Harmonisation of internal model approaches, a new era for banks?

Tue 21 Jun 2016

Basel 4 and Single Supervisory Mechanism act to reduce the excessive variability in the results of internal model approaches to credit risk.

For more than ten years, Basel reform has encouraged the development and use of internal models designed to better place risk management at the heart of banks’ control arrangements. Basel II saw massive banking recourse to internal model approaches. The successive crises experienced since 2008 have led, in the framework of Basel III, to reinforced requirements involving improved attention to risks such as securitisation and derivatives , as well as specific recognition of liquidity risk and a greater solvency requirement via the determination of capital.

During the same period, internal models have become more sophisticated and have progressively evolved by integrating the specific features of individual banking institutions reinforced by historical (backward-looking) observation and/or by expert judgement. But recent work by the EBA, culminating in its November 2015 report on banks’ transparency, has disclosed excessive variability in the results for the calculation of RWAs (risk-weighted assets) from one bank or jurisdiction to the next. Those differences hinder the comparability of institutions the Basel Committee strives for. So in March 2016, the Committee launched a consultation on reducing variation in credit risk-weighted assets by imposing constraints on the use of internal model approaches.

Banks have until 24 June to provide input to the consultation with a view to finalising the text by the end of 2016.

A far more restrictive regulatory framework

To limit variability in the calculation of credit risk-weighted assets, the Basel Committee proposes restricting the use of internal rating-based approaches and models to those activities, presenting sufficient significant data and historical loss experience to achieve due assessment of the applicable risks. The proposal thus automatically sets aside the application of internal models to LDPs (Low- Default Portfolios)^[1], equities and specialised lending. In particular, the Committee thinks that:

• For the LDPs, “the low-default nature of the assessed portfolios, and the consequent lack of appropriate data for risk parameter estimation, was likely one of the key factors leading to differences across banks”;
• For equities, “it is unlikely that banks will have specific knowledge concerning the issuer over and above public data…, it is difficult to justify capital requirements for these exposures varying between banks”;
• For specialised lending, “banks are typically unlikely to have sufficient data to produce reliable estimates of PD and LGD” (only the standardised approach and the current IRB supervisory slotting approach are thus left).

It equally sets aside the internal models approach for credit valuation adjustment risk (IMA-CVA), finally judged inopportune despite the consultation on the subject launched in July 2015^[2]. The Basel Committee further defines the criteria to be respected by banks wishing to develop their own internal models:

• Both quantitative and qualitative sufficiency of the data correlated to the bank’s risk parameters;
• The availability of particularly useful information or of specific knowledge related to each activity and risk to be measured;
• The use of robust, generally recognised and easily verifiable modelling techniques.

In order to limit the scope and heterogeneity of internal ratings-based approaches, the Committee suggests the imposition of a number of constraints for the estimation of model parameters including:

• The definition of rating systems offering greater stability over time (thereby favouring a forward-looking approach with changes in rating generally due to idiosyncratic or industry-specific changes rather than to business cycles subject to inherent fluctuation);
• Greater weighting of recessionary or downturn phases in the data used to calculate PDs and LGDs;
• A minimum level of granularity when calculating PDs by rating grade;
• Rules for simplification of the calculation of LGDs (for example, by removing the minimum collateralisation requirement for secured LGDs);
• New prudential LGD values;
• Additional floor requirements for PD, LGD and EAD/CCF;
• New constraints for internal modelling of CCF;
• A precise definition of the maturity parameter under the A-IRB approach.

The criteria for the eligibility and recognition of guarantees and collateral for CRM (credit risk mitigation) purposes are also revisited with:

• Removal of the partial substitution approach to calculating PDs;
• Removal of the double default treatment;
• The exclusion of conditional guarantees under the A-IRB approach;
• Removal of the option to use own estimates of haircuts under the F-IRB approach;
• Removal of any recognition of CRM arising from first-to-default credit derivatives other than for A-IRB banks.

Finally, to limit still further the potential variability in results, the Committee proposes to define more granular additional RWA floors to complete the overall floor of between 60% and 90% of total RWAs calculated in accordance with the new standardised approach.

After AQR, TRIM

Following its asset quality review (AQR), and in accordance with the general guidance provided by the Basel Committee, the ECB’s objective is to harmonise the use of the internal model approaches of banks placed under its jurisdiction. Within this framework, the ECB envisages undertaking a targeted review of internal models spread over several years. The review would be largely complemented by quantitative impact studies (QISs) and would serve in particular to refine and calibrate the Basel Committee’s proposals. The ultimate aim of the work is to pose the foundations of a level playing field for the use of internal model approaches and to provide the ECB, in its role of Single Supervisor, a better vision of the risks at stake whilst at the same time reinforcing its control. Baptised the targeted review of internal models (TRIM), there is no doubt that the comparability and transparency of banks will be improved as a result.

What are the issues at stake and the potential consequences for banks?

The issues at stake in the framework of such ambitious harmonisation are potentially significant in terms of RWAs and capital requirements. The panorama of the risks with which banks are confronted will be modified. Banks’ strategies will also be impacted: risk appetite will need to be reviewed in order to take account of changes in profitability/risk spread ratios and in the risk sensitivity of activities in the event of crisis via the modelling of stress test scenarios. This may require banks to review their commitment policies for particular industries, geographical zones etc. This calibration exercise is of crucial importance. The regulatory aim is to find the right modelling balance between robustness of RWAs, simplicity and risk-sensitivity. The impacts in terms of RWAs need to be scrupulously assessed. This is why, on 5 May 2016, the Basel Committee published additional guidance for QISs – on IRB (Internal Rating Based) approaches began in April. In the framework of the Basel Committee’s twice-yearly monitoring exercise, the QIS requires publication of the results of the study in September 2016 based on data as of 31 December 2015.

The operating consequences of the harmonisation will equally be significant.

Internal practices, processes and procedures will need to be reviewed and adapted in line with the new requirements. The applicable models will need to be redefined and revalidated^[3], as will the applicable rating systems. Mapping, data collection and back-testing will also require adjustment. The ex-ante calculation of risks developed in tools for assisting in credit underwriting, as well as limits and indicators with impacts for operational monitoring and risk control, will equally be affected. It is also important to note that the possibility of discontinuity of historical data, and the calibrations applied to the data, would result in a significant increase in the associated modelling risk. What solutions might then be capable of minimising that risk?

The process of adaptation will thus take time and require, on the part of the financial institutions, heavy investment which will need to be spread over several years.

To sum up the situation, banks currently applying internal model approaches are going to be obliged to unravel their systems, processes and organisation, which represent a significant part of the solutions they have implemented during the last ten years, in order to substitute new,more restrictive and standardised rules.

Banks incited to ever greater rigour and transparency

New Guidelines and RTSs (Regulatory Technical Standards), outlining rules for the implementation and use of internal model approaches, will be published by the EBA during 2016 and 2017. Discussions are in progress. Procedures for evaluating the quality of those methods will equally be implemented by the supervisor with a view to any appropriate adjustment of requirements based on the discussions in progress and on banks’ impact studies which will be renewed in 2017. As a result, banks must now prepare to engage in TRIM and begin preparations for the review of selected models by . performing tests, draft documentation, organise teams of experts, and respond to questionnaires and on-site visits .

However a certain number of questions have yet to be resolved:

• How will the supervisor select the models to be reviewed, and on the basis of what criteria?
• What methodology will be retained for TRIM performance?
• What process will be applied for the validation or revalidation of the internal models selected, and in accordance with what timetable?

Banks may expect to be subject to much greater rigour, yet less freedom in  internal model approaches at their disposition in line with the current trend to implement less sophisticated models, but requiring greater transparency in their use. The shift represents a major preoccupation with reinforcing models’ specifications and documentation and with the availability and quality of the information treated, rather than with the design of algorithms and formulae which tend to be subject to massive simplification.

Reminder of the importance of governance and internal control

In this context, it becomes crucial for banks to have defined and implemented appropriate governance for effective steering of the design and review of their internal models including validation of parameters and models, monitoring of the rating system and Basel compliance, and specialist committees . In the same way, banks’ systems of internal control will need to be reinforced through, for example, verification of the existence, completeness and pertinence of models’ documentation, and adequacy of the historical timeline and quality of data. In particular, systems will need to be capable of reliably confirming at all times the adequacy of the internal model approaches used in terms of their compliance with the new requirements for harmonisation. In this respect, it may be noted that the importance of governance and internal control is also explicitly highlighted in the framework of ICAAP and ILAAP^[4] of which the harmonisation has also become an imperative component of the SSM.

The review of internal model approaches to credit risk may result in profound modification of banking strategies

In continuation of the new requirements imposed in the framework of Pillar 2 and with regard to internal risk management arrangements, the current aim of the Basel Committee and the ECB to harmonise the internal model approaches deployed in the framework of Pillar 1 clearly reflects the SSM’s philosophy of regulation of financial institutions’ functioning modes. But by imposing a restrictive framework, the authorities also limit the scope for banks to take due regard of their individual characteristics, thereby indirectly jeopardising their very DNA relating to the content of their strategy, their activities, their historical heritage and their experience. In contrast, up until now the internal model approaches which were accepted were largely based on those elements and clearly reflected their substance. For example, the disparity from one bank to another of loss given default is largely the reflection of differences in banks’ approaches to debt recovery and in the specific features of the related sureties and surety practices. Moreover, those mechanisms take account of legal specifics that could be different from one country to another. In this respect, what are the potential consequences of the imposition of LGD floors?

More generally, the imposition of minimum requirements on banks raises the question of how to adapt their business models to the new constraints, in particular in France and in Europe. For example, is there a risk of speculative bubbles for certain asset classes such as real estate?

If the Basel Committee’s proposals remain unchanged, it may be expected that banks’ strategies will undergo significant and lasting evolution. Banks will have to review their risk appetites and draw consequences from the present harmonisation in terms of their choice of activities for a given level of targeted return. Further, the onset of new risks associated with banks’ business models and profitability, both subject to stress as a result of significant asset write-downs and the prolonged period of low interest rates, as well as the arrival of new players such as FinTechs and other operators engaging in specific activities previously dominated by banks as well as digital transformation, equally serve as powerful accelerators of change. It may be added that uncertainty remains as to the potential consequences of European harmonisation for the banking industry’s competitiveness with regard to markets such as the USA and Asia subject to different forms of regulation, but also in the event of Brexit. Also, what may be the fate of the medium-sized institutions which will appear as undercapitalised and weakened by the new constraints?

Ultimately, within the current context of reassessment of banking risk in Europe, we must expect radical change in the banking landscape.

^[1] Comprising banks and other financial institutions, large corporates and sovereign instruments (NB: the treatment of sovereign risk exposures will be the subject of a holistic review by the Basel Committee and of a separate dedicated process of consultation).

^[2] See Review of the Credit Valuation Adjustment Risk Framework (July 2015), www.bis.org/bcbs/publ/d325.htm

^[3] Supervisors will require systematic revalidation of all models modified since the changes made will probably always appear material.

^[4] Internal Capital Adequacy Assessment Process and Internal Liquidity Adequacy Assessment Process (cf. article entitled “ICAAP / ILAAP: what will change in 2016?” equally published on the Mazars Financial Services blog).