Banco BMG Q2 2026 slides: profitability surges as portfolio shifts

Banco BMG reported Q2 2026 net income of R$157 million (+25.6% YoY) and ROAE of 15.5% (up from 14.3% a year earlier), with results meeting analyst expectations (EPS 0.2339; revenue R$1.65B). The bank’s turnaround is attributed to a portfolio remix toward higher-margin personal credit and private payroll loans, lifting net interest margin after cost of credit to 11.1% of assets (from 10.7% YoY), while the efficiency ratio improved to 52.4% (toward a 50% target). Asset quality weakened with provision expense annualized at 6.3% (from 4.6%) and NPLs >90 days rising to 4.2% (from 3.8%), but liquidity remains strong (LCR 429%; NSFR 124%). Shares barely moved to $4.92 from $4.91 as management highlighted an AI-driven operating transformation and a dividend yield of 10.7%.

Analysis

The investable signal is not the reported ROE improvement; it is whether that ROE is durable after the credit cycle catches up. A lender rotating into higher-yield consumer products can look like a quality compounder for 2-3 quarters, but once loss content rises, the market usually stops capitalizing earnings on peak margins and starts capitalizing them on normalized credit costs. That matters most for subscale financials: if funding stays cheap but credit costs keep creeping up, the real risk is not a collapse in earnings but a lower multiple on a still-improving business.

The second-order winner is any platform that can lower acquisition and servicing cost without sacrificing underwriting speed. The loser set is the long tail of branch-heavy or manually underwritten consumer lenders, because digital self-origination plus automated contract review compresses their edge on distribution and collections. Over 6-18 months, the AI claim only matters if it translates into a persistent efficiency advantage; if competitors replicate the workflow, the benefit gets competed away into lower pricing rather than higher returns.

The contrarian read is that the dividend story may be more fragile than the headline yield suggests. Falling coverage and a richer product mix can be fine in a benign macro, but a modest uptick in unemployment or delinquency would hit this kind of book disproportionately and likely arrive before management can reprice enough. The key falsifier is not one quarter of solid profit; it is whether NPLs and provision cost stabilize over the next 1-2 reporting cycles while capital ratios remain comfortably above targets.

Key Decisions for Investors

• No direct trade in AMZN/TGT/OZK/HCSG/CBSU/GRO from this note; the read-through is too indirect for a position over the next 1-3 months.
• If you want the cleaner expression, buy ITUB or BBD on weakness versus a basket of Brazilian financials for 3-6 months; the setup is a relative-quality rerating if Brazil consumer delinquencies stay contained.
• Avoid initiating longs in higher-risk Brazilian payroll/consumer lenders for now; wait for 1-2 quarters of stable NPLs and provisioning before paying up for the growth story.
• Set an alert on any 1Q/2Q follow-up that shows coverage ratio continuing to fall below the low-160s or provision expense staying above the mid-6% area; that would be the first sign the ROE story is peaking.