South Korea's Births Rise for a 25th Straight Month — And Four Sets of Numbers That Don't Fit One Table

South Korea registered 24,275 births in July 2026, up 11.1% from the same month a year earlier. That's the 25th consecutive month of year-on-year increases. The figure sits in the July population trends release that Korea's national statistics agency, the Ministry of Data and Statistics (국가데이터처), published on September 23. Over a roughly comparable stretch, Taiwan registered 61,742 births in January through August, down 15.2%, while China reported 7.92 million births for 2025, down from 9.54 million the year before.

Read that far and a chart draws itself: Korea in the black, its neighbors in the red. Try to build the chart and it snags almost immediately. Korea's number is a provisional monthly count. China's is an annual communiqué. Japan's 2025 confirmed count fell, while its 2026 preliminary count is up 0.5% — and that preliminary series has foreign nationals mixed into it. Taiwan counts only people on its domestic household registry.

For readers outside Asia, Korea has become the reference point in the global argument about population decline — the case people reach for when they want to show how far a birth rate can fall. So a 25-month run of rising births lands as a test of whether the number can be moved at all, which is why it travels well beyond Seoul. The trouble is that the comparison everyone wants to run, Korea against its neighbors, breaks before the data even gets to the question.

The order has to change. Before the four sets of numbers go side by side, each one has to be sorted by what it is actually measuring. Only after that does the next question open: was Korea's rebound built by population structure, by policy, or by marriage?

July's number holds a rise and a slowdown at once

Four figures carry the statistics agency's July release: 24,275 births, up 11.1%; 22,282 marriages, up 9.3%; 28,921 deaths, up 3.0%; and 7,764 divorces, down 0.8%.

Natural change came to minus 4,647, per the agency's own publication. With that month, Korea's population has now shrunk naturally for 81 consecutive months. Births rose and the population still fell, in the same release, on the same page.

The cumulative figures are wider. Births for January through July totaled 170,079, up 14.7% from the same period a year earlier and the highest for that window in seven years, since 2019. The Korea Economic Daily, Financial News and Dailian all carried the same numbers. The 25-month streak itself rests on reporting rather than on a single agency series we were able to pull; multiple Korean outlets state the same count, and they agree.

One phrase needs careful handling. Korean coverage of this year's birth data has repeatedly reached for a superlative — a record increase. What that phrase rests on is the 15.4% cumulative rise in the first half of the year. Attach it to July's single-month 11.1% and the basis underneath it no longer matches. July was also a month that slowed: June's increase was 15.6%. The streak extended to 25 months, and the growth rate came down 4.5 percentage points in the space of a month. Both facts belong to the same month, and both belong in the same sentence.

July's total fertility rate was 0.89, up 0.08 from the same month last year, as reported by Financial News. That number cannot be set beside anyone else's annual total fertility rate. A monthly total fertility rate annualizes the childbearing level of that one month; the full-year rates Taiwan and Japan publish are calculated over a different interval. Where this article compares fertility rates across the four, it uses annual figures only.

Three things have to travel with every number

We collected the latest birth statistics from the four places into a single table — with the statistical series and the counted population written next to each number, and the period folded into the first column. Delete those two columns and what's left is a ranking, and the ranking doesn't hold.

Two of the labels in that table are worth unpacking in their original languages. Japan's 月報年計 is the series built by summing monthly reports into a yearly total. Taiwan's 按登記日期 means the count is compiled by the date a birth was registered, not the date it occurred.

No two of the four have been confirmed to be counting the same group of people. Korea's definition we could not confirm. Japan's preliminary series takes in foreign nationals living in Japan and Japanese nationals living abroad. Taiwan counts only people on its domestic household registry. China's communiqué excludes residents of Hong Kong, Macao and Taiwan along with foreign nationals. Absolute totals can't be matched against each other.

Nor can Japan's 2026 cumulative figure be converted into a Japanese-nationals-only series. The count that covers Japanese nationals alone arrives about five months after the survey month, so July's edition lands around December — which makes rebuilding January–July 2026 on that basis impossible for now. China is a different problem again. It publishes birth figures once a year, so no number exists for the same window as the other three.

The sources are four agencies. Japan's come from the Ministry of Health, Labour and Welfare's Vital Statistics: the 2025 confirmed count published September 30, 2026, the 2025 approximate count published June 3, 2026, and the July 2026 preliminary count published September 25, 2026. Taiwan's come from the Ministry of the Interior's quarterly population statistics bulletin, the Winter ROC 114 and Spring ROC 115 editions. China's come from the National Bureau of Statistics' Statistical Communiqué on National Economic and Social Development for 2025, published February 28, 2026. Korea's come from the Ministry of Data and Statistics' July population trends, published September 23, 2026.

Here's how far the verification went. Japan's two figures, Taiwan's 2025 annual total and China's 2025 number were checked directly against original PDFs and release pages. China's 2024 figure of 9.54 million, carried alongside in the table, came from multiple Chinese outlets, and we did not check it against the primary document. Taiwan's January–August 2026 cumulative also fell short of primary sourcing; it rests on reports from the Central News Agency (CNA) and the public broadcaster PTS citing Ministry of the Interior data.

Lay the table out this way and only two cells share a period: Korea's January–July cumulative and Japan's January–July cumulative. China's 2026 number won't exist until January 2027 at the earliest. A sentence like "this year's birth scoreboard for four Asian economies" cannot be written at this point in time.

Japan's +0.5% includes people the confirmed count leaves out

Japan's Ministry of Health, Labour and Welfare publishes vital statistics in three forms. The preliminary series (速報, sokuhō) arrives about two months after the survey month. The monthly approximate count (概数, gaisū) arrives about five months out. The annual confirmed count (確定数) arrives in September of the following year. The three differ in release date and in who they count, and the ministry pins those differences down in a table of its own, in the section on survey outline and notes for users.

The preliminary series counts survey forms filled out. Japanese nationals in Japan, foreign nationals in Japan and Japanese nationals living abroad all go in, and so do events that occurred in earlier years. The approximate series counts Japanese nationals in Japan and strips out events from earlier years. Run the same year through both series and you get two different numbers, and both of them are correct.

The latest confirmed value covers 2025. In the confirmed count the ministry published on September 30, 2026, births came to 671,270, down 14,903 from the previous year, or -2.2%. That is the lowest since the survey began in 1899, and the tenth consecutive annual decline. The total fertility rate was 1.14, a record low and also a tenth straight year of decline. Deaths came to 1,589,533, down for the first time in five years. Natural change was minus 918,263, negative for a 19th consecutive year — though the gap narrowed from the previous year's minus 919,205.

Set against the approximate count from June, the revision is small. Births moved by 34, deaths by 44, and the rate of change stayed at -2.2%. The provisional series published four months earlier was confirmed almost exactly as it stood. Worth noting what still doesn't exist: the rates that use population in the denominator were published this time only as national totals, and the prefecture-level rankings and year-by-year series arrive in November 2026.

The ministry's own language on the -2.2% was that it is "moderate compared with the recent trend of around -5%." One sentence carrying both the decline and the slowing of the decline.

Then there's 2026. The only series that can speak to January–July 2026 is the preliminary one. Open it and Japan's January–July births come to 404,547, up 0.5% from the same period last year. That preliminary figure includes foreign nationals living in Japan and Japanese nationals living abroad. Reading the same window through the count that covers Japanese nationals alone is not yet possible: July's approximate count arrives about five months out, around December, and the 2026 annual approximate count arrives in June 2027. There is no Japanese-nationals-only figure for 2026 in existence today, which is the reason the 2026 preliminary number and the 2025 confirmed number should not be stacked against each other.

Sponsored

Advertise with Us

[email protected]

Sponsored

The source document adds one more note worth carrying over: the year-earlier comparison value is also a preliminary figure. So +0.5% is preliminary measured against preliminary, and the two sides of that comparison do match.

To get a sense of the size of that 0.5%, set it against the gap between the series. On the preliminary basis, births for the 12 months from August 2025 through July 2026 came to 707,661. On the confirmed basis, Japanese-national births for January through December 2025 came to 671,270. The periods don't fully overlap, but the gap between the two series runs to 36,391 a year. The increase of 1,852 births in January–July 2026 is a little over 5% of that gap — under 10% even if you scale it to seven months. And the preliminary count doesn't separate Japanese nationals from foreign nationals. Where those 1,852 births came from is not answerable with this data.

Stay inside the preliminary series and Japan's position looks like this: +0.5% for January–July, -0.1% across the most recent 12 months, -1.5% for July alone. It has moved from decline toward flat, by a margin too small to settle the direction. The run of ten consecutive record lows that usually gets cited in discussions of Japan's falling birth count is carried by the confirmed series, not the preliminary one. For 2026, the approximate count arrives in June 2027 and the confirmed count in September 2027.

Japan also has its own version of the single-month trap. Marriages in July 2026 came to 39,767 couples, down 32.3% from the same month a year earlier. Before hunting for a shift in institutions or in sentiment, check the calendar. The comparison month contains July 7, 2025 — the seventh day of the seventh month of Reiwa 7, a triple-seven date that comes around once in 30 years, and Tanabata besides. Registrations piled onto that day. According to Nikkei and others, July 2025 marriages were up 22.5% year-on-year. July 2026's plunge sits on the far side of that base effect.

Korea and Japan read the same factor with opposite signs

The four explain their own movements in four different vocabularies. Korea and Japan are the interesting pair, because they reach for the same factor.

Korea's statistics officials start from population structure. Park Hyun-jung, who heads the population trends division at the Ministry of Data and Statistics, pointed to marriages postponed during COVID-19 now coming through, and to the population in its early 30s — the main childbearing ages — rising continuously since 2021. That account reaches us through media coverage; we did not check the briefing transcript against it. In the agency's commentary on the 2024 marriage statistics, the same axis carried two more items: a broader positive view of marriage, and government policies encouraging it. The early-30s group in question is what Korea calls the echo-boom cohort — people born between 1991 and 1995, the children of the second baby boom of 1964 to 1974, from years when annual births ran to roughly 700,000, per reporting in Kyunghyang Shinmun, the Seoul Economic Daily and News1.

Japan's health ministry explains its decline with a shrinking young female population plus later marriage and later childbearing, per Nikkei and others. Japanese demography has a standard tool for this, splitting movement in the birth count three ways: a population factor, meaning the number of women itself; a marriage factor, meaning the share of people who are married; and a marital fertility factor, meaning births per married couple. Since 2005, the heaviest downward pressure on Japanese births has come from the first of the three.

Then open the 2025 confirmed statistical tables by mother's age, and they point at one thing.

Of nine age brackets, three rose. The 30–34 bracket added 2,234. Under-15 added 10, and 50-and-over added 5. Those last two come to 15 between them. In practice one bracket moved, and it is exactly the bracket Korea names as the engine of its rebound.

The other rows explain the outcome. The 35–39 bracket fell by 7,333, 25–29 by 4,691, and 20–24 by 3,103. The early 30s gained 2,234, and 12,024 drained out of the brackets immediately above and below. Japan's total fell even as the bracket Korea points to grew.

Diagnoses split inside Japan as well. The Japan Research Institute located the main driver of the first-half 2025 decline in falling marital fertility rather than in the population factor. That's a single private think-tank analysis, so it carries a low grade as evidence — but the fact that one country, using one standard tool, produces two conclusions is itself a reference point.

On the China side, what shows up first is the base. Behind 2025's 7.92 million sits 9.54 million in 2024, and 9.02 million in 2023 before that. In 2024 alone the figure rose by 520,000. The 2025 drop has an unusually high prior year sitting in its denominator. This article does not rule on the cause of that prior-year spike. The 2023 and 2024 figures came from multiple Chinese outlets, and we did not check them against the communiqué itself.

More conspicuous is a gap in the explanatory coverage. The population commentary the National Bureau of Statistics issued on January 19, 2026 is headlined "China's total population stood at 1.40489 billion in 2025 as high-quality population development continues to advance." The body is built around the qualitative development of the population and labor force resources. In a year when births fell by 1.62 million, the commentary carries no account of the decline. We opened that page and confirmed it directly. Whether the bureau explained the causes somewhere else — in a press conference Q&A, for instance — we could not confirm. The dragon-year pull-forward argument and the shrinking-childbearing-age-women argument that circulate inside China come mainly from portal self-published media, so this article does not use them as factual grounds.

The diagnosis coming out of Taiwan points at marriage. Cheng Yen-hsin, an associate research fellow at the Institute of Sociology at Academia Sinica, locates the core of Taiwan's falling fertility in a sharp rise in the unmarried population across every age group. Among married people, by her reading, childbearing stays close to two children on average. People who marry have children; fewer people marry. Housing costs, low wages and the burden of childcare come up alongside. That is a single-researcher citation. We did not reach a document in which Taiwan's authorities officially explain the 20.05% decline recorded in 2025.

A baby born in July 2026 was conceived around October 2025

The simplest way to test the policy-effect argument is a calendar. A baby born in July 2026 was conceived around October 2025. Set that as the baseline, lay this year's measures against it, and the sequence becomes visible.

Start with the most recent, the three-part package of spousal support measures. According to a Ministry of Employment and Labor (고용노동부) release dated September 17, 2026, it took full effect on September 18. A worker whose spouse has a miscarriage or stillbirth can take leave of up to five days, with the first three paid. Spousal childbirth leave was renamed and restructured; the window widened from within 120 days after the birth to starting as early as 50 days before the due date, for 20 days in total. Parental leave, previously available only after a child was born, can now begin before the birth where the spouse faces a risk of miscarriage or premature delivery. Effective date: September 18 — two months after the babies born in July. Any sentence that explains July's increase with this package inverts the order of cause and effect.

Other items changed as of January 1, 2026. The ceiling on maternity leave benefits rose from 2.1 million won to 2.2 million won a month. Reduced working hours for childcare widened from children up to age 8 to children up to age 12, and the benefit ceiling for that program rose from 2.2 million won to 2.5 million won a month. These items are compiled from policy-explainer outlets. We did not reach primary documents such as the Ministry of Economy and Finance's (기획재정부) annual "What Changes in 2026." The extension of child allowance eligibility to age 9, frequently mentioned alongside them, is still awaiting action in the National Assembly as of a September 30 check. It cannot be written up as a measure in force.

Even the January measures fall after the decision point of October 2025. They work on the burden of pregnancy and the period after birth; they don't reach the moment when someone decides whether to have a child at all. The local-government birth incentive experiments carry the same lag problem. Support added this year does not show up in this year's birth count.

The candidate that clears the lag test is marriage. Korea recorded 193,657 marriages in 2023. In 2024 the figure was 222,412, an increase of 28,755, or 14.8%. In 2025 it rose a further 17,914 to 240,326, up 8.1%. The 2024 increase was the largest since 1996, and the growth rate the highest since the series began in 1970. July 2026 marriages were up 9.3%, so the trend is still running. We confirmed the agency's marriage and divorce statistics through the KDI Economic Information and Education Center and a Korea Policy Briefing repost; we did not check the original release PDF.

Pinning the trend down numerically is harder. We did not obtain statistics on the average interval between marriage and first birth. How many births 28,755 extra marriages produce a few years later is not something that can be written without a basis for it. So the judgment stays conditional. The reading that rising marriages worked as a leading indicator is not ruled out by timing. The reading that measures enacted this year lifted July's births does not line up in time. One more disclosure: this lag check applies only to measures that took effect in 2026. Measures from 2024 and 2025 were not catalogued in this round of reporting, and weighing policy's share of the result requires that list first.

Three media markets read one rebound three ways

The same rebound picked up three different headlines in three language markets. Line the headlines up and the frames separate on their own.

Japan holds acknowledgment and reservation inside a single sentence. Nikkei wrote that Korea's birth rate rose to 0.80 in 2025, a second consecutive annual increase, and still among the world's lowest. Jiji Press attached its own qualifier — risen for a second year, but. The NLI Research Institute published a report comparing the background to Korea's two-year rise with low-fertility measures in Japan and Korea. Japan reads Korea as a policy reference case.

That "world's lowest" formula wobbles as soon as you check it against annual figures. Korea's annual total fertility rate for 2025 was 0.80, up from 0.75 in 2024, back above 0.8 for the first time in four years. Taiwan's total fertility rate for 2025 was 0.695, per the United Daily News. On single-year figures, Taiwan's is the lower of the two. Both numbers came to us through media reports; we did not verify either against a primary source.

Taiwanese outlets use Korea as a mirror. Storm Media (風傳媒) asked whether pouring government money in was what worked. Hakka News (客家新聞) was blunter: "Taiwan's monthly newborns break below 7,000, Korea hits 23,000 for a seven-year high." Business Weekly layered on a National Development Council projection — "Taiwan may be left with only 94,000 this year." On that projection, Taiwan's annual newborns would fall below 100,000 for the first time; we did not check the council's original document. The three causes Taiwanese outlets list for Korea's rebound are the marriage increase of the past two years, the growing population of women in their 30s, and improving social attitudes toward childbirth. That is close to the Korean statistics agency's official account, almost item for item.

Taiwan's own numbers, meanwhile, look like this. August 2026 produced 7,492 births against 16,470 deaths — deaths running about 2.2 times births. Total population at the end of August was 23,224,721, down for a 32nd consecutive month, with the share aged 65 and over at 20.62%. Those figures were reported by CNA and others citing Ministry of the Interior data.

Mainland Chinese outlets are the most skeptical of the group. Yicai (第一財經) ran a question mark — has Korea's fertility rate really turned a corner? — and Globe (《環球》), published under Xinhua, went looking for causes with a piece on what sits behind the "counterattack" in Korea's birth rate. On portal self-published media, headlines along the lines of "isn't this a last flicker?" and "Korea is our earlier edition" circulated. Those are low-grade outlets, carried here only as examples of framing. English-language coverage stayed closer to reporting the fact of a second straight annual rise; we couldn't reach the English originals and verified them through Japanese-language versions.

That leaves three readings. The first is that policy worked. Its grounds are that Korea's statistics agency included government policies encouraging marriage in its account of the causes, and the frame shows up most sharply in Taiwanese headlines. The second is that population structure produced the result. In Japanese demography's factor decomposition, the heaviest downward pressure since 2005 came from the shrinking female population, and applying the same tool to Korea explains the increase as a population factor too. The third is that marriage is the variable, and this diagnosis turns up in all four places: Korea's marriage surge in 2024 and 2025, the concentration of marriages in 2022 and 2023 that Chinese outlets point to, and the rising unmarried share that Taiwan's researcher identifies all point at the same variable.

The cohort behind the rebound starts shrinking in 2027

One of the three readings can be tested against a calendar year. The population-structure one.

The female population aged 30 to 34 — the core childbearing ages — peaks in 2026 at 1,732,115 and declines from 2027. The projected path runs 1,724,869 in 2027, 1,672,673 in 2030, and 1,375,412 in 2035. The 35–39 bracket passes its peak in 2031 and declines from 2032. Financial News reported these figures on August 3, 2026, citing the medium scenario of the statistics agency's population projections. We did not open the projection document itself in this round of reporting.

If the projection holds, a test date comes into view. The 30–34 female population falls by 7,246, or 0.4%, in 2027, and by 2030 it is 3.4% smaller than in 2026. Total births also depend on the other age brackets and on fertility rates, so the figures only acquire discriminating power around 2030. And because the 35–39 bracket keeps growing until 2031, the point at which both brackets shrink together falls after 2032.

That is where the directions split. If births are still rising around 2030, the increase will have been produced while the early-30s population was contracting — grounds for saying a variable outside the cohort was doing the work. If they turn down instead, the explanation for the 25-month run leans toward population structure.

The first number to watch arrives in late October: Korea's August population trends. That timing is an estimate drawn from the release pattern — May's data on July 29, June's on August 26, July's on September 23. Three things to check in it. Whether the streak extends to 26 months. Whether the growth rate falls further from July's 11.1%. And whether marriages are still rising. The marriage count sits in the same release, printed right beside the births.