Response to consultation on draft RTS on IRRBB supervisory outlier tests
Go back
Under the new proposal, the bound would be set to a level far below any level that interest rates have reached in the past. In particular, we highlight a misalignment between the calibration period of the updated lower bound and its entry into force since, as shown in Chart 1, the current interest rate curve is well above the existing lower bound. This would seem to make unnecessary the recalibration. Besides, the current EBA 2018 GL (EBA/GL/2018/02) already envisage the possibility to adapt the lower bound in case of interest rate curves were lower than the bound.
Chart 1 – Current IR curve and lower bounds
We also point out that the impacts on IRRBB metrics of the proposed new lower bound has never been properly tested within previous QIS or IRRBB dedicated supervisory exercises. As a matter of fact, the QIS data used to support the calibration of SOT assumptions only include metrics calculated according to the current bound and the unconstrained negative scenario. Moreover, the EBA itself admits that only a limited number of banks participating to the QIS exercise filled in the IRRBB sections and that data provided had some quality issues.
In our view, instead, the impacts would be extremely material, both on EVE and NII measures under negative shocks scenarios, being the latter particularly sensitive to changes in short term interest rates. In the answer to Question 2, we provide a benchmark analysis that was carried out on a sample of more than 25 Italian banking groups, to support discussion about the NII threshold calibration, together with the potential implications of the new lower bound on both NII and EVE. From the analysis conducted, results reported in Table 1 show that introducing the new proposed lower bound would increase the number of outlier banks from 14% to 39% in the case of EVE SOT, while for NII SOT the increase in outliers would raise from 21% to 68% and from 18% to 50%, for Metric1 and Metric2, respectively. NII metrics are computed according to the definition of NII as the difference between interest income and expenses, not considering fair value changes.
We also highlight that the implications for the management of IRRBB in banks could be significant. In facts, all else equal, such regulatory changes might cause a limit breach, triggering managerial actions to adjust the position and the IRRBB strategy of the bank, causing possible negative impacts on profitability. Similarly, in terms of market disclosure (Pillar III), a regulatory change could have an impact on the level of the indicators publicly disclosed with consequent reputational effects.
As for the use of one-year time horizon and the constant balance sheet assumption, we agree with the EBA proposal, as it is consistent with internal practices.
We recognize that the proposal to consider current commercial margins aims at obtaining more precise NII projections, although we observe that it implies the use of a simulation approach, thus potentially requiring the evolution of current models and operational processes. It should also be noted that for managerial purposes, when projecting NII for instance for planning processes, different assumptions might be considered for commercial margins, so that the NII projections carried out for risk measurement might still not be aligned with the internal one.
Finally, regarding the choice to include changes in market value in the calculation of NII SOT, we highlight that accounting approaches are not consistent across EU; therefore the above-mentioned inclusion may cause a loss of the comparability, which we understand is one of supervisors’ goal. Besides, we observe that a “wider NII” metric would differ from current managerial NII metrics and would also differ from the accounting measures of NII.
Hence, our preference is to exclude market value changes from the NII SOT calculation. Nonetheless, in case the regulatory choice will include this component, we suggest that a more detailed instruction for the calculation would be useful. To avoid unintended double counting of the impact of NII and EVE over the first year of FV instrument, it should be clarified that the changes in fair value should be calculated at the end of the NII time horizon. We also note that this statement seems to be contradicted by the disposals set in the RTS for the standardized methodology, where the computation of market value changes is explicitly calculated at the beginning of the time horizon (i.e., at analysis date).
On the opposite, Metric 2 could be strongly influenced by extraordinary events (M&A, pandemic …) and multi-annual budget planning which are, in most of the cases, not related to the specific risk factor. For these reasons, we believe that this formulation would add an unwanted volatility to the indicator and moreover it would reduce the comparability across banks.
As for the calibration of the NII SOT, we believe that the proposed thresholds are too strict.
They could lead to a relevant number of outliers, with significant implications for the management of the IRRBB position and, ultimately, for the bank’s strategy.
Aifirm IRRBB Commission carried out an analysis on a sample of 28 Italian banking groups, representing more than 70% of Italian banks’ total assets. Main results are reported in Table 1. NII metrics are computed according to the definition of NII as the difference between interest income and expenses, not considering fair value changes.
Applying the calibration methodology proposed by the EBA, based on the number of outliers observed for the EVE SOT, the NII threshold should be set at -4,2% in case of Metric 1 (instead of the proposed -2,5%), or at -37% for Metric 2 (instead of -35%), thus showing that the calibration is significantly dependent on the sample considered.
As mentioned before in the answer to Question 1, we are also particularly concerned that the calibration of the NII threshold should be done in close connection with the setting of the lower bound for post shock interest rates.
Current proposed thresholds on NII appear to have been calibrated based on BCBS scenarios 1 and 2 (the latter being simulated with the floors currently in force). However, the switch to the new proposed floor would impact severely the measures, as already highlighted in the previous answer and as shown by results reported in Table 1 below: given the proposed thresholds and the proposed new lower bound, NII outliers would raise from 21% to 68% and from 18% to 50%, for Metric1 and Metric2, respectively.
Table 1 – Benchmark analysis on SOT metrics
Where:
Therefore, we reaffirm our request not to change the current lower bound, or, as a second-best choice, we ask that the threshold should be set based on the NII results of the unconstrained downward scenario of the QIS, given the same identified percentile.
In both cases, a wider data set seems to be necessary to apply the proposed calibration methodology based on outliers’ distribution.
As a last comment on this topic, we also point out that, in case of confirmation of the new lower bound, at such levels of interest rates (-1,5%) it would be questionable whether the zero floor legal provision on contracts with customers would be still reasonable, so that different assumptions for this products should be consistently envisaged.
Finally, as for currency aggregation examples, they are clear, but we believe that examples with exposure in EUR being positive should be reported for better understanding the process.
As an additional remark, we point out that banks are also expecting the introduction of the new accounting framework for macro fair value hedge, which will introduce a further constraint to the management of IRRBB. So, given the increased complexity for banks to manage simultaneously the two regulatory indicators (NII SOT and EVE SOT), we argue that it would be desirable for banks to dispose of all the elements, both in terms of prudential regulation and of accounting rules, to properly define their IRRBB strategy and control framework.
Finally, regarding article 4(n) ("In assessing the risk of interest rate-sensitive products that are linked to inflation or other market factors, prudent assumptions shall be applied. These assumptions shall be based on the current/last observed value, on forecasts of a reputable economic research institute or on other generally accepted market practices and shall be generally scenario-independent") we kindly request for more clarifications regarding the treatment of the inflation risk factors in the calculation of the IRRBB measures.
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.
Regarding the recalibrated lower bound for post-shock IR levels, we believe that the proposed floor is too low, and that current bound should be maintained.Under the new proposal, the bound would be set to a level far below any level that interest rates have reached in the past. In particular, we highlight a misalignment between the calibration period of the updated lower bound and its entry into force since, as shown in Chart 1, the current interest rate curve is well above the existing lower bound. This would seem to make unnecessary the recalibration. Besides, the current EBA 2018 GL (EBA/GL/2018/02) already envisage the possibility to adapt the lower bound in case of interest rate curves were lower than the bound.
Chart 1 – Current IR curve and lower bounds
We also point out that the impacts on IRRBB metrics of the proposed new lower bound has never been properly tested within previous QIS or IRRBB dedicated supervisory exercises. As a matter of fact, the QIS data used to support the calibration of SOT assumptions only include metrics calculated according to the current bound and the unconstrained negative scenario. Moreover, the EBA itself admits that only a limited number of banks participating to the QIS exercise filled in the IRRBB sections and that data provided had some quality issues.
In our view, instead, the impacts would be extremely material, both on EVE and NII measures under negative shocks scenarios, being the latter particularly sensitive to changes in short term interest rates. In the answer to Question 2, we provide a benchmark analysis that was carried out on a sample of more than 25 Italian banking groups, to support discussion about the NII threshold calibration, together with the potential implications of the new lower bound on both NII and EVE. From the analysis conducted, results reported in Table 1 show that introducing the new proposed lower bound would increase the number of outlier banks from 14% to 39% in the case of EVE SOT, while for NII SOT the increase in outliers would raise from 21% to 68% and from 18% to 50%, for Metric1 and Metric2, respectively. NII metrics are computed according to the definition of NII as the difference between interest income and expenses, not considering fair value changes.
We also highlight that the implications for the management of IRRBB in banks could be significant. In facts, all else equal, such regulatory changes might cause a limit breach, triggering managerial actions to adjust the position and the IRRBB strategy of the bank, causing possible negative impacts on profitability. Similarly, in terms of market disclosure (Pillar III), a regulatory change could have an impact on the level of the indicators publicly disclosed with consequent reputational effects.
As for the use of one-year time horizon and the constant balance sheet assumption, we agree with the EBA proposal, as it is consistent with internal practices.
We recognize that the proposal to consider current commercial margins aims at obtaining more precise NII projections, although we observe that it implies the use of a simulation approach, thus potentially requiring the evolution of current models and operational processes. It should also be noted that for managerial purposes, when projecting NII for instance for planning processes, different assumptions might be considered for commercial margins, so that the NII projections carried out for risk measurement might still not be aligned with the internal one.
Finally, regarding the choice to include changes in market value in the calculation of NII SOT, we highlight that accounting approaches are not consistent across EU; therefore the above-mentioned inclusion may cause a loss of the comparability, which we understand is one of supervisors’ goal. Besides, we observe that a “wider NII” metric would differ from current managerial NII metrics and would also differ from the accounting measures of NII.
Hence, our preference is to exclude market value changes from the NII SOT calculation. Nonetheless, in case the regulatory choice will include this component, we suggest that a more detailed instruction for the calculation would be useful. To avoid unintended double counting of the impact of NII and EVE over the first year of FV instrument, it should be clarified that the changes in fair value should be calculated at the end of the NII time horizon. We also note that this statement seems to be contradicted by the disposals set in the RTS for the standardized methodology, where the computation of market value changes is explicitly calculated at the beginning of the time horizon (i.e., at analysis date).
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?
We state a preference for Metric 1, since we believe it is simpler to manage by the risk-taking centers, easier to communicate to Top Management and more aligned to EVE metric, making the two measures able to provide better integrated information. At this regard we remark that during the RAF process, the limits setting should be made ensuring coherence between EVE and NII limits, so that managing IRRBB may impact both measures without any other noisy factors. Having the thresholds set with the same metric would help this process.On the opposite, Metric 2 could be strongly influenced by extraordinary events (M&A, pandemic …) and multi-annual budget planning which are, in most of the cases, not related to the specific risk factor. For these reasons, we believe that this formulation would add an unwanted volatility to the indicator and moreover it would reduce the comparability across banks.
As for the calibration of the NII SOT, we believe that the proposed thresholds are too strict.
They could lead to a relevant number of outliers, with significant implications for the management of the IRRBB position and, ultimately, for the bank’s strategy.
Aifirm IRRBB Commission carried out an analysis on a sample of 28 Italian banking groups, representing more than 70% of Italian banks’ total assets. Main results are reported in Table 1. NII metrics are computed according to the definition of NII as the difference between interest income and expenses, not considering fair value changes.
Applying the calibration methodology proposed by the EBA, based on the number of outliers observed for the EVE SOT, the NII threshold should be set at -4,2% in case of Metric 1 (instead of the proposed -2,5%), or at -37% for Metric 2 (instead of -35%), thus showing that the calibration is significantly dependent on the sample considered.
As mentioned before in the answer to Question 1, we are also particularly concerned that the calibration of the NII threshold should be done in close connection with the setting of the lower bound for post shock interest rates.
Current proposed thresholds on NII appear to have been calibrated based on BCBS scenarios 1 and 2 (the latter being simulated with the floors currently in force). However, the switch to the new proposed floor would impact severely the measures, as already highlighted in the previous answer and as shown by results reported in Table 1 below: given the proposed thresholds and the proposed new lower bound, NII outliers would raise from 21% to 68% and from 18% to 50%, for Metric1 and Metric2, respectively.
Table 1 – Benchmark analysis on SOT metrics
Where:
Therefore, we reaffirm our request not to change the current lower bound, or, as a second-best choice, we ask that the threshold should be set based on the NII results of the unconstrained downward scenario of the QIS, given the same identified percentile.
In both cases, a wider data set seems to be necessary to apply the proposed calibration methodology based on outliers’ distribution.
As a last comment on this topic, we also point out that, in case of confirmation of the new lower bound, at such levels of interest rates (-1,5%) it would be questionable whether the zero floor legal provision on contracts with customers would be still reasonable, so that different assumptions for this products should be consistently envisaged.
Finally, as for currency aggregation examples, they are clear, but we believe that examples with exposure in EUR being positive should be reported for better understanding the process.
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?
As already mentioned in previous responses, we are mainly concerned about the potential impacts of the proposed changes to the assumptions and the proposed thresholds, as it seems that those impacts have not been thoroughly assessed. As a closely related issue, we point out that at least a proper time lapse should be granted to banks to take necessary actions, consistent with the level of the foreseen overall impacts. We suggest considering a 1-year period, or at least envisaging transitional provisions for the most critical aspects, such as for instance the lower bound for post-shock IR levels.As an additional remark, we point out that banks are also expecting the introduction of the new accounting framework for macro fair value hedge, which will introduce a further constraint to the management of IRRBB. So, given the increased complexity for banks to manage simultaneously the two regulatory indicators (NII SOT and EVE SOT), we argue that it would be desirable for banks to dispose of all the elements, both in terms of prudential regulation and of accounting rules, to properly define their IRRBB strategy and control framework.
Finally, regarding article 4(n) ("In assessing the risk of interest rate-sensitive products that are linked to inflation or other market factors, prudent assumptions shall be applied. These assumptions shall be based on the current/last observed value, on forecasts of a reputable economic research institute or on other generally accepted market practices and shall be generally scenario-independent") we kindly request for more clarifications regarding the treatment of the inflation risk factors in the calculation of the IRRBB measures.
Upload files
2022-03-31_CP-RTS-SOT.pdf
(292.08 KB)