The Quality of Education and Cohort Variation in Black-White Earnings Differentials: Comment.
使用30-55岁男性调查数据，检验控制教育质量后，黑人与白人教育回报率的差异是否缩小。结果发现教育质量对收入有显著影响，但控制质量后种族间回报率差距仍然很大，且不同年龄组间差距基本不变。
Historically, empirical estimates of the returns to years of education have consistently indicated that the percentage returns to whites exceed those to blacks. Possible explanations for the gap include discrimination and differing quality of education obtained during the years of attendance. In his 1973 article in this Review, Finis Welch suggested that differing quality of education is a major factor causing the observed gap. He provided evidence that education quality of young blacks and whites had been more equal than for older cohorts, and that young blacks who entered the labor force in the 1960's were experiencing a larger percentage return to an added year than were similar young whites. An implication of acceptance of Welch's hypothesis, combined with the reality of essentially equal returns for the equally educated young blacks, would be equal estimated returns to older cohorts in an empirical model correctly specified to control for socioeconomic backgrounds, experience, education years and education quality, Unfortunately Welch had no quality data for his sampled earners and therefore was unable to test directly for a major reduction in the gap for older cohorts that would be provided by correcting for the quality of their education. Charles Link, Edward Ratledge, and Kenneth Lewis (hereafter, L-R-L) empirically tested a model in which quality is included by interacting it with years of education. Unfortunately they do not include either years or quality in a noninteracted form.' Most important, they use a sample containing 17-27-year-old males and are therefore unable to provide much insight into the effect of quality on older groups. In this paper we use survey data for a group of men ages 30-55 to test whether an empirical model including quality of education information will predict a smaller gap in returns to years of education for blacks and whites than will a model with no controls for quality. Because of the age range of our sample we are also able to compare the differences in rates of return by race for different age groups. Our results suggest that quality of education is a significant determinant of earnings for both races, but that quality-controlled returns to years of schooling still differ considerably for blacks and whites. We find a sizable gap between young blacks and whites which remains more or less constant over age groups. Our best point estimates suggest that the gap may narrow slightly for older groups.