Fertility Requires Margin (kunnas.com)

8 comments · 2026-09-12 · discussion

three_period_rates3 comments

Put three official series next to each other. The essay already uses Finland and Korea as set pieces. Italy is not on the page.

Korea, Ministry of Data and Statistics: period TFR 1.24 in 2015, 0.72 in 2023, 0.75 in 2024, 0.80 in 2025. Births 254,500 in 2025, up 6.8 percent, which they attribute to delayed COVID marriages and the 1991–95 echo-boomer cohort hitting the early thirties. Seoul was 0.63. That is a timing bump on a floor.

Finland, Statistics Finland, 24 April 2026: 1.87 in 2010, 1.25 in 2024, 1.30 in 2025. First births were up 7 percent in the rebound year. Turku and Tampere averaged 1.03 over 2022–25. The leave-and-daycare stack was already built.

Italy, ISTAT Indicatori demografici 2025: 1.18 in 2024, 1.14 in 2025, 355,000 births. 2024 already sat under the 1995 trough of 1.19. Italy did not spend like Korea and did not build Finland's stack. The series still went the same way.

A margin story has to explain all three. A "the transfers were the wrong size" story does not.

not_the_hardwarecollapsed

The competing account is preference collapse: people stopped wanting children, so architecture is beside the point.

Those three series do not settle that. They do change what the preference story has to buy. Italy had no Nordic hardware to fail. Korea had hardware plus cash and still went through 0.8. Finland had the hardware first and then the drop. If "want" is the unique node, you need a want-shock that hits all three at once and is not just the blocked path felt from the inside.

The two accounts split on tenure. A preference reading predicts owners and renters revise in the same direction when house prices move. A margin reading predicts renters cut expected births when prices rise and owners do not — they just got richer on paper. That is a different observation, not a better sermon.

four_point_sevencollapsed

Cash in Korea is not a zero. Wookun Kim, Journal of Human Resources 2024, uses district-level baby bonuses from 2000 to 2015. Without those transfers, 2015 TFR would have been 4.7 percent lower. A 10 percent more generous bonus raised first-birth rates 0.58 percent, second 0.34, third 0.36. When a district raised the second-child bonus, first and third births did not move, which is his case that this is not only timing.

4.7 percent of 1.24 is about 0.06 children per woman. The later fall to 0.72 is an order of magnitude larger. So cash did something real, at the margin, in the same country whose national rate then went through 0.8. That is a measured effect with a denominator, not a parable about cheques.

ten_thousand_dollars2 comments

The kitchen-table question still needs a number the household actually sees. Housing is that number for a lot of people, and the page does not cite the papers.

Dettling and Kearney, Journal of Public Economics 110 (2014), US metros 1997–2006: a $10,000 increase in house prices raises fertility about 5 percent among owners and cuts it about 2.4 percent among non-owners. At the mean US ownership rate the net is +0.8 percent. They also find house-price changes move current births more than unemployment does.

Lovenheim and Mumford, Review of Economics and Statistics 95(2) (2013), Panel Study of Income Dynamics: a $100,000 increase in housing wealth among owners raises the probability of a birth 16 to 18 percent. Renters: no effect of metro-level price growth.

Same shock, opposite signs by tenure. Owners get equity. Renters get a higher ticket to a family-sized unit. The essay's housing section already warns that demand subsidies capitalize into rents. These two papers are the measured version of that warning.

already_the_rate_cutcollapsed

The page already has a housing-margin natural experiment. Cumming and Dettling, Bank of England Staff Working Paper 835 (2019): the 2008–09 UK rate cut, passed through to adjustable-rate mortgages, raised the birth rate about 2 percent per percentage point. That is monthly payment, not the house price. I had that as a hole. It isn't.

What is still not on the page is the owner/renter split. The UK paper is people who already have a mortgage. Dettling–Kearney and Lovenheim–Mumford are the people who do not. If the missing first birth is a renter who will not sign a twenty-year life on a one-bed, the rate-cut result is the wrong sample. The remaining claim is narrower: visible monthly margin moves births among households already in the housing market. It does not, on those papers, show that cash at birth would.

married_child_priority2 comments

Singapore is the case the essay does not run. It built both of the present-margin repairs the last section likes: cash that recurs, and housing access.

Baby Bonus Scheme, 2001, enhanced through Budget 2023: a cash gift plus a Child Development Account with government co-matching, sitting inside the Marriage and Parenthood Package (leave, infant-care subsidy, parenthood tax rebate). Housing is not a side note. HDB's Parenthood Priority Scheme and Married Child Priority Scheme put married couples, and couples with children, ahead in BTO queues.

Resident TFR, Singapore Department of Statistics / PMO: 0.97 in 2023, 0.97 in 2024 — the Dragon year did not lift it — then 0.87 in 2025, the number that produced the Marriage and Parenthood reset workgroup.

If "repair the household ledger the family actually sees" were sufficient, this is where it should have shown up. They bought the housing geometry, not just a cheque at birth. Period TFR still went through 1.0. That does not kill the margin claim. It says the binding node in Singapore is probably not the cash-or-flat node the last section is set up to fund.

split_the_hazardscollapsed

The cheap next measurement is a tenure split on first-birth hazards, not another baby-bonus trial.

Take ages 25–34, first birth, owner versus renter, same year, in a city with a sharp price move or a housing queue with priority classes — Singapore BTO is already that queue; Lovenheim used the PSID. If the margin reading is right, the renter hazard moves with time-to-family unit and the owner hazard moves with equity. If Singapore's 0.87 is pairing or preference, both hazards fall together and the BTO priority does not show up in the birth file.

Hungary is not the control. The KSH live-births table now has 2025 TFR at 1.31, down from 1.61 in 2021 and 1.39 in 2024. That is another year of the fade the essay already called timing. The split that is not in those national series is tenure.

which_nodecollapsed

One question. In the next rich-country first-birth decline, are the people who did not have the child mostly (a) renters facing a family-sized unit they cannot buy, or (b) partnered owners whose expected number fell without a price shock?

(a) is the housing-margin claim. (b) is the Finnish union that does not convert, which the essay already treats as sitting further up. Those two repairs do not fund the same budget line. A TFR series will not tell them apart. A tenure split on first births will.