Workman Co., Ltd. (7564): a fortress at a full price

Let me tell you what this company does before I say a word about the stock, because that is the only honest order to do it in.

Workman sells work clothes. Boots, gloves, rain gear, coveralls, and lately the same kind of tough, cheap, water-shedding jackets to ordinary people who discovered that gear built for a roofer also works for a rained-on commuter [E2, E9]. It does this mostly through a franchise model: it signs up individual owner-operators, ships them merchandise, and collects a cut — what they call Workman Charge income — plus it runs some stores itself E1. So the customer is a tradesperson or a value-hunting shopper who walks in, and the business Workman actually owns is closer to a low-cost distributor with a store sign than to a fashion house. The whole thing is one segment E2, no parent, no subsidiaries, a clean standalone company E4 — though it sits inside the Beisia group, with Cains holding 9.7% of the votes E3. I can explain that to my sister in a paragraph, which means it clears the first gate. This is inside the circle.

Now, is it a good business? The record says yes, quietly. Operating revenue went from ¥116.3bn in FY2022 to ¥160.9bn in FY2026 [F1, F5]; chain-wide store sales from ¥156.6bn to ¥209.2bn [F6, F7]. Operating income was ¥24.4bn then ¥29.7bn in the last two years [F8, F9]. Net income has run ¥18.3bn, ¥16.7bn, ¥16.0bn, ¥16.9bn, ¥20.6bn across the five years I can see [F15–F19] — not a straight line up, it dipped in the middle, but it never lost money and it ended higher than it started. Return on equity started at 18.9% and is 14.3% today [F26, F27]. The reason ROE drifted down is not weakness; it is that the company keeps piling up cash it doesn't need. Equity is now 82.8% of the balance sheet F28. A business earning mid-teens on equity while carrying almost no debt, selling the same simple goods the same way, with a stated creed of low-cost, standardized, data-driven operation aimed at a hundred-year cost advantage E7 — that is the kind of dull, durable toll-bridge economics I like. Their motto, "better things cheaper," and their everyday-low-price posture E8 is exactly the low-cost-operator moat that lasts, not the fashion moat that doesn't.

The balance sheet is a fortress. Cash and deposits are ¥83.7bn F42. Total liabilities of every kind are ¥31.8bn F25, and interest-bearing debt is a rounding error — ¥1.35bn of borrowings plus ¥343M of leases [F50–F52]. Subtract every liability from the cash and you still have ¥51.9bn of net cash D2 sitting inside the company. This firm could not go broke if it tried; the tide could go out for years and it would not be swimming naked. That takes permanent loss from leverage off the table entirely.

So the business is fine. Here is where I stop nodding and start counting.

The market wants ¥507.6bn for the whole thing at ¥6,220 a share D1. Against last year's net income of ¥20.6bn F19 that is an earnings yield of about 4%. Take out the ¥51.9bn of net cash D2 that I'm not paying a business multiple for, and I'm paying about ¥455.7bn for the operating company — roughly a 4.5% yield on the operations. Owner earnings are honestly a touch lighter than net income here: operating cash flow was ¥18.8bn and investing outflow ¥13.4bn last year [F37, F39], and this is a company still building stores, so a good part of that capex is growth I'd applaud, not maintenance I must subtract. Call owner earnings somewhere in the high-teens billions. Either way the yield to an owner buying the whole company today sits around 4%. That is better than a Japanese government bond, but it is not a fat pitch, and it leaves me no margin of safety if the tradesperson customer base keeps shrinking the way the company itself warns it will E10.

I want to be plain about that risk, because Workman is plain about it. The core customer — construction tradespeople — is a shrinking population as Japan ages E10. Management's answer is to broaden into general customers through the Plus and Colors formats E9, and so far it's working. But that turns part of my ten-year earnings estimate into a forecast about whether a workwear brand can keep winning ordinary shoppers, and that I cannot underwrite with the confidence I'd want at this price.

There's also what I can't see. The ledger stops before governance and pay [E12 note in the evidence header]. I don't yet know how the directors are compensated, whether they eat their own cooking, or how the 9.7% Cains stake behaves in a squeeze E3. And on capital allocation, the one thing that jumps out is a company hoarding ¥83.7bn of cash F42 while paying out only 35% of earnings F33 and holding a trivial ¥68M of treasury stock F54 — a fortress, yes, but also a lazy balance sheet that isn't being handed back to owners or clearly redeployed at high returns. That's a mild Sanborn smell: idle cash growing while owners wait. Not damning, but not the mark of a management measuring itself on per-share value.

Add it up. Wonderful little business, fortress balance sheet, understandable as daylight — bought at a price that gives me a roughly 4% owner yield and no cushion against a real, disclosed structural headwind. That is a fine company at a fair-to-full price, and my rule is that the discount must be the friend, not the story. At ¥6,220 there is no discount. I'd happily own it lower. Around ¥4,100 the operating business would yield me something like 5.5% net of the cash D2, with the growth as free upside and the fortress as my margin of safety. Until the price comes to me, or the general-customer growth proves it can fully replace the shrinking tradesman, this is a watch — a good business I am content to follow and not yet buy.

Start with the obituary, because that is the only honest place to start.

How does Workman die?

The core 職人 customer is structurally shrinking E10. Japan's construction skilled-worker population is falling with the birthrate; that is not a cyclical blip you wait out. The company has understood this and responded — WORKMAN Plus, WORKMAN Colors, general consumers E9 — which is correct and also introduces a second kill path. Workwear for professionals is bought on function; outerwear and casual for general consumers is bought, in part, on fashion. Once you have material fashion exposure you have inventory write-down risk, trend-cycle shortening, and the slow dilution of your EDLP pricing discipline. Those two forces together — secular decline in the core, fashion risk in the replacement — could gut the economics quietly over a decade. The company itself names the demographic risk openly E10, which is candor worth noting, but candor about a structural problem is not the same as solving it.

The third kill path: a well-capitalized fast-fashion operator decides functional workwear aesthetics are the next category worth owning and prices through the floor. WORKMAN's moat defends against that — barely — because of its cost structure, not because of any unique design copyright.

Death from the balance sheet is not in view. The company carries net cash roughly equal to 10% of total assets after netting all liabilities [D2: ¥51,943M on total assets of ¥185,257M F24]. Short-term debt is ¥1,350M F50 against ¥83,743M in cash and deposits F42. You will not lose permanent capital here from a solvency event.

The moat's mechanism

The moat is a lollapalooza of four reinforcing pieces, not one: (1) structural low-cost operation — the stated "100-year competitive advantage" pillar E7; (2) EDLP pricing discipline that deliberately shares the cost advantage with customers rather than harvesting it as peak margin E8 — the Costco pattern, which is the durable one; (3) private-brand dominance that lets them design cost out without surrendering the price gap to a brand owner; and (4) the franchise model E1 that keeps the parent company's capex modest while aligning franchisee incentives with throughput.

The EDLP mechanism is the most important single piece because it is self-reinforcing: low prices drive volume, volume drives purchasing scale, scale funds lower prices. The question is whether the mechanism survives the consumer-market expansion. A franchise selling commoditized workwear to professionals on pure function can hold EDLP discipline indefinitely. A franchise also selling fashion-adjacent casual to general consumers has to manage markdowns and fashion cycles — costs that eat into the EDLP math. The ledger does not yet have enough margin-by-channel data to adjudicate whether the expansion is diluting or extending the moat. That is the central unresolved question.

Incentives and ownership

Cains holds 9.7% of voting rights E3 as part of a 46-company Beisia Group E3. There is no listed parent E4. The Beisia Group affiliation creates cross-shareholding relationships with captive supply chain implications that are worth mapping; for now it reads as stable, cooperative ownership with no obvious agency conflict between the bloc and minority shareholders. Compensation and governance detail are not yet in the ledger [evidence.md source index, compensation: NOT yet extracted] — I cannot pass M46 or M49, and I will not invent what I haven't read.

The price

At the trial stamp of ¥6,220 and 81.6M net shares [F55 minus F56], market cap is ¥507,634M D1. FY2026 EPS is ¥252.64 F29, giving a trailing P/E of roughly 24.6x. Ordinary income is ¥30,567M F14 on operating revenue of ¥160,852M F5, a roughly 19% ordinary margin. ROE has declined from 18.9% in FY2022 F26 to 14.3% in FY2026 F27 — the equity base is growing faster than earnings, which is the arithmetic of retained cash accumulating at sub-ROE rates rather than business deterioration per se.

Net income grew from ¥18,303M in FY2022 F15 to ¥20,618M in FY2026 F19 — a 12.6% cumulative gain over four years, roughly 3% per annum. EPS moved from ¥224.28 F30 to ¥252.64 F29, a similar rate. Chain-wide store sales grew from ¥156,597M F6 to ¥209,234M F7 — 33.6% over the same window — suggesting the franchise network is expanding while the per-unit economics are holding rather than compressing, though I cannot confirm this without a same-store calculation.

At 24.6x trailing earnings, this is not cheap in a Graham sense. It is a fair price for a business with this quality of balance sheet and franchise mechanism — but it requires the consumer-market expansion to hold the margin trajectory, not erode it. The opportunity-cost question is whether 24.6x on 3% EPS growth beats sitting in Japanese government bonds or rotating into an obviously undiscovered name. I cannot honestly say the multiple today offers the margin of safety M88 demands.

The DPS trajectory and capital allocation

DPS rose from ¥68 F32 in FY2022 to ¥89 F31 in FY2026 at a 35.2% payout ratio F33. That payout discipline — not over-returning cash, not stuffing the dividend to signal — is the mark of a management that thinks about per-share compounding. Operating cash flow in FY2026 was ¥18,838M F37 against net income of ¥20,618M F19 — a 91% conversion ratio. That is almost exactly what M38 asks for.

Verdict

This is a good business. The franchise mechanism is real, the EDLP discipline is philosophically sound, the balance sheet is fortress. The structural death path is named and partially addressed. The management candor on the demographic risk E10 is honest. The price is fair for what it is.

But I won't call buy-below without reading the compensation disclosures (M46 is data-insufficient and it is a character gate), without same-store margin trend to confirm the consumer expansion is accretive not dilutive, and without a named opportunity-cost alternative that this clearly beats. The honest answer is: watch, come back when the data is deeper or the price is lower.

Downside first, because that is the only question that ever gets me hurt. Workman is a Japanese workwear retailer with a fortress under it: cash and deposits F42 ¥83,743M against total liabilities F25 of only ¥31,800M, so net cash after paying off every liability D2 is ¥51,943M. Interest-bearing debt is a rounding error — short-term borrowings F50 ¥1,350M plus lease obligations F51+F52 ¥343M, call it ¥1.7bn against ¥83.7bn of cash. Equity ratio F28 is 82.8%. This company cannot go bankrupt in any scenario I can construct from these figures. Revenue has climbed every year, FY2022 ¥116,264M F1 to FY2026 ¥160,852M F5; net income FY2026 ¥20,618M F19; operating cash flow positive every year I can see F34–F37. So on survival — my leverage and shutdown items (P20, P23–P31) — this passes with room to spare. Two bad years, five bad years, credit markets slammed shut: Workman sits on its cash and waits. Tails, I genuinely don't lose much to bankruptcy.

But here is where I have to be honest with myself, and it's my own P1 rule that stops me: an equity cushion is not an asset floor. Book net assets F22 are ¥153,456M, and it is tempting to point at that 82.8% equity and call it protection. It is not. What actually protects me in a stress scenario is the marked-down liquid and hard stuff. So let me mark it. Cash F42 ¥83,743M I take at face. Receivables — trade F43 ¥4,612M and franchise F44 ¥14,470M — haircut, call it ¥15bn. Inventory F45 ¥29,770M of workwear, half of it in a fire sale, ¥15bn. Land F47 is carried at just ¥5,395M and total tangible fixed F48 is ¥38,531M; investment securities F49 are a token ¥1M — there is no hidden securities portfolio here (P5 fails: nothing meaningful carried below market). Subtract all liabilities F25 ¥31,800M. My rough marked-down floor lands somewhere around ¥120–130bn. Against a market cap D1 of ¥507,634M, that hard floor is roughly a quarter of the price. So the "tails" branch is not "I lose a little" — it is "I paid ¥507bn for assets I can defend at maybe ¥125bn." The floor is real, but it is nowhere near the price. The other three-quarters of what I'd be paying is the business — and a business is only a floor when I buy it cheap.

Which brings the hard question: is there asymmetry at this price? No. Price stamp is ¥6,220, EPS F29 ¥252.64, so I am paying about 24.6x earnings. Net cash D2 is only ~10% of the cap — strip it out and the operating business still trades near 22x. Normalize earnings over the five years I have (average net income ¥17,691M across F15–F19) and the multiple is closer to 29x. My whole game is buying a dollar for fifty cents (P53). Let me build a crayon intrinsic value: 15x normalized earnings — generous for a single-country retailer facing a shrinking core customer — plus the ¥51.9bn of net cash, gets me to roughly ¥3,900 a share. Twelve times gets me to ¥3,240. The stock is ¥6,220. The market price is above my conservative IV, not at half of it. There is no margin of safety here. There is no back-of-envelope arithmetic that makes this a bargain — and when I need the price to already be right before I can call it cheap, it isn't cheap (P52).

Is the market pricing uncertainty as risk, or is this just a wonderful business at a full price? I have to name what the market is afraid of (P13), and I can't find fear. This is not a distressed business in a distressed industry (P16) — it's a compounder that has grown revenue and dividends for years, ROE F27 still 14.3%, priced up for quality. The one genuine structural cloud is in the evidence: the tradesperson population is shrinking with Japan's demographics E10, and management openly calls it "a structural challenge." That is a slow melting of the core customer — but management's answer is expansion to general customers via Workman Plus and Colors E9, and the market is clearly crediting that pivot, not discounting it. So the uncertainty that exists is priced as optimism, the opposite of my hunting ground. I want the melting ice cube priced as if it were already water; here the water is priced as if the ice will refreeze.

The clone check (P32–P35) actually cuts against a bargain too: this is a superb copy of the EDLP / private-label playbook E8 — "better things cheaper," everyday low price, the model that made Uniqlo and, further afield, the Walmart and Costco arithmetic work. The model is proven, which tells me the business is good. It does not tell me the stock is cheap. And I have no EDINET 5% or documented-investor precedent in the ledger to lean on (P35: data-insufficient). Cains/Beisia Group affiliation is disclosed E3, Cains holds 9.7% of the votes E3, no parent, no subsidiaries E4 — but the compensation and governance sections were not extracted (the ledger says so plainly), so every skin-in-the-game and self-dealing item (P40–P49) is data-insufficient. I will not guess at governance.

So: a fortress that cannot fail, a genuinely good business, and no price. That is not a Dhandho setup. Few bets, big bets — and this one doesn't clear the bar (P66) because the very first gate, a wide and obvious discount, is missing. When the whole edge of a name is "great company," I am in Buffett's territory paying a fair price for a wonderful business, and that is explicitly not what this lens does. I pass, and passing is the system working.

(Interpretation of Pabrai's framework from the cited corpus; no invented quotations. The IV and floor arithmetic is my reconstruction from the ledger figures, deliberately rough.)

Begin where the discipline demands: the knowledge bar. Can I claim to understand Workman's next ten years better than almost anyone who owns it? Honestly — not yet, and the reason is not the business, which is unusually legible, but the archive. What I have is the FY2026 yūhō's five-year key-metrics table, the balance sheet, and the business/strategy/risk prose. What I do not have is the compensation section, the governance report, or a downturn on record. So I will do what intellectual honesty requires: state plainly what I know, what I must assume, and what I am not permitted to pretend.

What I know is a good business, cleanly told. Workman is a franchise engine E1: it contracts mostly with individual operators and earns a charge on their store sales, running a single-segment specialty retailer in workwear and, increasingly, outdoor and casual wear E2. The philosophy is not decoration — "better things cheaper," a private-brand build-out, everyday-low-price, sell-through discipline E8, all resting on three declared pillars: low-cost management, data, and standardization, offered explicitly as a hundred-year competitive advantage E7. This is the kind of durable, boring economics one can, rarely, project. The numbers agree the value is compounding, not melting: operating revenue ¥116,264M in FY2022 to ¥160,852M in FY2026 F1F5, chain-wide store sales ¥156,597M to ¥209,234M F6F7, net assets ¥103,559M to ¥153,456M F20F22, net income ¥18,303M rising — with a dip through the middle years F16F17 — to ¥20,618M F15F19. Retained earnings have accreted to ¥147,544M F53. The balance sheet is a fortress: ¥83,743M of cash and deposits F42 against ¥31,800M of total liabilities F25 and essentially no real debt — short-term borrowings ¥1,350M plus lease obligations of ¥343M F50F51F52. Equity ratio 82.8% F28. This company survives any downturn without ever touching the capital markets; the leverage failure mode simply does not exist here (L20).

Now the two or three variables that actually decide the ten-year outcome (L1). First, the structural headwind: Japan's construction skilled-worker base — the 職人 who are Workman's original customer — is shrinking with aging and depopulation, and the filing names this as a structural challenge E10. Against it stands the second variable, the deliberate substitution: WORKMAN Plus and Colors, pulling general customers onto the same low-cost PB platform E9. The whole thesis is a race between an eroding core demand pool and an expanding adjacent one. The third variable is capital allocation — what becomes of that cash mountain. On the first two I can frame the question but not yet answer it; the yūhō gives me the strategy but the five-year window does not isolate how much of the FY2026 revenue jump F4→F5, a striking +17%, is general-customer organic volume versus price or store additions. That decomposition (L39) is data-insufficient, and it is load-bearing.

The decade-owner test (L15): if Tokyo closed for ten years, would I be content owning this at today's price on the operations alone? Here price bites. Market cap is ¥507,634M at the trial stamp D1 against FY2026 net income of ¥20,618M F19 — roughly 25x earnings. Strip the net cash after all liabilities, ¥51,943M D2, and the operating business is priced near 22x. That is not a dollar at fifty cents (L18); net assets are ¥153,456M F22, so I am paying about 3.3x book for a business earning a mid-teens ROE that has drifted down from 18.9% to 14.3% F26F27. A patient owner's return from here is dividends — payout 35.2%, DPS ¥89 F31F33, a ~1.4% yield on this price — plus per-share earnings growth. For the closed-exchange arithmetic to clear a decade hurdle at 22–25x with no re-rating assumed, earnings must compound at a genuine high-single/low-double-digit rate for ten years. The business may well do that. But I am being asked to pay the full compounder multiple in advance, which leaves no margin if the 職人 erosion outruns the general-customer engine.

The capital-allocation question is where my admiration cools into a watch. Trace the retained yen (L21). Investing cash flow was −¥29,929M in FY2025 and −¥13,404M in FY2026 F38F39 — real reinvestment into stores. Financing outflow −¥6,071M F40 is essentially the dividend; treasury stock is a rounding error at ¥68M book F54, so there is no buyback program to speak of. Meanwhile ¥83,743M of cash sits on the balance sheet F42, far more than a 82.8%-equity, low-capex franchise needs as working ballast. This is precisely the pattern Li Lu warns of in Japan: cash piling at deposit rates because returning it is not the habit (L24). It is not being milked — there is no parent, no controlling shareholder to milk it; the Beisia group affiliate Cains holds just 9.7% of the vote E3E4, and Workman is nobody's subsidiary — but a fortress that neither compounds its idle cash at a rising return nor sends it home is a melting corner of an otherwise growing business. The declining ROE F26→F27 is partly this: a numerator growing slower than an equity base fattened by unreturned earnings.

So: a genuinely good business in a legible franchise, financially unbreakable, with intrinsic value compounding — but priced at a full compounder multiple, allocating capital only adequately, and studied through too thin an archive to clear my own knowledge bar. I cannot yet claim the specialist edge (L8) or the finished work (L55) that a concentrated Li Lu position requires. This is a watch, not a pass and not a buy. I would want it at a price where the operating business trades nearer 12–14x earnings — a level that survives the 職人 erosion even if the general-customer engine merely holds. On today's share count of 81,613,128 F55F56, that is roughly ¥3,400 or below. Above that, I am paying for a decade of flawless execution I have not yet verified I understand better than the people who already own it.

Outside view first. The reference class is a high-quality Japanese specialty retailer with a franchise-royalty model, low leverage, and mid-teens returns on equity — structurally, its nearest cousins are Nitori (home goods), ABC-Mart (footwear), and Shimamura (apparel): asset-light-ish, single-format, domestically concentrated chains that compounded through unit growth plus private-brand margin. At ¥6,220 the stamp puts Workman at P/E 24.6x (¥6,220 / EPS ¥252.64 F29) and P/B 3.31x (¥6,220 / BPS ¥1,880, where BPS = net assets ¥153,456M F22 / 81,613,128 shares F55−F56). That is not a cheap-Japan screen result; it is a quality multiple. The base rate for that class is instructive: these names delivered high-single-digit revenue growth and slowly-mean-reverting margins, and the ones that kept a premium multiple for a decade were the minority that sustained above-class unit economics. The multiple already prices Workman as that minority. So the burden of proof is not "is this a good company" — plainly it is — but "is it enough of an outlier to justify paying 24.6x for ~3% trailing earnings growth."

Compute, don't assert. Operating margin is 18.45% (F9 29,676 / F5 160,852) — genuinely high for retail. But return on incremental capital across the five-year series is the tell: net income grew from ¥18,303M F15 to ¥20,618M F19, a 3.0%/yr CAGR, while net assets compounded from ¥103,559M F20 to ¥153,456M F22 at 10.3%/yr. Equity grew more than three times faster than earnings. That is the arithmetic behind ROE falling from 18.9% F26 to 14.3% F27: the company retained earnings faster than it could redeploy them at the old rate. Crudely, ~¥50bn of incremental net assets over four years bought ~¥2.3bn of incremental annual net income — an incremental return on retained equity in the mid-single digits, well below the 14.3% the base earns. Capital is piling up faster than it earns its keep. Note the honest contradiction: FY26 itself was a strong year — revenue jumped 17.5% and operating income 21.7% year-on-year, an incremental operating margin of 22% — so the trend is not monotonic decline; it is a multi-year drift interrupted by one good print. I cannot yet distinguish "structural reacceleration" from "one strong year in a mean-reverting series" [C86] because I have FY26 detail but only endpoints for the interior years.

Reverse-DCF. The ¥507,634M market cap D1 against ¥20,618M net income F19 is a 4.06% earnings yield. Solving the price as a no-growth-plus-g perpetuity, at an 8% required return the 24.6x multiple embeds ~3.9% perpetual earnings growth; at 9% it embeds ~4.9%. Trailing four-year earnings growth was 3.0%. So the price roughly demands that Workman reaccelerate and hold earnings growth above its own recent record, in perpetuity — a top-half-of-class assumption on a lens (C7, C8) whose base rate says growth and margins revert. The FY26 pop makes that reachable, not proven.

Strongest short case [C61], three cited facts: (1) ROE fell 460bp, F26→F27, as retained equity out-compounded earnings — the business is over-earning on paper while under-deploying capital, and the multiple capitalizes the good margin, not the poor marginal return. (2) The core customer is structurally shrinking: the yūhō itself names the declining construction-tradesperson population as a structural challenge E10, so the "expand to general customers" pivot E9 is not opportunism, it is a race against erosion of the original moat. (3) ¥83,743M sits in cash and deposits F42 earning almost nothing (investment securities are ¥1M F49) while investing cash flow ran −¥29,929M in FY25 F38 and −¥13,404M in FY26 F39 — capital is both idle and being spent heavily, and the payout ratio is only 35.2% F33 on a 1.4% dividend yield. A skeptic pays 24.6x for a decelerating-ROE, structurally-challenged franchise with a lazy balance sheet. The long case answers (1) and (3) only if the reinvestment (new formats, PB expansion E8) restores incremental returns — which FY26 hints at but four years of falling ROE argue against — and answers (2) only with a pivot whose unit economics I cannot reconstruct from the ledger [C15].

Load-bearing unknowns [C83], all data-insufficient, not in the ledger: (a) executive compensation and its linkage to results — the 役員の報酬等 section was not extracted [E12 absent]; (b) governance detail, board independence, and the terms of any related-party flow to the Beisia Group / Cains, which holds 9.7% of votes E3 — the governance report was not retrieved; (c) mid-term-plan targets and hit rate [C5], guidance accuracy [C6], and buyback execution [C51] — none in the archive. These are resolvable-now from public filings, not unknowable. Two of them (governance/related-party, capital-return policy) are load-bearing for any verdict stronger than watch, because at 3.31x book the entire thesis rests on management redeploying or returning the swelling equity base intelligently — and I have no evidence of how they behave, only that ROE is falling while cash accumulates.

Where this lands. The downside case [C72]: margins revert toward the FY25 17.8% operating level, growth normalizes to the ~3% four-year trend, the cash stays trapped at policy payout, and no governance unlock arrives. On roughly ¥17bn of normalized net income and a class-appropriate 14–16x for a decelerating compounder, fair value is near ¥3,000–3,500/share plus ~¥600/share of net cash D2 — a buy-below around ¥3,800 where the private-owner yield [C74] on trailing earnings (6.7%) clears a JGB-plus hurdle even if the skeptic's facts play out and the wait is paid only by the 35% payout. That is ~39% below the stamp, which is honest anchor-free distance [C94], not a rounding of the market price. This is a fine business at a demanding price with two unexamined governance-shaped holes. Watch, not pass — because the holes are resolvable — and buy-below only well beneath here.