Burying Basel

It is far too early to pronounce the bloated body of Basel 2 dead, but at least the patient is ailing. In the FT Harald Benink and George Kaufman have some suggestions:

First, we urge the Basel committee to conduct another quantitative impact study using observations from the recent turmoil before allowing banks to use their internal models for calculating regulatory capital.

That's a good idea, but remember that Basel 2 is not based on bank's internal models. It is based on bank's ratings, which then go into the supervisor's formula. It is the supervisors who are at fault for any capital underestimate, not the banks.

Second, we advocate the additional adoption of a meaningful non risk-weighted leverage ratio requirement, as currently applicable in the US, to supplement Basel II risk-weighted capital requirements.

There is a (tiny) constraint already in Basel 2, in that PDs have a 0.03% floor. But the suggestion is a good one - a crude leverage ceiling would act as an additional control.

Third, we recommend that the Basel II approach using banks' own risk models should be complemented by a credible and effective form of market discipline.

I remain unconvinced that this will help much. It can't hurt, but I'm afraid that greed will outweigh fear most of the time despite the reasons to be scared that might be known.

Let me add in a few more ideas.

Fix the procyclicality of Basel 2. A risk sensitive Accord is necessarily procyclical. There is no way around this. And procyclicality is likely to intensify the depth and intensity of asset price bubbles and recessions. Basel 2 is based on one year PDs. Instead it should be based on across the cycle PDs.

Fix the incentive for less advanced banks to take the worst risks. The menu of approaches in Basel 2 makes it more expensive for advanced banks to take high PD exposures vs. standardised approach banks. Unlike VAR, which gives a reduction in capital for pretty much any market risk portfolio, the IRB is only cheaper than the standardised approach for better quality assets. This creates a perverse incentive which should be addressed.

Fix the IRB formulae. The correlation assumptions are just wrong, as is the assumption of constant default correlation. Based on the suggested QIS, they should be rewritten with much more conservative default correlation assumptions.

Cap the benefit for getting assets off balance sheet either via securitisation or into a conduit. This preserves alignment of interest and again acts as a cap on leverage.