The Jar That Does Not Open

On 6 April 2020, Nikkei BP Consulting published a ranking of the world's long-lived companies. Japan had 33,076 firms with a business history of at least a hundred years, or 41.3 percent of a world total of roughly 79,900. The United States, in second place, had 24.4 percent. The underlying data was as of October 2019. The 41 percent has been in circulation ever since, usually without the date, usually without the denominator, and usually presented as a standing fact about a country rather than a count of rows in a file.

The same publisher ran the survey again. In the 2022 edition Japan has 37,085 century-old firms and 50.1 percent of a world total of 74,037, with the United States at 21,822 (29.5 percent) and Germany at 5,290 (7.1 percent). The 200-year count moves from 1,340 to 1,388, and its share from about 65 percent to 65.2. Between the two editions Japan gained roughly 4,000 firms and the world lost roughly 5,900. Of the nine-point rise, about five points come from Japan's own count going up and about three from the world total coming down. Neither movement is a claim about firms. Both are claims about rows. The 2022 release is no longer retrievable at its original address; the figures here come from several secondary outlets that agree with one another, which is a weaker form of confirmation than one would like for a statistic this widely repeated.

Establishment Year, Not Founding Year

The method is the first place to look for the movement. The counts are built from Teikoku Databank and Bureau van Dijk's Orbis, with business age computed from the year of establishment, not the founding year that Japanese longevity statistics normally use and which can precede incorporation. A firm qualifies only if it is active and has an address and revenue of at least one million yen (about $6,100) on file. Establishments, public bodies, foreign companies, religious corporations and primary and secondary schools are excluded. Five countries were dropped entirely because their data was judged unreliable.

The Japanese side of the ledger is fed directly by a domestic credit bureau with unusually complete founding-year fields. The rest of the world is fed by whatever Orbis happens to know. None of that proves what moved the number, but it is where a nine-point shift in a national attribute is likely to live: in coverage rather than in firms.

The OECD has published a validation study of Orbis, and it is not reassuring for cross-country league tables. Firms in Orbis are disproportionately larger, older and more productive than the populations they are drawn from. Japan's value-added coverage is 10 percent or less, the worst band among the twenty countries examined and one it shares with the United States, and Japan was excluded from the study's hand-picked sample because only 2002 to 2004 had stable coverage. The authors conclude that Orbis is better suited to studies taking a global perspective than to comparisons across countries. That paper was measuring financial variables, not founding-year fields, so the warning is adjacent rather than direct. It still lands on a ranking whose ordering is decided by who is in the file.

The numerator does not settle either. Teikoku Databank counted 46,708 Japanese firms of a hundred years or more as of December 2025, out of a base of about 1.5 million, and put the incidence at 3.11 percent, the first reading above three. Tokyo Shoko Research, screening about 3.3 million firms for a founding year of 1923 or earlier, counted 42,966 on a 2023 basis. Nikkei BP's 2022 edition counted 37,085. The spread is roughly ten thousand companies, about a fifth of the largest count, and it exists entirely within one country and one language, across three proprietary databases with three different as-of dates and, in Nikkei BP's case, incorporation year in place of founding year.

The depth figures belong to whichever file you happen to be holding. Teikoku Databank has 1,836 firms at 200 years, 905 at 300, 47 at 500 and 11 at a thousand. Tokyo Shoko Research has 869 between 200 and 300 years and 228 at 500 or more, which is nearly five times as many five-century firms from the file that finds fewer century ones. The incidence number deserves particular care. The prior survey, as of September 2024, found 45,284 firms and 2.75 percent on a base of about 1.64 million. The numerator rose by 1,424. The denominator fell by about 140,000. Most of the celebrated jump above three percent happened underneath the fraction.

The 200-year claim runs on two incompatible worlds. Nikkei BP's 2020 edition put Japan at 1,340 of a global 2,062, or about 65 percent; the 2022 edition revised Japan to 1,388 and 65.2 percent, with no world total for that band in the materials that survive, so the comparison that follows is with the 2020 numbers. The Bank of Korea's 2008 study, which is the origin of the older and more famous version of this claim, found 5,586 bicentenary firms across 41 countries, of which Japan had 3,146 (56.3 percent), Germany 837 (15.0), the Netherlands 222 and France 196. The same phenomenon, world totals differing by a factor of 2.7.

Neither is a census, and the two sampling frames share almost nothing; the Korean study's extraction rules are not documented in the materials available here. It also put Japan's century-old firms at about 50,000, which is above every domestic measurement taken since. Its internal breakdown is confirmed here through the central bank's press materials and Korean newspaper coverage from May 2008 rather than the report itself, and the numbers are eighteen years old.

The Domestic Side of the Ledger

Which leaves the question the ranking cannot answer: what longevity does to capital. Here the data is domestic, current, and disagrees about magnitude rather than existence. Teikoku Databank puts the share of Japanese companies without an identified successor at 50.1 percent in 2025, down 2.0 points, the seventh consecutive year of improvement and the lowest in the twelve editions of the survey. That is about 138,000 firms. The peak was 66.5 percent in 2017, so the decline is 16.4 points, though the series is discontinuous before 2016 and past years are revised as new data arrives.

The gradient by size is steeper than the trend: 24.9 percent at large firms, 51.2 at mid-sized, 57.3 at small ones, with the smallest firms improving least. By prefecture the range runs from 33.9 percent to 73.7. This is a private credit database of about 270,000 analyzable firms, not an official statistic, and the smallest operators are the ones it sees least well.

There is a mechanical reason a successor-absence rate falls that has nothing to do with successors being found. A firm with no successor eventually stops filing anything at all. Tokyo Shoko Research counted 67,210 closures and dissolutions in 2025, up 7.2 percent and the third consecutive record, with services at 21,961 (32.7 percent), construction at 10,283 and retail at 7,903. Teikoku Databank, publishing on the same day, 9 January 2026, counted 67,949 for the same year and called it a decline of 1.6 percent and the second-highest of the past ten years. The levels are within about one percent of each other. The signs are opposite.

The definitions differ on whether private workouts are excluded, and Tokyo Shoko Research revised its dissolution detection in 2024 to key off statutory public notices, a change that sits inside the 25.9 percent jump recorded from 2023. Anyone quoting a growth rate here is quoting a methodology.

The profitable-closure statistic has quietly stopped saying what it is usually used to say. Of the closing firms whose final accounts are known, Tokyo Shoko Research finds 52.8 percent were profitable in 2025, following 54.9, 52.4 and 51.5 in the three prior years. Teikoku Databank puts the profitable share at 49.1 percent, the first sub-50 reading in a series it can only retrieve back to 2016. The widely repeated line that six or seven closures in ten are profitable describes the period from roughly 2000 to 2020; it fell below 60 percent in 2021. The two current figures sit within a few points of each other on opposite sides of a round number.

Tokyo Shoko Research computes its share only on firms with a retrievable final statement, reaching back up to two years to find one, which skews toward the larger end of a population defined by its smallness. Teikoku Databank does not disclose the equivalent denominator.

The age data is the least ambiguous thing in the file, provided you stay inside one file. Among closing firms whose representative's age is known, Tokyo Shoko Research finds 90.6 percent aged 60 or over, the first time above nine in ten, with 38.6 percent in their seventies and 34.0 percent at 80 or above. The average is 74.9 years, against 72.6 a year earlier; the median is 76, against 74. The 80-and-over share was 26.2 percent in 2024. Teikoku Databank, same year, reports an average of 71.5, up 0.2 on its own prior reading, and an 80-and-over share of 24.4 percent, which is another reason not to mix the two series.

On Tokyo Shoko Research's two measures the exiting cohort aged more than a year in a year. That is either a change in who exits or a change in what the survey catches, the 2024 revision to dissolution detection sitting directly underneath the move from 26.2 to 34.0, or some of both.

Succession itself is becoming less familial. Among firms that changed president in 2025, provisional figures put internal promotion at 36.1 percent against family succession at 32.3, with M&A and related routes at 20.6 and outside hires at 7.6. In the confirmed 2024 numbers family succession still led, 35.7 to 35.0. The compiler hedges its own headline, writing that internal promotion may also lead once 2025 is final, so the crossover is not yet a fact. Among identified successor candidates, non-family stands at 41.0 percent, past 40 for the first time, against 29.7 percent for children, 24.6 for other relatives and 4.7 for spouses. Whatever else this is, it is the separation of ownership from management happening one retirement at a time.

A Clock in the Tax Code

There is also a clock, and it has been widely misread. The special measure suspending tax on transfers of unlisted shares requires a succession plan filed with the prefecture, and that filing deadline was extended by one year and six months in the FY2026 tax reform outline adopted by cabinet on 26 December 2025, moving it to 30 September 2027. What was not extended is the transfer itself: the gift or inheritance must occur by 31 December 2027, unchanged. The Small and Medium Enterprise Agency states both dates plainly. The National Tax Agency's own explanatory page, written on the law as of 1 April 2025, still carried the pre-amendment deadline when checked on 26 July 2026. The paperwork got more time. The deed did not. As of today that leaves about seventeen months.

The public alternative to a family successor is running at record volume and a very small average size. The 48 government succession-support centers completed 2,265 third-party transfers in FY2025, bringing the cumulative total to 14,571, with new inquiries above 24,000 for the first time and 1,903 completions of family-internal support. The scale is the point. Among FY2024 transferors, 36.1 percent had annual revenue of 30 million yen ($183,000) or less and another 31.9 percent fell between that and 100 million yen ($610,000), so 68.0 percent were at or under 100 million yen. Firms above 500 million yen (about $3.1 million) were 5.0 percent.

The private side of this market has grown to about 3,400 registered support organizations as of March 2026, of which roughly 1,200 are dedicated M&A intermediaries or advisors, and the ministry notes that fee rates rise as deal size falls because minimum fees bind. A qualification exam and a register of those who pass it are under consideration. Nothing in that sentence has a date attached yet.

The Front Door of the Listed Market

Meanwhile the listed market is shrinking through the front door. There were 125 delistings from the Tokyo Stock Exchange in 2025, the most in the eleven-year series METI compiles from exchange data, after 94 in 2024 and 61 in 2023. That series runs 66, 67, 40, 61, 42, 57, 86, 77, 61, 94, 125, excluding TOKYO PRO Market and share warrants. Three of the 125, or 2 percent, were failures to meet listing standards. The rest were transactions: 55 acquisitions by another company (44 percent), 34 buyouts by a controlling shareholder (27 percent), 26 management buyouts (21 percent), six holding-company reorganizations and one intra-group restructuring. The exchange is evicting almost no one: three companies. It is being bought out from underneath.

Three counts of the going-private wave circulate, and they come from two data sets. On a Recof-based series compiled by Daiwa Institute of Research, MBOs involving a going-private transaction run 6, 6, 4, 4, 7, 11, 19, 12, 18, 18 from 2015 through 2024, then 28 through November 2025. The Nikkei reports 30 for the full year, up about 70 percent and a record; the compiler and the definition behind that figure are not stated in what is publicly readable, though 18 to 30 is the same shape as 18 to 28.

Tokyo Shoko Research, counting disclosures rather than completions, finds 112 listed companies disclosed a going-private tender offer or MBO during 2025, split 80 and 32, with funds including activists making up 27.5 percent of tender-offer buyers, parent-company entities 22.5, same-industry buyers 18.7 and large shareholders 17.5; by market, 49 Prime, 47 Standard, 16 Growth, and five companies where the deal lapsed. METI, counting delistings actually completed, records 26 MBOs in the same year. Announcement, disclosure, completion: three definitions, one direction.

The money series is worse documented than the count series. In 2023 the annual value of MBOs passed one trillion yen ($6.1 billion) for the first time, reaching 1.4153 trillion yen ($8.6 billion) across 17 deals as of 11 December, a provisional figure. A separate tally taken about two weeks earlier put the same year at 16 deals and over 1.1 trillion yen ($6.7 billion), which is what year-end counts do. One transaction, announced on 24 November 2023 at roughly 710 billion yen ($4.3 billion), accounted for about half the annual total on its own. Comparable yen totals for 2024 and 2025 could not be located in primary or secondary sources for this piece. The deal count is complete through last year. The value series stops in 2023.

What a going-private candidate looks like has been written down. A securities research house published a four-condition screen in April 2025: founder or management holding 10 percent or more, price-to-book below 1.0, share price at or below half its post-1980 high, and net cash of at least 0.4 times market capitalization. The conditions were reverse-engineered from ten MBOs in its own small-cap universe, of which four traded below 1x book and seven below 1.5x, eight sat below half their high, and the average net-cash ratio was 0.4x with three at 0.8x. Ten cases is a very small sample and a screen fitted to it is a description, not a prediction. Thirteen companies met the criteria; two of them announced going-private tender offers during 2025.

Separately, Daiwa notes that MBOs have increased among firms trading below book and among firms it describes as at elevated risk of activist shareholder proposals, and puts the average premium across the sixteen most recent tender offers in its sample at 39.3 percent before price revisions and 46.7 percent after, in a January 2026 report covering deals through November 2025. The screen and the premium describe the same situation from opposite ends: assets carried at a value the market declines to pay for, held by people who do not have to sell.

That is also, in the language of capital rather than culture, a description of a very old company. Cash with no scheduled use. Shares held by people with no need for a public price. Records that stay accurate because nothing ever forces them to be restated. The counting problem and the capital problem turn out to have the same shape, which is why the league table moves nine points when the same publisher counts again two years later from the same two databases, and why two credit bureaus counting the same year's closures land about nine points apart on the growth rate with the signs reversed, 7.2 percent up against 1.6 percent down, on definitions, populations and detection methods that differ, the statutory-notice trigger adopted in 2024 among them.

A jar that has not been opened in two hundred years is either full or empty, and the only way to find out is an event that nobody holding it has any reason to schedule. Last year 67,210 of them were opened, or 67,949, depending on who was counting.

A note on the currency. Every yen figure here is converted at ¥163.8 to the dollar, the rate on 24 July 2026, including a revenue threshold set against 2019 data and deal values from 2023. That is a convenience for readers who think in dollars and nothing more; the dollar figures above are not a time series, because they contain no exchange-rate movement at all. Where the two disagree, the yen is the original.