Yes, partly we agree. However, we have several implementation questions:
We agree that non-credit risk related technical defaults should be solved within 90 days. However, in some cases this is simply not possible. These technical defaults should not blur the data set of internal models, nor the capital calculations.
In any case technical defaults (beyond 90 days past due without a credit risk link and eventually without a loss), should not be taken into account in the internal model data sets, non-credit risk related defaults should not lead to an increase of credit risk related capital calculations.
If these non-credit risk related technical defaults would be taken into account in the data set of internal models, due to LGD floors, the increase in PD often cannot fully be compensated by lower LGD values (LGD values will go down due to higher cure rates). This might become even more relevant once BCBS decides to increase input floors on LGD.
We assume that technical defaults as defined above do not need to be registered into a central (risk) data base, but should be registered at least in the local data base, in order proper checks can take place (audit trail).
The audit trail of technical defaults (that did not lead to actual defaults) should only be required for new cases.
We welcome further harmonization of the accounting and regulatory frameworks. Unnecessary difference between the two frameworks in terms of credit impaired and defaulted might confuse the investors.
On the other hand, regulatory rules should never be influenced by interpretation changes of the accounting framework. Therefore the CRR text should have no text references (meaning: opposite to “copying the text”) to the accounting framework.
We seek additional guidance on the purchased or originated credit impaired financial assets, as there might be no credit justification to keep it defaulted for the remaining lifetime. The accounting of these items do not follow the general IFRS 9 requirements for recognizing a loss allowance (i.e. do not follow the three stages approach as referred to in the consultation document). If these cases would need to stay in default until maturity the economic rational of the LGD input would be missing.
No, we do not consider the proposed treatment of the sale of credit obligations appropriate for the purpose of identification of default.
If an asset would deteriorate from AA to BB, the asset is still quite remote from default, but could be sold for a discount (larger 5%) due to credit deterioration, this should not trigger a default (for IRB modelling).
In most cases the sale of (a group of) asset(s) will partly be credit risk related and partly be motivated by portfolio (concentrations) or business strategy considerations. It could be very difficult to split all these components (or isolate credit risk components).
Limiting the freedom to sell (without taking additional default in your modelling, although the assets are not near default) would make the financial market less liquid.
All in all, we argue that the 5% threshold should be increased significantly (to facilitate the cases where obligors deteriorate without entering into default e.g. from AA to BB).
We are of the view that the appropriate interest rate should be the agreed interest rate applicable at the moment of default, which normally is the customer’s original effective interest rate . The restructuring agreements can be signed well after the default moment, so this is hard to use as from the moment of default.
However, penalty interest (if part of the agreed interest rate clauses applicable at the moment of default) should be excluded from the expected cash flows based on the new agreement, as these has less or no economic value. The main purpose of such a rate is to have a stronger opening position in the negotiation process with the defaulted client.
We seek additional guidance on the interest rate to be used for purchased or originated credit impaired financial assets (i.e. the credit-adjusted effective interest rate?).
We do not agree,
Asymmetry between the buying and the selling of assets is not required, as the risk of (intentionally) influencing the data set used for IRB modelling (selling leg), is not present at the buying side.
For every asset that you purchase (with or without material discount) a proper due diligence should be made, also along the lines of 90 days and the unlikelihood to pay assessment. At the moment an asset or a group of assets is purchased the normal rules should apply. That is days past due or the unlikelihood to pay (the amount that the buying party took on their books, can well be lower than the face value). It can well be that the seller originally started the loan to a client when the client was rated AA. In the meantime the client is downgraded to BB. Now the seller wishes to sell the assets for a discount, the risks has increased, however, there are no signs of a default.
The assets can have fixed interest rates, while the floating rates went up. Buying such an asset will be done with a discount, without a deterioration of the credit profile.
Also, there can be all kinds of strategic or business reasons to sell assets against a loss, which do not need to have a relation to a possible default of the transferred assets.
The suggested EBA guidelines could lead to banks not selling these assets against a discount, and therefore other banks not buying these assets at all, which has a clear negative effect on the important intermediary role that the banking industry should fulfil in favor of their clients.
Market sentiment will influence the level of discount, which is sometimes difficult to separate from credit deteriorations (or credit risk changes).
Most important, the buyer is paying a price that the buyer at least expects to get back from the client. Therefore it makes no sense putting such an exposure in default from the start. If the buyer expects a lower yield, it would have paid less in the first place.
It is preferred to align the definitions w.r.t. defaults, (non-)performing and impairment where this is possible. It is not preferred to apply several comparable but different definitions, which will become confusing.
The proposal in this consultation document is not aligned with the probation periods suggested in the EBA reporting paper (w.r.t. FINREP). In that paper only a 12 month probation period was proposed for distressed restructuring. Including a 90 day probation period seems too conservative for arrears which were 90 days past due and repaid fully immediately afterwards.
In addition it also is not consistent with article 178 paragraph 5 where it states that in case no trigger of default continues to apply, the institution shall rate the obligor/facility as they would for non-defaulted exposure. This in fact does not even allow room for a probation period.
If EBA wishes to minimize the re-defaults, by stretching the probation period, EBA is kindly asked to articulate what this minimum excepted amount of re-defaults is. Then, a sound probation period decision could be made.
We agree with the option given to the institutions as long as it reflects the institutions internal risk management practice.
A potential issue might arise when the default definition for a retail SME client is based on obligor level, while the owner of the retail SME company – which provided a personal guarantee – also has a personal credit card line, for which the default definition is based on facility level. For these cases the individual institution should convince their supervisor through the ICAAP / SREP process that their practice is sound and consistently applied.
We agree, as this questions relate to Retail, the answer is in line with the answer given to question 8: As the EBA document clearly states “may” it is a choice made by the institutions. Dutch institutions do not consider this.
This should be considered as a rebuttable trigger. With respect to retail exposures there are various examples, where for one facility the payments are according to schedule, whereas the other is in default.
Yes, we agree to the proposed application of materiality threshold, as it is proposed that this would be rebuttable trigger.
Yes, we agree. The requirements are in line with the current CRR.
We fear that given all the possible (Basel) changes to IRB, including all kinds of possible input (PD and LGD floors) or output floors (SA Capital Floors), supervisory pressure in the model approval process and additional levels of conservatism, the Basel Use Test will become under extreme pressure. It will become difficult to evidence that institutions fulfil to the Use Test.