In most high-income countries, girls now outperform boys on a majority of academic measures. They achieve higher grades on average, are less likely to leave education early, and enrol in and complete university at higher rates.
This is a genuine reversal and a fairly recent one. In the 1960s the direction ran the other way in almost every one of those countries.
It has not produced earnings parity. The median woman still earns less than the median man across essentially every developed economy. The two facts are frequently presented as a paradox, and they are not one — but resolving them requires being specific about what each measure captures.
What the education advantage consists of
The attainment advantage is real but uneven, and its shape matters.
It is largest in verbal and language-based assessment, present in most humanities and social sciences, roughly neutral in mathematics in most recent measurements, and reversed in advanced physical sciences and computing where enrolment gaps persist.
It is also, importantly, an advantage in grades and completion more than in test-score ceilings. Girls' advantage in continuous assessment and coursework tends to be larger than in high-stakes timed examination, and the gap in the extreme upper tail of some mathematics assessments has historically favoured boys, though it has narrowed substantially over time.
Whatever produces the advantage, it looks more like sustained conscientious performance than a difference in peak capability — which is exactly the kind of advantage that schools reward heavily and labour markets reward considerably less.
Where it stops converting
Four points of leakage account for most of the gap between educational advantage and earnings outcome.
Field of study. Degrees are not interchangeable. Graduates of engineering, computing and finance earn more than graduates of education, social work and the humanities, in most markets by a very large margin. The distribution of women and men across those fields is skewed, and it is skewed in the direction that costs. Studies decomposing the graduate pay gap routinely find field of study is the single largest identifiable component.
Occupational sorting after graduation. Even within a field, graduates distribute unevenly across roles. Within medicine, specialty choice varies considerably by gender and specialties differ in pay by multiples. The same is true within law and within business.
The first decade. Earnings trajectories for men and women in the same entry-level roles track closely for several years and then diverge. The divergence coincides with the period in which promotion to first management level occurs and, for many, with the arrival of children. Analyses of administrative earnings data in several countries find the majority of the eventual gap opens in this window.
Hours and continuity. Part-time work and career interruption both carry earnings penalties that are far larger than the proportional reduction in hours, and both are distributed unevenly.
What this rules out
It rules out the version of the story in which the pay gap is primarily an educational deficit that will resolve as cohorts age. Women graduating today have been out-attaining men for decades; the cohorts have aged; the gap has narrowed but not closed.
It also rules out the version in which the gap is purely direct discrimination in pay for identical work. The like-for-like gap in most rigorous studies is much smaller than the raw gap — usually a few percentage points rather than fifteen or twenty. That residual is not nothing, and it is real, but it is not where most of the difference lives.
The difference lives in sorting: who ends up in which field, which role, and on what schedule. That is a less satisfying answer than either of the simple ones, because it implicates a large number of small decisions made across two decades rather than a single villain.
Why this framing is useful to a student
Because it makes visible which decisions are load-bearing.
Grades matter for access. They do not, past a threshold, determine earnings. Field determines earnings to a far greater extent than performance within field, which is not what school rewards teach students to expect.
The first decade after graduation is disproportionately consequential and is the period during which most people are given the least guidance. Almost all careers advice is concentrated at eighteen, when the highest-leverage decisions come at twenty-four to thirty-two.
And continuity has a price that is worth knowing in advance rather than discovering afterwards. This is not an argument against interruption; it is an argument for knowing the arithmetic before making the choice, and for arranging matters — savings, pension contributions, return terms — with the arithmetic in view.
The part that is not the individual's to solve
It would be a misreading of all of this to conclude that women should simply choose differently.
The pay difference between an engineering degree and an education degree is not a natural fact. It is a set of choices about what societies pay for, and the persistent finding that occupations lose relative pay as they become female-dominated — documented across several decades of occupational data — suggests the causation does not run only from field to pay.
Both things are true at once. An individual navigating this system benefits from knowing where the money is. A society interested in the outcome would have to ask a harder question about why it is there.