The Dale-Krueger Study

In the AI Story

The study addressed a problem that had plagued earlier research on the returns to college selectivity: the observed earnings premium for graduates of elite institutions could be driven either by the institutions' educational contribution or by the pre-existing qualities of the students they admitted. Dale and Krueger used a novel identification strategy — comparing students admitted to the same set of elite institutions who then made different attendance choices — to isolate the institutional effect from the selection effect.

The finding was stark. Once the admission decisions were held constant, the institution attended had little or no effect on subsequent earnings. Students who were admitted to Harvard and chose Penn State earned roughly the same as students who were admitted to Harvard and attended Harvard. The exception was for disadvantaged students — Black, Hispanic, and first-generation college students — for whom attending the more selective institution did produce a measurable earnings benefit, likely through the network effects that provided access these students lacked in their pre-college lives.

The implications were uncomfortable for elite universities. If the educational contribution of Harvard versus Penn State was negligible for most students, then Harvard's price premium reflected not superior education but superior signaling — and the institution's admissions office was its most economically important unit, not its classrooms. The finding reinforced signaling theory as the dominant economic account of higher education's value and complicated every subsequent claim that elite institutions justified their costs through pedagogical excellence.

The study's relevance to the AI era is direct. If the degree's value was always primarily about selection rather than education, then AI's threat to the educational function is less existential for elite institutions than for the broad middle of higher education that depended on the bundled package of moderate selection and moderate training. The credential reckoning hits hardest where the Dale-Krueger logic applies least — at institutions whose credentials never commanded the network-effect premium that pure selection provides.

Origin

Dale was a researcher at the Andrew W. Mellon Foundation; Krueger was a Princeton economist who later served as chair of the Council of Economic Advisers under Obama. The paper used the College and Beyond database, which tracked students who applied to selective colleges in 1976 and 1989, with follow-up earnings data extending into the early 2000s. Krueger extended the analysis in a 2014 paper with Dale that confirmed the findings on a more recent cohort.

Key Ideas

Admission beats attendance. Students admitted to elite schools earned the same whether they attended them or less selective alternatives.

Exception for the disadvantaged. Black, Hispanic, and first-generation students did benefit from attending more selective institutions — likely through network access.

Selection, not education. The institutional contribution to earnings was negligible once pre-existing student qualities were controlled for.

Signaling theory confirmed. The empirical finding reinforced Spence's theoretical account of how credentials actually work.

Implications for AI era. Elite institutions' credential value is relatively insulated; middle-market institutions, selling primarily education-as-bundled-training, face sharper displacement.