C — Q2 2026 Financial Report Analysis

Citi's best quarter in years, and most of it is real

Citigroup earned $5.83 billion, or $3.15 per diluted share, in the three months to 30 June 2026, against $4.02 billion and $1.96 a year earlier — net income up 45%, earnings per share up 61%. Revenues, net of interest expense, rose 14% to $24.77 billion, and all five of Citi's operating businesses grew. Expenses rose only 5%, to $14.22 billion, so the gap between revenue growth and cost growth (what Citi calls operating leverage) was 960 basis points — 9.6 percentage points in the bank's favour.

The single most-watched number at Citi is return on tangible common equity, or RoTCE — the profit the bank generates for every dollar of shareholder money once goodwill and intangible assets are stripped out. It reached 13.0%, up from 8.7% a year ago. That is the first quarter in which Citi has printed a number in the range management has spent years promising.

Two things flatter the year-on-year comparison and are worth naming up front: the prior-year quarter carried an unusually heavy credit-reserve build, and the prior-year tax line carried a one-off benefit. Both are discussed below. The underlying operating improvement survives them.

The quarter in numbers

Two of those terms carry most of the weight. The efficiency ratio is simply operating expenses divided by revenue — the share of every dollar of income eaten by running the bank; lower is better, and a 5.3-point improvement in a year is large for an institution this size. The CET1 capital ratio measures the highest-quality loss-absorbing capital against risk-weighted assets; it is the constraint regulators impose on how much Citi can pay out to shareholders.

For the first six months, revenues were $49.40 billion (+14%), net income $11.62 billion (+44%), diluted EPS $6.21 against $3.92, RoTCE 13.1% against 8.9%, and the efficiency ratio 57.7% against 62.4%.

Where the growth came from

Services — cash management, payments and custody for corporate and institutional clients — is the closest thing Citi has to a franchise moat, and it delivered a 44% efficiency ratio. Revenue rose 18%, with net interest income up 18% on a 19% increase in average deposit balances, "largely driven by higher operating deposits, as Citi continues to deepen relationships with existing clients and onboard new clients." Operating deposits are the balances corporates must hold to run payroll and settle trade; they are stickier and cheaper than deposits chasing yield, which is why this line matters more than its size suggests. Securities Services fees benefited from a 22% rise in assets under custody and administration, part market appreciation and part new mandates. Cross-border transaction value rose 13% and U.S. dollar clearing volume 5%.

Markets grew 17%, but the composition needs care. Net interest income in Markets was $4.0 billion against $2.9 billion, up 42%, while non-interest revenue fell 5% to $3.0 billion. Citi explains why in the filing itself: it assesses Markets on total revenue "as security inventory is often hedged by derivative instruments, creating offsetting gains and losses across revenue lines." Much of the apparent NII surge is accounting geography, not new earnings. The genuine drivers are elsewhere: Equity Markets revenue rose 45% to $2.3 billion, on equity derivatives and prime brokerage, where "prime balances were up nearly 60%"; Spread Products and Other Fixed Income rose 25% on financing, credit trading and commodities. Fixed Income overall grew only 7%, with Rates and Currencies up just 1% — foreign exchange volumes were strong, "primarily offset by lower revenues in rates." So Citi's trading quarter was an equities-and-financing quarter, and financing-led revenue is balance-sheet-hungry: Markets average loans rose 29% and average assets 15%, to $1.41 trillion.

Banking is the biggest percentage mover: revenue +34%, net income $350 million against $93 million. Investment banking revenues rose 44%, with debt capital markets up 65% on leveraged finance and investment-grade issuance, and equity capital markets up 92% "amid very strong market conditions... with strength in IPOs and follow-on activity." Advisory fell 4% against a strong comparison. This is the most cyclical earnings stream Citi has, and a 92% ECM gain is a statement about the issuance window, not about durable share gains. Corporate lending, stripped of loan-hedge marks, actually declined 4% on "lower loan spreads and balances."

Wealth revenue rose 13% and net income 51%, with net interest income up 18% on better deposit spreads. Non-interest revenue rose only 4%, and the filing gives the reason: the absence of "an approximate $80 million gain on sale of an alternative investments fund platform in the second quarter of 2025, as well as the loss of fee revenue from the sale of a trust business in the third quarter of 2025." Underlying fee momentum is better than the 4% headline. Client investment assets rose 14%; net new investment assets were about $16 billion in the quarter and over $56 billion in the last twelve months, which Citi puts at 9% organic growth. At a 75% efficiency ratio, Wealth remains the weakest of the five on cost, even after a 7-point improvement.

U.S. Consumer Cards is the one business that did not really grow: revenue up 1%. Net interest income rose 5% on higher balances, but non-interest revenue swung to negative $659 million from negative $447 million. Card revenue is reported net of what Citi pays its co-brand partners, and that line — card rewards and partner payments — widened to $3.56 billion from $3.01 billion, alongside higher new-account acquisition costs. Interchange fees themselves rose 11% and card spend volume rose 12% to $162 billion. New credit card account acquisitions were 4.0 million against 1.7 million, up 135%, reflecting the completed purchase of the additional American Airlines co-branded card portfolio. Citi is buying growth in cards, and the cost is showing up immediately in revenue while the balances build more slowly — the segment's efficiency ratio worsened to 40% from 36%.

Credit costs: better, but read the reserve lines

Total provisions for credit losses and for benefits and claims were $2.52 billion, down 12%. That is worth decomposing, because the improvement is almost entirely in reserves rather than in actual losses.

A bank's provision line has two parts: net credit losses (loans actually written off, minus recoveries) and the change in the allowance for credit losses — the reserve set aside against loans expected to sour later. Citi's actual write-offs rose: net credit losses of $2.4 billion were up 8% from a year ago, "driven by increases in Banking and Legacy Franchises." What collapsed was the reserve build: $118 million this quarter against $638 million a year ago. The prior-year build was driven by "transfer risk, portfolio growth and changes to certain macroeconomic variables" — transfer risk being the danger that a country's controls stop a solvent borrower repaying in hard currency. Services alone took $286 million of provisions on other assets and held-to-maturity securities in Q2 2025 on that basis, against $7 million this year. Strip that out and the year-on-year provision improvement largely disappears.

Two smaller items deserve flags:

• In Banking, net credit losses jumped to $138 million from $16 million, but the filing says these "were driven by loan sales, which were previously reserved for," offset by "reserve releases on the loan sales." That is a balance-sheet cleanup passing through the loss line, not fresh deterioration.
• In U.S. Consumer Cards, the provision line shows a $272 million release on unfunded lending commitments. Per footnote 3, this is the reserve built in Q1 2026 against the pending American Airlines portfolio purchase, released when the deal closed and "re-established as a reserve for the loans that were acquired." It is a transfer between reserve lines, not a credit-quality improvement — and it means the $40 million reserve build shown against USCC loans contains that re-established amount, so the underlying seasonal release was considerably larger than it looks.

The consumer credit trend itself is genuinely improving. The annualized net credit loss rate on U.S. Consumer Cards fell to 4.19% from 4.43%, and on the general-purpose card book to 4.01% from 4.20%, with 90-day-plus delinquencies at 1.21% against 1.30%. Early-stage 30–89 day delinquencies ticked up to 1.17% from 1.12% — the one number moving the wrong way, and the one to watch next quarter given how fast Citi is adding accounts. Total allowance for credit losses on loans stood at $19.96 billion.

The tax line, and why pre-tax profit grew faster than net income

Pre-tax income from continuing operations rose 54%, to $8.03 billion. Net income to Citigroup rose 45%. Two things account for the gap.

First, tax: the effective rate rose to 25% from 23%, "largely due to the absence of a benefit recognized in the prior-year period related to a resolution of a tax audit." The prior-year quarter was flattered by that benefit, so the underlying earnings improvement is larger than the net income growth implies, not smaller.

Second, minority interests: income attributable to noncontrolling interests was $193 million against $14 million. This is a direct consequence of the Banamex sell-down — Citi completed the sale of a further 22.6% of Banamex's common stock on 29 April 2026, so a growing slice of that Mexican business's profit now belongs to other shareholders and is deducted before Citi's reported net income. Expect this deduction to persist and grow.

On Banamex more broadly: the sale of the remaining 1.4% of the committed 24% stake is expected in Q3 2026. Citi still carries roughly $9 billion of unrealized currency-translation losses attributable to Banamex, which will hit earnings when the business is classified as held-for-sale. Management says it "may recognize the CTA losses in earnings and deconsolidate Banamex in early 2027," and that the cumulative effect "will ultimately be regulatory capital neutral." That is a large, known, non-cash earnings hit sitting ahead of the stock — capital-neutral, but not headline-neutral.

Capital and payout

Citi returned $5.0 billion to common shareholders in the quarter: $4.0 billion of buybacks under the 2026 $30 billion repurchase programme, plus $1.0 billion of dividends. Year to date, repurchases are $10.3 billion against $3.75 billion in the first half of 2025. The buyback is doing real work on the per-share line — net income rose 45% while diluted EPS rose 61%, a gap explained almost entirely by a shrinking share count.

It is also visibly consuming capital. CET1 fell to 12.78% from 13.48% a year ago, which Citi attributes to "common share repurchases, the payment of common and preferred dividends and an increase in RWA, largely offset by net income." That leaves roughly 120 basis points of headroom over the regulatory requirement. Tangible book value per share rose 7% to $100.89 — buying back stock below tangible book adds to that figure, and Citi has been doing so. The total payout ratio for the half-year was 113% of net income available to common shareholders: Citi is currently returning more than it earns.

Management raised the quarterly common dividend from $0.60 to $0.67 per share, declared on 21 July 2026.

Takeaway: The 13.0% RoTCE is the headline, but the quality check is what produced it. Roughly half the improvement is durable — Services' 44% efficiency ratio on sticky operating deposits, a 5.3-point drop in the group efficiency ratio on 5% expense growth, and a genuinely improving card loss rate. The other half rests on things that will not repeat on the same terms: a prior-year quarter burdened by transfer-risk reserve builds, an investment banking window in which equity capital markets revenue nearly doubled, and a buyback converting 45% net income growth into 61% EPS growth while CET1 fell 70 basis points. Citi has finally shown it can hit its return target; it has not yet shown it can hit it in a quarter where markets are quiet and reserves have to be built rather than released.

What to watch

Citi does not publish formal financial guidance in its Form 10-Q, and none appears here. What the filing does commit to is the dividend increase to $0.67, the $30 billion repurchase authorisation, completion of the residual 1.4% Banamex sale in Q3 2026, and possible Banamex deconsolidation with the associated currency-translation charge in early 2027. Management's own stated risk framing is unchanged: "elevated inflation; conflicts in the Middle East; changes in U.S. laws or policies; and changes in interest rates and monetary policies" could affect results "during the remainder of 2026."

Our read on trajectory:

• Net interest income is less solid than the 13% headline. Non-Markets net interest income — the part driven by lending and deposits rather than trading-book funding — rose only 6%. Corporate/Other revenue went to negative $316 million from positive $25 million on "actions taken... to reduce Citi's asset sensitivity due to a lower interest rate environment," meaning Citi has deliberately traded away some rate upside for protection if rates fall. Net interest margin of 2.54% is up only 3 basis points year on year.
• Cards is the swing factor into 2027. Citi has added 4.0 million accounts in a quarter and absorbed a large co-brand portfolio. The partner-payment and acquisition costs land now; the interest income lands over the following years. If the 30–89 day delinquency uptick continues as those cohorts season, the provision line gets harder before the revenue arrives.
• The reserve tailwind is close to exhausted. With the allowance at $19.96 billion and macro assumptions already reflecting a softer outlook, there is limited room to keep flattering provisions with releases. From here, earnings growth has to come from revenue and cost, which is the harder version of the story — and the version that would prove 13% is a floor rather than a peak.