Remittances and the children left behind
A Socratic walk-through of remittances and the children left behind — reasoned out one step at a time, not lectured.
The question we started with
THE QUESTION #Why can money sent home by a migrant parent lift a child's school enrolment while lowering what they learn?
A parent goes abroad and sends money home. The child, who might otherwise have left school, stays enrolled — fees paid, uniform bought, no need to work the market stall. And in several studies that same child's test scores sit below a comparable non-migrant household's child.
Enrolment up, learning down. The reflex is to treat one as the error. But consider whether they are two effects of one event arriving through different channels — and if so, whether "does migration help the children" was ever a question with a single answer.
Reasoning it through
REASONING #Separate what a departure actually changes in a household, because it is not one thing.
The first change is the budget. Schooling is not free even where tuition is: fees, uniforms, books, transport, and above all the forgone income from a child who could be working. Remittances relax all of that at once, so enrolment and attendance should rise, most for children closest to dropping out. This part is well supported.
The second change is that an adult has left. Whatever that adult did at home is now not being done, or is done by someone with less time. Two components matter for learning. One is direct: supervision, help with homework, noticing a child has stopped attending, keeping the household to a routine. The other is easy to overlook — the remaining caregiver absorbs the departed adult's domestic work, and where they cannot, the older children absorb it. So a child's hours in school can rise while their hours actually studying fall, and a girl in particular can end up enrolled and simultaneously running a household.
A third channel is subtler and less firmly established: what the household now believes schooling is for. Where migration is the visible route to income and the qualifications it rewards are not academic, the perceived return to doing well in school can fall even as the return to staying in it rises. That depresses effort without depressing enrolment, which is the observed shape — but I flag it as plausible with mixed support rather than settled.
Notice that the first channel and the rest push in opposite directions on different outcomes. Money buys presence in a classroom. Presence in a classroom is not learning, and the thing that converts one into the other — adult time — is precisely what the migration removed.
That also predicts when the pattern should not appear. If lost supervision is the mechanism, a household with another adult who can take it over should show the enrolment gain without the learning loss. Whether a grandmother lives in the home therefore matters a great deal, and so does which parent left: several literatures report larger deficits when the mother is the migrant, which fits an account built on lost caregiving time rather than absence in general. I recall that direction with reasonable confidence and attach no number to it.
The analogy
THE ANALOGY #Think of a farm whose owner takes a job in the city and wires money back. The money buys better seed and a tractor — inputs that were unaffordable and now are not. But the person who walked the fields daily, noticed the blight early and decided when to irrigate is no longer walking them. Input use goes up; yield per hectare can go down. Nobody made a mistake, and no amount of extra seed substitutes for the walking.
A farm's monitoring can be hired at a known price, whereas a parent's attention is not straightforwardly purchasable and the substitute is usually another family member with their own tasks; and a farm has no equivalent of a child's response to the absence itself.
Clarifying the model
THE MODEL #The account so far is a story about channels, and it is only as good as the evidence that any of it is causal. Here that evidence has a serious problem.
Households that send a migrant are not a random sample. They are richer than the poorest — migration costs money — and better connected, since it usually needs someone already at the destination. They are often more ambitious about their children. And frequently something prompted the departure: a lost job, a bad harvest, a debt, an illness. Every one of those independently affects a child's schooling, so comparing migrant with non-migrant households compares two populations that differed before anyone left.
The confounds do not even point the same way, which makes the raw comparison useless rather than merely biased: selection on wealth and ambition inflates the apparent benefit, while selection on crisis deflates it.
What separates them? Following the same child across the departure, so each is compared with themself. Comparing siblings who were different ages when the parent left, holding the household constant. And best, a shock to migration the family did not choose — a visa ballot admitting applicants at random, or an exchange-rate move altering what a remittance is worth without altering who migrated. Ballot-based studies exist and are the strongest evidence here; the wider set of estimates rests on weaker designs.
So the honest position is layered. That remittances raise enrolment is well supported. That an adult's absence costs supervision is well supported in its own right. That the net effect on learning is negative is where the disagreement lives, and it varies by who left, for how long, who took over, and what the destination pays. Anyone reporting a single number is reporting a context rather than a constant — which is why there are no effect sizes here.
One more caution. These are population averages over households deciding under constraint, and describe none of them individually. Nothing here says whether any family should migrate, and the tradeoff they face is not one a finding resolves for them.
A picture of it
THE PICTURE #How to readEach point is a household arrangement rather than a measured statistic — the placements illustrate the argument's logic, not data, which is why no axis carries units. Read across for a household's cash and up for the adult attention remaining in it. The lower-right quadrant is what the question is about: enrolled because the money arrived, unhelped because the person who supervised is gone. The two migration points separate vertically rather than horizontally, which is the whole claim — what differs when the mother rather than the father leaves is the time, not the money — while the grandmother point sits high because a resident adult restores the lost supervision.
What became clearer
WHAT CLEARED #The puzzle dissolves once you stop treating a parent's migration as one intervention. It is at least two, arriving together and acting on different things. Money removes the constraint keeping a child out of school; the departure removes the adult time that turns being in school into learning. So enrolment and attainment can move in opposite directions without contradiction, and the size of each depends on who left and who remained to take over.
The second thing that became clearer is how thin the causal ground is. Migrant households differ from non-migrant ones before the migration, and the differences pull the estimate both ways at once, so the ordinary comparison cannot be read. Only a design following the same child through the departure, or exploiting a ballot or a shock nobody chose, can say what the migration itself did — and outside those, the fair answer to "is it worth it for the children" is that it is not established, and probably has no single answer.
Where to go next
ONWARD #- What happens on return, and whether the schooling gap closes when the migrant parent comes home.
Key terms
TERMS #| Term | What it means |
|---|---|
| Remittance | money sent home by a migrant to their household of origin. |
| Selection bias | the distortion arising when the group studied differs systematically from the comparison group for reasons unrelated to the treatment. |
| Visa ballot | an immigration lottery admitting applicants at random, used by researchers as a naturally occurring randomised assignment. |
| Enrolment versus attainment | being registered at a school, as distinct from what is learned there; the two can move in opposite directions. |
Every term the collection defines is gathered in the glossary.