The Improvement Period

On 23 July 2026, the Tokyo Stock Exchange's own tally of listed companies read 3,900. On 31 January 2026, the same page read 3,933. The page carries a last-updated date and is revised as listings change, which makes it a running tally rather than an annual fact; it counts companies rather than issues, and five of the 3,900 are foreign. The composition that day was 1,552 Prime, 1,564 Standard, 597 Growth, and 187 on TOKYO PRO Market. The middle tier now contains more companies than the top one.

The year-end series is where the direction becomes legible. Counting all markets including TOKYO PRO Market: 3,869 at the end of 2022, 3,933 at the end of 2023, 3,975 at the end of 2024, and 3,945 at the end of 2025 — the first year-end decline in that run, by 30 companies. TOKYO PRO Market, a venue for professional investors, is usually left out of these conversations; it held 163 companies at the end of 2025 against 133 a year earlier. Strip it out and the three main markets went from 3,842 to 3,782, down 60, the second consecutive annual decline, as the Japan Research Institute noted in January 2026. The question of when Japanese listings started shrinking has two defensible answers depending on which market you decline to count.

The flows underneath: 59 IPOs on the three main markets in 2025 — 7 Prime, 12 Standard, 40 Growth — against 80 in 2024 and 88 in 2023. That 59 is an addition of the exchange's own segment rows; the headline JPX figure, 105, includes TOKYO PRO Market. Almost all of the decline sits in Growth, which went from 65 new listings in 2023 to 63 to 40. The Japan Research Institute attributes part of that to the April 2025 tightening of Growth's continued-listing bar, from a market capitalization of 4 billion yen (about $24 million) after ten years to 10 billion yen (about $61 million) after five; that reading is the institute's, not the exchange's.

Delistings, machine-counted from JPX's annual lists, ran 42 in 2019, 57, 86, 77, 61, 94, and then 125 in 2025 — 59 Standard, 45 Prime, 20 Growth, plus one foreign Growth issue. Fifty-nine minus 125 is negative 66, and the year-end tables show negative 60. The exchange publishes no reconciliation in that document; migrations up from TOKYO PRO Market and technical listings are the usual candidates.

The Exchange's Own Stated Reasons

Sorting the 2025 delistings by the exchange's stated reason, classified on the first twenty characters of each string: acquisition by another company via tender offer or share consolidation, 35; MBO, 24; squeeze-out through a cash-out demand, 14. That is 73 of 125, and the method is literally string matching, so variations in wording will shuffle a few items between buckets. The shape survives the crudeness. The most common way to stop being listed in Tokyo in 2025 was that somebody bought the shares.

Counting the buying is where the sources start disagreeing with each other in interesting ways. On Recof's data, 136 tender offers were filed with regional finance bureaus in 2025, against 100 in 2024 and 59 in 2022, beating the previous high of 104 set in 2007. Those are filings: a two-step structure counts twice.

Tokyo Shoko Research counts companies instead, and finds 112 that disclosed a take-private tender offer or MBO during 2025 — 80 tender offers and 32 MBOs, of which 5 failed — with 49 on Prime and 47 on Standard and one on Growth, a segment breakdown that covers 97 of the 112. TSR also states that its survey is a first edition with no prior-year comparison, so the record language has to come from somewhere else. And the MBO count for the same year is 32 on a company basis and 30 on Recof's deal basis. Neither number is wrong; they are answers to different questions that look like the same question.

Value tells a similar story with fewer participants. Recof's count as of 12 December 2025 had Japanese tender offers passing 10 trillion yen ($61 billion) for the first time — a provisional figure taken nineteen days before year-end. One transaction, a take-private led by the target's own corporate group rather than by management, accounted for roughly 4.684 trillion yen ($28.6 billion). Slightly under half of a record was a single deal. For scale, MBO value cleared 1 trillion yen ($6.1 billion) in 2023 and set a record on the same count, and even that was a running total published on 26 November of that year.

The management-led series, compiled by Daiwa Institute of Research in January 2026 from Recof data and company disclosures, covers take-private MBOs only: 6 in 2015, 6, 4, 4, 7, then 11 in 2020, 19, 12, 18, 18, and 28 through November 2025. Recof's separate tally, of MBOs by listed companies, put the full year at 30, about 70 percent above 2024 and a record on that basis — a different population, so the 28 and the 30 are not two points on one line. The Daiwa line is not a clean upward one either: 19 fell to 12 between 2021 and 2022, and it excludes tender offers launched by funds rather than management, which means it measures a slice of take-private activity rather than the whole.

The Rulebook with a Clock

Now the rulebook, which is the part with a clock in it. The exchange's transition measures, whose end was announced in January 2023, stopped applying to base dates falling on or after 1 March 2025. That is not a calendar event; it is a queue ordered by fiscal year-end. A company that misses a continued-listing standard gets an improvement period of one year, six months for the trading-volume test, then designation as a Security Under Supervision and, once a delisting decision is made, as a Security to Be Delisted — generally six months for the two together — then delisting.

For a March fiscal-year company the earliest complete path runs: base date March 2025, improvement period ending March 2026, delisting 1 October 2026. Companies that had disclosed, by 31 March 2023, a compliance plan running past their first base date on or after 1 March 2026 keep their supervision designation until the plan's own deadline, which can extend into 2027.

As of 20 May 2026, JPX's list of issues under transition measures says there are none. The relief is finished as an operating fact, not merely as a policy announcement. It also means nothing about whether the companies are compliant. They are on a different page.

On that different page, as of 30 November 2025 and presented to the exchange's follow-up council on 14 January 2026, 161 companies were failing a continued-listing standard: 46 Prime, 82 Standard, 33 Growth. The dominant tests are tradable-share market capitalization below 10 billion yen ($61 million) on Prime, 41 companies; below 1 billion yen ($6.1 million) on Standard, 62 companies; and market capitalization below 4 billion yen ($24 million) after ten years on Growth, 24 companies.

The exclusions are more interesting than the total: companies outside the transition framework are left out, and ten of those are also non-compliant; breaches of the net-asset test alone are left out; companies already designated because a take-private is pending are left out; and the base dates are staggered, with tradable-share data running through August 2025 and market-cap data only through October fiscal-year companies. The figure was also down 11 from the previous quarter.

Forty-eight of those improvement periods expired in March 2026. As of 1 April 2026, one company had been decided for delisting and moved to the to-be-delisted designation, on the Growth market-cap test, and 21 were under supervision — 1 Prime, 20 Standard, none on Growth. Forty-eight became 22, the difference being companies that came back into compliance or changed segment during the period. Of the 21, one Prime company is in a segment-change review, and one Prime and eight Standard companies already carry a second listing on another Japanese exchange, so a Tokyo delisting does not end trading in their shares; since only one Prime company is in the 21 at all, those two Prime descriptions may well be the same company, and the document lists the attributes without saying.

The published version of that council document still carried presenter notes from an editing stage, naming individual companies and marking two of them as special cases to be excluded. The file does not say the tally is assembled by hand. It shows the exclusions being decided one company at a time.

Behind them: 64 companies with improvement deadlines falling from April 2026 onward — 22 Prime, 27 Standard, 15 Growth — with a peak of 24 in December 2026. Separately, the exchange's live improvement-period list, dated 23 July 2026, shows 70 companies, 26 Prime, 24 Standard and 20 Growth, with 42 tradable-share market-cap breaches, 18 tradable-share ratio, 10 market cap and 5 net assets. Strip the net-asset-only cases, which have nothing to do with the transition regime, and it is 67. That list excludes companies already under supervision or already scheduled for delisting, which is why it cannot be subtracted from the 161 or the 64 to produce anything meaningful. Three counts, three populations, one phenomenon.

The Removals and the Migrations

The removals themselves are, so far, modest. The exchange's 2026 delisting list, as of 24 July 2026, carries 13 issues whose stated reason is failure to meet continued-listing standards: 11 Standard, one Growth, one Prime, the last already delisted on 1 July. Ten of the thirteen are dated 1 October 2026, which is the earliest date the machinery permits for a March fiscal-year company. Against 161 non-compliant companies in November, the standards have directly produced exits in the low double digits. The rest complied, moved, were bought, or are still inside the clock.

Moving is the underrated outcome. In the one-off re-selection window that ran from April to September 2023, 177 Prime companies — all former First Section constituents — shifted to Standard effective 20 October 2023, in a single batch. Ordinary segment changes recorded in JPX's tables from June 2022 to 2 July 2026, machine-counted, add 31 Prime-to-Standard and 36 Growth-to-Standard moves, with 13 of the Prime moves in 2025 and 18 in 2026 through 2 July. Traffic runs the other way too: 11 Standard-to-Prime and 26 Growth-to-Prime over the same window. This is an independent count of the exchange's own lists rather than a JPX-published aggregate, which is a caveat and also, unavoidably, the only way to get the number.

The Other Side of the Transaction

On the other side of the transaction, Bain & Company put Japanese private equity deal value at 4.8 trillion yen ($29.3 billion) for 2025, the fifth consecutive year above 3 trillion yen (about $18 billion); 2024 was 3.1 trillion yen ($18.9 billion). Exits reached 2.4 trillion yen ($14.7 billion), a record, against 1.9 trillion yen ($11.6 billion) the year before.

Deals above 100 billion yen ($610 million) made up roughly 70 percent of value, and take-privates of listed companies accounted for around half of it — Bain gives that as a share, not a yen figure, so the yen figure should not be reverse-engineered. Premiums on take-privates, per the same report, typically run 60 to 80 percent, measured from a pricing date the report does not specify. There is no official statistic for Japanese private equity volume; this is a proprietary count of disclosed deals, and undisclosed consideration is invisible to it.

There is also a second count, and it does not agree. Dealogic's PE-backed buyout series, denominated in dollars, showed $27.6 billion from 1 January to 20 August 2025, against $9.5 billion in the same window of 2024, with $40.3 billion in 2023 as the standing record. At 163.8 yen to the dollar, that 2023 record converts to roughly 6.6 trillion yen, comfortably above Bain's 4.8 trillion yen ($29.3 billion) for 2025 — a comparison the sources themselves warn against putting on one time series, the definitions, the populations and the currency of record all differing.

Bain's own record claim is about exits, 2.4 trillion yen ($14.7 billion), not deal value; on deal value it says only that Japan cleared 3 trillion yen (about $18 billion) for a fifth straight year and remains a standout market. The article carrying the Dealogic figures described the same $40.3 billion as both 2023 and as last year's record, in different sentences. Two measurements of one phenomenon, mutually unreconcilable, folded into the same story about a record year.

The number most often wanted here — undrawn commitments sitting at Japanese buyout funds — has no public statistic behind it. What is published is an estimate from the Japan Investment Corporation, dated April 2026: venture capital dry powder of 1.3 trillion yen ($7.9 billion) at the end of 2025, down 12 percent year on year, equivalent to 4.1 years of that year's VC investment of 317.1 billion yen ($1.9 billion). That is roughly back to the 2019 level, which was 4.0 years, and exactly 2024's 4.1.

It is venture capital, not buyout. It is an estimate derived from a benchmark survey whose base date is December 2024, applied to prior years' fundraising. It is nonetheless the figure that circulates when people want to say how much money is waiting to bid for listed companies, and it describes a different asset class entirely.

What the paperwork actually describes, then: a set of standards that has so far reallocated companies between tiers more than it has removed them, a middle tier that now holds more companies than the top tier, and roughly 58 percent of 2025's exits — 73 of 125 delistings, sorted by the exchange's own stated reasons — occurring because somebody made an offer. In the private equity take-privates Bain counts, those offers have typically carried a premium of 60 to 80 percent; that is Bain's population, not the whole delisting list.

As of 24 July 2026, the exchange's 2026 delisting page already listed 114 issues for the year, plus one dated 1 January 2027, against 125 for all of 2025 — a figure that includes dates which have not yet arrived and excludes announcements which have not yet been made. Leaving, like everything else in this market, is filed some months before it happens.

A note on currency: yen figures here are converted at ¥163.8 to the dollar, the rate on 24 July 2026, applied uniformly to every year without adjustment. The yen is the original unit in these disclosures and the dollar is a convenience, so the dollar series says nothing whatsoever about exchange-rate movement over the period. Where a source published its own dollar figure, that figure is reproduced as published, which is why not every dollar number in this piece sits on the same scale as every other one.