Supply Pressure and the Export-Import Performance in the Japan-U.S. Bilateral Trade
  • 1, International Monetary Fund

The paper examines the effects of the supply pressure of the exports in the Japan-U.S. bilateral trade. A simultaneous equation approach with a Almon lag structure is adopted. Two factors of supply pressure, i.e., full-employment capacity and inventory are specified, and positively-sloped export supply functions with these two shift factors are successfully estimated. While capacity is positively correlated with exports, the inventory is often negatively correlated. It is also shown that export supply pressure is much stronger in Japan’s exports than in the U.S. exports, and that supply pressure often affects exports with a lag structure spreading over twelve quarters.


The paper examines the effects of the supply pressure of the exports in the Japan-U.S. bilateral trade. A simultaneous equation approach with a Almon lag structure is adopted. Two factors of supply pressure, i.e., full-employment capacity and inventory are specified, and positively-sloped export supply functions with these two shift factors are successfully estimated. While capacity is positively correlated with exports, the inventory is often negatively correlated. It is also shown that export supply pressure is much stronger in Japan’s exports than in the U.S. exports, and that supply pressure often affects exports with a lag structure spreading over twelve quarters.

I. Introduction

In the 1970 and early 80s, Japan’s current account surplus continued to expand in spite of temporary setbacks caused by the two oil crises and repeated large appreciations of yen. It decreased substantially again during the second half of the 1980s on account of an abnormal expansion of domestic demand induced by an expansionary monetary policy and subsequent speculative price increases of land and stock. However, it expanded sharply again and reached a new record level of about 118 billion dollars in 1992. On the other hand, the current account of the United States whose surplus declined in the 1960s and 70s has started to show large deficits in the 80s and 90s. The most important factor that accounts for this difference is the private and public investment-saving balances; Japan’s absorption fell short of her GNP and U.S. absorption exceeded her GNP by a margin of 3-4 percent in the mid 1980s. However, the relationship that links the I-S balance and the current account needs more investigation. In general, the relation that investment minus saving (minus government deficits) equals the current account does not necessarily imply that the movements of I-S balance explains those of the current account, but only imply an equilibrium condition under which income and prices have to be such that the above relationship holds.

Whenever the domestic private saving exceeds the sum of domestic private investment and government deficits, there arises downward pressure on domestic prices through increased competition among suppliers. This pressure may cause changes not only in the general price level but also in relative prices including those between domestic and export prices. Similarly, there is pressure to expand exports through increased non-price competitiveness in the domestic and export markets. In other words, whenever total effective demand falls short of the full-employment supply capacity, there will be various kinds of supply pressures on quantities to be sold in the export and domestic markets even if the relevant prices remain unchanged. These adjustments of the supply side affect the level of equilibrium income, although the demand side also exerts influences on income in a well-known fashion. One of the important implications of such adjustment mechanism is that exchange rate appreciation is likely to have a very limited effect on the current account surplus of an economy in which supply pressure is very strong. 1/

Supply pressure consists of two aspects. One aspect of supply pressure is price performance. Exporting firms tend to keep domestic or foreign export prices as low as possible in spite of the upward pressure to push those prices up at the times of such cost-push inflation as oil price increases, or large appreciations of the exchange rate 1/. Cost-reducing rationalization and low price policy belong to this aspect. The second aspect is the efforts to promote exports without changing export prices per se. Export promotion through product differentiation and intensified marketing may be of this type. In the actual estimation of supply schedules, however, it is difficult to distinguish the two aspects, because both aspects are reduced to estimating the factors that make the supply schedule shift downward.

In order to measure the magnitude of the supply pressure in terms of the shift of supply curves at given relative prices, we consider two factors explicitly, the level of inventory and the full-employment productive capacity of exporting firms. There are some studies that explore a certain negative correlation between exports and domestic demand pressure in exporting countries. Eaton and Steuer (1966), Bridge (1966), Artus (1970), Winters (1974) estimate “export functions” along this line. Mintz (1967) discusses the “export supply pressure” hypothesis without engaging in econometric analysis. Econometric analyses that treat supply pressure or “capacity pressure hypothesis” are Branson (1968), Basevi (1973), Goldstein and Khan (1978), Dunlevy (1979), and Onitsuka (1984). Economic Planning Agency (EPA) of Japan sometimes includes level of inventory as a factor accounting for supply pressure in its estimation of expect functions in its annual reports (Keizai-hakusho). Asako and others (1992) recently examined the supply pressure of both productive capacity and inventory.

Most of these contributions, however, do not distinguish export supply schedules from import demand schedules, but estimate a single equation that is in general a demand function with negative responses to price changes. Other works including Citrin (1985) that distinguish the two schedules do not address themselves explicitly to the supply pressure hypothesis. The only exceptions are Goldstein and Khan (G-K) and Dunlevy.

While the G-K is successful in estimating meaningful supply functions of exports of several industrial countries, it is not in the estimation of the price effect of Japan’s supply schedule of exports. Dunlevy confines his analysis to those of the U.K. and United States. Both contributions do not take into account the inventory of finished goods as a supply pressure factor. Moreover, Dunlevy’s result is ambiguous; the level of capacity utilization is not negatively correlated with exports, although an increase in capacity utilization is.

The purpose of this paper is, first, to investigate the magnitude of these supply pressures by specifying explicit supply of export as distinguished from import demand of an importing country, with a view to comparing the magnitudes of supply pressures of Japan and the United States. In view of lifetime employment in large firms and similar practice among medium and small-sized firms in Japan, it is reasonable to export the supply-pressures to be greater in Japan than in the United States. The analysis is confined, however, to the bilateral trade between Japan and the United States.

The second purpose is to examine the difference in export performance from the demand side. If exporting firms have a medium and long-run strategy of developing productive capacity of these products that have high income elasticity of demand in importing countries, the demand for import of importing countries tends to have high income elasticity. When the supply and demand schedules of exports are more properly specified and estimated, the comparison of price and income elasticities of Japan and the U.S. can be done more accurately.

1. The theoretical framework

We first discuss the conceptual framework (the basic model) and then modify it so that we can apply it to the actual data.

a. The basic model

Supply and demand functions for two countries:

Xit=Xi(Pit,Cit,Zit-1,ΔCit-1,ΔZt-1)(i=J, U)(1)

where suffixes J and U stand for Japan and U.S..

Xit : supply of country i in period t

Mit : import demand of country i

Sit : supply of domestic goods in country i

Dit : exogenous domestic demand

Qit : the relative price for importers of country i (– price of imports relative to that of domestic goods)

Pit : the relative price for suppliers of exports of country i (– price of exports relative to that of domestic goods)

Cit : productive capacity at the full employment level.

ΔCit : a change in productive capacity (ΔCit – Cit−Cit−1)

The equilibrium conditions for exports in two countries:

XJt=MUt(Japan's export supply=U.S.demand for Japan's exports)(5)
XUt=MJt(U.S. export supply=Japan's demand for U.S. exports)(6)

Inventory adjustment process,

Zit=Zit-1+Sit-1-Dit-1(stock of unintended and planned inventory of country i)(7)

The definitions of GNP (measured in the units of domestic goods),

Yit=Sit+PitXit-QitMit(i=J, U)(8)

The exogenous variables that are not determined by the above equations at period t are Cit, Dit, and relative prices of domestic goods. These variables are predetermined by the variables not explicitly treated in the above model. We have equations (1), (2), (3), (4), (7) and (8), for two countries, thus giving rise to 12 equations. By adding equations (5) and (6), we have 14 equations, which determine 14 endogenous variables, XJ, PJ, ZJ, MJ, QJ, YJ, SJ, DJ, XU, ZU, MU, YU, SU, and DU. Note that, although there are four prices PJ, QJ, PU and QU that correspond to four markets (exports and domestic goods of two countries), only two prices are independent in our model. The reason is as follows: if we choose one of the four goods as the numeraire, the independent price variables are reduced to three. Since we can assume the Japan’s export price approximately equals U.S. import price if measured in common units, QU is not independent from PJ. Similarly PU is not independent from QJ. The relationship between these prices are as follows:


where Pd is the relative price of Japan’s domestic goods in term of U.S. domestic goods i.e., Pd – PJd/E•PUd. E is the yen-denominated exchange rate of the U.S. dollar, PUd is the dollar price of U.S. domestic goods and PUd D yen-denominated price of Japan’s domestic goods. We have assumed that Pd is exogenous, or that Pid and E are exogenous. The inventory (Zit) and domestic demand (Dit) are predetermined at each period, together with potential productive capacity (Cit) according to equations (4) and (7).

b. Explanation of the basic model

The new feature of our model is the export supply function. First, it is assumed that, as equation (1) shows, firms (suppliers) face the choice problem of producing exports or domestic goods. Hence the relevant price is the price of exports relative to domestic goods. The price of exports relative to the import price that is often adopted in specifying export functions is not appropriate for our purpose because the latter are the relative price faced by the importing country whose export price happens to equal its domestic price. A rise in the former relative price of exports (in terms of domestic goods) is expected to increase the willingness to supply exports, other things being constant.

Second, the level of potential supply capacity (or production capacity) is assumed to exert an expansionary pressure in the export supply. In addition, not only the level of the capacity, but also an increase in the capacity is assumed to put expansionary pressure on exports.

In the case of inventory, its effect on exports depends on whether it consists of unintended or planned inventory. If it is unintended inventory, its level as well as its rate of increase is likely to put pressure to expand exports in order to reduce undesirable inventory. If it consists largely of planned inventory, however, inventory may not be positively correlated with exports. Suppose a firm decides to reduce the costs of production and distribution and successfully reduces inventory as part of rationalization. Then a reduction in inventory and other costs of operation is likely to lead to an export expansion with certain time lags. In this case the level of inventory is controlled over time through sales promotion and production control. Note, however, there can be unintended inventory investment even in this case due to uncertainty, although unintended may be relatively small. Since actual inventory consists of both kinds, we can have a situation where the effect of the level of inventory is largely dominated by planned one but the effects of the changes in inventory is sometimes dominated by unintended one. The net effect of the actual inventory can be thus negative or positive.

These variables (Ci and Zi) give us some measure of so-called export-drive or additional export expanding efforts at the time of actual or anticipated recessions. We expect that, owing to the life-time employment, the company union system, and other factors such as higher rate of economic growth, additional export augmenting effects are larger in Japan than in the United States.

Equation (2) is an import demand function. The relative price that importers face is the price of imports relative to that of domestic goods. The rise of this relative price will reduce the demand under the normal circumstances. The demand also depends on GNP.

Equation (3) is the supply function of domestic goods. This supply is the other side of the coin of the export supply (equation (1)). We abstract, however from the possible effects of the inventory i.e., it may or may not increase the supply. Equation (4) specifies domestic demand. This demand has to be consistent with import demand, but we assume that domestic demand behaves like an exogenous variable in the short-run due to its low speed of adjustment.

Equations (5) and (6) indicate the equilibrium conditions that determine two export prices PJ nd PU. Whereas the demand and supply are equated in the markets of exports, this is assumed not to be the case in the domestic good markets. The prices and quantities adjust very slowly so that there are unintended and planned increases (decreases) in the inventories.

2. The revised model

In the real world, the world does not consist of only Japan and the U.S.. We treat, therefore, Japan and the U.S. as two separate blocks in the world in the following empirical analysis. The GNPs of the two countries thus become exogenous. We also adopt (lag) linear approximation. The variables in the following equations are the same as those in the previous section. The two blocks are as follows:

a. Japan Block

Japan’s export supply to the U.S.


U.S. import demand for Japan’s exports


Japan’s domestic shipment of manufactured goods


b. U.S. Block

U.S. export supply to Japan


Japan’s import demand for Japan’s exports


U.S. domestic shipment of manufactured goods


Ui, Vi, Wi are error terms of each equation of country i.

Furthermore, variable logxt−logxt − 1 or logxt/Xt − 1 is used as a proxy for variable Δxt. For example, export supply function (equation(1)) is expressed as,


where aJ1, aJ2, aJ3, eJ1 and eJ2 are positive constants.

In other words, higher relative price of exports (PJt) is associated with higher exports. The larger the productive capacity, the larger the export. In addition, it is assumed that not only the level of capacity but also the rate of change of the capacity tends to exert expansionary pressure on exports: when the capacity is increasing, the incentive for a firm to promote exports will be greater for a given level capacity. Similarly, we assume that not only the level of inventory but also the rate change of inventory exert pressure on exports. Equation (17) can be rearranged as follows.

logXjt+aj 0+aj 1logPjt+aj 2logCjt+ej 1logCjt-1ej 1Cjt-1ej 1Cjt-2aj 4logZj+ej 2logZjt-1ej 2logZjt-2+Ut(18)

Furthermore, those independent variables Pjt Cjt, Cjt − 1/Cjt − 2, Zjt, Zjt − 1/Zjt − 2 influence exports with lags. In other words, those coefficients and variables are taken to be vectors. For example, the terms referring the effects of production capacity, aj2logCt + bj1log(Ct − 1/Ct − 2) in equation (17) may be written as,

aj2logCt = αoct + α1ct − 1 + α2ct − 3 + …… + αnct − n

bj1log(Ct−1/Ct−2) = β0(ct−l/ct−2) + βl(ct−2/ct−3) + β2(ct−3/ct−4) + β3(ct−4/ct−5) + …… + βn(ct − n/ct − n - l)

where ct is an element in vector logCt, and α i and βi are constants, By adding and rearranging, we get,

aj 2logCt+bj 1log(Ct-1/Ct-2)=αoct+(α1+β0)ct-1+(α2+β1β0)ct-2++(αn+βn-1βn-2)Ct-n(19)

Similarly, these terms referring to the effects of inventory are aj3logZjt + bj2log(Zjt-1), can be expressed as

aj 3logZjt+bj 2log(Zjt-1/Zjt-2)=γ0zt+(γ1+δ0)zt-1+(γ2+δ1+δ0)zt-2+......+(γn+δn-1δn-2zt-n(20)

where zi is an element of vector logZt.

After all, equation (18) are written as follows:


where xjt, pjt, cjt, zjt are elements in vectors logXjt, logPjt, logCjt and logZjt, kt, dt and mt are constant coefficients.

Other equations (12) - (16) can be also expressed in a manner similar to equation (21).

Note that, although α1, βl, γi and δi can be safely assumed to be positive in the case of unintended inventory, lt-1 and mt-1 (i≥1) are not necessarily positive. They can be zero or negative if the effect on exports of the rate of change of capacity or inventory is larger than that of the level of capacity or inventory. For example, the coefficient of zjt-2 of equation (21), mt-1, is γ210, and can be negative if γ2 + δi < δ0. This is likely to be the case for most of mt-1 as will be discussed later. If γ1 is zero or close to zero, a negative value for mt-1 implies that δ1 < δ0. If γ1 - 0 for all i(i-1,2,3, ...... n), then mt-1 < 0 for all i implies.

δ012δ3, ...... δn-1n>0

The economic implication of this particular case is that while the level of inventory does not affect exports, the rate of change does, and that an increase in inventory put pressure for export expansion with such a lag structure that the effects diminishes as time passes on. If γi < 0 for some i, however, the above inequality may not hold.

In the case of planned inventory, αi, βl, γ1 δ1 may be positive or negative. Even if some of δ1 are negative, however, the coefficients of Zjt-1 i.e., γi + δi-1δi-2, can be zero or positive depending on the relative magnitudes of δ1 and δ(i≠l).

The sample period is 1974. I - 1992.111. For each block, common instrumental variables are applied. Two stage square method is adopted. A Almon lag structure with no restriction is assumed. All variables are natural logarithm values. The estimated coefficients are shown in Tables 1 and 2 below.

Table 1.

Estimated Coefficients: Japan Block

(1976. I - 1992.II)

article image

Significant at 5 percent level.

Significant at 10 percent level.

The figures for the sum of the coefficients in the parentheses indicate the sum of coefficients which are significant at 10 percent or better.

Table 2.

Estimated Coefficients: U.S. Block

(1976. I - 1992. III)

article image

Significant at 5 percent level.

Significant at 10 percent level.

The figures for the sum of the coefficients in the parantheses indicate the sum of coefficients which are significant at 10 percent or better.

II. Data

Data’s are all seasonally adjusted and natural logarithm values are used.

article image

The relative prices are approximated by the following data:

article image

All the Japanese data are obtained from Nikkei NEEDS. Regarding the U.S. data, (f) and (g) are obtained from the Economic Report of the President. Price data for the U.S. are from IMF’s International Financial Statistics.

III. Major Findings and Interpretation

a. Exports

Most of the estimated coefficients for Japan show that those of Japan’s export supply have expected signs from the theoretical point of view, except for those of inventory. The higher the relative price of exports, the larger the volume of exports. The full employment capacity has, by and large, expansionary effects on exports. It may be also noted that the capacity effects are negative for the lagged period of 5.6, and 7 quarters. As discussed earlier, however, this may mean that the effects of the rate of increase of the capacity may be larger than those of the level of capacity of these periods.

The estimated coefficients of inventory indicates that the level of inventory are, by and large, negatively correlated with exports. As pointed out earlier, however, this fact does not necessarily imply that an increase in the level of inventory decreases exports. Is it quite possible, again, that the effects of the rate of increase in inventory are greater than those of the level of inventory, and that the former effects increase exports but the magnitude of the effects diminish as time passes. On the other hand, the possibility can not be precluded that the level of inventory itself is in fact negatively correlated with exports. If this is the case, it may give rise to a puzzle because it means that an increase in unintended inventory would reduce exports, provided that the bulk of actual inventory consists of unintended inventory. 1/

The estimated coefficients of the export supply function of the United States are largely significant and have the expected signs except for inventory. The volume of supply and the relative price of exports is positively correlated. An increase in the full employment production capacity tends to promote exports. The inventory, however, is negatively correlated as in the case of Japan.

b. Imports

The coefficients of the import functions of the two countries are successfully estimated with an exception of Japan’s price effects. The income elasticities of the U.S. is more than five times as large as the income elasticity of Japan’s imports from the U.S. The relative price of Japan’s imports has a wrong sign, but the effect is very small (0.13). 2/ This is the combined effects of earlier positive effects and later negative effects. The wrong sign for the net effects may be due to an aggregation problem because Japan’s imports from the U.S. is the mixture of both agricultural and industrial products.

c. Domestic goods

The estimated coefficients of the price of the supply of domestic goods (shipment) have wrong signs in the case of both Japan and the U.S. This may be due to the fact that the data include those of exports. Most of the U.S. coefficients are not significant and the magnitudes tend to be small. 3/

The capacity effects are present both in Japan’s and U.S. shipment of domestic goods, but again quite a few coefficients are not significant and t-values tends to be small. The sum of significant coefficients of the capacity on the U.S. domestic shipment is again less than that of Japan. In both countries, however, the capacity pressure in which weaker in the domestic market than in the export market.

d. Comparison of estimates

Tables 3 and 4 list the results of the estimation by Goldstein-Khan, Dunlevy and the present author for the sake of comparison. Note that, as pointed out earlier, our results are not comparable to those o G-K and Dunlevy in a strict sense because we estimated the coefficients of Japan’s (bilateral) exports to the U.S. and the U.S. (bilateral) exports to Japan, whereas G-K and Dunlevy estimated those of each country’s total exports to the rest of the world. Nevertheless, we believe that the comparison is useful and highly suggestive to judge the reasonableness of the magnitude and sign of the estimated coefficients. Table 3 indicates the price elasticity of Japan’s export is 0.7. This does not seem unreasonable in view of G-K’s estimate of the same coefficient for Japan itself is ∞, which is inferred from an insignificant estimate. The price elasticity of U.S. export supply to Japan is 4.6, or about 30 percent less than G-K’s and four times as large as Dunlevy’s.

Table 3.

Coefficients of Export Supply

article image

See the footnote (1) of Tables 1 and 2.

over the 1955 through 1975 period.

over the 1954 through 1975 period.

Table 4.

Coefficients of Import Demand

article image

See the footnote (1) of Tables 1 and 2.

Income elasticities of the demand of the rest of the world for the exports of respective country listed above.

Regarding the difference between the two countries, the following can be pointed out. The two kinds of supply pressure, the capacity effect and inventory effect are present both in Japan’s and U.S. export behavior. However, the Japan’s supply pressure is much larger than of Japan’s exports is 9.26 and that of the U.S. is 9.1, the sums of the those coefficients that are between than or equal to “significant at 10 percent level” are 9.27 and 2.87 respectively (see Tables 1 and 2). 1/

IV. Conclusion

With the specification of the supply sides of exports, the supply and demand functions of bilateral exports between Japan and the U.S. are estimated more or less successfully almost all the coefficients are significant and have correct signs from the view-point of the hypothesis discussed in section 1a. In particular, it is shown that the ratio of the export price to that of domestic goods and the two non-price factors of supply pressure i.e., the productive-capacity at full utilization, and that inventory do play important roles in Japan’s and U.S. exports.

It is also shown that, while the two supply-pressure factors are present in the export performance of both countries, they are much stronger Oin the case of Japan’s exports. U.S. producers of domestic goods seem to be also less responsive to a change in capacity in the domestic market than their Japanese counterparts. The inventory of finished goods, however, turned out to be by and large negatively correlated with exports in both countries. This may be due to the rate of change in inventory playing a greater role than the level of inventory itself. It is also possible for planned inventory to have played a more important role than unintended inventory.

These findings suggest that we need to take into account those supply-side factors more explicitly than the traditional Keysian approach in order to have clearer understanding of the determination mechanism of income and the current account in the market. In particular, the presence of strong supply pressure in Japan’s exports suggests that the effect on the current account surplus of sharp appreciations of the exchange rate are likely to be limited. This is particularly the case when appreciations occur after the boom in which both investment and productive capacity expand at a high rate. It is also likely that such demand shocks as contractionary monetary-fiscal policies, or recessions caused by the oil crises tend to give rise to a large current account surplus through the supply pressure in the market in the countries where there exists strong supply pressure.

However, our conclusions have to be carefully qualified in view of the following limitations. The micro-economic foundation of the model needs to be examined with a view to clarify the role of inventory in the firms’ adjustment process. In addition, the empirical studies of export supply behavior at the firm-industry levels have to be incorporated into the aggregate analysis. The margins of error due to those approximations of data of such variables as relative prices, supply pressure and shipments of domestic goods that are adopted here may not be negligible.