By ERIC A. HANUSHEK
DENNIS D. KIMKO*
Direct measures of labor-force quality from international mathematics and science test scores are strongly related to growth. Indirect speciﬁcation tests are generally consistent with a causal link: direct spending on schools is unrelated to student performance differences; the estimatedgrowth effects of improved labor-force quality hold when East Asian countries are excluded; and, ﬁnally, home-country quality differences of immigrants are directly related to U.S. earnings if the immigrants are educated in their own country but not in the United States. The last estimates of micro productivity effects, however, introduce uncertainty about the magnitude of the growth effects. (JELO40, I20, J24)
Recent theoretical analyses of international differences in growth rates have focused attention on the role of human capital. Most crosscountry empirical studies of long-run economic growth now include some proxy for human capital, and these are invariably signiﬁcant. Data limitations have, however, forced severe compromises. Paralleling analyses of wage determination, empiricalimplementation virtually always employs some readily available measure of the quantity of formal schooling to reﬂect human capital, but this appears inadequate. The analysis of international differences in growth rates here suggests that math and science skill is a primary component of human capital relevant for the labor force. Such cognitive skill of a population is not well proxied by measures ofschool quantities or measures of resources devoted to schools. Accounting for differences in labor-force quality signiﬁcantly improves our ability to explain growth rates. Two issues arise in considering the effect of human capital on economic growth: how should any relationship be speciﬁed and how should
* Hanushek: Hoover Institution, Stanford University, Stanford, CA 94305, and NationalBureau of Economic Research; Kimko: Institute for Defense Analyses, 1801 North Beauregard Street, Alexandria, VA 22311. We have beneﬁted from comments and discussion with Mark Bils, Byron Brown, Stanley Engerman, Douglas Hodgson, Peter Klenow, Paul Romer, Sergio Rebelo, Michael Wolkoff, two very helpful referees, and participants in seminars at Boston University, Columbia University, the RandCorporation, the University of Oregon, and the University of Rochester. Jin-Yeong Kim provided excellent research assistance. 1184
human capital be measured? The focus of this paper is the second issue. This paper does not consider alternative formulations of the underlying growth relations but instead is a direct application of models of endogenous growth developed theoretically by a variety of people(e.g., Richard R. Nelson and Edmund Phelps, 1966; Paul Romer, 1990a; Sergio Rebelo, 1991). In the simplest formulation, growth rates are affected by ideas and invention, which in turn are related to the stock of human capital either through research and development (R&D) activities or through adoption behavior. These formulations indicate not only why the level of output is higher when a countryhas more human capital but also why the growth rate is higher. Previous investigations of growth have concentrated on various measures of formal schooling activities as proxies for relevant human capital. The most frequently employed measure is either the primary- or secondary-school enrollment rate, used, for instance, in Romer (1990b), Robert J. Barro (1991), and N. Gregory Mankiw et al. (1992)and highlighted in the inﬂuential sensitivity studies of Ross Levine and David Renelt (1992) and Levine and Sara J. Zervos (1993). These schooling ﬂow variables, however, will not accurately represent either the relevant stock of human capital of the labor force or even changes in the stock during periods of educational and demographic transition. To deal with these problems, Barro and JongWha...