Stress testing regimes have come a long way since the 2007-9 global financial crisis, however, they are still evolving with regulators on both sides of the Atlantic signalling potential changes. By Justin Pugsley  

Over the next few years, regulators on both sides of the Atlantic are likely to make changes to their stress testing regimes to keep pace with evolving risks and to conform with political priorities. 

Beverly Hirtle, executive vice president and director of research at the Federal Reserve of New York, believes that stress testing and the US Comprehensive Capital Analysis and Review (CCAR) programme were the most significant and impactful regulatory and supervisory changes to emerge from the 2007-9 global financial crisis (GFC).  

“I think the only reason banks are so well capitalised is down to the stress test process," says Damien Burke, a partner at specialist risk consultancy, 4most, adding that it gives supervisors considerable insight into a bank’s risk management processes.

As such, most industry sources expect stress testing to continue growing in importance and are likely to include new scenarios reflecting new risks.  

“I think the future direction of stress testing is being shaped by politics reflecting what the government is worried about. The general capitalisation of banks is less in the crosshairs versus conduct and operational resilience,” says David Biggin, a financial expert at PA Consulting.

The US has gone the furthest with stress testing making it central to supervision and has become a means for regulators to shape bank balance sheets and capital buffers.

The US is currently re-evaluating its stress testing regime as explained in November 2018  by Randal Quarles, the vice chairman of supervision at the US Federal Reserve Board.

In line with the administration’s deregulation agenda, US regulators are making the tests more proportionate to a bank’s size and risk profile and are reducing compliance burdens.

One global framework?

Meanwhile, jurisdictions are diverging their stress testing regimes.

On November 16, 2018, The Program on International Financial Systems (PIFS), an organisation fostering dialogue on financial issues, published a report outlining some of the differences.

For instance, in their scenario designs the Fed, the Bank of Japan (BoJ) and Bank of England (BoE) incorporate counter-cyclical factors, while the European Banking Authority (EBA) does not. The time horizon for the scenarios ranges from a nine-quarter planning horizon based on a 13-quarter scenario in the US, through to three years for the EBA and BoJ and five years for the BoE.

The Fed and BoJ rely heavily on their own models, while the EBA uses bank models (with constraints) and the BoE takes a hybrid approach. All four have a global shock component, but the BoJ does not have a counterparty default component unlike the other three. The same three use stress testing to help inform bank specific capital requirements, but not the BoJ and according to PIFS only the EBA and BoE use them as part of system-wide capital requirements.

“You have different systems and you have real disparity between them,” says Angus Dent CEO at ArchOver, a UK P2P lending platform, noting that the tests can create different results for the same banks.

“It is creating inconsistency. It undermines the credibility of the tests,” says Mr Dent, who believes there should be a global standard for all regulators rather than simply a set of high level principles as set out by the Basel Committee on Banking Supervision. He also thinks there is a lack of transparency.

Indeed, many global banks would agree with Mr Dent’s assertion, saying that different stress testing regimes create inconsistencies and complicate management decisions.

“The methodology, disclosure and actions are very different across jurisdictions,” says Fernando de la Mora, managing director and the head of Alvarez & Marsal Spain and Portugal.

He cites the very different common equity tier 1 (CET 1) depletion rates used in the latest batch of EU, UK and US stress tests coming in at 410, 480 and 600 basis points respectively. “There would be benefits for convergence on these different regimes and approaches,” he says, but is not optimistic that this will occur.  

But some believe diversity has some pluses.

“I think it is helpful to have multiple sets of regulators coming up with their own ideas as to what is effective,”  says Mark Turner, managing director, compliance and regulatory consulting practice at Duff & Phelps, and who was also once a senior risk consultant at the UK’s Prudential Regulation Authority.

Mr Turner believes banks should engage with national regulators and use their relationships with these authorities to iron out conflicts that might arise from different stress testing regimes.

US regime under review

In 2018, the authorities themselves delivered hints over how stress testing regimes could evolve while acknowledging some limitations.

Speaking at the Federal Reserve stress testing research conference on October 9, the New York Fed’s Ms Hirtle spelt out some of the limitations saying: “Capital ratios are measured separately for individual institutions, with no attempt to capture dynamic interactions, spillovers or specific cross-firm exposures.”

She added that outcomes for the banking sector do not interact dynamically with the macroeconomic scenarios, which are inputs to the calculations, which are standalone. “Results for the banking system are therefore simply the sum of results for individual institutions, rather than capturing interactions among them,” she said.

Therefore, there is only a limited attempt to capture the extent of interconnectedness, which under stressed conditions can quickly morph into contagion as happened during the GFC. It seems natural for supervisors to stress test for contagion.

“Working that out would probably be messy,” says Julian van Kan, head of financial institutions coverage for EMEA at Mitsubishi UFJ Financial Group (MUFG). He reflects a common view that attempting to work out how contagion might spread is fiendishly difficult to simulate given the complexity of the financial system.

“In a perfect world where the analysis of all the data of every bank could be done concurrently, there could be some benefit,” says Mr Turner adding that banks and supervisors are not at a stage where that can be done anyway – as stated by Ms Hirtle.

“One thing we have got now is central clearing and that has taken away some of that risk,” says Mr Van Kan. Also, well capitalised and properly managed banks should be able to withstand contagion.

Ms Hirtle dedicated much of her speech to addressing the question of “what’s missing” in the US stress tests. She noted that the current regime does not explicitly integrate capital and liquidity stress testing at the firm level or attempt to assess the probability or impact of a firesale of banking sector assets. Furthermore, being capital focused, the regime does not address firm-specific or systemic issues related to resolution.

“The ‘stand-alone’ design choice accounts for this outcome,” she said.

Mr Van Kan believes there is insufficient focus on liquidity as the tests tend to look more at credit stress.

“Your cash is sitting in the LCR (liquidity coverage ratio); it is also invested in HQLA (high quality liquid assets). Available cash is a big thing. We have to make sure there is available cash in order to disburse obligations under our loans," he explains.

He stresses the importance of understanding a bank’s ability to realise some of its assets under stressed conditions, which feeds into liquidity.

Ms Hirtle explained that designing a regime that takes into account possible liquidity pressures, bank runs and firesale risks would be complex involving greater abstraction and simplification at the expense of precision.

She then addressed criticisms around the limited number of scenarios used in the stress tests implying a lack of thoroughness. Ms Hirtle said generating supervisory projections is resource and time intensive and this limits the number of scenarios that can be used at any one time. Banks, meanwhile, are supposed to stress test their own particular vulnerabilities.

Nonetheless, she acknowledged that multiple common scenarios could provide a more robust assessment of the banking sector’s capital strength.

Ms Hirtle discussed the merits of using simpler and easier models for estimating net income and its key components. The Fed has already developed benchmark models for producing loss and revenue estimates as a comparison to the projections from the more sophisticated and complex production models.

She believes this approach produces more robust assessments of supervisory results over time.

As for further research, Ms Hirtle said there is scope to analyse whether the tests take enough account of the “first-mover problem” a bank might face when cutting dividends or doing capital raisings during emerging periods of stress

An important area of research, in her view, is lending by non-banks and non-stress tested banks that could substitute CCAR covered institutional lending.

P2P lending platforms are a new type of funding channel, particularly for smaller businesses and for some consumer lending, which many banks shun due to derisking. However, P2P platforms given they are relatively new remain untested during times of stress – albeit with one exception.

The world’s first internet-based P2P platform is UK-based Zopa, which launched in March 2005. The firm grew during the GFC and delivered positive returns for lenders using its platform, despite a big spike in defaults in 2008.

The company claims to have learned plenty from that experience generating unique data sets for its risk management processes. It believes that lenders using its platform would still make money during another recession.

Though Zopa’s example is encouraging, the sector remains largely untested and small.

Ms Hirtle called for more research into how stress testing effects the cyclicality of capital requirements and the appropriate degree of cyclicality. And finally, she questioned risks around the Fed’s approach creating model monocultures, which could drive banks to behave in ways that create systemic risks by pursuing very similar hedging or capital allocations strategies.

EU under scrutiny

In June 2018, the European Parliament published a report identifying many weaknesses in the EU’s stress testing regime. It questioned whether its methodology provides a consistent enough picture of the European banking systems’ vulnerabilities.

For example, banks use their own models to project potential losses from operational risks, but with regulatory constraints to maintain consistency. The report highlights that some of those constraints might be too binding on some banks transforming deposits into loans with a degree of maturity mismatch.

“The EBA approach is aimed at enforcing comparability such as having very strict methodology constraints and assumptions, like static balance sheets,” says Mr de la Mora at Alvarez & Marsal.

He contrasts this with the US and UK approach which allows for a more dynamic balance sheet approach. “UK and US stress tests are much more aligned to the internal planning of the banks and their ability to take mitigating actions to replenish capital over time in case a crisis is triggered," he explains.

Also, he believes the comparability of the EBA’s results is undermined by the  arbitrary nature of the European Central Bank’s quality assurance processes. “I don't think the comparability goal is being accomplished,” he says.  

The Parliament report expressed concern that benchmark models used by National Competent Authorities (NCAs) to scrutinise the banks’ results are not subject to public consultation and may miss certain bank specific issues.

These issues are potentially exacerbated with the Parliament estimating that the the EBA does not have enough stress testing staff relying considerably on the NCAs. Industry sources warn this could lead to NCAs influencing the stress testing results to flatter weaker institutions.

The Parliament is calling for more resources for the EBA and for an independent oversight body that is “less prone to the interests of individual authorities and more inclined to guarantee a uniform application of the EBA methodology.”

Other areas Mr de la Mora thinks need addressing is the torrent of disparate and unnecessary data requests banks are bombarded with by the authorities during the tests along with unsynchronised processes. “I think it could be much more efficient for all parties,” he says.  

Meanwhile, the European authorities take great pride in the transparency of their tests generating some 100,000 data points. Mr de la Mora believes there is too much transparency and wonders what the authorities do with so much data.

Indeed, the European authorities are aware there are shortcomings.

In November 2018, EBA chairman Andrea Enria, who became head of  banking supervision at the ECB in January 2019, suggested the EU could learn from the Fed’s approach to testing bank resilience.

He said the main deficiency in the EU approach versus the US is the opaqueness of the EU’s supervisory actions, which contrasts with the transparency of its tests.

He gave a damning analysis of the EU regime, saying the EBA lacked powers to directly challenge bank responses to the stress scenarios and that Europe is slower than the US with follow up actions.

Mr Enria said there is a lack of clarity around corrective measures, supervisory judgements and communications.   

He said the authority could take a more dynamic approach over how bank balance sheets evolve over the course of the test cycle as done by the BoE. He even floated the idea of combining the EBA’s ‘bottom up’ approach with the Fed’s ‘top down’ approach, but said this would take several years to develop.

Coming changes?

Given recent comments from regulators, it is clear that the EU and US stress testing regimes will experience changes.

“I think they will stress test more of what I would call the credit market and the impact of rising interest rates, and I think they might take more interest in asset managers and hedge funds. I think there will be a lot more scrutiny on them," says MUFG’s Mr Van Kan. “There might be more stress testing outside of the bank market and on the insurance markets because banks rely quite heavily on the credit insurance market." The EU is set to run a liquidity stress test later this year.

He thinks this could see a more joined up approach among regulators for stress testing different types of financial institutions.

Mr de la Mora does not expect much immediate change to the UK regime, but sees the prospect of less banks in the US being exposed to CCAR requirements and thinks the Fed could become more flexible on some of its model approaches. For the EU he foresees some methodology changes possibly involving relaxing some of the constraints on bank models and for the exercise to become more aligned with bank internal process. “This will take some time to adopt,” he says.

4most’s Mr Burke believes a hybrid approach using a single scenario of published data and a bank’s own internal accounts combined with a move from a static balance sheet approach to a more dynamic one is the way to go. So supervisors can compare the static and dynamic approaches to get a better idea of how a bank’s balance sheet might evolve while under stress.

“This  allows an understanding of the shape and size of the portfolio going forward, its risk profile and also where those stresses occur and what the bank would do about them and their history of managing such risks,” he says. “You also get the comparability.”

He explains that the tests should account more for the actions management can take during a period of severe stress. However, he acknowledges that this would require more supervisory resources to conduct it at a time when they may be already stretched.

Meanwhile, there are growing concerns around cyber risks and climate change and their potential impact on banks. Climate related models are being developed with the UK’s Prudential Regulation Authority (PRA) and the industry.

In terms of cyber resilience the US authorities and industry have been conducting simulation exercises and the this year the PRA is to stress test the resilience of bank payment systems to a cyber attack.

Mr de la Mora says these exercises are useful to gain an understanding of these risks, but thinks they need to evolve and mature more before including them in a comprehensive stress test. “It is hard to estimate the impact of a cyber attack on the financial system," he says.

PA Consulting’s Mr Biggin thinks these factors should be included in the stress tests, but thinks it is difficult at this stage to say whether they should form part of existing tests, looking at their impact on GDP for example, or if they should be conducted as dedicated standalone tests.

“I think we're more likely to see a proliferation of other types of stress test, which go after specific issues, such as cyber and operational resilience more generally,” he says, adding that the UK authorities have placed cyber resilience and operational resilience at the top of their agenda.

“The challenge is to frame that stress in way that it can be modelled, otherwise it becomes very much a game of crystal ball gazing,” he says.

Mr Turner thinks it inevitable that cyber and weather related risks will find their way into stress tests given such risks have already been manifested.

“What the Bank of England has tried to do with the mis-selling stress component, as in push the onus on firms to determine what's right for them, I think it would make sense to take the same approach for cyber type scenarios,” he says.

Stress testing regimes have come in leaps and bounds since the GFC, but clearly the methodologies that support them still have considerable scope to evolve.