TBC Bank: The Regulator Blinked, the Collections Reform Didn't Land, and the Loan Book Still Hasn't
TBC Bank: The Regulator Blinked, the Collections Reform Didn't Land, and the Loan Book Still Hasn't
Every quarter for the last two years, TBC Bank Group has come to market with the same shape of story: Georgia prints exceptional returns, Uzbekistan is being "recalibrated" against a moving regulatory target, and the analyst community asks the same question in a slightly different way — when does the fog lift? The third quarter of 2026 delivered a record: “record group net profit of GEL 404 million, up by 10% year-on-year and a return of equity of 23.2%” — Vakhtang Butskhrikidze, Group CEO · 2026-11-06. That is the fifteenth consecutive quarter above 23% ROE. Georgia alone ran at 24.8% with net profit up 17%. On the numbers, nothing broke. What changed is subtler and, I'd argue, more consequential.
The regulator didn't defer — it cancelled
In August, management told investors the Uzbek central bank's proposed hike in consumer risk weights — which would have taken unsecured lending risk weights from roughly 100% to upwards of 200% — was off. The language was careful: “at the current time, it is not being implemented. So we consider this as a potential pause” — Guy Richard Stevens, Group CFO · 2026-08-06. A pause is a thing you plan around. It implies a sword still hanging.
On this call, Group CFO Guy Stevens removed the ambiguity entirely. Asked to confirm, he stated flatly that “the regulator in Uzbekistan does no longer intends to implement them” — Guy Stevens, Group CFO · 2026-11-06 — not delayed, not paused, not kicked down the road. The proposed changes in risk weightings are dead. This is a genuine, company-specific de-risking event, and it is the single most important sentence in the release. It dismantles the mechanical justification for the forced march out of instant cash loans — and yet management explicitly said the 50%-of-bank-book cap for year-end 2026 and the 25% cap for 2027 remain in force.
The relative composition of loan books... by the end of 2027, our bank's balance sheet having to have no more than 25% of loans in the form of instant cash loans, that remains. There hasn't been a change in the position of the regulator.
So the carrot has been withdrawn and the stick left standing. That is a very different regulatory posture to the one investors were pricing eighteen months ago, and it reframes the recalibration from a capital-management exercise into a pure strategic choice by the bank.
The reform that keeps arriving next January
The second tension is the one that matters for near-term earnings. Restrictions on auto collections — the mechanism by which Uzbek banks automatically debit a borrower's linked card on a missed payment — have been the recurring bogeyman in this company's guidance since at least the first quarter. In August, Stevens guided Q3 cost of risk in Uzbekistan to "low mid-teens" and declared “the third quarter is going to be the peak” — Guy Richard Stevens, Group CFO · 2026-08-06.
Actual Q3: 10.8%. Below guidance. And the reform itself? Same sentence, different quarter: “further restrictions in respect of auto collections are now expected to come into effect in January of next year” — Guy Stevens, Group CFO · 2026-11-06. A rule that was going to bite in Q3 now bites after year-end — which conveniently means the year-end print will be the first clean look. Management is explicit that they cannot size it: "we're looking at all possible scenarios" and clarity comes only with the full-year numbers in February. This is the third consecutive call where the most material risk in the Uzbekistan book is described as unquantifiable and just over the horizon. Treat the deferral as a genuine tailwind to the Q4 print and a genuine vacuum in the 2027 model.
The NPL ratio tells you why nobody is relaxed: group NPL sits flat at 3.3%, but Uzbekistan's is rising on the seasoning of 2025 vintages plus the mechanical effect of extending write-offs from 270 to 360 days — against a denominator that is still roughly 11% smaller than it was a year ago. Analysts asked twice when it peaks. Twice the answer was effectively after January.
Georgia is now doing all of the work
With Uzbekistan contributing 4% of profits, the group story is unapologetically a Georgian one, and the franchise is compounding in ways that have nothing to do with credit. Group digital monthly actives passed 7.5 million, up 350,000 in the quarter. In Georgia, 83% of retail deposits and 87% of unsecured retail loans are issued fully digitally with an average acceptance time of 30 seconds. The in-app AI assistant handles 6,500 queries a day and resolves 60% without a human — a small but real instance of the personal AI agents theme that is now threaded through CEO commentary across global financials.
Two product moves are worth flagging. The American Express acquiring partnership, live this quarter, should add 1–2% to the acquiring business next year plus a nudge to fee income — “we will increase our business in acquiring business, it depends, but in the range of 1% to 2%” — Vakhtang Butskhrikidze, Group CEO · 2026-11-06 — and this week TBC became the first Georgian bank to launch palm payments. Both are card payments economics dressed as novelty; the strategic point is share of wallet in a duopolistic market.
The offset is fee income. Georgia's fee line has been suppressed for two to three quarters by deliberate investment in cards and loyalty, and Stevens would only commit to "a positive trajectory" into Q4 — carefully non-committal, and directly echoing the August framing that full-year fee and commission income would be flattish. The cost income ratio ticked up to 38.7% at group level precisely because Uzbekistan is being re-staffed for growth, while Georgia's improved to 32.7% with positive jaws. Management was unusually blunt that group jaws get worse before better: "you shouldn't be surprised in terms of the operating jaws not being positive over the next kind of couple of quarters."
The small print nobody asked about, and the one they did
Buried in the P&L is a reclassification: deposit insurance and resolution fund expenses pulled out of interest expense and administrative costs into a standalone line to match European and Georgian peer practice. The CFO's own quantification — NIM would have been 13bps higher, cost-to-income 30bps higher — means it is immaterial to the bottom line but material to comparability. It is a presentational tell: TBC is now managing the optics of a 23% ROE business against a peer set rather than in isolation.
And then the recurring non-answer. For the second consecutive quarter, an analyst asked whether the $75 million AT1, callable from November, will be called. For the second consecutive quarter the response was effectively we are mindful of market precedents. Also recurring: the confusion around the Uzbek bank's standalone accounts versus the consolidated four-balance-sheet picture — one analyst this quarter asked why IFRS profit and the local-statutory result diverge so sharply, prompting a reference to “the fog, as I call it, of Uzbek accounting standards” — Guy Stevens, Group CFO · 2026-11-06. When a group has to teach the buy side how to read its second market every single quarter, that discount is structural.
Capital return is the counterweight. A third-quarter dividend of GEL 1.75 takes the nine-month total to GEL 5.25, up 5%, and comes alongside a GEL 50 million buyback and cancellation programme — inside a stated 25–45% payout ratio range, with the bank holding 200bps of CET1 headroom in Georgia and roughly 13 percentage points above the minimum in Uzbekistan. The ambition to lift Uzbekistan to 20%+ of group profit by 2030 is unchanged in destination, explicitly changed in pace.
The bottom line: TBC just had the capital rule it feared removed and the collections reform it provisioned for postponed — two tailwinds — and the shares are backed by a 23% ROE and a growing buyback. But the return to growth in Uzbekistan is three quarters old and still single-digit, the reform arrives in January unquantified, and the entire earnings engine is a 3.7-million-person market growing GDP at 7.4%. That is a wonderful place to be long. It is also a very concentrated one.