Record Year, Below-Consensus Guide, and a Buyback the CEO Had to Explain Twice: Initiating Mizuho at Hold
Key Takeaways
- Profit attributable to owners of parent reached ¥1,248.6bn for the year ended March 2026, up 41.0% and past ¥1tn for the first time in the group's history. That is 4.8% above the ¥1,191.5bn Street consensus and 110.5% of Mizuho's own ¥1,130bn forecast, with ROE at 11.4% and the expense ratio down 3.0 points to 59.4%.
- Management does not treat the headline as the earnings power. Its own normalized bridge marks FY2025 at roughly ¥1,150bn, some ¥100bn below the print, and normalized growth at 15% rather than 41%. Two items explain most of the gap: net gains related to stocks rose ¥190.8bn on cross-shareholding disposals, and a ¥65.3bn deferred-tax benefit pulled the effective rate to 22.7% from 25.3%.
- Strip out the Markets banking book and the franchise story shrinks. Of the ¥316.8bn increase in consolidated net business profits, ¥104.5bn came from Global Markets' banking book swinging from a ¥3.0bn loss to a ¥101.5bn profit while sales and trading went backwards 1%. Customer Groups grew a genuine ¥201.1bn, but retail and global corporate banking both delivered lower net income than last year on rising credit costs.
- The FY2026 guide of ¥1,300bn sits 0.9% below the ¥1,311.2bn consensus and is the weakest of the three megabanks (MUFG +12.5%, Sumitomo Mitsui +7.4%, Mizuho +4.1%). The buyback opened at ¥100bn against ¥400bn repurchased last year, which the CEO conceded four days later had caused "some surprise."
- The Rakuten Bank capital alliance announced this morning gives Mizuho Bank a 10.52% voting stake for its Rakuten Card holding rather than cash, and points Rakuten's retail deposits at Mizuho's corporate origination. It is a structural answer to the single weakness management named on the call, and it costs nothing today.
- Rating: Initiating at Hold. A bank earning 11.4% on equity at 1.50x book with genuine rate optionality is not expensive, but it guides to less growth than either peer, its two largest customer franchises are earning below cost of capital, and the profit that cleared ¥1tn leaned on a finite stock-sale engine. We would rather own the option after the second-half buyback decision than before it.
Results vs. Consensus
Mizuho reports under Japanese GAAP on a cumulative basis and publishes a full-year forecast that it revises through the year, so the print is judged against three yardsticks: the sell-side poll, the company's own most recent forecast, and the prior year. All three were cleared, and the size of the clearance differs meaningfully depending on which one you pick.
One convention matters before any table. Mizuho's investor deck reports consolidated gross profits and consolidated net business profits including net gains related to ETFs and others of ¥38.3bn, while the same captions in the Tanshin exclude them. Deck ¥3,515.6bn equals Tanshin ¥3,477.2bn plus ¥38.3bn; deck ¥1,461.1bn equals Tanshin ¥1,422.7bn plus the same. Where a figure below carries the ETF gains, it is labelled.
Full year ended March 31, 2026
| Metric | Actual | Consensus / forecast | Beat/Miss | Magnitude |
|---|---|---|---|---|
| Profit attributable to owners of parent | ¥1,248.6bn | ¥1,191.5bn | Beat | +4.8% |
| Profit attributable to owners of parent vs. company forecast | ¥1,248.6bn | ¥1,130.0bn | Beat | 110.5% of forecast, +¥118.6bn |
| Basic EPS | ¥502.92 | ¥453.49 (company forecast) | Beat | +10.8% |
| Ordinary income | ¥9,085.4bn | n/a | +0.6% YoY | from ¥9,030.4bn |
| Ordinary profits | ¥1,573.2bn | n/a | +34.6% YoY | from ¥1,168.1bn |
| Consolidated net business profits (incl. ETF gains) | ¥1,461.1bn | ¥1,350.0bn (company outlook) | Beat | 108% of outlook, +27.6% YoY |
| ROE (earnings on own capital) | 11.4% | n/a | +2.9ppt YoY | from 8.5%; FY27 target of over 10% hit two years early |
| Expense ratio | 59.4% | n/a | (3.0)ppt YoY | G&A ¥2,091.7bn on gross profits ¥3,515.6bn |
| Credit-related costs | ¥133.0bn | n/a | +¥81.4bn YoY | incl. ¥54.7bn of forward-looking reserves |
| Annual dividend per share | ¥145.00 | ¥140.00 prior year | +¥5.00 | in line with the stated policy step |
| CET1 ratio (fully-effective, excl. unrealized gains) | 9.9% | operational range mid 9–10% to mid 10–11% | (0.4)ppt YoY | from 10.3%; within range |
Year-over-year comparison, full year (¥bn)
The income statement is the arbiter here, not the call. Every line below traces to the consolidated statement of income in the Tanshin.
| Line | FY ended Mar 2025 | FY ended Mar 2026 | Change | % |
|---|---|---|---|---|
| Ordinary income | 9,030.4 | 9,085.4 | +55.1 | +0.6% |
| Interest income | 6,000.2 | 5,851.6 | (148.6) | (2.5)% |
| Interest expenses | 4,954.9 | 4,474.5 | (480.4) | (9.7)% |
| Net interest income | 1,045.3 | 1,377.1 | +331.8 | +31.7% |
| of which interest on loans and bills discounted | 2,739.3 | 2,785.7 | +46.3 | +1.7% |
| of which interest and dividends on securities | 861.1 | 929.7 | +68.6 | +8.0% |
| Trust fees | 62.3 | 67.0 | +4.7 | +7.6% |
| Fee and commission income | 1,115.4 | 1,311.9 | +196.5 | +17.6% |
| Fee and commission expenses | 208.6 | 231.5 | +22.9 | +11.0% |
| Net fee and commission income | 906.8 | 1,080.4 | +173.6 | +19.1% |
| Trading income | 1,047.5 | 898.9 | (148.6) | (14.2)% |
| Other operating income | 394.6 | 430.5 | +35.9 | +9.1% |
| Other operating expenses | 536.0 | 372.6 | (163.4) | (30.5)% |
| Other ordinary income | 410.4 | 525.5 | +115.1 | +28.0% |
| General and administrative expenses | 1,840.7 | 2,103.5 | +262.8 | +14.3% |
| Other ordinary expenses | 322.0 | 326.2 | +4.2 | +1.3% |
| of which provision for allowance for loan losses | 36.2 | 105.6 | +69.4 | +192.0% |
| Ordinary profits | 1,168.1 | 1,573.2 | +405.0 | +34.6% |
| Extraordinary gains | 58.4 | 93.8 | +35.5 | +60.8% |
| of which gains on cancellation of employee retirement benefit trust | 12.4 | 69.8 | +57.4 | n/m |
| Extraordinary losses | 36.4 | 44.7 | +8.3 | +22.7% |
| Profit before income taxes | 1,190.1 | 1,622.3 | +432.2 | +36.3% |
| Total income taxes | 301.4 | 368.1 | +66.7 | +22.1% |
| of which deferred | (0.1) | (65.3) | (65.2) | n/m |
| Profit | 888.7 | 1,254.2 | +365.5 | +41.1% |
| Profit attributable to non-controlling interests | 3.2 | 5.5 | +2.3 | +70.8% |
| Profit attributable to owners of parent | 885.4 | 1,248.6 | +363.2 | +41.0% |
Two lines carry the year. Net interest income grew ¥331.8bn, and it grew because funding costs fell faster than asset yields rose: interest income actually declined ¥148.6bn while interest expenses fell ¥480.4bn. That looks strange for a bank levered to a rising policy rate until you see the composition. Almost the entire expense reduction sits in one line, interest on payables under repurchase agreements, down ¥462.9bn as the collateral-financing book repriced and shrank, while the corresponding asset-side line fell only ¥13.1bn. What is left is the banking franchise: interest on loans and bills discounted up ¥46.3bn on a modestly larger balance at a much better spread, and interest and dividends on securities up ¥68.6bn. Net fee and commission income added another ¥173.6bn, a cleaner and more repeatable number.
The March quarter, derived
Mizuho publishes cumulative periods only, so the fourth quarter has to be isolated by subtracting the nine months reported through December from the full year. It is worth doing, and it is worth reading with a warning attached.
| Line (¥bn) | Q4 FY2024 (Jan–Mar 2025) | Q4 FY2025 (Jan–Mar 2026) | Change | % |
|---|---|---|---|---|
| Ordinary income | 1,956.9 | 2,472.0 | +515.2 | +26.3% |
| Net interest income | 308.0 | 386.7 | +78.7 | +25.6% |
| Net fee and commission income | 253.8 | 323.9 | +70.1 | +27.6% |
| Trading income | 224.4 | 241.5 | +17.2 | +7.7% |
| Other operating income | 92.6 | 163.3 | +70.7 | +76.4% |
| Other operating expenses | 240.7 | 232.6 | (8.2) | (3.4)% |
| Other ordinary income | 59.6 | 212.7 | +153.1 | +256.8% |
| General and administrative expenses | 482.9 | 594.3 | +111.4 | +23.1% |
| Ordinary profits | 41.6 | 318.5 | +276.9 | +665.6% |
| Profit before income taxes | 22.8 | 299.1 | +276.2 | n/m |
| Total income taxes | (7.5) | 68.8 | +76.3 | n/m |
| Profit attributable to owners of parent | 30.1 | 228.7 | +198.7 | +661.0% |
| Derived EPS (¥) | 12.56 | 93.71 | +81.15 | n/m |
Quality of the beat
- Revenue: consolidated gross profits of ¥3,515.6bn including ETF gains rose 18.5%, and the composition is defensible. Roughly half is the domestic rate cycle showing up in the loan and deposit margin, and a large slice is fee income, which reached a record ¥1,463.4bn on a group basis. Yen depreciation helped: management sizes it at ¥4.5bn of net business profits per ¥1 of depreciation. The deck also discloses that approximately ¥150bn of losses were realized in the securities portfolio as a forward-looking provision, which means the reported top line understates the underlying by that amount.
- Margins: the expense ratio fell 3.0 points to 59.4% even though G&A rose 14.3% in absolute terms. That is operating leverage, but it is leverage delivered by a gross-profit line that absorbed roughly ¥150bn of realized securities losses. On the customer-group businesses alone, the picture is more mixed: retail still runs a 76.8% expense ratio and global corporate banking's rose 2.3 points to 59.6%.
- EPS: ¥502.92 against ¥350.20, but the effective tax rate fell to 22.7% from 25.3% on a ¥65.3bn deferred-tax benefit, and the share count shrank because the group repurchased ¥400bn of stock during the year. Neither is operating income. Adjust the tax rate back to FY2024's 25.3% and net income would have been roughly ¥1,206bn, still a beat on the ¥1,191.5bn consensus but a 1.2% one rather than a 4.8% one.
Segment Performance
Mizuho runs five in-house companies. The group aggregate figures below are the company's own management-accounting presentation, with the prior year recalculated on FY2025 rules so the comparison is like for like. This is the table that matters most, because it is where the difference between a good year and a good franchise becomes visible.
| Segment (¥bn) | Gross profits FY25 | YoY | Net business profits FY25 | YoY % | Net income FY25 | YoY % | ROE | Expense ratio |
|---|---|---|---|---|---|---|---|---|
| Customer Groups | 2,654.4 | +316.3 | 1,124.7 | +22% | 878.3 | +14% | 10.8% | n/a |
| Retail & Business Banking (RBC) | 984.6 | +152.4 | 237.5 | +69% | 118.7 | (4)% | 5.9% | 76.8% |
| Corporate & Investment Banking (CIBC) | 739.3 | +102.5 | 499.8 | +23% | 521.2 | +29% | 16.0% | 33.9% |
| Global Corporate & Investment Banking (GCIBC) | 857.0 | +47.6 | 367.7 | +1% | 219.5 | (9)% | 8.0% | 59.6% |
| Asset Management (AMC) | 73.6 | +13.7 | 19.7 | +66% | 18.8 | +1,885% | 15.3% | 62.9% |
| Global Markets (GMC) | 664.9 | +156.2 | 260.0 | +66% | 177.3 | +65% | 7.9% | 60.9% |
| of which Banking | 154.1 | +102.1 | 101.5 | n/m | n/a | n/a | n/a | n/a |
| of which Sales & Trading | 510.7 | +54.1 | 158.5 | (1)% | n/a | n/a | n/a | n/a |
A second table is more revealing than any of the growth rates: how much internal risk capital each business consumes against what it returns.
| Segment | Internal risk capital, avg. (¥bn) | Share of segment risk capital | Net income (¥bn) | Share of group net income | ROE |
|---|---|---|---|---|---|
| RBC | 2,001.0 | 19.3% | 118.7 | 9.5% | 5.9% |
| CIBC | 3,265.5 | 31.4% | 521.2 | 41.7% | 16.0% |
| GCIBC | 2,753.3 | 26.5% | 219.5 | 17.6% | 8.0% |
| GMC | 2,248.9 | 21.6% | 177.3 | 14.2% | 7.9% |
| AMC | 122.7 | 1.2% | 18.8 | 1.5% | 15.3% |
Shares of group net income do not sum to 100% because ¥193.1bn sits outside the five companies in consolidation adjustments and unallocated items.
Retail & Business Banking: the best top line and the worst bottom line
RBC produced the largest absolute gross-profit increase of any customer segment, ¥152.4bn, and the largest proportional increase in net business profits, 69%. Interest income inside the segment rose ¥111.9bn as the deposit franchise finally started paying for itself in a positive-rate world, and non-interest income added ¥40.0bn. The expense ratio improved 7.5 points to 76.8%. Every operating measure moved the right way.
Net income fell 4%, to ¥118.7bn. The entire gap is credit: RBC's credit-related costs swung from a ¥6.7bn reversal in FY2024 to a ¥64.6bn charge in FY2025, a ¥71.3bn deterioration that swallowed three-quarters of the segment's net business profit improvement. Segment ROE fell to 5.9% from 6.3%.
"In mass retail, Gross Profits increased by 11% year-on-year. The number of new accounts opened also grew significantly by 26% in FY25. Furthermore, our app MAU increased by 13%."
— Masahiro Kihara, President & Group CEO
Assessment: RBC consumes 19.3% of the group's internal risk capital and returns 9.5% of its profit at an ROE that is comfortably below any plausible cost of equity for a Japanese bank. Management has responded organizationally rather than financially, appointing a new co-head of the business and naming deposit gathering as the top priority. That is the right diagnosis. It is not yet a plan with a date on it, and the credit-cost swing is a reminder that a mass-retail and SME book leveraged to a rising rate cycle collects the revenue and the defaults in the same year.
Corporate & Investment Banking: the engine, and the only segment beating its cost of capital by a wide margin
CIBC is where the year was actually earned. Gross profits rose ¥102.5bn on a 33.9% expense ratio, credit costs improved ¥42.7bn to a ¥24.1bn charge, and the segment booked ¥192.2bn of net gains on stocks, which is where most of the cross-shareholding disposals land. Net income rose 29% to ¥521.2bn, 41.7% of the group total on 31.4% of the risk capital. Segment ROE reached 16.0%, up 4.3 points, and it did so while internal risk capital fell ¥216.0bn.
The domestic corporate solutions business grew from ¥264.1bn to ¥330.1bn, with investment banking fees at ¥74.4bn, real estate at ¥79.7bn and credit-related fees at ¥176.0bn. Solution-business income inside CIBC rose ¥54.2bn to ¥226.9bn.
Assessment: this is a genuinely good franchise operating in a genuinely good market, and it is the reason to look at Mizuho at all. The caution is that ¥192.2bn of its net gains on stocks is the disposal programme, not the customer business, and CIBC's ROE without it would look materially different. Management has three more years of cross-shareholding sales budgeted; after that, this segment has to earn 16% on client revenue alone.
Global Corporate & Investment Banking: growth stalled and costs did not
GCIBC grew gross profits ¥47.6bn but G&A rose ¥47.3bn against it, leaving net business profits up ¥2.3bn, or 1%. Credit costs swung ¥35.0bn to a ¥31.3bn charge. Net income fell 9% to ¥219.5bn, ROE dropped a point to 8.0%, and the expense ratio rose 2.3 points to 59.6%, the only segment where it went the wrong way. The group notes a correction to specific credit-related costs from FY2024 in this segment's comparatives.
The underlying business is not weak. Non-interest income rose ¥31.4bn to ¥476.4bn, with the Americas contributing ¥269.0bn (up ¥26.1bn) and investment banking fees within it up ¥17.0bn. Mizuho Americas gross profit reached USD 5.2bn with average monthly revenue of approximately USD 435M and monthly volatility of approximately 20%, against USD 190M and approximately 60% in FY2019.
"We believe there is still significant potential in the U.S. business. In terms of gross profit, we have reached USD 5.2 billion. We have significantly deepened our client relationships and have established a solid product lineup. Through global collaboration, we believe we can grow this business even further."
— Masahiro Kihara, President & Group CEO
Assessment: a quarter of the group's risk capital earning 8% with a rising cost base is the second structural problem on this page. The Greenhill acquisition is producing league-table evidence (second in Japan-involved cross-border M&A, a first for the group; global CIB rank from 17th in FY2022 to 15th in FY2025), so the revenue thesis is intact. The issue is that the cost of building a global CIB is currently arriving faster than the revenue, and the segment absorbed the year's overseas credit deterioration on top.
Global Markets: the entire improvement is the banking book
GMC net business profits rose ¥103.1bn to ¥260.0bn, a 66% increase and the second-largest contributor to the group's growth. Of that, the banking book contributed ¥104.5bn, swinging from a ¥3.0bn loss to a ¥101.5bn profit. Sales and trading net business profits fell ¥1.4bn to ¥158.5bn, a 1% decline, on gross profits that rose ¥54.1bn and a cost base that rose ¥55.5bn.
Assessment: the customer-facing markets business was flat. What improved was the securities portfolio, recovering from a year in which it absorbed rebalancing losses. That improvement is real cash, and the current portfolio is positioned conservatively (JGB average remaining life 0.9 years, foreign bonds 1.2 years, with USD 27.4bn of held-to-maturity foreign bonds built up against the possibility of US rate cuts). But it is a base effect, and the same ¥104.5bn will not repeat. Roughly a third of the group's ¥316.8bn net business profit improvement sits in this one line.
Asset Management: small, and finally working
AMC gross profits rose ¥13.7bn to ¥73.6bn and net income went from ¥0.9bn to ¥18.8bn, an ROE of 15.3% on 1.2% of the group's risk capital. Asset Management One's share of in-house funds ranking in the top quartile of a one-year institutional peer comparison rose from 25% in FY2024 to 52% in FY2025. Group retail assets under management rose 15% and NISA accounts 17%.
Assessment: the numbers are too small to move the group and the base effect on net income is enormous, but the investment-performance statistic is the one that matters, because distribution follows performance with a lag. This is the segment where the Rakuten Securities relationship should eventually show up.
Key Topics & Management Commentary
Overall Management Tone: confident on delivery, unusually candid on shortcomings, and visibly stung by the market's reaction to the capital-return announcement four days earlier. The prepared remarks spent more time on Japan's macro turn and the group's history than on the quarter, which is the posture of a management team that believes the operating question is settled and the re-rating question is not. The one place the tone broke was capital return, where the chief executive volunteered regret for having caused "some surprise" and then restated an unchanged policy twice in the same breath.
1. The record is real, and management will not let you use it as the earnings base
Profit attributable to owners of parent cleared ¥1tn for the first time. Management opened the presentation by discounting it.
"The Key Messages are as you already know. While we recognize that the figure is slightly above our normalized earnings capacity, Profit Attributable to Owners of Parent was JPY 1,248.6 billion for the previous fiscal year."
— Masahiro Kihara, President & Group CEO
The normalized bridge in the deck puts FY2024 at ¥1,000.0bn against ¥885.4bn reported (adding back ¥300bn of forward-looking preparations, net of one-time items) and FY2025 at ¥1,150.0bn against ¥1,248.6bn reported (removing ¥240bn). Normalized growth was therefore 15%, not 41%.
Assessment: a bank that voluntarily marks down its own record is telling you something useful, and the ¥1,150bn is the number to build a model on. It also reframes the FY2026 guide: ¥1,300bn is +4.1% on the print but +13% on the normalized base. That is the strongest argument available for the guidance, and management made it in the first Q&A exchange rather than in the prepared remarks, which suggests it was a response to the Street's reaction rather than a planned message.
2. Where the growth actually came from
Consolidated net business profits including ETF gains rose ¥316.8bn to ¥1,461.1bn. Customer Groups contributed ¥201.1bn and Markets ¥103.1bn. Within Markets, the banking book contributed ¥104.5bn and sales and trading contributed nothing. Within Customer Groups, RBC contributed ¥97.2bn and CIBC ¥93.8bn, while GCIBC contributed ¥2.3bn.
Assessment: two businesses and one base effect produced essentially the entire improvement. The concentration is not fatal, because CIBC's contribution is high-quality and RBC's is the first real evidence that the retail deposit franchise has value in a positive-rate world. But an investor underwriting FY2026 needs to replace ¥104.5bn of banking-book recovery with something, and the guide's answer is another ¥73.1bn of stock gains plus ¥23.0bn of lower credit costs.
3. Net interest income and the domestic margin
Group net interest income rose 31.7% to ¥1,377.1bn. The domestic loan and deposit rate margin at the two banks widened from 0.92% to 1.10%, with returns on loans up 34bp to 1.32% and the cost of deposits up 15bp to 0.21%. Loan spreads improved far more modestly than the base rate: RBC Corporate widened from 0.58% in FY2021 to 0.64% in FY2025, and CIBC from 0.52% to 0.59% across the same five years. The margin story is the policy rate, not pricing power.
Average domestic loan balances rose from ¥56.9tn to ¥57.8tn, with RBC Corporate up ¥1.0tn and RBC Individual down ¥0.3tn as the mortgage book continued to run off. Period-end balances tell a more dramatic story: ¥60.4tn against ¥56.3tn, a ¥4.1tn jump concentrated in CIBC (¥34.6tn period-end against a ¥32.2tn average), which is the bridge financing management later described.
Assessment: the margin expansion is real but it is being shared with depositors faster than the headline sensitivity implies. A 15bp increase in the average cost of deposits against a 25bp increase in the Bank of Japan current account rate over the same window is a 60% aggregate pass-through, which is high by the standards of what Japanese banks assumed when they modelled the exit from negative rates. Management flagged the same concern about its own disclosed sensitivity, which is the honest read.
4. The fee engine reached a record
Group non-interest income rose to ¥1,463.4bn from ¥1,330.2bn, the fifth consecutive annual increase from ¥1,106.4bn in FY2021. The composition: RBC ¥502.6bn (+¥40.0bn), CIBC ¥373.8bn (+¥60.3bn), GCIBC ¥476.4bn (+¥31.4bn), AMC ¥74.7bn (+¥14.2bn). Domestic corporate solutions rose from ¥264.1bn to ¥330.1bn, of which real estate contributed ¥79.7bn (+¥19.7bn) and credit-related fees ¥176.0bn (+¥35.3bn).
"For large corporates, product-related loans have grown significantly, and IB income has shown a CAGR of 24%. For mid-cap companies as well, high-return loans, which we define as loans of JPY 1 billion or more with spreads of 100 basis points or more, have expanded significantly."
— Masahiro Kihara, President & Group CEO
Assessment: this is the highest-quality line in the results and the one least dependent on the rate cycle or the disposal programme. It is also the line most exposed to Japanese corporate-action volumes, which management concedes could stall if the Middle East situation turns corporate sentiment conservative. The credit-related fee component at ¥176.0bn is worth watching separately, because fees earned on arranging credit tend to peak alongside the credit cycle.
5. Credit costs tripled, and next year is guided lower
Credit-related costs rose ¥81.4bn to ¥133.0bn, comprising ¥78.3bn of base cost and ¥54.7bn of newly recorded forward-looking reserves. By segment: RBC ¥64.6bn, CIBC ¥24.1bn, GCIBC ¥31.3bn. The provision for allowance for loan losses inside other ordinary expenses nearly tripled, to ¥105.6bn from ¥36.2bn. Against that, non-performing loans fell to ¥0.92tn from ¥1.04tn and the NPL ratio fell to 0.80% from 0.97%.
The FY2026 outlook assumes credit costs of ¥110.0bn, a ¥23.0bn improvement.
Assessment: the asset quality data and the provisioning move in opposite directions, which is exactly what a forward-looking reserve is supposed to look like. The tension is with the guide. Management booked ¥54.7bn of Middle East reserves in FY2025 and then guided credit costs down for FY2026 while separately saying the plan does not incorporate the Middle East at all. Both can be true if you believe FY2025's reserve was sufficient. If it was not, the ¥110bn assumption is the first thing that breaks.
6. Capital: the ratio fell because the balance sheet grew
The CET1 ratio on a Basel III fully-effective basis excluding net unrealized gains fell to 9.9% from 10.3%. CET1 capital rose 8.6% to ¥9,355.2bn while risk-weighted assets rose 12.8% to ¥93,876.6bn. Management's own bridge attributes +1.3 points to profit, −0.8 points to shareholder return, and −0.9 points to inorganic growth and net RWA change.
The growth was not in loans alone. Loans and bills discounted rose 6.0% to ¥99,753.1bn, but securities rose 24.3% to ¥42,632.5bn and total assets rose ¥18,919.6bn to ¥302,240.0bn.
"Loan demand was unexpectedly strong in the previous fiscal year, leading to an increase in RWA, but we maintained our discipline on return. As shown here, compared to the end of March 2019, RWA increased from JPY 78 trillion to JPY 97 trillion, and RORA improved from 2.4% to 3.6%."
— Masahiro Kihara, President & Group CEO
Assessment: 9.9% sits inside the stated operational range of mid 9–10% to mid 10–11%, which is a comfortable position against its own 9.5% to 10.0% operating range. Gross profit RORA improving from 2.4% to 3.6% across seven years is a credible defence of the RWA growth. The cost is that capital ceased to accrete this year, and the buyback is what absorbs that.
7. Shareholder return, and the apology
The dividend goes to ¥150 from ¥145, consistent with the stated policy of approximately ¥5 per year. The buyback was set at up to ¥100bn and up to 25,000,000 shares, running May 18 to August 31 with all repurchased shares to be cancelled on September 24. FY2025 saw ¥400bn of buybacks and a 60% total payout ratio, against 51% in FY2024.
"Regarding share buybacks, we are beginning initially with JPY 100 billion, since there is still a degree of uncertainty regarding the Middle East conflict. I may have caused some surprise during our earnings announcement, and I regret if that was the case. Of course, as we watch the Middle East situation, we plan to flexibly increase this."
— Masahiro Kihara, President & Group CEO
The policy itself is unchanged: a total payout ratio of 50% or more as a guide. On the ¥1,300bn guide, a 50% total payout implies approximately ¥650bn, against approximately ¥366bn of dividends at ¥150 per share. That leaves roughly ¥284bn of implied buyback capacity, of which the ¥100bn announced represents about 35%.
Assessment: the arithmetic says this is a timing decision, not a policy change, and the chief executive said so twice in a single passage. That he felt the need to is the story. A stock whose re-rating case rests on the durability of a 50%-plus payout cannot afford ambiguity in the announcement, and the market read the ¥100bn headline against the ¥400bn precedent rather than against the policy. Expect a second tranche; the question is whether it arrives at the half-year or later.
8. The FY2026 guide, and what it assumes
Profit attributable to owners of parent is guided to ¥1,300bn, +4.1%. Consolidated net business profits are guided to ¥1,630.0bn (+¥168.8bn), credit costs to ¥110.0bn (+¥23.0bn of improvement) and net gains related to stocks to ¥360.0bn (+¥73.1bn). The assumptions are a Bank of Japan policy rate of 0.75%, Nikkei 225 at ¥57,000 and USD/JPY at 150.
"Our guidance of JPY 1,300 billion does not take the Middle East situation into account and is in this sense “full tilt”. However, if you ask whether there is truly no downside risk, as you all know, it is a fact that we are in a very difficult environment."
— Masahiro Kihara, President & Group CEO
Assessment: "full tilt" is an unusual admission. It means the guide has no macro cushion in it, which cuts both ways: it is not sandbagged, and it has nothing to absorb a shock. Two of the three assumptions are already favourable. The policy rate assumption of 0.75% embeds no further hikes despite an environment in which management itself expects them, and the USD/JPY assumption of 150 sits well below where the currency has been trading, worth roughly ¥25bn of additional net income at ¥3.0bn per yen if spot holds. The offsetting risk is that ¥360bn of the ¥1,860bn ordinary profit guide is stock disposals, which is 19% of the plan sitting in a line item that has nothing to do with banking.
9. The FY2028 target: ROE over 12% without further rate hikes
The medium-term target was raised from over 10% ROE by FY2027 to over 12% by FY2028, with consolidated net business profits of approximately ¥1.8–2.0tn against the previous ¥1.4–1.6tn. Both are set on an assumed policy rate of 0.75% and USD/JPY of 150. The FY2027 target was achieved two years early.
"As for our FY28 target, we aim to achieve an ROE of 12% assuming no interest rate hikes. While this is quite an ambitious target, we view it as a level we absolutely must achieve."
— Masahiro Kihara, President & Group CEO
Assessment: setting the target without rate help is the right way to set it, and it converts every future hike into upside rather than into the plan. The path is back-end loaded, though. On our arithmetic, the ¥1,300bn guide against a book that compounds at anything like this year's rate implies FY2026 ROE of roughly 11.1%, below the 11.4% just delivered, because equity grows faster than the profit guide when the payout ratio is not fully deployed. The entire move to 12% therefore has to happen in FY2027 and FY2028. Management's own framing on a normalized basis (approximately 10.5% in FY2025) makes the target look more achievable, which is presumably why it is framed that way.
10. Cross-shareholdings: the engine that flattered the year
Net gains related to stocks excluding ETF gains reached ¥286.8bn, up ¥190.8bn. Against a three-year target of over ¥350bn of listed-stock reduction between March 2025 and March 2028, FY2025 delivered ¥114.6bn, or ¥152.4bn including sales already accepted, a 45% progress rate in year one. Deemed holdings fell ¥274.4bn against a ¥200bn outlook. Market value of holdings against net assets stands at 30.1%, against a stated goal of under 20%.
Assessment: the disposal programme is running well ahead of plan, which is good governance and a P&L crutch at the same time. Two consequences follow. The FY2026 guide of ¥360bn of stock gains is 28% above the ¥286.8bn just delivered, so the crutch gets heavier before it gets lighter. And at 30.1% of net assets against a sub-20% goal, the reserve is large enough to fund several more years of this, which means the day the group has to earn its ROE without disposal gains is far enough out that no one is discounting it. That is precisely when it is worth thinking about.
11. Rakuten Bank: the deposit problem gets a structural answer
This morning Mizuho Bank and Rakuten Bank announced a capital and business alliance. Mizuho Bank receives 23,559,673 Class A shares of Rakuten Bank through a share delivery in which Rakuten Card and Rakuten Securities Holdings become Rakuten Bank subsidiaries, converts them to common shares, and ends with 10.52% of the voting rights, effective October 1, 2026. The consideration is Mizuho's existing Rakuten Card stake, not cash. The agreement carries a standstill preventing Mizuho from acquiring more, and transfer restrictions preventing it from selling, with a stated purpose of preserving Rakuten Bank's management autonomy.
The commercial substance is that Rakuten Bank will acquire corporate loans and securitized receivables originated by Mizuho Bank on an ongoing basis, with Mizuho supporting Rakuten Bank in building the credit review framework required to hold them. The filing frames it against "intensifying competition for deposits following the return to positive interest rates."
Assessment: this is the most strategically interesting thing in the whole reporting cycle and it was announced the day after the analyst meeting, at which the chief executive had said there was "absolutely no concern" that shareholder returns would be constrained by the Rakuten reorganization. Read together, the sequence is a company that knew a deal was close and chose not to signal it. On the merits it is well designed: Mizuho's structural weakness is a ¥48.5tn retail deposit base and a yen loan-to-deposit ratio around 50%; Rakuten Bank's structural weakness is asset origination. Trading a minority card stake for access to the other side of that mismatch, with no capital outlay and an explicit "no material impact" on the ratio, is a cheap option. The standstill caps the upside as much as it caps the risk, and the ¥73.1bn Rakuten Bank earned last year is not going to move a group guiding to ¥1,300bn.
12. Deposit competition and the ¥120bn sensitivity that management no longer believes
The disclosed sensitivity is approximately ¥120bn of annual pre-tax benefit per 25bp of policy rate, with cumulative FY2026 pre-tax impact from past hikes of approximately ¥345bn, of which approximately ¥90bn is incremental to FY2025. Time deposits now represent 26.5% of the deposit base. Base rates on the deposit book moved to 0.300% for current deposits, 0.400% for one-year time deposits and 0.900% for ten-year time deposits.
"Page 27 shows the impact when rates rise by 25 basis points, which is currently indicated as JPY 120 billion. However, because we may need to pay slightly higher costs for time deposits and other products, I suspect the actual impact might be slightly below JPY 120 billion. We did not revise the figure this time, but we intend to conduct a more precise analysis and revise it in the future."
— Masahiro Kihara, President & Group CEO
Assessment: a chief executive pre-announcing a downward revision to his own headline rate sensitivity is a disclosure worth taking seriously, because it is the single number the equity case for every Japanese bank is built on. The candour is welcome. The consequence is that models carrying ¥120bn per 25bp should carry less, and the ROE sensitivity of 0.6–0.7 points per 25bp should probably carry less too.
Guidance & Outlook
| Metric (¥bn) | FY2025 result | FY2026 outlook | Change | Direction |
|---|---|---|---|---|
| Consolidated net business profits (incl. ETF gains) | 1,461.1 | 1,630.0 | +168.8 | +11.6% |
| Credit-related costs | (133.0) | (110.0) | +23.0 | Improvement assumed |
| Net gains (losses) related to stocks (excl. ETF gains) | 286.8 | 360.0 | +73.1 | +25.5% |
| Ordinary profits | 1,573.1 | 1,860.0 | +286.8 | +18.2% |
| Profit attributable to owners of parent | 1,248.6 | 1,300.0 | +51.3 | +4.1% |
| EPS (¥) | 502.92 | 533.10 | +30.18 | +6.0% |
| Annual dividend per share (¥) | 145.00 | 150.00 | +5.00 | Raised |
| Share buyback | 400.0 | 100.0 initial | n/a | Opened lower |
| Total payout ratio | 60% | not guided; policy is 50% or more | n/a | Maintained |
The gap between the ordinary-profit guide (+18.2%) and the net-income guide (+4.1%) is the tax line normalizing. FY2025's effective rate of 22.7% carried a ¥65.3bn deferred benefit; a return toward the mid-20s absorbs most of the pre-tax improvement. This is the mechanical reason the headline growth rate looks so weak, and it is a better explanation than any narrative about caution.
Street at: the ¥1,300bn guide sits 0.9% below the ¥1,311.2bn consensus for the year ending March 2027. Among the three megabanks reporting the same day, Mizuho guided to +4.1%, Sumitomo Mitsui to +7.4% and MUFG to +12.5%. On a normalized base of ¥1,150bn the same guide represents roughly +13%, which is the framing management chose in the first Q&A answer.
Implied build: starting from the normalized ¥1,150bn, the deck's own waterfall adds approximately ¥160bn from consolidated net business profits and approximately ¥110bn from net gains related to stocks, against a tax and other drag, to reach ¥1,300bn. Roughly 40% of the gross positive contributions in that bridge therefore come from selling equity holdings.
Guidance style: Mizuho set its FY2025 forecast at ¥1,130bn in November and delivered ¥1,248.6bn, a 10.5% overshoot, and delivered 108% of its net business profit outlook. The house pattern is to guide below and beat. Whether that pattern applies to a guide management explicitly describes as "full tilt" is the open question, and it is the reason we would not simply mark the ¥1,300bn up by 10% and call it conservative.
Analyst Q&A Highlights
Mizuho publishes its own summary of the analyst Q&A and does not identify the questioners. Attributions below reflect that convention; the answers are the chief executive's.
Whether the FY2026 outlook shows any growth at all
The first and dominant question of the session went straight at the guide. If the year just closed included roughly ¥150bn of realized securities losses, the questioner reasoned, then removing that alone should lift the base, and ¥1,300bn barely moves off it. Management's answer was to reject the base entirely and substitute its own normalized figure.
Q: "Financial targets — Outlook for FY26."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "Considering that we realized losses in the Bond portfolio of approximately JPY 0.15T in FY25, we understand the view that the FY26 Outlook of 1.30T does not show particularly strong growth. However, both upside and downside factors are being considered here: upside (P/L positive) items such as the absence of loss realization, and downside (P/L negative) items such as last year's one-off income items and tax related items. We see our Net Income (Profit Attributable to Owners of Parent) for FY25 as approximately JPY 1.15T on a normalized basis and, when viewed from that level, the Outlook of JPY 1.30T represents substantial growth and is appropriately ambitious."
— Masahiro Kihara, President & Group CEO
Assessment: the answer is analytically correct and rhetorically defensive. Management is right that the reported base is inflated by one-off income and a tax benefit, and right that ¥1,300bn off ¥1,150bn is 13%. It is also true that the company chose to publish the ¥1,300bn against the reported ¥1,248.6bn on the results day, and the Street reacted to what was published. The normalized argument would have landed better on May 15 than on May 19.
Why the buyback opened at ¥100bn
The second capital question was whether the Rakuten reorganization was quietly constraining the return. Management denied it in unusually absolute terms and pointed at the Middle East instead.
Q: "Shareholder Return in FY26 (specifically share buybacks)."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "Our Shareholder Return Policy remains unchanged: we will continue to use a total payout ratio of 50% or more as a guide. On the other hand, the situation in the Middle East remains uncertain, so we will implement returns flexibly while monitoring the impact on our financial performance and corporate demand for financing. There is absolutely no concern that returns would be constrained by Rakuten Group's Fintech Business reorganization."
— Masahiro Kihara, President & Group CEO
Assessment: the denial was accurate and, twenty-four hours later, incomplete. The Rakuten Bank alliance announced this morning consumes no cash and does not constrain returns, so the letter of the answer holds. But management knew a transaction was imminent and let an analyst's question about Rakuten pass with a denial about capital rather than a signal about strategy. That is defensible disclosure practice and poor investor relations, and it is the second capital-communication misstep in five days.
Risk-weighted asset discipline after a 12.8% increase
With risk-weighted assets up ¥10.7tn and the CET1 ratio down 40bp, the natural question was whether the growth was intentional or a loss of control. The answer identified bridge financing as the driver and reaffirmed an originate-and-distribute response.
Q: "Discipline on RWA allocation."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "The increase in RWAs toward the end of FY25 was largely driven by an increase in bridge financing. Providing financing to customers looking to secure liquidity in light of the Middle East situation has been building since April, but at this point in time, not at particularly notable scale. We are reaffirming, group-wide, our 'origination and distribution' approach—an area that we have a strong track record. That includes using capital markets to rotate, shrink, and replace RWAs."
— Masahiro Kihara, President & Group CEO
Assessment: bridge financing is self-liquidating by construction, which makes this the most reassuring answer of the session if you believe it. The period-end domestic loan balance of ¥60.4tn against a ¥57.8tn average corroborates a year-end surge rather than a structural build. The Rakuten Bank alliance announced today is the same answer expressed as a transaction: a permanent distribution channel for originated assets.
Competition for deposits and what it costs
A recurring theme across the session was whether the domestic deposit franchise can be grown without giving the margin back. Management was explicit that corporate deposits have become a seller's market and that the disclosed rate sensitivity may not survive it.
Q: "Domestic competition for deposit acquisition."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "As for corporate deposits, competition has also clearly intensified. It is increasingly a seller's market, and we are seeing more cases where the bank that is willing to pay up wins the deposit—so costs could rise here. Regarding sensitivity to the BOJ Policy rate, we maintain for the time being a pre-tax benefit of JPY 120B per +25 bps, but if costs for corporate deposits and the like increase, we may need to consider revising that going forward."
— Masahiro Kihara, President & Group CEO
Assessment: the most important answer of the day and the one least covered. Every valuation case for a Japanese megabank runs through a deposit-beta assumption. Management has now said, on the record, twice, that its published sensitivity is probably too high. New account openings grew 26% and the group is running acquisition campaigns, but it is also explicitly declining to buy time deposits at the market clearing price, which is a choice to protect margin over volume.
What management thinks is broken
Asked to name the group's weaknesses, management produced an unusually specific list rather than the usual generalities, and attached a named executive to the largest one.
Q: "Key challenges currently recognized."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "First, in mass retail business (Japan) we need to take deposit balances up another notch—that's a major priority. From FY26, I've appointed Naoshi Inomata as the Head of Domestic Retail Business & Co-Head of RBC and have trusted him with the tasks of laying out all required initiatives, clarifying the timelines to address them, and driving their execution. In Wealth Management & Asset Management (Japan), we recognize we are trailing some of the major securities firms. ... For corporate wholesale business, we see comparatively fewer issues."
— Masahiro Kihara, President & Group CEO
Assessment: the self-assessment maps precisely onto the segment table. RBC earns 5.9% on 19.3% of the risk capital and is named as the priority; wealth and asset management is the second gap; the wholesale businesses, which are where CIBC's 16.0% ROE lives, are described as having comparatively fewer issues. What is missing is a number and a date. "Another notch" is not a deposit target, and no timeline for RBC to reach a cost-of-capital return was offered.
The Americas, and whether more acquisitions are coming
With Greenhill integrated and the US business at USD 5.2bn of gross profit, the question was whether Mizuho intends to buy more. The answer was a clear no for now, with a specific reason attached.
Q: "Outlook for the Americas business and the possibility of inorganic growth."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "Regarding inorganic growth, it is not something we envision in the Americas at this moment. Even though integration of Greenhill has progressed considerably, there may still be areas not fully completed. We think the proper order to follow here is to first complete this PMI, fully generate synergies, confirm that Americas earnings step up a notch, and then—if an opportunity arises—consider further inorganic growth. Also, since we already have CIB capabilities situated in-house, even if we were to bring in another firm's CIB function through investment, functions would likely overlap..."
— Masahiro Kihara, President & Group CEO
Assessment: capital discipline stated plainly, and welcome from a group whose CET1 ratio just fell 40bp. It also implicitly concedes that GCIBC's 8.0% ROE is an execution problem rather than a scale problem, which is the harder of the two to fix but the cheaper. The cross-border M&A league-table result is the evidence that the Greenhill logic works; the segment's cost line is the evidence that it has not yet paid for itself.
The path to a 12% return on equity
The final substantive exchange asked what the group is actually aiming for beyond the FY2028 target, and got a rare glimpse of the internal ambition alongside a candid admission that there is no timeline for it.
Q: "Medium- to long-term ROE outlook."
— Analyst question, FY2025 results presentation, May 19, 2026
A: "As an immediate commitment, we are targeting ROE of 12% or more for FY28, without assuming any further BOJ policy rate hikes. Achieving ROE over 12% would require roughly JPY 1.5T in Net Income, and we are aiming to achieve that as an absolute must across the Group. ... As a mid-term 'desired state,' we do believe the main path is to aim for ROE around 15% while incorporating interest-rate factors, but we do not yet have a clear timeline at this point."
— Masahiro Kihara, President & Group CEO
Assessment: the ¥1.5tn figure is the most useful number management disclosed all session, because it converts an ROE target into a profit target an investor can track. From ¥1,300bn guided for FY2026, ¥1.5tn by FY2028 implies roughly 7.4% compound growth over two years with no rate help. That is achievable and it is not heroic, which is the right description of this whole plan.
What They're NOT Saying
- No first-half guidance. The Tanshin's 1H FY2026 row for profit attributable to owners of parent is blank. For a group that has just told the market its full-year number is "full tilt" with no macro cushion, declining to frame the half is a meaningful omission; it removes the only interim checkpoint investors would have before November.
- No second-half buyback framework. The ¥100bn tranche runs to August 31. Management said it plans to "flexibly increase" the amount but gave no decision date, no trigger, and no capital-ratio threshold. Management gave no date at which it would revisit the size.
- The credit-cost guide is not reconciled to the Middle East reserve. FY2025 carried ¥54.7bn of forward-looking reserves for a conflict management describes as an ongoing and worsening risk, and FY2026 is guided to ¥110bn, ¥23bn lower. Either the FY2025 reserve is expected to be sufficient or the guide is optimistic, and management did not say which.
- No quantified Rakuten Bank impact. The alliance filing says the financial effect is "currently under review." Given an October 1 effective date and a 10.52% equity-method stake in a bank that earned ¥73.1bn last year, the order of magnitude is knowable today and was not given.
- The ¥120bn rate sensitivity was flagged as too high and left unrevised. Management said it "did not revise the figure this time" and intends to revise it later. The disclosure therefore remains in the deck at a level its author has publicly disowned.
- Nothing about ROE after the disposal programme. ¥360bn of stock gains sits inside the FY2026 guide and cross-shareholdings are still 30.1% of net assets against a sub-20% goal, so the programme has years to run. What the FY2028 12% ROE looks like once it stops was not addressed.
- No date or target for RBC to earn its cost of capital. A new co-head was named and the priorities were listed; a deposit target, an expense-ratio target and a timeline were not.
- The Orient Corporation shareholder proposal went unmentioned. The board opposed a proposed amendment to the articles of incorporation concerning "risks associated with the consolidation of Orient Corporation" on the same day as the results, and the presentation discussed the Orico stake sale to Muninova without reference to it.
Market Reaction
Mizuho released after the Tokyo close on Friday May 15, which split the reaction across two markets and four sessions, and layered a leaked story on top of it.
- Pre-print setup: the ordinary shares closed at ¥6,912 on May 15, up 21.3% year to date from ¥5,700 and trading at 1.49x the ¥4,640.23 book value per share reported at March 31. The ADR closed at $8.86 on May 14, up 21.0% year to date and 68.8% over twelve months, against a 52-week closing range of $5.14 to $10.09. The S&P 500 was up 9.6% year to date entering the print.
- ADR reaction session (May 15): the ADR opened at $8.59, a 3.0% gap down, traded $8.59 to $8.82 and closed at $8.69, down 1.9% on 4.2M shares against a 4.0M 30-day average. The S&P 500 fell 1.2% the same session.
- Tokyo reaction session (May 18): the shares fell 5.60% to ¥6,525 on 21.4M shares, 2.45x the 30-session average, with an intraday low of ¥6,379, down 7.71% from the pre-release close. The Nikkei 225 fell 0.97%.
- Recovery (May 19–20): the shares rose 5.53% to ¥6,886 on the day of the investor presentation and a further 1.39% to ¥6,982 the following session, recovering 7.00% from the trough and finishing 1.01% above the pre-release close. The Nikkei fell in both sessions, by 0.44% and 1.23%. The ADR closed at $8.69 on May 19, down 1.92% cumulatively from its pre-print close.
The May 18 selloff was not one thing. Two arrived together: a buyback that the market read against the ¥400bn precedent rather than against the unchanged policy, and a weekend press report that Mizuho planned further investment in Rakuten Bank, which the company answered on Monday morning by confirming it was examining options but had made no decision. Rakuten Bank shares rose as much as 10.4% the same session, their largest gain in three months, which is the market pricing a transfer of value from the acquirer to the target. The Nikkei fell less than a point that day; this was idiosyncratic.
What happened next is more informative than the selloff. The shares recovered the entire decline across the investor presentation and the alliance announcement. The presentation supplied the normalized ¥1,150bn framing, the explicit statement that the payout policy was unchanged, and the chief executive's regret over the announcement. The alliance supplied the actual terms, which turned out to involve no cash and a capped 10.52% stake rather than the open-ended commitment the leak implied. A stock that round-trips a 7.7% drawdown in two sessions on clarification rather than on new earnings is telling you the drawdown was a communication event, not a fundamental one.
At the May 20 close of ¥6,982 the shares trade at 1.50x book, 13.1x the guided FY2026 EPS of ¥533.10 and a 2.15% yield on the guided ¥150 dividend. That is essentially where they started the week, and it is not a distressed entry point.
Street Perspective
Debate: Is the ¥100bn buyback a policy change or a placeholder?
Bull view: the policy language is verbatim unchanged, the chief executive restated it twice and apologised for the confusion, and the arithmetic leaves roughly ¥284bn of buyback capacity under a 50% total payout on guidance. The ¥100bn is a first tranche sized for an unresolved geopolitical risk, and it runs only to August, which is a short leash by design.
Bear view: the CET1 ratio fell 40bp on 12.8% risk-weighted asset growth, and management has said it is actively considering strategically increasing risk-weighted assets further. Capital that goes into the balance sheet does not come back as buybacks. The Middle East framing is a convenient cover for a group that has decided to grow rather than return.
Our take: the bulls have the better of the policy argument and the bears have the better of the capital argument, and both can be right. Our expectation is a second tranche, sized to keep the total payout near 50% rather than near last year's 60%. That is a step down in return intensity from FY2025 without a policy breach, and it is what the shares now discount.
Debate: How much of the FY2025 result is repeatable?
Bull view: the repeatable part is the part that grew. Net interest income up 31.7% and net fee income up 19.1% are franchise revenue, the domestic margin has widened 18bp with more policy rate to come, and the expense ratio has fallen three points. Management's normalized ¥1,150bn is a conservative construction that strips out the securities-loss realization as well as the one-off gains.
Bear view: ¥104.5bn of the ¥316.8bn improvement in net business profits is a banking book recovering from a self-inflicted prior-year loss, ¥190.8bn of the year-over-year improvement is stock disposals, and a ¥65.3bn deferred-tax benefit did the rest. Adjust the tax rate to last year's and the consensus beat shrinks from 4.8% to 1.2%. The customer businesses that are supposed to carry the story delivered lower net income in two of four cases.
Our take: both sides are describing the same accounts. The reconciliation is that the revenue improvement is largely genuine and the profit improvement is largely not, because credit costs and the cost base absorbed the revenue in retail and global corporate banking. That is why we anchor on the customer-groups net business profit line of ¥1,124.7bn, up 22%, and treat the ¥1,248.6bn headline as an accounting outcome.
Debate: Should Mizuho trade at its peers' multiple?
Bull view: at 1.50x book with an 11.4% ROE and a credible path to over 12% by FY2028 that assumes no further rate hikes, the multiple is undemanding. Every 25bp of policy rate adds 0.6 to 0.7 points of ROE on management's arithmetic, and none of it is in the plan. The disposal programme funds the transition.
Bear view: Mizuho is guiding to a fraction of the growth rate its largest peers have set out, and both of its largest customer franchises earn below cost of capital. Same price, less growth, worse mix.
Our take: the bear case is the one that decides the rating. A relative-value argument does not require Mizuho to be a bad bank, only to be a worse expression of the same trade. The domestic rate cycle, the corporate-action cycle and the cross-shareholding unwind are sector themes available at every megabank; Mizuho's idiosyncratic contributions are a retail franchise earning 5.9% and a global CIB earning 8.0%. Until one of those inflects, the multiple should not lead.
Debate: Is the Rakuten Bank alliance a fix or a distraction?
Bull view: it is the cheapest possible solution to the group's stated number-one problem. Mizuho swaps a minority card stake for 10.52% of a digital bank with ¥16.6tn of assets and a ¥73.1bn profit, gains a permanent distribution channel for originated corporate loans and securitized receivables, and pays no cash. The alliance directly serves the originate-and-distribute answer management gave on risk-weighted asset discipline.
Bear view: a 10.52% stake with a standstill and transfer restrictions is not control, the deposits belong to Rakuten Bank rather than to Mizuho, and the earnings contribution is immaterial against a ¥1,300bn guide. Meanwhile the group has now accumulated Rakuten Securities, Rakuten Card and Rakuten Bank relationships across three separate transactions without yet demonstrating a P&L effect from any of them.
Our take: the bear case on materiality is correct and the bull case on optionality is correct, and optionality that costs nothing is worth having. What the deal does not do is change FY2026 or FY2027. We would treat it as a genuine improvement to the FY2028-and-beyond structure and give it no weight in near-term numbers, which is also what the "impact currently under review" language invites.
Model Update & Valuation Framework
This is an initiation, so what follows establishes the base case rather than revising one.
| Item | Company guidance / FY2025 actual | Our base case | Reason |
|---|---|---|---|
| Profit attributable to owners of parent, FY2026 | ¥1,300.0bn guided | ¥1,330–1,360bn | The house pattern is to guide low and beat; USD/JPY at 150 assumed against a materially weaker spot is worth roughly ¥25bn alone. |
| Consolidated net business profits, FY2026 | ¥1,630.0bn guided | ¥1,600–1,650bn | Achievable, but it needs to replace ¥104.5bn of non-repeating banking-book recovery with customer-group growth. |
| Credit-related costs, FY2026 | ¥110.0bn guided | ¥130–150bn | Guiding costs down while carrying an unresolved geopolitical overlay and a retail book that just took a ¥71.3bn swing is the least defensible line in the plan. |
| Net gains related to stocks, FY2026 | ¥360.0bn guided | ¥330–360bn | Achievable given 45% of the three-year listed-stock target was completed in year one, but it is 25.5% above what was just delivered and depends on the equity market. |
| Effective tax rate | 22.7% in FY2025 | 24–26% | The ¥65.3bn deferred benefit does not repeat; this is the main bridge between the +18.2% ordinary-profit guide and the +4.1% net-income guide. |
| ROE, FY2026 | 11.4% delivered in FY2025 | approximately 11.0–11.3% | Our arithmetic, not the company's: equity compounds faster than the profit guide when the total payout runs near 50%, so reported ROE likely dips before the FY2028 target. |
| Total payout ratio, FY2026 | 50% or more (policy) | 50–55% | Down from 60% in FY2025. Implies a second buyback tranche of roughly ¥150–200bn on top of the ¥100bn announced. |
| Rate sensitivity per +25bp | approximately ¥120bn pre-tax (disclosed) | ¥95–115bn | Management has publicly said the disclosed figure is probably too high and intends to revise it. |
Valuation framework. At the May 20 close of ¥6,982 the shares trade at 1.50x the ¥4,640.23 March-2026 book value, 13.1x guided FY2026 EPS of ¥533.10 and a 2.15% dividend yield. Book value per share compounded 11.5% in FY2025 while the group paid out 60% of earnings, which is the single most attractive statistic in the results. Applying the FY2028 target of over 12% ROE to a book value compounding at high single digits gets to a materially higher intrinsic value than 1.5x, but only on a three-year view and only if the customer franchises inflect.
The bridge that decides the rating: for the shares to re-rate above 1.5x book on anything other than the sector's rate trade, RBC has to move from 5.9% toward a cost-of-capital return and GCIBC has to stop losing ground at 8.0%. Between them they consume 45.8% of the group's internal risk capital and generate 27.1% of its net income. That is the arbitrage inside Mizuho, and it is worth more than any assumption about the Bank of Japan.
Thesis Scorecard Post-Earnings (Initiating Coverage)
No prior Aardvark Labs coverage exists for Mizuho, so this quarter establishes the thesis rather than grading one. The pillars below are the ones we will carry forward and score each quarter.
| Thesis Point | Status | Notes |
|---|---|---|
| Bull 1: The domestic rate cycle re-rates the deposit franchise | Confirmed | Net interest income +31.7%; domestic loan and deposit margin 0.92% to 1.10%; cumulative FY2026 pre-tax benefit from past hikes of approximately ¥345bn. The pass-through to depositors is running higher than modelled, which is why this is confirmed rather than exceeded. |
| Bull 2: Fee income is a structurally growing, rate-independent engine | Confirmed | Non-interest income to a record ¥1,463.4bn, fifth consecutive annual increase; domestic corporate solutions ¥264.1bn to ¥330.1bn; CIBC ROE 16.0%. |
| Bull 3: The cross-shareholding unwind funds the transition to a 12% ROE | Neutral | ¥286.8bn of stock gains delivered and ¥360bn guided, with 45% of the three-year target done in year one. Works as designed, but it is now 19% of the FY2026 ordinary-profit plan, which makes it a dependency as much as a tailwind. |
| Bear 1: The two largest customer franchises earn below cost of capital | Challenged | RBC net income −4% at 5.9% ROE; GCIBC net income −9% at 8.0% ROE with the expense ratio up 2.3 points. Together 45.8% of internal risk capital for 27.1% of net income. Management named RBC as the priority without attaching a target or a date. |
| Bear 2: Capital return steps down from FY2025's 60% total payout | Challenged | Buyback opened at ¥100bn against ¥400bn; CET1 down 40bp to 9.9%; management actively considering further strategic risk-weighted asset growth. The policy is intact; the intensity is not. |
| Bear 3: Guidance credibility and communication | Neutral | The FY2026 guide came in below consensus and last among the megabanks, and required a normalized-basis defence four days later. Offsetting this, the FY2025 forecast was beaten by 10.5% and the FY2027 ROE target was hit two years early, so the delivery record is good even where the messaging is not. |
Overall: a strong operating year inside a franchise with two unresolved return problems, guided forward at the slowest rate in its peer group. The thesis is established as balanced rather than positive.
Action: hold. We would revisit on evidence of a second buyback tranche that keeps the total payout near 60%, on a dated commitment for RBC, or on a share price back toward the 1.41x book struck on May 18.