The Shallow End of the Filing

On 12 June 2026, someone at Daiwa Institute of Research counted 101 Japanese listed companies that had received a shareholder proposal for that month's annual meeting season. Fifty-two of those 101 had received theirs from activist investors and other institutional shareholders, which was a record on a company basis for the firm's series, against 51 the year before. As of the 12 June count, a further six companies — five of them on the activist side — had proposals withdrawn, retracted, or rejected, and are excluded from both totals. So is every meeting held after June, because the tally only covers June. This is a preliminary number from a private research house, not an exchange statistic; Daiwa's October season-review report has in past years carried the confirmed June figures.

The June 2026 figure was the second-highest ever. The highest was June 2025, at 111 companies. In that year, seven of the 111 saw a shareholder proposal actually pass, which Daiwa noted as a record number of approvals. Seven out of 111 works out to about 6 percent, though that ratio is arithmetic performed here rather than a number Daiwa published, and the denominator counts companies rather than resolutions. In June 2026 the count of companies where a shareholder proposal passed was two. Nikkei, running its own tally, put the prior decade's typical range at roughly one to four, which makes 2025 the outlier and 2026 the return to form.

So the mechanism worth describing is not the approval. Note also what the record actually was: 101 companies is second to 111, and it is only the activist-side subset that set one. Proposals keep arriving at that volume into a process whose best year on record still ended with one carrying at seven companies out of 111 — and Daiwa's reading of even those seven is that they owed more to large shareholders voting yes than to any broad institutional shift, and that proposals likely to pass are frequently absorbed into the company's own agenda before the meeting. Something other than the vote count is doing the work.

The Disclosure Clock and the Threshold

Start with the disclosure clock, because it sets the tempo of everything that follows. Under Article 27-23 of the Financial Instruments and Exchange Act, anyone whose holding ratio in listed shares exceeds 5 percent must file a large shareholding report with the Prime Minister within five days, subject to a proviso exempting cases where the total number of shares held has not increased and others specified by cabinet office ordinance. The statute does not say five business days; it says five days, excluding Sundays and the administrative holidays specified by cabinet order, which under Article 14-5 works out to Saturdays, public holidays, and 29 December through 3 January. Practically, five business days.

The report must state the holding ratio, the source of acquisition funds, and the purpose of the holding. Thereafter a change report is due within the same five-day window whenever the holding ratio moves by one percentage point or more, with carve-outs: nothing is required where the move does not involve a change in the total number of shares held, where it falls within the minor-change categories of Article 14-7-2 of the cabinet order, or where a report stating that the ratio has fallen to 5 percent or below has already been filed.

The threshold, not the deadline, is why the first thing about a campaign that is guaranteed to be public is a form, and why the form's "purpose of holding" field carries so much freight. The field connects to a defined term: juuyou teian koui-tou, "important proposal acts," set out in Article 14-8-2(1) of the enforcement order. Twelve enumerated items, concerning the issuer or its subsidiaries, addressed to the general meeting or to officers, with matters meeting a minor-significance standard excluded: disposal or acquisition of material assets; substantial borrowings; selection or dismissal of representative directors and representative executive officers and the appointment or removal of executive officers; the appointment of a specified person as an officer; material changes to board composition; share exchanges, share transfers, share deliveries, company splits and mergers; transfers, acquisitions, suspensions or terminations of business; material changes to dividend policy; material changes to capital increase or reduction policy; delisting; listing; and a catch-all delegated to cabinet office ordinance.

The FSA's Q&A, applied from 1 May 2026, reads the trigger as three cumulative conditions: an act of proposing to the issuer, falling within an enumerated item, with the purpose of causing material change or effect on business activities.

This term matters because of what it disqualifies. The special reporting regime under Article 27-26 lets qualifying institutions — Type I financial instruments business operators, investment management firms, banks — report on reference dates notified to the authorities in advance, at least twice a month, rather than within five days of crossing a threshold. Under the Monday option that means the second and fourth Mondays, plus the fifth in months that have one. The price of the convenience is not holding the shares for the purpose of important proposal acts, and then a set of exclusions in the cabinet office ordinance: a holding ratio above 10 percent, a purpose of acquiring above 10 percent, a joint holder outside the qualifying categories above 1 percent. Touch any of them and the fast clock resumes.

And if a holder intends to perform an important proposal act within a defined window after crossing 5 percent or after a 1 percentage point increase, the report must be filed five days before the act. The regime therefore prices intention, including the intention merely to buy more. A fund that wants the option of proposing something pays for it in disclosure speed, and pays before it acts, not after.

The 2024 amendments, effective 1 May 2026, rearranged the list. The enumerated items went from thirteen to twelve: the old entry on selecting or dismissing representative directors was widened to cover representative executive officers and executive officers, a new item on appointing a specified person as an officer was added, and two old items — important employees, important organizational units — were deleted.

In an August 2025 explanatory document the FSA sorted the items into those whose impact is relatively high, where the purpose condition is normally satisfied, and those whose impact is relatively low, where it is satisfied only if pursued in a manner that does not leave the decision to management's own discretion: exercising the shareholder proposal right, publicizing a demand without the company's agreement, or signalling that one will do so. That taxonomy is not in the statute. It is the agency's interpretive tidying, and the FSA says plainly that its Q&A binds neither prosecutors nor courts.

Read that sorting rule from the filer's side and it says something precise, and narrower than it first looks. What is regulated is the act of proposing. For the items the FSA rates as lower-impact, what tips them into a reportable purpose is the manner of pursuit — the shareholder proposal right exercised, the demand publicized without agreement, the signal that one or the other is coming — the manners that do not leave the decision to management. For the higher-impact items, the manner matters much less; the purpose condition is normally satisfied anyway. Publicity, then, is one of the instruments the regime notices rather than the thing itself, and a private letter that leaves the board to make up its own mind is, on this reading, the opposite of one.

The Second Clock, Filing to Meeting

Which brings us to the second clock, the one running from filing to meeting, and to what happens inside it. Japanese proxy advice has become unusually legible because both major firms publish thresholds. ISS's Japan benchmark policy, published 16 December 2025 and effective for meetings from 1 February 2026, recommends against the top executive of a company with an audit-and-supervisory-board structure where the five-year average ROE is below 5 percent without improvement in the most recent year, or where cross-shareholdings amount to 20 percent or more of net assets — subject to carve-outs, including companies listed under five years and top executives newly appointed in a rescue or turnaround.

Glass Lewis may recommend against the board chair where cross-shareholdings reach 10 percent of net assets, and may withhold that recommendation where the company discloses a concrete plan, with scale and timing, to fall below 20 percent of net assets by the end of fiscal 2030. Two firms, one variable, thresholds a factor of two apart, each with room to change its mind. On gender, ISS moves from one female director to 10 percent of the board, but the 10 percent standard applies only from 1 February 2027; the 2026 season still runs on one. Glass Lewis asks Prime Market companies for 20 percent.

The single most consequential line in either document is not a threshold at all. ISS recommends against the top executive who fails to implement a majority-supported shareholder proposal that ISS judges to be in the interest of shareholders, and who also fails to put an equivalent item on the following year's agenda as a company proposal with a for recommendation. A footnote records the Japanese context: shareholder proposals here often take the form of articles amendments, which require a special resolution of two-thirds. So a proposal can be supported by a majority of votes cast and still lose. The policy catches exactly that gap and converts it into next year's director election risk. Majority support without approval is not a failed proposal. It is a deferred one, with the deferral cost assigned to a named individual.

The distribution says how rare that is. Of the 450 sub-resolutions attached to the 111 companies in June 2025 — sub-resolution meaning each director candidate counted individually, a different unit from the parent-resolution basis used elsewhere in the same report — 58, roughly a tenth, cleared 30 percent support. Three hundred thirty-two, about seven-tenths, came in under 20 percent.

In June 2026, per a Mitsubishi UFJ Trust tally cited by Nikkei, nine companies and eleven resolutions were defeated with 40 percent or more support, up from four and four the prior year. The two that passed did so at roughly 63 and 80 percent. Note the direction of travel: approvals fell from seven to two, on two different houses' counts, while high-support defeats more than doubled. Those two series are not moving together, and with samples this small neither one is a trend yet.

There is also the arithmetic reason approvals are rare that nobody in the process finds mysterious. Daiwa's own reading of the 2025 approvals is that they owed more to large shareholders voting yes than to broad institutional support, and the firm notes that proposals likely to pass are frequently absorbed into the company's own agenda beforehand. A settled proposal disappears from the numerator. The approval rate therefore measures, in part, the failure of pre-meeting negotiation. Filing volume and approval rate can move in opposite directions without anything odd happening.

The float underneath all of this has been changing, and it is being measured. Both advisory firms now score cross-shareholdings as a fraction of net assets, which is a statement that unwinding them is a governance variable and not a balance sheet detail. And the Tokyo Stock Exchange, in a January 2026 study-group document, set out a plan to require companies with a shareholder holding 40 percent or more of voting rights to disclose the minority-shareholder breakdown of votes on company-proposed director elections, with a further requirement to publish an engagement policy and, within six months, feedback and additional measures where a majority of minority shareholders voted against.

That document is a proposal as of 26 January 2026, scheduled for a formal outline and public comment in spring; the final rule text has not been checked here. If it lands as drafted, applying from annual meetings for fiscal years ending after December 2026, the effect is to split the reported approval rate into two numbers, one of which has never been public.

The Counting Problem and Its Definitions

The counting problem is worth stating plainly, because the phrase in circulation — Japan as the world's second-largest activist market — is doing more work than its sources support. Barclays' Shareholder Advisory Group counted 255 campaigns globally in 2025, a record, with 141 in the United States and 56 in Japan, or 22 percent, and wrote that Japan continued to dominate non-U.S. activism at roughly half of all activity outside the United States. Lazard counted the same 56 in Japan but 297 globally, which puts Japan's share at about 19 percent.

Diligent Market Intelligence counts something else entirely — companies subjected to activist demands, 1,040 worldwide, 579 in the United States, about 246 in Asia — and describes Japan as the second-most active market globally, ahead of Europe, which is a comparison against a region rather than a country. For 2024, Diligent counted 64 Japanese companies as targets, against Barclays' 51 campaigns, as reported from its 2024 review, the original of which was not obtained here. A gap of roughly a quarter, and not even measured in the same unit.

The spread is definitional, and the definitions are mostly undisclosed. Barclays states one criterion clearly, repeated in a footnote on every page: campaigns at companies with market capitalizations above $500 million at announcement, with limited exceptions for companies that cleared $500 million at some point in the prior twelve months. What counts as a campaign, beyond the threshold, is not defined in the document, and Barclays' source line lists Bloomberg, FactSet, Diligent, 13D Monitor, Dealogic, and press reports — so Barclays and Diligent are not two independent verifications of the same fact, they share inputs. Lazard publishes no methodology on the relevant page. Diligent provides its methodology on request. No Japanese public body appears to publish a campaign count at all; none could be identified at the exchange or at the FSA.

The Japan Research Institute, in a May 2026 paper, said the quiet part with commendable directness: there is no single, established definition of an activist. It then defined one anyway — 27 funds trackable on Bloomberg — and looked at 295 companies those funds touched between 2022 and the end of 2025. Of those 295, some 130, or 44 percent, had received nothing but a large shareholding report. No formal proposal, no public campaign. A filing, and then whatever happened next happened somewhere that does not generate a document.

That filing-only question is one driver of the gap between 51 and 64, alongside Barclays' $500 million threshold and each provider's choice of which funds count as activists. It is not the whole of it, and the numbers cannot be netted against each other in any case: JRI's 130 span four years and a 27-fund universe of its own construction, while 51 and 64 are single-year counts.

The Visible Artifacts and Their Denominators

Which returns the question to where it started, from the other end. The visible artifacts are the filings, the resolutions and the vote percentages, and they arrive from three houses on three different units and periods, none of which divide into each other. JRI: of the 295 companies its 27 chosen funds touched between 2022 and the end of 2025, 130 produced nothing but a large shareholding report. Daiwa: of the 450 sub-resolutions at the June 2025 meetings, 332 drew under 20 percent. Nikkei, on its own tally of the June 2026 season: two companies saw a proposal carry, against the roughly one hundred that received one on Daiwa's 12 June preliminary count.

What these numbers share is not a denominator. It is that each exists because a rule requires a document. The five-day report exists because someone crossed 5 percent. The vote percentages exist because each company files its tallies in a statutory extraordinary report, and they matter twelve months later because a proxy adviser wrote a sentence about them. And the FSA's 2025 guidance treats the manner of pursuit — exercising the proposal right, going public without consent — as what tips a lower-impact matter into the important-proposal purpose test, which is not a prize but a bill: satisfying it costs an investor the monthly reference-date regime. The proposal is not filed in order to pass that test. Filing it is how you fail the exemption. This is the shallow end because it is the part the statute requires to be written down.

Lazard's first-half 2026 figures show Japan at 52 campaigns, up 53 percent year over year, against 184 globally and 94 in North America. Half a year at 52 against a full prior year at 56 looks explosive until you remember that Japanese annual meetings cluster in June, which loads the first half structurally; doubling it would be wrong. The honest summary is that the volume keeps rising and the approval count does not — with the caveat that the seven is Daiwa's and the two is Nikkei's, different houses on different bases, and that Nikkei puts the prior decade's approvals at roughly one to four, which makes seven the anomaly rather than two.

Individual outcomes are public: each company files its vote results in a statutory extraordinary report available on EDINET, and the exchange proposes to add minority-shareholder breakdowns for annual meetings covering fiscal years ending after December 2026. What no public body publishes is the aggregate. Every count in circulation — Daiwa, Nikkei citing Mitsubishi UFJ Trust, Barclays, Lazard, Diligent — is a private house counting on a definition of its own.

Currency note: yen amounts, where they appear, are converted at ¥163.8 to the dollar, the rate on 24 July 2026, applied uniformly regardless of the period the yen figure describes. Dollar figures from different years therefore carry no information about the exchange rate and should not be read as a time series. The yen is the original; the dollar is a convenience.