The Cheapest Pro-Natalist Policy Nobody Wants, Least of All Nigel Farage

Alarmists & Technocrats agree on two things: birth rates are too low, & workers (who've had it far too good for too long) shouldn't allowed to work from home. You've probably spotted the contradiction

Nigel Farage on a delightful TV show called “Spitting Image”

Elon Musk has called declining birth rates a threat of “mass extinction.” JD Vance wants “more babies in America.” Nigel Farage, at a Birmingham rally, warned of demographic decline. All three are also among the most aggressive opponents of working from home. Musk banned remote work at Tesla, SpaceX, and X, called it “morally wrong,” and spearheaded the push to force federal employees back to the office five days a week. Farage dismissed remote work as “a load of nonsense” and demanded “an attitudinal change to hard work rather than work-life balance.”

This isn't a partisan issue. Both Biden and Trump administrations, Canada's Liberal government, Labour governments in the UK and New South Wales, centre-right and centre-left alike across Europe: all are dismantling flexible work. The one thing the entire political spectrum agrees on is that someone working from home is getting away with something, results be damned!

A new paper (on a mountain of papers which Lyman Stone own paper is the bedrock of) suggests they are actively undermining their own stated goals.

“Work from Home and Fertility,” by Aksoy, Barrero, Bloom, Cranney, Davis, Dolls, and Zarate, draws on original survey data from 38 countries and U.S. Census data to investigate the relationship between remote work and having children. The team spans Stanford, Princeton, the European Bank for Reconstruction and Development, Mexico’s ITAM, and Germany’s Ifo Institute. The central finding is striking: working from home is associated with substantially higher fertility, the relationship is robust across multiple datasets and identification strategies, and the implied contribution to national birth rates is large enough to matter. Interpreted causally, their estimates imply that WFH accounts for roughly 291,000 U.S. births per year, about 8.1% of all American births as of 2024. By comparison, government spending on early childhood care and education appears to deliver less, at far greater cost. We’ll return to that comparison with full numbers later.

This paper can’t definitively rule out every alternative explanation for the patterns it documents. But it is a pretty comprehensive examination of the WFH-fertility relationship to date, the broadest in geographic scope and the most careful in its use of multiple empirical strategies.

What the paper finds

The study draws on two surveys of the authors’ own design: the Global Survey of Working Arrangements (G-SWA), covering 38 countries with data from late 2024 and early 2025, and the U.S. Survey of Working Arrangements and Attitudes (SWAA), a monthly survey with observations from December 2022 to December 2025. Both contain questions on working arrangements, demographics, and a detailed module on realized and planned fertility. The analysis focuses on respondents aged 20 to 45.

The pattern is consistent in both datasets. People who work from home at least one day per week have higher realized fertility (children actually born since 2023), higher planned future fertility, and higher total lifetime fertility (children to date plus plans for more). For couples, the relationship strengthens when both partners work from home. In the 38-country G-SWA data, estimated lifetime fertility is 0.32 children per woman higher when both partners WFH at least one day a week than when neither does, a 14% difference relative to mean total fertility. In the U.S. SWAA data, the gap is even larger: 0.45 children per woman, or 18% of the mean.

These patterns hold after controlling for age, education, marital status, number of children before 2023, own and partner’s employment status, and country or state fixed effects. They hold separately for men and women. They hold in the raw data and in conditional regressions. They hold across 38 countries on four continents.

The extensive margin matters most

One of the paper’s most policy-relevant findings concerns the shape of the WFH-fertility relationship. When the authors replace their binary WFH indicator with a continuous measure (how many days per week?), the continuous measure does show a positive relationship with fertility, statistically significant in the U.S. data, marginally significant in the 38-country sample. More WFH days, more fertility. So far, so intuitive.

But the picture changes when you let the two measures compete. In specifications that pit the binary indicator against the continuous variable, the binary indicator wins decisively: it remains positive and statistically significant while the continuous variable goes small, insignificant, or sometimes wrong-signed. The authors' conclusion is direct: "the data suggest that some ability to WFH over the week is the key feature of working arrangements that relates to fertility. Given the ability to WFH at least one day a week, we find no evidence that fertility rises with each extra WFH day." And when both partners have that flexibility, the effect compounds: estimated lifetime fertility is 0.32 children per woman higher when both partners WFH at least one day a week than when neither does.

The standalone positive result for WFH days is largely picking up the extensive margin, the jump from zero days to one. That jump is where the action is. Once you let the binary indicator absorb it, the additional contribution of a second, third, or fourth day fades. This doesn’t mean extra WFH days are harmful to fertility, or that they confer no other benefits. It means the bulk of fertility benefit concentrates heavily at the threshold: the difference between some flexibility and none.

For anyone designing workplace policy, or anyone currently dismantling existing flexibility, this is critical. You don’t need to go fully remote to get the fertility benefit. You just need to break the five-day-in-office mandate. One day of flexibility changes the calculus. Conversely, the move from a hybrid schedule to a rigid five-day mandate, which is precisely what many governments and corporations are now trying to impose, eliminates the one feature of working arrangements most clearly linked to higher fertility.

How the evidence is built

What makes this paper more than another “interesting correlation” story is the care with which the authors triangulate their result. The argument is best understood as three concentric rings of evidence, each addressing the causal question with increasing force.

The first ring is the individual-level association between WFH status and fertility outcomes, documented in the G-SWA and SWAA data. The positive relationship is strong and consistent. The obvious objection is selection: perhaps people who want children (or already have them) sort into jobs that offer remote work. The authors don’t deny this possibility at this stage. They establish the pattern and move on.

The second ring is more ambitious. Turning to the U.S. Current Population Survey from 2023 to 2025, the authors examine how individual fertility responds not to a person’s own WFH status, but to the occupation-level prevalence of WFH jobs. To measure this, they use the Hansen et al. (2026) index, derived from a large language model applied to roughly half a billion online job postings to flag those that advertise hybrid or fully remote work. The model achieves 99% accuracy by human audit, greatly outperforming dictionary methods.

The results are clear. One-year fertility rates rise with the share of WFH jobs in one’s own occupation. For partnered respondents, fertility also rises with WFH opportunities in the partner’s occupation. For a sense of magnitudes: raising both own and partner’s occupational WFH shares by one standard deviation (about seven percentage points) is associated with a 13.8% increase in one-year fertility for women and a 10% increase for men, relative to the mean. These are large effects, more than half the cross-sectional standard deviation of one-year fertility for partnered women.

Why does this matter for causal interpretation? Because a person’s occupation-level WFH prevalence is far less under individual control than their personal WFH status. You can choose to work from home (if your employer allows it); you have much less say over the WFH norms of your entire occupation. Selection becomes a weaker explanation.

The third ring pushes the identification strategy further still. Since WFH opportunities rise steeply with educational attainment in the post-pandemic era, and since people aged 30 to 45 chose their education levels many years before the pandemic made WFH widely available, educational attainment provides a proxy for WFH opportunities that is effectively immune to reverse causality concerns. Nobody earned a college degree in 2010 in anticipation of hybrid work policies that would arrive a decade later.

The results confirm the pattern. In the CPS data, college-educated respondents aged 30 to 45 have substantially higher one-year fertility in 2023–2025. Relative to respondents who did not attend college, women with a Bachelor’s degree have a one-year fertility rate 29% higher (in the sample of partnered women). If her partner also holds a college degree, fertility rises by a further 24%. The magnitudes are similar for men.

The authors are forthright about what this ring of evidence can and cannot establish. They cannot rule out the possibility that other characteristics of college-educated couples (income, relationship stability, cultural attitudes) independently raise fertility in this period. But the consistency across all three approaches is telling. Individual WFH status, occupation-level WFH opportunities, and education as a WFH proxy all point in the same direction. Either WFH raises fertility, or there exists some powerful confound that simultaneously correlates with all three measures and operates independently of the extensive controls the authors include. Not impossible, but increasingly implausible.

The time-series logic

A methodological choice worth noting: the authors deliberately exclude fertility data from 2020 to 2022. Two reasons, both sensible. First, the pandemic itself (health shocks, contagion fears, social distancing, economic uncertainty) had potentially large and confounding effects on fertility that cannot be easily separated from WFH effects. Second, fertility decisions are forward-looking. People who contemplated having children in 2020 or 2021 could not have been confident that the surge in remote work was permanent. By 2023, it had become apparent that WFH rates would not revert to pre-pandemic levels. U.S. monthly data show WFH rates stabilizing by early 2023 and fluctuating in a narrow band since then. Data from 2023 onward are therefore more informative about the longer-term relationship between WFH and fertility, and that is the period the authors focus on.

The cross-country picture

The paper’s 38-country scope yields some of its most vivid material. WFH rates among workers aged 20 to 45 vary enormously: from 21% in Japan to 60% in Vietnam. (The high rates in some lower-income countries, including Vietnam, Egypt, the Philippines, and Thailand, partly reflect farming, craft work, and small-scale production that takes place at or near home, rather than laptop-based knowledge work.) Among high-income countries, the spread is wide. The United States (39%), the United Kingdom (54%), and Canada (53%) stand well above Japan (21%), South Korea (24%), and Italy (24%).

This variation matters because the paper translates it into fertility consequences. Using a straightforward accounting framework, the authors calculate the contribution of WFH to each country’s total fertility rate (TFR) in the post-pandemic era. The results range from a WFH contribution of 0.038 children per woman in Japan and 0.037 in Italy to 0.101 in Canada and 0.131 in the United States. As a share of TFR, that’s 3.1% in Japan and Italy, 7.6% in Canada, and 8.1% in the United States.

The more provocative exercise asks: what if countries with low WFH rates raised them to the average of the U.S., U.K., and Canada, about 45% of working women? The implied fertility gains are modest but nontrivial. Japan would gain an estimated 0.057 children per woman, translating to roughly 31,800 extra births per year. South Korea would gain 0.033 children per woman, or about 10,500 extra births per year. France and Italy would each gain about 0.042 children per woman, some 17,000 extra births per year in France and 12,800 in Italy.

These numbers won’t reverse decades of fertility decline, but it would reverse a few years. Not to mention, these fertility gains come at essentially zero cost to the public purse.

The missing intercept

There is a subtlety in the paper’s accounting that sophisticated readers should note. All of the estimates rest on cross-sectional variation: they capture whether your WFH status or opportunities predict your fertility relative to others. They do not capture what the authors call a “missing intercept,” the possibility that living in a society where WFH is broadly available raises fertility for everyone, including people who don’t currently work from home. The logic is intuitive. If a woman knows that WFH jobs are widely available in her economy, and will likely remain so, the prospect of combining employment with motherhood looks less daunting even if her current job is fully on-site, because she could switch to a more flexible role later. Economic reasoning suggests this aggregate effect has a positive sign. If so, the paper’s estimates are lower bounds on the true contribution of WFH to fertility. The authors flag this honestly; we flag it because it bears on the policy implications.

Mechanisms: what’s driving this?

The authors identify three stories that could underpin the positive relationship, and they resist the temptation to declare a winner.

The first is straightforward causation: WFH jobs make it easier to combine child-rearing with paid employment (managing school pickups, being present when a child is sick, avoiding long commutes) and this directly leads women and their partners to choose higher fertility.

The second is pure selection: families with children (or plans for children) sort into WFH jobs for the flexibility they offer, but the availability of WFH doesn’t actually change fertility decisions. Under this story, WFH and fertility are correlated but causally unrelated.

The third, and most interesting, is selection operating as a causal force: the availability of WFH jobs raises fertility by expanding current and future opportunities to select into parent-friendly work. Under this story, selection is itself the mechanism through which WFH affects fertility. People have more children because they know flexible jobs exist, even before they take one.

All three stories are consistent with the underlying idea that WFH makes it easier for parents to combine work and family. The occupation-level analysis (ring two) and the education-as-proxy analysis (ring three) are designed to put pressure on the pure selection story, and they do. In the CPS sample, respondents aged 30 to 45 are restricted precisely because occupational mobility at these ages is low: nearly nine in ten don’t switch occupations from one year to the next. It’s hard to argue that a 35-year-old woman chose her occupation in response to a fertility event when she’s been in that occupation for years. The pure selection story isn’t impossible, but it looks increasingly strained.

The comparison that reframes the debate

Olivetti and Petrongolo (2017), using annual data for 22 high-income countries from 1970 to 2010, find that additional government spending on early childhood care and education equal to one percent of GDP is associated with 0.2 extra children per woman. The United States spends about 0.4% of GDP on such programs. Interpreted causally, that implies a fertility contribution of roughly 0.08 children per woman. The paper’s estimate of WFH’s contribution to U.S. fertility is 0.131 children per woman, more than 60% larger.

The comparison is not entirely clean. The Olivetti and Petrongolo estimates come from panel data with country and year fixed effects; the Aksoy et al. estimates come from cross-sectional variation. Different methods, different data, different periods. We should be cautious about drawing too precise a conclusion. But the broad implication is hard to ignore: the fertility benefit of a workplace practice that costs the government nothing appears to be in the same ballpark as, and may exceed, the fertility benefit of programs that absorb meaningful shares of GDP.

What we don’t know

First, the three-ring evidence structure is persuasive in aggregate, but each ring has vulnerabilities. The individual-level analysis is subject to selection bias. The occupation-level analysis could reflect unobserved occupation-level characteristics that correlate with both WFH prevalence and fertility norms. The education analysis can’t rule out other features of college-educated couples. The authors know all of this and say so explicitly. What they have is a strong circumstantial case, not a randomized experiment.

Second, the planned fertility measures rely on stated intentions. People say they plan to have two more children; whether they actually will is another matter. The authors partially address this by examining realized fertility (children actually born since 2023) alongside planned fertility, and the patterns hold for both. But the total fertility measure, which drives many of the paper’s headline numbers, is a composite that depends on intentions.

Third, the cross-country estimates rest on a common treatment effect (β) for all countries except the United States, because the G-SWA isn’t large enough to support reliable country-specific estimates. The actual fertility response to WFH could differ substantially across cultures. In East Asian societies, where status-driven competition in children’s education is a major driver of low fertility, WFH may help less than in contexts where the binding constraint is primarily logistical.

Fourth, the paper says nothing about WFH’s effects on productivity, collaboration, innovation, or any of the other outcomes that (not so) legitimately concern employers and policymakers. Other academic research on productivity points that remote work has no effect or boosts productivity, and most employers on investor calls will admit that RTOs is often used to push people to quit as a soft layoff, rather than productivity.

The policy contradiction

We come back to where we started. Governments of every political complexion are simultaneously lamenting falling birth rates and tightening return-to-office mandates. But the paper's extensive-margin finding makes these incremental rollbacks more damaging than they appear. A mandate that goes from two required office days to three, then three to four, doesn't just reduce flexibility. It tells every worker that the last remaining WFH day is next. And people making fertility decisions don't plan around what they have today. Each step looks like it only removes one day, but what it actually removes is the confidence that any flexibility will survive, which is where the fertility effect lives.

In the United States, RTO started under the Biden Administration, with significant support from city mayors and center governors, but it was reinforced by the Trump administration’s executive order requiring full return-to-office for federal employees was among its earliest acts. Musk and Ramaswamy, in a Wall Street Journal op-ed, explicitly welcomed the “wave of voluntary terminations” they expected it to produce, framing it not as a side effect but as a feature: a tool for shrinking the federal workforce by making conditions unpleasant enough that people quit.

In the United Kingdom, the Labour government maintained the previous Conservative administration’s 60% office attendance mandate for civil servants, set out in the Civil Service People Plan 2024–2027. Staff at the Land Registry and the Office for National Statistics voted for strike action over compulsory attendance. Metropolitan Police civilian staff walked out over similar mandates. This is not a right-wing government; it is a centre-left one, enforcing the same policy its predecessor introduced.

Farage went further than any mainstream British politician at his Birmingham rally on February 9th, dismissing the entire concept of work-life balance and calling for remote work to end entirely. He offered no evidence. The Spectator noted that his audience was overwhelmingly pensioners and that his own party had previously advertised remote-working roles. Even Reform’s own council leader in Nottinghamshire contradicted him, saying their hybrid policy “works very well.”

Canada tells a similar story. The Liberal government mandated three days a week in the office for federal public servants starting September 2024, up from two. By February 2026, Prime Minister Mark Carney’s government announced a further tightening to four days per week, with executives required on-site five days. This despite the fact that Carney himself has written extensively about the economic benefits of flexible work. The union representing federal scientists called it “a slap in the face”; multiple unions are exploring legal action. Meanwhile, Ontario and Alberta, conservative provinces, went further still, requiring full five-day return-to-office for their public servants in early 2026.

In Australia, the Labour premier of New South Wales ordered all 430,000 public servants back to office-based work in August 2024, giving staff one day’s notice. The union warned of 20% vacancy rates in critical agencies like child protection. The neighbouring state of Victoria, also Labour-run, took the opposite approach, offering to poach disgruntled NSW workers with its maintained flexible work policies.

In France, public administration workers are mandated to be in the office an average of 3.7 days per week, nearly double the mandate in London. The corporate sector is moving in the same direction. Amazon, JP Morgan, Goldman Sachs, and Meta have all imposed five-day return-to-office policies. Some of these mandates are openly designed to trigger attrition. Amazon’s CEO denied it was a “backdoor layoff”; few were persuaded.

This is not a left-versus-right issue. It is a cross-partisan consensus that remote work is a perk workers don't deserve. Evidence that it also raises fertility doesn't soften the position; it enrages it. The idea that letting people work from home one day a week might be good policy is intolerable to leaders who have already decided that flexibility is something to be clawed back.

Research from King’s College London finds that 55% of women say they would seek a new job if required to return full-time, compared to 43% of men. If RTO mandates disproportionately push women out of the workforce, or into less flexible, lower-WFH occupations, the fertility implications run in both directions. More WFH means more babies. Less WFH may mean fewer.

The Aksoy paper suggests that work-life balance, operationalized as even one day of remote work per week, is associated with 0.32 extra children per woman across 38 countries. One of the paper’s co-authors, Cevat Giray Aksoy, a lecturer at King’s College London, told LBC directly that “the research does not support his blanket claim that working from home is nonsense.” The Spectator observed that Farage’s position makes him “not a conservative, but a malcontent of boomer liberalism.” Even the TaxPayers’ Alliance cautioned that a party claiming to support free markets “cannot, in the same breath, start micromanaging how private firms organise their workforce.”

If you think that there are tensions in the WFH debate that the paper does not resolve. That some organizations may suffer from reduced in-person collaboration. Some roles genuinely require physical presence. One-size-fits-all mandates in either direction are probably a bad idea. But if you believe, as Musk says he does, that low fertility rates threaten civilization itself, and he has donated $10 million to population research at the University of Texas to back up that belief (which isn’t enough in the grand scheme of things), the evidence increasingly suggests you should be expanding remote work opportunities, not eliminating them.

The cheapest pro-natalist policy available is guaranteeing that people work from home one day a week, if the role allows for it. It costs the taxpayer nothing. It requires no new legislation. And a growing mountain of evidence, including this study encompassing 38 countries and multiple identification strategies, suggests it raises fertility by a nontrivial amount. The question is whether anyone who claims to care about birth rates is willing to take the evidence seriously, or whether the culture war over remote work will continue to trump the demographic concerns that supposedly animate it.