The Greatest Hoax on European Bank Capital Shortfalls ever?
Acharya and Steffen (A&S), authors of “Falling short of expectations? Stress-testing the European banking system”, estimate capital shortfalls of EU banks mounting up to €767bn. By any standard this is a large amount. However, we notice that the results of A&S are biased: their most prominent results are the result of awkward choices regarding the way they set up their research.
In an attempt to demonstrate the capital shortfalls of European banks, A&S apply a veritable kitchen-sink approach: any percentage and any definition of solvency that delivers a high capital shortfall figure – in whatever context – seems to work for A&S.
However, in reality, European banks this year face two distinct tests: i) the asset quality review (AQR) which requires banks to satisfy an 8% CET1 ratio requirement, and ii) the EBA stress test, which involves an elaborate set of assumptions that define stress scenarios. A&S should have focused on these tests only. Instead A&S present a confusing paper.
Below follow our main comments, which make us wonder why A&S published their high-profile results. It is a sloppy paper that does not show much news. However, it does highlight a lack of understanding of solvency ratios – something that worries us.
A sloppy paper.
First and foremost is the quality of the paper: it bears many of the hallmarks of sloppiness. This affects the authority of the paper: why trust the authors regarding complex matters if accomplishing simple tasks turn out to be a challenge? For example, not all tables are self-explanatory; the execution of tests is imperfect; alternative data form the European Banking Authority could have been used. Important variables, such as C Tier 1, are not well explained: is the “C” in C Tier 1 the “C” for “Common”, or the “C” for “Core”? What matters is CET1: Common Equity Tier 1, as defined by the CRR.
Sloppy definitions of solvency metrics.
The shortfall measures are ill-chosen. For example, an important solvency measure in A&S is C Tier 1, as this variable is probably the closest to the CET1 measure for the AQR. That measure should tick the 8% box. Using that measure, the deficit is under 8bn euros. We replicated this result using data from EBA’s transparency exercise: we show a €7.77bn deficit for 64 banks, where A&S show a €7.553bn deficit for 109 banks; see the tables below.
The second main ratio featuring in A&S is the leverage ratio, which under Basel III and the CRR is defined as Tier 1 divided by the exposure measure. For the numerator of the leverage ratio, A&S use various exotic measures that, as far as we know, are not defined by the relevant regulation. Instead, A&S should have used Tier 1 capital divided by their currently used denominator tangible assets minus derivative liabilities. The other measures of leverage may not be relevant.
A&S also use a market solvency measure jointly with a 4% or 7% market leverage ratio requirement. The market leverage measure forces a bank to have in issue 4 or 7 Euros of market equity for every 100 euros of book debt + market equity. Given that no bank can credibly write a contract on market values, the market measure is useless for bank supervision purposes.
It is also unclear why A&S use such an odd market leverage measure. Its denominator may be inflated: it is the sum of the book value of liabilities and the market value of equity. Highly levered firms may have liabilities that trade below par. On the other hand, equity values of these firms are relatively high because of the value of the option to default. Using book liability values and market equity values biases leverage ratios downward. A proper market measure of the denominator would rely on the market value of both equity and debt. Further, a simple test of a market capital shortfall would be the extent to which the market value of equity dips below book value. Given that banks have survived for years with market values below book values, one may wonder why particularly today this measure is problematic.
Regarding the leverage ratio, all countries, except for Greece and Cyprus, tick the 3% leverage ratio box, see A&S Table 1. The 3% leverage ratio requirement will last for some years. It will be evaluated by EBA at the end of 2016, beginning of 2017. Therefore, it is unclear why A&S mention a 7% leverage ratio requirement, which clearly is over the top. A 7% requirement just shows a large capital deficit. Furthermore, also using 4%, within the European context, is odd: the proper leverage ratio requirement should be 3%.
The OECD, in its 2013 report, examined European leverage ratio shortfalls. Using a 5% leverage ratio requirement, the OECD singled out France, Germany, Finland, and the Netherlands as countries with leverage ratio weaknesses. A&S, however do not single out Finland and the Netherlands. Therefore we are left with conflicting messages about leverage ratio weaknesses in Europe.
Sloppy execution of analyses.
Some of the analyses are poorly executed. For example, Table 1 shows that the C Tier 1 Ratio of the Netherlands is 28.88%. This is very high, especially given that the same table reports an Equity/Assets ratio of 3.19% for the Netherlands. The reason for the extremely high value of 28.88% is one small bank with a C Tier 1 ratio of 111.06%.
So, the authors used the average of the solvency ratios, per country. This biases results in favour of the smaller banks. For example, using this approach, the bank of Valleta in Malta carries the same weight as BNP Paribas. A better way of presenting solvency ratios is to divide the total of C Tier 1 by the total of Risk-weighted assets – per country. Just assume there is one big bank in each country and that single bank has all equity, assets, Tier 1, CET1, RWA’s, etc. This is still not perfect, as it biases the results in favour of small countries, i.e. countries with few banks. However, for countries with many banks of different size, this approach mitigates the bias.
More importantly, the authors may or may not be aware of the effect these sampling choices have on their inferences. They may just have not spotted the very high C Tier 1 solvency rate of the Netherlands. Researchers that don’t do a sanity check on the data should worry us.
Mixing up the context of tests.
Although the AQR is about achieving the 8% CET1 ratio, this measure appears late in the paper and also in the wrong context. See Table 7, where the CET1 ratio measures the effect of a capital shortfall after write down. But note, this is in stressed conditions; and in stress, Basel III and CRD IV allow banks to relax the conservation buffer requirement. Instead of 8%, the minimum 4.5% CET1 requirement may apply, or perhaps the 5.125% CET1 ratio that triggers a Tier 1 CoCo write-down (which implies that perhaps in stress, Total Capital, the sum of Tier 1 and Tier 2, could serve as a proper measure of loss absorbency).
Another example of a mix-up is the use of the leverage ratio, which, for the AQR is not the relevant measure. A&S use the 4% and the 7% leverage ratio requirements to demonstrate the capital shortfalls. However, within the context of the AQR one should look at the quality of assets, not at the effects of various solvency percentages.
Following on from the last point, the stress assumptions appear harsh: a 40% stock market decline and a full write off of non-performing loans. A&S could have more elaborately compared their stress assumptions against those that are available in the public domain.
More importantly, we wonder if a full write off of non-performing loans can be executed under current accounting rules. This is in the light of comments on the powers of the ECB to direct banks to write down assets.
No news, so what?
The results that A&S present are hardly new. Shortfall amounts of hundreds of billions of Euros are not new. PWC recently attracted attention by mentioning a shortfall of €282bn.
But more worrying is the lack of international standards for bank capitalisation. A Google search on “shortfall capital European banks” offers a wide range of figures, all equally reliable and meaningful. Seven years into the global financial crisis, and despite the new Basel III definition of capital, there is still no international agreement regarding what it means to be a well-capitalised bank. This should worry us all.
The Tables below replicate some of the results of A&S. It is not our aim to demonstrate more accurate capital shortfall figures. Instead, the tables show the sensitivity of shortfall figures with respect to the data source, the choice of solvency definition, the choice of required solvency percentage, and the sample choice.
Bankscope data, end of 2012.
Tier 1 ratio is Tier 1/RWA, T1 LR is the Tier 1 leverage ratio: Tier 1/TA, T1 deficit 8% is the Tier 1 deficit at a Tier 1 requirement of 8% of RWA. LR deficit 3% is the leverage ratio deficit at 3% of total assets (TA). Data is European Bank data from Bankscope at end of December 2012. Amounts in bn.
Using data from the EBA Transparency exercise 2013, per end of 2012.
This table is comparable to the table above, in addition, it shows the deficits in terms of CET1 at various CET1 requirements. Amounts are in bn euros. Data is from EBA’s transparency exercise, measured at the end of December 2012.