China: The Number Changed and the Instrument Did Not

Every society governs reproduction. Most do it without admitting to it — through the price of a flat, the shape of a working week, the length of a lease, what a clinic will and will not do without a husband beside you, what a form asks and what it has no box for. The instruction is real and it is never written down as an instruction. China is the case where it was written down as one. Between 1980 and 2021 the same state told the same population, in documents anybody can still read, that the correct number of children per couple was one, and then that it was two, and then that it was three.

This is why China belongs in a chamber that uses the world as a mirror, and it is not because it is an extreme. It is because it is legible. A society that leaves its preferences implicit can always say it has none: nobody was told to do anything, people simply chose. A society that published a number, enforced it, and then published a different number cannot say that. The reversal is an admission, made by the state itself and dated, that reproduction had been an object of policy all along. Almost no other country has ever been obliged to make that admission in public.

What follows is not an indictment and not a warning, and a reader who wants either will have to find it elsewhere. It is an argument about what the reversal shows, and the argument runs against the obvious reading. The obvious reading is that a state able to push births down ought to be able to push them back up. Ten years of published statistics say it has not. Two things follow, one of them uncomfortable for everybody. The first is that the original policy was probably working with the demographic current at least as much as against it. The second, and harder, is that a generation of women instructed that one child was correct are under no obligation to find the new instruction credible. This Library does not treat a birth figure as a score and does not treat a woman without children as a person with a deficit. That position is not suspended here. It is the reason this essay can be written at all.

The spine is four instruments and it is worth setting down in order, because the dates do the argument. On 25 September 1980 the Central Committee of the Communist Party of China published an Open Letter to Party and Youth League members on the problem of controlling population growth, calling on them to take the lead in having one child. It is treated as the beginning of the policy because it was the first central directive to apply a single-child standard to the whole country, and it set a goal: to hold the population below 1.2 billion by the end of the century.

In December 1982 family planning entered the Constitution. Article 25 provides that the state shall promote family planning so that population growth is consistent with economic and social development plans. Article 49 provides that both husband and wife shall have the obligation to practise family planning. Both survived the constitutional amendment of March 2018 and are in the text today. On 29 December 2001 the National People’s Congress Standing Committee adopted the Population and Family Planning Law, in force from 1 September 2002. Its Article 18 advocated one child per couple, with a second where provincial rules allowed. Its Article 41 required a citizen who gave birth outside Article 18 to pay a social maintenance fee.

Then the reversal, in two moves. On 29 October 2015 the communiqué of the Fifth Plenum announced that couples would be allowed two children; on 27 December 2015 the law was amended, and from 1 January 2016 Article 18 advocated two. On 26 June 2021 the Central Committee and the State Council issued a Decision on optimising fertility policy, and on 20 August 2021 the law was amended a second time. Article 18 now provides that the state advocates marriage and childbearing at an appropriate age, and that a couple may have three children.

Now the precision point, which is the whole essay in one sentence. The number moved; the instrument did not. The statute is still called the Population and Family Planning Law. Its Article 2 still states that family planning is a basic national policy, and that the state takes comprehensive measures to regulate population size, improve population quality and promote the achievement of an appropriate fertility level. Appropriate. The word that gives a state the standing to have a view about how many children a citizen should have was never removed. Only the figure behind it was changed.

The enforcement history has to be stated, and it has to be stated from documents rather than from atmosphere. The cleanest evidence is the state’s own paperwork, and there are four pieces of it.

The first is the fee. Article 41 of the 2001 law made a birth outside the limit a payable event, with provincial governments setting the amount and late payment attracting a surcharge enforceable through the courts. It was abolished by the amendment of 20 August 2021, and on 26 September 2021 the State Council repealed the 2002 Measures for the Administration of the Collection of Social Maintenance Fees along with two other family-planning regulations. The word no longer appears in the statute.

The second is the Decision of 26 June 2021 itself, which instructs authorities to abolish the fee, to clean up and repeal related punitive provisions, and — the sentence that matters — to fully decouple household registration, school enrolment and employment from an individual’s childbearing status. A state does not order the decoupling of a child’s school place from the circumstances of that child’s birth unless the two had been coupled. The instruction is the record.

The third is a set of official health statistics. In 1983, the year in which enforcement was at its most intense, the Chinese health authorities recorded on the order of twenty million sterilisations, close to eighteen million intrauterine device insertions and over fourteen million abortions, in a year of roughly twenty-one million births. Two of those three procedures can only be performed on a woman’s body. The figures do not separate the consented from the compelled, and this essay will not pretend they do; what they establish is the scale of the surgical apparatus, not the proportion of it that was voluntary.

The fourth is that the 2001 statute anticipated its own abuse. Article 4 required officials to enforce the law civilly and not to infringe citizens’ lawful rights and interests, and Article 39 set out sanctions for functionaries who infringed personal rights, abused power or falsified population statistics. A legislature writes that clause because it knows what it is legislating for. And the public record includes at least one case, in 2012, in which a city government in Shaanxi publicly apologised for a late-term termination and disciplined the county officials responsible. This essay records the official response and declines to narrate what was done to the woman, because she is not an illustration of anything.

The National Bureau of Statistics publishes annual births, and the series since the limit was raised is the plainest thing in this essay. In 2016, the first year of the two-child law, there were 17.86 million births — the highest figure of the decade and, read in isolation, a policy success. In 2017 there were 17.23 million. Then 15.23 million, then 14.65 million, then 12.02 million in 2020, then 10.62 million in 2021, the year the limit went to three. In 2022 there were 9.56 million births and China recorded its first population decline since 1961. In 2023 there were 9.02 million.

In 2024 the figure rose, for the first time in seven years, to 9.54 million — an increase of about 520,000. Then in 2025 it fell to 7.92 million, with a crude birth rate of 5.63 per thousand and a population down 3.39 million to 1,404.89 million, the fourth consecutive annual fall. That is the lowest number of births recorded since the founding of the People’s Republic.

One more pair of figures, because it is the sharpest thing on the record. Marriage registrations, which in China are closely tied to births by both law and custom, fell to 6.106 million couples in 2024 and then rose to about 6.76 million in 2025, an increase of roughly eleven per cent. In the same year births fell by seventeen per cent. Marriages up, births down, in one calendar year. Even that comparison needs a caution: a revised Marriage Registration Regulation took effect on 10 May 2025 removing the requirement to register in the location of a spouse’s household registration, which makes a marriage easier to record as well as easier to hold, so part of the rise is administrative.

That caution is not a quibble. It is the method, and the next section is about nothing else.

A birth figure counts registered live births in a calendar year in a territory. It is a good measurement of exactly that and a poor measurement of almost everything it is used for, and an essay that leans on it owes the reader the list.

It does not show how many children a woman will have across her life. Annual births are the product of two independent things: how many women of childbearing age there are, and how many children each of them has in that year. China’s cohort of women in their twenties and thirties has been shrinking for a decade because of how many girls were born thirty years ago, so the annual figure can fall while per-woman behaviour holds perfectly still. Roughly half of any recent fall is arithmetic inherited from the past.

It does not distinguish postponement from refusal. A year in which couples decide to wait produces a lower figure without producing smaller completed families, and some of what was postponed comes back later. It does not control for the zodiac: 2024 was a dragon year, and the timing of Chinese births has a documented sensitivity to the cycle, so the one increase in the series should be read with both the zodiac and the post-pandemic marriage rebound inside it rather than as an effect of policy.

The crude birth rate is worse still, because its denominator is the whole population, which is ageing — 23.0 per cent of it was sixty or over at the end of 2025. The denominator moves for reasons that have nothing to do with anybody’s decision about anything. And the measure that would actually answer the question, the total fertility rate, is not published annually by China. The 2020 census gave 1.3. Estimates for later years circulate and some of them are very low; this essay does not use them, because it cannot verify them and does not need them.

What the series does show, and it shows it robustly, is direction. In nine of the ten years since the one-child limit was lifted, births have fallen. That is not a measurement of anybody’s desire. It is a measurement of what happened after an instruction was changed.

The surface reading of China is that a state proved it could control reproduction and is now discovering the cost. The counter-reading, which this essay holds, is that the control was always less than it looked, and the reversal is the experiment that shows it.

Start with sequence. The fertility decline did not begin in 1980. It began in the early 1970s with a campaign known as later, longer, fewer — later marriage, longer intervals between births, fewer children — and during that decade the total fertility rate fell by more than half, from about 5.8 births per woman in 1970 to about 2.7 in 1979. By the time the Open Letter was published the steep part of the fall had already happened. Peer-reviewed work by Martin King Whyte, Wang Feng and Cai Yong, published in The China Journal in 2015, puts the share of the decline since 1970 that preceded the one-child policy at as much as three-quarters.

Then the arithmetic of the official claim. The state has long held that the policy prevented more than 400 million births. The figure was produced by extending the fertility trajectory of 1950 to 1970 forward into decades that looked nothing like them, and comparing the projection with what happened. The same demographers who dated the decline put the averted figure nearer 200 million for 1979 to 2009. The honest description is that the comparison is with a counterfactual nobody observed, and that the further you push it the less it is a measurement and the more it is an assumption with a number attached.

Put sequence and arithmetic together with what happened after 2016 and the counter-thesis is simply what is left. A great deal of what the policy is credited with is what industrialisation, urbanisation, the schooling of girls, later marriage and the cost of a city childhood do everywhere they occur, including in countries that never told anybody anything. The policy pushed in the direction the water was already running. It has never been shown able to push the other way — and since 2016 the state has tried: raised the limit twice, abolished the fee, repealed the penalties, ordered the decoupling, and in 2025 put cash on the table. Births that year were the lowest on record.

This is a finding about an instrument, not a verdict on a country, and it is worth saying plainly that there is nothing here to feel superior about. Every society in the same demographic position is finding the same thing, including the one this Library writes from.

The second half of the counter-thesis is the harder one, and it is about credibility rather than demography.

Consider what the state is now offering. From 1 January 2025 a national childcare subsidy pays 3,600 yuan per year for each child under three, exempt from individual income tax and excluded from the income tests used for social assistance, with around 90 billion yuan allocated centrally in the first year and more than twenty million families expected to benefit. It is the first large-scale universal direct cash benefit in the history of the People’s Republic. Alongside it, local authorities have been directed to plan for free preschool education, and the 2021 Decision asks that housing allocation give some consideration to the number of minor children in a household.

These are serious measures and this essay does not sneer at them. But consider who is being asked. A woman of thirty-five in China today was born around 1991, into a household to which the state had already said one. Her parents built a life around that number — the flat, the schooling, the savings, the assumption that a single child would be educated to a standard the family could just about afford. The number was not only obeyed. It was internalised as a standard of good parenting: one child, properly raised. That standard is now the largest obstacle to the new figure, and the state authored it.

Three consequences follow. First, an instruction that has been reversed once within a lifetime carries less weight the second time, because the reversal is itself evidence that instructions change. Second, an instruction that was once enforced and is now merely encouraged is a weaker thing by design: the state gave up its grip in 2021 and kept its standing, which is exactly the combination least likely to move a household. Third, and this is the sentence to hold, a state that has demonstrated it will govern reproduction has also demonstrated that its current preference is not permanent — which is a reason for a rational person to decline to organise a life around it.

Two pieces of residue belong here, stated and not over-read. The sex ratio at birth recorded by the 2020 census was 111.3 boys per 100 girls, which is a demographic fact with a generation behind it and a generation ahead of it. And the Outline for Women’s Development 2021 to 2030, issued by the State Council in September 2021, lists among its health measures the reduction of induced abortions that are not medically necessary. Its predecessor document had spoken of reducing unintended pregnancy and abortion, so the change is one of framing rather than of a new prohibition; but the framing arrived within months of the limit being raised to three, and this essay records that sequence without claiming to know what it will mean in practice, because that is not yet on the record.

Japan has never set a limit on children and has never set a target for a household. It governs reproduction the way most countries do: by incentive, by silence, and by one statute this Library has already read closely in its essay on motherhood as a decision with no procedure.

The statute is the Basic Act on Measures for Society with Decreasing Birthrate, Act No. 133 of 2003, promulgated on 30 July of that year. Its preamble concedes, in so many words, that marriage and childbirth are of course matters of individual decision — and then continues with a however, declaring that halting the decline is something strongly required of the country now. Article 2 sets out the basic principles, and its fourth paragraph provides that measures in every field, social, economic, educational and cultural, must be taken with regard to the declining-birthrate situation.

That is a rule you cannot see. There is no permit, no fee, no number. There is a standing instruction to every field of policy to bear the birth figure in mind, which reaches a woman not as a rule addressed to her but as a weather system — in what a workplace assumes, in what a form implies, in what a relative says at a particular age. Japan’s own figures for 2025, from the Ministry of Health, Labour and Welfare, are 671,236 births and a total fertility rate of 1.14, a tenth consecutive record low, in a year when marriages rose for a second time to 489,119.

Here is the comparative finding, and it is stated without a ranking because a ranking would be false. A rule you can see has a date, an author, a text and a repeal clause. It can be argued with. It was argued with, and in 2021 it was changed. A preference you cannot see has none of those things, and therefore cannot be repealed, because there is nothing to hand to a legislature. That is the cost of the Japanese arrangement. The cost of the Chinese one is the other half of the same fact: a visible rule can be enforced on a body, and an invisible preference cannot. Neither of these is the better arrangement to live inside, and this essay declines to say which it would choose.

What can be said is the thing both cases have in common, and it is the finding the mirror is for. In both countries, the figure the state watches is produced entirely by decisions taken by individual people about their own bodies and their own years, and in neither country is the person taking the decision consulted about the target. This Library has argued elsewhere that motherhood is a decision with no procedure, that a life outside marriage can be imagined on its own terms rather than as a residue, and that the right to your own body is the ground everything else stands on. The Chinese case does not add a new principle to those. It makes the old one impossible to miss.

This house sells evenings to adult women, and it should state its interest before it states its finding. It sells into a market in which a woman’s undivided hours are scarce, and they are scarce in large part because a great deal of unpriced work — of children, of parents, of households, of the emotional maintenance of other people — is expected of her and is not counted anywhere. A house that sells a few hours of undivided attention is a beneficiary of that scarcity. It would sell less if the hours were less rationed.

There is a second interest to declare, and it is specific to this essay. A house of this kind has an obvious commercial interest in readers who do not have small children, and it should not pretend that its reading of a demographic record is disinterested. The reading has been checked against documents rather than against preference, and the limits are set out in the next section; but the interest exists and the reader should hold it while reading.

What the house may not claim follows from that. Nothing here is advice about whether to have a child, at what age, or at all — that is a decision with no procedure and no adviser, and a service business is the last party with standing to offer one. Nothing here is a position on any country’s population policy. And nothing here treats a woman without children as having solved a problem or as having missed something, because the whole argument of the essay is that neither of those is a thing a figure can establish about a person.

It does not claim to know what China’s births will do next. Ten years is a reasonable test of an instrument and it is not a forecast. It does not claim that the incentives introduced since 2021 will fail; it claims only that after a decade of raised limits, abolished penalties and, most recently, direct cash, the births have not arrived — which is a statement about what has happened and not about what will.

It does not claim that the one-child policy caused the fertility decline, and it does not claim that it had no effect. It claims what the peer-reviewed work establishes about sequence: that a large share of the decline preceded the policy, and that the official figure for births averted rests on a projection rather than an observation.

It does not claim that coercion was universal, uniform, or the whole of the period. Enforcement was provincial, uneven and changed over four decades, and the official health statistics quoted here count procedures without separating the consented from the compelled. It does not offer a figure for how many women were subject to compulsion, because it does not have one it can verify and because a number of that kind, stated loosely, would be a rhetorical device rather than a finding.

It does not claim anything about the character of any nation or the beliefs of any woman in either country. It does not treat a birth figure as a scoreboard, a falling one as a verdict, or childlessness as a deficit. It does not rank China against Japan, and a reader looking for permission to do so has read a different essay than the one written.

Its limits are these. The legal spine is described by principal provisions rather than article by article, and a reader wanting the operative wording should take it from the statute. The 1983 procedure figures are given at an order of magnitude, because published versions differ slightly and the essay makes no claim that turns on the third digit. The 2024 increase in births is read with the zodiac and the marriage rebound inside it, which is an interpretation. The 2025 marriage figure is affected by a registration reform that took effect in May of that year, which the essay says rather than hides. No total fertility rate later than the 2020 census is used. And the essay is written from outside both societies it compares, from documents, by a reader of documents.