Banks' liquidity risk: what policy could do
Speech by Mr Fernando Restoy, Chair, Financial Stability Institute, at the XXIII Annual Conference on Risks, Club de Gestión de Riesgos España, Madrid, 22 November 2024.
Many thanks to the organisers for their kind invitation to come back to the Club de Gestión de Riesgos to share with you a few thoughts on current regulatory policy issues.1 Last time I was here, I had the opportunity to discuss the prudential treatment of climate-related financial risks. Recent catastrophic episodes in this country clearly and dramatically illustrate the relevance of what we discussed at that time.
Introduction
Today, my presentation will cover a different but also relevant topic: the control of liquidity risk. I think this theme requires little motivation in this forum, particularly after the 2023 banking turmoil.
Indeed, what we observed at that time were bank runs of an unprecedented magnitude. For example, in the case of Silicon Valley Bank, the volume of deposits that were withdrawn or planned to be withdrawn within just two days reached 85% of its total deposit base. While not so extreme, deposit outflows in the case of other banks that failed in 2023, such as Credit Suisse or First Republic, were substantially larger than the ones observed during the Great Financial Crisis (BCBS (2024)).
Those developments may well be linked – at least partially – to a structural reduction in the degree of stability of deposits at weak banks. As has often been argued, that can be the consequence of the application of new technologies – such as online banking – by commercial banks. In addition, there is already evidence that social media and the internet can contribute to and have already contributed to intensifying bank runs, by helping to disseminate news and rumours very rapidly2.
There are therefore clear grounds to reflect on how policy should respond in an environment where episodes of liquidity stress in weak banks could become more frequent and severe, thereby accelerating the collapse of those banks and generating or exacerbating financial stability risks. That reflection should cover, at least, four different policy areas: regulation, deposit insurance, supervision and central banks' emergency liquidity support. Let me share with you some views on these four policy domains.
Regulation
Liquidity requirements are a cornerstone of our prudential frameworks. By mandating that banks use sufficiently stable funding sources and maintain a significant stock of liquid assets, authorities mitigate risks associated with liquidity and maturity transformation activities. Such requirements also reduce the likelihood of firms taking drastic, potentially damaging procyclical actions to address liquidity shocks during stress periods. Importantly, liquidity buffers provide banks and supervisors with additional time to prepare for the orderly resolution of unviable banks.
The regulatory instrument which has been more directly challenged by the banking turmoil is the liquidity coverage ratio (LCR). This is designed to ensure that banks have a sufficient amount of high-quality liquid assets (HQLA) to face a liquidity stress scenario characterised by the runoff of a portion of liabilities with a maturity of less than 30 days. The runoff rate for each type of liability is a function of its perceived instability: eg low for fully insured deposits and high for funding from other banks or corporates.
Recent episodes show that the calibrated runoff rates for specific liabilities in the LCR are considerably lower than the ones actually observed. For example, the deposits that Silicon Valley Bank lost in a single day were still higher than what the stress scenario underlying the LCR calibrations would assume for a whole month.
At the same time, the definition of HQLA has been subject to some controversy. As you all know, the current eligibility criteria do not consider the accounting treatment of the instruments. Thus, the conditions for debt instruments in the amortised cost (AC) category to be eligible as HQLA are the same as similar instruments in fair value (FV) categories. The argument has been made than AC instruments would not be as readily available as FV instruments to cover liquidity outflows, as the sale of AC assets would often imply the need to recognise potentially sizeable capital losses in both the profit and loss account and regulatory capital. That does not happen when AC instruments are used to obtain liquidity via repo transactions with private counterparties or the central bank. Yet, private repo markets are often dysfunctional in stress situations, and a large reliance on central bank liquidity could carry some stigma.
Against that background, some regulators3 are considering reviewing the runoff rates assumed for specific liabilities in the calibration of the LCR and also introducing constraints on the eligibility of AC instruments as HQLA. The Basel Committee on Banking Supervision has those issues on its analytical agenda, although no policy adjustment has been proposed so far.
The temptation to strengthen regulatory requirements in the aftermath of such episodes is understandable. However, there is a limit on what regulatory requirements can achieve and on the degree of regulatory stringency that can be imposed without impairing banks' intermediation business.
Thus, minimum liquidity requirements are inherently narrow in scope and cannot anticipate all possible scenarios of liquidity stress. Importantly, regulatory requirements are not and should not be designed to ensure that all banks (not even all solvent banks) would be able to address any conceivable run of deposits or other liabilities.
Going back to the case of the Silicon Valley Bank failure. That bank suffered the withdrawal of 25% of its deposits, with another 60% expected for the following day. Arguably, any bank required to hold a sufficient amount of safe and liquid assets to cover that extreme liquidity stress would have been unable to conduct any meaningful commercial activity.
Therefore, targeted adjustments to the LCR may be warranted, provided that such changes are supported by sufficient evidence on their effectiveness and proportionality. However, there appears to be no compelling case for a complete overhaul of existing liquidity requirements. That would not be the most effective way to address all challenges posed by the apparent reduction in the degree of deposit stability that can be expected in weak banks.
Deposit insurance
Deposit insurance is another critical pillar of the policy framework aimed at preserving financial stability. In particular, insurance programmes for banks' liabilities were introduced in the United States almost two centuries ago precisely with the objective of mitigating the risk of runs and helping to prevent contagion (FDIC (1998)). As you are aware, in the light of lessons learned from the 2023 banking turmoil, some observers and authorities have suggested a review of the adequacy of current coverage levels in existing deposit insurance frameworks. For example, a proposal has been made to consider expanding deposit insurance coverage to firms' operational payment accounts (FDIC (2023)). By their very nature, these accounts can have large balances. As firms use these accounts in their day-to-day operations, they find it costly to diversify funds in smaller accounts held at different institutions. At the same time, blocking firms' access to those accounts when the bank fails could be highly disruptive to their business continuity, as their ability to pay employees and regular providers could be severely impaired.
However, here again the case for significant reform of existing frameworks is unclear. Moderate increases in coverage levels are unlikely to substantially alter the volume of uncovered deposits and, therefore, the frequency or intensity of possible runs, as large-volume deposits are concentrated in accounts held by a very limited number of clients. At the other extreme, more substantial changes could have significant implications for the size of deposit insurance funds and, by extension, for the costs to the banking sector and, eventually, taxpayers. Moreover, a substantial increase in coverage would raise the usual concerns about moral hazard, redistribution of resources from sound banks to weaker competitors and distortions in resource allocation within the financial system.
Supervision
As you can see, I am somewhat sceptical about the net benefits of introducing much more stringent liquidity requirements or extended deposit coverage. In contrast, I am much more confident in the role that strengthened supervision can play in preserving banks' stability.
Through the combined outputs of various components of the supervisory review and evaluation processes, including onsite and offsite inspections, supervisors can develop an informed and comprehensive assessment of banks' exposure to liquidity risk. Moreover, supervisors may use liquidity stress tests to take a forward-looking perspective on banks' liquidity risk. At the entity level, stress tests allow authorities and banks to review how banks' liquidity positions evolve under adverse conditions and to test their resilience. At the macro level, authorities can conduct liquidity stress tests to assess the overall resilience of the banking sector to severe liquidity shocks, considering the relevant interdependencies among individual banks and, potentially, between banks, markets and other types of financial institutions (Baudino et al (2024)).
Yet, as clearly illustrated during the banking turmoil, liquidity stress episodes are often directly linked to banks' structural vulnerabilities, affecting their governance, business models and internal controls. Identifying those vulnerabilities and the interaction among them should be the main outcome of the regular supervisory review and evaluation processes. Thereafter, early stage supervisory dialogue and moral suasion may suffice to resolve many of the issues arising from the supervisory analysis. Beyond that, supervisors may escalate matters and deploy formally binding requirements commensurate with the nature and severity of identified deficiencies. Available measures can include not only liquidity add-ons but also qualitative measures aimed at correcting structural deficiencies in governance, risk management, contingency funding plans and even business models. I believe that well defined qualitative measures are essential. Or in other words, I do not see how supervision can be fully effective if it relies only on the imposition of quantitative capital or liquidity requirements. As was documented in a recent external assessment report commissioned by the ECB,4 there is clear scope to improve the formulation, prioritisation, scalability and monitoring of the qualitative supervisory measures employed in Europe. I am convinced that the same could be applied to the supervisory framework employed in other jurisdictions.
It is true, however, that the success of supervision hinges on adequate resources, legal powers, independence and a supervisory culture that fosters early intervention when necessary. The cultural element is a real challenge, as it requires the supervisory agency to put in place and apply, at all levels of the organisation, a risk tolerance framework that could assume a certain amount of policy mistakes and successful litigation against supervisory actions.
All that said, as long as banks continue to engage in risk transformation, they will remain vulnerable to runs, even if they satisfy regulatory and supervisory requirements.
Central bank liquidity support
This brings me to my final point: the role of central banks. Central banks play a vital role in reducing the probability and impact of bank runs by acting as lenders of last resort to the banking system.
The main constraint for central bank liquidity support to solvent banks is the availability of acceptable collateral. However, for a typical commercial bank, runnable liabilities, including non-covered deposits and short-term market funding, represent around 20–30% of total assets. This seems to suggest that, even with large asset haircuts, there is normally a significant margin to generate collateral for the required loans from the central bank in emergency situations.
Yet, the process of pledging collateral to obtain a central bank's liquidity support entails some complexity. Banks need to have ready all the adequate documentation corresponding to each asset pledged, conduct adequate due diligence from a legal and operational point of view and ensure the availability of accurate valuations. Central banks need to review all those actions taken by firms and apply the right haircuts. Logically, that process – which might be particularly cumbersome for non-traded assets – should be conducted in an agile way within the limited time available for central banks to draw emergency loans in the middle of a liquidity stress scenario.
Therefore, central banks need to put in place adequate measures to ensure banks are fully prepared operationally and willing to use central bank facilities if needed. Thus, at the very least, authorities must require all banks, or at least systemically important ones, to have operational arrangements in place to pledge collateral for central bank liquidity support. Such requirements could also mandate regular testing and simulation exercises.
To ensure the availability of sufficient collateral to provide emergency liquidity support, central banks could go a step further They could require or incentivise firms to pre-position collateral. Pre-positioning can take different forms but, at the very least, it entails transmitting to the central bank details on collateral assets and all necessary documentation to assess eligibility, transferability and valuation. At present, different forms of pre-positioning are possible or even encouraged by many central banks.5
While this is not the current practice of any major central bank, some observers have put forward proposals to formally require banks to pre-position a certain volume of collateral at the central bank, calibrated as a proportion of liabilities that are considered unstable.
The most extreme proposal is the so-called "pawnbroker for all seasons" approach advocated by Mervyn King and Paul Tucker.6 Such a proposal goes as far as replacing the key elements of the current regulatory, supervisory and deposit insurance frameworks with a requirement for all banks to pre-position with the central bank sufficient assets to collateralise (after conservative haircuts) loans from the central bank for an amount equivalent to the bank's total amount of runnable liabilities. In this proposal, runnable liabilities would include all deposits and short-term market funding.
A more moderate alternative is the one presented by the Group of Thirty (2024), which would calibrate pre-positioning requirements as a function of a set of liabilities that excludes insured deposits. Along those lines, the Federal Reserve Board is considering introducing pre-positioning requirements as a proportion of non-covered deposits (Barr (2024)).
There appear to be arguments to at least consider some form of pre-positioning requirements and to make them depend on the existing amount of liabilities – such as non-insured deposits – that are more likely to run in a crisis episode. Yet, since the amount of pre-positioned collateral needs to be measured after applying conservative haircuts over accounting values, those new requirements might constrain banks' ability to lend unless they increase in parallel the amount of stable market funding and capital, thereby constraining their ability to conduct traditional intermediation business.
Therefore, requirements should be carefully calibrated so that they do not directly undermine the sustainability of the risk, liquidity and maturity transformation business that most commercial banks perform.
Conclusion
In conclusion, recent events have underscored the increasing relevance of commercial banks' liquidity risk and suggested a structural reduction in the degree of deposit stability that can be expected in weak banks. This calls for a commensurate evolution in the relevant policy framework. However, instead of advocating for sweeping changes to regulatory and deposit insurance frameworks, authorities should focus on the implementation of agreed international standards, bolstering banking supervision, enhancing emergency liquidity arrangements and ensuring banks are operationally prepared to access necessary liquidity facilities. By taking these steps, authorities can significantly reduce both the likelihood and impact of bank runs.
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.
Baudino, P, P de Carvalho and J P Svoronos (2024): "Liquidity stress tests for banks – range of practices and possible developments", FSI Insights on policy implementation, no 59, October.
Basel Committee on Banking Supervision (2024): The 2023 banking turmoil and liquidity risk: a progress report, October.
Coelho, R , M Drehmann, D Murphy and R Walters (forthcoming): "Navigating liquidity stress: operational readiness for central bank liquidity support", FSI Insights on policy implementation.
Cookson, J A, C Fox, J Gil-Bazo, J F Imbet and C Schiller (2023): "Social media as a bank run catalyst", Université Paris-Dauphine Research Paper, no 4422754, April.
Dahlgren, S, R Himino, F Restoy and C Rodgers (2023): Assessment of the European Central Bank's supervisory review and evaluation process, report by the Expert Group to the Chair of the Supervisory Board of the ECB.
Federal Deposit Insurance Corporation (FDIC) (1998): A brief history of deposit insurance, September.
--- (2023): Options for deposit insurance reform, 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 (2023): "The quest for deposit stability", speech at EFDI International Conference, Budapest, Hungary, 25 May.
Tucker, P (2023): Regimes for lender of last resort assistance for illiquid monetary institutions: lessons in the wake of Credit Suisse, report to the Swiss Finance Ministry
1 I'm grateful to Rodrigo Coelho for very helpful comments.
2 See eg Cookson et al (2023).
3 See eg Barr (2024).
4 See Dahlgren et al (2023).
5 See Coelho et al (forthcoming).
6 See King (2023) and Tucker (2023).
