On Birthrates, Revealed Preferences and the Decline of the West
On the possibility that humans, when presented with choice, choose to not have enough children and how AI can allow managed population decline.
Ends are not bad things, they just mean that something else is about to begin. And there are many things that do not really end anyway; they just begin again in a new way.
It is hard to follow politics now without running into population decline, and behind it the immigration fight and the pensions that a shrinking workforce can no longer fund. The changing shape of our populations is one of the central facts of the present debate, and yet ask why the change is happening and the answers diverge sharply. The Right names the decay of family norms, women drawn into the workforce, a secularised public, a thinning of social cohesion. The Left answers with housing priced out of reach, childcare too dear to afford, economic insecurity, and the unequal burden parenthood still imposes. A third party takes a few items from each list and scolds both for the extremity of theirs. Each of these surely bears on whether children are wanted, and, to a degree, the degree itself in question on whether they are feasible. But none of them explains the size of the gap between the fertility rate we observe and the rate a stable population requires. And the experiments run around the world point the same way: address one of these factors, or several, and the effect on the birth rate is usually small. So the question is whether the decline uniform across the OECD and well beyond it is mainly a failure of policy, or whether something is being revealed. It may be that the low fertility rate is not a bug to be fixed but a preference being expressed: the natural result of letting people choose, in a modern life that offers them so much else to choose instead.
Begin by setting aside the idea that this is a Western complaint. Almost every OECD country has fallen below replacement, and the decline has long since left the rich world behind it. By the United Nations' 2024 reckoning, 131 of 237 countries, home to more than two-thirds of the living, are already below the 2.1 births per woman a population needs merely to replace itself. An effect this universal does not keep a local cause. Whatever is at work is spread thin across every kind of country, and is for that reason very hard to reach. The lever that actually sets fertility is birth control. Take it away and the rate climbs; the data say as much, since wherever contraception is shunned the children are many. For the first time, fertility is something chosen rather than suffered. And once it is chosen, a confusion at the centre of the whole debate comes into view. Replacement is treated as a moral number, a quota each generation owes. It is nothing of the kind. There is no correct number of children to have. Replacement is an accounting figure and only that: the rate at which a population holds level, no more sacred than the rate at which a reservoir holds its water. The morality draped over it is borrowed. It is the morality of obligation owed not to anything higher, but to the pension funds and labour markets that need bodies to keep running. A thin morality, and a hired one.
Demographers have long drawn a line between two kinds of cause, and the line is the first thing to get right. Behind every birth rate stand the background determinants, the wealth of a country, the schooling of its women, the migration from the field to the city, the price of a child reckoned in money and in year. These press on fertility from a distance. They do their work only at second hand, by passing through a far smaller set of switches, the proximate determinants, the handful of biological and behavioural channels through which every distant cause must finally run. The age at which unions form, the practice of contraception, the recourse to abortion, the long infecundity that follows a birth. Wealth and schooling move fertility by moving on of these few gates, and in the modern world the widest of them, the one that carries most of the traffic, is contraception. Fertility limitation is older than its modern instruments. Withdrawal, abstinence within marriage, a daughter kept from marrying young, an abortion sought in quiet; the will to have fewer children is ancient, and in places such as provincial France the birth rate was already falling generations before any clinic opened its doors. What the modern methods supplied was reliability. They took a capacity that had always been intermittent and uncertain, dependent on a partner's restraint or a confessor's permission or plain luck, and made it routine, private, and largely the individual woman's own, repeatable month after month and answerable to no one. A thing once occasional became ordinary, and a thing once shared became hers.
The causal evidence for even this bounded role is better than the debates assume, because history left behind something closer to controlled experiments. Martha Bailey, exploiting the accidents of nineteenth-century anti-vice laws that left some American states able to ban the sale of contraceptives and others not, found that access to the pill measurably hastened the decline in martial fertility after 1960. Grant Miller, following the haphazard spread of Colombia’s family planning programme across towns that received it years apart, found that the arrival of supply let women postpone a first birth and trimmed their completed families by something on the order of half a child. Where the capacity arrived fertility moved, and it moved in a direction and by an amount that can be measured.
Contraception is the gate through which the modern decision passes, the mechanism that made a smaller family cheap to choose and easy to hold to. The choosing happens elsewhere, in people weighing a life against the cost of enlarging it. To call the gate the cause of the cause would be to mistake its opening for the decision to walk through; to deny it any part would be to throw away every clean experiment we have. Which carries the question forward by one step. Once the capacity to decide is reliable, ordinary and lodged with the individual, the live question stops being whether people are able to hold their fertility and becomes what they do once nothing stops them. A mechanism only establishes the lever exists and turns free. What people reveal is the pulling. Return to that older world for a moment and try to read it. A woman in it bears six children across her twenties and thirties, the number is written down, and we are tempted to treat the number as verdict, as though she had been asked how many she wanted and had answered six. She was asked no such thing. The six are a residue of a hundred other wants; of wanting a husband, and a marriage bed, and the regard of a village that counted childlessness a misfortune and barrenness a curse; of nights that carried a consequence no one had a reliable way to refuse; of a faith and a custom that left the matter to providence. Somewhere inside that six there may sit a genuine wish for a large family. There sits everything else beside it, and from the outside the wish and the everything else arrive as a single figure, and cannot be told apart. Now bring the same reading to a woman today who stops at one, or at none. Her figure carries its own freight of circumstance: the price of a flat, the shape of a career, the willingness or refusal of a partner, the years that ran out while she waited for the conditions to come right. But a refusal sits inside it that the older number could never hold. To arrive at one child now is to have declined, month upon month, the second that was available the whole time. The count has become, in part, an answer. Where the six bundled a wish together with everything that pressed upon it, the one has had much of the pressure drawn off, and what remains shows through. This is the quiet revolution contraception worked beneath the demographic one, and it is worth stating modestly. Modern fertility stays shot through with circumstance: accidents happen, partners disagree, the body fails its owner, and many people end with fewer children or more than they ever meant to. The old births, for their part, held as much love as any since; a great many were longed for, prayed over, and adored. It made one kind of reproductive refusal legible for the first time, by draining away the noise that had always drowned it out. The low modern number is the first in history that can be read, with some care, as a kind of testimony.
And here the argument meets its first real obstacle, the one any honest version has to walk straight into. If the modern number is testimony, it should agree with what people say when you simply ask them. Ask them, and they do not say one.
They say two.
Take the objection at its full strength, because it is a strong one. When the surveys ask people how many children they would like, the answer across much of the western world has become remarkably stable, and remarkably specific. It is two, a figure set close to the line a population needs to hold itself level. If anything the surveys cut against the argument so far, because women today tend to reach the end of their fertile years having had fewer children than they once named, and childless more often than they meant to be, the shortfall running to the better part of a child in parts of southern Europe. Put plainly, people say they want about two and end up with rather less. That looks less like a hidden preference for emptiness than like a wish being thwarted, a want the world refuses to let them satisfy. So begin with what the number is made of. A survey is a sentence spoken in a particular room, by a particular kind of person, under a particular set of pressures, and all three shape what gets said. The person most likely to be asked what she wants is young and has no children yet, and the lifetime figure such a person reports is a tentative thought, volatile, lightly held and a weak guide to what she herself will later do. She is naming an intention before any of the conditions that test it have arrived; the partner, the job, the flat, the body’s own schedule. She is also naming it inside a culture that has an answer ready for her. Two is the respectable number, the one that means neither selfishness nor excess, and when a person reaches for the expected reply she reaches for it. There is also a second pressure, and it bends the number in one direction only. Once a child exists, the arithmetic of the question changes, because to name an ideal below the family you already have is to say, in front of another living person, that you would have preferred them not to be. A mother of two who has privately found two to be too many cannot easily report an ideal of one, since the one she would subtract has a name and sleeps down the hall. The socially sayable answer is anchored at or above the family she has, whatever the quieter account would be. So stated ideals drift upward and resist falling, held in place by a decency that has nothing to do with how many children a person, knowing what she now knows, would choose again. This does not tell us how large the hidden gap is, or prove there is one in every case. It tells us that the instrument reads high, and that a number which can only comfortably move up makes a poor thermometer for a preference that may be moving down.
What we can watch instead is something narrower and more trustworthy: the number a person arrives at after the wanting has been pushed through every real condition of a life. Call it the implemented preference. It is worse than the survey at telling us what people dream of and far better at telling us what they choose, and it is the only one of the two that is paid for in years. Here the decisive thing is a matter of information. The first child is chosen blind, or nearly so. Whatever a person believes about parenthood beforehand, she believes it from the outside, assembled from other people's lives and her own imagining. The second child is the first decision she makes with the facts in hand. By then she has met the pregnancy, the birth, the months without sleep, the narrowing of a day around a dependent creature, the particular way a small child empties a person of time and returns love in a currency that does not pay the same bills. The step from one to two is the first reproductive choice made in possession of the experience, and a great deal of the variation between countries in final family size turns out to live exactly there, in whether the parents of one go on to a second. When you look at what governs that step, you find learning. Where the well-being of new parents falls hardest in the passage through a first birth, the second child is the least likely to follow, and the size of that fall foreshadows the size of the family that comes after it. Sharper still, the pattern bites most among the older and the better educated, the people most able and most willing to revise a plan once they have lived a stretch of it. That is the signature of an estimate being corrected against reality. A number was carried into parenthood, and a truer one was carried out. Whatever this fall is, regret is mostly the wrong word for it. Most of these parents love the child entirely and would, asked directly, make the same choice again, and for many the dip is passing weather rather than a verdict. The revision sits alongside the love instead of against it. Knowing now, in the body, what a single child costs in sleep and self and hours, a great many parents quietly decline to double the figure, and they do it without ever saying a disloyal word about the child they have. Which is the last piece, and the reason behaviour deserves more trust here than speech. The one statement almost no one will make aloud is that a child was more than they wanted. So the wanting goes unsaid, and the answer is given in the only register left open, in what is done next rather than what is declared. The low rate at which parents of one proceed to two may be the plainest confession the system permits: a preference revised downward, in private, and reported only in the having.
None of this is to say that money and time count for nothing. The Left's half of the catalogue we began with is real. When a country makes childcare cheap and abundant, completed family size tends to rise, and the result holds up across the better-controlled studies. Cash in the hand has a softer and more temporary touch: payments to parents lift the birth rate for a while, largely by moving births earlier rather than adding them, and the bump fades. Conditions enter the decision. A person weighing a second child against a flat too small for one, a career that punishes the interruption, and a partner unwilling to carry half the load is a person whose constraints are doing part of the choosing. Ease those constraints and some of the children the constraints were suppressing arrive. The lever exists, and it moves. The trouble is how far. Look at the governments that have pushed hardest, the ones that declared a demographic emergency and spent at a scale the rest only discuss. Russia's Maternity Capital, Poland's Family 500+, Hungary's lavish tax relief for mothers: flagship programmes, generous by any measure, sustained for years, and watched by every government with the same fear. Their effect on fertility has been modest and largely fleeting, and where the money landed hardest the clearest result was a fall in child poverty rather than a return of the missing births. Hungary, which has devoted some of the largest family budgets in the world to the task, has stayed well below replacement throughout. A lever can move a number at the margin and still be unable to move it across the distance that matters. These programmes can buy a tenth of a child and a lighter childhood for the children already wanted. None of them has bought its way back to two. Each time a flagship programme underperforms, its defenders reach for the same reply: the support was never quite enough, the transfers too small, the leave too short, the next increment the one that would have worked.
It is tempting to read all this as a story about the West, the loss of nerve of one tired civilisation. The geography refuses the reading: the same descent now runs through more than half the countries on earth, across rich and poor and every creed, which makes it hard to lay at the door of any single Western ideology or welfare regime, and points instead to the one thing these societies share, reliable control over reproduction arriving alongside cities, schooling, women's autonomy, and an abundance of other things a life can now be spent on.
Suppose, then, that the residual possibility is the real one, and voluntary fertility settles below replacement and stays there even as the material excuses fall away. I do not offer this as a forecast; the projections fan out too widely for that. Take it as a scenario worth planning for, and ask what planning would mean. The reflex is to read sustained decline as catastrophe, and for most of history it would have been: an old population leaning on a young one too small to carry it, every comfort thinning as the workers who supplied it vanished. That arithmetic assumed output rose and fell with the number of bodies. That assumption is the part now coming loose, and a serious demographic policy should be built on the gap it opens.
The move is to treat automation as the demographic strategy itself, funded with the seriousness governments have spent, to so little effect, on baby bonuses. A state that cannot conjure the missing workers can build the capability that makes their absence survivable. Machines that draft the document, write the code, read the scan and run the back office raise output per person in exactly the work an ageing society sheds first, the lever that lets a smaller workforce carry a growing weight of dependents. Aimed deliberately, the same investment reaches the hardest corner of an old society, its care, where assistive robotics and monitoring can keep frail people independent longer and let a scarce carer attend more lives, even as the held hand at the relational core of that work stays stubbornly human. The recommendation is plain. Redirect the demographic budget from the increasingly hopeless project of raising the birth rate to the achievable one of raising what each remaining worker can do, and treat productivity, longer working lives, migration and automation together as the central response rather than the consolation prize.
Honesty about the limits keeps the prescription from turning into a fantasy. Automation raises no tax base on its own, holds no emptying village together, shares none of its gains fairly unless it is made to, and answers nothing in the longest run, where a rate forever below replacement carries a people, however richly, toward vanishing. It can make a contraction slower, wealthier and easier to govern across the horizons we plan within. It cannot, alone, make decline harmless, and it cannot end it.
Which returns us, with less certainty than either camp would like, to one hard thing seen clearly. The number people live by is pressed down by every real constraint, so it understates what they would freely choose; the number people say is pushed up by decency and the scripts a culture lends them, so it overstates it. The truth sits between two readings we cannot take cleanly, and the claim made here is the narrow one: the faith that free people, left alone, drift back to the rate their society needs has little ground beneath it. Beneath every policy sits a conflict no policy dissolves. On one side stands a freedom we will not surrender and should not, the power to decide whether and when to become a parent, among the plainest goods our world has made. On the other stands the continuity of the very people who hold it, which seems to rest, in the aggregate, on a great many of them choosing the larger family that fewer and fewer want. No villain stands between the two. There is only the sum of free and reasonable choices arriving at a total no one intended.
The thing we built to free people from a fate once chosen for them by the body, the village and God may turn out to set the population quietly falling. If it falls, it will fall by consent, with no disaster to mark the year. It will look like a great many people, at last, having exactly the families they choose.