Introduction

Do Labour Standards Affect Comparative Advantage?

The role of labour standards in fostering international trade has been in the forefront of the trade policy agenda in recent years. It has attracted the interest of humanitarian organisations, governments, and international organisations as production of labour-intensive goods has continued to move to developing countries with increased globalisation. 

In a 2002 article, Matthias Busse analysed econometrically whether different categories of labour standards do, in fact, have an effect on comparative advantage in developing countries.

• He addressed the question of whether a developing country can derive comparative advantage in unskilled-labour-intensive goods by employing low labour standards and thus increase its world exports. 

• A number of different groups believe that this is happening and thus are urging the implementation of import barriers against countries with notably lower standards for both humanitarian reasons and to ensure a more “level playing field”. 

• Busse distinguished between “core” labour standards (important human rights such as union rights, freedom from forced labour, abolition of child labour, equal opportunity) and labour standards often referred to as “acceptable conditions of work” (e.g., minimum wages, safety and health standards). 

• He focused on the effect of core standards on comparative advantage and exports utilising the Heckscher-Ohlin theoretical framework developed in this chapter to provide the underpinnings for his empirical work. 

• Because comparative advantage is determined by relative factor endowments, he hypothesised that lower core labour standards would lead to a relative increase in unskilled labor and thus increase relative exports of unskilled-labor-intensive goods. 

• Five indicators of core labor standards:

  • discrimination against women

  • presence of child labour

  • use of forced labour

  • basic union rights

  • number of ratifications of the eight International Labour Office (ILO) conventions of core labour standards

• These were key variables used to explain the share of unskilled-labour-intensive goods in total exports in an 83-country cross-section regression analysis. 

• Several of the more interesting conclusions were that:

  • Greater discrimination against women weakened comparative advantage.

  • Whereas weaker union rights, greater child labour, and greater use of forced labour increased export share.

  • Interestingly, the number of ratifications of the ILO conventions appeared to have no significant effect.

• It is, however, important to note that in all cases educational attainment and the overall relative labour endowment had relatively stronger influences on the trade patterns than did the labour standard variables.

Interesting and useful policy analysis, such as that contained in the Busse paper, necessitates the use of a more formal structure in which the complexities of international comparative advantage can be sorted out in a consistent way.

In previous chapters, we demonstrated that a country will gain from trade anytime that the terms of trade differ from its own relative prices in autarky. The country gains by expanding production of and exporting the commodity that is relatively more valuable in the foreign market and reducing production of and importing the good that is relatively less expensive in the foreign market. These adjustments permit the consumption of a combination of goods that lies outside the production-possibilities frontier at a higher level of consumer well-being.

It was further demonstrated that the underlying basis for the relative price differences that led to international trade could be traced to differences in supply and/or demand conditions in the two countries. This analysis assumes that there is no intervention in the markets to alter prices from these general equilibrium results. Clearly, taxes and subsidies can cause autarky prices to be more or less different prior to trade. 

• This chapter will examine in greater detail the factors that influence relative prices prior to international trade, focusing on differences in supply conditions. 

• We first examine how different relative quantities of the factors of production can influence product prices and produce a basis for trade. 

• We then discuss how the resulting trade will in turn affect factor prices and the distribution of income within the trading countries.

• Finally, the implications of various assumptions employed in the analysis are presented. The purpose of this chapter is to provide a deeper understanding of the critical factors underlying relative cost differences and therefore comparative advantage. 

Factor Endowments and the Heckscher-Ohlin Theorem

The effects of factor endowments on international trade were analysed early in the twentieth century by two Swedish economists, Eli Heckscher (in 1919) and Bertil Ohlin (in 1933).

As employed by modern economists, this analysis makes a number of simplifying assumptions, specifically that: 

1. There are two countries, two homogeneous goods, and two homogeneous factors of production whose initial levels are fixed and assumed to be relatively different for each country. 

2. Technology is identical in both countries; that is, production functions are the same in both countries. 

3. Production is characterised by constant returns to scale for both commodities in both countries. 

4. The two commodities have different relative factor intensities, and the respective commodity factor intensities are the same for all factor price ratios. 

5. Tastes and preferences are the same in both countries. Further, for any given set of product prices, the two products are consumed in the same relative quantities at all levels of income; that is, there are homothetic tastes and preferences.

6. Perfect competition exists in both countries. 

7. Factors are perfectly mobile within each country and not mobile between countries.

8. There are no transportation costs.

9. There are no policies restricting the movement of goods between countries or interfering with the market determination of prices and output. 

Most of these assumptions are common to the models examined in previous chapters. However, the two that are especially critical to the Heckscher-Ohlin (H-O) explanation of the emergence and structure of trade are (assumption 1) that factor endowments are different in each country and (assumption 4) that commodities are always intensive in a given factor regardless of relative factor prices. These two assumptions need to be examined in greater detail prior to discussing the H-O theorem. 

Factor Abundance and Heckscher-Ohlin

It is important to understand that the phrase different factor endowments refers to different relative factor endowments, not different absolute amounts. 

Crucial to the H-O analysis is that factor proportions are different between the two countries. 

Relative factor abundance may be defined in two ways: the physical definition and the price definition:

1. The physical definition explains factor abundance in terms of the physical units of two factors, for example, labour and capital, available in each of the two countries. 

• Country I would be the capital-abundant country if its ratio of capital to labour exceeded the ratio of capital to labour in country II:

• It must be stressed that the relative amount of the factors is critical, not country size. 

• A country with fewer absolute units of physical capital than a larger country could still be the capital-abundant country as long as the amount of capital relative to labor was greater than the same ratio in the larger country.

• Finally, in the two-country, two-factor case, if country I is the capital-abundant country, then country II must by definition be the labor-abundant country. 

2. The price definition relies on the relative prices of capital and labor to reveal the type of factor abundance characterising the two countries. 

• According to this definition, country I would be the capital-abundant country as long as:

• That is, the ratio of the price or rental rate of capital (r) to the price or wage of labor (w) in country I is less than in country II. This definition views relative abundance in terms of the factors’ relative scarcity prices. 

• The greater the relative abundance of a factor, the lower its relative price. 

One definition focuses on physical availability (supply), and the other focuses on factor price. 

What is the link, if any, between the two? On the surface, there does not seem to be a problem because the greater or smaller the supply of a factor, the lower or higher its price tends to be. 

• In countries with large populations such as India and China, the price of labour is relatively low while that of capital is relatively high. 

• The converse tends to be true in a country such as Germany or the United States. 

• However, the problem is that the factor price reflects not only the supply of available factors but also the demand. 

• Because factors of production are not consumed directly but are used to produce final goods and services that are consumed, demand for the factors results from the structure of demand for final goods and services. 

• Demand for a factor of production is thus often referred to as a derived demand resulting from producers meeting final consumption preferences.

• Factor prices reflect not only the physical availability of the factors in question but also the structure of final demand and the production technology employed. 

Fortunately, because the H-O model assumes that technology and tastes and preferences are the same in both countries, the two definitions will produce the same result. 

• With technology and demand influences neutralised between the two countries, the country with the relatively larger K/L ratio also will have the relatively smaller r/w ratio. The link between the two definitions is unambiguous unless technology or demands differ between the two countries. When this happens, the price definition may differ from the physical definition; for example, physically abundant capital may be relatively high priced. We refer to this possibility later in the chapter. 

Commodity Factor Intensity and Heckscher-Ohlin

A commodity is said to be factor-x-intensive whenever the ratio of factor x to a second factor y is larger when compared with a similar ratio of factor usage of a second commodity. 

• For example, steel is said to be capital intensive compared with cloth if the K/L ratio in steel production is larger than the K/L ratio in cloth production.

• H-O assumes not only that the two commodities have different factor intensities at common factor prices but also that the difference holds for all possible factor price ratios in both countries. This means that at all possible factor prices, the isoquants reflecting the technology used in steel production are more oriented toward the capital axis, compared with the isoquants reflecting cloth production, so that the capital/labor ratio for steel will always be larger than that for cloth (see Figure 1).

• It is important to note that this assumption does not preclude substituting labor for capital if capital becomes relatively more expensive, or substituting capital for labor if the relative price of labor rises. While such price changes would indeed change the capital/labor ratios in both commodities, they would never cause cloth to use more capital relative to labor compared with steel. This is a strong assumption and it is critical to the H-O analysis. We will later examine some possible conditions when it would not hold and the resulting implications for international trade.

Relative Factor Endowments in Selected Countries

Relative factor endowments differ considerably across countries. 

• Table 1 gives three factor ratios for a number of selected countries to provide some indication of the degree of difference in endowments in 1992.

• Table 2 then provides more recent estimates for a smaller number of countries for 2010. The wide variety of relative factor endowments supports the idea that underlying factor supply conditions continue to vary from country to country, as Heckscher and Ohlin posited many years ago.

The Heckscher-Ohlin Theorem

The set of assumptions about production leads to the conclusion that the production-possibilities frontier (PPF) will differ between two countries solely as a result of their differing factor endowments. 

With identical technology in both countries, constant returns to scale, and a given factor-intensity relationship between final products, the country with abundant capital will be able to produce relatively more of the capital-intensive good, while the country with abundant labor will be able to produce relatively more of the labor-intensive good. 

The shape and position of the PPF is thus determined by the factor intensities of the two goods and the amount of each factor available. This is obvious if one compares the Edgeworth boxes for two countries with different factor endowments (see Figure 2). 

The two boxes show that country I is the capital-abundant country. This is evident since the height of the box (amount of capital) is greater for country I, whereas the length of the box (physical amount of labor) is greater for country II.

In more general terms, the slope of the diagonal reflects the K/L ratio and therefore the relative endowment of the country. 

• This slope is greater in country I, making it clearly the capital-abundant country.

• The analysis in Chapter 5, which discussed how the PPF was obtained from the Edgeworth box, leads us to conclude that the PPF for each country will be different.

• Country I’s PPF is oriented more toward steel, and country II’s PPF is oriented more toward cloth. If these two differently shaped PPFs are now combined with the same set of tastes and preferences, two different sets of relative prices will emerge in autarky, as shown in Figure 3(a). 

• The relative price of steel will be lower in country I (the capital-abundant country) as reflected in a steeper autarky price line, while the relative price of cloth will be lower in country II (the labor-abundant country) as evidenced by a flatter autarky price line.

• Because relative prices in autarky are different between the two countries, a clear basis for trade results from the different factor endowments.

• The trade implications of this situation can be seen in Figure 3(b). The international terms of trade must lie necessarily between the two internal price ratios, being flatter than the autarky price line in country I and steeper than the autarky price line in country II. 

• In this situation, country I will export steel to and import cloth from country II, reaching a higher community indifference curve in the process. 

• Country II also will find itself better off by exporting cloth and importing steel.

• The common equilibrium international terms of trade that produce this result are drawn between the autarky prices of both of the countries.

• A single international terms-of-trade (TOT) line, (PC/PS)int, tangent to both PPFs, is used in panel (b) for convenience, although all that is required is an equally sloped TOT line, not necessarily a common line. 

• For equilibrium to occur, country I’s desired exports of steel (S1S0) and imports of cloth (C2C1) must equal exactly country II’s desired exports of cloth (C1C0) and imports of steel (S2S1) at the prevailing international terms of trade. 

• When this occurs, both countries find themselves on the higher indifference curve, IC1, indicating the mutual gain from trade.

RELATIVE FACTOR INTENSITIES IN CANADA 

Heckscher-Ohlin importantly assumed that relative factor intensities were different across commodities.

To provide an indication of the degree of variation within a given country, we calculated in Table 3 the capital/labor ratios for a selected group of industries, mostly in manufacturing, in Canada in 2006. (The data for constructing this table are not available for years after 2006 as the relevant series were discontinued.) 

As can be observed, there are huge differences in the ratios, going from a high of $2,675,267 per worker to about $9,000 per worker. 

Then, in Table 4, we present Canadian capital/labor ratios for a range of more broadly defined industries in 2014. The figures are the ratio of net capital stock at the end of 2013 (beginning of 2014) to the number of workers employed in 2014. Again, there is a very large degree of variation across industries.

The previous discussion used the physical definition of factor abundance. A similar result would have occurred if we had used the price definition. 

• Because country I is the capital-abundant country, 

With identical technology and constant returns to scale, country I will be able to produce steel relatively more cheaply than country II, and country II can produce cloth relatively more cheaply than country I.

This relationship between relative factor prices and relative product prices can be developed more formally through isoquant-isocost analysis. 

• Consider Figure 4(a).

• Given isocost line MN in country I, whose slope reflects (w/r)I, steel would be produced at point X and cloth at point Y. Thus, the same factor cost in the two industries yields S1 units of steel and C1 units of cloth.

 Isoquants represent all the same output on their slope

• On the other hand, (w/r)II < (w/r)I, so a flatter isocost line is present in country II (M′N′). Given this isocost line, country II’s producers would select points Q and T. Hence, in country II, C2 units have the same cost as S1, while in country I only C1 units (a smaller quantity than C2) could be produced for the same cost as S1. Thus, cloth is relatively cheaper in country II and steel is relatively cheaper in country I

• The conclusion is that a higher w/r leads to a higher relative price of cloth. As cloth is a more labour-intensive industry

The H-O relationship is illustrated in Figure 4(b). 

• Note that if steel had been the relatively labor-intensive good rather than cloth, the relationship would be reflected in a downward-sloping line. It is now clear that different relative factor prices will generate different relative commodity prices in autarky. Consequently, there is a basis for trade, and each country will export the product it can produce less expensively: steel in country I and cloth in country II.

• This same conclusion was reached in the graphical PPF analysis that utilized the physical definition of factor abundance. In both cases, each country expanded production of and exported the good that made the more intensive use of its relatively abundant factor of production.

Titans of International Economics: Paul Anthony Samuelson (1915–2009) 

Paul A. Samuelson was one of the most widely known economists in the United States, due not only to his prodigious research for the last half-century but also for the success of his principles book, Economics, which introduced millions of students to the subject and which has been in print for well over 60 years.

He was born in Gary, Indiana, in 1915 and later attended some 14 different secondary schools, eventually entering the University of Chicago at age 16. Upon graduation in 1935, he studied economics at Harvard University for five years. Samuelson published his first article in 1937 as a 21-year old graduate student and averaged more than five articles a year during his career. 

His 1941 doctoral dissertation is still regarded as the groundbreaking work on the mathematical foundations of theoretical economics. He accepted a position on the economics faculty at the Massachusetts Institute of Technology in 1940 and remained there until his death in 2009. His contributions were in the areas of:

•  microeconomic theory

• consumer theory

• welfare economics

• capital theory

• dynamics

• general equilibrium

• public finance

• macroeconomics

• international trade

He received the National Medal of Science in 1996 from President Bill Clinton, and he advised Presidents Kennedy and Johnson. Samuelson once said, “Our subject puts its best foot forward when it speaks out on international trade,” and his own contributions in the area have had a lasting impact. In the study of trade, he is well known for his seminal work on Heckscher-Ohlin (sometimes referred to as the Heckscher-Ohlin-Samuelson model), focusing on factor price equalisation and the Stolper-Samuelson theorem on the distributional effects of trade (discussed later in this chapter). He was awarded the Nobel Prize in Economics in 1970 and has received all the major honours in the economics profession. His many mathematical and theoretical contributions have had a profound effect on the discipline and the profession. His MIT colleague Robert Solow (also a Nobel laureate) noted that Samuelson “was producing new ideas into his 94th and last year.”

 Sources: Stanley Fischer, “Paul Anthony Samuelson,” in John Eatwell, Murray Milgate, and Peter Newman, eds., The New Palgrave: A Dictionary of Economics, Vol. 4 (London: Macmillan, 1987), pp. 234–41; Adrian Kendry, “Paul Samuelson and the Scientific Awakening of Economics,” in J. R. Shackleton and Gareth Locksley, eds., Twelve Contemporary Economists (London: Macmillan, 1981), chap. 12; “Paul A. Samuelson, Nobel Laureate,” MIT Department of Economics website, econ-www.mit.edu; Robert M. Solow, “Paul A. Samuelson, (1915–2009),” Science, vol. 327, January 15, 2010, p. 282.

With this H-O analysis in mind, one of its major conclusions, commonly referred to as the, 

Heckscher-Ohlin theorem: a country will export the commodity that uses relatively intensively its relatively abundant factor of production, and it will import the good that uses relatively intensively its relatively scarce factor of production. 

This conclusion follows logically from the initial assumptions. While the Heckscher-Ohlin theorem seems to be consistent in a general way with what we observe, violations of H-O assumptions can lead to different behaviour by a nation in terms of the commodity structure of its trade. The extent to which the H-O theorem is supported by empirical tests is discussed in the next chapter. 

Trade, Factor Prices, and Income Distribution

The Factor Price Equalization Theorem

Different relative prices in autarky are sufficient to generate a basis for trade in trade theory.

Further, as trade takes place between two countries, prices adjust until both countries face the same set of relative prices. Our discussion of H-O demonstrated that this convergence of product prices takes place as the price of the product using the relatively abundant factor increases with trade and the price of the product using the country’s relatively scarce factor falls.

This change in final product prices has implications for the prices of factors in both of the participating countries as well, as was rigorously pointed out by Paul A. Samuelson in 1949. Let’s again consider the two countries producing cloth and steel, with cloth the labor-intensive good and steel the capital-intensive good. Country I is the capital-abundant country and country II the labor-abundant country. With the opening of trade, the price of cloth rises and the price of steel falls in country II, signaling producers to produce more cloth and less steel. Assuming perfect competition, production will shift along the PPF toward more output of cloth and less of steel. For this to happen, resources must be shifted from the production of steel to cloth. However, the bundle of resources released from steel production is different from the bundle required for increased cloth production because the relative factor intensities of the two goods differ. As the capital-intensive good, steel uses a bundle of resources that contains relatively more capital than the bundle of resources desired by cloth producers at the initial factor prices. Alternatively, the bundle released from steel production does not contain the required amount of labor relative to capital to satisfy the expanding cloth production. There is thus an increase in the demand for labor and a decrease in the demand for capital as this adjustment takes place. Assuming fixed factor supplies (see Figure 5), these market changes will lead to an increase in the price of labor and a decrease in the price of capital. The change of factor prices will cause the factor price ratio, (w/r)II, to rise and induce producers to move to a different equilibrium point on each respective isoquant (see Figure 6). Note that these price and production adjustments lead to a higher K/L ratio in both industries in this labor-abundant country. In country I, a similar adjustment takes place. With the initiation of trade, the relative price of steel rises, signaling producers to produce more steel and less cloth. The expansion of steel production and the contraction of cloth production lead to an increase in the overall demand for capital and a decrease in the overall demand for labor. With factor supplies fixed, an increase in the price of capital and a decrease in the price of labor will occur. The resulting decline in the factor price ratio, (w/r)I, means that producers will substitute labor for capital in both industries until the ratio of factor prices is again equal to the slope of the production isoquants. As a result of this change in relative prices, the K/L ratio in country I will fall in both industries.

Combining the general equilibrium results of country I and country II reveals an interesting phenomenon. Prior to trade, (w/r)I < (w/r)II. However, with trade the factor price ratio in country I falls while that of country II rises. Trade will expand until both countries face the same set of relative factor prices. The result is what is known as the factor price equalization theorem, often referred to as the second important contribution of the H-O analysis (the first being the H-O theorem): in equilibrium, with both countries facing the same relative (and absolute) product prices, with both having the same technology, and with constant returns to scale, relative (and absolute) costs will be equalized; the only way this can happen is if, in fact, factor prices are equalized. Trade in final goods essentially substitutes for movement of factors between countries, leading to an increase in the price of the abundant factor and a fall in the price of the scarce factor among participating countries until relative factor prices are equal (see Figure 7). Although the implications of trade for factor prices seem logically correct, we do not observe in practice the complete factor price equalization suggested by H-O. This is not surprising because several of the assumptions of H-O are not realized, or not realized as fully as stated in the model. Transportation costs, tariffs, subsidies, or other economic policies contribute to different product prices between countries. If product prices are not the same, then relative factor prices certainly cannot be expected to be the same although the tendency to equalize can still be present. In addition to the failure of goods prices to equalize, imperfect competition, nontraded goods, and unemployed resources also cause problems for factor price equalization. In addition, the factors of production are not homogeneous. If one acknowledges that the relative structure and quality of factors can vary between countries, the equalization of factor prices—in the sense that they are discussed here—is much less likely to come about. Further, technology is not everywhere identical, so that the rewards given the factors of production may well vary from country to country and inhibit the equalization of factor prices.

Despite these limitations, the H-O model provides some helpful insights into the likely impact of trade on relative factor prices. Trade based on comparative advantage should tend to increase the demand for the abundant factor and ultimately exert some upward pressure on its price, assuming that the presence of unemployed resources does not entirely absorb the price pressure. Thus, for the labor-abundant country, trade can offer a way to employ more fully the abundant factor and/or to increase its wages, and at the same time earn scarce foreign exchange necessary to import needed capital goods. The experiences of economies such as Taiwan support this view and demonstrate that in a general way, the factor price movements described earlier do occur. Finally, as economist Robert Mundell (1957) noted, the same result would obtain with respect to commodity prices and factor prices if factors were mobile between countries and final products were immobile internationally. In this instance, the relatively abundant factors would move from relatively low-price countries to high-price countries, causing factor price movements similar to those described. These factor movements would continue until factor (and commodity) prices are equalized, assuming that such movement of factors is costless. Thus, concerning their impact on prices, goods movements and factor movements are indeed substitutes for each other. 

The Stolper-Samuelson Theorem and Income Distribution Effects of Trade in the Heckscher-Ohlin Model

Wolfgang Stolper and Paul Samuelson developed the Stolper-Samuelson theorem in an article published in 1941. The initial article focused on the income distribution effects of tariffs, but the theorem was subsequently employed in the literature to explain the income distribution effects of international trade in general. The argument builds upon the changes in factor prices that accompany the opening of trade, which were discussed in the previous section. The argument proceeds as follows: assume that a labor-abundant country initiates trade. This will lead to an increase in the price of the abundant factor, labor, and a decrease in the price of the scarce factor, capital. Assuming that full employment takes place both page 136before and after trade, it is clear that labor’s total nominal income has increased, because the wage has increased and the labor employed remains the same. Similarly, the nominal income share of capital will have fallen since the price of capital has fallen and the capital employed remains the same at full employment. To this point the argument seems very straightforward. However, it is important to remember that the ability to obtain goods and services, that is, real income, depends not only on changes in income but also on changes in product prices. Thus, workers who consume only the cheaper, imported capital-intensive good are clearly better off, because their nominal income has increased and the price of the capital-intensive good has fallen. Their absolute and relative command over this product has increased. But what about those workers who consume only the labor-intensive export good? This case is not so clear, since both their nominal income and the price of the good they consume have increased. If their income has increased relatively more (less) than the price of the labor-intensive good, then their real income has increased (decreased). Is it possible to reach a definitive conclusion about the real income of this group with trade? Using the equilibrium condition that comes about in competitive factor markets, we can demonstrate that the wage rate in the labor-abundant country will rise relatively more than the price of the export good. Remember that, in equilibrium, labor’s wage equals the marginal physical product of labor (MPPL) times the price of the export good. Because both the wage and the price of the export good are increasing, the answer to the question, “Which is rising relatively more?,” rests on the nature of changes in MPPL. If labor is becoming more productive, then wages will be rising more than the price of the export good, and real income will be rising. If labor is becoming less productive, then wage increases will be outpaced by rising export-good prices. With trade, the labor-abundant country will find the price of capital falling and the wage rate increasing (as noted earlier), and its producers will respond by using relatively more capital and relatively less labor in production; that is, the capital/labor ratio in production will rise. This will increase the productivity of labor at the margin (i.e., MPPL increases), resulting in an unambiguous increase in the real income of labor. We can therefore conclude that the real income share of the owners of the abundant factor increases with trade. Because a similar argument can be used to demonstrate that the price of capital is falling relatively more than the price of the capital-intensive import (because with an increase in the capital/labor ratio the marginal product of capital is falling, as each unit of capital has less labor to work with), it is clear that the real income of the owners of the scarce factor is decreasing with trade. This result—that the price of a factor changes relatively more than the price of the good intensive in that factor—is often referred to as the magnification effect. Thus, the third aspect of the Heckscher-Ohlin analysis regarding the income distribution effects of trade is explained in the following more formal way by the Stolper-Samuelson theorem: with full employment both before and after trade takes place, the increase in the price of the abundant factor and the fall in the price of the scarce factor because of trade imply that the owners of the abundant factor will find their real incomes rising and the owners of the scarce factor will find their real incomes falling. Given these conclusions, it is not surprising that owners of the relatively abundant resources tend to be “free traders” while owners of relatively scarce resources tend to favor trade restrictions. For example, within the United States, agricultural producers and owners of technology and capital-intensive industries have tended to support expanding trade and/or dismantling trade restrictions, while organized labor has tended to oppose the expansion of trade. Finally, we may not see the clear-cut income distribution effects with trade because relative factor prices in the real world do not often appear to be as responsive to trade as the H-O model implies. In addition, personal or household income distribution reflects page 137not only the distribution of income between factors of production but also the ownership of the factors of production. Because individuals or households often own several factors of production, the final impact of trade on personal income distribution is far from clear. Conclusions The initial work by Heckscher and Ohlin has had a profound effect on the theory of international trade. This seminal work led not only to the famous H-O theorem but also to three additional propositions. Two of these, the factor price equalization theorem and the Stolper-Samuelson theorem, have been discussed. The final theorem, the Rybczynski theorem, which focuses on changes in factor endowments and the accompanying changes in final products produced, will be developed in Chapter 11.

Theoretical Qualifications of Heckscher-Ohlin

Several of the assumptions in the H-O model are not always applicable to the real world. For that reason, it is useful to examine assumptions that seem especially critical to the results of the model and to determine the impact of their absence on the H-O result. 

Demand Reversal

A strong assumption in the H-O model is that tastes and preferences are identical in the trading countries. If this is not true, it is no longer possible to predict the pretrade autarky prices and thus the structure of trade. The reason is that each country’s tastes and preferences could cause it to value the products in very different ways. An extreme example of this is often referred to as demand reversal. In Figure 8, demands in the two countries are so different that in country I the price of the good (steel) that intensively uses the relatively abundant factor is actually higher than its price in the trading partner, country II. With the opening of trade, country I would find itself exporting cloth and importing steel from country II because steel is relatively cheaper at international prices. This is illustrated in Figure 8 by the international terms-of-trade line, (PC/PS)int, which is steeper than autarky prices in country I and flatter than autarky prices in country II. This pattern of trade is, of course, the opposite of that predicted by H-O. It will cause the relative price of capital to fall in country I and that of labor to fall in country II. The difference in the nature of demand between these two countries, with each tending to prefer the good intensive in its physically abundant factor, has caused them to trade in a manner opposite to that anticipated by the H-O analysis. While demand patterns seem to be similar throughout the world, especially among similar socioeconomic income classes, differences in tastes and preferences certainly exist. If the differences are sufficiently strong, they can reduce the ability of the H-O model to predict trade and the movement of factor prices. Note, however, that the H-O model would still hold even in this instance if the analysis is restricted to the price definition of relative factor abundance. This occurs because home demand for the product using the abundant factor intensively leads to such a high price for that product and the factor used intensively in its production that the physically abundant factor is the scarce factor from the standpoint of the price definition.

Factor-Intensity Reversal

A second assumption crucial to the H-O conclusions is that a commodity is always relatively intensive in a given factor regardless of relative factor prices (the strong-factor-intensity assumption). Critical to this exercise is drawing the curvature of the isoquants so that each possible pair intersects only once (see Figure 1, page 126). Without this assumption, the H-O model cannot always accurately predict the structure of trade, even if technology is the same between countries. A violation of the assumption appears in Figure 9. The degree of substitution between the two factors is sufficiently different between industries (labor and capital can be substituted for each other more easily in cloth production than in steel production) so that we cannot guarantee that a given product will always be relatively intensive in the same factor. To see why this is so, look at panel (b). At (w/r)1, capital is relatively expensive; that is, the price of capital is high and the price of labor is low. With relatively flat isocost lines, producers will minimize costs by using KS1 amounts of capital and LS1 amounts of labor in steel production and KC1 amounts of capital and LC1 of labor in cloth production. The K/L ratio in steel production is greater than that in cloth production, suggesting that steel is the capital-intensive product. Next, suppose that labor is relatively more expensive. This results in a higher w/r ratio, (w/r)2, and a steeper isocost line. Producers attempting to minimize cost will employ KS2 amounts of capital and LS2 amounts of labor in steel production and KC2 amounts of capital and LC2 amounts of labor in cloth production. With this set of relative prices, the K/L ratio for cloth is now larger than the K/L ratio for steel. The factor-intensity comparison between steel and cloth has reversed with this large change in relative factor prices. These commodities are thus not always relatively intensive in the same factor. The H-O model may no longer be able to predict the export good based on relative factor abundance.

Suppose that (w/r)1 applies to country I and (w/r)2 to country II. From H-O, we expect country I (the labor-abundant country in this example) to export cloth (the labor-intensive good) and country II (the capital-abundant country) to export steel (the capital-intensive good). However, when capital is abundant (country II), cloth is the capital-intensive good. We would expect cloth to be country II’s export as well. Predicting trade flows in this two-country case is problematic because factor-intensity reversal (FIR) exists. FIR occurs when a commodity has a different relative factor intensity at different relative factor prices. With one country exporting cloth and the other exporting steel in actual trade, one of them will match the H-O prediction, but the other will not. Factor-intensity reversal could also interfere with factor price equalization, because one of the two countries can end up exporting the good that intensively uses its relatively scarce factor. For example, in Figure 9, country I might end up exporting steel and importing cloth. This will produce upward pressure on the price of capital and downward pressure on wages in country I, much like that occurring in country II. If this happens, relative factor prices in both countries (w/r) will move in the same direction (both will be falling) instead of converging toward each other. In the larger context, FIR helps us understand why it might be possible for a labor-abundant country such as India and a capital-abundant country such as the United States to export the same commodity, steel, for example. The question remains open as to whether FIRs actually exist to any important degree, but most economists doubt that these reversals alone are likely to explain trade patterns that appear to be inconsistent with H-O. 

Transportation Costs

A third assumption that is not valid in the real world is that of no transportation costs. Product prices will differ between two locations by the cost of transportation. This is demonstrated page 140by the two-country market graphs for corn in Figure 10. The market price in autarky for corn is lower in France than in Norway. Consequently, Norway has an incentive to buy corn from France. Ignoring transportation costs, these two countries should trade at a common (international) price for corn that will be lower than the price in Norway and higher than the autarky price in France. That will cause the amount of excess supply available for export in France (Q1Q2) to be exactly equal to the excess demand for imports (q1q2) in Norway.

Suppose that we now include transportation costs. If France attempts to pass the entire transportation cost on to Norway, the price of corn in Norway will rise and Norway’s excess demand (and imports) will fall. This leaves France with an inventory of corn it does not want. France will lower its price and, thus, the price including transportation costs in Norway. As this happens, France will find that it now has a smaller amount of corn available for export, an amount more in line with the new quantity demanded in Norway at the higher transportation-inclusive price. Ultimately, the price in Norway will rise above the original international equilibrium price, and the price in France will fall below that price until the amount of France’s desired exports is exactly equal to the amount of Norway’s desired imports. The difference between the price of corn in the two countries will equal exactly the transportation costs involved. Another point is that the participating countries will not necessarily share the transportation costs equally. Ultimately, the incidence of the transportation cost will depend on the elasticities of supply and demand in each country. The more inelastic supply and demand are in the importing country and the more elastic demand and supply are in the exporting country, the larger the relative amount of transportation costs paid by the importing country. Similarly, the more inelastic market conditions are in the exporting country and the more elastic in the importing country, the greater the amount of transportation costs borne by the exporting country. The costs associated with moving goods between countries have been undergoing change in recent years because of new transport technologies and emerging market page 141concerns. Until recently, transportation costs had demonstrated a downward trend because of the use of larger ocean vessels, new cargo handling techniques, and an expanded use of air transport.3 However, in the new globalized world where “quick response” is replacing the holding of large inventories, increased attention is being directed to transport time as well as distance. David L. Hummels and Georg Schauer (2013), using U.S. merchandise import data, estimate that “ocean shipping costs are equivalent to a 3 percent tariff and air shipping costs are equivalent to an 8 percent tariff” (p. 2943). They also estimate that one extra day in transit would be equivalent to an additional tariff in the 0.6–2.1 percent range, depending on the product, because time is valuable to sales. Hummels and Schauer suggest that timely delivery is like an increase in product quality to a consumer, a feature that, in conjunction with a dramatic fall in air transport costs, is consistent with the fact that in recent decades air shipments have increased 2.6 times faster than ocean shipments. The implications of transportation costs do not alter H-O conclusions about the composition of trade. However, the amount of trade and specialization of production will be reduced, and the relative structure of trade could be changed as a result of differences in transportation costs for different types of goods. However, because relative product prices do not equalize between countries, relative factor prices will not equalize and complete factor price equalization cannot be attained. Finally, if transportation costs are sufficiently large, they can prevent trade from taking place, even though commodity autarky prices are clearly different between countries; that is, transportation costs can lead to the presence of nontraded goods. 

Imperfect Competition

A fourth assumption important to the H-O analysis has been the presence of perfect competition. This assumption was necessary to guarantee that product prices and factor prices would equalize with trade. Again, we know in the real world that imperfect information, barriers to entry (both natural and contrived), and so forth, lead to imperfect competition of many different forms. Let’s examine briefly how imperfect competition such as monopoly can alter the H-O conclusions about several of the effects of trade. A first case is a variant of the traditional domestic monopoly model. The monopolist maintains the monopoly position at home but at some point chooses to export at world prices. In other words, the monopolist continues to act as a price setter at home but becomes a price taker on the world market. This can occur only if imports are prevented from coming into the country. Assume that the monopolist is maximizing profits at the quantity of output where marginal cost (MC) equals marginal revenue (MR), that is, at P0 and Q0 in Figure 11. If it now becomes possible to export as much as desired at the world price, Pint, what should the monopolist do to maximize profits? Profits are still maximized by equating MC with MR. However, the MR curve now consists of the domestic MR curve down to the Pint curve and the Pint curve beyond that point. Because the monopoly firm can sell all that it wishes at the international price, it has no reason ever to sell at a lower MR. Consequently, production is Q1, where MC = Pint. The quantity Q2 is sold in the domestic market at price P2, and Q1 − Q2 is exported. Because the international price puts a floor on the marginal revenue at all quantities to the right of Q2, the monopolist actually reduces the amount sold in the local market and charges a higher price. In this case, international trade leads to an increased difference between the domestic price and the world price, not a convergence to a single commodity price. The production and factor price effects tend to approach the H-O result as the monopolist acts as a price taker in the world market. However, the domestic market distortion that permits the monopolist to sell with trade at page 142a higher price at home than in autarky inhibits product price equalization and thus factor price equalization.

A second case is simply the application of pure monopolistic price discrimination to international trade. In this case, we have a single world supplier faced with how to distribute output among several countries and what price to charge in each. It is assumed that the monopolist is a profit maximizer, that the markets in the various countries can be kept separate (i.e., arbitrage cannot take place between markets), and that the elasticities of demand differ between the various markets. Assume that the monopolist is faced with two markets (see Figure 12). Assume also that the marginal cost of the monopolist is constant. What is the optimal amount to sell in each market? The level of total output is equal to the sum of the optimal amount of sales in each market. To maximize profits, the monopolist locates the quantity in each market at which MC = MR. The profit maximization criterion thus indicates that at a marginal cost of MC*, the quantity QI should be sold in country I at PI, and QII in country II at PII. A higher price will be charged in the market where demand is less elastic. Pure price discrimination leads to the charging of different prices in different markets and tends to reduce the degree of factor price equalization that takes place. (Remember that this type of discrimination exists only as long as there is no arbitrage between markets.) Although the presence of a single world supplier of a product is very rare, it is not uncommon for several major suppliers to band together and form a cartel. This arrangement allows them to behave economically as a single world supplier and to price-discriminate between markets.

Immobile or Commodity-Specific Factors

It has been assumed up to this point that factors are completely mobile between different uses in production within a country. This assumption permits production adjustments to move smoothly along the PPF in response to changes in relative product prices. Often, page 143however, it is not easy or even possible for factors to be moved from the production of one product to another (e.g., from wheat to automobile production). With some degree of factor immobility, at least in the short run, the nature of the adjustment to international trade is altered from that suggested in the H-O framework. Adjustment to trade has been analyzed in this instance through the use of the specific-factors model.

The specific-factors model (SF model) is an attempt to explore the implications of short-run factor immobility between sectors in an H-O context. In the short run, it assumes that there are three factors of production, not two. The three factors in industries X and Y are: (a) labor, which is mobile and can be used to produce either good X or good Y; (b) capital in industry X, KX, which can be employed in that industry but not in industry Y; and (c) capital in industry Y, KY, which cannot be used in industry X. The SF model acknowledges that, in practice, it takes time for capital to be depreciated in one industry and reemployed in another. The contrast between the assumptions of the SF model and those of the H-O model can be shown in an Edgeworth box diagram (see Figure 13). In panel (a) the factors are freely mobile between sectors, and the production contract curve has smooth curvature and connects the lower-left and upper-right origins of the box. This is the typical Edgeworth box. Panel (b) illustrates the Edgeworth box in the context of the SF model. Because capital is fixed in industry X, that amount of fixed capital is shown by the vertical distance No matter what amount of good X is produced, quantity of capital is used in industry X. Similarly, the fixed capital used in industry Y is shown by the vertical distance Hence, in the SF model, the contract curve is the horizontal liThe different contract curves in the two situations will be associated with different PPFs, because PPFs are derived from contract curves (see Chapter 5). In Figure 14, the PPF labeled RA′S is the PPF associated with the normal contract curve of both panels of Figure 13. The PPF labeled TA′V represents the PPF associated with the specific-factors contract curve This “new” PPF is coincident with the “normal” PPF only at point A′. A movement from point A to point B on the normal contract curve in Figure 13(b) yields a movement from point A′ to point B′ on the normal PPF in Figure 14, while a movement from A to C on the specific-factors contract curve results in a movement from point A′ to point C′. But the output of good Y at B or B′ (amount y2) in the normal situation will be associated with less output of good X under the specific-factors situation than under the normal or mobile-factors situation. This difference in output of X reflects the fact that isoquant y2 in Figure 13(b) is associated with isoquant x2 on the normal contract curve but only with the smaller quantity of X on isoquant x3 (at point C) on the specific-factors contract curve. Because analogous contrasts can be made with all other points on the two different contract curves in Figure 13(b), it follows that the specific-factors PPF in Figure 14 will lie inside the normal PPF except at point A. In fact, in the United States at present, a case can be made that not only is capital a specific factor, but labor also may be immobile in the sense of possessing particular skills. In this situation there may be potential for the country to be even further “inside” its normal PPF, as evidenced by the recent recession. We do not pursue this additional immobility complication in this chapter, however. 

IN THE REAL WORLD: MONOPOLY BEHAVIOR IN INTERNATIONAL TRADE

Economists know from their understanding of market behaviour that imperfect competition leads to a reduction in quantity sold and an increase in price, as well as to the possibility of price discrimination. An historical example of this latter behavior was the incandescent electric lamp cartel, where firms in different countries acted jointly as a monopoly. A 1939 U.S. Tariff Commission study indicated the severe price discrimination practices of this cartel, as the accompanying table shows. The cartel clearly kept prices from equalizing across countries, inhibiting any tendency toward factor price equalization. A current example of monopolistic behavior came to light when the European Union (EU) in 2015 levied formal charges against Russian government–controlled natural gas supplier Gazprom, accusing the firm of blocking competition and charging unfair prices in its sales to countries in southern and eastern Europe. Monopolistic price discrimination was alleged, because, in comparison with the $341 average price per cubic meter of natural gas charged to the EU, Gazprom charged, for example, 21.7 percent more in Estonia, Latvia, and Lithuania, 11.1 percent more in Poland, 5.3 percent less in Germany, and 9.7 percent less in the Slovak Republic. Such price differences clearly could not exist if free resale of natural gas between countries were possible. Until very recently, the Organization of Petroleum Exporting Countries (OPEC), mostly Middle-Eastern countries, has been a relatively strong international cartel. It was successful in raising prices dramatically through production controls and the exercise of market power in the two oil shocks of 1973–1974 and 1979–1980. The average price for oil (the benchmark Brent crude oil price) was $3.61 per barrel in 1972, $4.25 in 1973, and $12.93 in 1974. By 1978 the price was $14.26 per barrel, and it rose to $32.11 in 1979 and $37.89 in 1980. Oil prices then fell because of consumer conservation, new sources of supply from Mexico, the North Sea, and Alaska; and switches to alternative fuels. By 1986 crude prices had fallen to $14.43; they then fell, after some increase in the early 1990s, to $12.72 in 1998. OPEC then attempted to restrict output, and the price averaged $28.31 in 2000. The decision by OPEC to cut output by 10 percent in early 2004, as well as growth in demand, led to an average price of $54.43 in 2005. The price rose to $97.66 by 2008 but then decreased due to the Great Recession to $61.86 in 2008 and $79.63 in 2009. However, the average oil price subsequently rose to above $100 per barrel in 2011, 2012, and 2013. It reached about $115 in June 2014 but then, with the slowdown of GDP growth in many countries and the emergence of huge new supplies of shale oil in the United States, a glut occurred and the price fell to about $50 per barrel in early 2015. Effective cartels clearly are not permanent in nature.

The implication of the immobility of capital in an H-O context can be seen by comparing the impact on the rates of return to the factors of production when a country moves from autarky to trade. Suppose we have the normal situation of a country with full mobility of all factors of production. If the country is located at point A in Figure 13(b) in autarky, then an opening of the country to trade involving specialization in the labor-intensive good X will result, for example, in a movement from A to B in Figure 13(b) or from A′ to B′ in Figure 14. This movement will bid up the price of labor and reduce the price of capital as expanding industry X seeks to acquire relatively more labor and contracting industry Y is releasing relatively more capital. After the adjustment, at point B in Figure 13(b), the ratio of capital to labor in each industry has risen. With the rise in the K/L ratio in each industry, each worker in each industry has relatively more capital to work with, so the worker is more productive; thus, productivity and wages rise. The flip side of the rise in wages is the fall in the real return to capital. The important policy implication in the traditional Heckscher-Ohlin, full-factor mobility situation is that the country’s abundant factor prefers free trade to autarky and that the country’s scarce factor prefers autarky to free trade. Even though the country as a whole gains from trade, some part of the economy (the scarce factor) will have an incentive to argue for protection. What is the consequence of trade for factor returns in the specific-factors model? With autarky at point A in Figure 13(b) and the opening of the country to trade, labor-intensive industry X expands because of the higher price of X and capital-intensive industry Y contracts because of the lower price of Y. Production tends to move to the right from point A in the direction of a point such as point C on the SF contract curve. The increased demand for labor will bid up the money wage of all labor. However, the direction of movement of the return to capital depends on which industry is being considered. Industry X has increased its demand for capital, but the supply of capital is fixed at Hence, the return to capital in X rises with the opening of the country to trade. However, the demand for capital in industry Y falls because some of good Y is now imported rather than produced at home. Thus, the demand for capital in industry Y decreases, while its supply of capital is fixed. The return to capital in Y therefore falls. These income distribution effects of trade are obviously different from those of the traditional H-O model. There, capital as a whole suffered a decline in its return, but in the SF model capital in X gains and capital in Y loses. The scarce factor of production will not be unanimously opposed to moving from autarky to trade. Owners of capital in industry Y will argue against free trade, while those in industry X will argue in favor of it. This situation may indeed be more realistic than the traditional H-O type of model, especially in the short run, where all of one factor was opposed to trade and all of the other factor favored it. A final note is necessary about the return to labor. Saying that the money wage for labor has risen does not mean that labor’s real wage has risen. Consider the money wage in industry X, which equals the money wage in industry Y with competition. Remember that the money wage equals the price of the product times the marginal physical product of labor, thus, w = (PX)(MPPLX). With trade, MPPLX falls, because more labor is being used with the fixed amount of capital In other words, each worker has less capital to work with. Because w = (PX)(MPPLX), this means that w/PX has fallen because (w/PX) = MPPLX. The fall in w/PX simply indicates that money wages have not risen as much as the price of page 147good X. Workers who consume only good X are worse off because their real wages have declined. Analogously, w/PY rises; thus, if workers consume only good Y, their real wages have risen. The direction of the real return for a worker therefore depends on the bundle of goods being consumed. The SF model yields conclusions on the winners and losers from free trade that may be more consonant than traditional H-O trade theory with what we observe in the real world. (See Concept Box 1 for a more thorough discussion of the impact of trade on labor’s real wage in the specific-factors model.) 

Other Considerations

It is evident that several other assumptions, such as constant returns to scale, identical technology, and the absence of policy obstacles to trade, also are not always applicable to the real world. To the extent that they are not, the conclusions of the H-O model are compromised. Several of these conditions will be examined further in Chapter 10 and in later chapters on trade policy.

The Specific-Factors Model and the Real Wage of Workers

As noted in the text, the direction of movement of the real return to a worker when a country is opened to trade in the specific-factors model depends on that worker’s consumption pattern. This conclusion is illustrated in Figure 15. The figure portrays the demand for labor by the X industry and by the Y industry. Each industry’s demand for labor reflects the fact that a firm will hire a worker as long as the worker’s contribution to the firm’s revenue (the marginal physical product of labor times the price of the output) exceeds the cost to the firm (the wage rate). Thus, in the figure, curve DLX is the demand curve for labor by firms in the X industry (with DLX = MPPLX × PX), and DLY is the demand curve for labor by firms in the Y industry (with DLY = MPPLY × PY). These curves are downward sloping. As the wage rate falls, industry X hires more labor (reading rightward from origin 0); alternatively, as more labor is added to the fixed amount of other inputs, the MPPLX falls and the additional labor will be hired only if the wage rate is reduced. Curve DLY indicates the demand curve for labor by the Y industry but, for this industry, the quantity of labor is read in the leftward direction from the origin 0′.

Introduction

Theories, Assumptions, and the Role of Empirical Work

It is common to hear trade theories criticized for not being “relevant to the real world” or for having “unrealistic assumptions.” This has led researchers such as Leamer and Levinson (1995) to assert that empirical work on trade has had very little influence on trade theories and to encourage their colleagues to “Estimate, don’t test.” They propose that researchers should attempt to learn from real-world data rather than simply accepting or rejecting an abstract hypothesis. Rather than using the lack of relevance to the real world as a reason to ignore trade theory, it should serve as a motivation for empirical testing. Davis and Weinstein (1996, p. 434) offer a more encouraging view of empirical work. “…look for ways of weakening the strict assumptions of the theory …to find a version that does in fact work …[W]e should learn which of the assumptions …are most crucial, and to which types of data sets one can sensibly apply the various versions of the theory.”1 This chapter examines the empirical tests of the Heckscher-Ohlin predictions in an attempt to find which of the often strict assumptions are crucial. You will see that there is not a consensus among economists on the degree to which relative factor endowments explain international trade flows and the consequences of trade flows. As developed in Chapter 8, the Heckscher-Ohlin theorem states that a country will export goods that use relatively intensively the country’s relatively abundant factor of production and will import goods that use relatively intensively the country’s relatively scarce factor of production. In this chapter, we review some empirical tests of this seemingly straightforward and commonsense hypothesis. The literature has produced some conflicting results on the real-world validity of the H-O theorem. The most surprising result of one early test was that the world’s largest trader, the United States, did not trade according to the Heckscher-Ohlin pattern. Explanations are given on why this surprising result might have occurred. We then review tests for other countries and more recent work on trade patterns. In addition, we survey the current controversy regarding the extent to which H-O-type trade has contributed to the increasing income inequality in developed countries in recent years, especially in the United States. 

The Leontief Paradox

The first major test of the H-O theorem was conducted by Wassily W. Leontief and published in 1953. This comprehensive test has influenced empirical research in this area ever since. Leontief made use of his own invention—an input-output table—to test the H-O prediction. An input-output table provides details, for all industries in an economy, of the flows of output of each industry to all other industries, the purchases of inputs from all other industries, and the purchases of factor services. In addition, the table can be used to indicate not only the “direct factor requirements” of any given industry—the capital and labor used with intermediate goods in the particular stage of production—but also the total factor requirements. The total requirements include the direct requirements as well as the capital and labor used in the supplying industries of all inputs to the industry page 152(the “indirect factor requirements”). The table is very useful for calculating the aggregate country requirements of capital and labor for producing a bundle of goods such as exports and import substitutes. To evaluate the H-O prediction for the United States, Leontief imagined a situation where, using 1947 data, the United States simultaneously reduced its exports and imports proportionately by a total of $1 million each. The input-output table made it possible to determine how much capital (K) and labor (L) would be released from producing exports and how much capital and labor would be required to produce at home the $1 million of goods no longer being imported. (Leontief confined his analysis to “competitive imports,” meaning that he did not include goods that the United States did not produce at home, such as bananas.) Given the estimates of the K and L released from reducing exports and required to reproduce imports, a comparison could be made between them. Because the United States was thought to be a relatively capital-abundant country, the expectation from the statistical analysis was that the K/L ratio of the released factors from the export reduction would be greater than the K/L ratio of the factors required to produce the forgone imports. This expectation could be evaluated as to its validity through the concept of the Leontief statistic, which is defined as

where (K/L)M refers to the capital/labor ratio used in a country to produce import-competing goods and (K/L)X refers to the capital/labor ratio used to produce exports. According to the H-O theorem, a relatively capital-abundant country would have a Leontief statistic with a value less than 1.0 (since the denominator would be larger than the numerator) and a relatively labor-abundant country would have a Leontief statistic greater than 1.0. Leontief’s results were startling. He found that the hypothesized reduction of U.S. exports would release $2.55 million worth of capital and 182.3 years of labor-time, for a (K/L)X of approximately $14,000 per labor-year. On the import side, to produce the forgone imports would require $3.09 million worth of capital and 170.0 years of labor-time, yielding a (K/L)M of approximately $18,200 per labor-year. Thus, the Leontief statistic for the United States was 1.3 totally unexpected for a relatively capital-abundant country. A disaggregated analysis of his results also supported these findings. The most important export industries tended to have lower K/L ratios and higher labor requirements and lower capital requirements per dollar of output than did the most important import-competing industries. Thus, the seemingly commonsense notion that a country abundant in capital would export capital-intensive goods and import labor-intensive goods was seriously called into question. The doubt cast on the widely accepted Heckscher-Ohlin theorem by this study became known as the Leontief paradox.

Initial Explanations for the Leontief Paradox

Leontief’s results have produced many studies seeking to explain why these unexpected findings might have occurred. In this section, we briefly discuss the more well-known “explanations.”

Demand Reversal 

The concept of demand reversal was introduced in Chapter 8. In demand reversal, demand patterns across trading partners differ to such an extent that trade does not follow the H-O pattern when the physical definition of relative factor abundance is used. The relative preference of a country for goods made with its physically abundant factor (called “own-intensity preference”) offers one explanation for the Leontief paradox page 153if we hypothesize that the United States has relative preference for capital-intensive goods and that U.S. trading partners have relative preference for labor-intensive goods. The U.S. demand for capital-intensive goods bids up the price of those goods until the U.S. comparative advantage lies in labor-intensive goods. A similar process occurs in trading partners, giving them a comparative advantage in capital-intensive goods. The validity of demand reversal as an explanation of the Leontief paradox is an empirical question. However, considerable dissimilarity (which seems unlikely in practice) is required for the demand reversal explanation to be of value in understanding the paradox. Further, the presence of demand reversal would imply that demand within the United States for labor-intensive goods would be relatively low and therefore U.S. wages also would be relatively low—which is not consistent with observed wage rates across countries. Thus, other reasons for the Leontief result need to be explored.

Factor-Intensity Reversal

As noted in Chapter 8, factor-intensity reversal (FIR) occurs when a good is produced in one country by relatively capital-intensive methods but is produced in another country by relatively labor-intensive methods. It is not possible to specify unambiguously which good is capital intensive and which is labor intensive, and the Heckscher-Ohlin theorem can be valid for only one of the two countries. For example, consider a situation where X is the K-intensive good in country I but is the L-intensive good in country II (and hence Y is the L-intensive good in country I but is the K-intensive good in country II) and country I is the relatively K-abundant country. If country I is exporting X to II, then the Heckscher-Ohlin prediction is correct for country I but it cannot be correct for country II, because country II, the L-abundant country, must be exporting good Y to country I (because both countries cannot be exporting X in this two-country H-O model). However, good Y is the K-intensive good in country II, and therefore country II is not conforming to the H-O theorem. In the context of the Leontief paradox, the FIR suggests that, although U.S. import goods might have been produced labor intensively overseas, the production process of these goods in the United States was relatively capital intensive. The trading partners (being labor-abundant) were conforming to H-O when they exported the goods, but the United States was not conforming to H-O. The validity of this explanation for the occurrence of the Leontief paradox is also an empirical question. The literature is somewhat divided on the matter of whether factor intensity reversals occur with any frequency, and we cannot rule them out altogether. The most famous test was conducted by B. S. Minhas (1962) for the United States and Japan using 1947 and 1951 data for 20 industries. Suppose that we consider the same 20 industries, that we have the K/L ratio employed in each country in each of the 20 industries, and that we rank the 20 industries in descending order in each country according to K/L ratios (as Minhas did). For example, in the United States (using total capital and labor requirements) petroleum products constituted the most capital-intensive industry (it had the highest K/L ratio), coal products ranked number 2, iron and steel ranked number 8, textiles ranked number 11, shipbuilding ranked number 15, leather ranked number 19, and so forth. If there are no FIRs, then the rankings for Japan would be the same as those for the United States. Statistically, this means that the rank correlation coefficient between the U.S. ranking and the Japanese ranking would be 1.0. (Note: If two rankings are identical, the correlation coefficient between them is 1.0; if they are perfectly opposite to each other, the coefficient is −1.0; and if there is no association between the two rankings whatsoever, the rank correlation coefficient is 0.) When Minhas calculated this correlation coefficient using total factor requirements, he obtained a rank correlation coefficient of only 0.328. (This reflects that in Japan iron and steel was number 3 instead of 8, shipbuilding was number 7 instead of 15, etc.) For “direct” requirements only, the coefficient was higher but still only 0.730. Thus, doubt can be cast on the “no FIRs” assumption of Heckscher-Ohlin. However, economists such as G. C. Hufbauer (1966) and D. S. Ball (1966) pointed out that if the differences in land availability and agriculture in the two countries and the influence of these differences on the relative employment of K and L are allowed for, then the rank correlation coefficients are much closer to 1.0. page 154 

Testing for Factor-Intensity Reversals

Another study that examined the potential likelihood of factor-intensity reversals in the real world is that of Matthias Lücke (1994). Lücke notes that empirical work on the causes of trade patterns often relies on the assumption that U.S. relative factor intensities across industries are representative of relative factor intensities across industries in other countries. To test whether this assumption is valid, Lücke gathered data on 22 manufacturing industries in 37 industrialized and developing countries (the same 22 industries in each country). He then calculated the dispersion across these industries in each country of what he called “total capital intensity” (value added per employee), “human capital intensity” (wages per employee), and “physical capital intensity” (non-wage value added per employee). The resulting dispersion values within each country were then correlated with the dispersion in intensity of these same industries within the United States. Lücke’s hypothesis was that a significantly positive correlation coefficient (a linear correlation to represent the structure of intensities in general, not a rank correlation) between the dispersion in any given country and the dispersion in the United States would suggest that relative factor intensities in manufacturing industries are broadly similar between the countries and thus relative factor-intensity reversals in general are unlikely. Using the three different measures of intensity (total capital intensity, human capital intensity, and physical capital intensity) and two different time periods, Lücke was able, given data limitations, to calculate 192 different correlation coefficients between the factor-intensity dispersion in the 36 other countries and the factor-intensity dispersion in the United States. All but eight of the coefficients were positive in a statistically significant sense. He regarded his results as highly supportive of the notion that the structure of factor intensities in manufacturing industries in general in the 36 other countries was indeed very similar to that structure in the United States. Hence, factor-intensity reversals across countries in the manufacturing sector seemed very unlikely. A more recent investigation (2011) pertaining to factor-intensity reversal is that of Yoshinori Kurokawa (2011). Kurokawa’s work takes a different approach from most factor-intensity reversal studies in that it focuses on a possible factor-intensity reversal between relatively high-skilled labor and relatively low-skilled labor rather than on a possible reversal between relative capital- and labor-intensity. He dealt with U.S.–Mexican trade in 1994 and in 2000 (as NAFTA—the North American Free Trade Agreement—was being implemented). He divided manufacturing industries into two mutually exclusive categories—the “electronics” industry and all other industries, with the latter being very broadly classified as the “non-electronics” industry. Between 1994 and 2000, the electronics industry in the United States experienced a very large increase in its net exports to Mexico, at the same time that net imports into the United States from Mexico of non-electronic goods increased dramatically. Within this two-industry context of the electronics industry and the non-electronics industry, Kurokawa then examines the relative skill intensity in the two industries in 1994 and 2000 by calculating the ratio of the number of non-production workers in an industry to the number of production workers in the industry. Non-production workers are roughly designed to represent high-skill workers, and the production workers are roughly designed to represent low-skill workers. Non-production workers are generally viewed as having greater educational attainment than production workers. Kurokawa then compares this ratio in the two broad industries in each of the two countries with the average skill intensity figure for each country as a whole. Unsurprisingly, Kurokawa finds that the skill intensity in electronics in the United States is above the U.S. average skill intensity and that the skill intensity in non-electronics is below the U.S. average. Hence, for the United States, exports to Mexico conform to Heckscher-Ohlin since the United States is generally thought to be relatively more well-endowed with high-skill labor than is Mexico and the United States is exporting the relatively high-skill intensive product to Mexico. With respect to Mexico, the results are indeed surprising. Kurokawa determines that, in both 1994 and 2000, the ratio of non-production workers to production workers in non-electronics was higher than the average for Mexico, while that ratio was lower in electronics than the Mexican average. Hence, by this measure, the non-electronics industry in Mexico utilized relatively high-skill labor compared to the electronics industry (and thus the electronics industry utilized relatively low-skill labor compared to the non-electronics industry). Mexico was thus exporting its relatively high-skill-intensive product to the United States whereas the country is generally viewed as low-skill abundant relative to the United States—not in conformity with Heckscher-Ohlin. How is this surprising page 155result possible? Kurokawa’s explanation is that the result reflects a factor-intensity reversal—the electronics industry is relatively high-skill intensive in the United States compared to non-electronics, while at the same time electronics is relatively low-skill intensive compared to the non-electronics industry in Mexico. This conclusion is consistent with the earlier discussion in this chapter that, with a factor-intensity reversal, only one of the two countries can conform to Heckscher-Ohlin. It is also consistent with Kurokawa’s observation that the wages of high-skill workers relative to low-skill workers increased in both the United States and Mexico during this period—a phenomenon which fits with our earlier point that, with a factor-intensity reversal, the tendency for trade to lead toward factor price equalization disappears. The Kurokawa study would suggest that further analysis is needed, especially with regard to the point that the electronics industry is less skill-intensive in Mexico than the skill-intensity level of other industries. Sources: Matthias Lücke, “A Note on Empirical Evidence on Factor Intensity Reversals in Manufacturing,” Economia Internazionale 47, no. 1 (February 1994), pp. 51–54; Yoshinori Kurokawa, “Is a Skill Intensity Reversal a Mere Theoretical Curiosum? Evidence from the US and Mexico,” Economics Letters 112, no. 2 (August 2011), pp. 151–54.

U.S. Tariff Structure 

This explanation for the Leontief paradox focuses on the factor intensity of the goods that primarily receive tariff (and other trade barrier) protection in the United States. From Heckscher-Ohlin and the accompanying Stolper-Samuelson theorem we learned that the opening of a country to trade increases the real return of the abundant factor and decreases the real return of the scarce factor. This suggests that, in the United States, labor will be more protectionist than will owners of capital (which is the case). Therefore, U.S. trade barriers tend to hit hardest the imports of relatively labor-intensive goods. With these goods restricted, the hypothesis is that the composition of the U.S. import bundle is relatively more capital intensive than would otherwise be the case because labor-intensive goods are kept out by protection. Thus, the Leontief result might have been in part a reflection of the tariff structure of the United States and not an indication of the free-trade pattern that might conform to Heckscher-Ohlin. Can this argument account for the occurrence of the Leontief paradox? In a 1971 study, Robert Baldwin recognized the possible role of tariffs and estimated that the K/L ratio in U.S. imports would be about 5 percent lower if this effect were incorporated. While allowance for the tariff structure works toward reducing the extent of the paradox, it seemed unable to remove it fully.

Different Skill Levels of Labor

In this explanation of the Leontief paradox, the basic point is that the use of “labor” as a factor of production may involve a category that is too aggregative, because there are many different kinds and qualities of labor. One early test involving this approach was conducted by Donald Keesing (1966). Keesing divided labor into eight different categories. page 156For example, category I (scientists and engineers) was regarded as the most skilled labor (we have unsuccessfully searched for a listing of economists in this category!), while category II (technicians and draftsmen) was regarded as the second most skilled. This listing continued through category VIII (unskilled and semiskilled workers). Keesing then compared U.S. labor requirements in export- and import-competing industries with those of 13 other countries for 1962. He found that U.S. exports embodied a higher proportion of category I workers and a lower proportion of category VIII workers than the exports of other countries. Similarly, on the import side, the United States used the smallest fraction of category I workers and the largest fraction of category VIII workers. This type of test suggests that the Leontief paradox may have occurred because a two-factor test was employed instead of a test with a larger number of factors (with each skill category of labor regarded as a distinct factor of production). Perhaps the United States is skilled labor abundant (as well as capital abundant) and unskilled labor scarce in its factor endowments. If so, the U.S. trade pattern conformed to Heckscher-Ohlin because the United States was exporting goods that were relatively intensive in skilled labor and importing goods that were relatively intensive in unskilled labor. Other tests have confirmed the general impressions from the Keesing analysis. For example, Robert Baldwin’s study (1971) found that, compared with import-competing industries, export industries had a higher proportion of workers with 13 or more years of schooling. On the other hand, compared with export industries, import-competing industries had a higher proportion of workers with eight years of schooling or less. Using data from the 1970s and early 1980s, Staiger, Deardorff, and Stern (1988) estimated that a move to free trade by the United States would lead to a reduction in demand for operatives and an expansion of demand for scientists, engineers, and physical capital. These results of eliminating trade restrictions are also consistent with an H-O multifactor explanation of trade that has no Leontief paradox. Indeed, many tests of this type have suggested that it is necessary to go beyond a two-factor model to test whether the U.S. trade pattern conforms to Heckscher-Ohlin.

The Role of Natural Resources

This explanation also builds around the notion that a two-factor test is too restrictive for proper assessment of the empirical validity of the Heckscher-Ohlin theorem. In this case, the additional factor is “natural resources.” In the context of the Leontief paradox, many of the import-competing goods labeled as “capital intensive” were really “natural resource intensive.” Leontief was assessing the factor requirements for producing imports at home and found that this production required the use of capital-intensive production processes; but in industries such as petroleum products, coal products, and iron and steel, domestic production of the goods involves a great deal of natural resources as well as capital. For Leontief, production of these import-competing goods involved capital-intensive production [raising (K/L)M in the calculation of the Leontief statistic] because his was a two-factor test. However, the “true” intensity of the goods produced might not be in capital but in natural resources. If we were able to identify the true factor intensity, we might conclude that the United States was importing natural-resource-intensive products. If the United States is relatively scarce in its endowment of natural resources, then there is no paradox with Heckscher-Ohlin.

The importance of natural resources has been confirmed in some empirical tests. For example, James Hartigan (1981) performed Leontief-type tests for U.S. trade for 1947 and 1951. A paradox existed in general, but not when natural-resource-intensive industries were deleted from the tests. Without natural-resource industries, U.S. trade yielded a Leontief statistic of 0.917 for 1947 and 0.881 for 1951. These results are not “paradoxical.” Leontief himself (1956) also discovered that adjustment for natural resources could reverse the paradox. On the other hand, Robert Baldwin (1971) found that accounting for page 157natural resources reduced the paradox but did not eliminate it. Thus, there is uncertainty about the relative importance of the “natural resources as a third factor” explanation of the Leontief paradox. 

More Recent Tests of the Heckscher- Ohlin Theorem

The paradox found by Leontief has spawned many, many investigations into the question of the validity of the Heckscher-Ohlin theorem in predicting trade patterns. We will mention only a few of these studies here, but even our limited discussion should suffice to indicate that the question of the empirical validity of H-O continues to be an unanswered one. This ambiguity was established early on when some tests utilizing Leontief’s approach—Tatemoto and Ichimura (1959) for Japan, Stolper and Roskamp (1961) for East Germany, and Rosefielde (1974) for the Soviet Union—found support for the theorem, because factor intensities of trade flows matched expectations from factor endowments, while other studies—Wahl (1961) for Canada and Bharadwaj (1962) for India—yielded “surprising,” unexpected results. (For example, India’s exports to the United States were found to be relatively capital intensive and India’s imports from the United States were found to be relatively labor intensive!) With respect to the United States, Robert Baldwin (1971, p. 134) found, for 1962 trade, a Leontief statistic of 1.27; when agriculture was excluded, the figure rose to 1.41, and when natural resource industries were excluded, the Leontief statistic fell to 1.04. Thus, in all cases (although only barely so when natural resource industries were excluded), the paradox still existed. However, utilizing a different approach, Harkness and Kyle (1975) found that, if natural resource industries are excluded, a U.S. industry had a greater probability of being a positive net exporter (where net exports = exports minus imports of the industry’s product) if, among other characteristics, it utilized a higher ratio of capital to labor in production. This does not suggest a paradox because the United States was thought to be relatively capital abundant and relatively labor scarce. In addition, an industry had a higher probability of being a positive net exporter if it had a higher ratio of scientists and engineers in its labor force, which lends support to a “labor skills” or “human capital” explanation of the U.S. trade pattern. An important role for human capital was also uncovered by Stern and Maskus (1981). They attempted to explain the net export position of U.S. industries over the years 1958–1976, and they found the size of net exports of an industry to be positively correlated with the amount of human capital used in the industry. They also found the size of net exports to be negatively correlated with the amount of labor in the industry and sometimes, although not always, negatively correlated with the amount of physical capital used in the industry. (In affirmation of the traditional H-O prediction, positive net exports were in fact positively associated with physical capital in some years.) A recent study by Julien Gourdon of the World Bank (2009) examined relative factor endowments as predictors of whether countries would be net exporters or net importers of goods in various product categories. He examined the endowments and trade patterns of 71 countries over the period from 1960 to 2000. Factor endowments for any given country were measured relative to world factor endowments, and factors were broken into the categories of arable land per inhabitant, capital stock per worker, and three types of labor based on differing levels of skill. Commodities were classified into groupings such as agricultural products, processed food products, manufactured goods intensive in capital, and manufactured products intensive in unskilled labor. The results with respect to the predictions of the correct sign between the relevant endowments and net exports or net imports of the given products were encouraging—for example, the correct sign was obtained 56 percent of the time for unskilled labor-intensive manufactured products, 70 percent of the time for page 158agricultural products, 71 percent of the time in capital-intensive manufactured products, 72 percent of the time in skilled labor-intensive products, and 85 percent of the time in what he called technology-intensive products. Gourdon then added variables associated with newer, post–Heckscher-Ohlin theories, such as consumer preferences and economies of scale (discussed in Chapter 10), and the results were even better for these endowments-plus-newer-variables tests. In an overview, Gourdon indicated that, over time, endowments are indeed important for determining trade and that an increasing role in trade and specialization is being played by different types of labor skills. An interesting study that seemed to provide strong support for the Heckscher-Ohlin theorem was conducted by Nicholas Tsounis (2003). This study focused on whether Heckscher-Ohlin is valid empirically on a bilateral basis (i.e., in the trade of a country with individual trading partner countries rather than in the country’s total trade). Greece was the “home” country whose bilateral trade patterns were being examined, and the selected trading partners were the 11 other European Union countries during the trading time period being considered, 1988 and 1989. (The trading partners were Belgium, Denmark, France, West Germany, Ireland, Italy, Luxembourg, the Netherlands, Portugal, Spain, and the United Kingdom.) First, utilizing capital stock/investment data and an assumed depreciation rate of capital together with information on the number of working persons, Tsounis calculated the factor endowment ratio, K/L, for each country. He then divided this K/L result for each trading partner country by the Greek K/L endowment ratio. For example, the 1988 ratio for France was 2.82, meaning that France’s endowment of capital to labor was nearly three times Greece’s endowment of capital to labor. For every trading partner European Union (EU) country except for Portugal, the ratio always exceeded 1.00 (Portugal’s ratio was 0.72 in 1988 and 0.75 in 1989). Thus, Greece was relatively labor-abundant in comparison with each of its EU trading partners except for Portugal. Using an input-output table for each country, Tsounis calculated the Leontief statistic for Greece in its trade with the partner countries. In all cases except in Greek trade with Portugal, the Leontief statistic indicated that Greek imports from the EU trading partner were more capital-intensive (less labor-intensive) than were Greek exports to that trading partner. In the case of Portugal, the opposite (by a very slim margin) was found. Hence, on a bilateral basis, Greece’s trading pattern with each of its EU partners seemed to conform to Heckscher-Ohlin. 

Factor Content Approach with Many Factors

Other studies have also moved beyond calculating Leontief statistics and often beyond examining only the United States. Many factors of production in many countries have been included. Calculations are first made (using input-output tables) of the quantity of any given factor needed to produce the goods contained in any given country’s aggregate output bundle (i.e., the supply of the factor’s services embodied in production). This requirement is then compared with the demand for the given factor embodied in the country’s existing aggregate consumption bundle. If the total production requirement for any given factor exceeds the total consumption requirement, then (with assumed full employment of the factor) the country must, on balance, be exporting the services of that factor; if the total consumption requirement exceeds the total production requirement, then the country must, on balance, be importing the services of the given factor. In effect, then, a country with positive net exports of the services of a given factor must be relatively abundant in that factor, and a country with negative net exports (i.e., positive net imports) of the services of a given factor must be relatively scarce in that factor.2page 159