Q&A: EBA’s Vaillant on Basel IV, FRTB and CVA

Isabelle Vaillant speaks quietly, doesn’t get flustered and gives the impression of meaning what she says – no bad thing for someone in her position.

As director of regulation at the European Banking Authority (EBA) – one of three industry-focused authorities that help the European Union flesh out its financial regulation – Vaillant and her team will play a critical role in the final stages of the post-crisis regulatory cycle. In the transatlantic stand-off over the last components of the Basel III package, for example, the EBA provided the technical analysis behind the European position that capital should be risk-sensitive most of the time.

“Non-risk-sensitive items should be a non-dominant part of the framework,” says Vaillant. “And by ‘non-dominant’, we mean roughly a third of the capital requirements should be driven by those non-risk-sensitive metrics, while the rest should be the dominant part, at least 60%. This has been our key goal.”

The US stance is that banks cannot all be trusted to get these calculations right, so variability should be constrained by a leverage ratio, and an output floor that stops modelled capital dropping below some percentage of the regulator-set standardised capital formulas.

In press coverage of this tussle – extended after international regulators called off scheduled meetings in January and again in March – the negotiation has been reduced to a simple arm-wrestle about where the floor should be set, with the US arguing for higher and the EU in favour of lower. But the logic behind this is complex, taking in arguments about structural differences in the respective banking systems, and whether standardised formulas can accurately capture them. It may not always be appropriate to insist on a small gap between standardised and modelled capital, says Vaillant.

“You also have another understanding of the output floor, which is that this distance exists because the calibration of the standardised and internal models approach reflects your actual risks. That means structural differences in risk may produce a greater distance in one jurisdiction than another, which happens to be the case for segments like mortgages, which are typically structured differently – here, you can have a larger distance between the two, and a lower output floor. This is the whole debate we are having now,” she says.

Most banks will applaud the EBA for this stance, and the industry will also find plenty to approve in Vaillant’s approach to incoming market risk capital rules – the Fundamental review of the trading book (FRTB) – which will be transposed into EU regulation via a fifth round of updates to the Capital Requirements Directive (CRD V) and a second round for the Capital Requirements Regulation (CRR II), the proposed text for both of which was published by the European Commission on November 23 last year.

It’s not obvious at all that if you relieve capital you create more room for lending – quite the contrary. There have been so many studies showing robustness in capital levels is a support to the flow of credit

Isabelle Vaillant, EBA

The text gives the EBA the job of specifying the technical details of the profit and loss (P&L) attribution test, which acts as the gateway to the internal models approach (IMA) via a regulatory technical standard (RTS), but imposes some constraints. For example, it states that the test should be based on a comparison of P&L figures calculated by a bank’s front-office pricing models and its risk management models. Despite this, Vaillant says the authority is considering “several possible solutions”. Banks have been protesting the FRTB’s take on the test since the rules were finalised in January last year. 

But on other topics, the EBA is pushing arguments that are opposed not just by banks, but also by EU legislators. For example, the authority has made the running in attempts to remove Europe’s credit valuation adjustment (CVA) capital exemption for trades with non-financial corporates – a split from the international framework that was dubbed “materially non-compliant” by the Basel Committee on Banking Supervision in 2014.

The exemption was won by members of the European Parliament (MEPs), on the argument that the capital charge would make it more costly for corporates to hedge – two tough groups to oppose, and so it has proved. Vaillant says the EBA’s efforts have suffered two blows: first, in its overhaul of the CVA framework, the Basel Committee was unable to agree that capital modelling should be allowed, which would likely add to the costs corporates face if the exemption was removed; second, the authority’s plans to advise EU supervisors to apply a Pillar 2 capital add-on were challenged by MEPs in particular.

This fight is not over – Vaillant insists the risks remain and should be capitalised – but for now, the EBA is having to review its options.

The authority had also advised the EC to stick to the letter of Basel’s net stable funding ratio (NSFR), for example – another bit of international standard-setting wrapped up in CRD V and CRR II – but the commission went its own way on a number of points, including two designed to make it easier for banks to hold and finance European Economic Area government debt. The EC claimed the EBA’s stance could have hurt this market.

Vaillant’s take: “We did not find evidence of that; we did not see evidence of it from the EC either.”  

Elsewhere in the following interview, Vaillant addresses criticism of international regulation by US politicians, last year’s struggle to agree on credit risk capital modelling standards, and what Italian state support for Banca Monte dei Paschi di Siena (MPS) means for the EU’s bank recovery and resolution directive (BRRD).

She does it, for the most part, with a smile. These are the kind of challenges that have punctuated Vaillant’s career: she spent a decade as an economist after joining the Banque de France (BdF) in 1986, but was then appointed head of European regulation at the central bank and has since spent time in high-level regulatory and supervisory roles. She joined the EBA in her current role in 2011.

Are you happy with progress towards the G20’s reform goals?

Isabelle Vaillant: It is important we close the G20 reform package, and for that, an agreement is needed soon. The good thing is that we now have a much wider range of tools – leverage, liquidity, bail-in. This has enhanced our ability to protect depositors and taxpayers. We have also made improvements in banks’ capital robustness, which is extremely important – going from 9% to 14% is a 50% increase.

Some US politicians are now arguing the capital reforms have overshot. Is there any evidence of this?

IV: We’ve heard this kind of argument in Europe, too, and we have to leave behind this theory, that the cost of capital is an impediment to lending. It’s not obvious at all that if you relieve capital you create more room for lending – quite the contrary. There have been so many studies showing robustness in capital levels is a support to the flow of credit.

But what we most need to defend is the idea of a risk-sensitive capital framework, and for this, an accurate ranking of risks is crucial.

If the evidence is clear, why does the argument persist?

IV: It appeals to the intuition: if you ask for more capital, then there is less capital available for lending. So I think we’ll continue to hear that view. You have to take two or three steps to explain it doesn’t work in that way.

Comparing the Basel text and CRR, you have different difficulties. Basel is not even consistent from one piece to another – so there we still have to push for some aspects to be closed, and we are doing that

Isabelle Vaillant, EBA

Is there any danger of the US withdrawing from standard-setting bodies?

IV: The Basel Committee has been working effectively for more than 40 years and this has benefited the entire community. Not being part of this standard-setting process would have a very negative impact.

But there is another dimension here, which is that of transparency in the rulemaking process. In Europe, this is also being requested of those participating in the standard-setting process. The EBA strongly supports enhanced transparency.

Where can improvements be made?

IV: It’s a very technical process. I would not enter politics here, but when prudential rules are embedded in a law voted on by politicians, then it’s good to speak to politicians and be clear on the underlying rationale. So it’s not just public disclosure, it’s liaising in a transparent manner with the legislative channels. I understand this is also what’s being called for in the US.

Are you imagining a future in which US regulators would need to go to Congress before calibrating their version of the leverage ratio?

IV: We need to explain the rationale and produce the evidence. Then, it’s for every nation to decide whether the adoption of the rule is contingent on the say of the Parliament. I wouldn’t advocate that; you would be mixing technical decision-making with politics at an early stage.

So when you say transparency, you really do just mean transparency?

IV: Yes: I stand ready to explain.

CRR II gives the EBA the responsibility to produce at least two-dozen RTSs. Does the authority have the resources it needs?

IV: I counted more than that. The commission’s proposal foresees around 37 new technical standards and a number of new guidelines and reports.

But just in terms of statistics – because I’m responsible for our technical rulemaking activities – based on what we have delivered in the past, we’re looking at two years of work on the CRR if we keep our current staff levels.

Have you been following discussions on the P&L attribution test?

IV: Yes. Institutionally, we are observers of the Basel Committee rather than full members, but it doesn’t make a difference, as we are involved in the Basel groups and bring our contribution with the European perspective.

Comparing the Basel text and CRR, you have different difficulties. Basel is not even consistent from one piece to another – so there we still have to push for some aspects to be closed, and we are doing that.

CRR offers some clarity on the P&L attribution test, but it’s obviously a goal of ours to make sure any progress in Basel is achieved in parallel with the CRR. In addition, it has been decided it would be helpful for the EBA to provide an opinion to the council while it is still negotiating, and before we deliver our opinion, we thought it would be good to issue a discussion paper. We have a list of issues to discuss with the industry, to gather evidence, to make sure we dig sufficiently into the operational issues – because a lot of this debate is about how you put the P&L attribution test into practice. 

So we’ll take time. We’ll issue our discussion paper, we’ll prepare an opinion, the council will still be discussing the text at that point; this is the best contribution we can make.

It’s not a mystery that the EBA has been defending models with the idea I mentioned earlier – that risk sensitivity is crucial

Isabelle Vaillant, EBA

When will we see the discussion paper?

IV: By the end of the year.

You said some things need closure in Basel. Are you thinking of the split FRTB definition of risk-theoretical P&L?

IV: Exactly. This is exactly the point.

But that was clarified in CRR II.

IV: Well, to some extent – CRR II requires a comparison and for that, first, we need to make sure Basel understands it that way, and then we need to negotiate CRR II.

But there is a dilemma here. We want to investigate the solution with our discussion paper, but on the other hand we believe this should be kept open in the level one text – because these practices are going to evolve, they are not yet settled within banks. So we think it would be inappropriate to set it in stone, in the form of regulations as any change requires a three-year process. In our view, a level two solution – meaning technical standards or guidelines – would be more appropriate. 

So what might the level one text say about the test?

IV: It would give the principle – what you have to compare. So, whether it is your risk management model being compared to your front-office model, or something else – there are several possible solutions. And then there will be technical standards to specify it. As a matter of fact, this is the actual plan in the commission proposals.

Banks liked the appendix version of the test, in which two sets of risk factors were run through a single model. You’re not dismissing that right now, then?

IV: This is exactly what we want to investigate, but it’s our option – you could have a closer set of factors, and different ways of comparing them.

But CRR II seems to close off some of those options.

IV: I have no doubt this is open, and that’s what we want to contribute with the discussion paper. The legislative proposal is a starting point and then the negotiations take place.

Will you judge the design of the test by looking at pass/fail rates?

IV: Yes, but not only that. One could be more ambitious in trying to achieve more parallelism in the input, rather than just the output. But this is high ambition.

Do you want a healthy proportion of desks to be able to use models?

IV: It’s not a mystery that the EBA has been defending models with the idea I mentioned earlier – that risk sensitivity is crucial. Of course, it depends whether the modelling is robust, otherwise you stop talking about risk sensitivity and start calling it undue variability, so this is what we have to sort out.

What we face with the P&L attribution test is that it’s not yet very well settled – so we are not facing years of practice, where banks know how to deal with it and are very sure they have stable practices. We don’t want to be advisers to the banks, but we have to push for practices to settle down.

What we’ve sought to do at EBA is make sure the bottom-up and top-down approaches can match at some point. Our main stance – which we have tested continuously – is that non-risk-sensitive items should be a non-dominant part of the framework

Isabelle Vaillant, EBA

It’s been said the Basel Committee lacks the data needed to judge whether the test is too tough. Do you agree with that?

IV: Yes, that’s why we want to take six months to put this discussion paper together, then have three months of discussion – we’re already looking at a year of work there.

If the outcome of the P&L attribution test is that 10% of desks pass – do you see any justification for that?

IV: It’s difficult to tell: either the rule is not well calibrated or the bank’s practices are not sufficiently prudent.

Can you see a reason to clamp down on trading book model use?

IV: There could be. You have to say there is no sufficiently stable practice, which is even more complex in this case because we don’t even know how the desks will be rearranged – so it is a chicken-or-egg problem to a certain degree.

In this case, we have to be pragmatic. We already have a backstop in the form of the SBA [sensitivities-based approach]. We’ll observe two or three years of practice and will then try to incorporate it in our standards.

But I don’t expect it would be as bad as you suggest – only 10%? Come on – they will be better than that. While the calibration of the rule has to be conservative, it intends to reflect best practices, and banks’ practices will adapt quickly.

CRR II softened the Basel version of the NSFR, against EBA advice. How do you feel about the proposed text?

IV: Well, this goes hand in hand with transparency – every organisation takes responsibility for its own decisions. We provided advice – the evidence was there, our rationale was there – and the EC took its own responsibility. There can be some degree of difference or distance.

Our feeling is those differences are not so many, they are being factored in cautiously, and the objectives are something other than prudential policy.

The EC was upfront about the justification – it said it wanted to protect the functioning of eurozone government bond markets. Do you buy the argument that setting the required stable funding at 5% rather than 0% would affect the market?

IV: We did not find evidence of that; we did not see evidence of it from the EC either. We understand this is why the EC added a review clause – it’s very hard to estimate this kind of thing.

The EBA has advocated removing Europe’s CVA exemption for corporate exposures. What is happening on that front?

IV: This is a case where the recommendations we had were contingent on Basel developments that did not happen, so we have to reconsider the situation – so, that’s life. That’s why we are going to issue another monitoring report and provide benchmarks, offering quantitative evidence of the range of current practices.

The guidelines have been put on hold, so we would rather stay at the stage of producing evidence – developing best-practice guidelines is not yet there.

We would not find it right if more than a third of the capital requirements were driven by non-risk-sensitive metrics – we want the risk-sensitive metrics to drive around two-thirds of the capital, but that’s in terms of constraints

Isabelle Vaillant, EBA

When will the monitoring report come out?

IV: I think just after summer.

What were your recommendations contingent on?

IV: It’s the same story as with market risk. The models being developed were very young and – collectively, at the global table – we couldn’t conclude we should rely on what exists today, so then you need to have a more standardised approach. Maybe in a few more years it will have evolved. But we couldn’t find something sufficiently stable – and convergent – across the planet, which would allow us to build a rule across it.

So that leaves the standardised and basic approaches?

IV: We have to wait for the final conclusion, but probably the idea will be to have the standardised and basic approach.

Do you still believe there should be CVA charges for corporates?

IV: Yes. Again, this is one of those cases where the prudential objective goes against a more political objective, with the view that if you charge a risk, you make things tougher for the economy – that’s the basic issue.

In our view, the risk has not disappeared because you decide to have a less costly requirement – the risk is there. Our task is to make people aware of what is the size of this risk, the amount, and then if by convention the capital charge is not there, banks still have the opportunity to cover for this if they think it’s correct.

So, your aim with the monitoring report is to give supervisors the data to impose their own add-ons?

IV: It’s not aimed at the national supervisors, because we are in a legal setting where you can’t have this. This is why we interpreted there could be room for a Pillar 2 solution, even though this was disputed by some – in particular, members of the European Parliament. We never issued the guidelines, so in the end it means it’s only for the bank to decide.

I think, if we are transparent in producing benchmarks, at least an outlier bank will think twice – if I was on the board of a bank, I would think twice about this situation.

How has the internal ratings-based (IRB) framework changed since the March 2016 proposal?

IV: The idea was to remove variability by reducing modelling scope. We call this a top-down approach. Here in Europe we’ve been of the view that models should be repaired on a bottom-up basis.

What we’ve sought to do at EBA is make sure the bottom-up and top-down approaches can match at some point. Our main stance – which we have tested continuously – is that non-risk-sensitive items should be a non-dominant part of the framework. That means components of regulation such as an output floor and the leverage ratio. And by ‘non-dominant’, we mean roughly a third of the capital requirements should be driven by those non-risk-sensitive metrics, while the rest should be the dominant part, at least 60%. This has been our key goal.

Initially, there was a wish to exclude institutions from IRB models. From an EU perspective, we pointed out that there are a large number of banks and – as we’ve seen – there can be defaults, so you can think of modelling this

Isabelle Vaillant, EBA

So, when you apply that principle to the IRB debate, what are the policy outcomes? We’ve heard the threshold for modelling corporate exposures has been raised – meaning more exposures are in scope.

IV: We were working on this last year. In the end, we went for the idea of striking a balance via the size of the corporate. Then you negotiate with your peers and, in the end, you find a figure. It’s never going to be perfect, but you need to find consensus.

Is that figure higher than the original?

IV: Yes, we made progress.

We also heard it will be possible to model default probabilities for banks and other financial institutions. Is that right?

IV: Initially, there was a wish to exclude institutions from IRB models. From an EU perspective, we pointed out that there are a large number of banks and – as we’ve seen – there can be defaults, so you can think of modelling this. Again, we got progress at the global table. But as we all know, this is contingent on the final conclusion, which is itself contingent on the output floor.

On Brexit, has there been any talk about the capital impact if the UK loses equivalence to the EU CRD?

IV: Frankly, no. The EU wanted to be extremely disciplined about taking this step by step. Article 50 has only just been activated, so we need to see where the discussion on the content will lead.

I think we can praise the EC for its communication giving us a picture of the provisions under review for equivalence when the time comes. In banking, this is different from other sectors; there are various provisions, but limited areas to pronounce equivalence about. I think it’s also in the European supervisory authorities’ [ESAs] review consultation paper that the ESAs should be involved in this – perhaps more involved – but I suspect this will be set up after spring.

What is the right outcome in the ongoing Basel talks? You said one-third of capital should come from non-sensitive pieces …

IV: To be more precise, it’s a notion of constraint. We would not find it right if more than a third of the capital requirements were driven by non-risk-sensitive metrics – we want the risk-sensitive metrics to drive around two-thirds of the capital, but that’s in terms of constraints. This has been our guiding principle, but I can’t guarantee it will be the final outcome.

Is this based on EBA analysis?

IV: This was EBA analysis that was provided to the European supervisory community, including the one negotiating in Basel. It was the consensus view.

So when people talk about setting the floor at 60% or at 70%, the argument for you is slightly different?

IV: There are two logics. On the output floor, the first rationale is to measure the gap between the standardised and internal models approach. With this in mind – and I suspect this is the US approach – the gap between the two should be rather small.

But you also have another understanding of the output floor, which is that this distance exists because the calibration of the standardised and internal models approach reflects your actual risks. That means structural differences in risk may produce a greater distance in one jurisdiction than another, which happens to be the case for segments like mortgages, which are typically structured differently – here, you can have a larger distance between the two, and a lower output floor. This is the whole debate we are having now.

What does the MPS case say about BRRD?

IV: This is the perfect illustration that what was in BRRD can be implemented, despite all sorts of comments that it would never be applied. On the contrary, this is a very good illustration of how it can be applied, using the underlying rationale of all the provisions – precautionary measures and recapitalisation are part of the menu.

What if precautionary recapitalisation is the only approach ever used?

IV: You are suggesting bail-in may never be used? Well, bail-in is only just implemented. It’s a very young framework, not all of the banks have all their plans and all their eligible liabilities in place. So it’s difficult to draw lessons on that.

But it’s always going to be young until someone goes first, isn’t it?

IV: Well, when banks have their plans in place and their eligible liabilities settled, there will also be clarity on how the tool can be used in a safe manner.

Once everything is in place, do you expect some brave resolution authority to eventually say ‘We’re not going to jump in, we’re going to let this run its course and allow bondholders to take the pain’?

IV: The provisions are there. The eligible liabilities are not all set up – there’s a little bit of work to do so people have the confidence to use all the tools, also regarding the interaction of the total loss-absorbing capacity requirement with capital – this is still needed. I think Basel and the EU are working on that this year, so I think that will complete the picture.