Citigroup Q3 2026 Earnings: $2.85 EPS on the Best Q3 Revenue in a Decade

On October 13, 2026, Citigroup reported third-quarter net income of $5.2 billion, or $2.85 per share — up 40% and 53% from a year earlier, and a clean beat against the $2.66 consensus — on revenue of $23.85 billion (+8%), which chief executive Jane Fraser called "the best third-quarter revenue in a decade." The market's response was to sell: shares fell about 5% as investors looked past the beat to a provision for credit losses that rose 6% on a $369 million reserve build tied to "changes in the macroeconomic outlook," and to fee income that softened across trading and cards. Strip out the year-ago quarter's $726 million Banamex goodwill impairment and the growth is a still-solid 17% in net income and 27% in EPS — but the quarter's real story is the tension between a booming spread business and a management team quietly stockpiling for a downturn.

The Headline Numbers​

Every figure below comes from the Q3 2026 financial supplement (8-K Exhibit 99.2) and the earnings release (Exhibit 99.1):

Three deltas explain both the beat and the selloff. First, the revenue mix rotated hard toward spreads: net interest income surged 14% on 12% loan growth and wider deposit spreads, while non-interest revenue fell 4% as principal transactions dropped 18% and card fee economics deteriorated. Second, expenses fell 2% only because the year-ago quarter carried the $726 million Mexico goodwill impairment tied to the Banamex 25%-stake sale; excluding it, expenses rose 3% on compensation, transactional costs, and deposit insurance — still positive operating leverage of over 1,000 basis points against 8% revenue growth, but not the cost decline the headline suggests. Third, credit cost $136 million more than last year even though net charge-offs were flat at $2.2 billion: the entire increase is a $369 million net reserve build, up from $236 million, that management attributes to the macroeconomic outlook and portfolio growth. A bank that builds reserves while losses hold flat is either prudent or prescient, and the market spent the day debating which.

Revenue Deep Dive​

Five of the six businesses grew; the consolidated +8% is net of a shrinking All Other, where exit markets continue to wind down:

Banking revenue is shown including the $9 million gain on loan hedges, the basis on which the release reports its +6%; All Other is on the managed basis that excludes divestiture-related impacts, so the segments sum to approximately — not exactly — the reported total.

Services is now unambiguously the crown jewel: $6.6 billion of revenue (+15%) at a 31.2% return on tangible common equity, with Treasury and Trade Solutions up 16% to $4.9 billion and Securities Services up 13% to $1.7 billion. Fraser's verdict — "Services delivered another standout quarter, up 15%, with momentum in every driver" — checks out against the drivers: cross-border transaction value up 10%, commercial card spend up 10%, dollar-clearing volume up 5%, and assets under custody up 16% to $35 trillion. Average Services deposits grew 16% to just over $1 trillion, which is the funding engine for the whole bank's loan growth.

Markets grew 6% to $6.1 billion, but the mix is the message. Equities rose 14% to $1.7 billion on prime-services growth, with prime balances "up over 40%," while Fixed Income rose 4% entirely on spread-products financing (+13%) — rates and currencies were flat, with FX volume gains offset by lower rates revenue. Markets net income barely moved (+2%) because expenses rose 7% and the provision more than doubled to $77 million on portfolio growth. Wealth had the cleanest quarter in the bank: revenue up 15% across all three businesses on wider deposit spreads, net income nearly doubling (+93%) to $585 million, and $38 billion of net new investment assets — more than double last year's inflow. Banking's 9% investment-banking growth hides a split market: ECM surged 77% as Citi "gained share in follow-ons and convertibles" and "participated in the quarter's six largest transactions," while Advisory fell 4% "amid a lower wallet" and DCM slipped 3% on soft investment-grade issuance.

The two soft spots are both fee stories. U.S. Consumer Cards revenue was flat at $4.7 billion despite 6% NII growth, because non-interest revenue collapsed 62% to $(758) million on "higher partner payment accruals and rewards costs" — the co-brand and rewards economics eating the spread gains — and net income fell 8%. All Other's managed loss widened to $(955) million as Legacy Franchises revenue slipped 1%: growth in Mexico was more than offset by "continued lower revenues from closed exit and wind-down markets."

Management's growth language is confident but narrowly sourced. The release's quotable signals are all volume-and-share claims — prime balances up 40%+, six of the largest ECM deals, cross-border and clearing volumes up — plus one piece of forward guidance: "We expect to exceed this year's return target, even as we pull forward investments toward our 11-13% near-term goal — the next waypoints on our path to sustainable returns of 14-15%." What is absent matters as much: there is no "demand exceeds supply," no industry-upcycle language, and no tight-supply wording. For a bank, the demand signal is deposit spreads, and the release cites "higher deposit spreads" in Services, Wealth (three times), and cards-adjacent commentary — pricing power confirmed by the ledger's 14% NII growth, not just claimed.

The Margin Story​

The margin story is operating leverage finally arriving — with an asterisk. The efficiency ratio improved more than 600 basis points year over year to 58.5%, but roughly half of that is the absent $726 million impairment: on an ex-notable basis the year-ago ratio was 61.4%, so the true improvement is about 290 basis points. That is still real leverage — 8% revenue growth against 3% ex-notable expense growth — driven mechanically by NII, which carries almost no marginal cost. Every incremental spread dollar falls at a ~60% pre-tax margin, which is why pre-tax income grew 37% on 8% revenue growth.

Return metrics tell the same story in two speeds. RoTCE of 11.5% is up 350 basis points year over year and sits squarely inside the 11–13% near-term goal Fraser reaffirmed — but it is down 150–160 basis points sequentially from the 13%+ first half, as revenue normalized from record quarters and the share count stopped shrinking as fast. The tax line helped year over year (26% effective rate versus 29%), while the preferred-dividend and NCI drags grew: $353 million of preferred dividends (+29%) and $175 million of noncontrolling interests (+361%, the Banamex minority share) now stand between pre-tax income and common shareholders. Net income available to common still grew 41%, but the minority slice is a new structural headwind worth watching — it was $38 million a year ago.

The One Big Question: Is the $369 Million Reserve Build Prudence or Prophecy?​

No line in the quarter moved the stock more than the allowance. Net charge-offs were flat at $2.2 billion, yet Citi added $369 million to reserves — 56% more than last year's $236 million build — explicitly "driven by changes in the macroeconomic outlook and portfolio growth." When a bank's realized losses say "steady" and its models say "build," investors have to decide whether management sees something the charge-offs don't show yet. The release gives both sides of the argument:

¹Loan reserve build plus unfunded-commitment provision; the release's $369M/$236M "net reserve build" adds the other-provisions row.

The prudent-read evidence is strong. Non-accrual loans — the pipeline of future charge-offs — fell 12% to $3.3 billion, with corporate down 16% on "upgrades and repayments in Markets." Card losses actually improved: U.S. Consumer Cards net credit losses fell 4% on better private-label performance. And the 12% loan growth mechanically demands reserves — a bank growing loans $86 billion year over year that didn't build would be under-reserving, not outperforming. The coverage ratio slipped from 2.7% to 2.5% precisely because loan growth outran the build.

The prophecy-read evidence is concentrated in two places. First, the U.S. Consumer Cards build of $208 million — nearly quadruple last year's $58 million — is "primarily driven by changes in the macroeconomic outlook," the same macro wording that covers the consolidated build. Cards are where a consumer downturn lands first, and Citi is fortifying there while losses still look fine. Second, Mexico Consumer is deteriorating in real time, not in models: All Other net credit losses jumped 46% on "higher volume and portfolio seasoning in Mexico Consumer," with another $99 million build driven by policyholder benefits and claims there. Seasoning + volume + a macro overlay is exactly how a consumer-credit problem starts.

Our read: mostly prudence, with one genuine watch item. A 2.5% coverage ratio on a loan book growing 12%, against falling non-accruals, is a bank leaning into growth with a macro hedge — not one hiding losses. But Mexico Consumer's 46% loss growth inside a franchise Citi is trying to separate (the Banamex exit) is the number to track each quarter: if U.S. card delinquencies follow Mexico's seasoning curve, today's $369 million build will look small. If they don't, it reverses into earnings.

Tracking a $2.94 Trillion Bank in Plain Text​

A money-center bank is where double-entry accounting earns its keep: every dollar of $2.94 trillion in assets must reconcile to a dollar of liabilities or equity, and Beancount enforces that identity mechanically. Our ledger models Citigroup exactly the way we model every company — read how we model every company for the shared conventions — with a bank chart: revenue is total revenue net of interest expense, the provision for credit losses and for benefits and claims is the cost of revenue, and operating expenses are operating expense. Income is negative (credit) and expenses positive (debit), so the transaction below sums to zero:

; Check: (-23845) + 2586 + 13952 + 1887 + 175 + 5245 = 0
 
2026-09-30 * "Citigroup Inc." "FY2026Q3 Income Statement"
  Income:Revenue                             -23845 MUSD
  Expenses:CostOfRevenue                       2586 MUSD
  Expenses:OperatingExpenses                  13952 MUSD
  Expenses:IncomeTax                           1887 MUSD
  Expenses:OtherNet                             175 MUSD
  Equity:Adjustments                           5245 MUSD

One posting deserves attention: the $175 million in Expenses:OtherNet is the filing's own noncontrolling-interests deduction — overwhelmingly the Banamex minority share after the December 2025 and April 2026 equity sales — so the $5,245 million offset is Citigroup's attributable net income, matching the face of the statement. A year ago that line was $38 million; the minority slice has nearly quintupled, and the ledger carries it explicitly rather than burying it.

The balance-sheet number that tells the narrative is the loan book against its allowance: loans of $819,705 million, up 12% year over year, covered by a $20,091 million allowance that grew only 5% — the 2.5% coverage ratio at the heart of the quarter's credit debate, above. A second artifact worth noticing: Treasury stock of $(102,418) million now approaches the combined $125,532 million of preferred stock, common stock, and paid-in capital, so years of buybacks have nearly hollowed out the contributed-capital account — retained earnings of $227,733 million is what holds equity up. And deposits of $1,505,322 million (+9%), funded overwhelmingly by 16% growth in Services transaction deposits, show a bank funding its loan growth with the stickiest deposits in the franchise.

The Multi-Year Arc​

Five full years plus the new quarter, every figure from the ledger's filed statements:

*Q3 is a single quarter; balance-sheet columns are period-end snapshots.

The arc is a bank compounding its spread engine while shrinking its share count through a divestiture program. Net interest income grew 41% from FY2021 to FY2025 ($42.5 billion to $59.8 billion) as higher rates repriced the loan book, while revenue compounded steadily to $85.2 billion and net income traced a V: the FY2021 peak ($22.0 billion) was flattered by $3.8 billion of net reserve releases, the FY2023 trough ($9.2 billion) was depressed by the $1.7 billion FDIC assessment and restructuring charges, and the recovery since is the real operating story — RoTCE climbing from 4.9% in FY2023 to 7.7% in FY2025 to 11.5% this quarter. Meanwhile the share count fell every year, from 1,984 million shares outstanding at the end of FY2021 to 1,748 million at FY2025, turning earnings growth into faster per-share growth — tangible book value per share compounded from $79.16 to $97.06 over the same stretch. One comparability note: each annual in the ledger is modeled from its own 10-K as filed, so the FY2022 premises breakout and the FY2025 interchange-fee netting (which restated FY2024/FY2023 revenue and opex down equally, net income unchanged) step the series — see the Comparability note in main.bean.

Capital return is the second compounding engine, now running near full throttle. Citi returned $4.1 billion to common shareholders this quarter alone — an 85% payout ratio — after $13.3 billion of buybacks in FY2025, and Fraser says Citi will "exceed this year's return target." The constraint is visible: the CET1 ratio of 12.8% is flat sequentially but down from 13.3% a year ago and 13.18% at FY2025 year-end, as 11% asset growth and buybacks consumed the retained profit. Book value per share reached $115.61 (+7%) with tangible book at $101.59 (+6%) — both held back by the $(41,438) million AOCI hole of unrealized securities losses. The 11–13% RoTCE goal and the buyback pace can coexist only while credit stays benign — which is exactly why the market spent earnings day staring at the allowance.

The Verdict: Bull vs. Bear​

Bull Case

• The spread engine compounds on volume: NII up 14% on 12% loan growth and wider deposit spreads, with $1 trillion of Services transaction deposits (+16%) funding it — even a flat margin on this balance sheet grows NII every quarter.
• Operating leverage is real and targeted: 8% revenue growth against 3% ex-notable expense growth, efficiency down to 58.5%, and management guiding to exceed this year's return target on the path to 14–15% sustainable returns.
• Credit is benign where it counts: charge-offs flat at $2.2 billion, non-accruals down 12%, corporate non-accruals down 16% on upgrades and repayments — the $369 million build is a macro hedge on a growing book, not a response to losses.
• Fee engines are accelerating where it is structural: Wealth net income +93% on $759 billion of client assets (+15%) with record $38 billion inflows, ECM +77% with share gains, prime balances up 40%+.
• Capital return is relentless and affordable: $4.1 billion in one quarter at an 85% payout ratio, the share count down 12% since FY2021, CET1 at 12.8% still well above requirements — every dollar of earnings buys more EPS than the year before.

Bear Case

• The provision leads charge-offs by $369 million and the build is explicitly macro-driven: if Citi's economists are right that conditions are turning, today's build is the first of several — and the 2.5% coverage ratio, down from 2.7%, leaves less cushion than a year ago.
• Fee income is decaying in the two biggest consumer lines: Markets non-interest revenue down, principal transactions down 18%, and U.S. Cards fee revenue down 62% to $(758) million as partner payments and rewards devour the spread — the "soft fee income" the market sold on.
• Mexico Consumer is deteriorating in real time: net credit losses up 46% on volume and seasoning, inside a franchise Citi is trying to exit — if U.S. cards follow that curve, the U.S. reserve build quadruples again.
• Returns peaked for now: RoTCE down from 13%+ in the first half to 11.5%, the low end of the 11–13% goal, while the new $175 million quarterly NCI drag (from $38 million) permanently diverts Banamex earnings from common shareholders.
• The balance sheet is outgrowing its capital: assets up 11% against a CET1 ratio down 50 basis points year over year, with a $(41.4) billion AOCI hole and an 85% payout ratio — management's own signals (a macro-driven reserve build, no upcycle language) concede the cycle is turning.

Our Take

This is a bank executing well into a late cycle, and the market is right to pay for the spread compounding while charging for the credit ambiguity — but $369 million of macro-driven reserves against flat charge-offs and falling non-accruals reads as prudence, not prophecy, and 11.5% RoTCE with a 58.5% efficiency ratio is a franchise working, not one peaking. Buy the Services deposit machine and the Wealth inflows; underwrite the buyback pace and the 14–15% return target against Mexico Consumer's 46% loss growth rather than assuming it. Citi has earned the benefit of the doubt — it just has to re-earn it every quarter until the macro outlook stops deteriorating.