East West Bancorp, Inc. (NASDAQ: EWBC) — A Great Bank at Peak Optics, Drifting Into Private Credit

Report date: 18 July 2026 · Initiating coverage · Sector: Financials — Regional Banks Price at analysis: $134.52 (17 July 2026) · Market capitalization: ~$18.7bn · Shares outstanding: 137.0m

This is general information and analysis, not investment advice. The main body of this article (Sections 1–15) contains no investment recommendation and no price target — it analyses embedded expectations and scenarios only. The single, deliberate exception is the Claude's Take block immediately below, which is clearly labelled as such.

⚡ Claude’s Take

This block is the author’s own subjective opinion. It is general information, not investment advice, and represents an independent view only. Everything from the Executive Summary onward carries no position and no price target.

Verdict: HOLD — the cohort’s best operator at cohort-average pricing, which sounds like a bargain and is actually just fair. Fully priced with no margin of safety. Accumulate on weakness toward ~$105–120 (roughly 1.7–1.95x tangible book), with genuine enthusiasm below ~$100. Not a short.

East West is the best-run bank in its cohort and it is not close. Its 36.2% efficiency ratio beats the next-best peer by 400bps and the cohort median by 2,400bps. Return on average tangible common equity has not fallen below 12.4% in six years — through COVID, through a 200bp rate round-trip, and through the March 2023 funding panic that briefly took this stock down 56% intraday. Tangible book per share has compounded 12.8% a year since 2021 while the share count shrank, with goodwill unchanged at $465.7m since 2020 — every dollar of that growth was earned, not bought. Capital is a fortress at 15.1% CET1, some ~$16 a share of it arguably surplus. The commercial real-estate book, which is what people fear about a California bank, carries a 49% blended loan-to-value on a $3m average loan size. This is not a fragile institution, and the bear case is not that it breaks.

Let me be precise about the valuation, because the lazy version of this argument is wrong. At 2.19x tangible book on FY2025 figures — 2.16x on the newer Q1-2026 balance sheet — EWBC sits at the 97th percentile of its own ten-year price-to-book range, but it is not expensive against peers. Regressing cohort price-to-tangible-book on ROTCE gives a fitted value of ~2.27x for EWBC against 2.19x actual: it trades marginally below the line its peers are priced on. Regions Financial sat at the 99.2nd percentile of its own range a month ago. The whole regional-bank complex is at the top of its band. So the multiple risk here is sector-beta risk, not an EWBC-specific overvaluation — which matters, because it is not a risk management can fix by executing well.

What I do object to is paying a full price for an earnings profile in which four separate quality inputs are peaking at once. First, the celebrated 35.7% efficiency ratio is roughly 40.8% restated on the pre-2024 presentation: adopting ASU 2023-02 moved ~$90m a year of tax-credit amortisation out of operating expense and into the tax line, and the company quietly stopped publishing the adjusted efficiency ratio that would have revealed it. Second, every basis point of margin expansion since 2024 came from the liability side — earning-asset yields actually fell 52bps — and the CD repricing tailwind is roughly one year from exhaustion as the book converges on the 3.60% special rate. Third, the “record” fee quarter was 81% wealth management, which the filing attributes to fixed annuity and fixed-rate corporate bond sales: not diversification against rate risk, but a second lever on the same bet. Fourth, and most important, the growth engine has quietly changed — non-depository financial institution lending supplied half of all 2025 loan growth while core C&I contracted, and inside that book the safe, over-collateralised capital-call piece grew 6% over five quarters while the opaque residual grew 57%. Management’s “virtually no charge-offs in a decade” was earned by the capital-call book and is now used to describe a portfolio that is 70% something else. Meanwhile classified loans rose 15% in a single quarter and management raised full-year charge-off guidance to 1.7–2.8x the current run rate. The leading indicator turned; the lagging ones people quote have not.

The arithmetic that settles it: at 2.19x tangible book with a 10% cost of equity and 5% growth, the market requires ~15.95% sustainable ROTCE. EWBC earns ~17%. That is one point of cushion — about 6% of the multiple — and the entire answer sits in an unobservable cost-of-equity assumption. At 9% COE the stock is cheap; at 11% — defensible for a bank with this China-corridor concentration and a 1.178 industry beta — it requires 19.3% and does not clear. I am not willing to underwrite the low end of that range while credit is turning and insiders are selling.

Framing: a quality compounder inside a crowded sector trade, and the second fact dominates the first. With an 81.6% factor R² and negative loadings on Quality and Low-Volatility, you are not buying idiosyncratic excellence — you are buying levered regional-bank beta with a much better operator attached. The insider record is the tiebreaker: every one of the 18 open-market purchases in five years happened in a nine-week window in spring 2023, and nobody has bought a share since, while the CEO, Vice Chairman, Chief Risk Officer and three directors sold into the 2026 high outside any 10b5-1 plan. Conviction: medium. Flips bullish: classified-asset migration reversing with the NDFI mix stabilising, plus real deployment of the excess capital (a large buyback at a sane price, or a well-priced East Coast deal). Flips bearish: charge-offs breaching the 25bp ceiling, a named credit event in the private-credit book, or Dominic Ng departing with no successor named. Note the timing: Q2 2026 reports on 21 July, three days after this report.

Tag: the best house on the street, priced as if the street never floods.

📈 Stock Price Action — Five-Year Event Map

East West has round-tripped from rate-cycle darling to funding-panic casualty and back to an all-time high. Over the trailing five years the stock ran from roughly ~$63 (July 2021) to a February 2022 peak near ~$78, collapsed to an intraday low of $30.90 on 13 March 2023 during the Silicon Valley Bank week, and has since compounded to $134.52 (17 July 2026) — +335% off that low, within ~1.3% of its all-time high of $136.24 set on 16 July 2026. The 52-week range is $90.90–$136.24; the 200-day EMA sits at $115.33, leaving the stock ~17% above its own long-term trend, above the 21-, 50- and 200-day EMAs in strict ascending order.

Cycle narrative. (1) The 2021–22 advance was a rates trade — EWBC is asset-sensitive and the market paid up ahead of liftoff. (2) It gave the gain back as the curve inverted and the group re-rated on recession odds. (3) March 2023 is the defining event in this stock’s modern history: East West fell 56% intraday from its February close, not on anything it did, but because it is a California commercial bank whose uninsured deposits were then, as now, roughly 55% of the book — in a funding panic that arithmetic is the whole story regardless of a 1.7% ROA. (4) The recovery began when the deposits simply did not leave, validating the franchise only after the market had priced its destruction. (5) The 2024 advance rode the easing cycle and post-election expectations of lighter regulation and revived M&A. (6) November 2024–April 2025 surrendered half of it to a hawkish Fed repricing and then the April 2025 tariff shock, where EWBC took a double hit as both a regional bank and the US–China corridor bank. (7–8) The last fifteen months are a clean, uninterrupted re-rating on genuinely strong results: record loans ($58.1bn), deposits ($68.9bn) and fee income, Q1-2026 net income +23% year-over-year, an NII guidance raise the CFO attributed entirely to the Fed no longer cutting, and the March 2026 capital-framework re-proposal that management estimates would free ~$7bn of risk-weighted assets. Seven of the eight news items on the tape in the last month are sell-side price-target raises — every one published after the move it justifies.

1. Executive Summary

East West Bancorp is a $81.5bn-asset California-chartered commercial bank, the largest US bank oriented to the Chinese-American community, and the only US regional holding a locally-incorporated Chinese commercial banking licence. It earns a return on average tangible common equity of ~17.0%, an efficiency ratio it reports as 35.7%, and a return on assets of 1.70% — against a US banking industry that has averaged roughly 0.75% ROA since 1935 and exceeded 1% in only fourteen years over that span. On any long-horizon base rate, this is an outlier franchise, and the outperformance is real rather than an artifact of leverage: CET1 is 15.1%, tangible common equity 10.3%, and the loan-to-deposit ratio 84%.

The moat is genuine but narrower than the popular story. It is not a deposit-franchise advantage: EWBC’s FY2025 cost of deposits was 2.46% and 38% of the book is time deposits — it funds itself more expensively than several peers, and average noninterest-bearing deposits have slipped from 29% to 24% of the base in two years. The advantage is an expense structure — occupancy at 0.08% of assets, ~$613m of core deposits per US branch, $875k of revenue per employee — layered on economies of scale inside a demographically bounded market, reinforced by a China banking licence no US entrant can obtain today. The decisive control experiment is Cathay General (CATY): same city, same era, same customer base, one-third the size, 11.25% ROE versus 17.0%. Cultural affinity is table stakes within the niche; scale within it is the moat.

Four findings materially qualify the quality narrative, and together they form the core of this report. First, the efficiency ratio is flattered by accounting geography: adopting ASU 2023-02 on 1 January 2024 shifted ~$90m a year of tax-credit amortisation from operating expense into the tax line, and restating FY2025 on the prior presentation yields ~40.8%, not 35.69% — the company discontinued its adjusted-efficiency disclosure in the same year. Second, margin expansion is entirely liability-driven; earning-asset yields fell 52bps since FY2024 while interest-bearing deposit costs fell 99bps, and with the CD book at 3.36% against a 3.60% special rate, that tailwind is roughly one year from reversing. Third, the “record” fee quarter was 81% wealth management, which the filing attributes to fixed annuity and fixed-rate corporate bond sales — a rate-amplified line, not a rate hedge. Fourth, and most consequential, non-depository financial institution lending provided half of all FY2025 loan growth while core C&I contracted ~$350m; inside that book the capital-call piece grew 6.3% over five quarters while the residual private-credit and sponsor exposure grew 57.3%, to a total of $8.3bn — 14% of loans and ~93% of common equity.

Credit is pristine on lagging measures (net charge-offs 9bps, non-performing assets 26bps) and reserved at 1.51% of loans, or 6–10x management’s own guided losses. But the leading measure moved: classified loans rose 15% in Q1-2026 alone, to the highest level in the disclosed series, and management raised FY2026 charge-off guidance to 15–25bps — 1.7–2.8x the current run rate. Simultaneously, management guides FY2026 expenses (+7–9%) above revenue (~+7%), i.e. zero-to-negative operating leverage.

At $134.52 the stock trades at 2.16x tangible book (2.24x marking held-to-maturity securities), 14.1x trailing earnings, and the 97th percentile of its own ten-year price-to-book range. Solving the standard bank identity, that multiple requires ~17.0% sustainable ROTCE at a 10% cost of equity and 4% growth — precisely what EWBC earns today, leaving no margin for the four peaking inputs above. Sell-side price targets ($131–154) now bracket the spot price, and the stock sits 1.3% below an all-time high after a +335% run off the March 2023 low. Capital is the offsetting asset: ~$3.5–4bn of excess CET1 (roughly 20% of market capitalisation), potentially growing under the March 2026 capital re-proposal — though six years of non-deployment have coincided with ROE decaying from 24.5% to 16.0%.

No recommendation or price target is expressed in this section or in the sections that follow.

2. Business Overview

East West Bancorp is the holding company for East West Bank, a California state-chartered commercial bank founded in 1973 as a savings institution serving Chinese immigrants in Los Angeles. At 31 March 2026 it held $81.5bn of assets, $58.13bn of loans and $68.92bn of deposits, funded by $9.0bn of equity. It employs 3,350 full-time equivalents and operates just over 110 locations, of which 96 are US branches across California, Texas, New York, Washington, Georgia, Massachusetts and Nevada, plus branches in China and Hong Kong and representative offices in China and Singapore. Roughly 300 employees sit in Asia.

The structurally distinguishing asset is East West Bank (China) Limited (“EWCN”), a locally-incorporated Chinese commercial banking subsidiary. The FY2025 10-K states plainly that this licence “makes it unique among U.S.-based regional banks,” permitting the bank to operate branches, make loans and accept deposits inside mainland China. No other US regional holds one. The company is accordingly supervised not only by US regulators but by the People’s Bank of China, China’s National Financial Regulatory Administration, the Hong Kong Monetary Authority and the Monetary Authority of Singapore — a regulatory surface no domestic peer carries.

2.1 Segments

The split is misleading if read as “half consumer.” Consumer & Business Banking is overwhelmingly a deposit-gathering engine — $33.4bn of average deposits against only $20.3bn of average loans. Commercial Banking is a loan-deploying engine at 125% loan-to-deposit. The branch network’s economic function is to fund the commercial book, not to earn its own spread. That framing matters for the moat discussion below.

A quality-of-earnings observation that belongs here rather than buried in the financials: both operating segments saw net income decline in FY2025 — Consumer & Business Banking −11%, Commercial Banking −8%. All of the consolidated +14% EPS growth came from Treasury & Other, where net interest income swung +$444m on securities-portfolio repositioning plus $32m of purchased-credit-impaired discount accretion. FY2025’s headline “record year” is, at the segment level, a securities and rate outcome rather than a franchise outcome.

2.2 Loan mix

Loans of $56.90bn at year-end 2025 divide into C&I $18.65bn (33%), total commercial real estate $21.26bn (37%), single-family residential $15.00bn within total consumer of $16.97bn (30%). Commercial credit is therefore ~70% of the book. Within C&I the largest concentrations are real-estate investment and management (13%), capital call lending (12%) and media & entertainment (12%). Loans to non-depository financial institutions total $7.6bn, up from $6.0bn — the single most important number in this report, developed below. C&I utilisation is 67% on $27.7bn of commitments, and 58% of loans held for investment are variable-rate.

2.3 Deposits and the cross-border question

Deposits of $68.92bn at Q1-2026 comprise noninterest-bearing demand $17.48bn (25%), interest-bearing checking $8.07bn (12%), money market $16.23bn (24%), savings $1.73bn (2%) and time deposits $25.41bn (37%). Uninsured deposits are $37.19bn, or 55.4% of the total; on management’s adjusted basis — domestic uninsured excluding $4.46bn collateralised and $0.13bn affiliate balances — the figure is $28.84bn, or 45% of domestic deposits.

The most commonly misunderstood aspect of this franchise is how little of it is directly cross-border. Foreign-exchange income was $59m in FY2025 — 2.0% of total revenue. International branch deposits are $3.88bn, or 5.8% of the total. Total noninterest income is $379.2m, 12.9% of revenue. The US–China corridor is best understood as a customer-acquisition and relationship-anchoring device rather than a profit centre: the China capability is the reason the US commercial relationship exists, and the US relationship is where the spread is earned. That is a defensible inference, and it is an inference — the company does not disclose a clean China-linked revenue figure.

Verdict: a commercially-oriented, deposit-funded, relationship-lending bank whose distinguishing feature is a demographically-defined customer base and a unique Chinese banking licence — not, on the numbers, a trade-finance house.

3. Industry Dynamics

3.1 Structure and the consolidation wave

Regional-bank consolidation reached a seven-year high in the first half of 2026 — 25 deals year-to-date totalling $15.11bn, including Fifth Third/Comerica, PNC/FirstBank, Santander/Webster and Huntington/Cadence. The stated rationale is uniform: scale to amortise compliance and technology cost, plus deposit acquisition.

This matters less than it appears. In capital-cycle terms it is capacity consolidation without pricing-power creation — the supply of charters is falling while the supply of credit is rising. Consolidation tightens competition for cost absorption, not for lending. EWBC has sat the entire wave out: goodwill has been unchanged at $465.7m since 2020.

Against the long sweep of banking history, the industry’s economics are poor. Sell-side primer work documents that US banking-industry return on assets has averaged roughly 0.75% since 1935, exceeding 1% in only fourteen years (all between 1993 and 2006), with industry return on common equity averaging ~10%. EWBC earns a 1.70% ROA. That is not “a good bank” — it is a ~2.3x-industry-average outlier, and the entire analytical burden of this report is to decide whether that reflects a structural advantage or a mean-reversion candidate being capitalised at a peak multiple.

3.2 The $100bn threshold — a datable forward cost

At $80.4bn of assets growing 6–7% a year, and with the test applied on a four-quarter average so the trigger precedes the headline, EWBC crosses $100bn in roughly 2028–2029 — with pre-build spending pulled forward to 2027–2028. Category IV status brings biennial supervisory stress tests, replacement of the fixed 2.5% conservation buffer with a model-driven and volatile Stress Capital Buffer, FR Y-14/Y-15 reporting build-out (the largest single data-infrastructure cost), capital and recovery planning, and FDIC insured-depository resolution plans. Reduced LCR is unlikely to bind given deposit funding.

The read-through is direct and recent. Webster Financial — $84.1bn of assets growing 6–8%, a near-exact structural twin — told investors it expected to incur “a significant portion of those compliance costs as we approach the $100 billion tier,” and prior published peer research identified that cost curve as a core rationale for its sale to Santander. A bank of EWBC’s size and trajectory faces a regulatory cost step that materially altered a direct peer’s strategic calculus. EWBC has not quantified its own pre-build spend. FY2026 expense guidance of +7–9% against 5–7% loan growth suggests some of it has begun.

3.3 Basel III Endgame — rescinded, re-proposed, and a tailwind

On 19 March 2026 the Federal Reserve, OCC and FDIC issued three proposals overhauling the US capital framework and formally rescinding the 2023 Basel III Endgame proposal. Comments closed 18 June 2026; finalisation is expected in Q4 2026 with implementation from 2027. The re-proposal is capital relief: risk-weighted-asset reductions of ~4.8% for G-SIBs, ~5.2% for Category III/IV large regionals and ~7.8% for smaller institutions. Peer estimates corroborate — M&T ~+90bps of CET1, KeyCorp ~9% RWA reduction.

EWBC’s CFO guided to a ~$7bn RWA reduction and +1.6–1.8 percentage points of capital ratios. On a ~$62–65bn RWA base that implies an ~11% reduction — roughly double the cohort average. It is plausible for a book heavy in 46–52%-LTV commercial real estate and collateralised C&I, both of which receive favourable revised risk weights, but it is a management estimate of a proposed, not final, rule and should be treated as interpretation.

The consequence is the more interesting question. EWBC already runs 15.1% CET1 against a cohort clustering at 10.2–11.8%. Add 1.6–1.8pp and it approaches ~16.8%, roughly 500–600bps above cohort median. The capital-allocation section quantifies what that costs.

3.4 The capital cycle in non-bank financial lending — the central industry risk

Bank lending to non-banks has outpaced every other loan category, comprising roughly 40% of all US bank loan growth since January 2026. The Federal Reserve’s chief regulator has stated that private credit’s systemic footprint “remains a mystery” and that “it’s been very difficult for us to have a clear understanding of where those funds have been flowing”; the Fed expanded non-depository-financial-institution call-report granularity in 2025 precisely because it could not see the exposure, and the Financial Stability Board published a dedicated Report on Vulnerabilities in Private Credit on 6 May 2026.

The cycle is already turning. Private-credit default rates rose from 8.1% in 2024 to 9.2% in 2025, and “bad PIK” — distressed payment-in-kind deferrals — reached 6.4% of total private-debt volume in Q1 2026. The September 2025 failures of First Brands and Tricolor exposed >$500m at UBS and $715m of “questionable receivables” at Jefferies, with collateral double-pledging alleged. Deutsche Bank disclosed $30bn of private-credit exposure in March 2026.

EWBC’s own disclosure, decomposed, is the finding:

Over five quarters, capital-call lending grew 6.3% while the residual non-capital-call book grew 57.3% — against ~6% total loan growth. The portfolio management describes as “approximately 30% capital call lines” is 30% capital call because the other 70% is growing roughly nine times faster. The Q1-2026 10-Q states the book is “diversified across business credit, private equity, and mortgage credit facilities” — the growth is in direct private-credit and sponsor lending, not subscription lines. At $8.3bn, NDFI is 14.3% of loans, ~93% of common equity, and over 41% of the entire C&I book.

Management’s claim that “99.99% of our NBFI loans are current” and that “the past decade there have been virtually no net charge-offs in this portfolio” is almost certainly true and almost entirely uninformative. Capital-call lines are secured by uncalled limited-partner commitments from institutional investors; they do not default until an LP base defaults, which has not happened at scale since 2008. A decade of zero losses on that product measures the absence of a stress event, not the presence of underwriting skill. That record was earned by the capital-call book and is now being used to characterise a portfolio that is 70% other things and getting more so. This is the single most important question to put to management.

Nor is EWBC evidently being paid for the shift. Spreads are compressing in exactly the segments it is growing — Q4-2025 saw ~SOFR+300 for broadly syndicated and ~SOFR+480–500 for direct lending amid “heated competition among lenders,” while capital-call lines themselves price at only 3–6%. Management’s own guidance ratifies the direction: FY2026 net charge-offs guided to 15–25bps, raised, against 9bps actual in Q1-2026.

3.5 Commercial real estate — the defensible part

CRE is 37.4% of loans at a 49% blended loan-to-value on a $3m average loan size, 20% owner-occupied. Geographically, California is $14.37bn or 68% of CRE, of which Southern California is $10.55bn — 50% of all CRE. California CRE alone equals 25.3% of total loans and ~161% of common equity.

Los Angeles office vacancy stood at 25.1% in Q1-2026, range-bound at 24–25% since 2024; multifamily vacancy was 4.2% with 3.8% rent growth, and industrial vacancy 3.8% with the Inland Empire still the tightest US market. The characterisation across asset classes is “a reset story, not a growth story.”

This is the most defensible part of EWBC’s risk profile, and it is probably not where the problem is. Office at 11% of CRE and 3.9% of total loans, at a 52% LTV on granular non-trophy West Coast collateral, would require an ~48% decline from appraised value before the first dollar of principal on the average loan is impaired. Multifamily and industrial — 44% of the book — are California’s two healthiest classes. Two caveats stand: LTVs use the “most recent available appraisal,” and with California transaction volume collapsed, stale appraisals overstate collateral coverage, so the true mark is worse than 49%; and a California-specific shock — regulatory, seismic, insurance-market or fiscal — would hit 68% of CRE and the deposit franchise simultaneously, a concentration the 10-K flags explicitly.

3.6 Rates and the policy calendar

The FOMC held at 3.50–3.75% on 17 June 2026 and removed the prior 2026 cut from the dot plot, pushing reductions to 2027–28; futures price ~4% at year-end with one 25bp hike by October, on inflation elevated by supply shocks. This validates EWBC’s asset sensitivity and its NII guidance raise — but note the construction: the CFO attributed the raise exclusively to the rate outlook, not to volume, pricing or share. That is a macro windfall booked as performance, and it reverses symmetrically.

Deposit competition remains structurally elevated post-2023, and the FY2025 10-K adds a novel channel: stablecoins, following the GENIUS Act of July 2025, as potential competition for “payments and liquidity management” — that is, transaction deposits, the cheapest funding EWBC has.

On geopolitics, mid-2026 is a détente window inside a secular deterioration. The Supreme Court struck down the IEEPA tariffs on 20 February 2026; the replacement 10% Section 122 global tariff took effect 24 February and expires by statute on 24 July 2026 — six days after this report’s date — with only Congress able to extend it and no extension passed. USTR faces a 20 July 2026 deadline on two Section 301 investigations proposed at 12.5% across 46 countries including China. The COINS Act of 2025, signed 18 December 2025, codifies “reverse CFIUS” but does not take effect until Treasury issues regulations by 13 March 2027. Average applied rates were ~34% (US) and ~31% (PRC) as of February 2026. The realistic tail for EWBC is not a single event but client-side attrition across cross-border trade finance and Chinese-national US real-estate and deposit activity. The company has partially hedged by following clients’ manufacturing relocation into Southeast Asia and Mexico.

Verdict: a structurally below-average industry, and EWBC’s sub-segments are at a late and deteriorating point in the capital cycle. The product is a commodity funded by a commodity; there is no pricing power in the core spread; capital intensity is high; a regulatory cost step arrives in 2028–29; and consolidation removes cost capacity rather than competitive capacity. Cohort returns cluster in a narrow 12–17% ROTCE band at 1.3–2.3x tangible book — a narrow band is itself the signature of weak differentiation. EWBC sits at the top of it, which is a genuine and non-trivial operating achievement. But the band is the band.

4. Competitive Position

4.1 The gap, quantified

EWBC reports a FY2025 efficiency ratio of 35.69% (36.23% in Q1-2026) against a peer median near 60%:

Peer figures from prior published peer research (WBS 2026-06-26, MTB 2026-06-27, HBAN 2026-06-19, BOH 2026-06-07, GBCI 2026-06-08, KEY 2026-06-21, CFG 2026-06-21); CATY from ROIC.ai. The CATY efficiency estimate is not primary-sourced and should be treated as indicative only.

Applied to FY2025 revenue of $2,931.9m, closing a 24-point gap to peer median would cost roughly $700m of pretax income — about 40% of pretax profit. The gap is the entire earnings differential. Two qualifications follow, and both are essential.

4.2 What the moat is not: the deposit franchise

The popular account of EWBC holds that a culturally-anchored Chinese-American depositor base provides cheap, sticky funding. The filings do not support this. FY2025 average cost of deposits was 2.46%; interest-bearing deposits cost 3.24%; cost of funds 2.56%. That is expensive — above Webster’s 2.05% blended and Glacier’s 1.40% cost of funds, level with Huntington’s interest-bearing deposits, above Bank of Hawaii’s interest-bearing liabilities. Noninterest-bearing deposits at 25% are mid-pack (WBS 29%, GBCI 30%, BOH 27%, MTB ~27%; only KEY at 18.6% and HBAN at 18% are materially lower). And 38% of the book is time deposits — the least sticky, most rate-shopped category in banking.

Worse, the mix has deteriorated: average noninterest-bearing deposits fell from 29% of average deposits in FY2023 to 24% in FY2025, and 53% of FY2025’s $3.91bn of deposit growth came from time deposits. A franchise whose growth is majority CDs and whose noninterest-bearing share fell 500bps in two years is buying its growth, not earning it.

EWBC pays up for deposits. Whatever the moat is, it is not a funding-cost moat, and anyone underwriting this stock on deposit-franchise quality is underwriting something the numbers do not show.

4.3 What the moat is: expense structure and scale within a bounded market

FY2025 noninterest expense of $1,046.4m is 1.30% of assets, decomposing into compensation and benefits $618.8m (21.1% of revenue), occupancy and equipment just $66.1m — 2.3% of revenue and 0.08% of assets — computer and software $54.7m, deposit account expense $35.2m, FDIC and regulatory $31.7m, other operating $165.0m, and tax-credit/CRA amortisation $74.8m. Derived: revenue per employee $875k; assets per employee $24.0m; core US customer deposits per US branch ~$613m.

The mechanism is a high-density, low-touch delivery model. A $613m-per-branch deposit base is several times a typical US regional’s; occupancy at 0.08% of assets is a rounding error. The branch network is small and concentrated in dense urban Asian-American corridors — the San Gabriel Valley, Flushing, Houston. On the asset side the lending is wholesale in character even where ticket sizes are modest: $7.6bn of non-depository financial institution lending, $2.2bn of media and entertainment, $1.0bn of broadly syndicated loans, $503m of art finance — relationship-sourced, low-headcount, high-balance businesses. Low branch cost plus low origination headcount plus high per-unit balances produces the gap, and it is visible in the line items rather than asserted.

The decisive control experiment is Cathay General Bancorp. Same Los Angeles origin, same Chinese-American customer base, same era, same geographies, same products. FY2025: $24.23bn of assets, 11.25% ROE, 1.33% ROA, against EWBC’s $80.43bn, 16.01% ROE and 1.70% ROA. This refutes the naive proposition that the Chinese-American niche is itself a moat — two banks in an identical niche earn 11.3% and 17.0%. Cultural and linguistic affinity is table stakes within the niche, not a barrier to entry. What it supports is a scale thesis: EWBC is 3.3x Cathay’s size in the same customer pool, amortising fixed costs that do not scale with assets — the China licence and its compliance apparatus, BSA/AML and OFAC infrastructure for cross-border flows, treasury and FX platforms, technology, and the $7.6bn non-bank lending capability — over 3.3x the revenue base.

In Greenwald’s taxonomy the moat is therefore economies of scale combined with customer captivity, inside a geographically and demographically bounded market. It is not a supply-side cost advantage in the input-price sense; it is not brand; it is not a network effect. It is the classic Greenwald configuration — a firm dominant within a defined market it does not attempt to leave. The regulatory component is genuine and is the strongest single element: the EWCN China banking licence is not obtainable by a new US entrant today, given Executive Order 14117’s data restrictions (effective 8 April 2025) and the PRC’s cyber and data framework, both discussed at length in the FY2025 10-K. That is a licence-based barrier — the most durable kind — and the one thing a JPMorgan or Wells cannot simply buy past.

Captivity is real but exists in a middle band: customers too small or too idiosyncratic for the global banks to underwrite at that documentation profile, too complex for Cathay-scale community banks. That band is defensible and finite — not expandable into a national franchise, which is precisely why there are 96 branches and no acquisitions.

4.4 The durability caveat — a third of the gap is the rate cycle, and part is accounting

Two adjustments materially shrink the headline advantage.

First, the efficiency ratio has not been ~36% for long. The 10-K series reads 44.23% (2019), 44.42% (2020), 43.80% (2021), 36.65% (2022), 39.22% (2023), 36.65% (2024), 35.69% (2025). Roughly 8 of the ~24 points is a rate-cycle denominator effect — NII rose 12% in FY2025 alone against a fixed cost base.

Second, and more consequential, roughly 5 points is accounting geography. Effective 1 January 2024 EWBC adopted ASU 2023-02, moving ~$89–90m a year of tax-credit amortisation out of noninterest expense and into income tax expense. Restating FY2025 on the FY2023 presentation — all $223.8m of amortisation in noninterest expense — puts expense at $1,195.4m and the efficiency ratio at ~40.8%, not 35.69%. The reported improvement from 39.22% in 2023 to 35.69% in 2025 is therefore largely presentational.

A corroborating tell: EWBC disclosed an adjusted efficiency ratio in the FY2022 (31.74%) and FY2023 (31.63%) 10-Ks — where the add-back was precisely this tax-credit amortisation — and discontinued the disclosure in FY2024 and FY2025. Once the reclass did the work automatically, the non-GAAP metric was retired.

Netting both, the durable structural gap is nearer 16 points than 24 — still an enormous, real advantage, but materially less than the headline. And management’s own FY2026 guidance points the wrong way: expenses +7–9% against NII +6–8% implies the ratio deteriorates.

To EWBC’s credit, there are no definitional games in the reported number itself: it is computed on GAAP noninterest expense over total revenue, including the $74.8m of tax-credit and CRA amortisation that many peers route through the tax line, and the company leads with GAAP rather than its own lower adjusted figure. The ratio is if anything conservatively stated on its current basis. The problem is comparability across time, not manipulation.

4.5 Key-person risk

Dominic Ng, age 67, has been Chairman and Chief Executive since 1992 — 34 years — and a director since 1991. The roles are not separated; the 2026 proxy justifies this by reference to “the strong independence of the Board with 10 of the 11 directors being independent.” There is no named successor and no succession timeline; disclosure is entirely process-level. Ng beneficially owns 924,071 shares — under 1% (~0.68%); all seventeen directors and executive officers combined hold 1,285,692 shares, also under 1%. The largest holders are institutions: BlackRock 9.5%, Capital International 7.2%, Invesco 6.8%, FMR 6.2%.

The senior bench is external and recent: CFO Christopher Del Moral-Niles joined October 2023 from Associated Banc-Corp; Head of Global Banking Deborah Leerhsen joined December 2024 from CBA, Mizuho and HSBC; COO Parker Shi joined 2021 from McKinsey; Vice Chairman Douglas Krause is 69. Long-tenured insider Irene Oh moved from CFO to Chief Risk Officer in October 2023 after thirteen years in the finance seat.

This is moat risk, not generic CEO-transition risk, for three specific reasons. The 10-K itself grounds the competitive claim in “senior management and Board of Directors’ ties to Asian business opportunities and Asian American communities” — management tells you the advantage is embodied in people. The PRC-regulator standing that supports EWCN is not a transferable asset in the way a deposit base is. And the succession bench consists of capable outside professionals with two to five years of tenure, none of whom carry the community or regulatory standing at issue; the 2023 lateral of the thirteen-year CFO removed the most obvious internal continuity candidate from the finance seat. The proxy’s phrasing — crediting Ng with “the thoughtful expansion of its leadership team to support robust succession planning” — suggests the Board regards those external hires as the plan.

Verdict: a real and durable competitive advantage, but narrower, more cyclical and more person-dependent than the headline numbers imply — call it a moat of medium width and questionable transferability. The affirmative case is strong and financially verified: an expense advantage visible in occupancy at 0.08% of assets and $613m of deposits per branch, producing ROTCE at or above 16.9% for four consecutive years and never below 12.4% in six, entirely organically, reinforced by a China licence no US peer can replicate. The Cathay comparison isolates the mechanism as scale rather than ethnicity. But the deposit franchise is not the moat and may be widely mistaken for it; roughly a third of the current advantage is rate cycle and accounting presentation rather than structure; and the whole rests on a 67-year-old Chairman-CEO with no disclosed successor and less than 1% of the stock.

5. Growth History and Forward Opportunities

5.1 The record

Loans grew from $41.69bn (2021) to $58.13bn (Q1-2026) — +39%, an ~8.0% CAGR. Deposits grew 29%, a ~5.9% CAGR. Diluted EPS compounded ~10.2% a year since 2018. And tangible book per share compounded 12.8% a year from 2021 to 2025, before $2.40 a share of FY2025 dividends, while the share count shrank. Goodwill has been $465.697m in every balance sheet from 31 December 2020 through 31 March 2026: this is entirely organic growth.

That combination deserves emphasis because it is genuinely rare. There is no purchase accounting, no accretion income masquerading as margin, no integration risk, no goodwill impairment exposure, and no serial-acquirer adjusted-earnings bridge. An 8% loan CAGR funded by a 12.8% tangible-book CAGR with a shrinking share count, producing ROTCE that never fell below 12.4% through COVID, the 2023 regional-bank crisis and a 200bp+ rate round-trip, is something almost no regional bank of this size can claim. It is the evidentiary core of any bull case.

5.2 The composition problem

The form of the growth is excellent. The substance of the most recent growth has changed, in three ways that are systematic rather than random.

Loan growth has migrated to non-differentiated lending. FY2025 total loan growth was $3.17bn. Non-depository financial institution lending alone grew $1.6bn — half of all loan growth. Because total C&I grew only $1.25bn, core C&I excluding NDFI contracted by roughly $350m. The 10-K characterises FY2025 as “well-balanced growth across major loan types”; the balances do not support that description. This matters beyond mix: capital-call and sponsor lending is a commoditised, low-spread, private-equity-cycle-correlated product in which EWBC has no niche advantage whatsoever — it is competing against every large bank and the private-credit funds themselves, on price. It is the clearest available signal that the differentiated book is not growing fast enough on its own to reach the 5–7% guide.

Deposit growth has migrated to the most expensive category. Of FY2025’s $3.91bn deposit increase, $2.07bn — 53% — was time deposits, and average noninterest-bearing deposits fell from 29% of average deposits in FY2023 to 24% in FY2025.

Fee growth is one line item. Of the +$10.63m increase in core fee income in Q1-2026, wealth management contributed +$8.58m, or 81%; excluding it, core fees grew $2.05m, or 2.3%. Two of the five core lines — lending/servicing and foreign exchange — declined year over year.

5.3 Forward opportunities

The credible ones are: continued organic share gain in the Chinese-American commercial niche, where EWBC’s 3.3x scale advantage over Cathay is compounding; geographic extension into East Coast Chinese-American population centres (management has hinted at this, and it is the most natural use of surplus capital); further build-out of wealth management, if it can be converted from transactional product distribution into recurring advisory revenue; and following clients’ supply-chain relocation into Southeast Asia and Mexico, which converts part of the geopolitical risk into a nearshoring opportunity. The constraint is structural: organic-only growth means there is no inorganic lever left to pull if the niche saturates, and the adjacency management actually chose in FY2025 was the commoditised one.

Verdict: high-quality in form, deteriorating in substance. The organic, self-funded, book-compounding character of the growth is real and should not be undersold. But its 2025–26 composition is lower-quality than its 2021–23 composition on all three of loans, deposits and fees, and the deterioration is directional rather than noise. Weighing the disconfirming evidence honestly: 6% loan growth produced at 49% CRE loan-to-values with a 15.1% CET1 is a conservatively-produced 6%, and the noninterest-bearing deposit share stabilised at 25% in Q1-2026. That is enough to reject a “the franchise is breaking” reading. It is not enough to call the mix shift benign.

6. Financial Quality

6.1 Returns and margin

A note on definitions, because it caused confusion in our own work: EWBC’s headline “return on equity” of ~17% is return on average tangible common equity. Return on average common equity is 16.01% for FY2025 — the figure EWBC itself reports in the proxy’s Pay-versus-Performance table. Both are used below, labelled.

6.2 Margin is a liability-side story

Every basis point of margin expansion since 2024 came from funding costs, not asset yields:

Earning-asset yield has fallen 52bps from FY2024’s 6.01% to Q1-2026’s 5.49%. Interest-bearing deposit cost has fallen 99bps over the same span. Net interest income growth is a deposit-repricing story with an earning-asset-growth kicker — not a franchise-pricing story.

Management’s claim that interest-bearing deposit costs are down “111bps since cuts started” checks out: peak was 3.93% in Q3-2024 against 2.84% in Q1-2026, a 109bp decline. The related claim of “comfortably exceeding our 50% beta guidance” is true but flattered by mix — the 10-K discloses a 56% weighted-average total-deposit beta, a blend that includes ~25% noninterest-bearing deposits whose beta is zero by construction. The noninterest-bearing book does no repricing work; it dilutes the average. This is disclosed correctly and is not a deception, but it sounds like pricing power and is partly arithmetic.

The tailwind is roughly one year from exhaustion. Time deposits are $25.4bn at a 3.36% average rate; management prices CD specials at 3.60%. Maturing 2025-vintage CDs carried ~3.9–4.0%, so the current roll still saves 30–55bps — but once the book converges on 3.60%, rolling a CD becomes cost-increasing rather than cost-decreasing. With ~$10bn rolling in Q1-2026 alone, convergence is about a year away. Simultaneously the asset side is diluting: available-for-sale securities grew 22% to $13.2bn and now represent ~17% of average earning assets at a 4.42% yield against 6.11% on loans.

Rate sensitivity is modest and asymmetric-friendly at present: the 10-K’s twelve-month dynamic model shows +200bp instantaneous +5.6% / ramped +3.4%; −100bp −3.2% / −1.5%; −200bp −5.9% / −3.0%. On ~$2.75bn of run-rate NII, a 100bp cut costs ~$88m instantaneous or ~$41m ramped — roughly $0.33–$0.46 of after-tax EPS. This is the whole explanation for the FY2026 guidance raise, which the CFO attributed exclusively to the shift to a no-cuts outlook. The raise was a rate assumption, not an operating improvement.

6.3 Quality of earnings — the tax-credit reclass

EWBC is a very large tax-credit investor: $969.5m of affordable-housing, tax-credit and CRA investments plus $337.5m of unfunded commitments.

On 1 January 2024 EWBC adopted ASU 2023-02, applying proportional amortisation to new-markets, historic, production and renewable-energy credits — moving ~$89–90m a year of amortisation out of noninterest expense and into income tax expense. This single reclassification explains both of the metrics management most wants you to look at.

First, the efficiency ratio is not comparable to its own history. Restating FY2025 on the FY2023 presentation — all $223.8m of amortisation in noninterest expense — gives expense of $1,195.4m and an efficiency ratio of ~40.8%, not 35.69%: a 511bp difference. The reported improvement from 39.22% (2023) to 35.69% (2025) is largely geography, not operating leverage. And EWBC disclosed an adjusted efficiency ratio in the FY2022 and FY2023 10-Ks — where the add-back was precisely this amortisation — and discontinued it in FY2024 and FY2025, once the reclass performed the same function automatically.

Second, the effective-tax-rate increase is the same reclass in reverse, and the common narrative about it is wrong. The rise from 20.5% to 23.2% looks like a $103.5m ($0.74/share) headwind. But $149.0m of amortisation now sits in the tax line, accounting for 8.6 percentage points of the 23.2% rate against 2.9 points in 2023. Strip it out and the underlying tax burden is $251.3m on $1,725.5m of pretax income — 14.6% in FY2025 against 17.5% in FY2023. It went down, not up. Economically the tax-credit programme contributes more than before: credits of $318.2m less total amortisation of $223.8m is a $94.5m net benefit in FY2025 against $22.0m in FY2023. Net income is unaffected by the geography. The genuinely economic drivers of the higher headline rate are California state apportionment and a one-time deferred-tax-asset revaluation on adopting the California single-sales-factor method in 2025.

6.4 Other quality-of-earnings items

One-time items across the five-year set roughly offset: an ~$60m FDIC special assessment (FY2023); +$32.3m of loan-payoff discount accretion and purchased-credit-impaired interest recoveries in FY2025; +$18m from a Q3-2025 reversal of credit losses on a PCI payoff; −$31m from a Q3-2025 change in equity-award expense recognition for retirement-eligible employees; and ~+$4m of DC Solar recovery and Rayliant fair-value gain. Net pretax ~+$23m, about $0.12 a share. The earnings level is close to clean; the distortion is in geography and in the margin narrative. Note in particular that the 10-K’s own non-GAAP reconciliation shows an adjusted average loan yield ex-accretion of 6.34% against 6.40% reported — 6bps of FY2025’s 14bp NIM expansion was a one-time payoff recovery.

On securities marks: accumulated other comprehensive income improved from −$620.6m (2023) to −$345.6m (2025), a ~$275m tailwind that flattered tangible-book growth — and it reversed in Q1-2026, worsening $42.6m to −$388.2m. Held-to-maturity securities carry an unrealised loss of $406.0m pretax at 31 March 2026 ($2,859.0m amortised cost against $2,453.0m fair value), on a 5.9-duration book yielding 1.70%, all government or agency guaranteed. Marking it at the 10-K’s own 28.02% blended statutory rate reduces tangible book by $2.13 a share — a 3.4% haircut, taking adjusted tangible book to $60.14 and price-to-tangible-book to ~2.24x. This is modest and not a thesis risk — held-to-maturity is only $2.9bn, 3.5% of assets, versus the double-digit marks that broke banks in 2023 — but it is widening, and marked tangible common equity is ~10.0% rather than 10.3%.

6.5 Credit

On the measures the market quotes, credit is pristine: net charge-offs of 9bps, non-performing assets of 26bps, and an allowance of $835.9m plus a $47.0m unfunded-commitment reserve — $882.9m total, 1.51% of loans, or 6.1–10.1x management’s own guided FY2026 losses of $87–145m. Against $913m of classified loans the allowance is 0.92x, and classified assets do not become total losses; at a 30–40% severity the embedded loss content is $274–365m, comfortably covered. The reserve is adequate.

But those are lagging measures, and the leading measure moved. In Q1-2026 alone, classified loans rose $117m, or 15%, to $913.4m — 1.57% of loans, the highest in the disclosed series. Total criticized rose 8%, nonaccruals 9%, accruing past-due CRE 223% to $45.2m, and HELOC nonaccruals 69% to $29.0m. And management raised FY2026 net-charge-off guidance to 15–25bps, or 1.7–2.8x the current 9bp run rate. Bank managements do not raise loss guidance for sport; they raise it when internal risk-rating migration tells them to — and that migration is visible in the filing. Provisions would need to rise $30–85m against FY2025’s $59.8m of charge-offs, worth $0.16–$0.44 of after-tax EPS, or 2–5% of earnings. This is the most informative disclosure of the quarter and it is receiving the least attention.

Verdict: do economics improve with scale? They did, and they have stopped. Efficiency went from 44.4% (2020) to 35.7% (2025) and ROATCE from 12.4% to 17.0% as assets grew from $52bn to $80bn — real operating leverage over the full period. But roughly half of the reported improvement since 2023 is accounting geography, and the company retired the metric that would have shown it. The forward evidence is worse: Q1-2026 revenue +12% against expenses +11% is 19bps of improvement, effectively none, and FY2026 guidance of ~+7% revenue against +7–9% expenses is management guiding to zero-to-negative operating leverage while simultaneously guiding charge-offs to 1.7–2.8x the current rate. The balance sheet is genuinely strong — 15.1% CET1, 10.3% tangible common equity, 84% loan-to-deposit, a 49% blended CRE loan-to-value, a held-to-maturity mark worth only 3.4% of tangible book, and 1.29x liquidity coverage of adjusted uninsured deposits — and the earnings level is clean. This bank is not fragile. But returns of ~16% ROAE and ~17% ROATCE are now being defended rather than extended, by a deposit-repricing tailwind a year from exhaustion, a fee line levered to the same rates, an efficiency ratio flattered by a reclass, and a credit book whose classified assets just rose 15% in a quarter.

7. Capital Allocation

7.1 The excess-capital question

At 31 March 2026 EWBC reported CET1 of 15.1%, Tier 1 15.1%, total capital 16.4% and Tier 1 leverage 11.0%, against a well-capitalised CET1 threshold of 6.5% and a minimum-plus-conservation-buffer requirement of 7.0%. The bank runs at more than double its buffer-inclusive requirement, and against a peer range of 9.9–13.5%.

Sizing the surplus depends on the target assumed. Against an 11% CET1 operating target the excess is roughly $2.2–2.4bn, or 12–13% of market capitalisation, about $16 a share. Management’s estimated Basel relief (~$7bn of risk-weighted-asset reduction, +1.6–1.8 percentage points) would take the ratio toward ~16.8% and the surplus to ~$3.2bn, or ~17% of market capitalisation.

The cost is measurable. Excess capital earns the securities yield, not the franchise yield. Returning ~$2.2bn — and forfeiting ~$68–83m of after-tax investment income on it — would lift ROTCE from ~17% to roughly 20%, which at a 10% cost of equity and 5% growth would justify a multiple above 3x tangible book. On this arithmetic EWBC’s reported returns understate its earning power, and the current multiple is cheap. It is the strongest quantitative argument in the bull case.

The rebuttal is decisive, and it is behavioural rather than analytical. EWBC has carried a 14–15% CET1 for years with no evident urgency to deploy it. Goodwill has been frozen since 2020 — six years without an acquisition. The payout ratio is ~25% on FY2025 and ~31% on the raised dividend. Internal capital generation of ROTCE × (1 − payout), roughly 11.8% a year against 5–7% asset growth, means the capital pile compounds faster than the balance sheet and management is content to let it. Management’s stated hierarchy on the Q1-2026 call — organic growth, then dividend, then “inorganic at the right price,” then opportunistic buybacks, with repurchases ranked last — does not clear the surplus; it accumulates it. Capital generation is running roughly 3x capital consumption.

Capitalising a distribution that management has declined to make for six years is precisely the error this framework exists to prevent. The excess capital is a genuine option with an indefinite exercise date. It belongs in the bull case, not the base case. And the ROE series tells the story plainly: 24.5% (2018) → 16.01% (2025), a decline that is not deteriorating profitability but capital accumulating faster than it is deployed.

7.2 Buybacks — the 2023 test

Credit where it is due: the March 2020 purchase of 4,471,682 shares at $32.64 was outstanding — $146m deployed at the bottom, now worth $601m, accounting for 53% of all cumulative buyback gains and dragging the six-year blended average down to $59.92.

And then the 2023 failure. EWBC repurchased zero shares in the first, second and third quarters of 2023. The FY2023 10-K’s full-year figure of 1,506,091 shares at $54.56 is identical to the Q4-2023 issuer-purchases total. The stock bottomed at $30.90 on 13 March 2023 — cheaper than the COVID low the company had bought three years earlier, and roughly 0.7x tangible book — with $254.0m of authorisation already approved and unused.

The steelman defence is real: repurchasing during a system-wide deposit run signals capital depletion to depositors and regulators at the worst possible moment. Three facts undercut it. EWBC was simultaneously issuing 8-Ks on 13 and 15 March 2023 publicly reiterating its capital and balance-sheet strength — it wanted the market to believe it was fortress-capitalised while declining to act on that belief. It entered the crisis at ~13% CET1, not 8%. And it waited seven months: by the time buying resumed in October the stock had already rallied ~76% off the low. Waiting until the risk is gone means paying for the absence of risk.

The asymmetry is the tell: in the very same window the company sat out, its own officers and directors bought with personal money (see below). Management’s private capital said the stock was cheap at $41–45; management’s corporate capital did not act until $54.56.

And the pace has inverted. Repurchases now accelerate as the price rises, culminating in the largest quarterly outlay in company history in Q1-2026, at ~2.2x tangible book — the 97th percentile of the stock’s own ten-year range. At that multiple every dollar spent retires only ~$0.45 of tangible equity; the repurchase is immediately tangible-book dilutive, and is EPS-accretive only because a 17% ROTCE exceeds a ~7.4% earnings yield — a real but far weaker argument than the one available at $31.

7.3 Dividends and M&A

Dividends are the clearest positive. The quarterly rate has gone $0.40 (Q1-2022) → $0.48 → $0.55 → $0.60 → $0.80 (Q1-2026, +33%), with annual dividends per share of $1.92 (2023), $2.20 (2024) and $2.40 (2025) against a 2026 run-rate of $3.20. The FY2025 payout was 25.2%; the forward yield is ~2.38%. Growth was uninterrupted through COVID and through the 2023 crisis, with a payout never above the mid-20s. The 33% raise is the first substantive acknowledgment that the capital position is excessive — but even at $3.20 the payout is only ~31%, and 2026 total capital return of ~$740m against ~$1.45bn of likely earnings still retains roughly half of profits into an already-bloated stack. The dividend is a pressure-release valve, not a solution.

M&A: sixteen years of nothing, and a loaded gun. The only deployment since 2020 is a 49.99% interest in Rayliant Global Advisors (Q3-2023, $95m cash plus 349,000 performance-based RSUs contingent on Rayliant’s revenue and EBITDA through September 2028, carried as a derivative liability at ~$15m). The 49.99% structure is deliberate — one basis point below control, avoiding consolidation and goodwill. It is a toe-dip, not a strategy.

The franchise-defining transaction was the 6 November 2009 FDIC-assisted acquisition of United Commercial Bank’s banking operations — ~$5.90bn of loans, $599.0m of cash, $1.56bn of securities, under a shared-loss agreement in which the FDIC absorbed 80% of losses on the first $2.05bn of covered loans and 95% thereafter. It roughly doubled the bank and delivered the China footprint that defines it today. It was a genuinely great transaction — and it was a government-subsidised distressed purchase seventeen years ago. It is not evidence of repeatable M&A skill, and nobody currently in management has integrated a bank acquisition.

This produces the largest unpriced risk in the name. At 2.19x tangible book EWBC’s stock is excellent acquisition currency: a target at 1.2–1.5x tangible book is immediately accretive to EWBC’s tangible book per share. Combined with $2.4–3.2bn of surplus capital, the arithmetic genuinely favours paper-funded M&A over repurchase — and that same arithmetic is how banks destroy capital. A management team with $3bn of excess capital, a record-high currency, no integration experience in sixteen years, and a compensation plan that pays 20% of the annual bonus on average loan growth has every structural incentive to overpay for balance sheet. A deal announcement would be a thesis event.

7.4 Insider transactions

All 176 Form 4 filings in the five-year window were retrieved and parsed — 252 discrete transactions: 75 grants (code A), 66 sales (S), 44 tax withholdings (F), 25 exercises (M), 24 gifts (G) and 18 open-market purchases (P).

Every open-market purchase happened in one nine-week window. All 18 code-P transactions fall between 13 March and 11 May 2023, totalling ~$2.13m at $41.57–$53.37. The buyers were broad-based and self-funded: Dominic Ng (17,600 shares, $751k, at $41.57–$43.94), Irene Oh (then CFO, 11,000 shares from 13 March), Gary Teo, Parker Shi, and five directors — Sussman, Campbell, Alvarez, Deskus and Kay. That was a real signal and it was correct.

There has not been a single open-market purchase since 11 May 2023. In more than three years, across a price range of $54 to $136, no insider has bought one share.

Dominic Ng sold 260,000 shares for $28.36m across 2025–26 at $90.01–$125.34, against 17,600 shares bought for $751k in 2023; his direct holdings fell from 1,011,492 (March 2024) to 707,930 (May 2026) — a 30% reduction — and on 18 March 2026 he gifted a further 190,730 shares into trusts, an estate-planning transfer executed at a record valuation.

The procedural detail matters most. An 8-K dated 20 November 2024 disclosed Ng’s Rule 10b5-1 plan for up to 120,000 shares effective March–December 2025, and his 2025 sales carry the corresponding footnotes. Every one of the eleven 2026 filings containing a sale carries aff10b5One = 0 — none is executed under a trading plan. The 2026 discretionary sellers are Ng (50,000 shares in February at $117–118; 60,000 in May at $123–125), Vice Chairman Douglas Krause (10,000 at $123.50), Chief Risk Officer Irene Oh (12,511 shares in June at $129.00–$133.19, the highest prices any insider has ever received), directors Liu, Campbell and Deskus, and EVP Gary Teo (7,000 shares, leaving him 5,252).

The shift from planned to discretionary selling, at all-time highs, by the Chief Executive, Vice Chairman, Chief Risk Officer and multiple directors simultaneously, is the most informative datapoint in the filing corpus. These are the same people who demonstrated in 2023 that they will buy with personal capital when they believe the stock is cheap. Note especially that the Chief Risk Officer sold at the all-time high, discretionarily, in the same quarter classified loans rose 15%. (The May 2026 grant of 1,156 shares to each of ten non-employee directors is the routine annual equity retainer and is not a signal.)

7.5 Compensation and governance

Dominic Ng’s FY2025 total compensation was $9,683,884 (salary $1,275,000, frozen three years; stock awards $5,094,193; non-equity incentive $3,179,850), against $9,213,633 in 2024 and $8,358,550 in 2023 — ~3.2x the next-highest NEO, with a CEO pay ratio of 83:1. “Compensation actually paid,” which marks unvested equity to market, was far higher at $18,512,734 in 2025 and $27,525,426 in 2021; Ng’s realised economics are strongly levered to the share price, which is genuine alignment while he holds it.

What he is paid on is the problem. The annual bonus is 65% financial, 20% strategic, 15% risk, and the financial component is 30% adjusted diluted EPS, 30% pre-provision pre-tax income, 20% average total loan growth, 10% criticized-loans ratio, 10% net-charge-off ratio. There is no ROE, no ROTCE, no ROA and no efficiency ratio anywhere in the annual bonus. A bank chief executive sitting on $2.4bn of excess capital is paid 80% on growth and volume and 0% on returns — a structure that directly rewards balance-sheet accumulation and is silent on the capital inefficiency that is this section’s central criticism. It also explains the M&A risk above.

Targets have been progressively de-stretched. FY2023 targets were set 10% above prior-year actual; FY2024 targets were set below 2023 actual; and for FY2025 the proxy states that target adjusted diluted EPS and pre-provision pre-tax income were “$8.30 and $1,661 million, respectively, both of which were the same as 2024 results” — zero growth required to earn 100% of target. FY2025 paid out at 172% corporate, 181% financial, and Ng’s bonus at 249% of salary. A 181% financial score is the arithmetic consequence of setting the bar at last year’s result. The qualitative 35% is unfalsifiable — including a 200% score on “Deepen Customer Relationships” and 172% on risk management in a year the proxy itself describes as having “continued to make progress in addressing regulatory matters affecting three lines of defense” — and the table quantifying goal achievement is published as an image rather than text.

The long-term plan is better: 100% performance-based units weighted ROA 37.5% / ROE 37.5% / relative TSR 25%, scored on percentile rank against the KBW Nasdaq Bank Index. Returns do appear here. Two flaws: the nominally three-year award is measured as three separate one-year periods merely settled at year three, removing any sustained-outperformance test; and payouts have run 187.7%, 180.4% and 169.3% for the 2023, 2024 and 2025 awards — with the 2025 award paying 169.3% despite total shareholder return at the 41st percentile, i.e. below median, because ROA and ROE carry 75% of the weight.

On governance, the mitigants are real — 10 of 11 directors independent, a Lead Independent Director (Lester Sussman), annual elections, no classified board, no poison pill, proxy access adopted in 2023, related-party dealing clean, no loans to NEOs, hedging and option repricing prohibited. Against that: Ng is Chairman and CEO, a director since 1991, and owns 924,071 shares — under 1% (~0.68%); all seventeen directors and officers together own under 1%; the CFO owns zero shares more than two years in; Compensation Committee Chair Jack C. Liu has been a director since 1998 — 28 years — and also sits on the Nominating Committee that renominates him; the clawback is the Dodd-Frank restatement-only minimum with no misconduct trigger, which matters when 35% of the bonus is discretionary; and say-on-pay support fell to 95.56% in 2025, the lowest of four years.

A new flag worth monitoring: the 2025 and prior proxies stated that “it is against Company policy for NEOs to pledge shares of common stock in the Company for any purpose.” The 2026 proxy replaces that absolute prohibition with a case-by-case exception regime, capped at 50% of holdings above guideline levels and “less than 5% of the Company’s outstanding shares” collectively — roughly $900m of stock, for a group owning under 1%. No insider has pledged anything today. But an absolute prohibition is not rewritten in the abstract.

Verdict: partially — with a clear failure at the moment it mattered most. In favour: an excellent dividend record, an outstanding 2020 buyback, no value-destroying acquisitions in six years (a real achievement in a sector where banks routinely destroy capital buying each other), trivial dilution, clean related-party dealing and an admirably boring 8-K record. Against, in ascending severity: the balance sheet is being hoarded at a cost of ~200bp of ROE, with buybacks ranked last while capital generation runs 3x consumption; the 2023 test was failed outright, with $254m of authorisation unused at 0.7x tangible book while insiders bought personally; and the repurchase pace has inverted into buying high, at 2.2x tangible book. The incentive structure explains all three. This is a management team that protects capital well and deploys it poorly — and the insider record closes the argument.

8. Changes and Headwinds — Last Two Years

Conspicuously absent across 42 8-Ks: no M&A, no restatement, no auditor change, no material litigation, no enforcement action, no capital raise. For a bank that navigated 2023, an unusually quiet filing record is a genuine positive, and the CFO transition was orderly and pre-announced.

The headwinds that matter are, in order: the migration of loan growth into non-bank financial lending at exactly the point the private-credit cycle is turning; the 15% quarter-over-quarter rise in classified loans against raised charge-off guidance; the approaching exhaustion of the CD repricing tailwind; guided negative operating leverage for FY2026; the Category IV regulatory cost step arriving in 2028–29; and an unusually dense near-term policy calendar in the US–China corridor that defines the franchise’s customer base.

Verdict: on balance these developments weaken the thesis at the margin rather than strengthen it. The dividend raise, Basel relief and continued record results are genuine positives. But the two changes with the most forward information content — the raised charge-off guidance and the NDFI mix shift — both point the same way, and neither is reflected in a 97th-percentile multiple or in a sell-side target range that brackets the spot price.

9. Risk Analysis

The shape of the risk is unusual and worth stating plainly: the high-likelihood risks are mostly valuation and margin risks, while the high-impact risks are mostly tail risks that are currently well-defended. EWBC is unlikely to break; it is quite likely to disappoint relative to a multiple that requires everything to keep working. Risk 1 and Risk 2 are the two that should govern position sizing, and they are correlated — a private-credit accident would de-rate the whole cohort and impair EWBC’s fastest-growing book simultaneously.

10. Valuation Discussion — Embedded Expectations

10.1 Framework and the governing numbers

A bank is not a discounted-cash-flow exercise. It is a levered spread book whose value is a multiple of tangible book justified by the return it earns on that book against its cost of equity:

Justified P/TBV = (ROTCE − g) / (COE − g)

Everything else is a cross-check. Enterprise value, EV/EBITDA and EV/Sales are meaningless for EWBC and are not reported here — deposits are the raw material, not a financing item.

At $134.52 against FY2025 tangible book of $61.27 a share, EWBC trades at 2.19x tangible book, 2.08x stated book and 14.1x trailing earnings. On the newer Q1-2026 tangible book of $62.27 the multiple is 2.16x; marking held-to-maturity securities takes it to ~2.24x. The analysis below uses the FY2025 basis (2.19x) for consistency with the peer set, which is struck on the same fiscal year.

A necessary honesty about the numerator. EWBC’s sustainable ROTCE is less precise than it appears. Tangible common equity was $8,433.5m (FY2025) and $7,257.4m (FY2024), a two-point average of $7,845.5m. Net income of $1,325.2m gives 16.9% on average tangible common equity, 15.7% on ending tangible common equity, and 18.0% if one grosses up the reported return on equity for goodwill. We could not reproduce EWBC’s reported 16.98% from two-point average equity, which yields 15.9% — the company appears to use a quarterly-average convention we could not replicate. We use ~17.0%, computed identically to every peer, and flag that the commonly-cited “18–19%” sits at the top of a defensible 15.7–18.0% band. The entire valuation turns on this number.

10.2 Peer comparison

*Excluded from the median. Webster is not a valid comparable — it is under a definitive agreement to be acquired by Santander and trades at deal value, not fundamental value. Glacier’s FY2025 earnings are depressed by Bank of Idaho/Guaranty acquisition expense, making its 27.0x P/E and 9.5% ROTCE artifacts rather than signals. Peer ratios were derived by hand from FY2025 balance-sheet line items because ROIC.ai’s return-on-common-equity field is unreliable for banks (it returns HBAN 36.0%, GBCI 21.3%, WBS 23.9%).

Three conclusions follow.

EWBC’s efficiency ratio is best in the cohort by ~400bps and ~2,400bps versus the median, with Cathay next at 40.4%. That the two structural comparables are the two cost leaders is meaningful: it suggests the branch-light, commercial-deposit, ethnically-anchored model is a genuine cost architecture rather than one management team’s discipline. This is the mechanical source of the ROA advantage — 1.69% against a 1.10% cohort median.

EWBC is not expensive against peers. Regressing price-to-tangible-book on ROTCE across the nine clean names gives:

P/TBV = 0.632 + 0.0964 × ROTCE(%), R² = 0.62

Each incremental point of ROTCE is worth ~0.096x of tangible book to this cohort. The fitted value for EWBC at 17.0% ROTCE is 2.27x against 2.19x actual — EWBC trades marginally below the line its peers are priced on. At 2.19x versus a 2.06x cohort median it carries a ~6% premium while earning 17.0% ROTCE against a 15.0% median. On a relative basis this is fair value, arguably a shade cheap.

The richness is against its own history — and that is a sector-wide condition. EWBC’s price-to-book sits at the 97.1st percentile of its own ten-year range. But prior published peer research found Regions Financial at the 99.2nd percentile of its own range one month earlier. Two structurally different banks both at the 97th–99th percentile of decade-long ranges is not an idiosyncratic signal — it is the regional-bank complex trading at the top of its multiple band. This reframes the risk entirely: EWBC’s de-rating exposure is sector-beta exposure, which the factor-positioning analysis independently confirms.

10.3 Embedded expectations — what must be true

Rearranging: Required ROTCE = P/TBV × (COE − g) + g.

Working the central cell, with growth anchored to the 5–7% loan-growth guide: 2.19 × (0.10 − 0.05) + 0.05 = 15.95%.

The market is underwriting something achievable, but with no margin of safety. Against a derived ~17.0% sustainable ROTCE, the required 15.95% leaves roughly one point of cushion — about 6% of the multiple. And that cushion exists only if you accept a cost of equity at or below 10%. Push it to 11% — defensible for a bank with EWBC’s China-corridor concentration and a 1.178 bank-industry beta — and the required return at 4% growth becomes 19.3%, which EWBC does not earn on any of the three derivations. At 9% it falls to ~14% and the stock is comfortably cheap.

Essentially the entire answer sits in the cost-of-equity assumption, which is unobservable. That is the honest conclusion, and it is less satisfying than either the bull or the bear would like.

The inverse grid shows why the risk is asymmetric:

Note the convexity. At a 10% cost of equity and 5% growth, a three-point ROTCE decline from 18% to 15% cuts the justified multiple from 2.60x to 2.00x — a 23% valuation haircut for a deterioration a normal credit cycle delivers routinely. The Gordon denominator is small, so the multiple is violently levered to the numerator.

10.4 Scenarios

The base case implies FY2026 EPS of ~$10.38, inside the $10.28–$10.59 Street consensus range — the model is calibrated, not optimistic. Tangible-book builds retain EPS less the $3.20 dividend, less ~$1.08 a share of dilution from repurchasing stock at 2.19x tangible book, a real and frequently-ignored cost of buying back an expensive bank.

Credit arithmetic underpins the spread: on ~$58.1bn of loans, 9bps of charge-offs is ~$52m, the guided 15–25bps is $87–145m, and a genuine recession at 60–100bps is $349–581m. Moving from 9bps to 80bps costs ~$317m after tax, or ~$2.30 a share, before any reserve build.

The critical decomposition. Taking the bear case from $134.52 to ~$91.00 (1.40x on $65 of tangible book):

• Earnings and book effect alone — tangible book grows to $65 while the multiple holds at 2.19x: $142.35, i.e. +5.8%.
• Multiple effect alone — de-rate to 1.40x on unchanged $65 of book: $91.00.

Virtually 100% of the bear-case loss is the multiple, not the earnings. Even in a genuine recession EWBC’s tangible book grows: a bank earning a 9.4% trough ROTCE at a 31% payout still accretes capital. What breaks is the market’s willingness to pay 2.19x for it. Sharper still: a de-rate to the 2.06x cohort median with zero earnings deterioration costs 6%; a de-rate to a ~1.6x own-history median costs ~27% — with the franchise entirely intact. That is what the 97.1st percentile is warning about, and it requires no operational bad news whatsoever.

Bear −32%, base +14%, bull +32% is roughly symmetric in magnitude but not in probability: the bear requires a normal credit cycle plus mean-reversion of a decade-high multiple — two base-rate events over a multi-year hold — while the bull requires a cycle peak to persist indefinitely.

10.5 What the market is pricing correctly, and incorrectly

Correctly: the cost structure — a 36.2% efficiency ratio is best in cohort and durable, and the regression shows the market pays for the resulting ROTCE at the same rate it pays every peer; EWBC’s relative position, at 2.19x actual against 2.27x fitted; and current credit, which genuinely is excellent.

Potentially incorrectly, in both directions: it is capitalising a peak-cycle ROTCE at a peak-cycle multiple, and the Gordon convexity means those two errors compound rather than offset. Conversely it gives nothing for ~$16 a share of excess CET1 — a genuine asymmetry in EWBC’s favour if it is ever deployed, though six years of non-deployment argue against capitalising it. The guided step-up in charge-offs from 9bps to 15–25bps is management telling you the credit trough is behind them, and the multiple does not reflect it. And the China-corridor concentration does not appear to be priced into the cost of equity at all — at 10% the stock is fine, at 11% it is expensive, and nothing in the tape suggests the market is applying the higher number.

11. Variant Perception

11.1 Consensus

Seven of the eight items in the full news feed since 25 June 2026 are sell-side price-target raises: Citi $154 (Buy), Barclays $150 (Overweight), Cantor $150 (Overweight), Wells Fargo $140 (Overweight), Truist $136 (Hold), Morgan Stanley $131 (Equal-weight). Three observations. Every raise trails the price — the stock reached $136.24 on 16 July; the targets were set after the move. The $131–154 range brackets the $134.52 spot, so consensus upside is arithmetically exhausted. And not one analyst is out: Buy/Overweight/Overweight/Overweight/Hold/Equal-weight, with no Sell.

This is the signature of a crowded, backward-looking consensus underwriting the extrapolation of a completed move. It is not a catalyst and it is not evidence. Where consensus is most likely offsides: it is modelling continued net-interest-income beats off the no-rate-cuts outlook that drove the Q1 guidance raise — an outlook the CFO attributed exclusively to the rate path. That is a macro call dressed as franchise momentum, and if the Fed cuts it reverses mechanically with no offsetting franchise driver in the models.

11.2 The strongest bull case

EWBC is the most efficient bank in the American regional cohort, running a 36.2% efficiency ratio no peer approaches, and it earns that through a genuinely differentiated franchise that competitors cannot replicate through branch density or price. That produces 1.69% ROA and ~17% ROTCE on a 15.1% CET1 — best-in-class returns on a fortress balance sheet, a combination that essentially does not exist elsewhere in the table. Return on equity has not fallen below ~15% in eight years including COVID, and earnings per share have compounded ~10.2% annually since 2018 with zero acquisitions: organic, un-levered, un-financially-engineered compounding, with tangible book per share growing 12.8% a year while the share count shrank. Credit is pristine at 9bps of charge-offs and 26bps of non-performing assets against 1.51% of reserves. The China banking licence is a genuine, unrepeatable regulatory barrier. And the market prices all this at 14.1x earnings and marginally below the peer regression line, while giving no credit for ~$16 a share of excess capital that, if deployed, lifts ROTCE toward 20% and justifies a multiple above 2.8x. You are buying the cohort’s best operator at cohort-average pricing, with a free option on capital return.

11.3 The strongest bear case

You are paying the 97th percentile of a decade of price-to-book for a bank at the top of its credit cycle, and the arithmetic says nearly all your downside is multiple rather than earnings. Management has told you credit is normalising — guiding charge-offs from 9bps to 15–25bps — while guiding expenses to grow 7–9% against 6–8% net interest income, which is negative operating leverage eroding the one metric the entire premium rests on. The 2026 guidance raise was attributed purely to the rate outlook: this is a rate-cycle earnings peak, not a franchise inflection. The celebrated efficiency ratio is ~40.8% restated on the prior presentation, and the company retired the disclosure that would have shown it. The growth engine has migrated into non-bank financial lending — half of all 2025 loan growth, with the safe capital-call piece static and the opaque residual compounding 57% over five quarters — at exactly the moment private-credit defaults are rising and regulators concede they cannot see the exposure. The Gordon convexity is unforgiving: three points of ROTCE erosion removes 23% of the justified multiple, and a de-rate to EWBC’s own historical median costs ~27% with the franchise undamaged. The excess capital that supposedly rescues the valuation has sat idle for six years. The concentration that generates the returns argues for a higher cost of equity than the ~10% the price implies, and at 11% the stock does not clear. And not one insider has bought a share in three years while the CEO, Vice Chairman and Chief Risk Officer sell into the high outside any trading plan.

11.4 The assumptions that matter, and what falsifies each

Assumption 4 is the fulcrum. At a 9% cost of equity EWBC requires 13.8% ROTCE and is cheap; at 11% it requires 18.1% and is expensive. Nothing else in the analysis moves the answer as far.

11.5 Factor positioning — what the tape actually says

Within the “All Factors” model (R² 0.816, 17 July 2026), EWBC’s largest loadings are Industry: Banks +1.178 and DividendYield +0.963, followed by Market +0.903, Financials +0.530, Beta +0.499, SmallSize +0.487, Value +0.474 and CreditRisk +0.420 — while Quality is negative (−0.112) and Low-Volatility is negative (−0.135). The simpler “Base” model is starker still: DividendYield +1.081, Beta +0.898, CreditRisk +0.445, Value +0.319, against Growth −0.284 and Quality −0.198. (Betas are comparable only within a single nested model, never across models.)

This is the report’s empirical variant perception. EWBC’s fundamentals are unambiguously high-quality. The market does not trade it that way. It trades it as a high-beta, dividend-paying, credit-cyclical value bank, loading positively on Value and Credit Risk and negatively on Quality and Low-Volatility. The tape and the income statement disagree about what this company is.

Two readings, and the report holds both. Either the market is wrong and a quality franchise is being priced as generic bank beta — the bull case. Or the market is right that, whatever the return on assets, an $81bn balance sheet levered roughly nine times into California commercial real estate and capital-call lending is a credit-cyclical instrument, and the quality shows up in good years by construction. The March 2023 drawdown is the evidence for the second reading: a genuinely elite bank still lost 56% intraday in three weeks on a funding scare, because that is what levered credit does regardless of a 1.7% ROA.

An R² of 0.816 is extraordinarily high for a single name — 81.6% of EWBC’s return variance is explained by common factors, leaving under a fifth for anything idiosyncratic, with annualised idiosyncratic volatility of 13.4%. Stock selection here is largely a bet on the regional-bank complex, not on East West’s execution. That is corroborated by the cross-read: EWBC at the 97.1st own-history percentile alongside Regions at the 99.2nd means the whole cohort re-rated together, exactly as a 0.816-R² world predicts.

A striking omission: the Country: China loading is just +0.096 — de minimis. The market assigns EWBC almost no China factor exposure. The geopolitical tail risk that defines the franchise is not being priced as a factor at all, which means a corridor shock would arrive as idiosyncratic rather than as something holders are already compensated for.

One tension, flagged rather than resolved: the Base model shows an InterestRate beta of −0.145 — EWBC has tended to fall when the rate factor rises — while management states that “higher for longer is net better for East West Bank” and raised guidance on exactly that basis. Macro spreads are not orthogonalised in this model, so this is closer to a raw correlation than a clean loading, and it likely reflects the 2022–23 episode when rising rates hit bank funding and accumulated other comprehensive income. It should not be over-read; it does not corroborate the asset-sensitivity story either.

On the record: three-year return +35.6% annualised at a 1.09 Sharpe with a −35.8% maximum drawdown; one-year +27.5%; three-month +73% annualised (~+14.7% actual quarter). Price sits 17% above its 200-day exponential moving average, +335% off the March 2023 low, ~1.3% off the all-time high. A 1.09 three-year Sharpe is strong but not anomalous, and the lifetime maximum drawdown of −92% in 2008–09 stands as a reminder of what this business model can do in a genuine credit event.

Synthesis: this is neither a falling knife nor an idiosyncratic melt-up. It is a quality compounder inside a crowded sector trade — and the second fact dominates the first. A high-quality bank at a full multiple, at the top of a sector-wide re-rating, whose price behaviour is ~82% determined by factors management does not control, with a Q2 print three days away and a sell-side that has already raised its targets to meet the price.

12. Fact vs. Interpretation

13. Open Questions

1. What is the composition of the $5.93bn non-capital-call NDFI book? The 10-Q says only “business credit, private equity, and mortgage credit facilities.” This is the fastest-growing exposure in the bank, ~93% of common equity in aggregate, and the disclosure is a single sentence. The most important unanswered question in the file.
2. What is the yield and spread on the NDFI book? Undisclosed. It determines how much of the C&I yield compression (7.85% in FY2024 to 6.43% in Q1-2026) is mix-shift into low-spread lending rather than benchmark rates.
3. Why did classified loans rise 15% in one quarter, and in which industries? The 10-Q says “classified C&I and CRE” without naming exposures. This is the gating question for the raised charge-off guidance.
4. What is the CD maturity ladder by quarter, and at what average rate? It determines precisely when the deposit-repricing tailwind flips to a headwind.
5. How much of the guided 7–9% expense growth is Category IV threshold preparation versus ordinary investment? Management has not quantified it.
6. What share of revenue is genuinely China-linked? Foreign-exchange income is 2.0% of revenue and international deposits 5.8% of the total, but the indirect, relationship-driven contribution is undisclosed — and it is the whole moat argument.
7. Is there a designated internal successor to Dominic Ng? The proxy’s language suggests the Board regards the 2021–24 external hires as the succession plan, which is a materially different risk profile from an identified heir.
8. Why is EWBC’s reported 16.98% return not reproducible from two-point average equity (which yields 15.9%)? The convention appears to be quarterly-average; the resulting ROTCE band of 15.7–18.0% is wide enough to matter to the valuation.
9. Will any insider pledge shares under the new 2026 exception regime that replaced an absolute prohibition?
10. Management’s “15 straight years of deposit growth” does not appear in the FY2025 10-K and could not be verified from primary sources beyond the last two years. It should be attributed to management, not repeated as fact.

14. What Must Be True

14.1 For the bull case

1. Sustainable ROTCE stays at or above ~16–17% once the deposit-repricing tailwind ends and charge-offs normalise to the guided 15–25bps. Falsification test: two consecutive quarters of clean ROTCE below 15%, or a full year in which the efficiency ratio prints above 40% while revenue still grows.
2. The NDFI book performs through a private-credit downturn, validating that the decade of zero losses reflects underwriting rather than the absence of a stress event. Falsification test: net charge-offs above 40bps in any quarter, or any single named NDFI/sponsor credit event material enough to require 8-K disclosure.
3. The cost of equity really is ~10%, i.e. the market does not re-rate the corridor concentration. Falsification test: any tariff, capital-control, sanctions or deposit-flight shock that visibly widens the sector’s required return.
4. Excess capital is eventually deployed accretively — a large repurchase at a sane multiple, a special dividend, or a well-priced acquisition. Falsification test: a seventh consecutive year with CET1 at or above 14%, no buyback acceleration and no deal.
5. The efficiency advantage is structural, not presentational — that the restated ~40.8% still beats a ~60% peer median by enough to justify the premium. Falsification test: the gap to peer median narrowing below ~15 points on a like-for-like basis.

14.2 For the bear case

1. Credit normalises beyond management’s guidance, with classified-asset migration continuing rather than reversing. Falsification test: net charge-offs staying at or below 20bps through FY2027 with non-performing assets below ~40bps, and classified loans falling back below 1.4% of loans.
2. The sector multiple mean-reverts from the 97th–99th percentile toward historical medians. Falsification test: the cohort holding above ~2x tangible book through a full year including a credit-normalisation quarter.
3. Operating leverage stays negative, with the guided expense build proving permanent rather than investment-phase. Falsification test: the efficiency ratio holding at or below 38% through a full year of the guided 7–9% expense growth.
4. The NDFI mix shift is a warning rather than a diversification success. Falsification test: the capital-call share of NDFI stabilising or rising while total NDFI growth moderates toward total loan growth, with charge-offs still near zero.
5. The insider distribution is informed rather than incidental. Falsification test: renewed open-market insider buying, or a formal 10b5-1 plan disclosed covering the 2026 sales.

The honest summary: the bull and the bear are arguing about the same five numbers, and Q2-2026 — reporting three days after this report’s date — will move at least three of them.

15. Source Appendix

Full source detail, with URLs and access dates, is provided in Appendix B — Source Appendix below. Primary reliance was on EWBC’s SEC filing corpus for the trailing 60 months (5 × 10-K, 15 × 10-Q, 42 × 8-K, 5 × DEF 14A and 176 Form 4s, all mirrored locally), the Q1-2026 earnings-call transcript, the March 2026 federal banking-agency capital proposals, and prior published peer research on eleven peer banks. All quantitative figures were reconciled to the filings; where third-party aggregators disagreed with the filings, the filings governed and the discrepancy is documented.

Sections 1–15 contain no investment recommendation and no price target. The Claude's Take block at the head of this article is a labelled exception and represents the author’s own independent opinion. This is general information, not investment advice.

APPENDIX A — Standard Diligence Questionnaire

East West Bancorp, Inc. (NASDAQ: EWBC) · 18 July 2026

A standard diligence questionnaire applied to East West Bancorp. Answers are labelled Fact / Interpretation / Assumption where it matters. Where a question does not map to a bank’s business model, the correct sector analog is given rather than a forced answer.

General

What thoughtful questions have other investors asked about this company?

The most sophisticated question in the file belongs to Li Lu’s Himalaya Capital, which initiated an EWBC position in Q1-2023 — into the SVB panic — at a cost of ~$183m (roughly $65 a share on average), added ~21% in Q2-2023, and has held unchanged since. As of 31 March 2026 the position was ~2.8m shares, 1.99% of shares outstanding and ~9.26% of Himalaya’s 13F portfolio. (Fact.) The implicit question he answered correctly was: is a 55%-uninsured California commercial bank with a concentrated ethnic depositor base a funding-run candidate, or is that base unusually loyal? The deposits did not leave, and the stock has more than doubled. (Interpretation: this is powerful evidence about the durability of the franchise — and no evidence at all about today’s price. Li Lu underwrote it at roughly 0.9x tangible book; the question this report addresses is what it is worth at 2.19x. An owner’s cost basis is not a valuation argument.)

Recurring questions from the sell side on recent calls, and the honest answers:

• “Why is loan-growth guidance only 5–7% when you just grew 8% annualised?” (Casey Haire, Autonomous). Management pointed to a 7.0% year-over-year print. The unasked follow-up is better: how much of that growth is non-bank financial lending, and what is core C&I doing? Answer: NDFI supplied half of FY2025 loan growth and core C&I ex-NDFI contracted ~$350m.
• “What are the parameters around earn-back and tangible-book dilution if you do a deal?” (Haire). Management deflected to “organic growth is our #1 priority.” Given ~$2.4–3.2bn of surplus capital and a record-high acquisition currency, this deserved a real answer and did not get one.
• “Where are you in the ACL build?” (Timur Braziler, UBS). Management noted a two-basis-point build and that the downside scenario “did change quite substantially” from year-end. The more informative disclosure — classified loans rising 15% in the quarter — was not raised by anyone on the call.
• “Is the drawdown on capital-call lines stressed?” (Ebrahim Poonawala, BofA). Management said no, and that a third would repay in Q2. Reasonable. But nobody asked what the other 70% of the NDFI book is.

The question we would ask that nobody has: Your NDFI portfolio grew from $6.0bn to $8.3bn in five quarters. Capital-call lines within it grew 6%. What is the other $5.9bn, who are the borrowers, what is the average spread, and how would it behave if a large sponsor’s portfolio companies began defaulting?

Cyclicality & Earnings Nature

Are earnings at a cyclical high or low? A cyclical high. (Interpretation, strongly supported.) Four separate inputs are peaking simultaneously: (i) net interest margin at 3.49%, expanded entirely from the liability side, with the CD repricing tailwind ~one year from exhaustion as the book converges on the 3.60% special rate; (ii) an efficiency ratio of 35.69% that is ~40.8% restated on the pre-2024 presentation and that management guides to deteriorate; (iii) fee income at a record, 81% of the increase from rate-dependent annuity and fixed-rate-bond distribution; and (iv) net charge-offs at 9bps against management’s own guidance of 15–25bps. Against a US banking industry that has averaged ~0.75% ROA since 1935 and exceeded 1% in only fourteen years, EWBC’s 1.70% is a ~2.3x outlier — real, but not a normal-state number.

Driven by the external environment or internal actions? Both, but the marginal driver in 2026 is external. (Fact + Interpretation.) The CFO stated that the FY2026 NII guidance raise from 5–7% to 6–8% was attributable “exclusively to the change in the rate outlook.” That is a macro windfall booked as performance. The genuinely internal achievements are the expense structure, the organic deposit gathering, and eight consecutive years of ≥15% ROE.

How stable are revenues? Reasonably stable and structurally recurring: ~87% of revenue is net interest income from a diversified, granular, mostly-variable-rate loan book (58% variable), with ~13% fee income. Spread revenue reprices with rates but does not disappear. The instability is in margin, not in volume.

Outlook for products/services? Core commercial lending and deposit gathering will grow with the Chinese-American commercial economy and EWBC’s share of it. Wealth management is growing fastest but from a transactional base. The risk is not demand; it is that incremental growth is being sourced from a commoditised adjacency.

How big will this market be — growing, shrinking, domestic or international? Predominantly domestic: international branch deposits are 5.8% of the total and foreign-exchange income 2.0% of revenue. The addressable market is the US Chinese-American and Asian-American commercial banking niche plus adjacent commercial lending — growing modestly faster than US GDP on demographic tailwinds, but finite and geographically bounded, which is precisely why EWBC operates 96 branches and has made no acquisitions in six years.

Business Quality & Competitive Moat

Is the industry getting more or less competitive? More. The FY2025 10-K’s own competition disclosure names commercial banks, savings institutions, finance companies, brokerages, insurers, credit unions, mortgage banks, non-bank financial institutions and — newly — stablecoins following the GENIUS Act of July 2025, as competition for “payments and liquidity management,” i.e. the cheapest funding EWBC has. It concedes competition “may put pressure on the pricing for our products and services.” Consolidation is running at a seven-year high but is removing cost capacity, not competitive capacity.

How profitable is the business (ROIC, ROE)? ROIC is not a meaningful metric for a bank — the correct analogs are ROA, ROE and ROTCE. FY2025: ROA 1.70%, ROAE 16.01%, ROATCE 16.99%. ROAE by year: 11.17% (2020), 15.70%, 19.51%, 17.91%, 15.93%, 16.01% (2025). Never below ~15% in eight years including COVID. (Fact.)

How profitable is the industry — how many competitors, what barriers to entry? The industry is structurally poor: ~0.75% long-run ROA, ~10% long-run ROCE, thousands of chartered competitors, and a commodity product funded by a commodity. Cohort returns cluster in a narrow 12–17% ROTCE band at 1.3–2.6x tangible book — a narrow band is itself the signature of weak differentiation. Barriers to entry (charter, regulatory capital, a granular deposit franchise) are real but protect the deposit franchise, not the lending franchise — and NDFI lending, where EWBC is growing, is a national market contested by every large bank and the private-credit funds themselves.

Can the business be easily understood? Mostly yes — it is a plain-vanilla commercial bank with unusually good cost control. Two parts are not easily understood and are not adequately disclosed: the $5.93bn non-capital-call NDFI book, and the true economics of the China subsidiary.

Can it be undermined by foreign low-cost labour? No. Deposit-taking and relationship lending are licensed, local and regulated. The relevant analogous threats are technological (digital banks, stablecoins, fintech deposit competition) and regulatory/geopolitical (US–China corridor restrictions), not labour arbitrage.

Do brands matter? Moderately, and in a specific way. “East West Bank” carries genuine standing within Chinese-American business communities — but the Cathay General comparison proves that brand-plus-affinity alone yields 11.25% ROE, not 17%. Brand is table stakes; scale within the bounded niche is the moat. (Interpretation.)

What is the nature of competition? Price competition on deposits (intense, and EWBC pays up — 2.46% cost of deposits, 38% time deposits); relationship and service competition on commercial lending, where EWBC’s language, cultural and cross-border capability genuinely differentiate in a middle band of customers too small for the megabanks and too complex for community banks; and pure price competition in NDFI/capital-call lending, where EWBC has no differentiation.

Customers’ switching costs? Genuine but modest for depositors (direct deposit, payments, cash-management integration) and higher for commercial borrowers with revolving facilities, treasury services and multi-year relationships — particularly where the relationship depends on Mandarin- or Cantonese-language service and cross-border capability a competitor cannot staff. The tell that switching costs are not extreme: EWBC has to pay above-peer deposit rates to hold the money.

Financial Condition & Balance Sheet

Assets not fully recognised on the balance sheet? The East West Bank (China) licence — described in the 10-K as making the bank “unique among U.S.-based regional banks” — carries no balance-sheet value but is a genuine, currently unobtainable regulatory asset. The deposit franchise itself (the ability to gather $68.9bn at 2.46%) is unrecognised. The $465.7m of goodwill from the 2009 UCB deal is carried at cost and is almost certainly worth far more.

Off-balance-sheet liabilities? $27.7bn of C&I commitments at 67% utilisation (i.e. ~$9.1bn undrawn) plus unfunded NDFI/capital-call commitments — a genuine contingent draw risk in a stress, and the reason the $47.0m unfunded-commitment reserve exists. Also $337.5m of unfunded tax-credit investment commitments, and standard letters of credit and derivatives. The Rayliant earnout is carried as a derivative liability (~$15m). Nothing exotic or hidden. (Fact.)

How conservative is the accounting? Mixed, and this is the memo’s central quality-of-earnings finding. Conservative in substance: CRE loan-to-values at 49% blended, an allowance at 1.51% of loans (6–10x guided losses), reserve coverage of non-accruals at 463%, and — to its credit — an efficiency ratio computed on GAAP expense including the $74.8m of tax-credit amortisation many peers route through the tax line. Less conservative in presentation: the ASU 2023-02 adoption moved ~$90m a year of amortisation out of operating expense into the tax line, improving the reported efficiency ratio by ~511bps without any operating improvement, and EWBC discontinued the adjusted-efficiency-ratio disclosure it had published in FY2022 and FY2023 — precisely the metric that would have made the change visible. Not a misstatement; the reclass is disclosed and GAAP-compliant. But the resulting headline is not comparable to its own history, and the company stopped providing the bridge.

How CapEx-hungry is the business? The bank analog is premises, technology and regulatory investment. Very light: occupancy and equipment is $66.1m, 2.3% of revenue and 0.08% of assets, with net fixed assets of only $207.7m on an $80bn balance sheet. The real “capital intensity” of a bank is regulatory capital, and here EWBC is the opposite of hungry — it is over-supplied at 15.1% CET1 with ~$2.4bn of surplus. The forward capital call is the Category IV compliance and data build arriving ~2027–29.

Capital Allocation & Management

How much free cash flow does the business generate, and how is it used? Free cash flow is not meaningful for a bank; the analog is earnings less the capital required to support balance-sheet growth. EWBC earned $1,325m in FY2025 and needed roughly $450m of CET1 to support 5–7% loan growth — capital generation is running at roughly 3x capital consumption, or ~11.8% a year against 5–7% asset growth. Uses: ~$334m of dividends (rising to ~$440m), ~$115m of buybacks in 2025 and $99m in Q1-2026, and the residual accumulates. (Fact + Interpretation.)

Management’s philosophy? Stated on the Q1-2026 call: organic growth first, then dividend, then “inorganic at the right price,” then opportunistic buybacks — repurchases ranked last. (Fact.) In practice this is a policy of accumulation: CET1 has run 14–15% for years, and the ROE decline from 24.5% (2018) to 16.01% (2025) reflects capital piling up faster than it is deployed rather than deteriorating profitability.

Significant acquisitions recently? None. Goodwill has been $465.697m in every balance sheet from 31 December 2020 through 31 March 2026 — six years without an acquisition. The only deployment is a 49.99% interest in Rayliant Global Advisors (Q3-2023, $95m plus performance RSUs); the 49.99% structure is deliberately one basis point below control, avoiding consolidation and goodwill. The franchise-defining transaction remains the 2009 FDIC-assisted acquisition of United Commercial Bank under an 80%/95% shared-loss agreement — an excellent deal, but a government-subsidised distressed purchase seventeen years ago. Nobody currently in management has integrated a bank acquisition.

Buying back shares? Yes, and the record is instructive. Cumulative: 11.46m shares for $686.5m at a $59.92 blended average. The March 2020 purchase at $32.64 was outstanding and accounts for 53% of all cumulative gains. But EWBC repurchased zero shares in Q1, Q2 and Q3 of 2023 — with $254m of authorisation already unused, while the stock traded to $30.90, roughly 0.7x tangible book — and resumed only in Q4 at $54.56, after a ~76% rally. The pace has since inverted into buying high, culminating in the largest quarterly outlay in company history in Q1-2026 at ~2.2x tangible book, where each dollar spent retires only ~$0.45 of tangible equity. (Fact.)

Issuing large amounts of new shares to insiders? No. Stock compensation was $76.2m in 2025, but ~$31m of that is a one-time recognition change for retirement-eligible employees adopted in Q3-2025 — normalised SBC is ~$45m, essentially flat with 2024, or ~3.4% of net income and ~0.25% of market capitalisation. Share count fell from 145.6m (2019) to 137.0m (Q1-2026), a 5.5% reduction. Dilution is not a concern.

Compensation policy of directors/management? Dominic Ng’s FY2025 total compensation was $9,683,884 (~3.2x the next NEO; 83:1 pay ratio); “compensation actually paid,” marking unvested equity to market, was $18,512,734. Non-employee directors receive an annual grant (1,156 shares, ~$145k, in May 2026). The design flaw is material: the annual bonus contains no ROE, ROTCE, ROA or efficiency metric — it is 30% adjusted EPS, 30% pre-provision pre-tax income, 20% average loan growth, 10% criticized ratio, 10% charge-off ratio, plus 35% qualitative. A CEO holding ~$2.4bn of excess capital is paid 80% on growth and volume and 0% on returns. FY2025 targets for adjusted EPS and PPNR were set, in the proxy’s own words, “the same as 2024 results” — zero growth required for a 100% payout; the plan paid 172% corporate / 181% financial. The long-term plan is better (ROA 37.5% / ROE 37.5% / relative TSR 25% against the KBW index) but is measured as three separate one-year periods, and the 2025 award paid 169.3% with total shareholder return at the 41st percentile — below median.

Motivations of management? (Interpretation, evidence-based.) Ng has run this bank for 34 years and clearly identifies with its independence, its community standing and its fortress balance sheet. That has produced genuine, durable outperformance and an admirably boring filing record. But the incentive structure rewards size and safety rather than returns on capital, and his personal economics point in a direction worth noting: he owns under 1% (924,071 shares) after 35 years, has cut direct holdings 30% since March 2024, sold 260,000 shares for $28.36m across 2025–26, gifted a further 190,730 shares into trusts in March 2026, and his 2026 sales are discretionary, not under a 10b5-1 plan.

Valuation & Market Data

Is the stock an ADR, MLP, or K-1 issuer? No. EWBC is a US-domiciled Delaware bank holding company filing 10-Ks with the SEC, listed on NASDAQ, issuing a standard Form 1099-DIV. No K-1, no ADR, no MLP structure, no preferred stock outstanding (total equity is common equity).

Dividend policy? Quarterly, raised annually, with a deliberately conservative payout. The rate has gone $0.40 (Q1-2022) → $0.48 → $0.55 → $0.60 → $0.80 (Q1-2026, +33%); annual dividends per share $1.92 / $2.20 / $2.40 for 2023–25 against a 2026 run-rate of $3.20. FY2025 payout 25.2%, forward yield ~2.38%. Uninterrupted growth through COVID and the 2023 regional-bank crisis. This is the cleanest positive in the capital-allocation record.

How profitable is the business? See above — ROA 1.70%, ROAE 16.01%, ROATCE 16.99%, efficiency ratio 35.69% reported (~40.8% restated). Best-in-cohort on every one of these except the restated efficiency ratio, where it is still comfortably first.

Is net income diverging from cash from operations? The bank analog is whether earnings quality is supported by pre-provision pre-tax income and whether reserves are being released to flatter results. No material divergence. FY2025 cash flow per share was $10.85 against diluted EPS of $9.52 — operating cash flow exceeds earnings, as is normal for a bank with growing non-cash provisions. Reserves were built, not released (ALLL 1.31% → 1.42% → 1.44%), so earnings are not being flattered by reserve releases. Net one-time items across the five-year set net to ~+$23m, about $0.12 a share. The earnings level is clean; the distortions identified in the memo are in geography (the tax-credit reclass) and in the margin narrative (6bps of the 14bp FY2025 NIM expansion was one-time payoff accretion), not in the bottom line.

Risks & Downside

What factors would cause the stock to decline? In descending order of likelihood: (1) a sector-wide multiple de-rating — the cohort sits at the 97th–99th percentile of its own historical price-to-book range, and a reversion to a ~1.6x own-history median costs ~27% with the franchise entirely intact; (2) credit normalisation in the NDFI/private-credit book, now $8.3bn, ~93% of common equity, where the non-capital-call residual grew 57% in five quarters; (3) continued classified-asset migration after a 15% quarterly rise; (4) margin compression as the CD repricing tailwind exhausts; (5) negative operating leverage on guided expenses of +7–9% against ~+7% revenue; (6) a rate reversal, since the entire FY2026 guidance raise was a rate assumption; and (7) a US–China corridor shock, which the factor model suggests is not priced at all (Country: China loading +0.096).

Risk of a catastrophic loss? Low, but not zero, and the mechanism is well-identified: a funding run. Uninsured deposits are 55.4% of the total (45% on management’s adjusted domestic basis), and $7.31bn of uninsured time deposits mature within three months. This risk is not theoretical — it was realised in March 2023, when the stock fell 56% intraday in three weeks, and the lifetime maximum drawdown of −92% in 2008–09 shows what this business model can do in a true credit event. The defences today are substantially stronger: 1.29x liquidity coverage of adjusted uninsured deposits (~$37bn of sources against $28.8bn at risk), an 84% loan-to-deposit ratio, 15.1% CET1, a 49% blended CRE loan-to-value, and a held-to-maturity mark worth only 3.4% of tangible book.

Chance of a total loss? Very low. It would require simultaneous catastrophic credit losses and a funding run. To wipe out $8.4bn of tangible common equity, losses would need to approach ~14% of loans — roughly 15x the worst annual charge-off rate of the last five years and several multiples of anything experienced in 2008–09 by this institution at its current capitalisation. The realistic severe case is not insolvency but a 30–40% drawdown from multiple compression plus a credit cycle, as modelled in the memo’s bear scenario.

Recent News & Events

Has the business environment changed recently? Yes, in four dated respects. (i) The Federal Reserve, OCC and FDIC re-proposed the US capital framework on 19 March 2026, formally rescinding the 2023 Basel III Endgame proposal; comments closed 18 June, finalisation is expected in Q4-2026, and management estimates ~$7bn of RWA relief and +1.6–1.8pp of capital ratios. (ii) The FOMC held at 3.50–3.75% on 17 June 2026 and removed the prior 2026 cut from its projections, with futures now pricing ~4% at year-end — the direct cause of EWBC’s raised NII guidance. (iii) The US–China tariff architecture is mid-reconstruction: after the Supreme Court struck down the IEEPA tariffs on 20 February 2026, the replacement 10% Section 122 tariff expires by statute on 24 July 2026, with a USTR Section 301 deadline on 20 July and proposed 12.5% duties across 46 countries including China. (iv) Private-credit conditions are deteriorating industry-wide — default rates 8.1% → 9.2%, “bad PIK” at 6.4% of Q1-2026 volume, and the First Brands/Tricolor failures.

Significant acquisitions? None (see above).

Change in accounting policies? Yes — two, both material to reported metrics and both flagged in the memo. (i) ASU 2023-02, adopted 1 January 2024, applying proportional amortisation to new-markets, historic, production and renewable-energy tax credits — moving ~$89–90m a year of amortisation from noninterest expense into income tax expense. This improved the reported efficiency ratio by ~511bps and raised the headline effective tax rate, with no effect on net income; EWBC also discontinued its adjusted-efficiency-ratio disclosure in the same year. (ii) A Q3-2025 change in equity-award expense recognition for retirement-eligible employees, adding ~$31m of one-time compensation expense. Separately, a 2025 California single-sales-factor deferred-tax-asset revaluation affected the tax line.

Recent changes — new markets, facilities, management? Management has signalled interest in East Coast expansion into Chinese-American population centres (no transaction announced) and is following clients’ supply-chain relocation into Southeast Asia and Mexico. On leadership: Christopher Del Moral-Niles joined as CFO in October 2023 with Irene Oh moving to Chief Risk Officer in a “planned executive succession process”; Deborah Leerhsen joined as Head of Global Banking in December 2024; director Rudolph I. Estrada retired in March 2025; Peter Babej joined the board in December 2025, taking it to eleven. Governance changed in two ways in 2026: the absolute prohibition on NEO share pledging was replaced with a case-by-case exception regime, and say-on-pay support fell to 95.56%, the lowest in four years.

The single most important near-term event: EWBC reports Q2-2026 results on 21 July 2026 — three days after this report’s date. The two lines to read first are the NDFI balance and its capital-call share, and the classified-loan and net-charge-off trajectory.

APPENDIX B — Source Appendix

East West Bancorp, Inc. (NASDAQ: EWBC) · 18 July 2026 All sources accessed 18 July 2026 unless otherwise stated. SEC CIK 0001069157.

B.1 Primary — SEC filing corpus (trailing 60 months)

The full corpus since 18 July 2021 was enumerated and mirrored locally to output/EWBC/sources/ — 84 saved documents plus a complete insider-filing index, comprising 5 × 10-K, 15 × 10-Q, 42 × 8-K, 5 × DEF 14A, 5 × 11-K, 4 × S-8, ARS and Schedule 13G/13D filings, with 176 Form 4 filings retrieved and parsed separately (252 discrete transactions). Structured-note noise (424B*, FWP, 144) was excluded.

Corpus assembled with scripts/edgar.sh since EWBC 2021-07-18 and scripts/fetch_sources.sh EWBC 2021-07-18.

B.2 Primary — management commentary

B.3 Regulatory and policy sources

B.4 Market, factor and pricing data

B.5 Industry and market-condition sources

B.6 Comparative and framework sources

B.7 Data-quality notes and reconciliations

Every material figure was reconciled to the filings. Where third-party data disagreed with a filing, the filing governed. Four defects are documented because they materially affect the analysis:

1. ROIC.ai per-share book values are wrong for EWBC — not used. ROIC reports FY2025 book_val_per_sh of $60.01 and tang_book_val_per_sh of $60.96 — tangible book above stated book, arithmetically impossible against $465.7m of goodwill, with the same inversion in 2018–2021 and 2023. The stated figure is also understated: equity of $8,899.2m ÷ 137.579m shares = $64.68. All book-value work in this report is derived from the balance sheet and independently corroborated by AZI’s $64.78.
2. ROIC.ai’s return-on-common-equity field is unreliable for banks — not used. It returns HBAN 36.0%, GBCI 21.3% and WBS 23.9%, far above those banks’ own reported figures. All peer ROE/ROTCE figures in the valuation comparison were recomputed by hand from FY2025 balance-sheet line items on two-point average equity.
3. “ROE” definitional conflict, resolved. EWBC’s frequently-quoted ~17% return is return on average tangible common equity (16.99%). Return on average common equity is 16.01%, independently confirmed by the company’s own Pay-versus-Performance table in the 2026 proxy. Both are used in this report, labelled. Separately, the reported 16.98% could not be reproduced from two-point average equity (which yields 15.9%), implying a quarterly-average convention; the resulting sustainable-ROTCE band of 15.7–18.0% is disclosed in the valuation section rather than papered over.
4. Executive-compensation figure, resolved. Two figures circulate for Dominic Ng’s FY2025 compensation — $9,683,884 and $8,358,550. The proxy’s Summary Compensation Table settles it: $9,683,884 (2025), $9,213,633 (2024), $8,358,550 (2023) — the lower figure is the 2023 row.

Additionally noted: management’s claim of “15 straight years of deposit growth” does not appear in the FY2025 10-K and could be verified from primary sources only for the most recent two years; it is attributed to management throughout rather than stated as fact. Management’s “111bps” deposit-cost decline was verified at 109bps (3.93% in Q3-2024 to 2.84% in Q1-2026) and is accurate.