THE MOVEMENT of the terms of trade of a country or an area is measured by means of the ratio of two index numbers: an index number of the prices received for exports and an index number of the prices paid for imports. Variations in the terms of trade may be said to measure changes in the quantity of imports which can be obtained in exchange for a given quantity of exports. When export prices have risen more or fallen less than import prices and when, accordingly, the quantity of imports that can be obtained for the same quantity of exports is greater than in a given base period, the terms of trade of a country or an area are said to be more favorable than in the base period. The converse is true when export prices have risen less or fallen more than import prices. It should be borne in mind that the term “favorable” (and, conversely, “unfavorable”) has significance only if used in conjunction with a specified base period; this point is of particular importance because of the sharp difference in prices between the two prewar years, 1937 and 1938, which are most commonly used as base years.

Abstract

THE MOVEMENT of the terms of trade of a country or an area is measured by means of the ratio of two index numbers: an index number of the prices received for exports and an index number of the prices paid for imports. Variations in the terms of trade may be said to measure changes in the quantity of imports which can be obtained in exchange for a given quantity of exports. When export prices have risen more or fallen less than import prices and when, accordingly, the quantity of imports that can be obtained for the same quantity of exports is greater than in a given base period, the terms of trade of a country or an area are said to be more favorable than in the base period. The converse is true when export prices have risen less or fallen more than import prices. It should be borne in mind that the term “favorable” (and, conversely, “unfavorable”) has significance only if used in conjunction with a specified base period; this point is of particular importance because of the sharp difference in prices between the two prewar years, 1937 and 1938, which are most commonly used as base years.

THE MOVEMENT of the terms of trade of a country or an area is measured by means of the ratio of two index numbers: an index number of the prices received for exports and an index number of the prices paid for imports. Variations in the terms of trade may be said to measure changes in the quantity of imports which can be obtained in exchange for a given quantity of exports. When export prices have risen more or fallen less than import prices and when, accordingly, the quantity of imports that can be obtained for the same quantity of exports is greater than in a given base period, the terms of trade of a country or an area are said to be more favorable than in the base period. The converse is true when export prices have risen less or fallen more than import prices. It should be borne in mind that the term “favorable” (and, conversely, “unfavorable”) has significance only if used in conjunction with a specified base period; this point is of particular importance because of the sharp difference in prices between the two prewar years, 1937 and 1938, which are most commonly used as base years.

Movements in the terms of trade of a country constitute an important element in changes in a country’s balance of payments position. But they are not the sole factor of importance. The volume of exports which can be produced or for which a market can be found abroad, the availability of foreign loans, the possibility of obtaining imports promptly in the quantities desired at current prices, the flow of foreign exchange from the tourist trade, and other invisible items are also among the many factors that play an important, and often decisive, role in determining the position of a country’s balance of payments. This paper, however, is limited to the terms of trade of Latin American countries and their measurement; no attempt has been made to discuss other factors. The period studied was undoubtedly an exceptional one. For the reasons indicated below, it would in any event be a mistake to attach too much importance to the apparently precise figures which emerge from the analysis that has been made. The margin of error is necessarily great; the study may, however, be accepted as on the one hand indicating the widely divergent experiences of the Latin American countries and on the other hand illustrating the effectiveness of the techniques available for studying the terms of trade.

Statistical Problems

The paucity of data on the terms of trade for most countries in the world is due primarily to the statistical and conceptual difficulties involved in establishing adequate statistics. These difficulties are of extraordinary complexity. A rather detailed treatment of the problems has been presented in “Relative Prices of Exports and Imports of Underdeveloped Countries: a Study of Postwar Terms of Trade between Underdeveloped and Industrialized Countries,”1 prepared by the Secretariat of the United Nations for the Subcommission on Economic Development; and for most of the problems, reference to this study may be sufficient.

The question of weighting, however, requires further attention. In the computation of an index of prices for a series of years, e.g., 1938, 1946, 1947, and 1948, there are two alternative methods of weighting. According to one method, weights are chosen from the base year (for instance, 1938), in which case the index is called a “Laspeyre index,” i.e., “an index with base year weights.” According to the other, a different set of weights is used for each year, in which case the index is called a “Paasche index,” i.e., an “index with current year weights.” The two types of index measure two different things. If 1938 and, say, 1946, are the years under consideration, a Laspeyre index of import prices would compare the cost of the imports of 1938 with what they would have cost if purchased at 1946 prices, while a Paasche index would compare the cost of the imports of 1946 with what they would have cost if purchased at 1938 prices. The two indices normally yield somewhat different results, and a preference between them can be established only by reference to the use to be made of the indices. In normal times, and especially if two successive years are compared, the difference between the two indices is usually very small; but in many comparisons over a longer period, especially between prewar and postwar years, when great changes have occurred both in the composition of trade and in the relative prices of many commodities, the difference turns out to be quite large. For this reason both types of index have been computed in this study, whenever the statistical material permits.

Methodology

An analysis is first made of changes between 1938 and 1946 in the terms of trade of a number of Latin American countries. This information is then generalized for Latin America as a whole, is compared with other information on the prices of exports and imports of Latin America, and is extended both backward (to 1925) and forward (to 1948).

The year 1938 has been selected as the base year, since it was the last complete prewar year. Moreover, it is the base year used in other studies on the same subject, and comparisons are therefore possible. It was, however, a year of unusually low raw-material prices, and cannot be considered as a “normal” prewar year. The year 1946 is the latest for which a reasonably complete set of statistics was available at the time the study was undertaken (1948). It is in many respects unfortunate that 1946 data, rather than 1947 data, had to be used, not only because the 1946 data refer to a situation more than two years past, but also because the movement of commodities in international trade and of the corresponding prices then reflected, in many cases, unusual wartime shortages and stopgap measures to overcome such shortages which have since disappeared, or at least have become of much less significance.

Fairly complete data are available for calculations of the terms of trade for eight countries: Argentina, Brazil, Chile, Colombia, Cuba, Guatemala, Mexico, and Peru.2 The use of price indices for individual export commodities and for groups of import commodities obtained from the analysis of these eight countries, together with appropriate weights, made possible estimates of the terms of trade for five additional countries: Bolivia, Costa Rica, Ecuador, Nicaragua, and Venezuela. Because statistics were not available in sufficient detail, estimates of the terms of trade for the other Latin American republics could not be made.

In most cases, export and import price indices were computed on the basis of unit values in terms of dollars derived from official trade statistics. Unit values are the quotients of reported values and reported quantities. Fluctuations in unit values may differ substantially from actual price fluctuations in exports and imports for a number of reasons: because of the classification of the original trade statistics in categories which often make it impossible to take account of changes in the relative importance of the groups covered by each category; because of inaccurate declarations of value and quantity by importers or exporters; because of statutory valuations (irrespective of market values) prescribed by the government for the purpose of trade statistics; or because of the inappropriate treatment of values in countries where there are multiple currencies. In the calculations which underlie the present study, only such data have been used as are believed to be sufficiently specific and accurate to permit significant conclusions, and the resulting indications of price movements have been checked against other relevant information.

Even if data on the terms of trade are computed with the greatest care, however, the figures will still, in many cases, be subject to a margin of error of 10 or perhaps 20 per cent. Thus if the terms of trade are found to be 105 compared with a certain base period, there is some presumption of an improvement but the evidence is definitely weak; similarly a figure of 95 provides only a weak presumption of a deterioration. Figures of the order of 115 or 85, on the other hand, might be accepted with a much higher degree of confidence as indicating improvement or deterioration.

In the trade statistics used, imports are valued c.i.f. and exports f.o.b. Thus shipping and other transportation costs, which have risen more than the cost of imported commodities, are included in the import data but not in the export data. Nevertheless, the statistics in this form provide terms of trade data that are relevant to Latin American countries, since both the imports and exports of these countries are for the most part carried in foreign bottoms.

For each country the commodities were grouped in classes which seem to be meaningful from an economic point of view. Obviously, however, any classification of this nature cannot provide homogeneous groups containing the same goods for different countries. The classes selected are as follows:

article image

The approximate content of most of these groups is self-evident. Among imports, radios and automobiles and their parts, as well as household equipment, are included in group (c). The general classifications could not be followed exactly for Bolivia, Costa Rica, Ecuador, Nicaragua, and Venezuela, but the classifications that were used approximate those indicated above.

Price indices for individual commodities were combined into group indices according to both the Laspeyre and the Paasche methods (to the extent that the necessary information was available), and the group indices were combined, also by both methods, into indices for total exports and total imports.

Country Comparisons

The changes in the terms of trade of the Latin American countries show wide divergencies from country to country. In the period covered, i.e., between 1938 and 1946, the data indicate a substantial improvement for some countries and a significant deterioration for others, while for a third group the changes are not large enough to justify any conclusions. The differences among the various countries can be explained largely in terms of the composition of their exports and imports and the specific markets to which each country is related. It is useful, therefore, to consider first the movements in export prices and import prices separately, and then the net result of the two, that is, the terms of trade.

Export prices

Between 1938 and 1946 the export prices of foodstuffs and of “other” raw materials rose, on the average, more than the export prices of minerals and fuels (Table 1). Very large increases of the food price indices occurred in Peru, Chile, Ecuador, and Brazil. In Peru, the rise was due almost exclusively to a rise in sugar prices, which in some markets other than the United States rose by more than 600 per cent. In Chile, it resulted from a sharp increase in bean prices, whose index rose to 881 in 1946 (the second largest increase for foodstuff prices found in this study). In Ecuador, it was due to the increase in world cocoa prices and in Ecuador’s export price for coffee. This rise in coffee prices, which exceeded that of any other coffee exporting country, can be explained in terms of shifts in markets and apparently because the price of low quality coffee rose faster than that of high quality coffee. Before the war, Ecuadoran coffee, which is of low quality, was sold in Spain, a low-price market, but in 1946 the level of world demand made it possible for Ecuador to export its coffee to high-price markets. This shift did not affect other coffee exporters to the same extent as those in Ecuador.

Table 1.

Export Price Indices, 1946, and Ratios of Commodity Groups to Value of Total Exports, 1938, for Selected Latin American Countries

article image

Based on data in terms of dollars; 1938 weights.

In the computation of these data, minor items have been disregarded in most cases. A dash in column indicates that the value of the commodity group was less than 5 per cent of total exports in 1938. Dots indicate that data are not available.

Export price indices (1938 = 100) for manufactures and for “other exports” in 1946 were 341 and 274, respectively. These two groups accounted for 4 per cent of total exports in 1938.

For the group, “Other Raw Materials,” which includes such agricultural materials as cotton, hides, wool, timber, and oilseeds, the indices of all countries except Chile showed substantial increases. The relatively low figure for Chile is a consequence of the heavy weight of wool exports, whose price index rose to only 131 (1938=100) in 1946.

There is close agreement among the indices of prices of minerals, the low figure for Nicaragua being due to the large weight of gold exports of that country.

Import prices

In general, import prices in those countries whose imports were largely foodstuffs and textiles rose more than the prices in countries where these imports were relatively small (Table 2). In some countries, however, such as Chile and Peru, where the proportion of food and textiles imported wag not particularly large, the prices of these commodities rose so substantially that the general import price indices of these countries were also high.

Table 2.

Import Price Indices, 1946, and Ratios of Commodity Groups to Value of Total Imports, 1938, for Selected Latin American Countries

article image

Based on data in terms of dollars; 1938 weights.

In the computation of these data, minor items have been disregarded in most cases. In this table and Table 3, dashes indicate that data are not available.

Includes durable consumers’ goods.

There were considerable differences among countries with respect to changes for the same food products. For instance, the price index of wheat in 1946 rose to 337 (1938=100) in Brazil, to 331 in Peru, and to 167 in Colombia. Generally the countries which could import wheat from the United States paid lower prices than those which imported it from Latin American countries.

The textile price indices for 1946 show a rather close agreement for all countries except Brazil. Brazilian textile imports are mainly clothing, and very little cotton cloth is imported. Furthermore, Brazilian statistics are not available in sufficient detail to yield a reliable index. For most of the other textile importing countries, Japan was an important supplier in 1938, but this source had been replaced by certain Latin American countries in 1946. Mexico and Cuba, on the other hand, imported most of their textiles from the United States both before and after the war; this explains the relatively small increase in the price indices for Mexican and Cuban textile imports.

The machinery and equipment group contains many commodities which are not classified in sufficient detail in the trade statistics to assure the significance of the price comparisons obtained. Nevertheless, for each country except Chile the group index is considered reliable. A check of the results against indices of U. S. export prices of machinery and equipment suggests that the rise indicated by the price index for Chilean imports of machinery may be underestimated.

The relatively low indices for import prices of raw materials and semimanufactures in Brazil, Chile, Guatemala, and Mexico are in large part due to imports of fuel and steel products, while the relatively high figures for some other countries are explainable in terms of imports of textile raw materials and of timber.

Terms of trade

The most pronounced improvement in the terms of trade (Table 3) was in Brazil, a country with large exports of food and nonmineral raw materials, and with relatively small imports of textiles and foodstuffs. On the other hand, Bolivia, Chile, Peru, and Venezuela, which have large exports of fuels and minerals and large imports of foodstuffs and textiles, show a tendency toward a deterioration in their terms of trade. For Cuba there was no significant change, although exports of food were high in relation to imports of food and textiles. Sugar exports were sold in a controlled market while food imports were bought in noncontrolled markets. Furthermore, Cuban food imports are heavily weighted by such items as rice, whose price rose substantially in relation to other food prices.

Table 3.

Indices of Terms of Trade of Selected Latin American Countries, 1946

(1938 = 100; data in terms of dollars)

article image

Index with base year weights.

Index with 1946 weights.

Weighted average of country indices.

As shown in Table 3, there are only minor differences between the indices for export prices calculated according to the Laspeyre and the Paasche methods. For import prices, on the other hand, the Paasche index is systematically lower than the Laspeyre index, by as much as about 20 per cent for Brazil, Colombia, and Peru. The terms of trade measured by the Paasche indices are therefore more favorable than those based on the Laspeyre indices. But while the Laspeyre formula may tend to give a downward bias to the results of this study, this is offset somewhat by the upward bias resulting from the use of 1938 as the base year.

Latin America as a Whole

An estimate of the terms of trade for Latin America as a whole is of interest, in view of the divergent results obtained for individual countries. One estimate has been obtained by weighting the import and export price indices computed for individual countries by the values of their 1938 imports and exports. In the absence of Paasche indices for most countries, the index for Latin America as a whole has been computed on the basis of Laspeyre indices only.3

The result of this calculation is an export price index for Latin America as a whole, in terms of dollars, of 219 (1938 = 100) for 1946, and an import price index of 202. The resulting terms of trade in 1946 on the 1938 basis would therefore be 108.

Both the import and export figures include intra-Latin American trade. It may be assumed that price changes between 1938 and 1946 in such trade have similar effects on both the import and the export price indices for Latin America as a whole. Elimination of this equal element from the two indices would yield a figure of 110 (1938 = 100) for the terms of trade of Latin America with the outside world. In this connection, it is worth mentioning that prices in intra-Latin American trade appear to have increased more than prices in the trade of Latin American countries with the outside world. Countries such as Mexico and Venezuela, which drew only a very small percentage of their imports from other Latin American countries in 1938,4 show a very low import price index (Mexico 169, Venezuela 168), whereas Bolivia, the country with the highest percentage of imports from Latin America,5 showed the highest import price index (237).

A second, somewhat rougher, computation of the terms of trade for Latin America as a whole, was made by the following method.

Export prices

The bulk of Latin American exports to the outside world is made up of a relatively small number of commodities. For these commodities the relevant changes in world market prices between 1938 and 1946 may be estimated on the basis of U. S. wholesale price quotations (Table 4).6 On this basis the average price increase is 100 per cent, against the average increase of 119 per cent in the export price index derived above. The difference appears to be due mainly to exports by Latin American countries to each other and to countries other than the United States at prices far above U. S. wholesale prices, especially in the first half of 1946, when ceiling prices still applied in the United States. Thus the Cuban export price index (1938 = 100) of sugar in 1946 averaged 42 per cent, the Brazilian coffee export price 17 per cent, and the Chilean nitrate export price 8 per cent higher than the corresponding U. S. wholesale prices used. The discrepancies were probably less important in 1947 and 1948.

Table 4.

Price Indices of Principal Latin American Export Commodities, 1946, and Ratios of Commodity Values to Total Value of Latin American Exports, 1938

article image

Based on data in terms of dollars.

For these commodities, which are the main exports of Argentina, the U. S. price in the last quarter of 1946, after the lifting of controls, was used rather than the average for the year. Even these prices are probably lower than those obtained by Argentina in 1946.

Import prices

Since Latin American imports from the outside world are mainly manufactured products from the United States and the United Kingdom, an indicator of the changes in prices of these imports can be obtained by averaging, with appropriate weights, the index numbers of the unit values of exports of manufactured products from these two countries. The proper weights would be approximately 3 for the United States and 1 for the United Kingdom for recent years, and relatively larger weights for the United Kingdom in earlier years. This calculation indicates a 53 per cent increase between 1938 and 1948 in the import price index for Latin America, which is considerably lower than the 102 per cent increase indicated by the weighted average import price index calculated above.

The discrepancy may be attributed to several factors. The index based on U. S. and U. K. export prices does not take into account the trade among Latin American countries at much higher relative prices. Furthermore, the import unit values used for Latin American countries necessarily take account of the increase in cost due to the compulsory substitution of more expensive U. K. or U. S. commodities for commodities that had been imported from Germany or Japan before the war. Finally, a small part of the discrepancy may be due to the fact that the export price indices for the United States and the United Kingdom are based on f.o.b. prices, and that freight rates, which are included in the c.i.f. prices for imports used above, have increased somewhat more than the prices of imported commodities.

The ratio between the two indices calculated in this section gives an estimate of 130 for the terms of trade. This figure is too high (the direct estimate was 110), primarily because it is based on an import price index which apparently underestimates seriously the increase in Latin American import prices.

Extrapolation of Results

Whereas it would be very laborious and perhaps impossible to extend the calculations by countries to other years, it is comparatively easy to make an extrapolation, both backward and forward, of the terms of trade for Latin America as a whole calculated according to the second method developed in the preceding section. The results of this extrapolation for a number of significant years are shown in Table 5.

Table 5.

Terms op Trade for Latin America

(1938 = 100)

article image

Based on U. S. and U. K. export prices of manufactures.

Index a raised by ratio of 202 (the 1946 index of import prices computed from Latin American statistics) to 153 (the 1946 index of import prices computed on the basis of U. S. and U. K. export prices of manufactures).

Figure found by direct measurement.

With respect to the export price index it may be assumed that, although the figure for 1946 in Table 4 tends to be too low, similarly derived figures for 1947 and 1948, when U. S. prices were no longer controlled, are less likely to be unreliable. Likewise, the movement for the years before 1938, as derived from U. S. wholesale prices, would seem to be reasonable.

For import prices, the rise between 1938 and 1946, when measured by changes in U. S. and U. K. export prices, is undoubtedly underestimated. On the other hand, the figures for 1947 and 1948, obtained by extrapolating the 1946 figure, no doubt overestimate the rise since 1938; and, in particular, these figures used in conjunction with an export price index which includes only some but not all of the high prices charged in inter-American trade, overestimate changes in the terms of trade.

The terms of trade in 1947 may therefore be estimated to have been somewhere between 153 and 116 (on the 1938 base) and probably nearer the lower of those two figures, and hence somewhat higher than in 1946. For two countries—Argentina and Brazil—for which fairly adequate data are available, the terms of trade in 1947 were better than in 1946, by about 17 per cent for Argentina and 11 per cent for Brazil.7 Estimates for 1948 indicate little change from 1947.

According to Table 5, the terms of trade of Latin America were more unfavorable in 1938 than in any other year in the period covered. The terms of trade for other individual years are thus more favorable when computed on a 1938 base than when computed on any other base. The terms of trade in 1946, which were 10 per cent above the 1938 level, were 14 per cent below the level of 1937.

1

E/CN.l/sub.3/W5.

2

Detailed calculations have been made for these countries, but they are not included in this paper.

Argentine import and export values in 1938 were based on arbitrary official valuations which did not correspond to market values. It has therefore been impossible to calculate satisfactory group price indices though the recorded dollar values of imports and exports have made possible the calculation of general price indices.

3

The countries covered accounted for 90 per cent of Latin American trade in 1938.

4

One per cent of Venezuela’s imports, and less than one per cent of Mexico’s imports, in 1938 were from other Latin American countries.

5

23 per cent in 1938.

6

With the exception of copper for which the average U. S. import price excluding import duty was used, it would have been slightly preferable to use the net, rather than the gross, exports of Latin America, i.e., after deduction of exports from one Latin American country to another; the figures of net exports were, however, more difficult to obtain and their use would not significantly affect the weighting coefficients.

7

The relevant data are as follows (1946 = 100):

article image