Subaru Corporation7270.T
Normalise Subaru at the pod's ¥130 yen and the margin disappears entirely — operating profit lands at ¥7.6bn on ¥4,785bn of revenue, which is neither profit nor loss but exactly break-even. What is left is not a carmaker but an option: a claim on the dollar staying above the level where the export spread pays for itself, backed by a balance sheet holding ¥728bn of surplus cash and buying back 8% of its own stock. Weighted across the scenarios, that option prices at fair value to within a fifth of a percent. There is nothing to buy, and — with the cash floor and the yield in the way — not enough to sell either.
Excess cash of ¥728bn — 39.6% of the market capitalisation — is credited in full. The market does not discount the cash ; it discounts the factory.
The operating net asset is ¥2,025bn of book value. Normalised at USD/JPY 130 it earns nothing, so it cannot carry an earnings multiple ; valued at 59.2% of book it is worth ¥1,199bn.
The two sum to ¥1,927bn of equity, or ¥2,734 per share on the pro-forma divisor of 704,997,107 shares.
The market prices Subaru at ¥2,611.50. Weighted across the three scenarios, fair value is ¥2,607. The option prices at par.
The margin Subaru was famous for was never really its own. At its FY March 2016 peak the group printed a 17.5% operating margin — the highest a Japanese mass-market carmaker had ever posted — and the market paid 2.31x book for it, the richest premium in the bucket. Ten years on the same business prints 0.84%, on 48% more revenue. That is not a company that lost a competitive edge ; it is a company whose margin was a function of two things it does not control, the yen and the tariff, and both have turned. Normalise the last full year to the pod's ¥130 and 15% tariff and the operating result comes out at ¥7.6bn — 0.16% of sales, break-even to the yen. The question the dossier turns on is therefore not whether Subaru is cheap on its P/E of 10.4x. It is what you are actually buying once the earnings power is priced at zero.
The honest answer is that you are buying a currency option wrapped in a balance sheet. Subaru builds 59.7% of its cars in Japan in yen and sells 79.0% of them in North America in dollars ; the profit is the spread between the two, minus the tariff that now taxes exactly that transfer. At ¥130 the spread pays for nothing. At ¥140 the normalised result is ¥117.6bn ; at ¥150 it is ¥227.6bn. The exposure is convex, and the market prices it as if it were linear — which is the one thing in the file that is genuinely mispriced, in both directions. Below ¥130 the earnings power turns negative and the case is a short ; above ¥145 it restores a real margin and the case is a long. The stock is the value between those two strikes.
The second thing you are buying is the reason the downside is bounded rather than permanent. Subaru has never issued equity in eleven years, runs a 50.6% equity ratio, and holds ¥1,305bn of pro-forma net cash of which roughly ¥728bn is genuinely surplus. That cash earns ¥36.4bn a year at the current 10-year JGB, and it funds a buyback of 8% of the shares that is already 18.2% executed. A business earning nothing on its factory but 2.8% on its cash, handing that cash back below book, is a strange object — but it has a floor, and the floor is why a nil-earnings carmaker is a watchlist name and not a short.
What the consensus is doing instead is pricing a recovery. The sell-side carries ¥234.6 of FY March 2027 EPS, 30.6% above the company's own guidance, on a tariff run-rate that is probably understated and while ignoring a ¥130bn materials assumption the company itself flags as conservative. That is the market pricing the ¥150-yen margin as if it were structural — the exact mistake the sector has punished for a decade. The variant reading is simpler : the normalised earnings power is nil, the balance sheet will be returned, and at ¥2,611.50 the market has, almost to the yen, already agreed.
The position framing is patient observation, not ownership. Weighted fair value of ¥2,607 sits on the spot, the asymmetry ratio is 0.71 against a long, and the quality grade is 12.0 of 25 — enough of a short bias to note, not enough to trade against 39.6% surplus cash and a 12.67% prospective yield. Conviction is moderate. The one variable worth watching resolves on a published calendar : the realised yen over the half-year to 30 September 2026, printed in November.
The decade reads as two peaks and two troughs, and all four are macro. The FY March 2016 peak — 17.5% margin, 77% return on capital ex-cash, 2.31x book — was the Abenomics yen doing the work while the company let the change flow through and lifted the dividend. That eroded into FY March 2019 as quality-recall and inspection costs took operating profit down 52% on revenue down 2%, a charge no volume or mix can explain. Then COVID and the semiconductor shortage cut revenue by a fifth, drove the margin to 3.3%, and pushed management to cut the dividend 61% — from ¥144 to ¥56 — with ¥692bn of net cash on the balance sheet and no solvency pressure. The weak yen after 2022 reopened the margin mechanically, revenue jumped past ¥4.7tn and the result reached ¥385bn, the second-best of the decade. And then the Section 232 tariff landed in April 2025 and took operating profit from ¥405bn to ¥40bn in a single year. Every inflection in the table below is a currency or a policy, not an execution.
| Inflection | FY mar 2016Abenomics peak | FY mar 2019Quality-cost erosion | FY mar 2022COVID trough | FY mar 2024FX reprieve | FY mar 2026Tariff wall |
|---|---|---|---|---|---|
| Revenue (¥bn) | 3,232.3 | 3,156.2 | 2,744.5 | 4,702.9 | 4,785.0 |
| EBIT (¥bn) | 565.6 | 181.3 | 91.8 | 469.1 | 40.0 |
| EBIT margin | 17.50% | 5.75% | 3.34% | 9.97% | 0.84% |
| Return on capital ex-cash | 77.2% | 17.0% | 4.9% | 28.8% | 2.3% |
| FCF (¥bn) | 487.5 | 120.1 | 94.4 | 579.5 | 123.9 |
| Net cash (¥bn) | 832.9 | 883.3 | 676.0 | 1,387.7 | 1,332.5 |
| EPS reported (¥) | 559.5 | 184.4 | 91.3 | 509.2 | 125.5 |
| P/B (close) | 2.31x | 1.15x | 0.79x | 1.01x | 0.64x |
Source: workbook 7270_Subaru.xlsm (Income Statement, Balance Sheet, Multiples) and data pack 22 July 2026, close-year "FY mar YYYY" convention. EBIT = reported IFRS/J-GAAP operating income. Net cash = −net debt. Return on capital ex-cash = NOPAT / (equity + net debt), reconstructed. The market last held Subaru durably above book in FY March 2020 ; the FY March 2024 print of 1.01x was brief and immediately given back.
Three management decisions sit inside the U. The Abenomics cash was never converted into the one asset that would have insulated the model — US production capacity — so local cover sits at 50.1% today against roughly 90% at Honda, at an observable cost of ¥320,000 of tariff per North American vehicle. The FY March 2021 dividend cut, taken with ¥692bn of net cash and no solvency need, saved 5% of the cash pile and cost the credibility of the distribution ; the de-rating accelerated across that window. And the BEV programme was capitalised, impaired by ¥57.8bn, and then formally postponed in the same reporting window — impairing and postponing together is the signature of a decision taken before it was disclosed. The discipline since is real but narrow : SG&A cut 11.2%, ten sales subsidiaries merged, and a buyback now running. It is corrective, and it is financed on the cash stock rather than the cash flow.
The consolidated margin measures the performance of no real operation, and the origination data shows why. In FY March 2025 the Japanese entity carried 20.1% of revenue and generated 75.5% of operating profit, at a 33.1% margin ; the North American entity carried 77.4% of revenue and generated 24.4% of the profit, at 2.78%. The profit is booked where the cars are built, which is the accounting signature of an exporter — but the analytical consequence is that the 9.97% group margin of FY March 2024 was a transfer margin, not an operating one. The entity that sells is not the entity that earns.
Read the two entities separately and the real economics appear. The North American margin has averaged 4.69% over five years with a range of 3.23 points, and it has never once exceeded 6.01% — including at the top of the weak-yen cycle. That is the true ceiling of Subaru's commercial model with the currency stripped out. All of the group's apparent volatility — a 38.4-point swing in the Japanese entity margin — lives in the entity that barely sells anything, because that entity is the accounting receptacle for the currency, the tariff and the one-off charges. A model that forecasts the consolidated margin is forecasting a translation variable.
The critical cost is double, and the second half is the one nobody had flagged. The tariff explains 63% of the deterioration in cost of goods ; the other 37% is a unit-cost drift that ran ahead of price — cost per unit up 8.22% against price up 6.68% ex-tariff, a 1.54-point squeeze that has nothing to do with trade policy and everything to do with fixed-cost absorption on 12.8% less domestic production. On top of that sits a materials drag the company itself sizes at ¥130bn for the year ahead — precious metals, Middle East freight — but flags as a conservative assumption ; the sell-side models ¥75–80bn, which would leave the guidance with a 35% cushion. The debate has quietly inverted : the risk is no longer that materials break the guidance, it is that they were over-provisioned and the guidance is beaten.
The cash bridge is the part that protects the downside and the part that is quietly deteriorating at once. Reported FCF/OP of 3.10x is an artefact of the collapsed denominator, not a conversion strength — the operating profit fell to ¥40bn while depreciation held the cash flow up. Underneath, the cash conversion cycle has lengthened 40.4 days over the decade, from −0.6 to +39.9, tying up roughly ¥530bn on inventory that grew 188% against revenue up 48%. Set against that drift is a balance sheet doing a great deal of quiet work : ¥728bn of surplus cash, a 50.6% equity ratio, no dilution in eleven years. At the 2.79% 10-year JGB that cash earns ¥36.4bn a year — 1.98% of the market cap, and in a nil-AOP year more than the operating asset itself produces. The balance sheet is the part of Subaru that still earns a return ; normalised, the factory does not.
This is the pillar that decides the case, and it is the weakest in the bucket, because it determines whether the industrial asset is worth anything above its replacement value — and normalised, it is not. Return on capital ex-cash is 2.33% against a reconstructed WACC of 6.98–8.69%, a spread of −4.65 to −6.36 points. This is not a tariff accident : it had already fallen to 4.94% in FY March 2022, before any trade shock. Invested capital ex-cash has multiplied 2.8x over the decade for a NOPAT divided by 11.8. The second finding is the harder one because it is exogenous to nothing : ex-tariff, unit cost grew 8.22% against price at 6.68%. A model whose costs drift faster than its prices with no shock to blame does not create value in any weather.
The second cardinal is the balance sheet, because it is the floor under the downside and the only mechanism of per-share value the file contains. Shareholder yield is 7.62% trailing and 12.67% prospective ; the share count has fallen 8.3% over the decade and 6.7% in the last three years alone ; the buyback of 8% of the capital is executing at 18.2% of its cap on 22% of its calendar, and below book, which is accretive to BVPS. The equity ratio is 50.6% with no issuance in eleven years. The limit is that the pillar cannot rescue the model : a capital return financed on the cash stock — the distribution runs at 188% of FCF — creates value only while the underlying asset does not destroy it faster. And the FY March 2021 cut sits on the record as proof the distribution is an adjustment variable.
Resilient but single-market : North American units fell only 3.3% against domestic production down 12.8%, so the constraint was supply. But 79.0% of units sit on one market with no recurring revenue, and the resilience measured in wholesale is not final demand — retail was down 5.9% on 47 days of stock.
The symmetrical-AWD boxer niche is not quickly replicated and holds a loyal US base at below-market incentives. But the North American entity margin is capped at 6.01%, R&D is sub-scale at 3.5% of revenue, and there is no captive, no motorcycle, no aftersales — none of the recurring pockets the bucket rewards.
The FY March 2026 cost discipline is genuine — SG&A cut ¥47bn, ten sales subsidiaries merged. The decade record is not : no US localisation during the yen rent, a 61% dividend cut on ¥692bn of cash, and a ¥57.8bn BEV impairment followed by the programme's postponement.
A lower-tier profile, with the two points of dispersion sitting exactly between the model that does not earn and the governance that returns cash. The grade is consistent with the classification — a trap/illusion resting on a balance sheet — and with the market read : nothing in the grid earns a premium, and the cash floor is what keeps a nil-earnings carmaker off the short list. Above Nissan (8.5/25), below Toyota (17.5/25), mid-way in a bucket where the best grids carry the smallest asymmetries.
Is the normalised earnings power recovering, or structurally nil ?
The North American margin : a depressed floor, or the true ceiling ?
The origination data shows the commercial entity earning 2.78% in FY March 2025, on a five-year average of 4.69% and a maximum of 6.01% reached at the very top of the weak-yen cycle. If 2.78% is a depressed floor, the consensus recovery is possible ; if 4.69% is the real level and 6.01% the ceiling, then no yen and no tariff resolution reconstructs a consolidated 8–10% margin, because the transfer spread that used to supply it is gone. The bipolarity — a stable, low-margin US distributor and a volatile, high-margin Japanese exporter — means the group has no operating margin of its own to recover to.
Is the net cash distributable, or working capital in disguise ?
The surplus has been read three ways — ¥1,333bn published, ¥383bn after a strict working-capital deduction, and ¥728bn once the provisions are treated by nature. On the ¥728bn reading the cash is 39.6% of the cap and the buyback is its realisation mechanism, executing in rhythm at ¥27.2bn. But the prospective 12.67% yield is covered only 57% by normalised FCF : the programme consumes the stock, taking surplus cash toward ¥605bn by year-end. The floor exists ; it erodes as it is used, and it is good for one or two turns, not structurally.
At ¥2,611.50 the market pays ¥1,113bn for an operating net asset of ¥2,025bn — 55.0% of book — after crediting the surplus cash in full. On a nil-AOP asset that 45% discount is coherent, which is the definition of a stock at fair value : neither the excess nor the deficit that would make it a trade. The metric is P/B, not P/E, because the sector rewards book compounding and looks through peak-FX margin, and because on a name in net cash the P/E is an artefact — it went from 5.77x to 19.79x in FY March 2026 while the shares fell, as the EPS collapsed faster than the price. What the cash floor and the buyback do is cap the downside without creating upside : the value is the currency option, and at spot the option is at par.
The yen strengthens to 120 and the tariff reverts to 27.5% on a margin base already at zero, turning the normalised result to −¥102.4bn ; incentives drift to $3,000, confirming the niche is eroding. The buyback is suspended and the dividend cut a second time — the FY March 2021 precedent stands. P/B compresses to 0.50x, 21.8% below the decade-low close of 0.639x. The floor holds at ¥1,953 because the surplus cash plus 32% of the operating book underpin it, on a 50.6% equity ratio with three years of Bear-rate cover. A timing disappointment, reversible.
The yen normalises to the pod's 130 and the tariff settles at 15%. The company hits its normalised break-even exactly — it neither earns nor loses in economic terms, and all of the value is the balance sheet and the progressive return of the surplus cash. The market does not re-rate, per the sector rule, but does not de-rate further either, the 45% discount on the operating book being already coherent with a nil rent. P/B of 0.70x on the ¥3,905 pro-forma BVPS ; the operating asset valued at 59.2% of book. Total return with the 4.44% dividend : +9.1%.
The yen holds at 145, restoring a ¥172.6bn normalised result and a 6.45x implicit multiple on today's operating value ; the materials drag prints at the sell-side ¥77.5bn rather than the company's ¥130bn, freeing ¥52.5bn ; and the buyback executes in full while absorbing the 8.55% of dissolvable cross-holdings, giving accretion without a placement overhang. P/B of 0.95x — above the 5-year average, still below book. The path needs the currency to hold and the programme renewed ; neither is operational improvement.
| KPI | Latest value | Status | What it tells us |
|---|---|---|---|
| Realised USD/JPY · H1 to 30 Sep 2026 | prints Nov 2026 | Cardinal | The variable that governs everything. Above ¥145 restores a ¥172.6bn normalised result and forces a switch to long ; below ¥130 turns it negative and forces a short. Each 5 yen shifts fair value ~15%. |
| Quarterly tariff charge · Q1 FY mar 2027 | prints Aug 2026 | Trigger | A charge above ¥30bn confirms the ¥160bn annualised run-rate at 15%. The first hard read on whether the reconstructed run-rate holds ; back-test due August 2026. |
| US incentives per unit | $2,868 Jun 2026 | Watch | Up 63.9% in five quarters, 46.7% above the sector proxy. Rising incentives on 47 days of stock with retail down 5.9% is the channel-stuffing red flag ; the pricing power of H1 reversed in H2. |
| North American origination margin | 2.78% FY mar 2025 | Watch | The true commercial ceiling, capped at 6.01% over five years. Above 6.5% on FY March 2027 breaks the ceiling reading ; holding near 4–5% confirms the model has no consolidated margin to recover to. |
| Gross margin · Q2 FY mar 2027 | prints Nov 2026 | Watch | Below 16% would confirm the materials drag as a structural cost regime ; above 18% would confirm it is cyclical and the ¥130bn company assumption over-provisioned. |
| Buyback execution | ¥27.2bn / ¥150bn | Trigger | 18.2% executed on 22% of the calendar. Execution below ¥100bn at 31 December 2026 invalidates the distributable-cash reading ; the linear ¥93bn trajectory puts that within reach. |
| P/B (close) | 0.669x | Reference | 4.7% above the decade-low close of 0.639x, 20.4% below the 5-year average of 0.84x. A close below 0.60x (price ¥2,343) means the market is de-rating the balance-sheet floor itself. |
The case turns long if the currency holds. A realised half-year yen above ¥145 to 30 September 2026, confirmed at the November print, restores a ¥172.6bn normalised result and makes the ¥3,710 bull the central case — the watchlist would convert to a long on the currency alone, without any operational improvement. This is observable on a dated calendar and is not signalled today.
The case turns short if the yen breaks the other way. A realised average below ¥130 turns the normalised earnings power negative and, combined with the 12.0/25 grid, the sub-WACC return on capital and the 63.9% incentive drift, would justify a short — were it not for the 39.6% surplus cash and the 12.67% prospective yield, which make the carry cost of the position exceed its expected return. The short bias is documented but un-exploitable at this level ; a close below 0.60x book would reopen it.
The allocation risk is the one to watch most carefully, because the company has made a version of it before. A restarted internal electrification programme committing more than ¥300bn, on a 2.33% return on capital ex-cash, or a second dividend cut outside any solvency constraint, would convert the timing disappointment into a permanent impairment — the 45% discount on the operating book would extend to the whole balance sheet, and the floor would fall toward ¥1,500 or below. Currently not signalled ; both force a return to R1 or a full 2b build.
The information provided on this website is for informational and educational purposes only and should not be construed as financial, investment, legal, or tax advice. All content reflects the personal opinions, interpretations, and analyses of the author at the time of writing and is subject to change without notice. Nothing contained herein constitutes, or should be interpreted as, a recommendation, solicitation, or offer to buy or sell any securities, financial instruments, or other investment products. The author is not a licensed financial advisor, broker, or investment professional. Any references to specific assets, markets, or strategies are illustrative in nature and do not constitute personalized investment advice. Investing in financial markets involves risk, including the potential loss of capital. Past performance is not indicative of future results. Readers are solely responsible for their own investment decisions and should conduct their own independent research and due diligence before making any financial commitments. You are strongly encouraged to consult with a qualified financial advisor, legal professional, or other relevant specialist before making any investment or financial decisions. By accessing and using this blog, you agree that the author shall not be held liable for any direct or indirect losses, damages, or consequences arising from the use of, or reliance on, the information presented herein. All content is provided "as is" without any warranties of completeness, accuracy, or reliability.