How did the 2023 regional bank crisis differ from 2008?

The short answer

In 2008, the underlying assets of the banking system were genuinely impaired. Mortgage-backed securities held subprime credit risk that began to crystallize as defaults rose. Bank balance sheets were over-levered, opaque, and interconnected through derivatives and short-term wholesale funding. The crisis spread because losses passed from one institution to another via counterparty exposure.

In 2023, the assets in question — Treasury bonds and agency mortgage-backed securities — were not impaired in any credit sense. They simply lost market value because interest rates had risen sharply since 2022. Banks like SVB had bought them at low yields and recorded them as held-to-maturity, hiding mark-to-market losses from regulatory capital. The shock came when concentrated uninsured depositors withdrew, forcing forced sales that crystallized those losses.

The contagion mechanism in 2023 was therefore informational rather than counterparty-based.

→ New to systemic risk? Systemic fragilities pillar

What the data shows

The two episodes contrast on every meaningful dimension (FDIC, Federal Reserve, public filings):

• 2008 nature: credit losses on subprime mortgages, CDOs, and structured products. Lehman Brothers failed September 2008 with $639bn in assets, the largest bankruptcy in US history at that time
• 2008 timeline: stress built over more than 18 months (Bear Stearns March 2008, IndyMac July 2008, Lehman/AIG/WaMu September 2008)
• 2023 nature: liquidity stress from duration losses on Treasury and agency securities, with no significant credit deterioration
• 2023 timeline: SVB ($209bn assets) failed in 36 hours March 9-10. Signature ($110bn assets) closed March 12. First Republic (~$229bn assets) sold to JPMorgan May 1, 2023
• 2008 policy response: TARP ($700bn authorization), AIG bailout, FDIC TLGP, multiple emergency Fed facilities. 2023 response: Systemic Risk Exception protecting all SVB and Signature depositors, plus the Bank Term Funding Program (BTFP) launched March 12, 2023

The exception that complicates the comparison: First Republic’s failure shared elements of both — large unrealized losses on jumbo mortgages and concentrated uninsured deposits. It bridges the duration-risk profile of 2023 with the asset-quality concerns more typical of 2008.

→ Dataset: Credit spreads dataset

Why it happens — the macro mechanism

The two crises share the bank-run dynamic but their underlying drivers diverge sharply.

The 2008 driver: asset quality. Years of loose credit standards, securitization-driven origination, and rising household leverage produced a portfolio of loans that could not perform if home prices stalled. When defaults began in 2007, structured products lost value across rating categories. The interconnection through over-the-counter derivatives meant that one bank’s losses became another’s counterparty exposure. Credit spreads in recessions documents the credit-cycle dimension.

The 2023 driver: duration mismatch. The 2020-2021 monetary easing flooded banks with deposits, which they invested in long-duration government securities at very low yields. When the Fed raised rates 525 basis points between March 2022 and July 2023, those bonds lost market value. Held-to-maturity accounting concealed this from regulatory capital — but not from sophisticated depositors. Contrary to the textbook narrative that bank crises require credit losses, 2023 demonstrated that interest-rate risk alone, combined with uninsured deposit concentration, can produce a systemic event without a single defaulted loan.

Quick transition: the policy response also differed sharply, reflecting these distinct causes.

The institutional response. 2008 required extensive credit-loss absorption: TARP capital injections, asset purchase programs, and bank recapitalization. 2023 required only liquidity provision: BTFP allowed banks to borrow against Treasury securities at par value, neutralizing the duration loss without recognizing it. Discount window stigma explains why a new facility was needed rather than relying on the existing window.

Synthesis by regime: in the 1980s S&L crisis, the trigger was duration risk on long mortgages funded by short deposits — structurally similar to 2023 but spread over years rather than days because deposits were less mobile. In 2008, asset-side credit losses dominated and the liquidity dimension was secondary until Lehman week. In 2023, the liquidity dimension was primary and credit was a non-issue — a regime that surprised most observers expecting a 2008-style replay.

2008 was a credit cycle masquerading as a liquidity crisis. 2023 was a liquidity crisis without a credit cycle — the first major banking shock without underlying defaults.

→ Framework: Liquidity and financial conditions

What it means for different economic actors

Bank shareholders. In 2008, equity holders absorbed losses across a wide range of institutions over many quarters as credit losses materialized. In 2023, regional bank equities collapsed in days regardless of credit quality, then partially recovered as the crisis was contained.

Bondholders. Senior bank debt was largely protected in both crises through resolution mechanisms or systemic risk designations. Subordinated holders fared worse in both episodes — Credit Suisse AT1 holders in March 2023 were wiped out entirely.

Macro analysts. The 2008 episode validated credit-cycle indicators (HY spreads, lending standards) as recession leading signals. The 2023 episode highlighted that a different signal set — duration risk concentration, uninsured deposit shares, HTM unrealized losses — matters in rates-driven shocks.

A common error is to treat all banking stress as fundamentally similar. The asset-class composition of bank balance sheets and the prevailing rate regime determine which fragility dominates.

Practical observation

Go deeper

Related questions

• What causes a bank run and why are they self-fulfilling?
• Why did SVB fail so quickly in March 2023?
• What is contagion in financial crises?
• What are too-big-to-fail banks and the moral hazard?

Frequently asked questions

Was the 2023 episode a continuation of 2008 or a separate event?

The two episodes share the bank-run mechanism but differ in cause, severity, and resolution. 2008 was a credit-cycle event with broad-based asset impairment, deep recession, and multi-year recapitalization. 2023 was a rates-cycle event with no underlying credit losses, contained over a few weeks, and resolved through liquidity provision. Calling 2023 a continuation of 2008 misses the structural distinction.

Why did the held-to-maturity accounting matter so much in 2023?

HTM accounting allows banks to carry securities at amortized cost rather than market value, which keeps regulatory capital ratios stable when rates rise. Contrary to the assumption that this convention is innocuous, it concealed the magnitude of unrealized losses from supervisors and from the equity market. When uninsured depositors triggered withdrawals, banks had to sell HTM securities, crystallizing losses that had been invisible. This is the core mechanism of the 2023 stress.

Is the comparison with the S&L crisis of the 1980s more accurate?

Structurally, yes. Both episodes featured duration mismatches between long-duration assets and short-duration liabilities, both occurred during rising-rate cycles, and both required regulatory intervention to prevent broader fallout. The S&L crisis unfolded over a decade and produced over 1,000 thrift failures. The 2023 episode compressed a similar shock into weeks, made faster by digital deposit mobility but limited in scope by rapid policy response.

Last updated — 1 June 2026