) (9€° 7) (c9°T) . 1g00° 720°? €66° 7zZ°- 690° gy" - TL* 42° 07° TS°2 9 , (69°-) (€L") St° 1) (z1°9) (9S°T) (98°17) (cet) zgoo’ Le°t 766° 760° 1° 67° 69° 9T° 9e° 10°? € (€6°-) , 9€0°- € (LI°1-) L70°~ Z (€6°1-) Teo" T (Z1°) L600° 0 sae] (€6°T-) (1L°€-) (17° OT) (zE°€) (€s°€) (6€°9) ; 9g00° 8i'T €66° (0) ie Gyo LL° €2° (Aa 99°€ cA (0S°1-) (L0°9-) (S6°0T) (67°€) (78°) (18°9) ggoo’ €eL"T 766° gl0°- 8y7°- 8L° €2° 7° €7°e T —- OT —_- kK, an =) -3 as Md zu a Iya Fog HNO M t own t no queqsu0) uoyssei3eu

(TIEL6T - I€96T) SPOCD peanqoegnuey yo sqzodxq *S°N xoj xepul entea 3FuN VO uopyepoeaded ABTTOd Jo 399534 Zz etd

-17-

and a weighted average of competitive export prices expressed in foreign currency, PX, The current values of these variables enter in regression 3, and in regression 4 they enter with the same distributed lag used for PC. in regression 2. Since the exchange rate variable--which increased by 22 percent between 19711 and 19731I--has a negative coeffi-~ cient, the results in Table 2 are not consistent with the hypothesis that U.S. exporters of manufactured goods raised their dollar prices to foreign buyers as a result of the recent depreciation of the dollar.?/ This result will be used in section III in computing the change in the value of U.S. exports of manufactured goods that can be attributed to

the exchange rate adjustments between 1971 and 1973.

2. Impact of the Dollar Depreciation on the Prices of U.S. Imports of Finished Manufactures

All that remains is to estimate the effect of the 1971-1973 adjustments in exchange rates on the dollar prices of U.S. imports of finished manufactures, which include capital goods, consumer durables and nondurables, and imports of automobiles and parts other than from Canada. The methodology described in the previous section can be used to obtain an estimate of this effect. It is first necessary to note that the dollar import price, PM, is identically equal to the exchange rate, Res times the, foreign export price, PX?

(14) PM = Re ° PXee In the preceding section it was argued that a country's export price for manufactured goods will be a function of its exchange rate. There-

fore in order to estimate the impact of the dollar depreciation on the

price of U.S. imports of finished manufactures, one must calculate the

-18-

extent to which foreign exporters adjust the foreign-currency prices, PXp> of the goods they ship to the United states.+°

One can proceed in either of two ways. One can measure the impact of Rg on PX, and then use (14) to calculate the effect on PM. Alternatively, one can substitute the variables determining PXe> which include R, directly into (14) and estimate the effect of Re on PM in one step. In the absence of measurement and specification errors the two methods would yield identical results. However, since such errors are present, both estimates have been made.

One approach to estimating the impact of Re on PX, would be to calculate for each country exporting to the United States the effect of the change in its exchange rate on its own export price. The change in the U.S. import price could then be computed as a weighted average of these export price changes. This disaggregated approach would allow for differences in the response to exchange rate changes across countries. For simplicity, however, an aggregative approach was used instead. This involved estimating an equation using a weighted average of foreign countries' export prices as the dependent variable. The same variables -used in this equation were then substituted into (14), where PM is the dependent variable, to obtain a second estimate of the rise in U.S. {import prices of finished manufactures resulting from the dollar depreciation.

In the initial estimation involving PX, as the dependent variable a modified version of equation (13) was used:

(15) in PX, = By) + By in Cys * Bo In CU, + 8, In UVM

+ B, in We + Be ln Re + Bet +e°

cece ee mee gmp on mabecimme tn

~-19—

where

PX. = weighted average of export prices or unit values for manufactured goods, expressed in local currency.

CU... = Wharton aggregate capacity utilization index.

US

cu, = weighted average of Wharton aggregate capacity utilization indexes. .

UVM, = weighted average of overall import unit value indexes.

We = weighted average of wage rates in the manufacturing sector.

Re = weighted average of exchange rates (dollars per unit of foreign currency).

t = linear time trend.

€ = disturbance term. °

The weighted averages include the same seven countries used to construct PC pe Here the weights are the shares in U.S. imports of finished manufactures in 1965-1966. The overall import unit value index was used as a proxy for the cost of inputs of raw materials in these countries. A time trend, t, was used as a proxy for increases in output per worker. In this aggregate export price equation there is no obvious competitive export price similar to PC, in equation (13). One possi- -bility would have been to use the U.S. unit value series for exports of manufactured goods. This would put one in the anomalous position of explaining the U.S. import price with its export price. Rather than include the U.S. export price, an exchange rate variable was retained in order to measure the impact of the depreciation of the dollar on foreign export prices. .

In the initial estimation of equation (15) the proxy for U.S.

- demand pressure, Wyo» was consistently insignificant. In addition,

™ Ss nlomareec alae Z ATER eC ONL IT AE X Be OST ETTORE I FPO OT SEE EERONE E08 SS RRA TE A

~20-

the UVM, variable had an unrezsonably high coefficient, which probably

reflects the fact that the components of PX, make up a large proportion

f£ of UVM,. Therefore in the results reported in Table 3 both CUiyc and

f UVM. were dropped from the regression equations. Seasonally~adjusted quarterly data from 19641 to 1972IV or 1973II were used in these regressions.

Regression 1 in Table 3 shows that the current value of Re has the correct (negative) sign, but the estimated coefficient is small (-.11) and is not significant at the .05 level. However, in regression 25 Re was entered as a six-quarter second degree polynomial distributed lag constrained to zero at the far end. The individual lag coefficients and their standard errors are given in Table 4. The sum of the lag coefficients is -.32, indicating that a 1% depreciation of the dollar against the seven countries whose currencies are included in Re results in a cumulative -.32% decline in manufactured goods’ export prices of these same countries. This decline in foreign export prices implies a longrun increase in dollar import prices, or pass-through of the exchange rate change, of .68%.

In both regressions 1 and 2 foreign cost conditions are represented by wage rates in manufacturing and a time trend. As an alternative to measuring costs directly, a weighted average of foreign wholesale prices of manufactured goods, WPI., was used as a proxy for the total costs of producing the exported good. 7? In regressions 3 and 4 in Table 3 WPI.

has been substituted for W, and the time trend. In regression 3 the

£

current value of Re was entered, whereas in regression 4 a five-quarter

~ o

cee a nn Ao aepes Kem oe cA RS ee ERT a eR eT Rat Tamme pment

-21-

sywnpoo asset aa up, aAoge Saver” een ainps pereut ase ayy ate" poqsodad suopssatsad qeuty e473 Ut pesordue sem ginpasoad NjT-YAe4Prrnm — te *g-T suotsseisor JO soqeutyse TeTITUT ut stenpTsezt eu ut uote [et1090 Ne queopyyustTs 5° asneseq (7) {

‘pue rez oud qe poufez3su0? “t'prd zoqaenb-xTs +40 -oAfi ® Jy pue ‘pue 1235 aur qe pouterysuo? “rt pid yaqaenb-aAF¥s ‘goa8ap puosves # wats Jzam ‘spue y7o4d ye pouferqsuo0s Cpe Bet peanqFszastP yepwoustod qaqaenb-xTs ‘g918ap puores & yatn yotsse13e1 344 peraque atqeraes “MN auL (€)

eee

,

*pesn st 3a jo anzeA qweainy ay azoyh L pue ¢ ‘€ ‘T guopsseiset UT qdaoxe gyuatotyjeoe ZeT ey? jo wns ay} 105 aiesy pue stam ‘3m roy Sx10119 paepue3s pue gyvatotsje09 (Z)

-quapoTyzaoo Yow? MOoT2q gosoyqueied Ut UdsATs 312 so73st3e3s—-4 (1) +S930N

(16°9 (81°72) (02° 2) (7€° 97) (r1eL - 179) 110° ZL°T 766° zs" 6e°T 47° 16° 7- Wd °8 (ze"L) (90°92) | (18°T) (et 97) (11eL - 179) i TT0° €9°T 766° 94° ge°T 02° gS°t7 Wa “L (3 (22° 7) (96°€) (€8°-) (€S°?-) (6€°T) (AIZL - 149) i Ov". 110° 166° gy" 73° 61'- Z10°- os’t: Wa °9 ‘ (61° T) (LS° 4) (61°) (16°77) (62°) (AIZL — 149) on" TT0° 066° ZZ" 00°T 120° y10°~ "Z° “Wa °S (08° 7-) (05°02) (ss0’-) — (86°-) (11€L,~ 179) 08° ze00° 866° Oc’ g0°T €£00°- gt’- Xd °% (09° T-) (ge ° OT) (7e°) (gt’) - (11€L ,~ 179) 06° g£00° 866° c80°~ 66° 770° oT’ Xd “Ef (10°S-) (27° TT) (ST’) (e47°L-) (el°o) (AIZL,~ 1479) os’ 6z00° 866° Ze°- 18° 7900° Z10°- ct°T Xa ‘2 (0S°T) (g8°S) (91° T) (oe €-) (1L°€) (AIZL,~ 179) OL’ 9€00° 866° tl'- 79° TLO° ¢800°- 9s‘T xd “TT (1-3 6 as na a 5, jaan 7M T-)3qq 0 OWEL queqsuog aTqeFzeh é qyuapueded

\

spoo) peangoe ynuen jo santea qyug 310dxa uspero0g pue jzodwy *s*n uo yop epoeaded aeqttoa 5° 309334

. 14d

-22=

polynomial distributed lag constrained to zero at the far end was used for Ree It is reassuring that the estimates of the effect of a dollar depreciation on foreign manufactured goods" export prices (-.082 using

the current Re and -.30 for the long-run effect) are very close to those

obtained in regressions 1 and 2, indicating that these estimates are

not highly sensitive to the manner in which foreign costs are measured. Table 4

Distributed Lag of Exchange Rate on U.S. Import and . Foreign Export Manufactured Goods Unit Values

Dependent Current Variable Quarter t-1 t-2 t-3 t-4 t-5 PX, -.83 -.073 -.062 -.048 ~-.034 -.018 (regression 2) (-2.28) (-4.07) (-4.66) (-2.9) (-1.97) (-1.51) (regression 4) (-1.39) (-4.50) (-4.00) (-3.07) (-2.63)

PM oil ell 093 076 2055 .030 (regression 6) (.83) (1.68) (2.04) (1.21) (.84) (.65)

PM ~20 14 -10 056 026 (regression 8) (2.04) (6.78) (2.37) (1.00) (.57)

t-

The estimated increase in the dollar prices of U.S. imports of

‘finished manufactures is roughly seventy percent when a weighted average

of foreign export prices is the dependent variabie. Tf the variables

used to explain PX are substituted into equation (14), however, the

estimate for the pass-through obtained from the resulting equation is

somewhat loweT. The equations underlying regressions 1-4 in Table 3 are:

B, B, B Bt _ 2°45. 6 (16a) PX = ByCUs We Re e €

~23-

B, B B 8 28s P7 (16b) PX, = ByCUs Re WPT ¢ €

Substituting these expressions for PX- into equation (14) gives:

(17a) PM = B Us We Re B 146. Bo (17b) PM = BCU. Re WPI, Ee

Equations (17a) and (17b) were estimated in log-lincar form and the results are given in regressions 5-8 in Table 3. The distributed lag coefficients and their standard errors appear in Table 4. Since -1< 8. < 0, the coefficient of Re in these regressions, 148.5 should be less than unity but greater than zero. It therefore provides a direct estimate of the effect of a devaluation of the dollar on the prices (as represented by a uit value index) of U.S. imports of finished manufactures. Regressions 6 and 8 indicate that the long-run, i.@+, five or six quarter, impact is about fifty percent of the weighted average change in the dollar exchange rate.”

Thus the estimated increase in the prices of imported finished manufactures is somewhat lower when equations (17a) and (17b) are used instead of (16a) and (16b). One reason for this discrepancy may be that ‘foreign exporters discriminate between the United States and other export ’ markets in terms of the export price that is charged to foreign purchasers. In other words, an exporter may reduce the price of his own currency by a greater amount for U.S. buyers, since over the 1971-1973 period the dollar depreciated to a greater extent than all other currencies except the pound. Buyers in other countries may receive a smaller reduction

in price. Thus the change in the export price or unit value index,

egy co cmgre SEFENPERO TTIE TN 2 PRY pt SATE see eer SRT IEE TO OS II II FO wee ec Neematee coer

~2h-

expressed in the exporter's currency, would not properly measure the fall in price to U.S. buyers because it is a weighted average of export~ price changes to many different countries. These considerations argue in favor of using the results from the regressions where the unit value index for imports of finished manufactures appears as the dependent variable. Therefore the distributed lag results for the pass-through obtained from regressions in Table 3 with PM as the dependent variable were used in the calculation of the impact of the dollar devaluation on the quantity and value of imports of finished manufactures. II, Estimation of the Responsiveness of U.S. Imports and Exports to

Exchange Rate Changes

In this section regression analysis is used to obtain estimates of the price or exchange rate responsiveness for five categories in the U.S. trade account. On the import side regressions are estimated for finished manufactures, foods, feeds, and beverages, and industrial materials and supplies (exclusive of fuels and lubricants). On the export side, because unit values indexes are not available before 1967 on an end-use breakdown, regressions were estimated for two broad categories of commodities, namely, agricultural and non-agricultural goods.

A. Finished Manufactures

For both imports and exports estimates of price elasticities were obtained by the common procedure of explaining 4 real trade flow in terms of an aggregate activity variable and the ratio of the price of imports to the price of domestic substitutes. Following Marston {1971},

for imports of finished manufactures the activity variable was decomposed

eee ~ . a" pert Sowers

-25-

into trend and cyclical components in order to allow the response in

imports to vary depending on the extent to which the activity variable . 21 . :

differs from its trend value. The basic equation for imports of fin-

ished manufactures is given by:

FM _ oy A> (14T)PM, "3 aa, Bie agry G/T) “Mr” US where: FM = imports of finished manufactures, seasonally adjusted annual rate PM = unit value index for finished manufactures, not seasonally adjusted XY = real GNP, seasonally adjusted annual rate, 1963$ TY = trend value of real GNP, computed by fitting an exponential time trend to the series and using the predicted value from the regression equation T = tariff rate on imports of finished manufactures WPL ic = U.S. wholesale price of manufactured goods, not seasonally adjusted Vv = disturbance term.

Equation (18) was transformed into log-linear form and estimated over the period 19631 - 197311. Four dummy variables were added to take account of dock strikes and the closing of the Suez Canal. Current values of Y and y/TY were used, whereas 4 distributed lag was used for the relative price variable. The results are given in regression 1 of Table 5.

What is noteworthy about this equation is the very nigh long-run price elasticity of -4,72, which is the sum of the lag coefficients.

In this regression the relative price term was entered as a second degree,

~26-

*0°0 09 jenba 29S aion 4a - 1d

‘papypoads asTaseuqe ssotun S11IZL96T pue TIZ96T UF OT = pq :eansot9 qeueg zeng 105 kuump = 7d

“IZL6T UF OTT POP AITL6T UF O'T ‘TTITL61T UF S°7 ™ Ea .:e4xFzIS AVOP TL6T 203 Aumunp = €a “T1696T UF O° T- PUP 16961 UF OT = 2d :oyfFas YoP 696T aoz Auump = 7d “TIS96T UE O° T- PUP {S961 UF O'T = Id soypaas YoP S96T zoz Auump = 1d (€)

ssatqepaea aseyy 103 gquetotsyaco Set out yo uns aya 105 aie si0z1e prepuers pue $quatot3sJe00 . “satqeTaeA 18420 TT? toy S3eT poangF2astP qefwoukspod pue AL/A pue AL 203 pasn aie sentes quazang) = ((Z) “yueToFgye00 YOR? MOT?4 gsosoujuesed Ut UdAFZ 912 soT3stie3s - 9 (1)

2:S030N

(ogt-) (yz TTL") (6L"~) cern) (ySt@) (82) (LL°@-) (96°S-) (12° €-) (70° T-) 620° €0°% 966" OL*Z- LOT ty- £9°T- 09°Z cyey = TLOT™ SSO ET LL0°- OT z SO an Sram Fa 7n0 Gat) (19°87) (6ot) (62) (197) (oo*e-) (29° 9-) (10° 9) (¢* L9-) 1z0° 0z*%~—sL 66" ZL° 9 06'T core 160" = SOT SAT g10°- 6 12> I Sy an as hd zt fla > C1FT) XAL/) a af va ea za tq qeysuog suoTssess3°y

(II€L6T - I€96T)

saanjoeynuey pousTurd yo sqiodwy 103 guofssa13ey

G eTqeL

~27~

20-quarter polynomial distributed lag constrained to zero at both ends. Estimates using shorter lags (16 and 12 quarters) gave lower long-run elasticities and slightly lower R's. This particular polynomial distributed lag generates a symmetrical inverted "U" shaped lag distribution with a minimum response in the current and 20th quarters of -0.061 and

a maximum response of -0.337 in the 10th and llth quarters. Such leng lags in the adjustment of trade flows to relative price changes have been found by Hutton and Minford [1972] and Junz and Rhomberg [1973].

This estimated price elasticity for imports of finished manufactures, together with the estimated response of PM to exchange rate changes described in the previous section, can be used to obtain. an estimate of the impact of a depreciation of the dollar on this category of imports. Alternatively, such an estimate can be obtained directly by substituting the expression for PM given by either equation (17a) or (17b) into equation (18). The resulting reduced-form expression for imports of finished manufactures will then include the exchange rate variable, Ree Substituting (17b) into (18), for exemple, and taking logarithms of both

sides gives: "(19) In & = In(a 0 ‘tay In(TY)+e,1n(¥/TY) +0, inl A) NE ys s] + a 38 pin(CU,) +0, (A+B. )in(R, )ta, Bs in(WPI, eine. 344n(v).

In estimating equation (19) the total length of the lag for the variables other than TY and Y/TY is equal to the sum of the length of the lag of that variable in (17b) plus the 20-quarter lag for the relative

price variable in equation (18). Thus (1+T)/WPL¢ has a 20-quarter

eee mE A ORE GON RE DIO AA! TIES ee ee

Ae ST TO ee Pe IE AOE AES NE OS eee ts

-28—

lag, CUg a 2i-quarter lag, and Re and WPI. both have a 25-quarter lag. A second degree polynomial constrained at both ends was used for these four variables. The empirical results were, however, far inferior to those reported for regression 1] in Table 5: the estimated coefficients for TY and WPI, were unreasonably high (8.90 and =-14.21 respectively), and the coefficient for Re had the wrong sign. An alternative regression was estimated with all four variables entered with the same 20-quarter polynomial distributed lag, and these results are reported in regression 2 in Table 5. The overall explanatory power of the equation is about the same as regression l. However, except for the dummy variables, the estimated coefficients have much larger standard errors. The coefficients of all three of the variables used to explain the unit value index for finished manufactures have the correct sign but are not significant at the .05 level. The lack of precision in these estimates is probably due to multicollinearity in the explanatory variables. Therefore the estimated relative price coefficient in regression 1 was employed in calculating the exchange rate impact on U.S. imports of finished manufactures.

It is nevertheless interesting to note that the estimated coefficient of Re in regression 2 is less than the coefficient estimated for PM in regression 1. As a comparison of equations (18) and (19) shows, this result is consistent with the theoretical specification that Ja5\ > |u,(1+8,) |-

In using an exchange rate variable directly in an import (or export)

regression equation it is therefore not the case that the coefficient

of this variable is necessarily the same as coefficient for the import

yen emma TART SE TE me gates MASS RT SS te ee oem HE

A I mmr mR aR GRR ET TNA RNC camera oe

-29-

price variable (PM). The coefficients will be equal only if one substitutes the right-hand side of the identity, PM = Re . PXes for PM in equation (18). In this case one would indeed expect that the estimated coefficients for Re and PX, would be the same as that for PM, namely G3.

However, if one substitutes the determinants of PXe into (18) and thereby

obtains equation (19), it is apparent that the coefficient of Ry is now

less than a The reason is that in (19) the induced reduction in PX,

3° caused by a depreciation of the dollar is taken into account in the Be term that enters the coefficient of Ree

Thus whether the estimated effect of an exchange rate change on imports is the same as that brought about by a variation in import prices depends on whether the foreign export price appears in the equation. If it does appear as a separate variable, one obtains an estimate of the exchange rate impact on imports based on the ceteris paribus assumption of unchanged foreign export prices. If, on the other hand, the variables determining the foreign export price (in particular, the exchange rate) appear in the equation, one will obtain a lower estimate of the exchange-rate effect on imports because the induced response (in an offsetting direction) in the foreign export price caused by the exchange rate change is taken into account.”

B. Industrial Materials and Supplies

The following equation was used to explain imports of industrial

materials and supplies exclusive of fuels and lubricants:

a IMS _ «IPT 1

_ (14T)uvIMS "2 UVIMS 0

(20) WPIM

ce OO ee IE AR RE I Rm SRR RAR URTE e one ROS RLgRINRNNIeNRC US Un Tt aS nee om

-30—

where:

IMS = total imports of industrial materials and supplies minus imports of fuels and lubricants, seasonally adjusted annual rates

T = tariff rate on IMS

UVIMS = unit value index for total imports of industrial materials and supplies, seasonally adjusted

IPL = aggregate Federal Reserve industrial production index, seasonally adjusted

WPIM = wholesale price index of industrial materials, seasonally , adjusted € = disturbance term

4

Equation (20) was estimated in log-linear form over the period 19661 - 197311. Data limitations precluded a longer sample. Dummy variables were added for dock strikes and the closing of the Suez Canal.

No lag was used for IPI, whereas a second degree, eight quarter polynomial

(1+T) -UVIMS

distributed lag constrained to zero at both ends was used for WIM .

Because the initial estimates indicated significant autocorrelation in the residuals, the final results reported below were estimated using the Hildreth-Lu procedure. The estimated value of the autocorrelation

coefficient was 0.50.

UVIMS s2

Constant D2 D3 D4 IPL. WIM R SE -2.00 -0.122 -0.079 -0.075 96 -1.25 0.889 0.0430

(-1.47) (-5.32) (-3.83) (-2.07) (3.31) (-2.26)

The long-run price elasticity, which is the sum of the lag coeffi-

- cients, is equal to -~1.25. The lag distribution is a symmetrical inverted

eee tmnt ce nm enenpemearempgtiiie mm tam tte gee mA ES ee OYA ay SORE

-31-

"U" shaped pattern with a response of -0.0836 in the current and eighth quarters, and response equal to -0.209 in the fourth and fifth quarters. The equation was initially estimated with IPI decomposed into trend and cyclical components. This approach yielded a positive price elasticity and consequently it was abandoned.

C. Foods, Feeds and Beverages

The following equation was used to estimate the price coefficient

for imports of foods, feeds and beverages:

_—_——— = 14+T)e (21) GHEE 7 agYD “[1+T) -UVFFB] -DDFFB 2c where: * FFB = imports of foods, feeds and beverages, seasonally adjusted

annual rates

UVFFB = unit value index for foods, feeds and beverages, seasonally adjusted

YD = real personal disposable income, defined as personal disposable income divided by the deflator for personal consumption expenditures

T = tariff rate on FFB

DDFFB = deflator for personal consumption expenditures on foods, feeds and beverages, seasonally adjusted

€ = disturbance term

Dummy variables were added to take account of dock strikes. The dummy variable for the Suez Canal closure was found to be insignificant and consequently was excluded in the final results reported below. In the initial estimation both the trend value of YD and the ratio of YD to its trend value were included as explanatory variables. Since the

latter proved to be insignificant, both variables were repiaced with the

Sil eet eee ee

Set et comers ng ee, ee ee em we ew me ee et eye ct = te Sn Cw es REP POE IIR TER ER SR aed atl oe ond . - ke Sp raQRer a et ee ast ase weet gs ' ete ~ ‘ A+ eR te 6 oe wx . att eee apes i EN tee meee

~32-

actual value of YD. Since there was some indication that the response to a price change differed depending on whether the price variable was

domestic or foreign, UVFFB and DDFFB were entered separately.

Each

price variable was estimated using a second degree, six quarter polynomial.

distributed lag constrained to zero at the far end. The equation was

estimated in log-linear form over the period 19601 - 197311 with the

following results:

Constant D1 D2 D3 YD. (1+T)UVFFB DDFFB x DW SE -4.24 -.l11 -.14 -.19 .38 -1.14 1.85 .897 1.66 .052 (-9.90) (-3.07) (-3.91) (5.33) (2.76) (-3.86) (5.54)

The distributed lag coefficients and standard errors are:

Current t-1 t-2 t-3 t-4 t-5

(1+T)UVFFB -.28 -.26 -.22 ~.18 -.13 -.069

(-1.86) (-3.49) (-3.42) (-2.16) (-1.54) (-1.22)

DDFFB 61 47 34 23 14 063

(1.81) (3.53) (4.39) (1.68) (.91) (.57)

D. Non-Agricultural Exports

On the export side an end-use breakdown was not used, as was the

case with imports, because unit value series are not available on this

basis. Rather, the series constructed by Parrish and Dilullo {1972] was

used instead. This series includes all non-agricultural exports with

the exception of automotive products shipped to Canada and aircraft.

- Also excluded are military shipments. Finally, the series was adjusted

to exclude the effects of major industrial and dock strikes and certain

other disturbances.

fein a eee . pie meee ee a ee a ee:

ey mea ete ie ae

-33-

For exports of non-agricultural goods the same procedure was adopted as in the case of imports of finished manufactures: an equation was first estimated with a relative price term and then re-estimated using an explicit exchange rate variable. The following equation was

used to estimate the coefficient of the relative price term:

a, oO a’ NX _ 1... 2, UVNX , 3 (22) TVX = a, FIP CU, qPK, v where: NX = value of U.S. non-agricultural exports, seasonally adjusted annual rates. UVNX = unit value index for non-agricultural exports, not geason-

ally adjusted.

FIP = weighted average of the seasonally adjusted industrial production indexes for Canada, Japan, the United Kingdom and continental Western Europe, the weights being the annual shares of these areas in U.S. exports. This index, together with NX, was taken from Parrish and Dilullo [1972].

CU, = weighted average of foreign capacity utilization variables.

R.*PX. = weighted average,of foreign export unit values converted f°" £ 28 . to dollar terms.

Vv = disturbance term.

In initial estimates of equation (22), which was transformed into log-linear form, the FIP variable was decomposed into trend and cyclical components. However, it was found that the actual level of FIP gave a better statistical fit when it was combined with cu, as the foreign cyclical demand variable. Initial estimates of (22) also indicated considerable autocorrelation in the residuals, and consequently. this

equation was estimated over the period 19601 - 197311 using the

SO Re ee ARR NT MNT LENT IES OLE OE ONTENTS 6 ETN I ERS. -

-34—

Cochrane-Orcutt transformation. The estimated value of the autocorrelation

coefficient in the following regression was 0.340:

UVNX —2 Constant FIP. (CUP) a Ro PX BR SE £ £ -2.02 1.03 0.933 -0.906 0.980 0.022 (-1.78) (22.0) (4.51) (-4.61)

In this regression the current value of FIP was entered whereas the lagged value of CU, was used. The relative price term was entered with a second degree, 12 quarter polynomial distributed lag constrained to zero at both ends. Longer or shorter lags gave a somewhat lower BR’. This particular lag structure yields a symmetrical inverted "Uy" shaped distribution with the lag coefficient equal to -0.030 in the first and twelfth quarters and equal to -0.105 in the sixth and seventh quarters.>

In order to estimate a direct exchange rate effect on non-agricultural exports one should take into account the impact of a dollar devaluation on both the dollar price of U.S. manufactured exports, UVNX, as well as the induced response in the foreign currency price of foreign goods that compete with U.S. exports, PX The results reported above ‘in Table 2, however, indicate that prices of U.S. exports of manufactured goods do not rise as a result of a dollar devaluation. There is therefore no need to substitute for UVNX in equation (22). The possibility remains, however, that foreign exporters may reduce the prices of those commodities that compete with U.S. exports, since a devaluation of the

‘dollar lowers the foreign currency cost of these exports. To take account

=35-

of the possibility that PX, may be a function of Res equation (16b) was

£

used to substitute for PX, in equation (22). Expressing the resulting equation in log-linear form gives:

rol NX, _ 3 yy

(23) InGuy® = 1n (a8 y+ a, In (FIP)+(o, - a.,8,) In (CU,)+a,1n (UNV) 0

+ a.,871n (WPI) -+(1+8,)a,1n (Rg) +1n(v)+1n(e

3),

Equation (23) was estimated with CU es UVNX, WPI. and Re all entering with the same 12-quarter polynomial distributed lag used for the relative price term in equation (22). The results were quite poor, most of these four variables having the wrong sign. Equation (23) was then constrained in two ways. First, By was assumed to be zero and therefore cure entered with only a one period lag, as in equation (22). Second, since we know from Table 3 that Bo is close to 1.0, UVNX and FWP were combined in ratio

form. These changes gave the following empirical results:

UVNX =2 Constant FIP (CU,) 4 WI, Re R SE -2.45 0.880 0.907 -0.633 -0.875 0.990 0.021 (-1.71) (40.4) (4.49) (-2.17) (-3.00)

Initial estimates of equation (23) indicated autocorrelated residuals and therefore the Cochrane-Orcutt transformation was used to obtain the above results. The estimated value for the autocorrelation coefficient was .38. In these results Re was expressed in foreign currency units per dollar and therefore its estimated coefficient has the proper (negative)

f£ lag coefficients, and again the shape of the lag distribution was a

sign. The coefficients reported for UVNX/WPT ¢ and R, are the sum of the

ey omens Oe et CG a MRR RE TRE eS RMR RR Rp Rn ER A OA ET ee RE Ee ee enema A Oo

[07 SS RNC REERER Ie eo RO a M oe ET E cn

-36-

12~quarter symmetrical inverted "U". The estimated coefficient for Ree -0.875, is essentially the same as that for the relative price term in equation (22), -0.903, so that the implied value of Be is zero. This means-that foreign exporters competing with U.S. exporters do not cut their prices in response to the lower foreign currency prices of U.S. exports caused by the devaluation of the dollar. This behavior is not necessarily inconsistent with the results reported in Table 3 which show

that 8. ranges between .3 and .5; in the former case it is the price =

5

behavior of substitute goods that is relevant, whereas in the latfer case @

it is the same good that is involved.-° In other words, a dollar deval-

vation may cause exporters in other countries to reduce the foreign

currency prices of the goods they ship to the United States, but it need

not induce them to lower the prices of those commodities that are sub-

stitutes for U.S. exports.

The finding that the estimated coefficient for the exchange rate term is essentially the same as that estimated for UVNX in the regression explaining the quantity of U.S. exports is consistent with the earlier finding that the dollar prices of U.S. exports of manufactured goods do not increase as a result of a devaluation of the dollar. This implies that the foreign currency prices of U.S. exports fall by the full amount of the devaluation since there is no offset in the form of higher dollar prices.

E. Agricultural Exports

The following equation was used to estimate the price responsive~

ness of agricultural exports:

-37-

a. a FX wy L UVFX," 2 (24) TEX * a TFY Gr. “V where: FX = value of agricultural exports, seasonally adjusted annual

rates, and adjusted for dock strikes and other disturbances; from Parrish and Dilullo [1972].

UVFX = unit value index for agricultural exports, seasonally adjusted; constructed by taking a weighted average of the unit values for crude foods and manufactured foods, with the average shares of these two components over 1965-1966 used as weights.

TFY = trend value of the weighted average of seasonally adjusted real GNP of Canada, Germany, Japan, and the U.K., with weights equal to shares in U.S. agricultural exports in 1965-1966; trend value equal to fitted value from an exponential] time trend.

a

FPF = weighted average of seven countries’ (Canada, France, Germany, Italy, Japan, Netherlands and U.K.) consumer price indices for food, seasonally adjusted, with weights equal to shares in U.S. agricultural exports in 1965-1966, and converted into dollars.

Vv = disturbance term.

This equation was estimated over 1959III -— 1972IV using the current

values of both explanatory variables. Experiments with a polynomial

t.

distributed lag on the relative price term were unsuccessful in obtaining a negative coefficient. Since the initial estimates indicated severe autocorrelation, the equation was estimated using a Hildreth-Lu trans-

formation. The estimated autocorrelation coefficient in the results

given below was 0.700.

UVFX 52

Constant TFY Grr R SE | 0.58 0.25 -0. 38 0.89 0.048 (0.82) (1.68) (-1.32)

Both coefficients are only marginally significant and the results must

therefore be considered highly tentative.

eee: me eee ie Fee ne RR ABO Se SS pea: ~ ame eenane eee eae ee a ee Oe ene aes ~~ women. ee ” a Ant ee Sor -- ee rromer. meme me on one eD eee eer WER mere a a eee ne RE NT

~38-

III. Exchange Rate Effects on the U.S. Trade Balance: 197111 - 197311

Table 6 presents the calculated effects of exchange rate adjustments between 197111 and 1973II on the current and constant dollar trade flows for the five categories of U.S. imports and exports considered in this paper. The figures in this table are based on the computations in the previous two sections.

Looking first at imports of finished manufactures, the impact of the depreciation of the dollar on this category of imports was computed in two steps. First, the effect of the dollar depreciation on the unit value index for imports of finished manufactures was computed using the estimated coefficients from regression 8 in Tables 3 and 4. In this computation the exchange rate variable, Res was held constant at its 19711 value. This estimate of what the unit value index for finished manufactures would have been in the absence of a devaluation was then combined with the distributed lag coefficients that were estimated for the relative price term in equation (18). This calculation, reported in column 1 of Table 6, provides an estimate of the decline in real imports (expressed in 1963 dollars) that can be attributed to the depreciation ‘of the dollar between 1971II - 1973II on the assumption that domestic

- prices and incomes are fixed. The fairly modest figure of only a $920 million reduction in imports by 1973II reflects both the very long 20-quarter lag for real imports to adjust and the 5-quarter lag for the full pass-through to occur, as well as the fact that the estimated pass-through is only 50 percent. The long lag in the adjustment of real

imports to relative price changes also causes imports of finished

TT ase rman coe” wwe eee ER RE peg RT On Taner eet ORES A GEG Tee On RMR IRE mero eer ee omen ees eam manne sarc >

—39--

Oe 6°€-

0°L-

L°e-

(€1) § quezIND

eT stseq °‘d JO “Gd <$ uezing Jo suof{Itd) souBeyeg sperzayl [BIOL

7€°OT

Zl6°‘T Ste‘e B9T‘T

EZI'T LLESZ 758

G6E L6S‘T 609 Zy- IZ1*t Lzs c6I- col 9S¥

LSI- 864 Z9€

cen Lez 002 €o~ 89 19 8- 9 9

(Zt) (11) (oT)

07°8

Ble 9S¢ 9ST 86T UC VAT 9ST 0S 9 (6)

$ queTIND $E961 $°42ND $E96T

Sjaodwy Uy ssueyD $q100dxy

-sjziodxq uy o3ueyD Leanq[Noy23y

:$J0a9FJ_T peutquog

(saje1 Tenuue ps

CZ°VE 21°97

OLLST O8z‘T

O€z‘T 026 oss 0479 OLS O44 Ove 092 OLT O€T 09 ~3=s oS: Ol Ol ) 0 (g) (2)

$°and S€96T $}100xy Teany (squeof.

-[NOFAZY-UON -qnyT pue sTEN;

TE°ST 8T°ET

9949 - 694 9L4 607 78C 94T

cv

0

(9)

$*anp $€96T

G2S~ 89e~ 1 A LeI- 8Scc~ c- I- 0 (¢)

ssaT) settddng 9 slTet1937eW letaqsnpul jo sjz10dw]

ec*

8? 99 96 OOT cL 94 rat v7 0 (7)

L It’s

Z1z- 17a Lz1z8- 7 1 ¢z- 0 (€)

$°and $€96T

SOoeABAOgY

spooj ‘spood

jo sj10du]

qasnfpe ATTeuosees ‘sieT[Op Fo SUOT TT TW) Zueyoxy quesey JO 6399FFA PouTquioo

sjiodxq pug qiodwy *§*n uo saSueug o7eY 2

9 1981

(SUOTTTT4) TI€Z-1iITZ :“oTd ZS°61 LO°YT opery asersay

cSy =: OZ 6~ IT€L6T €€v 299- IC L6T 06% Tv7- AIZL6T 0c9 Hl7- =ILIcL6T

we9—s«sIGT= =: LIZLOT

L467 =«TL- IZL61 8%Z = Ez AITL61 18 S- _«ITITL6T 41 0 IIIL61T (7) (1) $°4n> $€96T SIN OBFNUSHL poystura

jo sqiz0dwy

A TERR oF nC I IS RE TTT RIEDEL ST PRIN sm, — . ~ - . . ta vert ab akan Perea N te eo PT SF ies ot TE RC ARE

Sa a co a

-40-

manufactures in current dollars to increase as a result of the dollar depreciation, as seen in column 2. This initial perverse effect simply reflects the fact that the increase in the dollar price of imported finished manufactures was not offset by a reduction in the quantity imported during the period covered by Table 6.

In computing the effects of the exchange rate adjustments on the other two categories of imports, use was made of the results given in Table 1. The figures in columns 3-6 of Table 6 were computed on the basis of the price effects that were generated using the mid-poipt estimates of the supply and demand elasticities, ives, column 2, in Table 1.27 Because demand is inelastic in the short run for both foods, feeds, and beverages and industrial materials and supplies, the reduction in the quantities imported is outweighted by the rise in the dollar prices of these imports, so that their dollar value has risen as a result of the depreciation of the dollar. Since the long-run elasticities for these imports exceed unity in absolute value, the depreciation (in this partial equilibrium context) will ultimately reduce the value of the goods imported.

A similar calculation was performed to obtain estimates of the devaluation effects on U.S. exports of agricultural goods. The price effects reported in column 2 of Table 1. were combined with the estimated price elasticity in equation (24) to yield the results reported in columns 9 and 10 in Table 6. As can be seen froma comparison of these two columns, the fairly large improvement in this sector of the trade balance is due mainly to the increase in the dollar prices of these

exports, part of which was brought about by the depreciation.

~4l~

The increase in non-agricultural exports was comput ed by setting the exchange rate in equation (22) equal to its value in 19711 and then using the estimated lag coefficients of a, to calculate the increase in this category of imports. Since the evidence in Table 2 suggests that export prices of manufactured goods did not rise as a result of the depreciation of the dollar, the increase in the value of non-agricultural exports reported in column 8 (over and above that due to the increase in quantity) is due solely to the rise in prices caused by factors other than exchange rate changes.

The figures at the bottom of Table 6 show the average value of the trade flows (in current and constant dollars) for the five categories of imports and exports between 1971II and 1973II. These figures are provided in order to facilitate comparison between the exchange rate effect in each category and the absolute magnitude of trade in that category. They show that in real terms the greatest percentage improvement occurred in imports of finished manufactures, whereas in current dollars the largest percentage gain was in agricultural exports.

The combined effects for the five categories of imports and exports are reported in columns 11 and 12, The improvement in the "real" trade balance was not sufficient to overcome the perverse effect of the higher dollar import prices, so that the time path of the response in the trade balance to a devaluation looks something like a "J" curve. However, the dip in the early part of the "J" is very modest indeed, reaching a maximum of only -$200 million in 1972II. This small deter-

joration reflects both the slow rate at which dollar prices of imports

RMI MRT Sone a RN TE am

fen ee RO REI FR RR RET CR EERO NRNRE RE oo RENE I COPE UMN ermine STAT

-42-

have been estimated to increase in response to a devaluation as well as the fact that the estimated long-run passthrough for imports of finished manufactures is only about 50%. Thus it appears unlikely that the U.S. trade account suffered dramatically as a result of transitory negative effect of the 1971-1973 exchange rate realignments. And after this perverse phase had run its course, there is an improvement that reaches $2 billion (seasonally adjusted annual rate) in the trade balance in the second quarter of 1973, This figure represents the cumulative impact

of the exchange rate adjustments between 1971II and 197311 on the trade balance in 1973II.

The estimate for the "J" curve in the U.S. trade balance in Table 6 is quite different from the "J" curve for the U.K. trade balance obtained by simulation of the London Business School Model of the United Kingdom as described in Ball et al. [1972]. Using this model, the trade balance effect of the November 1967 devaluation is estimated as:

1967 1968 | 1969 Change in U.K. QIV Ql QII QIII QiV Ql QII QIII QV trade balance . (million current &, -128 -204 -228 -128 40 128 £172 196 208

seasonally adjusted ‘annual rate)

The initial deterioration in the trade balance in the first two quarters of 1968 was about equal to the eventual improvement that occurred by

the fourth quarter of 1969. The primary reason for the accentuated U.K. "3" and the shallow U.S. "J" is. that the prices of imports

appear to have adjusted much faster to exchange rate adjustments in the

United Kingdom.

-43-

In both absolute and relative terns, however, the trade balance response in the United Kingdom was quite small. This is evident when the above figures are compared with 86,400 million in merchandise exports and £7,100 million in merchandise imports in 1968. There was a large improvement in exports (61,044 million, S.A.A.R., in 1969Q4), but this was offset to a great extent by a significant rise in the value of imports (4836 million, S.A.A.R., in 1969Q4) because the demand for imports is very price inelastic. Nevertheless, the 1967 devaluation did provide a substantial boost to the current account. As described by Ball et al. {1972} and the National Institute of Economic and Social Research [1972], the service account benefited. by an amount at least as great as the trade account.

The results in Table 6 cannot be considered definitive for the entire trade balance, shown for the sake of comparison in column 13, because they do not include all exports and imports. Left out are shipments of military goods, trade in automobiles and parts with Canada, exports of aircraft, imports of fuels and lubricants, and imports and exports not elsewhere specified. The exchange rate impact on the value of imports ‘of petroleum products would somewhat reduce the overall trade-balance ‘ effect of the devaluation below the figures reported in Table 6, but it is very difficult to say how much. Nevertheless, the results reported here suggest that a portion of the improvement in the overall trade

balance between 1972 and the first half of 1973 can be attributed to

the depreciation of the dollar.

-44-

It needs to be emphasized that these estimates pertain to only the direct or first-round effects of the dollar devaluation. They therefore provide an estimate of the maximum beneficial impact on the U.S. trade balance resulting from the multilateral exchange rate adjustments that took place between 1971 and 1973. The size of this estimated improvement in the trade balance depends on the ultimate magnitude and timing of the response in the quantities of imports and exports to price changes. These price changes in turn depend crucially on the speed with which import and export prices react to exchange rate changes. Thus one lesson that can be drawn from the results and methodology in this paper is that for a country as important in world trade as the United States, an understanding of how adjustments in exchange rates affect U.S. import and export prices is just as essential as knowing how the quantities of these goods respond to price changes. Clearly much more research is needed in this area.

An appraisal of the full effects of a devaluation on the U.S. trade balance of course requires that the induced or second-round domestic effects be taken into account. These include increases in expenditures on domestic output and in prices and wages. The degree to which these second-round effects offset the positive impact effect of a devaluation depends on the level of unemployment and capacity utilization, and as emphasized by Alexander [1952], on the extent to which domestic output rises above domestic absorption. Thus for a full analysis of the

impact of a devaluation one must look at the savings behavior of an

- economy.

~45-

The results in this paper therefore cannot provide an answer to the full long-run effects on the U.S. trade balance resulting from the exchange rate realignments that took place between 1971 and 1973. They May nevertheless provide some guidance as to the plausible magnitude of the short-run impact before the second-round effects are complete. In addition, to the extent that the general~equilibrium savings and expenditure decisions depend on price level developments, the paper may be helpful in that it provides a framework for analyzing the linkage between exchange rate changes and import and export prices. It is, of course, not a substitute for a full general equilibrium analysis in which domestic

money supply considerations are explicitly taken into account. This

requires a fully-specified macroeconomic model.

Re Or im . * eee spe cam ne om nm HEE ARE POE EN UP ee ROR EEE RR EER MOE RI I UE SAR PERNA SET CTT SIRS ae ORGS ERIN AE TEC TRO I A ESS YT

Footnotes

“The author would like to thank Peter Isard, Barbara Lowry and Peter Hooper for helpful discussions relating to the paper. He would also like to thank Henry Goldstein, Sam Katz, Sung Kwack, P.A.V.B. Swany, Ted Truman and Christopher Yandle for comments on a previous draft. The author also benefited from having presented earlier drafts of the paper at the University of Wisconsin in Madison and Milwaukee and at the University of Rochester. A version of this paper (Clark [1973}) was presented at the Annual Meeting of the Econometric Society, December 28-30, 1973 in New York City. Finally, the author thanks Gary Schlossberg for the considerable research assistance required to produce this paper.

lohis measure of the depreciation of the dollar is derived from the figures given on pp. 15-16 of World Financial Markets (October 23, 1973), Morgan Guaranty Trust Co., which in turn reflect bilateral export and import weights for fourteen countries. In this paper all referen¢ges to a depreciation of the dollar refer to weighted average or effective exchange rate changes. .

2 rhe analysis in this paper is thus subject to the usual strictures against the elasticities approach to devaluation analysis leveled by those who adopt a general equilibrium perspective. For this latter, generally monetarist approach, see Dornbusch [1973] and Johnson [1972].

3The domestic price increase induced by the devaluation would in any case be attenuated by the price freeze and other restraints on price rises that were part of the New Economic Policy initiated by Nixon in August, 1971. It should be noted that these restraints did not apply to either import or export prices, and consequently the price freeze and the other phases of wage and price controls would tend to slow the rate at which domestic prices would respond to the higher import and export prices. For attempts to estimate the impact on the U.S. rate of inflation of the dollar devaluation, see the article by Kwack in this. volume, "The Effects of Foreign Inflation on Domestic Prices and the

Relative Price Advantage of Exchange Rate Changes," and Clark [1974].

4ohis qualification is the same as that which applies to a priori

‘ estimates of the effect of the 1971 realignment on the U.S. trade

balance, as contained in Branson [1972], for example.

Sror an attempt to estimate supply elasticities for U.S. exports and imports, see Magee [1970].

Srhis methodology is also used by Isard in his paper in this volume.

"This may appear to be a very restrictive assumption, but it is

_the same assumption that is used to derive aggregate import and export

equations. For a discussion of aggregation problems of this type, see Leamer and Stern [1970].

~46~

-47~

8 since this measure of world trade in each commodity class is less than the true world total, the calculations using (10) may overestimate the rise in the dollar prices of these commodities because M in the denominator of (10) is too small. Such an overestimate can only occur, however, if there has been no net appreciation of the non-OECD countries vis-a-vis the United States.

"The author is indebted to Christopher Yandle, Commodities Division of the Research Department in the International Monetary Fund, for providing him with a range of export supply and import demand elasticities that would be appropriate for the kinds of agricultural commodities that a) the United States imports and b) that it exports. ‘ne range of elasticities in case a) was used for foods, feeds and beverages, and the range of elasticities for case b) was employed for those commodities that approximate U.S. agricultural exports, namely foods, feeds and beverages plus agricultural industrial supplies. A considerably wider range of elasticities was used for industrial supplies and materials. The results given in Table 1 show, however, that the calculated price changes are not very sensitive to large variations in the elasticities used to make the calculations. r

0p or particular commodities, however, the price effect of multilateral exchange rate changes may be significantly higher than indicated in Table 1. Bernstein [1973, pp. 5-6] for example, weights the exchange rate changes of the Group of Ten plus Switzerland by each country's consumption of four nonferrous metals (copper, lead, zinc and tin) and finds a weighted average price effect of about 11.5 percent from the end of December 1972 to the end of July 1973. This high figure stems from the fact that the sample includes countries that both appreciated the most against the dollar and consume most of the metals, For the impact of recent exchange rate changes on the dollar price of wheat, see the contribution by Yandle in this volume.

I sce Clark [1973]. Jacques Artus has developed a model of export price determination (see his contribution to this volume) that is also based on profit-maximizing assumptions.

12 or a discussion of this markup hypothesis, see Eckstein and

’ Fromm [1968].

13enis link is on the demand side. There is an additional relationship that comes through the impact of exchange rate changes on, domesti cost conditions. Thus a devaluation will tend to increase ULC and UMC in equation (11) in the text, thereby increasing export prices. This link between exchange rate changes and export prices through repercussions on domestic costs is ignored in this paper. For an examination of this link, see the paper by Kwack in this volume entitled, "The Effects of Foreign Inflation on Domestic Prices and the Relative Price Advantage of Exchange Rate Changes." There is in addition a third link that arises if exports are denominated in foreign currency. In this case a devaluation would immediately raise the domestic currency prices of

EAN NN FRO RE ERTIES INTRA ROT ERATOR OER EE ST SERRE ESTEE, 1 ORIOL SC alae Rel in als eae

-48~

those exports invoiced in foreign currency. This link would not appear to be of much importance for the United States, since at least up until the recent exchange rate changes most U.S. exports were probably denominated in dollars.

laonis result is derived in Clark [1973] and Nordhaus [1972].

1545 a rough correction for the fact that export price indexes record contract prices, whereas export unit values are a proxy for prices recorded at the time of shipment, export price indexes were lagged one quarter when combined with export unit values to construct this index. These export price or unit value indexes were first converted into dollars by using an exchange rate index based on quarterly averages of spot exchange rates.

16 . :

R was constructed by taking a weighted average of exchange rate indexes for the seven countries. The weights used to construct R and

PX, are the same as those used for PC ee ‘

the fact that the sum of the lag coefficients for R in regression

4 is negative and significant at the .05 level would appear to be a statistical fluke, since there is no theoretical reason to expect a negative relationship between R and the export price. The paper by Artus in this volume and the OECD [1973] find that U.S. export prices

of manufactured goods are only weakly related to the export prices of foreiznm competitors and thus they tend to support the conclusion that U.S, manufactured export prices are not substantially affected by demand shifts induced by exchange rate changes. This conclusion depends on the use of unit value indexes that admittedly may be poor proxies for the actual price behavior of U.S. exports. (See Lipsey [1963] for a discussion of the deficiencies in unit value indexes.) The close association between movements in the wholesale prices and in the export unit value indexes for manufactured goods described in Ripley [1974] may indicate, however, that unit values are not altogether inappropriate proxies for export prices.

Ban exchange rate change has an immediate impact on the dollar import price if the imported commodity is invoiced in foreign currency. Since most U.S. imports have been invoiced in dollars, at least up until the recent exchange rate adjustments, this direct impact of a dollar depreciation on U.S. import prices has probably been small. On this point see Magee [1973] and Grassman [1973]. There is, unfortunately, an additional difficulty raised by U.S. Customs valuation practices. Imports invoiced in dollars are converted into foreign currencies at the rate prevailing on the date of the purchase contract and then reconverted into dollars at the date which prevailed on the date of shipment. Thus it would appear that imports nominally invoiced in dollars are effectively invoiced in foreign ‘‘rrency. ‘Apparently, however, less than ten percent of import documeuts z°2 being changed by the Customs Bureau for valuation purposes. The zuthor is indebted to Daniel Roxon for this unofficial estimate, He is also indebted to Stephen Magee for

ee ORC TENE or Mm ag thet ba UTE SE eR go pene SI OCR. Tae I

~49-

pointing out the implications of U.S. Customs valuation practices for the impact of exchange rate changes on U.S. import unit values. Since this paper was written, an article by Magee [1974] has appeared in which he presents some evidence on the currency denomination of U.S. imports and describes in detail the valuation procedures of the U.S. Customs.

19 molesale prices of manufactured goods have also been used in a recent analysis by the OECD of the determinants of export prices of manufactured goods. See OECD [1973].

20nnis estimate is lower than the figure of eighty percent calculated by Branson [1972, pp. 20-22] that is based on the supply elasticities estimated by Magee [1970]. Branson's result relies on a perfectly competitive model of price determination, and as mentioned above, this model is not appropriate for trade in finished manufactures. For this reason, and because there is considerable uncertainty regarding estimates of supply elasticities for U.S. imports, the estimates given in Table 3 would appear to be more reliable than Branson's calculations.

Using regression analysis similar to that in this paper, Llewellyn

[1974] found that a devaluation of the pound would raise the prices of imports entering the United Kingdom by about 60% of the devaluation. This result is consistent with the figure of 50% for the United States reported in this paper, and does not support Branson's conclusion that the figure should be 80%. The reason is that since the United Kingdom has a smaller share in world trade than the United States, one would expect that the exchange rate impact on its import prices would be larger, not smaller, than for the United States.

aloe income, for example, is above its trend value, then this would tend to be associated with longer waiting times for the delivery of domestic goods and other manifestations of supply constraints, leading to increased purchases from abroad.

22, quation (17b) has been used in the following discussion because the WPI. variable gave a better statistical fit than did the combination of the W. and time trend variables in (17a). It should be pointed out that one implication of the methodology underlying equation (19), and especially the fact that the export price is a function of the exchange rate (as shown in equation (l6a) and (16b)), is that foreign wholesale prices, expressed in dollar terms, are not a good deflator of nominal imports. The reason is that exchange rates changes break up the normal or usual link between foreign wholesale prices and foreign export prices.

For an example of deflation using foreign wholesale prices, see Parrish and Dilullo [1972]. , 23 quation (19) also shows that the impact of the WPI variable on imports, a,8,, is not equal to the exchange rate effect, a (1+8.). Only if there is perfect arbitrage such that PX, = WPI. will ?t be’ true that changes in foreign wholesale prices and in exchange rates will _ have the same effect on imports.

sash essen, a rene ine, ‘ . meer 7 . a ARPES RERS Sete ere . alee aa ieee a Re as eee ee RE RET SRN ON AE YL EAE ELH LE 10 SIRE I T— /TO RROORE NS RR MT, ARN Re IMEI IM,

~50-

24... : . . The PX. variable in this equation was constructed as a weighted

average of thé overall export unit value indexes for eight countrics (the previous seven plus Italy), with shares in exports of manufactures used as weights.

25 yen equation (22) was estimated using a weighted average of both export prices and unit values for manufactured exports, the results were very similar to those reported in the text.

Cone difficulty arises in estimation because the same export unit values are used in both cases, although the country weights are different in each case. Clearly what one would like to have are separate price indexes for those goods exported to the United States and for those exports that compete in world markets with U.S. exports, but such data unfortunately are not available.

27 . . P ._ The price increases reported in Table 1 were lagged one period in calculating the effects of the depreciation of the dollar. This is designed as a rough adjustment for the fact that because of delivery delays and long-term contracts, imports and exports in the current quarter reflect the spot prices of previous quarters.

a te RL EMR eT aera etree

ERIE LRT OI An SE NR LTR ee OB ARES OO

References

Alexander, Sidney S., "Effects of a Devaluation on a Trade Balance," International Monetary Fund Staff Papers, Vol. 2 (April 1952).

Artus, Jacques R., "The Behavior of Manufactured Export Prices," this volume.

Ball, R.J., T. Burns and G. Miller, "Preliminary Simulations of the London Business School Model," paper prepared for the Conference

on Modelling of the U.K. Feonomy at the London Business School, July 1972.

Bernstein, E.M., "The Inflation Problem in the United States," in EMB (Ltd.) Research Economists, Report No. 73/17, August 17, 1973.

Branson, William H., "The Trade Effects of the 1971 Currency Realignments," Brookings Papers on Economic Activity No. 1, 1972, The Brookings Institution, Washington, D.C. s

Clark, Peter B., "Some Early Estimates of the Price and Quantity Effects of the Smithsonian Agreement on the U.S. Trade Account," unpublished manuscript, 1973.

, "An Aggregative Approach to the Domestic Price Effects of Recent U.S. Exchange Rate Changes," paper presented at the Western Economic Association Annual Conference, Las Vegas, Nevada, June 10, 1974.

Coombs, Charles A., "Treasury and Federal Reserve Foreign Exchange Operations," Federal Reserve Bank of New York Monthly Review Vol. 54, Nos. 3 and 9 (March and September, 1972), and Vol. 55, Nos. 3 and 9 (March and September, 1973).

Dornbusch, Rudiger, "Devaluation, Money, and Nontraded Goods," American Economic Review, Vol. 63, No. 5 (December, 1973). me

’ Dunn, Robert M., "Flexible Exchange Rates and Oligopoly Pricing: A Study of Canadian Markets," J. of Political Economy, Vol. 78, No. 1 (January and February, 1970).

Eckstein, Otto and Gary Fromm, "The Price Equation," American Economic Review, Vol. 58, No. 5, Part 1 (December, 1968).

Grassman, Sven, "A Fundamental Symmetry in International Payment Patterns," J. of International Economics, Vol. 3, No. 2 (May, 1973).

Hutton, John and Patrick Minford, "A Model of U.K. Manufactured Exports and Export Prices," unpublished manuscript, U.K. Treasury, 1972.

~51-

OTD TPR rene er ree ye TRIBE Ne RENE NRE REED woo nw

~52—

Isard, Peter, "The Price Effects of Exchange Rate Changes," this vclume.

Johnson, Harry G., "The Monetary Approach to the Balance of Payments," J. of Financial and Quantitative Studies, Vol. 7 (March, 1972).

Junz, Helen and Rudolf Rhomberg, "Price Competitiveness in Export Trade

Among Industrial Countries," American Economic Review Papers and Proceedings, Vol. 63, No. 2 (May, 1973).

Kwack, Sung Y.,; "The Effects of Foreign Inflation on Domestic Prices and the Relative Price Advantage of Exchange Rate Changes," this volume.

Leamer, Edward E., and Robert M. Stern, Quantitative International Economics (Boston, Mass.: Allyn and Bacon, 1970).

Lipsey, Robert E., Price and Quantity Trends in the Foreign Trade of the United States (Princeton, N.J.: Princeton University Press

for the National Bureau of Economic Research, 1963). .4

Llewellyn, G.E.J., "The Determinants of United Kingdom Import Prices," Economic Journal, Vol. 48, No. 333 (March, 1974).

Magee, Stephen P., “A Theoretical and Empirical Examination of Supply and Demand Relationships in U.S. International Trade," unpublished manuscript for the Council of Economic Advisers, October, 1970.

» "Currency Contracts, Pass-Through, and Devaluation," Brookings Papers on Economic Activity, The Brookings Institution, Vol. l, 1973.

, "U.S. Import Prices in the Currency-Contract Period," Brookings Papers on Economic Activity, The Brookings Institution, Vol. l, 1974.

SN

Marston, Richard, "Income Effects and Delivery Lags in British Import Demand: 1955-1967," J. of International Economics, Vol. 1, No. 4 (November, 1971).

National Institute of Economic and Social Research, "The Effects of the Devaluation of 1967 on the Current Balance of Payments," Economic Journal, Vol. 82 (March, 1972), Supplement. Nordhaus, William D., "Recent Development in Price Dynamics," in The Econometrics of Price Determination, Board of Governors of the Federal Reserve System (June, 1972).

OECD, "The International Transmission of Inflation," OECD Economic Outlook, No. 13 (July, 1973).

1 ee Ne wee ae oo neg. ae . a, DEO HS SEEN OE TIRE AECL SY REELS: RENT TOE ASNT IN CTR FT I TOI AE TIE Ter SRI PETRIE TEP oF SLES ONE, TEN, ONE UT I I RE EI LOE OIE OT

~53=

Parrish, Evelyn and Anthony Dilullo, "U.S. Merchandise Trade Projecticns," Survey of Current Business (May, 1972).

Ridler, Duncan and Christopher A. Yandle, "A Simplified Method for Analyzing the Effects of Exchange Rate Changes cn Exports of

a Primary Commodity," International Monetary Fund Staff Papers, Vol. 14, No. 2 (November, 1972).

Ripley, Duncan, "The Movement of Export and Domestic Prices in the United

States, Japan and Germany," International Monetary Fund Departmental Memorandum (DM/74/66), July 10, 1974.

SRE Neem nae came mene et “ aL pang Sen Me A oe

en RES FR peep ee ery men se ate inet i were ae ait e ainda th. on aa echans’ ctzeaiadltescaiaoaciceinemenean aaa atten timate a tcamieh wos cntaie aah aanataniaal

Data Appendix

Imperts:

Data on the value of imports by end-use breakdown were obtained from the Survey of Current Business. In this paper imports of finished manufactures are defined to include four end-use categories: consumer manufactured durables, consumer manufactured nondurables, capital goods except automotive, and motor vehicles, parts and engines other than from Canada,

Data for the unit value series for foods, feeds and beverages and industrial materials and supplies were obtained from the Bureau of International Commerce (Dept. of Commerce) unit value computer runs. For finished manufactures the unit value index for the economic class called "finished manufactures" was used, For 1958-71 this index was taken from Index Numbers of U.S. Exports and Imports 1919-1971 (U.S. Bureau of the Census, U.S. Government Printing Office, Washington, D.C., 1972). For 1972-197311 the index was taken from various issues of Export-Import Indices, an unpublished memorandum put out by the Foreign Trade Division of the Bureau of the Census.

Exports: : .

Data on the value of agricultural and non-agricultural exports, ex= clusive of automotive exports to Canada, aircraft and military equipment, were adjusted for dock strikes and other disturbances. These

two series were obtained from the Balance of Payments Division of the Bureau of Economic Analysis, Department of Commerce. The series for non-agricultural exports is described in Parrish and Dilullo [1972].

The unit value series for these two categories of exports were obtained from the from the same source which provided the unit value series for imports of finished manufactures. :

Import _end Export Overall Unit Values: International Monetary Fund, International Financial Statistics, country pages. .

Price Index for Canadian Exports of Manufactured Goods (exclusive of

automotive products): Bank of Canada.

Wholesale Prices of Manufactured Goods:

._ For Belgiun, Canada, France, Germany, Italy, the Netherlands, and the United Kingdom, OECD, Main Economic Indicators: Historical Statistics, 1955-1971, and Main Economic Indicatozs, March 1973. The series for the United Kingdom was updated through the second quarter of 1973 using Economic Trends, Central Statistical Office. For all countries except Canada, it was necessary to make an adjustment for the introduction of the value-added tax. The author wishes to thank Donald Curtis (U.S. Treasury Department) for making these adjustment factors available to him. The source for Japan is Basic Data for Economic Analysis, Bank of Japan.

Export Price and Unit Values Indexes for Manufactured Goods: These series were obtained from national sources.

-54~

vege ne LIMA TI Crary epee te Tt

=55=—

Exchange Rates: For most countries, Federal Reserve Bulletin; otherwise International

Financial Statistics.

Hourly Earning or Wage Rates_in Manufacturing: For non-U.S. countries, OECD, Main Economic Indicators: Historical

Statistics 1955-1971 and Main Economic Indicators, various issues. For the United States, compensation per man-hour was defined as wages and salaries of employees plus employers' contributions for social insurance and private benefit plans. Both compensation per man-hour and output per man-hour were obtained from the Bureau of Labor Statistics, U.S. Department of Labor.

Aggregate Capacity Utilization Rates: Wharton, EFA, Inc.

Index of Foreign Industrial Production:

Weighted average of industrial production indices for Canada, Japan, United Kingdom, and continental Western Europe, weighted by the annual shares of these areas in U.S. experts. The index for continental Westcrn Europe is a’ weighted average of the indices for Germany, France, Italy and the Netherlands, with weights equal to the 1963 values ar their gross domestic products. This series was obtained from the same source as the export value data. :

U.S. Industrial Production Index: Federal Reserve Bulletin.

Wholesale Prices of Total Manufactures and Industrial Commodities: Survey of Current Business.

GNP, GNP Deflator, Disposable Personal Income and Deflator for Personal

Consumption Expenditures: Office of Business Economics.

Deflator for Personal Consumption Expenditures on Foods, Feeds and

Beverages: Department of Commerce, Mr. Bassett.

Foreign Consumer Price Indices for Food: OECD, in Economic Indicators, Historical Statistics, 1955-1971

Nea ne a Naha an a

and Main Economic Indicators, September 1973.

Foreign Real GNP:

Canada, National Income and Expenditure Accounts; Japan, Annual Report of the Economic Planning Agency (value in 197311 estimated on assumption of 10% growth rate in GNP); West Germany, Viertel jahrsheft (values in 19731 and 197311 estimated on assumption of a 6.2% growth rate in GNP); United Kingdom, Economic Trends (value for 1973II estimated on assumption of a 6% growth rate in real GNP); weights derived fron 1970 value of, agricultural exports, which were given in Overseas Business Reports, Dec., 1970, Department of Commerce. The author is

“indebted to Sung Kwack for making this data available to him.

Wholesale Price Index Excluding Crude Foodstuffs, Feedstuffs, Etc.; Monthly Labor Review.

ee ee rte

ap ern IES REE ERIE NORE ne Eni erm eo

Tee ma capes

~56-

Tariffs:

Calculated using fixed 1964 import weights for all items for which global imports exceeded $1 million, and aggregated into end-use cate~ gories. Tariff levels for the 1958-1967 period were assumed to be constant. Starting in 1968 and ending in 1972, the tariff levels computed for 1964 were cut by 20% of the Kennedy Round concessions. The tariff rate on imports of finished manufactures was computed as a weighted average of the tariffs on capital goods, automotive goods and consumer goods, with 1964 import weights. The author is indebted to John Wilson for supplying him with this tariff data.

yee ae ae

FN TATE NN Oe ORRIN I