How Not to Stress EU Banks and Their Supervisors Mario Tonveronachi

One crucial parameter for computing the sufficiency of capitalisation is the average risk weight (RW) applied to bank assets. Banks with the same value of capital and total assets but different RWs may end in red or black for capital sufficiency. If the RWs were to reflect the true degree of bank risk that would be what prudential regulation was designed for. If, on the contrary, the regulatory RWs were distorted by inappropriate rules and heterogeneous supervisory practices, the result would be a complete regulatory failure. After the recent crises, the official authorities had to recognise that many rules were distorting the RWs (especially those relating to the trading book) and that the discretion attributed to national supervisors was in many cases responsible for the light touch that started or deepened the crisis.

Regarding the rules, two European Directives (CRD II and CRD III) have addressed the problem of the low risk weight for the trading book and securitisation, potentially tripling the value of the RW applied to such activities. The objective of the new calibration is to re-equilibrate the RWs of different parts of the portfolio, abolishing the incentives favouring the trading book. Regarding discretion, three new European supervisory authorities have been created with the precise objective of homogenising supervisory practices across the EU Member Countries.

In the banking sphere, one of the first initiatives taken by the new European Banking Authority (EBA) has been to promote a new stress test for 91 EU banks, adopting a common and supposedly stringent methodology. The recent publication of the results of the test offers much food for thought. This brief note only focuses on the RWs in order to suggest that the two above mentioned reforms are not producing the desired results.

In short, the European Banking Authority has substantially accepted the projected RWs that were transmitted by the banks, supposedly with the seal of their national supervisory authorities, for the years 2011 and 2012 under the two baseline and adverse scenarios. Starting from January 2011, these coefficients should also incorporate by the beginning of 2012 all of the RWs’ increases dictated by the two above mentioned Directives. It is worth noting is that if the banks were not to send their calculations on their perspective RWs for market risks, the EBA would have applied a floor for their increase with multipliers of 1.1 and 1.4 for banks utilising the standardised or IEB approach respectively. Not much, if we think that these multipliers only apply to market risks.

The RWs for 2011 and 2012 shown in the Tables 1 and 2 are computed dividing the amount of the risk weighted assets by the level of total assets. Since both quantities are the result of a complex set of evaluations, including the opposite effects due to the higher risks included in the baseline scenario and some write-offs, of which no detailed data are given, a very neat evaluation is difficult. However, we hardly observe for the following two years a significant general increase of the average RWs with respect to 2010, even in a stressed scenario. This result may be due to the fact that although the RW for market risk increases substantially, its very low starting level renders the overall effect almost negligible.

Deutsche Bank is a good example. Although its risk weighted assets for market risk as a percentage of total risk weighted assets almost triple, going from 7% in 2010 to 20% in 2012, the average RW only increase of 25.5%. Starting with the lowest average RW, Deutsche ends in the same relative position. Even if Deutsche were punished with a 9.5% of Tier 1 capital requirement for being a Global SIFI, it could go on in a 2012 stressed scenario with a maximum leverage of 46, well above the 33 that the Basel Committee generously suggested as a tentative cap. Since Deutsche accounts for the more significant increase from among the sample, it seems that we cannot count on the new rules on RWs to decrease leverage and improve resilience.

Let’s then consider the re-balancing issue. While Table 1 shows the data for the banks of the sample that are among the 20 world largest, Table 2 refers to the country averages of the rest of the sample. We may easily suppose that the largest banks generally operate with higher market risks. A large difference in RWs exists between the largest banks and the average of the other sample, and that the gap is going to persist. No re-balancing is, then, is going to come from the new rules. If they were wrong at the onset of the crisis, they remain firmly so.

The heterogeneity in national supervisory practices too is still alive and well. First, the two tables, which list banks and country averages with RW in increasing order, present a good correspondence at country level. Second, the variability of the two samples is much the same. The data thus suggest that at least a large part of the variability of RWs is due to national factors, supervisory ones in the first place. As in the previous cases, the old distortions are not going to disappear.

It is easy to understand how different should have been the results of the stress test if the re-balancing toward the higher levels of RW were homogeneously in place. The German banks, among others, have much to celebrate.