Banco do Brasil's Moral Hazard Confession
Banco do Brasil's third quarter of 2026 produced BRL 1.3 billion of net income, or BRL 18.3 billion year to date — numbers the CEO herself refused to defend. “This result is much lower than we expect for a bank with the strength of Banco do Brasil. We know that, and you know it, too.” — Tarciana Gomes Medeiros, Executive (likely CFO or similar) · 2026-11-12 But the real event of the call was not the P&L. It was a single sentence that reclassified the entire Agri business cycle from an act of God to an act of man.
The confession
For six quarters management carried a tidy narrative: a once-in-a-generation margin squeeze in soy and corn, a record harvest on the way, and a delinquency curve that would bend in the first half of 2026. The harvest arrived. The curve did not. Collection on the ag book, which they had told the market would recover toward 90% between August and December, instead slid to roughly 70% in September and October — below August, the wrong direction of travel.
Some mid- to large-sized growers chose neither to pay nor to renegotiate, or even refused to provide collateral... We did not anticipate the persistence and deliberate nature of moral hazard. This behavior comes at a cost.
That is the new vocabulary, and it is the story. The keyword moral risk has climbed into the bank's own top ten themes this quarter, and the arithmetic behind it is damning: of the BRL 13.4 billion renegotiated under Provisional Measure 1376, only 15% involved delinquent borrowers, against an internal expectation of 50%. “Initially, we signaled that we could operate with up to BRL 30 billion, of which 50% for delinquent growers.” — Gilson Bittencourt, Executive (likely risk or credit related) · 2026-11-12 The gap was not a credit problem. It was a willingness problem — farmers betting on a better deal from Brasília, and, for a while, right to bet.
Cost of risk hit BRL 21.2 billion in the quarter and BRL 58.6 billion over nine months, up 33.2% year over year. The bank is now doing what it spent two years avoiding: 20,000 out-of-court notices, 236 of more than 400 eligible properties in repossession, and — genuinely new — a decision to sell non-performing loans outright, with more than 40 buyers already in the data room. That option was first floated as a hypothetical on the November 2024 call; it is now live. The forbearance loan book, at roughly BRL 60 billion with a 22% expected loss, is the visible next shoe.
Capital: three answers to one question
Analysts pushed on capital from four different angles, and got three different answers on the dividend, which is its own kind of disclosure. The CEO framed it as a floor with a contingency: payout cut to the mandatory minimum of 25% for 2026 and 2027, “with distribution contingent on the capital projections verified at the end of every fiscal year.” The CFO was blunter. “for 2027, we decided not to have any distribution or payout of dividends because we want to reinforce our capital.” — Marco Geovanne da Silva, Executive (likely CFO or similar) · 2026-11-12 And in the final exchange he said the 10.5% CET1 target “does not contemplate payout” and would be reassessed at year-end 2027. Three framings, one direction.
What is genuinely new is the toolkit. The bank intends to convert roughly BRL 12–14 billion of deferred tax assets into presumed credit under Law 12838 — worth potentially 120 basis points of CET1, a law it has never previously been able to trigger. Add two portfolio companies queued for IPOs and a reworking of tax shields at affiliates. Explicitly rejected: selling down BB Seguridade despite the CFO's own estimate that the control surplus is worth about BRL 15 billion. “BB Seguridade is not for sale, and renegotiation of the contract will happen at the expected time within our projections.” — Tarciana Gomes Medeiros, Executive (likely CFO or similar) · 2026-11-12 That refusal is the most revealing line in the transcript: management would rather grind out 120bps from a tax statute than crystallise a BRL 15 billion gain and lose the earnings stream.
A compass that keeps moving
This is where the prior calls matter. In August 2025 the CRO set a finish line: “what we aim at is this curve to meet as soon as possible in the beginning of 2026 close to this level of 2%, 2.5%” — Felipe Guimaraes Geissler Prince, Executive (likely CRO or similar) · 2025-08-15 — the historical ag NPL. In November 2024 the CFO promised, “And we will continue to pay dividends to our shareholders at 45% payout, irrespective of this implementation.” — Geovanne Tobias, Executive (likely CFO, CRO, or similar) · 2024-11-14, having already stepped down from 40%. The same year the bank called “The ideal number would be around 11%.” — Marco Geovanne Tobias da Silva, Executive (likely CFO or similar) · 2025-08-15. Today the ag book is nowhere near 2.5%, the payout is 25%, and the capital target is 10.5%. Each individual step was defensible; the sequence is a pattern, and the market has watched it for two years.
To management's credit, the Q&A did not hide this. The CEO took the unusual step of running a counterfactual: normalise ag risk and 2025 ROE would have been ~19% rather than 11.4%, and the 2026 nine-month figure would rise from 6.2% to ~16%. That is an honest way of saying the franchise works and the cycle is the whole story. The phrase “soft landing of Banco do Brasil” was used about a dozen times.
What survives — and what the tape is saying
The growth engine management will actually fund is private payroll loan lending — origination steered into public payroll, secured pension-backed credit, high-income cards, and receivables, with the ag book deliberately shrinking in risk-adjusted terms. The problem is that the market is already voting the other way on the Brazilian consumer-credit complex. In the global ninety-day tape, Private payroll loans sit among the sharpest decliners, dragged by NU, BBD, ITUB and BBSE3 — the very cohort whose private-payroll expansion BB is counting on. The bank is pivoting into a lane the market has spent a quarter de-rating.
And the cycle itself is not purely Brazilian. el Niño was the second-highest-momentum theme in the entire global keyword set a quarter ago, and it echoed across this reporting season from BAP and AVD to DOLE and AUSS — ag lenders and growers on two continents describing the same water and margin shock. Banco do Brasil is the most levered listed expression of that weather trade, and the difference is that it lent against it.
The tactical question for the next two quarters is narrow and unpleasant. The CFO pre-announced it: “Fourth quarter '26 and first quarter '27, due to the past-due loans that will become Stage 3.” An estimated 45% of the December receivable flow now sits under the new resilience-matrix methodology, rising to 57% — better collateral, better covenants, but not yet a better P&L. The bank has bought itself time with the dividend, 120bps of tax law and a refusal to sell its crown jewel. What it has not bought is a borrower who is willing to pay.