Per capita income, human capital, and inequality convergence: A latent-variable approach

Journal of Agricultural and Applied Economics, 2003 by Deepak, Sri Devi, Seale, James L Jr, Moss, Charles B

The purpose of this paper is to empirically analyze determinants of income-level convergence. Specifically, the effect of human capital on per capita income is estimated for 22 countries of the Organization for Economic Cooperation and Development (OECD). Additionally, the effects of openness in international trade and investment and government expenditures on per capita income are estimated and evaluated. Human capital is modeled as a latent variable, and results indicate that it is a significant factor in explaining the variation of per capita income levels among the OECD countries. Further, the entire time path of human capital is utilized to explain deviations in per capita income.

Key Words: convergence, human capital, inequality, latent variable, OECD

Whether or not countries are converging in terms of per capita income levels is an important question.1 A direct way to answer this question for a group of countries is to measure income inequality among countries over time. Much of the recent work directly measuring convergence or divergence of per capita income levels has been done by Theil and associates at the University of Florida (Theil 1989; Theil and Deepak 1994, 2002a,b; Seale, Theil, and Deepak; Moss, Theil, and Deepak; and Theil and Seale). Of these, Theil (1996) and Theil and Seale are the most comprehensive, and Theil and Seale document that, over the period 1950-1990, the relatively rich northern countries are converging, South-East Asian countries are diverging, and tropical Africa has tended to converge during periods of income growth and to diverge during periods of negative growth.

Although answering the convergence question directly, the above studies do not analyze the determinants of convergence.2 Several recent papers have looked at what factors determine income growth using regression analysis (e.g., Barro; Barro and Sala-i-Martin; and Mankiw, Romer, and Weil). Others including Lucas (1988, 1993) have developed theoretical models of endogenous growth to study the determinants of income growth. (See also Rebelo; Tamura; and Romer 1994.)

Barro analyzes convergence in 98 countries during 1960-1985 by studying the relationship among growth rates in per capita income to levels of per capita income and initial levels of human capital. He uses school enrollment rates during 1960 as a proxy for the level of human capital and finds that, holding levels of human capital constant, the growth rate in per capita income is inversely related to the level of per capita income. Further, holding the initial level of per capita income constant, results indicate there is a positive relationship between the growth rate of income and the level of human capital. Therefore, convergence is evident only in countries with high levels of initial human capital.

This paper departs from Barro's work in several important ways. While Barro's study was based on cross-sectional data, this study utilizes pooled data (1955-1990) for 22 OECD countries.3 Second, unlike Barro, we hypothesize that income convergence is conditional on convergence of its determinants and not just on initial starting values. Third, we construct a multiple-indicator index of human capital utilizing a latent-variable approach (Bollen) and estimate per capita income over time as a function of human capital, international openness, investment expenditures, and government expenditures. Finally, we answer the convergence question directly for the 22 OECD countries over the years 1955-1990 by utilizing Theil's inequality measure.

In the following sections, we describe the data, discuss the latent-variable model, and present the latent-variable-estimation results. We also present evidence on income convergence, discuss inequality measures over time for the hypothesized determinants of real per capita income levels, and draw conclusions based on the evidence.

Data

The three sources of data for this study are the Supplement to Mark 5 or Penn World Tables (PWT5.5) compiled by Summers and Heston; Basic Facts and Figures compiled by the United Nations Educational, Scientific and Cultural Organization (UNESCO 1952-1962); and the Statistical Yearbook compiled by UNESCO (1963-1993). The data on income, population, international openness, government expenditure, and investment expenditure are extracted from the Summers and Heston data. All the expenditures are in real terms (1985 international prices) and span the years 1955-1990. The unique feature of these data is that the expenditure entries are denominated in a common currency so that real international quantity comparisons can be made both among countries and for a particular country over time.

Real per capita income (Y) is taken directly from the Supplement, as is population (POP). Real per capita government expenditure (G) and real per capita investment (I) are computed from the Supplement. Data for openness (O), which is represented by the real per capita sum of exports and imports, is also computed from the Supplement.