According to ESBG’s experience, ‘technical defaults’ are usually corrected within a few days after they occurred. Thus, they will not last for 90 days or more.
Option 2 is less strict and – as the EBA stated in its draft RTS – better identifies technical defaults than option 1. Therefore, ESBG prefers option 2.
Furthermore, we would like to ask the EBA for a clarification with regard to option 1. According to Art. 2 draft delegated Regulation, which is based on Art. 178 CRR, reference is made to an ‘absolute’ and a ‘relative’ component. While the ‘absolute’ component (Art. 2(2)(a) draft delegated Regulation) is very clearly worded, we have one doubt concerning the ‘relative’ component (Art 2(2)(b) draft delegated Regulation). ESBG would very much appreciate if the EBA could clarify whether the 90-days (or 180-days) threshold also applies to the ‘relative’ component. We assume that for both the ‘absolute’ and the ‘relative’ components the 90-days (or 180-days) limit needs to be fulfilled in order to consider an obligor as defaulted. However, the EBA’s proposal is not very clear in this respect.
Due to the fact that only very little data is available on this aspect it seems at present hardly possible to us to precisely assess suitable maximum levels of the thresholds.
However, we cannot agree with the proposed maximum level of the thresholds. Several ESBG members have implemented well-calibrated thresholds which are different from the proposed ones. The possible process of implementing the necessary adjustments (data, risk parameters) would be fairly costly, time consuming and not necessarily justified by the benefits. Therefore, ESBG would rather advocate maintaining the existing thresholds (Please see also the answer to question 4).
First of all, ESBG would like to hold that the start of the implementation period needs to be aligned with the definition of the applicable materiality threshold by respective NCAs, not with the finalisation of the RTS.
In ESBG’s opinion, an implementation period of less than two years is not feasible due to the reason that the implementation process is expected to be very complex. In fact, there are several factors relevant to the implementation period which can only be more precisely determined after both the RTS are finalised and the thresholds by the NCAs are adopted.
Re-developing internal models usually takes around four years. Therefore, if adjusting IRB models is required, ESBG would ask for a sequential adjustment of models over a period of four years.
ESBG agrees with the assessment of costs for IRB institutions, but does not agree with the assessment of benefits.
With regard to the costs, it is our estimation that the implementation of the necessary adjustments will indeed be very costly and time consuming. These costs will be significant and, in our view, not entirely justified by the stated benefits.
In respect of the benefits, we would like to state the following two reflections:
• Greater comparability of own funds requirements: ESBG believes that the reasons of poor comparability of own funds requirements are to be found well beyond the unified definition of the materiality of default. This is also documented by a BCBS study on the ‘Regulatory consistency assessment programme (RCAP) – Analysis of risk-weighted assets for credit risk in the banking book’ (July 2013). In our opinion, introducing unified thresholds will only marginally improve comparability of own funds requirements for IRB institutions.
• The reduction in administrative and operational burden for cross-border institutions to comply with different regulatory frameworks in different Member States: No benefit can be seen for IRB banks that have already implemented own materiality thresholds and have already borne the costs of the implementation. On the contrary, changing the thresholds would impose significant additional operational burdens. Additionally, the initial intention of the different thresholds was to take into consideration the specific situations in the different markets (e.g. average consumer loan volumes in different Member States). This flexibility should be maintained.
Following option 1 instead of option 2 (see question 1) could lead to a significant increase in defaulted clients, which would not reflect very precisely the (risky) situation of the client or his financial difficulties. As a consequence, the PDs would be increased, the LGD decreased, and the increased cure rate would show that the default rating from those clients was not adequate.
At the same time it is true that, without simulating, it is rather difficult to predict what the effect in the long run will be on the RWA and risk-cost side.