Response to consultation on draft RTS on IRRBB supervisory outlier tests
Go back
In principle, we support these requirements in the interests of comparability and believe they will also simplify calculations. In accordance with the CRD requirement (cf. Article 98(5a)(c)(iii) of CRD V), the period over which future NII will be measured must also be specified. The ECB opted for a 12-month period for the purpose of its short-term exercise (STE). We support this decision on the condition that institutions only have to carry out the calculation for complete calendar years.
It should also be borne in mind that it is not possible for certain business models (e.g. German building and loan associations, development of the building savings pool, entitlement to loan drawdowns from a building savings contract) to comply with the constant balance sheet assumption in a way that makes sense. In such cases, the requirement should be applied less strictly and on a best-effort basis. Alternatively, such business models could be permitted to opt for a dynamic balance sheet assumption.
In addition, we see a danger of the consideration and replacement of embedded termination rights and swaptions potentially generating disproportionate costs under the constant balance sheet assumption. Given the lack of relevance and materiality of these positions, relief should be considered so that costs do not outweigh benefits. It might make good sense to omit them from the NII simulation or at least allow the use of a bank’s internal model.
NII definition:
◼ To begin with, the additional consideration of the valuation result will lead to increased complexity and less comparability.
◼ Write-downs and write-ups depend on national GAAP or IFRS and on the options exercised by institutions.
◼ For banks that do not use IFRS, some of the explanations will be difficult to interpret (e. g. “market value changes of instruments shown in the profit and loss account or directly in equity”).
◼ The NII definition is inconsistent with that for the purposes of FINREP, where the amount of realised NII continues to be the difference between interest income and interest expense. This discrepancy and potential for confusion is problematic, in our view, especially when disclosing risk measures to the general public.
– It can result in arbitrarily high limit utilisation even if the NII risk is low in absolute terms (small denominator problem).
– It connects unrelated items (the reason for administrative costs may be totally unrelated to NII exposure) and creates volatility and dependency in NII limit utilisation on non-interest rate items.
◼ In addition, use of the broader definition will entail disproportionate time and effort. It requires the interest-induced valuation result to be isolated from other valuation corrections. A range of assumptions (e. g. about exchange rates, interest rate volatility, etc.) are needed to do so. This unnecessarily increases complexity and creates additional work without generating any meaningful added value.
◼ For instance, differentiating between interest-related and non-interest-related net commission income generates a disproportionate operational burden. The inclusion of interest-dependent commission – especially if there is a strong focus on financing through retail deposits – requires complex analysis of the interest rate sensitivity and transmission to customers in different interest rate shock scenarios. At the same time, the interest-related portion of net commission income is generally quite small compared to an institution’s total interest rate risk, so there is no justification for the massive time and effort required.
◼ Under German law, moreover, institutions without a trading book, for example, would have to value the liquidity reserve at fair value, while many other items are valued at amortised cost when performing the same check of whether provisions need to be set up in accordance with BFA 31 due to a lack of fair values at banking book level. The valuation of pension provisions under German GAAP depends on the interest rate level without fair value changes having any real direct influence on the calculation of profit and loss under German GAAP. As things stand, the regulatory framework for IRRBB takes no account of German GAAP-specific effects. This may result in the broader NII definition requiring banks to model effects that do not represent a relevant risk.
◼ Last but not least, the inclusion of valuation effects in the NII is not appropriate from an economic point of view. Changes in present value can only be captured appropriately using the EVE method. They are fundamentally different from the P&L valuation effects following price movements of the valued products since, unlike under the EVE approach, valuation or accounting options in the P&L context may distort the actual price movements.
◼ This would, for instance, be the case if securities are held for the purpose of hedging interest rate risk that arises from liabilities recognised at amortised cost. Using the EVE method, the change in value of both “sides” is recorded, while a distorted picture would arise under the NII method due to the inclusion only of the change in the value of the securities.
◼ The information content of valuation effects is therefore often lower than that of changes in present value. Consequently, an assessment of the change in the present value should continue to be based on the EVE approach while, at the same time, the NII method should be restricted to NII. On no account should the EVE and NII methods be combined.
Metric 1 is
◼ simple to implement and keep up to date (no updating of an alpha factor as in option B),
◼ structurally similar to the metric in the EVE SOT, thus easier to understand, more transparent and with results that are easier to communicate,
◼ comparable even across different business models and cost structures (in contrast to option B, which would only allow peer group comparison),
◼ not dependent on the accounting standards used (unlike option B),
◼ more consistent over time since a change in the alpha factor (e. g. following a restructuring) will not, as in option B, possibly trigger a transition from non-outlier to outlier status,
◼ better suited to comprehensively reflect the ability of institutions to use their capital to absorb interest rate risks that arise.
Metric 2
◼ A major disadvantage of option B, as we see it, is that it cannot normally be assumed that NII-related expenses are proportional to total expenses.
◼ Option B requires the alpha factor to be regularly updated; the effects of these updates are difficult to forecast, however, and will thus create an “uncertainty factor”.
◼ Option B is inconsistent by virtue of its very design as forecast revenues are set against historical costs. We have particular reservations about this approach because it mixes different periods, some of which will possibly contain one-off accounting effects. A combination may set risk management incentives based on past accounting effects that have no economic significance.
◼ Option B may thus generate implausibilities in the context of restructuring and other structural changes.
◼ The consideration of FINREP data is operationally challenging, especially when it comes to reports submitted after the annual financial statements have been approved.
◼ Moreover, an interest rate risk metric should assess interest rate risk and not the bank’s entire business model.
◼ In our view, the inclusion of administrative expenses in option B for the metric of a “large decline” in NII will potentially lead to less transparency and to problems in comparing the resulting metrics of banks, especially if they have different business models.
◼ We would like to point out that this combines the effects of changes in the interest rate risk positions of institutions with effects of expenses and income positions that go beyond net interest income and will significantly increase the complexity of the NII metric. As a result, it will be difficult to maintain a consistent view of the interest rate risk of institutions due to differences in their business models and to compare the development of interest rate risks over time.
◼ In addition, the definition of α in option B refers to FINREP positions which banks preparing German GAAP accounts do not have available in this form (NIIhist with fair value changes, gains/losses). Since there is no direct equivalent in every accounting regime, the calibration of outliers, which is already based on a very small sample at present, cannot be applied here.
For this reason, fulfilling at least one of the two outlier criteria set out in Article 98(5) of the CRD should not automatically trigger the imposition of capital surcharges by supervisors. The criteria should only be seen as indicators of a need for supervisors to monitor the institution more closely. This should be followed by a review of the institution in question and further investigation. Supervisors should begin by using all the information available to them and only then, in a second step, carry out further investigations with the involvement of the institution. The institution must be given an opportunity to comment, and it should have sufficient time to remedy any shortcomings that may exist. We would ask the EBA to add wording to this effect.
We are also opposed to the requirement in para 6 on page 5 and para 85 of the draft guidelines, which says that the SOT should be fully integrated into the internal framework for measuring and managing IRRBB and used as a complementary tool. This may set undesirable risk management incentives (e. g. to engage in hedging activities that make little sense). For internal management purposes, it makes sense to treat SOT results merely as a secondary indicator.
With respect to the calibration, we would like to stress that the sample used is too small for a decision of this importance. What is more, it is dependent on the interest rate environment and therefore clearly too conservative as we are currently in one of the extreme situations that are supposed to be reflected in shock scenarios. The calibration should therefore be reviewed after the first reports have been made. Moreover, using a snapshot as at a specific date to calibrate “outlier thresholds” based on observed variation across banks will not yield any information on potentially excessive NII risk. The methodology should be modified to make more economic sense.
In addition, it should be clarified how the baseline global interest rate shock parameters have been determined and why they are deemed appropriate from a regulatory perspective.
It should also be clarified why the scenarios defined in ANNEX 1 are appropriate given that they were calibrated on the basis of a time series from 2000 to 2015, when there was a large decrease in interest rates.
Clarification is needed in Article 1(3) that the currency-specific scenarios do not have to be applied to all currencies accounting for less than 5% of the non-trading book assets once the 90% threshold is reached. A concrete requirement specifying how these volumes should be treated is also needed.
Further details of how to apply Article 4(l) should be provided. Currently, the factor of 50% has to be applied once the absolute value of 80% of the ERM II currency gains is larger than the absolute value of the EUR loss. Further specification is required of whether the less favourable recognition would actually have to be used immediately or whether it would be possible to apply a pro rata reduction of the positive effects down to the 50% factor. Under the current wording, very small changes in the portfolio could result in highly disproportionate changes in the regulatory ratios (cliff effect).
Finally, it is unclear how to apply positive effects in EUR. It should be clarified that the home currency always allows a 100% recognition of gains.
Respondents are also kindly requested to express whether they find an inclusion of market value changes in the calculation of the NII SOT clear enough.
One-year time horizon and constant balance sheet assumption:In principle, we support these requirements in the interests of comparability and believe they will also simplify calculations. In accordance with the CRD requirement (cf. Article 98(5a)(c)(iii) of CRD V), the period over which future NII will be measured must also be specified. The ECB opted for a 12-month period for the purpose of its short-term exercise (STE). We support this decision on the condition that institutions only have to carry out the calculation for complete calendar years.
It should also be borne in mind that it is not possible for certain business models (e.g. German building and loan associations, development of the building savings pool, entitlement to loan drawdowns from a building savings contract) to comply with the constant balance sheet assumption in a way that makes sense. In such cases, the requirement should be applied less strictly and on a best-effort basis. Alternatively, such business models could be permitted to opt for a dynamic balance sheet assumption.
In addition, we see a danger of the consideration and replacement of embedded termination rights and swaptions potentially generating disproportionate costs under the constant balance sheet assumption. Given the lack of relevance and materiality of these positions, relief should be considered so that costs do not outweigh benefits. It might make good sense to omit them from the NII simulation or at least allow the use of a bank’s internal model.
NII definition:
◼ To begin with, the additional consideration of the valuation result will lead to increased complexity and less comparability.
◼ Write-downs and write-ups depend on national GAAP or IFRS and on the options exercised by institutions.
◼ For banks that do not use IFRS, some of the explanations will be difficult to interpret (e. g. “market value changes of instruments shown in the profit and loss account or directly in equity”).
◼ The NII definition is inconsistent with that for the purposes of FINREP, where the amount of realised NII continues to be the difference between interest income and interest expense. This discrepancy and potential for confusion is problematic, in our view, especially when disclosing risk measures to the general public.
– It can result in arbitrarily high limit utilisation even if the NII risk is low in absolute terms (small denominator problem).
– It connects unrelated items (the reason for administrative costs may be totally unrelated to NII exposure) and creates volatility and dependency in NII limit utilisation on non-interest rate items.
◼ In addition, use of the broader definition will entail disproportionate time and effort. It requires the interest-induced valuation result to be isolated from other valuation corrections. A range of assumptions (e. g. about exchange rates, interest rate volatility, etc.) are needed to do so. This unnecessarily increases complexity and creates additional work without generating any meaningful added value.
◼ For instance, differentiating between interest-related and non-interest-related net commission income generates a disproportionate operational burden. The inclusion of interest-dependent commission – especially if there is a strong focus on financing through retail deposits – requires complex analysis of the interest rate sensitivity and transmission to customers in different interest rate shock scenarios. At the same time, the interest-related portion of net commission income is generally quite small compared to an institution’s total interest rate risk, so there is no justification for the massive time and effort required.
◼ Under German law, moreover, institutions without a trading book, for example, would have to value the liquidity reserve at fair value, while many other items are valued at amortised cost when performing the same check of whether provisions need to be set up in accordance with BFA 31 due to a lack of fair values at banking book level. The valuation of pension provisions under German GAAP depends on the interest rate level without fair value changes having any real direct influence on the calculation of profit and loss under German GAAP. As things stand, the regulatory framework for IRRBB takes no account of German GAAP-specific effects. This may result in the broader NII definition requiring banks to model effects that do not represent a relevant risk.
◼ Last but not least, the inclusion of valuation effects in the NII is not appropriate from an economic point of view. Changes in present value can only be captured appropriately using the EVE method. They are fundamentally different from the P&L valuation effects following price movements of the valued products since, unlike under the EVE approach, valuation or accounting options in the P&L context may distort the actual price movements.
◼ This would, for instance, be the case if securities are held for the purpose of hedging interest rate risk that arises from liabilities recognised at amortised cost. Using the EVE method, the change in value of both “sides” is recorded, while a distorted picture would arise under the NII method due to the inclusion only of the change in the value of the securities.
◼ The information content of valuation effects is therefore often lower than that of changes in present value. Consequently, an assessment of the change in the present value should continue to be based on the EVE approach while, at the same time, the NII method should be restricted to NII. On no account should the EVE and NII methods be combined.
Question 2: Do respondents have any comment related to these two metrics for the specification and the calibration of the test statistic for the large decline in Article 6 for the purpose of NII SOT? Specifically, do respondents find the inclusion of administrative expenses in metric 2 clear enough? Do respondents have any comment on the example on currency aggregation for metric 1 and metric 2?
The main basis for assessing the suitability of a benchmark should be the extent to which the metric can be used to help answer the question of whether the strength and stability of the earnings stream and the level of income are sufficient to generate and maintain normal business operations (Basel Standard para 91). We believe the advantages of the capital-based metric 1 clearly outweigh metric 2, especially taking into account the suitability criterion cited above. We see the advantages and disadvantages of the two options as follows:Metric 1 is
◼ simple to implement and keep up to date (no updating of an alpha factor as in option B),
◼ structurally similar to the metric in the EVE SOT, thus easier to understand, more transparent and with results that are easier to communicate,
◼ comparable even across different business models and cost structures (in contrast to option B, which would only allow peer group comparison),
◼ not dependent on the accounting standards used (unlike option B),
◼ more consistent over time since a change in the alpha factor (e. g. following a restructuring) will not, as in option B, possibly trigger a transition from non-outlier to outlier status,
◼ better suited to comprehensively reflect the ability of institutions to use their capital to absorb interest rate risks that arise.
Metric 2
◼ A major disadvantage of option B, as we see it, is that it cannot normally be assumed that NII-related expenses are proportional to total expenses.
◼ Option B requires the alpha factor to be regularly updated; the effects of these updates are difficult to forecast, however, and will thus create an “uncertainty factor”.
◼ Option B is inconsistent by virtue of its very design as forecast revenues are set against historical costs. We have particular reservations about this approach because it mixes different periods, some of which will possibly contain one-off accounting effects. A combination may set risk management incentives based on past accounting effects that have no economic significance.
◼ Option B may thus generate implausibilities in the context of restructuring and other structural changes.
◼ The consideration of FINREP data is operationally challenging, especially when it comes to reports submitted after the annual financial statements have been approved.
◼ Moreover, an interest rate risk metric should assess interest rate risk and not the bank’s entire business model.
◼ In our view, the inclusion of administrative expenses in option B for the metric of a “large decline” in NII will potentially lead to less transparency and to problems in comparing the resulting metrics of banks, especially if they have different business models.
◼ We would like to point out that this combines the effects of changes in the interest rate risk positions of institutions with effects of expenses and income positions that go beyond net interest income and will significantly increase the complexity of the NII metric. As a result, it will be difficult to maintain a consistent view of the interest rate risk of institutions due to differences in their business models and to compare the development of interest rate risks over time.
◼ In addition, the definition of α in option B refers to FINREP positions which banks preparing German GAAP accounts do not have available in this form (NIIhist with fair value changes, gains/losses). Since there is no direct equivalent in every accounting regime, the calibration of outliers, which is already based on a very small sample at present, cannot be applied here.
Question 3: Do respondents consider that all the necessary aspects have been covered in the draft regulatory standard? Do respondents find the provisions clear enough or would any additional clarification be needed on any aspect?
Regulatory “one-size-fits all” standard metrics for identifying outlier institutions will inevitably have methodological weaknesses and are therefore not a suitable means of determining a bank’s individual risk situation.For this reason, fulfilling at least one of the two outlier criteria set out in Article 98(5) of the CRD should not automatically trigger the imposition of capital surcharges by supervisors. The criteria should only be seen as indicators of a need for supervisors to monitor the institution more closely. This should be followed by a review of the institution in question and further investigation. Supervisors should begin by using all the information available to them and only then, in a second step, carry out further investigations with the involvement of the institution. The institution must be given an opportunity to comment, and it should have sufficient time to remedy any shortcomings that may exist. We would ask the EBA to add wording to this effect.
We are also opposed to the requirement in para 6 on page 5 and para 85 of the draft guidelines, which says that the SOT should be fully integrated into the internal framework for measuring and managing IRRBB and used as a complementary tool. This may set undesirable risk management incentives (e. g. to engage in hedging activities that make little sense). For internal management purposes, it makes sense to treat SOT results merely as a secondary indicator.
With respect to the calibration, we would like to stress that the sample used is too small for a decision of this importance. What is more, it is dependent on the interest rate environment and therefore clearly too conservative as we are currently in one of the extreme situations that are supposed to be reflected in shock scenarios. The calibration should therefore be reviewed after the first reports have been made. Moreover, using a snapshot as at a specific date to calibrate “outlier thresholds” based on observed variation across banks will not yield any information on potentially excessive NII risk. The methodology should be modified to make more economic sense.
In addition, it should be clarified how the baseline global interest rate shock parameters have been determined and why they are deemed appropriate from a regulatory perspective.
It should also be clarified why the scenarios defined in ANNEX 1 are appropriate given that they were calibrated on the basis of a time series from 2000 to 2015, when there was a large decrease in interest rates.
Clarification is needed in Article 1(3) that the currency-specific scenarios do not have to be applied to all currencies accounting for less than 5% of the non-trading book assets once the 90% threshold is reached. A concrete requirement specifying how these volumes should be treated is also needed.
Further details of how to apply Article 4(l) should be provided. Currently, the factor of 50% has to be applied once the absolute value of 80% of the ERM II currency gains is larger than the absolute value of the EUR loss. Further specification is required of whether the less favourable recognition would actually have to be used immediately or whether it would be possible to apply a pro rata reduction of the positive effects down to the 50% factor. Under the current wording, very small changes in the portfolio could result in highly disproportionate changes in the regulatory ratios (cliff effect).
Finally, it is unclear how to apply positive effects in EUR. It should be clarified that the home currency always allows a 100% recognition of gains.
Upload files
GBIC comments on IRRBB SOT.pdf
(129.98 KB)