# The Price of Another Child
## When fertility policy becomes cheaper than population ageing

# Part I: The finding

A government cannot buy a birth in the same way that it buys a bridge. It can make parenthood less expensive, less risky or less exhausting. It can pay people who were already going to have children. It can move a birth from 2031 to 2028 without changing the eventual number of children. It can also fail entirely.

That uncertainty is the centre of this study.

The model follows one million prospective households for five years, assumes 350,000 births without a new policy and tests seven deliberately stylised interventions.

The result is not that the largest payment produces the cheapest birth. The result is that programmes become efficient only when they reach the households whose desired child is blocked by the particular constraint being removed.

In the average model run, the median cost of one permanently additional birth ranges from about R720,000 for free childcare to R4.18 million for a parent tax exemption. The targeted family package costs about R780,000 per added birth and produces nearly 99,000 additional births per million prospective households. A cash top-up costs about R1.76 million per added birth and produces only about 12,000.

| Policy | Five-year programme cost | Median additional births | Median cost per additional birth | 10th-90th percentile cost |
|---|---:|---:|---:|---:|
| Cash grant top-up | R21.0bn | 11,800 | R1.76m | R0.81m-R3.96m |
| Free childcare | R32.0bn | 44,100 | R0.72m | R0.41m-R1.28m |
| Family housing | R54.0bn | 51,100 | R1.05m | R0.57m-R1.97m |
| Parent tax exemption | R42.0bn | 10,000 | R4.18m | R1.83m-R9.81m |
| Debt forgiveness | R20.0bn | 18,200 | R1.10m | R0.56m-R2.18m |
| Shorter, flexible work | R36.0bn | 44,000 | R0.81m | R0.46m-R1.45m |
| Targeted family bundle | R77.5bn | 98,600 | R0.78m | R0.48m-R1.27m |

The second result is harsher. Discounted to birth, an additional person's lifetime production is worth R1.45 million in a weak labour market, R3.73 million in an average one and R5.83 million in a strong one. Government captures only a share as tax and also pays for education, health, services and old age.

Net lifetime fiscal value is negative R610,000 in the weak labour market, positive R190,000 in the average labour market and positive R960,000 in the strong one. None of the programmes repays itself in the average case. Under the strong outcome, childcare, flexible work and the bundle frequently do.

> Fertility policy becomes cheaper than ageing only when three conditions hold together: the policy causes completed births rather than subsidising or rescheduling them; the future economy employs those children productively; and government starts early enough for the new workers to arrive before the fiscal emergency.

# Part II: South Africa is ageing before it is rich

South Africa is not yet an old country, but the direction has changed. [Statistics South Africa's 2025 population estimates](https://www.statssa.gov.za/?p=18613) put the population at 63.1 million. About 26.2 percent was younger than 15 and 10.5 percent was 60 or older. The total fertility rate fell from 2.78 children per woman in 2008 to 2.21 in 2025.

A fertility rate near replacement does not create an immediate population decline. Population momentum, mortality and migration continue to matter. It does, however, narrow the future base of workers relative to older people if the decline persists.

This creates a distinctive South African problem. The country can age while a large share of the working-age population remains excluded from employment. In the second quarter of 2026, [Stats SA reported](https://www.statssa.gov.za/?p=19804) an official unemployment rate of 33.6 percent and a youth unemployment rate of 47.4 percent. [Women faced 37.5 percent unemployment](https://www.statssa.gov.za/?p=19808), compared with 30.3 percent for men.

The usual ageing story imagines a shrinking workforce that is fully employed. South Africa may instead have too few future workers and too few jobs for the workers it already has. Producing more births without fixing employment can enlarge the population while leaving the fiscal base weak.

Current child support is important poverty policy but is not designed as a purchase price for additional fertility. [National Treasury's 2026 Budget Review](https://www.treasury.gov.za/documents/National%20Budget/2026/review/FullBR.pdf) sets the child support grant at R580 a month in 2026/27. That payment can improve nutrition, schooling and household stability even if it causes no extra births. Judging it only by fertility would miss its actual purpose.

The question in this report is narrower: if government deliberately wanted one more completed birth, which constraints would be cheapest to remove, and when would the future return justify the present cost?

# Part III: A birth has three counterfactuals

Every programme-funded birth belongs to one of three groups.

The first group would have happened without the programme. The payment may still reduce poverty or improve childhood, but the birth is not additional. This is the largest group in almost any broad programme because most recipients are not deciding at the margin.

The second group is brought forward. A couple that planned a child in three years has the child next year because the subsidy is temporary. The programme raises births during the policy window but leaves completed family size unchanged. This can briefly improve age structure, but it does not create a permanently larger generation.

The third group is permanently additional. A household that expected one child ends with two, or a household that expected none ends with one. Only this group can be credited as a lasting demographic response.

The distinction is easy to state and extremely difficult to observe. A birth certificate records what happened, not what would have happened without policy. A rise in births after a grant may reflect the economy, migration, the age structure of women, a recovery after an earlier shock or simple timing.

The model therefore refuses to give any policy a single response. It runs 30,000 alternative worlds. Programme cost varies. The number of induced births varies. The share that is permanent varies. Male employment, female employment, housing pressure, care availability, time pressure and relationship stability move together and separately.

The whiskers in Figure 1 are not confidence intervals from South African programme data. They are decision ranges generated by declared assumptions. Their purpose is to prevent a point estimate from pretending that government knows the behavioural counterfactual.

![Cost per additional birth](assets/cost-per-additional-birth.png)

*Figure 1. The black line marks the R0.96 million net fiscal value of an additional birth in the strong labour-market case. A policy to the left can pass that fiscal test; one to the right cannot. Wide whiskers are the price of behavioural uncertainty.*

The cheapest median estimates belong to childcare, the integrated bundle and flexible work. Their rankings overlap because the uncertainty is large. Housing and debt relief can also be efficient when they are accurately targeted. The parent tax exemption performs worst because it spends heavily on a relatively small group of formal taxpayers and assumes only a modest change in completed fertility.

# Part IV: What the seven policies actually buy

The policies are stylised national experiments, each offered to the same population of one million prospective households over five years. The costs are incremental public or employer costs in constant 2026 rand. They are not estimates of a current South African bill.

| Intervention | Model design | Constraint it tries to remove | Main reason for deadweight |
|---|---|---|---|
| Cash grant top-up | R1,000 a month for five years on all births | Daily cost of a child | Paid to nearly every birth, including those already planned |
| Free childcare | Subsidised care for 65% of births for five years | Care fees and maternal career loss | Some families already use informal or paid care |
| Family housing | 150,000 deeply subsidised family homes | Space, rent and tenure insecurity | Housing is valuable even when it changes no birth decision |
| Parent tax exemption | Five-year relief for 300,000 taxpayers | Formal-sector opportunity cost | Excludes people below the tax threshold and rewards many existing plans |
| Debt forgiveness | R80,000 for 250,000 constrained adults | Student and household balance sheets | Debt may not be the binding fertility constraint |
| Shorter, flexible work | Four-day or flexible schedules for 400,000 workers | Time, commuting and career conflict | Many jobs cannot be reorganised at the assumed cost |
| Targeted family bundle | Housing, childcare, debt and work support for 250,000 | Several constraints at once | High upfront cost and demanding administration |

The bundle is not cheapest because bundling is magic. It performs well because the assumed recipients are carefully selected households with a stated desire for another child and more than one binding constraint. If the targeting system cannot identify them, its cost rapidly approaches that of a universal package.

The cash programme has a different weakness. It is broad and simple, but small relative to the lifetime cost and risk of parenthood. A payment that is valuable for household welfare can still be too small to change completed fertility. Making the payment larger increases the response, but it also increases the windfall paid to every family that would have had a child anyway.

The classic evidence does not imply that cash never works. Kevin Milligan's [study of Quebec's newborn allowance](https://www.nber.org/papers/w8845) found that fertility responded to substantial tax incentives. The more general lesson is that response varies across family type and income. A government importing the headline effect into another country would be assuming away the very uncertainty that matters most.

# Part V: Why ordinary child grants often fail to restore fertility

An ordinary child grant is usually designed to support a child after birth. The fertility decision is made before birth and concerns far more than the next month's cash.

Prospective parents consider whether they can form and maintain a household, find an adequate home, preserve employment, arrange care, survive illness or separation, and carry years of expenses. A small grant can make life materially better without changing that decision.

There are five reasons the measured fertility effect may be weak.

First, the grant is inframarginal. It reaches households whose plans were already settled. Second, it is small relative to housing and lost earnings. Third, it may be uncertain: a temporary or politically vulnerable promise does not insure an eighteen-year obligation. Fourth, it arrives after some decisive costs, including partnership formation, housing deposits and career planning. Fifth, it may relieve poverty without making parenthood compatible with work.

This is not failure in the programme's social purpose. It is a mismatch between the instrument and the demographic objective.

In the model, 350,000 births occur without the new programme. Cash creates a median 12,000 permanently additional births and roughly 20,000 births that are mainly shifted forward. The state pays R21 billion across the entire programme to change the completed-fertility decision of a relatively small group.

![Birth decomposition](assets/birth-decomposition.png)

*Figure 2. Coral represents births that remain additional after the timing effect is removed. Grey represents births brought into the five-year window but not added to completed family size. The dashed line shows the much larger baseline cohort that would arrive without any new policy.*

# Part VI: Housing versus childcare

Housing and childcare solve different stages of the same problem.

Housing matters before conception. A couple may postpone living together, marriage or a second child because the dwelling is too small, tenure is insecure or moving would make commuting impossible. Housing support can therefore unlock household formation. It is also a durable asset or service with benefits that extend beyond fertility.

Childcare matters most after the birth is imagined. It lowers the expected conflict between employment and care. Its value is not only the fee. Reliable care can preserve job tenure, earnings progression, pension contributions and bargaining power inside the household.

South African household data suggests that formal care is far from universal. The [2025 General Household Survey](https://www.statssa.gov.za/publications/P0318/P03182025.pdf) reports that 36.3 percent of children aged zero to four attended an early childhood development facility and 50.2 percent stayed at home. Only 31.4 percent of children lived with both biological parents, while 45.9 percent lived with their mother only. Those figures do not prove a fertility response, but they show why a model based only on a stable two-parent household and formal childcare would miss much of the country.

The same report notes that 42.6 percent of households were female-headed. [Stats SA's 2026 care-work analysis](https://www.statssa.gov.za/?p=19882) finds that female-headed households were much more likely than male-headed households to have no employed member and that unpaid care remains disproportionately performed by women.

The model gives childcare a median cost of R720,000 per permanently additional birth, compared with R1.05 million for family housing. That is not a universal ranking. Childcare wins when employment is available and a parent fears losing it. Housing wins when the household cannot form or cannot obtain adequate space.

The practical answer is diagnostic targeting. Ask whether the blocked household lacks money, space, care, time or confidence in the relationship. Paying the wrong constraint creates a generous programme with a weak birth response.

# Part VII: Male work, female work and the time bargain

Employment affects fertility through income and risk, but not symmetrically.

Male unemployment can reduce household formation and the confidence that a long obligation can be financed. Female unemployment can reduce income too, but employment may also raise the opportunity cost of time away from work. A secure job with inflexible hours can make a child harder to accommodate; no job can make the child unaffordable. The relationship is therefore not a simple line from more employment to more births.

The model treats male employment mainly as household security and female employment as both security and exposure to career-care conflict. Childcare and flexible work respond more strongly to female employment because their value rises when a job can be preserved. Housing and debt support respond more to combined household stability.

Working hours matter even when childcare is nominally affordable. Commutes, school closing times, sick days and unpredictable shifts create coordination costs that money alone may not solve. A four-day week can help if the reduction is genuine and income is preserved. It can hurt if five days of output are compressed into four longer days or if wages fall.

Recent international evidence is suggestive rather than transferable. A 2026 [CEPR study of work from home and fertility](https://cepr.org/publications/dp21250) reports higher realised and planned fertility among people who work remotely, especially where both partners do. The association and research design do not establish that the same response would occur for South African workers, many of whose jobs cannot be done remotely.

Flexible work in this model therefore means more than remote work. It includes predictable shifts, usable parental leave, reduced commuting, four-day schedules where production permits them, and protection from career penalties. Its median cost per added birth is R810,000, but the true cost could sit anywhere from roughly R460,000 to R1.45 million under the simulation.

The employer may also receive part of the return through lower turnover and retained experience. The fiscal comparison in this report does not credit those private benefits to government. A cost-sharing design could therefore be more rational than a fully public subsidy.

# Part VIII: Relationship stability and delayed family formation

Fertility policy often starts too late in the chain of decisions. A child normally requires a relationship or a credible solo-parenting plan, a household, and enough certainty to make an irreversible commitment.

South Africans are formalising relationships later. [Stats SA's 2024 marriages and divorces release](https://www.statssa.gov.za/?PPN=P0307&SCH=73981&page_id=1856) reports median ages of 39 for bridegrooms and 35 for brides in civil marriages. Registered marriages and unions fell 2.6 percent from 2023, and 41.7 percent of divorces in 2024 came from marriages lasting less than ten years.

Marriage records do not capture every cohabiting partnership, and marriage is not a requirement for parenthood. They still reveal that a policy aimed only at costs after birth may ignore delayed union formation and perceived separation risk.

A household considering a second child asks what happens if one partner leaves, becomes unemployed or becomes ill. Free childcare may protect the working parent in that state. Housing tied to the relationship may make separation more dangerous. A tax exemption received mainly by the higher earner may not insure the caregiver at all.

For that reason, the model gives relationship instability a larger negative effect on the bundle, housing and debt response than on a simple cash payment. The intended policy lesson is not that government should subsidise marriage. It is that benefits should remain usable through household change, and that individual entitlements can be more credible than benefits conditional on a couple remaining together.

The age of motherhood is also moving later. [Stats SA recorded](https://www.statssa.gov.za/publications/P0305/P03052024.pdf) an increase in the median age of mothers from 26.6 in 2005 to 28.3 in 2024. Later timing can reflect education, improved choice and better career opportunity. It can also reduce the time available to recover a postponed birth.

# Part IX: Postponed is not the same as forgone

A temporary programme can produce a baby boom and no long-run increase in family size. Households pull births forward to qualify before an expiry date, then have fewer births later.

That timing effect is not worthless. Earlier births change the spacing between generations and may slightly improve a near-term dependency path. They can also ease individual biological constraints. But they should not be valued as if they create extra lifetime taxpayers.

The model's average case assumes that 38 percent of cash-induced births, 67 percent of childcare-induced births and 82 percent of bundle-induced births remain additional after timing is removed. Those shares are assumptions, not findings. The simulation varies them on every run.

| Policy | Worst-case added births | Average-case added births | Best-case added births | Worst / average / best cost per added birth |
|---|---:|---:|---:|---:|
| Cash grant top-up | 3,800 | 14,400 | 25,200 | R6.25m / R1.45m / R0.77m |
| Free childcare | 18,300 | 49,600 | 78,600 | R1.96m / R0.65m / R0.37m |
| Family housing | 22,700 | 58,500 | 91,500 | R2.66m / R0.92m / R0.54m |
| Parent tax exemption | 3,300 | 12,500 | 21,200 | R14.04m / R3.37m / R1.82m |
| Debt forgiveness | 7,100 | 21,300 | 34,600 | R3.15m / R0.94m / R0.53m |
| Shorter, flexible work | 19,000 | 49,700 | 78,100 | R2.12m / R0.72m / R0.42m |
| Targeted family bundle | 43,600 | 108,200 | 167,600 | R1.99m / R0.72m / R0.43m |

The fixed scenarios differ from the Monte Carlo medians because the simulation combines many skewed uncertainties. The average-case table is a clean reference scenario. Figure 1 reports the distribution of simulated outcomes.

The table shows why evaluation must continue long after the first birth spike. A programme should track completed family size, births by age and parity, and a credible comparison group. It should also record whether the same families would have qualified without changing their plans.

# Part X: The lifetime account

An additional child creates costs immediately and tax revenue much later. Discounting makes that timing decisive.

The model follows the person from birth to age 100. It includes stylised public spending on childhood health, education, working-age services, old-age support and healthcare. It gives the person age-specific survival, employment, production, real earnings growth and an effective tax share. Every future amount is discounted to the date of birth at 3 percent a year.

The weak labour market employs 42 percent of working-age capacity and begins with output of R255,000 per employed worker. The average case uses 70 percent and R345,000. The strong case uses 86 percent and R410,000. Real productivity growth ranges from 0.7 to 1.9 percent. The effective share of production collected as tax ranges from 24.5 to 33 percent.

These scenarios are not forecasts of an individual salary. They compress formal work, informal work, unemployment, wages, consumption taxes and payroll taxes into a lifetime average. Their purpose is to show that fertility and employment policy are complements.

![Lifetime worker value](assets/lifetime-worker-value.png)

*Figure 3. Public cost is almost R0.97 million at birth in every scenario. Taxes depend on employment and productivity. Production is much larger than tax revenue but belongs to workers, firms and consumers as well as government.*

| Lifetime outcome at birth | Weak labour market | Average labour market | Strong labour market |
|---|---:|---:|---:|
| Discounted taxes | R0.36m | R1.16m | R1.92m |
| Discounted public cost | R0.97m | R0.97m | R0.97m |
| Net fiscal value | -R0.61m | +R0.19m | +R0.96m |
| Discounted economic production | R1.45m | R3.73m | R5.83m |

The public cost is held constant to isolate the labour-market effect. In reality, better employment can reduce household grant dependence and improve health, while high productivity may raise expectations for public services. Those feedbacks could move the result in either direction.

The model also avoids a common accounting trick. It does not call all production a government return. If an additional worker produces R5.83 million in discounted output, most of that value becomes wages, profits and consumer benefit. Only R1.92 million arrives as modelled tax.

## What the lifetime account cannot decide

A positive fiscal value does not prove that a larger population is socially optimal. More people can require housing, water, electricity, transport and capital before they produce at the assumed level. If investment fails to keep pace, output per person can fall even while total output rises. Environmental limits and congestion are also absent from the account.

A negative fiscal value does not prove that the person is a loss. Children, family life and the continuation of communities have value that is not a tax receipt. Unpaid care and household production are economic activity even when national accounts and tax systems barely see them. An extra child can also alter parents' employment, saving, health and transfers to other relatives.

The model values a marginal person while holding institutions constant. A much larger generation could change wages, housing prices, politics, innovation and the capital stock. A much smaller generation could induce automation, immigration or a reorganisation of care. These general-equilibrium responses belong in a national population model, not in a simple programme test.

The narrow account remains useful because it imposes discipline. A minister who claims that a birth incentive will pay for pensions is making a fiscal claim. That claim should be tested against net tax revenue, timing and public cost. Broader social reasons can justify a policy, but they should be stated as broader social reasons rather than hidden inside an inflated estimate of future tax.

# Part XI: The break-even map

The fiscal break-even point depends on two uncertain quantities: how many births are permanently additional and what those future people contribute after public costs.

At an average net fiscal value of R190,000, a R20 billion programme needs about 105,000 permanently additional births merely to repay its public cost. None of the average-case policies reaches its required threshold. At the strong value of R960,000, the same programme needs about 21,000 additional births. Several policies become plausible.

![Break-even map](assets/break-even-map.png)

*Figure 4. Points above a break-even line repay their modelled public cost. Under the strong labour-market line, free childcare, flexible work and the bundle clear the threshold in the clean average scenario. Under the average line, none does.*

The uncertainty simulation gives the same lesson. Against the strong worker value, childcare breaks even in 74 percent of runs, the bundle in 71 percent, flexible work in 64 percent, housing in 43 percent, debt forgiveness in 40 percent, cash in 16 percent and the tax exemption in 1 percent. Against the average value, the probability is effectively zero.

This does not mean that childcare has no case in the average economy. The fiscal test ignores several benefits: children's development, maternal earnings, lower poverty, household wellbeing, gender equality and the option value of allowing people to realise desired families. Some of those benefits can also raise tax revenue but are not modelled here.

An ageing argument cannot justify any price. A programme that claims a narrow budget return must state the required number of permanently additional births and measure whether they appeared. Better targeting, a stronger permanent response or a more productive labour market can pass the test; calling a transfer an investment cannot.

# Part XII: Can fertility decline become effectively irreversible?

No fertility rate is mechanically irreversible. Preferences, institutions, migration and economic conditions can change. The danger is economic and social lock-in.

After a long period of low fertility, the population contains fewer prospective parents. Schools, housing, workplaces and neighbourhoods adapt to fewer children. Partnership norms and expectations change. Employers design careers around uninterrupted work. People learn from peers that late or no parenthood is normal. A temporary postponement can become permanent when biological and relationship windows close.

The model illustrates this lock-in by starting the same cash and bundle policies in 2026, 2040 and 2060. It gradually shrinks both the prospective-parent pool and its assumed response. It also reduces the share of induced births that remain permanent.

![Delay and irreversibility](assets/delay-and-irreversibility.png)

*Figure 5. This is an illustrative adaptation path, not a demographic forecast. By 2060 the same bundle produces about 39,000 additional births rather than 108,000, while its real cost per added birth rises from R0.72 million to R2.40 million.*

The cash programme's cost per added birth rises from R1.45 million in 2026 to R5.02 million in 2060. The bundle rises from R0.72 million to R2.40 million. At that point, even the strong-labour fiscal value of R0.96 million is too small.

This is the practical meaning of effective irreversibility: not that births cannot recover, but that the subsidy required to restore them becomes larger than the fiscal value government can plausibly recoup. A country may still act for social or cultural reasons, but it can no longer claim an easy budget return.

Starting early has another advantage. It permits experimentation. Government can compare regions or eligible cohorts, observe completed fertility, stop ineffective instruments and expand those that remove genuine constraints. A late emergency encourages large universal promises made before the response is understood.

# Part XIII: The policy design that survives the model

The model favours a targeted family option rather than a universal baby bonus.

Eligibility would begin with stated fertility intention, age, household formation and identifiable constraints. The package would be modular. A household blocked by care receives care. A household blocked by space receives a housing pathway. A household blocked by rigid hours receives a work arrangement. Benefits would follow the caregiver or child through relationship change and would not depend on remaining with a partner.

The programme should not promise a free house to every birth. That would create enormous deadweight and invite strategic timing. Housing support can instead take the form of priority access, rental guarantees, additional space, serviced land, long leases or a recoverable equity stake. Debt relief can be conditional, capped and spread over time rather than written off at conception.

Childcare should be reliable, geographically accessible and compatible with actual shifts. A nominal place that closes before a parent's workday does not remove the constraint. A childcare programme also needs quality safeguards. Earlier evidence from Quebec shows why utilisation, maternal employment and child outcomes must be evaluated separately; the [Baker, Gruber and Milligan study](https://www.nber.org/papers/w11832) found large changes in childcare use and maternal labour supply alongside concerns about some child and parent outcomes.

Flexible work should be occupation-specific. Remote work cannot help a nurse, miner, driver or retail worker in the same way it helps an office worker. Predictable schedules, shift exchanges, reduced commuting, part-time career protection and compressed weeks may be more relevant than home working.

The evaluation contract should be written before launch.

1. Publish the five-year cost, not only the monthly benefit.

2. Estimate baseline births for a credible comparison group.

3. Separate births brought forward from completed-family additions.

4. Report response by age, parity, income, employment, sex of household head and province.

5. Measure maternal and paternal employment, earnings and pension contributions.

6. Follow children for health, care quality and education outcomes.

7. Re-estimate cost per additional birth every year and stop automatic expansion when the denominator is unknown.

8. Pair the programme with employment and productivity policy, because the future worker creates no fiscal dividend without work.

9. Keep a near-term ageing plan. The first policy-induced workers enter the labour force only around the mid-2040s.

The preferred pilot in this model is not the cheapest grant. It is a constrained package aimed at households that already want a child but face a measurable housing, care or time barrier. Its budget should be released in stages, with randomised or quasi-experimental evaluation where ethical and administratively feasible.

# Part XIV: The answer

Subsidising fertility becomes cheaper than financing population ageing only under a demanding joint test.

The policy must produce enough permanently additional births that its cost per changed birth falls below the discounted value of the future worker. The future economy must employ that worker productively enough for lifetime taxes to exceed lifetime public costs. The birth must arrive early enough to contribute before the ageing deficit becomes acute. And the programme must deliver benefits beyond fertility well enough that failure to change births is not pure waste.

Under this study's average labour market, the additional person has a net fiscal value of only R190,000 at birth. None of the modelled fertility programmes is cheaper than that. Under the strong labour market, the value rises to R960,000. Free childcare, flexible work and a targeted bundle then become plausible; cash top-ups and parent tax exemptions remain weak because too much spending goes to plans that were already going to happen.

The deepest result is that fertility policy is really institutional policy. People do not merely buy children from disposable income. They decide whether a child fits inside housing, work, care, partnership and time. A small grant can protect a child without changing fertility. A carefully targeted service can change fertility without looking like a birth payment.

> The fiscally rational pro-natal policy is not the policy that pays most. It is the policy that finds a genuinely blocked desired birth, removes the binding constraint, preserves the parent's productive life and gives the future child an economy in which work exists.

Even then, the new generation is not a substitute for present reform. Fertility support can improve the workforce of the 2040s and 2050s. It cannot pay next year's pensions.

# Notes: Scope, assumptions and sources

This is an armchair scenario model, not a demographic forecast, programme evaluation, benefit-cost appraisal or official costing. All money is constant 2026 rand. The model follows one million prospective households for five years and assumes 350,000 baseline births without a new policy.

The seven programme budgets are constructed assumptions, listed in Part I. Programme costs vary by 12 percent around them in the simulation.

Every policy has an assumed central number of gross induced births and an assumed share that remains additional after timing is removed. Gross response uncertainty is lognormal, with policy-specific dispersion. Permanence is drawn from a beta distribution. Six shared conditions also vary: female employment, male employment, housing, care, working-time compatibility and relationship stability. Policy-specific weights determine sensitivity to each condition. The model runs 30,000 draws using a fixed seed for reproducibility.

The fixed worst case reduces response to 55 percent of the central case, lowers permanence by 22 points and raises cost by 12 percent. The best case raises response to 138 percent, permanence by 10 points and lowers cost by 8 percent. These are stress tests, not confidence intervals.

The lifetime account follows ages zero to 100. Survival is a stylised schedule. Public cost is R28,000 annually at ages zero to four, R42,000 at ages five to seventeen, R34,000 at ages eighteen to twenty-two, R22,000 at ages twenty-three to sixty-four, R67,000 at ages sixty-five to eighty-four and R82,000 thereafter, all before survival and discounting. The model applies a 3 percent real discount rate.

The weak, average and strong labour scenarios vary employment, initial output, productivity growth and the effective tax share. Public costs are deliberately held constant. Production includes output during partial employment at ages eighteen to twenty-two and limited work after sixty-five. It excludes unpaid care, household production, environmental costs, capital deepening, congestion, migration, intergenerational transfers and the effect of population size on institutions or innovation.

The fiscal break-even probability compares each simulated programme cost per permanently additional birth with the deterministic net fiscal value in the average or strong labour scenario. It does not include the welfare value of children, parental preferences, distribution, maternal career gains, lower child poverty or strategic national value. Production is reported separately and is not treated as public revenue.

The delayed-start illustration reduces the modelled pool of prospective parents by 1 percent and the institutional fertility response by 1.4 percent for every year after 2026, compounds a small real delivery-cost increase, and reduces permanence by 0.45 percentage points per year subject to a floor. These values are intentionally transparent assumptions used to show lock-in, not estimates of South Africa's future fertility.

Primary South African context comes from [Stats SA's Mid-year Population Estimates 2025](https://www.statssa.gov.za/?p=18613), [QLFS Q2 2026](https://www.statssa.gov.za/?p=19804), its [gender labour-market analysis](https://www.statssa.gov.za/?p=19808), [General Household Survey 2025](https://www.statssa.gov.za/publications/P0318/P03182025.pdf), [Recorded Live Births 2024](https://www.statssa.gov.za/publications/P0305/P03052024.pdf), [Marriages and Divorces 2024](https://www.statssa.gov.za/?PPN=P0307&SCH=73981&page_id=1856), [Gender Disparities in Care Work 2026](https://www.statssa.gov.za/?p=19882) and the [National Treasury 2026 Budget Review](https://www.treasury.gov.za/documents/National%20Budget/2026/review/FullBR.pdf).

The international evidence is used only as mechanism context, not as a South African calibration. It includes Milligan's [Quebec tax-incentive study](https://www.nber.org/papers/w8845), Baker, Gruber and Milligan's [universal childcare study](https://www.nber.org/papers/w11832), and Aksoy and co-authors' [2026 work-from-home and fertility study](https://cepr.org/publications/dp21250).