It makes sense, as Becker points out, that increased earnings of graduates of top colleges will attract more applicants to those colleges, enabling the colleges to be more selective. Furthermore, as colleges stopped discriminating against Jews and Asians, who tend to be the best students, the quality of the applicant pool rose. In addition, colleges want to have wealthy alumni, and if the financial return to IQ rises because of the increased complexity and sophistication of the economy, high IQ students become increasingly attractive as future alumni and therefore potential donors.
Another causal factor may be the correlation between parental income and kids’ IQs. Obviously not all wealthy people are smart and have smart children, but in general there is a positive correlation between income and IQ and between parental IQ and children’s IQ. As the advantages of attending a top college became more apparent, well-to-do parents began investing more in tutoring for their kids and in sending their kids to top (and very expensive) private schools, and this further increased the quality of the college applicant pool.
But this does not explain the relation between having attended a top college and having a higher income than students with similar intellectual promise who attended a lower-ranked college. The possibility explored by Becker, and by Caroline Hoxby, whose study he cites, is that a student benefits from the company of very bright students. One can learn from one’s fellow students as well as from one’s teachers, and moreover the presence of very bright students may stimulate the teacher to greater exertion in teaching. But at least three alternative possibilities should also be considered. The first is that potential employers consider graduation from a top college a certification of the graduate’s ability. The second is that what students mainly learn from their fellow students is social rather than intellectual—the students at these schools tend to come from prosperous, sophisticated households, which equips them to succeed in fields in which social sophistication (charm, poise, self-confidence, articulateness, etc.) is an employment asset, as it is in many fields. And third is that students at the top schools make valuable contacts—with their classmates who are marked for success. These contacts are business assets.
Becker makes the interesting point that insofar as a student’s future earnings are bolstered by reason of his attending a college that has many bright students, the future earnings of those students are reduced because the students face more competition than they would if less-bright students were not admitted. But this effect could be offset by the external benefits generated by those students. The student whose employment prospects are enhanced from what he learns from his fellow students may turn out to be a highly successful entrepreneur who business activities generate lucrative employment or investment opportunities for other top-college graduates.
Even though the greater selectivity of the top colleges increases the earnings of their graduates, and so, prima facie, their contribution to aggregate economic welfare, a question remains whether some alternative method of sorting applicants to colleges would be more productive. I am dubious. Suppose high-school graduates who passed a test designed to determine whether a student would benefit from a college education were assigned to colleges randomly. Parents would invest less in tutoring and other forms of college preparation for their children. Colleges would employ tracking to segregate the brighter students from the less bright ones, which would reduce the opportunities for the less bright students to learn from the brighter ones. Colleges would compete less vigorously for top-flight faculty, as better faculty would not attract better students.
Moreover, if the benefits (for example in higher lifetime earnings) of association between bright students are multiplicative rather than additive, random assortment will reduce those benefits. (Becker has made a similar point with respect to assortative mating, in his work on the economics of the family.) Suppose that A is a 4 in intelligence or other attributes and B a 3, while C is a 2 and D a 1.5. If B studies, works, or converses with A, their total productivity will be 12 (3 x 4); then D will be assigned to C and their total productivity will be 3 (1.5 x 2). The grand total will thus be 15 (12 + 3). If instead D were assigned to A and C to B, the grand total would be only 12 (1.5 x 4 + 2 × 3). I think this is likely to be the case in the sorting of students. When there is a significant disparity in ability, communication will be difficult and neither will learn much from the other.
I conclude that the current system is superior to one that would try to homogenize the quality of students accepted to different colleges.