Ally Pays the CECL Bill for Finally Growing
A 16% EPS beat and the first dividend hike since 2022 — but also the first quarter in a year where delinquency bent the wrong way, and the two tailwinds that carried 2026 credit went quiet.
ALLY · Earnings Call · 2026-10-20
The credit line bends
On the surface this was the quarter Ally's management has been promising for two years. Adjusted EPS of $1.34 rose 16% year over year, revenue rose 9%, noninterest expenses were flat, and the efficiency ratio fell to 69.3%, down 6.4 points year over year. Michael Rhodes opened with the two-year scorecard: core ROTC from 5% to 13%, retail NCOs down more than 30 basis points, retail auto and corporate finance assets up nearly 15%.
Then Ryan Nash of Goldman asked the question that reframes the whole call: “we did see delinquencies slip from down year over year to up” — Ryan Nash, Analyst · 2026-10-20. Thirty-plus-day delinquencies of 4.96% were up 6 basis points year over year. Retail auto net charge-offs of 189 basis points were up 32 basis points sequentially, roughly flat annually. And the late-2025 and early-2026 vintages are running ahead of the 2024 vintage on early-life delinquency.
The defense is coherent, and it is the same defense Ally has used for three quarters: the 2024 vintage outperformed, so some normalization was always coming, and first-quarter 2026 was deliberately rich in both credit content and yield — a credit content decision management flagged transparently at the time. What is new is the framing of what growth costs. Retail auto grew $2.4 billion in the quarter, and that produced nearly $90 million of CECL reserve build — “a 3Q headwind of roughly $0.20 to EPS and 200 basis points to ROTCE” — Russell Hutchinson, Chief Financial Officer · 2026-10-20. That is the trade Ally is now asking shareholders to underwrite: book the charge today, harvest the spread tomorrow.
So far the market has not paid for it. The stock sits 19.6% below its June peak and 32% below its 2021 high, at roughly 1.0x book and under 10x earnings. A 16% EPS beat into a 9% ninety-day drawdown is a skepticism problem, not a results problem.
The lease-shaped pothole in the NIM path
Net interest margin of 3.66% rose 3 basis points sequentially and 34 year over year — quietly one of the better margin performances in the regional-bank complex. Guidance was tightened to 3.6%–3.65% for the year with a fourth quarter around 3.7%. The structural drivers are real: $154 billion of deposits, up 4%, a 70th consecutive quarter of customer growth, and a balance sheet remixing out of legacy mortgages into retail auto and corporate finance.
But this is also where the call got uncomfortable. NIM included a $16 million loss on lease termination because of “continued pressure on the Stellantis' auction prices and residual values stemming from broader brand and recall related overhang” — Russell Hutchinson, Chief Financial Officer · 2026-10-20, and management guided that pressure to persist into the fourth quarter. Compare that to April, when the same item was a $10 million loss described as "a little favorable" and tied to what Russell then called "a handful of PHEV models." The lease yield problem has migrated from a model-specific footnote to a named brand overhang with a 2027 recovery timeline.
The other wrinkle: the rate world moved against Ally's January assumption by about 100 basis points of Fed funds, and management now describes roughly $60 billion of floating-rate exposure repricing higher before the balance sheet inflects liability-sensitive. This is the same directional story every regional bank told this week — loan beta and deposit repricing dominate the shared keyword set across reporting banks — so the differentiator is not the setup, it is the honesty about the path. Asked whether portfolio yield steps back before stepping forward, Russell declined the step function and offered physics instead: “do not expect any step function changes in portfolio yield. Think about more of a drift” — Russell Hutchinson, Chief Financial Officer · 2026-10-20. Originated yields exited the quarter in the high 9s, above the 9.44% printed for the period, so the input is improving even as the stock of loans lags.
Capital: the "and" finally pays a dividend
CET1 of 10.2% was up 10 basis points year over year, or 8.8% fully phased under the proposed standardized approach with another ~30 basis points of IRBA benefit available. Ally executed its second credit risk transfer of 2026 — the largest North American retail auto CRT management is aware of — worth more than 25 basis points at issuance. It bought back $148 million of stock in the most recent reported quarter and nearly $500 million over the trailing year. And it raised the quarterly dividend to $0.35, the first increase since 2022.
That combination is why capital levels sits atop Ally's keyword set this quarter, ahead of the auto-credit vocabulary that usually dominates. The story of "and, not or" — grow, hold capital, return capital — was a talking point in January and April. It is now a cash flow. Rhodes was pointed about the milestone: “we have returned meaningful capital to shareholders while also building our capital base” — Michael G. Rhodes, Chief Executive Officer · 2026-10-20. Given a 14% increase in effective revenue, the capital-generation arithmetic is doing most of the work.
What Ally stopped saying
Here is the tell. For five straight quarters, "flow to loss" and used vehicle prices were the load-bearing beams of Ally's credit argument. In July: “we have seen good support. From used vehicle prices” — Russell Hutchinson, Chief Financial Officer (CFO) · 2026-07-21. In April: “flow-to-loss rates were solid. We got support from used vehicle prices” — Russ Hutchinson, CFO · 2026-04-17. In October 2025 it was the same refrain. This quarter, flow to loss has dropped out of the company's keyword set entirely, and so have used car prices. Management said the quiet part out loud: those tailwinds were expected to be supportive but "less supportive as we work through the year," and that is precisely how it played out.
Equally telling is what vanished from the tone. The word "measured" — the rhetorical hedge that had defined every call since tariffs arrived — is gone from the current transcript, and the keyword tracker flags a pronounced decline in measured posture. Ally is no longer hiding behind caution; it is leaning into application flow of 4.8 million decisions, up 20%, and it is originating at the high end of its 1.6%–1.8% risk appetite. That is a genuine posture shift, and it is being paid for in the reserve line.
The odd note is agentic commerce. Asked whether AI shopping agents change consumer behaviour, Rhodes was dismissive — “I think so far the activity level. We have seen activity, but it is actually been quite low” — Michael G. Rhodes, Chief Executive Officer · 2026-10-20 — while adding that Ally already shows up "pretty favorable" in AI search. Synchrony, reporting the same day, treats Agentic Commerce as a live strategic theme. One of these two is early; it is not obvious which.
The verdict
The quarter produced the numbers and the capital return that the bull case required. What it did not produce was corroboration on credit: provision expense of $430 million rose 12% year over year, and the delinquency trend line has stopped falling. The thesis now rests on whether growing into a 9%-plus originated yield can outrun a 189-basis-point loss rate while a single OEM's residuals take a $16 million bite out of a $2.3 billion revenue quarter. Rhodes framed the destination as a drift toward mid-teens returns, and the drift is real. It is just slower, and more expensive per dollar of growth, than the multiple at 1.0x book is willing to admit.