Source: Bank for International Settlements
Good morning, everyone.
Introduction
The events of 2023 were a stark reminder of the evolving nature of financial risks. The digitalisation of finance and the influence of social media have amplified the speed and severity of bank runs, creating new challenges for regulators and institutions alike. In response, two key avenues have emerged in the debate on improving liquidity risk management.
First, there is the potential refinement and strengthening of liquidity requirements, particularly the Liquidity Coverage Ratio (LCR). Second, there is a renewed focus on ensuring banks’ operational readiness to access central bank liquidity support during periods of stress.
To date, these approaches have largely been pursued independently. However, I believe that integrating these two dimensions offers a more comprehensive framework for addressing liquidity risk. In doing that, there would be more chances to improve the control of liquidity risks without introducing overly restrictive regulatory requirements that could undermine commercial banks business models. Today, I will outline how that integration could take place, the challenges it entails and a potential framework to address them.
The limitations of current prudential regulation
Let us begin by examining the current regulatory framework for liquidity risk. In the aftermath of the Great Financial Crisis, liquidity requirements became a key component of the new regulatory standards, Basel III. In particular, the LCR was created with the purpose of ensuring that banks maintain a sufficient stock of high-quality liquid assets (HQLA) to withstand a severe liquidity stress scenario over a 30-day period.
The LCR has proven to be an effective tool in many respects. It asks banks to put in place a sort of self-insurance that reduces the likelihood that they will resort to drastic and potentially destabilising measures during periods of liquidity stress. It also gives banks and supervisors critical time to prepare for the orderly resolution of institutions that are no longer viable.
However, recent events have exposed limitations in the LCR calibration. During the 2023 turmoil, actual runoff rates far exceeded the assumptions underlying the LCR. For instance, Silicon Valley Bank experienced deposit outflows in a single day that surpassed what the LCR stress scenario assumes for an entire month.
Moreover, the definition of HQLA has come under scrutiny. Current eligibility criteria do not differentiate between instruments based on their accounting treatment. This raises questions about the practical availability of certain – theoretically liquid – assets during stress scenarios. In particular, as the sale of instruments held at amortised cost may generate solvency-weakening capital losses, the suitability of those assets to meet liquidity requirements can be questioned.
In the light of these challenges, some have called for more stringent LCR calibration, entailing higher assumed runoff rates of certain deposits and/or constraints in the eligibility of assets that are not measured at fair value in the calibration of LCR. While this response is understandable, it is important to recognise the limits of self-insurance. Excessively stringent requirements could impair banks’ ability to perform their core intermediation function, which, by definition, typically implies assuming a fair amount of liquidity risk.
The case of Silicon Valley Bank illustrates this dilemma. The bank faced deposit withdrawals amounting to 25% of its total deposits in a single day, with an additional 60% expected the following day. If banks were required to regularly hold sufficient liquid assets to fully cover such extreme scenarios, most would struggle to engage in any meaningful commercial activity1. At the same time, that approach would assume that banks can only resort to their own holding of liquid assets in stress situations, thereby ignoring any external source of liquidity support.
This brings us to a second component of the current policy framework: central bank liquidity facilities.
The role of central bank liquidity support
Central banks play a crucial role as lenders of last resort, providing liquidity support to solvent banks during periods of stress. But it is true that the availability of this support depends on the holdings of acceptable collateral which, for most central banks, include non-tradable assets, after imposing adequate haircuts.
For a typical commercial bank, runnable liabilities – such as uninsured deposits and short-term market funding – represent 30–50% of total unencumbered assets. This suggests that, even with significant haircuts, sound banks generally have sufficient assets that could in principle be used as collateral to secure emergency loans from central banks.
Yet accessing central bank liquidity support is not without challenges. The process of pledging collateral involves legal, operational and valuation complexities, particularly for non-traded assets. In severe liquidity stress scenarios, when time is of the essence, these challenges can become significant obstacles.
To address these issues, central banks must ensure that banks are operationally prepared to use their facilities. This includes requiring them to have the necessary arrangements in place to pledge collateral, along with regular testing and simulation exercises to ensure readiness.
An additional measure is the introduction of prepositioning requirements. Prepositioning involves banks providing central banks with detailed information about their collateral assets, along with all necessary documentation to assess eligibility, transferability and valuation. While many central banks encourage prepositioning, few mandate it.
Some proposals go further. For example, the “pawnbroker for all seasons” approach advocates that banks preposition sufficient collateral with the central bank to fully back their runnable liabilities.2 These liabilities would include all deposits and short-term market funding, with the collateral amount determined after applying conservative haircuts. In its original formulation, this proposal was presented as a possible substitute of key elements of the current regulatory, supervisory and deposit insurance frameworks. A more moderate alternative is proposed by the Group of Thirty, which recommends calibrating prepositioning requirements based on a narrower set of liabilities, excluding insured deposits.34
A tiered framework for liquidity controls:
As I mentioned before, the policy debate has thus far dealt with two issues in parallel: recalibrating banks’ existing liquidity requirements, and strengthening banks’ operational readiness to access central bank liquidity support during stress situations. However, these two debates should be more interconnected. Specifically, there appears to be a tension between making the stress scenario underlying the calibration of the LCR more severe while simultaneously ignoring the possibility that banks could obtain liquidity from central banks in such adverse scenarios.
Given the complementary roles of regulatory liquidity requirements and central bank liquidity support, in a recent Financial Stability Institute (FSI) paper5 we propose a framework that integrates these two dimensions. This framework introduces a tiered approach to asset eligibility, corresponding to different levels of liquidity stress.
In moderate stress scenarios, it seems reasonable to rely on self-insurance and require banks to hold sufficient HQLA to manage their needs without relying on central bank facilities. This is partly because using central bank liquidity support may carry a stigma.
However, as the severity of the stress increases the “anticipatory” stigma associated with central bank support becomes a less important consideration, while large-scale asset sales by banks could become even more destabilising for markets.
The criteria for asset eligibility under central bank liquidity facilities are generally less stringent than the HQLA requirements. For instance, non-tradable assets – such as bank loans – are often eligible as collateral for central bank lending. Central banks also tend to apply even more flexible collateral eligibility criteria for emergency liquidity assistance compared with that for their standing lending facilities.
This suggests a framework with three tiers of asset eligibility, corresponding to different levels of liquidity stress:
- Type 1 assets: HQLA, which banks are expected to hold to address moderate stress scenarios without relying on central bank facilities.
- Type 2 assets: HQLA plus other assets that, after standard haircuts, could be used as collateral for central banks’ standing lending facilities.
- Type 3 assets: HQLA plus additional assets that could be used to collateralise either standing facilities or, with more conservative haircuts, emergency liquidity support in extreme stress scenarios.
Therefore, in order to better monitor banks’ liquidity risks, in addition to the current regulatory controls (based on the notion of self-insurance), taking into account the availability of collateral that could be used to obtain liquidity from the central bank in alternative stress scenarios with different degrees of severity could be considered.
Arguably, the way in which central bank support could be factored in should be jurisdiction-specific, reflecting the significant variations in central banks’ operational frameworks across countries. In this context, given its flexibility, Pillar 2 emerges as a natural choice to enhance the effectiveness of banks’ liquidity risk controls. Additionally, Pillar 2 measures could take into account bank-specific characteristics, such as funding concentrations and, possibly, the extent to which banks rely on amortised cost instruments to meet HQLA requirements.
Importantly, Pillar 2 measures based on the availability of eligible collateral should take the form of guidance or supervisory expectations and avoid being over-prescriptive. As such, they could function as complementary indicators to monitor banks’ liquidity situation. More formal and rigid requirements could be subject to disclosure obligations. This would potentially exacerbate the stigma effect that may be associated with central bank borrowing, hence reducing those Pillar 2 measures’ effectiveness.
In this framework, the three tiers of asset eligibility could be used to define three indicators for liquidity control, which would be used either for Pillar 1 requirements or Pillar 2 supervisory guidance:
- The first indicator would be a Pillar 1 minimum liquidity requirement consistent with the current LCR in terms of both eligible assets and the stress scenario.
- A first supplementary liquidity ratio under Pillar 2 would be designed as a reformulation of the LCR. It would show the level of liquidity that banks hold, or are able to obtain, to cope with a stress scenario that is more severe than what the LCR assumes. This suplementary liquidity indicator would therefore include not only holdings of HQLA but also assets which would be eligible (after haircuts) as collateral of central banks’ standing facilities.
- A second supplementary liquidity ratio under Pillar 2 would be designed to measure the bank’s ability to address extreme liquidity stress. For this ratio, eligible assets will include those that are eligible for LCR and the first suplementary ratio but will also include assets which could be acepted by the central bank (normally after severe haircuts) when providing emergency liquidty support.
From an operational perspective, when computing the two supplementary ratios, the proposed framework would require that eligible non-tradable assets be prepositioned with the central bank to ensure their swift mobilisation in times of need. As such, if the stress scenario underpinning the second supplementary ratio were to assume a run on all uninsured deposits and short-term funding, supervisory expectations about the level of this ratio would closely align with the recommendations outlined in the Group of Thirty report.
In keeping with the principles of Pillar 2, authorities would have the discretion to implement guidance on one or both supplementary ratios, depending on their specific needs and circumstances, including with regard to the characteristics of domestic frameworks for central bank liquidity support. They would also be responsible for calibrating the severity of the stress scenarios and for determining the range of eligible assets for each supplementary ratio.
The simulations we have conducted at the FSI suggest that covering significantly more stringent stress scenarios than the one currently underpinning the LCR solely with HQLA would be challenging for most banks. At the same time, sound banks would generally be well positioned to comply with reasonable supervisory expectations for the supplementary ratios if they were to preposition non-HQLA, particularly in jurisdictions with broad collateral frameworks. In contrast, banks with a high volume of runnable liabilities would probably struggle to meet these expectations.
Conclusion
Let me conclude.
As policymakers, regulators and industry participants, it is our collective responsibility to ensure that the lessons of 2023 translate into meaningful reforms. At the same time, we must ensure that prudential controls do not unduly challenge the sustainability of otherwise sound business models.
The 2023 banking turmoil underscored the need for a more integrated approach for controlling banks’ liquidity risk. While the current regulatory framework provides a robust foundation, current requirements need to be complemented with an assessment of banks’ ability to cope with more severe liquidity scenarios. That assessment should factor in the availability of sufficient assets that can be expeditiously used to collateralise access to central bank liquidity facilities.
By introducing a tiered approach to asset eligibility and incorporating central bank facilities and collateral prepositioning, we can enhance the robustness of the existing control framework for banks’ liquidity risks in the current environment. This integrated framework should help ensure that sound banks remain resilient to severe liquidity shocks without requiring a fundamental reshuffling of their balance sheets.
Thank you.
References
Barr, M (2024): “Supporting market resilience and financial stability”, speech at the 2024 US Treasury Market Conference, Federal Reserve Bank of New York, New York, 26 September.
Coelho, R and F Restoy (2025): “Rethinking liquidity requirements”, FSI Insights on policy implementation, no 25, May.
Group of Thirty (2024): Bank failures and contagion: lender of last resort, liquidity and risk management, January.
King, M (2023): “We need a new approach to bank regulation”, Financial Times, 12 May.
Restoy, F (2024): “Banks’ liquidity risk: what policy could do”, speech at the XXIII Annual Conference on Risks, Club de Gestión de Riesgos de España, Madrid, 22 November.
Tucker, P (2014): “The lender of last resort and modern central banking: principles and reconstruction”, BIS Papers, no 79, September.