CMB hits its NIM target and refuses to celebrate it

The quarter the poetry turned into plumbing

For two years, China Merchants Bank sold a story: the intelligent bank, the AI-first moat, the 1,500 identified work items and the Golden Sunflower assistant. In the first-quarter 2026 call, large model was the single highest-ranked keyword in CMB's own transcript. It is absent from this quarter's top thirty entirely. So is Intelligent Bank, so is AI model. In their place: loan pricing, liability management, policy rate, base rate, asset yield, liability cost. The vocabulary of a trading desk, not a technology conference.

Not because management stopped being asked. JPMorgan spent a full question on AI deployment and got a careful answer that AI is no longer supporting isolated tasks but reshaping full workflows — “First, AI is no longer supporting isolated task is reshaping our full workflows before it helps our staff find information, organized documents, write reports” — Unknown Executive, Various senior executives including General Manager from Financial Accounting Department, Risk Management Department, Digital Finance Development Office, Corporate Banking business, etc. · 2026-11-01 — plus a 13% lift in average in-person relationship-manager visits. But the balance of the call moved. When a bank's own language shifts from ambition to arithmetic, the arithmetic has become the binding constraint.

Delivered the goal, declined the credit

Peng Jiawen opened with the number everyone wanted: “the NIM for the first 3 quarters was 1.82%, down 5 basis points year-on-year, flat versus the first half” — Jiawen Peng, Senior Executive (likely General Manager or similar senior role) · 2026-11-01 — and, in Q3 alone, flat quarter-on-quarter.

That is precisely the outcome CMB asked the market to hold it to. Back in March, Peng set three targets and said of NIM stability: “We wish that we could achieve this goal in the second half of the year.” — Jiawen Peng, Senior Management · 2026-03-30 By the August call he was more expansive still — “the most difficult time for NIM decline has already passed” — Jiawen Peng, Senior Management (likely CFO or similar) · 2026-08-30 — with a caveat that the industry would still face pressure.

So: target met, tone reversed. Asked directly whether 1.82% is now a stable band, Li Li of Asset & Liability Management answered: “we cannot say it has already been stabilized. There are still uncertainties” — Li Li, Asset and Liability Department · 2026-11-01. She then listed them with unusual precision — a possible LPR cut, the policy benchmark shift pressuring market rates, new bond yields below the existing book, and the structural fact that CMB's deposit cost has less room left than peers' because it started so low. Asset yield fell 28bp year-on-year to 2.80%; liability cost fell 27bp to 1.04%; average customer deposit cost 0.95%. When both sides fall at the same speed, a flat margin is a victory of management, not of the environment — and management knows the next 27bp of liability relief will be harder to find than the last.

DR001: a genuinely new anchor for Chinese loan pricing

The most important thing said on this call was buried in the third question. The PBOC is reforming its policy-rate framework, moving toward DR001 as the anchor and away from a single LPR-plus-spread convention. CMB is not observing this — it is participating. Management disclosed that in July it launched its first loan priced off DR001, that more than 20 such businesses have now been implemented, and that it has issued its first floating-rate bond anchored to the same rate.

The mechanics matter. Large corporates migrate to a diversified base rate anchored on DR001; retail and smaller clients stay on LPR. DR001 sits below LPR today, so the near-term effect is downward pressure on corporate loan pricing — but only for the slice of the book that reprices, and only gradually. What CMB gets in exchange is a market-linked asset base it can hedge, and an explicit seat in the design of China's rate liberalisation. That is a multi-year structural story, and it is why DR001 is the fastest-rising new keyword in CMB's own momentum history this quarter. It is also the theme most likely to be misread as noise.

The retail core shrinks; the corporate book carries the quarter

The uncomfortable arithmetic: retail loans fell 1.28% year-to-date to CNY 3.67 trillion, while corporate loans rose 10.59% to CNY 3.56 trillion and financial investments grew 9.19% to CNY 4.48 trillion. The pivot is visible in the keywords too — retail side and loan side both spike in this transcript, as analysts and management circle the same structural turn. Wang Xiaoqing was blunt in August that this pressure hits CMB harder than peers, given a retail book above half of all loans and less room on funding cost than rivals; nothing this quarter contradicts that. The leading position in retail is now measured in AUM and customers — CNY 18.71 trillion, up 9.52%, and 236 million retail clients — rather than in credit growth.

Asset quality is where the trade-off shows up. NPL held at 0.94% but NPL formation rose CNY 8.16 billion year-on-year to CNY 56.16 billion, an annualised formation ratio of 1.07%, up 11bp. Retail NPL ran 1.19%, up 11bp, while property-development NPL fell 75bp to 3.89% and allowance coverage eased 2.4 points to 389.39%. A credit cost of 0.72%, up 5bp, is essentially the whole explanation for net profit growing 2.38% against revenue growth of 5.37%. Hui Liu argued the marginal trend is improving — “the increment of NPL formation is decreasing. As compared with 2025, we are seeing a slowdown in risk accumulation” — Hui Liu, General Office of Retail Finance · 2026-11-01 — while conceding that special-mention loans rose on stricter classification and negotiated instalment rules, not on new rot.

Capital is the other half of that argument, and it connects CMB to a theme the whole market was chewing on this quarter. CET1 of 14.38% under the advanced measurement approach rose 0.22 points from year-end 2025. The IRB model was the market's single top-ranked keyword of the period — a Nordic peer spent its own call on a new corporate IRB model and on risk weights before the output floor — and CMB's local version of that conversation is that heavier corporate risk weights, as the loan mix tilts away from retail, eat into the endogenous capital engine that funds its dividend.

Which brings the payout question, asked once more and answered with a framework that is rhetorically fresh even if the substance is not:

I call it 3R balance. One is the ROAE. The second one is the CAR ratio. The third one is the dividend payout ratio... It should be a balance and check balance among this.

Read that as a soft no to a near-term increase, dressed as governance. A 13.49% ROAE and a payout already lifted from 33% to roughly 35% are presented as a triangle to be balanced, not a lever to be pulled.

Where this sits in the wider tape

CMB is not alone in talking about deposit franchises. The market's own top keywords this quarter are laced with core deposits, cost of deposits and lending rates, and a Korean bank reporting the same week framed its quarter around funding rate and lending rate compression. What differentiates CMB is the starting point: a 0.95% deposit cost and a 49.75% demand-deposit share of daily average balances. That is arguably the best funding franchise in Chinese banking and simultaneously the smallest remaining cushion — the paradox Li Li spent an entire answer explaining.

So the story of this report is a bank executing well against a target it no longer wants to celebrate, buying time on the margin while it re-wires the plumbing underneath: a new policy-rate anchor, a corporate-led balance sheet, and a retail franchise that grows in customers and AUM but shrinks in credit. The AI narrative went quiet not because it failed, but because it stopped being the thing that moves the numbers. At a 1.82% net interest margin, the seminar is over and the arithmetic has begun.