Put to the test

Finansinspektionen became the first European regulator to publicly reveal the results of its stress tests on Sweden's four largest banks in June. But how much do such exercises reveal about the ability of the country's financial institutions to withstand serious stresses? Alastair Marsh reports

p21-andersson-jpg

If the turmoil of the past two years has proved anything, it is that the reliance on quantitative risk management techniques by financial institutions has its limitations. As events unfolded, it became apparent the firms who performed best supplemented quantitative tools with a strong emphasis on qualitative measures, including stress-testing.

It is no wonder, then, that stress-testing has seen its stock soar with regulators lately, particularly in their assessments of capital adequacy at banks. Following the exit from the market in 2008 of two major dealers, Bear Stearns and Lehman Brothers, regulators and other market participants want to know whether banks have enough capital to survive future stresses.

While both the Financial Stability Forum and the US Senior Supervisors Group made reference to the importance of stress-testing practices in their 2008 reports on the causes of the crisis, the Basel Committee on Banking Supervision took matters a step further with the release in January 2009 of a consultation paper Principles for sound stress-testing practices and supervision.

The Committee followed that up with a finalised set of principles in May, in which it concluded: "Not only was the crisis far more severe in many respects than was indicated by banks' stress-testing results, but it was possibly compounded by weaknesses in stress-testing practices in reaction to the unfolding events." In response, the Committee recommended banks adopt an enterprise-wide approach to stress-testing, that there should be greater buy-in on the results by senior management, and that more extreme scenarios should be stress-tested.

National and regional regulatory bodies have been quick to latch on to these ideas. In the US, banking regulators embarked on a programme in February to assess the capital requirements of the country's 19 largest banks by running their portfolios through two macroeconomic scenarios of differing severity. The results were publicly released on May 7, and 10 banks were collectively advised to find a combined $74.6 billion in additional capital.

In Europe, the Committee of European Supervisors (Cebs) has also identified stress-testing as a top priority. However, the two EU-wide stress tests it will conduct this year will look at the aggregate banking system rather than "identify individual banks that may need recapitalisation". Cebs considers such exercises the responsibility of national supervisors and, in contrast to the US example, it will not make the results public. National regulators in the UK, Switzerland, Denmark, Italy and Norway have conducted stress tests on individual firms, but also declined to publish the results.

In fact, the only European country that has published its findings on stress tests is Sweden. On June 10, Finansinspektionen, the country's financial regulator announced the results of its stress tests on the country's four largest banks: Nordea, Svenska Handelsbanken, Swedbank and Skandinaviska Enskilda Banken (SEB). The banks were assessed on whether they were sufficiently capitalised to withstand three scenarios of differing severity over a three-year period between 2009 and 2011.

The conservative base scenario assumed bank lending will continue to increase over the period, albeit at a lower rate than the past few years; while a second scenario envisages extreme stress in eastern Europe, where Swedish banks have lent heavily, rising unemployment, a huge fall in demand and no new loans being originated. The third, and most stressed, scenario factors in extreme stress in eastern Europe, but also assumes a prolonged recession in western Europe.

Credit risk, which, according to Finansinspektionen, typically represents 85% to 90% of bank capital requirements, was the main variable assessed in the tests. It was assumed the capital requirement for market risks and operational risks would remain constant during the three-year time horizon. The results reflect expected losses in banks' credit portfolios based on previous economic downturns, while taking into account Finansinspektionen's discussions with the banks.

The banks' credit portfolios were divided into 30 different exposure classes, including residential mortgages, corporate loans and exposures to foreign markets. Each was ranked according to the level of risk across Sweden, Finland, Norway and Denmark. For each exposure type, the level of credit losses was assumed to be equal for each bank so that the differences in loss estimates only reflected differences in the composition of their loan portfolios.

The conservative base scenario assumes aggregate credit losses across the four banks of 202.6 billion Swedish krona (SKR) and earnings of SKR296.1 billion. The second scenario assumes the same level of earnings but losses of SK^R249.2 billion. In scenario three, credit losses will rise to SK^R350 billion and earnings will fall to SKR260.3 billion.

Rating migrations within banks' internal rating systems will also have an effect on their capital requirements. Migrations mean exposures are moved between different risk classes. In the base scenario, banks' capital requirements are assumed to increase by 5% in 2009 and by an additional 2.5% in 2010 due to migrations.

Under the base scenario, all of the banks would report an overall profit for the three-year period. However, both SEB and Swedbank would incur a loss in 2010 as a result of increased credit deterioration in the Baltic region and the Ukraine, where they have large exposures. (In this scenario, the Baltic states account for 15.1% of credit losses, while Ukraine makes up 38%). Over the three years, Swedbank would suffer the highest credit losses in percentage terms of 3.68% (SK^R52.2 billion in nominal terms), SEB's losses would total 2.81% (SK^R47.3 billion), Nordea's are forecast to be 2.33% (SK^R77.3 billion) and Handelsbanken's would amount to 1.41% (SK^R25.8 billion). In terms of the effect on banks' Tier I capital ratios, the change is most noticeable at Swedbank, which sees its capital ratio fall from 10.5% to 9%.

In the second scenario, where credit conditions are predicted to significantly deteriorate in the Baltic states and eastern Europe, the differences between the banks become more marked. Only two of the banks, Nordea and Handelsbanken, are predicted to have earnings in excess of credit losses between 2009 and 2011, while losses would exceed earnings by SKR21.9 billion at Swedbank and by SKR7.6 billion at SEB. Over the period, the losses result in Swedbank's Tier I ratio falling from 10.5% to 6% and SEB's dropping from 12% to 8.2%

In the most adverse scenario, only one bank - Handelsbanken - is predicted to have earnings higher than credit losses. Conversely, Swedbank would record an overall loss of SKR38.1 billion, SEB would report a SKR35 billion loss and Nordea would be hit by a SKR20.4 billion loss. In this scenario, Swedbank's Tier I capital ratio would fall to just under 6%, SEB and Nordea's Tier I ratios would be marginally under 8%, and Handelsbanken's would be below 10%.

As with any qualitative exercise, the first question that needs to be asked is: what do the tests prove? The Swedish regulator is adamant the tests adequately demonstrate the ability of the four banks to weather severe storms. Finansinspektionen concluded that "with regard to regulatory requirements, there is currently no need for the major Swedish banks to further strengthen their capital adequacy". However, the regulator cautioned that rising loan losses in the Baltic region and a higher market demand for capital will pose real challenges for Swedish banks. Consequently, the four institutions must demonstrate 'good capital preparedness' by putting in place concrete plans to improve their capital adequacy.

Shortly after the results were published, the International Monetary Fund (IMF) on June 15 agreed the results should be viewed with caution. In a report on Sweden, the IMF noted the severity of the underlying scenario assumptions, the bank-by-bank analysis and the transparency in reporting the results reflected best international practice. But it added that concerns remain over the adequacy of Swedish banks' capital, asserting "the regulatory minima (capital standards) are regarded as insufficient".

There are other reasons to scrutinise the results closely. The tests took into account changes in credit risk, but implied that market risks and operational risks would remain constant. However, the turmoil of the past two years has demonstrated a close correlation between market and credit risks, exposing weaknesses in the traditional hypothesis that these two risk types are unrelated.

In a May report, entitled Findings on the interaction of market and credit risk, the Basel Committee came to the same conclusion. "The development of credit risk transfer markets and the moves to mark-to-market accounting for portions of held-to-maturity banking book positions, however, have blurred distinctions between them and raise questions regarding approaches that treat the two types of risks separately," the Committee said.

Finansinspektionen insists it was reasonable to exclude market risks when stress tests are conducted over a longer time horizon, because market positions could be hedged or closed out in a shorter time period. Martin Andersson, general director of Finansinspektionen, believes that in the Swedish context this is a suitable approach.

"This was appropriate considering the structure of the Swedish banks. We haven't had as much of a market risk problem as other countries," he says. "We don't have this drama because there isn't a very big push for increasing the trading book of the banks at the expense of the banking book. Swedish banks are more traditional with a large banking book."

Liquidity - or, more pertinently, the lack of it - is another risk many banks have struggled to manage during the crisis. Andersson acknowledges the point, and adds that it was underestimated by regulators in all jurisdictions. He advocates a more 'hard-wired' approach to liquidity risk, but said that factoring this in to the stress tests was not necessary. "Addressing liquidity concerns with capital requirements is, at best, an imperfect approach. It is better to focus stress tests on capital (solvency) and require significant buffers above minimum levels," he asserts.

In Sweden, as in many other parts of the world, regulators are engaged in serious dialogue concerning the best way of assessing capital adequacy. Andersson believes that any solution must include countercyclical measures, but is unsure of the best way to achieve this given the enormous market pressure for capital as well as stricter regulatory demands.

"I'm not sure that, if we introduce a countercyclical capital regime to build up buffers in good times and to be drawn down in bad times, we will address the issue," he says. "Markets are procyclical anyway: even though banks have these very high buffers from the good times, the market reaction may not allow them to draw down on the buffers in the bad times because this will be seen as a weakness".

The relationship between regulatory demands and those of the market is a new phenomenon, which has only emerged during the current crisis. As a result, more discussion is needed before countercyclical regulations are enforced, Andersson adds. "We absolutely need to have less cyclicality, but we must do it in a robust way so that the regulatory countercyclical measures are not overridden by the market's procyclicality"

Numerous solutions have been put forward to address this, including the Spanish approach of dynamic loan loss provisioning through-the-cycle or a more countercyclical capital charge. However, he is an advocate of a 'third way': a through-the-cycle risk classification system embedded in banks' internal ratings-based models. This would work particularly well in the Swedish case, Andersson believes.

"In many other countries with point-in-time risk classifications, there can be more cyclicality. We don't have as much migration as them; in our case the models are calibrated over-the-cycle against the 1990s regional financial crisis to include countercyclical elements".

In fact, the four banks involved in the stress-testing exercise employ different internal rating models for calculating the probability of default and loss given default of their counterparties. For SEB and Swedbank - which use a more through-the-cycle approach - the capital requirements were assumed to increase less than at Nordea and Handelsbanken, which favoured a point-in-time approach. In addition, SEB and Swedbank will be affected more by the negative scenario, meaning that they will have more counterparties in inferior risk classes that default.

In the official stress-test report, Finansinspektionen confirms "the choice of rating methodology affects the banks' capital requirements". However, Andersson does not believe that a uniform approach was necessary, and says the discrepancies do not invalidate the results. "We have looked into both models and I would not say that two are through-the-cycle and two are point-in-time, I would say that two are a bit more through-the-cycle and the others are a bit less. We have accepted both of these set-ups and they are both appropriate. It is a balancing act that we have here," he adds.

Banks give their feedback on stress tests

The results of Finansinspektionen's stress tests have, in general, been viewed positively from the banks involved. "The tests strike a good balance between a satisfyingly good stress scenario and not pushing the parameter to the point of impossibility with the extreme scenario. In general, the banks feel they are tough but fair," says Philip Winckle, Stockholm-based head of group credit risk control at Skandinaviska Enskilda Banken (SEB).

Perhaps the main reason for the positive response from banks is that the loss expectations under the different scenarios are roughly in line with their own estimates. Rolf Marquardt, chief risk officer at Svenska Handelsbanken in Stockholm, tells Nordic Risk the regulators' assumptions were "not far off" the results of its own stress tests.

However, some discrepancies were noted between the parameters of the scenarios used by regulators and those used by the banks themselves. Marquardt says the regulatory scenarios are "less applicable" to Handelsbanken because they have a heavy focus on losses in the Baltic countries, where the bank has only a relatively minor exposure of approximately 2 billion Swedish krona. For proprietary reasons, Marquardt could not disclose the full list of scenarios employed in Handelsbanken's stress tests, but says two scenarios are based on the Swedish financial crisis of the 1990s and an extended deflationary period, such as that experienced by Japan in the same decade.

Winckle says that "in the absolute stressed scenario there was no big divergence [between SEB's scenarios and the regulators'] because we use aggressive parameters. But we would never call their first scenario a base case".

The severity of scenario one has been a common complaint among the other banks. "We have been asked by management and the board of directors if it is in fact a forecast and, if it is, we don't believe it is a fair forecast. I would think of a base case as the best estimate for the year. Instead I view this as a mild stress scenario," says Winckle.

Other concerns have also arisen. "We think the tests were excessive in that the average loss rates are only appropriate for banks with a variety of size of counterparties. If a bank had a portfolio of exposures to small, medium and large corporates, it would have been appropriate. However, Nordea and SEB mainly have exposures to large corporates with better risk grades, and that hasn't been taken into account," Winckle argues.

Overall, Marquardt believes the stress tests were a useful exercise, despite their limitations. "It is hard to know how large credit migrations will be and how demand for capital may develop. We think the regulators' estimates are as good as can be, but these kinds of tests are not an exact science," he remarks.

Winckle feels the tests would have been more accurate had the regulator factored in more than publicly available data, which do not give a complete picture of a bank's overall portfolio. "We know our portfolio contains a mixture of exposures, and they haven't taken full account of that," he says.

Only users who have a paid subscription or are part of a corporate subscription are able to print or copy content.

To access these options, along with all other subscription benefits, please contact info@risk.net or view our subscription options here: http://subscriptions.risk.net/subscribe

You are currently unable to copy this content. Please contact info@risk.net to find out more.

You need to sign in to use this feature. If you don’t have a Risk.net account, please register for a trial.

Sign in
You are currently on corporate access.

To use this feature you will need an individual account. If you have one already please sign in.

Sign in.

Alternatively you can request an individual account here