Response to consultation on draft RTS on IRRBB supervisory outlier tests
Go back
o Definition of NII SOT based on a NII “narrow” or an “Earnings – including mark to market” perspective: We recommend maintaining the NII definition aligned to the common (narrow) definition of net interest income and not including market value changes of fair value instruments. The narrow perspective of NII is more aligned with the internal management and control of IRRBB.
The inclusion of market value changes of fair value instruments in the banking book would change the commonly accepted NII perspective, making the risk exposure very dependent on the projection period and the accountancy criteria, which may lead to a misinterpretation of the results. Furthermore, it introduces a much higher complexity in the calculation, and it overlaps with the EVE approach.
Positions at fair value are generally sovereign bonds or derivatives used in the banking book as stable hedging instruments that do not affect NII following the accounting rules. Additionally, these portfolios are already incorporated in EVE metrics, and including them in NII would lead to a double accounting. Besides, it could lead to undesirable incentives for maintaining positions based on accounting criteria, such as HTC instead of economic motivations.
Lastly, operationally, the calculation of the broader NII metric is highly complex. There is an intersection between the NII and the FV changes during the projection period (e.g., 12M). Thus, the calculation of FV changes requires excluding the instruments maturing during the first year and estimating the future MtM impact at the end of the projection period (1 year) on the fair value instruments with longer maturities (>1Y) which will be computationally very demanding. In other words, it must be considered that the residual maturity of the MtM instruments at the end of the projection period is reduced by 1 year.
Likewise, the inclusion of fees and commissions would be very burdensome. Besides, calibrating the SOT NII under an earnings perspective and option B entails difficulties since the income budget does not include forecasted changes in MtM portfolios.
In any case, we suggest monitoring and controlling the effect in fair value instruments in a separate (secondary) metric and not replacing the standard NII definition
o Floor level: The current floor levels have been occasionally broken by the market risk free rates for the longer tenors (>20Y). However, the new floor levels proposed are too extreme, far below the minimum observed market rates for reference risk free rates (swap curves). Even when bond yields are considered, the current floor level has been only significantly breached in the long-term tenors. Moreover, considering the expected evolution of interest rates, where the minimum negative rate levels have been left behind.
The proposed floor seems to be too low given the historical analysis and tendency of interest rates, and it could lead to important changes for both the EVE SOT and NII SOT estimations. Consequently, the threshold levels defined based on QIS 2020 (applying the old floor) must be reviewed.
Accordingly, we strongly recommend softening the proposed floor by keeping the current level, or, if not, setting a higher initial level (e.g. -100bp) and adapting the slope of the current lower bound for the longer tenors (flattening the slope of the floor).
Market Swap rates vs floor: [Please find the graph regarding this question in the uploaded PDF document]
Moreover, we would like to warn on the effect of aggressive slope changes on implied forwards that affect the estimation of cash flows in variable products mainly for SOT EVE. When creating scenarios from shock parameters, the effect on the implied forwards in the new curve is usually not checked to observe the reasonableness of such estimates in the presence of spikes. This effect can be further magnified by the inclusion of a floor if the base scenario is close to the floor level, as it introduces artificial slope changes. The problem exists both with the current definition of the floor and with the proposed new one, as it depends to a large extent on the shape and level of the yield curve at any given point in time. The following graph shows an example of the type of problem we are referring to.
[Please find the graphs regarding this question in the uploaded PDF document]
We would appreciate it if the guidelines could include a mention of the possibility of smoothing the function to correct this type of problem, or another type of solution, when these problems arise in the generation of scenarios for the SOT.
o Inflation treatment: We agree on the general proposal to use prudent assumptions in relation to inflation behaviour. However, we consider that restricting the inflation modelling by setting a scenario-independent assumption would be a fatal flaw. The evolution of inflation, as a component of nominal interest rates, is highly correlated to the interest rates levels, especially in those jurisdictions with high market interest rates where the Spanish banks operate and important positions in inflation indexed products exist. To consider constant inflation in combination with high interest rate shocks it is a very unrealistic scenario, and not necessarily prudent.
Besides, treating real interest rates as an independent currency would have material impact in the SOT aggregation of significant currencies and could lead to artificial changes in the risk profile in some balance sheets. Real and nominal rates are highly correlated, so they should be treated as a single currency for aggregation purposes.
We strongly recommend eliminating from the text the reference to scenario-independent assumptions, and do not prescribe the inflation treatment which should be done according to the banks’ internal models. Additionally, for the SOT impacts aggregation, we believe that it should be stated that exposures to real and nominal currencies should be treated as a single currency position.
Net Interest income SOT common modelling assumptions:
o 12M Horizon: We welcome the proposal of setting a one-year horizon, as this enables to get a much more accurate measure, reducing the uncertainty of balance sheet evolution in longer periods, and simplifying the operational workload.
Nevertheless, the horizon must be carefully assessed if combined with a broad NII definition including FV changes. A very short horizon may lead to misleading results due to the lower contribution of the income factor to the metric (vs MtM impacts). We recommend anchoring NII SOT to the narrow definition of Net Interest Income.
o Commercial margins: We consider appropriate the inclusion of commercial margins in NII measures, as it enables the projection of the products’ total rate levels in order to correctly apply the repricing restrictions (floors/caps) of the balance sheet items. Besides, margins are essential to accurately estimate the NII projection levels.
o Constant balance sheet: We understand the motivation for applying a constant balance sheet as it eases the comparability among banks. Nevertheless, if applied “literally” as it is established in the Guidelines, we consider that it may lead to very unrealistic results. The current balance sheet structure obeys to a specific interest rates environment (e.g., low/negative rates), and it cannot be assumed that this profile will remain unchanged under a different interest rates scenario, especially when extreme shocks (e.g., +200bp) are applied.
This is especially relevant when applied to customer deposits that will evolve in a very different way in higher interest rates periods. The same consideration can be made regarding other balance sheet items, such as central banks deposits or TLTRO funding.
We suggest allowing flexibility to Banks to decide on the most appropriate approach to be used, dynamic or constant balance sheet, according to their internal risk management. Alternatively, we recommend relaxing the constant balance sheet assumption by allowing the conditional modelling to project the evolution of certain specific balance sheet items, according with the internal planning. This will also promote a higher consistency with the internal metrics that are disclosed by the bank.
o CET1 instruments exclusion: From an NII perspective, we consider that this assumption must be eliminated to avoid any confusion. Equity is part of the funding structure of the bank and should be taken into account in any case for NII projections.
Metric 2 is conceptually more meaningful and a better representation of the risk under assessment as it relates NII shocks to the NII generation capacity (NII base projection as denominator). Thus, metric 2 recognizes the effect of changes in the interest rates level and balance evolution in the base projection, and not only in the variability of NII (NII shock). Nevertheless, we would appreciate more simple rules regarding its calculation. We would like to highlight that FINREP PL reports (F02.00) are only reported at the level of reporting currency but the breakdown by material currencies is not binding. Therefore, there are operational implications in the aggregation of NII baseline and Admin expenses by currency. The calculation of this metric can only be done at the level of reporting currency.
In addition, we identify a potential mismatching between FINREP and SOT perimeter, when significant currencies represent less than 100% of projected NII. Consequently, it should be allowed to adjust this percentage in the metric calculation.
As an alternative, and to simplify the approach,
we propose to use consolidated accounted NII and total expenses as denominator as reported in FINREP, instead of using the same aggregation method as for sensitivity, and/or
we suggest using a simplified metric 2, by eliminating the administrative costs deduction: (NII shock- NII baseline)/ NII baseline < [Threshold]
Metric 1, although simpler, is less accurate and representative for NII assessment, as TIER 1 is not an appropriate reference to evaluate changes on NII, which should be put in relation to the NII generation.
As an example, if interest rates rise, the variability of NII will be increased but also the NII generation capacity. This effect will only be correctly recognized by Metric 2.
Threshold level: Whatever the metric chosen; the threshold level should be re-calibrated taking in consideration the new lower bound levels. Additionally, the potential effects on the threshold level of exigency, under a higher (and/or positive) interest rates environment should be taken into account. Consequently, we recommend extending the analysis over time by collecting more data before defining a large decline. The threshold should be set based on a single (punctual) observation.
Furthermore, we consider that it is not appropriate to set the NII threshold based on the number of outlier banks obtained in the EVE SOT. In the first place, the relationship between NII generation and TIER1 is not consistent among banks. If this effect is not considered in the calibration, it will unreasonably penalize banks with greater income generation capacity and, therefore, an absolute (not necessarily relative) higher amount of income variation. This can be seen in the fact that the output of the 2 metrics proposed may result in a different set of outlier banks, which does not make sense. Following this, the management of IRRBB across banks may be very different from both perspectives, leading to wrong conclusions. For example, if banks are prioritising NII vs EVE management, this should generate a higher number of outlier banks in the EVE SOT. By forcing the NII threshold to get a target number of outlier banks would lead to an excessively low reference.
NII threshold should be calibrated independently from EVE according to a prudent risk level.
● SOTs would be conducted applying internal models, and parameters not specifically mentioned in the RTS should follow internal calibrations. We would appreciate it if this aspect could be explicitly mentioned in the RTS.
● The definition of NII shock scenarios (equal to EVE SOT parallel scenarios) is extremely unrealistic, especially for developed economies such as EUR and USD currencies, where the occurrence of sudden shocks of +/-200 bps is very unlikely. We suggest recalibrating downwards the size of the shocks for the SOT according to volatility levels, and/or applying gradual shocks (instead of sudden shocks) as it is more appropriate for dynamic 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 common modelling assumptions for the purpose of SOT calculation, we’d like to point out the following considerations:o Definition of NII SOT based on a NII “narrow” or an “Earnings – including mark to market” perspective: We recommend maintaining the NII definition aligned to the common (narrow) definition of net interest income and not including market value changes of fair value instruments. The narrow perspective of NII is more aligned with the internal management and control of IRRBB.
The inclusion of market value changes of fair value instruments in the banking book would change the commonly accepted NII perspective, making the risk exposure very dependent on the projection period and the accountancy criteria, which may lead to a misinterpretation of the results. Furthermore, it introduces a much higher complexity in the calculation, and it overlaps with the EVE approach.
Positions at fair value are generally sovereign bonds or derivatives used in the banking book as stable hedging instruments that do not affect NII following the accounting rules. Additionally, these portfolios are already incorporated in EVE metrics, and including them in NII would lead to a double accounting. Besides, it could lead to undesirable incentives for maintaining positions based on accounting criteria, such as HTC instead of economic motivations.
Lastly, operationally, the calculation of the broader NII metric is highly complex. There is an intersection between the NII and the FV changes during the projection period (e.g., 12M). Thus, the calculation of FV changes requires excluding the instruments maturing during the first year and estimating the future MtM impact at the end of the projection period (1 year) on the fair value instruments with longer maturities (>1Y) which will be computationally very demanding. In other words, it must be considered that the residual maturity of the MtM instruments at the end of the projection period is reduced by 1 year.
Likewise, the inclusion of fees and commissions would be very burdensome. Besides, calibrating the SOT NII under an earnings perspective and option B entails difficulties since the income budget does not include forecasted changes in MtM portfolios.
In any case, we suggest monitoring and controlling the effect in fair value instruments in a separate (secondary) metric and not replacing the standard NII definition
o Floor level: The current floor levels have been occasionally broken by the market risk free rates for the longer tenors (>20Y). However, the new floor levels proposed are too extreme, far below the minimum observed market rates for reference risk free rates (swap curves). Even when bond yields are considered, the current floor level has been only significantly breached in the long-term tenors. Moreover, considering the expected evolution of interest rates, where the minimum negative rate levels have been left behind.
The proposed floor seems to be too low given the historical analysis and tendency of interest rates, and it could lead to important changes for both the EVE SOT and NII SOT estimations. Consequently, the threshold levels defined based on QIS 2020 (applying the old floor) must be reviewed.
Accordingly, we strongly recommend softening the proposed floor by keeping the current level, or, if not, setting a higher initial level (e.g. -100bp) and adapting the slope of the current lower bound for the longer tenors (flattening the slope of the floor).
Market Swap rates vs floor: [Please find the graph regarding this question in the uploaded PDF document]
Moreover, we would like to warn on the effect of aggressive slope changes on implied forwards that affect the estimation of cash flows in variable products mainly for SOT EVE. When creating scenarios from shock parameters, the effect on the implied forwards in the new curve is usually not checked to observe the reasonableness of such estimates in the presence of spikes. This effect can be further magnified by the inclusion of a floor if the base scenario is close to the floor level, as it introduces artificial slope changes. The problem exists both with the current definition of the floor and with the proposed new one, as it depends to a large extent on the shape and level of the yield curve at any given point in time. The following graph shows an example of the type of problem we are referring to.
[Please find the graphs regarding this question in the uploaded PDF document]
We would appreciate it if the guidelines could include a mention of the possibility of smoothing the function to correct this type of problem, or another type of solution, when these problems arise in the generation of scenarios for the SOT.
o Inflation treatment: We agree on the general proposal to use prudent assumptions in relation to inflation behaviour. However, we consider that restricting the inflation modelling by setting a scenario-independent assumption would be a fatal flaw. The evolution of inflation, as a component of nominal interest rates, is highly correlated to the interest rates levels, especially in those jurisdictions with high market interest rates where the Spanish banks operate and important positions in inflation indexed products exist. To consider constant inflation in combination with high interest rate shocks it is a very unrealistic scenario, and not necessarily prudent.
Besides, treating real interest rates as an independent currency would have material impact in the SOT aggregation of significant currencies and could lead to artificial changes in the risk profile in some balance sheets. Real and nominal rates are highly correlated, so they should be treated as a single currency for aggregation purposes.
We strongly recommend eliminating from the text the reference to scenario-independent assumptions, and do not prescribe the inflation treatment which should be done according to the banks’ internal models. Additionally, for the SOT impacts aggregation, we believe that it should be stated that exposures to real and nominal currencies should be treated as a single currency position.
Net Interest income SOT common modelling assumptions:
o 12M Horizon: We welcome the proposal of setting a one-year horizon, as this enables to get a much more accurate measure, reducing the uncertainty of balance sheet evolution in longer periods, and simplifying the operational workload.
Nevertheless, the horizon must be carefully assessed if combined with a broad NII definition including FV changes. A very short horizon may lead to misleading results due to the lower contribution of the income factor to the metric (vs MtM impacts). We recommend anchoring NII SOT to the narrow definition of Net Interest Income.
o Commercial margins: We consider appropriate the inclusion of commercial margins in NII measures, as it enables the projection of the products’ total rate levels in order to correctly apply the repricing restrictions (floors/caps) of the balance sheet items. Besides, margins are essential to accurately estimate the NII projection levels.
o Constant balance sheet: We understand the motivation for applying a constant balance sheet as it eases the comparability among banks. Nevertheless, if applied “literally” as it is established in the Guidelines, we consider that it may lead to very unrealistic results. The current balance sheet structure obeys to a specific interest rates environment (e.g., low/negative rates), and it cannot be assumed that this profile will remain unchanged under a different interest rates scenario, especially when extreme shocks (e.g., +200bp) are applied.
This is especially relevant when applied to customer deposits that will evolve in a very different way in higher interest rates periods. The same consideration can be made regarding other balance sheet items, such as central banks deposits or TLTRO funding.
We suggest allowing flexibility to Banks to decide on the most appropriate approach to be used, dynamic or constant balance sheet, according to their internal risk management. Alternatively, we recommend relaxing the constant balance sheet assumption by allowing the conditional modelling to project the evolution of certain specific balance sheet items, according with the internal planning. This will also promote a higher consistency with the internal metrics that are disclosed by the bank.
o CET1 instruments exclusion: From an NII perspective, we consider that this assumption must be eliminated to avoid any confusion. Equity is part of the funding structure of the bank and should be taken into account in any case for NII projections.
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?
NII sensitivity is primarily used to assess the risk of volatility in forecasted NII due to interest rate variations. For that purpose, our view is that Metric 2 is more appropriate than metric 1 to evaluate a “large decline in the net interest income”.Metric 2 is conceptually more meaningful and a better representation of the risk under assessment as it relates NII shocks to the NII generation capacity (NII base projection as denominator). Thus, metric 2 recognizes the effect of changes in the interest rates level and balance evolution in the base projection, and not only in the variability of NII (NII shock). Nevertheless, we would appreciate more simple rules regarding its calculation. We would like to highlight that FINREP PL reports (F02.00) are only reported at the level of reporting currency but the breakdown by material currencies is not binding. Therefore, there are operational implications in the aggregation of NII baseline and Admin expenses by currency. The calculation of this metric can only be done at the level of reporting currency.
In addition, we identify a potential mismatching between FINREP and SOT perimeter, when significant currencies represent less than 100% of projected NII. Consequently, it should be allowed to adjust this percentage in the metric calculation.
As an alternative, and to simplify the approach,
we propose to use consolidated accounted NII and total expenses as denominator as reported in FINREP, instead of using the same aggregation method as for sensitivity, and/or
we suggest using a simplified metric 2, by eliminating the administrative costs deduction: (NII shock- NII baseline)/ NII baseline < [Threshold]
Metric 1, although simpler, is less accurate and representative for NII assessment, as TIER 1 is not an appropriate reference to evaluate changes on NII, which should be put in relation to the NII generation.
As an example, if interest rates rise, the variability of NII will be increased but also the NII generation capacity. This effect will only be correctly recognized by Metric 2.
Threshold level: Whatever the metric chosen; the threshold level should be re-calibrated taking in consideration the new lower bound levels. Additionally, the potential effects on the threshold level of exigency, under a higher (and/or positive) interest rates environment should be taken into account. Consequently, we recommend extending the analysis over time by collecting more data before defining a large decline. The threshold should be set based on a single (punctual) observation.
Furthermore, we consider that it is not appropriate to set the NII threshold based on the number of outlier banks obtained in the EVE SOT. In the first place, the relationship between NII generation and TIER1 is not consistent among banks. If this effect is not considered in the calibration, it will unreasonably penalize banks with greater income generation capacity and, therefore, an absolute (not necessarily relative) higher amount of income variation. This can be seen in the fact that the output of the 2 metrics proposed may result in a different set of outlier banks, which does not make sense. Following this, the management of IRRBB across banks may be very different from both perspectives, leading to wrong conclusions. For example, if banks are prioritising NII vs EVE management, this should generate a higher number of outlier banks in the EVE SOT. By forcing the NII threshold to get a target number of outlier banks would lead to an excessively low reference.
NII threshold should be calibrated independently from EVE according to a prudent risk level.
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?
Common modelling and parametric assumptions given in the guidelines at a very high level are clear. Nevertheless, we would appreciate a more concise position or clarifications about the following points:● SOTs would be conducted applying internal models, and parameters not specifically mentioned in the RTS should follow internal calibrations. We would appreciate it if this aspect could be explicitly mentioned in the RTS.
● The definition of NII shock scenarios (equal to EVE SOT parallel scenarios) is extremely unrealistic, especially for developed economies such as EUR and USD currencies, where the occurrence of sudden shocks of +/-200 bps is very unlikely. We suggest recalibrating downwards the size of the shocks for the SOT according to volatility levels, and/or applying gradual shocks (instead of sudden shocks) as it is more appropriate for dynamic measures.
Upload files
AEB CP on SOT (EBA-CP-2021-36) + EX SUM.pdf
(219.5 KB)