Nonlinearity in the Return to Education
This study estimates marginal rates of return to investment in schooling in 12 countries. Significant systematic nonlinearity in the marginal rate of return is found. In particular, the marginal rate of return is increasing significantly at low levels of education, and decreasing significantly at high levels of education. This may help explain why estimates of the return to schooling are often... (ver resumen completo)
I. IntroductionThe rate of return to education has been estimated in literally hundreds of studies (see the surveys by Psacharopoulos. 1985, 1994; Ashenfelter et al., 1999; and Harmon et al. 2000). The vast majority of this work implicitly assumes that the marginal rate of return is constant over all levels of education. Some studies, however, found significant nonlinearity in the rate of return to schooling. Most of this work focused on deviations from nonlinearity at particular levels of education; that is, sheepskin effects (see, for example, Hungerford and Solon, 1987; Belman and Heywood, 1991; and Jaeger and Page, 1996). Perhaps as a result, evidence on the general nonlinearity in the return to schooling appears inconsistent. Mincer (1974), Psacharopoulos ( 1985, 1994), and Harmon and Walker (1999) showed significant diminishing returns to education.1 Heckman and Polachek (1974), Card and Krueger (1992), and Card (1995, 1999) argued that the rate of return appears roughly constant. One could, however, interpret Card and Krueger's (1992) results as indicative of increasing returns at low levels of education. The results in Heckman et al. (2003) suggest increasing returns at low levels of education followed by diminishing returns at high levels of education. The general nature of possible nonlinearity in the return to education is unclear.This study tests for the general nonlinearity in the (private) rate of return to education for working-age men using comparable micro data in 12 countries. The data indicate that the marginal rate of return is essentially nil for the first several years of schooling, it then increases rapidly until about year 12, and then it declines.II. DataData from the International Social Survey Programme (ISSP) are used. The ISSP contains comparable cross-sectional data on individuals in 33 countries from 1985 through 1995 (most of the countries, however, only participated in a few of the years). Only 13 of the countries have at least 1,000 observations of labor-market data for men, and measured schooling is truncated between 10 and 14 in one of these countries (Great Britain). Thus observations from Great Britain are excluded, leaving samples from 12 countries.The 12 samples consist of men within the ages of 18 to 64; without missing information on wage rates or education; and not self-employed, retired, or in school. A handful of observations with more than 22 years of measured education are also excluded.2 Table 1 lists for each country its: sample size, number of cross sections, mean years of education, and standard deviation...
Para continuar leyendoSOLICITA TU PRUEBA GRATIS
COPYRIGHT GALE, Cengage Learning. All rights reserved.