Transmission of Business Fluctuations from Developed to Developing Countries
Author: Rudolf Rhomberg

An earlier version of this paper was presented to a conference on the topic “Is the Business Cycle Obsolete?” organized by the Social Science Research Council and held in London, England, on April 3–7, 1967.

Abstract

An earlier version of this paper was presented to a conference on the topic “Is the Business Cycle Obsolete?” organized by the Social Science Research Council and held in London, England, on April 3–7, 1967.

The years since world war II have been an era of remarkable economic stability in comparison with earlier periods for which economic data are recorded. Declines in economic activity in industrial countries have been rare, and business cycle analysts have had to direct their attention to periodic advances and retardations of growth rates—or periods of expansion and of “pause”—rather than to actual booms and recessions of the old-fashioned kind. Cumulative processes of the type that led to severe and protracted depressions in the past have been prevented in the postwar period by appropriate economic policies, by a number of fortunate coincidences, or by an institutional environment more conducive to economic stability than was that of earlier periods.1

The postwar stability of the world economy is particularly striking if attention is directed toward such broad international aggregates as world production or world trade. But the broad view hides many more or less isolated instances of instability, e.g., in individual countries, in individual commodities, or in particular sectors of one or more economies. These instances are, of course, of great importance to those affected directly; but as long as their incidence is random, and the response pattern of the world economy is not such as to transform these individual random shocks into cyclical swings, they do not constitute a threat to world economic stability. They may require action in the affected country, or policies influencing particular commodities or sectors, but they do not ordinarily require concerted policy efforts to sustain or to retard the rate of expansion of world demand as a whole.

The degree of over-all economic stability observed over the past 15 or 20 years should not, however, give rise to complacency. Fortuitous circumstances may not always favor the maintenance of a stable world economy. And while the institutional framework has certainly changed in such a way as to foster economic stability, there is not, as yet, sufficient evidence to indicate with certainty whether economic policy in the main industrial countries has reached a level of sophistication sufficient to ensure stability in all circumstances, or whether it merely has not yet been put to a test comparable to that which faced the policymakers in the 1930’s.

The transmission of fluctuations of economic activity from developed countries2 to developing countries—with which this paper is concerned—is a broad subject. The present study therefore has been limited in a number of ways, to confine itself to certain aspects of the topic.

First, an aggregative approach is chosen in which only two regions are distinguished, the developed countries as one group and the developing countries as another. While this approach is appropriate for assessing the magnitude of business fluctuations in developed countries, the transmission of these fluctuations to the developing countries, and the response of the latter, it does not show the effects which changes in economic activity in individual developed countries (or in subgroups of such countries) may have on the economies of individual developing countries (or subgroups of them). This is a severe limitation of the present paper; it is intended to supplement, at a later stage, the findings reported here by studies of smaller groups of countries.

Second, the paper deals only with short-term variations in economic activity and their international transmission. This excludes discussion of long-run variations in the production and prices of individual classes of commodities. Long-run fluctuations of this type may indeed have cyclical characteristics; however, they are usually not on a world-wide scale, but are specific to individual commodities or sectors. To whatever extent they are universal, moreover, they are more profitably studied in the context of economic growth rather than in that of short-term fluctuations.

Third, attention is directed to the transmissions of business fluctuations from developed to developing countries and not to transmission in the reverse direction. This is not to deny that individual developing countries may experience economic fluctuations with cyclical characteristics. But these countries, as a group, have not been subject to autonomous short-term fluctuations of sufficient magnitude to affect to a noticeable degree the economies of the developed countries as a group.3

Fourth, two limitations of a statistical character have had to be accepted: (1) The analysis is based on annual data. It would, of course, be more satisfactory to use semiannual or quarterly values for an investigation of a cyclical process. Unfortunately, however, suitable data on regional trade flows and trade by commodities for time periods of less than 1 year are not available for a sufficiently long period. (2) The indicator of changes in economic activity in the developed countries chosen for this study is the combined (weighted) index of industrial production in these countries. For more detailed investigation of the topic of this paper, it would be appropriate to base the analysis on a variety of cyclical indicators (such as changes in inventories, the rate of unemployment, and changes in orders) in different countries or regions, and to ascertain the influence which each of these may have on imports from, and capital exports to, the less developed countries. This study is, however, limited by the nature and quality of the annual data on trade and capital flows. For this reason, refinement of the data (indicating variations in economic activity in the developed countries) may not be warranted. The paper confines itself, with two exceptions to be mentioned later, to an analysis of the apparent responsiveness of trade and capital flows between developed and developing countries to variations in economic activity, as indicated by the combined annual index of industrial production for the developed countries as a group. Its aim is to reach some broad generalizations regarding the dependence of developing countries on short-run variations in economic activity in developed countries.

After a brief discussion of the channels through which business fluctuations may be transmitted from developed to developing countries, and of the timing and intensity of the fluctuations experienced in the postwar period, a statistical analysis of the influence of short-term fluctuations in aggregate economic activity in developed countries on their imports from developing countries is presented. Then the influence of changes in economic activity in developed countries on capital flows to developing countries is discussed. This is followed by consideration of the response mechanism of developing countries as a group to cyclical influences transmitted to them. The final sections of the paper show why the effect of variations in economic activity in developed countries upon the economies of developing countries and their aggregate balance of payments has been relatively slight and diffuse; it is argued that the response mechanism in developing countries has in fact contributed to the dampening of fluctuations in the world economy.

I. Channels of Transmission

Trade in goods and services is the principal channel through which changes in economic activity in developed countries affect the economies of developing countries. Actual or anticipated changes in production in developed countries result in variations in demand for raw materials imported from developing countries. Since the supply of such materials is generally not very elastic, especially in the short run, variations in demand should be expected to cause changes in the same direction in the prices of these commodities. The more inelastic are supply and demand with respect to price, the larger will tend to be the induced variation in prices, and the smaller that in quantities, of exports of materials from developing countries. Demand for food, on the other hand, is less responsive to variations in economic activity, and cyclical changes in developed countries should not be expected to have a pronounced effect on exports of food products from developing countries. During the postwar period there has been a small, though rapidly growing, amount of exports of manufactured goods from developing to developed countries. The export volume of this class of products should be expected to show some cyclical responsiveness, although perhaps less than manufactured exports from developed to other developed countries, which consist to a large extent of investment goods.

Fluctuations in the exports of developing countries will tend to be mitigated by two factors which have the character of built-in stabilizers. One factor is the extent to which exportables are produced by foreign-owned companies. A change in such exports will be associated with a change in the same direction in repatriated earnings and will therefore reduce the variations in the current account balance resulting from fluctuations in exports. This influence is quite pronounced in some countries. From a sample of 29 countries, it has been estimated that it may amount, on average, to as much as 7 per cent of the value of exports; and in some oil-producing countries, it may range from small values to more than 40 per cent.4 The second factor is the extent to which the production of exportables requires imports from the developed countries. The import content of exports of developing countries is difficult to assess, but it has been estimated at perhaps 5 per cent of the value of exports.

The transmission of business fluctuations through the current account of the balance of payments has been widely discussed in the literature. Less attention has been devoted to the possible transmission of fluctuations through capital movements, especially in the context of relations between developed and developing economies. Since changes in economic activity in developed countries affect both the availability of capital and domestic investment opportunities, and since the transmission of these fluctuations to developing countries affects the business situation there, the possibility of a systematic variation of capital flows from developed to developing countries, in response to cyclical business fluctuations, cannot be excluded. If such an influence is found to exist, developing countries may be harmed by the drying up of the private capital inflow from developed countries in addition to the reduction of export receipts which they experience as a result of a decline in economic activity in developed countries.

Changes in public loans and grants from developed to developing countries would, in principle, affect the latter group of countries in the same way as variations in private capital flows. It is plausible to assume, however, that changes in public capital flows are not subject, to any great extent, to cyclical factors in the developed countries. It may be that decisions concerning the flow of public capital from individual developed countries are subject to considerations with regard to the balance of payments positions of these countries. However, longer-run viewpoints often prevail over these considerations, and public capital flows are not ordinarily altered abruptly in response to short-term changes in the budgetary or balance of payments situation. Moreover, it may be argued that cyclical influences will not greatly alter the over-all balance of payments of the developed countries as a group. While some of these countries may experience balance of payments difficulties when there is a recession in economic activity in other countries of the group, the latter would tend to find their external payments position eased. The effect which such changes may have on the flow of public loans and grants to developing countries depends on the incidence of these balance of payments shifts among countries with different propensities to make such loans and grants.

A number of minor channels of transmission will be left out of account in this discussion, e.g., profits made, or losses sustained, by nationals of developing countries as a result of investment in the stock markets of developed countries.

II. Economic Fluctuations in Developed Countries

Fluctuations in economic activity in the main industrial countries during the postwar period were not only mild but also dissynchronous. In the years covered by the study, 1950–65, the United States has had recessions in 1953–54, 1958, and 1960–61. Each of these downturns was short-lived, lasting about 1 year or less. Industrial production on an annual basis fell in 1954 and 1958, but not in 1960 or 1961. Recessions were even milder in industrial Western Europe. Only two recessions, or pauses, in economic growth in that area can be clearly identified: one was in 1951–52 when the annual index of industrial production failed to rise, and the other in 1958 when the index was only 2 per cent higher than in 1957 (Table 1). Western Europe did not share with North America (i.e., United States and Canada) either the recession of 1953–54 or that of 1960–61; in fact, in 1953–54 it experienced a vigorous recovery from the 1951–52 recession. On the other hand, the European recession of 1951–52 was not shared by the United States, where there was only a slight retardation in the growth of economic activity.

Table 1.

North America, Western Europe, and All Developed Countries: Indices of Industrial Production

article image
Source: United Nations, Monthly Bulletin of Statistics.

United States and Canada.

OECD, Australia, New Zealand, and South Africa.

As a result of these divergent movements, the combined index of industrial production for the developed countries, which is taken in this study as an indicator of variation in economic activity in the entire group, showed two slight declines—one from 1953 to 1954 and one from 1957 to 1958. In addition, it showed particularly small increases from 1951 to 1952, from 1956 to 1957, and from 1960 to 1961. This indicator does therefore reflect, even on the annual basis, all the recessions or pauses experienced by either North America or Western Europe.

III. Transmission Through Foreign Trade

A number of recent studies deal with the dependence of exports of developing countries upon economic activity in developed countries during the postwar period. For instance, the world trade model of the International Monetary Fund (IMF)5 contains equations for imports by the United States and Western Europe from one another and from the rest of the world.6 The relevant income and price elasticities, evaluated at the means of the variables, are shown in Table 2. It appears that the U.S. demand for imports from the rest of the world is somewhat more elastic with respect to gross national product (GNP) than is that of Western Europe. The weighted GNP elasticity of imports by these two industrial regions combined from the rest of the world is slightly higher than unity. Estimated price effects on the volume of exports of the rest of the world are, in this model, low or nil.

Table 2.

Incomes and Price Elasticities as Derived from IMF World Trade Model1

article image

The original model is described in Rudolf R. Rhomberg and Lorette Boissonneault, “Effects of Income and Price Changes on the U.S. Balance of Payments,” Staff Papers, Vol. XI (1964), pp. 59–124. The elasticities given here resulted from a re-estimate of the model, based on data for 1953–65.

Elasticity with respect to gross national product (GNP).

Elasticity with respect to the ratio of import prices to the domestic GNP price deflator.

Test of lag pattern

The question arises whether the use of annual data obscures any lag in the response of imports to changes in economic activity. In work with quarterly import functions, it had been found that there is either no lag or a very short lag between changes in economic activity and the changes in imports induced by them.7 The lag pattern was tested once more by the present author; he computed equations with (seasonally unadjusted) quarterly data from the first quarter of 1948 to the first quarter of 1966 for imports of merchandise and of total goods and services into the United States from developing areas. Industrial production (seasonally adjusted) was used as the independent variable. It was again found that the best results are achieved if imports are unlagged, or follow with a lag of one quarter behind the variable expressing changes in economic activity. Lags of two or more quarters give distinctly worse results, and are in fact unacceptable when computations are carried out in first differences of logarithms. (These tests are shown in Appendix I.) It may be concluded that in the United States imports lag behind industrial production with an average lag of 1 to 3 months, and that the corresponding relation in terms of annual data should be unlagged.

Cyclical dependence of exports to developed countries

In order to assess the dependence of export receipts of developing countries on year-to-year changes in economic activity in developed countries, annual changes in the logarithms of export values, export volumes, and export prices were regressed on annual changes in the logarithms of the index of industrial production in developed countries.8

Equations of this form relate proportionate changes in the dependent variable to those in the independent variable; the regression coefficients therefore indicate apparent elasticities of response of the former to changes in the latter. The estimated relationships are independent of the common upward trends observed in all the variables in question, since they are based on year-to-year changes.

Our present aim is not to derive the best possible explanatory equations for imports of developed from developing countries. If that were the objective, a number of other explanatory variables, such as relative prices, changes in inventories, and indicators of changes in demand by economic sector, would be introduced. Our aim is merely to establish the degree of dependence of exports to developed countries on our selected indicator of variations in economic activity in those countries.

Particular interest attaches to the question of the separate effects of changes in economic activity in the developed countries on the quantities and prices of their imports from developing countries, and to a comparison of these results with the corresponding responsiveness of quantities and prices of imports of developed countries from other developed countries. This response pattern was tested for total merchandise trade as well as for four commodity categories, namely, food and beverages, raw materials, fuels, and manufactures. Annual data for the period 1953-65 were used; the results are shown in Table 3, and further details are given in Appendix II.

Table 3.

Elasticity Coefficients of Value, Quantum, and Unit Value of Exports to Developed Countries from Developing Countries and Developed Countries with Respect to Industrial Production in Developed Countries1

article image

The elasticities were determined from regressions of first differences of the logarithms of value, quantum, or unit value on first differences of the logarithms of the index of industrial production of the developed countries. Elasticity coefficients with an asterisk (*) are significant at the 5 per cent level; others are not. Annual data for 1953–65 were used. The commodity groupings are the United Nations Standard Industrial Trade Classifications (SITC). For details, see Appendix II.

According to traditional views, the export earnings of primary producing countries are more unstable than those of industrial countries. In the short run, the supplies of the commodities exported by primary producers are inelastic, and so is the demand for these products. Shifts in demand tend, therefore, to induce some variation in the volume of exports but larger variation in export prices. Supply shifts will produce offsetting changes in volume and price, but they usually lead to changes in value in a direction opposite to those of the underlying shifts in supply. By contrast, shifts in demand for exports of industrial countries are expected to induce larger variations in volume than in price, and shifts in supply of these exports are not very pronounced.

The findings presented here are not intended to support or to contradict these views. In fact, they are not directly comparable with the findings of studies on the instability of prices and volumes in international trade. In the first place, the present study is not concerned with the over-all instability of export proceeds, but merely with the effect of changes in economic activity in the (developed) importing countries on export proceeds of developing and developed countries. In other words, short-term changes in demand for reasons other than variations in economic activity, and short-term changes in supply in the exporting countries, are left out of account. Second, the distinction made is not between industrial and primary exporting countries, but rather between developed and developing countries. Among the former are some, like the United States, whose exports consist to a considerable extent of primary products. At the same time, exports of manufactured goods from developing countries, though initially small, have been rapidly expanded.

The extent to which exports of developing and developed countries have responded to variations in economic activity in the developed countries is shown in Table 3 for all commodities and for certain subgroups. The figures for total merchandise exports support the view that the volume of exports from developed countries is much more responsive than the volume of exports from developing countries. The estimated elasticities of the volume of exports with respect to industrial production in developed countries are, respectively, 0.37 for developing countries and 0.90 for developed countries (first line of Table 3). The expected difference in price responsiveness is, however, not found. Table 3 shows that the unit values of both groups of countries are only weakly responsive to changes in economic activity in the developed countries, and that there is no significant difference in the two estimated coefficients. As a result, changes in total export proceeds of developed countries are much more responsive to changes in industrial production in their own area (with an elasticity of 1.06) than are the export proceeds of developing countries (with an elasticity coefficient of 0.48). From this, it may be concluded that a change of 1 per cent in industrial production in developed countries would tend to change export proceeds of that group in the same direction by about 1 per cent, but those of developing countries by only about ½ of 1 per cent.

For the four commodity groups listed in Table 3, there is a clear indication that developing countries’ export proceeds from raw materials are strongly responsive to variations in economic activity in developed countries, and that this responsiveness is to a great extent one of prices rather than one of quantities. Interestingly, there is a similar indication for manufactured products. Their price responsiveness is particularly high, whereas the elasticity coefficient for the volume of exports is small and statistically uncertain. For these two groups taken together—which account for not quite one half of the developing countries’ exports to developed countries—export proceeds would tend to change by about 1¾ per cent in response to a change of 1 percentage point in the industrial production index of developed countries. Most of this change would be accounted for by price variations. For the other two commodity groups, food and fuels, quantities exported from developing countries to developed countries are found to be only weakly responsive to economic activity in the recipient area, and prices appear to be inversely affected. But as these elasticity coefficients are statistically not significant, not much weight should be given to this finding. If the true elasticity of export proceeds from these two commodity groups were to be set equal to their apparent quantum elasticities, the elasticity of developing countries’ total export proceeds with respect to economic activity in developed countries would be close to unity. Even in that case, however, it would not exceed the responsiveness of total export proceeds of developed countries from other developed countries.

For the developed countries, the expected responsiveness of export volumes are found for all commodity groups, with the elasticity of food being very low and that for each of the three other commodity groups being about unity. Price responses are found to be small and statistically uncertain.

The equations for total merchandise exports could be computed on the basis of a somewhat longer time period, namely, 1950-65. In this instance, the apparent elasticity of the volume of developing countries’ exports to developed countries is almost the same as that found for the period 1953-65, but the price responsiveness is somewhat higher. As a result, the elasticity of total export proceeds of developing countries from developed countries, computed for the longer period, is 0.71.9 The corresponding elasticity of export proceeds of developed countries is somewhat lower than that computed for the shorter period, namely, 0.87.

Although more intensive study would be necessary to reach firm views on this matter, it must be taken as doubtful that export proceeds of developing countries—as a group—to developed countries are more strongly affected by short-term changes in economic activity in developed markets than are the export proceeds of the developed countries themselves. If anything, the indication is the reverse.10 As stated earlier, these conclusions are drawn for the aggregate of developing countries. Export proceeds of individual members of this group may be very severely affected by declines in economic activity in their main export markets, or in the group of developed countries as a whole. Moreover, the conclusion refers only to short-term changes; although the fluctuations in economic activity in North America and Europe did not occur at the same time and were rather quickly reversed, it can nevertheless be argued that export proceeds of developing countries would have shown not only greater stability but also a higher growth rate if even these minor setbacks to economic growth in the developed countries could have been avoided.

IV. Transmissions Through Private Capital Flows

Some general considerations that would lead to expectations that capital flows from developed to developing countries may respond to changes in economic activity have been advanced above. There is, of course, the question of whether these influences are strong enough to be detected in the available data. This question is particularly acute, since the response of capital flows to changes in economic activity is unlikely to be a simple and simultaneous one.

Capital movements may be expected to respond to factors affecting the supply of funds—as reflected in past retained earnings of corporations and the general ease or tightness of the capital market—and to considerations of alternative investment opportunities at home and abroad. When retained earnings are high relative to investment at home, the incentive to invest abroad should be relatively strong, and vice versa. At the start of a recession, domestic investment opportunities, in relation to given investment opportunities abroad, tend to appear less promising than earlier; but earnings accumulated over the recent past are still high.11 Investment abroad should therefore remain high, or even rise, for some time after a recession has started. After the recession has continued for a while, retained earnings will decline and thus lower the supply of funds available for foreign investment. At the same time, the incentive to invest in foreign export industries, whose profits may be affected by the domestic recession in economic activity, may be impaired. Nevertheless, if the recession is still in progress, the continued reduction in domestic investment opportunities may offset these adverse effects on foreign investment. Shortly after the beginning of a recovery, past accumulated earnings are presumably at their lowest point, and domestic investment opportunities improve. As a result of both of these factors, investment abroad should continue to decline. Only after the recovery has gone on for some time will earnings improve and begin to counteract the negative effect of better domestic investment opportunities on foreign investment. Still later, earnings will have fully recovered, and domestic investment opportunities will tend to level off. At that point, foreign investment will again rise, and the cycle will have been completed. This hypothesis leads to the expectation that changes in foreign investment will follow changes in domestic economic activity with a substantial lag.

To test the actual lag pattern, it was again necessary to use quarterly data. Data for the United States were employed. Quarterly U.S. direct investment in developing countries, and alternatively the total net flow of private U.S. capital to these countries on a quarterly basis, were related to the quarterly index of industrial production, with various discrete lags. On the basis of this test it was concluded that a lag of capital flows behind economic activity of approximately 1 year was appropriate.12

Unfortunately, continuous time series on the flow of capital from all developed to developing countries are not available for a very long period. Data beginning with 1956, published by the Organization for Economic Cooperation and Development (OECD), show that the flow of total private long-term net capital from industrial OECD countries to developing areas in 1958 was less than in 1957, and that the decline continued in 1959. After some recovery in 1960 and 1961, the capital flow declined again in 1962 and further in 1963.13

In attempting to relate the hypothesis outlined above to annual data (the only data available for all countries), the trade data of developing countries were combined with changes in international reserves to compute a residual series which represents net capital flows to developing countries plus net credit balance on service account minus net deficit with the Soviet countries and Mainland China on trade account. The data are shown in Table 4. Although the resulting series does not represent capital movements exclusively, its short-term variations can be considered representative of the changes in net capital flows.14 The changes in the series shown in column 4 of Table 4 indicate reductions in the net flow of foreign exchange on service and capital account to developing countries in 1953 and in 1954, again in 1958 and 1959, and once more in 1961 and 1962; in addition, there was a decline in the rate of net inflow in 1964. Here again, the evidence is not clear as to the lag by which capital flows to developing countries follow economic activity in the industrial countries. The data may be consistent with a lag of anywhere from 6 to 18 months. But there is some indication, even in these impure data, that a relation of the type postulated exists.

Table 4.

Developing Countries: Trade Balance with Developed Countries, Reserve Changes, and Residual Balance of Payments Items, 1950–65

(In billions of U.S. dollars)

article image

Merchandise exports of developing to developed countries less merchandise exports of developed to developing countries. Source: United Nations, Monthly Bulletin of Statistics.

Change in reserves of the group described in the source as Less Developed Areas. Source: International Monetary Fund, International Financial Statistics.

Column (2) minus column (1). This residual represents (a) the net flow of capital and aid into developing countries from all other countries plus (b) the net receipts on service account by developing countries, minus (c) the merchandise trade deficit of developing countries with the Soviet countries and Mainland China, plus (d) a statistical discrepancy reflecting the difference between developing countries’ imports from developed countries and the exports to them reported by developed countries.

Change from preceding year; based on data in column (3).

When this series (Cdl) is related to the index of industrial production in developed countries lagged 1 year, the following results, in terms of first differences, are obtained:

ΔCdlt=0.23+0.06(1.3)ΔAt1(1)R¯2=0.05d=2.16

The coefficient of Δ At-1 is not significant at the 5 per cent level, but the magnitude of the effect, a change of $60 million in the annual capital flow for a change of 1 percentage point in industrial production, is not inconsistent with the value found for the response of U.S. capital alone ($6 million quarterly, i.e., $24 million annually, for a change of 1 percentage point in U.S. industrial production).

It may be tentatively concluded that, in addition to the trade effect of fluctuations in economic activity in developed countries on the economies of developing countries, there may be an influence through systematic variations in capital flows, but that this influence is exercised with a lag which may be as long as 1 year.

V. Response Mechanism of Developing Countries

The two preceding sections have indicated the intensity and time pattern of the transmission, through the balance of payments, of variations in economic activity in developed countries to the economies of developing countries. The purpose of this section is to consider the response pattern in the developing countries. Here again, considerable differences among members of the group of developing countries would be expected, and these differences and their explanation merit further study; but this paper concerns itself only with the global response pattern of the countries taken as a group.

The main features of the response pattern have been discussed extensively in the literature. Changes in export receipts, whether they occur through variations in the volume of exports or price changes, result in income changes in the export sector of the affected economies. These changes tend to lead directly to changes in the same direction in imports, as well as to a multiplier effect on the domestic economy which will result in further changes in the demand for imports. In addition, there may be a more direct connection between the demand for imports and changes in export receipts. Since foreign exchange reserves in most of these countries are not ample, and do not admit of large downward variations, the authorities may be compelled to tighten import restrictions fairly promptly after a decline in export receipts, unless this decline is offset by an inflow of capital or foreign aid. When export receipts increase, import restrictions may be, and often are, relaxed as soon as some increase in foreign exchange reserves has occurred.

The response to changes in the net capital inflow is similar, in more than one way, to the response to changes in export receipts. In the first place, an increase in receipts on capital account may lead to a relaxation of restrictions; and a reduction in the inflow, to a tightening of restrictions. Moreover, an increased capital inflow is often accompanied by, or is the financial equivalent of, increased investment expenditure, and has a multiplier effect similar to that of a change in exports.

As a result, it is not inappropriate to lump together export receipts and net capital inflows into a total of “gross foreign exchange receipts,” and to expect changes in import expenditures to be related partly to the change in this total in the current year and partly to the change in the preceding year.15

The reaction of imports to present and recent past foreign exchange receipts from exports and capital inflows may not exhaust the response mechanism. The private responses may be such that, with a given set of foreign exchange restrictions and financial policies of the countries concerned, import demand tends to remain behind, or to run ahead of, gross foreign exchange earnings. In addition, the authorities may find that, in adjusting their financial policies and foreign exchange measures to achieve balance of payments equilibrium, they either have insufficiently compensated or have overcompensated for the changes in gross foreign exchange receipts which have occurred. In such instances, foreign exchange reserves tend to change by amounts different from those anticipated, and the authorities may take additional corrective action. It is reasonable, therefore, to test the hypothesis whether, in addition to current and lagged gross foreign exchange receipts, the past change in foreign exchange reserves (that is, the change in the preceding year) helps to explain current import expenditures.

Since the levels of import expenditures and gross foreign exchange receipts from exports and capital inflows are obviously related over time, the response mechanism is appropriately tested not in terms of the levels of these variables but in terms of annual changes. As a result, the variable for the change in foreign exchange reserves of the preceding year is expressed in first differences; that is, it is the change in the change in foreign exchange reserves which enters the equation. The result is as follows (data period 1952-65):

ΔVdlt=0.173+0.795(6.3)ΔGt1+0.410(4.4)ΔGt1+0.407(2.1)ΔΔRlt1(2)R¯2=0.87d = 1.95

In this equation, Vdl represents the value of exports of developed to developing countries (i.e., the imports of developing from developed countries); G equals Vld+ Cdl, which is the value of gross foreign exchange earnings of developing countries (i.e., the value of their exports plus the net capital inflow from developed countries, including developing countries’ net balance on service account and their net trade balance with the Soviet countries and Mainland China); and Rl is the developing countries’ international reserves (gold, foreign exchange, and reserve positions in the IMF). The coefficients of ΔGt and ΔG t-1 are significant at the 1 per cent level, while the coefficient of ΔΔRlt-1 barely misses significance at the 5 per cent level. This equation has a standard error of $450 million, or less than 2 per cent of the recent value of developing countries’ imports from developed countries.

The addition of an indicator of the change in economic activity in developing countries, namely, the index of industrial production in these countries, did not add to the explanatory power of the equation; neither was that indicator by itself significantly related as an explanatory variable to changes in developing countries’ imports from developed countries. This finding does not contradict the existence of a multiplier process or of a tendency of imports to respond to changes in economic activity in developing countries. It merely means that, in view of the generally low international reserves of these countries, the process of balance of payments adjustment—brought about through market forces, through financial policies, or through variations in import restrictions—is necessarily very rapid. As a result, changes in foreign exchange receipts are a better predictor of changes in imports than are changes in economic activity.

The interpretation of the response mechanism of developing countries to variations in their gross foreign exchange receipts from developed countries, as it emerges from this equation, is as follows. A change in gross foreign exchange receipts tends to result in a concurrent change in import expenditures of about four fifths of the change in receipts. One year after the change in foreign exchange receipts has occurred, there is a further induced change in the same direction in import expenditures by about two fifths of the change in receipts. If the rate of reserve accumulation in the preceding year is higher, or lower, than the rate was 2 years earlier, two fifths of this difference will tend to be added to, or subtracted from, import expenditures of developing countries. In every year, import expenditures tend to increase less or to decrease more—by about $170 million—than they would have done on the basis of the influences expressed in the three variables just discussed.

The nature of the adjustment mechanism may be appraised by adding to equation (2) an identity for the current change in international reserves, which is defined as the value of gross foreign exchange receipts (exports plus net capital inflows) minus the value of imports:

ΔRlt=Vldt+CdltVdlt.

Thus, the change in the change in international reserves is

ΔΔRlt=ΔVldt+ΔCdltΔVdlt=ΔGtΔVdlt(3)

By substituting (3) in (2), the following expression is obtained for the change in developing countries’ imports from developed countries:

ΔVldt=0.173+0.795ΔGt+0.817ΔGt10.407ΔVdlt1(4)

The change depends about equally on changes in gross foreign exchange receipts in the current year and the preceding year, and is negatively affected by its own value in the preceding year and by a constant trend decline.

Table 5 shows the response of imports, the foreign balance, and international reserves in three cases. In Example A, gross foreign exchange receipts increase, after a period of constancy, by $1 billion in year 1 and remain at that new level without further change; in Example B, they increase by $1 billion in every year; and in Example C, they fluctuate.

Table 5.

Response Pattern of Developing Countries’ Imports and Reserve Changes to Changes in Their Receipts from Exports and Capital Inflows1

(In billions of U.S. dollars)

article image

Computed from equations (2) and (3). Changes in the year preceding year 1 are assumed to be zero.

These examples illustrate the tendency, inherent in equation (4), of imports to overadjust to changes in gross foreign exchange receipts. Balance is restored through the effect of reserve developments on imports.16 Column 8 of Table 5 shows the effects of the assumed changes in gross foreign exchange receipts on the over-all balance of payments. Example C illustrates the rapid balance of payments adjustment and the correspondingly weak effect of fluctuations in gross receipts on reserves. This may help to explain the small variation in reserves of developing countries, which remained virtually unchanged from 1953 to 1965 and varied within this period in a relatively narrow range, i.e., between $8.4 billion (1962) and $10.7 billion (1956).

VI. Dampening the Effect of Business Fluctuations

The findings of this paper cover (1) the effect of fluctuations in economic activity in developed countries on export proceeds of developing economies, through their effect on both quantities and prices; (2) the effect of these changes in economic activity on the flow of capital from developed to developing countries; and (3) the response mechanism of the developing countries as a group to these impulses. It is possible to gather these findings into a small model with the following five equations (with the variables as previously defined):

ΔVldt=0.13ΔAt(5)
17
ΔCdlt=0.06ΔAt1(6)
ΔVdlt=0.17+0.80Gt+0.41Gt1+0.41ΔΔRlt1(7)
ΔGt=ΔVldt+ΔCdlt(8)
ΔΔRlt=ΔGtΔVdlt(9)

The solutions for changes in the developing countries’ imports, trade balance (B), and over-all balance of payments (which equals the change in their reserves) are as follows:

ΔVdlt=0.41ΔVdlt1+0.10ΔAt1+0.16ΔAt1+0.05ΔAt20.17(10)
ΔBt=0.41ΔVdlt1+0.03ΔAt0.16ΔAt10.05ΔAt2+0.17(11)
ΔΔRlt=0.41ΔVdlt1+0.03ΔAt0.10At10.05ΔAt2+0.17.(12)

Each of these magnitudes depends on economic activity in the developed countries in the current period and the two preceding periods, as well as on the change in imports in the preceding period. As long as changes in economic activity in the developed countries reverse themselves frequently, with each phase being of short duration, the influence of these fluctuations on the economies of the less developed countries, taken as a group, will be spread out over time and weakened through mutual offsetting. The developing countries will still experience the effect of previous prosperous years during a year of recession in the developed countries. At a time when they are responding to a past recession, they will already be under the influence of the subsequent recovery in the industrial countries.

The trade and over-all balances are affected chiefly by lagged variables, and only to a small extent by current industrial production. This suggests the possibility of forecasting the magnitudes of the cyclical components of these dependent variables—though not the variables themselves, which may be affected by factors left out of account in this study—for the year following the latest 12-month period for which the requisite data are available.

As a result of the response mechanism discussed in the preceding section, and of the brevity of recessions in the developed countries, the developing countries have, as a group, exerted a stabilizing influence on the world economy. Declines in their exports have not been fully reflected in concurrent declines in imports, and part of the adjustment of imports has tended to occur after the recessions have ended. While it is true that developing countries have returned impulses emanating from the capital account, changes in developed countries’ imports from developing countries and in their capital exports to these countries have not been synchronous. Consequently, the effect on developed countries’ exports of the response by developing countries to these impulses has been, at least to some extent, offsetting rather than reinforcing.

The model also shows, however, that the pattern would have been different, had there been a protracted decline in economic activity in developed countries. In this case, the effects of declines in their imports from developing countries in successive years would begin to reinforce one another, and the effect of reductions in capital exports to developing countries would, some time after the beginning of a prolonged decrease in economic activity, start to reinforce the trade effects. As far as the economic relations of developed and developing countries are concerned, there would then be a cumulative process in which developing countries would, after some time, return the full impact of a business decline in developed countries.

VII. Repercussions on Developed Countries’ Exports

In order to determine the extent to which, over the period 1953-65, fluctuations in exports from developed to developing countries were the apparent echo effect of variations in economic activity in the developed countries themselves, the model can be solved, taking as given changes in industrial production in the developed area and also the initial change in these exports for 1952-53. The computed (simulated) changes are then compared with those actually observed (Table 6). The discrepancies can be interpreted as having arisen from causes other than shortterm variations in economic activity in the developed countries. They may be the result of other factors operating in the developed economies, such as (1) autonomous changes in private and public capital flows to developing countries, i.e., changes that are unrelated to changes in economic activity as measured in this study, or (2) changes in developed countries’ imports independent of variations in economic activity. To some extent, however, they may be due also to factors operating within the developing region.

Table 6.

Exports of Developed to Developing Countries: Actual and Computed Changes, 1952–65

(In billions of U.S. dollars)

article image

Computed from equation (10); computed values of exports in year t were used to calculate exports in year t+1.

Actual minus computed values.

During the 1953-54 recession, the difference between actual and computed exports of developed to developing countries was not large. In the 1958 recession (1957-58), actual exports declined much more than computed exports, and exceeded, therefore, the decline that could be accounted for by the echo effect. The same is true of the 1960–61 recession, when actual exports rose very much less than computed exports.

The marked declines, or reductions in the rate of growth, of exports of developed to developing countries in 1952-53, 1957-59, and 1960-62 cannot be explained entirely as a reflection of declining business activity in the developed countries. They seem also to be related to substantial reductions in the rate of net capital flow from developed to developing countries, which exceeded by far the reductions that we have been able to ascribe to cyclical causes. Similarly, the large rise in exports from 1956 to 1957 seems to have been associated with a large increase ($1.6 billion) in the net capital flow to developing countries.

It is generally agreed that developing countries do not contribute to cyclical instability of economic activity in developed ones. It is possible, however, to go further, and argue that they have contributed to stability of aggregate demand in developed countries. Indeed, this proposition is supported by inspection of the timing of changes in the international reserves of developing countries (see Table 4, page 15). Reductions occurred in 1952, 1957, 1958, 1961, and 1962. To the extent of these reductions, developing countries helped to support aggregate demand in the developed countries during these recessive periods.

APPENDICES

I. Test of Lag of Imports Behind Industrial Production

Equations with lags of varying length were computed for (undeflated) quarterly U.S. imports of goods and services and merchandise imports. The independent variable is the seasonally adjusted index of U.S. industrial production. Since imports are seasonally unadjusted, quarterly dummy variables are included in these equations. The data cover 73 observations from the first quarter of 1948 to the first quarter of 1966. The equations without lag and with lags of one and two quarters, computed with first differences of logarithms, are shown in Table 7.

Table 7.

Quarterly U.S. Imports from Developing Countries and U.S. Industrial Production Unlagged and Lagged One and Two Quarters1

article image

Regressions of the change in the logarithm of the value of imports on the change in the logarithm of industrial production (Δ log A) with indicated lags. Quarterly data from first quarter 1948 to first quarter 1966 (73 observations) are used. Figures in parentheses below the regression coefficients are their t ratios.

Durbin-Watson test statistic for serial correlation of residuals. The values indicated by an asterisk (*) fall in the indeterminate range of the test.

For imports of goods and services, and also for merchandise imports alone, the equation with a lag of one quarter is slightly superior to that without lag. But with a lag of two quarters the correlation is significantly reduced; and with longer lags, the results become progressively worse.

II. Regression Equations

Table 8 presents the detailed regression results discussed in connection with Table 3 on page 10 above, as well as the alternative regression equations referred to in footnote 8.

Table 8.

Elasticity Coefficients of Value, Quantum, and Unit Value of Exports of Developing and Developed Countries to Developed Countrieswith Respect to Industrial Production in Developed Countries1

article image

Annual data, 1953–65. The equations are of the form ΔlogXt = a + b ΔlogAt + ut and log Xt = a + b logAt + ct + vt, where Xt stands, respectively, for value, quantum, or unit value of exports, At for the index of industrial production of developed countries, t for a linear trend, and ut and vt for the residuals of the two equations.

Ratio of coefficient to its standard error. An asterisk (*) indicates that the coefficient is significant at the 5 per cent level.

R2 is the coefficient of determination corrected for degrees of freedom.

Durbin-Watson statistic for serial correlation test.

As a result of the correction for the number of degrees of freedom, R2 equals zero.

Excluding fuels.

Smaller than 0.005.

La transmission des fluctuations de l’activité économique des pays développés aux pays en voie de développement

Résumé

Cette étude s’efforce d’évaluer les effets des fluctuations à court terme de l’activité économique dans l’ensemble des pays développés sur la balance des paiements de l’ensemble des pays en voie de développement. L’auteur examine deux circuits de transmission des fluctuations de l’activité économique : l’influence des variations de la demande d’importations dans les pays développés sur les recettes d’exportation des pays en voie de développement et l’influence par l’intermédiaire des variations provoquées dans les mouvements de capitaux privés vers les pays en voie de développement. Les éléments de l’analyse sont rassemblés dans un petit modèle à partir duquel il est possible de calculer les variations des exportations des pays développés à destination des pays en voie de développement en se fondant sur des données relatives aux fluctuations de l’activité économique dans les pays développés, et ces résultats calculés peuvent être comparés aux variations observées.

En évaluant les résultats de cette étude, il convient de se souvenir que la méthode globale adoptée tend à dissimuler l’effet des fluctuations de l’activité économique dans certains pays développés sur les économies de pays en voie de développement déterminés. Compte tenu de cette restriction, ces résultats peuvent se résumer comme suit : 1) Les fluctuations de l’activité économique dans l’ensemble des pays développés pendant la période de l’après-guerre ont été limitées et non synchronisées. 2) Les recettes d’exportation des pays en voie de développement n’ont pas été atteintes plus gravement par les fluctuations à court terme de l’activité économique dans les pays développés que les recettes d’exportation des pays développés eux-mêmes. 3) Avec un décalage d’environ un an, les fluctuations de l’activité économique dans les pays développés semblent exercer un effet (d’une ampleur indéterminée) sur les mouvements nets de capitaux des pays développés vers les pays en voie de développement. 4) Les importations des pays en voie de développement semblent s’adapter aux variations des recettes en devises provenant des exportations et des entrées de capitaux, en partie immédiatement et en partie avec un certain décalage. 5) Les répercussions des fluctuations à court terme de l’activité économique dans les pays développés sur leurs propres exportations vers les pays en voie de développement n’ont pas été immédiates et se sont réparties sur plusieurs années, de sorte que les fléchissements de l’activité économique dans les pays développés n’ont pas été renforcés par des fléchissements simultanés de ces exportations; le mécanisme de réaction des pays en voie de développement a généralement joué un rôle stabilisateur dans l’économie mondiale.

Propagación de las fluctuaciones económicas de los países desarrollados en los países en desarrollo

Resumen

Este estudio procura evaluar los efectos que las fluctuaciones económicas a corto plazo ocurridas en los países desarrollados, considerados en conjunto, producen en la balanza de pagos de los países en desarrollo, tomados éstos también en conjunto. Analiza dos vías de propagación de las fluctuaciones económicas: la de la influencia que ejercen las alteraciones en la demanda de importación de los países desarrollados sobre los ingresos de exportación de los países en desarrollo, y la de los efectos que surten sobre estos últimos países las variaciones inducidas en la corriente del capital privado que reciben. Valiéndose de los elementos del análisis, el estudio presenta un pequeño modelo que permite calcular, sobre la base de conocimientos acerca de las fluctuaciones económicas de los países desarrollados, las variaciones que experimentan las exportaciones de estos últimos a los países en desarrollo; los resultados de estos cómputos pueden ser comparados con los cambios efectivamente observados.

Al evaluar las conclusiones de este estudio debe tenerse presente que el enfoque global adoptado tiende a ocultar el impacto que las fluctuaciones económicas de ciertos países desarrollados surten sobre las economías de determinados países en desarrollo. Hecha esta aclaración, cabe sintetizar las conclusiones del estudio de la siguiente manera: (1) Las fluctuaciones económicas ocurridas durante el período posbélico en los países desarrollados, considerados en conjunto, han sido leves y asincrónicas. (2) Los ingresos de exportación de los países en desarrollo no han resultado alterados en grado mayor que los de los países desarrollados por las fluctuaciones económicas a corto plazo de estos últimos. (3) Al cabo de algo así como un año, las fluctuaciones económicas de los países desarrollados parece que causan repercusiones (de magnitud no determinada) en la corriente de capital neta que fluye de los países desarrollados hacia los países en desarrollo. (4) Las importaciones de los países en desarrollo parecen adaptarse—en parte simultáneamente y en parte con cierta diferencia de tiempo—a las variaciones que experimentan sus ingresos en divisas provenientes de sus exportaciones y de la corriente de capital que afluye a ellos. (5) Las repercusiones de las fluctuaciones económicas a corto plazo de los países desarrollados sobre las propias exportaciones de éstos a los países en desarrollo, no se han producido sino después de transcurrido cierto tiempo y en forma difusa a lo largo de varios años; de ese modo, vale decir, al no ser concomitante con los períodos de desanimación económica de los países desarrollados, la declinación de estas exportaciones no ha sido un factor que agudizara dichos recesos; el mecanismo de reacción en los países en desarrollo en general ha desempeñado una función estabilizadora en la economía mundial.

*

Mr. Rhomberg, Chief of the Special Studies Division of the International Monetary Fund, is a graduate of the University of Vienna and of Yale University and has been a member of the faculty of the University of Connecticut. He has contributed chapters to several books on economic subjects and articles to economic journals.

1

See, for example, A. Lamfalussy, “International Trade and Trade Cycles, 1950–60,” in International Trade Theory in a Developing World, ed. by Roy Harrod and Douglas Hague (London, 1963), pp. 241–76.

2

In this study, the term “developed countries” will refer to member countries of the Organization for Economic Cooperation and Development (OECD) plus Australia, New Zealand, and South Africa.

3

See, for instance, A. Lamfalussy, op. cit., and J. J. Polak and R. R. Rhomberg, “Economic Instability in an International Setting,” American Economic Review, May 1962 (reprinted in Readings in Business Cycles, ed. by R.A. Gordon and L. R. Klein, Homewood, Illinois, 1965).

4

See Marcus Fleming, Rudolf Rhomberg, and Lorette Boissonneault, “Export Norms and Their Role in Compensatory Financing,” Staff Papers, Vol. X (1963), p. 144, and Alasdair I. MacBean, Export Instability and Economic Development (Cambridge, Massachusetts, 1966), pp. 93–95.

5

A general description of this model is given in Polak and Rhomberg, op. cit., and a more detailed account is found in Rudolf R. Rhomberg and Lorette Boissonneault, “Effects of Income and Price Changes on the U.S. Balance of Payments,” Staff Papers, Vol. XI (1964), pp. 59–124. See also Grant B. Taplin, “Models of World Trade,” Staff Papers, Vol. XIV (1967), pp. 433–55.

6

This residual region includes not only developing countries but also Australia, Canada, Japan, New Zealand, South Africa, the Soviet countries, and Mainland China.

7

R.R. Rhomberg and Lorette Boissonneault, “The Foreign Sector,” in The Brookings Quarterly Econometric Model of the United States, ed. by J. S. Duesenberry and others (Chicago and Amsterdam, 1965); R.J. Ball and K. Marwah, “The U.S. Demand for Imports, 1948–1958,” Review of Economics and Statistics, November 1962, pp. 395–401.

8

These equations are of the following form:

ΔlogXt=a+bΔlogAt,

where Xt refers, alternatively, to the value of exports, the volume of exports, or an index of export unit values, and At stands for the index of industrial production in developed countries. In addition, similar equations in terms of logarithms and including a linear trend term were computed. These are of the following form:

logXt=a+blogAt+ct,

where t (= 1, 2, 3, …) is a trend variable. The two sets of computations are compared in Appendix II.

9

The equation, cast in first differences of logarithms, is as follows:

ΔlogVldt=0.0016+(0.707)(2.0)ΔlogAtR¯2=0.17d=2.18

where Vld is the value of exports from developing to developed countries, and A is the index of industrial production of developed countries. R2 is the coefficient of determination adjusted for degrees of freedom; d is the Durbin-Watson statistic for serial correlation; and the number in parentheses below the regression coefficient is the t ratio.

10

To repeat, this does not contradict the view that developing countries’ export proceeds may be more unstable than those of developed countries, particularly since the former may be affected to a greater extent than the latter by instability resulting from supply shifts, or from demand shifts unrelated to changes in aggregate economic activity.

11

If current profits are synchronous with the cycle in economic activity, the peaks and troughs of accumulated retained earnings, i.e., the funds available for investment at home or abroad, will tend to lag behind the peaks and troughs of the cycle.

12

The best results were obtained with capital movements (five-quarter moving averages) lagged four and five quarters behind industrial production (seasonally adjusted). For instance, the equations for U.S. direct investment in less developed countries (D) are as follows (third quarter 1948 to third quarter 1965):

Dt=330+6.5(3.81)At44.7(2.95)t,R¯2=0.235Dt=339+6.7(3.85)At54.8(2.98)t,R¯2=0.237

(Residuals are serially correlated in both equations.) The t ratios (in parentheses below the coefficients) and R2 are highest for the equation with a five-quarter lag. Inferior results were obtained with shorter and longer lags (tested up to seven quarters). The implied elasticity (at the means of the variables) of U.S. direct investment in less developed countries with respect to U.S. industrial production of the preceding year is approximately + 5.

13

OECD, The Flow of Financial Resources to Less-Developed Countries, 1956–1963(Paris, 1964). The data (in billions of U.S. dollars) for the years 1956–63, respectively, are as follows: 2.58, 3.23, 2.72, 2.44, 2.58, 2.59, 1.99, 1.87. These figures might be consistent with a distributed lag response, in which part of the reduction in capital flows to the developing countries occurs in the year of the recession and part in the year following the recession.

14

It was possible to determine from data for part of the period that the trade deficit of developing countries with the Soviet countries and Mainland China was not subject to cyclical variations, although it appears to show an upward trend.

15

In the monetary model of imports and income designed by J. J. Polak chiefly for application to developing countries, income and imports are related to current and past values of export receipts plus capital inflows plus the change in domestic credit creation. Application of this model has shown that the influence of lagged values of this causative variable diminishes rapidly as longer lags (beyond 1 or 2 years) are considered, and that a large part of the influence is accounted for by the current value and that of the preceding year. See J. J. Polak and Lorette Boissonneault, “Monetary Analysis of Income and Imports and Its Statistical Application,” Staff Papers, Vol. VII (1960), particularly pp. 358–59.

16

As a result of the constant term, exact balance is not restored in Examples A and B. In Example A, the foreign balance (Δ Rl, which is the cumulation from year 1 to the year in question of ΔΔ R1) would continuously improve as long as gross foreign exchange receipts remained constant, whereas in Example B it would steadily worsen as long as gross receipts continued to grow by a constant absolute amount.

17

This linear equation is substituted for the logarithmic one shown in footnote 9. The equation is estimated over the period 1951–65; the implied elasticity of the value of developing countries’ exports with respect to industrial production in developed countries at the means of the variables is 0.7.