We partially agree with the proposal. However, in practice we regularly encounter defaults that, although technically in default, are not caused by a credit risk event. One could argue that these defaults should therefore not be taken into account in credit risk internal models and subsequent credit capital requirements. We do take note of EBA’s stance that banks should be able to solve these issues within 90 days. However, in practice this is not always possible. For instance, additional guidance is sought on how to deal with disputes of payments between the bank and a borrower. For example, within leasing, the borrower and the bank may have a dispute on to the quality of the leasing object. The dispute may cause the borrower to default based on the 90 days past due trigger.
We are of the opinion that only defaults that are caused by credit risk events should be captured. Therefore, we request EBA to reconsider the definition of technical defaults to exclude these cases from the default portfolio.
Some more clarification might be needed on the exact definition of “full transfer of risks and benefits of the ceded receivables to the factor”. Generally, in non-recourse factoring transactions there may always be a certain recourse element left, mainly concerning the dilution risk. Notwithstanding these recourse elements, auditors in general accept such structures as off-balance for the client.
The difficult part at present is what the implications of these requirements could be. In many non-recourse factoring transactions the servicing of the underlying receivables remains with the seller which means that the factor would not be able to identify a default of an underlying receivable.
We agree with the approach for the treatment of specific credit risk adjustments and welcome the included harmonisation between the frameworks. We would like to note, however, that ‘Stage 3’ is not formally defined in IFRS 9. To reduce the chance of discrepancies in interpretation, we kindly propose to remove the references to Stage 3 and solely refer to credit-impaired instead.
Also, additional guidance is sought on the treatment of ‘purchased or originated credit-impaired financial assets’ as defined under IFRS 9, under the proposed Guidelines. One could argue that these assets remain in Stage 3 as a credit-impaired asset for its remaining lifetime, as a lifetime expected loss will be recognised throughout the remaining lifetime of the asset. This raises the question if ‘purchased or originated credit-impaired financial assets’ should remain classified as defaulted until full maturity, i.e. would not be able to cure regardless of the client’s payment behaviour. In our view, this could entail input data in LGD models that do not reflect true economic reality, which could potentially lead to LGD models that might be less useful for internal risk management and decision making within the bank. Therefore, we kindly propose to exclude ‘purchased or originated credit-impaired financial assets’ from the criteria that impaired exposure should be classified as defaulted.
We would also like to point out to EBA that IFRS 9 requires banks to apply a ´symmetrical´ approach to recognise impairment , i.e. favourable changes in credit risk should be considered consistently with unfavourable changes in credit risk. This principle is consistent with the current CRR requirements, as an exposure can be considered cured as soon as “no trigger of default continues to apply” . However, the proposed Guidelines would eliminate this alignment, due to the introduction of minimum probation periods in the Guidelines. Although we recognise that the accounting and regulatory framework should be considered separately, we are of the opinion that the elimination of this alignment could be considered a fundamental choice that should therefore be substantiated as such. An reduced alignment between the accounting and regulatory frameworks would entail a less transparent to investors and analysts.
During periods of stress the selling price of credit obligations tend to decrease due to the lack of appetite from investors. In order to limit the amount of defaults to be taking into account in internal models in a period of already increased capital demands caused by decreasing credit quality of the overall portfolio and economy, banks may become hesitant to sell credit obligations that include a decrease in credit quality (but have not defaulted yet). This may subsequently reduce flexibility in balance sheet management and could lead to a reduced credit supply to the real economy during periods of stress. This potentially has a pro-cyclical effect on the real economy.
In addition, we have concerns of the level of the threshold. A 5% threshold might be easily breached due to a minor deterioration in the external rating or change in interest rates. We kindly propose to increase the level of the threshold to a more suitable level.
We agree that using the original effective interest rate would be most useful to discount the cash flows, as this is also used to assess the modification with IFRS 9. We are also of the opinion that the use of the original effective interest rate allows for a “fair” comparison between the expectations at origination and the expectations after the restructuring.
In addition, we have some concerns on the level of the threshold against which the diminished financial obligation should be measured. We kindly request EBA to reconsider the threshold, taking into account the results of the Quantitative Impact Study.
We do not agree. Discounting on the face value of a financial asset may have numerous reasons outside the underlying credit worthiness:
• Some markets may be more shallow and selling parties may be driven by wishes of diversification, strategic re-orientation or liquidity demands (e.g. 2008 crisis) in selling off at a discount.
• Differences in assessment of pre-payment risks, liquidity risks, transfer risks between the buyer and seller may also lead to offering material discounts.
Classifying assets bought at a discount due to one of both reasons outlined above as defaulted would lead to defaults that are not linked with the actual ability to pay of the counterparty. An alternative may be to consider adopting the criteria of “Purchased or originated credit-impaired financial assets” under IFRS 9 as unlikely to pay triggers. This would also increase harmonisation between both frameworks.
We would also like to mention that the requirement to classify credit frauds as default could, within some banks, lead to double counting between credit- and operational risk. In addition, one could argue that, for instance, identity thefts are not representative to the actual risk present within the portfolio and should therefore be excluded from the defaults used for IRB modelling.
We understand that, although banks are required to monitor the re-defaults of the portfolio, minimum probation periods are prescribed by EBA. Banks are expected to revise their internal policies if the number of re-defaults is not limited. This could lead to a longer probation period than the absolute minimum set by the Guidelines.
The Guidelines, however, do not prescribe what a limited number of re-defaults would actually be. This could lead to an inconsistency in supervisory practise, as the limited number of re-defaults would be interpreted differently by the various European supervisors. Therefore, we kindly propose to set a maximum relative amount of re-defaults instead of a minimum probation period. Banks would be required to increase the probation periods until the re-defaults are below the threshold. This would ensure a more risk-sensitive default definition, at the same time increasing harmonisation as all banks would use the same ‘re-default threshold’. In addition, banks would have an additional incentive to provide sufficient aftercare for cured clients.
We would also like to point out to EBA the inconsistency between the probation periods and CRR article 178 (5), which states that an exposure should be reclassified to a non-default status as soon as the default trigger no longer applies. The application of minimum probation periods would therefore, in our view, require a change in the CRR, most notably with regard to the probation period for non-distressed restructured defaults.
The CRR uses “more than 90 days past due”, while the Guidelines uses a minimum 3 month probation period as set out in paragraph 58. For the sake of simplicity, we would like to propose to align the two definitions to three months.
We agree with the option of banks to determine the level of application for retail exposure to best fit internal risk management practices. However, we would like to note that a further clarification of ‘retail’ with regard to the level of application could deviate from the internal segmentation that is used for internal risk management. This could lead to a number of process and implementation challenges.
Paragraph 80 of the draft Guidelines states that banks may consider it unlikely that all obligations of an obligor will be paid in full where a significant part of the exposure of that obligor is in default. Could EBA please confirm that this rule is optional for banks to adopt for retail portfolios where the level of application is at facility level, similar to the “other indications of unlikeliness to pay” listed in paragraphs 46 to 48?
Yes, we agree with the proposed approach for the application of materiality threshold to joint credit obligations.
Yes, we agree with the requirements on internal governance as this is already in line with the CRR.
We would, however, like to note that banks often use a default definition that best fits its internal risk management and decision making process, within regulatory requirements. This is also necessary to comply with the requirements as laid down by CRR article 144. We therefore appreciate the opportunity to use different definitions of default to best fit internal risk management and decision making process, as provided by paragraph 69 of the draft Guidelines.