2011-06-15

Instruction No. 2011-I-10 on the Monitoring of Internal Models Used for Calculating Capital Requirements

The Prudential Control Authority issued Instruction No. 2011-I-10 to establish reporting requirements for credit institutions and investment firms using internal models for regulatory capital calculations. The instruction mandates the periodic submission of detailed information regarding extensions, significant evolutions, and governance changes of these models to ensure prudent supervision. Institutions must submit annual reports by April 30, 2012, utilizing a standardized framework defined in the instruction's annexes to facilitate the supervisor's oversight of model integrity and performance.

Autorite de Controle Prudentiel et de Resolution logo

France

Autorite de Controle Prudentiel et de Resolution

Click to view thumbnail

PRUDENTIAL CONTROL AUTHORITY Instruction No. 2011-I-10 On the Monitoring of Internal Models Used for Calculating Capital Requirements

The Prudential Control Authority, Having regard to the Monetary and Financial Code, particularly Article L. 612-24; Having regard to the Decree of February 20, 2007, as amended, relating to capital requirements applicable to credit institutions and investment firms; Having regard to the Regulation of the Committee for Banking and Financial Regulation No. 2000-03 of September 6, 2000, as amended, relating to prudential supervision on a consolidated basis; Having regard to Instruction No. 2007-02, as amended, of the Banking Commission relating to capital requirements applicable to credit institutions and investment firms; Having regard to the opinion of the Consultative Commission for Prudential Affairs dated May 26, 2011; Decides:

Article 1 This Instruction applies to institutions subject to the aforementioned Decree that have been authorized by the Prudential Control Authority, or previously by the Banking Commission, to use internal models for the calculation of their capital requirements.

Article 2 Subject institutions are required to communicate to the Prudential Control Authority any useful information regarding changes made to the authorized internal models used for calculating capital requirements. Without prejudice to the provisions of the aforementioned directives and Decree, this Instruction defines the periodic information expected by the Prudential Control Authority in cases of extension or evolution of internal models used for regulatory purposes.

Article 3 For internal models used for regulatory purposes, extension is defined as the application of a model, authorized by the supervisor, to an entity, activity, or portfolio not foreseen in the scope of the initial authorization. Model evolutions are defined as any change impacting the conditions for prudential recognition of internal models for the calculation of capital requirements.

Article 4 Annex 1a of this Instruction provides a list of examples of evolutions requiring particular attention in the information addressed to the Prudential Control Authority as early as possible before any implementation for regulatory purposes. In addition, Annex 1b provides a list of examples of evolutions that must be included in this report as part of the information enabling the Prudential Control Authority to have a global view of changes made to the models. Annex 1a is intended to guide institutions in organizing the information to be transmitted to the Prudential Control Authority. Based on feedback, it will be regularly updated by the General Secretariat of the Prudential Control Authority in consultation with the industry. It does not exhaustively and normatively list all possible scenarios. In this regard, it does not in any way exempt institutions from ensuring compliance with applicable regulations through their own criteria and experience.

Article 5 Significant extensions or evolutions involving a parent credit institution in the Union and its subsidiaries, or subsidiaries of a parent financial holding company in the Union, fall under the provisions of Article 129-2 of Directive 2006/48/EC.

Article 6 Subject institutions shall annually, and no later than six months following the December 31 cutoff, submit all relevant information elements regarding the monitoring of the use of models used for regulatory purposes and their current and future evolutions, following the framework defined in Annex 2 of this Instruction. Based on feedback, this framework will be regularly updated by the General Secretariat of the Prudential Control Authority in consultation with the industry. Subject institutions shall transmit the aforementioned reporting to the General Secretariat of the Prudential Control Authority in two copies at the following address: General Secretariat of the Prudential Control Authority 61, rue Taitbout 75436 Paris Cedex 09 and in electronic form. The reporting on the monitoring of the use of models used for regulatory purposes must be accompanied by a letter signed by one of the managers responsible for the institution as defined in Article L. 511-13 of the Monetary and Financial Code. This letter must include the institution's commitment that the documents and information provided present an accurate and faithful summary and omit no significant facts.

Article 7 This Instruction enters into force on the date of its signature. The first submission of the report referred to in Article 6 shall be made no later than April 30, 2012. Paris, June 15, 2011 The President of the Prudential Control Authority, [Christian NOYER]

Annex 1 Lists of Examples of Evolutions Requiring Information in the Framework of Monitoring Validated Internal Models This Annex aims to provide two lists of examples (non-exhaustive) to guide institutions in organizing the information to be transmitted to the General Secretariat of the Prudential Control Authority.

Annex 1a Prior Authorization Any change related to the regulatory method used for calculating capital requirements (non-exhaustive list) For credit risk: – Transition from the Standardized Approach to an Internal or Advanced Approach – Transition from the Foundation IRB (IRBF) to the Advanced IRB (IRBA) – Treatment of double default – Use of the regulatory formula for calculating capital requirements on investor securitization positions – Return to the Standardized Approach for part of the authorized scope For operational risk: – Transition from the Standardized Approach to the AMA – Inclusion of insurance For market risk: – Transition from the Standardized Approach to an internal model approach for an activity or a type of risk (i.e., specific risk) – Return to the Standardized Approach for a portfolio or activity that is part of the initially authorized regulatory VaR scope Extension of model use (i.e., roll-out, application of an authorized model to a new activity, etc.), examples (non-exhaustive list) For credit risk: – Use of a model developed for the portfolio of Bank X applied to the exposures of Bank Y following a group reorganization. The institution must then explain and justify its validation procedures and the relevance of the extension. For operational risk: – Extension or significant change relative to the initially authorized scope of use of the AMA method. For market risk: – Extension or significant change relative to the initially authorized scope for the internal model. New definition of the main model parameters (non-exhaustive list) For credit risk: – Implementation of a new default indicator. For operational risk For market risk

Major methodological overhaul or new methodology for estimating fundamental model parameters (non-exhaustive list) For credit risk: – Modification of the rating model – Major change or recalibration impacting the segmentation of risk classes – New methodology for estimating Basel parameters (PD, LGD, FCEC, EAD) For operational risk: – Transition from a statistical approach to a scenario-based approach – Major change in distribution law assumptions and dependence structure or parameter estimation (inclusion of correlations) For market risk: – Significant changes in selected risk factors or their distribution law assumptions leading to a significant decrease in calculated capital requirements – Significant change in the method for calculating potential loss distributions (e.g., transition from full revaluation to an approximation using sensitivities) Major change in internal model input data (non-exhaustive list) For credit risk: – Evolutions in the scope of accounting or financial data significantly affecting the measurement of risk dimensions covered by the model. For operational risk: – Major change in the structure and characteristics of source data (use of a new external database, modification of the historical depth used, etc.) Migration/change in data stock or data quality having a major impact on the main model parameters or the fundamental conditions for operational integration. Significant modification of governance, organization, or internal model validation methodologies. Removal or evolutions of add-ons or prudential margins.

Annex 1b Annual Post-facto Information Implementation of recommendations formulated by the Prudential Control Authority. Adaptation of the definition or measurement of observed risk parameters (default rate, loss rate, EAD) following environmental evolutions (accounting standards, payment systems, law, etc.). Non-major modification of risk class segmentation, risk parameter estimators, prudential margin determination methods, etc. Routine model modifications: rebalancing of weights, data quality improvement program, etc. Addition of an additional add-on on RWA or a prudential margin on non-significant portfolio parameters. Non-significant modifications of governance, organization, or internal model validation methodologies.

Annex 2 Framework for Annual Reporting on the Monitoring of Internal Models Used for Calculating Capital Requirements The annual reporting on the monitoring of internal models used for calculating capital requirements aims at three objectives. It is intended to present annually: (i) the updated mapping of scope evolutions and models (sections 1 and 2), (ii) evolutions in the model environment and their performance (sections 3, 4 and 5), and (iii) the future work program regarding monitoring and evolutions (section 6). The framework below specifies the main sections of the reporting illustrated with some examples to facilitate the preparation of said reporting. Institutions are invited to include any additional information deemed relevant. Based on this information, and particularly that presented in section 6, the General Secretariat of the ACP will establish, where appropriate, a schedule for the transmission of additional information with the institutions. In particular, the validation method for a model evolution planned by the institution may be determined based on the results of the institution's internal audit, which notably includes impact studies. N.B.: Due to regulatory evolutions affecting the measurement of market risks (IRC and stressed VaR), the transmission of information for the 2011 fiscal year is optional for market risk models within the framework defined by this Instruction. Section 5, which presents the prospective work program, does not refer to the 2011 fiscal year and must therefore be completed.

  1. Institution's Situation for "Roll-out" Status of subsidiaries' entry and exit, other scope evolutions, evolutions inducing changes in home-host relationships. A synthetic snapshot of the roll-out process is desired; this can be done by referring to documents already sent elsewhere to the SGACP.

  2. General Summary of Evolutions Occurred The main evolutions carried out in application of supervisor requests will be highlighted. a. Evolutions occurring on credit risk and operational risk models i. Scope: entity or portfolio concerned by the model evolution ii. Model name: designation of the model as appearing in the authorization request file iii. Affected exposures (for credit risk): initial gross amount and risk-weighted exposure of exposures impacted by the model evolution (in absolute value and in relative value compared to total exposures of the business line, entity, or reference Basel category) iv. Date of the last validated version of the model v. Reminder of the main evolutions occurred (since the authorization to use the model for regulatory purposes) vi. Synthetic description of the evolution: objectives of the evolution, translation into modeling choices, etc. vii. Impacted parameters: analysis of the main impacted parameters (in static and dynamic terms, backtesting and benchmarking elements, etc.) viii. Impact measurement: estimation of the impact of the model evolution on main risk parameters, risk-weighted assets, expected losses, and capital requirements, impact on operational steering tools and operational integration if applicable, etc. ix. Dates of internal audit review and conclusions of the last audit x. Body and validation methods: body or committee in charge of validation, examination methods (content of the validation file) xi. Date of production deployment of the evolution b. Evolutions occurring on market risk models (according to the institution's format) The reporting should indicate at least the significant modifications made (addition, removal of risk factors, modifications of the length of the history retained for the calibration of risk factor evolution laws...). Backtesting results must also continue to be transmitted.

  3. Evolutions in Governance and Operational Integration Methods a. Governance and model validation Roll-out monitoring process, model validation procedures, monitoring of model performance and proper use, decision-making to evolve models (according to the type of evolution) and validation of evolutions, application calendar, etc. b. Evolutions occurring in "Operational Integration" Indicative and non-limitative list. i. Evolutions in contributors (to ensure the entry of qualitative data, for example, or the inclusion of expert judgment on certain modules) ii. Evolutions in users (more decentralized or centralized use of the model...) iii. Role in rating in the credit granting decision, in the delegation threshold, in risk pricing, in the determination of limits, in the collateral policy, in the definition of provisions, in recovery management, in the remuneration policy of staff... iv. Use of the model in the design of stress tests

  4. Quality Assurance Actions Implemented Concerning Models Institutions are invited to provide a general presentation of their quality assurance approach. a. Controls related to data freshness b. Controls related to the responsiveness of indicator updates Examples of indicators for credit models: time between the inclusion of a rating-determining piece of information and the update of the rating, frequency of rating updates (commented descriptive statistics). c. Other indicators at the institution's choice, with commentary d. Presentation of operational incidents related to the implementation of models e. Quality improvement axes selected for the following fiscal year

  5. Key Events in Performance Institutions are invited to provide a synthetic presentation of the main key events related to model performance. The presentation can be made by type of risk (credit / market / operational) or by model family.

Type of risk or model family Main indicators followed Results: overall view of performance for the year General trends across all models (factors generally robust, factors calling for attention) Models that raised specific difficulties – due to the difficulty of interpreting backtesting results – due to "negative" backtests

  1. Multi-year Work Program and Projects for Model Evolutions and Extensions Institutions are invited to present their work program regarding the monitoring and evolutions of models. This presentation will strengthen communication modalities with the supervisor, best anticipate cases requiring prior authorization, and determine at the appropriate time the examination methods for the prior authorization request. a. Introduction of new models b. Model extensions c. Model revisions d. Model quality control e. Evolutions of operational integration methods f. Internal audit programs for models