The proposed changes to the Basle Accord include a major overhaul of the regulatory capital framework, allowing some banks the option of calculating their own capital in-house. In this interview, ERisk talks to Ashish Dev, senior vice president of risk management at KeyBank, one of the large US banks that has been testing the impact of the current proposals on capital levels.

Can you tell me what Key has been doing with regard to testing the impact of the current proposals on the corporate credit portfolio?

We have been involved in different working groups, coordinating industry and individual bank responses to the consultative document. Through the auspices of RMA, we are filling out a questionnaire devised by the regulators on the quantitative impact of the proposals, and most large US banks have been a part of that, as well as a few non-bank organisations such as Providian. The idea is to derive some general conclusions and look for inconsistencies concerning the effects of the standardised, foundation IRB and advanced IRB on the regulatory capital of banking institutions at large.

What are the main messages so far?

One very major point to have surfaced is that the vast majority of commercial banks in the US, probably excluding only the few money centre banks, have a corporate portfolio that includes a large percentage below BBB, below investment grade. That's true of us as well, because most of our corporate credits are in the middle market – both upper and lower middle market – which tends to be below BBB.

Typically anywhere from 35 per cent up to 60 per cent of the loans in the corporate portfolio will have a probability of default corresponding to BBB or below, and there are community banks with up to 90 per cent of their loans BBB-equivalent, because they are often dealing with very small customers who are definitely not investment grade.

For these portfolios, the foundation IRB is likely to require more regulatory capital than the standardised method. There will be little incentive to go to the foundation IRB. At the same time, it may not be possible to satisfy the conditions for adopting the advanced IRB. It appears that the banking industry as a whole in the US will end up keeping higher capital under the new standardised method than under the current one-size-fits-all accord – and that’s not just because there is approximately 20 per cent extra capital required for operational risk.

Why is BBB such a significant grade?

BBB marks the transition from investment grade to non-investment grade. The probabilities of default for investment grade credits are very low. In fact, they are much lower than 0.7 per cent. When you plug a probability of default of 0.7 per cent into the IRB formula you end up with a risk weight of 100 per cent. If you use anything lower you end up with risk weights of less than one, anything greater and you end up with risk weights greater than one.

It’s also important to note the shape of the default curve. As you move away from investment grade ratings, the probability of default increases substantially at a growing rate. Not only do you end up with a risk weight greater than one, you end up with risk weights significantly greater than one.

In summary, in the standard method, credits rated BBB+ to BB– are risk weighted 100 per cent. Those below BB– are weighted 150 per cent. In the IRB method, credits rated BB+, BB, or BB– will end up with risk weighting well over 100 per cent. Those rated below BB– will end up with risk weights well above 150 per cent. Clearly banks whose portfolios contain more than a trivial percentage of non-investment grade assets will not have an incentive to use the IRB method.

What part of the current proposals is driving the capital up for the foundation IRB?

The foundation IRB assumes a loss given default (LGD) of 50 per cent, and that's primarily why the capital is so high. The 50 per cent LGD figure is stated to be for unsecured corporates – it also states that the figures could be as low as 40 per cent if secured by residential real estate. While 50 per cent may be justified for unsecured credit, it is not justified for secured credit; if the portfolio is secured by something, there should be some relief.

To illustrate, internal capital allocation models would be using something like 35 per cent LGD for a secured portfolio – and I’m not talking just about commercial real estate. A commercial real estate portfolio could have LGDs as low as 25 per cent. That’s typically what most people are assuming.

What will the group’s message on the 50 per cent LGD assumption be?

We are likely to say that 50 per cent for unsecured credit may be reasonable but we’ll also be saying that it is unreasonable for portfolios secured by properties, inventory or accounts receivable.

So if a bank is looking for capital relief, and could move to advanced IRB, that option would make the most sense?

Yes, for corporates. It seems that, for most banks, the foundation IRB would not make much sense because it may not mean much, or any relief, over the standardised method.

Would that always be the case? Could you achieve a capital relief incentive if you had a large proportion of your portfolio in highly rated credits, and a similar proportion in BBB and lower?

Possibly. If you suppose a portfolio that has 20 per cent in highly rated credits – AA types, and then 40 per cent in BBB and then the remaining 40 per cent in below BBB – you would not benefit from the foundation IRB. On the other hand, if you had 60 per cent in A and AAA and another 20 per cent in BBB and only 20 per cent in below BBB, you would benefit.

Do you think it’s likely that the formula for the foundation IRB will be changed – and what do you think the options are?

For a bank like us, one option is to go to the advanced IRB, but that would depend on what happens in retail. You have to choose the same approach for both corporates and retail, and we don’t know for certain where the retail IRB will come out.

So it’s conceivable that, even though using the foundation IRB for the corporate portfolio may increase the capital you hold, a sufficiently beneficial approach to retail could still result in a net reduction in capital?

Yes. And it could work the other way round, too. A bank with a large retail portfolio and a large high-quality, pristine corporate portfolio, may have to think twice about the approach it takes, because while you may get a benefit on the corporate side, you may also miss out on the retail side. But that squarely depends on how the new Accord treats retail and now we’re beginning to look at the retail treatment in more depth, but the picture is still somewhat clouded. The regulators have made clear time and again that the formula in the consultative paper is simply a placeholder, a starting-point for discussion. It’s also clouded by the fact that the starting-point given by the paper is the equivalent of the advanced IRB for corporates – there is currently no standardised approach.

Why not?

As we understand it, a more comprehensive treatment will be arrived at following the comment period and during the ongoing consultative process. In the 1999 proposal discussion of retail was kept to a simple ratio-based approach, with a 100 per cent weight for everything, except mortgages, which got a 20 per cent weight. In the current document, there is a mention of a standardised approach, but it is not spelt out, so since February, we’ve really been pushing everyone to address the retail IRB.

The internal rating risk weights suggested in the proposals – when they’ve been resolved or decided upon – will be used to drive decisions about the standardised approach as well as a foundation IRB approach. Compared with the corporate piece, it will be a case of working backwards.

That could be a good thing for us. It would almost seem logical to assume that if you’re knowingly setting your standardised approach based upon the advanced IRB, you would expect incentives across the board – I think it will be a lot easier to work backward when building incentives, and I’d be surprised to see any disincentives in the way it’s being constructed … unlike the current iteration for corporates.

What do the retail proposals look like at the moment?

We’re in the process of doing calculations so we can’t offer a fully cooked meal, and we also know that the proposed figures are just illustrative, but if the current formula prevails, a prime auto portfolio could get capital of between 10–12 per cent. We would consider that as far too high even compared with the 8 per cent capital we currently use, and compared also with what is happening in the corporate world. But we have to take it with a grain of salt because they’ve said that formula is only illustrative.