Loading field map
Why do some workers earn more than others? Why do investments in education and training pay off differently across countries and over time? These questions have driven a subfield that began with a simple but powerful idea—that skills are a form of capital—and then fractured into competing accounts of what skills really measure, how they are shaped by technology and institutions, and whether their value can be captured by market earnings alone. The history of human capital and skill formation is a story of successive challenges, each framework emerging from the blind spots of its predecessors, and of a field that today operates as a pluralistic toolkit rather than a single orthodoxy.
Human Capital Theory, formalized by Theodore Schultz, Gary Becker, and Jacob Mincer in the 1960s, treated education and on-the-job training as investments that raise a worker’s productivity. The Mincerian earnings function—log earnings as a function of schooling and experience—became the workhorse of empirical labor economics. The framework’s core claim was that wage differentials reflect genuine productivity differences created by skill acquisition. For two decades, this view dominated the subfield.
But the 1970s brought two direct challenges. The Screening Hypothesis (also called the signaling model, associated with Michael Spence and Kenneth Arrow) argued that education does not necessarily enhance productivity; instead, it serves as a costly signal of pre-existing ability. Employers use diplomas to sort workers, and the correlation between schooling and wages may reflect sorting rather than causation. The Job Competition Model (Lester Thurow) went further: wages are attached to jobs, not individuals, and workers queue for positions based on their trainability. Education helps a worker move up the queue but does not change the job’s wage. Both frameworks preserved the empirical pattern that Human Capital Theory explained—more schooling, higher earnings—but offered radically different causal stories.
The resulting debate was not resolved by a knockout blow. Instead, the frameworks entered a state of living disagreement. Empirical strategies (twins studies, instrumental variables, natural experiments) attempted to isolate the causal effect of schooling on earnings, and the consensus that emerged is that both productivity-enhancing and signaling effects exist, with their relative importance varying by context. Human Capital Theory remains the default framework for estimating returns to education, but the Screening Hypothesis permanently complicated its interpretation.
In the 1980s, the Capability Approach, developed by Amartya Sen and Martha Nussbaum, introduced a fundamentally different yardstick. Rather than asking how much education raises earnings, it asked what education enables a person to be and do—the freedoms and functionings that matter for human flourishing. This framework did not reject Human Capital Theory so much as narrow its scope: human capital is valuable, but only as one input into a broader set of capabilities. The Capability Approach shifted attention to non-market outcomes (health, political participation, social inclusion) and to the distribution of opportunities within households and communities. Methodologically, it resisted the single metric of earnings, advocating for multidimensional indicators. While it never displaced the economic frameworks in mainstream labor economics, it created a persistent normative critique and influenced policy evaluation, especially in development and education.
The 1990s brought a new puzzle: why did wage inequality rise sharply in the United States and other advanced economies, even as educational attainment increased? Skill-Biased Technical Change (SBTC) provided an answer. Building on the human capital tradition, SBTC argued that new technologies—especially computers—complement highly educated workers while substituting for less skilled labor. The framework extended Human Capital Theory by making the returns to skill depend on the pace and direction of technological change. SBTC became the dominant explanation for rising inequality, supported by evidence that the college wage premium grew alongside computer adoption.
Yet SBTC treated technology as a largely exogenous force and paid little attention to national institutions. The Varieties of Capitalism (VoC) framework, articulated by Peter Hall and David Soskice in 2001, offered a complementary institutional layer. VoC distinguished between liberal market economies (e.g., the US, UK) and coordinated market economies (e.g., Germany, Sweden), arguing that firms and workers invest in different types of skills depending on the institutional environment. Coordinated economies foster industry-specific and firm-specific skills through apprenticeship systems and employment protection, while liberal economies encourage general skills and labor mobility. VoC did not replace SBTC; rather, it absorbed the technology story into a broader institutional account, showing that the same technological shock can produce different inequality outcomes depending on how skill formation is organized nationally.
By the early 2000s, SBTC’s simple dichotomy between high- and low-skill workers began to show cracks. The Task-Based Approach, pioneered by David Autor, Frank Levy, Richard Murnane, and later Daron Acemoglu and Pascual Restrepo, refined and partially replaced SBTC by shifting the unit of analysis from skills to tasks. Instead of asking whether a worker is high- or low-skill, it asks what tasks a job involves—routine or non-routine, manual or cognitive, abstract or service-oriented. This granular lens explained a phenomenon SBTC could not: job polarization, where middle-skill routine jobs (clerical, production) declined while high-skill abstract and low-skill service jobs grew. The Task-Based Approach also made technology endogenous: automation replaces routine tasks, but it can also create new tasks that complement labor. It transformed the subfield’s vocabulary, making “routineness” and “task content” standard empirical tools.
Today, no single framework dominates. Instead, researchers select tools based on the question. Human Capital Theory remains the workhorse for estimating private returns to education and for analyzing training investments. The Screening Hypothesis and Job Competition Model inform studies of credential inflation and labor market queues. The Capability Approach guides evaluations of education’s broader social benefits and is influential in development economics. SBTC is still used for long-run inequality trends, though it is often combined with the Task-Based Approach for more precise accounts of automation and polarization. Varieties of Capitalism provides the leading comparative framework for understanding cross-national differences in skill systems and their interaction with welfare states.
What the leading frameworks agree on is that skills matter for labor market outcomes, but their value is mediated by technology, institutions, and measurement. They disagree on whether education primarily builds or signals productivity, whether technology is an independent driver or shaped by policy, and whether institutional variation is best captured by VoC’s binary typology or by more continuous measures. The subfield’s vitality comes from this pluralism: each framework highlights a different facet of skill formation, and the most compelling empirical work often draws on several at once.