27 March 2011

EU Bank Stress Test 2011: Severity of Scenario

On 18 March 2011 EBA, the European Banking Authority published the details of EU-wide Bank Stress Test scenarios and methodology in 2011. Compared to last year, the early announcement is quite a progress of course, and among others allows analysts to assess and comment the severity of stress test scenario long before the results come out. One of the first questions coming into mind is: “How severe is this year’s adverse scenario compared to  the last year’s scenario?”

In its Q&A page EBA provides an answer: “The adverse macro-economic scenario, designed by the ECB, will incorporate a significant deviation from the baseline forecast, which is larger than the one assumed in last year exercise. For example, in the 2011 scenario the GDP shock for the EU has a cumulative shift from the baseline over two years of 4 percentage points. This compares to just 3 percentage points in the 2010 scenario. It will also include country-specific shocks on real estate prices, interest rates and sovereigns.”  The answer continues: “However, a stress test should be judged on a number of factors including the scenario to be assessed but also the harshness of the assumptions used and the capital threshold against which banks are assessed.” Sounds like the 2011 scenario is a lot more severe, eh? Let’s see.

A first quick look to the scenario reveals that it still doesn’t include the sovereign default, although we think that at current stage this would definitely classify if no more likely than that, then at least as “severe but plausible hypothetical event”, which the adverse scenario pretends to reflect.

Further, we can also show that in terms of GDP, the severity of the scenario compared to last year’s exercise is just a matter of presentation. Look at the assumed evolution of GDP figures for EU27 and Euro area in the Table below. When calculating the differences between GDP growth rates in adverse scenario and in base scenario (as EBA does it) we can see that 2011 exercise is more severe indeed. More precisely, the first scenario year is more severe (e.g. for EU27 -0.4% vs 1.7% in 2011 exercise, which compares to 0.0% vs 1.0% in 2010 exercise), while in the second year the difference between GDP growth rates in base and in adverse scenario is more-less the same in both exercises. But is that the comparison that we should make?


What we just compared, is the relative level of stress, stress relative to base scenario. In this type of comparisons the adverse scenario looks the more severe the more positive we are about base scenario. As you can see, GDP growth rates are almost the same or even slightly better for 2011 exercise when compared to 2010. The only thing that makes 2011 scenario more severe is that more adverse developments occur during the first scenario year, i.e. by the end of 2011 the banks have had less time to build up so-called first line of defence, profits before losses.

We also calculated two indexes for GDP. First we equalised GDP for 2009 to 100 (see the row “GDP, 2009=100”) to see what the scenario looks like in relation to the worst year during the most recent downturn. The result: in 2010 exercise the GDP level for adverse scenario went slightly below to that in 2009, but in 2011 exercise it remained above. Thus, in that sense the 2011 scenario is even less severe than the much-criticised last year’s scenario. Secondly we thought that ok let’s relax the criteria for comparison a bit (justified by long-term upward trend in GDP) and equalise GDP for the year immediately preceding the first scenario year to 100 (i.e. for 2010 exercise GDP for 2009=100, see row “GDP, 2009=100” and for 2011 exercise GDP for 2010=100, see row “GDP, 2010=100”). As we already concluded from the comparison of GDP growth rates, even in that case the 2011 stress scenario looks only slightly more severe.

What’s the most appropriate comparison? Well, it depends on the purpose of the analysis, but if we really want to assess the strength of the banks, we should base the analysis to the absolute level of stress rather than relative. When putting it otherwise: we shouldn’t compare the adverse scenario with baseline scenario (as the EBA does it – why?), but rather compare the developments of scenario variables with some absolute levels. The low-point of the latest severe downturn would suit well. Ok, if the downturn was long ago, this low-point might be corrected with some reasonable long-term trend in real GDP. If we take base scenario as point of reference, the stress test results are likely to be most optimistic just before everything collapses.

Of course, GDP is not the only important macro variable to look at. The conclusion for unemployment, for example, would not be different. In 2010 exercise the unemployment was actually higher than it is in 2011 exercise. Where the adverse scenario for 2011 exercise really seems to be at least somewhat more severe is real estate prices, see Table 2 below. Notably, real estate price growth rates for most of the PIIGS countries except Spain have been revised downwards. For some countries like Portugal (commercial real estate), Ireland (residential real estate), Italy (commercial real estate), Greece (commercial real estate), Malta, Poland and United Kingdom (commercial real estate) the difference is even that significant that the question arises: “What were they thinking when preparing the scenario for 2010 exercise?” In other countries, however, the 2011 scenario also for real estate is more optimistic (see Germany and Netherlands) – and this sometimes despite of more conservative assumed GDP development.


Harness of the assumptions used for stress testing purposes as well as the capital threshold against which banks are assessed are matters of methodology and thus separate topics that shouldn’t be mixed up with assessing the severity of the scenario. So we will not comment them right now.

To summarise, again the stress test scenario seems to be fitted to some unspoken expectation or “target” of how many banks will fail the test and which ones of them. When just speculating then for example, Deutsche Bank whose Core Tier 1 capital formed only 1.6% of  the total assets as of 31 December 2010 (source: Annual Report of Deutsche Bank), is probably not supposed to fail. On the other hand, we could assume more failures in smaller banks and where risks are anyway in the spotlight.

No comments:

Post a Comment