Measuring banking resilience to adverse outcomes – Bank Underground

Giovanni Covi and Tihana Škrinjarić

The ability of the banking system to absorb shocks and continue providing vital financial services is important because it underpins the smooth functioning of the broader economy. We propose a methodology that serves as a valuable tool for monitoring banking system stability. It quantifies the resilience of the banking system given the prevailing macrofinancial risk environment. The main measure we derive is the probability that one or more banks will fail to meet regulatory capital or liquidity requirements within a given horizon.

What we do

Maintaining banking stability is challenging, as it requires a clear quantifiable definition and a proper measurement. Macroprudential regulators (authorities that monitor and manage systemic risk across the entire financial system) must accurately assess banking stability and set capital requirements so banks can absorb severe shocks. As such, it is important to understand how susceptible banks are to different shocks, such as credit, market, and liquidity shocks. At the same time, it is important to consider the banking sector’s capacity to provide credit that supports the real economy. Setting capital requirements too high could risk hampering economic growth.

In our recent paper (Covi and Škrinjarić (2025)), we extend the capital at risk (CaR) methodology of Covi et al (2022) that quantifies the resilience of banks to those shocks. CaR can be used as a policy tool for tail risk monitoring, scenario, and sensitivity analysis. CaR looks at banks’ balance sheet, capital, and liquidity positions – based on supervisory returns – and how they could change in the prevailing macrofinancial environment exploiting a network perspective. It can also evaluate those changes to a specific shock or stress scenario and assess how shocks propagate throughout the network of bilateral relationships (loans, securities, and funding exposures).

First, we construct the supervisory granular banking exposure data set covering both the asset and liability sides of the seven major UK banks’ balance sheets. In that way, we can see their exposures on both sides of the balance sheets, with the amount and their potential risk (measured by probability of default (PD), and loss given default (LGD)). Then, we observe how banks’ balance sheet, capital, and liquidity positions could change, given a shock in the real economy. To do so, we produce Monte Carlo simulations of banks’ counterparty defaults, based on the correlation structure of their PDs. In that way, we can track and evaluate how risk propagates from within the economic system to banks’ balance sheets.

In the third step, we recalculate banks’ balance sheet, capital, and liquidity positions. Here, we account for the initial Monte Carlo shock, and potential behavioural reactions of banks in several subsequent steps. These include funding withdrawals, accessing the secured and unsecured money markets and engaging in fire sales (forced, rapid sale of assets at prices significantly below their fundamental value).

We then quantify how many times a bank breaches its minimum capital and leverage requirements within one-year horizon over total Monte Carlo simulations. And we weight together the probability that individual banks falling below their minimum regulatory requirements by their relative size to produce and derive the banking system level indicator, 1Y-WALMin. This is our main measure which is used to track how banking stability evolves across quarters allowing the identification of key risk drivers.

What we find

Our results reported in Chart 1 show that the 1Y-WALMin (black curve) started around 1.8% in 2015, at the tail end of banks building capital post global financial crisis (GFC). Subsequently, the value decreased over time, showcasing the benefits of an improved loss-absorbing capacity of the banking system. As of 2024 Q4, the 1Y-WALMin indicator stands at 0.9%, highlighting that banks currently have a high degree of resilience.

We look at the potential impact of a GFC-type event by stressing the risk parameters such as PD and LGD of banks’ exposures and re-estimating our indicator conditional on this adverse scenario for four quarters ahead – up to 2025 Q4 (shaded red area in Chart 1). We find that banking stability (measured by higher 1Y-WALMin) would deteriorate, pushing the likelihood at the peak of the stress to 6.6% (black curve) that is, seven times higher than in the absence of shocks (0.9%).

Chart 1: Weighted average likelihood of banks falling below minimum regulatory requirements

Notes: 1Y-WALMin is weighted by the bank’s size measured by total assets. It is estimated according to a bank’s Common Equity Tier 1 (CET1) ratio falling below 7% (of risk-weighted asset) or leverage ratio below 3.25%. Thresholds are kept constant among banks and over time for comparability purposes. Shaded area refers to estimates of the 1Y-WALMin in the case of a GFC-type event adverse scenario.

How changes in capital affect likelihood of banks falling below minimum regulatory requirements

Besides tracking the historical values of 1Y-WALMin with respect to the actual capital that the banking system had over time (black curve in Chart 1), we can also produce counterfactual values of 1Y-WALMin if the capital would have been higher or lower (orange and red curves in Chart 1). This exercise can tell us how the likelihood of banks falling below minimum regulatory requirements could change, ie how sensitive it is to changes in bank capital.

We perform this counterfactual exercise – Table A – showcasing what would be the system’s equilibrium if banks’ loss-absorbing capacity is to be reduced or increased by 100 basis points (bps) and 200 bps of CET1 ratio, ranging between 12% to 16%. We find that increasing the loss-absorbing capacity by 100 bps and 200 bps would reduce the estimated 1Y-WALMin indicator in normal times by 21 and by 35 bps, and in bad times (GFC-type event) by 122 bps (multiplier = 1.2 ~ 122 bps/100 bps) and by 202 bps (multiplier = 1 ~ 202bps / 200 bps).

Table A: Impact of higher/lower CET1 ratio capital on banking stability

Notes: We report deviations from current levels. GFC columns refer to a hypothetical adverse scenario resembling a financial crisis stress. BCST refers to the Bank of England’s stress test scenario we apply. AVG stands for average effect, NORMAL refers to normal times of our sample, ie without Covid-19 shock and the stressed scenario BCST, whereas COVID refers to the period of Covid-19 shock. Peak refers to the effect in 2025 Q4, ie when the peak of the stressed scenario is assumed.

The regulator may opt to increase banks’ capital buffers by 100 bps to push up banks’ capital over time. This higher capital base would have a limited positive effect on reducing 1Y-WALMin under current conditions as of 2024 Q4 (21 bps). But the benefit of that additional capital would increase if the macroeconomic environment subsequently became stressed. In case a stressed event (as the GFC-type described above) happened, the peak 1Y-WALMin of 6.6% (from Chart 1) could be mitigated to 5.4% because of the previous 100 bps increase of the CET1 ratio. If the regulator would choose to reduce the buffers by 100 bps, this would increase the 1Y-WALMin by 32 bps. If subsequently the macroeconomic environment would to become stressed, the peak 1Y-WALMin would worsen to around 8%.

This exercise shows us that a countercyclical approach builds resilience during stable periods, ensuring banks are prepared before stress emerges, rather than reacting only after trouble begins. Building resilience (loss-absorbing capacity) in good times is key to making the system more resilient in bad times. However, strengthening bank resilience must be weighed against its potential effects on economic growth. It is therefore beyond the scope of this analysis to be able to fully understand the costs and benefits of changing the capital requirements.

The effects of raising or reducing capital in the system are non linear

In both cases, in the good and bad states, we find that increasing loss-absorbing capacity has positive marginal decreasing returns, that is, the first 100 bps increase in CET1 ratio is more effective (multiplier = 1.22) in decreasing the 1Y-WALMin than the latter 100 bps increase. Hence, increasing the loss-absorbing capacity is still an effective tool in building resilience into the system, although the marginal benefits seem to decrease.

Lowering banks’ CET1 ratio by 100 bps and 200 bps would increase on average the probability of default by 32 bps and 81 bps under normal conditions, and by 144 bps (multiplier = 1.44) and 322 bps (multiplier = 1.6) under stress conditions at the peak of the hypothetical GFC-severity crisis.

This result suggests that the likelihood of banks falling below minimum regulatory requirements – keeping everything else equal (like the severity of the shock, banks’ liquidity positions, exposures to CPs, balance sheet positions, etc) – is more sensitive and affected by a reduction in banks’ loss-absorbing capacity (proxy by changes in CET1 ratio) than an equal increase. This is because the same shock will consume a larger amount of capital in the case of lower capital and the non-linear effects we capture (initial shock and subsequent losses due to banks’ reactions) increase the further we go into the tail of possible outcomes. This result holds in bad times as well as in good times.

Giovanni Covi is an independent researcher, who previously worked in the Bank’s Stress Testing and Resilience Division, and Tihana Škrinjarić works in the Bank’s Bank Stress Testing and Resilience Division.

If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.

Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.