David Benamou +33 1 84 16 03 61 david.benamou@axiom-ai.com Philip Hall +33 1 44 69 43 91 p.hall@axiom-ai.com Jérôme Legras +33 1 84 17 36 10 Jerome.legras@axiom-ai.com Adrian Paturle +33 1 44 69 43 92 adrian.paturle@axiom-ai.com Stress tests finally credible? The eurozone is at a turning point in its short and eventful history, with the ECB taking up an ambitious challenge: cleaning up banks balance sheets in order to prevent a Japanese-style deflationary scenario from happening in Europe, without bringing more volatility to the markets. The ECB is in an excellent position to convert this tough balancing act into a tremendous achievement for Europe which is having some trouble finding new growth drivers. There are three prerequisites to this success: 1. Restoring confidence First of all, the ECB must restore confidence in the soundness of European banks to reduce financing costs so banks can offer more reasonable interests rates to their customers. In peripheral countries, reducing financing costs is a key growth factor. For equivalent credit risks, Spanish SMEs are obtaining funding at rates 250bps higher than their French or German counterparts! How to convince investors that the banking crisis is finally over? The answer is simple, although we are aware of its quite circular nature: some scapegoat must be offered to the market. Some banks have to fail for the market not to disregard this new stress test, like it did with previous stress tests. The ECB is aware of that, as demonstrated by Ms. Nouy, in charge of European banking supervision, who dared to say that some zombie banks would have to disappear. Everyone is now wondering who will be concerned. Even if no one can give a definite answer, market expectations are clear and the ECB s decisions should be pretty much in line with them. In a recent survey, 68% of investors believed that 6 to 20 banks would fail, with an average of 11. However, failing does not mean going bankrupt, but only identifying additional capital needs, which are likely to be met by the private sector. According to the respondents, total capital needs would amount to 50bn to 100bn. In our opinion, most of the banks concerned should be able to implement a credible recapitalisation programme, with or without public offering. Three to five banks, at most, could be forced to a organized bankruptcy..axiom Alternative Investments 39, Avenue Pierre 1 er de Serbie 75008 Paris Tel: +33 1 44 69 43 90 contact@axiom-ai.com
2. Adopting a credible methodology to finally clean up balance sheets Spotting the victims is, however, not an end in itself: if the market believes regulators have not targeted the right banks (Dexia s success in the EBA s stress tests is still fresh on everyone s mind!), then the whole exercise will have been in vain. European banks could end up in the same downward spiral as their Japanese counterparts, and would have to reschedule NPLs to artificially prop up profits, rather than leaving past mistakes behind and financing the economy again. As a result, stress tests will have to be economically meaningful, which is a far more complex objective, given how many factors are to be taken into account. To date, four main types of stress tests have been conducted in Europe: The EBA s stress tests, which have no credibility since they have been incapable of pointing out the banks weaknesses. Scenarios identified by the European Commission as part of its review of state aid programmes, which did offer an economic and general approach, but only under moderate stress conditions. Stress tests performed by consulting firms at the request of the Troika, in countries that had received an aid (and in Slovenia). These consultants have done an excellent job but their action was strongly limited by an institutional steering committee. They focused exclusively on estimating provisions and overlooked entire aspects of bank solvency: RWA calculations, trading portfolios, impact of new regulations, etc. Targeted portfolio reviews (SMEs in Italy, shipping and commercial property in Germany, etc.) by some financial regulators. Although the German BaFin seems to be trustworthy, its review was not conducted transparently and it is difficult, in these conditions, to make a judgment. Like most analysts, we were initially convinced that the stress tests of the Comprehensive Assessment would be similar to those implemented in Spain, but the documents published recently are forcing us to reconsider our position. The methodology the ECB and the EBA intend to implement is way more ambitious. It is based on the following two pillars. 2.1. Asset quality review An asset quality review (AQR) was conducted by the ECB and a whole army of consultants. According to the initial feedback, the amount of data needed for this review is absolutely bewildering: for each bank, about one million data are collected and 20,000 credit files are reviewed. Although the relevance of such a massive workflow is questionable, it is almost certain that the hundreds of millions of euros spent in consulting fees will not have been in vain as substantial impacts are to be expected, maybe even higher than those of stress tests. Regarding the AQR, the ECB has nothing to feel guilty about, except maybe being over-zealous! 2
This review would allow, on the one hand, reclassifying sound debts as doubtful debts (based on new standard EBA definitions) and, on the other hand, estimating adequate provisioning rates. We believe that investors can get an idea of the first effect by reading banks Pillar 3 reports, which offer a breakdown of exposures by asset class. The second consequence is much less transparent, since banks are very reluctant to disclose any information about their provisioning techniques. Increased transparency will be a key factor in this respect. 2.2. Stress tests Portfolios reviewed will then be subject to a stress test which is a major advance carried out by the banks themselves under the EBA s supervision. The first idea that comes to mind is that trusting the banks with this self-stress test is doomed to failure as they will be able to come to whatever figure they want but the reality is more complex. All technically advanced banks have established as required by regulations a stress test methodology that is undoubtedly more robust than the one implemented so far by EBA. The EBA will, however, add many constraints to the exercise by setting the parameters for these models. The test will be based on the following three main pillars, described by increasing order of relevance: Market risk modelling, including moderate stress (unlike what was done in 2011!) on the value of government bonds held in the available-for-sale portfolio. Operating profitability modelling, including stress on financing costs borne by the bank and on LTRO maturities. By combining these two factors (stress on govies and LTRO maturities), the ECB intends to reduce the golden opportunities generated by the widely known carry trade (3-year LTRO loans, purchase of govies with maturities of less than 3 years). Credit risk modelling, for which banks are required to adopt a radically new approach, as summarised below: 3
Current situation Post-AQR Post-AQR & ST Sound debts Sound debts Sound debts RWA = f( PD Reg-ST ) Generic provisions EL = PD Reg-ST * LGD Reg-ST RWA = f( PD Reg ) Generic provisions Expected loss (EL) = PD Reg * LGD Reg RWA = f( PD Reg ) Generic provisions EL = PD Reg * LGD Reg Potential doubtful debts Amount = PL* PD Pit- ST Provisions = LGD pit- ST Doubtful debts NPL=Loans*PD Pit Provisions = LGD pit Doubtful debts NPL=Loans * PD Pit -AQR Provisions = LGD pit -AQR Doubtful debts Provisions = LGD Pit- ST In order to understand these tests, the following concepts have to be explained: PD Reg and LGD Reg refer to probability of default and loss given default, respectively, which are used in regulatory models. These are average values over long economic cycles. LGD Pit and PD Pit refer to the same values, taken at a certain point in time (PiT) during the economic cycle. Under normal conditions, these values result from doubtful debts actually recorded by the banks and from corresponding provisions. Therefore, an AQR is a simple recalculation of LGD Pit and PD Pit (which become LGD Pit-AQR and PD Pit-AQR, respectively) correcting for classification differences between former and current definitions. Apart from the AQR, stress tests performed so far in Europe consist in calculating new potential provisions through LGD Pit-ST and PD Pit-ST and using simple macro-economic models (e.g. correlation between probability of default and GDP growth). It is also possible to use probabilities of default calculated by credit rating agencies, taking for example the worst value in ten, twenty or thirty years. For instance, depending on the method, the probability of default used for a BBB-rated credit could range from 0.27% (historical average) to 0.57% (in 2009). The EBA will most likely leave it up to banks to choose either method. 4
The EBA has imposed two main new requirements for treating sound debts: on the one hand, banks will have to calculate changes in ratings under stress conditions (e.g. loan ratings changing from A to BBB) and to adjust capital needs; on the other hand, they will also have to adjust the probability of default (PD Reg-ST ) used for RWA calculations! The EBA is thus imposing a quadruple punishment on banks: increased portfolios of doubtful debts, potential new doubtful debts, lower ratings for sound debts and higher consumption of capital for a given rating! Therefore, the exercise will undeniably be robust and conservative. Two question marks remain: 1) will smaller banks have the operational resources to undertake such a tremendous amount of work? 2) how will the major differences between banks applying an internal method (sensitive to ratings and, therefore, unfavourable) and those using the standard method (less sensitive to ratings and, therefore, more favourable) be taken into account? 3. Avoiding any systemic risk If the two previous conditions are met (i.e. if exercise is based on a credible methodology and meets the market s expectations), is there a risk that an unforeseen failure might lead to a new period of volatility? We consider this risk to be very low for the following reasons. Banks subject to these stress tests can be classified into five categories: 1- national champions (BNP, Santander, Intesa, etc.): these will not have any trouble passing the tests as they are solid corporations; 2- banks currently being restructured (HSH, Dexia, etc.) or recently restructured (Bankia, etc.): these are now completely isolated from the financial sector, they no longer participate in the interbank market and only attract distressed investors who are aware of the risks they are exposed to; 3- smaller banks (IKB, Aareal, etc.): their failure would have no systemic impact at all; 4- banks that fall outside the scope of this exercise (Axa Banque, Volkswagen Financial, etc.): their participation is pointless; 5- more complicated cases. The truth is, excluding complicated cases, a failure following these stress tests would have broadly similar consequences as the default of Spanish banks in 2013, SNS s nationalisation or the Cyprus bank bail-in, that is hardly any implications. The more complicated cases are the few large banks holding low-quality assets: Monte dei Paschi, Commerzbank, Raiffeisen, Banco Espírito Santo Except maybe for the Italian bank, which could 5
soon find a new strategic investor, none of these institutions is faced with the risk of a recapitalisation exceeding the money it could raise in the markets. Investors are thus more concerned by the risk of dilution than by systemic risk. In any case, the ECB has already demonstrated its capacity and willingness to take strong action whenever necessary. Everything seems to be ready for the selective power of stress tests to help revive the European banking industry. 6