Bank of Hangzhou: The Best Balance Sheet in China's Riskiest Sector

I. Introduction & Episode Roadmap

On the afternoon of April 28, 2026, a conference room inside the headquarters of 杭州银行 Bank of Hangzhou filled with analysts from Morgan Stanley, HSBC, Guangfa, Changjiang, and half a dozen domestic fund houses. The bank had just closed a year in which its revenue grew 1.09% — the weakest showing in a decade — while its net profit grew 12.05%.⁷⁸ The first question of the session did not come from a sell-side analyst. It came from a retail shareholder, and it was blunt: if net interest income grew more than 12% and fee income grew more than 13%, why did total revenue barely move? And was the profit number being manufactured out of provisioning releases and cost control?¹

That question is the whole story of this bank in a single sentence.

Bank of Hangzhou is a roughly RMB 2.36 trillion (about USD 330 billion) city commercial bank headquartered in the city that produced 阿里巴巴 Alibaba, that hosts one of the densest private-enterprise clusters on earth, and that in 2025 became internationally famous for the "六小龙 six little dragons" — the cohort of Hangzhou hard-tech startups including DeepSeek and Unitree Robotics that briefly made the city the most-discussed address in Chinese technology. Bank of Hangzhou banks four of those six.¹ Almost no global investor has heard of it.

Here is the hook. For thirteen consecutive quarters through the end of 2025, this bank reported a non-performing loan ratio of 0.76% — flat, quarter after quarter, the lowest of any listed Chinese bank, in a banking system that has spent five years absorbing a property collapse, a local-government debt restructuring, and the sharpest net interest margin compression in its modern history.⁸⁹ In the same window, three other things happened. Its foreign anchor shareholder of twenty years, Commonwealth Bank of Australia, quietly sold out entirely.⁴ Its president resigned abruptly in April 2025, more than a year before his term expired, with no explanation beyond "personal reasons."⁶ And in May 2026, minority shareholders threw roughly 120 million shares against each of two executive-compensation resolutions at the annual meeting — an unusually loud protest vote for an A-share bank.³

An investor looking only at the asset-quality line would conclude this is the best-run bank in China. An investor looking only at the governance line would conclude something is wrong. Both readings are supported by the disclosures. The job of this episode is to work out how they fit together.

Three threads run through what follows. The first is institutional: how a 1996 merger of 33 city credit cooperatives in a provincial capital became a specialist lender to hard-tech startups, and whether that specialization is a genuine competitive advantage or a well-marketed version of ordinary relationship banking. The second is about ownership: what it means when a bank trades a Western strategic investor for a shareholder register dominated by the municipal government's own investment platforms — several of which are also among its largest borrowers. The third is about earnings quality: why profit growth has been running four to eleven times faster than revenue growth, what is actually producing that gap, and whether the two business lines management publicly designated as its growth engines are delivering or quietly failing.

That last thread matters most, because "best asset quality in China" and "best bank to own" are not the same claim, and the distance between them is where the money is made or lost.

↑ Back to Top

II. Origins: From City Credit Cooperatives to Zhejiang's Flagship Bank

In the mid-1990s, China had a problem it had created itself. Across every city of any size sat hundreds of urban credit cooperatives — small, loosely supervised deposit-taking institutions that had proliferated during the 1980s reform decade to serve the private traders and workshop owners the state banking system ignored. By 1995 they were a systemic hazard: undercapitalized, badly documented, frequently insolvent, and politically untouchable because each one was somebody's local franchise. Beijing's solution was consolidation. Merge them, city by city, into a single capitalized entity with a real balance sheet and a real regulator.

Hangzhou's turn came in 1996. On May 9 of that year the People's Bank of China authorized the formation of a city cooperative bank in Hangzhou; on September 25, 1996, Hangzhou City Cooperative Bank formally opened for business, assembled from the shareholders of 33 urban credit cooperatives, the five directly-administered offices of the municipal credit union, and the Hangzhou Municipal Finance Bureau.²⁰ It was the ninth such institution in the country. Two years later, in May 1998, it was renamed Hangzhou City Commercial Bank; a decade after that, in 2008, it took the name it carries today.²⁰

The pre-2005 history is table-setting rather than plot. What matters is the geography, because geography is destiny for a city commercial bank. A Chinese bank's loan book is not really a portfolio choice — it is a photograph of the local economy it is licensed to serve. And Zhejiang's economy is unlike almost anywhere else in China.

Zhejiang did not industrialize on the back of state-owned heavy industry. It industrialized through merchant clusters: the shoe and lighter workshops of Wenzhou, the small-commodity trading city of Yiwu, the textile complexes of Shaoxing, the electronics towns of Ningbo. The unit of production was the family firm, not the ministry subsidiary. By the 2000s Hangzhou itself layered a second economy on top — the platform businesses that grew up around Alibaba, and later the payments, logistics, and cloud ecosystems that clustered near them.

For a bank, this creates a specific set of conditions. Borrowers are numerous, small, and opaque: they have no bond rating, patchy financials, and collateral that is often a leased workshop. Underwriting them cannot be done from a spreadsheet in a head office. It requires what economists call soft information — the loan officer knowing the borrower's customers, the borrower's cousin who supplies the components, whether the order book is real. Zhejiang banks that survived multiple credit cycles did so by industrializing that soft-information process. The ones that did not survive lent against collateral values and guarantee chains, and were destroyed when Wenzhou's mutual-guarantee networks collapsed after 2011.

That last point deserves a moment, because it is the formative trauma of banking in this province and it explains a great deal about how Bank of Hangzhou behaves today. In the boom years before 2011, small Zhejiang firms without hard collateral obtained credit by guaranteeing each other's loans. Firm A guaranteed Firm B, which guaranteed Firm C, which guaranteed Firm A. On paper every loan was secured. In practice the entire structure was one balance sheet wearing many hats, and when export demand collapsed and the first links failed, the defaults propagated through the chains like current through a circuit. Banks that had congratulated themselves on secured lending discovered they held nothing. Wenzhou's credit market seized, entrepreneurs left the country, and the province spent years working through the wreckage.

The banks that came out of that episode intact learned a specific and durable lesson: in small-business lending, collateral and guarantees are decoration, and the only thing that matters is whether the borrower has real cash flow from real operations. That is precisely the framing Bank of Hangzhou's executives still use in public — deputy president Zhang Jingke described the essence of credit risk control as identifying the actual source of repayment and monitoring the borrower's cash position, with a specific carve-out for technology companies where venture funding constitutes a second legitimate repayment source.² Investors evaluating a Chinese regional bank should always ask what credit cycle taught the current management team, because that is usually a better predictor of behaviour than any published risk policy.

The upside of the Zhejiang franchise is that private-sector credit demand is genuinely abundant and pricing is better than in state-dominated provinces. The downside is concentration: a bank whose entire book sits in one province is a leveraged bet on that province. Chairman 宋剑斌 Song Jianbin said as much at the 2026 briefing, noting that most of the bank's credit assets sit in Zhejiang and framing the province's "415X" advanced manufacturing cluster programme — four target industries aimed at trillion-yuan scale — as the bank's addressable market.¹ That is an honest description of both an opportunity and a correlation risk.

By the mid-2000s the bank had a defensible local position and a structural problem shared by every Chinese city commercial bank of the era: it did not really know how to manage credit risk at scale, and it did not have enough capital. The fix for both arrived from an unlikely direction — Sydney.

↑ Back to Top

III. The Foreign Anchor Era: Commonwealth Bank of Australia (2005–2021)

In 2005 the Chinese banking system was being pried open. Beijing had decided that the fastest route to modernizing domestic banks was to sell strategic minority stakes to foreign institutions and let the governance and risk technology flow in — Bank of America into China Construction Bank, HSBC into Bank of Communications, Goldman Sachs into ICBC. Further down the food chain, mid-sized city banks were paired with mid-sized foreign partners.

Commonwealth Bank of Australia — a retail-heavy, conservatively run institution with no meaningful investment-banking culture and an obsession with credit process — bought into Hangzhou. The entry was cheap. CBA subscribed for 250 million shares at RMB 2.50 apiece, roughly RMB 625 million, for what became a 19.91% position and the status of largest shareholder and designated strategic investor.⁴ By 2016, on the eve of the bank's public listing, that stake stood at about 20%.⁵

What does a foreign strategic investor actually deliver in an arrangement like this? Not capital, primarily — RMB 625 million was not transformational. What it delivers is process. Board seats mean somebody in the room asks why a credit was approved. Secondments mean risk-rating models get rebuilt to a template that has survived an Australian property cycle. A foreign partner with reputational exposure has a standing incentive to say no to the deals a local bank's relationship managers want to say yes to.

It is genuinely difficult to prove the counterfactual, and investors should be careful with the tidy story that CBA's presence explains today's 0.76% NPL ratio. Many Chinese banks took foreign strategic investors in this period and few of them ended up with best-in-class asset quality. Bank of Ningbo, the closest and most successful Zhejiang comparator, took a smaller foreign stake and built its underwriting culture largely on its own. What can be said with more confidence is that Bank of Hangzhou spent fifteen years operating with a permanent, well-resourced, commercially motivated skeptic on its board — and that the risk-management vocabulary its executives still use in public today, with its emphasis on identifying repayment sources and refusing to trade risk for growth, is recognizably the vocabulary of a conservative retail bank rather than that of a growth-chasing Chinese lender.²

The public milestone came on October 27, 2016, when Bank of Hangzhou listed on the Shanghai Stock Exchange under 600926.SS, the first Zhejiang-headquartered financial institution to reach the SSE main board.²⁰ A listing does several things at once for a Chinese city bank: it supplies core Tier 1 capital, it forces quarterly disclosure, and it creates a currency — a traded share price — that later becomes the mechanism by which the bank solves its next capital problem. Hold that thought.

The listing itself is worth a sentence of context, because 2016 was an unusual window. Chinese regulators had frozen bank IPOs for four years after 2010, and reopened the queue in 2016 for a cohort of regional lenders that had grown too fast for their capital bases. Bank of Hangzhou came to market at a moment when domestic investors were being told that well-run city commercial banks were the growth end of Chinese financials. The framing was largely correct, and it has shaped how the bank has funded itself ever since: as a listed institution with a rising share price, it has consistently been able to convert market confidence into regulatory capital.

It also creates an exit. And this is the part of the strategic-investor story that Chinese banks and their analysts consistently underweight. A foreign anchor shareholder is read domestically as a vote of confidence, a permanent endorsement. It is nothing of the sort. It is a position on somebody else's balance sheet, subject to somebody else's capital allocation review, in somebody else's currency, under somebody else's regulator. The question was never whether CBA believed in Bank of Hangzhou. The question was when the Australian bank's own strategy would make a locked-up minority stake in a Chinese regional lender look like a non-core asset.

The answer turned out to be 2021.

↑ Back to Top

IV. The Great Unwind: CBA's Exit and the New (State-Anchored) Shareholder Base (2021–2025)

The unwind was executed the way these things always are — in tranches, at fair prices, with warm language attached.

The first tranche was announced in March 2021 and completed in February 2022. CBA transferred approximately 593.6 million shares — 10% of the bank — at RMB 13.94 per share, generating about RMB 8.275 billion in proceeds.⁴ The buyers were not financial investors. They were 杭州城投 Hangzhou Urban Construction Investment and 杭州交投 Hangzhou Communications Investment: two wholly municipal infrastructure and transport platforms, each taking 5%. CBA kept the balance, agreed to a three-year lock-up on the remainder, and the accompanying commentary framed the sale as portfolio management rather than retreat.

The second tranche ended the relationship. On January 24, 2025, CBA agreed to transfer its final 329.64 million shares — 5.45% — to 新华保险 Xinhua Insurance at RMB 13.095 per share, for approximately RMB 4.32 billion.⁵ Twenty years after buying in at RMB 2.50, the Australian bank walked away having realized roughly RMB 12.5 billion in cumulative gains, before counting two decades of dividends.⁴ CBA's stated rationale was the one it had been giving for several years: refocus on core banking in Australia and New Zealand, divest non-core international holdings. Notably, it did not exit China entirely — it retained its 15% position in Bank of Qilu in Shandong.⁵ The decision was specific to this asset, not to the country.

By any conventional measure it was a superb investment. It was also, for the remaining shareholders of Bank of Hangzhou, a regime change.

Look at the register that emerged after the dust settled and after the 2025 convertible-bond conversion diluted everyone proportionally. The largest holder is Hangzhou Financial Development Investment Group at about 16.12% — the municipal finance bureau's investment arm. Then 红狮控股集团 Hongshi Holding Group at roughly 9.84%, a privately held cement and building-materials conglomerate. Then Hangzhou Urban Construction Investment at about 6.08%, and Xinhua Insurance at 5.09%.⁴ Add the transport platform and the picture is unambiguous: Bank of Hangzhou is now anchored by the Hangzhou municipal government and its operating vehicles, with one large private industrial group and one domestic insurer alongside.

Read plainly, the bank traded imported governance rigor for local political alignment. Both have value, and it is worth being precise about the trade rather than moralizing about it.

What the state-anchored register buys: stability of ownership, essentially zero risk of a disruptive block sale, privileged access to the municipal fiscal ecosystem, and a shareholder base whose interests are aligned with the bank remaining a durable local institution rather than maximizing a three-year return. In a system where a bank's franchise depends heavily on relationships with local government, this is not a trivial asset. The bank's policy-and-government finance business — lending to municipal state-owned enterprises for infrastructure, industrial parks, and urban renewal — carried no risk assets at all as of the 2026 briefing, according to deputy president 李炯 Li Jiong.¹

What it costs: independent challenge. A municipal finance bureau's investment arm and two city infrastructure platforms are not natural adversaries of a bank's management on credit standards, executive pay, or dividend policy. Nor are they neutral parties on the composition of the loan book, since the same platforms sit on both sides of the relationship — shareholders of the bank and borrowers from it. The 2025 half-year and 2026 annual briefings both featured extended management discussion of policy-and-government finance demand pipelines in Zhejiang, described as ample and sustainable.²¹ That may well be true. But it is a related-party exposure of a size and structure that would attract sustained questioning in most developed markets, and it is disclosed in China at a level of granularity that makes independent verification hard. We will return to it when we get to the loan book.

There is one more piece of timing that investors should not wave away. CBA's final exit was agreed in January 2025. On April 2, 2025, president 虞利明 Yu Liming resigned with immediate effect.

↑ Back to Top

V. The Core Business: Banking Zhejiang's Private Economy

Start with the shape of the thing. At the end of 2025 Bank of Hangzhou held about RMB 2.36 trillion in total assets, up 11.96% on the year; RMB 1.072 trillion in loans, up 14.33%; and RMB 1.44 trillion in deposits, up 13.20%.⁸¹³ A year earlier those figures had been RMB 2.11 trillion, RMB 937.5 billion and RMB 1.27 trillion respectively.⁹ Roughly 14,000 employees run it.¹

In the league table of listed Chinese city commercial banks, that places Bank of Hangzhou in the upper-middle. 江苏银行 Bank of Jiangsu displaced Bank of Beijing as the largest by profit in 2025, earning RMB 34.5 billion on revenue of RMB 87.9 billion. Bank of Beijing, 宁波银行 Bank of Ningbo, 上海银行 Bank of Shanghai and 南京银行 Bank of Nanjing each cleared RMB 20 billion of net profit. Bank of Hangzhou earned RMB 19.03 billion.¹⁷⁸ Across all 17 listed city commercial banks, combined 2025 revenue ran about RMB 530.8 billion and combined net profit about RMB 203.5 billion — meaning Bank of Hangzhou produces roughly 9% of the sector's earnings on about 7% of its revenue.¹⁷ That efficiency gap is real, and by the end of this episode it should be clear how much of it is structural and how much is accounting.

Where the money actually comes from

Corporate lending is the engine, and it is not close. In 2024, corporate banking generated roughly RMB 8.4 billion of segment profit against about RMB 850 million from retail and RMB 800 million from small and micro enterprise banking.¹¹ Infrastructure and urban-construction lending accounted for roughly 44% of the total loan book.¹¹ Strip away the strategy language and Bank of Hangzhou is, at its core, a corporate and government-project lender in the wealthiest urban cluster in China, with a retail business attached.

That matters because retail and small-micro were not supposed to be an appendix. Under the bank's 2021–2025 five-year plan, retail finance and small-micro lending were explicitly designated the growth engines. Section X deals with how that turned out.

The genuine bright spot in fee income is 杭银理财 Hangyin Wealth Management, the bank's wealth-management subsidiary. Its outstanding product balance passed RMB 430 billion at the end of 2024, up about 17% year on year, and exceeded RMB 510 billion by mid-2025, putting it in the first tier of city-commercial-bank wealth subsidiaries.¹⁴ In 2025 the bank restructured wealth management into a standalone first-tier head-office department and reported retail client assets under management up 15.73%.¹ Fee and commission income grew 13.10% in 2025 to RMB 4.21 billion.⁸ For a bank staring at permanent margin compression, fee income that compounds at double digits is the single most valuable thing it can build, because it consumes almost no capital.

The differentiated bet: 科创金融 science-and-innovation finance

Now the part of the story management most wants investors to focus on.

For more than fifteen years Bank of Hangzhou has been building a specialist lending model for technology companies, organized around a three-word slogan: 做早、做小、做硬科技 — get in early, lend small, back hard tech. By the end of 2024 the bank operated a "1+7+N" specialist structure, served more than 28,000 technology enterprises, and carried a technology loan balance above RMB 115 billion.¹⁴ In the first half of 2025, technology loans grew 21% and manufacturing loans 23% from the start of the year — far ahead of total loan growth.²

Why is this hard, and why might it be a moat? Consider what a hard-tech startup looks like to a traditional credit officer. It has heavy upfront investment, capacity that has not yet ramped, and therefore terrible book financials. It has no meaningful fixed-asset collateral. Its value sits in patents, in engineers, and in a production process that may or may not work. Li Jiong described the requirement precisely at the 2026 briefing: the bank has to be able to form a judgement on industry research, patents and human capital — assets that never appear on a balance sheet — and then layer on policy-backed risk compensation, intellectual-property pledges, and investment-lending linkage to share the risk it cannot underwrite alone.¹

The organizational answer has been to stop treating this as normal lending. Science-innovation finance runs as a business division with its own profit and loss, dedicated science-innovation centres in key regions, embedded credit approval inside those centres rather than at head office, and specialist teams for healthcare, intelligent manufacturing, and artificial intelligence.¹ In 2025 the bank launched a platform called 投融通 to provide fund-raising-to-exit services for government funds and market investors, positioning itself inside the local venture ecosystem rather than merely adjacent to it.¹

The economics are the most interesting disclosure. Asked at the 2025 half-year briefing whether technology lending was a reputational business that does not make money, then-deputy president 张精科 Zhang Jingke gave an unusually concrete answer: pricing on these loans is set by market competition and the bank simply takes the market rate, so profitability has to come from the cost side — funding cost and risk cost. His evidence was that within the science-innovation specialist units, loans and demand deposits run at roughly a 1:1 ratio.² In plain English: the bank lends to a startup, the startup parks its venture funding and working capital in a current account at the same bank, and that near-zero-cost deposit funds the loan. The spread is not in the loan rate; it is in the deposit.

That is a real mechanism, and it is the correct way to think about relationship lending. It is also exactly the mechanism that made Silicon Valley Bank enormously profitable for two decades before it failed — a comparison worth holding, not because Bank of Hangzhou has SVB's duration mismatch, but because it clarifies that the model's returns depend on a client base whose deposits are correlated with venture funding cycles.

Two honest caveats belong here. First, the bank does not disclose segment-level profitability or default rates for the science-innovation book separately, so the claim that "doing technology lending can make money" rests on management assertion plus a deposit-ratio data point, not on published portfolio economics. Second, every large Chinese bank is now under policy instruction to lend to technology firms, and the big state banks can undercut on price indefinitely. Management effectively conceded the competitive pressure, describing "involution" — ruinous price competition — in lending to quality manufacturers and tech firms, and noting that only the central bank's pricing self-discipline mechanism has put a floor under loan rates.²

Where the 0.76% comes from, and what it hides

The headline is genuinely exceptional. NPL ratio flat at 0.76% and provision coverage of 502.24% at end-2025, down from 541.45% at end-2024.⁸⁹ Net NPL formation ran 0.72% in 2025, slightly lower than the prior year, and fell about 10 basis points further to an annualized 0.6% in the first quarter of 2026.¹

But the blended ratio is an average, and averages hide distributions. In 2023 the bank's corporate real-estate loan book carried a 6.36% NPL ratio — real estate NPLs represented 52.66% of all corporate bad loans and 38.3% of the bank's total NPLs, despite real estate being only 4.56% of the loan book.¹⁵ By the 2024 report that ratio had risen further to 6.65%.¹⁶ The bank's overall book is clean because property is a small slice of it, not because the bank underwrote property better than anyone else.

Myth versus reality

Three consensus statements about this bank are worth testing against the disclosures.

Myth: the 0.76% NPL ratio proves Bank of Hangzhou is the best underwriter in China. Reality: it proves the bank is the best underwriter of the mix it actually holds. That mix is roughly 44% infrastructure and urban construction lending to municipal state-owned borrowers in China's wealthiest province, where implicit government support has so far been absolute, plus a small property book that has performed poorly and a retail book that is now deteriorating.¹¹¹⁶¹³ The skill is real; the portfolio composition is doing a lot of the work.

Myth: the science-innovation franchise is a proven moat. Reality: it is a proven capability with an unproven economic return. The bank has published client counts, loan balances, and an organizational structure. It has not published segment profitability, segment NPL formation, or client retention, and it has explicitly said it has no pricing power in the segment.²¹⁴ A moat that cannot be measured is a hypothesis.

Myth: the convertible bond conversion was a costless capital solution. Reality: it was a 22% equity issuance at a price fixed four years earlier, executed at the moment the share price rose.¹² The mechanism was elegant. The dilution was ordinary.

The same logic now applies on the retail side, and the direction of travel there is the opposite of reassuring. That is the subject of Section X. First, the industry structure that constrains what any bank in this position can do.

↑ Back to Top

VI. Industry Structure: Porter's Five Forces and the Regional-Bank Power Set

Here is a war-game. Suppose you are the treasurer of a profitable Hangzhou machinery exporter with RMB 400 million of revenue, and you need a RMB 120 million facility. Who calls on you?

Bank of Hangzhou, obviously. Also Bank of Ningbo, which is bigger and has the better digital retail franchise. Also Bank of Jiangsu and Bank of Nanjing, both hunting Yangtze River Delta corporate business from adjacent provinces. Also China Merchants Bank and every other national joint-stock lender. Also the local branches of the Big Four, whose cost of funds is structurally below everyone else's and who have been instructed by policy to grow manufacturing and technology lending. That is eight to twelve credible bidders for one loan.

This is the single most important fact about the industry, and it dictates the shape of all five of Porter's forces.

Rivalry is brutal and getting worse. The competition is not for deposits, which are rate-capped and regulated; it is for the shrinking pool of creditworthy borrowers. Chinese banks call the resulting condition an "asset shortage" — too much lending capacity chasing too few good credits. The consequence shows up in the numbers with unusual clarity: banking-system net interest margin fell 22 basis points in 2023, 17 in 2024, and 10 in 2025, with Bank of Hangzhou's own margin falling 19, 9, and 5 basis points over the same three years.¹ The bank's margin ended 2025 at 1.36%, down 5 basis points, and recovered to 1.42% in the first quarter of 2026 — the first sequential improvement in years, driven mostly by deposit repricing rather than asset yields.⁸¹ Deputy president 章建夫 Zhang Jianfu was careful not to overclaim, saying the inflection point still needed confirmation.¹ That is the right posture; margin recovery driven by falling deposit costs is a one-time repricing benefit, not a structural improvement.

Buyer power is asymmetric and rising on the corporate side. Large corporates and municipal investment platforms can and do run competitive processes. Management said as much: local state-owned enterprises have explicit mandates to cut their financing costs, and peer banks have been using long-tenor, low-price offers to displace incumbents on quality assets.¹ Retail depositors have almost no leverage under regulated deposit ceilings — which is precisely why the deposit franchise is worth so much — but retail investors increasingly do, because wealth-management products compete on visible returns.

Substitution is a slow structural leak. The best borrowers increasingly bypass banks entirely through the bond market. Bank of Hangzhou has responded by becoming the underwriter rather than the lender — management noted it has held the top interbank-market bond underwriting share among banks in Zhejiang for several years running.¹ That is the correct adaptation: if disintermediation is inevitable, capture the fee. But it is a lower-revenue business than holding the loan, and the residual pool of borrowers who still need bank credit is, by definition, the pool that could not access the bond market.

Barriers to entry are close to absolute. Banking licences in China are not issued to newcomers; the regulator has spent five years consolidating small banks, not creating them. That protects incumbents completely. It also means growth in this industry is a share-shift game, not a market-expansion game — and it means the competitive equilibrium is set by whoever is willing to accept the lowest return, because nobody exits.

Regulatory intensity is the dominant external variable. PBOC policy rates and LPR set asset yields; the National Financial Regulatory Administration sets capital and provisioning rules; and increasingly, fiscal policy directly determines competitive position. The clearest example: in 2025 the Ministry of Finance introduced interest subsidies for consumer loans, and Bank of Hangzhou was left off the first list of eligible banks. Deputy president 陈岚 Chen Lan described the effect candidly at the half-year briefing — a widening rate gap versus listed banks, risk of losing existing customers, and greater difficulty acquiring new ones — and said the bank was lobbying local authorities for a parallel municipal scheme.² By the first quarter of 2026 the bank had been brought into the second batch and had signed subsidy agreements with more than 24,000 consumer-loan customers.¹ The episode is a useful reminder that in this industry, a policy list can matter more to a segment's growth than any amount of product work.

One more structural feature deserves attention, because it distinguishes Chinese banking from most industries. Nobody in this business can meaningfully differentiate the product. A one-year working-capital loan from Bank of Hangzhou and a one-year working-capital loan from Bank of Ningbo are the same instrument. What varies is price, speed of decision, and whether the relationship manager understands the borrower's business. That is why the competitive battle has migrated so completely into service bundling — Bank of Hangzhou's answer being a suite of digital platforms it markets under names like 财资通 for corporate treasury, 政务通 for institutional and government clients, 票证通 for trade documentation, and 投融通 for fund and investment services.¹ The strategic logic, as Song Jianbin explained it, is to stop asking whether a thin-margin transactional product is profitable in isolation and start asking whether it anchors the customer relationship that makes everything else profitable.¹ That is the right answer to a commodity problem. It is also the answer every competent competitor is giving.

The 7 Powers reading

Run Hamilton Helmer's framework against this and most of the boxes come back empty. Scale economies: no — Bank of Hangzhou is a fraction of the size of the national banks it competes against, and Chairman Song was explicit that "we are a small bank."¹ Network economies: no. Branding: marginal, and in banking, brand mostly translates into deposit cost, where the Big Four win. Cornered resource: no. Process power: partially, in the sense that fifteen years of building science-innovation credit process is genuinely hard to replicate quickly, but process power is the hardest of the seven to prove from outside.

The two that hold up best are counter-positioning and switching costs, and they are related. Counter-positioning: a large national bank could build a Hangzhou hard-tech lending division, but doing so at the ticket sizes involved — small loans to pre-revenue companies requiring specialist judgement and embedded approval authority — is uneconomic against its own cost structure and inconsistent with its centralized credit process. It is not that they cannot; it is that the return on effort is poor relative to lending RMB 2 billion to a provincial SOE. Switching costs: once a startup's operating accounts, payroll, foreign-exchange settlement, IP pledges and government-subsidy applications all run through one bank, moving is genuinely painful. The 1:1 loan-to-demand-deposit ratio is the quantitative fingerprint of that entanglement.

The honest assessment is that this is a thesis, not a settled moat. A moat should show up as pricing power, and Bank of Hangzhou has explicitly disclaimed pricing power in exactly this segment — management said it takes the market rate because the market sets it.² What the franchise appears to deliver instead is lower risk cost and lower funding cost at the same price, which is a real and defensible edge if it persists across a full cycle. It has not yet been tested by one. Hangzhou's hard-tech cluster has never experienced a serious venture funding winter with this loan book outstanding.

That structural squeeze — no pricing power, compressing margins, capital-hungry growth — is what makes the next chapter necessary.

↑ Back to Top

VII. Capital Deployment: The 杭银转债 Convertible Bond and the Quiet Capital Raise (2021–2025)

Every fast-growing Chinese bank eventually runs into the same wall. Loan growth consumes risk-weighted assets; risk-weighted assets consume core Tier 1 capital; core Tier 1 capital can only come from retained earnings or from shareholders. Grow loans at 15% with a return on equity in the mid-teens and a payout ratio around a quarter, and the arithmetic more or less works — until it does not.

Bank of Hangzhou hit that wall around 2020. Its answer, in 2021, was a RMB 15 billion convertible bond: 杭银转债.

The instrument deserves a plain-English explanation, because it is the most misunderstood tool in Chinese bank finance. A convertible bond is debt that turns into equity if the share price rises enough. For the issuer, the appeal is threefold. It raises money immediately at a low coupon. It does not dilute existing shareholders on day one, so it avoids the price hit that follows an announced share placement. And — critically for a bank — if it converts, the proceeds count as core Tier 1 capital, the highest-quality regulatory capital there is. Until conversion, they do not. So the bank is carrying the cost of the debt while waiting for a share price it does not control.

That is the trap. If the shares never rally, the bond just matures and the bank repays cash, having achieved nothing on capital. Bank of Hangzhou spent four years in exactly that limbo.

The tension was visible at the 2024 annual results briefing on April 16, 2025. Board secretary 王晓莉 Wang Xiaoli disclosed that as of April 15, only 30.76% of the bond had converted, leaving RMB 10.385 billion outstanding against a March 2027 maturity — and said the bank was "actively seeking support from major shareholders" on conversion.¹⁸ That phrasing is worth pausing on. It means management was asking its largest shareholders to buy the bonds and convert them voluntarily, effectively arranging a capital injection by persuasion. She also disclosed the core Tier 1 ratio at end-2024: 8.75%, uncomfortably close to regulatory minimums for a bank growing loans at 16%, and estimated that full conversion would add roughly 0.8 percentage points.¹⁸

Then the market did the job for them. Chinese bank shares rallied hard through the first half of 2025 as investors rotated into high-dividend financials. Bank of Hangzhou's share price cleared the trigger — 130% of the RMB 11.35 conversion price for the required number of trading days — and the bank called the bond for mandatory redemption. Holders faced the standard choice: convert at RMB 11.35 or accept a redemption price barely above par. Essentially everyone converted. Between May 9 and June 30, 2025 alone, a further RMB 7.641 billion turned into 673 million shares.¹² By the end of June, 99.39% of the issue had converted, creating about 1.311 billion new shares — equivalent to 22.11% of the pre-conversion A-share count — and the bond was delisted on July 7, 2025.¹²²¹

The capital effect was immediate and large. Core Tier 1 rose to 9.74% at the end of the second quarter of 2025, and finished the year at 9.59%, up 0.74 percentage points on 2024, with total capital adequacy at 14.37%.²⁸ Total shares outstanding rose to 7.25 billion.² The capital constraint that had been capping loan growth was, for the moment, relieved — and the bank promptly grew loans 14.33% in 2025 and added another RMB 81.2 billion in the first quarter of 2026 alone, a 7.6% increase in three months.⁸¹

Now the analytical part, because this is the cleanest capital-allocation case study in the story.

Management engineered a capital raise that never looked like one. There was no rights issue, no announced placement at a discount, no headline that read "bank taps shareholders for RMB 15 billion." Instead the share price rose, a bond converted, and the capital ratio went up. Shareholders experienced it as good news.

It was still dilution. Roughly 22% more shares now claim the same earnings stream. The bank issued equity at RMB 11.35 per share — a price set back in 2021 and adjusted for dividends since — which is a materially different thing from issuing equity at the prevailing 2025 market price. Whether that was a good deal for existing holders depends entirely on whether RMB 11.35 was above or below intrinsic value, and reasonable people can disagree.

What is not debatable is the alternative that was foregone. The bank could have run slower loan growth and paid out more of its earnings. It chose to grow the balance sheet and rebuild capital instead. That is a defensible choice for an institution that believes it has an underexploited franchise — Chairman Song's framing is that 10–20% asset growth is what the bank's internal capital generation can support, and that the strategy is to make expansion as capital-light as possible.² But it is a choice with a visible cost, and the shareholders paying it noticed.

There is also a footnote that has not closed. The bank has a separate private placement approved and pending: up to 900 million new shares raising up to RMB 8 billion. The board extended the plan on April 11, 2025 and shareholders approved the extension on June 25, 2025, but as of the half-year briefing in August 2025 it had still not been filed with the Shanghai Stock Exchange.² Wang Xiaoli confirmed the pricing formula — the higher of 80% of the 20-day average price or audited book value per share — meaning it cannot be issued below book.² The instrument's continued existence tells investors something important: management does not regard the capital question as settled.

Which brings us to the people making these decisions.

↑ Back to Top

VIII. Current Management: A Young, Dual-Track Leadership Team

On the morning of April 2, 2025, Bank of Hangzhou filed a short announcement. President Yu Liming had resigned for personal reasons, effective immediately. Chairman Song Jianbin would assume the president's duties in the interim.⁶

Chinese corporate resignation announcements follow a template, and the template includes a paragraph thanking the departing executive for their contributions. This one did not have it.⁶

Yu Liming was born in October 1966 and had spent more than 35 years in finance — roughly two decades at Bank of Communications, then senior roles across investment firms and trusts, including eleven years as chairman of Hangzhou Industrial Trust.⁶ He had joined Bank of Hangzhou as president in December 2022. He lasted two years and three months against a term that ran to July 2026. Press reporting at the time noted, without official confirmation, that his previous employer had accumulated severe real-estate-related losses, with a non-performing ratio reported at 52.77% by 2023 and cumulative losses exceeding RMB 1 billion over three years.⁶ The bank has never explained the departure beyond the two-word formula.

The seat then stayed empty for the better part of nine months. 张精科 Zhang Jingke, born August 1978, was elected president by the board on January 13, 2026, with regulatory approval following in March.¹³³ Unlike Yu, Zhang is an insider: he joined the bank in 2001 and worked his way up through the branch network, including a stint as deputy head of the technology branch and a career weighted toward science-innovation finance.¹³ He had been serving as a deputy president, and it was Zhang who fielded Morgan Stanley's question at the 2025 half-year briefing about whether technology lending actually makes money — the answer quoted in Section V.²

The elevation of a science-innovation specialist to the top operating job is the clearest signal management has sent about where it thinks the franchise's future lies. It is also, for investors, a useful consistency check: the person now running the bank is personally identified with the strategy the bank has been marketing, which makes it harder to quietly abandon and easier to hold to account.

The full team is now structured as one president and six deputy presidents, and for the first time every member of senior management was born in the 1970s, with four born between 1975 and 1980.³¹³ That is a genuine generational handover at an institution where the chairman, born in July 1971, is himself a relatively young executive by Chinese banking standards.

Song Jianbin is the continuity. He holds a doctorate in economics from the Chinese Academy of Social Sciences, joined Bank of Hangzhou in 2002, and rose through roles including deputy president, chief risk officer and chief information officer before becoming president in November 2014 and chairman in September 2022, with regulatory approval in December of that year.¹⁹ Twenty-four years at one institution, including six as its chief risk officer — that biography is consistent with the bank's actual behaviour, which is what an investor should look for.

His public style is unusually discursive for a Chinese bank chairman. Asked at the 2026 briefing to lay out the new five-year plan, he did not read a slide; he talked about the difficulty of pricing transactional assets like discounted bills, which yield less than deposit costs, and argued the bank should stop treating them as profit centres and start treating them as customer infrastructure.¹ Asked about credit growth, he reframed the question as a challenge to management's discipline — whether the team can hold the balance between growth, return and risk when growth slows.² Asked what changes in the new plan, he said the previous cycle's priorities of risk management and technology stay constant, and the new emphasis is on employee development, adding that younger staff have generational expectations his own cohort did not.¹ His summary formulation: commercial banking in the coming decade will test endurance, not explosiveness, and the industry will shrink in volume while improving in quality.¹

That is a coherent, non-promotional worldview, and it is consistent across briefings. It is also worth noting what he does not do: he does not give quantified guidance he can be held to. Asked directly by Industrial Securities for a quantitative medium-term credit growth forecast, he offered that mid-sized banks could plausibly achieve above 10% loan growth over five years "for your reference" — a deliberately loose framing.¹

The dual-track pay problem

Now the governance structure that generates the friction. Song Jianbin, as the designated head of a municipally controlled state enterprise, has his compensation set under state-enterprise pay rules by the municipal authorities rather than by the bank's own remuneration committee. Zhang Jingke, as an executive hired under market terms, does not. The result is arithmetically striking: for the reporting year disclosed in 2026, Zhang's pre-tax pay was RMB 2.538 million while Song's was RMB 762,500 — the chairman earning less than a third of the president he supervises.³

Dual-track pay is common across Chinese state-linked financial institutions and is not itself evidence of wrongdoing. But it creates two real problems. First, it decouples the most senior decision-maker's incentives from shareholder returns entirely; the chairman's pay does not move with the share price, the dividend, or the return on equity. Second, it makes the bank's aggregate management compensation politically awkward to defend, because the number is driven by the market-track executives. Total senior management compensation reached RMB 21.744 million — the highest among Yangtze River Delta city commercial banks — with each of the six deputy presidents paid above RMB 2.4 million.³

Set that against a payout ratio in the mid-twenties and you have the ingredients for a fight. In May 2026, shareholders had one.

↑ Back to Top

IX. The Shareholder Revolt: Pay, Payout, and the 2026 AGM

Dividend policy at Bank of Hangzhou has been the persistent sore point since the day it listed.

Across ten cash distributions from the 2016 IPO through 2024, the payout ratio exceeded 30% exactly once — 31.44% in 2019.¹⁰ The 2023 payout was 21.44%; the 2024 declared payout was 23.5%.¹⁰ Against a Chinese market convention that treats 30% as the threshold for a "quality" bank dividend, and against a trailing dividend yield of 4.28% that ranked 30th out of 42 listed A-share banks in mid-2025, the bank has been a consistent laggard on distribution despite being a consistent leader on earnings growth.¹⁰

Management has a real counter-argument, and it should be stated fairly. A low payout ratio on a rapidly compounding earnings base can still deliver strong absolute dividend growth. Wang Xiaoli made exactly this case at the 2025 half-year briefing, and with a nice detail: when the convertible bond expanded the share count by 22%, the bank did not cut the dividend per share to hold the total constant. It kept dividend per share unchanged and let the total rise — RMB 4.25 billion paid for 2024, up 38% year on year — explicitly on the principle that new and old shareholders should share equal returns.² That is a shareholder-friendly decision that cost the bank real money, and it is why the 2024 payout that was declared at 23.5% ended up landing closer to 25% of net profit.

The bank also committed in an April 2025 valuation-enhancement plan to distribute cash dividends twice a year in principle, and shareholders authorized the board in June 2025 to determine a 2025 interim distribution.² For 2025, the total proposed distribution reached RMB 4.784 billion, a payout in the mid-twenties — around 25% on one calculation and 26.43% on another.⁸³

It was not enough.

At the annual general meeting in May 2026 — the bank's thirtieth year — two resolutions concerning executive compensation, the pay-management framework and the 2026 director compensation plan, each drew roughly 120 million shares in opposition, about 9.5% of the votes cast by shareholders holding less than 5% of the company.³ Both resolutions passed. In an A-share bank whose register is dominated by municipal platforms that vote with management, minority opposition on that scale is not a rounding error; it is a coordinated statement.

What were they objecting to? The juxtaposition, not the absolute numbers. Roughly RMB 21.7 million of senior management compensation, the highest in the peer group, paid in a year when the bank distributed roughly a quarter of its profits, revenue grew 1.09%, and — as the next section shows — the composition of that profit had become considerably less flattering than the headline.³⁸

This is the activist stress test, and it is worth running properly. A skeptical long-short investor looking at Bank of Hangzhou would build the case around four points, all of them supported by disclosure rather than inference. One: the company has raised equity capital twice in five years, once through the convertible bond and once pending through a private placement, while paying out less than peers — meaning shareholders have funded growth twice over. Two: management pay is peer-leading while shareholder distributions are peer-lagging, and the board's remuneration committee does not even set the chairman's pay, so there is no single body accountable for the total. Three: a president departed abruptly with no explanation and the seat sat empty for nine months, during which the chairman held both roles. Four: the largest shareholders are municipal platforms that are also counterparties in the loan book, which means the register that would normally discipline management on pay and payout has an interest in the bank's continued cooperation on credit.

None of these is a scandal. Collectively they describe a governance structure in which minority shareholders have limited leverage and have started using the only one they have — the protest vote. Whether management responds is one of the more informative things to watch over the next two AGMs.

And the reason the pay question became flammable in 2026 rather than 2023 is that the earnings underneath it stopped telling a simple story.

↑ Back to Top

X. The Profit-Quality Problem: Where the Growth Engines Stalled

Start with the gap, because the gap is the entire issue.

In 2024, revenue grew 9.61% and net profit grew 18.07%.⁹ In 2025, revenue grew 1.09% and net profit grew 12.05%.⁸ In the first quarter of 2026, revenue grew 4.29% and net profit grew 10.09%.⁷ Three consecutive periods in which profit grew roughly two to eleven times faster than the revenue that is supposed to produce it.

There are only three places that gap can come from in a bank: operating costs, credit provisions, or one-off gains. Bank of Hangzhou's cost base is not shrinking dramatically. So the answer is provisions — and the disclosure is explicit. The bank's credit impairment charge fell 38.47% in 2025.¹³ Provision coverage dropped from 541.45% to 502.24%, a decline of 39.21 percentage points.⁹⁸ In the first quarter of 2026 the provisioning charge fell a further 51.5% and coverage fell again to 481.39%.⁷

This is not fraud and it is not even aggressive by the standards of the industry. A bank carrying five times as much reserve as it has bad loans has an enormous cushion, and drawing it down when credit costs are genuinely low is defensible accounting. But it is important to be precise about what it means: Bank of Hangzhou's reported profit growth in 2025 was substantially a decision, not an outcome. Management chose to release reserves built in prior years. That is a lever with a finite number of pulls.

The revenue side explains why the lever was needed. Net interest income grew 12.82% in 2025 to RMB 27.59 billion and fee income grew 13.10% to RMB 4.21 billion — both genuinely strong.⁸ But other non-interest income, which is overwhelmingly bond trading and fair-value gains, fell 31.40% to RMB 7.00 billion, dragging total non-interest income down 19.51% and holding group revenue essentially flat.⁸

Note what that means, because it is the opposite of the common assumption. In 2025, investment gains did not prop up Bank of Hangzhou's revenue — they sank it. The core banking business grew at double digits and a shrinking bond-trading contribution absorbed the entire benefit. The 2024 comparison base had been inflated by an exceptional bond rally, and when Chinese government bond yields backed up, that income evaporated.

Management's own account is consistent with this and, to its credit, was not evasive. Zhang Jianfu told the 2026 briefing that bond-market rate volatility and a high prior-year base compressed current revenue growth, and argued the effect washes out across accounting periods.¹ Wang Xiaoli disclosed the portfolio surgery undertaken in response: the bank cut the share of fair-value-through-profit-and-loss assets by nearly 5 percentage points in the first half of 2025, reduced that book's interest-rate sensitivity by RMB 15 million of PV01, shifted new bond purchases into the fair-value-through-OCI account, and let the amortized-cost book run down as reinvestment yields fell.²¹ Her characterization of the 2026 environment — low rates, high volatility, wide range-trading, with renminbi bond investment reverting to a liquidity-management function rather than a profit centre — is a clear-eyed statement that this income line is not coming back soon.¹

So the reported picture is: strong core lending revenue, collapsing trading revenue, and provisioning releases bridging the gap to a double-digit profit number.

The engines that were supposed to be running

Now the uncomfortable part.

Under the 2021–2025 five-year plan, retail finance and small-micro lending were designated the bank's growth engines. In 2024, retail banking segment profit fell 44.8% and small-enterprise banking profit fell 47.4%.¹¹ Both segments continued to receive substantial credit allocation while their profitability halved. Corporate banking — the business that was supposed to be the mature base, not the engine — produced roughly ten times the profit of either.¹¹

The credit metrics moved the same way. Individual loan NPL ratios rose 0.18 percentage points across mortgages, personal operating loans and consumer credit in 2024, and a deputy president acknowledged that mortgage defaults among certain customer segments had been increasing since the first quarter of 2025.¹¹ By the end of 2025 the individual loan NPL ratio had reached 1.02%, and the individual loan balance had actually contracted 3.24% — the bank shrinking a book it had called a growth engine.¹³

Management has not disputed the trend. Deputy president 潘华富 Pan Huafu told the 2026 briefing that corporate NPL formation remained stable and low while small-micro and retail formation rose, and then said the thing analysts most needed to hear: retail NPL exposure has not reached its inflection point. Even with group-wide NPL formation falling to an annualized 0.6% in the first quarter of 2026, he said the large-retail risk profile was essentially unchanged from 2025 and that a genuine turning point depends on household employment and income expectations recovering and on the property market stabilizing.¹ Consumer lending and online lending he expected to hold steady or improve; personal operating loans and mortgages he expected to stay under pressure.¹

That is an unusually direct admission, and it is worth crediting. It is also an admission that the bank does not control the variable that determines when this improves.

Chen Lan's account of the tactical response is granular and revealing about how hard the segment has become. In the first half of 2025 the bank disbursed RMB 12.7 billion of mortgages for a net balance increase of just RMB 4.3 billion — most new lending simply replacing repayments.² New mortgage pricing hit a ten-year low.² By 2025, all six large state banks reported outright declines in their personal mortgage balances, which tells you the pool is shrinking industry-wide.¹ In inclusive small-micro lending, the balance reached RMB 167.6 billion by mid-2025, up 7.7% year-to-date, but at an average new-loan rate of 4.22%, down year on year.² The one genuinely encouraging data point: newly originated small-micro credit of RMB 1.9 billion carried a yield above 7% with an NPL ratio of 0.56%.² Small volumes, but evidence that selective underwriting at attractive pricing is still possible.

Management's diagnosis of why small-micro deteriorated is the correct one and matters more than the numbers: weak end-demand, compressed margins at the borrower level, and falling collateral values, hitting simultaneously.¹¹ Those are not underwriting errors. They are what happens to small business lending when nominal growth slows.

The synthesis

Put the pieces together and the case looks like this. Bank of Hangzhou's core lending franchise is performing well — net interest income up 12.82% in a year when the industry's margin fell again is a strong result, and first-quarter 2026 loan growth of RMB 81.2 billion at a new-loan rate of 3.27% with margins recovering 6 basis points suggests the corporate engine is intact.⁸¹ But the two businesses the bank told investors would carry it into its next phase have gone backwards, both in profit and in credit quality, and the reported headline has been smoothed by reserve releases and depressed by trading losses in ways that make the underlying trend hard to see.

The bank now enters a new five-year plan — Song Jianbin's "三三六六" framework, built around a three-dimensional objective of customers, returns and scale, and three directional shifts toward transaction-flow banking, digital-and-intelligent operations, and internationalization.¹ The strategic language has changed. Whether the segment economics change with it is the question that determines what this business is worth.

↑ Back to Top

XI. Risk Radar

Risk lists are usually generic. This one is not, because Bank of Hangzhou's risks are unusually specific and mostly already visible in the disclosures rather than hypothetical.

Retail and small-micro credit deterioration is the live risk, and management says so. The mechanism was laid out above, but the thing to internalize is the sequencing. Retail bad loans in China surface slowly: a borrower's income falls, they draw on savings, they refinance, and only then do they default. Bank of Hangzhou is early in that sequence, not late — the individual loan NPL ratio has roughly one-third more to travel before it reaches the level at which most peers already sit, and the bank's own head of risk declined to call a turning point.¹ The offsetting factor is scale: retail is a minority of the loan book and produced under RMB 1 billion of segment profit in 2024, so even a severe retail outcome is survivable.¹¹ The real cost is not credit losses; it is that the growth engine stays broken.

Policy-and-government finance concentration is the largest structural exposure, and the hardest to independently assess. Infrastructure and urban-construction lending is around 44% of the book.¹¹ The counterparties are municipal state-owned enterprises in Zhejiang, and several of the bank's largest shareholders are exactly such entities. Management stated that this business currently carries no risk assets and that the main risk is policy-compliance rather than credit, describing the residual stress in the sector as platforms caught between insufficient marketization and expectations of implicit debt repayment.¹ The pipeline they describe is real — Zhejiang's "千项万亿" major-project programme, special-purpose bond co-financing where the bond funds only about half a project's cost, and the securitization of operating assets held by local SOEs.¹ But investors should be clear about the structure of the bet: these loans perform because local government fiscal capacity supports them, and Zhejiang's fiscal capacity is among the strongest in China. If that changes, the bank takes the hit on the asset side while its largest shareholders take it on their own balance sheets simultaneously. The exposures are correlated by construction.

Net interest margin compression is chronic rather than acute. The margin recovered 6 basis points in the first quarter of 2026, but the driver was deposit costs falling 24 basis points against loan yields falling 22 — a symmetrical repricing where the liability side simply moved faster.¹ Deposit repricing benefits are finite; loan yield pressure is not. Zhang Jianfu's claim that the bank's superior asset quality and provisioning give it a 30–40 basis point margin buffer versus the market average is plausible but unaudited, and it is a buffer against a lower level, not against further decline.¹

Real estate concentration inside an otherwise clean book. Corporate real-estate NPLs at 6.65% in 2024 against a blended 0.76% is the clearest illustration in the disclosures that this bank's risk is not evenly distributed.¹⁶ It is contained because the exposure is small. It is a reminder that "0.76%" is a portfolio-weighted average and tells you nothing about any individual segment.

Capital remains a ceiling, not a solved problem. Core Tier 1 at 9.59% is comfortable rather than abundant for a bank growing risk-weighted assets at low-to-mid teens.⁸ The pending RMB 8 billion private placement has been kept alive for years without being filed.² Management's stated intention is to shift toward capital-light growth — flow business, settlement, custody, wealth management, bond underwriting — which is strategically correct but slow. If loan growth continues at 14%, another capital event is a question of timing.

Governance and succession uncertainty is a discount factor rather than an event risk. An unexplained president departure, a nine-month vacancy, a dual-track pay structure that leaves the chairman's compensation outside board control, and a documented minority-shareholder protest are not individually disqualifying. Together they describe an institution where the accountability chain to outside shareholders is weak.

Technology and competitive disruption sits in an unusual place here. The obvious framing — fintech lenders eating bank lunch — is largely stale in China, where regulators reined in 蚂蚁集团 Ant Group and 微众银行 WeBank-style consumer credit several years ago and capped what non-banks may do. The live version of the risk is different and runs the other way: the largest state banks have industrialized digital small-business lending to a degree that erodes the informational advantage a relationship lender used to hold. When a national bank can price a small-business loan off tax records, invoicing data and payment flows in minutes, the value of a loan officer knowing the borrower's cousin declines. Bank of Hangzhou's response is visible in its disclosures — data feeding risk approval and early-warning models, AI assistants in development, and a "iron triangle" front-line structure pairing relationship, risk and product managers on each client.¹² Whether that preserves the edge or merely slows its erosion is not yet answerable.

Second-layer items worth monitoring without overweighting. Reporting in 2025 flagged repeated foreign-exchange compliance penalties alongside the bank's infrastructure concentration — small in financial terms, but a recurring compliance pattern is a management-quality signal. The bank has continued issuing perpetual capital bonds to support Tier 1, with credit rating documentation available through the interbank market — routine for the sector, but it confirms the bank is a habitual user of the capital markets rather than a self-funding compounder. And the ownership shift itself has consequences for index and institutional flows: with Xinhua Insurance and other domestic insurers now anchoring the register, the shareholder base is more sensitive to insurance-sector regulatory capital rules than it was under a foreign strategic holder.

Two of these — retail credit and capital — are things management can influence. The others are structural features of being a Zhejiang city commercial bank in 2026. Which sets up the final argument.

↑ Back to Top

XII. Bull vs. Bear

The bull case

The bull case does not require heroic assumptions, which is its main strength.

Start with the asset quality, because it is the one claim that has been proven rather than asserted. Thirteen straight quarters at 0.76%, provision coverage above 500%, and net NPL formation of 0.72% in 2025 falling to an annualized 0.6% in early 2026 is not a snapshot — it is a multi-year track record spanning the worst credit environment Chinese banking has faced in three decades.⁸¹ Through the property developer defaults, the local government debt restructuring, and a consumer slowdown, this bank did not blink. The internal explanation is a phrase management repeats across every briefing — refusing to trade risk for growth — backed by a normalized risk-screening and portfolio-adjustment mechanism the bank says it has run for more than a decade, and extended in 2026 with head-office direct escalation for large exposures.¹²

Geography is the second pillar. Zhejiang has the most vigorous private economy in China, the deepest household wealth accumulation outside the top-tier metros, and — through Hangzhou's hard-tech cluster — a genuine claim on the country's most valuable emerging industries. Chen Lan's framing of the wealth-management opportunity rests on this: high-net-worth density built from company founders and small business owners, in a province where the bank already has complete branch coverage plus outposts in Beijing, Shanghai, Shenzhen, Nanjing and Hefei.¹

Third, the science-innovation franchise is a real differentiator, whatever one thinks of its durability. Fifteen years, 28,000-plus technology clients, RMB 115 billion of technology loans, a divisional structure with embedded credit approval, and a 1:1 loan-to-demand-deposit relationship that converts lending into cheap funding.¹⁴² Serving four of Hangzhou's six most famous startups is a marketing fact, but the ecosystem position underneath it is not.¹

Fourth, capital allocation on the convertible bond was, in execution, about as shareholder-friendly as a RMB 15 billion equity raise can be — no discounted placement, no forced rights issue, and a decision to hold dividend per share constant through a 22% share count increase rather than diluting the payment.²

Fifth, the state-anchored register, viewed from the bull side, is stability. There is no overhang of a foreign holder waiting to sell, no realistic scenario in which control changes hands, and privileged access to the fiscal ecosystem of one of China's richest municipalities.

And sixth, the operating momentum entering 2026 was better than 2025: revenue growth back to 4.29%, margin up 6 basis points, loans ahead of budget, consumer loan balances up 10.73% in a quarter, and new consumer-loan customer signings more than double the prior year.⁷¹

The bear case

The bear case is not that this is a bad bank. It is that the reported numbers describe a better bank than the underlying trend does.

Earnings quality is the core of it. Profit grew 12.05% on revenue growth of 1.09% because the credit impairment charge fell 38.47% and provision coverage came down nearly 40 percentage points.⁸¹³ A bank can do that for two or three years. It cannot do it indefinitely, and every year of doing it lowers the cushion available for the retail credit cycle that management has said has not yet turned.¹ The bear does not need to allege anything improper; the bear only needs to point out that the trajectory of coverage — 541%, 502%, 481% across five quarters — is the trajectory of a buffer being spent.⁹⁸⁷

The strategic failure is the second leg. Two segments were designated growth engines for a five-year plan. At the end of that plan, retail segment profit had fallen 44.8%, small-micro profit had fallen 47.4%, the individual loan book had shrunk 3.24%, and individual loan NPLs had risen to 1.02%.¹¹¹³ The bank's response has been to write a new five-year plan with new language. An investor is entitled to ask what specifically will be different, and the answer offered so far — flow banking, digitalization, internationalization, better employee development — is directional rather than operational.

The distribution record is the third. Below 30% payout in every year but one since listing, a dividend yield ranking in the bottom third of listed A-share banks, peer-leading management compensation, and a minority protest vote that management has not yet publicly answered.¹⁰³ The bull says growth justifies retention. The bear says shareholders have now funded a RMB 15 billion convertible conversion, face a pending RMB 8 billion placement, and are being asked to fund a third round of growth in segments that just halved in profitability.

The fourth leg is governance opacity. A president left mid-term with no explanation, in the same window that a twenty-year anchor shareholder exited completely.⁶⁵ Neither event has been connected publicly and it would be irresponsible to assert a link. But the pattern — unexplained senior departure, extended vacancy, no narrative from the board — is the sort of thing that reduces the multiple an outside investor should be willing to pay, not because of what is known but because of what is not disclosed.

And the fifth is correlation. This is a single-province bank whose largest borrowers include entities controlled by its largest shareholders, in an economy whose fiscal strength has never been tested by a genuine downturn. The bank has no overseas branches at all — management confirmed it serves clients going abroad only through domestic comprehensive services.¹ There is no geographic diversification to fall back on.

Reconciling the two

The competitive framework helps arbitrate. Porter says this is a low-power industry: brutal rivalry, rising buyer power, structural substitution through bond markets, and no exit. In such an industry the only sustainable advantages are cost position and risk selection, because pricing power is unavailable to everyone. Bank of Hangzhou has demonstrably achieved superior risk selection — that is what 0.76% across thirteen quarters means — and has a credible cost-of-funds edge inside its technology franchise. That is a genuine, if narrow, source of excess return.

The 7 Powers reading says the same thing more sharply: the bank's power is counter-positioning plus switching costs inside one specialized niche, sitting on top of an otherwise undifferentiated commercial banking business. The niche is real. It is also small relative to the RMB 1.07 trillion loan book, and it does not currently produce pricing power.

So the honest synthesis is that Bank of Hangzhou is a well-underwritten bank with one genuine specialty, operating in a structurally unattractive industry, in a strong but concentrated economy, under a governance structure that gives outside shareholders limited voice — and reporting profit growth that is currently flattered by reserve releases while its designated growth businesses shrink.

Whether that is attractive depends almost entirely on whether the retail and small-micro segments recover, because that determines whether the reserve-release bridge leads somewhere or simply runs out.

↑ Back to Top

XIII. Durable Business & Investing Lessons

Decompose bank profit growth before believing it. Bank earnings have three engines — net interest income, non-interest and trading income, and the provisioning charge — and only the first is a recurring operating result. Bank of Hangzhou's 2025 is an almost laboratory-grade case: strong core income (net interest up 12.82%, fees up 13.10%), collapsing trading income (other non-interest down 31.40%), and a provisioning charge cut 38.47% that produced the double-digit headline.⁸¹³ Any of those three could have been the story. Only by separating them can an investor tell that the core lending business was actually performing well while the reported revenue line looked terrible, and that the reported profit line looked better than the sustainable earning power.

A convertible bond that force-converts during a rally is a capital raise wearing a disguise. The mechanics are worth learning because Chinese banks use this instrument constantly. The bank issued RMB 15 billion in 2021, watched it sit largely unconverted for four years while it was of no use as core capital, was reduced to asking major shareholders for help in April 2025, and was then rescued by a share-price rally that triggered mandatory redemption and forced 99.39% conversion within weeks.¹⁸¹² Shareholders experienced a rising share price and an improving capital ratio and felt good about both. They were simultaneously diluted by 22%.¹² The lesson generalizes: when a capital constraint disappears without a visible capital raise, find the instrument that did the work.

Relationship-based, soft-information lending can be a real moat, but it is unverifiable from outside and must be evidenced over cycles. The Bank of Hangzhou science-innovation model has a plausible mechanism, a fifteen-year history, an organizational structure that matches the claim, and a specific economic explanation for why it is profitable despite unattractive loan pricing. What it does not have is published segment-level default and return data, or a completed downturn in its client base. Investors should hold this kind of thesis provisionally and demand the data — portfolio yields, NPL formation within the segment, deposit balances per client — rather than accept the narrative.

Low payout combined with peer-leading management pay is a reliable trust-eroding combination. It does not matter whether the retention is justified; the optics compound. The informative variable is not the protest vote itself but what management does next. A board that responds with a higher payout, a clearer capital plan, or an explanation has one kind of relationship with its minority holders. A board that lets the resolutions pass and moves on has another.

Swapping a foreign strategic anchor for a domestic state shareholder base is a regime change in governance culture, even when it happens quietly. It happened here across four years and two transactions, was reported as a series of block trades, and never produced a single day of dramatic news. But the institution that emerged has a fundamentally different accountability structure than the one that existed in 2020. When a long-standing strategic shareholder starts selling, the question to ask is not what price they got. It is what disappears from the boardroom when they leave.

↑ Back to Top

XIV. What to Watch

Three metrics carry most of the information about this business, and one of them is not on the front page of the results release.

First: the retail and small-micro NPL formation rate. Not the headline 0.76% NPL ratio, which is a stock measure that provisioning and write-offs can manage, but the flow — how much new bad debt the retail and small-micro books generate each period. Management disclosed group-wide net formation of 0.72% for 2025 and an annualized 0.6% in the first quarter of 2026, and explicitly said corporate formation is low and stable while retail and small-micro formation rose.¹ The bank does not routinely publish the segment split, which is itself worth noting. This is the number that determines whether the credit cycle turns before or after the provisioning cushion runs down.

Second: the provision coverage ratio. It has fallen from 541.45% at end-2024 to 502.24% at end-2025 to 481.39% at the first quarter of 2026.⁹⁸⁷ Read it as the fuel gauge on reported profit growth. As long as it keeps falling, a portion of each period's earnings growth is a transfer from the balance sheet rather than a result from operations. When it stabilizes, reported profit growth will converge toward revenue growth — and that convergence, whenever it comes, will reveal the bank's true earning power.

Third: the gap between revenue growth and net interest income growth. In 2025 that gap was enormous — 1.09% against 12.82% — and it was entirely bond trading.⁸ Watching whether the gap narrows tells an investor whether the balance-sheet restructuring management described, shifting from fair-value-through-profit-and-loss into OCI accounts and shortening duration, has actually removed the volatility from the revenue line.²¹

Deliberately excluded from that list: the NPL ratio itself, which is too managed to be informative at this level of provisioning; total asset growth, which management can produce at will and has told investors it intends to hold in the 10–20% range; and net interest margin in isolation, which is an industry variable more than a company one.²

Beyond the metrics, three narrative questions decide the outcome. Does the core Tier 1 ratio hold near 9.5% or does the pending RMB 8 billion placement finally get filed? Does the payout ratio move at all in response to the 2026 protest vote? And does the new five-year plan produce a visible change in retail and small-micro segment profitability, or does the bank quietly redesignate its growth engines again?

↑ Back to Top

XV. Epilogue & Close

Thirty years after 33 credit cooperatives were stitched together in a provincial capital, Bank of Hangzhou has become something none of the original participants would recognize: a RMB 2.4 trillion institution with the cleanest reported loan book in Chinese banking, a specialized franchise lending to robotics and AI companies, a wealth-management subsidiary running half a trillion renminbi, and a shareholder register that reads like a directory of the Hangzhou municipal government.

Where the story stands in mid-2026 is genuinely mixed, and the temptation to resolve it in one direction should be resisted. The credit underwriting is excellent and proven — thirteen quarters of evidence through the worst possible test conditions is not luck, and it is the hardest thing in banking to fake for that long. The core lending franchise is growing faster than the industry, funding itself increasingly well, and finally showing margin stabilization. The management team is young, technically credible, unusually candid in public about what it does not control, and now led operationally by someone who built the differentiated business himself.

Against that: the two businesses this bank told the market would carry it forward went backwards over the plan period; the profit line has been supported by reserve releases that cannot repeat forever; a president left without explanation and the board never provided one; minority shareholders have started voting against management on pay; and the ownership structure that replaced twenty years of foreign board-level scrutiny is composed largely of the bank's own borrowers.

The real test over the next few years is not credit quality. That question has been answered about as well as it can be. The test is whether a bank that has proven it can avoid losses can also prove it can generate growth — whether the science-innovation niche can be scaled into something that moves a trillion-yuan loan book rather than decorating it, whether retail and small-micro can be made to work in an economy where household balance sheets are still repairing, and whether reported profits can grow without help from the provisioning account.

And underneath all of that sits the question that Commonwealth Bank of Australia answered with its feet and that 120 million shares echoed at the 2026 annual meeting: whether the people who own this bank from the outside have any reason to believe management is working for them.

References

1. 杭州银行股份有限公司2025年度暨2026年第一季度业绩说明会问答纪要 — 上海证券交易所披露, 2026-05-06 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

2. 杭州银行2025年半年度业绩说明会问答实录 — 上证e互动, 2025-08-29 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

3. 杭州银行30周年高管变阵，薪酬方案遭1.2亿股反对 — 界面新闻, 2026-05 ↩↩↩↩↩↩↩↩↩

4. 20年赚了125亿，外资股东高位清仓杭州银行 — 界面新闻, 2025 ↩↩↩↩↩

5. 43亿，杭州银行遭澳洲联邦银行"清仓"离场 — 新浪财经, 2025-01-24 ↩↩↩↩

6. 2.2万亿杭州银行行长虞利明"中途退场"，究竟有何隐情 — 界面新闻, 2025-04 ↩↩↩↩↩↩

7. 杭州银行营收增1.09%创十年低，拨备压降26%助推净利涨12% — 新浪财经, 2026-04-28 ↩↩↩↩↩↩

8. 年赚190亿、分红近48亿！非息收入承压，杭州银行营收结构分化 — 新浪财经, 2026-04-23 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

9. 杭州银行2024年年报解读：营收净利双增，多项指标现积极变化 — 新浪财经, 2025-04-11 ↩↩↩↩↩↩↩

10. 净利润增速领跑银行股，杭州银行分红还是不到30% — 新浪财经, 2025-05-07 ↩↩↩↩

11. 新增长点乏力？杭州银行零售利润大跌，中小微风险压力增大 — 南方都市报, 2025-04-18 ↩↩↩↩↩↩↩↩↩↩

12. 杭银转债今日摘牌 6只银行转债年内到期 — 新浪财经, 2025-07-07 ↩↩↩↩↩

13. 杭州银行迎"75后"新行长！2025年营收增长仅1% — 东方财富, 2026-02-02 ↩↩↩↩↩↩↩↩↩↩

14. 杭州银行2024年营收、归母净利双增，理财公司存续产品规模超4300亿元 — 北京商报, 2025-04-11 ↩↩↩↩

15. 去年房地产不良贷款增逾七成，杭州银行23%的净利增幅下有何隐忧？ — 证券之星, 2024-04-28 ↩

16. 杭州银行去年净利增幅降至18%，房地产业不良贷款率升至6.65% — 澎湃新闻, 2025 ↩↩↩

17. 北京银行居首，5家银行资产规模突破3万亿元 — 新浪财经, 2026-05-08 ↩↩

18. 直击杭州银行业绩会：正积极寻求大股东可转债转股等方面支持 — 新浪财经, 2025-04-17 ↩↩↩

19. 杭州银行董事长任职资格获批，70后"老杭银人"宋剑斌正式掌舵 — 界面新闻 ↩

20. 杭州银行首次覆盖报告 — 东方证券, 2019-03-05 ↩↩↩

21. 股价走高触发强赎 银行可转债接连触发强赎 核心资本充足率获支撑 — 新浪财经, 2025-06-01 ↩