Commission Delegated Regulation (EU) 2024/397 of 20 October 2023 supplementing Regulation (EU) No 575/2013 of the European Parliament and of the Council with regard to regulatory technical standards on the calculation of the stress scenario risk measure

COMMISSION DELEGATED REGULATION (EU) 2024/397 of 20 October 2023 supplementing Regulation (EU) No 575/2013 of the European Parliament and of the Council with regard to regulatory technical standards on the calculation of the stress scenario risk measure (Text with EEA relevance)

THE EUROPEAN COMMISSION,

Having regard to the Treaty on the Functioning of the European Union,

Having regard to Regulation (EU) No 575/2013 of the European Parliament and of the Council of 26 June 2013 on prudential requirements for credit institutions and amending Regulation (EU) No 648/2012 (1), and in particular Article 325bk(3), fourth subparagraph, thereof,

Whereas:

(1) To ensure a level playing field among institutions in the Union and to minimise regulatory arbitrage, the methodologies for developing extreme scenarios of future shock for non-modellable risk factors should be based on the international standards agreed in January 2019 by the Basel Committee on Banking Supervision (BCBS) (Basel framework), and should take into account the materiality of the own funds requirements for non-modellable risk factors. Therefore, specific and detailed methodologies for developing extreme scenarios of future shock for non-modellable risk factors should be laid down.

(2) The quality of data and the number of observations that are available to determine future shocks for non-modellable risk factors may vary significantly from one non-modellable risk factor to another. It is therefore necessary to ensure that extreme scenarios of future shock cover a wide range of cases. For that reason, it is necessary to provide for alternative sets of methodologies that institutions may use depending on the quality and the number of observations that are available for each non-modellable risk factor. Furthermore, institutions should reflect in their calculations the fact that fewer available data leads to a higher uncertainty of the estimates or values used to determine the extreme scenarios of future shock and should therefore become more conservative.

(3) Given its accuracy, one method to determine the extreme scenario of future shock for a non-modellable risk factor should consist of directly calculating the expected shortfall measure of the losses that would occur when applying a shock to that non-modellable risk factor with the historically observed levels during the relevant stress period. However, such method would provide reliable results only where the institution has a significant amount of data for that stress period, and would require many loss calculations per risk factor leading to a high computational effort. It is therefore necessary to provide for an alternative method that requires a significant lower number of loss calculations and applies a stepwise approach. Under that alternative method, institutions should first calculate an expected shortfall measure on the returns observed for a non-modellable risk factor, and then calculate the loss that corresponds to the movement in the risk factor identified by that expected shortfall measure. Such stepwise approach should also address the specific case where the number of observations for a non-modellable risk factor in the stress period concerned is insufficient to obtain accurate and prudent estimates. Since that situation can be expected to occur only in a limited number of cases, those cases should be addressed by leveraging on methodologies that institutions have implemented for other non-modellable risk factors for which they have more observations or, where possible, on the alternative standardised approach.

(4) The Basel framework requires that the market risk own funds requirements for non-modellable risk factors have to be calibrated to a period of stress that is the same for all non-modellable risk factors that belong to the same broad category of risk factors. To determine extreme scenarios of future shock on the basis of data observed during that identified period, institutions should collect data for non-modellable risk factors for that identified stress period.

(5) To ensure harmonisation of the calculation of the stress scenario risk measures across institutions in the Union, it is necessary to specify how institutions should identify the stress period. Those specifications should be proportionate to the purpose, and should neither require an excessive computational effort, nor the implementation of specific pricing methods. The global financial crisis of 2007–2008 has been a major stress event for the financial system. The stress period to be identified should therefore start at least from 1 January 2007. To ensure that the stress period remains relevant to an institution’s trading portfolio, institutions should periodically review that stress period. However, to limit the administrative burden on institutions, it should be only required that the frequency of such review follows at least the same quarterly frequency as the corresponding supervisory reporting.

(6) The Basel framework requires that institutions determine extreme scenarios of future shock by using the pricing methods of their risk measurement model, as those methods are used in the context of the back-testing and the profit and loss attribution test. There may be scenarios of future shock for which those pricing methods cannot determine the corresponding loss for some financial instruments or commodities. Where that is the case, institutions should act in a prudentially sound manner and target only those instruments that are affected by the pricing failure. The methodologies implemented by the institution to address those cases are not to affect in any way the results of the back-testing and the profit and loss attribution requirements laid down in Commission Delegated Regulation (EU) 2022/2059 (2).

(7) Article 325bk(3), second subparagraph, of Regulation (EU) No 575/2013 requires that the level of own funds requirements for market risk for a non-modellable risk factor is to be as high as the expected shortfall measure for that risk factor referred in Article 325bb of that Regulation, i.e. an expected shortfall of losses at a 97,5 % confidence level over a period of stress. The statistical estimators and the parameters for determining that expected shortfall measure should therefore be set in such a way that that confidence level is met.

(8) According to the Basel framework, the regulatory extreme scenario of future shock should be the one leading to the maximum loss that may occur due to a change in the non-modellable risk factor. It should therefore be specified what institutions should consider as maximum loss in cases where the maximum loss is not finite.

(9) To ensure consistency with the Basel framework, institutions should be able to determine the stress scenario risk measure for more than one non-modellable risk factors where those non-modellable risk factors are part of a curve or a surface, and where those risk factors belong to the same non-modellable bucket among those set out in Commission Delegated Regulation (EU) 2022/2060 (3), and provided that institutions have assessed their modellability in accordance with the standardised bucketing approach referred to in that Delegated Regulation. Institutions should therefore be allowed to compute a single stress scenario risk measure for more than one non-modellable risk factor under those conditions only.

(10) To ensure the adequacy of the own funds requirements for non-modellable risk factors with the risk profile of institutions, institutions should reflect in the aggregation of the stress scenario risk measures those risks that were not yet captured when determining the extreme scenario of future shock, including the liquidity horizons of the non-modellable risk factors. To ensure a level playing field, the stress scenario risk measures should be aggregated by applying the aggregation formula agreed in the Basel framework.

(11) This Regulation is based on the draft regulatory technical standards submitted to the Commission by the European Banking Authority.

(12) The European Banking Authority has conducted open public consultations on the draft regulatory technical standards on which this Regulation is based, analysed the potential related costs and benefits, and requested the advice of the Banking Stakeholder Group established in accordance with Article 37 of Regulation (EU) No 1093/2010 of the European Parliament and of the Council (4),

HAS ADOPTED THIS REGULATION:

CHAPTER 1

DEVELOPMENT AND APPLICATION OF THE EXTREME SCENARIOS OF FUTURE SHOCK

Article 1

Development of extreme scenarios of future shock and their application at risk factor level

Institutions shall develop the extreme scenarios of future shock for non-modellable risk factors by applying either of the following methods:

(b) the stepwise method set out in Article 3.

Article 2

Direct method – non-modellable risk factors

(b) they shall calculate the estimate of the right-tail expected shortfall in accordance with Article 11(2) for the time series of losses obtained in accordance with point (a).

Article 3

Stepwise method – non-modellable risk factors

(a) they shall, in accordance with Article 7, determine the time series of 10 business days returns for the non-modellable risk factor for the stress period determined in accordance with Article 12;

Article 4

Development and application of the extreme scenarios of future shock at standardised bucket level

Where institutions calculate a stress scenario risk measure for more than one non-modellable risk factor, they shall determine the extreme scenario of future shock for the non-modellable standardised bucket to which those risk factors belong in accordance with Delegated Regulation (EU) 2022/2060 by applying either of the following methods:

(b) the stepwise method set out in Article 6.

Article 5

Direct method – non-modellable standardised buckets

(b) they shall calculate the estimate of the right-tail expected shortfall in accordance with Article 11(2) for the time series of the losses obtained in accordance with point (a) of this paragraph.

Article 6

Stepwise method – non-modellable standardised buckets

(a) for each non-modellable risk factor within the non-modellable standardised bucket they shall, in accordance with Article 7, determine the time series of 10 business days returns for the stress period determined in accordance with Article 12;

For the purposes of point (c), institutions shall multiply the upward and downward calibrated shocks by the parameter β in two cases, with β = 1 and β = ⅘.

Article 7

Determination of the time series of 10 business days returns

(a) they shall determine the time series of observations for the non-modellable risk factor for the stress period and include in that time series only one observation per business day that shall represent actual market data;

(b) they shall extend the time series referred to in point (a) by including the observations available within the period of 20 business days following the stress period; where the reference date for the calculation of the stress scenario risk measure is less than 20 business days after the end of the stress period, institutions shall include those observations that are available from the end of the stress period to the reference date;

Article 8

Downward and upward calibrated shock with the historical method

where:

— Ret denotes the time series of 10 business days returns of the non-modellable risk factor;

— is the estimate of the left-tail expected shortfall for the time series Ret calculated in accordance with Article 11(1);

— N is the number of returns in the time series Ret.

where:

— Ret denotes the time series of 10 business days returns of the non-modellable risk factor;

— is the estimate of the right-tail expected shortfall for the time series Ret calculated in accordance with Article 11(2);

— N is the number of returns in the time series Ret.

Article 9

Downward and upward calibrated shock with the asymmetrical sigma method

Under the asymmetrical sigma method, institutions shall determine the downward and upward calibrated shock from a time series of 10 business days returns for a non-modellable risk factor by applying the following steps in the following order:

(b) for each subset referred in point (a), they shall compute the mean of the 10 business days returns in the subset;

Article 10

Downward and upward calibrated shock with the fallback method

(a) they shall identify the risk-weight assigned to that risk factor in accordance with Part Three, Title IV, Chapter 1a, of Regulation (EU) No 575/2013;

(c) the downward and upward calibrated shock shall be the result of point (b).

(a) from those risk factors defined in Part Three, Title IV, Chapter 1a, Section 3, Subsection 1, of Regulation (EU) No 575/2013 differing from the non-modellable risk factor only in the maturity dimension, they shall identify the risk factor that is the closest in the maturity dimension to the non-modellable risk factor;

(b) they shall identify the risk-weight assigned in accordance with Part Three, Title IV, Chapter 1a, of Regulation (EU) No 575/2013 to the risk factor identified in accordance with point (a);

(d) the downward and upward calibrated shock shall be the result of point (c).

(a) it belongs to the same broad risk factor category and broad sub-category of risk factors referred to in Article 325bd of Regulation (EU) No 575/2013 of the non-modellable risk factor;

(b) it is of the same nature as the non-modellable risk factor;

(c) it differs from the non-modellable risk factor for features that do not lead to an underestimation of the volatility of the non-modellable risk factor, including under stress conditions;

(d) its time series of 10 business days returns referred to in paragraph 6, point (a), contains at least 12 returns.

(a) for the selected risk factor, institutions shall, in accordance with Article 7, determine the time series of 10 business days returns for the stress period determined in accordance with Article 12;

(a) the historical method set out in Article 8, where the number of returns in the time series of 10 business days returns referred to in paragraph 6, point (a), is equal to or greater than 200 for all the selected risk factors;

(b) the asymmetrical sigma method set out in Article 9, where the condition referred to in point (a) of this paragraph for applying the historical method is not met.

Article 11

Estimators of the expected shortfall

where:

— N is the number of observations in the time series;

— α = 2,5 %;

— ] denotes the integer part of the product α ×N;

— denotes the i-th smallest observation in the time series X.

where:

— ( ) is the estimate of left-tail expected shortfall for the time series  calculated in accordance with paragraph 1.

Article 12

Determination of the stress period

where:

— i denotes the broad risk factor category;

— j is the index denoting the non-modellable risk factors or the non-modellable standardised buckets for which the institution calculates the stress scenario risk measure belonging to the broad risk factor category i;

— is the rescaled stress scenario risk measure for the non-modellable risk factor or the non-modellable standardised bucket j calculated in accordance with Article 16.

Article 13

Computation of the losses

(a) they shall identify those financial instruments or commodities and the cause of the failure of the pricing calculation;

(b) they shall use sensitivity-based pricing methods, including at least the material first order and material second order terms of Taylor series approximations, to reflect the change in the price of those financial instruments or commodities due to changes in the non-modellable risk factors in this scenario of future shock.

CHAPTER 2

REGULATORY EXTREME SCENARIO OF FUTURE SHOCK

Article 14

Determination of the regulatory extreme scenario of future shock

(a) they shall use an expert-based approach using qualitative and quantitative information available to identify a loss due to a change in the value taken by the non-modellable risk factor that will not be exceeded with a level of certainty equal to 99,95 % on a 10 business day horizon in a future period of financial stress equivalent to the stress period identified for the non-modellable risk factor. When doing so, institutions shall take into account the skewness and the excess kurtosis that may characterise the returns of the non-modellable risk factor in a period of financial stress and shall justify any distributional or statistical assumptions taken for identifying that loss;

(c) they shall identify the regulatory extreme scenario of future shock as the shock leading to the loss resulting from points (a) and (b).

(a) they shall use an expert-based approach using qualitative and quantitative information available to identify a loss due to a change in the values taken by the non-modellable risk factors that will not be exceeded with a level of certainty equal to 99,95 % on a 10 business day horizon in a future period of financial stress equivalent to the stress period for the non-modellable risk factors. When doing so, institutions shall take into account the skewness and the excess kurtosis that may characterise the returns of the non-modellable risk factors in a period of financial stress and shall justify any distributional or statistical assumptions taken for identifying that loss;

(c) they shall identify the regulatory extreme scenario of future shock as the scenario leading to the loss resulting from points (a) and (b).

CHAPTER 3

CIRCUMSTANCES UNDER WHICH INSTITUTIONS MAY CALCULATE A STRESS SCENARIO RISK MEASURE FOR MORE THAN ONE NON-MODELLABLE RISK FACTOR

Article 15

Circumstances for the calculation of a stress scenario risk measure for more than one non-modellable risk factor

The circumstances under which institutions may calculate a stress scenario risk measure for more than one non-modellable risk factor as referred to in Article 325bk(3), point (c), of Regulation (EU) No 575/2013 shall be the following:

(a) the risk factors belong to the same standardised bucket as referred to in Article 5(2) of Delegated Regulation (EU) 2022/2060;

(b) institutions assessed the modellability of those risk factors, by determining the modellability of the standardised bucket to which they belong in accordance with Article 4(1) of Delegated Regulation (EU) 2022/2060.

CHAPTER 4

AGGREGATION OF THE STRESS SCENARIO RISK MEASURES

Article 16

Aggregation of the stress scenario risk measures

where:

— ICSR denotes the set of non-modellable risk factors or non-modellable standardised buckets for which institutions determined a stress scenario risk measure that was classified as reflecting idiosyncratic credit spread risk only, in accordance with paragraph 3;

— k is an index denoting the non-modellable risk factors or non-modellable standardised buckets belonging to ICSR ;

— EIR denotes the set of non-modellable risk factors or non-modellable standardised buckets for which institutions determine a stress scenario risk measure that was classified as reflecting idiosyncratic equity risk only, in accordance with paragraph 4;

— l is an index denoting the non-modellable risk factors or non-modellable standardised buckets belonging to EIR ;

— OR denotes a non-modellable risk factor or non-modellable standardised bucket for which institutions determine a stress scenario risk measure that was neither classified as reflecting idiosyncratic credit spread risk only, in accordance with paragraph 3, nor idiosyncratic equity risk only, in accordance with paragraph 4;

— j is an index denoting the non-modellable risk factors or non-modellable standardised buckets belonging to OR;

— are respectively the rescaled stress scenario risk measures for the non-modellable risk factors or the non-modellable standardised buckets k,l, j calculated in accordance with paragraph 1;

— .

(a) the nature of the risk factor is such that it reflects idiosyncratic credit spread risk only;

(b) the value taken by the risk factor is not driven by systematic risk components;

(c) the correlation among risk factors is negligible;

(d) institutions perform and document the statistical tests used to verify the condition set out in point (c).

(a) the nature of the risk factor is such that it reflects idiosyncratic equity risk only;

(b) the value taken by the risk factor is not driven by systematic risk components;

(c) the correlation among risk factors is negligible;

(d) institutions perform and document the statistical tests used to verify the condition set out in point (c).

Article 17

Non-linearity coefficient for a single risk factor

Where the stress scenario risk measure for which institutions are determining the non-linearity coefficient has been determined for a single risk factor, such non-linearity coefficient shall be determined as follows:

Article 18

Non-linearity coefficient for a bucket

Where the stress scenario risk measure for which institutions are determining the non-linearity coefficient has been determined for a non-modellable standardised bucket, the non-linearity coefficient shall be determined as follows:

Article 19

Calculation of the estimate of the tail parameter

Institutions shall calculate the estimate of the tail parameter for a given non-modellable risk factor as follows:

Article 20

Calculation of the uncertainty compensation factor

where:

— N is the number of losses in the time series referred to in Article 2(1), point (a)(iii), from which the extreme scenario of future shock has been determined for the non-modellable risk factor in accordance with that Article.

where:

— N is the number of losses in the time series referred to in Article 5(1), point (a)(iv), from which the extreme scenario of future shock has been determined for the non-modellable bucket in accordance with that Article.

CHAPTER 5

QUALITATIVE REQUIREMENTS

Article 21

Documentation of the criteria and methods

For the purposes of developing extreme scenarios of future shock, determining the regulatory extreme scenario of future shock, and aggregating the stress scenario risk measures, the set of internal policies referred to in Article 325bi(1), point (e), of Regulation (EU) No 575/2013, shall include documentation of any information necessary to demonstrate that the applicable criteria and methods laid down in this Regulation are complied with, in particular in relation to criteria on the application of choices, assumptions made, conditions, required steps for applying the derogations, and justifications, where applicable.

CHAPTER 6

FINAL PROVISIONS

Article 22

Entry into force

This Regulation shall enter into force on the twentieth day following that of its publication in the Official Journal of the European Union.

This Regulation shall be binding in its entirety and directly applicable in all Member States.

Done at Brussels, 20 October 2023.

For the Commission The President Ursula VON DER LEYEN

(1) OJ L 176, 27.6.2013, p. 1.

(2) Commission Delegated Regulation (EU) 2022/2059 of 14 June 2022 supplementing Regulation (EU) No 575/2013 of the European Parliament and of the Council with regard to regulatory technical standards specifying the technical details of back-testing and profit and loss attribution requirements under Articles 325bf and 325bg of Regulation (EU) No 575/2013 (OJ L 276, 26.10.2022, p. 47)

(3) Commission Delegated Regulation (EU) 2022/2060 of 14 June 2022 supplementing Regulation (EU) No 575/2013 of the European Parliament and of the Council with regard to regulatory technical standards specifying the criteria for assessing the modellability of risk factors under the internal model approach (IMA) and specifying the frequency of that assessment under Article 325be(3) of that Regulation (OJ L 276, 26.10.2022, p. 60).

(4) Regulation (EU) No 1093/2010 of the European Parliament and of the Council of 24 November 2010 establishing a European Supervisory Authority (European Banking Authority), amending Decision No 716/2009/EC and repealing Commission Decision 2009/78/EC (OJ L 331, 15.12.2010, p. 12).