In 1970, economists Jacob Mincer and Gary Becker faced a problem that would haunt their field for half a century. They'd shown that people with more education earned more money—a lot more. But skeptics had a devastating question: What if smarter, more motivated people simply stayed in school longer? What if education was just a signal, not the cause?
The causality debate has now been settled. A 2026 analysis by Harry Patrinos and George Psacharopoulos examined 191 estimates from 145 studies across 54 countries, using the strongest causal methods available. The verdict: each additional year of schooling increases lifetime earnings by roughly 10%, and this is a genuine causal effect, not just correlation.
The Measurement Challenge
The problem with measuring education's true impact is deceptively simple. Students who finish college differ from dropouts in countless ways—family wealth, innate ability, work ethic, connections. Traditional studies comparing their earnings might capture these pre-existing differences rather than education itself.
Enter instrumental variables, a technique that exploits natural experiments. When Indonesia built 61,000 primary schools in the 1970s, children in regions with more construction got more education—but not because they were smarter or richer. When Britain raised its school-leaving age from 14 to 15 in 1947, students born just after the cutoff date had to stay in school longer through no choice of their own. These policy changes act as randomizers, isolating education's causal effect from confounding factors.
The Patrinos-Psacharopoulos analysis synthesized decades of such studies. The finding that instrumental variable estimates average 9.7%—38% higher than conventional methods—upends what researchers expected.
The Ability Bias Reversal
For years, economists assumed conventional estimates overstated education's benefits. The theory seemed intuitive: more capable students stayed in school longer, inflating the apparent returns. Causal methods should reveal lower, truer effects.
The data tells the opposite story. In nearly 80% of cases, causal estimates exceed conventional ones. The pattern is strongest in developing countries, where instrumental variable estimates often dwarf traditional calculations.
Two forces explain this reversal. First, measurement error. When students misreport their education levels or when credentials don't perfectly capture learning, conventional methods underestimate true returns. Instrumental variables correct this downward bias.
Second, and more profound: the students most affected by policy interventions—compulsory schooling laws, school construction programs, scholarship expansions—are precisely those who benefit most from additional education. These "marginal" students, often from disadvantaged backgrounds, weren't planning to continue their education. When policy changes pushed them to stay in school, their earnings jumped dramatically.
Who Gains Most
A first-generation university student in rural Indonesia sees larger returns than a doctor's child in Jakarta. A teenager in 1947 Britain forced to stay in school until 15 experiences bigger earnings gains than a peer who would have attended anyway.
This pattern aligns with economic theory about scarcity rents. Where education is rare, an additional credential commands a premium. In wealthy countries with near-universal secondary education, one more year of high school adds less value. In countries where most workers lack basic schooling, that same year can transform lifetime prospects.
The implications for economic mobility are direct. Education doesn't just raise average earnings; it disproportionately benefits those at the bottom. A construction worker's daughter who finishes secondary school doesn't just earn more than her parents—she often earns substantially more than conventional estimates would predict.
This finding contradicts the popular narrative that education has become a mere sorting mechanism, shuffling the privileged into high-paying jobs while leaving others behind. The causal evidence suggests education still functions as an engine of mobility, particularly for those who wouldn't otherwise access it.
The Diminishing Returns Gradient
The global pattern reveals an economic gradient. Low-income countries show the largest gaps between causal and conventional estimates. Middle-income countries fall in between. Wealthy nations show smaller differences.
This isn't just about data quality. It reflects genuine economic forces. In Ethiopia, where average schooling remains below six years, each additional year generates substantial returns. In South Korea, where tertiary enrollment exceeds 70%, marginal returns are smaller.
The pattern has a policy corollary: education investments yield highest returns where they're currently scarce. Expanding access in underserved regions doesn't just promote equity—it's economically efficient.
Beyond the Wage Premium
The 10% figure captures only direct earnings effects. Education's full causal impact extends further. Better-educated workers create spillovers, raising productivity of those around them. They're more likely to innovate, start businesses, and adapt to technological change. They experience better health outcomes, longer lives, and greater civic participation.
These broader effects resist precise measurement but compound over lifetimes and generations. A parent's education affects children's outcomes independent of income. Communities with higher education levels show lower crime, stronger institutions, and faster economic growth.
Rethinking Education as Infrastructure
The causal evidence reframes education policy. If education merely signaled pre-existing ability, expanding access would be redistributive—helping individuals but not growing the overall pie. The causal findings show education genuinely increases productivity, making it an economic investment comparable to physical infrastructure.
This distinction matters for policy design. Infrastructure investments require upfront costs but generate returns over decades. The same logic applies to education. Countries that expanded schooling in the 1970s and 1980s—South Korea, Taiwan, Indonesia—saw sustained economic growth as those cohorts entered the workforce.
The causality debate consumed five decades because the stakes were high. Correlation could justify education spending on equity grounds alone. Causation makes it an economic imperative. The 2026 evidence closes the case: education doesn't just sort winners from losers. It creates winners.