Stress-testing the banking system: what lies ahead?

A new round of stress tests was completed just before the summer break: the Federal Reserve’s in June, and the EU-wide and the Bank of England ones in July. The three exercises concluded, unambiguously, that banking sectors in these economies have the capacity to support lending to households and businesses in times of severe stress. In addition, banks demonstrated strong resilience to weather very adverse conditions, mainly reflecting the solid solvency position at the start of the exercise.

As in the UK, the EU-wide adverse scenario incorporated persistently higher inflation, translating into higher interest rates, exploring the interest rate risk channel for the economy and for the financial sector. The US exercise, by contrast, considered short-run interest rates very close to zero over the whole horizon of the exercise, characterised by a global recession but with lower, not higher, interest rates. In a recent book, NYU Stern’s Viral Acharya makes the case for a stagflation stress test – including higher rates – to restore confidence in the US banking system.¹

When compared with the UK and US stress tests, the severity of the impacts on capital of the EU-wide exercise was remarkably stronger. The adverse scenario of the EU-wide stress test – combining severe EU and global recessions, high inflation, increasing interest rates and aggravated credit spreads – produced a sizeable depletion of Common Equity Tier 1 (CET1), which fell 459 basis points, exceeding by far the equivalent results for the UK and the US (340bp and 250bp, respectively). CET1 stressed figures were, in all the three cases, comfortably above their pre-global financial crisis (GFC) levels.

The EU-wide 2023 stress test presented relevant improvements:² introduction of top-down elements, a more detailed sectoral breakdown and an increased sample coverage. The European Central Bank conducted a parallel stress test, covering further medium-sized institutions that are directly supervised by the ECB. This means roughly 80% of total banking sector assets in the euro area were covered in the exercise.

Market analysts note that some banks expressed disappointment with the quality assurance process of the euro-area exercise, as the supervisor’s view on the effects on capital depletion prevailed over banks’ own estimates. But this is good news: the additional role played by top-down elements brings a reinforced ownership of the exercise by the supervisor. The main features of the forthcoming 2025 capital stress test in Europe are expected to further enhance the diagnostic capacity of the exercise, namely through a further increase in the use of top-down analytical instruments.

Risk of reverse engineering

Stress-testing gained a compelling relevance during the financial crisis and, since then, it has played a prominent role in the capitalisation of banking sectors in most geographies. As a recent International Monetary Fund working paper stated:³ “Wider adoption of stress tests as a supervisory tool represents a great advance in risk management and oversight since the GFC, broadening supervisors’ views of threats to individual banks and the sector beyond historical data and past experiences.”

It is also the case, however, that the sequence of stress tests over the last 15 years creates the risk of adaptation. In a recent interview, former Fed governor Dan Tarullo⁴ suggested that banks are possibly in a condition to reverse-engineer the supervisory model, limiting the benefits of stress-testing; he even mentioned that, “we may need to rely more on conceptually second-best, but in practice more transparent, point-in-time capital requirements”.

In the recent EU stress test, only an extremely reduced number of banks failed to meet either the total supervisory review and evaluation process (SREP) capital requirement or the total SREP leverage ratio requirement. This is not at all surprising as the results of stress-testing – which is not designed as a pass-fail test – are used as an input for SREP and therefore for the definition of capital needs. Under those circumstances, the additional benefits of repeating stress-testing with similar characteristics may be declining over time.

The banking sector benefited enormously from reinforced regulation (the Basel III reforms) and more intense and better focused supervision. This has determined prudent levels of capital and liquidity, at the individual bank level.

The banking sector benefited enormously from reinforced regulation (the Basel III reforms) and more intense and better focused supervision (both micro- and macro-prudential). This has determined prudent levels of capital and liquidity, at the individual bank level.

But two other significant changes occurred in the financial sector since the GFC. The non-banking sector – which has remained largely unregulated, as it lacks a macro-prudential framework in most jurisdictions – expanded sizably. Assets of non-bank financial institutions reached about half of the assets of the global financial system. In addition, digitalisation produced drastic changes in the banking landscape. These two developments bring important system-wide risks, which exceed, by large margin, the sum of individual risks.

System-wide approach

How should stress-testing adapt to this rapidly changing financial reality? There are three – independent but complementary – priorities: a system-wide approach covering both banking and non-banking; liquidity stress-testing; and cyber security stress-testing.

First, financial system-wide stress tests should be developed and implemented. The increasing interconnectedness between different segments of the financial sector justifies a predominant system-wide approach covering banks, insurers, pension funds and the asset management industry in the same exercise.

The Bank of England is taking the lead on this. Governor Andrew Bailey announced in December 2022 that the bank would run an “exploratory” stress test to assess how liquidity stress in the non-banking sector could be transmitted to the rest of the financial sector. This exercise, launched this June, has a special focus on liquidity, modelling firms’ actions under liquidity stress, allowing for fire sales, as well as limits to exposures to some counterparties or segments of the financial market. It is also public that the European Commission invited the European supervisory agencies to conduct a one-off climate risk scenario, incorporating the contagion and second-round effects across firms and segments of the financial sector.

Both the BoE and EU exercises are presented as having two rounds: the first one at the level of each individual institution or sector, the second one with a top-down nature to model contagion and amplification effects. A staff working paper of the Bank of England, published in August, presents a very thorough survey of the literature on macro-prudential stress-testing⁵ – with a focus on contagion between banks, between banks and non-banks, and between the financial sector and the real economy – with the identification of key lessons for the development of system-wide stress tests.

Such financial system-wide stress-testing – in which the stress comes from the interactions (contagion, amplification, spillover) between the different segments of the financial sector – would probably benefit from having the usual two-three year horizon. If liquidity stress results in persistently higher risk premiums, the existing vulnerabilities arising from the current high corporate indebtedness could take some time to materialise; highly leveraged companies would face difficulties in servicing the debt after facing rating downgrades and reaching the maturities of the existing financial instruments. Assessing, under a predominantly forward-looking perspective, the capacity of the leveraged large firms – which should be explicitly included in the exercise – to refinance and repay debt should constitute the key new relevant feature of the stress test.

Liquidity in focus

Second, liquidity stress-testing may be very useful in the current juncture of rising interest rates. Andrea Enria, head of the Single Supervisory Mechanism (SSM), in an interview⁶ with Milano Finanza this July, explained why the SSM decided to move the reporting frequency from monthly to weekly: to monitor more closely funding and liquidity developments. The US banking turmoil this year revealed unpleasant evidence of unprecedentedly fast bank runs accelerated by social media and instant news. Since then, suggestions for regulatory change have ranged from adjustments in Basel liquidity metrics (the liquidity coverage ratio, for instance) to the more drastic ex ante collateral coverage of possible liquidity gaps.

The supervisory process requires banks to run their own liquidity stress-testing, including reverse stress-testing, on a regular basis. A possible improvement would be to conduct liquidity stress-testing based on a set of common guidelines defined by the supervisor that would apply to all the institutions but with a systemwide approach,7 with an explicit focus on contagion between banks and non-banks under liquidity stress.

The horizon should be very short, one to three months, perhaps. Aspects to be covered should include the speed of the run-off of retail funding, depletion of wholesale funding, impacts of fire sales on collateralised positions, the impact of a broad deterioration in credit ratings, debt buy-back calls, feasibility of adjustments in banks’ funding profile under stress, and downgrade of the institution itself. The results of reverse stress-testing – identifying in which extreme circumstances a fatal bank run could effectively occur – should be reassessed following the significant scale and speed of outflows experienced in the recent US banking turmoil.

Liquidity stress-test results should not be made public; but they would provide the supervisor with a deeper knowledge of the overall liquidity and funding position of the institution. The environment has become more challenging – characterised by increasing interconnectedness, digitalisation of finance and the role of social media – which could prompt the adoption of preventive risk management measures. In addition, liquidity stress tests could also provide some guidance on possible regulatory adjustments.⁸

The cyber dimension

Finally, cyber security stress-testing is more needed than never, as the economy is much more digital and incidents in the financial sector are rising faster than in any other sector, particularly for payment firms and insurers. A lot of progress has been made since the pioneer experiments run by the Monetary Authority of Singapore in 2016 and by the Bank of Israel in 2019 (for the insurance and banking sectors, respectively). Also in 2019, the European supervisory authorities issued to the Commission joint advice on the costs and benefits of developing a coherent cyber resilience-testing framework. Similarly, the UK’s financial policy committee has conducted a pilot cyber stress test. More recently, the Bank of England conducted a cyber stress test for the retail payment system and the ECB announced that a cyber stress test is being designed to see the light of the day in 2024.

What should be the priorities for cyber stress-testing? The answer is not easy. Two aspects are, however, advisable: the stress test should be bottom-up, under strict guidelines defined by the supervisors, and the results should not be released at an individual level. The European Insurance and Occupational Pensions Authority (Eiopa)⁹ has produced a very thorough discussion on methodological principles to govern the cyber component in stress-testing, having in mind the nature of the effects – operational and financial – and the wide variety of possible losses (direct financial losses, interrupted business activity, restoration costs, legal claims, fines, and reputational costs). These principles apply to all the segments of the financial sector. In addition, for the insurance sector, the cyber stress test should consider underwriting risk.

Cyber stress-testing has a very distinctive advantage: it provides relevant insights to address the cyber protection gap. Insurance-based transfer of cyber risk still lacks any material relevance: the cyber risk-protection gap can be as high as 90% (that is, 90% of the losses are not covered). A sound cyber risk insurance market is of the utmost relevance. Enhancement of cyber resilience requires, simultaneously, sound supervision across the financial sector, appropriate underwriting practices and suitable cyber insurance coverage. Cyber stress-testing could contribute to assessing where insurance protection levels fall below socially optimal levels. It could also provide some guidance on how higher risk-sharing could be achieved. Reducing the cyber risk-protection gap is essential to increasing the resilience of the economy.

Pedro Duarte Neves is adviser to the board of directors of Banco de Portugal. The views, opinions and conclusions expressed in this policy note are those of the author and do not necessarily reflect those of Banco de Portugal or the Eurosystem.

Notes

1. “Restoring confidence in the banking system with a stagflation stress test”, Viral V. Acharya, chapter 6 in “SVB and beyond: the banking stress of 2023”, NYU Stern Business School, July 3, 2023.

2. Jose Manuel Campa, the chairperson of the EBA, introduced these improvements at the Handelsblatt Conference on Banking Regulation: “Our path to the future EU-wide stress-testing”, Frankfurt, March 29, 2023.
3. “Good supervision: lessons from the field”, Tobias Adrian, Marina Moretti, Ana Carvalho, Hee Kyong Chon, Katharine Seal, Fabiana Melo, and Jay Surti, Working Paper WP/23/181, IMF, September 2023.
4. “Talking to Dan Tarullo about bank mergers, stress tests, and supervision”, by David Wessel, Hutchins Center on Fiscal & Monetary Policy, August 10, 2023.
5. “Macroprudential stress-test models: a survey”, by David Aikman, Daniele Beale, Adam Brinley-Codd, Giovanni Covi, Anne-Caroline Hüser and Caterina Lepore, Staff Working Paper No. 1,037, Bank of England, August 11, 2023.
6. Interview with Andrea Enria, conducted by Francesco Ninfole, Milano Finanza.
7. See for instance “Systemwide liquidity stress-testing tool”, Hiroko Oura, Working Paper WP/22/252, IMF, December 2022.
8. See “Reflections on the 2023 banking turmoil”, Pablo Hernández de Cos, chair of the Basel Committee on Banking Supervision and Governor of the Bank of Spain, at the Eurofi Financial Forum 2023, Santiago de Compostela, September 14, 2023.
9. “Methodological principles of insurance stress-testing. Cyber component”, Eiopa, July 11, 2023.