Fertility Requires Margin

Why family policy fails when it buys the child but not the life around the child

Elias Kunnas

Fertility requires margin. A child is long-term demographic capital for society and a short-term margin shock for the household — money, housing, sleep, time, health, career, relationship stability, and cognitive bandwidth all tighten at once. Fertility collapses when the act that increases public margin becomes locally margin-negative for the agents expected to perform it. Pronatalist policy fails when it buys the long-horizon asset without repairing the short-horizon margin the household actually sees.

Standard objections addressed in this essay
  • "Cash works — look at Hungary, Israel, France" — §IV (cash can move births; treating cash as the universal binding constraint is the error)
  • "People just don't want children anymore" — §III (stated preferences are an integrator of architecture, not a separable binding input)
  • "This reduces to 'restore marriage'" — §VI (commitment technology is a structural argument about time horizons, not a moral claim about union types)
  • "You can't engineer fertility" — §VI (old reproductive cultures engineered it; modern equivalents are the open question)
  • "This ignores immigration" — separate replenishment mechanism with its own integration, housing, trust, and fiscal ledgers; doesn't repair the fertility production function analyzed here (see Full Accounting)
  • "This is just what happens when women get educated and enter the workforce" — §III, §V (the education and labour-force shifts act through inputs 3, 4, and 5; France has comparable rates of both with a meaningfully higher TFR, which says the same gradient produces different outcomes under different architectures)

I. The Two Ledgers

A child appears on two ledgers at once. On the public ledger, the child is demographic capital — a 20-to-30-year asset whose payoff arrives in tax base, care capacity, institutional renewal, and the simple continuity of the society. On the household ledger, the child is a present-tense draw against money, housing, sleep, time, health, career momentum, and the relationship's bandwidth. The two ledgers describe the same human, but they integrate over wildly different horizons.

In many agrarian and kin-based societies the two ledgers ran much closer together. Children were household labor, insurance, identity, lineage, and old-age support on the same timeline as the household itself. The state did not need to coax fertility in the same way because reproductive architecture already embedded children inside household economy, kinship, status, and old-age security. This wasn't uniformly margin-positive — famine, infant mortality, dowry pressure, and inheritance fragmentation all complicated the picture — but the structural alignment between household interest and demographic outcome was tighter than today's.

The modern child sits differently. Pensions, healthcare, schools, and dependency-ratio arithmetic moved the economic return on children onto the public ledger. The cost stayed on the household ledger; the benefit accrued elsewhere. The household paying the bill is no longer the household collecting the return.

Pronatalist policy notices the public ledger and tries to fix the household ledger by writing checks. The bet is that a transfer at birth is sufficient to flip the household sign from negative to positive. Sometimes that bet works. Often it does not. The essay is about why.

II. The Short Horizon

People make life decisions by looking at the visible near-term metric — the part of the integral they can actually see — and asking whether that part stays positive. Long-horizon optimization is what economists assume in models, not what households do at the kitchen table.

Behavioral finance has been pinning this down for decades. In consumer installment debt — auto loans, most clearly — demand responds more strongly to monthly payment size and maturity than to long-run present-value cost: borrowers target the legible part, not the integral.

The same machinery runs in the fertility decision. The question a prospective parent actually answers is not the one the state's projections imply:

"Is this child net-positive for the national demographic stock over twenty-five years?"

It is the one their life actually puts in front of them:

"Will this break the next year of my life?"

The first question has an obvious answer and almost no operational weight. The second has a difficult answer and almost all of the operational weight. The state's instrument operates on the long integral; the household's on the short. Policy that pays out on the long ledger does not automatically change the number the household uses.

III. The Fertility Production Function

Births are the output of a production function with several inputs, all of which must clear for the birth to happen. The scope here is bounded: the production function below describes the missing, intended, stable-family births in modern rich societies — the gap between what people say they want and what they end up with. It does not claim to explain every human birth across history.

Seven inputs must hold simultaneously:

  1. Exposure. A plausible co-parent exists. The mating market has produced a person with whom a household could be formed at all.
  2. Commitment horizon. The union is stable enough that an irreversible twenty-year investment makes sense. A child is the canonical relationship-specific asset; the discount rate the household applies to it tracks the perceived probability the cooperation survives.
  3. Timing runway. The first birth happens early enough that a second or third child remains feasible. Delay does not preserve completed fertility; it compresses it.
  4. Present margin. Money, housing, sleep, time, health, career slack, and relationship bandwidth do not go negative when the child arrives.
  5. Cultural permission. Parenthood is legible as a normal life stage rather than as either a heroic sacrifice or a status forfeit.
  6. Cognitive legibility. The path feels executable now. Not merely beneficial in the abstract — actually doable starting Monday.
  7. Biological feasibility. The reproductive window has not been consumed by delay in any of the upstream inputs.

Each input is a real binding constraint in some household somewhere. The inputs multiply rather than add: a household with comfortable income, good housing, and an excellent daycare option but a broken commitment horizon will produce far fewer births than a household with a stable union, modest income, and modest housing. In the extreme — when one input is completely absent, like no plausible co-parent at all, or a reproductive window already closed — no amount of generosity in the other six produces the birth. Inside the normal range there is real substitution at the margins (more cash genuinely loosens the cognitive load of weak cultural permission), but the seven inputs are not freely interchangeable.

The diagnostic question becomes: for the marginal non-birth, which input is binding? A policy that relaxes a non-binding constraint produces nothing. A policy that relaxes the binding one produces a birth. Which input is binding matters more than how generously the others are funded.

The serious version of "people just don't want children" comes from Becker-style revealed-preference economics: as women's earning potential rose and contraception became reliable, the opportunity cost of time spent in childrearing rose with them, and households rationally chose fewer children. This is a real mechanism. It operates squarely on inputs 3, 4, and 5 of the production function — and France has similar female labor-force participation and education to Finland with a TFR meaningfully higher than Finland's (about 1.56 vs 1.30 in 2025, and historically a wider gap). The same opportunity-cost gradient produces different outcomes under different architectures. Opportunity cost is one of the inputs; it is not the unique binding one.

The colloquial reading that "people just don't want children" is best understood as a downstream report. When several inputs are chronically near zero, prospective parents stop reporting a desire for children. Much of what appears as preference collapse is the production function felt from the inside. Some of it is genuine preference change — the claim is not that every childfree preference is secretly a blocked birth, but that policy should first distinguish genuine non-desire from desire made non-executable.

A 2026 survey by the Finnish Institute for Health and Welfare (THL) — the FinChildren release — points the same way. The survey asked parents who had a child in 2024 whether they wanted more. Among those who were unsure, the most common reason for hesitating was coping: exhaustion, sleep deficit, the load of present caretaking — not absence of desire. The data describes what the present-margin input looks like once the first child has actually arrived. Whether it generalizes to the still-childless deciding about a first birth is a different question, but the mechanism the survey surfaces — desire blocked by load, not absence of desire — is exactly what the production-function frame predicts.

IV. Why Cash Buys Timing, Not the Channel

Per-child cash transfers act on a single input — present margin — and even there, only at the level of one item in the basket. The size and durability of the effect depends on whether the missing birth was actually blocked by a cash-at-birth threshold.

One technical note before the evidence. The headline fertility statistic everyone quotes — "Finland's TFR was 1.30 in 2025" — is the period total fertility rate: a snapshot built from one year's age-specific birth rates and projected onto a hypothetical woman. It is a noisy diagnostic, because if real women shift the timing of their births earlier or later, period TFR moves without their completed family size changing at all. The number that actually matters for population continuity is completed fertility per birth cohort — knowable only decades after the fact. Period TFR is what policy debates fight over; completed cohort fertility is what determines whether a society reproduces itself.

The OECD's 2024 synthesis is the cleanest summary of the international evidence: most cash-benefit interventions have no or only modestly positive effects on fertility, and positive effects are often transitory. The literature distinguishes two effect types:

When cash transfers have visible fertility effects, they most cleanly move tempo. Hungary's package raised TFR from 1.23 to a 2021 peak of 1.61 before falling back to 1.38 by 2024 — tempo exhaustion, not a structural win. South Korea is the sharpest warning: after years of substantial pronatalist spending across cash and institutional hardware (daycare, youth housing loans, fertility services), TFR fell to 0.72 in 2023, with a modest 2024–25 rebound to ~0.80 attributable to marriage catch-up rather than structural recovery. Cash and services together failed to repair the binding upstream architecture.

France is the most frequently cited cash-friendly architecture. TFR has run noticeably above Finland's for the past two decades under a tax and benefit system that includes the quotient familial — a continuous tax-relief structure that scales with the number of children and runs across each child's dependent life. France too is now declining: TFR fell to about 1.61 in 2024 and to 1.56 in 2025, a post-WWI historic low. The current France–Finland gap is about 0.26 children per woman; the historic gap was closer to half a child.

The plausible read is that France's architecture relaxes the present-margin input more durably than a one-time baby bonus does, because it operates as a continuous tax structure across the child's life rather than as a single transfer at birth. That is exactly the form a margin-translation argument predicts can work. The recent French decline is consistent with the broader production-function frame: a well-architected cash lever helps, but it cannot compensate alone when upstream inputs erode under the same modern pressures everywhere else.

Cash moves what cash can reach: visible household margin at the moment of decision and across the years following. Cash underperforms when the missing birth is not blocked by a cash-at-birth threshold. In rich, high-support societies, most of the missing births are blocked elsewhere.

The form of a cash transfer determines what it can buy. A one-time bonus at birth touches the present-margin input once; a continuous, parity-scaled tax structure touches it across the child's whole dependent life. Intergenerational baby-fund proposals that lock returns until the child reaches their own parenthood operate on a third horizon — they move the public ledger now and the household ledger only two generations later, which means the household deciding today does not literally see the future payout when it weighs the decision. Such instruments can plausibly affect fertility through indirect channels (expectation effects, status signals, public-ledger booking of demographic capital), but those channels would have to be demonstrated rather than assumed. The mechanism-analysis framework calls this Type 3-risk: a proposal whose implicit causal model treats a non-binding input as the binding one. The risk applies to any pronatalist instrument whose target node is asserted rather than mechanism-tested.

V. The Nordic Diagnostic

Finland is one of the cleanest diagnostics for the limits of conventional family-policy hardware. Over 3% of GDP flows through family benefits and services — placing Finland among the OECD's high-spending family-policy systems — with 14 months of paid parental leave, near-free daycare, comprehensive healthcare, and low maternal mortality. Total fertility was 1.25 in 2024 and 1.30 in 2025 — historic lows, despite the family-policy infrastructure being already built.

Finland has solved several visible family-policy constraints. The constraints that remain are less legible, and they sit upstream of where conventional policy looks.

The Flux Consortium's 2026 brief is the sharpest published diagnostic on the residual constraints. About 82% of the 2010–2024 fertility decline is attributable to declines in first births. Lifetime fertility is projected below 1.56 for early-1990s female cohorts. Postponement of first births is not being compensated later. The European Journal of Population analysis of the Finnish decline finds that lower fertility within unions explains roughly three-quarters of the first-birth drop, while union-formation declines and rising cohabitation dissolution explain roughly the remaining one-quarter. The decline reflects both fewer unions and existing unions less often converting to first births.

The failure of existing unions to convert to first births is the load-bearing finding. Even when exposure exists and a union is in place, the channel from union to first birth does not consistently open. The blockage sits in some combination of commitment horizon, timing runway, present margin, and cultural permission — the upstream inputs — rather than at the cash-at-birth threshold.

A follow-up by the same research group (Hellstrand, Nisén, and Myrskylä, European Sociological Review, 2024) sharpens this further: the fertility decline was weaker in fields with stable lifetime-earnings paths (health, welfare, education, agriculture) and stronger in fields characterized by economic uncertainty (ICT, arts, humanities). Uncertainty-related factors explain roughly two-fifths of the first-birth decline. The blockage localizes exactly where the production function predicts it would — in the inputs carrying long-horizon commitment capacity, not at the cash window.

The internal Finnish variance supports the same reading. Same laws, same benefits, same currency: in 2024, the highest-fertility Finnish region recorded a TFR of 1.72 while the lowest sat at 1.06. The variation cannot be explained by national cash policy alone. It points toward upstream ecology — partner formation, family-compatible housing, intergenerational density, religious and cultural norms, and permission for early parenthood. Where that ecology is stronger, the same national cash lever sits on top of a more functional production function. Where it has eroded, the cash lever pays for the wrong half of the system.

VI. Structural Empathy

Two strategies are available for getting humans to do future-oriented things under short-horizon cognition. Call them cognitive empathy and structural empathy.

Cognitive empathy says: imagine the future strongly enough to override the present. Picture the child, picture the society, picture the demographic capital — and let the picture move you to bear the present cost. This is the strategy of exhortation. It is what pronatalist op-eds and political appeals to civic duty severed from institutional architecture have always asked of households. It is also, empirically, weak. People do not make the marginal life decision by solving the long-horizon optimization. The picture is not a lever the household reaches for.

Structural empathy says: build the defaults, norms, and material arrangements so that the present self does not need heroic long-horizon imagination to do the future-oriented thing. The clearest analogue is retirement saving. Workers did not become better intertemporal optimizers between 1990 and 2010. Auto-enrollment changed the default, and the future-oriented behavior followed without any cognitive upgrade. The architecture compensated for the absence of long-horizon cognition by making the future-oriented action locally easy.

This is the same move Ethics Is an Engineering Problem makes for ethics: stop expecting saints, build the game so the incentive-compatible action is the correct action. Applied to fertility:

A society cannot build demographic capital by asking parents to be reproductive martyrs. Where present-tense parenthood demands heroic sacrifice, low fertility is not a moral failure by households. It is an architectural failure by the society.

This frame also dissolves the "future-belief" reading of the fertility crisis. Future-belief functions as an integrator variable — a summary statistic households produce when the architecture either does or does not make the future executable in the present. People who say they cannot project a livable future for their children are reporting that the production function around them is broken in inputs 1 through 6, not making a metaphysical claim about civilization.

The historical reproductive architecture did structural empathy by default. Early pairing, marriage as a public commitment-extension technology, religious framing, kin networks, intergenerational caretaking, lower parenting standards, gender roles, inheritance arrangements, communities enforcing family norms. None of these had to be argued for inside a 28-year-old's head. They were features of the environment. The 28-year-old inherited them as defaults.

Industrialization, modern markets, and the welfare state replaced many of those defaults — often for good reason, sometimes for none. What replaced them rarely performed the same architectural function. The result is a society that needs demographic continuity but expects each prospective parent to construct the entire commitment apparatus from scratch, while running a full career and a private life under high cognitive load.

A small example of the same mechanism running in the opposite direction: ubiquitous phone cameras turn bounded social contexts into permanent public surfaces. The teen at a dance is no longer moving in front of the room; she is moving in front of a possible future audience. This is structural anti-empathy — environments that make normal experimentation more surveilled, more reputationally risky, and less reversible. The downstream effect on adult life is what the production function predicts: when the cost of looking awkward in any one bounded context goes up, people practice less and arrive at intimacy, commitment, and fertility decisions with less embodied rehearsal of the moves they require.

Marriage in this framing is a historical commitment-extension technology — a public arrangement that extends the household's effective planning horizon enough to make irreversible co-investment rational. The function still has to be performed by something. Modern arrangements — late household formation, cohabitation without a public commitment step, atomized parenting — perform that function unevenly, and on average less reliably for first-birth conversion than the architectures they replaced. This is a narrow functional claim about fertility and first-birth conversion, not a claim that the older architecture was morally superior or that exit rights, contraception, education, or women's labour-market autonomy should be reversed. The Finnish data on declining first-birth conversion inside existing unions is consistent with the functional reading. The question is not whether to restore the old form; it is whether a post-traditional society can build an equivalent function without restoring the old constraints. Commitment architecture only works if it shares the risk without taking away autonomy; if it turns into lock-in, it does not lower the risk of fertility — it raises it.

VII. What Would Move the Margin

Specifying a full policy package is past the scope of one essay. The essay's contribution is the invariant by which proposed packages can be judged:

The invariant. Does this intervention change the margin the decision-maker actually sees, at the moment of decision, on the household ledger? Or does it move a number on the public ledger and hope the household notices?

Applied to the standard menu:

Cash at birth moves the household ledger weakly and briefly. Most of the spend transfers to inframarginal recipients — couples who would have had the child anyway. Buys tempo more reliably than quantum.

Continuous parity-scaled tax relief, of the French quotient familial type, moves the household ledger durably and visibly across the child's life. Probably the most defensible cash-form intervention, because it is the form a household actually integrates over.

Housing geometry — supply of family-compatible, near-work, affordable space — touches the durable present-margin input directly. The trap to avoid: demand-side housing subsidies that capitalize into rents without changing supply, leaving the household ledger unchanged at higher fiscal cost. The Goodhart hazard is also real — a housing agency tasked with "increasing supply" will optimize the legible metric (unit count) rather than the latent one (family-compatibility). The intervention has to specify what it is buying or it will buy generic units instead.

Care-time and load reduction — daycare quality not just availability, parental sleep infrastructure, employer flexibility norms, pension credits that recognize unpaid caretaking. Targets the present-margin and cognitive-legibility inputs. Plausibly the highest-leverage cash-adjacent lever for the marginal hesitating household. Same Goodhart hazard as housing: agencies will measure what is easy (availability, hours, ratios), and parents will live what is hard (the actual daily load).

Commitment-architecture repair — removing welfare cliffs that penalize household formation, joint-taxation neutrality, pension accrual that recognizes the family as a productive unit. Removes architectural disincentives rather than creating new transfers. Politically cheaper because it is undoing a distortion. Touches the commitment-horizon and exposure inputs.

Timing-runway expansion — anything that makes the first birth available earlier within partnered, prepared couples: housing accessible to people in their twenties rather than thirties, educational architectures that leave a reproductive window intact, employer norms that protect rather than penalize early-career parenthood. Acts as a mechanical multiplier on completed fertility, because earlier first births leave time for second and third children.

Cultural-permission rebuilding — slow, indirect, mostly outside the policy aperture. The diagnosis here is honestly larger than the prescription: religion, kin density, marriage norms, and the social legibility of parenthood are architectural variables that policy can nudge at the margin but cannot legislate. Public discourse that depicts parenthood as load-bearing rather than as a survival story does structural-empathy work even when it looks like commentary; that's most of what the cultural layer is reachable through.

None of these is sufficient alone, and the marginal binding input varies by household. The Finnish evidence in §V points to the upstream commitment-and-timing cluster — existing unions failing to convert to first births — as the load-bearing blockage in that case; other rich-country fertility-decline corridors point to closely related upstream inputs. The constructive claim is narrower than a single magic lever: every pronatalist intervention should be tested against the invariant — does this change the margin the decision-maker sees, at the input where the decision is currently blocked? Interventions that pass that test are worth funding. Interventions that fail it are paying for activity on the public ledger that the household ledger never sees.

The deeper repair is at the accounting layer. As long as demographic capital is booked at zero on the public ledger, every cost of depleting it is a cost no one is forced to recognize. A measurement institution that actually books the depreciation — and a constitutional architecture that creates consequences when it crosses thresholds — is what makes the binding-input analysis decision-relevant in the first place. Cash continues to win every budget cycle because cash is the only lever the existing accounts can see.

The argument in three sentences. A child appears on two ledgers — long-horizon demographic capital for the society and short-horizon margin shock for the household — and modern societies pay out on the wrong one. Pronatalist cash transfers act mostly on the present-margin input and move tempo more reliably than quantum, while the production function for missing births is bottlenecked upstream in commitment horizon, timing runway, cultural permission, and care load. Restoring fertility requires structural empathy — repairing the architecture that lets a present-self do the future-oriented thing without heroic imagination — rather than buying the long-horizon asset and hoping the household ledger updates.

Sources and Notes

The fertility numbers.

  • Statistics Finland reports TFR 1.30 in 2025 after 1.25 in 2024 — the historic Finnish lows. The pre-2010 Nordic fertility consensus has unwound; Finland, Norway, and Iceland have converged toward Southern European levels despite retaining their family-policy infrastructure.
  • The Flux Consortium's 2026 brief on Finnish first births is the sharpest published diagnostic: first births account for about 82% of the 2010–2024 decline, lifetime fertility is projected below 1.56 for early-1990s female cohorts, and postponed first births are not being compensated later. Flux Consortium.
  • Hellstrand, Nisén, and Myrskylä, "Less Partnering, Less Children, or Both? Analysis of the Drivers of First Birth Decline in Finland Since 2010" (European Journal of Population, 2022) finds that lower fertility within unions explained most of the Finnish first-birth decline, with union-formation declines and rising cohabitation dissolution as smaller but real contributors. This is the empirical backbone of the §V claim that the channel from union to first birth is what is failing.
  • Hellstrand, Nisén, and Myrskylä, "Educational field, economic uncertainty, and fertility decline in Finland in 2010–2019" (European Sociological Review 40(5), 2024) sharpens the §V cluster claim: the decline was weaker in fields with stable lifetime-earnings paths (health, welfare, education, agriculture) and stronger in fields with high economic uncertainty (ICT, arts, humanities); uncertainty-related factors explain roughly two-fifths of the first-birth decline.

The behavioral microfoundation.

  • Argyle, Nadauld, and Palmer, "Monthly Payment Targeting and the Demand for Maturity" (NBER Working Paper 25668, 2019), document monthly-payment targeting in consumer auto debt: borrowers respond more strongly to monthly payment size and loan maturity than to long-run cost. The auto-loan market shows what §II claims people do generally — they target the visible part of the integral, not the integral itself.
  • Madrian and Shea, "The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior" (Quarterly Journal of Economics 116(4), 2001), is the canonical default-architecture paper. Auto-enrollment shifts retirement outcomes more than any amount of long-horizon investor education. This is the analogue the structural-empathy section in §VI rests on.

The international cash-lever evidence.

  • OECD, "Fertility Trends Across the OECD: Underlying Drivers and the Role for Policy," Society at a Glance 2024. The synthesis cited in §IV: most cash benefits have no or only modestly positive fertility effects, and positive effects are often transitory; housing costs, opportunity costs, and changing norms recur as drivers.
  • Hungary's package is frequently cited as evidence that cash works at scale. The tempo-versus-quantum critique is summarized in N-IUSSP. The post-2021 trajectory has now made the tempo reading harder to dispute: Hungarian Central Statistical Office (KSH) figures show TFR falling from the 2021 peak of 1.61 back to 1.50 in 2023 and roughly 1.38 in 2024. Some quantum effect plausibly remains, but "Hungary proved cash works" does not survive the post-peak data.
  • South Korea is widely cited as evidence that pronatalist spending fails at scale. TFR fell to 0.72 in 2023, then modestly rebounded to about 0.75 in 2024 and around 0.8 in 2025 — partly on an echo-boomer cohort effect and post-COVID marriage catch-up, not a structural recovery. OECD's "Korea's Unborn Future" (2025) is a useful single overview. Korean policy is a mixed package: a large share of the spend went to daycare infrastructure, youth housing loans, and institutional services rather than direct transfers, while supply-side housing geometry in Seoul stayed constrained. The case is useful as evidence that state hardware plus cash together failed to repair the binding upstream architecture in Korea (intensive parenting, housing geometry, mating-market geometry, gendered care load) rather than as a clean indictment of cash specifically.
  • France's quotient familial is the strongest counter-example to a blanket cash-skeptic position. The relevant mechanism is that the relief is continuous and parity-scaled across the child's dependent life, not concentrated at birth: the French income tax divides taxable income into "shares" (parts fiscales), with additional shares granted per dependent child (an extra half-share for the first two, a full additional share for the third child, and further child-related shares thereafter according to the French schedule), durably lowering tax liability for higher-parity families. The recent French TFR trajectory (INSEE Bilan démographique: ~1.61 in 2024, 1.56 in 2025 — a post-WWI historic low) shows that even a well-architected cash lever is not sufficient alone when other production-function nodes erode; the architecture remains visible in France's persistent gap above Finland (~0.26 children per woman in 2025, historically wider).
  • Quebec's Allowance for Newborn Children (1988–1997) is one of the cleanest natural experiments on parity-tiered cash incentives, analyzed in Milligan, "Subsidizing the Stork: New Evidence on Tax Incentives and Fertility" (NBER Working Paper 8845, 2002 / Review of Economics and Statistics, 2005). The program produced sharp short-run birth-rate response — especially at higher parities — that decayed after the program ended.
  • Cumming and Dettling, "Monetary Policy and Birth Rates: The Effect of Mortgage Rate Pass-Through on Fertility" (Bank of England Staff Working Paper No. 835, 2019). The 2008–09 UK policy-rate cut passed through differentially to households on adjustable-rate vs fixed-rate mortgages; each 1 pp drop raised the overall UK birth rate by about 2%, with larger effects for exposed households. Direct natural-experiment evidence that visible monthly margin in the household budget can shift fertility when it lands on the binding constraint — complementing the Argyle/Madrian behavioral microfoundation in §II.
  • González and Trommlerová, "Cash Transfers and Fertility: How the Introduction and Cancellation of a Child Benefit Affected Births" (Journal of Human Resources 58(3), 2023). Spain's 2007 €2,500 universal child benefit produced a ~3% birth-rate rise; the 2010 cancellation announcement produced a 4% transitory pre-deadline spike; the post-cancellation period showed a ~6% sustained drop. Lifetime-program estimate: roughly 69,000 additional births over the window. Effects were strongest among lower-income households. Cleaner natural experiment than Quebec for distinguishing tempo from a bounded quantum effect.

The post-first-birth evidence.

  • The Finnish Institute for Health and Welfare (THL) released the 2026 FinChildren survey reporting that the most common reason parents with a baby in 2024 gave for hesitating about more children was coping as a parent, with financial situation, work–family balance, birth experience, climate change, and global uncertainty also cited. Exhaustion was common, and support for sleep issues and parental wellbeing was often insufficient. THL release. This is the strongest single recent source for the §III claim that the binding inputs are upstream of money, with the scope caveat that it surveys parents already inside the system, not those still deciding about a first birth.
  • Osmo Soininvaara's 2024 essay on the collapse of fertility is a useful framing of the same incentive-inversion problem — children's future benefits socialized through pensions while their costs stayed private to parents.

Counter-thesis to engage.

  • The strongest case against the margin-translation reading is preference collapse — the claim that a meaningful share of would-be parents has stopped wanting children at all, in which case no architecture closes the gap. The §III treatment is that this is best read as a downstream report of the same architecture rather than as a parallel cause, but the case deserves a fuller specific essay. Recent Finnish cohort data on rising "ideal-zero-children" reports among men born in the late 1980s is the cleanest data point on this side of the question.

Related reading: