Response to consultation on draft Guidelines on IRRBB and CSRBB
Go back
We strongly disagree with the choice to introduce the five-year cap. Disallowing the modelling of NMDs severely constrains the actual IRRBB risk management of institutions, this would lead not only to results that are too conservative but also to potential inconsistencies, particularly where e.g. hedging of the positions stretches over a say 6 years horizon. In such cases it would in fact not be prudent to have the 5 years cap. We also strongly disagree with the exclusion of non-maturity deposits from financial customers from NMD. This prevents the models from mirroring reality and is not in line with BCBS 368. Moreover, the wording is not consistent and mentions a cap on the maturity itself rather than the average.
We also highlight the following:
The explanatory box clarifies that the 5-year cap applies to the full amount of NMDs – as opposed to just core deposits. Provided that we disagree with the cap, as noted above, such clarifications should be also included in the regulatory text.
While it is clear from the explanatory box that the 5-year cap applies to the combined volume of core and non-core deposits, it remains somewhat ambiguous if the 5-year cap would apply to
the aggregate NMD portfolio, or
to portfolios individually.
This should be clarified in the regulatory text. Moreover, we observe that it is not uncommon for retail transactional deposits portfolios to have modelled average maturities beyond 5 years – in line with the nature of this product. Hence, the 5-year cap apply could at most be applied to the aggregate portfolio of modelled non-maturing deposits and not on each portfolio individually. Alternatively, at least a higher cap should be considered for retail transactional deposits.
The criteria are comprehensive. However, we suggest adding further wording so that in the supervisory evaluation process authorities can duly consider the relevant proportionality aspects and allow simplifications and adjustments in line with the underlying, bank-specific risk materiality. Examples where such proportionality should be applied, are the assessment of elasticities in 109(b) which small, non-complex institutions are mostly not able to perform and the isolation of behavioral effects under 112(h), which is of very limited informative value.
We also suggest to indicate that, where a supervisor is of the opinion that an institution does not satisfactorily comply with all elements required for internal models, in particular 119b, the supervisor should start a dialogue with the institutions to restore compliance rather than immediately imposing the (S)SA.
Besides, we propose to streamline the text in some points and to avoid duplications (e.g. 154 vs 155 and 146 vs 149).
For the inclusion of fair value changes in the definition of the NII we see challenges arising from the diverging accounting definition of NII (as it is for example also reported in FINREP) in comparison to the NII definition for IRRBB purposes. We also point to the fact that the complexity for small institutions to measure those risks significantly increases as internal risk measurement and steering is often based on the narrow NII definition. Furthermore, the CRR and CRD mention NII in contrast to an Earnings perspective named as NII, which is being proposed by the EBA. We therefore strongly recommend that fair value effects continue to be considered under an earnings metric and that a materiality threshold can be established for the consideration and limitation of these effects.
Moreover, to cater for banks that are not accounting according to IFRS (e.g. in Germany, Austria), the requirements should be presented in such a way that they are meaningful for all institutions and the wording should be adjusted accordingly.
We agree with the fact that IRRBB from an NII perspective is an important risk that needs to be measured and reported. However, given that NII is part of the normative perspective, the individual limitation of this specific risk is not in line with the general normative assumptions and underlying management actions. Like other factors in the normative perspective, NII should only be assessed periodically under defined overarching scenarios and analyzed in the context of the entire normative requirement. We would therefore ask the EBA to align the definition of limits in the normative context with the new ICAAP requirements to allow banks to streamline their efforts and to prevent diverging steering implications.
Finally, we ask the regulator to specify in the Guidelines that banks cannot solely rely on the (simplified) standardized approaches but need internal measurement systems to fully reflect their bank-specific behavior. Standardized approaches for IRRBB are no alternative to adequate internal risk management and should not be used as a benchmark. We emphasize that there should be no incentive to neglect building and retaining internal competences in the context of interest rate risk management.
There is currently no industry standard on how to measure CSRBB. On the contrary, the existing approaches differ strongly. Hence, no industry standard should be "enforced" by the EBA. The specifications should be "modeling neutral", which is not the case. For example, there are credit risk models that are extended to include components for measuring CSRBB, which allow integrated measurement of spread changes based on changes of creditworthiness as well as changes in the market price for credit spreads. Especially with such integrated models, the proposed approach can quickly lead to double counting, which must be avoided. The existing models are all well understood and proven, and their use should continue to be possible.
Overall, guidance on instruments – other than fair valued ones – to be included in the calculation of CSRBB is unclear and may be interpreted differently by the banks. CSRBB is defined as relating to changes of market credit and market liquidity risk, yet the definition of market is undefined. A usual interpretation would be the one referring to instruments with a deep, liquid and active secondary market, yet this is not specifically stated.
The scope includes the liability side of the balance sheet but the GL does not provide guidance on potential netting from similar instruments on both sides of the balance sheet. Further, the consultation states that a deterioration of an institution’s credit quality should not have any positive impact on the credit spread risk measure. Notwithstanding idiosyncratic risk, some clarification is needed on whether a deterioration in own credit due to changes in general market credit spreads is in scope.
According to para. 157, idiosyncratic spread components can only be included in the CSRBB measurement if evidence is provided that the inclusion is conservative. The obligation to provide evidence should be deleted, since, depending on the reporting date considered, an inclusion can lead to an increase in risk at one time and a reduction in risk at another, due to the diversification effects that may be present in the case of an inclusion. Since any inclusion of additional components of the total credit spread risk tends to increase double counting, it should be considered conservative and allowed per se without the requirement for evidence. In contrast, a precise measurement of the individual components of credit spread risk (market spread, liquidity spread, separate consideration of idiosyncratic spread) may be theoretically desirable but it is not possible in practice. This corresponds with the associated explanation box (on page 46). Even if the spread components were to be separated, it is still questionable whether it would offer any added value at all and lead to meaningful steering input in the institution. For example, the default premium could already be included in the credit risk and the liquidity premium could, if necessary, be included in the institution's liquidity risk. Expected rating changes, i.e. expected migrations to other rating classes, can in turn affect the default premium and, if applicable, the liquidity premium.
Usually, credit spread data is sector and/or country specific and there is no general market credit spread curve for each rating class. If the final version still requires the use of a general curve, clarification is needed on how the general curve can be obtained. When considering credit spread risk, institution-specific circumstances (portfolio composition, design of internal systems, availability of reference data, etc.) should be taken into account. The specific design including integrated or isolated measurement should be up to the banks. For example, the analysis of spread components clustered by currency proposed in paragraph 123 is not appropriate in every case; on the other hand, other clusters (e.g., by rating class or sector or even a more granular, issuer-specific mapping) may become relevant. Here, it should be up to the institutions to choose an adequate type of clustering. In our view, this is the only way to achieve meaningful results and to derive sensible steering input in the institutions.
Backtesting does not always deliver meaningful results or is not possible at all. For example, for a customer loan - e.g. a loan to corner shop – there is no market where the actual result can be observed (see note 147). It remains unclear which value should be backtested and which meaningful implications (management input) should be derived from this theoretical exercise.
The EBA has widened the scope for the measurement and monitoring of CSRBB, which includes not only fair valued instruments but also other, non fair valued, instruments that are not part of existing guidance on CSRBB. The scope states that no instruments in the banking book should be excluded from the scope of CSRBB (e.g. off balance sheet loan commitments). However, by definition, the scope is limited to those instruments with exposure to market credit and market liquidity spread risk and observable credit spread activity. The definition of the perimeter should therefore be explicitly reduced to this. Further, the GLs do not provide clear guidance on the instruments to be included for CSRBB measurement. This will require further clarification for proper implementation.
The scope includes the liability side of the balance sheet but the GLs do not provide guidance on potential netting from similar instruments on both sides of the balance sheet. Further, the consultation states that the deterioration of an institution’s credit quality should not positively impact the credit spread risk measure.
Moreover, the inclusion of idiosyncratic elements for CSRBB calculation is not very clear as in 5.1.3 it is mentioned as preferred option idiosyncratic elements should not be included for CSRBB measure. Only market and related liquidity elements should conform CSRBB definition. In practice, we see considerable technical difficulties in isolating forward-looking market expectations regarding the development of idiosyncratic credit risk from observable credit spreads.
We also recommend explicitly allowing the introduction of threshold to ensure that small banks, and more generally banks with immaterial risk exposure, can be excluded from the calculation process. Furthermore, it should be possible to exclude certain positions from the risk assessment if it can be shown that the corresponding risks are already assessed and managed elsewhere (e.g. within combined credit risk and credit spread risk models). An example of this is balances of mortgage lending funded by covered bonds, where the values of mortgage loans have a close correspondence with the values of the corresponding covered bonds cf. Article 33 (3-4) of CRR. Furthermore, while CSRBB must be measured and steered, the specific design including integrated or isolated measurement should be up to the banks. We suggest reflecting this aspect in the final text.
We support the approach to focus the CSRBB on assets at fair value and we agree that there might be particular assets at book value to be included into the CSRBB framework. However, we do not consider generally including assets at book value and requiring banks to demonstrate exceptions a pragmatic approach. Instead, we would welcome if assets at book value were excluded from the CSRBB in the first place and if cases in which assets at book value might be exposed to credit spread risk could be clarified in the Guidelines.
Market prices are not available for traditional customer lending business. This is therefore outside the scope of application as defined in para. 7. For example, co-operative banks across various Member States fund a large number of small and medium-sized enterprises (e.g. corner shops), for which no market prices are available and therefore no credit spreads can be determined. The same applies to private mortgages. Where meaningful market information is available (e.g. corporate bond portfolio), such positions should be included in the CRSBB measurement. Specifically, this applies to transactions for which credit spreads are relevant, i.e. securities in particular.
Hence, we advocate a narrower interpretation of CSRBB, which we believe is also in line with the Basel rules. Based on the sensitivity of the position to credit spread risk, fair value positions would generally be included. However, positions for which changes in the balance sheet cannot be observed, e.g., loans accounted for at amortized cost, would generally be excluded from the CSRBB analysis, as they do not (materially) affect the credit spread risk and would also incur immense expenses.
In line with our understanding of the definition of CSRBB (see answer to Question 3), only positions for which meaningful market information is available should be included in the measurement of CSRBB. Reference prices are not sufficient, position-specific prices have to be available. This includes marketability and a certain minimum market liquidity. Prices from secondary markets, on the other hand, are not suitable for an adequate measurement of CSRBB (e.g. bonds traded on secondary markets with only issuer spreads available would also have to be excluded). As a result, only liquid fair value positions in the banking book should be included in the measurement. In our view, this is the only way to achieve an appropriate management effect in the institution.
In particular, we reject the requirement in para. 124 that institutions should not be allowed to exclude certain positions ex ante but must provide detailed documented evidence of an absence of sensitivity to credit spread risk each time a group of similar positions is excluded. In the examples given for certain product types, it would make more sense to generally assume that CSRBB is not material (negative lists at least at national level).
In special cases and depending on the respective business model, the competent authority could further review the respective transactions regarding the relevance of CSRBB.
Furthermore, pragmatic procedures should be envisaged to allow institutions to exclude non-material items from the scope on an individual basis. Sensitivity analyses should also be allowed to be carried out on a qualitative basis, e.g. for the exclusion of certain derivatives (if not on the negative list).
The distinction is clear.
Question 1: In the context of the measurement of the impact of IRRBB under internal systems, paragraph 111 envisages a five year cap repricing maturity for retail and non-financial wholesale deposits without a specified maturity. Would you foresee any unintended consequence or undesirable effect from this behavioural assumption in particular on certain business models or specific activities? If this is the case, please kindly provide concrete examples of it.
Please refer to attached paper for further details.We strongly disagree with the choice to introduce the five-year cap. Disallowing the modelling of NMDs severely constrains the actual IRRBB risk management of institutions, this would lead not only to results that are too conservative but also to potential inconsistencies, particularly where e.g. hedging of the positions stretches over a say 6 years horizon. In such cases it would in fact not be prudent to have the 5 years cap. We also strongly disagree with the exclusion of non-maturity deposits from financial customers from NMD. This prevents the models from mirroring reality and is not in line with BCBS 368. Moreover, the wording is not consistent and mentions a cap on the maturity itself rather than the average.
We also highlight the following:
The explanatory box clarifies that the 5-year cap applies to the full amount of NMDs – as opposed to just core deposits. Provided that we disagree with the cap, as noted above, such clarifications should be also included in the regulatory text.
While it is clear from the explanatory box that the 5-year cap applies to the combined volume of core and non-core deposits, it remains somewhat ambiguous if the 5-year cap would apply to
the aggregate NMD portfolio, or
to portfolios individually.
This should be clarified in the regulatory text. Moreover, we observe that it is not uncommon for retail transactional deposits portfolios to have modelled average maturities beyond 5 years – in line with the nature of this product. Hence, the 5-year cap apply could at most be applied to the aggregate portfolio of modelled non-maturing deposits and not on each portfolio individually. Alternatively, at least a higher cap should be considered for retail transactional deposits.
Question 2: Do respondents find that the criteria to identify non-satisfactory IRRBB internal models provide the minimum elements for supervisors’ assessment?
Please refer to attached paper for further details.The criteria are comprehensive. However, we suggest adding further wording so that in the supervisory evaluation process authorities can duly consider the relevant proportionality aspects and allow simplifications and adjustments in line with the underlying, bank-specific risk materiality. Examples where such proportionality should be applied, are the assessment of elasticities in 109(b) which small, non-complex institutions are mostly not able to perform and the isolation of behavioral effects under 112(h), which is of very limited informative value.
We also suggest to indicate that, where a supervisor is of the opinion that an institution does not satisfactorily comply with all elements required for internal models, in particular 119b, the supervisor should start a dialogue with the institutions to restore compliance rather than immediately imposing the (S)SA.
Besides, we propose to streamline the text in some points and to avoid duplications (e.g. 154 vs 155 and 146 vs 149).
For the inclusion of fair value changes in the definition of the NII we see challenges arising from the diverging accounting definition of NII (as it is for example also reported in FINREP) in comparison to the NII definition for IRRBB purposes. We also point to the fact that the complexity for small institutions to measure those risks significantly increases as internal risk measurement and steering is often based on the narrow NII definition. Furthermore, the CRR and CRD mention NII in contrast to an Earnings perspective named as NII, which is being proposed by the EBA. We therefore strongly recommend that fair value effects continue to be considered under an earnings metric and that a materiality threshold can be established for the consideration and limitation of these effects.
Moreover, to cater for banks that are not accounting according to IFRS (e.g. in Germany, Austria), the requirements should be presented in such a way that they are meaningful for all institutions and the wording should be adjusted accordingly.
We agree with the fact that IRRBB from an NII perspective is an important risk that needs to be measured and reported. However, given that NII is part of the normative perspective, the individual limitation of this specific risk is not in line with the general normative assumptions and underlying management actions. Like other factors in the normative perspective, NII should only be assessed periodically under defined overarching scenarios and analyzed in the context of the entire normative requirement. We would therefore ask the EBA to align the definition of limits in the normative context with the new ICAAP requirements to allow banks to streamline their efforts and to prevent diverging steering implications.
Finally, we ask the regulator to specify in the Guidelines that banks cannot solely rely on the (simplified) standardized approaches but need internal measurement systems to fully reflect their bank-specific behavior. Standardized approaches for IRRBB are no alternative to adequate internal risk management and should not be used as a benchmark. We emphasize that there should be no incentive to neglect building and retaining internal competences in the context of interest rate risk management.
Question 3: Is there any specific element in the definition of CSRBB that is not clear enough for the required assessment and monitoring of CSRBB by institutions?
Please refer to attached paper for further details.There is currently no industry standard on how to measure CSRBB. On the contrary, the existing approaches differ strongly. Hence, no industry standard should be "enforced" by the EBA. The specifications should be "modeling neutral", which is not the case. For example, there are credit risk models that are extended to include components for measuring CSRBB, which allow integrated measurement of spread changes based on changes of creditworthiness as well as changes in the market price for credit spreads. Especially with such integrated models, the proposed approach can quickly lead to double counting, which must be avoided. The existing models are all well understood and proven, and their use should continue to be possible.
Overall, guidance on instruments – other than fair valued ones – to be included in the calculation of CSRBB is unclear and may be interpreted differently by the banks. CSRBB is defined as relating to changes of market credit and market liquidity risk, yet the definition of market is undefined. A usual interpretation would be the one referring to instruments with a deep, liquid and active secondary market, yet this is not specifically stated.
The scope includes the liability side of the balance sheet but the GL does not provide guidance on potential netting from similar instruments on both sides of the balance sheet. Further, the consultation states that a deterioration of an institution’s credit quality should not have any positive impact on the credit spread risk measure. Notwithstanding idiosyncratic risk, some clarification is needed on whether a deterioration in own credit due to changes in general market credit spreads is in scope.
According to para. 157, idiosyncratic spread components can only be included in the CSRBB measurement if evidence is provided that the inclusion is conservative. The obligation to provide evidence should be deleted, since, depending on the reporting date considered, an inclusion can lead to an increase in risk at one time and a reduction in risk at another, due to the diversification effects that may be present in the case of an inclusion. Since any inclusion of additional components of the total credit spread risk tends to increase double counting, it should be considered conservative and allowed per se without the requirement for evidence. In contrast, a precise measurement of the individual components of credit spread risk (market spread, liquidity spread, separate consideration of idiosyncratic spread) may be theoretically desirable but it is not possible in practice. This corresponds with the associated explanation box (on page 46). Even if the spread components were to be separated, it is still questionable whether it would offer any added value at all and lead to meaningful steering input in the institution. For example, the default premium could already be included in the credit risk and the liquidity premium could, if necessary, be included in the institution's liquidity risk. Expected rating changes, i.e. expected migrations to other rating classes, can in turn affect the default premium and, if applicable, the liquidity premium.
Usually, credit spread data is sector and/or country specific and there is no general market credit spread curve for each rating class. If the final version still requires the use of a general curve, clarification is needed on how the general curve can be obtained. When considering credit spread risk, institution-specific circumstances (portfolio composition, design of internal systems, availability of reference data, etc.) should be taken into account. The specific design including integrated or isolated measurement should be up to the banks. For example, the analysis of spread components clustered by currency proposed in paragraph 123 is not appropriate in every case; on the other hand, other clusters (e.g., by rating class or sector or even a more granular, issuer-specific mapping) may become relevant. Here, it should be up to the institutions to choose an adequate type of clustering. In our view, this is the only way to achieve meaningful results and to derive sensible steering input in the institutions.
Backtesting does not always deliver meaningful results or is not possible at all. For example, for a customer loan - e.g. a loan to corner shop – there is no market where the actual result can be observed (see note 147). It remains unclear which value should be backtested and which meaningful implications (management input) should be derived from this theoretical exercise.
Question 4: As to the suggested perimeter of items exposed to CSRBB, would you consider any specific conceptual or operational challenge to implement it?
Please refer to attached paper for further details.The EBA has widened the scope for the measurement and monitoring of CSRBB, which includes not only fair valued instruments but also other, non fair valued, instruments that are not part of existing guidance on CSRBB. The scope states that no instruments in the banking book should be excluded from the scope of CSRBB (e.g. off balance sheet loan commitments). However, by definition, the scope is limited to those instruments with exposure to market credit and market liquidity spread risk and observable credit spread activity. The definition of the perimeter should therefore be explicitly reduced to this. Further, the GLs do not provide clear guidance on the instruments to be included for CSRBB measurement. This will require further clarification for proper implementation.
The scope includes the liability side of the balance sheet but the GLs do not provide guidance on potential netting from similar instruments on both sides of the balance sheet. Further, the consultation states that the deterioration of an institution’s credit quality should not positively impact the credit spread risk measure.
Moreover, the inclusion of idiosyncratic elements for CSRBB calculation is not very clear as in 5.1.3 it is mentioned as preferred option idiosyncratic elements should not be included for CSRBB measure. Only market and related liquidity elements should conform CSRBB definition. In practice, we see considerable technical difficulties in isolating forward-looking market expectations regarding the development of idiosyncratic credit risk from observable credit spreads.
We also recommend explicitly allowing the introduction of threshold to ensure that small banks, and more generally banks with immaterial risk exposure, can be excluded from the calculation process. Furthermore, it should be possible to exclude certain positions from the risk assessment if it can be shown that the corresponding risks are already assessed and managed elsewhere (e.g. within combined credit risk and credit spread risk models). An example of this is balances of mortgage lending funded by covered bonds, where the values of mortgage loans have a close correspondence with the values of the corresponding covered bonds cf. Article 33 (3-4) of CRR. Furthermore, while CSRBB must be measured and steered, the specific design including integrated or isolated measurement should be up to the banks. We suggest reflecting this aspect in the final text.
We support the approach to focus the CSRBB on assets at fair value and we agree that there might be particular assets at book value to be included into the CSRBB framework. However, we do not consider generally including assets at book value and requiring banks to demonstrate exceptions a pragmatic approach. Instead, we would welcome if assets at book value were excluded from the CSRBB in the first place and if cases in which assets at book value might be exposed to credit spread risk could be clarified in the Guidelines.
Market prices are not available for traditional customer lending business. This is therefore outside the scope of application as defined in para. 7. For example, co-operative banks across various Member States fund a large number of small and medium-sized enterprises (e.g. corner shops), for which no market prices are available and therefore no credit spreads can be determined. The same applies to private mortgages. Where meaningful market information is available (e.g. corporate bond portfolio), such positions should be included in the CRSBB measurement. Specifically, this applies to transactions for which credit spreads are relevant, i.e. securities in particular.
Hence, we advocate a narrower interpretation of CSRBB, which we believe is also in line with the Basel rules. Based on the sensitivity of the position to credit spread risk, fair value positions would generally be included. However, positions for which changes in the balance sheet cannot be observed, e.g., loans accounted for at amortized cost, would generally be excluded from the CSRBB analysis, as they do not (materially) affect the credit spread risk and would also incur immense expenses.
In line with our understanding of the definition of CSRBB (see answer to Question 3), only positions for which meaningful market information is available should be included in the measurement of CSRBB. Reference prices are not sufficient, position-specific prices have to be available. This includes marketability and a certain minimum market liquidity. Prices from secondary markets, on the other hand, are not suitable for an adequate measurement of CSRBB (e.g. bonds traded on secondary markets with only issuer spreads available would also have to be excluded). As a result, only liquid fair value positions in the banking book should be included in the measurement. In our view, this is the only way to achieve an appropriate management effect in the institution.
In particular, we reject the requirement in para. 124 that institutions should not be allowed to exclude certain positions ex ante but must provide detailed documented evidence of an absence of sensitivity to credit spread risk each time a group of similar positions is excluded. In the examples given for certain product types, it would make more sense to generally assume that CSRBB is not material (negative lists at least at national level).
In special cases and depending on the respective business model, the competent authority could further review the respective transactions regarding the relevance of CSRBB.
Furthermore, pragmatic procedures should be envisaged to allow institutions to exclude non-material items from the scope on an individual basis. Sensitivity analyses should also be allowed to be carried out on a qualitative basis, e.g. for the exclusion of certain derivatives (if not on the negative list).
IRRBB only
Please refer to attached paper for further details.CSRBB only
Please refer to attached paper for further details.Both IRRBB and CSRBB
Please refer to attached paper for further details.The distinction is clear.