Mary Jean Bowman

Schultz, Denison, and the Contribution of "Eds" to National Income Growth

N 1961, as part of his pioneering work on education as an investment, T. W. Schultz estimated that education accounted for 21-40 per cent of national income growth in the United States over the period 1929--56; increases in education per member of the employed labor force accounted for 17-33 per cent of income growth over the same period.1 In 1962, Edward Denison published a study that attempted to assess the roles of a long list of factors contributing (positively or negatively) to national income growth in this country from 1907 to 1929 and from 1929 to 1957, and to project the growth potentials from these various sources over the period 1957-80. For the interval 1929-57 he credited increased education per member of the employed labor force with 23 per cent of the national income growth rate, a figure that falls between the 17 and 33 per cent estimates reached by Schultz.2 Both of these studies are being widely quoted and misquoted, and all too often misused, around the world. Meanwhile

An Integrated Framework for Analysis of the Spread of Schooling in Less Developed Countries.

The first and main task of this paper is the laying out of a framework in which microeconomic decision theory and elements of information and communication theory drawn from human geography and sociology are joined in an integrated approach to the analysis of the spread of schooling among the populations of less developed countries. Given the diversity of socioeconomic structures and cultures, this requires a framework that is at once of broad generality and yet sufficiently selective and precise to permit the specification of well-focused, testable hypotheses. The conceptual generality of the main features of the model serve, meanwhile, to sharpen the conceptualization of its components in terms of commonalities running through the differences among societies. The second major part of this paper illustrates and tests the performance of the model in analyses of area variations in enrollments and school attainments in India, Brazil, and Mexico. Although some of this research has dealt with changes over time (for India, over a half-century), the comments here will refer only to analyses for single points in time. This makes it possible to compare some of the findings quite directly. It means, however, that longitudinal dimensions lying behind an essentially dynamic development process (and in some cases its stagnation) get less attention than they deserve. The problem of moving between analysis focused on variations among individuals (using households as the units of observation) and variations among aggregated subpopulations (in the present case, using observations on populations classified by area of residence) is fully recognized. It is avoided here by careful specification and selection of variables, by the fact that our interest is in population aggregates, by the fact that observations are on fairly small area populations, and by the explicit adaptation of the model to such empirical treatment. Ideally, one would wish to work with data from individual households along with observations on entire subpopulations, but such matched data are rare, and for our purposes the