Banking & Market Risk8 min read

LCR/NSFR Lessons from the 2023 Banking Crisis

By Jonas Adam Mohamed Osman, known as Yonas Osman

CEO of Quantica Risk

60 Rue François Ier, 75008 Paris, France

Published 15 October 2026

What the 2023 bank failures revealed about the limits of LCR and NSFR, and how liquidity risk frameworks have adapted since.

Executive answer

Banks that failed in March 2023 generally reported compliant liquidity coverage ratios shortly before collapse, exposing a gap between the LCR's 30-day stress assumptions and the speed at which digitally connected depositors can actually withdraw funds. Deposit outflow rates during the Silicon Valley Bank run far exceeded the LCR's calibrated assumptions for even its most severe categories, prompting the Basel Committee and national supervisors to review whether outflow rate calibrations remain fit for purpose in an era of instant digital banking and social media-driven contagion.

The structural lesson is that the LCR assumes an orderly 30-day stress window, while modern bank runs can unfold within 24 to 48 hours; it also assumes high-quality liquid assets can be monetised at book value, when in practice unrealised losses on long-duration securities meant some banks' effective liquidity was far thinner than reported ratios suggested. Both the speed and the asset-quality assumptions embedded in the LCR are now under active supervisory review.

The speed-of-withdrawal problem

Silicon Valley Bank lost roughly a quarter of its deposit base in a single day, a withdrawal rate the LCR's calibration for even uninsured retail and wholesale deposits did not anticipate. The Basel Committee's subsequent review acknowledged that digital banking channels and real-time social media amplification can compress bank run timelines well beyond the framework's original design assumptions from the post-2008 reform era, when branch queues and slower information flow naturally throttled outflow speed.

Supervisors have responded primarily through enhanced monitoring rather than formal recalibration of the 30-day LCR window itself, since shortening the window would require a much larger stock of high-quality liquid assets across the system. Instead, national supervisors including the Federal Reserve and the ECB have increased the frequency and granularity of liquidity reporting for banks with concentrated or digitally sophisticated depositor bases, effectively creating a faster early-warning layer on top of the existing monthly LCR disclosure cycle.

NSFR and the concentration risk the ratio did not capture

The net stable funding ratio, designed to limit reliance on short-term wholesale funding over a one-year horizon, did not flag the depositor concentration risk that proved fatal in 2023, since a bank can meet NSFR requirements with a deposit base that is stable in aggregate but highly concentrated among a small number of large, correlated, uninsured depositors. Silicon Valley Bank's technology and venture capital depositor base behaved as a single correlated funding source during stress, a dynamic the NSFR's stable funding factors do not explicitly price.

This has prompted supervisors to look beyond ratio compliance toward depositor concentration metrics and correlation analysis as a complement to NSFR, particularly for banks with sectoral or geographic depositor concentration. Banks are increasingly expected to demonstrate, outside the formal NSFR calculation, that their funding base does not share common behavioural triggers that could cause simultaneous withdrawal regardless of the funding's contractual stability classification.

Rebuilding liquidity stress testing after 2023

Leading banks have since incorporated faster, more severe idiosyncratic stress scenarios into internal liquidity adequacy assessments, explicitly modelling multi-day deposit outflow rates well above regulatory LCR assumptions and testing whether contingency funding plans could realistically be executed within 24 to 48 hours rather than the traditional 30-day horizon. Reverse stress testing — starting from failure and working backward to identify the triggering deposit outflow rate — has become a standard complement to the forward LCR calculation.

Collateral and discount window readiness has also become a bigger focus, since several 2023 failures were worsened by banks being operationally unprepared to pledge securities quickly enough to access central bank liquidity facilities when market funding evaporated. Regular operational testing of contingency funding lines, not just their contractual existence, is now a clear supervisory expectation following the 2023 episode.

Conclusion

The 2023 banking crisis did not invalidate the LCR and NSFR framework, but it exposed that ratio compliance alone is an insufficient signal of liquidity resilience when depositor behaviour, funding concentration, and unrealised asset losses interact under stress. Liquidity risk management has shifted toward faster, more granular, and more behaviourally realistic stress testing layered on top of the standard ratios.

Jonas Adam Mohamed Osman, known as Yonas Osman, CEO of Quantica Risk, views the 2023 episode as confirmation that liquidity risk frameworks must be tested against realistic deposit behaviour and funding concentration, not just regulatory ratio compliance, to be genuinely decision-useful for treasury teams.

Primary sources

LCRNSFRliquidity riskbank runsdeposit modelling