HIST363: Global Perspectives on Industrialization

For factor accumulation to have a major impact on the structure of output and exports, two conditions need to be satisfied. The first relates to the structure of the $\pi_{\mathrm{pv}}$ matrix and requires that changes in relative factor endowments must result in substantial changes in the composition of output à la Rybczynski, rather than on changes in factor proportions within sectors as in the neoclassical growth model.

The second is that there must be sizeable changes in relative factor endowments, that is that that ∆ν must be nonuniform. In this section, the evidence on the impacts of changes in factor endowments is first examined, and then the evidence on changes in relative factor endowments. Finally, attention turns to the extent to which the Rybczynski assumption of exogenous, or at least pre-determined, factor endowments can be taken to be realistic. 

Whether changes in relative factor endowments will affect the composition of output is a question that can only be resolved through empirical studies. If, for instance, the factor intensities of different sectors were not greatly different, or if different factors were near-perfect substitutes, this effect would not be expected to be large. The empirical impact of factor accumulation on the share of output and hence on export patterns has received considerable attention in recent years. A number of studies using quite different approaches have concluded that changes in factor endowments can have quite strong impacts on the composition of output and exports, rather than on factor prices, confirming the potential empirical importance of the Rybczynski theorem. 

Martin and Warr examine the determinants of the rapid decline in the share of agriculture in Indonesian GDP. They estimated, using time-series data, a profitfunction that incorporates the factor endowment and technological change effects discussed in this study as well as relative price changes that include the impacts of changes in trade policy. Their conclusion was that the most important determinant of the reduction in the share of agriculture in the Indonesian economy was increases in the capital-labor ratio. The output price effects that take into account the effects of factors such as changes in protection policy and in worldwide technical change played a much smaller role. Technological advance was found to be biased towards agriculture, and hence tending to increase agriculture's share of output, other things equal.

Gehlhar, Hertel and Martin used a completely different analytical tool - the GTAP computable general equilibrium model of the world economy - to examine the changing structure of the world economy. This model incorporates the non-homotheticity in consumer demand that plays such an important role in discussions of the decline in agriculture's share of output in the world economy. It also includes input-output tables with the differences in the factor intensities of different sectors that drive the Rybczynski effects when relative factor endowments change. Further, it includes forward and backward linkages through its input-output structure and the transport costs that loom large in the new economic geography. The model was first validated over the 1980s to ensure that it could realistically replicate the changes in sectoral shares of exports in the Asia-Pacific region. Then, the structure of output was projected from 1992 to 2002. A key conclusion of the analysis was that the most important determinant of likely changes in agricultural output and trade patterns, and particularly a sharp decline in reliance on agricultural exports in East Asian developing economies, was likely to be differential rates of factor accumulation, rather than non-homotheticity in consumer demand. 

Harrigan examines the impact of technological changes and changes in relative factor endowments on the structure of manufacturing output in a panel of OECD countries. His econometric results caused him to conclude that factor endowment changes, as well as technological changes, have large effects on output shares. Kee reaches the same conclusion in a study of the manufacturing sector in Singapore.

In a completely different literature, Hanson and Slaughter examine the implications of changes in the supply of workers with different skill levels in states of the United States. They find that a key part of the adjustment to changes in the supply of workers of a particular type is a change in the structure of output of the type suggested by Rybczynski effects. 

There remains some controversy about the relevance of the Rybczynski theorem in some cases. Cohen and Hsieh focused on the very large immigration of Russian Jews into Israel in the early 1990s and found results more in line with the single-sector neoclassical model: a short-run fall in the wages of native Israelis and a rise in the return to capital. Equilibrium was restored through an increase in the capital stock associated with increased external borrowing. This was, however, something of a special case. Investigation of the response of the output response to this shock, and the response itself, was complicated by the ambiguous skills endowment of the immigrants. While they were much more highly educated than the native population, they suffered substantial occupational downgrading which made it difficult to assess whether the output response should have involved outputs intensive in skilled or unskilled labor. 

If one accepts the potential validity of the Rybczynski theorem as a potential cause of structural change, a key question is whether there have been major changes in relative factor endowments that would cause changes in the composition of developing country output away from dependence on resources. Because equation (1) is linearly homogeneous in pre-determined factors such as capital, labor and human capital endowments, the key is to find changes in relative factor endowments. Equal rates of accumulation of all factors would not be expected to result in changes in the output mix. Recent data on accumulation of human and physical capital suggest that there have been quite sharp changes both between developed and developing regions, and between different developing country regions. 

The most comprehensive such database of which I am aware is that by Nehru and his coauthors. Some of their results on the growth rates of human and physical capital are presented in Table 1. These data refer to the two key measures of reproducible capital. They ignore natural resources such as land, mineral resources and well-located ports. If we make the initial working assumption that these resources are constant, then changes in the volume of capital and human capital represent increases in these resources relative to natural resources. This is a particularly suspect assumption where mineral resources are concerned as the stock of these resources is subject to dramatic change in response to investments in mineral exploration. 

Table 1. Annual Changes in Factor Endowment Ratios

As a first approximation, we scale the increases in measured capital stocks by the growth rate of the labor force for major regions. The first column of Table 1 points to quite rapid increases in physical capital per worker (K/L) in both industrial and developing countries. The 5.1 percent per year growth rate for East Asia implies more than a quadrupling of capital per worker over the thirty-year period of observation. The 2.4 percent a year increase in Latin America implies more than a doubling of the capitallabor ratio over the period. Even the 2.1 percent per year increase in Sub-Saharan Africa implies a near doubling of capital per worker.  The stock of education per worker, measured by years of schooling completed, grew at quite high rates in most developing country regions, although it grew very slowly in the industrial countries.  This was particularly the case for secondary and tertiary education stocks, which grew extremely rapidly in most developing country regions. The 9.2 percent annual growth in the stock of secondary school education in East Asia, for instance, implies a fourteen-fold increase in this stock over thirty years. 

Before placing too much emphasis on the apparent increases in capital and in education per worker in developing countries as indicators of changes in factor endowments, it is important to examine the capital-output ratio (K/Q). One of Kaldor's key stylized facts of economic growth was a constant capital/output ratio and a rising capital/labor ratio. This is frequently interpreted to imply that technical change is Harrod-neutral, with capital per worker increasing in line with effective labor. If true, this would imply an absence of changes in effective factor endowments, implying no long run changes in factor endowment ratios, and hence no role for Rybczynski effects. 

In fact, it appears from Table 1 that the physical capital-output ratio increased quite substantially over the period in both developing and industrial countries. For human capital, the education to output ratios have increased substantially in developing countries, but fallen quite rapidly in developed countries. These results have potentially important implications for our interpretation of the process of growth and structural change. Before going too far, however, it is important to check the Nehru et al data against other data sets to ensure that these results are not merely artifacts of the data construction process. A check against the well-known Penn World Tables data (see www.nber.org) for a range of countries suggests that physical capital/output ratios were generally rising quite rapidly in the 1970-1990 period for which these capital accumulation data are available. The fact that the growth rates of K/Q and Edn/Q are generally lower than their growth relative to the labor input does, however, give reason for caution about common assumptions, such as Hicks-Neutral technical change in all sectors used by Harrigan and Kee.

Despite the evidence from many different types of empirical studies on the potential role of Rybczynski effects, the coincidence of high rates of accumulation of physical and human capital over the period, and the rapid shift of developing countries into exports of manufactures and services, is clearly not definitive evidence of causation.

However, it is strongly suggestive, and needs to be examined in conjunction with changes in trade policy and in technology.