Risk-sensitivity, simplicity, comparability – these are the three attributes regulators are trying to balance in an ongoing review of the bank capital framework. Balance is the operative word, because it's generally thought impossible to promote all three of them, to the same extent, at the same time. Clearly, if bank capital is very risk-sensitive, it will not be simple.
Comparability, though, has a slightly more slippery relationship with the other attributes. Could bank capital be simple and comparable? Yes. Could it be risk-sensitive and comparable? Yes, although it might be more difficult. And while the other two attributes are inherently desirable – risk-sensitivity is a good thing, as is simplicity – comparability is not necessarily helpful to anyone if the things being compared are false or if the comparison yields no useful information. In fact, it could be harmful.
As part of their efforts to strike a balance, regulators plan to give standardised approaches a key role – these fixed formulas have always been seen as the poor relation of internal capital models, but are now being revamped to make them suitable for a new, expanded function.
The idea is for banks that have modelling approval to calculate standardised capital numbers as well. Both sets of numbers will be disclosed alongside each other; the modelled numbers will also be floored at some percentage of the appropriate standardised approach.
It has a lot to recommend it. The disclosures, in tandem with the floors, will promote the virtues of simplicity and comparability; risk-sensitivity will suffer – depending on precisely where the floors are struck – but the overhaul of the standardised approaches is intended to make them more sophisticated.
It certainly makes it easier to compare Bank A and Bank B, but what does it say about the actual level of risk each bank is running?
Despite that, the scheme is catching a lot of – predictable – flak. Modelling banks fear the loss of risk-sensitivity and a possible jump in capital levels. Standardised banks say the new approaches are too complex.
The more awkward questions, again, arise when considering comparability. If the capital numbers produced by the standardised approach for market risk really are 13 times higher than those obtained by internal models – as a study of unpublished data from the latest impact study suggests – what will that tell analysts and investors? It certainly makes it easier to compare Bank A and Bank B, but what does it say about the actual level of risk each bank is running? The conclusion, when confronted with that kind of gulf, is that both numbers are wrong, and the truth lies somewhere inbetween. But where?
Regulators should be applauded for trying to fix the problems with the capital regime, but there is a risk the results will undermine, rather than strengthen, the framework.