This Week's Bank Earnings Tell the Consumer Story Through Provisions, Not IB Fees

Bank of America CEO Brian Moynihan told investors in mid-September that Q3 investment-banking fees would fall more than 10% from a year earlier, landing somewhere between $1.6B and $1.8B versus roughly $2.0B in Q3 2025. The KBW Nasdaq Bank Index has already absorbed the message, sliding more than 13% from its mid-August peak as global M&A activity slowed to about $1.23T in Q3 2026 from $1.8T in Q2, and IPOs like Oura and SB Energy pulled their listings in late September. The first read on whether the IB cliff materializes is therefore the least interesting question facing the prints — markets have been discounting exactly that for six weeks.

What hasn't been priced with the same conviction is the consumer. JPMorgan raised its FY2026 net interest income guide to roughly $105.5B and adjusted expense to about $107.5B after Q2, but also lowered its full-year card-services net charge-off (NCO) guidance to ~3.2% from 3.4% — a quiet admission that the consumer is performing better than the macro would suggest. Q2 2026 card NCO came in at 3.34% for JPMorgan, and Wells Fargo's company-wide Q2 NCO was 0.34%, down 10 bps YoY. The cycle's biggest test is whether those benign prints survive a quarter in which the 10-year Treasury sits near 5.23% and ICE BofA's high-yield OAS has blown out to 3.15% from 3.03% in two sessions.

Q3 2025 set a high bar for JPMorgan's provision line. The firm booked a $3.4B provision for credit losses, $2.6B of net charge-offs (up $506M YoY), and a $810M net reserve build, with $608M of that build in Consumer and $205M in Wholesale; card services ran a 3.15% NCO rate. Q2 2026 normalized sharply: provision fell to $2.5B, NCO to $2.4B, and the net reserve build shrank to just $149M. That sequence is the playbook investors should walk into Tuesday carrying — Q3 2026 will tell them whether the Q2 calm was a one-quarter reprieve or the start of a benign run.

Three forces argue for watching provisions more carefully than usual this cycle. First, Citigroup's U.S. Consumer Cards unit set aside $1.6B in Q2 2026 against $1.9B of net credit losses — a near one-for-one relationship that any swing in delinquencies will amplify into the parent provision. Second, Capital One, which just absorbed Discover, posted a $3.0B provision in Q2 2026 ($1.1B lower YoY) on $3.6B of NCO, signaling that even a structurally subprime lender can see reserve releases when delinquencies ease. Third, Synchrony Financial reported a $1.2B provision in Q2 2026 with a $163M reserve release, the kind of release that quietly reverses when early-stage delinquencies roll forward.

Credit-card charge-offs at the industry level are actually improving — the FRED series for all commercial banks fell from 4.18% in Q3 2025 to 3.82% in Q2 2026. But the Fitch U.S. Auto Loan ABS Indices tell a different story in the pockets that matter. Subprime 60+ delinquencies reached 6.74% in 2H25, eased briefly to 5.80% on tax-season relief, and climbed back to 6.13% by July 2026. Fitch's August outlook explicitly expects prime and subprime auto loan ABS performance to weaken further in 2H26, driven by tariff uncertainty and oil-price stress. The prime index has stayed near 0.43% — the divergence is what matters.

Citi's setup is the most leveraged. Management set a medium-term 13%+ RoTCE target at its May 2026 Investor Day and is targeting 10–11% RoTCE for 2026; reported RoTCE was 7.7% in 2024 and 8.8% in 2025, so the climb back to target requires both an ACLL release and benign cost of credit. The Q2 2026 USCC business already booked a $232M net ACL release — if Q3 prints a build instead, that path to 13% gets materially longer, and the stock's 2026 multiple starts to compress.

Jamie Dimon has spent 2026 sharpening a credit-cycle warning. In April he told shareholders the next credit crisis could be worse than 2008; in private-credit commentary through Q2 he argued defaults would move modestly higher; and in the first week of October he sharpened the corporate-credit language, warning that rising government borrowing costs and AI-related capex could spill into corporate spreads, with leveraged loans trading below 60 cents on the dollar climbing to $65B — the highest level since March 2020. That is wholesale credit, not consumer, but it primes the lens he uses for the consumer provision line.

The cleanest proxy for what Dimon is signaling comes from Q2's own commentary. Card NCO fell to 3.34%, but management simultaneously raised the FY26 expense guide to ~$107.5B from $105B — a tacit acknowledgment that it costs more to run this book at the current loss rate than it did six months ago. Watch Q3 expense discipline alongside the provision: any meaningful upward revision to either is the signal the consumer story is bending.

The consensus expects JPMorgan to deliver EPS of $5.93 on $51.19B of revenue — 17% YoY EPS growth on 12 analysts — but the stock is down 13% from its August peak with Citi and BofA the worst hit. That price action reflects two distinct bets: that the IB-fee warning is the start of a deeper capital-markets slowdown, and that the consumer will follow. Q3 prints will resolve the first; the second depends almost entirely on the provision line, where Dimon has already telegraphed caution and where subprime auto delinquencies at 6.1% argue the cycle hasn't turned yet. The setup favors investors who treat this week's print as a credit-data event more than an earnings event.