Mitsubishi Corp.: The share price assumes an ROE target not yet reached
Fair value ¥3,114 vs. price ¥5,148: the market pays for a permanent 12% ROE, the company delivered 8.5%. Quality yes, return no.
In March 2020 a Mitsubishi share cost 0.65 times book value, with a return on equity of 9.8% in the fiscal year just ended. In March 2026 it cost 2.06 times, with a return on equity of 8.5% [1]. The valuation has more than tripled, while earning power has slipped slightly and sits at the cost of capital.
A note on dates: Mitsubishi names its fiscal years by their start. Fiscal 2025 ended on 31 March 2026; the current fiscal 2026 ends on 31 March 2027.
Mitsubishi owns stakes in around 220 companies, from copper mines to supermarkets, and earns roughly what the money invested is allowed to cost. What is left over the group reliably returns to shareholders: the dividend has risen for years and the share count keeps shrinking. But the share now costs about twice its book value, as if the 12% return-on-equity target had already been reached for good. I calculate a fair value of ¥3,114 per share and a return of 6.5% a year. Rating: Reduce.
For the first half Mitsubishi reports net income of ¥619.7bn (+74%) and raises its full-year forecast to ¥1.2tn, adjusted to ¥930bn [6], [8]. It adds a buyback of up to ¥1tn [7] and an ROE forecast of 12.7% [8]. The share rose 7.9% to ¥5,148 [9]. My fair value barely changes; the gap to the price widens.
1. The business model: a portfolio manager with its own balance sheet
Mitsubishi is no longer a trader but an investment house. The CEO describes his role as that of a portfolio manager who directs capital and people to where the group creates the most value [1]. The money comes from three sources: cash flows of the holdings, sales of mature holdings, and debt on the terms of an A rating [1].
No segment contributes more than a third of profit, and none earns more than 4.1% on its assets:
| Segment | Net income FY2025 (¥bn) | Forecast FY2026 (¥bn) | ROA FY2025 |
|---|---|---|---|
| Energy & Power Solution | 210.0 | 383.0 | 3.5% |
| Mineral Resources | 215.2 | 230.0 | 4.1% |
| Smart-Life Creation | 98.4 | 130.0 | 3.7% |
| Urban Development & Infrastructure | 87.0 | 90.0 | 3.9% |
| Mobility | 60.4 | 100.0 | 3.2% |
| Food Industry | 86.7 | 65.0 | 4.0% |
| Materials Solution | 29.0 | 43.0 | 1.5% |
| Group | 800.5 | 1,200.0 | 4.8% |
Sources: FY2025 and ROA [1], pp. 51–57; forecast of 4 November 2026 [8], p. 26. The difference to the segment total runs through “Other”.
In other words: the profit jump comes mainly from energy and resources, i.e. from prices Mitsubishi does not set.
2. Unit economics: what a holding earns
Mitsubishi's economic unit is the individual holding. Its economics are simple: a thin return on assets, leverage through debt, and an extra gain on sale.
- Return on assets: 1.5% to 4.1% per segment, 4.8% for the group [1].
- Leverage: ¥24.2tn of assets on ¥9.4tn of equity give an 8.5% ROE [1].
- Recycling: almost 30% of fully divested holdings returned more than five times the investment [1].
- What reaches shareholders: about ¥572bn a year (calculation below).
| Step | Value (¥bn) | Source |
|---|---|---|
| Underlying operating CF FY2025 | 1,048.1 | [1] |
| − Sustaining capex (¥1.3tn ÷ 3 years) | −433.3 | [1], p. 26 |
| × (1 − minority share 6.9%) | ×0.931 | [6] |
| Owner cash flow | 572.4 | [own calculation] |
Underlying operating CF is operating cash flow without working-capital swings. The minority share is, as a simplification, the minorities' portion of equity at 30 September 2026: 724.5 ÷ (724.5 + 9,795.6).
3. Capital returns: the dividend rises, the share count shrinks
Mitsubishi returns a large and growing share of its cash flow to shareholders. Next to the business model, this is the most important part of the story for shareholders.
| Element | Status | Source |
|---|---|---|
| Dividend per share | ¥26.67 (FY2016) → ¥125 (FY2026e), +16.7% p.a. | [1], [6] |
| Dividend in the crisis year FY2020 | slightly raised to ¥44.67 on EPS of only ¥38.95 | [1] |
| Shares outstanding | 4,760.9m (March 2019) → 3,663.4m (Sept. 2026), −23% | [1], [6] |
| Buybacks | ¥1tn each in FY2025 and FY2026; the shares are cancelled | [4], [7] |
| Total payout | 176% (FY2025) and 121% (FY2026e) of net income | [8] |
Dividend growth formula: (125 ÷ 26.67)^(1/10) − 1. The policy is called a “progressive dividend”: the yardstick is the dividend per share, which is meant to rise with earning power and not fall [1], [8].
What the buyback does per share. Because the repurchased shares are cancelled, earnings per share rise permanently. The interest on the money spent has to be deducted:
| Step | Value |
|---|---|
| ¥1tn ÷ price ¥5,148 | 194m shares = 5.3% of the total |
| EPS, from fewer shares alone | +5.6% |
| After-tax interest cost (3.9% rate, 20% tax) | about ¥31bn a year |
| EPS, net | about +2.9% |
Formula: (1,200 − 31) ÷ 1,200 ÷ (1 − 5.3%) − 1. Interest: first-half finance costs of ¥114.5bn [6], annualised over average gross interest-bearing debt of ¥5,895bn [8]; first-half tax rate [6] [own rough estimate].
In other words: the buyback raises EPS because the share's earnings yield (6.4%) exceeds the after-tax interest cost (3.1%). Whether it also raises value per share depends on the price. If Mitsubishi buys above intrinsic value, remaining shareholders hand value to those selling. In my base case value per share falls arithmetically by 4%, from ¥2,980 to ¥2,858 [own calculation, excluding interest]. If the price were at book value, the same buyback would clearly create value.
The Japanese background. The payout policy is not a solo effort but part of a restructuring of Japan's capital market. Three developments interlock:
- Pressure from the exchange. Since March 2023 the Tokyo Stock Exchange has asked listed companies for management conscious of the cost of capital and the share price, with published improvement plans [11]. In July 2022, 43% of TOPIX 500 companies traded below book value [11]; Mitsubishi stood at 0.84 in March 2023 [1]. The CFO names the reforms as one reason for the higher valuation [1]. Japanese companies bought back a record ¥22.32tn of shares in fiscal 2025 [13].
- Unwinding cross-shareholdings. The Corporate Governance Code revised in 2026 pushes companies to reduce strategic shareholdings [1]. Mitsubishi still holds 32 such positions worth about ¥0.8tn, down from 128 in FY2019 [1]. Every sale frees capital for buybacks or investment.
- Wages and interest rates. Japan is leaving deflation. Unions won wage increases above 5% for the third year in a row in 2026 (5.01%) [10], and the central bank holds its policy rate at 1.00%, the highest since 1995 [12]. For Mitsubishi this cuts both ways: in its domestic business, for example Lawson, the group itself names labour shortages and inflation-driven weaker consumption as risks [1]. And finance costs rose in the first half from ¥81.8bn to ¥114.5bn, mainly because of higher borrowings [6]. Rising rates also raise the cost of capital I discount with.
Mitsubishi has aligned its strategy with this. Long-term executive pay depends on total shareholder return relative to the TOPIX, and the bonus falls by 20% if ROE stays below 12% [1]. By FY2027 the group plans at least ¥2.0tn of buybacks and ¥1.5tn of dividends [8]. A signal for afterwards: free capital is to go primarily into growth investments [8]. Buybacks of this size are therefore not the default. The CFO also admits that the dividend yield has become less competitive relative to the TOPIX after the share price rise [1]; today it is 2.4%.
“The buyback lifts earnings per share. It lifts value per share only if the price is right.”
4. The moat is access, not pricing power
Mitsubishi has a real advantage, but it works when buying, not when selling. The group invests its own money without a fund life and can hold for decades and add capital [1]. That makes it the partner of choice for large projects such as Copper World in Arizona or the gas business in Haynesville [1].
| Question | Assessment |
|---|---|
| Source | Cheap capital (A rating, own balance sheet) and a network of ~220 holdings open access to large projects [1] |
| Durability | No rule-setter can abolish it. But sovereign funds, private equity and the other trading houses also have cheap capital. |
| Pricing power | None at the core: copper, coal and gas trade at world prices. For FY2027 management itself expects about ¥140bn less tailwind from commodity prices [8]. |
| Direction | Stable. Early indicator: whether Mitsubishi wins auctions without paying above its cost of capital |
In other words: the moat protects access to good projects, not the margin afterwards.
5. Growth and ROE: returns at the cost of capital
What matters for the valuation is what Mitsubishi earns on additional retained capital. Between the averages of FY2016–2018 and FY2023–2025, net income rose by ¥374.7bn and equity by ¥3,969.1bn [1]. That is an incremental return of 9.4% (374.7 ÷ 3,969.1) [own rough estimate], against a cost of capital of 8% to 10% according to the company [1]. Growth has therefore created hardly any value.
Haynesville fits the picture. The largest acquisition in the company's history is expected to earn about ¥50–60bn on a proportional purchase price of about ¥600bn, i.e. 8% to 10% [1] [own calculation]. In the first half the gas business was among the largest drags, and the gas price assumption fell from US$3.90 to US$3.10/MMBtu [4], [8].
How permanent is the 12.7% ROE? The new forecast contains special effects:
| FY2026 ROE, read as … | Calculation | ROE |
|---|---|---|
| Company figure (incl. buyback) | [8] | 12.7% |
| Adjusted for ¥270bn of divestment gains and one-time items | 930 ÷ 9,795.6 | 9.5% |
| Adjusted and without commodity tailwind | (930 − 140) ÷ 9,795.6 | 8.1% |
| Ten-year average FY2016–2025 | [1] | 10.5% |
Sources: forecast and commodity headwind [8], equity at 30 September [6].
Part of the gap is legitimate. Divestment gains are part of Mitsubishi's model, with divestitures of at least ¥2.6tn planned by FY2027 [8]. And the buyback raises ROE not only arithmetically through smaller equity but permanently through higher earnings per share. History argues against extrapolating: in the commodity years FY2021 and FY2022 ROE was 15.0% and 15.8%, three years later 8.5% [1].
And if the transformation succeeds? Mitsubishi wants a more resilient portfolio and explicitly plans from 2027 without commodity tailwinds [8]. The goal is not new: the strategy for 2016 to 2018 already aimed to rebalance resource and non-resource businesses [1], p. 61. So far little of this shows. Energy & Power and Mineral Resources delivered about 54% of segment earnings in FY2025, and 59% in the new FY2026 forecast [1], [8] [own calculation]. The largest new investments also go into gas, copper and LNG. And the more stable segments do not earn more on their assets: Food Industry 4.0%, Smart-Life Creation 3.7%, Mineral Resources 4.1% [1].
If the restructuring succeeds nonetheless, it works mainly through risk: a less cyclical portfolio lowers the cost of capital and thus justifies a higher multiple, even at the same ROE.
| Scenario | ROE | Cost of capital | Growth | Fair P/B | Value per share | vs. price |
|---|---|---|---|---|---|---|
| Assumed today | 12% | 9% | 3% | 1.50x | ¥4,011 | −22% |
| Transformation succeeds | 12% | 8% | 3% | 1.80x | ¥4,813 | −7% |
| … with more growth | 12% | 8% | 4% | 2.00x | ¥5,348 | +4% |
Fair P/B = (ROE − g) ÷ (k − g), multiplied by book value of ¥2,674 per share [6]. The 8% cost of capital is my assumption for a clearly more stable portfolio, at the low end of the company's estimate [1].
In other words: today's price is justified if Mitsubishi holds a 12% ROE through the whole commodity cycle and the portfolio becomes stable enough for the cost of capital to fall to about 8%. That is the real bull case. It requires a restructuring announced ten years ago to show measurable results now, while new money flows mainly into gas and copper.
6. Valuation: ¥3,114 is fair, the price requires peak ROE forever
I value the company with an FCFE model that allows growth only through reinvestment: distributable is E × (1 − g ÷ ROIIC). The discount rate is 9%, the midpoint of the company's own cost-of-equity estimate [1].
Bear case (30%)
¥2,230
- Earnings fall back to ¥800.5bn (FY2025)
- Incremental return 7%, growth 2%
Base case (50%)
¥2,980
- ¥1.2tn FY2027 target with 0.8x haircut: ¥960bn
- Incremental return 9.4%, growth 3%
Bull case (20%)
¥4,777
- ¥1.2tn target fully reached
- Incremental return 12% (ROE target), growth 5%
| Scenario | E (¥bn) | × (1 − g ÷ ROIIC) | = FCFE | ÷ (k − g) | per share | Weight |
|---|---|---|---|---|---|---|
| Bear | 800.5 | 0.714 | 571.8 | 7% | ¥2,230 | 30% |
| Base | 960.0 | 0.682 | 654.9 | 6% | ¥2,980 | 50% |
| Bull | 1,200.0 | 0.583 | 700.0 | 4% | ¥4,777 | 20% |
| Weighted | ¥3,114 | 100% | ||||
Formula: E × (1 − g ÷ ROIIC) ÷ (k − g) ÷ 3,663.4m shares [6]; k = 9%. Weighted: 0.3 × 2,230 + 0.5 × 2,980 + 0.2 × 4,777.
Return at today's price. At a constant multiple a shareholder earns the FCFE yield plus growth: 654.9 ÷ 18,859 + 3% = 6.5% a year (market cap: ¥5,148 [9] × 3,663.4m shares). My 15% annual hurdle is clearly missed.
Does the higher ROE justify a higher multiple? Yes; that is pure mechanics. The fair price-to-book ratio is (ROE − g) ÷ (k − g):
| Permanent ROE / growth | Fair P/B | Value per share | vs. price |
|---|---|---|---|
| 9.0% / 3% (= cost of capital) | 1.00x | ¥2,674 | −48% |
| 10.5% / 3% (ten-year average) | 1.25x | ¥3,334 | −35% |
| 12.0% / 3% (target) | 1.50x | ¥4,011 | −22% |
| 12.0% / 5% (target, more growth) | 1.75x | ¥4,679 | −9% |
| 12.7% / 5% (forecast) | 1.93x | ¥5,147 | 0% |
Example, last row: (12.7% − 5%) ÷ (9% − 5%) = 1.925; × book value of ¥2,674 per share [6] = ¥5,147.
In other words: the price of ¥5,148 is fair exactly if Mitsubishi holds the forecast ROE of 12.7% forever while growing 5% a year. I consider a higher P/B than in the past justified, more like 1.25x to 1.5x instead of 0.65x to 1.0x. Today's 1.93x, by contrast, assumes that an ROE from a year of high commodity prices and divestment gains becomes the norm. Even if I take the ¥1.2tn as sustainable without a haircut, I get ¥3,724 per share, 28% below the price.
Floor and payback. If Mitsubishi earns only its cost of capital, book value is fair: ¥2,674 per share (¥9,795.6bn ÷ 3,663.4m shares, at 30 September [6]), 48% below the price. The purchase price of ¥18,859bn takes 33 years to come back through the owner cash flow of ¥572.4bn, and 21 years on the new cash flow forecast of ¥1.38tn [6] [own calculation].
“The higher ROE justifies a higher multiple. But the price already pays for the peak ROE as a permanent state.”
7. Checklist and risks
| Test | Result | Reason |
|---|---|---|
| Return permanently above cost of capital | marginal | Avg. ROE 10.5%, adjusted FY2026 9.5%, cost of capital 8–10% |
| Understandable moat | yes, narrow | Access to projects, no pricing power |
| 15% annual hurdle | no | 6.5% at today's price |
| Growth above cost of capital | neutral | Incremental return 9.4% vs. 9% |
| Debt | yellow | Net interest-bearing debt ¥4,431.8bn [6] = 7.7 years of owner cash flow; net D/E 0.42x, ceiling 0.6x [8] |
| Liquidity, balance sheet, cash | green | Current ratio 1.49; goodwill and intangibles 8.2% of equity; cash flow above net income [1], [6] |
| Management | yellow | Bonus tied to cash-flow size, 20% ROE penalty, 500% base-salary holding requirement [1] |
The debt is manageable for an A-rated investment house because it is spread across many individual assets. But it is rising: from ¥3,888bn in March to ¥4,431.8bn in September, mainly because of Haynesville [6]. The buyback will raise it further unless divestments such as the partial sale in LNG Canada for about ¥250bn [8] fund it.
- Reliable, growing capital returns
- ROE forecast 12.7%; the target is expected a year early [8]
- LNG Canada expansion and Copper World approved, contribution from FY2028 [8]
- Broad diversification across seven segments
- Valuation: P/B of 1.93x above every historical range
- Commodity cycle: ¥140bn headwind expected for 2027 [8]
- Haynesville: ¥3.4bn of net income per 10 US¢ of gas price [8]
- Rising interest rates and wages in Japan [6], [10], [12]
- … Mitsubishi reaches an ROE of at least 12% in fiscal 2027 and adjusted net income excluding divestment gains exceeds ¥1tn. As of November: management is planning exactly this direction [8].
- … the price falls below my fair value of about ¥3,100. Then I review the rating upwards.
- … Haynesville has to be written down or the gas price stays far below US$3.90/MMBtu. As of November: assumption US$3.10, first half US$2.91 [8], an early warning.
What I watch
- Adjusted earnings without commodity tailwinds. Will at least ¥1.2tn of group net income be achieved in 2027 despite a ¥140bn headwind [8]?
- Buyback price. Will Mitsubishi keep buying back on a large scale well above book value, or will free capital go into growth as announced [8]?
- Haynesville. Restructuring, partial sale down to 75%, and the gas price [1], [8].
- Commodity share of earnings. If the share of Energy & Power and Mineral Resources falls below today's 59% without ROE falling, the transformation is working. Then I will revisit the cost of capital and the rating.
In brief
In September Mitsubishi announced it would raise its stake in the Philippine Ayala Corporation to 15% for about ¥120bn [8]. The Middle East buffer has been moved into the segment forecasts; according to the company the burden stays within the original assumption [8].
Conclusion
Mitsubishi is a solid investment house with good access to large projects and a capital-return policy that fits the restructuring of Japan's capital market. The buybacks lift earnings per share permanently. But a good company is not automatically a good stock: at ¥5,148 I pay 1.93 times book value and expect 6.5% a year. The higher ROE deserves a higher multiple than in the past, more like 1.25x to 1.5x, but not the peak ROE as a permanent state. My fair value is ¥3,114, the floor is book value at ¥2,674. The gap would already point to “Sell” on my scale. I stay with “Reduce” because the ROE trend is right and the gap shrinks to 9% to 22% should Mitsubishi hold 12% through the cycle. For an existing position this means: reduce the weighting, do not add. Rating: Reduce.
- Cost of capital 9%: midpoint of the company's 8–10% estimate [1]. Rising Japanese interest rates could raise it.
- Sustaining capex spread evenly over three years; Haynesville may require more.
- Scenario parameters, weights and the interest rate in the buyback effect are my own assumptions.
- I tested the higher guidance and a successful transformation through variants, not through new scenarios.
The accompanying Excel model contains assumptions, scenarios, DCF, P/B check, buyback effect, sensitivities and the segment cross-check.
Update log
- 04.11.2026 (3) — Introduction corrected: valuation and return on equity now compared for the same dates (March 2020: 9.8%, March 2026: 8.5%); the previous comparison mixed two periods.
- 04.11.2026 (2) — Added: test of whether the shift to more stable businesses justifies a higher ROE or a lower cost of capital (table in section 5); commodity share as fourth watched variable. Fair value and rating unchanged.
- 04.11.2026 — Revision: article condensed; new section on capital returns with the Japanese background (exchange reform, cross-shareholdings, wages and interest rates) and the buyback's effect on earnings per share (+2.9% net). Corrected: because shares are cancelled, the buyback raises earnings per share permanently; the earlier wording that it only raises ROE through smaller equity was incomplete. Q2 data from the flash report [6] and presentation [8]; P/B check added. Price ¥5,148 (previously ¥4,715), fair value ¥3,114 (previously ¥3,115), return 6.5% (previously 6.8%), floor ¥2,674 (previously ¥2,631). Rating unchanged.
- 30.10.2026 — First draft (data: Integrated Report 2026 [1], Q1 FY2026 [2], [3], price as of 23 October 2026 [5]).
Sources
- Integrated Report 2026, Mitsubishi Corporation, 2026. CEO and CFO letters, capital allocation, segment data (pp. 51–57), financial history (p. 59), remuneration (p. 47), strategic shareholdings (p. 26), Smart-Life Creation risks (p. 57), Haynesville and Copper World. → https://www.mitsubishicorp.com/jp/en/ir/library/ar/
- Financial Highlights for the Three Months Ended June 30, 2026, Mitsubishi Corporation, 3 August 2026. → https://www.mitsubishicorp.com/jp/en/ir/library/earnings/pdf/202608e.pdf
- FY2026 Q1 Earnings Presentation, Mitsubishi Corporation, 3 August 2026. → https://www.mitsubishicorp.com/jp/en/ir/library/meetings/pdf/260803/20260803e.pdf
- FY2025 Results and FY2026 Forecast, Mitsubishi Corporation, 1 May 2026. Commodity assumptions, FY2025 buyback. → https://www.mitsubishicorp.com/jp/en/ir/library/meetings/pdf/260501/20260501e.pdf
- Share price data Mitsubishi Corp. (TYO: 8058), stockanalysis.com, close 23 October 2026 (first draft).
- Financial Highlights for the Six Months Ended September 30, 2026, Mitsubishi Corporation, 4 November 2026. Half-year result, forecast, dividend, balance sheet and share count at 30 September, finance costs, tax rate. → https://www.mitsubishicorp.com/jp/en/ir/library/earnings/pdf/202611e.pdf
- Notification of Share Repurchase and Cancellation of Treasury Stock, Mitsubishi Corporation, 4 November 2026. Buyback of up to ¥1tn or 420m shares, 5 Nov 2026–31 Jul 2027, cancellation by 31 Aug 2027. → https://www.mitsubishicorp.com/jp/en/news/release/2026/files/mcpr261104ze.pdf
- FY2026 Q2 Earnings Presentation, Mitsubishi Corporation, 4 November 2026. Adjusted net income, ROE forecast, FY2027 outlook, capital allocation, payout ratios, dividend policy, segment forecasts, gas price, net D/E, gross debt, LNG Canada, Copper World, Ayala. → https://www.mitsubishicorp.com/jp/en/ir/library/meetings/pdf/261104/20261104e.pdf
- Market report, 財経日報, 4 November 2026. Closing price ¥5,148 (+7.9%). → https://news.jkn.co.kr/editions/japanese/post/1070086
- Japan Unions Clinch 5.01 Pct Wage Hikes in 2026 Shunto, Jiji Press via nippon.com, 3 July 2026. Final result of the wage round. → https://www.nippon.com/en/news/yjj2026070300852/
- Tokyo Stock Exchange Initiative on Cost of Capital and Stock Price Conscious Management, Harvard Law School Forum on Corporate Governance, 21 October 2025. Exchange initiative since March 2023, share of companies below book value. → https://corpgov.law.harvard.edu/2025/10/21/tokyo-stock-exchange-initiative-on-cost-of-capital-and-stock-price-conscious-management/
- Japan Interest Rate, FocusEconomics, as of the central bank meeting of 31 July 2026. Policy rate 1.00%, highest since 1995. → https://www.focus-economics.com/country-indicator/japan/interest-rate/
- Japan Buybacks Hit Record 22.32 Trillion Yen in Fiscal 2025, MT Newswires citing Nikkei, 8 May 2026. Secondary source. → https://www.itiger.com/news/2633432106