Yes, the specified three main categories are a clear and logical way of organising the matters that should input into the NCA decisions.
Yes, the conditions, and the range of possible adjustments, are sensible. But the NCA should be required to publish, with appropriate supporting evidence, the matters which it has determined.
The explanation of the alternative lines of argument derived from other provisions of CRR – set out in the CBA – is clear. We think the first argument (that maximum losses of 2.8%, corresponding to a loss expectation of 1.5%, are covered by the 35% RW) is by far the more convincing, and the result is more consistent with CRR RWs generally.
Moreover, we are not sure if the second argument is correctly applied anyway – 0.3% (derived from Article 125) is stated as the benchmark for increasing risk weights above 100%, when in fact the effect of losses exceeding 0.3% is that the RW for residential mortgage loans that do not satisfy the independence criterion in Art. 125.2(a) moves from 35% to 100%. So 0.3% is in fact the threshold - in the Article 125 context – for increasing the RW above 35%. In that case, the EBA should perhaps have stated the lower bound of its indicative range as 0.3%, not 0.10%.
However, these thresholds can never be a matter of exact science. Since the available time series of loss data may also be short, we think (independently of the two arguments above ) to begin with at least the benchmark should anyway be set at the upper bounds of the indicative range – i.e. for residential mortgages – loss expectation of up to 1.5% (corresponding to maximum losses of 2.8%) is adequately catered for by 35% RW, loss expectation above 1.5% but no more than say 5% catered for by higher RWs between 35% and 100%, and only where loss expectation exceeds say 5% should the RWs be set in the top range of 100% to 150%. (It may also be necessary to define loss expectation more specifically.)
The other reason for setting the benchmark at the upper end of the range consulted on is to mitigate, or at least prevent aggravation, of the general problem that the 35% RW already tends to overestimate credit risk in conventional residential mortgage lending, as explained above.
Broadly, yes – the specification of financial stability considerations covers the right ground, and we particularly support the need to take account of procyclicality. But we caution against the following unintended effect: if one G-SIB, or several smaller banks, in a particular member state mismanage lending of a particular type and thereby incur serious losses, the right remedy might be to increase Pillar 2 add-ons for these banks (only), rather than increase Pillar 1 RWs for all credit institutions, including those who have managed the same types of lending successfully. We suggest the EBA includes a provision in these RTS to make clear that Pillar 1 RWs should not be increased to deal with idiosyncratic issues at one, or a few, banks, when the right tool is Pillar 2.
Again, for these reasons, transparency is paramount – unless the NCAs publish their evidence and reasoning, it will be difficult to challenge what may be the wrong use of this Pillar 1 tool, skewed by one or two hard cases.
Yes, these conditions are broadly fine. The NCA should have to publish whatever it has determined.
As previously indicated, the requirement for NCAs to consider ‘the expected evolution in immovable property market prices and the expected volatility in those prices…’ and ‘the time horizon over which the forward-looking property market developments are expected to materialise…’ should be accompanied by a requirement for the NCAs to make explanations public as to the reasoning behind their views on the future direction and timing of property price movements. Without such explanations it will be difficult to assess the level of conservatism being applied in the NCAs’ thinking and avoid suspicion of over-conservatism and subjectivity.
The logic for not setting indicative benchmarks for higher minimum LGD values appears sound.