Battle of Ideas

We should pay people to have children

A steelman map of the public fight over baby bonus — arguments for and against, with sources and assumptions.

AI-generated · paired steelman agents · independently red-teamed · Pass-1 source spot-checks only · framing-fidelity not independently verified · single model family

Direct cash or tax for having children (pronatal transfers). Childcare, housing, and parental leave are cousins, not the cheque. Whose children, and whether immigration is the alternative, are in the fight. Specimens (Hungary, France, South Korea, Singapore) are not the claim here's country.

AGAINST 5

no further strong arguments at this depth

FOR 5

no further strong arguments at this depth

People also ask

Questions people actually type. The two columns above are the cases — open a card for sources, assumptions, and counters.

Baby bonus

Should we pay people to have children?

Pronatalist policy

Baby bonus pros and cons

Do baby bonuses work?

Why do countries pay people to have babies?

Ordering within each column: strongest first — validation tier, then source quality, then representativeness.

AGAINST · We should pay people to have children
Empirical — moderateP1

The extra baby is expensive because you also pay for the ones who were coming anyway

Milligan (REST, 2005) is the cleanest cheque experiment on this page, and AGAINST should not deny the response. Quebec paid up to C$8,000 for a third child and fertility moved — more at higher parity. The C.D. Howe backgrounder on the same programme said it achieved extra births at a high cost per additional birth, because you cannot write the cheque only to the woman who was on the fence. Drago et al. (2009) on Australia's 2004 Baby Bonus: modest. Demographers have spent a career separating tempo (have the child this year instead of next) from quantum (have one more in a lifetime). A bonus announced in May that moves September conceptions is partly timing. FOR's 'the lever exists' is true and not enough. The claim is that we should pay. 'Should' has a price. If you must pay ten infra-marginal parents to buy one extra child, the fiscal externality that was supposed to justify the policy has to be enormous to survive. Childcare and leave are cousins that might change the opportunity cost rather than the sticker. This page is the sticker. The sticker's yield, where it has been measured, is modest and dear.

Key assumptions

  • A large fraction of observed TFR bumps are timing rather than extra lifetime children partial
  • Cost per additional lifetime birth, not the existence of a positive elasticity, is the right test of 'should' untestable

Red team — the strongest counters

Pigou does not require cheap — it requires the externality to exceed the cost

A high cost per extra birth is the start of the calculation, not the end. If the present value of a future taxpayer is large enough, dear is still worth it. C.D. Howe did not close that sum.

Timing still has value if the support ratio is this decade

A birth pulled forward is a worker pulled forward. Tempo is not quantum, and it is not nothing. The claim can be justified as bringing forward the cohort you were going to get anyway.

Sources

Confidence, decomposed

Logical validity●●●●●
Premise support●●●●○
Representativeness●●●●●
Source quality●●●●●

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · We should pay people to have children
Empirical — moderateP1

Korea and Singapore already wrote very large cheques — the rate did not come back

Statistics Korea's TFR was 0.72 in 2023, 0.78 in 2022, with a small later tick. The Guardian, reporting the 2023 release, noted more than 360 trillion won (about USD 270 billion) in schemes since 2006 — cash, babysitting, infertility support. Jales and co-authors, on an earlier Korean baby-bonus programme in the Journal of Asian Economics, estimated that most of the disbursements paid for births that would have happened anyway. Singapore has run a Baby Bonus (cash gift plus co-matched Child Development Account) for years; the National Population and Talent Division puts resident TFR at 0.97 in 2023 and 2024, with a preliminary 2025 figure still lower. That is not 'we never tried a cheque.' It is the two richest, most administratively capable East Asian states pulling the lever this page is about, plus cousins (housing, leave), and remaining at the floor of the world table. Hungary is the European specimen: TFR rose from about 1.23 around 2011 to about 1.61 in 2020–21 under a large family package, then fell to 1.39 in 2024 (KSH, as reported). A bounce that gives back is not replacement. AGAINST's claim is not that prices are zero. It is that the cheque, at sizes governments actually write, does not buy a 2.1.

Key assumptions

  • Korea and Singapore are fair tests of 'pay people to have children,' not unique culture boxes where money cannot speak partial
  • A TFR stuck near 1.0 after large spend is a failure of the claim, not a reason to write a still larger cheque untestable

Red team — the strongest counters

Korea is a housing-and-hours bundle, not a clean cheque test

Work culture, apartments, and gender are in the residual. Treating 0.72 as a finding about a baby bonus is how you ignore the cousins the scope left out — and those cousins may be the actual constraint.

Hungary's rise is the lever; the later fall is not a zero

1.23 to 1.61 is a move. Giving some of it back by 2024 is a bounce, not proof the package never worked. FOR's 'the lever exists' survives a partial give-back.

Sources

Confidence, decomposed

Logical validity●●●●●
Premise support●●●●○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · We should pay people to have children
Logically validP1

If the object is workers, immigration is the cheaper instrument

National Transfer Accounts make the timing brutal. Children consume; workers produce; the old consume again. A birth worsens the support ratio for years before it helps. A working-age migrant improves it on arrival. If the claim is justified by pensions and wards, AGAINST is allowed to pick the cheaper way to fill a roster. Convergence of immigrant TFR toward the host rate is real and cuts the 'import a high-fertility culture' fantasy — and it also means you can keep the inflow targeted on workers without pretending you have bought a birth rate. Whose children is in the fight, as the scope says. Paying for births is often, in practice, a preference about who the next cohort is. That preference should be argued as a preference, not laundered through NTA arithmetic that immigration can meet at lower cost. This is not the immigration-economics page. It is the claim that a baby bonus is not the instrument the fiscal story implies.

Key assumptions

  • The live political object of the cheque is the support ratio, not a named people's reproduction untestable
  • A large working-age inflow is politically and administratively available on the relevant horizon partial

Red team — the strongest counters

Immigrants age, and the inflow has to be permanent

A 25-year-old today is a 75-year-old in the NTA. Unless you keep the tap open forever, you have imported a pension, not solved one. Births compound inside the population.

A roster is not a political community

The immigration page already fights the economics. This page's 'whose children' is allowed to be the object. Calling the border cheaper assumes the object was only this year's tax base.

Sources

Confidence, decomposed

Logical validity●●●●●
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · We should pay people to have children
Logically validP1

The opportunity cost is the point — a cheque does not buy a second shift back

UN WPP's global TFR decline is the fact FOR starts with. AGAINST's reading: development worked. Child mortality collapsed, contraception spread, schooling lengthened, women's wages rose, housing in the cities that have the jobs got dear. Those are not market failures to be Pigouvian-taxed with a baby bonus. They are why a second child lost to a career in Seoul or Singapore. A cheque that does not buy a second bedroom, a grandmother, or a job you can leave at 5 p.m. is a sticker on the wrong price. France's relatively higher European TFR, when it had it, was more cousin than cheque — childcare, schools, allocations. The scope of this page leaves those cousins out of the claim on purpose. That is a problem for FOR. If the real constraint is time and housing, the policy that might work is not this claim, and the policy that is this claim (Korea's cash, Singapore's Bonus, Australia's lump sum) is the one that stayed modest. Reproductive autonomy is the moral restatement: a state that pays for births is a state with a preferred family size. People who wanted zero should not have to fund a natalist target. The pension shortfall can be met with later retirement, higher productivity, and migration — tools that do not write a price on a uterus.

Key assumptions

  • Low TFR is mostly desired given current prices of housing, time, and careers, not a massive unmet demand for children that a cheque would reveal partial
  • Using public money to raise births is a illegitimate preference about other people's families untestable

Red team — the strongest counters

If opportunity cost is the point, the cheque is how you change the relative price

AGAINST's development story is FOR's demand curve. A payment is what you do when the alternative got better. Saying 'other lives got better' is the mechanism, not a stop.

A bonus is optional; a ban on contraception is not

Autonomy objections that treat a cheque like a command collapse the difference. People who want zero keep their time. The pension can still be everyone else's problem — that is what a public good looks like from the person who does not provide it.

Sources

  • World Population Prospects 2024 Global TFR decline with development, education, and urbanisation as the long backdrop. Pass-1: WPP exists. AGAINST uses the development-fertility association, not a single causal paper. P1 checked
  • Singapore Marriage & Parenthood package vs TFR 0.97 Standing Bonus plus leave and housing; TFR still ~1. Pass-1: population.gov.sg already used. Specimen of cousins-plus-cheque together, still at the floor. P1 checked

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
AGAINST · We should pay people to have children
Plausible, low testabilityP1

A bonus is a selection device dressed as a public good

Becker's quantity-quality trade-off is the standard demographic-economics objection: a price that raises the number of children can lower the resources per child. You do not need the 1960 paper to see the live version. A lump-sum bonus is a larger share of income at the bottom of the earnings distribution. Higher-earning parents were less likely to be on the fence. Quebec's larger third-child payment moved higher-parity births — which is what you would write down if the cheque bit on households already having children in harder circumstances. AGAINST does not need a dysgenic cartoon. It needs the admission that 'pay people to have children' is not 'pay a representative parent.' Hungary's package layered marriage rules, loan forgiveness, and tax holidays for mothers of several children — a design that names who should have the babies. The claim on this page is the cheque, not Orbán. The cheque still selects. Childcare and leave, the cousins, at least buy time for people who already wanted the child and were blocked by a job. The sticker buys the child from whoever's reservation price you hit.

Key assumptions

  • The marginal extra births under a bonus have, on average, lower resources per child than infra-marginal births partial
  • That selection is a reason not to pay, not only a design problem to be taxed away untestable

Red team — the strongest counters

Higher-parity responses can be people who already wanted the child

A third-child bonus that moves third children is not automatically a quality drop. It can be the household that was blocked by the sticker. Selection is a distribution of reservation prices, not a moral ranking of parents.

Paying infra-marginal people is every universal benefit

Schools, child benefit, and pensions all pay people who would have used them anyway. Cost-per-extra is a standard that would kill half the welfare state. It is not unique to the claim.

Sources

Confidence, decomposed

Logical validity●●●●○
Premise support●●●○○
Representativeness●●●●○
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · We should pay people to have children
Empirical — moderateP1

Most rich countries are already below replacement, and the map is still filling in

UN World Population Prospects 2024 puts the global total fertility rate a little above two (UN DESA summary: 2.25 live births per woman; the 2024 World Fertility report uses about 2.2). Replacement is about 2.1 in low-mortality countries. In 2024 more than half of countries and areas — 131 of 237 in the World Fertility 2024 count — were already below 2.1, covering about two-thirds of world population. East Asia is the sharp edge: Statistics Korea's TFR was 0.72 in 2023, the lowest national rate in the tables, after a long decline despite large pronatal budgets. Southern Europe and East Asia are not a US panic. Latin America has been joining them. FOR's first move is not a forecast of extinction. It is that below-replacement fertility, sustained, is how workforces, armies, and pay-as-you-go pensions shrink relative to the old. The claim is a cheque or a tax for births. Childcare, housing, and leave are cousins. Specimens (Hungary, France, Korea, Singapore) are not the claim's country — they are where the policy has been tried.

Key assumptions

  • Below-replacement TFR is a problem that policy should try to move, not only a revealed preference to be accepted untestable
  • National TFRs from UN WPP and statistical offices are the right object, not completed cohort fertility a decade later partial

Red team — the strongest counters

Desired fertility may already be the TFR

If people wanted more children than they have, the cheque would be a revealed-preference gift. A lot of the collapse is wanted — smaller families after child mortality fell. Paying against a preference is not a census fact.

Productivity and later retirement also close a support-ratio gap

Below-replacement is a problem for a pay-as-you-go formula that nobody is willing to rewrite. That is not the same problem as 'we should pay for births.' The formula is the cheaper edit.

Sources

  • World Population Prospects 2024 / World Fertility 2024 UN DESA: global TFR about 2.2–2.25 in 2024; 131 of 237 countries/areas below 2.1. Pass-1: population.un.org/wpp and World Fertility 2024 PDF exist. Use the UN range, not a fake single decimal. P1 checked
  • Statistics Korea / reporting of 2023 TFR 0.72 Korea TFR 0.72 in 2023 (0.78 in 2022); later 2024 tick-up around 0.75 in subsequent releases. Pass-1: Guardian 28 Feb 2024 reporting of Statistics Korea, and later official releases. Specimen of the floor, not the claim's country. P1 checked

Confidence, decomposed

Logical validity●●●●●
Premise support●●●●●
Representativeness●●●●●
Source quality●●●●●

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · We should pay people to have children
Empirical — moderateP1

Quebec and Australia already wrote the cheque — births moved

Quebec's Allowance for Newborn Children (1988–1997) paid a few hundred dollars for a first birth and up to C$8,000 for a third. Milligan (Review of Economics and Statistics, 2005) uses the introduction as a natural experiment and finds a fertility effect in the high-single-digit to low-double-digit percent range, larger for higher-parity births — later citations put third-child responses up to about 25%. The C.D. Howe backgrounder on the same programme said the quiet part: it worked, at a high cost per extra birth. Australia announced a Baby Bonus in 2004 (initially A$3,000). Drago, Sawyer and co-authors (Melbourne Institute, 2009), using HILDA, estimate that fertility intentions and then births rose modestly. FOR should not turn 'modest' into a baby boom. The steelman is narrower: the quantity demanded of children is not a cultural constant. It moved when the cheque moved. Design — size, duration, indexation, whether it is a lump sum or a tax credit — is the live argument. Childcare and leave are cousins; they are not this claim. Korea's later collapse despite huge spend is the specimen AGAINST will use. FOR's reply: Korea is a bundle of housing, work culture, and gender, not a clean test of a cheque.

Key assumptions

  • The extra births were additional lifetime children, not only births pulled forward a year partial
  • A modest percentage rise is worth the fiscal cost of paying for the births that would have happened anyway untestable

Red team — the strongest counters

Tempo is not quantum

A bonus that moves a birth from 34 to 32 does not add a lifetime child. Milligan and Drago have to be read with that split. Modest plus timing is not a warrant for 'should.'

High cost per extra birth is the finding, not a footnote

C.D. Howe already said Quebec worked and was dear. You pay the infra-marginal majority to buy the marginal extra. Pigou that fails a cost test is a subsidy to people who were going to parent anyway.

Sources

  • Subsidizing the Stork: New Evidence on Tax Incentives and Fertility Milligan, Review of Economics and Statistics 87(3):539–555, 2005. Quebec ANC; fertility responses, larger at higher parity. Pass-1: REST/NBER w8845 exist. Magnitudes commonly cited 5–12% overall, up to ~25% for third children — treat as the paper's range, not a precision fetish. P1 checked
  • Did Australia's Baby Bonus increase the fertility rate? Drago, Sawyer, Sheffler, Warren & Wooden, Melbourne Institute Working Paper 1/09, 2009. HILDA; modest rise in intentions and estimated births after the 2004 announcement. Pass-1: Melbourne Institute / APO page exists. P1 checked

Confidence, decomposed

Logical validity●●●●○
Premise support●●●●○
Representativeness●●●●●
Source quality●●●●●

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · We should pay people to have children
Logically validP1

Other people's children pay other people's pensions

National Transfer Accounts — Lee, Mason, and the NTA network — measure how consumption in childhood and old age is paid for by labour in the middle. The economic support ratio falls when fertility has been low for a generation, because there are fewer workers per retiree. That is not a metaphor. It is the arithmetic of pensions, health, and long-term care in every country that taxes workers to pay the old. A couple that has a third child produces a future taxpayer whose benefits spill to people who had zero. The parents pay the nights and the lost wages. FOR's cheque is Pigou, not sentiment: if you want the fiscal externality internalised, you pay for it. Immigration is in this fight. A working-age arrival can fill a slot this year. That arrival also ages, often brings family, and does not, by itself, restore the age structure a generation out unless you keep the inflow large forever. Paying for births and recruiting workers are different policies. The claim is the first.

Key assumptions

  • Pay-as-you-go old-age systems will still be how rich countries fund retirement on the relevant horizon partial
  • The fiscal externality of a birth is large enough, after education costs, to warrant a cheque rather than only a tax cut on work partial

Red team — the strongest counters

Education and years of consumption can swamp the pension gain

NTA is a two-edged sheet. A child is expensive for a long time in a rich country. The fiscal externality is an empirical magnitude, not a sign you can read off the existence of pensions.

Immigration is the instrument that moves this decade's ratio

A birth worsens the support ratio for years. If the brochure is pensions, AGAINST is allowed to pick the worker who already exists. 'Whose children' is then the real claim, and it should be named.

Sources

  • National Transfer Accounts Project Lee–Mason NTA: age profiles of labour income and consumption; support ratios and fiscal support ratios across countries. Pass-1: ntaccounts.org exists. P1 checked
  • Support Ratios and Demographic Dividends: Estimates for the World Mason, Lee, Abrigo & Lee, UN DESA Technical Paper 2017/1. Support-ratio estimates for 166 countries. Pass-1: UN DESA/PD landing page exists (the guessed 2020/Feb PDF path 404'd — replaced with the Population Division record). P1 checked

Confidence, decomposed

Logical validity●●●●●
Premise support●●●●○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · We should pay people to have children
Logically validP1

Immigration fills a roster this year; it is not the same policy as a birth

The alternative to a baby bonus is almost always 'then let more people in.' FOR has to take that alternative at full strength. A 25-year-old who arrives tomorrow is in the tax base tomorrow. A baby born tomorrow is a net fiscal cost for years. If the only object is next year's support ratio, immigration wins the race. The claim is not next year. Below-replacement fertility for a generation is a shrinking native-born youth cohort, a political community that did not choose to replace itself, and — in every NTA country — an immigrant TFR that, after a lag, looks like the neighbours'. You can run a large inflow forever. That is a different country, and a different claim, already fought on the immigration-economics page. Paying people to have children is the attempt to have the next generation without making the age structure a permanent function of the border. Both can be done. Treating them as one policy is how the cheque gets talked out of the budget.

Key assumptions

  • Who has the children, not only how many workers this year, is a legitimate object of the policy untestable
  • Immigrant-origin fertility converges far enough, fast enough, that inflows do not permanently clone a high-TFR society partial

Red team — the strongest counters

If the object is workers, the border is a substitute

FOR's 'different country' is a preference. For the NTA arithmetic it is using, a working-age inflow is the same slot. Calling that not-a-substitute is how you smuggle the preference.

Immigrant TFR convergence cuts both ways

Yes, the second generation looks like the host. That is a reason a baby bonus for 'our' births is not required to keep a high-fertility culture — the culture was not going to be cloned by the border anyway.

Sources

Confidence, decomposed

Logical validity●●●●●
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.
FOR · We should pay people to have children
Logically validP1

Parents already pay a tax that non-parents do not — the bonus is the rebate

Opportunity cost is the real price of a child in a rich country: years out of the labour market, slower wage growth, housing that has to be bigger. That cost lands on parents, and in every labour-force survey it lands harder on women. Childcare and parental leave are the cousin policies that try to shrink the cost. The claim is the cash or tax for the birth itself. FOR's consistency claim is simple. If the rest of us collect the fiscal dividend of that child at 30, we are already in the parents' business. A bonus is the honest form of that business. It is not a command to give birth. People who do not want children keep their money and their time. People who are on the margin of a second or third child — Quebec's higher-parity response is the empirical hint — face a smaller private loss. Quality-versus-quantity (Becker) is the standard objection: pay for births and you get more births from households with less of whatever you wanted. FOR can grant selection and still say: the current policy is to select for parents who can afford the whole private cost, which is also a selection, just an accidental one. Hungary's loan-and-tax package is a specimen of a large cheque plus marriage rules; it is not the only design, and it is not this page's country.

Key assumptions

  • A voluntary bonus is not coercive in the way a ban on contraception would be untestable
  • Paying the people on the margin does not, on net, lower the human capital of the next cohort enough to cancel the fiscal gain partial

Red team — the strongest counters

Cash does not buy a second shift or a second bedroom

Opportunity cost is time, career, and housing. A lump sum that does not move those prices is a sticker on the wrong good. Korea and Singapore already wrote larger stickers.

A natalist rebate is still a state with a preferred family size

Optional on paper; budgeted as a target in practice. People who wanted zero fund a fertility rate. That is a political choice about other people's bodies, not a rebate of a tax that nature levied.

Sources

  • Subsidizing the Stork (higher-parity response) Milligan 2005: larger fertility response at higher parity under Quebec's larger third-child payment. Pass-1: already used; this arg uses the 'margin of the extra child' not the headline percent. P1 checked

Confidence, decomposed

Logical validity●●●●●
Premise support●●●○○
Representativeness●●●●●
Source quality●●●●○

Provenance

Generated by a paired steelman agent (single model family) · red-teamed by an independent adversarial agent · sources Pass-1 spot-checked (existence and rough fit) — framing-fidelity not independently verified. Judged on merit: per the founding rule of this project, AI authorship is disclosed at site level and arguments stand or fall on their content.