UOB: the Street is provisioning for a cycle that already happened
Consensus has FY26 credit cost at 32 bps. The book that could justify that number was written down in 2023 and 2024. I have 27 bps, and the difference is worth roughly 6% of earnings.
Sample note with illustrative figures. This file is the template: copy it, keep the seven headings in this order, and replace the content.
Thesis in three sentences
UOB’s Citi-ASEAN consumer book has finished seasoning, and the specific provisions that went with it are rolling off faster than the Street’s model assumes. Margin is going to be worse than consensus, but the credit line is a bigger number and it moves the other way. The stock is priced for a mediocre bank at 1.2x book; it is a mediocre bank with a one-off provision tailwind that lasts three more halves.
The three drivers
| Driver | My FY26 | Consensus | Why it matters |
|---|---|---|---|
| Credit cost | 27 bps | 32 bps | Decides the stock. Each 5 bps is about 4% of FY26 net profit. |
| Net interest margin | 1.94% | 1.99% | I am below the Street. Deposit repricing lags the cut cycle by two quarters. |
| Fee income growth | +8% | +8% | In line. Not part of the view; included so the reader knows I checked. |
What the price already embeds
At S$35.10 the stock discounts a sustainable ROE of about 12.5% against a 9% cost of equity. That is consistent with consensus credit cost and consensus margin. If my credit line is right and my margin line is also right, the netted ROE is 13.4%, which on the same framework is worth S$38 to S$39.
Bear case
A second wave of Thai and Indonesian unsecured stress would push credit cost back above 35 bps and the whole argument disappears. The 1H26 disclosure on stage 2 loans in Thailand is the number to read first.
What makes me wrong
Two consecutive halves of credit cost above 30 bps, or a management guide above 30 bps for FY27. Either one closes the call as wrong regardless of where the share price is.
Who is on the other side
The marginal seller is foreign long-only money trimming Singapore banks as the rate cycle turns, and it sells all three names together. That is why this is a pair against DBS rather than an outright long: the flow that hurts UOB hurts DBS at least as much, and DBS has the more expensive margin assumption.
Size and horizon
3% of a model portfolio, long UOB against short DBS, twelve months, reviewed at each results date. The pair is sized so that a 5 bps miss on credit cost costs about 25 bps of portfolio.
Data integrity notes
Credit cost is specific plus general allowances over average gross loans, annualised, from the 1H26 results pack. Consensus is the Bloomberg mean as at 11 August 2026. Share count is the diluted figure from the same pack.