Marius ComperDemography · economics · gender Română
Claudia Goldin · NBER Working Paper 35425 · July 2026

Before the baby, a promise with no guarantee

Money, housing and care services shape the decision to have children. Claudia Goldin follows a constraint that is harder to reimburse: before a birth, a woman must decide whether her partner and public institutions will credibly absorb part of the care and career risk. Fertility can fall when women’s autonomy advances faster than the sharing of family work, even when the desire for children remains.

This is an NBER working paper and has not been peer reviewed. Its model describes heterosexual couples and isolates one mechanism; it is not a complete account of fertility.

1.34total fertility rate in the EU, 2024
1.39Romania, 2024
1.39England and Wales, provisional 2025 estimate
28%childless among US college graduate women born in 1955
The thesis

Fertility can fall because risk is allocated before it becomes observable

In Goldin’s model, some women can invest in education and earn a higher wage. After a birth, that professional premium may disappear if the father does not provide enough care. His type is learned only after the child exists, while the decision to have the child must be made beforehand.

That sequence turns fertility into a credible commitment problem. A partner may promise to share nights, sick days and schedule sacrifices. The state may promise childcare, leave and workplace protection. The woman has to assess those promises before they are tested.

The more professional investment is at stake, the more reliable future support must appear.
The central instrument

Where does the decision land in Goldin’s model?

Move the model’s three variables. The chart shows the frontier between a configuration in which college-eligible women have children and one in which they avoid the risk. It is an aggregate schematic, not a test for individuals or couples.

The difference between earnings in a career that uses education and the earnings available if childcare prevents that career.
The utility the model assigns to having a child. It is not a cash amount and cannot be directly measured.
The aggregate probability that a father provides enough care for the mother to retain her career. It is not an assessment of a particular partner.
Frontierp > 1 − K / Δ
0%10%20%30%40%50%60%70%80%90%100%406080100120140160 career and child career without child share of dependable fathers, p potential career penalty, Δ
selected configurationcalculated frontier
The calculated threshold is 65%. The selected value of p is 60%, so the point lies in the “career without child” region.

The model assumes one birth, random matching, two father types and a penalty that affects only educated women. Its simplicity makes the mechanism visible, while limiting how literally it can be read.

The US history

Autonomy moved faster than men’s expectations

The contraceptive pill allowed young women to delay marriage, finish college and enter professions requiring long investments. Goldin shows that change arrived in stages: fertility control first changed the timing of marriage, then education, then the credibility of women’s careers in employers’ eyes.

Men’s views began to change in the 1980s. The demographic outcome moved as well: childlessness among college graduate women fell from 28% in the 1955 cohort to 17.5% in the 1974 cohort. Goldin offers several possible explanations, including better timing, assisted reproduction and more cooperative partners. The data do not isolate each contribution.

A preschool child is likely to suffer if his or her mother works
college graduate women
60%
college graduate men
66%

Over 13 years, agreement among women fell by 28 percentage points. Among men, it fell by 3 points.

Four countries, four days

The unpaid-work gap sits beside the fertility gap

Goldin’s figure for 22 countries places higher fertility where the difference in household and care time is smaller. The text reports exact figures for four countries. The bars below show how many additional hours per day women do relative to men.

The relationship is descriptive. Time-use surveys come from different years, and countries differ in income, housing, labour markets, norms and family policy. The correlation supports the mechanism but does not identify it on its own.

Policy X-ray

A policy can pay the cost while leaving the risk untouched

Choose an intervention. The map shows the channel through which it may act in Goldin’s theory and adjacent research. It is a conceptual translation, without estimates of effect size.

p · credibility of support no direct channel
Δ · career penalty no direct channel
λ · autonomy and access to investment no direct channel
C · direct costs direct channel

It reduces the immediate cost of a child. In Goldin’s model, a bonus does not automatically make a partner’s promise more credible or shrink the career penalty. The OECD finds varied results across reforms, often small for isolated cash transfers.

How far the evidence reaches

The paper combines causal results, historical description, correlation and a stylised model

The strength of the conclusion varies across its components. Keeping them separate preserves the useful idea without turning it into a single-cause theory.

stronger

Contraceptive access changed women’s education, marriage and work

Goldin builds on earlier studies using legal variation across US states. They link early access to the pill with later marriage, longer education and greater labour-force participation.

rich description

US cohorts show a fall and then a partial recovery

CPS data track completed fertility and childlessness by education. They document the timing and groups in which change was concentrated, but do not identify the cause by themselves.

association

Countries with more equal household work tend to have higher fertility

The 22-country chart is consistent with the mechanism. It does not isolate every difference between societies and uses time surveys from different years.

model

The commitment frontier makes the problem legible

The model shows why a more valuable career can require a more credible promise. Binary assumptions and random matching make it too simple for individual prediction.

separate hypothesis

The iPhone may explain part of the post-2007 decline

Goldin cites another working paper exploiting AT&T’s initial monopoly. Its result concerns mostly unintended births in 2007-2011 and does not show that technology explains the decline in completed fertility.

Outside the frame

Several fertility constraints can reinforce one another

Housing, job insecurity, education costs, intensive parenting, partnership formation, reproductive health and migration can change both desired family size and the timing of births. Recent research treats compatibility between career and family as an important core, not the only mechanism.

· · · ·

Several fertility constraints can reinforce one another
housing and costsjob securitycare and schedulesnorms and partnershipsintensive parentingunintended fertility
The test for any pronatalist policy

Can a woman believe, before pregnancy, that motherhood will not cancel the investment she has made in her own life?

A country can pay for a birth while leaving one parent to absorb much of the lost sleep, interruption and career risk. Goldin’s theory directs attention to the quality of the promise: who shares care, how predictable support is, and what happens to a career after the first child.

Sources and status

The central document and the research around it

  1. Claudia Goldin, “The Rise of Female Autonomy and the Decline of Fertility: The Role of Mismatch”, NBER Working Paper 35425, July 2026. Argument and model: pp. 1-8; US history: pp. 9-16; international comparison: pp. 17-20; conclusion: pp. 21-23; Figures 3, 7, 9 and 12.
  2. Eurostat, “Fertility statistics”, data extracted 19 February 2026.
  3. Office for National Statistics, “Births in England and Wales: 2025”, 27 May 2026.
  4. Matthias Doepke and Fabian Kindermann, “Bargaining over Babies”, American Economic Review, 2019.
  5. Henrik Kleven, Camille Landais and Gabriel Leite-Mariante, “The Child Penalty Atlas”, NBER Working Paper 31649.
  6. Giulia Briselli and Libertad González, “Are Men’s Attitudes Holding Back Fertility and Women’s Careers?”, BSE Working Paper 1506, 2025.
  7. OECD, “Fertility trends across the OECD: Underlying drivers and the role for policy”, Society at a Glance 2024.
  8. Caitlin K. Myers and Ezekiel Hooper, “Is the iPhone Birth Control?”, NBER Working Paper 35310, June 2026.

Goldin 2026 is an NBER working paper circulated for discussion and comment, without peer review. Local figures come from official statistics published in 2026. The policy map is an editorial synthesis of mechanisms, not a quantitative meta-analysis.