First Republic Bank: relationship banking after the funding model broke

The appeal and the vulnerability of the relationship model

First Republic built a franchise around affluent customers, personalized service, residential lending and wealth relationships. A competitively priced mortgage could help attract a household’s deposits and broader financial business. The model depended on the combined relationship economics: a modest loan yield could work while funding remained inexpensive and stable.

The FDIC inspector general describes jumbo mortgages with initial interest-only periods, generally ten years, followed by amortization. Such loans represented 61% of the single-family residential portfolio at December 31, 2022. This was a cash-flow structure, not evidence that those borrowers were delinquent. The same report found the bank remained committed to its longstanding strategy as rates rose. [1]

The economic vulnerability was that the customer could retain a valuable low-rate mortgage while moving deposits to another institution. A banking relationship is not a contractual requirement that every product stay together. The bank could lose cheap funding without being able to reprice the asset or recover the associated principal on demand.

A balance sheet dominated by loans

The FDIC’s supervisory review reports year-end 2022 assets of $212.6 billion, loans of $166.9 billion, deposits of $176.4 billion and cash and balances due from depository institutions of $4.3 billion. Estimated uninsured deposits were $119.5 billion. At March 31, 2023, assets had increased to $232.9 billion while deposits had fallen to $104.5 billion. Growth in assets during a run was therefore not evidence that the deposit franchise had recovered. [2]

Original calculations from the rounded year-end values put loans at 78.5% of assets and 94.6% of deposits. Cash and balances due were only 2.0% of assets. These ratios describe the reported mix, not a complete liquidity coverage test: saleable securities, pledgeable collateral, encumbrances, borrowing agreements and timing also matter.

The structure differs from an institution holding much of its excess cash in marketable securities. A high-quality mortgage is not automatically a same-day cash asset. Potential buyers must price its coupon, remaining maturity, credit risk, servicing and funding. A loan can remain current yet command materially less than its recorded principal in a higher-rate market.

Scroll horizontally to see all columns.

Rate risk was embedded in the loan book

The FDIC’s May 2023 testimony notes that the amortized-cost basis of mortgage and other loans, net of allowances, exceeded their fair value by about $22 billion at December 31, 2022. The unrealized difference illustrated interest-rate sensitivity beyond securities portfolios. It was not a contemporaneous charge-off or a claim that borrowers would default on $22 billion of principal. [4]

Consider an explicitly hypothetical $100 fixed-rate loan yielding 3% funded by deposits costing 0.5%. Before operating costs, credit costs and capital, the simple annual spread is $2.50. If replacement funding costs 5% while the loan coupon stays fixed, the spread becomes negative $2.00. That $4.50 change occurs without a missed borrower payment. It is an illustration of funding repricing, not First Republic’s actual marginal rates.

A fair-value shortfall and a negative current spread are related but not identical. The former capitalizes expected future cash-flow economics using market assumptions. The latter describes income over a period. Selling assets can recognize losses and shrink future funding needs; keeping them can avoid an immediate sale while extending exposure to expensive funding. Either route can constrain a private rescue.

March: the denominator behind the $100 billion headline

First Republic’s April 24 earnings release reported deposits of $173.5 billion on March 9 and $104.5 billion on March 31. The latter included $30 billion deposited by eleven major banks on March 16. The issuer said activity stabilized in late March and reported $102.7 billion of deposits on April 21. Those were management’s dated descriptions, not evidence that stability would persist. [3]

Using the rounded quarter-end values, the reported net decline was $176.4 billion minus $104.5 billion, or $71.9 billion, equal to 40.8% of December 31 deposits. Removing the new $30 billion support gives a March 31 residual of $74.5 billion and a $101.9 billion decline, or 57.8%, against the same year-end base. These are net changes in deposit stocks, not a sum of gross withdrawal transactions. [2, 3]

Against the March 9 base instead, $173.5 billion minus $74.5 billion equals $99.0 billion, or 57.1%. Different start dates explain why superficially similar “run” numbers differ. It would be wrong to label $104.5 billion both the remaining deposit balance and the amount withdrawn.

The bank-support announcement identifies four $5 billion placements, two $2.5 billion placements and five $1 billion placements, totaling $30 billion. They were explicitly uninsured deposits, not common equity and not a permanent government guarantee of the bank. The support supplied cash and confidence, but also created liabilities that had to be repaid or assumed. [5]

What a smaller uninsured percentage could conceal

The inspector general reports that uninsured deposits were approximately 68% of total deposits at year-end 2022. After the initial run, the ratio excluding the consortium’s $30 billion fell to about 28%. That improvement partly reflected the departure of uninsured funds. It did not establish that the surviving bank had repaired its asset-liability economics. [1]

Insurance shares are fractions with two moving parts. If high-balance customers leave and replacement official borrowing rises, an institution may show a lower uninsured-deposit share while depending more heavily on expensive wholesale funding. Conversely, adding uninsured rescue deposits may raise the reported share while temporarily improving cash available for withdrawals.

The funding question is therefore broader than one percentage: what replaced the deposits, at what cost, for how long, and against which collateral? A less runnable residual deposit base can coexist with a business model whose earnings no longer cover its funding expense. This explains why the passage from emergency liquidity to durable recovery was difficult.

The reprieve and the renewed loss of confidence

The April earnings release made the scale of the deposit loss public and described efforts to strengthen the balance sheet. The FDIC’s September review identifies the loss of market and depositor confidence after the March failures as the primary cause, with reliance on uninsured deposits and interest-rate exposure central vulnerabilities. First Republic failed on May 1 after weeks of attempted stabilization. [3, 6]

That sequence makes First Republic different from SVB’s compressed final two days. Time and external liquidity were available, yet the market still needed a convincing answer to who would fund the assets and absorb the economic shortfall. An industry deposit package can slow a run without generating the capital or sustained earnings necessary for independence.

A private buyer assessing the bank before receivership would have had to consider the whole liability structure and recognition of asset marks. A receiver-arranged sale could instead specify assets, deposits, financing and loss sharing. The eventual transaction’s feasibility does not prove that an equivalent stand-alone private rescue was readily available earlier.

Supervision: a strong franchise was not a complete risk assessment

The FDIC’s internal review found opportunities for a more forward-looking supervisory approach and more forceful challenge to the bank’s assumptions. Its account did not reduce the failure to a sudden deterioration in borrowers’ credit quality. It focused on the interaction of the business model, funding confidence and interest rates. [6]

The inspector general separately identified missed opportunities for earlier supervisory action and rating downgrades. The case demonstrates a recurring tension: favorable earnings, long customer relationships and low historical credit losses can coexist with growing sensitivity to funding and rates. Historical franchise quality was relevant information, but not sufficient evidence that deposits would remain during a panic. [1]

GAO’s September 2026 report adds that the three failed banks’ 2021 and 2022 disclosures described risk thresholds without disclosing breaches or their remediation. For First Republic and Signature, GAO also highlights the disclosure-review structure for publicly traded banks without holding companies. That is a finding about transparency and oversight, not a court determination of securities-law liability. [10]

May 1: receivership and a purchase-and-assumption transaction

California closed First Republic on May 1 and appointed the FDIC receiver. JPMorgan Chase Bank assumed all deposits and substantially all assets in a transaction the FDIC described as the least-cost resolution. No systemic-risk determination was made for First Republic. Full transfer of deposits therefore had a different legal route from the emergency protection used at SVB and Signature. [7, 4]

The FDIC announcement cited $229.1 billion of assets and $103.9 billion of deposits as of April 13. JPMorgan’s closing announcement used approximately $173 billion of acquired loans, $30 billion of securities and $92 billion of deposits. The figures reflect different dates and transaction scope. Combining the earlier FDIC deposit stock with the later acquired asset values would create a synthetic balance sheet that neither source reported. [7, 8]

JPMorgan explicitly did not assume First Republic’s corporate debt or preferred stock. Deposit customers moved to a continuing bank; common and preferred investors did not receive that same protection. Nonassumed obligations remained matters for the receivership and their legal priority, rather than claims automatically transferred to the buyer. [8]

How financing and loss sharing changed the deal

JPMorgan’s transaction presentation says it would pay $10.6 billion to the FDIC and receive $50 billion of five-year fixed-rate financing. The buyer would repay $25 billion of deposits from other large banks and eliminate its own $5 billion deposit on consolidation. Loan-loss sharing covered 80% of specified losses for seven years on single-family mortgages and five years on commercial loans, including commercial real estate. [9]

These terms solve different problems. Financing supplies funding; loss sharing allocates specified credit losses; the purchase price is consideration; elimination of an intragroup deposit is consolidation accounting. None is by itself the total public cost. The receiver’s retained liabilities, recoveries, financing income, expenses and future contractual losses all affect the eventual insurance-fund result.

The structure also explains why the acquiring institution’s economics need not match the old bank’s. A new acquisition basis, different financing and a larger franchise can make an acquired portfolio useful after the original funding model has failed. That is not proof that the old securities or deposits were misreported merely because the buyer recognizes an accounting gain.

Costs and creditor outcomes remain dated statements

The FDIC initially estimated a $13 billion insurance-fund cost on May 1. GAO’s September 2026 report cites an approximately $15.8 billion First Republic estimate as of December 31, 2025. These estimates were prepared at different stages; neither should be described as the final cash loss simply because it appeared in an official publication. [7, 10]

The FDIC’s bank-specific receivership materials distinguish deposit assumption from creditor claims. The normal priority structure means that a shareholder or unsecured creditor cannot infer a recovery from customers’ uninterrupted account access. This article does not assert a final distribution rate for every class or a termination of the receivership. [11]

First Republic also should not be confused with Republic First Bank, the separate Philadelphia institution that failed in April 2024. The similar names identify different banks, failures, buyers and estates. First Republic’s distinctive case is the failure of a loan-heavy relationship model when deposit portability exposed the cost of retaining long-duration assets.

A lasting distinction between loyalty and funding economics

The 2026 FDIC transaction-level study reinforces the role of large depositors and uninsured balances in run behavior. That evidence does not mean every affluent client ran or that all commercial accounts behave identically. It does demonstrate why historical loyalty cannot be assumed to offset a sudden change in perceived safety. [12]

The central finding is not that personalized banking or jumbo mortgages are inherently unsound. It is that those services generated a joint economic proposition which customers were free to unbundle. Once the low-cost funding left, the remaining assets and emergency liabilities had to work on their own terms. First Republic’s weeks-long struggle and the final structured acquisition show how difficult that transition became.

Sources

1. FDIC OIG, Material Loss Review of First Republic Bank, November 2023Official source · PDFBack to text: ↑1↑2↑3
2. FDIC, Supervision of First Republic Bank, September 8, 2023, financial-information tableOfficial source · PDFBack to text: ↑1↑2↑3↑4↑5↑6↑7
3. First Republic, first-quarter results, April 24, 2023, issuer releaseSourceBack to text: ↑1↑2↑3↑4
4. FDIC, Oversight of Financial Regulators testimony, May 17, 2023Official sourceBack to text: ↑1↑2
5. Eleven banks, $30bn uninsured deposit announcement, March 16, 2023SourceBack to text: ↑
6. FDIC, First Republic supervisory-review findings, September 8, 2023Official releaseBack to text: ↑1↑2
7. FDIC, JPMorgan assumption of First Republic, May 1, 2023Official releaseBack to text: ↑1↑2↑3
8. JPMorgan Chase, acquisition announcement, May 1, 2023SourceBack to text: ↑1↑2
9. JPMorgan Chase, acquisition presentation filed May 1, 2023SourceBack to text: ↑
10. GAO, Bank Financial Disclosures, GAO-26-107719, September 3, 2026Official sourceBack to text: ↑1↑2
11. FDIC, First Republic receivership informationOfficial sourceBack to text: ↑1↑2
12. FDIC, transaction-level depositor-flight research release, May 14, 2026Official releaseBack to text: ↑