The Skill Premium

In The You On AI Field Guide

The rising skill premium was the central fact of labor economics in the 1990s and 2000s. Goldin and Katz's influential 'race between education and technology' framework argued that the premium rose when technology demand for skills outpaced education supply. The policy conclusion — invest in college education to match technology's demand — shaped a generation of educational policy. The framework explained rising inequality as a consequence of educational lag, remediable by producing more college graduates.

AI disrupts this framework in two directions. First, it automates tasks that previously required college-level cognitive skills (legal research, medical image analysis, software development at the junior level), potentially reducing demand for college graduates in these fields. Second, it complements tasks that require skills not captured by the college credential (judgment, creativity, interpersonal ability, entrepreneurial initiative), potentially increasing the premium for these specific capacities regardless of formal credentials. The result may be a redefinition of the skill premium: no longer keyed to education, but to specific non-routine capabilities that education may or may not develop.

Segal's observation in You On AI that the developer in Lagos now has access to the same coding leverage as the Google engineer is a case in point. The coding skill that commanded a premium because it was scarce is becoming abundant; the judgment skill that determines which code to write is becoming the binding constraint. The skill premium is not disappearing; it is migrating to a different skill. Whether educational institutions can identify and develop the new skill is an open question.

Origin

David Autor

The systematic measurement of the skill premium and its historical trajectory is most associated with Claudia Goldin and Lawrence Katz's 2008 The Race Between Education and Technology. Autor's contribution has been to connect the skill premium to the task-based framework, showing that the premium reflects which tasks technology is currently complementing versus substituting.

Key Ideas

The premium is task-specific. The college premium reflects demand for the non-routine cognitive tasks college graduates disproportionately perform — when AI automates those tasks, the premium for those graduates falls.

The premium has plateaued. After four decades of growth, the US college wage premium stopped rising around 2015, suggesting the race between education and technology may be entering a new phase.

The premium is shifting in composition. Within the college-educated workforce, the premium is now accruing differentially to workers whose specific skills (judgment, creativity, interpersonal ability) complement rather than compete with AI.

Credentials may lag capabilities. The skill premium may increasingly detach from formal education and attach to demonstrated capacity in AI-complementary tasks, accelerating a credential reckoning.