The Committee of European Banking Supervisors has released a position paper on counter-cyclical capital buffers, favouring discretionary supervision (Pillar 2) over Capital Rules (Pillar 1):
While the mechanisms identified might be alternatively employed in Pillar 1, its use under the Pillar 2 umbrella is still considered the most sensible option at this stage. Pillar 2 allows for flexibility in testing new prudential tools; moreover, an application in Pillar 1 would require further work and refinements.
…
With regard to this last point a meeting with rating agencies was organized. They stated very clearly that transparency on capital adequacy is a key issue and it is a precondition for market acceptance of time-varying capital buffers. Rating agencies seem to prefer Pillar 1 solutions, considered more transparent [and] less prone to national discretions; however, they seem also aware that Pillar 2 would allow quicker responses and may be used for testing tools to be subsequently improved and, possibly, implemented under Pillar 1.
I suggest it’s not a matter of awareness: it’s a matter of trust. In Canada, of course, we have OSFI with its demonstrated willingness to short-circuit Pillar 1 on the basis of a panicky ‘phone call, as well as contemptuous opacity towards the concerns of investors (Pillar 3).
Essentially, the position paper aims at a different methodology for calculating Expected Losses (EL) – see Expected Losses and the Assets to Capital Multiple. EL is calculated by the formula
EL = PD * EAD * LGD
where PD = Probability of Default
EAD = Exposure at Default
LGD = Loss Given Default (a percentage)
What C-EBS is aiming at is:
the use of mechanisms that rescale probabilities of default (PDs) estimated by banks, in order to incorporate recessionary conditions.
Currently:
The input to the IRB formula is the annual PD expected to be incurred in that grade (computed as the long-run average of one-year default rates).
…
As for the LGD, banks are requested to use LGD estimates that are as much as possible estimated for an economic downturn (where these are more conservative than the long-run average).
One problem I see with the approach is there does not appear to be an allowance for the term of the exposure. Would a bank dealing exclusively in mortgages with a 5-year term be expected to use the same recessionary PD as a bank with a portfolio of exclusively 30-year mortgages?