Japan's 2025 Births at 671,000: A 15-Year Acceleration and the Marriage-Birth Paradox

On June 3, 2026, the Ministry of Health, Labour and Welfare (MHLW) released the preliminary 2025 vital statistics. The birth count came in at 671,236, down 14,937 (-2.2%) from the previous year. This is the lowest figure since records began in 1899, extending the record low to ten consecutive years. The total fertility rate (TFR, the average number of children a woman is projected to have over her lifetime) stood at 1.14, the lowest reading since 1947.

The headline phrases — "tenth consecutive record low," "lowest TFR since 1947" — are familiar. The genuinely new substance of this release sits elsewhere.

In its 2023 release, the National Institute of Population and Social Security Research (IPSS) projected the 2025 birth count at 749,000 under its medium variant. The 671,236 reality sits 78,000 below the medium variant and even below the low-variant figure of approximately 697,000.

The heavier point is that 670,000 was not expected to appear until around 2040. A number meant for 15 years ahead has arrived in 2025. Our earlier article "Behind a 1.13 Fertility Rate" framed this as a roughly 16-year acceleration on preliminary 2025 estimates; the confirmed preliminary figure of 671,236 leaves the gap of roughly the same magnitude.

The other numbers released alongside the headline figure round out the structural shift.

The marriage figure deserves attention. 489,119 is the second consecutive annual increase, building on 474,000 in 2023 and 485,000 in 2024. Pandemic-postponed marriages account for part of the rebound, but the central fact is that rising marriages and falling births now coexist. The naive assumption that "more marriages will mean more births" no longer holds.

Another threshold has been crossed on the prefectural map. Tokyo's TFR fell to 0.96, the first sub-1.0 reading for any prefecture in Japan. Okinawa held the national high at 1.52, although it sits in the middle of a long decline from the 1.86 it recorded in the early 2010s.

What "15 Years Early" Actually Contains

The IPSS medium-variant projection is built on the long-run TFR trend and an assumption of convergence toward a long-term TFR of 1.36. It has been treated as the "main scenario" because the past three decades of trend largely fit inside that range. Three structural drivers explain why the projection has now slipped by 15 years.

First, the mother cohort is smaller. Women now in their early thirties were born in the late 1990s, when annual births had already fallen into the 1.2-million range. With cohorts this small, even modest fertility declines compound into large drops in absolute births.

Second, late childbearing is accelerating. According to the MHLW Vital Statistics, the average age of mothers at first birth in 2024 was 31.0, up 1.9 years from 29.1 in 2005. As more first births shift past 35, the probability of reaching second or third births falls.

Third, the structural decline in married fertility (the birth rate among married women) has continued. As laid out in "Behind a 1.13 Fertility Rate," married fertility shifted around 2015-2016 from a force pushing births up to a force pulling them down. A growing share of married couples now exit before reaching even the first birth.

These three drivers are not independent. Small cohorts plus late childbearing leave less time to reach second or third births; falling married fertility amplifies the cohort effect. The IPSS scenario in which TFR converges back toward 1.36 sits further from reality than at any point in the projection's history.

In other words, "15 years early" is less a single shock than the simultaneous surfacing of 15 years of demographic inertia. Pulling that back requires policy to fight inertia, not to react to a one-off jolt.

The Acceleration Plan's First Full Year

2025 is the first year in which the "Acceleration Plan" of the Children and Families Agency's National Strategy for Children's Future is in full implementation. The three-year plan moves through approximately 2.3 trillion yen in FY2024 (about 60% implementation), over 3.0 trillion yen in FY2025 (over 80% implementation), and a 3.6-trillion-yen baseline by FY2026. FY2025 is therefore the midpoint of the ramp-up.

The four main spending pillars are:

• Expansion of the child allowance (removal of income limits, 30,000 yen for the third child, extension through high school): about 0.6 trillion yen
• Higher-education cost relief (free tuition for households with multiple children): about 0.1 trillion yen
• Expansion of parental-leave benefits (post-birth support benefit at 80%): about 0.1 trillion yen
• Improved nursery staffing ratios and compensation: about 0.1 trillion yen

In this first full year of implementation, births fell 2.2%. Read superficially, that says "3.6 trillion yen and births still fell."

But this direct juxtaposition is immature as policy evaluation. Pregnancy and childbirth carry long lags between intention and outcome. With the average interval from marriage to first birth now around 2.8 years, the effect of an expanded child allowance will not appear in birth counts before 2026-2027. The 2025 figure is better read as a measurement of pre-Acceleration-Plan inertia, not of the plan itself.

The deeper issue lies in design. Professor Shintaro Yamaguchi notes that "cash transfers do raise the fertility rate, but the effect is not particularly strong," while OECD analysis by Associate Professor Yu Shibata finds that only "immigration and childcare expansion" carry statistically significant positive effects, with child allowances showing no significant effect. There is no clear indication that the 3.6 trillion yen is weighted toward what the evidence flags as effective.

The Acceleration Plan was designed against population assumptions close to the IPSS medium-variant projection. When the underlying projection has slipped by 15 years and the policy menu still embeds the old assumption, the design-to-reality gap will widen further.

The Marriage-Birth Paradox

Two consecutive years of rising marriages reflect pandemic-postponed weddings returning. Even so, 2025's 489,000 sits at just 44% of the 1972 peak of 1.099 million marriages. "Two years of growth" reads less as a turn than as a small rebound from a very low base.

The lag effects compound. With marriage-to-first-birth at 2.8 years and the proportion of second and third births continuing to decline, marriage gains in 2024 and 2025 will translate into births only in 2026-2027, and the translation will be partial. Japan Research Institute chief researcher Takumi Fujinami notes that young people's desire to have children is itself low, and that the share of married couples choosing not to have children continues to grow.

The structure in which more marriages do not translate into more births is being absorbed by falling married fertility and late childbearing. The earlier article's observation that "the fertility rate among married couples themselves has begun to decline" remains uncontested by the 2025 preliminary data.

Prefectural Gaps and Early-Warning Signals

Tokyo's 0.96 carries weight beyond a simple new minimum. The first sub-1.0 reading for any prefecture marks the formal entrenchment of a structure in which cities that attract young people fail to reproduce them. As mapped in "The Demographics of Tokyo Concentration," Tokyo absorbs young people from the regions but suppresses fertility through housing costs, education costs, and commuting burdens. The 0.96 figure now records that structure on the public balance sheet.

Okinawa's 1.52 remains the national high but is itself mid-decline from 1.86 in the early 2010s. The MHLW reports 13 prefectures with TFR readings above the previous year, led by Ishikawa at +0.07 points to 1.30.

The "13 prefectures rising" signal should not be over-interpreted as a reversal. TFR is the ratio of births to the female population aged 15-49. In regions losing young women, the denominator shrinks and TFR can rise mechanically even as the birth count falls. Ishikawa's rise overlaps with population shifts following the 2024 Noto Peninsula earthquake and should be read with care as an indicator of policy effect.

The 744-municipality discussion in "The Structure of Disappearing Municipalities" surfaces a layer the prefectural averages cannot reach. Beneath a prefecture that looks stable, individual towns and villages can be collapsing in parallel.

The East Asian Sub-1.0 Belt and the Wobble of the "Advanced-Nation Model"

East Asia is forming a belt clustered around 1.0. Intense educational competition, gendered wage gaps, megalopolis concentration, and declining marriages in the 20s recur across the region. Japan's 1.14 sits at the higher end of this belt but still represents barely more than half of the 2.07 needed for population replacement.

The "advanced-nation model" of fertility maintenance is also wobbling. France, long the model country with 2.9% of GDP in family policy spending, saw its 2023 TFR fall to 1.68 (the lowest since 1946) and an estimated 1.62 in 2024. Sweden, sustaining family policy spending at 3.4% of GDP, saw its 2024 TFR drop to 1.43.

The long-held premise that "3% of GDP in family policy will maintain fertility" is faltering across the advanced world at once. Japan's family policy spending stands at 1.6% of GDP, below the OECD average of 2.3%; but even raising it to French levels no longer offers a guarantee. Policy discourse that treats France or the Nordics as reference models needs to revise the reference points themselves.

Policy Lag, Demographic Inertia, and Social Security Design

A 2.2% decline in births during the first full year of a 3.6-trillion-yen Acceleration Plan illustrates the weight of policy lag and demographic inertia. In the short run, this year's spending cannot move next year's births.

But deferring evaluation in the name of policy lag creates its own problem. With the IPSS projection now 15 years off, the population assumptions baked into the pension, health insurance, and long-term care systems are off by the same margin.

Evaluating the Acceleration Plan will take 5-10 years. Updating the population assumptions of the social security system has to run in parallel. "Wait for policy effects" and "update system premises" need to proceed at the same time.

Sketching a Structural Prescription

The 2025 preliminary data point toward several axes for a structural prescription.

First, rebalance the 3.6 trillion yen toward measures the evidence flags as effective. Combining Shibata's OECD analysis (childcare expansion and immigration as the only significant positive levers) with Yamaguchi's caution on the limited effect of cash transfers suggests a shift away from a child-allowance-heavy composition. As laid out in "Children and Families Agency Budget Effectiveness," correcting the generational asymmetry of public spending (113.6 trillion yen for elderly-targeted programs versus 10 trillion yen for children) remains the core fiscal question.

Second, elevate the IPSS low variant — not the medium variant — as the base case for social security design. Dai-ichi Life Research Institute's Takuya Hoshino already frames "a breach of the low variant" as a realistic near-term scenario. Treating the medium variant as the "main scenario" will drift further from reality.

Third, read Tokyo's 0.96 as the self-undermining face of Tokyo concentration. If cities that absorb young people cannot reproduce them, regional population decline accelerates and Tokyo itself never recovers fertility. Regional dispersion policy and structural reform of Tokyo's housing and education costs belong to the same population policy.

As 『地方消滅２ 加速する少子化と新たな人口ビジョン』 (Vanishing Regions 2: Accelerating Population Decline and a New Demographic Vision) by the Population Strategy Council makes clear, regional disappearance and accelerating fertility decline are two faces of the same structure. "15 years early" calls simultaneously for earlier action, redesign of the policy mix, and updated projection assumptions. Neither chasing short-term results nor deferring everything in the name of policy lag will do.

Related Columns

• Behind a 1.13 Fertility Rate (the structural shift in married fertility and the wobble of the advanced-nation model)
• The Reiwa Year of the Fire Horse and Japan's Declining Birthrate (from one-off shocks to chronic shocks)
• The Structure of Disappearing Municipalities (744 municipalities and the black-hole pattern)
• The Demographics of Tokyo Concentration (young inflow alongside the lowest fertility)
• Children and Families Agency Budget Effectiveness (the generational asymmetry of public spending)

Related Practical Guide

References

Reference Books

• 『地方消滅２ 加速する少子化と新たな人口ビジョン』 (Vanishing Regions 2: Accelerating Population Decline and a New Demographic Vision) edited by Population Strategy Council (Chuko Shinsho, 2024): A decade after the original "Vanishing Regions" report of 2014, this volume updates the analysis of municipalities at risk of disappearance and proposes a new demographic vision. The foundational reference for the inter-prefectural and black-hole debates
• 『日本の少子化対策はなぜ失敗したのか? 結婚・出産が回避される本当の原因』 (Why Did Japan's Declining-Birthrate Countermeasures Fail? The Real Reasons Marriage and Childbirth Are Being Avoided) by Masahiro Yamada (Kobunsha Shinsho, 2020): Argues from family sociology that Japan's culture of risk avoidance and concern for appearances has neutralized the European childcare-centric model, providing context for a critical reading of the Acceleration Plan's design
• 『まちがいだらけの少子化対策: 激減する婚姻数になぜ向き合わないのか』 (Mistaken Countermeasures Against Declining Birthrates: Why Are We Not Confronting the Collapse of Marriages?) by Kanako Amano (Kinzai, 2024): Places the decline in marriages at the center of the declining-birthrate debate and critiques the over-focus on "married couples not having children." A complementary lens on the marriage-birth paradox