One of the questions we (probably too infrequently) get asked is around why banks have to hold eight percent of risk weighted assets as capital. This question is normally asked by people who are new to the Accords and is normally directed to the old hands. The risk weighted assets bit is easy to handle – normally a quick discussion of probability, classes of risk and other such areas causes an eyes glazing over effect and some nodding. The newbie then scuttles off to read a bit more.

The eight percent bit, though, is frequently left alone.

There is a good reason why it is normally glossed over. The reason is that the old hand to Basel knows exactly as much as the new person as to why the eight percent was chosen – no real idea. That’s right – there is no statistical or other reason why eight percent was chosen. It just seemed to be a good idea at the time. It was a number the regulators were happy with and the Banks were prepared to live with. There is also no science behind the four percent allocated to tier one capital – it is also a rule of thumb – AKA a guess.

The justification actually in the Accord is about as far as we can go. This is para. 44 of the original Accord:

In the light of consultations and preliminary testing of the framework, the Committee is agreed that a minimum standard should be set now which international banks generally will be expected to achieve. It is also agreed that this standard should be set at a level that is consistent with the objective of securing over time soundly-based and consistent capital ratios for all international banks. Accordingly, the Committee confirms that the target standard ratio of capital to weighted risk assets should be set at 8% (of which the core capital element will be at least 4%).

Over the eighteen years since the implementation of Basel I the number of bank failures in economies using Basel I has been fairly low, so, in hindsight, it has seemed to have been at least conservative (i.e. safe) and the very fact of measuring it has probably made a real difference.

There is, however, as far as I know, also little or no research on whether a lower number would protect almost as well without imposing high costs. If you know of some, please let us know.

Another possible subject for a PhD (or perhaps a DFin) thesis?

Advertisement