Journal Issue: Opportunity in America Volume 16 Number 2 Fall 2006
Does the Economic Value of Education Differ by Family Background?
We first examine whether education has a different value for people of different socioeconomic backgrounds. If children from more advantaged families receive larger gains from each additional year of schooling, they will have a greater incentive to stay in school. Because research on the economic value of education is extensive, while that on the extent to which that value varies by family background is more limited, we begin by discussing the overall relationship between education and income.
Estimating the Economic Value of Schooling Is Not Straightforward
Economists conventionally measure the economic value of additional schooling (or the “return to schooling”) as the average percentage difference in mean earnings for each additional year of education.11 Estimates based on the Current Population Survey, for example, suggest that on average for each year of schooling, a person's earnings increase by about 11 percent.12 While the economic value of education has been well documented, the question of why education increases income is more controversial. Nobel Laureate Gary Becker theorizes that education provides skills, or human capital, that make a worker more productive.13 If so, then because a worker's income reflects his or her productivity, education is a key determinant of upward social mobility. It follows that much of the gap between the rich and the poor arises from a lack of skills among the poor—with the policy implication being that education and training should form the cornerstone of programs aimed at reducing income inequality.
Other researchers, such as Nobel Laureate Michael Spence, argue that education may not generate higher incomes—that is, the relationship may not be causal.14 Instead, education and income may be linked because people with greater “ability” complete more schooling and would likely earn higher wages and salaries even without the additional schooling. In this case, as with the relationship between family socioeconomic status and a child's educational attainment, the schooling-income connection may mostly reflect the fact that more able people command a premium for their (innate) skills in the labor market. Thus empirical estimates of the return to schooling such as the one just noted are too large. In this view, increasing funding for educational programs for the disadvantaged will have little or no effect because schooling cannot change innate ability.
Again, researchers have developed several methods to isolate the economic value of education in an effort to disentangle these two hypotheses. To determine definitively whether more schooling raises income, an ideal experiment would involve randomly assigning one group of students to complete high school and another group to drop out, regardless of the students' innate ability or family background. Years later researchers would compare how the two groups fared in the labor market. On average the only difference between the two would be whether they had graduated from high school. Differences in the earnings of the two groups would provide an estimate of the economic value of education— how much completing high school causes earnings to increase. To determine whether this economic value varies by family background, the researcher could simply estimate the earnings difference for subgroups of students based on their family background at the start of the experiment.
Empirical Estimates of the Economic Value of Schooling
Recognizing that no such experiment will ever be conducted, researchers have developed two broad approaches to empirical estimation of the economic value of education. The first approach—so-called natural experiments— locates events or policies that might be expected to alter the schooling decisions of some people, but would not be expected to alter their income independently. The idea is straightforward. Suppose that researchers knew of an event or policy, such as an increase in the compulsory schooling age, that would increase a group's years of completed schooling. Suppose, further, that they were certain that the policy would have no direct effect on the group's earnings. They would then estimate the effect of education on earnings in two steps. First, they would estimate how much the policy increased the group's educational attainment. Next, they would measure how much the same policy affected their earnings. If they find that the group's earnings have increased, they can be sure that education caused the increase because they are certain the policy had no direct effect on earnings. The ratio of the increase in income to the increase in schooling is an estimate of the economic value of education. Many such studies estimate that the return to schooling is at least as large as estimates by conventional procedures that relate the level of schooling to income directly.15
Other researchers have used sibling or twin pairs to estimate empirically the return to schooling. Because siblings and twin pairs share genetic material and are raised in similar household environments, their “ability” and other unobservable characteristics are much more similar than those of randomly selected members of the population. As a result, when researchers relate differences in siblings' schooling to their earnings, they implicitly account for these unobserved factors. Although the estimated return to schooling varies because of the widely different time periods covered by the studies, the various sibling and twin studies find a significant link between schooling and earnings.16 Further, the more recent and more sophisticated estimates typically do not differ from conventional estimates of the return to schooling.17
The findings of all these empirical studies— those using natural experiments and those using family relationships—are surprisingly consistent: the return to schooling is not caused by an omitted correlation between ability and schooling. A conventional estimate of the economic value of education is thus likely to be quite close to that of the ideal experiment. In fact Nobel Laureate James Heckman, writing with Pedro Carneiro, concludes, “By now there is a firmly established consensus that the mean rate of return to a year of schooling, as of the 1990s, exceeds 10 percent and may be as high as 17 to 20 percent.”18
Do Differences in the Value of Education Explain Differences in Educational Attainment?
Although researchers consistently find that education has a causal effect on earnings— that education has economic value—they have not come to a consensus on whether that value varies depending on an individual's family background. Importantly, they have not established whether people from more advantaged families complete more schooling because it has greater value for them. One study, for example, concludes that individuals with higher “ability” or from more advantaged families do not enjoy greater returns to schooling.19 Other studies find no variation in the returns to schooling by the race or ethnicity of the individual, or by IQ.20 Still others, however, find higher returns to schooling for more able individuals.21 Another important question is why the return to schooling might differ by family background. Differences in school quality, which we address below, provide one possible explanation.