If you want to predict how much education a child will complete, and you are allowed one question, do not ask about the child. Ask how much education their parents completed. It will outperform almost anything you could learn about the individual — their early test scores, their school, their reported motivation.
This finding is old, robust, and international. It holds in countries with selective systems and comprehensive ones, in rich economies and poor ones, under very different policy regimes. The strength of the association varies — some societies are considerably more mobile than others — but the direction never does.
It is worth sitting with how strange this is. Formal education is the institution most explicitly designed to be a mechanism of mobility: universal, publicly funded, open to everyone regardless of birth. And it turns out to be one of the most reliable channels through which advantage is transmitted. Understanding how that happens is the point of this essay, because each channel suggests a different intervention.
Five channels
The transmission is not one thing. It is at least five, operating simultaneously and reinforcing each other.
1. Language, early and dense
The most consequential differences appear before any school is involved. Children in households where adults talk to them extensively — narrating, explaining, asking open questions, using varied vocabulary — arrive at school with substantially larger vocabularies and more developed grammatical understanding than children who received less of this.
The size of the gap has been the subject of vigorous methodological argument, and some early estimates were almost certainly overstated. What survives the argument is the direction and the significance: differences in early language exposure are real, measurable, and predictive of later reading comprehension, which in turn gates access to every subject taught through text.
Crucially, the difference is not about how much parents love their children or how much they want them to succeed — a distinction that gets lost with expensive frequency. It is about time, exhaustion, competing demands, and whether anyone ever suggested that talking to an infant who cannot answer is worth doing.
2. Knowing how the system works
This channel is invisible to the people who benefit from it, which makes it especially powerful.
An educated parent knows a set of things that are nowhere written down. That you can ask for a different teacher. That a diagnosis can be requested and what it is called. That streaming decisions can be appealed. That the deadline for the application is soft in practice. That "your child is doing fine" from a teacher may mean something other than fine. That there is a difference between courses that keep options open and courses that close them. Which questions to ask at a parents' evening, and in what tone, so that the school responds.
A parent who did not complete school themselves may be equally committed and entirely unable to do any of this. They may not know the categories exist. They may be deferential toward an institution that they experienced as authoritative and unfriendly. And schools, reading deference as disengagement, may quietly stop expecting much from that family.
Most of what an educated parent gives a child is not knowledge of subjects. It is knowledge of how the institution works, and the confidence to act on it.
3. Homework, and who can help
A homework system assumes an adult at home who can explain what was not understood in class. This assumption is unevenly true, and the unevenness is systematic.
When a parent cannot help — because they never studied the material, or because the method has changed, or because they are working an evening shift — the child either does the work wrong, does not do it, or seeks help elsewhere. Households with money buy tutoring. Households without it do not. Homework, intended as consolidation, becomes a mechanism by which the home environment is imported directly into the child's academic record.
4. Expectations that operate as forecasts
Children construct their sense of what is possible from what they observe. A child whose parents and their friends attended university experiences that as the ordinary next step — not an aspiration but a description of what happens after school. A child for whom nobody in their extended family has ever done it must construct the possibility from nothing, and then sustain it against the evidence of everyone around them.
This is why role models matter more than the phrase's overuse suggests. The function is not inspiration. It is evidence — proof that the path exists and is walkable by someone recognisably like you.
5. Money, and the geography it buys
The most direct channel and the least subtle. Educated parents earn more on average. Higher income buys housing in areas with better-resourced schools, tutoring, books, trips, equipment, quiet rooms, a reliable connection, and the ability to support a child through further study rather than requiring them to earn. In systems where school quality varies by neighbourhood, the housing market converts income into educational advantage automatically, without anyone having to intend it.
Why the channels matter separately
Each channel implies a different remedy. Language exposure suggests early years programmes. Institutional knowledge suggests navigation support and plain communication. Homework suggests supervised study at school. Expectations suggest contact with people who have walked the path. Money suggests funding formulas and housing policy. A strategy aimed at only one will underperform, because the others keep operating.
The good news, which is substantial
Everything above describes a mechanism of reproduction, and read alone it is bleak. The literature contains a second finding that is at least as important.
When a parent's education increases — through a policy change, a programme, a return to study — their children's outcomes improve. This has been examined using the same natural experiments described in the opening essay: compulsory schooling law changes that gave one cohort of parents an extra year, and school construction that gave one cohort access. The children of those parents, born years later, show measurably better outcomes than the children of the cohort just ahead.
In other words, the transmission runs in the desirable direction too. Educate one person and you have altered the starting conditions for people who do not yet exist. This is what makes educational intervention unusual among social investments: the returns do not terminate with the recipient.
Break the chain once and it does not have to be broken again. That is the entire case for treating education as an investment rather than a service.
It is also why adult education is chronically undervalued. A programme that teaches a thirty-year-old to read is scored on that person's employment outcomes. It should be scored on their children's school readiness as well, and in most accounting systems there is no line for that.
What actually interrupts the pattern
Approaches with a genuine evidence base, ordered roughly by how early they act.
| Intervention | Channel targeted | Evidence |
|---|---|---|
| High-quality early childhood education | Language, early skills | Strong, with effects traceable into adulthood — quality is decisive |
| Home visiting and parent coaching | Language, interaction | Positive where sustained and well delivered; fades if brief |
| Adult basic education for parents | All channels at once | Under-researched but promising; rarely funded as a child intervention |
| Supervised study at school | Homework support gap | Removes the home-environment dependency directly |
| Progressive school funding | Money, resources | Effective when the money reaches classrooms rather than administration |
| Navigation support and mentoring | Institutional knowledge, expectations | Cheap and surprisingly effective at transition points |
| Late second chances | Reversibility | Systems with open re-entry show higher mobility than early-tracked ones |
Two entries deserve a note. Early childhood provision has the strongest evidence of anything on the list, but the finding is conditional in a way that is often dropped: quality early education produces large effects, and low-quality childcare produces little. Expanding places without attending to what happens inside them buys the label and not the outcome.
And navigation support is the best value on the list. Much of what advantaged families provide is information — deadlines, options, consequences, the fact that a thing is possible at all. Supplying that information deliberately, at the moments when decisions are made, costs very little and repeatedly outperforms its price.
Two things this argument is not
It is not a claim that parents are to blame. Every channel described here operates through time, money, prior opportunity, and information — not through care. Parents with the least education are, in the main, doing what they can with what they were given. Interventions that proceed as though the deficit were one of effort tend to fail, and deserve to.
It is not a claim that mobility is impossible. Individuals cross these gaps constantly. The finding is statistical: the odds are steeply tilted, not the outcomes fixed. But the individual success stories should be read carefully. They are usually stories in which several of the channels above were substituted by someone else — a teacher who supplied the institutional knowledge, a relative who supplied the expectation, a programme that supplied the money. The exceptions are not exceptions to the mechanism. They are cases where something else performed its function.
The long view
There is a particular moment that the research describes in aggregate and that families experience as a single event: the first person in a family to complete a level of education that nobody before them reached.
What that person acquires is not only a qualification. They acquire the institutional knowledge, the confidence with officials, the vocabulary, the network, and the assumption that this level is normal. And they then transmit all of it, without any deliberate effort, to their children — who will experience as unremarkable what their parent had to fight for.
That is what the word "inheritance" in the title is doing. Educational advantage behaves like inherited property in how reliably it passes down. Unlike property, it cannot be lost, spent, seized, or devalued by a market. Once a person has it, it belongs to them and to everyone they raise.
Which is the strongest available argument for spending public money on it. Most social spending buys an outcome once and must be spent again next year for the next person. This buys a change in the starting conditions of a family line — and it only has to be bought once.