Cullen/Frost Bankers, Inc. (NYSE: CFR) — Texas’ Best Deposit Franchise, at the Richest Price It Has Ever Fetched

An independent analyst’s note. The body of this piece (Sections 1–15) contains no investment recommendation and no price target. The single deliberate exception is the clearly labeled Claude's Take block immediately below.

⚡ Claude’s Take

Claude’s own subjective opinion. This is general information, not investment advice, and not a recommendation to buy or sell any security. Do your own work. The body of the article (Sections 1–15) below carries no position and no price target.

HOLD at $170.08 — a genuinely excellent bank at a price that requires it to stay excellent. Accumulate on weakness toward ~2.1–2.3× tangible book, roughly $130–$142, which in practice means a Fed that cuts instead of hikes or a Texas credit scare, not a bad quarter. Fair-value zone: 2.4–2.8× reported tangible book value of $61.50/share, or approximately $148–$172, drifting up with book at roughly 9–10% a year now that the accumulated-other-comprehensive-income drag has stopped compounding against it. Not a short at any price I can see — the funding franchise is the real thing and the balance sheet has no hiding places.

Tag: You are paying for the deposits, and they are worth paying for — just not at any price.

Frost is the rarest thing in American banking: a bank that has grown by opening branches instead of buying them, and has been right to. Since the organic expansion began in late 2018 the consumer checking base is up 47% — a third of all Frost customers are new since then — and household growth accelerated to 5.7% year-over-year in the second quarter, the strongest quarter of customer additions since 2022, in a market where JPMorgan, Bank of America and a queue of acquirers are all building. The expansion branches now carry $3.0 billion of loans and $3.7 billion of deposits, 100,000-plus new households, and contributed 5.8% of second-quarter earnings per share while supplying 53% of loan growth and 72% of deposit growth. The payoff shows up exactly where a moat is supposed to show up: the June spot cost of total deposits was 1.11% with non-interest-bearing balances at 33.7% of a $43.3 billion base, against a fed funds target range of 3.50–3.75%. On Cullen/Frost’s own published comparison against the S&P Regional Bank Index, its blended deposit cost was 1.26% versus 2.02% for the peer group in 2025 — a 76 basis-point gap worth roughly $321 million of pre-tax income, or 41.6% of the company’s entire pre-tax income. That is why a bank with a mediocre 62.8% efficiency ratio still earns a 15.4% return on equity and 1.31% on assets. And the captivity behind it is not asserted: Frost has ranked highest in Texas in the J.D. Power retail banking study for seventeen consecutive years, scoring 757 against a Texas average of 681 and placing first in all eight dimensions.

So why not buy it? Three reasons, and none of them is that the bank is bad. First, price. At $170.08 the shares trade at 2.77× tangible book, 2.36× book, and 16.1× trailing earnings — the 95.8th percentile of Cullen/Frost’s own decade of price-to-book and the 86.3rd percentile of its own price-to-earnings. Zions delivers a 16.6% return on tangible common equity at 1.60× tangible book and Columbia 16% at 1.70×; Frost delivers about 18.6% (reported) at 2.77×. You are paying a 60–70% multiple premium for roughly a 12% return premium plus organic growth those two do not have. And the funding advantage is narrower against the banks that matter than against the index: in the second quarter of 2026 Frost’s total deposit cost was 1.08% against Commerce Bancshares at 1.11% and Regions at 1.18% — three and ten basis points — and both of those run efficiency ratios about two points better than Frost’s (58.4% and 58.3% against Frost’s ~60.4% on the same taxable-equivalent basis). Frost is at the top of a cluster, not alone at the top. Second, the reported returns flatter the truth, and the whole advantage is rate-contingent. Return on tangible common equity is computed against an equity base that a $852 million accumulated-other-comprehensive-income hole has hollowed out; add the mark back and the return is closer to 14.9%, the multiple on adjusted tangible book is 2.26×, and the FY2025 return on equity falls from 15.7% to about 12.5%. More importantly, this franchise earned 9.8–10.5% on equity every single year from 2009 to 2016 — the whole zero-rate decade — with the same branches, the same service culture and the same J.D. Power trophies. A funding advantage is worth nothing when funding is free. Third, and most concretely, the guidance embeds a rate hike that has not happened. Management’s raised full-year outlook explicitly assumes a 25-basis-point increase in September 2026; the chief financial officer put the sensitivity at about $2 million a month. Remove it and the beat-and-raise cadence the last 27% of the move was paid for gets harder.

Fourth, the incentive plan did not require a return on the capital management was rewarded for deploying. For FY2022 through FY2025 there was no return-on-capital metric anywhere in the programme: the annual bonus paid on net-income dollars against a self-set budget that was cut roughly 26% after the single year it was missed (FY2025’s celebrated “22% above budget” is still below the FY2023 budget the company failed to hit), and the long-term awards certifying in 2025–27 pay on absolute pre-provision-net-revenue growth — a metric the compensation committee stated in writing it chose “to align our performance award with our strategy of achieving organic growth by meaningful expansion.” It paid the maximum 150% for the 2022–24 cycle, i.e. executives were paid a maximum for the Federal Reserve’s tightening. The performance-contingent share of long-term pay was then halved, citing the same expansion. Relative return on assets has been restored, but no such cycle certifies until 2028. Meanwhile the buyback has not reduced the share count in a decade — diluted shares are up 1.8% since FY2016 — and was smallest ($39 million) when the stock was cheapest (~$97 in mid-2023) and largest ($160 million in six months) at the all-time high.

Fifth — and this is what the primary data added that the narrative does not — the expansion is not translating into statewide share. On FDIC Summary of Deposits data, Frost added 56 Texas offices between 2020 and 2025 (+36%), and its adjusted Texas deposit share fell from its 2022 peak of 4.16% to 3.93%, with deposits per office dropping from $267 million to $198 million — below the 2020 level of $210 million. Non-interest-bearing deposit dollars have compounded at 0.5% a year over five years while the branch count rose 51%; all the growth has come in interest-bearing and time form, and the non-interest-bearing share has fallen from 43.3% in 2021 to 32.9%. In Houston, after seven and a half years and seventeen new branches, Huntington’s pro-forma position has just passed Frost into sixth place by buying Veritex and Cadence. In Dallas-Fort Worth, twenty-eight new branches bought 29 basis points of adjusted share. The de novo model works — Frost’s branches out-earn the Texas de novo average by ~48% in years four and five — but a year-five branch reaches ~$62 million of deposits against a ~$295 million legacy average, so the programme dilutes average branch economics even as it adds absolute growth. Both things are true and only one of them is in the investor deck.

There is also a sixth thing worth saying plainly, because it is the least discussed and the most expensive lesson in this file. Between year-end 2021 and year-end 2025 Frost added $1,352 million to retained earnings and lost $1,190 million through accumulated other comprehensive income — the mark on a securities portfolio that management built to $19.1 billion of amortized cost while the loan-to-deposit ratio sat at 53%. AOCI consumed roughly 88% of four years of retained earnings. Tangible book per share was $56.89 at the end of 2021 and $59.62 at the end of 2025: four years of 15–18% reported returns produced 1.2% a year of tangible book growth. The bank was not reckless — it never sold, the securities are agency mortgage-backed paper and Texas municipals that are 68.9% Permanent-School-Fund-guaranteed or pre-refunded, capital never came close to a constraint, and the mark is now pulling back — but a shareholder who bought the 2021 story on reported returns waited four years to get anywhere. That is the risk of a 53% loan-to-deposit ratio in a bank that is structurally long duration, and it is the risk that recurs if long yields go up rather than down.

Framing: a quality compounder that has already been paid for — not a contrarian entry, not a falling knife, and, despite the tape, not a momentum trade either. The factor evidence is unusually clean on that last point. Over the 63 days in which Frost gained 27.2%, the Regional Banks industry factor returned −0.5% and the “Regional Bank Giants” basket −1.3%. This is an idiosyncratic re-rating on the company’s own results, not sector beta — a genuinely different setup from 2023, when Frost fell 47% for things other banks did. Yet its Momentum loading is zero to slightly negative (−0.024 to −0.042), because the twelve-month formation window still spans the April 2025 tariff drawdown; the stock’s actual style exposures are DividendYield (+0.89 to +1.09, its single largest), Value (+0.30), regional banks (+0.64), and SmallSize (+0.45), with Quality negative. The marginal holder owns this for income and value characteristics in a regime where DividendYield (z +1.77 over 252 days) and Value (z +1.61) are both working hard. That is a tailwind today and a specific, identifiable risk when the regime turns.

One thing genuinely cuts the other way and deserves to be weighed against all of the above. On 13 March 2023 — the first trading day after Silicon Valley Bank failed — the Chairman and Chief Executive bought 9,500 shares of his own company for $1.01 million of personal cash, joined the same day by two directors, the Chief Risk Officer and the Chief Credit Officer, with the company simultaneously repurchasing stock at roughly $97. That was the precise moment when Cullen/Frost’s AOCI-inclusive capital ratio was thinnest at 8.2%. And there are zero Rule 10b5-1 plans anywhere in five years of Form 4s, so every one of those purchases was a deliberate, discretionary act by exactly the people who would know if the securities book or the deposit base were a real problem. That is the best available evidence that the culture is what it claims to be — which matters a great deal here, because the pay plan is not.

Conviction: medium. Flips bullish if the expansion branches’ marginal economics are disclosed clearly enough to show the build-versus-buy arithmetic beating the roughly 15–20% deposit premiums being paid in Texas bank M&A and full-year 2027 delivers a second consecutive year of positive operating leverage without the September hike — that combination would make 2.8× tangible book a growth multiple rather than a cyclical peak. Flips bearish if the securities book takes a second duration hit (long yields up, AOCI re-widening) while deposit costs rise into the competition management already describes, because the same 53% loan-to-deposit ratio that funds the growth is what makes the balance sheet a bond portfolio with a bank attached.

📈 Stock Price Action — Five-Year Event Map

Over the trailing five years Cullen/Frost has round-tripped a full banking cycle and come out the other side at a record: from ~$116 in August 2021 it ran to ~$159 in November 2022, collapsed ~47% to ~$84 by October 2023 in the wake of the regional-bank failures, and has since compounded back to $170.08 on 2026-08-14 — a closing and intraday all-time high, 0.0% off its peak (prior nominal high $158.60, 2022-11-08). The 52-week range is $119.00 – $170.35, the shares sit above the 21-, 50- and 200-day exponential moving averages (~164 / ~158 / ~142) in strict ascending order, and the five-year total return is +70.1% (~+11.1%/yr) against a three-year +89.8% (~+23.8%/yr) — essentially all of the five-year gain was earned after the October 2023 trough.

Cycle narrative.

0. The March 2020 crash took the shares roughly in half in seven weeks; the single worst day in the entire price history is −21.4% on 2020-03-09, the day oil collapsed. It sits outside the sixty-month window but it frames the tail: Frost’s Texas energy and commercial-real-estate exposure is the market’s default panic button for this name.

1. The 2021–22 leg was pure rate beta. A balance sheet that is 60% floating-rate on the asset side and funded at close to zero on a third of its deposits repriced upward through the fastest tightening cycle in forty years, carrying the stock to what was then an all-time high of $158.60 on 2022-11-08.

2. That advantage inverted overnight in March 2023. Frost did not fail and was never close to failing, but it carried the two features the market punished — a very large securities portfolio and a majority-uninsured commercial deposit base — and lost 18% in the Silicon Valley Bank week alone, including −12.63% on 2023-03-13, the largest single-day decline of the last five years.

3. The damage compounded rather than resolved. First Republic’s seizure in May and the surge in long yields toward ~5% by October drove the shares to $83.66 on 2023-10-25 — a −47.3% peak-to-trough drawdown that took 38 months to recover, with the prior high not reclaimed until 2026-01-21.

4. The recovery began with the third-quarter 2023 print (+10.1% on 2023-10-26) and accelerated on the July 2024 CPI report, which rotated the market violently into small-caps and regional banks (+13.7% in five sessions to 2024-07-17).

5. The November 2024 election added a second leg — +12.6% in a single session on 2024-11-06 — as the market priced lighter regulation, a friendlier merger regime and a steeper curve for the whole group.

6. The tariff announcement of 2 April 2025 unwound most of that in four months, costing 26.5% peak-to-trough with −10.5% on 3 April, before the tariff pause produced an +8.2% snapback on 9 April. Neither move had anything to do with Frost.

7. The final leg is the most notable and the least dramatic: from the November 2025 low the shares have added ~42% with no single session better than +3.2% — a continuous re-rating rather than an event. The proximate catalysts are the June 2026 FOMC pivot that scrapped 2026 rate cuts and put a hike back on the table, and two consecutive beat-and-raise quarters ending with the 30 July 2026 second-quarter print: earnings per share of $2.70 against $2.39 a year earlier, return on assets 1.31%, return on equity 15.41%, margin 3.75%, with loan, net-interest-income and fee guidance raised and expense guidance cut. The print itself drew a −2.1% reaction — the market had largely anticipated it.

Price moves are FACT; attributed drivers are INTERPRETATION unless a dated filing or event is cited. This block is factual price history — it carries no recommendation and no price target; the opportunity judgment belongs to Claude’s Take above.

1. Executive Summary

Cullen/Frost Bankers is a $53.9 billion-asset bank holding company whose sole banking subsidiary, Frost Bank, has operated in Texas since 1868 and operates nowhere else. It runs 204 financial centers and about 1,760 automated-teller machines across seven Texas regions — San Antonio, Houston, Dallas, Fort Worth, Austin, the Gulf Coast and the Permian Basin — with 6,008 employees, two reported segments (Banking, and Frost Wealth Advisors), and $51.0 billion of trust assets of which $26.7 billion is managed.

The asset is the funding. At 30 June 2026 non-interest-bearing demand deposits were $14.58 billion, or 33.7% of a $43.33 billion deposit base. The average cost of interest-bearing deposits was 1.61% in the second quarter and the June spot cost of total deposits was 1.11% — against a fed funds target range of 3.50–3.75%. On Cullen/Frost’s own published series against the S&P Regional Bank Index, its blended deposit cost was 1.26% in 2025 against 2.02% for the peer group — a 76 basis-point gap worth roughly $321 million of pre-tax income, or 41.6% of the company’s $771.7 million of pre-tax income. That is, arithmetically, the whole of the excess return this bank earns. But the gap is rate-state dependent: it was zero in 2022 and averaged nineteen basis points from 2014 to 2017. The franchise is worth ~$320 million a year at a 3.75% funds rate and ~$30 million at zero.

It buys that funding with operating expense, and the trade is closer than the reputation suggests. Frost’s efficiency ratio was 62.8% in the second quarter of 2026 on a GAAP basis and roughly 60.4% on the taxable-equivalent basis peers report — against Commerce Bancshares at 58.4% and Regions Financial at 58.3% in the same quarter. That two-point gap is worth roughly $75 million a year on ~$2.3 billion of taxable-equivalent revenue, so the branch model gives back something like a quarter of the funding advantage it creates. Assets per employee are ~$9.0 million against $12–15 million at more centralized competitors. The branch-and-banker model is not a legacy inefficiency management has failed to fix; it is the mechanism that produces the deposits. The net of the two effects is positive but not overwhelming, and it is the single most important thing to understand about this company.

The organic expansion is working and is the reason to look at the stock at all. Since late 2018 Frost has opened branches into Houston, Dallas, Austin and, latterly, Fort Worth and infill San Antonio, rather than buying banks. Consumer checking accounts are up 47% over that period; a third of all Frost customers are new since the expansion began. Year-over-year consumer household growth accelerated to 5.7% in the second quarter of 2026, from 5.3% a year earlier, and 82% of new consumer customers over the last twelve months were 45 or younger. The expansion cohort now holds $3.0 billion of loans (13.4% of the total, up from 10.5% a year ago) and $3.7 billion of deposits (8.7%, up from 7.4%), delivered $0.16 or 5.8% of second-quarter earnings per share, and supplied 53% of loan growth and 72% of deposit growth. Management guides to five more branches in 2026.

Second-quarter 2026 was a clean, high-quality beat. Net income available to common shareholders of $170.4 million and diluted earnings per share of $2.70 (from $2.39) came with a 1.31% return on assets, a 15.41% return on equity, a 3.75% net interest margin — a fifteen-year high — and 140 basis points of positive operating leverage. Every guidance line moved the right way: net interest income growth raised to 4.75–5.25%, average loan growth to 7–8%, fee income growth to 7.5–8.5%, and expense growth cut to 4.5–5.0%.

Three things temper the picture, and all three are in the filings rather than the narrative. First, the balance sheet is a bond portfolio with a bank attached. Loans are $22.98 billion against $43.33 billion of deposits — a 53.0% loan-to-deposit ratio — with $21.48 billion of securities filling the gap. Between year-end 2021 and year-end 2025 Frost added $1,352 million to retained earnings and lost $1,190 million through accumulated other comprehensive income; the mark consumed roughly 88% of four years of retained earnings, and tangible book value per share compounded at 1.2% a year over that stretch despite 15–18% reported returns on equity. The unrealized loss is still $1.05 billion on the available-for-sale book plus $135 million unrecognized on held-to-maturity.

Second, credit is quiet but no longer improving in a straight line. Non-performing assets rose to $114 million, or 49 basis points of loans, from 33 basis points at the first quarter and 21 basis points a year ago, driven by a single $54–55 million Austin multifamily credit from the 2022 vintage with only a $1.5 million specific reserve. Criticized loans, encouragingly, fell to $917 million from $989 million. Commercial real estate is 47.8% of loans; non-owner-occupied plus construction and land is 29.1%. Management describes the competitive market as “a race to the bottom on structure” and says it is losing commercial-real-estate deals on terms it will not match — the right answer, and also an admission of where the marginal loan is being priced.

Third, the guidance rests on a rate hike that has not happened. The raised full-year outlook explicitly assumes a 25-basis-point September 2026 increase, worth roughly $2 million a month.

A fourth qualification comes from primary FDIC data rather than from the company: the branch build is running ahead of the deposits it produces. Frost added 56 Texas offices from 2020 to 2025 while its adjusted statewide deposit share fell from a 2022 peak of 4.16% to 3.93% and deposits per office declined from $267 million to $198 million. Non-interest-bearing deposit dollars have compounded at 0.5% a year over five years — flat — while the branch count rose 51%, and the non-interest-bearing share fell from 43.3% in 2021 to 32.9%. Frost has clearly outgrown its market on a five-year view (deposits +28% against a Texas market +9.6%, in a Texas deposit pool that is still 13.3% below its 2022 peak); it has not gained share since 2022.

The valuation is the crux. At $170.08 the shares trade at 2.77× tangible book value of $61.50, 2.36× book value of $72.04, and 16.1× trailing earnings of $10.58 — respectively the 95.8th and 86.3rd percentiles of Frost’s own multi-year price-to-book and price-to-earnings range, at a nominal all-time high. Adjusting tangible equity back for the AOCI mark gives $75.20 of tangible book per share, a 2.26× multiple, and a return on adjusted tangible common equity nearer 14.9% than the 18.6% headline. Against Zions at 1.60× tangible book on a 16.6% return and Columbia at 1.70× on 16%, Frost carries a 60–70% multiple premium. The premium is defensible — those two are shrinking or flat while Frost compounds households at 5.7% — but it is a premium for growth that is already in the price, not a discount to be closed.

Verdict on the business: a genuinely durable, narrow, well-run franchise with a real and measurable competitive advantage in deposit gathering, an above-average industry position in the best-growing large state in the country, and a capital-allocation record that is disciplined on acquisitions and questionable on duration.

2. Business Overview

2.1 What the company is

Cullen/Frost Bankers is a Texas financial holding company whose principal and sole banking subsidiary is Frost Bank, a Texas-chartered bank originally chartered as a national banking association in 1899 and traceable to a mercantile partnership formed in San Antonio in 1868. At 31 December 2025 Frost Bank had consolidated total assets of $53.1 billion; at 30 June 2026 the holding company had $53.9 billion. The company reports two segments:

• Banking — commercial and consumer banking, treasury management, and the insurance brokerage (Frost Insurance Agency) and brokerage (Frost Brokerage Services) subsidiaries. In the second quarter of 2026 this segment produced $167.1 million of the company’s $172.1 million of net income.
• Frost Wealth Advisors — private trust, estate administration, retirement services and investment management. Second-quarter 2026 revenue of $56.6 million and net income of $8.9 million, a pre-tax margin of 19.9%.

2.2 How it makes money

Frost is a spread lender, more so than most. Net interest income was 77.0% of total revenue in the first half of 2026 (77.7% in full-year 2025), which is high relative to diversified regionals and reflects both the size of the earning-asset base and the modest scale of the fee businesses.

The fee lines that are growing fastest — service charges on deposit accounts (+17.2%) and interchange (+16.5%) — are direct functions of customer count, not of markets or of pricing. Management made this point explicitly and it is the correct way to read the fee line: Frost has been steadily reducing the burden of overdraft charges (a $100 grace threshold, forgiveness levels that used to be double the industry’s) and overdraft revenue still grew at a double-digit rate, because the denominator of customers is growing 5.7% a year. That is an unusually clean read-through from franchise growth to fee income.

The wealth business is the weaker link. Trust assets were $51.0 billion at 31 December 2025 ($26.7 billion managed, $24.3 billion custody) against $51.4 billion ($26.2 billion / $25.2 billion) a year earlier — essentially flat in a rising equity market, with managed assets up modestly and custody down. Trust fee growth of 9% in the quarter is therefore mostly market appreciation plus a low-single-digit account-count gain, not share gain. And a 19.9% pre-tax segment margin is poor for a wealth manager; independent trust and investment businesses of this scale typically run 30–40%. Management has changed the leadership of the wealth business and expects benefits in “the back half of '27 and 2028,” while warning of adviser attrition in the interim. This is a real, unexercised option inside the company and it has been unexercised for a long time.

2.3 The loan book

Three observations. Commercial real estate at 47.8% of loans is high — but 18.7 points of that is owner-occupied, which behaves like commercial and industrial credit (the borrower’s business, not a tenant’s rent, services the debt). Non-owner-occupied plus construction and land is 29.1% of loans, or $6.69 billion, which is a normal-to-slightly-high figure for a commercial bank in a high-growth market and roughly 143% of total risk-based capital of $5.26 billion. Energy is 5.0% — the single largest industry concentration, ahead of automobile dealerships at 4.7% — and is far smaller than the market’s reflexive association with a San Antonio bank suggests; management describes energy borrowers as generating record cash flow and paying down revolvers.

The fastest-growing line is 1–4 family mortgage, up 49% in six months to $887 million, part of a deliberate consumer-acquisition push. Management disclosed the strategic logic and it is worth quoting in substance: Frost’s mortgage originations attract 45% new-to-bank customers, of whom 4,000 have added a checking or other account with average balances around $22,500 — above the average consumer deposit. This is a customer-acquisition channel financed by a low-yielding asset, and management said explicitly that it has been pricing aggressively to build the book to ~$1 billion and that “our pricing will tighten up” from here. That is a candid admission that the yield on this growth is temporarily subsidised, and it is the right way to read the guidance that loan yields are being held back by mix.

2.4 The deposit book and the funding gap

Deposits exceed loans by $20.36 billion. That gap is filled with $18.09 billion of available-for-sale securities, $3.39 billion of held-to-maturity securities, and $6.47 billion of cash and equivalents (funded also by $4.89 billion of customer repurchase agreements). Frost is therefore, structurally, a bank that lends about half its deposits and invests the rest — which is the source of both its safety and the balance-sheet problem discussed in Section 6.

The uninsured share of 52.6% looks high against the post-2023 benchmark, but it includes public funds that are collateralized by securities by law, and the composition is granular commercial operating balances rather than concentrated venture or crypto money. The $1.27 billion of Mexican-source deposits is a genuinely distinctive and sticky feature of a San Antonio franchise; it is small enough not to matter to the thesis and stable across the period examined.

Verdict. A simple, legible, single-state commercial and consumer bank with an unusually deposit-heavy balance sheet, a fee mix driven by customer count rather than markets, and one underperforming division (wealth) that has been underperforming for long enough to be treated as a structural feature rather than a near-term catalyst.

3. Industry Dynamics

3.1 The Texas deposit market — smaller and flatter than the narrative

The reflexive framing of Texas banking is that population growth, corporate relocations and energy make it a structurally expanding profit pool. The population and job data support that. The deposit data do not.

The Texas deposit pool shrank 13.3% from its 2022 stimulus-inflated peak and, at 30 June 2025, remained 13.3% below it. The statewide branch count is flat. The Dallas Federal Reserve’s own analysis of Eleventh District banks found district deposits grew 4.2% in 2025 against 4.8% nationally — Texas underperformed the nation on deposit growth. Any thesis that treats Texas as a rising tide for deposit gatherers is mis-specified: what Texas offers is a growing number of customers and businesses, competed for inside a deposit pool that is not growing faster than the country’s.

A second Dallas Fed finding cuts directly at the heart of the Frost story: its February 2026 research concluded that depositors are driven by rate, not by bank-specific risk or franchise quality, and that Eleventh District CD rates ran above national peers, peaking near 3% against a 2.5% national average. Texas is a higher-rate deposit market than the country. That makes Frost’s funding advantage more impressive, not less — but it also means the advantage is contestable by any competitor willing to pay up, which is precisely what the chief financial officer described on the second-quarter call.

3.2 Market structure — concentrated at the top, and Frost is not at the top

Raw Summary-of-Deposits data for Texas is unpublishable without normalization, because several institutions book enormous balances to a single branch: Charles Schwab’s three charters hold $240.6 billion — 16.1% of all Texas deposits — from three offices in Westlake; USAA holds $92.8 billion from one San Antonio office; Citibank $31.1 billion from two. Cullen/Frost’s own investor materials remove exactly these institutions, which is analytically correct and must be labeled. On that adjusted branch-banking basis (excluding the 207 charters with three or fewer Texas branches, removing $425.5 billion):

Frost is the fourth-largest branch bank in Texas and by a wide margin the largest bank headquartered in Texas — the next-largest Texas-domiciled institutions are Texas Capital at $26.5 billion and Prosperity at $26.3 billion. But the top three are money-centre banks holding 52% of the adjusted market between them.

The geographic distribution of Frost’s position is the central industry fact of this report:

Frost is dominant where the market is small and marginal where the market is large. San Antonio and the Permian Basin, where it ranks first, together represent roughly 5% of the Texas deposit market. Dallas-Fort Worth and Houston, where it holds 1.6% and 2.3% raw, together represent 61% of it. Everything in the growth story requires taking share in markets where Frost is a one-to-two-percent player against JPMorgan, Bank of America and Wells Fargo.

3.3 The consolidation wave — a verified, dated, one-time event

Texas was the most-targeted state for US bank M&A in 2025, with 21 deals proposed or completed through early November and seven of the twenty largest US deals by value involving Texas targets. The verified list of closings that matter to Frost’s competitive set:

(Corrections worth recording: FirstSun/HomeStreet was terminated in November 2024 after the Texas Department of Banking and the Federal Reserve declined it. Pinnacle/Synovus and PNC/FirstBank have no Texas branch footprint despite frequent inclusion in Texas M&A lists. Webster is being acquired by Banco Santander, not by Stellar — two unrelated deals. No credit-union acquisitions of Texas banks were found.)

Two structural consequences follow, and they point in opposite directions.

Favourable: essentially the entire $1–20 billion Texas mid-tier has been removed. Independent, Comerica, Cadence, Veritex, Stellar, Guaranty, Vista, Texas Partners, American Bank and Keystone are all gone as independent competitors. That is precisely the tier Frost fights for commercial relationships. San Antonio lost its largest independent commercial peer, Texas Partners, in February 2026. And there is hard, primary-source evidence that acquirer-led entry leaks deposits: the same 62 branches held $11.96 billion as Independent Bank at 30 June 2024 and $8.67 billion as SouthState a year later — a 27.5% decline in twelve months. (The deposit decline is fact from the FDIC; attributing all of it to post-merger runoff is interpretation, but it is the best available quantification of the mechanism Frost claims to harvest.)

Adverse: consolidation has imported two far better-capitalized competitors into exactly the markets where Frost is weakest. Fifth Third has committed $700 million-plus and 250 Texas financial centres by 2029 (from 108 post-conversion, with 60 in North Texas within three years), explicitly targeting top-five deposit share in Dallas, Austin and Houston. For scale: that is roughly 142 net new Texas branches in three years, against Frost’s ~69 in seven and a half. Huntington already holds the fifth-largest deposit share in both Dallas and Houston post-Veritex and post-Cadence and has pledged no Cadence branch closures. On a pro-forma basis Huntington ($8.17B) has edged past Frost ($8.01B) into sixth place in Houston — after Frost spent seven and a half years and seventeen new branches building there. That is the single sharpest rebuttal to the organic-versus-acquisition argument.

Timing matters for the near term. Most customer-facing conversions are still ahead: Fifth Third converts Comerica branches in September 2026, Huntington converts Cadence mid-2026, Prosperity/Texas Partners in November 2026 and Prosperity/Stellar in March 2027. The disruption dividend should therefore be larger in 2027 than in 2026 — and so should the competitive response.

One genuinely favourable and under-discussed data point: JPMorgan Chase’s February 2026 announcement of 160-plus new branches across 30-plus states in 2026 does not name Texas. Chase already has 481 Texas branches and roughly 20% raw share; it has largely completed its Texas build and is redeploying branch capital to the Southeast. The reflexive “Chase is coming for Texas” framing is not supported by Chase’s own disclosure.

3.4 Regulation — the most favourable band in US banking

At $53.9 billion of assets, Cullen/Frost sits above the scale at which compliance fixed costs bite and below every threshold that actually constrains a bank:

• Basel III Endgame is dead, and its replacement is relief. The FY2025 10-K states that the 2023 proposal “would not apply to us because Cullen/Frost and Frost Bank each have less than $100 billion in total consolidated assets.” That proposal was superseded by three separate proposals issued on 19 March 2026; the one covering non-Category-I/II firms including Cullen/Frost carries a stated reduction of roughly 8.6% in risk-weighted assets and in Common Equity Tier 1 requirements for covered depository institutions.
• The Liquidity Coverage Ratio and Net Stable Funding Ratio do not apply. Nor does supervisory or company-run stress testing (the company-run threshold is $250 billion), the supplementary leverage ratio, the countercyclical buffer, FR 2052a liquidity reporting, or a living will.
• The accumulated-other-comprehensive-income opt-out is preserved — and this is the sharpest regulatory point in the file. Every proposal from July 2023 through March 2026 confines mandatory AOCI recognition in regulatory capital to Categories III and IV, i.e. institutions above $100 billion. Cullen/Frost’s ~$1.05 billion net unrealized available-for-sale loss stays out of Common Equity Tier 1 solely because the company is under $100 billion in assets — the exact mechanism whose absence destroyed Silicon Valley Bank. Crossing $100 billion would recognize it. That is a quantified, structural disincentive to transformative acquisition, and it is a real argument for the organic strategy that Cullen/Frost has never made publicly.
• Operational-risk capital and the capital floor apply only to advanced-approaches organizations — not Frost.
• The one $50 billion rule that reached it — the FDIC’s June 2024 amendments requiring insured depository institutions between $50 billion and $100 billion to file resolution-planning informational filings on a three-year cycle — was suspended for Group B institutions including Frost Bank in December 2025 pending further amendments.

Growing organically at roughly 4% a year in assets, Frost has on the order of a decade before $100 billion becomes a live question. Against Fifth Third ($200B+) and Huntington ($200B+) now attacking Texas, this is a real and quantifiable regulatory cost advantage, and it is not discussed by the sell side.

3.5 Competitive intensity and the capital cycle

Applying the Marathon capital-cycle lens, the chief executive’s own description of the market is textbook: capital flooding in, new entrants buying share by relaxing covenants, the incumbent declining to match on structure while matching on price, and a two-year lag before the underwriting quality becomes observable.

“We’re seeing a pickup on competition on the lending side. And it’s mainly around structure… where I’ve seen more deals that we were unsuccessful on… it’s mainly been CRE. And that’s been some price, but a whole lot of structure. And it just seems like the market is continuing a bit of a race to the bottom on some of these structures.” — Phillip Green, 2026-07-30

“I think it will normalize after probably a couple of years. Some of these deals that are being made that are very structure-light, you’re never going to know if that’s a good loan or bad loan for another couple of years. And then if things soften up, they’ll see some things they wish they hadn’t done.” — Phillip Green, same call

The capital-cycle framework says the entrants’ losses arrive with a lag and the disciplined incumbent gains share on the far side. Frost’s own history is the strongest available evidence that it behaves this way: it was the only top-ten Texas bank to survive the 1980s without federal assistance or being merged. But the same framework warns that the damage shows up first in credits underwritten before the rate shock — and Frost’s own second-quarter non-performer was a 2022-vintage Austin multifamily loan, with the chief executive conceding “there are probably three more credits of that vintage.”

3.6 Verdict on the industry

Structurally mediocre, with one genuinely favourable local wrinkle. US commercial banking is a commodity business with high barriers to entry but no barriers to competition among incumbents; the long-run industry return on assets is 0.75% and return on common equity 10% (Banking 101, Deutsche Bank, 2011). Texas specifically offers strong customer-count growth inside a deposit pool that has been flat-to-down for three years, at rates above the national average, with the three money-centre banks holding half the market and two new $200 billion-plus entrants arriving with explicit share targets in Frost’s two largest expansion markets.

The favourable wrinkle is real and time-bounded: the wholesale removal of the Texas mid-tier by consolidation, plus a regulatory band that gives Frost roughly a decade of cost advantage over the entrants attacking it. A good place to be a well-run bank; not a good business in which to assume that being well-run is permanently rewarded.

4. Competitive Position

4.1 Naming the moat

In Greenwald’s taxonomy the correct classification is customer captivity (habit and search cost, not contractual switching cost) combined with local economies of scale — and the scale leg exists in San Antonio, the Permian Basin and Corpus Christi only.

The captivity mechanism is unusually well-evidenced for a soft-sounding claim:

• Frost Bank ranked highest in Texas in the J.D. Power US Retail Banking Satisfaction Study for the seventeenth consecutive year in the 2026 study — every year since J.D. Power began treating Texas as a separate region. The 2026 index score was 757 against a Texas regional average of 681, a 76-point gap, and Frost ranked first in all eight study dimensions (net promoter score, trust, people, account offerings, banking how and when you want, saving time and money, complaint resolution, digital channels). The study covers roughly 110,000 customers. Seventeen consecutive years is not noise.
• Coalition Greenwich awarded Frost fifteen 2025 excellence awards in small-business banking and twelve in middle-market banking, including Best Bank for Customer Service, Ease of Doing Business and Overall Satisfaction.
• The operating proof points are legible: 24/7 phone and chat support answered by humans; 53% of fourth-quarter 2025 consumer deposit account openings arriving through the online channel; a 4.9 App Store rating; 82% of checking households digitally engaged in the last month (from 79%); Zelle volume up 17.1%.
• Management’s own customer survey ranks the reasons customers choose Frost as (1) convenient locations, (2) reputation, (3) recommendation of a family member — with competitive interest rates near the bottom. (Label: management’s survey, not independently verifiable.)

4.2 The moat, priced in basis points

A moat claim that cannot be tied to a financial outcome is not a moat. This one can — and Cullen/Frost publishes the comparison itself, against the S&P Regional Bank Index peer group:

On $42.2 billion of FY2025 average deposits, a 76 basis-point gap is worth approximately $321 million of annual pre-tax income — 41.6% of Cullen/Frost’s $771.7 million of FY2025 pre-tax income, or roughly six to eight points of return on tangible common equity. That is the single most defensible quantification of the moat available, and it is the company’s own arithmetic against its own peer set.

A second, sharper caveat: the 76 basis points is measured against the index average, and the average includes badly funded banks. Measured against the best-funded regionals in the same quarter, the gap is far narrower:

(Deposit costs for CBSH and OZK computed from their Q2 2026 average-balance disclosures; RF discloses total deposit cost directly. Cullen/Frost discloses no efficiency ratio; 62.8% is computed on a GAAP basis and is roughly 60.4% on the taxable-equivalent basis the peers report — the like-for-like gap is about two points, not four. Efficiency-ratio definitions still differ across banks and are not strictly comparable.)

Frost is at the top of a cluster, not alone at the top. Against Commerce Bancshares its total deposit-cost advantage is three basis points; against Regions, ten. The 76 basis-point gap is real but it measures Frost against the whole regional index — which includes banks like Bank OZK at 2.86%, funded with 11.6% non-interest-bearing deposits. An investor paying a 60–70% multiple premium is paying it against banks that are also well funded, not against the index average. And note the corollary in the last column: Commerce and Regions achieve nearly the same funding cost with efficiency ratios about two points better than Frost’s on a like-for-like basis — and Commerce’s interest-bearing deposit cost of 1.57% actually beats Frost’s 1.61%.

And here is the caveat that the bull case usually omits: the advantage is rate-state dependent. In 2022 the gap was zero. In 2021 it was seven basis points. From 2014 to 2017 it averaged about nineteen. The franchise is worth roughly $320 million a year at a 3.75% funds rate and roughly $30 million a year at zero. Any valuation that capitalizes the current gap into perpetuity is capitalizing a rate regime, not a moat. The moat is the ability to be cheaper than peers; the size of the prize is set by the Federal Reserve.

The company’s own twenty-one-year return history makes the point unanswerably. Return on average common equity ran 18.8% and 18.0% in 2005–06, then 9.8% to 10.5% for the entire 2009–2016 zero-rate decade, then 14.2% (2018), 8.1% (2020), and 16.9–18.7% in 2022–23. Nothing about the franchise changed across those regimes. A funding advantage is worth nothing when funding is free. Any through-cycle earning-power estimate for Cullen/Frost must average across rate states rather than extrapolate the current one.

4.3 The pressure test the franchise partly fails: deposit mix

Frost’s non-interest-bearing deposit share has averaged 38% over the twenty years through 2025. The recent series is not encouraging:

Over five years, average non-interest-bearing balances grew at a compound rate of +0.5% while interest-bearing balances grew +9.6% and time deposits +42.4%. Non-interest-bearing dollars have been essentially flat while the branch count rose 51%. In 2025 the split of average growth was non-interest-bearing +0.6%, interest-bearing +4.3%, time +5.8%.

This is the same pattern visible at other deposit-rich regionals through this cycle, and it materially qualifies the sticky-cheap-deposit story. Frost is still cheaper than its peers at the margin — the 76 basis-point gap is measured, not asserted — but the free-money base is not growing; the growth is coming in paid form. A related tell: the chief financial officer put the interest-bearing deposit beta at 46% and guided it to the low 40s into a hiking cycle. A 46% up-beta is not the signature of captivity. The reconciliation is that Frost’s advantage lives in the level of interest-bearing pricing and in the 33% non-interest-bearing share, not in the beta. Whether that share stabilizes near 33% or continues toward the high twenties is the single most important open question on this franchise.

4.4 The share-stability test

Greenwald’s second test of a genuine advantage is market-share stability. The Texas record is mixed and, on the most recent stretch, negative:

Both readings are true and the article carries both. On a five-year view Frost has clearly outgrown its market: deposits +28% against a Texas market +9.6%. On a three-year view its adjusted share has fallen from the 2022 peak of 4.16% to 3.93%, and deposits per office have declined from $267 million to $198 million — now below the 2020 level of $210 million. Some of that is the 2022 stimulus peak washing out of everyone’s numbers. Some of it is real: the branch build is running ahead of the deposits it produces.

Market by market, the expansion returns are uneven:

• San Antonio — 29.0% adjusted share, first by a 2.5× margin over Bank of America, on a flat 34–35 branches. A genuine fortress, and mature.
• Austin — adjusted share up from 8.61% (2020) to 9.99% (2025) on 18 → 26 branches. The one market with a clean adjusted share gain.
• Houston — 1.97% → 2.30% raw on 46 → 63 branches. Seventeen branches bought 33 basis points of share, and Huntington has now passed Frost pro forma.
• Dallas-Fort Worth — 2.45% → 2.74% adjusted on 37 → 65 branches. Twenty-eight branches bought 29 basis points. The weakest return on expansion capital in the program.

4.5 The de novo economics, honestly stated

Frost’s own vintage curve (from FDIC data) shows its new branches reaching roughly $16 million of deposits in year one, then $27M, $37M, $50M and $62 million by year five, against all other Texas de novo branches at $18M / $29M / $36M / $41M / $42M. Frost’s branches out-earn the Texas de novo average by roughly 48% in years four and five — that is real evidence the brand travels.

But scale the result honestly. Legacy Frost branches average roughly $295 million of deposits; a year-five expansion branch reaches $62 million — about a fifth of legacy productivity. Implied expansion deposits per branch today are roughly $44 million across the cohort. The programme is simultaneously a genuine growth engine and a genuine dilution of average branch economics. Both statements are true, and the second is why deposits per office statewide have fallen from $267 million to $198 million.

4.6 Direct comparison with the competitive set

Frost is the only large Texas branch network meaningfully growing. Prosperity and Comerica shrank theirs. That is a genuine differentiator — and Frost’s 213 offices are still only 3.4% of the state’s 6,218.

4.7 Verdict

A durable advantage — narrower and more rate-dependent than the reputation implies, and geographically confined.

The moat is real. It has a named mechanism (customer captivity built on seventeen consecutive years of measured, externally audited service superiority), a measurable financial outcome (a 76 basis-point deposit-cost advantage worth roughly $321 million pre-tax, or 42% of pre-tax income), and a plausible reason it persists (a branch-and-banker cost structure competitors have chosen not to replicate because it depresses the efficiency ratio).

Three qualifications keep this from being a wide moat. First, it is rate-state dependent — worth $320 million at a 3.75% funds rate and $30 million at zero. Second, it is geographically confined — Frost is dominant in markets representing about 5% of Texas deposits and holds 1.6% and 2.3% of the two markets representing 61%. Third, the base is eroding at the edges — the non-interest-bearing share has fallen ten points since 2021, non-interest-bearing dollars are flat over five years while branches rose 51%, and adjusted statewide share has not grown since 2022.

The right characterization is not “crowded market with weak differentiation” — the J.D. Power record and the deposit-cost gap refute that decisively. It is: a genuine, measurable, local advantage that Frost is now spending capital to export into markets where it does not yet have one, with results that are good in Austin, adequate in Houston, and poor in Dallas.

5. Growth History and Forward Opportunities

5.1 The decade in numbers

Diluted EPS, ROA, ROE and DPS for 2023–2025 are as reported in the FY2025 Form 10-K (the two-class method produces slightly lower per-share figures than a simple net-income-over-shares calculation; the filing governs). Tangible book value per share is computed as common equity less $654.952M of goodwill, divided by period-end shares outstanding.

Two facts jump out of that table and they point in opposite directions.

First, earnings have compounded well. Diluted earnings per share grew from $4.70 in 2016 to $9.92 in 2025 — a 8.6% compound annual rate over nine years — with only two down years (2020, on COVID provisioning and rate collapse; 2024, on funding costs). Trailing-twelve-month earnings per share through the second quarter of 2026 are approximately $10.58, and the first half of 2026 was up 14.1% year over year ($5.35 vs $4.69). Revenue grew from $1.13 billion to $2.24 billion over the same nine years, a 7.9% compound rate.

Second, tangible book value per share went nowhere for four years. It was $56.89 at the end of 2021 and $59.62 at the end of 2025 — a compound rate of 1.2% a year while the company reported returns on equity of 10.3%, 15.7%, 18.7%, 15.8% and 15.7%. The entire gap is accumulated other comprehensive income. This is the single most important thing in the growth record and it is invisible in any earnings-based presentation of the company. Section 6 quantifies it.

5.2 The organic expansion — the growth that matters

Frost’s growth since late 2018 has been almost entirely de novo. There has been no bank acquisition. Instead the company committed capital to opening branches in markets where it already had a commercial presence but no retail distribution: Houston (announced 2018), Dallas (2022), Austin (2024), plus Fort Worth and infill San Antonio, and eleven further branches in trade areas outside the three announced expansions.

The disclosed results as of the second quarter of 2026:

And the franchise-level outputs:

• Consumer checking accounts +47% since the expansion began; roughly one-third of all Frost customers are new since 2018.
• Consumer household growth +5.7% year over year in the second quarter of 2026, accelerating from 5.3% a year earlier — described by management as the strongest quarter of customer growth since the second quarter of 2022.
• 82% of new consumer customers over the last twelve months are 45 or younger; 42% of the existing consumer base is millennial, Gen Y or Gen Z.
• New commercial relationships exceeded 1,000 for the fifth consecutive quarter; expansion markets supplied 33% of Houston’s new relationships, 39% of Dallas’s and 24% of Austin’s, and 22% of new commercial relationships company-wide.
• The 90-day weighted commercial loan pipeline reached $2.17 billion, the highest in the company’s history, up 11% sequentially and balanced roughly half commercial-and-industrial, half commercial real estate; 62% of it is existing-customer business.

Two qualifications on the household-growth figure, and they matter. First, it is not independently falsifiable. The 5.7% figure and the claim of “the best organic growth in the industry” appear only in the earnings-call script — not in any SEC filing — and the denominator has never been disclosed. The press release furnishes a different number (5.9%, including commercial) and the chief executive’s letter a third (5.3%). Of roughly twenty-one peers checked, thirteen disclose nothing comparable and every Texas competitor is silent, so no ranking can be constructed. The defensible restatement is: among the handful of US banks that disclose any household-growth figure at all, Frost’s is the highest — against Fifth Third at 3%, KeyCorp 3%, Huntington ~2% and Associated 1.4%.

Second, and more usefully, the metric management leads with is the one least connected to near-term earnings. Households grew 5.7–5.9%; average deposits grew 2.1% and average non-interest-bearing deposits 1.7%. The incremental household carries a far-below-average balance. That is entirely consistent with 82% of new customers being under 45 — a young household with a $2,000 balance is an excellent twenty-year option and a poor 2026 deposit — but an owner should price it as an option, not as current earnings.

Management’s own survey evidence on why customers choose Frost is worth recording precisely because it is testable and unglamorous: the number-one reason is convenient locations, second is reputation, third is recommendation of a family member, and competitive interest rates rank near the bottom. That is the profile of a business with genuine local customer captivity rather than a rate-shopping deposit base — and it is consistent with the observed funding cost.

5.3 Forward opportunities, ranked by credibility

1. Continued branch expansion (high credibility). Five more branches are planned for the balance of 2026, on top of five opened since the last call. The expansion cohort is only 8.7% of deposits after seven and a half years, so the runway inside Texas is long. The chief financial officer’s framing is the important one: adding 13–15 branches to a base of 130 is a smaller percentage cost increase than adding 10–15 to a base of 110, which is why expense growth guidance came down to 4.5–5.0% while revenue guidance went up. The operating leverage inflection is arithmetic, not aspiration — 140 basis points of positive operating leverage in the quarter and year to date, with management indicating 2027 should also deliver positive operating leverage.

2. Commercial share gain from market disruption (high credibility, finite duration). Management reports winning roughly 78–80% of opportunities against banks that have been acquired or are acquirers, and new relationships sourced from market disruption are up 65% year to date. This is a real, quantified benefit of being the largest Texas bank not participating in consolidation. It is also, by construction, temporary: it lasts as long as the current merger wave produces displaced bankers and unsettled customers.

3. Mortgage as a customer-acquisition channel (medium credibility). The 1–4 family book has grown from under $600 million to nearly $1 billion, attracting 45% new-to-bank customers. Management has already said pricing will tighten and refinance volumes will slow as long rates rise. Treat the deposit and relationship acquisition as the real product and the loan yield as the cost of acquisition.

4. Insurance brokerage cross-sell (medium credibility, long-dated). The chief executive was candid that Frost is below average penetration in selling commercial insurance to its own commercial customers, has put the business under the Chief Banking Officer, and is licensing bankers so they can share commissions. Insurance commissions were $14.2 million in the quarter, up 2.1%. The logic is sound and the base is small enough that success would be visible — but “we should be better at this” is a statement about the past as much as the future.

5. Wealth management (low credibility near-term). Trust assets are flat year over year in a rising market. New leadership is in place; management guides benefits to “the back half of '27 and 2028” and pre-announces possible adviser attrition. A 19.9% segment pre-tax margin against 30–40% for standalone peers is a large latent opportunity, but it has been latent for years.

6. Geography outside Texas (not a forward opportunity). Asked directly, the chief executive said he would consider it, believes the model would travel, and then said: “do not look for us to do that… in any foreseeable time.” Take him at his word. The addressable market is Texas.

Verdict: high-quality growth. It is organic, it is funded by customer acquisition rather than balance-sheet leverage, the incremental customers are young and bring deposits, and it is now producing measurable positive operating leverage. The only serious qualification is that earnings growth and tangible book growth have diverged violently for four years, and a shareholder is paid in the latter.

6. Financial Quality

6.1 Margin and the shape of the earnings power

The margin at 3.75% is a fifteen-year high and it is expanding for the right reasons: earning assets shifting out of cash held at the Federal Reserve into loans and securities, and fixed-rate assets repricing upward. The mechanics management disclosed for the back half of 2026 are specific and checkable — roughly $500 million of fixed-rate loans repricing at a spread of 120–125 basis points over what rolls off, approximately $1 billion of securities returning at 3.60–3.65% and reinvesting at 5.25–5.40% (a 170–180 basis point pickup), and a $250 million Treasury maturing in August at a sub-1% yield. Those are contractual, not forecast.

Two offsets. Deposit costs are rising again — up six basis points sequentially in the second quarter — and management guided the interest-bearing deposit beta down from 46% to the low 40s, which in a rising-rate environment means Frost expects to pass through less of a hike than it did in the last cycle. That is a bullish assumption resting on franchise strength; it should be tested each quarter. And the full-year guidance assumes a September 2026 hike worth roughly $2 million a month.

6.2 The efficiency ratio — the honest weakness

Cullen/Frost does not disclose an efficiency ratio, so it must be computed — and the basis matters here more than at most banks because roughly 40% of the securities book is tax-exempt Texas municipals. On a GAAP basis (non-interest expense over net interest income plus fees) the ratio was 62.8% in the second quarter and 63.2% for the half. On the taxable-equivalent basis that peers report, it is roughly 60.4%. Either way it is high: against Commerce Bancshares at 58.4% and Regions Financial at 58.3% in the same quarter — both taxable-equivalent — Frost runs about two points worse, and it has run in this range for years. Assets per employee are about $9.0 million (6,008 employees, $53.9 billion of assets) against $12–15 million at more centralized regionals.

This is not incompetence and it is not a fixable inefficiency. It is the cost structure of the strategy: 204 physical financial centers, 1,760 ATMs, human phone answering, and bankers calling on prospects for two years before winning them. The right way to evaluate it is as an explicit trade against the funding advantage:

(The deposit-cost comparison is Cullen/Frost’s own published series against the S&P Regional Bank Index, from its fourth-quarter 2025 investor presentation. Both efficiency ratios are stated taxable-equivalent so the comparison is like-for-like; definitions still differ across banks, so treat the expense line as an order-of-magnitude benchmark. Note that the funding advantage is measured against the index average while the expense disadvantage is measured against the best peers — the table is therefore generous to Frost on the revenue line and harsh on the cost line.)

On $771.7 million of FY2025 pre-tax income, a net ~$244 million is a genuine edge — on the order of four points of return on tangible common equity that Frost’s cost structure would otherwise forfeit. But read the table’s own asymmetry: the funding advantage is measured against the index average and the expense disadvantage against the best-run peers. Measured consistently against the best-funded peers on both lines — where Frost’s deposit-cost edge is three to ten basis points rather than seventy-six — the net advantage largely disappears. The correct statement is that Frost is comfortably better than the average regional bank and roughly level with the best ones, while paying more to get there. The moat is real and measurable; it is also expensive to operate, and the net is a fraction of the gross.

One further caveat, developed in Section 4.2 and repeated here because it belongs in any earnings-power discussion: the deposit-cost gap is rate-state dependent. It was 76 basis points in 2025, 64 in 2024, and zero in 2022; from 2014 to 2017 it averaged about nineteen. The franchise is worth roughly $320 million a year at a 3.75% funds rate and roughly $30 million at zero. The genuine excess return comes from the two effects plus the growth: the efficiency ratio is now improving (down 70 basis points year over year in the quarter) because the branch base has scaled, which is precisely the operating leverage inflection management is pointing at.

6.2a One earnings tailwind that is not operating leverage

A quality-of-earnings point that belongs beside the operating-leverage story: deposit-insurance expense fell from $76.6 million in 2023 to $37.3 million in 2024 to $18.8 million in 2025 as the FDIC’s special assessment for the 2023 bank failures rolled off (Cullen/Frost accrued $51.5 million of it in 2023, plus a further $9.0 million in 2024). That is a $57.8 million pre-tax swing over two years — roughly 7.5% of FY2025 pre-tax income — arriving in the same window in which the efficiency ratio began to improve. It is real money and it is non-recurring in the sense that it cannot repeat. Strip it out and the underlying expense discipline is good but less dramatic than the headline expense-growth guidance implies.

6.3 The securities book and the AOCI problem — the central finding

This is where the analysis of Cullen/Frost either becomes serious or stays at the level of the press release.

Net of $83.9 million of gross gains, the available-for-sale portfolio carries a $1.055 billion net unrealized loss; the held-to-maturity portfolio carries a further $135.2 million of loss that is not recognized anywhere. (On the second-quarter call management quoted “$1.15 billion” for the AFS book — that is the gross figure of $1.138 billion; the net is $1.055 billion. Immaterial, but the filing is the authority.)

Now the history that matters:

Read the bottom line carefully. Over four years Frost earned and retained $1,352 million and the securities mark took back $1,190 million of it — 88%. Common equity rose $133.5 million, or 3.1%, across four years in which the company reported returns on equity of 15.7%, 18.7%, 15.8% and 15.7%. Tangible book value per share rose $2.73, or 1.2% a year.

Three points of context, in fairness to management:

1. They did not sell. No losses were realized, no capital was destroyed in the accounting sense, and the securities pull to par as they mature. The 2026 reinvestment at 5.25–5.40% against 3.60–3.65% rolling off is precisely the mechanism by which the mark unwinds into earnings.
2. The credit quality is genuinely high. The mortgage-backed securities are all agency-issued. Every municipal bond in the available-for-sale book was issued by the State of Texas or a Texas political subdivision, and 68.9% are guaranteed by the Texas Permanent School Fund or pre-refunded. This is duration risk, not credit risk.
3. Capital was never a constraint. Common Equity Tier 1 was 13.95% at 30 June 2026 against a 7.00% requirement, and Frost has elected the accumulated-other-comprehensive-income opt-out, so the mark does not flow through regulatory capital. Even including it, CET1 would be roughly 11.4% — still comfortable.

But the fair criticism stands, and it is a capital-allocation criticism, not an accounting one. A 53% loan-to-deposit ratio is a choice. It means roughly $20 billion of a $43 billion deposit base is invested in securities rather than lent, and it makes the reported book value of the company a levered function of the long end of the yield curve. Duration was 4.9 years at the end of the second quarter (down from 5.2). If the ten-year Treasury rises 100 basis points from here — a scenario management’s own rate outlook makes more likely than a year ago — roughly $940 million of additional mark would appear in equity, or about $15 per share of tangible book, on a stock trading at $170.08. That is the mechanism by which a shareholder can be entirely right about the bank and lose money on the shares, and it is the same mechanism that produced the 47% drawdown in 2022–23.

6.4 Returns, correctly stated

The 18.6% figure is the one the market quotes and the 14.9% figure is the one an owner should use. Reported return on tangible common equity is inflated because the denominator has an $852 million hole in it that will close as the securities pull to par; as it closes, the numerator grows (higher reinvestment yields) but the denominator grows faster. Mechanically, reported ROTCE should decline over the next several years even if the bank performs perfectly. Any valuation anchored to a persistent 18–19% return is anchored to an artefact.

For context on the long-run standard: the US banking industry recorded a return on assets above 1% in only fourteen years between 1935 and 2010, with a long-run average of 0.75% and average return on common equity of 10% (Banking 101 — Large Cap Bank Primer, Deutsche Bank, May 2011). Frost’s 1.24% return on assets and 15.7% return on equity are well above that long-run norm. That is a compliment and a warning at the same time.

6.5 Credit quality

The headline deterioration is almost entirely one credit. Non-performing assets rose $41 million sequentially, of which $53.9 million is a single 2022-vintage Austin multifamily construction loan (offset by a $20 million paydown on a previously identified non-performer). Management’s account is that the sponsor is selling the property, a third-party equity partner declined to fund further, there is a guarantor, and the specific reserve is only $1.5 million — i.e. management expects “little if any” loss. The chief executive volunteered that roughly three more credits of the same 2022 vintage exist, mostly in Austin, and said he is not concerned about them because they have committed sponsors.

The genuinely encouraging datum is the one nobody asked about: criticized loans fell to $917 million from $989 million both sequentially and year over year, on resolutions anticipated in prior quarters. Criticized assets are a leading indicator; non-performing assets are a lagging one. A bank whose criticized book is shrinking while its non-performing book rises on one identified name is normalizing, not deteriorating.

Elsewhere: the office portfolio had a payoff and an upgrade in the quarter and, per the chief financial officer, now carries the highest debt-service coverage of any real-estate sector in the book. Energy borrowers are generating record cash flow and paying down revolvers — visible in the data, since $600 million-plus of newly booked revolving lines are drawn at under 10%. The one genuine soft spot management named is independent single-family homebuilders, squeezed by 6.25% mortgage rates at the entry and mid-tier and buying down rates out of margin; management expects risk-rating migration there but not losses, given strong balance sheets.

Full-year net charge-off guidance of 15–20 basis points is unchanged and is a benign number in absolute terms.

6.6 Capital and liquidity

Capital is abundant on any measure, including the punitive one. Risk-weighted assets are roughly $33.4 billion against $53.9 billion of total assets — a 62% density that reflects the very large holding of zero- and twenty-percent-weighted securities. Long-term borrowings are trivial for a bank this size: $123.3 million of junior subordinated debentures and $99.9 million of subordinated notes, plus $145.5 million of 4.45% Series B preferred stock. There is essentially no wholesale funding dependence; $4.89 billion of customer repurchase agreements is client sweep money, not market borrowing.

Liquidity is the mirror image of the balance-sheet criticism: $6.47 billion of cash and equivalents plus $18.09 billion of available-for-sale securities against $43.3 billion of deposits. The 2023 failure mode — an unmarked held-to-maturity book too large to sell — does not apply here in any meaningful degree, since held-to-maturity is only $3.39 billion, or 15.8% of the securities portfolio, with $135 million of unrecognized loss.

Verdict: do the economics improve with scale? Yes, and demonstrably so — but only recently and only in the operating-expense line. Positive operating leverage of 140 basis points in the quarter, an efficiency ratio down 70 basis points year over year, and a fixed branch base that new deposits scale over are all evidence that the expansion has passed its investment phase. The countervailing verdict is that the balance-sheet economics have not improved with scale: a bank that keeps half its deposits in securities compounds tangible book at the mercy of the long end of the curve, and the last four years are the proof.

7. Capital Allocation

7.1 The ten-year record

Over ten years Cullen/Frost earned $4,749 million, returned $2,489 million (52.4%) and retained $2,261 million. Dividends were 77.6% of everything returned; buybacks 22.4%.

The headline failure is in the last column. Diluted shares were 62.97 million in FY2016 and 64.10 million in FY2025 — up 1.8% over a decade despite $558 million of repurchases. For nine of the last ten years the buyback did no more than offset share-based-compensation dilution. Only in the last twelve months has the count genuinely fallen: 64.33 million in July 2025 to 62.15 million at 30 June 2026, a 3.4% reduction.

The dividend record is the genuinely good part. Dividends per share rose every year from $1.66 in 2008 to $3.95 in 2025, at a conservative 37–64% payout, with the most recent raises in Q2 2025 ($0.95 → $1.00, +5.3%) and Q2 2026 ($1.00 → $1.03, +3.0%), and the raise cadence moving a quarter earlier in each of the last two years. (On the streak: Cullen/Frost’s fourth-quarter 2025 investor presentation states 32 consecutive years of dividend increases. The proxies are internally inconsistent — “27 consecutive years” in the 2022 proxy and “30” in both the 2024 and 2026 proxies — and the figure appears in no 10-K or 10-Q. The investor-presentation number is the best-sourced of the three; the inconsistency is noted rather than resolved.)

7.2 Buyback execution — the discipline test, and a split verdict

The good news first, because it is real and it answers the most important question directly: Cullen/Frost did buy its own stock through the March 2023 regional-bank panic. The heaviest quarter of 2023 was the second — $28.0 million at roughly $96.7 a share, its cheapest print of the cycle — with a further $11.4 million in the third quarter at about $101.6. A bank that keeps buying while its own sector is being repriced for solvency is behaving correctly.

The criticism that survives is size, not price. The 2023 repurchases totalled $39 million — 6.5% of that year’s $598 million of net income — against a $100 million authorization that finished 61% unused. Cullen/Frost then spent $150 million at $124.67 in 2025 and $160 million at $137.61 in the first half of 2026, roughly four times the dollars at some 40% higher prices. Had the FY2025 $150 million been deployed at the second-quarter-2023 price it would have retired about 1.55 million shares instead of 1.20 million, 29% more. Buying most heavily into an all-time high is a momentum profile, not a value profile.

On multiples, every year’s repurchases were struck at roughly 2.2–2.4× reported tangible book — 2.4× in 2023, 2.3× in 2025, 2.25× in the first half of 2026. Adding the accumulated-other-comprehensive-income mark back, the 2026 buying is about 1.83× ex-AOCI tangible book and roughly 13× trailing earnings. That is not cheap, and it is defensible only if the mid-teens return on equity is durable.

7.3 The securities book — the decision that cost the most

This is the single largest capital-allocation act of the decade and it was not a good one.

Deposits rose from $27.6 billion at year-end 2019 to $42.7 billion at year-end 2021 — +$15.1 billion — while loans grew only from $14.6 billion to $16.1 billion. Management deployed the difference into securities, and the timing table is the crux:

Roughly $6.9 billion was deployed in 2021 and the first quarter of 2022 — at the bottom of the rate cycle — with a further ~$3.3 billion across the balance of 2022 at better yields, and ~$1.07 billion into held-to-maturity in the first quarter of 2023. The yield evidence confirms the timing: the securities portfolio’s taxable-equivalent yield fell from 3.29% in 2021 to 2.95% in 2022 — new money came on at rates low enough to drag the whole book down 34 basis points in a year — before recovering to 3.77% in 2025.

The price was severe. Available-for-sale gross unrealized losses went from $40 million at year-end 2021 to a peak of $2,212 million in the third quarter of 2023, and are $1,138 million today. Accumulated other comprehensive income swung from +$347.3 million to −$1,348.3 million during 2022 — a $1,695.6 million single-year move. Total shareholders’ equity fell from $4,439.6 million to $3,137.2 million in a year in which the company earned $579 million.

The regulatory capital ratios hide this, by election. Cullen/Frost has opted out of including most components of accumulated other comprehensive income in Common Equity Tier 1. Reported CET1 therefore ignores the mark entirely:

At the year-end 2022 trough the economic Common Equity Tier 1 ratio was roughly 460 basis points below the reported 12.85% — still above the 7.0% buffered minimum, but not the fortress the headline implied. A 4.47% tangible-common-equity-to-tangible-assets ratio is the same story in a different unit.

In fairness, the mitigations are real and they matter. Cullen/Frost never levered the trade — it still held $16.6 billion of cash at year-end 2021. 82% of the portfolio is available-for-sale, marked through AOCI where investors can see it, rather than parked in held-to-maturity to hide the loss — the opposite of what broke other banks in 2023. There is no credit risk: Treasuries, agency mortgage-backed securities, and Texas municipals of which roughly 69% are Permanent-School-Fund-guaranteed or pre-refunded. Duration of 4.9 years is mid-range, not a barbell. And management did not sell into the loss, so the mark accretes back — it has already improved from −$2,212 million to −$1,138 million.

Quantifying the opportunity cost as a bound: $6.9 billion deployed at a marginal yield low enough to pull the book to 2.95%, against roughly 5.3% available in cash at the Federal Reserve in 2023–24, is on the order of $138 million a year of foregone pre-tax net interest income across those two years — set against $1.0–1.4 billion of unrealized loss that never had to be realized. (ASSUMPTION: the counterfactual assumes cash rather than lending; it is a bound, not a point estimate.)

The honest verdict is a costly but survivable timing error, not a solvency error — and the “conservative bank” reputation does not survive it fully intact.

7.4 M&A: the discipline is management’s, not the plan’s

Cullen/Frost has made no bank acquisition in the period examined. Goodwill has been frozen at $654,952 thousand since at least the first quarter of 2020. The only acquisition line in five years of cash-flow statements is FY2024’s “net cash paid in insurance agency asset acquisitions — $703 thousand.” That is $0.7 million of M&A in five years, in the most-acquired banking market in America.

The 10-K’s own framing is worth quoting because it is not ideological: “While we are currently focused on organic growth, we may seek merger or acquisition partners that are culturally similar… Acquisitions typically involve the payment of a premium over book and market values, and, therefore, some dilution of our tangible book value and net income per common share may occur.” Cullen/Frost is not anti-M&A on paper. It is price-disciplined in practice, and its chief executive has said publicly he has “zero interest” and that acquirers “know I’m not in that game.”

In a market where Texas was the most-targeted state in the country in 2025 — 21 deals proposed or completed — refusing to participate is a genuine and rare act of capital discipline, and it is the strongest single item in this section.

7.5 The build: what it cost and what it earns

The alternative use of that capital was the branch programme. Purchases of premises and equipment over 2019 to the first half of 2026 total $1,002.3 million, taking net property, plant and equipment from $1,012 million to $1,370.2 million. Financial centres went from 157 at year-end 2021 to about 204 at year-end 2025.

What it produces, per management’s own disclosure: $3.0 billion of loans, $3.7 billion of deposits, 100,000-plus new households, and $0.30 of year-to-date earnings per share (5.9%).

The merits are substantial and specific. The deposits are core, with no acquired-deposit runoff, no goodwill, no tangible-book dilution, no integration risk and no regulatory-approval risk. Critically, the expansion branches run an 81% loan-to-deposit ratio ($3.0B / $3.7B) against 53% company-wide — they are materially improving the asset mix that the securities book degraded. And the programme is self-funding out of earnings, reversible and paced.

The costs are equally specific. By the company’s own disclosure the expansion was earnings-per-share dilutive every year from 2019 through 2024 and turned accretive only in 2025 — six years of dilution for the first tranche. Annualizing the disclosed first-half accretion of $0.30 gives roughly $0.60 a share, or about $37.5 million of net income; against $3.0 billion of expansion loans at 14% Common Equity Tier 1 on an 80% risk weight, allocated equity is roughly $336 million, implying an approximate 11% return on allocated capital — below the 15.7% consolidated return on equity and only marginally above a bank cost of equity. Include even a third of the $1.0 billion of cumulative capital expenditure as expansion-specific and the figure falls further. (ASSUMPTION-HEAVY: Cullen/Frost discloses expansion EPS accretion but never the capital invested in the programme or its return. That disclosure gap is itself a finding.)

(One comparison this report cannot complete: deposit premiums paid in recent Texas bank M&A were not obtained from a primary source, so the build-versus-buy head-to-head is qualitative here. The company’s own materials claim $101 million of cost per $1 billion of Houston deposits against a Texas M&A median of $175 million; that is management’s arithmetic and it excludes the six years of earnings dilution.)

7.6 Financing structure

• Series B preferred: 150,000 shares, $150.0 million liquidation preference, 4.450% non-cumulative perpetual, issued 19 November 2020 (net proceeds $145.5 million), replacing a higher-coupon Series A redeemed the same year. The first optional call date of 15 December 2025 passed and Cullen/Frost did not call it. That is the correct decision: the depositary shares traded at $16.93 against a $25 call price in May 2025 — a 32% discount — so retiring it would cost $25 to extinguish a liability the market values at $17. (The Chief Accounting Officer bought 1,870 of those depositary shares personally at $16.93.)
• Subordinated notes: $100.0 million of 4.50% notes issued March 2017, maturing 17 March 2027. Per the Q2 2026 10-Q, none qualified as Tier 2 capital at 30 June 2026 because the notes are within a year of maturity. A refinance-or-repay decision lands in roughly seven months with no 8-K yet.
• Junior subordinated debentures: $123.7 million to Cullen/Frost Capital Trust II (2004), three-month term SOFR plus ~1.81% (5.60% at 31 December 2025), maturing 2034.
• No common equity has been raised. Total holding-company debt is $223 million against $53.9 billion of assets — trivial.

7.7 Compensation and incentive alignment — the sharpest criticism in this report

Chief executive Phillip Green’s FY2025 compensation totalled $7,625,974 (salary $1,260,000; non-equity incentive $2,075,220; stock awards $4,000,081; other $274,582), a chief-executive pay ratio of 99:1. Chief financial officer Daniel Geddes totalled $1,700,948. Absolute pay is restrained for a $10.5 billion-market-capitalization bank: no employment agreements, no excise-tax gross-ups, a double-trigger change-in-control plan, a Rule 10D-1 clawback adopted October 2023, and both hedging and pledging prohibited. Say-on-pay support ran 96.6–99.1% across the last five years.

The design, however, does not require a return on the capital it rewards management for deploying.

The annual incentive’s only quantitative metric is net income against an internally set budget. The FY2025 budget was $527 million of net income available to common; actual was roughly $642 million, or 122% of budget, and every named executive was paid 122% of target. The chief executive’s plan is otherwise scored on seven purely qualitative measures with no disclosed weights; the plan for other executives states in terms that “the individual targets are not formula-driven and no specific weighting is targeted” and that “there is not a stated cap on this plan.”

The bar moves, and it moved down after the one year it was missed:

Two things follow. Upside tracks roughly one-for-one while downside is cushioned by discretion — a 17% miss produced only a 10-point pay cut. And after the miss the bar was reset roughly 26% lower: FY2025’s celebrated $642 million, “22% above budget,” is still below the FY2023 budget the company failed to hit.

The long-term plan is worse, and the company said why in writing. The performance-share metric history:

From the 2024 proxy, verbatim: “The Committee determined to maintain this growth metric in order to align our performance award with our strategy of achieving organic growth by meaningful expansion in existing markets.”

For the four years FY2022 through FY2025 there was no return-on-capital metric anywhere in Cullen/Frost’s incentive programme. The annual plan paid on net-income dollars against a self-set budget; the long-term cycles certifying in 2025 and 2026 paid on absolute pre-provision-net-revenue growth — a metric with no denominator, no per-share element, no cost of capital and no relative benchmark, chosen explicitly to reward the branch build. It paid the maximum 150% for the 2022–24 cycle, a period that captured the sharpest Federal Reserve tightening in forty years; pre-provision net revenue is dominated by net interest income, and net interest income rose mechanically on a deposit-funded balance sheet. Executives were paid a maximum for an exogenous rate move.

Relative total shareholder return is used nowhere. Return on equity and return on tangible common equity are used nowhere. Nothing anywhere penalizes tangible-book-per-share dilution or share issuance. And in October 2024 the performance-share mix was cut from 50/50 to 25% performance units and 75% time-vesting restricted stock, justified verbatim as appropriate “in this time of interest rate volatility, coupled with our expansion strategy” — so only about 19% of the chief executive’s FY2025 long-term grant value is performance-contingent. The fix — relative return on average assets — is real, but no such cycle certifies until 2028.

The mitigations, stated fairly: pre-provision net revenue nets out non-interest expense and charge-offs, so it is an earnings metric rather than a raw asset- or deposit-growth target; performance units are only a quarter of the long-term award; and the committee did restore relative return on assets for the 2024 and 2025 grants.

7.8 Ownership, board and succession

All Section 16 directors and executive officers as a group — 26 people — hold 718,791 shares, or 1.14% of the company. Green holds 151,217 shares (0.24%). Eight of fourteen director nominees hold under 10,000 shares and five hold zero or one share. Ownership guidelines exist (5× salary for the chief executive, 3× for other executives, 5× the cash retainer for directors) but unvested restricted stock counts toward them. Institutional holders dominate: Vanguard 10.0%, BlackRock 9.1%, Aristotle Capital 8.7%, State Street 5.2%.

The Frost family has exited entirely. Patrick B. Frost was replaced as President of Frost Bank by Paul H. Bracher in January 2024 and was last elected a director in April 2023. Form 4 data show he sold 128,762 shares and gifted 182,109 across the period. For a franchise whose entire competitive advantage rests on a service culture traceable to an 1868 partnership, the end of family representation in both management and the boardroom is a legitimate continuity question, even if no near-term effect is observable.

Succession disclosure is essentially absent. Phillip Green is 71; the nominal heir, Paul Bracher, is 69. Daniel Geddes, the chief financial officer since 1 January 2025, is 51 — and notably was Market President and Houston Expansion Lead from October 2018 to August 2021, i.e. he ran the programme that is now the growth story. Succession disclosure across five proxies consists of four boilerplate committee-charter references: no process narrative, no emergency plan, no timeline, no named candidate. Green’s perquisites rose from roughly $74,000 in FY2024 to $187,737 in FY2025 — security $112,532 (up 3.5×) and personal charter aircraft $55,797 — with no explanation given.

The board has fourteen nominees, thirteen independent, no classified board, and a combined Chairman/Chief Executive. Lead Independent Director Charles W. Matthews is 81, has sixteen years’ tenure, sits on all six committees and chairs three, and consistently draws roughly three times the board-average against-vote. No mandatory retirement age is disclosed in any of the five proxies. The stated rationale for combining the chairman and chief-executive roles is that “as we have traditionally combined these roles, separating them could cause significant disruption” — a tradition argument, not a governance one.

Two further flags, both minor individually: Jeffrey M. Rummel, elected 28 January 2026 to the Audit and Risk committees, was Ernst & Young’s San Antonio Office Managing Partner from 2020 to June 2025 — and Ernst & Young is Cullen/Frost’s own auditor, a roughly seven-month cooling-off period. And three of thirteen “independent” directors had commercial dealings with the bank in 2025: $1,006,471 to Pape-Dawson engineering (whose chief executive Samuel G. Dawson sits on the Compensation and Nominating/Governance committees) during a physical branch build-out; $282,090 of branch lease payments to an entity part-owned by director Crawford H. Edwards; and $217,767 to James Avery Craftsman.

7.9 Insider transactions — the best evidence in the file

A parse of all 264 Form 4, 13 Form 3 and 4 Form 5 filings in the mirrored corpus (integrity verified: 370 files on disk against 370 manifest rows) yields 544 transaction lines and one finding that is worth more than the rest of this section combined.

There were thirteen open-market purchases in five years, and twelve of them fall in a nine-week window:

Silicon Valley Bank failed on 10 March 2023. The Green, Avery, Willome and Severyn purchases are dated 13 March 2023 — the first trading day after. In total 28,200 common shares, roughly $2.9 million, at an average of about $104. The buyers include the sitting Chairman and Chief Executive, the Chief Risk Officer and the Chief Credit Officer — precisely the group that would know if the securities mark or the deposit base were a real problem. It is corroborated by the company simultaneously repurchasing stock, and by the fact that the economic (AOCI-inclusive) capital ratio was at its thinnest point in the decade at that moment.

A second finding raises the information content of everything above: there are zero Rule 10b5-1 plans. Of the 120 Form 4 filings carrying the SEC’s structured 10b5-1 checkbox (mandatory since February 2023), 100% report “0”, and a text search for “trading plan” across all 264 filings returns nothing. Every insider sale at Cullen/Frost in the last three and a half years is discretionary.

The counterweight is that selling has been larger. Over five years insiders sold 388,814 shares for roughly $51.3 million; Green alone sold 112,080 shares — 75,575 in November 2021 at about $136.70 and 34,505 in November 2024 at about $137.60 — against 9,500 bought. He is a net seller of shares by a factor of nearly twelve, though he ended the period with more shares than he began.

7.10 Verdict

Mixed, improving, and better than the balance sheet suggests but worse than the reputation.

Three things are genuinely well done. Refusing to overpay for Texas franchises while every peer consolidated — goodwill frozen at $655 million, zero tangible-book dilution, zero integration risk, in the most-acquired banking market in the country. A multi-decade dividend record funded at a conservative payout. And the decision to leave a 4.45% perpetual preferred outstanding when the market values it at a 32% discount to its call price — a small, correct, unglamorous choice that many boards get wrong.

Three things are not. The 2021–Q1 2022 deployment of roughly $6.9 billion of pandemic deposits into securities at the bottom of the rate cycle, which cost $1,695.6 million of accumulated other comprehensive income in a single year, took economic Common Equity Tier 1 to 8.23% and tangible common equity to 4.47% of assets, and still leaves a $1.19 billion pre-tax mark. A buyback that in a decade has not reduced the share count — up 1.8% since FY2016 — and that was smallest ($39 million) when the stock was cheapest and largest ($160 million in six months) at an all-time high. And an incentive plan that for four straight years contained no return-on-capital metric at all, paying maximum for absolute revenue growth delivered by the Federal Reserve, with the performance-contingent share of long-term pay cut in half in the same period, explicitly to accommodate the branch strategy.

The March 2023 buying — by the company and by the chief executive personally, three days after Silicon Valley Bank failed — is the single best evidence of genuine discipline in this file. The 2026 repurchase pace at $138 a share into an all-time high is the single best evidence against it. The alignment case ultimately rests on culture and personal restraint rather than on plan design, with 1.14% aggregate insider ownership across 26 people, a departed founding family, and a 71-year-old combined Chairman and Chief Executive whose named deputy is 69 and whose succession plan is not disclosed.

8. Changes and Headwinds — Last Two Years

8.1 Company-specific changes

Leadership. Daniel J. Geddes became Chief Financial Officer effective 1 January 2025 (announced by 8-K on 1 July 2024, compensation approved by 8-K/A on 1 November 2024), succeeding Jerry Salinas. Geddes joined Frost in 1998 and — the relevant detail — was Market President and Houston Expansion Lead from October 2018 to August 2021, then San Antonio Region President. The person now presenting the expansion’s economics is the person who built its first phase.

The Frost family exited. Patrick B. Frost was replaced as President of Frost Bank by Paul H. Bracher in January 2024 and was last elected a director in April 2023. He sold 128,762 shares and gifted 182,109 over the period.

Board turnover. Andrade joined in 2024; Engates in April 2025; Rummel and Shields on 28 January 2026. Alvarez and Patrick Frost departed in 2024; Avery in 2026. No reason is disclosed for any of the three non-renominations.

Capital return stepped up sharply. The buyback authorization doubled from $150 million (January 2025) to $300 million (January 2026), and deployment went from $50 million in 2024 to $150 million in 2025 to $160 million in the first half of 2026 alone. The dividend was raised in Q2 2025 (+5.3%) and again in Q2 2026 (+3.0%), each a quarter earlier than the historical cadence.

The expansion entered its accretive phase. By the company’s own disclosure the programme was earnings-dilutive every year from 2019 through 2024 and turned accretive in 2025; it delivered $0.16 (5.8% of earnings per share) in the second quarter of 2026. Eleven branches have now been opened outside the announced Houston, Dallas and Austin programmes. The Austin phase’s stated target of seventeen centres “by 2026” is a clean, testable management-credibility check at the fourth-quarter 2026 print — only one of the five branches opened in the second quarter was in Austin.

A financing decision lands in seven months. $100 million of 4.50% subordinated notes mature on 17 March 2027 and no longer count as Tier 2 capital. No 8-K has addressed refinancing or repayment.

8.2 Industry changes

The Texas mid-tier was consolidated out of existence. Between January 2025 and July 2026, Independent Bank Group, Comerica, Cadence, Veritex, Stellar, Guaranty, Vista, Texas Partners, American Bank and Keystone all ceased to exist as independent competitors. Texas was the most-targeted state for US bank M&A in 2025. San Antonio lost its largest independent commercial peer, Texas Partners, on 1 February 2026.

Two much larger competitors arrived in exchange. Fifth Third closed Comerica on 2 February 2026 and has committed $700 million-plus and 250 Texas financial centres by 2029, explicitly targeting top-five deposit share in Dallas, Austin and Houston. Huntington closed Veritex (October 2025) and Cadence (February 2026), already holds fifth place in both Dallas and Houston, has pledged no Cadence branch closures — and on a pro-forma basis has passed Frost into sixth place in Houston. Cullen/Frost’s own 10-K names the mechanism: deregulation “raised the asset thresholds at which more onerous requirements apply, which could cause certain large bank holding companies with less than $250 billion in total consolidated assets… to become more competitive or to pursue expansion more aggressively.”

Most of the customer-facing disruption is still ahead. Fifth Third converts Comerica branches in September 2026; Huntington converts Cadence mid-2026; Prosperity/American Bank in September 2026, Prosperity/Texas Partners in November 2026, and Prosperity/Stellar in March 2027. The disruption dividend should be larger in 2027 than in 2026 — and so should the competitive response.

One favourable non-event: JPMorgan Chase’s February 2026 announcement of 160-plus new branches across 30-plus states in 2026 does not name Texas. Chase has largely completed its Texas build and is redeploying branch capital to the Southeast.

8.3 Macro headwinds — the two-speed Texas economy

This is where the genuinely new information sits, and it is not in the investor deck.

The demographic engine is decelerating hard. Census Vintage 2025 shows Texas added 391,243 people in the year to 1 July 2025, down 30.5% from 562,941 in 2024; the growth rate fell from 1.8% to 1.2%, the slowest since 2021, and Texas fell from third to fourth among states by growth rate. Net domestic in-migration collapsed from 222,154 in 2022 to 67,299 in 2025 — roughly a 70% decline in three years — and Texas is now second to North Carolina on that measure. Net international migration fell 48%. In 2024 international migration supplied 56.8% of Texas’s population change, so the 2025 immigration decline hit the largest single driver. The state demographer’s assessment: “our economic growth isn’t probably going to be as hearty as what we’ve seen historically.” No Texas metropolitan area is in the national top ten by percentage growth, and San Antonio — Frost’s fortress — is the slowest-growing of the big four at 1.4%. Texas unemployment at 4.4% now equals the national rate; the historical Texas labour-market premium is gone.

Texas housing has fallen for eleven consecutive months. The statewide median price was $335,000 in April 2026, down 0.9% year over year; months of inventory 5.2 and rising; building permits down 7.4% year to date. By metro: Austin −3.3%, San Antonio −1.9%, Houston −1.7%, Dallas −1.3%. Austin’s average Zillow home value is −24.5% from its May 2022 peak. Homeowners insurance rates are rising 18–28% in 2026 on top of double-digit increases in 2024 and 2025; nine of the ten most financially significant US rate increases were in Texas.

Multifamily is the worst market in the country, and it is the fastest-growing loan category at Texas banks. Texas has the highest rental vacancy rate of any US state at 13.3%; Austin asking rents are down 16% from 2022; concessions run six to eight weeks free rent and up to twelve in some submarkets, with Austin leading and Dallas second. The Dallas Federal Reserve’s own March 2026 research states that multifamily loans are the fastest-growing loan category at Texas banks since 2021, notes rising delinquencies, and finds that loan modifications in Texas exceed the US rate overall. Cullen/Frost’s multifamily exposure is $795 million (3.5% of loans) — it is not overweight relative to peers — but its second-quarter non-performer was a $54–55 million Austin multifamily credit of the 2022 vintage, and its chief executive conceded there are “probably three more credits of that vintage… a couple of them are in Austin.” The macro data corroborate the company’s own disclosure precisely.

Against that, the commercial engines re-accelerated through mid-2026. Energy is a 2026 tailwind, not a headwind: West Texas Intermediate was $81–82 in mid-August 2026 with a Dallas Fed survey-period average of $87.27 in June; the Permian rig count of 263 is the highest since July 2025; the Dallas Fed energy activity index jumped to 46.1 in the second quarter from 21.0 in the first, the strongest since 2022. The Dallas Fed’s bank loan-demand index rose to 42.2 in June and commercial-real-estate loan volume jumped from 24.5 to 40.3 in a single quarter — directly supportive of Frost’s record $2.17 billion pipeline, and also a warning that everyone is chasing the same deals.

And one brand-new policy shock, twelve days before this report date. On 3 August 2026 Governor Abbott ordered a moratorium on new data-centre development pending a Public Utility Commission and ERCOT audit of every project in the interconnection queue. Texas had $89.5 billion of active data-centre construction and 248 planned facilities; the order puts roughly 49.8 gigawatts — about 20% of the entire US data-centre pipeline — at risk of delay, and the Energy Information Administration has already cut its Texas electricity-demand growth projection from 14% to 6%. This matters because the Dallas Fed’s 17 July 2026 upgrade of its Texas job-growth forecast to +2.0% leaned explicitly on “the data center buildout” — and that upgrade predates the moratorium. Cullen/Frost’s direct exposure is modest (construction and land is $2.72 billion; a named data-centre customer produced a ~$700 million deposit inflow and exit in the fourth quarter of 2025), but the second-order exposure to Texas construction and commercial capital expenditure is not.

Separately, the semiconductor construction wave is rolling off rather than building: Texas Instruments is cutting 2026 capital expenditure to $2–3 billion from $4.6 billion in 2025 as its Sherman project completes, and Samsung’s Taylor Fab 1 is finishing with volume production slipping to 2027. That is a headwind to North Texas construction lending — in the same Dallas market where Frost has spent the most expansion capital for the least share.

8.4 What did not happen — and it counts as evidence

Across sixty months of SEC filings there is no Item 4.01 (Ernst & Young was ratified every year), no Item 4.02 (no restatement), no Item 1.05 (no material cybersecurity incident, despite that disclosure regime taking effect in December 2023), no credit-rating action, no enforcement action, consent order or memorandum of understanding, and no Item 1.01, 2.03, 2.04 or 3.01. Item 3 of the FY2025 10-K is pure boilerplate with no named litigation.

That is real evidence for operational conservatism, in a way the securities book is not. A bank that has gone five years without a restatement, an enforcement action, a cyber disclosure or a rating downgrade is being run carefully at the operating level, whatever one concludes about its duration decisions.

(Two disclosure-practice notes for anyone working the filings: Cullen/Frost files no standalone dividend 8-K and puts buyback authorizations inside earnings-release exhibits — cite the 10-K, not the 8-K, for repurchase authorizations. And several senior officer appointments were never 8-K’d, including the Chief Credit Officer in January 2023 and an outside-hire Chief Wealth Officer in August 2023; they appear only in the 10-K officer table.)

8.5 A new structural threat, named by the company itself

The FY2025 10-K’s risk factors carry a sentence that deserves more attention than it has received: “In July 2025, President Trump signed into law the GENIUS Act, which establishes a regulatory framework for ‘payment stablecoins’ and their issuers. Consumers and businesses may view payment stablecoins as a substitute for traditional bank deposits, resulting in deposit withdrawals.”

A bank whose entire excess return rests on a 33% non-interest-bearing deposit share has named, in its own filings, a newly legislated substitute for non-interest-bearing transaction balances. The probability of near-term impact is low. The impact if it arrives is not, and it is the first genuinely new competitive threat to core transaction deposits in a generation.

8.6 Verdict

On balance these changes strengthen the near-term thesis and weaken the long-term one.

Strengthening: the removal of the entire Texas mid-tier competitive set; an expansion programme that has crossed from dilutive to accretive; a step-change in capital return; a margin at a fifteen-year high with contractual repricing still to come; energy and commercial loan demand re-accelerating; and Chase standing down from further Texas branch building.

Weakening: two $200 billion-plus competitors arriving with explicit share targets in the two markets that are 61% of Texas and where Frost holds 1.6% and 2.3%; a demographic engine decelerating by 30% year over year with domestic migration down 70% from its peak; a housing market down eleven consecutive months into which Frost has deliberately grown its consumer book 20%; the worst multifamily market in the country holding the 2022-vintage credits management concedes are still to work through; a data-centre moratorium that arrived twelve days ago and has already cut the state’s electricity-demand forecast by more than half; and a legislated substitute for the transaction deposit that is the whole basis of the franchise.

The single sharpest tension in the file: Cullen/Frost’s consumer loan book grew 20% year over year — driven by mortgage (+$533 million) and second-lien home equity (+$198 million) — into a Texas housing market that has now fallen for eleven consecutive months, with Austin down 24.5% from its peak. Management is candid that it has been “very price competitive” to build that book and that “our pricing will tighten up.” That is a small book being built for customer acquisition rather than for yield, and the strategy is defensible. It is still growth on price into the one asset class where Texas fundamentals are deteriorating.

9. Risk Analysis

9.1 Risk matrix

9.2 The risks that actually matter

Most of the nineteen rows above are ordinary banking risks disclosed and managed competently. Three deserve to be separated out because they are the ones that would change an owner’s outcome.

Risk 1 is the real one, and it is not a credit risk. Cullen/Frost’s most likely path to disappointing a shareholder is not a loan loss; it is another leg higher in long yields that re-widens the securities mark and freezes tangible book value per share for another multi-year stretch, exactly as happened from 2021 to 2025. The uncomfortable feature of the current setup is that management’s own base case makes this more likely, not less: the guidance assumes a rising short rate, and the company has added $2.1 billion of amortized cost to the available-for-sale book in six months. The reinvestment yields (5.25–5.40%) are attractive and the decision is defensible on an earnings basis. It is still the same bet, made larger.

Risk 7 compounds Risk 1. A stock at the 96th percentile of its own price-to-book has no cushion for a book-value stall. The 2022–23 experience is the direct precedent: earnings held up, book value fell, and the multiple compressed anyway — a 47% drawdown that took 38 months to recover.

Risks 12–14 are the ones that are not in the price. The Texas demographic engine decelerated 30% in a year and domestic in-migration is down about 70% from 2022; the statewide housing market has fallen for eleven consecutive months; and a data-centre moratorium landed twelve days before this report date that has already halved the state’s electricity-demand growth forecast. Cullen/Frost’s own investor materials project 2026–2031 Texas population growth of 6.5% against 3.5% nationally — using S&P projections published before the 2025 migration collapse was reported. The bank is not mispricing this; the market may be.

Risk 4 is the credit tail. Frost’s commercial real-estate underwriting appears genuinely better than the market’s — the evidence is that it is losing deals on structure and that its office book has the best coverage in its own portfolio. But a bank with 29.1% of loans in non-owner-occupied and construction real estate, in a market where competitors are lending against the same collateral on weaker terms, is exposed to the collateral values those competitors’ loans will eventually set.

On catastrophic loss: the probability of permanent capital impairment at Cullen/Frost is low. Common Equity Tier 1 of 13.95% (11.4% including the entire AOCI mark), a 53% loan-to-deposit ratio, a securities portfolio that is agency and Texas-Permanent-School-Fund paper, and $6.47 billion of cash make a solvency event very hard to construct short of a Texas-wide depression. The realistic bad outcome for a buyer at $170.08 is not loss of capital; it is a multi-year period of no return, which is exactly what a buyer at the 2022 high experienced.

10. Valuation Discussion — Embedded Expectations

No price target and no recommendation appears in this section. The purpose is to establish what the current price requires the business to deliver.

10.1 Where the stock actually trades

The percentile figures are the most important numbers in this section. On price-to-book — the multiple that matters most for a spread lender — Cullen/Frost has been more expensive than it is today in roughly 4% of its own multi-year history. On price-to-earnings it has been more expensive about 14% of the time. Price-to-sales, at the 66.6th percentile, is a near-meaningless metric for a bank and drags the composite down; the honest read is the 96th and 86th percentiles, not the 83rd.

For a sense of scale on the range: at the March 2023 trough the shares traded near 1.48× book, and at the 2020 COVID low near 0.72×. The current 2.36× is at the opposite end of that distribution.

10.2 Cross-sectional comparison

Two directly comparable anchors from recent work on the regional-bank cohort, both as of early August 2026:

*Columbia’s efficiency ratio is not directly comparable and is quoted on an operating basis in that report; the balance-sheet direction and multiple are the relevant comparisons.

Frost trades at a 62–73% premium on tangible book to two banks with essentially the same non-interest-bearing deposit mix and a comparable (or better) efficiency ratio, for a return-on-tangible-common-equity premium of roughly 12% on the reported figures — and no premium at all on the AOCI-adjusted figures, where Frost’s ~14.9% sits below Zions’ 16.6%.

The defensible justification for the premium is growth, not returns. Zions is roughly flat; Columbia is deliberately contracting its balance sheet and buying back stock; Frost is compounding households at 5.7%, loans at 7–8% and expansion deposits at 20%. In a dividend-discount framework, a two-point difference in sustainable growth is worth a great deal more than a two-point difference in return.

10.3 What the price requires — the embedded-expectations calculation

Using the standard justified price-to-tangible-book relationship for a bank, P/TBV = (ROTCE − g) / (COE − g):

Case A — solve for required return, on reported tangible book. At P/TBV = 2.77×, a cost of equity of 9.5% and a sustainable growth rate of 5.0%: ROTCE_required = 2.77 × (0.095 − 0.050) + 0.050 = 17.5% Delivered reported ROTCE is ~18.6%. The price is roughly fair on reported figures with a 5% growth assumption.

Case B — the same calculation on AOCI-adjusted tangible book. Adjusted tangible book is $75.20/share, so P/adjusted-TBV = 2.26×: ROTCE_required = 2.26 × (0.095 − 0.050) + 0.050 = 15.2% Delivered adjusted ROTCE is ~14.9%. On the adjusted basis the price is also roughly fair — but with essentially no margin.

Case C — the demanding assumption. At a cost of equity of 10.0% and a growth rate of 4.0% (a more conservative pair for a single-state bank at a cycle high in margin): ROTCE_required = 2.77 × 0.06 + 0.04 = 20.6% That is above anything Frost has delivered in the last decade on either basis. Under a conservative cost-of-equity and growth pair, the stock is expensive.

Case D — solve for required growth. Holding reported ROTCE at 18.6% and cost of equity at 9.5%, the growth rate implied by 2.77× tangible book is: g = (COE × P/TBV − ROTCE) / (P/TBV − 1) = (0.095 × 2.77 − 0.186) / 1.77 = 4.36% Frost is guiding to 7–8% average loan growth. Consistency requires only 4.4% sustainable growth — which the franchise clears comfortably today. The question is durability, not level.

The four cases together give the honest answer: at $170.08 the market is underwriting Cullen/Frost to sustain something close to its current returns and something better than mid-single-digit growth, indefinitely. That is not an absurd expectation for this franchise. It is also not a bargain, and it leaves no compensation for being wrong.

10.4 Scenario analysis

Assumptions common to all three: shares outstanding decline ~1.5%/year on the buyback; the dividend continues to grow with earnings; tangible book value per share is the anchor.

The distribution is wider on the downside than on the upside relative to the current price, principally because the multiple is already at the 96th percentile of its own history and because tangible book value is a levered function of the long end of the curve.

10.5 What the market is pricing correctly and incorrectly

Correctly priced:

• The deposit franchise. A 1.11% June spot cost of total deposits against a 3.50–3.75% fed funds range is not a fluke, is corroborated by seventeen consecutive years of J.D. Power leadership in Texas, and is worth roughly $321 million a year pre-tax on the company’s own peer comparison.
• The organic growth engine. It is disclosed quarterly with enough granularity to verify, it is accelerating, and it is now accretive to earnings per share.
• The credit book. A single identified Austin multifamily credit and a shrinking criticized book do not constitute a credit problem, and the market has not treated them as one.

Possibly mispriced, in the market’s favour:

• The persistence of reported ROTCE. As the accumulated-other-comprehensive-income hole closes, the denominator of return on tangible common equity grows faster than the numerator. Reported returns should fall toward the mid-teens even with flawless execution. A valuation that extrapolates 18–19% is extrapolating an accounting artefact.
• The rate assumption. The full-year guide contains a September hike. The stock’s last leg up coincided with the market pricing that hike in. Both the earnings and the multiple are exposed to the same variable.
• Duration risk in book value. A 100 basis-point rise in the ten-year would put roughly $15 per share of additional mark into equity. The market is treating the securities book as a source of margin expansion (which it is) without pricing it as a source of book-value volatility (which it also is).
• The rate-state dependence of the deposit advantage. The 76 basis-point deposit-cost gap was zero in 2022 and averaged nineteen basis points from 2014 to 2017. Capitalizing $321 million of annual funding advantage into perpetuity capitalizes a rate regime, not a moat. A return to a low-rate world shrinks the prize by roughly 90% without anything changing at the bank.
• The terminal-growth assumption versus the geography. The Case D calculation above requires only 4.4% sustainable growth — but 61% of the Texas deposit market sits in Dallas-Fort Worth and Houston, where Frost holds 1.6% and 2.3% raw share and where twenty-eight and seventeen new branches respectively bought 29 and 33 basis points. Meanwhile Frost’s adjusted statewide share has not grown since 2022 and deposits per office have fallen from $267 million to $198 million. The growth is coming, at the margin, at a declining return on incremental branch capital.

Possibly mispriced, against the market:

• The wealth and insurance businesses. A 19.9% pre-tax margin on a $56.6 million quarterly revenue base, with $26.7 billion of managed assets and an explicit management admission of below-average insurance penetration into its own commercial customer base, is a real option that carries no visible value in the multiple.
• The duration of the market-disruption tailwind. If Texas consolidation continues, Frost’s 78–80% win rate against acquired and acquiring banks is a multi-year share-gain engine, not a one-year event.

11. Variant Perception

11.1 What consensus believes

The consensus view, readable in both the sell-side questions on the second-quarter call and the price action, is that Cullen/Frost is a best-in-class Texas deposit franchise entering a multi-year period of positive operating leverage as its branch-expansion investment stops costing money and starts making it. Analysts asked about margin trajectory into 2027, about the sustainability of the guidance raises, and — tellingly — about whether the positive-operating-leverage moment they had “been waiting for” would persist. Management said yes without giving guidance. The stock has re-rated 42% from the November 2025 low on that thesis.

11.2 The strongest bull case

1. The deposit franchise is a structural, not cyclical, advantage, evidenced by a 1.11% spot cost of deposits, 33.7% non-interest-bearing balances, and customer survey data showing rate ranks near the bottom of the reasons customers choose Frost.
2. The growth is organic, young and self-funding. 5.7% household growth accelerating; 82% of new consumer customers 45 or younger; a record $2.17 billion commercial pipeline; five consecutive quarters above 1,000 new commercial relationships.
3. The operating-leverage inflection is arithmetic. A fixed base of 204 branches with 8.7% of deposits still sitting in the newest cohort means each incremental deposit dollar carries less overhead. 140 basis points of positive operating leverage delivered; expense guidance cut while revenue guidance was raised.
4. The margin has contractual upside. $500 million of fixed-rate loans repricing 120–125bp higher; $1 billion of securities rolling from 3.60% into 5.25–5.40%; a $250 million sub-1% Treasury maturing in August.
5. Texas has the best large-state demographic and employment growth in the country, and Frost is by a wide margin the largest bank headquartered in it — and the only large Texas branch network still growing (+10 offices in 2025, against Prosperity −6 and Comerica −6). (Note the qualification developed in Section 3.1: the Texas deposit pool is still 13.3% below its 2022 peak and grew more slowly than the nation’s in 2025. The customer growth is real; the deposit tide is not.)
6. Capital return has stepped up. $163 million of buybacks in the first half against $2.7 million a year earlier; a 1.8% reduction in shares in six months; a dividend that has grown every year for decades.
7. The Texas mid-tier competitive set has been removed by consolidation — Independent, Comerica, Cadence, Veritex, Stellar, Guaranty, Vista, Texas Partners, American Bank and Keystone are all gone as independents. Frost reports winning 78–80% of contested opportunities against acquired and acquiring banks, and the customer-facing conversions are mostly still ahead (Fifth Third/Comerica September 2026, Prosperity/Stellar March 2027), so the disruption dividend should be larger in 2027 than in 2026.
8. The regulatory band is a real cost advantage. Below $100 billion, Frost escapes Basel III Endgame, the Liquidity Coverage Ratio and the Net Stable Funding Ratio entirely, and the one $50 billion resolution-planning rule that reached it was suspended in December 2025. At ~4% asset growth it has roughly a decade before that changes — against $200 billion-plus entrants attacking the same markets.

11.3 The strongest bear case

1. You are paying the 96th percentile of the company’s own price-to-book history for a bank whose reported returns are flattered by a depressed equity denominator and whose adjusted returns are below a peer trading at 1.60× tangible book.
2. The expansion is not producing statewide share. Adjusted Texas deposit share fell from 4.16% (2022) to 3.93% (2025) while the office count rose from 173 to 213; deposits per office fell from $267 million to $198 million, below the 2020 level. Non-interest-bearing deposit dollars are flat over five years (+0.5% CAGR) against a 51% increase in branches, and the non-interest-bearing share has fallen from 43.3% to 32.9%. In Houston — the flagship expansion market — Huntington’s pro-forma position has passed Frost into sixth place by acquisition, after Frost spent seven and a half years and seventeen branches building there.
3. Tangible book value per share compounded at 1.2% a year from 2021 to 2025 despite reported returns on equity of 15–19%. The mechanism that did that — a 53% loan-to-deposit ratio and a $19.1 billion securities book at 4.9 years duration — is still fully in place, and management is adding to it ($2.1 billion of amortized cost in six months).
4. The guidance embeds a rate hike that may not arrive, and the last 27% of the stock’s move coincided with the market pricing that hike in. The exposure is doubled: earnings and multiple to the same variable.
5. The efficiency ratio is 62.8% and has been in the low 60s for years. If the operating-leverage inflection stalls — because expansion branch openings continue, or because management uses the flexibility to hire displaced bankers, which the chief financial officer explicitly flagged as an upside risk to expenses — the entire re-rating case weakens.
6. Commercial real estate is 47.8% of loans in a market management itself calls “a race to the bottom on some of these structures.” Frost is declining that business, which is correct — but it means the marginal Texas CRE loan is being underwritten by someone with worse standards, and Frost’s existing book sits alongside it in the same collateral market.
7. The wealth business is not growing. Trust assets fell from $51.4 billion to $51.0 billion year over year in a rising market. A 19.9% segment margin is poor. This is 10% of pre-tax income going nowhere.
8. Single-state concentration is a genuine risk, not a talking point. Every municipal bond in the securities portfolio is a Texas issuer. Every loan is essentially a Texas loan. A Texas-specific shock — energy, tariffs on Mexico trade, a Permian downturn, a property-tax or insurance shock to Texas real estate — hits assets, deposits and securities simultaneously.

11.4 The three to five assumptions that actually matter

11.5 The positioning read — where consensus may be offsides

The factor evidence points somewhere specific and slightly counter-intuitive. Cullen/Frost has no momentum loading. Its Momentum beta is −0.024 in the base model and −0.042 in the all-factors model, despite being up 38.7% over twelve months at an all-time high — because the twelve-month formation window still spans the April 2025 tariff drawdown. Its actual style exposures are DividendYield (+0.89 to +1.09, its single largest loading, larger than Market), Value (+0.30), Regional Banks (+0.64), SmallSize (+0.45) and CreditRisk (+0.34), with Quality negative (−0.08 to −0.21).

Two implications follow.

First, the re-rating is idiosyncratic, not sectoral, and that is genuinely bullish for the fundamental case. Over the 63 trading days in which Frost gained 27.2%, the Regional Banks industry factor returned −0.5% and the “Regional Bank Giants” basket −1.3% (−6.3% over 126 days, a z-score of −2.15). Idiosyncratic volatility is 11.96% annualized against an R² of 0.74 — roughly a third of the variance is stock-specific. The market is repricing this company on its own results, which is the opposite of 2023, when Frost fell 47% for things other banks did.

Second, the marginal holder owns this for income and value characteristics, not for the franchise story — and that is where consensus is most likely offsides. DividendYield has a 252-day z-score of +1.77 and Value +1.61; both of Frost’s largest style exposures are working hard in the current regime, while Momentum, which Frost does not own, is flat. A shareholder who believes they are underwriting a Texas growth compounder is, in factor terms, holding a low-beta dividend-and-value security. When the dividend/value regime turns, the position will lose money for reasons that have nothing to do with household growth in Fort Worth. That is the variant perception: the risk in this position is a style repricing, not a credit event, and the fundamental bull case does not protect against it.

The corollary is the bear’s weakest point: because the trend-following crowd is not in this name — the momentum loading is zero — the usual crowded-trade unwind risk does not apply. This is not a falling knife, not an abandoned value name, and not a crowded momentum trade. It is a low-beta dividend/value bank whose own results have driven a genuine idiosyncratic re-rating into the top decile of its own valuation history.

12. Fact vs. Interpretation

13. Open Questions

1. Does the non-interest-bearing deposit share stabilize near 33%, or continue toward the high twenties? This is the single most important unresolved question on the franchise. The share has fallen from 43.3% (2021) to 32.9% (Q2 2026) and non-interest-bearing dollars have compounded at only 0.5% a year over five years while the branch count rose 51%. No public data settles it.

2. What is the actual return on the expansion programme? Cullen/Frost discloses earnings-per-share accretion but has never disclosed the capital invested in the programme or its return on capital. The ~11% figure in Section 7.5 is an outside calculation on stated assumptions. The disclosure gap is itself the finding, and it is the single most useful thing management could remedy.

3. Does the Austin phase hit its stated seventeen financial centres by year-end 2026? Only one of the five branches opened in the second quarter was in Austin. This is a clean, dated management-credibility test at the fourth-quarter 2026 print.

4. What happens to the $100 million of 4.50% subordinated notes maturing 17 March 2027? They no longer count as Tier 2 capital. No 8-K has addressed refinance versus repayment.

5. Is the interest-bearing deposit beta really going to fall from 46% to the low forties into a hiking cycle? Management guided to it. It is a bullish assumption resting on franchise strength, testable every quarter, and a 46% up-beta is not, on its face, a captivity signature.

6. How many more 2022-vintage construction credits are there, and what are they worth? The chief executive named “probably three more,” mostly in Austin, and said he is not concerned. Against the worst multifamily market in the country and a Dallas Fed finding that Texas loan modifications exceed the national rate, this deserves quarterly tracking of the criticized balance, not the non-performing one.

7. What is the succession plan? Phillip Green is 71; Paul Bracher is 69; Daniel Geddes is 51 and ran the Houston expansion. Five proxies contain four boilerplate committee-charter references and nothing else — no process narrative, no emergency plan, no timeline.

8. Why did Green’s perquisites rise from roughly $74,000 to $187,737, with security up 3.5× to $112,532 and personal charter aircraft of $55,797? No explanation is given.

9. Can the wealth business be fixed, and on what timetable? Trust assets were flat year over year in a rising market and the segment earns a 19.9% pre-tax margin against 30–40% for standalone peers. Management points to “the back half of '27 and 2028” and pre-announces possible adviser attrition.

10. What is the durable size of the market-disruption benefit? The only clean level disclosure is Q1 2026’s statement that 8% of new relationships came from disruption. Everything else is growth rates off undisclosed bases. Most conversions are still ahead (Fifth Third/Comerica September 2026, Prosperity/Stellar March 2027), so 2027 is the year to measure it.

11. What did the 8-K/A of 3 March 2026 correct? It replaced the 31 December 2025 investor presentation “in its entirety” to add technology-strategy detail, update peer data and “correct certain items.” The corrected exhibit was not reviewed for this report.

12. Consecutive-dividend-increase streak. The proxies say “27 consecutive years” (2022, 2023) and “30” (2024 and 2026). The figure appears in no 10-K or 10-Q. No streak number is asserted in this report.

14. What Must Be True

14.1 For the bull case

14.2 For the bear case

14.3 The single most informative datum

If one number had to carry the thesis, it is the non-interest-bearing share of average deposits. It is the metric the moat is made of, it is the metric that would deteriorate first if the franchise were fictional, and it has been deteriorating for four years — from 43.3% in 2021 to 32.9% today. The bull case requires that decline to be a rate-cycle artefact that stabilizes. The bear case requires it to be the slow erosion of the only thing that justifies a 2.77× tangible book multiple. Four more quarters of data will separate the two.

Appendix A — Diligence Questionnaire

Cullen/Frost Bankers, Inc. (NYSE: CFR) · Report date 2026-08-15 · Price reference $170.08 (2026-08-14 close)

Supplemental to the article. Answers are labelled FACT / INTERPRETATION / ASSUMPTION where the distinction matters. Where a standard question does not map to a bank’s economics, the correct sector analogue is given instead.

General

What thoughtful questions have other investors asked about this company?

The sell-side questions on the Q2 2026 call cluster tightly, and they are the right ones. Cantor Fitzgerald asked what the September rate hike contributes to the raised net-interest-income guide (answer: ~$2M/month for one quarter) and — more usefully — how expense guidance could fall while revenue guidance rose. Barclays asked about deposit beta into a rising-rate environment and whether the step-up in buyback was opportunistic or a new run-rate. Autonomous asked the best question of the call: given that Texas competition is intensifying from new entrants, is this a permanent regime change? (Green: “it will normalize after probably a couple of years.”) KBW asked whether the positive-operating-leverage moment “we have been waiting for” persists into 2027. D.A. Davidson asked whether the margin, at a fifteen-year high, has further to run. TD Cowen asked directly whether Frost would ever expand outside Texas.

The questions nobody asked are more revealing, and they are the ones this article has tried to answer: why tangible book value per share compounded at 1.2% a year from 2021 to 2025 while reported return on equity ran 15–19%; why the diluted share count is higher today than in 2016 after $558M of buybacks; why the incentive plan contained no return-on-capital metric for four years; and why, after adding 56 Texas offices since 2020, the company’s adjusted statewide deposit share is lower than it was in 2022.

Cyclicality and the nature of earnings

Are earnings at a cyclical high or low? INTERPRETATION: at a cyclical high, and the cycle in question is interest rates, not credit. The net interest margin of 3.75% is a fifteen-year high. The deposit-cost advantage over the peer index is 76 basis points — the widest in the twelve-year series the company publishes, and it was zero in 2022. Credit is benign (net charge-offs 17bp annualized against a 15–20bp full-year guide). The clinching evidence is the company’s own twenty-one-year return history: return on average common equity was 9.8–10.5% in every year from 2009 to 2016, with the same branches, the same culture and the same J.D. Power awards. A funding advantage is worth nothing when funding is free.

Driven by the external environment or internal actions? Both, and the split is roughly identifiable. Externally: the level of rates sets the size of the deposit-cost prize, and the securities book’s repricing (~$1B rolling from 3.60% into 5.25–5.40%) is contractual. Internally: household growth of 5.7%, the record $2.17B commercial pipeline, the 78–80% win rate against acquired and acquiring banks, and 140 basis points of positive operating leverage from a branch base that has finished its build phase. The internal contribution is real and is what distinguishes Frost from peers; the external contribution is what makes the current return level unrepeatable in a lower-rate world.

How stable are revenues? Very. Net interest income is 77% of revenue and is generated from a granular deposit base and a diversified loan book with no industry concentration above 5.0% (energy) or 4.7% (automobile dealerships). Fee income is 22–23% of revenue and its fastest-growing components — deposit service charges (+17.2%) and interchange (+16.5%) — are functions of customer count, not of markets. The company has produced no annual net loss through the Great Recession, and was the first bank in the United States to refuse TARP funds.

Outlook for products/services; how big will this market be? Texas banking. The Texas deposit pool was $1,498.2B at 30 June 2025 — still 13.3% below its 2022 peak, and it grew 3.4% in 2025 against 4.8% nationally. FACT, and it contradicts the standard narrative: Texas is not a growing deposit market; it is a growing customer market inside a flat deposit pool. Population growth decelerated 30.5% year over year and net domestic in-migration is down roughly 70% from 2022. The market is domestic and entirely intrastate; management has explicitly ruled out geographic expansion “in any foreseeable time.”

Business quality and competitive moat

Is the industry getting more or less competitive? More, and structurally so. Consolidation removed essentially the entire $1–20B Texas mid-tier in eighteen months — Independent, Comerica, Cadence, Veritex, Stellar, Guaranty, Vista, Texas Partners, American Bank, Keystone. That is favourable. But it imported Fifth Third (committed to $700M+ and 250 Texas financial centres by 2029, targeting top-five deposit share in Dallas, Austin and Houston) and Huntington (already fifth in both Dallas and Houston; pro forma it has passed Frost into sixth place in Houston). Frost’s own 10-K names the mechanism: deregulation “raised the asset thresholds at which more onerous requirements apply, which could cause certain large bank holding companies with less than $250 billion in total consolidated assets… to become more competitive or to pursue expansion more aggressively.” One favourable non-event: JPMorgan’s 2026 branch-build announcement does not name Texas.

How profitable is the business (ROIC, ROE)? For a bank, “ROIC” in the industrial sense is meaningless; the correct measures are return on assets, return on equity and return on tangible common equity. FY2025: ROA 1.24%, ROE 15.66%. Q2 2026: ROA 1.31%, ROE 15.41%. Trailing return on tangible common equity ~18.6% as reported. The honest adjustments are large: adding back the $851.7M accumulated-other-comprehensive-income mark to the equity base gives ~14.9% on adjusted tangible common equity and ~12.5% on AOCI-neutral FY2025 average common equity. Reported ROTCE will mechanically decline as the mark accretes back, even with flawless execution.

How profitable is the industry — how many competitors, what barriers to entry? 466 institutions and 6,218 branches serve Texas. Long-run US banking economics are poor: return on assets above 1% in only fourteen years since 1935, long-run average 0.75% and return on common equity 10% (Banking 101, Deutsche Bank, 2011). Barriers to entry are low at the state level and high at the local level. Anyone can open a Texas branch; nobody has dislodged Frost from 29% of San Antonio in six years. The industry favours locally dominant incumbents and offers nothing to sub-scale entrants — which is exactly why Frost earns a mid-teens return in San Antonio’s economics and why its Dallas branches took six years to stop diluting earnings per share.

Can the business be easily understood? Yes — unusually so. Two segments, one state, one bank charter, no trading book, no material derivatives complex, no securitization, $223M of holding-company debt against $53.9B of assets, and a balance sheet that is 40% government and agency securities. The FY2025 10-K’s Item 3 legal proceedings is boilerplate with no named litigation.

Can it be undermined by foreign low-cost labour? Not materially. The competitive threat is domestic: better-capitalized national banks, and — newly — payment stablecoins. Frost’s own risk factors state that under the July 2025 GENIUS Act “consumers and businesses may view payment stablecoins as a substitute for traditional bank deposits, resulting in deposit withdrawals.” For a bank whose entire excess return rests on a 33% non-interest-bearing share, that is the first genuinely new threat to core transaction balances in a generation.

Do brands matter? Here, demonstrably yes — and this is the moat. Frost has ranked highest in Texas in the J.D. Power U.S. Retail Banking Satisfaction Study for seventeen consecutive years, every year since J.D. Power began treating Texas as a separate region. The 2026 index score was 757 against a Texas average of 681 and a national average of 657, first in all eight dimensions, on a study of ~107,000 customers — in a year when J.D. Power’s own headline was that industry satisfaction is deteriorating and 20% of customers had moved money away from their primary bank. Seventeen consecutive years of Coalition Greenwich awards corroborate it on the commercial side. Management’s customer survey ranks the reasons customers choose Frost as convenient locations, reputation, and family recommendation — with competitive interest rates near the bottom.

What is the nature of competition? Price and structure on the asset side; rate on the liability side. Management is explicit that it competes on loan price but not on loan structure: “we are not going to sacrifice our credit for the sake of growth,” and it describes the market as “a race to the bottom on some of these structures,” mostly in commercial real estate. On deposits the Dallas Fed’s own February 2026 research concludes that Texas depositors are driven by rate rather than by bank-specific quality, and that Eleventh District CD rates run above the national average — which makes Frost’s funding advantage more impressive and also more contestable.

Customers’ switching costs? Low in the contractual sense and high in the behavioural sense — and the article is careful about the distinction. There are no contracts, no lock-ups and no meaningful exit fees on either side of the balance sheet. What exists is habit, search cost and trust: the Greenwald category is customer captivity, not switching costs. The measurable output is a blended deposit cost of 1.26% (FY2025) against 2.02% for the S&P Regional Bank Index peer group. The important qualification: against the best-funded peers rather than the index average the gap is 3–10 basis points (Commerce Bancshares 1.11%, Regions 1.18%, Frost 1.08% in Q2 2026) — and on interest-bearing cost Commerce’s 1.57% actually beats Frost’s 1.61%. Frost is at the top of a cluster, not alone at the top. Constructed from FDIC call-report data across all US banks at 31 March 2026, Frost ranks roughly #3 of 49 charters with $50 billion or more of deposits, and #1 among traditional deposit-funded banks at that scale (only Charles Schwab and USAA are cheaper, and both are sweep models). The achievement is scaling a low-cost franchise, not the absolute level — 11% of sub-$1 billion banks achieve a comparable cost, against only 4% of banks over $50 billion.

Financial condition and balance sheet

Assets not fully recognized on the balance sheet? Three. (1) $51.0 billion of trust assets ($26.7B managed, $24.3B custody) generating $177.1M of FY2025 fees but carried at nothing. (2) The branch and deposit franchise itself — 204 owned or leased financial centres carried at $1.37 billion of net property, plant and equipment, generating a funding advantage worth roughly $321M of pre-tax income a year at current rates. (3) Frost Insurance Agency, generating $65.5M of FY2025 commissions, essentially unrecognized. Conversely, $654,952 thousand of goodwill from decades-old acquisitions is carried and is the appropriate deduction in the tangible-book calculations used throughout.

Off-balance-sheet liabilities? Ordinary for a bank and disclosed: unfunded loan commitments and standby letters of credit (Note 5), plus operating leases. The genuinely important off-balance-sheet item is not a liability but an unrecognized loss: $135.2 million of unrealized loss on the held-to-maturity securities portfolio, which does not appear in equity. Defined-benefit plans were frozen in 2001, which materially reduces pension volatility.

How conservative is the accounting? Conservative in every place that matters except one, and that one is an election rather than an estimate. Only 15.8% of securities are held-to-maturity, so 82% of the portfolio is marked through accumulated other comprehensive income where investors can see it — the opposite of the banks that failed in 2023. The allowance for credit losses is 1.23% of loans, or 1.45% including the unfunded reserve. Five years of filings contain no restatement (no Item 4.02), no auditor change (Ernst & Young ratified annually), no cyber incident disclosure and no enforcement action. The exception: Cullen/Frost has elected the AOCI opt-out for regulatory capital, so reported Common Equity Tier 1 of 13.95% excludes the securities mark entirely. Including it, CET1 is ~11.40% today and was 8.23% at year-end 2022, against a reported 12.85%. That is legal, standard and disclosed — but the headline ratio has not been the economic ratio for four years.

How CapEx-hungry is the business? Far more than a typical bank, and deliberately. Purchases of premises and equipment totalled $1,002.3 million from 2019 through the first half of 2026 — roughly 21% of cumulative net income over a comparable span — taking net property, plant and equipment from $1,012M to $1,370M. That is the physical cost of the organic strategy. A bank pursuing the same growth through acquisition would show it as goodwill instead.

Capital allocation and management

How much cash does the business generate, and how is it used? “Free cash flow” is not a meaningful bank metric; the analogue is net income less the capital required to support balance-sheet growth. Over ten years Cullen/Frost earned $4,749.4 million, returned $2,488.9 million (52.4%) — $1,930.8M in dividends and $558.1M in buybacks — and retained $2,261 million. The philosophy is: fund organic growth first, pay a steadily rising dividend at a ~40% payout, and repurchase opportunistically with what is left.

Significant acquisitions recently? None — and this is the strongest single fact in the capital-allocation record. Goodwill has been frozen at $654,952 thousand since at least Q1 2020. The only acquisition line in five years of cash-flow statements is $703 thousand of insurance agency assets in FY2024. In the most-acquired banking market in the United States — Texas was the most-targeted state in 2025, with 21 deals — Cullen/Frost bought nothing. Its 10-K frames acquisitions as tangible-book dilutive; its chief executive says he has “zero interest.”

Buying back shares? Yes, and the record is a split verdict. The good: Cullen/Frost bought through the March 2023 regional-bank panic — its heaviest quarter of 2023 was the second, at roughly $96.7 a share, its cheapest print of the cycle. The bad: the 2023 repurchases totalled only $39 million against a $100 million authorization left 61% unused, while 2025 saw $150 million at $124.67 and the first half of 2026 saw $160 million at $137.61. Four times the dollars at 40% higher prices. And the decisive number: diluted shares were 62.97 million in FY2016 and 64.10 million in FY2025 — up 1.8% over a decade despite $558 million of buybacks. For nine of ten years the programme merely offset stock compensation. Only in the last twelve months has the count genuinely fallen, by 3.4%.

Issuing large amounts of new shares to insiders? No. No common equity has been raised. Share issuance is limited to option exercises ($9.3M / $22.6M / $11.9M in 2023–25) and stock compensation, which the buyback has historically absorbed rather than more than absorbed.

Compensation policy of directors and management? Absolute levels are restrained; the design is the problem. CEO Phillip Green’s FY2025 total was $7,625,974 with a 99:1 pay ratio — modest for a $10.5 billion bank. There are no employment agreements, no excise-tax gross-ups, a double-trigger change-in-control plan, a Rule 10D-1 clawback, and both hedging and pledging are prohibited. Say-on-pay support ran 96.6–99.1%.

But for FY2022 through FY2025 there was no return-on-capital metric anywhere in the programme. The annual bonus’s only quantitative metric is net income against a management-set budget — and after the single year the budget was missed (FY2023, 83% of a $711M target, for which the CEO was still paid 90%), the bar was reset to $531M and then $527M, roughly 26% lower. FY2025’s celebrated “22% above budget” result of $642M is still below the FY2023 budget the company failed to hit. The long-term awards certifying in 2025–27 pay on absolute growth in pre-provision net revenue less charge-offs — no denominator, no per-share element, no cost of capital, no benchmark — and the compensation committee stated in writing that it chose the metric “to align our performance award with our strategy of achieving organic growth by meaningful expansion in existing markets.” It paid the maximum 150% for the 2022–24 cycle, i.e. for the Federal Reserve’s tightening. The performance-contingent share of long-term pay was then cut from 50% to 25%, citing the same expansion, so only ~19% of the CEO’s FY2025 long-term grant is performance-conditioned. Relative return on average assets has been restored — but no such cycle certifies until 2028.

Motivations of management? Culture, on the evidence — because the pay plan does not supply the alignment. Aggregate insider ownership is 1.14% across 26 directors and officers; Green holds 0.24%; eight of fourteen director nominees hold under 10,000 shares and five hold zero or one. The founding family has exited entirely (Patrick B. Frost left management in January 2024 and the board in 2024). Against that, the single best evidence in the file: on 13 March 2023 — the first trading day after Silicon Valley Bank failed, and the moment Cullen/Frost’s AOCI-inclusive capital ratio was at its thinnest — the Chairman and CEO bought 9,500 shares for $1.01 million of personal cash, joined the same day by two directors, the Chief Risk Officer and the Chief Credit Officer. There are zero Rule 10b5-1 plans in five years of Form 4 filings, so every one of those was a discretionary act. The counterweight: Green sold 112,080 shares over the same five years against 9,500 bought, including 34,505 in November 2024 at ~$137.60.

Two governance flags, minor individually. Director Jeffrey M. Rummel, elected in January 2026 to the Audit and Risk committees, was Ernst & Young’s San Antonio Office Managing Partner until June 2025 — and Ernst & Young is Cullen/Frost’s auditor (a ~7-month cooling-off). And three of thirteen “independent” directors had commercial dealings with the bank in FY2025, the largest being $1,006,471 to Pape-Dawson engineering, whose CEO sits on the Compensation and Nominating/Governance committees, during a physical branch build-out.

Valuation and market data

Is the stock an ADR, MLP or K-1 issuer? No. Cullen/Frost Bankers, Inc. is a Texas-incorporated domestic C-corporation filing US GAAP financial statements in US dollars with a 31 December fiscal year end. Common stock trades on the NYSE under CFR; a 4.450% Series B non-cumulative perpetual preferred trades as depositary shares under CFR PrB. Holders receive Form 1099, not Schedule K-1.

Dividend policy? A steadily rising dividend at a conservative payout. Dividends per share rose every year from $1.66 (2008) to $3.95 (2025); the current annualized rate is $4.12 following raises in Q2 2025 (+5.3%) and Q2 2026 (+3.0%), with the raise cadence moving a quarter earlier in each of the last two years. The payout ratio has run 37–64% over ten years, 39.8% in FY2025. The current yield is 2.37%; adding the annualized first-half buyback gives a total shareholder yield of roughly 5.5%. Two observations: raise sizes are decelerating (+16.0% in 2022 to +3.0% in 2026) even as earnings hit records, and in the first half of 2026 buybacks ($160.0M) exceeded common dividends ($128.8M) for the first time — capital return is shifting toward repurchase. (The company cites a multi-decade consecutive-increase streak; the proxies say “27 consecutive years” in 2022 and “30” in both 2024 and 2026, so no streak number is asserted here.)

How profitable is the business? See above: FY2025 ROA 1.24% and ROE 15.66%; Q2 2026 ROA 1.31% and ROE 15.41%; efficiency ratio 62.8% on a GAAP basis and roughly 60.4% on the taxable-equivalent basis peers report — about two points worse than Commerce Bancshares (58.4%) and Regions (58.3%), both of which achieve nearly the same funding cost. Cullen/Frost discloses no efficiency ratio itself.

Is net income diverging from cash from operations? Not in a way that signals anything. For a bank, operating cash flow is dominated by balance-sheet flows rather than accruals and is not a quality-of-earnings signal in the industrial sense. First-half 2026 net cash from operating activities was $290.3 million against $343.0 million of net income; the difference is working-capital movement in accrued interest and other liabilities. The genuine quality-of-earnings issue at Cullen/Frost is not accrual-versus-cash — it is income-versus-comprehensive-income. In FY2022 the company reported $579.1 million of net income and total shareholders’ equity fell by $1,302 million. That gap, not the cash-flow statement, is where an owner’s returns actually went.

Risks and downside

What factors would cause the stock to decline? In rough order of likelihood: (1) a rate path that turns down rather than up — the guidance embeds a 25bp September 2026 hike, and the deposit-cost advantage that produces 42% of pre-tax income was zero in 2022 and produced only a ~10% return on equity through the 2009–2016 zero-rate decade; (2) multiple compression from the 95.8th percentile of the company’s own price-to-book history at an all-time high; (3) a second duration hit — a 100bp rise in the ten-year would add roughly $940 million of mark, about $15 per share of tangible book; (4) deposit-cost competition pushing the interest-bearing beta above the guided low-forties; (5) Texas credit — the 2022-vintage construction credits, in a state with the highest rental vacancy rate in the country; (6) a style repricing, since the stock’s largest factor loading is DividendYield rather than anything to do with Texas banking.

Risk of a catastrophic loss? Low. Common Equity Tier 1 is 13.95% reported and ~11.40% including the entire securities mark, against a 7.00% buffered requirement. The loan-to-deposit ratio is 53%. The securities book carries essentially no credit risk — Treasuries, agency mortgage-backed securities, and Texas municipals of which 68.9% are Permanent-School-Fund-guaranteed or pre-refunded. Held-to-maturity is only $3.39 billion, so the 2023 failure mode (an unmarked portfolio too large to sell) does not apply. Cash and equivalents are $6.47 billion. The company survived the 1980s Texas collapse as the only top-ten Texas bank not to need federal assistance or be merged, produced no loss through the Great Recession, and was the first US bank to refuse TARP.

Chance of a total loss? Very low — constructing one requires a Texas-wide depression severe enough to impair a book that is 53% loans and 40% government-guaranteed securities, at a bank with 460 basis points of Common Equity Tier 1 above requirement even on the punitive AOCI-inclusive measure. The realistic bad outcome for a buyer at $170.08 is not loss of capital; it is a multi-year period of no return — precisely what a buyer at the November 2022 high experienced, waiting 38 months to recover a 47% drawdown.

Recent news and events

Has the business environment changed recently? Materially, in four ways, three of which post-date most sell-side models. (1) The Texas mid-tier was consolidated out of existence between January 2025 and July 2026 and replaced by Fifth Third and Huntington, both with explicit Texas share targets; most customer-facing conversions are still ahead (Fifth Third/Comerica September 2026, Prosperity/Stellar March 2027), so the disruption effect — both the benefit and the competitive response — is a 2027 story. (2) The rate regime turned: the June 2026 FOMC scrapped 2026 cuts and put a hike back on the table, which is what the last leg of the stock’s move was paid for. (3) Texas demographics decelerated sharply — population growth −30.5% year over year, net domestic in-migration down ~70% from 2022, unemployment now equal to the national rate, home prices down eleven consecutive months, and the highest rental vacancy rate of any US state. (4) On 3 August 2026 — twelve days before this report date — Governor Abbott ordered a moratorium on new data-centre development, putting ~49.8 GW (roughly 20% of the US pipeline) at risk and prompting the EIA to cut its Texas electricity-demand growth forecast from 14% to 6%. The Dallas Fed’s July 2026 upgrade of its Texas job forecast leaned explicitly on the data-centre build and predates the order.

Significant acquisitions? None. See above.

Change in accounting policies? None material. Five years of filings contain no Item 4.02 restatement and no auditor change. The company took the CECL transition capital exclusion and the AOCI opt-out for regulatory capital, both standard elections, both disclosed.

Recent changes — new markets, facilities, management? Facilities: 204 financial centres at year-end 2025, up from 157 at year-end 2021; five opened in Q2 2026 (one Austin, one Dallas, one San Antonio, two Fort Worth) with five more planned for the balance of 2026, plus eleven now opened outside the announced Houston/Dallas/Austin programmes. The Austin phase’s stated target of seventeen centres “by 2026” is a clean, dated credibility test at the Q4 2026 print. Markets: none new — Texas only, and management has explicitly ruled out geographic expansion “in any foreseeable time.” Management: Daniel J. Geddes became CFO on 1 January 2025, succeeding Jerry Salinas; notably, Geddes was Market President and Houston Expansion Lead from October 2018 to August 2021, so the executive presenting the expansion’s economics built its first phase. Paul H. Bracher replaced Patrick B. Frost as President of Frost Bank in January 2024, ending family representation in management. Board turnover: Andrade (2024), Engates (April 2025), Rummel and Shields (January 2026) in; Alvarez and Patrick Frost (2024) and Avery (2026) out, with no reason disclosed for any of the three. Capital structure: the buyback authorization doubled to $300 million in January 2026, and $100 million of 4.50% subordinated notes mature on 17 March 2027 with no refinancing plan yet announced.

Appendix B — Source Appendix

15. Source Appendix

All sources accessed 2026-08-15 unless otherwise stated. Primary filings were downloaded in full from EDGAR (370 documents across the trailing sixty months) and read directly rather than relied upon second-hand.

A. Primary — SEC filings (Cullen/Frost Bankers, Inc., CIK 0000039263)

B. Primary — peer SEC/FDIC filings (deposit-cost benchmarking)

• Commerce Bancshares (CBSH) Form 8-K, 2026-07-16, Exhibit 99.1 — https://www.sec.gov/Archives/edgar/data/22356/000002235626000176/cbsh06302026ex991.htm
• Regions Financial (RF) Form 8-K, 2026-07-17, Exhibits 99.1 and 99.2 (financial supplement; discloses total deposit cost directly) — https://www.sec.gov/Archives/edgar/data/1281761/000128176126000050/rf-2026630xexhibitx992.htm
• Bank OZK Form 8-K filed with the FDIC (OZK deregistered from EDGAR in 2017 and files with the FDIC), 2026-07-21 — https://securitiesfilings.fdicconnect.fdic.gov/api/instflng/11969/attachment/1

C. Primary — management commentary

• Cullen/Frost Q2 2026 earnings call transcript, 2026-07-30 (Phillip D. Green, Chairman & CEO; Daniel J. Geddes, Group EVP & CFO; A.B. Mendez, IR). Retrieved via a commercial transcript service and cross-checked against the company’s own earnings release.
• Cullen/Frost Q1 2026 (2026-04-30), Q4 2025 (2026-01-29) and Q2 2025 earnings calls — disruption-sourced relationship percentages, Permian growth, M&A commentary.
• Data caveat: the ROIC.ai transcript service mislabels one Cullen/Frost record — a request for FY2025 Q4 returns a transcript stamped 2026-01-29 / quarter 4 whose body is unambiguously the Q2 2025 call. Transcript identity was verified from body text, not metadata, before any quotation.

D. Primary — government and regulatory data

• FDIC Summary of Deposits, branch-level Texas files 2020–2025, pulled directly from the FDIC BankFind API (6,218 Texas branch records for 2025) — https://api.fdic.gov/banks/sod. The 30 June 2026 survey does not yet exist (due ~2026-09-30), so all deposit-share data pre-dates the 2026-02-01/02 Huntington/Cadence and Fifth Third/Comerica closings.
• FDIC, 2025 Summary of Deposits release, 2025-09-19 — https://www.fdic.gov/news/press-releases/2025/fdic-releases-results-summary-deposits-annual-survey
• Federal Reserve Bank of Dallas, “Eleventh District banks maintain growth, profitability heading into 2026,” Southwest Economy, 2026-05-08 — https://www.dallasfed.org/research/swe/2026/swe2610
• Federal Reserve Bank of Dallas, “Keeping bank depositors still comes down to rates,” 2026-02-19 — https://www.dallasfed.org/banking/pubs/dfb/2026/2602
• Federal Reserve Bank of Dallas, “Texas multifamily housing yet to stabilize; downside risks remain,” Southwest Economy, March 2026 — https://www.dallasfed.org/research/swe/2026/swe2605
• Federal Reserve Bank of Dallas, Texas Employment Forecast, 2026-07-17 — https://www.dallasfed.org/news/releases/2026/nr260717forecast
• Dallas Fed Manufacturing, Services, Energy and Banking Conditions Surveys, July / Q2 2026
• US Census Bureau, Vintage 2025 population estimates, 2026 — https://www.census.gov/newsroom/press-releases/2026/population-growth-slows.html
• Texas Workforce Commission, 2026-07-17 — https://www.twc.texas.gov/news/texas-job-market-grows-more-177000-positions-over-year

E. Industry, market and news sources

• J.D. Power 2026 U.S. Retail Banking Satisfaction Study, via PR Newswire 2026-03-26 — https://www.prnewswire.com/news-releases/for-17th-consecutive-year-frost-bank-ranks-highest-in-the-jd-power-retail-banking-satisfaction-study-in-texas-302726413.html (and the 2025 release for the 16th-year figures). jdpower.com’s own release page returns HTTP 403; figures are internally consistent across the PR Newswire and Businesswire summaries at 757 / 681 / 657.
• Frost Bank press release, “Frost Bank Announces Major Expansion in Austin,” 2023-06-27
• Frost Bank, “Frost Bank Expands Overdraft Grace, Eliminates NSF Fees,” June 2022 — https://www.prnewswire.com/news-releases/frost-bank-expands-overdraft-grace-eliminates-nsf-fees-for-consumer-banking-customers-301566457.html
• frostbank.com newsroom fact sheet and Coalition Greenwich award pages
• Banking Dive, “Frost Bank stays the course amid Texas M&A spurt,” 2026-02-02 — https://www.bankingdive.com/news/frost-bank-texas-growth-mergers-acquisitions-fifth-third-huntington/811110/
• Banking Dive / S&P Global Market Intelligence, Texas bank M&A tallies, 2025-10-08 — https://www.bankingdive.com/news/fifth-third-comerica-bank-mergers-acquisitions-2025-texas-first-horizon/802371/
• Deal confirmations (all primary company releases): SouthState/Independent Bank Group; Huntington/Veritex and Huntington/Cadence (https://ir.huntington.com/news-presentations/press-releases/detail/948/ and /967/); Fifth Third/Comerica (https://www.businesswire.com/news/home/20260202908949/en/); Glacier/Guaranty; Prosperity/Stellar, Prosperity/Texas Partners, Prosperity/American Bank; National Bank Holdings/Vista; Third Coast/Keystone; FirstSun/HomeStreet termination
• JPMorgan Chase, “Chase to open more than 160 branches in over 30 states in 2026,” 2026-02-18 — https://media.chase.com/news/chase-to-open-more-than-160-branches-in-over-30-states-in-2026
• Texas Real Estate Research Center (Texas A&M), Texas Housing Insight, April 2026 data published 2026-07-01
• Texas Tribune, “Texas population growth slows,” 2026-01-27 — https://www.texastribune.org/2026/01/27/texas-population-2025-census/
• Texas Tribune, data-centre moratorium, 2026-08-03 — https://www.texastribune.org/2026/08/03/texas-data-center-project-audit-greg-abbott/
• Utility Dive, “Texas hits pause on data center interconnections” — https://www.utilitydive.com/news/texas-hits-pause-data-center-interconnections/827046/
• Morrison Foerster, “Texas Governor Orders Data Center Moratorium,” 2026-08-11 — https://www.mofo.com/resources/insights/260811-texas-governor-orders-data-center-moratorium-disrupting
• Baker Hughes rig counts via BOE Report; Apartments.com Q2 2026 vacancy and rent data; CBRE HQ-relocation tracking, 2026-04-06

F. Quantitative data services (third-party; reconciled to filings)

• ROIC.ai — income statement, balance sheet, profitability ratios, per-share data, valuation multiples, enterprise value, latest stock price (FY2015–FY2025 annual; price 2026-08-14). Two data caveats found and worked around: ROIC’s FY2025 diluted EPS of $10.01 conflicts with the 10-K’s $9.92 (the filing governs — ROIC appears to omit the two-class allocation), and ROIC reports tangible book value equal to book value for FY2023–FY2025, incorrectly dropping the $654.952M of goodwill that remains on the balance sheet. Tangible book was recomputed by hand throughout. ROIC’s get_company_news returned an empty array for CFR.
• AZI Trading — five-year daily price/OHLCV history and the valuation_index own-history percentile ranks (P/E 86.3rd, P/B 95.8th, P/S 66.6th, composite 82.9th; pulled 2026-08-14, latest price $170.08 matching the live close exactly).
• FactorsToday — stock loadings (four nested models), leaderboard, stock-info, specific volatility, related stocks and factor-returns endpoints, pulled 2026-08-15. All leaderboard returns and Sharpe ratios are annualized including short horizons; the m3 figure was de-annualized and cross-checked against the AZI price series before use.

G. Analytical frameworks

• Bruce Greenwald and Judd Kahn, Competition Demystified — barriers-to-entry taxonomy, market-share-stability and profitability tests.
• Edward Chancellor (ed.), Capital Returns (Marathon Asset Management) — supply-side capital-cycle analysis and the asset-growth anomaly.
• Deutsche Bank, Banking 101 — Large Cap Bank Primer, May 2011 — long-run industry return norms (banking-industry ROA above 1% in only fourteen years since 1935; long-run average ROA 0.75%, ROCE 10%). Dated sell-side research used as framework context only, not as current data.
• Peer figures for Zions Bancorporation and Columbia Banking System are drawn from those companies’ own Q2 2026 earnings releases and financial supplements.

H. Explicitly not verified

The following could not be confirmed from a primary source and are either excluded from this article or flagged in place:

1. The consecutive-dividend-increase streak. Proxy bios state “27 consecutive years” (2022, 2023) and “30” (2024 and 2026) — internally inconsistent. No streak figure is asserted in this report.
2. Deposit premiums paid in recent Texas bank M&A. Not sourced; the build-versus-buy comparison is therefore qualitative and management’s own $101M-vs-$175M-per-$1B figure is labelled as company arithmetic.
3. Capital invested in the expansion programme and its return. Never disclosed by the company. The ~11% implied return on allocated capital is an outside calculation on stated assumptions.
4. The PPNR-growth payout scale (threshold/target/maximum growth rates). Outcomes are certified; the curve is never disclosed.
5. Any CEO succession narrative. Four boilerplate committee-charter references across five proxies.
6. The corrected 31 December 2025 investor presentation (8-K/A, 2026-03-03), which replaced the original “in its entirety” to “correct certain items” — not reviewed.
7. Cullen/Frost’s Texas deposit rank on a raw, unadjusted FDIC basis. The “#4 in Texas” figure is on a branch-banking-adjusted basis excluding single-branch bookings (Schwab, USAA, Citibank); this is stated wherever the figure is used.
8. Reasons for the non-renomination of directors Avery (2026), Alvarez (2024) and Patrick B. Frost (2024).
9. Austin versus Dallas-Fort Worth Q2 2026 office vacancy figures are transposed between two sources (19.9% versus 20.6%); neither is relied upon here.
10. Claims checked and found incorrect, and therefore excluded: that Frost did not close branches during COVID (it closed all Texas lobbies effective 2020-03-20, keeping drive-throughs open), and that Frost holds the highest number of Coalition Greenwich awards (Pinnacle Financial Partners claims more).