CP16/22 – Implementation of the Basel 3.1 standards: Credit risk – internal ratings based approach

Overview

4.1 This chapter sets out the Prudential Regulation Authority’s (PRA) proposals to implement the Basel 3.1 standards for the internal ratings based (IRB) approach to credit risk. It also proposes amendments to the PRA’s expectations in respect of the IRB approach and the definition of default used for both the IRB approach and the standardised approach (SA) to credit risk.

4.2 The proposals in this chapter would:

• complement HM Treasury’s (HMT) proposed revocation of certain Capital Requirements Regulation (CRR) articles and associated technical standards;
• introduce a new Credit Risk: Internal Ratings Based Approach (CRR) Part of the PRA Rulebook relating to the IRB approach, to replace CRR articles and associated technical standards that HMT plans to revoke and a technical standard that the PRA proposes to remove;
• amend the existing Credit Risk Part of the PRA Rulebook;
• remove the existing Standardised Approach and Internal Ratings Based Approach to Credit Risk (CRR) Part of the PRA Rulebook (in so far as it relates to the IRB approach); and
• withdraw Supervisory Statement (SS) 11/13 ‘Internal Ratings Based (IRB) approaches’ and introduce two SSs, namely: SS ‘Internal Ratings Based Approach’ (Appendix 13) and SS ‘Definition of Default’ (Appendix 14).

4.3 The IRB approach permits firms to use internal models as inputs for determining their regulatory risk-weighted assets (RWAs) for credit risk, subject to certain constraints. The Basel 3.1 standards introduce changes to the foundation internal ratings based (FIRB) approach and the advanced internal ratings based (AIRB) approach. The changes to the IRB approach are a key element of the Basel 3.1 package.

4.4 The Basel Committee on Banking Supervision’s (BCBS) empirical analyses, conducted following the global financial crisis, highlighted a significant degree of variability in RWAs calculated using internal models. While there was a high degree of consistency in firms’ assessments of the relative riskiness of obligors, the analysis identified material dispersion in the levels of estimated risk, as expressed in the probability of default (PD) and loss given default (LGD) that firms assigned to the same exposures.

4.5 Related to these findings, the BCBS also identified a number of other shortcomings in the IRB approach. These include excessive complexity, lack of comparability in firms’ internally modelled IRB RWAs, and a lack of robustness in modelling certain exposure classes.^{footnote [1]} Together, these findings suggest that a ‘one-size-fits-all’ approach to modelling where firms can and should model all exposure classes is unlikely to be the most prudentially sound approach.

4.6 In response to this, the Basel 3.1 standards make changes to the IRB approach, including a number of complementary measures that aim to:

• reduce the complexity of the approaches and improve comparability across firms;
• address excessive variability across firms in the calculation of RWAs for credit risk; and
• restore the credibility of the IRB framework among market participants.

4.7 The PRA agrees with the concerns identified by the BCBS and therefore proposes changes to the existing IRB framework that address them. By doing so, the PRA considers that the proposals in this chapter would promote the safety and soundness of firms with IRB permissions, and would reduce barriers to effective competition between SA and IRB firms.

4.8 Specifically, consistent with the Basel 3.1 standards, the PRA proposals include:

• removing the option to use the IRB approach for certain categories of exposures and restricting modelling within the IRB approach for certain other categories of exposures where it is judged that the model parameters cannot be estimated reliably for regulatory capital purposes. As such, firms using the IRB approach would no longer be required to model all material exposure classes;
• adopting exposure-level, model parameter floors (‘input floors’) to help ensure a minimum level of conservatism for portfolios where the IRB approaches remain available; and
• providing greater specification of parameter estimation practices to reduce variability in RWAs for portfolios where the IRB approaches remain available.

4.9 The PRA also proposes changes to improve the operation of the elements of the IRB framework that do not derive from the Basel 3.1 standards. These include a proposal to change the threshold for approving IRB model applications and IRB model changes from ‘full compliance’ with the IRB requirements to ‘material compliance’. The proposed new SS on the IRB approach would incorporate material from the existing SS11/13 as well as from the European Banking Authority’s (EBA) Guidelines related to the IRB framework that the PRA has adopted. The PRA also proposes to make a number of changes to existing expectations to improve the overall consistency and coherence of the PRA’s IRB framework. Where this chapter refers to existing expectations, this includes guidance currently located in EBA Guidelines as well as expectations currently located in SS11/13.

4.10 The proposals in this chapter are relevant to PRA-authorised banks, building societies, PRA-designated investment firms, and PRA-approved or PRA-designated financial holding companies or mixed financial holding companies (‘firms’). The proposals would not apply to UK banks and building societies that meet the Simpler-regime criteria and choose to be subject to the Transitional Capital Regime proposals^{footnote [2]} except for the case where a firm in the Transitional Capital Regime wishes to apply for an IRB model approval.^{footnote [3]}

4.11 In this chapter, the PRA has set out details of its proposals where it proposes substantive changes to requirements and expectations relative to the existing approach. The PRA also proposes to make a number of other amendments in order to enhance the clarity and coherence of the framework. This includes consolidating some existing PRA rules into new Rulebook parts. To the extent that the PRA does not propose to amend the existing approach, existing requirements and expectations would continue to apply.

Implementation timelines

4.12 This section sets out the PRA’s proposed timelines for implementing the changes to the IRB approach proposed in this chapter.

4.13 Implementation of the PRA’s proposals relating to the IRB approach would require a number of changes to firms’ IRB models. For some IRB models, the PRA recognises that the proposed changes would be in addition to other changes that need to be made in order to implement the PRA’s IRB roadmap.^{footnote [4]} This is particularly the case for wholesale LGD and exposure at default (EAD) models, and for all unsecured retail models.

4.14 The PRA has previously communicated to firms with existing IRB permissions^{footnote [5]} that it would minimise the extent to which they have to redevelop models in order to implement the IRB roadmap, only to then have to redevelop the same models a second time to implement the proposals set out in this chapter. Consequently, the PRA has delayed IRB roadmap submission deadlines for those models that are most likely to be affected by the proposals set out in this chapter to help ensure a co-ordinated and proportionate approach, and to reduce the burden on firms.

4.15 The PRA noted in its previous communication that it would give firms a sufficient period after the publication of the ‘near-final’ policy statement (PS) related to the proposals set out in this chapter^{footnote [6]} to submit delayed IRB roadmap model changes to the PRA. All other non-delayed IRB roadmap model change applications are expected to be submitted to the PRA by H2 2022 and implemented in advance of the proposals set out in this chapter.^{footnote [7]}

4.16 Consistent with the above, the PRA therefore clarifies below its proposed approaches to the:

• changes to the IRB framework that would need to be in place by the PRA’s proposed implementation date (see Chapter 1 – Overview);
• measures that would need to be in place where firms’ IRB models are not fully compliant by the PRA’s proposed implementation date; and
• timescales for IRB model submissions.

Changes required by the PRA’s proposed implementation date

4.17 The PRA proposes that firms implement non-modelling related changes and input floors to model parameters by the PRA’s proposed implementation date of 1 January 2025. The following changes would apply from this date:

• all restrictions on the scope of IRB models (eg restrictions on modelling EAD and mandatory use of the SA or FIRB approach, subject to the transitional arrangements for equity exposures);
• all changes to LGD and EAD under the FIRB approach and all changes to the maturity calculation; and
• all IRB input floors.

4.18 The PRA notes that the above non-modelling related changes would not need to be notified to the PRA under its notifications and approvals framework.

4.19 Firms’ existing IRB permissions were issued under the CRR. The PRA expects these permissions to be saved by HMT to avoid firms needing to re-apply for existing permissions. The PRA would then vary these permissions using its powers under section 144G of the Financial Services and Markets Act 2000 (FSMA) where necessary such that the saved permissions would operate in line with the proposals set out in paragraph 4.17 from 1 January 2025.

4.20 The PRA considers that these changes would improve the robustness of modelled risk weights and their timely implementation would promote the safety and soundness of firms and improve competition between SA and IRB firms.

Requirements where firms are not fully compliant by the PRA’s proposed implementation date

4.21 In contrast to the non-modelling changes, the PRA does not propose to require that the IRB models themselves be fully compliant at the proposed implementation date provided that:

• firms have an appropriate remediation plan in place that has been agreed with the PRA, including an appropriate date for the implementation of their new models, or demonstrate that the effect of the non-compliance is immaterial; and
• during this period, firms assess whether a post-model adjustment (PMA) is necessary in order to cover any shortfall in RWAs (further details about the proposals related to PMAs are provided in the section ‘Calculation of risk-weighted assets (RWAs) and expected loss (EL)’ in this chapter.

4.22 The above approach would apply in all cases of non-compliance, including in cases where the model submission deadlines communicated by the PRA (see paragraph 4.27) result in models that would not be approved by the PRA’s proposed implementation date. This approach would be in line with the approach that the PRA previously took to address non-compliance with the hybrid mortgage and IRB roadmap requirements and expectations.^{footnote [8]}

4.23 The PRA proposes to apply this approach to the residual parts of IRB models that would be split as a result of implementing the proposals set out in this chapter (eg where some exposures currently within scope of an LGD model are moved to the FIRB approach, while some exposures remain on the AIRB approach).

4.24 The PRA considers that the proposed application of PMAs would help ensure that RWAs are prudent, while maintaining proportionality as the PRA considers that it would be difficult for firms to ensure that all models would be fully compliant at the proposed implementation date. The PRA considers that remediation plans should include a clear timetable to bring the model into compliance.

Timescales for model submission

4.25 The following considerations have been taken into account by the PRA in determining its proposals for the timescales for IRB model submissions:

• the desirability that IRB models are as compliant as soon as possible after the PRA’s proposed implementation date;
• the time firms need to revise their IRB models following ‘near-final’ PS publication;
• the need to incentivise firms to begin model development work in a timely manner; and
• PRA resourcing considerations for reviewing model changes.

4.26 The PRA considers that firms would need a material amount of time between publication of the ‘near final’ PS and submission of the first set of model changes. In order to assist firms’ model development planning, the PRA proposes that it would not expect any model changes required to implement the proposals set out in this chapter to be submitted to the PRA before 1 July 2024.

4.27 The PRA proposes to communicate firm-specific timetables for submitting tranches of model change applications following publication of the PRA’s ‘near-final’ PS. This would align with the process used for the IRB roadmap where PRA supervisors communicated submission deadlines to individual firms in order to manage the flow of submissions to the PRA.

Question 16: Do you have any comments on the PRA’s proposed implementation timelines?

PRA objectives analysis

4.28 The PRA considers that the proposed timelines would advance the PRA’s primary objective of safety and soundness. The timelines would result in all IRB non-modelling changes being implemented by the PRA’s proposed implementation date, which would improve the robustness and risk capture of the IRB approach. In cases where model changes are not made by that date, firms would be required to assess whether PMAs would be appropriate to address any potential undercapitalisation of risk.

4.29 The proposals support the PRA’s secondary competition objective as the PRA considers that timely implementation of the IRB non-modelling changes would narrow the gap between SA and IRB approach risk weights. The proposed PMAs for model changes that are not in place by the PRA’s proposed implementation date would achieve a similar outcome and also help level the playing field between firms using IRB that are compliant and those that are not compliant, as well as between incumbent IRB firms and new IRB applicants.

‘Have regards’ analysis

4.30 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Relevant international standards (FSMA CRR rules):

• The proposals would result in the PRA requesting model change applications from firms to implement the IRB model changes relating to the PRA’s implementation of the Basel 3.1 standards as soon as reasonably possible. The proposal to require firms to develop a remediation plan and hold PMAs where necessary if they are not compliant would result in a level of capitalisation consistent with that required under the Basel 3.1 standards.

2. Relative standing of the UK as a place to operate (FSMA CRR rules) and competitiveness (HMT recommendation letters):

• The proposal to apply PMAs where firms are non-compliant and undercapitalised at the PRA’s proposed implementation date would help ensure a level playing field with other jurisdictions that have implemented the Basel 3.1 standards and therefore supports the relative standing of the UK as a place to operate and its global competitiveness.

3. Finance to the real economy (FSMA CRR rules):

• The proposals would support the provision of finance to the real economy by mitigating the potential risk of disruption to firms’ lending decisions if they were unable to continue using IRB models.

4. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA considers that the proposals set out in this section are a proportionate approach to the implementation of the proposals set out in this consultation paper (CP) and their interaction with other related changes to the IRB approach. The proposals recognise that firms would need time to update their IRB models to implement the proposed changes and, by aligning the timelines for implementation of such changes with those required for the IRB roadmap and by avoiding the need for firms to redevelop the same IRB models twice, the PRA aims to reduce the regulatory burden imposed on firms.

Permission to use the IRB approach

4.31 This section sets out the PRA’s proposals relating to IRB permissions.

Standards for IRB application approval for new models

4.32 Under the CRR, the PRA approves applications for IRB permissions if, and only if, it considers that all the requirements in the IRB chapter of the CRR are fully met. As a result, firms must remediate any CRR non-compliance in an IRB model application before an IRB permission can be granted by the PRA, even if the effect of the non-compliance is immaterial.

4.33 The PRA considers that this is not proportionate and places firms seeking IRB approval (‘IRB aspirants’) at a competitive disadvantage relative to firms with IRB approval. This is because firms with existing IRB approvals are not currently required to remediate immaterial non-compliance. The PRA considers this to have a negative impact on competition among UK firms.

4.34 The PRA therefore proposes to change the threshold for approval of IRB permission applications from ‘full compliance’ to ‘material compliance’ with the PRA’s requirements. For this purpose:

• non-compliance would only be considered immaterial if it results in a minimal impact on the quantitative and qualitative aspects of the firm’s IRB approach; and
• the materiality of the non-compliance would be assessed at model level and in aggregate to help ensure that immaterial non-compliance across multiple models does not become material overall.

4.35 The PRA proposes that applications that are materially non-compliant would not be approved as it considers that this could adversely impact the safety and soundness of firms.

Standards for IRB application approval for material model changes

4.36 Under the CRR, the PRA approves applications for material extensions and changes to IRB models if, and only if, it considers that the application does not introduce any new non-compliance. This applies even if the application would remediate other existing non-compliance in a firm’s IRB approach. The PRA considers that this may prevent a firm replacing a non-compliant IRB model with a more compliant model that nonetheless incorporates some new immaterial non-compliance.

4.37 The PRA considers that greater overall model compliance is more prudent, so proposes to change the standards for approval for IRB model change applications from full compliance to material compliance in line with the proposed approach to new applications.

4.38 In addition, the PRA proposes to change the standard for approval of IRB model change applications such that they may be approved where materially non-compliant if:

• the application is for changes to existing models only (applications to adopt more sophisticated modelling approaches for subsets of exposures would not be in scope);^{footnote [9]}
• approval of the application would reduce the overall level of non-compliance in the firm’s IRB approach; and
• the firm has a plan in place to remediate the remaining material non-compliance within a reasonable time period.

4.39 The PRA considers that this proposed change would be beneficial as it would enable IRB model improvements to be implemented sooner by firms. However, the PRA continues to consider it important that firms seek to promptly remediate non-compliance and would retain the right to take supervisory action in respect of remaining non-compliance, in cases where it grants approval for a materially non-compliant IRB application, for a material model change.

Other proposals

4.40 The PRA currently expects an appropriate individual in a Senior Management Function (SMF) holder to provide to the PRA, on an annual basis, written attestation as to whether their firm’s IRB models are compliant with CRR requirements and PRA SS expectations. The PRA also requests that firms submit an attestation of compliance when applying to make material model changes. See Appendix 13.

4.41 The PRA proposes to extend the annual attestation expectation so that it also covers compliance with PRA rules and covers implementation of an appropriate remediation plan where relevant. The PRA also proposes to introduce an explicit expectation that firms submitting applications and notifications relating to the IRB approach should provide the PRA with a self-assessment of whether it complies with relevant CRR requirements, PRA rules, and SS expectations. See Appendix 13.

4.42 The PRA also proposes to introduce a requirement that firms submit annually a list of their IRB models that are included within the scope of their IRB permissions. This would replace firm-specific requirements that apply to a number of IRB firms. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.43 The PRA proposes to move the criteria for determining whether IRB model change applications require ‘pre-approval’, ‘pre-notification’, or ‘post-notification’ and associated documentation requirements (which are currently set out in regulatory technical standards)^{footnote [10]} into PRA rules. As part of this, the PRA also proposes to align documentation requirements for notifications with those for ‘pre-approval’ applications to better reflect existing practice, to make a number of amendments to the criteria to improve the clarity of the framework, and to align with other proposals set out in this CP. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.44 The PRA proposes to align the approach for assessing the materiality of non-compliance which arises subsequent to an IRB permission approval being granted with the proposed approach for assessing the materiality of non-compliance for new IRB permission applications that is set out above. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.45 The PRA’s ‘overseas models approach’ (OMA) enables firms with overseas subsidiaries to use IRB models developed to non-UK standards and approved by local regulators to be used for calculating UK consolidated capital requirements when certain criteria are met. One of these criteria is that the exposures within the scope of the model must be located in an equivalent jurisdiction as determined by CRR Article 142(2). The PRA proposes to link the concept of an equivalent jurisdiction instead to CRR Article 114(7) given that it is proposed that CRR Article 142(2) would be revoked by HMT and not replaced in PRA rules. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.46 The PRA proposes to introduce a specific permission for use of the OMA and to introduce a savings provision so that firms with existing permission to use OMA models need not reapply. The PRA also proposes to transfer the majority of existing expectations regarding use of the OMA approach to PRA rules as requirements on firms. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

Question 17: Do you have any comments on the PRA’s proposals for permission to use the IRB approach?

PRA objectives analysis

4.47 The PRA considers that the proposed approach to granting permissions for IRB models would advance the PRA’s primary objective of promoting the safety and soundness of firms. The proposals would facilitate the timely remediation of existing non-compliance by enabling firms to replace non-compliant IRB models that may be a threat to safety and soundness with IRB models that, despite themselves being non-compliant, are more likely to ensure that firms’ RWAs better capture risk. In addition, the PRA considers that permitting immaterial non-compliance at the point of model approval would not be likely to have an adverse impact on safety and soundness.

4.48 The PRA considers that the proposals in this section would facilitate effective competition. The proposal that IRB model applications need to be materially compliant, instead of fully compliant, should promote a more level playing field between IRB firms and IRB aspirants, as firms with IRB approval are not currently required to remediate immaterial non-compliance. The proposals would also reduce the barriers to entry to using the IRB approach.

‘Have regards’ analysis

4.49 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Sustainable growth (FSMA regulatory principles and HMT recommendation letters):

• The PRA’s proposals to change the test for IRB model approval to ‘materially compliant’ would support sustainable growth as the PRA considers that firms which have the capability to apply the IRB approach would be able to implement it in a timelier manner. This additional flexibility should support sustainable growth to the extent IRB risk weights more accurately reflect the risk of exposures than SA risk weights.

2. Relative standing of the UK as a place to operate (FSMA CRR rules) and competitiveness (HMT recommendation letters):

• The proposal to change the test for IRB model approvals to ‘materially compliant’ should permit greater flexibility than the existing PRA approach and enhance the international competitiveness of the UK. Firms would be able to adopt the IRB approach in a timelier manner, enabling them to benefit from more risk-sensitive RWAs once they have sufficient modelling capabilities.

3. Relevant international standards (FSMA CRR Rules):

• The PRA considers that its proposed approach is broadly aligned with international standards. The PRA considers that its proposal to change the standard for approval of IRB model applications and model changes from full compliance to material compliance is not explicitly contemplated in the Basel 3.1 standards. However, given the PRA proposes that only immaterial non-compliance would be permitted, it considers the impact on alignment with the Basel 3.1 standards would be immaterial.

4. Efficient and economic use of PRA resources (FSMA regulatory principles):

• The PRA’s proposals to change the test for model approvals to ‘materially compliant’ would reduce the amount of resource that the PRA needs to deploy to assess immaterial issues.

5. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The proposals set out in this section are a proportionate response to non-compliance in IRB model applications. The proposals would enable the PRA to reach decisions on model applications more quickly and could reduce the number of model changes required by firms, while continuing to help ensure that models are robust.

IRB exposure classes and sub-classes

4.50 This section sets out the PRA’s proposals on IRB exposure classes and sub-classes.

4.51 The PRA has reviewed the existing IRB exposure classes and proposes a small number of definitional changes. The PRA also proposes to introduce new exposure sub-classes in some cases to bring greater clarity to the regulatory framework. There are instances where firms are required to apply different treatments to different categories of exposures within an exposure class. The introduction of exposure sub-classes would enable these requirements to be set out more clearly.

Central governments and central banks exposure class

4.52 The PRA proposes to restrict the scope of the ‘central governments and central banks’ exposure class to only include exposures to central governments and central banks. Under the CRR, some exposures to regional governments and local authorities, public sector entities (PSEs), multilateral development banks (MDBs), and international organisations can be assigned to this exposure class. The PRA proposes that these would be assigned to the ‘institutions’ exposure class instead.

4.53 The PRA considers that this proposal aligns the scope of this exposure class with the set of central government and central bank exposures which the PRA proposes would move to the SA as set out in the section ‘Restrictions on IRB modelling’ below.

Institutions exposure class

4.54 The PRA proposes to introduce two exposure sub-classes within the ‘institutions’ exposure class:

• ‘Quasi-sovereigns’ – this would include all exposures to regional governments and local authorities, PSEs, MDBs, as well as international organisations that are assigned a 0% risk weight under the SA.
• ‘Other institutions’ – this would include exposures to institutions and exposures treated as institutions under the SA.

Corporate exposure class

4.55 The PRA proposes to introduce three exposure sub-classes within the ‘corporate’ exposure class in order to align with the scope of the proposed approaches for modelling corporate exposures:

• ‘Specialised lending’ – this would include exposures to corporates that meet the definition of specialised lending. The PRA proposes to amend this definition to insert a condition that the borrowing entity has little or no other material assets or activities. Therefore, it would have little or no independent capacity to repay the obligation, apart from the income that it receives from the asset(s) being financed. The PRA proposes this change to align with the definition set out in the Basel 3.1 standards.
• ‘Financial corporates and large corporates’ – this would include all exposures to financial sector entities (FSEs), as defined in CRR Article 4(1)(27), that fall within the corporate exposure class (financial corporates), and all other exposures to corporates with total consolidated annual revenues greater than £440 million,^{footnote [11]} measured over a rolling average over the prior three years based on accounting consolidated data (large corporates).
• ‘Other general corporates’ – this would include all other corporate exposures.

Retail exposure class

4.56 The PRA proposes to introduce three exposure sub-classes within the ‘retail’ exposure class to align with the exposure sub-classes set out in the Basel 3.1 standards:

• ‘Qualifying revolving retail exposures’ (QRRE);
• ‘Retail exposures secured by residential immovable property’; and
• ‘Other retail’.

4.57 Currently, exposures to small and medium-sized enterprises (SMEs) can only be included in the retail exposure class if the total exposure to the obligor and related entities (excluding exposures secured by residential immovable property), not including undrawn amounts, does not exceed €1 million. The PRA proposes to amend this criterion so that exposures to SMEs would only be included in the ‘retail’ exposure class if the total exposure, including any undrawn limit, to the obligor and related entities (excluding exposures secured on residential property) does not exceed £0.88 million.^{footnote [12]} The PRA considers that its proposed inclusion of undrawn limits in the threshold would align better with the Basel 3.1 standards and reduce the likelihood that exposures are reclassified when facilities are drawn down and repaid.

Equity exposure class

4.58 The PRA proposes that exposures in the form of units or shares in collective investment units (CIUs) would be allocated to a separate sub-class within the ‘equity’ exposure class. The PRA also proposes to align the scope of the remainder of the ‘equity’ exposure class with those exposures that are treated as equity under the SA.

4.59 The PRA therefore proposes to introduce the following two exposure sub-classes within the equity exposure class which would align with the scope of the proposed approaches to risk-weighting these exposures:

• ‘Units or shares in CIUs’; and
• ‘Other equity’.

Question 18: Do you have any comments on the PRA’s proposed IRB exposure classes and sub-classes?

PRA objectives analysis

4.60 The PRA considers that the proposals to amend the IRB exposure classes and sub-classes are consistent with its primary objective of safety and soundness. The proposed change to the definition of SMEs would likely result in an increase in RWAs for some SME exposures to the extent that the existing definition fails to adequately capture the risk attached to undrawn credit facilities. The remaining proposals would not directly impact RWAs. Where different modelling approaches would be applied to different exposure classes or sub-classes, this is assessed in the relevant section of this CP.

4.61 The PRA considers that the proposals in this chapter are consistent with its secondary objective of facilitating effective competition. The proposals would result in more consistent allocation of exposures to exposure classes across firms, which should reduce unwarranted variation in RWAs across firms.

‘Have regards’ analysis

4.62 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Proportionality of costs and benefits (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The proposals in this chapter would create costs for firms initially, to the extent that they would need to adjust internal systems to reflect revised exposure class and exposure sub-class definitions. However, the PRA considers that these costs would not persist over time. The PRA considers that firms would benefit from a clearer exposition of requirements, which would justify any increase in costs.

2. Relevant international standards (FSMA CRR rules):

• The PRA considers that some of its proposed changes, such as the proposed revision to the scope of the ‘central governments and central banks’ exposure class, would result in some minor differences in how exposures are categorised relative to the Basel 3.1 standards. The PRA considers; however, that the proposed changes are desirable as they would result in exposures with more similar characteristics being grouped together.

3. Efficient and economic use of PRA resources (FSMA regulatory principles):

• Greater clarity regarding the exposures contained in each exposure class would result in more efficient use of PRA resources in assessing the appropriateness of firms’ allocation of exposures.

Restrictions on IRB modelling

4.63 This section sets out the PRA’s proposals to implement the restrictions placed on modelling under the IRB approach in the Basel 3.1 standards.

4.64 The BCBS identified concerns regarding the extent to which certain exposure classes, particularly those containing low default portfolios, can be modelled robustly for the calculation of RWAs. These exposure classes share similar characteristics, including insufficient data for the estimation of key risk inputs, a lack of information possessed by firms, insufficient comparative advantage of firms relative to regulators in assessing the risk of such exposures, and/or the lack of robust, generally accepted, and validated modelling techniques.

4.65 The PRA shares the concerns identified by the BCBS. The PRA also considers that some of the concerns identified by the BCBS apply to central government and central bank exposures. These concerns, as covered below in the ‘Roll-out, permanent partial use, and reversion’ section of this chapter, led the BCBS to conclude that firms do not need to either model all material exposure classes, or none of them (ie modelling should no longer be ‘all or nothing’).

4.66 In response to these concerns, the PRA proposes to introduce a number of restrictions on modelling risk weights, consistent with the Basel 3.1 standards, as set out in the paragraphs below.

Removal of IRB for central government and central bank exposures

4.67 Central government and central banks exposures can currently be modelled using either the FIRB or AIRB approach. While the Basel 3.1 standards do not introduce restrictions on modelling these exposures, the PRA considers that firms also face considerable challenges in modelling this exposure class.

4.68 Specifically, the PRA considers that some of the BCBS’s concerns regarding low default portfolios also apply to central government and central bank exposures, including a lack of modellability and an insufficient comparative advantage of firms relative to regulators in assessing the risk. The PRA also considers that UK firms have historically had difficulty building robust central government and central bank IRB models.

4.69 The PRA therefore proposes to prohibit modelling of central government and central bank exposures under IRB, and to require all central government and central bank exposures to be risk-weighted under the SA (see Chapter 3 – Credit risk – Standardised approach for the proposed SA approach for sovereign exposures).^{footnote [13]} The PRA considers that removing modelling of central government and central bank exposures would increase the consistency and comparability of risk weights for these exposures and facilitate effective competition.

4.70 The PRA recognises that firms could choose to continue to use central government and central bank IRB models for risk management purposes even if they are no longer permitted to be used for calculating regulatory RWAs.

Removal of the advanced IRB approach for exposures to institutions, financial corporates, and large corporates

4.71 Currently, RWAs for exposures to institutions, financial corporates, and large corporates can be modelled using either the FIRB or AIRB approaches. The Basel 3.1 standards remove the use of the AIRB approach for risk-weighting these exposures and, as a result, the FIRB approach is the only modelling approach that remains available. The BCBS concluded that this was justified based on the identified challenges in modelling LGD and EAD for these exposure classes and related concerns with the robustness of these models, as set out above.

4.72 The PRA shares the BCBS’s concerns and proposes to remove the AIRB approach for the following exposures:

• exposures to institutions and financial corporates, where financial corporates would comprise other financial sector entities (FSEs), as defined in CRR Article 4(1)(27); and
• exposures to large corporates, where the exposures are not classified as specialised lending. Modelling of exposures to corporates that are not large corporates, including exposures to SMEs, would still be permitted.^{footnote [14]}

4.73 The existing definition of FSEs includes ancillary services undertakings and holding companies; therefore, the PRA’s proposal would remove the AIRB approach from a wider set of exposures than is envisaged by the Basel 3.1 standards. However, firms are currently required to use the definition of FSEs to determine whether a 1.25 multiplication factor applies to the co-efficient of correlation parameter in the IRB approach risk weight formula, and the PRA considers that using a common definition to restrict AIRB modelling would reduce the implementation burden for firms. In addition, the PRA considers the additional exposures brought in scope by the wider definition would generally sit within low default portfolios, therefore supporting the PRA’s concerns regarding modellability.

4.74 In the section titled ‘IRB exposure classes and sub-classes’ above, the PRA proposes to expand the institutions exposure class to include regional governments and local authorities, PSEs, MDBs, and international organisation exposures that are assigned a 0% risk weight under the SA that currently fall within the ‘central governments and central banks’ exposure class. As the PRA proposes to remove AIRB modelling for exposures to institutions, the combined effect of these proposals would be that the AIRB approach would be removed for all exposures to regional and local governments, PSEs, MDBs, and international organisations that are assigned a 0% risk weight under SA. The PRA considers that this is appropriate due to the difficulty of modelling LGD and EAD for these exposures.

Removal of the IRB approach for equity exposures

4.75 There are currently three IRB methodologies for calculating RWAs for equity exposures: the ‘simple risk weight’ approach, the ‘PD / LGD’ approach, and the ‘internal models approach’. Under the simple risk weight approach, firms apply prescribed risk weights to different categories of equity exposures while the other two approaches involve some modelling.

4.76 The Basel 3.1 standards prohibit the modelling of equity exposures and withdraw all three IRB methodologies for equity exposures on the basis that such exposures cannot be modelled in a robust and prudent manner. The BCBS noted that, for many equity exposures, it is unlikely that firms would have specific knowledge concerning the issuers compared with that found in publicly available data. Additionally, the IRB approach for equities only applies to exposures in the banking book, and such exposures tend to only form a very small component of firms’ balance sheets. Furthermore, from a competition perspective, in cases where firms are using the IRB simple risk weight approach to risk weight equity exposures, there is little justification for different SA and IRB prescribed risk weights for the same exposures. As a result, the SA is the only credit risk approach remaining in the Basel 3.1 standards for risk-weighting equity exposures.

4.77 The PRA shares these concerns and proposes to remove the IRB approach for equity exposures and require RWAs for all equity exposures to be calculated using the SA.^{footnote [15]} The PRA considers that this should improve the robustness, simplicity, consistency, and comparability of the RWAs for equity exposures.

Transitional arrangements

4.78 The Basel 3.1 standards allow a five-year linear phase-in arrangement for firms using the IRB approach to move equity exposures to the SA, in order to give firms additional time to adjust to the revised approach. The PRA proposes to introduce transitional arrangements that are in line with the Basel 3.1 standards.

4.79 The PRA proposes two transitional approaches for equities: the SA transitional approach outlined in Chapter 3 and the IRB transitional approach outlined in this section. Together, the two proposed transitional arrangements are intended to facilitate a smooth change to the new SA equity risk weights by firms to avoid unnecessary volatility in RWAs. The PRA proposes that firms would apply the IRB transitional approach for all equity exposures subject to the IRB approach on 31 December 2024 and apply the SA transitional approach for all other exposures.

4.80 The PRA proposes to introduce a five-year IRB transitional period from the PRA’s proposed implementation date of 1 January 2025. Under the proposed IRB transitional approach, firms would, for each individual equity exposure, apply the higher of:

• the risk weight for the exposure calculated under the CRR (based on the firm’s IRB permission as of 31 December 2024); and
• the risk weight for the exposure calculated under the SA transitional approach (as outlined in Chapter 3).

4.81 In order to avoid excessive complexity in the capital framework, the PRA proposes that firms would calculate exposure values using the SA methodology (ie net of provisions) for the duration of the transition. The PRA does not propose that firms would be required to calculate any ‘expected loss’ (EL) amounts in respect of exposures subject to the IRB transitional approach.

4.82 The PRA recognises that the proposed transitional arrangements could nevertheless still result in some volatility in RWAs which it considers would be undesirable. To mitigate this risk, the PRA proposes to allow firms the option to ‘opt-out’ of the IRB equity transitional arrangements entirely and to move to the final SA risk weights for all equity exposures at any point before or during the transitional period by notifying the PRA. The PRA proposes that in these circumstances firms would not be able to revert back to the transitional arrangements.

4.83 Where firms apply the look-through or mandate-based approach for risk-weighting CIUs, they need to determine the risk weights of the underlying exposures. As an exception, where a firm is using the IRB approach, but equity exposures underlying a CIU are subject to the SA via a permanent partial use permission, the simple risk weight approach is used to determine the risk weights of the underlying equity exposures instead.

4.84 The PRA proposes that firms using the IRB transitional approach for equity would, when determining the risk weight of an equity exposure that would have been subject to the SA prior to the start of the equity transitional period, calculate the applicable risk weight as if the exposure had been subject to the simple risk weight approach prior to the start of the equity transitional period. This is to reduce undesirable volatility in risk weights arising from the introduction of the proposed transitional arrangements.

Question 19: Do you have any comments on the PRA’s proposed restrictions on the use of the IRB approach?

PRA objectives analysis

4.85 The PRA considers that the proposals relating to the removal of the AIRB approach for financial institutions and large corporate exposures would advance its primary objective. The PRA considers that the lack of default data for these exposures indicates that firms are typically unable to robustly model LGD and EAD. As a result, removing modelling of these parameters would promote safety and soundness by improving the robustness of, and reducing unwarranted variability in, RWAs.

4.86 Similarly, the PRA considers that its proposal to remove modelling of central government, central bank, and equity exposures would be consistent with the principle that firms should be permitted to model exposures only where they can do so in a robust manner. However, as noted in Chapter 3, the PRA has concerns that proposed SA risk weights for exposures to central governments and central banks can potentially result in an underestimation of RWAs. Such undercapitalisation can stem from the 0% risk weight applied to highly rated central government and central bank exposures, and the equivalence-based risk weight overrides. The PRA therefore intends to consult in the future, as part of the wider Pillar 2 review discussed in Chapter 10 – Interactions with the PRA’s Pillar 2 framework, on a potential Pillar 2 methodology to help ensure the adequate capitalisation of these exposures.

4.87 The PRA considers that the proposals set out in this section would advance its secondary competition objective. The proposed restrictions on modelling would apply to all firms using the IRB approach and, as a result, RWAs should be more comparable across firms for the same exposure. In the case of exposures to institutions, financial corporates, and large corporates, the PRA considers that the proposals would improve competition between those firms currently using the FIRB approach for these exposures and those currently using the AIRB approach.

4.88 The PRA considers that the proposal to remove modelling of central government, central bank, and equity exposures would also advance its secondary competition objective as all firms would risk weight these exposures under the SA, resulting in less variation in approaches across firms.

‘Have regards’ analysis

4.89 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Sustainable growth (FSMA regulatory principles):

• The PRA assesses that the proposals in this section would have little adverse impact on sustainable growth. There is some potential that the proposed removal of the AIRB approach for exposures to institutions, financial corporates, and large corporates could increase RWAs and reduce incentives for firms to lend, although the PRA considers this would depend on the conservatism of firms’ existing modelled LGD and EAD estimates. The PRA considers that where RWAs do increase, this would benefit sustainable growth to the extent that RWAs currently fail to adequately capture risk. This is because the PRA considers that adequate capitalisation of the risks of these low default portfolios is essential for sustainable lending and therefore sustainable growth.

2. Relevant international standards (FSMA CRR rules):

• The PRA considers that the proposed removal of the AIRB approach for institutions, financial corporates, and large corporates, and the proposed removal of all IRB approaches for equities, would be aligned with the Basel 3.1 standards. The PRA considers that the proposal to remove all IRB approaches for central government and central bank exposures would also be aligned with international standards as the Basel 3.1 standards envisage that national supervisors may choose not to implement all available modelling approaches.

3. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA considers that the proposals in this section would not adversely impact the competitiveness of the UK. The proposals relating to institutions, financial corporates, large corporates, and equity exposures are likely to be in line with the approach taken by other major jurisdictions. The PRA considers that the proposal to remove modelling entirely for exposures to central governments and central banks would likely result in lower RWAs for some exposures and higher RWAs for other exposures relative to those jurisdictions that retain the IRB approach. However, as the PRA considers that it is unlikely that overall RWAs would increase as a result of the proposals in this section, it does not consider there would be any adverse impact on competitiveness.

Roll-out, permanent partial use, and reversion

4.90 This section sets out the PRA’s proposals on roll-out, permanent partial use, and reversion to less sophisticated approaches.

4.91 Firms’ existing permissions for roll-out and permanent partial use were issued under the CRR. The PRA expects these permissions to be saved by HMT. The PRA would then vary these permissions using its powers under section 144G of FSMA where necessary, such that the scope of saved permissions would be restricted so that they are consistent with the proposals set out in this section. Firms would then treat any non-compliance in line with the approach set out in paragraph 4.21 and would be able to apply to extend the scope of existing permissions in line with the proposals set out in this section.

Roll-out and permanent partial use of SA by roll-out class

4.92 Firms are currently required to roll-out the IRB approach across all exposures once they have obtained an IRB permission. However, there are exceptions under certain circumstances. For example, an exception is made for ‘exposures in non-significant business units as well as exposure classes or types of exposures that are immaterial in terms of size and perceived risk profile’.^{footnote [16]}

4.93 This ‘full use’ requirement can act as a barrier to adoption of the IRB approach. The Basel 3.1 standards remove the ‘full use’ requirement and instead allow firms to adopt IRB for some exposure classes while allowing other exposure classes to remain permanently on the SA.

4.94 The PRA proposes to broadly align with the Basel 3.1 standards. It also proposes to withdraw its existing expectation regarding the maximum proportion of a firm’s RWAs that may remain on the SA when an IRB permission is in place, which would enable firms to depart from the full use requirement. The PRA considers; however, that it is important to avoid creating ‘cherry-picking’ opportunities where firms are able to optimise RWAs through their choice of which exposures should remain on the SA. Therefore, the PRA proposes to introduce a number of safeguards to mitigate this risk.

4.95 The PRA proposes to define a set of eight ‘roll-out classes’ for which firms are able to apply for permission to permanently use the SA. Where firms obtain this permission, they would be allowed to roll-out the IRB approach to some roll-out classes but apply the SA to others. The PRA proposes to introduce the following set of roll-out classes, which the PRA considers is broadly aligned with the Basel 3.1 standards:

• ‘institutions’;
• ‘specialised lending’;
• ‘corporate purchased receivables’;
• ‘general and financial corporates’;
• ‘qualifying revolving retail exposures’;
• ‘retail exposures secured on residential immovable property’;
• ‘retail purchased receivables’; and
• ‘other retail’.

4.96 Further detail on the proposed roll-out classes, and the interaction with the proposed exposure classes and sub-classes in the ‘IRB exposure classes and sub-classes’ section, is set out in the table below.

Table 1: Proposed roll-out classes

4.97 In order to mitigate the risk of ‘cherry-picking’, the PRA proposes to introduce a requirement that a firm should not permanently use the SA for a roll-out class if this would result in significantly lower credit risk RWAs than under the IRB approach unless:

• the firm cannot reasonably model exposures in the roll-out class; or
• the roll-out class is immaterial.

4.98 The PRA proposes that, for this purpose, ‘significantly lower credit risk RWAs’ would mean that SA RWAs are reasonably estimated to be less than 95% of group credit risk IRB RWAs for that roll-out class, prior to application of the proposed output floor.

4.99 The PRA also proposes to specify the following criteria to determine circumstances in which the PRA considers firms could not reasonably model exposures in a roll-out class, namely that either:

• the firm would not have sufficient data to model exposures in the roll-out class, nor could the firm be reasonably expected to obtain sufficient data in a timely manner, and that the deficiency in data did not arise due to historic non-compliance with the data collection and storage requirements in the CRR or in BIPRU;^{footnote [17]}
• the firm could not reasonably develop a compliant modelling approach due to the nature and complexity of the exposures in the roll-out class; or
• the use of the IRB approach for the roll-out class would not result in significant improvements in risk differentiation or risk quantification than if the SA were applied to that roll-out class.

4.100 The PRA proposes to specify that a roll-out class would be considered immaterial if total SA RWAs for the roll-out class do not exceed 5% of total group credit risk RWAs, prior to application of the proposed output floor.

4.101 The PRA has observed historically that SA RWAs for the QRRE and specialised lending roll-out classes could be materially lower than those calculated using the IRB approach. The PRA also considers that firms would typically be able to develop a compliant IRB approach for these roll-out classes (including the slotting approach for specialised lending exposures). As such, the PRA proposes to introduce an expectation, to supplement the proposed requirements, that firms with an IRB permission for any roll-out class should not permanently apply the SA for either the QRRE roll-out class or the specialised lending roll-out class unless the roll-out class in question is immaterial.

Permanent partial use of the SA within roll-out classes

4.102 As set out above, the Basel 3.1 standards no longer contain a full use requirement for adopting IRB across roll-out classes. However, the Basel 3.1 standards have introduced a requirement that once IRB has been rolled out to a roll-out class, IRB would be adopted for all exposures within that roll-out class, subject to a materiality exemption.

4.103 The PRA considers that to require firms to apply the IRB approach to immaterial exposures and to require firms to model RWAs for exposures where they lack the capability would be disproportionate and undesirable. The PRA therefore proposes that firms should be able to apply for permission to permanently apply the SA to subsets of exposures within a roll-out class. The PRA proposes however that this would only be permitted in the following cases:

a. the firm is unable to model the type of exposure;

b. the exposures are immaterial;

c. the exposures are to institutions relating to Bank of England minimum reserve requirements and certain other conditions (maintaining the CRR Article 150(1)(i) exemption); or

d. intragroup exposures meeting certain conditions (maintaining the CRR Article 150(1)(e) exemption).

4.104 For (a), the PRA proposes to apply the same criteria to determine that firms are unable to model a type of exposure within a roll-out class as those proposed for determining whether firms can model entire roll-out classes.

4.105 For (b), the PRA proposes to limit the exposures within a roll-out class that could be treated as immaterial to 5% of the total group credit risk RWAs for that roll-out class prior to application of the proposed output floor. The PRA considers this requirement would limit opportunities for ‘cherry-picking’ risk-weighting approaches.

4.106 The PRA has also considered whether to retain other existing CRR permanent partial use exemptions. The PRA considers that the majority of the existing exemptions are either no longer relevant, due to other changes in the framework such as restrictions on the scope of modelling; or no longer necessary, due to the proposed introduction of a more general exemption for immateriality as set out in (b) above. The PRA therefore proposes to remove these exemptions. The PRA proposes; however, to retain the exemptions currently in the CRR relating to intragroup exposures and exposures in the form of minimum reserves required by the Bank of England.

4.107 The PRA considers that to introduce an overall upper limit on the permanent partial use of the SA within a roll-out class is appropriate to limit ‘cherry-picking’ opportunities and to help ensure that a critical mass of exposures are modelled within roll-out classes for which firms adopt IRB. The PRA proposes to set this limit at 50% of total group credit risk RWAs for the roll-out class. Given the specific nature of exemptions (c) and (d), the PRA does not propose to restrict use of these exemptions. RWAs for these exposures would therefore be excluded from both the numerator and denominator in the calculation of the 50% RWA threshold.

Permanent partial use of the FIRB approach

4.108 There are currently no restrictions on the permanent partial use of the FIRB approach, except where firms wish to revert to a less sophisticated approach. As a result, there is a potential risk that firms may engage in ‘cherry-picking’ and select portfolios to remain on the FIRB approach in order to lower their RWAs.

4.109 The PRA considers it desirable to limit ‘cherry-picking’ opportunities for those exposures where both the FIRB and the AIRB approaches are available. The PRA therefore proposes that firms applying the AIRB approach for any of these exposures would need to apply to the PRA for permission to apply the FIRB approach to such exposures, and that use of the FIRB approach in these circumstances would not be permitted where the change is proposed to lower RWAs relative to applying the AIRB approach. The PRA proposes to create a ‘non-retail AIRB modelling roll-out category’ of exposures in its rules, which would consist of all exposure sub-classes that are eligible for both the FIRB and the AIRB approaches and that would be subject to these proposed restrictions.

Reversion to less sophisticated credit risk approaches

4.110 Firms are currently only able to revert to less sophisticated approaches (either from the IRB approach to the SA or from the AIRB approach to the FIRB approach) if they have permission from the PRA and the proposed reversion:

• is not proposed in order to reduce the RWAs of the firm;
• is necessary on the basis of the nature and complexity of the firm’s total exposures of this type; and
• would not have a material adverse impact on the solvency of the firm or its ability to manage risk effectively.

4.111 The Basel 3.1 standards contain a general provision that firms already using the IRB approach for an exposure class would continue to do so after implementing the Basel 3.1 standards, but it also permits reversion to less sophisticated approaches in ‘extraordinary circumstances’. The PRA considers that to implement a specific ‘extraordinary circumstances’ condition is unnecessary, and therefore proposes to retain the existing conditions for reversion and apply these to both entire roll-out classes and subsets of exposures within roll-out classes.

4.112 However, the PRA considers that one-off costs that relate to implementing the proposed requirements set out in this CP could occur and that, as a result, mandating firms to remain on either the FIRB approach or the AIRB approach could be unduly burdensome. Therefore, the PRA proposes to supplement the reversion conditions with expectations that it would consider one-off costs arising from implementing the requirements proposed in this CP as a relevant factor when it considers whether the conditions are met.

4.113 The PRA also proposes to clarify that firms using the AIRB or the FIRB approaches for specialised lending exposures could only adopt the slotting approach where they have the permission of the PRA and their proposed implementation of the slotting approach is materially compliant with the PRA’s requirements.

Question 20: Do you have any comments on the PRA’s proposed approach to roll-out, permanent partial use, and reversion?

PRA objectives analysis

4.114 The PRA considers that the proposals on roll-out and reversion advance its primary objective, as they are designed to reduce the prudential risk associated with firms ‘cherry-picking’ risk-weighting approaches in order to reduce RWAs without a corresponding reduction in risk. The PRA considers that the proposals would also advance safety and soundness by not requiring firms to model exposures for which they are unable to develop robust models.

4.115 The PRA also considers its proposals for partial use of the SA by roll-out class would also advance the PRA’s secondary competition objective. The proposals could reduce the barriers to entry for smaller firms adopting the IRB approach by no longer requiring firms to roll-out the IRB approach to all exposures. The PRA’s proposed conditions for permanent partial use of the SA by roll-out class may result in some barriers to entry remaining as some firms may still need to model certain roll-out classes; for example, QRRE. However, the PRA considers the benefits to its primary objective from applying these conditions justify any resultant negative impacts on its secondary competition objective.

‘Have regards’ analysis

4.116 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Sustainable growth (FSMA regulatory principles) and growth (HMT recommendation letters):

• The PRA considers that the proposals in this section would not adversely impact sustainable growth. The PRA considers that providing firms with greater flexibility to determine which portfolios could have RWAs calculated using internal models would enable firms to adopt the IRB approach in a more timely manner for those portfolios for which modelling approaches remain. The PRA considers the proposals would enable the firms to calculate more risk-sensitive RWAs and help ensure that RWAs do not fall to an imprudent level as a result of cherry-picking between risk-weighting approaches. The PRA considers that this would support sustainable growth.

2. Relevant international standards (FSMA CRR rules):

• The PRA considers that the proposals on roll-out and reversion are broadly in line with the Basel 3.1 standards, as the PRA’s proposed approach is stricter in some respects and provides more flexibility in others. The PRA proposes to introduce more stringent anti-cherry-picking measures in order to address the PRA’s concerns regarding the relative risk weights of certain exposures under the SA compared to under the IRB approach, such as for QRRE and specialised lending exposures, which could result in potentially inadequate RWAs at the aggregate level. The PRA also proposes to implement greater flexibility for firms to apply the SA to certain exposures within a roll-out class that is subject to IRB. Therefore, on balance, the PRA considers that its proposals are aligned with international standards.

Calculation of risk-weighted assets (RWAs) and expected loss (EL)

4.117 This section sets out the PRA’s proposals for the calculation of RWAs and EL.

The 1.06 scaling factor

4.118 Under the existing Basel standards and the CRR, the formula used to calculate IRB RWAs includes a scaling factor of 1.06 for non-defaulted exposures. This was included by the BCBS when the IRB approach was first introduced and was intended to maintain the aggregate level of minimum capital requirements compared to the previous Basel standards.

4.119 The Basel 3.1 standards remove the 1.06 scaling factor on the basis that it is no longer necessary given the enhancements to the IRB framework and the introduction of an aggregate output floor. The PRA shares the BCBS’s view and considers that, given the other changes the PRA proposes to implement in this CP, the 1.06 scaling factor is no longer required and proposes to remove the 1.06 scaling factor from the IRB formula.

The 1.25 asset value co-efficient of correlation multiplier

4.120 The CRR requires firms to apply a 1.25 multiplier (the ‘multiplier’) to the co-efficient of correlation in the IRB RWA function for exposures to ‘large financial sector entities’ and ‘unregulated financial sector entities’. The application of the multiplier results in higher RWAs for these exposures and is intended to capture the systemic risk associated with exposures to these entities.

4.121 The following definitions are currently used in the CRR:

• Large financial sector entities are defined as any financial sector entity (FSE) with total individual or consolidated assets greater than or equal to €70 billion (based on financial statements) where the entity (or one of its subsidiaries) is subject to UK prudential regulation or equivalent third-country prudential regulation.
• Unregulated financial sector entities are defined as entities that are unregulated but which perform one or more activities specified under The Financial Services and Markets Act 2000 (Regulated Activities) Order 2001 as their main business.

4.122 The PRA considers that the scope and application of the CRR requirements lack clarity in places.

4.123 The first shortcoming relates to the treatment of regulated third-country FSEs. While the CRR is clear on the treatment of regulated third-country FSEs that are equivalently regulated, the CRR does not explicitly set out the treatment of regulated third-country FSEs that are not equivalently regulated. In particular, the PRA considers that the exemption from the multiplier for large FSEs not subject to equivalent regulation is inappropriate as exposures to large FSEs not subject to equivalent regulation would have lower RWAs than would be the case if they were equivalently regulated. The PRA considers that exposures to large FSEs pose a systemic risk regardless of how they are regulated.

4.124 The PRA consequently proposes to extend the scope of application of the multiplier to all large FSEs, regardless of the nature of the regulation they are subject to. The PRA considers that this amendment would provide clarity to firms on the application of the multiplier, correct the counter-intuitive treatment in which non-equivalently regulated FSEs attract lower risk weights, and advance the PRA’s primary objective of safety and soundness.

4.125 The second shortcoming relates to the calculation of the €70 billion asset threshold in the large FSE definition. The CRR states that the €70 billion threshold is calculated on an individual or a consolidated basis. However, an EBA ‘Q&A’ issued while the UK was a member of the EU stated that only the assets of the entity and its subsidiaries should be considered, rather than the assets of the wider group.

4.126 The PRA considers that the CRR approach is potentially imprudent and could incentivise regulatory arbitrage. The PRA considers that systemic risks could arise from large groups as well as large entities. While measuring asset size based only on an entity, its parent company, and its subsidiaries, not taking the wider group into account, meets the explicit requirements of the Basel 3.1 standards, the PRA considers that this approach does not sufficiently reduce risks arising from the interconnectedness of large financial institutions, which is the purpose of the multiplier. The PRA therefore proposes to amend the CRR definition of large FSEs to explicitly include the total assets of the entire group. The PRA would continue to require firms to calculate total assets based on accounting assets.

4.127 The PRA considers that an additional benefit of this proposal is that it would reduce the potential for regulatory arbitrage. Such arbitrage could occur if firms choose to lend to specific subsidiaries within a group, in order to avoid the application of the multiplier and apply preferential risk weights to the same or very similar risk. In addition, groups could themselves transfer businesses into smaller subsidiaries to facilitate arbitrage or increase the number of parent entities in the group to avoid classification as a large FSE.

4.128 The third shortcoming relates to the definition of unregulated FSEs. The PRA considers that limiting the scope of unregulated FSEs to entities that are not subject to any regulation would result in the exclusion of entities only subject to limited prudential regulation from the requirements (eg financial intermediaries), which the PRA considers inconsistent with the Basel 3.1 standards.

4.129 The PRA therefore proposes to amend the definition of an unregulated FSE to include all FSEs that are not prudentially regulated as either a credit institution, investment firm, or an insurer. The PRA considers that this approach would result in the multiplier being applied to a set of FSEs that is consistent with those envisaged by the Basel 3.1 standards. The PRA also considers that the proposed approach would be simpler for firms to operationalise.

4.130 The PRA notes that the FSE threshold in the CRR is €70 billion and this corresponds to $100 billion in the Basel 3.1 standards. The PRA proposes in Chapter 13 – Currency redenomination, to redenominate the $100 billion threshold in the Basel 3.1 standards to £79 billion.

Question 21: Do you have any comments on the PRA’s proposals relating to the 1.06 scaling factor and to the 1.25 asset value co-efficient of correlation multiplier?

SME support factor

4.131 As noted in Chapter 3, the CRR SME support factor was originally introduced to limit disruption to the flow of credit to small businesses during the phase-in of stricter requirements following the global financial crisis. For exposures to businesses with a turnover below €50 million and total borrowings (excluding residential property) not exceeding €1.5 million, a support factor equal to 0.7619 is applied to Pillar 1 RWAs. This calibration was designed to broadly offset the additional capital required for the capital conservation buffer.

4.132 The SME support factor applies to all credit risk exposures included in the retail, corporate or secured by mortgages on immovable property exposure classes that satisfy the criteria in CRR Article 501 (exposures in default are excluded). It applies to the SA and IRB approaches. It was identified as a material deviation from the Basel standards in the 2014 BCBS Regulatory Consistency Assessment Programme (RCAP) report on the EU implementation of Basel standards.

4.133 CRR II expanded the scope of the SME support factor. The ceiling on total borrowings (excluding residential property) has been raised to €2.5 million, and a new support factor of 0.85 is now applied to any lending exceeding this threshold. These changes were onshored when the UK left the EU.

4.134 As set out in Chapter 3, the PRA has considered existing analysis on the effectiveness of the SME support factor and, while the PRA recognises that conceptually the SME support factor could be a stimulus for supporting sustainable lending to SMEs, the available evidence is inconclusive at a system level. EBA analysis concluded that the SME support factor could be imprudent, is not risk-based, and had not materially supported lending to SMEs.

4.135 The PRA considers that the risk weight treatment for SMEs under the IRB approach should already be appropriate without the support factor for the following reasons:

• IRB modelled risk weights should already be risk-sensitive and reflect a firm’s historic experience of lending to SME counterparties;
• the IRB corporate risk weight formula already includes a firm-size adjustment (an adjustment to the co-efficient of correlation) for exposures to SME obligors, which reduces the applicable risk weight for SME exposures relative to other corporate exposures (see Chart 1 below); and
• exposures to SMEs that qualify for the IRB retail exposure class are risk-weighted using the retail IRB risk weight formula rather than the corporate IRB risk weight formula which produces lower risk weights for retail SME exposures than for corporate exposures (see Chart 1 below).

Chart 1: Comparison of risk weights for retail SME, corporate SME, and non-SME corporate exposures^{footnote [18]}

4.136 Therefore, the PRA considers that modelled IRB risk weights for SME exposures should reflect the risk and that a further discount on the risk weights is not justified based on the evidence the PRA has available. It therefore proposes to remove the SME support factor.

4.137 However, the PRA considers it appropriate, in proposing to remove the support factor, to consider whether the IRB approach is appropriately calibrated for SME exposures, both the firm-size adjustment in the corporate IRB formula for corporate SME exposures and the retail IRB formula for retail SME exposures. The PRA would welcome feedback, including quantitative or qualitative evidence across a range of economic conditions, on whether the IRB approach would appropriately reflect the risk of SME exposures if the support factor is removed.

Question 22: Do you have any comments on the PRA’s proposal to remove the SME support factor under the IRB approach? Do you have any evidence – quantitative or qualitative – regarding the appropriateness of the IRB approach for SME exposures in the absence of the support factor?

Infrastructure support factor

4.138 As noted in the Chapter 3, the CRR infrastructure support factor allows firms to apply a 0.75 multiplier to RWAs for certain exposures that are allocated to the corporate exposure class or specialised lending exposure class. Defaulted exposures are excluded and the criteria in CRR Article 501a must be satisfied in order to apply it. The infrastructure support factor applies to exposures under the SA and the IRB approach.

4.139 The PRA proposes to not apply the infrastructure support factor under the IRB approach. In cases where a firm models risk weights for exposures that would currently qualify for application of the support factor, the modelled risk weights should be risk-sensitive and reflect the firm’s historic experience of lending to that type of exposure. For exposures that would currently qualify for the support factor that are assigned to the specialised lending exposure class and are risk-weighted using the slotting approach, the criteria for assigning exposures to slots would allow firms to take into account the specific risk characteristics of each exposure and result in risk-sensitive risk weights being assigned to the exposures. Therefore, the PRA considers that IRB risk weights for exposures qualifying for the infrastructure support factor should reflect the risk, and a further discount on the risk weights is not justified on the basis of the evidence the PRA has available. It therefore proposes to remove the infrastructure support factor.

4.140 However, similar to the above proposal for the SME support factor, the PRA considers it appropriate as part of its proposal to remove the infrastructure support factor to consider whether the IRB modelling approach and the slotting approach for infrastructure exposures are appropriately calibrated. It therefore particularly invites firms to present quantitative or qualitative evidence on this topic during the consultation.

Question 23: Do you have any comments on the PRA’s proposal to remove the infrastructure support factor under the IRB approach? Do you have any evidence – quantitative or qualitative – relating to the appropriateness of the IRB approach for infrastructure exposures in the absence of the support factor?

Treatment of expected loss amounts

4.141 Under the CRR, firms using the IRB approach are currently required to compare the total amount of eligible provisions (including various adjustments) (P) with the total EL amount. Where the EL amount exceeds P, firms must deduct the difference from CET1 capital. Conversely, where P is greater than the EL amount, firms may recognise the difference in Tier 2 capital (subject to a limit of 0.6% of IRB RWAs). Additionally, if specific provisions are greater than the EL amount for defaulted exposures, the difference cannot be used to cover EL amounts for non-defaulted exposures.

4.142 The PRA considers that the CRR currently lacks clarity as to whether excesses of specific provisions over EL amounts for defaulted exposures can be added to Tier 2 capital if they are not used to cover EL amounts for non-defaulted exposures. The PRA considers that adding these amounts to Tier 2 capital would be consistent with the objectives of the regulatory framework.

4.143 The PRA therefore proposes to introduce new formula for comparing P and EL amounts in PRA rules. The effect of the formulae would be that:

a. if (a) specific provisions for defaulted exposures are greater than the EL amount for defaulted exposures, and (b) the EL amount for non-defaulted exposures is greater than all other provisions, then:

• the difference in specific provisions over the EL amount for defaulted exposures may be included in Tier 2 capital (subject to the existing limit); and
• the difference in the EL amount for non-defaulted exposures over other provisions would be deducted from CET1 capital;

b. in all other cases:

• if P is greater than the EL amount, the difference may be added to Tier 2 capital (subject to the existing limit); and
• if the EL amount is greater than P, the difference would be deducted from CET1 capital.

4.144 The PRA considers that the proposed formulae would continue to help ensure that specific provisions for defaulted exposures cannot be used to cover EL amounts on other exposures. While this is not required by the Basel 3.1 standards, the PRA considers there is sufficient prudential justification for doing so.

Unrecognised exposure adjustment

4.145 The PRA currently would expect that firms using the AIRB approach should, where material, calculate RWAs on a portfolio basis (rather than at an exposure level) to reflect products or relationships that are not intended to result in a credit exposure, but where there is an exposure at default (EAD) nonetheless (‘adjustment for non-credit relationships and facilities’). An example of these products are current accounts without an overdraft facility that are nonetheless permitted to be overdrawn.

4.146 The PRA considers that some of the proposals set out in this CP would mean that this expectation is not sufficiently broad. In particular:

• The PRA proposes to define commitment under both the SA and IRB approach as any off-balance sheet contractual arrangement that has been offered by the firm and accepted by the obligor, including to extend credit, purchase assets, or issue credit substitutions (but which is not an issued off-balance sheet item). This could result in certain facilities that have an EAD falling outside the scope of the risk weight framework. These exposures may not be captured by the adjustment for non-credit relationships and facilities as currently defined; and
• The PRA proposes to significantly reduce the scope of exposures for which EAD could be modelled. The existing adjustment for non-credit relationships and facilities only applies where firms are applying the AIRB approach and modelling EAD. The result could be that this adjustment would potentially capture significantly fewer exposures going forward based on its existing scope.

4.147 In addition, there is currently ambiguity regarding how firms should reflect the existing adjustment for non-credit relationships and facilities, resulting in a lack of clarity and coherence in the regulatory framework.

4.148 In order to address the above considerations, the PRA proposes to remove the existing expectation and introduce a new requirement that firms using any of the IRB approaches would calculate an ‘unrecognised exposure adjustment’. The proposal would not require firms to calculate an unrecognised exposure adjustment where the amount of such exposures is immaterial.

4.149 The PRA’s rationale for making this proposal is to capture amounts outstanding at default arising from facilities or relationships that are not captured in exposure value measures. Outstanding amounts would not be otherwise captured because either (a) they were not intended to result in credit exposures, or (b) they are not classified as off-balance sheet items.

4.150 The proposed adjustment would be implemented as an RWA adjustment at the portfolio level applying to firms using any of the IRB approaches, regardless of whether EAD is modelled, and would aim to capture all credit exposure not otherwise captured by the IRB framework.

4.151 The PRA recognises that this proposal is additional to the Basel 3.1 standards. However, the PRA considers that to remove its existing expectations without a proposal for a more appropriate approach would result in a less prudent framework than is currently in place. The PRA considers that it is prudentially important that all exposures that would arise in the event of default are captured in IRB RWAs to help ensure that firms’ RWAs reflect the risks associated with all types of potential exposures.

Adjustment to RWAs for UK residential mortgage exposures in the retail exposure class

4.152 The PRA currently has an expectation that firms apply a 10% exposure-weighted average portfolio risk-weight floor to UK retail residential mortgage exposures in order to reflect risks that may not be fully captured by firms’ IRB models.

4.153 The PRA considers that attributing this floor the same status as other proposed floors would improve the coherence of the regulatory framework. The PRA therefore proposes to require firms to apply a 10% exposure-weighted average portfolio risk-weight floor to UK retail residential mortgage exposures through a PRA rule.

Post-model adjustments

4.154 The PRA currently has an expectation that firms address identified IRB model issues in a timely fashion with suitable model changes and that these changes are implemented in accordance with the appropriate model changes process. However, the PRA recognises that there are instances where it is prudent and correct for firms to adjust the RWAs produced by their models on a temporary basis through a post-model adjustment (PMA).

4.155 The PRA considers that to improve the coherence of the regulatory framework, PMAs should have the same status as the other proposals set out in this section relating to RWA and EL adjustments.

4.156 Where non-compliance with modelling standards results in a material understatement of RWAs and/or EL amounts for a particular IRB model, the PRA proposes to require firms to quantify and implement PMAs as an adjustment to RWAs and EL amounts through a PRA Rule.

4.157 The relevant detail about how firms calculate PMAs would continue to be set out in an SS. The PRA proposes to clarify that PMAs should be assessed following the application of all relevant floors.

Other minor change

4.158 As set out in Chapter 13, the PRA has proposed a methodology for redenominating certain references to Euros (EUR) and US Dollars (USD) into Pound Sterling (GBP) in the PRA rules proposed in this CP.

4.159 Two references to EUR that the PRA proposes to redenominate are currently in CRR Article 153(4). These are the maximum and minimum annual sales thresholds to be used in the firm size adjustment in the IRB formula for the corporate exposure class. The PRA proposes to replace the values currently in EUR in the formula to the GBP converted values as set out in Chapter 13. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.160 In addition to this, the PRA also proposes to calculate the difference between these converted thresholds, 39.6, and substitute this for 45 in the IRB formula for the corporate exposure class. This is to help ensure that the formula does not become inconsistent as a result of the PRA’s proposed currency redenomination. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

Question 24: Do you have any comments on the PRA’s proposed approach to calculation of risk-weighted assets and expected loss, not covered by the questions above?

PRA objectives analysis

4.161 The PRA considers that the proposals set out in this section would advance the PRA’s primary objective of safety and soundness. The proposals aim to reduce the prudential risk associated with firms’ IRB models not adequately reflecting risks and therefore resulting in underestimation of RWAs and EL. In particular, the proposed removal of the SME support factor and infrastructure support factor would help ensure RWAs for qualifying exposures are robust and not subject to potential underestimation. The proposal to remove the 1.06 scaling factor would mechanistically reduce IRB risk weights but the PRA considers this is prudentially justified on the basis that the scaling factor is no longer necessary given the proposed enhancements to the IRB framework and the introduction of an aggregate output floor. The other proposals in this section provide greater clarity on existing requirements and expectations, but in most cases do not represent a significant change to the existing approach.

4.162 The PRA considers that the proposals set out in this section would facilitate effective competition by ensuring that there is an appropriate level of conservatism in IRB modelled RWAs, therefore maintaining a more level playing field with firms that use the SA. The PRA proposes to remove the SME support factor and the infrastructure support factor under both the SA and the IRB approach, which it considers would facilitate effective competition between firms using the SA and firms using the IRB approach, particularly in light of the other proposed changes impacting the SA treatment of SME and project finance exposures set out in Chapter 3, which reduce the conservatism of the underlying SA risk weights.

‘Have regards’ analysis

4.163 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Growth (HMT recommendation letters), finance for the real economy (FSMA CRR rules), and sustainable growth (FSMA regulatory principles):

• The PRA does not expect the proposals set out in this section to adversely impact sustainable financing or growth. Some of the proposals in this section would increase RWAs, while others would result in lower RWAs. The PRA considers that the proposed changes would improve the risk-sensitivity of RWAs, which would benefit sustainable growth to the extent that RWAs are inappropriately calibrated in some areas.

2. Relevant international standards (FSMA CRR rules):

• The PRA considers that the proposals set out in this section are broadly aligned with the Basel 3.1 standards. The PRA notes that its proposals relating to the asset value co-efficient of correlation multiplier potentially diverge from the Basel 3.1 standards. The PRA proposes to define FSEs as unregulated if they are not prudentially regulated as a credit institution, investment firm, or an insurer, whereas the Basel 3.1 standards imply that entities would be treated as unregulated if they are not subject to prudential requirements consistent with international norms. However, the PRA considers that the increase in clarity and reduction of inconsistencies between the definitions of a FSE and an unregulated FSE is prudentially justified, and is less complex for firms to operationalise. The proposed removal of the SME support factor and the infrastructure support factor would remove existing CRR deviations from the Basel standards and align with the Basel 3.1 standards.

3. Relative standing of the UK as a place to operate (FSMA CRR rules) and competitiveness (HMT recommendation letters):

• The PRA considers its proposal to remove the 1.06 scaling factor would help ensure risk weights are appropriate for the risk and contribute to the relative standing of the UK as a place to operate. The PRA would expect other jurisdictions to implement this change.
• The PRA considers its proposals for the SME support factor and infrastructure support factor would improve safety and soundness, are aligned with the Basel 3.1 standards, and the PRA would expect them to be aligned with the approaches proposed by most other jurisdictions (which currently do not apply the support factors). The PRA understands that the European Commission proposes to retain the SME support factor and the infrastructure support factor, and considers this could have an impact on the UK’s competitiveness in the short-term.
• While acknowledging there could be some short-term impact without further adjustment to the IRB approach, the PRA considers that the overall proposals in this section would not materially impact the relative standing of the UK as a place for internationally active credit institutions and investment firms to operate. The proposed removal of the SME support factor and the infrastructure support factor could be perceived to negatively impact the UK’s relative standing, as it weakens a direct subsidy to lending. However, the PRA considers that IRB RWAs for SME and infrastructure exposures under its proposals would be risk-sensitive and appropriate to the risk firms are taking on, which would support UK competitiveness in the medium- to long-term. Additionally, the PRA considers that SME lending is predominantly undertaken within national markets, with limited cross-border lending. The PRA also observes that firms with IRB permissions are currently making limited use of the infrastructure support factor given difficulties in assessing the scope of eligible exposures. This reduces the potential impact of its proposed removal.

4. Efficient and economic use of PRA resources (FSMA regulatory principles):

• The PRA considers that while additional supervisory resources may be required in the short- to medium-term to monitor the implementation of the proposed new requirements in this section, the incremental increase in resource required would likely be immaterial.

General requirements for use of the IRB Approach

4.164 This section sets out the PRA’s proposals relating to general requirements for use of the IRB approach.

Minimum data requirements for parameter estimation

4.165 The Basel 3.1 standards set minimum data requirements for firms to adopt the IRB approach. These state that firms should use at least five years of data from at least one source to estimate all parameters, with the exception of LGD and EAD for non-retail portfolios where seven years of data is required.

4.166 UK firms are currently subject to a five-year minimum data requirement for all parameters, including LGD and EAD for non-retail portfolios, which can be met with internal, external, or pooled data. Firms are also expected to use data from a representative mix of good and bad economic periods to calibrate PD and data from downturn periods to calibrate LGD and EAD. The PRA does not propose to make any substantive changes to this approach.

4.167 However, the PRA proposes to make two minor amendments to its data requirements in order to better align with the Basel 3.1 standards and which the PRA considers removes complexity in its rules. The amendments which the PRA proposes are to:

• clarify in its rules that data from a representative mix of good and bad economic periods form part of the minimum data requirements for modelling PD; and
• remove CRR provisions, which the PRA considers redundant in practice, that enable firms to apply for permission to reduce the minimum data requirements from five years to two years for retail exposures, and for non-retail exposures under the FIRB approach, for up to 5% of a firm’s total credit risk exposures.^{footnote [19]}

4.168 In respect of the first proposal, the PRA currently expects PD estimates to reflect a representative mix of good and bad periods so that no additional data would be required in practice. The PRA proposes that this expectation becomes a PRA rule (as set out in the section ‘Probability of default (PD) estimation’ below), as it considers that to add a complementary minimum data requirement would increase the coherence of the regulatory framework. The PRA proposes that firms would continue to be permitted to use internal, external, or pooled data to meet the proposed new requirement.

4.169 In respect of the second proposal, the PRA notes that firms with limited internal data could use external or pooled data to meet the minimum five-year data requirement. As a result, the PRA considers that there is no need for these firms to apply to the PRA to reduce minimum data requirements and the PRA considers that the application process has become obsolete in practice. The PRA considers that this would be consistent with its broader aim of removing redundant complexity wherever possible.

4.170 The PRA would not expect that either of these proposals would adversely impact the ability of firms to adopt the IRB approach and would not expect that either proposal would have any impact on the facilitation of effective competition.

Data requirements and maintenance

4.171 The PRA proposes to make a number of amendments to its existing approach to data requirements and maintenance under the IRB approach. The PRA considers that these proposed amendments would improve the quality of data used in IRB models and would therefore advance its safety and soundness objective.

4.172 The PRA proposes to introduce more specific requirements for the collection and storage of:

• key borrower and facility characteristics;
• realised default rates;
• the components of loss for defaulted exposures; and
• the process of allocating exposures to grades or pools for retail exposures.

4.173 The PRA also proposes to introduce a requirement that firms submit an annual model inventory to the PRA. This proposal would replace existing requirements to submit model inventories where these have been applied to individual firms under s55M FSMA.

4.174 The PRA considers that its proposals would result in prudent and proportionate data standards, and that its proposals align with the Basel 3.1 standards. The PRA considers these changes would enhance clarity in the regulatory framework and would facilitate a consistent approach to collecting and storing of data among firms. This would enhance the PRA’s ability to effectively supervise firms.

IRB model governance and validation

4.175 In relation to standards for IRB model governance and validation, the PRA proposes to:

• require that senior management would approve all material differences between established procedures and actual practice for parameter rating and estimation processes;
• clarify that the firm’s management body or a designated committee thereof would be solely responsible for approving all material aspects of a firm’s rating and estimation processes;
• introduce more specific requirements for the summary reports that would need to be produced by the firm’s credit risk control unit (eg relating to historic default data and grade migration);
• explicitly require internal audit functions to document their findings;
• clarify an existing requirement relating to the consistency of quantitative validation through time by requiring that the methods and data used by firms should not vary systematically with the economic cycle; and
• add an expectation that firms take steps to remediate any deficiencies in the quality of the data used, or in their processes for maintenance of the data, in a timely manner.

4.176 The PRA considers that the proposals on IRB model governance and validation align with the Basel 3.1 standards, and the PRA considers that the proposals do so in a proportionate way that would increase clarity for firms.

Other proposed changes

4.177 The PRA proposes to remove an existing expectation that the rank-ordering of the IRB rating system should be exactly the same as the rank-ordering of the rating system used for internal risk management purposes. The PRA instead proposes to align with the Basel 3.1 standards and set an expectation that the rank-ordering of the IRB rating system should play an essential role in the rank-ordering used for internal risk management and decision-making purposes. The PRA considers this would remove an unnecessary barrier to firms’ ability to access the IRB framework and would therefore advance the PRA’s secondary objective of facilitating effective competition. See Appendix 13.

4.178 The PRA proposes to make two further minor changes to the framework as follows:

• to permit firms to apply a zero margin of conservatism in respect of the general estimation error where they can demonstrate that the general estimation error is immaterial. See Appendix 13; and
• to withdraw the existing provision that firms need not give equal importance to historic data if more recent data is a better predictor of loss rates or drawdowns. The PRA considers that this change would help ensure coherence and alignment with the Basel 3.1 standards.

Question 25: Do you have any comments on the PRA’s proposed general requirements for the use of the IRB approach?

PRA objectives analysis

4.179 The PRA considers that the proposals set out in this section would advance the PRA’s safety and soundness objective by ensuring that firms are subject to robust standards on the quality of data, data processes, and governance of IRB models.

4.180 The PRA considers the proposed additional data requirements would not be onerous for firms to implement and therefore considers that these would not have a detrimental impact on the facilitation of effective competition. By providing additional detail on what the PRA would expect with regards to data, processes, and governance of IRB models, the PRA considers that firms would be more likely to take consistent approaches, rather than potentially taking different interpretations of less detailed requirements. The PRA considers that this may help support a level playing field across firms and promote competition. The PRA considers that effective competition would be further promoted by its proposal to revise its existing expectation relating to rank-ordering of rating systems, as this would improve accessibility to the IRB framework.

‘Have regards’ analysis

4.181 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA would expect that the potential additional costs some firms could incur to help ensure data is stored and maintained in line with the proposals would be low and proportionate to the benefits of firms having sufficiently robust data.

2. Efficient and economic use of the PRA’s resources (FSMA regulatory principles):

• The PRA considers that by providing more detail on what it expects to be included in data, processes, and governance for IRB models, it would be able to process applications and conduct reviews of existing IRB models more efficiently. This is because it is more likely that the PRA would not have to use its resources to request additional information from firms.

3. Competitiveness (HMT recommendation letters):

• The PRA considers that the proposals set out in this section would increase external confidence in the quality of firms’ IRB models and enhance the ability of the PRA to assess the safety and soundness of firms. The PRA considers this would support the attractiveness of the UK’s financial market.

4. Relevant international standards (FSMA CRR rules):

• The PRA considers that the proposed changes set out in this section align with the Basel 3.1 standards.

5. Senior management responsibility (FSMA regulatory principles):

• The PRA has considered the impact of the proposals set out in this section on the responsibilities of senior management and considers that it is desirable to make the proposed clarifications in order to clearly set out the PRA’s expectations.

Definition of default

4.182 This section sets out the PRA’s proposals relating to the definition of default and is relevant to firms using the SA and the IRB approach.

4.183 The PRA’s existing expectations relating to the definition of default are set out in SS10/13 for firms using the SA and SS11/13 for firms using the IRB approach. These SSs include an expectation that firms should comply with the EBA Guidelines on the definition of default (EBA/GL/2016/07).

4.184 The PRA proposes to issue a new SS on the definition of default, which would replace existing material in SS10/13 and SS11/13 as well as the EBA Guidelines. The proposed new SS (see Appendix 14) would contain all existing expectations relating to the definition of default with the exception of material that the PRA proposes to move to PRA rules as set out below. The PRA proposes to make a number of minor changes to these expectations that are either consequential on other proposed changes or are considered to enhance the coherence and clarity of the regulatory framework.

4.185 The PRA proposes to move the following expectations relating to the definition of default to the PRA Rulebook to become formal requirements for firms:

• expectations relating to the ability for firms applying the SA to treat exposures as retail exposures for the purpose of applying the definition of default;
• certain provisions relating to the circumstances in which the counting of days past due could be suspended;
• the specific treatment for exposures to central governments, local authorities, and PSEs, which would enable firms to treat exposures relating to the supply of goods and services as non-defaulted for up to 180 days past due in certain circumstances. The scope of this treatment would continue to be limited and would not extend to bonds issued by such entities;
• the period over which defaulted exposures remain classified as being in default once the triggers of default cease to apply; and
• the definition of distressed restructuring, including clarifying the definition of forbearance according to which a distressed restructuring is considered to have occurred. This would align the concept of forbearance used with that set out in the CRR.

4.186 The PRA proposes to make a number of further minor amendments to the PRA rules relating to days past due, including:

• removing references to a PRA discretion to permit use of a 180 days past due criteria and instead requiring firms to use a 90 days past due criteria in line with existing PRA expectations;
• relocating PRA rules relating to the application of the materiality threshold for the counting days past due so they are located alongside other rules within the PRA Rulebook relating to the definition of default; and
• applying the materiality threshold for retail exposures to all exposures that meet the SA retail criteria where a firm applies the SA approach, including where it has an IRB permission for other exposures.

4.187 The PRA proposes to set an expectation to clarify that exposures should be classed as being in default where a trigger of default applies, regardless of any credit risk mitigation technique used. In particular, the PRA proposes to clarify that firms using the IRB approach and applying the ‘parameter substitution method’ described in Chapter 5 – Credit risk mitigation should class defaulted exposures that are guaranteed by an entity that is not in default as being in default for the purpose of the ‘EL – P’ calculation.

Question 26: Do you have any comments on the PRA’s proposed approach to the definition of default?

PRA objectives analysis

4.188 The PRA considers that the proposals set out in this section would advance the PRA’s primary objective of safety and soundness by enhancing the clarity and coherence of requirements and expectations relating to the definition of default.

4.189 The PRA considers that the proposals set out in this section are consistent with its secondary objective of facilitating effective competition as the PRA proposes to apply broadly consistent requirements and expectations to firms using the SA and the IRB approach.

‘Have regards’ analysis

4.190 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Relevant international standards (FSMA CRR rules):

• The PRA considers its proposals are materially aligned with the Basel 3.1 standards.

2. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA expects other jurisdictions to apply similar approaches, and therefore does not consider that the proposals set out in this section would adversely impact the competitiveness of the UK.

Input floors

PD, LGD, and EAD input floors

4.191 This section sets out the PRA’s proposed approach to introduce PD, LGD, and EAD input floors.

4.192 As previously noted, the BCBS has identified excessive complexity in IRB approaches and the resultant modelled RWAs were found to lack robustness, consistency, and comparability across firms. As a means to address this, the Basel 3.1 standards introduce exposure-level floors for firm estimated IRB parameters that are used as inputs to the calculation of modelled RWAs (‘input floors’). These include PD floors for the FIRB and the AIRB approaches, and LGD and EAD floors for the AIRB approach. In some cases, these floors consist of recalibrated values of floors in the existing Basel standards. In other cases, the floors represent new constraints on firms’ IRB models.

4.193 The PRA considers these input floors are an important backstop for improving comparability, reducing unwanted variability, and reducing the cyclicality of modelled RWAs. The PRA considers that application of the floors would help ensure a minimum level of prudence, in particular for low default portfolios where data are limited, in the cases where modelling of such portfolios continues to be permitted.

4.194 The PRA therefore proposes to introduce PD, LGD, and conversion factor (CF)^{footnote [20]} input floors as set out below. The proposed floors would align with the calibration in the Basel 3.1 standards, except for the proposed UK retail residential mortgage PD floor, for which the PRA proposes a modestly higher floor than in the Basel 3.1 standards.

4.195 The proposed floors are designed and calibrated to reflect the riskiness of the respective exposure classes as well as firms’ ability to robustly model different exposure classes. In determining the proposed floors, the PRA has sought to achieve an appropriate balance between ensuring the robustness of estimates while maintaining risk-sensitivity of models and to avoid creating incentives for firms to reduce investment in modelling capability or to increase the risk of their lending portfolio in order to reduce the impact of the floors.

4.196 The proposed PD floors would apply to all exposures capitalised under the IRB approach (except for exposures subject to the slotting approach), and the LGD and CF floors would apply to exposures to which the AIRB approach is applied (the CF floors would not apply where CF modelling is not permitted). The input floors would apply to non-defaulted and defaulted exposures but would not apply for the purpose of calculating ‘best estimate of expected loss’.

PD input floors

4.197 The PRA proposes to introduce the following PD input floors:

• a 0.1% PD floor for UK retail residential mortgage exposures and for QRREs categorised as transactors; and
• a 0.05% PD floor for all other exposures.

4.198 The PRA’s proposed 0.1% PD floor for UK retail residential mortgage exposures, which is somewhat more conservative than the 0.05% PD floor in the Basel 3.1 standards, would reflect the PRA’s longstanding concern that some IRB UK retail residential mortgage risk weights have been falling over recent years, may be too low, and may not fully reflect the potential for losses in unlikely, but plausible, tail scenarios. The PRA’s proposed floors for all exposures other than UK retail residential mortgage exposures would be aligned with the Basel 3.1 standards.

4.199 In PS16/21 – ‘Internal Rating Based UK mortgage risk weights: Managing deficiencies in model risk capture’, the PRA noted that calculating IRB mortgage risk weights is inherently uncertain and set out a number of observations that have cumulatively influenced the PRA’s concern, and which the PRA considers to still be relevant. These observations included:

• risk weights for low loan to value (LTV) mortgages can be difficult to calibrate due to limited historical experience of either extreme house price falls or the varying effects different types of economic downturn might have;
• average UK IRB mortgage risk weights are below, and in some cases significantly below, international peers;
• a number of other international jurisdictions have acted to address low mortgage risk weights generated from some IRB models;
• there is large variation in IRB risk weights between UK firms, including material variation between loans with similar LTVs;
• the significant difference between residential mortgage risk weights under the IRB approach and risk weights under the SA; and that
• IRB mortgage risk weights have been falling for a number of years. Although it was unclear whether the trend would continue in the short-term given the global pandemic, there remained the risk that the medium- and longer-term trend of falling mortgage risk weights would continue, increasing the likelihood that some IRB mortgage lenders were undercapitalised.

4.200 In response to these concerns, PS16/21 introduced an expectation that firms apply a 10% UK retail residential mortgage portfolio-level risk weight floor, which has applied since 1 January 2022. The PRA proposes now that application of this floor would instead become a requirement.

4.201 The PRA noted in PS16/21 that even after the application of the 10% portfolio-level floor, it continued to have concerns that some individual IRB mortgage risk weights were too low and were a source of prudential risk, and that the divergence between SA and IRB mortgage risk weights raised competition concerns. The PRA therefore noted that as part of its implementation of the Basel 3.1 standards, it would consider whether higher PD and LGD input floors would be introduced.

4.202 The PRA considers that its proposed PD floor of 0.1% for UK retail residential mortgage exposures would achieve an appropriate balance between its concerns regarding the adequate capitalisation of UK residential mortgage exposures, the need to preserve model risk-sensitivity and rank-ordering, and to maintain incentives for firms to develop and maintain high quality models. The PRA considers that introducing a PD floor of 0.1% would:

• address model risk and uncertainty in PD models caused by a lack of historic default data for low risk exposures (ie low LTV mortgages); and
• limit the extent to which a fall in default rates translates to falling RWs, as the PRA considers that this is a key driver of the recent observed decrease in UK residential mortgage risk weights.

4.203 The PRA proposes to apply the 0.1% PD floor to UK retail residential mortgage exposures only. Non-UK retail residential mortgage exposures would be subject to the 0.05% PD floor, in line with the Basel 3.1 standards, as the PRA does not have evidence that a higher PD floor would be justified for non-UK retail residential mortgage exposures.

LGD input floors

4.204 The PRA proposes to implement LGD input floors in line with the Basel 3.1 standards. These floors would be flat (ie non-variable) LGD floors for unsecured and retail residential mortgage exposures, and variable floors for other fully or partially secured exposures.

4.205 The proposed flat LGD floors are set out in the table below:

Table 2: Proposed unsecured and retail residential mortgage LGD floors

4.206 The variable LGD floors for other fully or partially secured exposures would be calculated at exposure level as a weighted average of the relevant unsecured floor (taken from Table 2 above) and prescribed secured floors (set out in Table 3 below). The weights would be determined by the value of the exposure covered by each type of collateral after application of haircuts specified in the ‘foundation collateral method’ (collateral not eligible under this method would not be recognised).^{footnote [21]}

4.207 The proposed secured floors that would be used to calculate these variable floors are set out in Table 3 below:

Table 3: Proposed secured LGD floors

4.208 The proposed flat 5% LGD floor for retail residential mortgage exposures would be consistent with the floor that the PRA currently expects firms to apply to these exposures. The PRA proposes that this floor would become a requirement in rules as part of these proposals and that the floor would continue to apply regardless of whether the mortgage exposure is UK or non-UK. This approach would be aligned with the Basel 3.1 standards.

4.209 The PRA proposes to remove the existing 10% exposure-weighted average portfolio LGD floor for residential mortgages in the retail exposure class and the existing 15% exposure-weighted average portfolio LGD floor for commercial mortgages in the retail exposure class. The PRA considers these floors are not required given the proposed retention of a 10% portfolio risk weight floor for retail residential mortgages, the relative low materiality of retail commercial mortgages, and the proposed input floors. Removing these floors would be consistent with the Basel 3.1 standards.

Conversion factor and exposure at default input floors

4.210 The PRA currently would expect that EAD is floored at current drawings and, consequently, that CF estimates are not less than zero. The Basel 3.1 standards introduce a new EAD input floor calculated as the sum of the on-balance sheet amount and the off-balance sheet exposure multiplied by 50% of the applicable SA CF.

4.211 The PRA considers that it is necessary to specify the floor separately for firms that provide own estimates of CFs and firms that provide own estimates of EAD. The PRA therefore proposes the following implementation:

• where a firm provides own estimates of CFs, the PRA proposes that these CF estimates would be floored at 50% of the SA CF; and
• where a firm provides own estimates of EAD, the PRA proposes that these EAD estimates would be floored at the current balance plus 50% of the SA CF multiplied by the off-balance sheet exposure.

4.212 Further details about the cases when firms would provide own estimates of CF or own estimates of EAD are set out in the section ‘EAD estimation’ below.

Question 27: Do you have any comments on the PRA’s proposed PD, LGD, and CF or EAD input floors?

PRA objectives analysis

4.213 The PRA considers that the proposed input floors would advance the PRA’s primary objective of safety and soundness because the floors should improve comparability, reduce unwarranted variability, and reduce the cyclicality of modelled risk weights. The PRA considers that the proposed floors would help ensure a minimum level of prudence, in particular for low default portfolios where data are limited.

4.214 The proposals would support the PRA’s secondary competition objective as the PRA considers that the input floors would narrow the gap between some IRB and SA risk weights, which would help firms using the SA compete with firms using the IRB approach. This is particularly the case for the proposed PD floor for UK retail residential mortgage exposures.

‘Have regards’ analysis

4.215 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Relevant international standards (FSMA CRR rules):

• The PRA’s proposed PD floor for UK retail residential mortgages would be higher than the minimum floor specified in the Basel 3.1 standards, which the PRA considers to be appropriate in order to address PRA concerns related to the current adequacy of some IRB UK retail residential mortgage risk weights. The PRA’s proposed implementation of input floors for all exposures other than UK retail residential mortgage exposures would be aligned with those specified in the Basel 3.1 standards.

2. Relative standing of the UK as a place to operate (FSMA CRR rules) and competitiveness (HMT recommendation letters):

• The PRA expects other jurisdictions to implement similar input floors and to remove exposure-weighted average portfolio LGD floors for retail mortgages, as this would be aligned with the changes introduced in the Basel 3.1 standards. The PRA also notes that it expects firms to use a 90 days past due definition of default for all exposure classes, which results in the existing exposure-weighted average LGD floor for retail mortgages binding on some firms. The proposed removal of this floor may enable firms to apply lower LGDs that better reflect the risk of their exposures and improve the risk-sensitivity of modelled risk weights. The PRA considers that this would support the UK’s relative standing.
• The PRA’s proposal to increase the PD input floor for UK retail residential mortgages to 0.1% is more conservative than the calibration in the Basel 3.1 standards. However, the PRA would not expect that this would have a material impact on the relative standing of the UK as retail residential mortgage lending is a predominantly domestic market with limited cross-border activity. In addition, several jurisdictions already set materially higher risk weight floors for mortgage exposures than implied by the relatively modest PD floor proposed by the PRA.

3. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA considers that the proposed input floors would be appropriately prudent, and recognises that there are drawbacks from setting overly conservative floors, such as loss of risk-sensitivity in modelled parameters. The PRA considers the proposed removal of the 10% exposure-weighted average portfolio LGD floor for residential mortgages in the retail exposure class and the 15% exposure-weighted average portfolio LGD floor for commercial mortgages in the retail exposure class would avoid excessive complexity in the input floors framework.

4. Finance for the real economy (FSMA CRR rules):

• The PRA considers that the proposals would support firms’ financing of the real economy by ensuring risks are appropriately capitalised. The PRA considers that the proposed floors would reduce cyclicality in risk weights, which should support firms’ lending to the real economy throughout the economic cycle.

Probability of default (PD) estimation

4.216 This section sets out the PRA’s proposals for PD estimation under the IRB approach to credit risk. For detail on the PRA’s proposals relating to PD input floors, please refer to the ‘Input floors’ section above.

Use of continuous rating scales

4.217 Under the CRR, firms are permitted to apply either discrete or continuous rating scales for PD estimation:

• for discrete rating scales exposures are grouped into rating grades based on risk characteristics, with a PD estimated for each grade; and
• for continuous rating scales exposures are not grouped together – instead, each exposure is assigned an individual PD estimate based on risk characteristics.

4.218 The PRA has observed that, for PD estimation, continuous rating scales are used relatively infrequently by firms, and that most firms adopt discrete rating scales in their hybrid mortgage model applications.

4.219 The PRA considers that the use of continuous rating scales for PD estimation could typically result in lower RWAs than the use of discrete rating scales. This is because total RWAs tend to reduce as the number of grades increases, due to the shape of the IRB risk weight function in respect of PD. While increasing the number of grades could potentially result in more accurate RWAs, the PRA considers that this is only justified where the extra grades result in a genuine increase in risk capture. The PRA also considers that adding additional grades does not increase risk-sensitivity beyond a certain point, and therefore a risk of continuing to allow continuous rating scales for PD estimation is that these reduce RWAs without increasing the overall risk capture of the model.

4.220 The PRA therefore proposes to prohibit the use of continuous rating scales in PD models and to require firms to use discrete rating scales instead. The PRA considers that this proposal is aligned with the Basel 3.1 standards.

4.221 The PRA recognises that a consequence of this proposal is that variable scalar approaches, where firms transform point-in-time (PiT) or hybrid rating systems to through-the-cycle (TtC) outcomes, would no longer be permitted. This is because the mechanics of variable scalar approaches effectively result in firms using a continuous rating scale. The PRA considers this would be an appropriate outcome given the concerns that the PRA has previously identified regarding the risk capture of variable scalar approaches, and considers that the number of models affected following the introduction of hybrid modelling for mortgage portfolios would be low. The PRA therefore proposes to withdraw its existing expectations relating to the use of variable scalar approaches.

4.222 The PRA does not consider that similar prudential concerns arise from the use of continuous rating scales for LGD and EAD models because the IRB risk weight formula is linear in LGD and EAD. The PRA therefore does not propose to restrict the use of continuous rating scales for LGD and EAD.

Obligor grade adjustments

4.223 For exposures to corporates and institutions, for which the exposures to the obligor are subject to a guarantee, the Basel 3.1 standards and the CRR generally require firms to assign all exposures to a particular obligor to the same obligor grade, irrespective of differences in the nature of each transaction. However, the CRR also sets out a number of exceptions to this requirement, including that the treatment of the guarantees could be reflected in an adjusted obligor grade assignment.

4.224 The PRA proposes to clarify that adjustments to obligor grade assignments could be made outside the credit risk mitigation (CRM) framework and therefore the CRM eligibility criteria would not apply.

4.225 In addition, the PRA considers that the purpose of these adjustments to obligor grades is to reflect reductions in default risk arising from the existence of guarantees rather than to reflect potential recoveries in the event that default occurs. In practice, the PRA recognises that firms could interpret existing requirements to permit firms to recognise a wider range of support arrangements including, for example, letters of comfort as well as verbal and implicit indications of support.

4.226 The PRA considers that it would be desirable for firms to reflect support arrangements in PD models where they are able to demonstrate a reduction in default risk, as the PRA considers that linking RWAs to risks advances the PRA’s primary objective. As such, the PRA proposes to continue to permit firms to reflect certain support arrangements in IRB obligor rating grade assignments.

4.227 However, the PRA considers that undocumented support arrangements are often unclear and not robust. In addition, the PRA considers that it is difficult for firms to demonstrate a reduction in default risk from these arrangements, and it is difficult for supervisors to challenge whether these arrangements are appropriately reflected in IRB models. The PRA therefore proposes that adjustments to obligor grades would only be permitted where the support arrangements are in writing.

4.228 In order to implement this proposal the PRA proposes to:

• exclude undocumented support arrangements from the requirements to incorporate ‘all available information’ in IRB rating systems and require firms to disregard this information for the purpose of assignment of exposures to obligor grades;
• require firms to disregard undocumented support arrangements when assessing model overrides; and
• clarify that all documented support arrangements and not just guarantees could potentially be recognised.

4.229 Firms reflecting unfunded credit protection (UFCP) in PD or LGD are required to floor risk weights at the risk weight that would apply to a comparable direct exposure to the protection provider. The PRA proposes to extend the scope of this floor to firms using obligor grade adjustments in order to provide a further safeguard against the effect of protection arrangements being over-reflected in RWAs.

4.230 The PRA considers that the Basel 3.1 standards are open to interpretation in respect of the interaction of obligor grade adjustment with CRM techniques, for example PD substitution (which the PRA proposes to retain) and PD adjustment (which the PRA proposes to withdraw). Further details of the PRA’s proposals regarding this interaction are set out in Chapter 5.

Other proposals

4.231 The PRA also proposes to make the following minor amendments to PD estimation:

• clarify that firms would be required to consider ‘seasoning’ as a risk driver for retail exposures, in line with the Basel 3.1 standards. See Appendix 13;
• move an existing expectation that PD estimates should be based on a representative mix of good and bad economic periods to PRA rules and make a number of clarifications to related expectations. See Credit Risk: Internal Ratings Based Approach (CRR) Part and Appendix 13;
• withdraw the option for firms to place greater weight on more recent observations for retail exposures, in line with the Basel 3.1 standards;
• clarify that a PD needs to be estimated for each rating grade and that this would need to be based on a count-weighted average of one-year defaults in line with the Basel 3.1 standards (the PRA considers that the proposal does not represent a substantive change to existing requirements). See Appendix 13;
• introduce an expectation that firms using the parameter substitution method (as described in Chapter 5) should collect information on the performance of the obligor and the exposure and use this information in PD estimation where appropriate. See Appendix 13;
• withdraw existing expectations relating to the consistency of time horizons in different stages of the modelling process (the PRA considers it can be desirable for firms to apply different time horizons for the purpose of assigning exposures to rating grades and for the purpose of model calibration in certain circumstances);
• withdraw existing expectations relating to making conservative adjustments due to old financial statements and external ratings (these relate to requirements set out in an EBA Final Draft RTS^{footnote [22]} that was not implemented in the UK);
• withdraw existing expectations related to the treatment of missing ratings because the PRA considers that all exposures within the scope of a rating system should be rated, and should be rated in a conservative manner where there is missing information; and
• clarify that the effect of any UFCP would be required to be excluded from PD models where the SA ‘risk weight substitution method’ is applied to exposures under the IRB approach. See Appendix 13.

Question 28: Do you have any comments on the PRA’s proposals on PD estimation?

PRA objectives analysis

4.232 The PRA considers that the proposals set out in this section would advance the PRA’s primary objective. The proposals aim to reduce the prudential risk that firms’ IRB models do not adequately reflect risks and therefore result in underestimation of RWAs. In addition, the proposal to prohibit the reflection of undocumented support arrangements in assignments to obligor rating grades would reduce the risk that support arrangements that do not result in the expected risk-mitigating effect are recognised. This would therefore promote the safety and soundness of firms.

4.233 The PRA considers that the proposals set out in this section facilitate effective competition. All the proposals in this section would apply equally to firms using the FIRB and AIRB approaches, which the PRA considers would facilitate effective competition between them.

‘Have regards’ analysis

4.234 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Growth (HMT recommendation letters), finance for the real economy (FSMA CRR rules), and sustainable growth (FSMA regulatory principles):

• Continuing to recognise documented support arrangements in PD models would support sustainable financing and growth by better linking RWAs to risks. The PRA considers that in order for lending to be sustainable and growth to be supported, RWAs should appropriately reflect risk.

2. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA would expect the net effect of the proposals in this section to be in line with other jurisdictions, and therefore considers that the proposals set out in this section would not have a significant impact on the competitiveness or relative standing of the UK.

3. Relevant international standards (FSMA CRR rules):

• The PRA considers that the Basel 3.1 standards are open to interpretation about how adjustments to borrower grades should be treated. However, the PRA considers that its proposed approach is broadly aligned with international standards. The PRA considers that the other proposals set out in this section are aligned with the Basel 3.1 standards.

4. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA recognises that its proposal to prohibit continuous rating scales for PD estimation may mean that some existing IRB models could have to be rebuilt. However, as outlined above, the PRA considers that the number of models affected is likely to be low. In addition, the PRA considers that the proposed IRB implementation timelines set out in this chapter would reduce the burden on firms by giving them an appropriate amount of time to make any required model changes.

5. Efficient and economic use of the PRA’s resources (FSMA regulatory principles):

• The PRA considers that certain proposals in this section, for example the removal of recognition of certain types of support arrangements, would reduce the amount of time and resources it spends reviewing models that are unlikely to be sufficiently robust. This would enable the PRA to focus its resources more efficiently.

Loss given default estimation

4.235 This section sets out the PRA’s proposals relating to LGD estimation, including the impact of CRM on LGD models.

FIRB LGD values for unsecured corporate senior claims

4.236 Firms using the FIRB approach currently apply a 45% LGD for all unsecured exposures to corporates that are senior claims. The Basel 3.1 standards reduce the FIRB LGD value for these exposures from 45% to 40%, with the exception of exposures to financial corporates where the FIRB LGD remains at 45%. The BCBS calibrated the new FIRB LGD value using international empirical data and the PRA does not have any evidence to suggest this is not an appropriate calibration.

4.237 The PRA therefore proposes to align with the Basel 3.1 standards and reduce the FIRB LGD value for exposures to non-financial corporates that are senior claims to 40%. The PRA does not propose to change the existing 45% FIRB LGD value for exposures to financial corporates that are senior claims.

LGD modelling collateral method for firms using the AIRB approach

4.238 Firms using the AIRB approach typically reflect the risk mitigating impact of collateral in their LGD models. As set out in Chapter 5, the PRA proposes a number of restrictions on where this approach (which the PRA proposes to call the ‘LGD modelling collateral method’) may be applied.

Approach to ineligible and disregarded collateral

4.239 When recognising collateral in LGD estimates while using the AIRB approach, firms are currently required to establish internal requirements for collateral management, legal certainty, and risk management that are generally consistent with those set out in the CRM chapter of the CRR. The PRA proposes to retain this approach for firms using the LGD modelling collateral method subject to clarifying that the relevant CRM standards are those that apply to firms using the FIRB approach, in line with the Basel 3.1 standards. The PRA also proposes to clarify that collateral that does not meet these requirements should be classed as ‘ineligible’ for the purpose of applying the LGD modelling collateral method.

4.240 In addition, there is currently some ambiguity in the CRR regarding whether firms using the AIRB approach have the option to disregard eligible collateral. This is because firms are required to use ‘all relevant information’ when developing their models.

4.241 The PRA considers that it is desirable to allow firms using the LGD modelling collateral method to disregard eligible collateral in order to be more consistent with the approach taken for firms using the SA and the FIRB approach in order to help limit complexities and operational costs for firms. The PRA considers that it is particularly desirable for firms to be able to disregard eligible collateral should they wish to in cases where collateral is difficult to model. The PRA therefore proposes to allow firms to disregard eligible collateral when using the LGD modelling collateral method. In such cases, a firm would treat the part of the exposure covered by disregarded collateral as being unsecured.

4.242 Firms are currently subject to an expectation that recoveries from ineligible collateral should be included in estimates of unsecured LGD, but with appropriate adjustments to avoid bias in LGD estimates. The PRA proposes to withdraw that expectation and to instead require firms applying the LGD modelling collateral method to exclude recoveries from ineligible and disregarded eligible collateral when calculating unsecured LGD. The PRA also proposes to clarify in PRA rules that firms would not include ineligible and disregarded eligible collateral as a risk driver in LGD models.

4.243 The PRA considers that if cash flows from ineligible and disregarded eligible collateral were included in LGD estimates, it would effectively negate the classification of the collateral as ineligible or disregarded and bias estimates. The PRA considers this could give rise to prudential risks.

4.244 Related to this, firms are also currently subject to an expectation that, where they do not regularly sell credit obligations as part of their recovery processes, and the allocation of the part of the price related to collaterals is too burdensome to make or too unreliable, they can decide not to take these observations into account in the model development process. In contrast, the Basel 3.1 standards permit firms to simply derecognise the collateral for these cases. Therefore, the PRA proposes to remove this expectation and align with the Basel 3.1 standards.

Proposed introduction of an alternative methodology

4.245 The Basel 3.1 standards introduce an alternative methodology for firms using the AIRB approach, to address the challenge of modelling robust LGD estimates where there are limited data. Under the alternative methodology, firms calculate LGD by combining modelled LGD estimates for the unsecured part of an exposure with FIRB LGD parameters for the secured part of an exposure.

4.246 The PRA recognises the modelling challenges identified by the BCBS and therefore proposes to introduce an alternative methodology as part of the LGD modelling collateral method, which the PRA proposes would be applied in cases where firms lack sufficient data to model collateral recoveries. The PRA proposes that LGD under the alternative methodology would be calculated using the following formula where a single type of collateral is recognised:

where:

• LGD* is the LGD applicable to a collateralised transaction;
• LGD_u is the estimated LGD of the exposure disregarding collateral (ie treating the exposure as unsecured);
• LGD_s is the foundation collateral method secured LGD applicable to the collateral type;
• E is the current value of the exposure after the effect of on-balance sheet netting;
• E_s is the current value of the collateral after the application of the applicable volatility adjustment (H_c);^{footnote [23]}
• E_u is the value of the unsecured exposure calculated as follows:

• H_c is the volatility adjustment applied to the collateral (as defined according to the financial collateral comprehensive method for financial collateral and according to the foundation collateral method for non-financial collateral); and
• H_E is the volatility adjustment applicable to the exposure.

4.247 The proposed formula is in line with that proposed for firms applying the foundation collateral method, as set out in Chapter 5 (an analogous formula would be applied where multiple collateral types are recognised). In particular, LGD_S, E, E_U, E_S, H_C and H_E would be calculated in accordance with the foundation collateral method and only LGD_U would be estimated by firms. As a result, only collateral eligible under the foundation collateral method would be recognised under the alternative methodology.

4.248 The PRA proposes that firms using the alternative methodology would be required to estimate LGD_U using an approved IRB model. The PRA also proposes that firms would not be permitted to take account of collateral recoveries in the model used to estimate unsecured LGD to avoid double counting the effect of the collateral.

4.249 The PRA proposes to require the use of the alternative methodology if a firm is using the LGD modelling collateral method to recognise the effect of collateral and does not have sufficient data to model the effects of the collateral. The PRA proposes to set an expectation that the data would be considered insufficient where firms have fewer than 20 relevant data points for any non-financial collateral that the firm wishes to recognise in their LGD models. This corresponds to a condition already in place in the PRA’s wholesale LGD framework (which the PRA proposes to remove as set out later in this section). The PRA does not propose to set an equivalent expectation for financial collateral, in line with its existing approach.

Other proposed minor amendment

4.250 The PRA proposes to introduce an expectation that firms estimating LGDs should, when calculating realised LGDs, reflect any recognised netting agreements in the EAD parts of the LGD calculation but should not treat any cash flows arising from netting as post-default recoveries in the economic loss part. This expectation would be consistent with the PRA’s proposals that netting agreements would only be recognised through exposure values as set out in the Chapter 5, and would complement existing guidance regarding the calculation of realised LGDs. See Appendix 13.

LGD adjustment method for unfunded credit protection (UFCP)

4.251 As noted in Chapter 5, the PRA proposes to describe recognition of UFCP in LGD estimates as the ‘LGD adjustment method’. Firms using UFCP would be able to use this method subject to the restrictions set out in that chapter.

4.252 The PRA proposes to clarify the eligibility requirements for recognising UFCP that would apply as part of the LGD adjustment method in order to bring these more in line with the eligibility requirements that would apply when firms use alternative CRM methods to recognise UFCP.

Combination with adjustments to obligor grades

4.253 As noted in Chapter 5, while the PRA proposes to withdraw the option of adjusting PD to reflect UFCP, it proposes to continue to permit firms to recognise support arrangements through the adjustment of obligor grades in some circumstances. The PRA notes that both the CRR and the Basel 3.1 standards are open to interpretation about whether firms can combine the LGD adjustment method with adjustments to obligor grades.

4.254 The PRA considers that for firms to simultaneously apply the LGD adjustment method and reflect the impact of a guarantee by adjusting obligor grades without double counting is challenging. The PRA therefore proposes to clarify that firms applying the LGD adjustment method would not be permitted to also reflect the effect of the guarantee by adjusting obligor grades.

Modelling standards

4.255 The PRA proposes to introduce enhanced expectations regarding modelling standards for firms applying the LGD adjustment method to help ensure that LGD estimates are sufficiently robust. The proposed expectations (also see Appendix 13) would include:

• that firms should have clear policies for assessing the effects of UFCP that are consistent with internal risk management practices; and
• that firms should reflect currency mismatches, correlation between the protection provider’s and the obligor’s ability to pay, and the defaulted status of the protection provider (where relevant).

Approach to ineligible and disregarded UFCP

4.256 In line with the proposals relating to collateral, the PRA proposes to clarify that recognition of UFCP within the LGD adjustment method is optional and therefore firms could choose to disregard UFCP. The PRA also proposes to prohibit the recognition of UFCP in LGD models if firms are not using the LGD adjustment method as well as where the UFCP is ineligible or has been disregarded.

Calculation of LGD when applying the parameter substitution method

4.257 As set out in Chapter 5, the PRA proposes that firms using the AIRB approach that wish to recognise the effects of UFCP would be required to apply the risk weight substitution method under certain circumstances, and would be required or permitted to use the parameter substitution method in specific other circumstances. In order to apply either of these methods, it would be necessary for firms using the AIRB approach to estimate LGD values for the exposures as if there were no UFCP.

4.258 The PRA proposes to clarify the framework by setting an expectation that, for the purpose of estimating LGDs as if there was no UFCP (also see Appendix 13):

• cash flows from the protection provider would not be taken into account;
• cash flows for funded credit protection (FCP) associated with the exposure could be taken into account in respect of the part of the exposure covered by the FCP;
• indirect costs would be taken into account in line with the principles and techniques that firms use in their own accounting systems;
• direct costs that are linked to the exercising of the UFCP would not be taken into account, but all other direct costs would be taken into account; and
• direct costs relating to the realisation of FCP would be taken into account in respect of the part of the exposure covered by the FCP.

Use of elements of the SA and the FIRB approach in LGD estimates

4.259 The PRA considers that it may be appropriate for firms applying the AIRB approach to incorporate elements of the SA and the FIRB approach within their LGD models in specific circumstances (for example by incorporating the supervisory haircuts used in the ‘financial collateral comprehensive method’ (FCCM)). The PRA proposes to introduce an expectation that firms should provide appropriate justification for their approach.

Removal of the PRA’s wholesale LGD framework

4.260 The PRA’s wholesale LGD framework was introduced by the Financial Services Authority (FSA) in 2012 to address modelling deficiencies for low default portfolios. The framework aims to help ensure that LGD estimates do not assume a level of recoveries that is not supported by data.

4.261 The PRA proposes in this CP to introduce three new constraints that would reduce the need for the PRA’s wholesale LGD framework:

• prohibiting LGD modelling for exposures to institutions, financial corporates, and large corporates (see section ‘Restrictions on LGD modelling’);
• for corporate exposures where LGD modelling would still be permitted, introducing LGD input floors (see section ‘LGD input floors’); and
• introducing the LGD alternative methodology under the LGD modelling collateral method where there is limited data to model collateral recoveries (as set out in this section).

4.262 The PRA has considered whether to retain the wholesale LGD framework in light of the above proposed modelling constraints. The PRA notes that the wholesale LGD framework was predominantly targeted at exposures to institutions, financial corporates, and large corporates, as these are typically the most significant low default portfolios. The PRA considers that the proposed prohibition of LGD modelling for these exposure classes would materially address the risks that the wholesale LGD framework targets.

4.263 The PRA also considers that removal of the wholesale LGD framework would strike an appropriate balance between ensuring that LGDs are not too low where firms have insufficient data and ensuring that the overall approach is not excessively conservative. The proposed introduction of the above constraints would increase conservatism in the LGD framework prior to application of the PRA’s wholesale LGD framework, which would reduce the need for application of the wholesale LGD framework. The PRA also considers that to remove the wholesale LGD framework would prevent an excessively complex regime.

4.264 As a result of the above considerations, the PRA proposes to withdraw its wholesale LGD framework.

Other LGD proposals

4.265 For the calculation of dilution risk for purchased receivables, firms that do not decompose their EL estimates into PD and LGD currently set their PD estimates equal to their EL estimate and apply a 75% LGD. The PRA considers that the existing approach is inconsistent and results in understated RWAs. EL is defined as the product of PD and LGD, so to set PD equal to the EL estimate implies that LGD should be set at 100%. The PRA therefore proposes to set LGD equal to 100% if the decomposed approach is not used. This is in line with the Basel 3.1 standards. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.266 The PRA proposes to permit firms to reflect post-default additional drawings in LGD instead of EAD for non-retail and retail exposures, and to require that firms reflect pre-default additional drawings in EAD rather than LGD. Further details about these proposals are set out in the EAD estimation section. See Credit Risk: Internal Ratings Based Approach (CRR) Part and Appendix 13.

4.267 The PRA proposes to permit firms to assume zero recoveries for incomplete workouts as an alternative to applying the approach to modelling of incomplete workouts that is currently set out in its expectations. The PRA considers that the proposal would improve the accessibility of the IRB framework for firms that are unable to model incomplete workouts robustly. The PRA considers removing this unnecessary barrier to those firms would advance the PRA’s secondary objective of facilitating effective competition. See Appendix 13.

4.268 The PRA also proposes to make the following minor changes to LGD estimation:

• withdrawing an expectation that long-run average (LRA) LGD should reflect a representative mix of good and bad economic periods. Instead, the LRA LGD would reflect all observed defaults within the data sources, in line with the Basel 3.1 standards. See Credit Risk: Internal Ratings Based Approach (CRR) Part and Appendix 13;
• clarifying that firms need only calculate LRA LGD at portfolio level if they are calibrating LGD estimates at portfolio level. See Appendix 13;
• amending the hierarchy of approaches for calibrating downturn LGD so that firms would be able to base LGD estimates on estimated impact without having to first show that they do not have sufficient and relevant loss data to base LGD estimates on observed impact. See Appendix 13;
• withdrawing the option for firms to calibrate downturn LGD as LRA LGD plus 15%. The EBA Guidelines on PD estimation, LGD estimation and treatment of defaulted assets currently permit this, but the PRA has an existing expectation that this approach should not be used. See Appendix 13;
• clarifying that firms which base estimates of ‘best estimate of expected loss’ on LRA estimates, should adjust these to reflect current economic conditions where necessary, and that in certain circumstances, no adjustment is necessary. See Appendix 13; and
• revoking PRA Standards Instrument: Technical Standards (Economic Downturn) 2021 and transferring its contents to the PRA Rulebook in order to increase the coherence and accessibility of the framework. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

Question 29: Do you have any comments on the PRA’s proposals to LGD estimation?

PRA objectives analysis

4.269 The PRA considers that the proposed approach to LGD estimation set out in this section would advance the PRA’s primary objective of safety and soundness. The proposed reduction in the FIRB approach LGD value for unsecured corporate exposures for senior claims (excluding financial institutions) was calibrated by the BCBS based on international empirical data. Therefore, the PRA considers that the proposal would support the safety and soundness of firms. The other proposals are designed to improve the robustness of LGD estimates. For example, the PRA considers that the proposed introduction of the LGD alternative methodology under the LGD modelling collateral method should result in more robust LGD estimates, as firms would not be permitted to model collateral recoveries when they have limited data to do so effectively.

4.270 The proposed removal of the PRA’s LGD wholesale framework reflects the proposed introduction of three new constraints on LGD modelling. The PRA considers that these proposed measures, on balance, would result in an appropriate degree of conservatism.

4.271 The PRA considers the proposed LGD value of 100% to be appropriate for dilution risk for exposures to purchased receivables where the decomposed approach is not used, as it considers the existing 75% LGD value to be inconsistent and insufficiently prudent as outlined above.

4.272 The PRA considers that the different changes proposed in this section would potentially have different impacts on competition but it does not expect that the proposals to adversely impact the facilitation of effective competition overall. The PRA considers that the proposal to permit firms to assume zero recoveries for incomplete workouts as an alternative to modelling incomplete workouts should make the IRB framework more accessible to firms that are unable to model incomplete workouts robustly, which could help facilitate firms using the SA to obtain IRB permission. The PRA considers that the proposal to allow firms using the LGD modelling collateral method to disregard eligible collateral would make the treatment under the AIRB approach more consistent with the treatment for firms using the SA and the FIRB approach, which should improve competition between firms using different approaches.

‘Have regards’ analysis

4.273 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA considers that the proposed removal of the PRA’s wholesale LGD framework would be a proportionate measure to reduce duplication and complexity in the regulatory framework, given the new constraints on LGD modelling that the PRA proposes to introduce. The PRA considers its proposals for the treatment of ineligible or disregarded eligible collateral under the LGD modelling collateral method could result in additional operational costs and complexities for firms and potentially higher RWAs. The PRA considers; however, that any potential costs or RWA increases are justified as it considers firms should not reflect cash flows from ineligible or disregarded collateral in their LGD models. This is because including such cash flows could bias the models and effectively result in the collateral being recognised, leading to imprudent RWAs.

2. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA expects other jurisdictions to also implement the reduction of the FIRB approach LGD value for exposures to unsecured corporates for senior claims (excluding financial institutions). The PRA considers that the proposed introduction of an alternative methodology under the LGD modelling collateral method and the proposed removal of the PRA’s wholesale LGD framework would improve the competitiveness and relative standing of the UK, as the PRA would expect that the result of these proposals is that the PRA’s requirements for low default portfolios are more closely aligned with other jurisdictions. The proposed introduction of the 100% LGD value for dilution risk for purchased receivable exposures is aligned with the approach that the PRA would expect to be taken by other jurisdictions.

3. Relevant international standards (FSMA CRR rules):

• There PRA considers that the Basel 3.1 standards are open to interpretation regarding whether firms using the AIRB approach could disregard eligible collateral, and whether firms could reflect recoveries from ineligible collateral in LGD estimates. However, the PRA considers that its proposals are broadly aligned with the Basel 3.1 standards. The PRA considers the other proposals set out in this section are aligned with the Basel 3.1 standards.

4. Efficient and economic use of PRA resources (FSMA regulatory principles):

• The PRA considers that while additional supervisory resources could be required in the short- to medium-term to monitor the implementation of the proposed LGD alternative methodology under the LGD modelling collateral method and the proposed removal of the PRA’s wholesale LGD framework, the incremental increase in resource required would be low.

Exposure at default estimation

4.274 This section sets out the PRA’s proposals on EAD estimation under the IRB approach.

4.275 CFs are used to calculate an exposure value (referred to as a credit equivalent amount) for off-balance sheet items. The PRA proposes to align the CFs in the FIRB approach with its proposed revised CFs that would apply under the SA (see Chapter 3). This proposal includes analogous treatments for undrawn purchase commitments for revolving purchased receivables.

4.276 The PRA proposes to split CRR Article 166 into four separate articles (166A to 166D) in order to more clearly reflect the methods it proposes for calculating exposure values under the FIRB approach and the AIRB approach.

4.277 The PRA also proposes to introduce an ‘unrecognised exposure adjustment’ to RWAs in order to reflect risks falling outside the scope of IRB approaches for EAD. This proposal is discussed in section ‘Calculation of risk-weighted assets (RWA) and expected loss (EL)’ above.

Scope of EAD modelling under the AIRB approach

4.278 Firms using the AIRB approach currently provide own estimates of CFs or EAD for most off-balance sheet exposures. The Basel 3.1 standards restrict the scope of EAD modelling under the AIRB approach to revolving commitments only, although the PRA notes the Basel 3.1 standards could be subject to differing interpretations about which exposures are intended to be included in the definition of revolving commitments. The PRA considers a reasonable interpretation of the Basel 3.1 standards is that modelling would be restricted under the AIRB approach to revolving loan facilities, given that the BCBS introduced this restriction because of its concerns that a lack of data and firm modelling capability resulted in unwarranted RWA variability for these types of exposures.^{footnote [24]}

4.279 The PRA shares the concerns identified by the BCBS and therefore proposes to align with the Basel 3.1 standards by restricting the scope of EAD modelling to revolving commitments in the form of revolving loan facilities only, meaning that:

• for issued off-balance sheet items, non-revolving commitments, and all commitments to issue off-balance sheet items or purchase assets, firms would apply the CFs set out in Chapter 3 in order to calculate exposure value; and
• for on-balance sheet exposures, firms would calculate exposure value in line with the FIRB approach (subject to one exception discussed below).^{footnote [25]}

4.280 The PRA considers that the proposed restrictions on modelling off-balance sheet exposures are appropriate as it considers firms’ existing modelling approaches for these exposures is inconsistent, resulting in unwarranted variation in EAD estimates between firms. The PRA considers that it is challenging for firms to produce robust EAD models for issued off-balance sheet items given the lack of relevant data available. In contrast, the PRA considers that the benefits of continuing to allow modelling of revolving commitments in the form of revolving loan facilities outweigh the disadvantages as the SA is less able to capture the dynamics of flexible drawdowns.

4.281 For on-balance sheet exposures, the PRA considers that prohibiting modelling would result in greater consistency across the framework. While firms typically have greater capability to model on-balance sheet exposures than non-revolving commitments and issued off-balance sheet items, the PRA considers that permitting modelling for on-balance sheet exposures while prohibiting modelling for non-revolving commitments would result in unwarranted complexity in the framework. This is because firms would potentially be required to start modelling EAD only once a commitment is fully drawn down, or alternatively apportion accrued interest on a facility between the on-balance sheet part (which would be modelled) and the off-balance sheet part (which would not be modelled).

4.282 The PRA does; however, consider that it is necessary to make an exception for on-balance sheet exposures that are connected to a revolving facility (eg a credit card exposure that is partly drawn down or is at, or over, its limit). The PRA considers that it is impractical for firms to model only the off-balance sheet part of revolving exposures and it therefore proposes that:

• if an on-balance sheet exposure and a revolving commitment relate to the same facility, firms’ models should incorporate increases in EAD arising from the on-balance sheet exposure as well as the revolving commitment; and
• if a revolving exposure is at or over its limit, firms should continue to model EAD.

4.283 The PRA currently has an expectation that enables firms to model EAD directly in place of the CF estimates that are required by the CRR. The PRA proposes to continue the substance of its existing approach but to formalise the approach into PRA rules. For revolving exposures that are at or over limit, firms would be required to model EAD directly as the PRA considers that CFs are not a meaningful concept for on-balance sheet exposures. The PRA proposes to make a number of related changes to its rules and expectations relating to the modelling of EAD and CFs.

4.284 Overall, the PRA considers that its proposals relating to EAD would advance the safety and soundness of firms and enhance effective competition among firms by ensuring that a consistent and transparent approach is taken to EAD estimation.

Removal of EAD modelling under the slotting approach

4.285 Firms are currently able to model EAD for specialised lending exposures that are risk-weighted using the slotting approach if they have received permission from the PRA. The PRA considers that the Basel 3.1 standards are open to interpretation regarding how EAD should be determined for slotted exposures.

4.286 The PRA considers that it is typically challenging for firms to model EAD robustly for slotted exposures because of a lack of relevant data. The PRA proposes to limit EAD modelling to revolving commitments, as set out above, and the PRA considers that as a result of that proposal, EAD modelling would likely only be available for a limited number of slotted exposures.

4.287 The PRA therefore proposes to prohibit modelling of EAD for exposures subject to the slotting approach as it considers that this would reduce unnecessary complexity in the regulatory framework.

Requirement for a 12-month fixed-horizon approach for EAD modelling

4.288 Firms are currently able to define the modelling horizon for EAD models in one of two ways:

• the ‘cohort approach’ where facilities are observed on a given date and default could occur at any point in the 12 months following the observation point (resulting in an average time horizon of 6 months); and
• the ‘fixed-horizon approach’ where the observation point is fixed at 12 months prior to the point of default.

4.289 The Basel 3.1 standards introduce a 12-month fixed-horizon approach for all EAD models. The BCBS took this approach in order to reduce unwarranted variation in RWAs and to avoid potential underestimation that could arise from EAD estimates being based on too short a time horizon. Underestimation could occur if estimates are based on a period where the rate of drawdown has decreased due to accounts approaching default (eg as a result of account management measures applied by the firm), and these estimates are then applied to accounts further from default where the rate of drawdown is greater.

4.290 The PRA proposes to align with the Basel 3.1 standards and require firms to use a 12-month fixed-horizon approach for EAD modelling. The PRA considers that there are benefits in standardising the time horizon for EAD modelling domestically and internationally as this would reduce unwarranted risk weight variability.

4.291 The PRA notes that firms could incur operational costs in redeveloping their EAD models in cases where the ‘cohort approach’ is currently used. However, the PRA considers that these costs would be reduced due to the PRA’s proposed timelines for model submission (set out in section ‘Implementation timelines’). The proposed timelines would enable firms, in many cases, to make changes arising from these proposals at the same time as other changes needed to implement the IRB roadmap.

Treatment of additional drawings

4.292 Firms are currently required to estimate post-default additional drawings for non-retail exposures in their EAD estimates. While this is aligned with the Basel 3.1 standards, the PRA would expect some jurisdictions to require post-default additional drawings to be reflected in LGD.

4.293 The PRA considers that post-default additional drawings could legitimately be reflected in either EAD or LGD. The PRA considers that taking a different approach from other jurisdictions could result in unnecessary costs for firms that operate cross-border, and considers that it would not be proportionate to require firms to redevelop existing approaches.

4.294 The PRA therefore proposes that firms would be permitted to recognise post-default additional drawings in either EAD or LGD for non-retail exposures as well as for retail exposures.

4.295 The PRA also proposes to:

• withdraw an existing PRA expectation that additional drawings beyond a 12-month time horizon need not be incorporated in model estimates, in order to align with the Basel 3.1 standards; and
• clarify in its rules that pre-default additional drawings would be required to be reflected in EAD estimates as currently outlined in an existing expectation.

4.296 The PRA considers that its proposed approach to the recognition of additional drawings is prudent and proportionate. The proposal to permit firms to recognise post-default additional drawings in EAD or LGD would provide firms with flexibility and avoid unnecessary costs being placed on firms.

Removal of PRA’s wholesale EAD framework

4.297 The PRA currently would expect firms to apply its wholesale EAD framework for low-default portfolios. Under this framework, firms with limited data could either:

• rank-order the off-balance sheet product types (separately for lending and trade finance) according to their drawdown risk. The CF for a product with 20 or more default observations could then be applied to low-default products with a lower drawdown risk; or
• use 50% of the CF for committed credit lines to determine the CFs for uncommitted credit lines; or
• apply the FIRB approach parameters.

4.298 The PRA considers that as a result of the other proposals set out in this CP, its wholesale EAD framework would no longer be necessary. This is because the PRA’s existing framework is mainly targeted at exposures to institutions, financial corporates, and large corporates, which the PRA proposes would move to the FIRB approach. In addition, the PRA proposes to implement input floors for EAD estimates, which would help ensure a minimum level of prudence of EAD estimates. The PRA therefore proposes to withdraw its wholesale EAD framework.

Other EAD modelling proposals

4.299 The PRA also proposes the following changes to EAD modelling:

• to withdraw an expectation that estimates of long-run average EAD reflect a representative mix of good and bad economic periods. Instead, the PRA proposes that long-run average EAD estimates would reflect all observed defaults within the data sources, in-line with the Basel 3.1 standards. See Credit Risk: Internal Ratings Based Approach (CRR) Part;
• to introduce an expectation that CF estimates should not be biased by exposures that are close to limit (where firms estimate CFs directly), in-line with the Basel 3.1 standards. See Appendix 13;
• to set an expectation that firms should not cap realised EADs or CFs in reference data to the principal amount outstanding. The PRA also proposes to introduce an expectation that accrued interest, other due payments and limit excesses should be included in EAD reference data. See Appendix 13;
• to clarify that firms modelling EAD should only make adjustments to remove the effect of limit increases from EAD data where this can be done in a robust manner. See Appendix 13;
• to introduce a rule that firms should collect data on limits and balances used to derive CF and EAD estimates and to produce realised CFs and EADs. See Credit Risk: Internal Ratings Based Approach (CRR) Part; and
• to revoke PRA Standards Instrument: Technical Standards (Economic Downturn) 2021 and transfer its contents to the PRA rulebook in order to increase the coherence and accessibility of the framework (as mentioned in the LGD sub-section).

Question 30: Do you have any comments on the PRA’s proposals to EAD estimation?

PRA objectives analysis

4.300 The PRA considers that its proposals for EAD modelling advance its primary objective. The PRA considers the proposals minimise complexity and enhance the transparency of the EAD modelling framework for firms. The PRA considers that its proposals to limit the scope of EAD modelling and to standardise the time horizon over which EAD is modelled would help ensure firms take consistent approaches and reduce RWA variability across firms. This would contribute to safety and soundness.

4.301 The PRA has assessed whether the proposals on EAD modelling would facilitate effective competition. The PRA considers that its proposals would reduce RWA variation across firms using the IRB approach, which would facilitate effective competition through a more level playing field between firms. In addition, the PRA considers that its proposals to reduce the complexity and enhance the transparency of its EAD modelling standards would increase the accessibility of the IRB framework, which could be beneficial for firms currently using the SA that are considering applying for IRB permissions.

‘Have regards’ analysis

4.302 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA has taken into account the practicalities and difficulties for firms in modelling EAD when arriving at its proposals to prohibit EAD modelling of on-balance sheet exposures, non-revolving commitments, and specialised lending exposures subject to the slotting approach. It considers that its proposals are proportionate.
• The PRA considers its proposal for the recognition of post-default additional drawings for non-retail exposures tis proportionate, as the proposal allows firms a choice about whether to reflect these in LGD or EAD estimates.
• The PRA considers that its proposal to remove its wholesale EAD framework is proportionate because when combined with other PRA proposals, for example moving exposures to institutions, financial corporates, and large corporates to the FIRB approach, retaining the wholesale EAD framework would add unnecessary complexity for limited prudential benefit.
• As a result of the above, the PRA considers that its overall package on EAD is proportionate.

2. Efficient and economic use of the PRA’s resources (FSMA regulatory principles):

• The PRA considers that its proposals to restrict the scope of EAD modelling would reduce the amount of time and resources it spends reviewing models that are unlikely to be sufficiently robust. This would enable the PRA to focus its resources more efficiently.

3. Growth (HMT recommendation letters), finance for the real economy (FSMA CRR rules), and sustainable growth (FSMA regulatory principles):

• The PRA considers that the proposals set out in this section would improve the robustness of RWAs and would help to ensure that they more accurately reflect risk that firms take. This would help to ensure that firms are able to continue lending throughout the economic cycle and support sustainable growth.

4. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA considers that most of proposals set out in this section support the competitiveness of the UK. The PRA proposes to broadly align with the Basel 3.1 standards on EAD modelling restrictions and to remove the PRA’s existing wholesale EAD framework, which is super-equivalent to the Basel 3.1 standards. The PRA would expect other jurisdictions to take a similar approach to implementing the Basel 3.1 standards, supporting the UK’s competitiveness and relative standing as the individual country regimes would become more closely aligned and there would be a more level playing field for firms that operate across them.

5. Relevant international standards (FSMA CRR rules):

• The PRA considers that the proposals set out in this section align with the Basel 3.1 standards. The proposed removal of the PRA’s wholesale EAD framework, which is not part of the Basel 3.1 standards, would also increase alignment between the PRA’s overall framework for EAD and the Basel 3.1 standards.

Maturity

4.303 This section sets out the PRA’s proposals for calculating the maturity parameter which is used in the IRB risk weight formula for exposures to corporates and institutions.

Calculation of maturity

4.304 The CRR currently sets out two methods for firms that apply the FIRB approach can use to calculate maturity:

• a fixed parameter approach, where maturity is set at 0.5 years for certain short-term transactions and at 2.5 years for all other exposures; and
• an effective maturity approach, where firms calculate effective maturity according to prescribed formulae. If a firm is unable to calculate effective maturity under this approach, the contractual maturity is instead applied. A one-year floor applies to the maturity calculated for most transactions, but certain transactions are subject to a reduced maturity floor.

4.305 The PRA currently specifies within IRB permissions that firms using the FIRB approach must calculate effective maturity rather than apply fixed parameters. This is because the PRA considers that calculation of effective maturity is a more risk-sensitive approach, which better reflects the economic substance of the exposures, and thus enhances the safety and soundness of firms. Furthermore, using effective maturity facilitates effective competition because firms using the AIRB approach are also required to apply the effective maturity approach.

4.306 The PRA proposes to maintain the substance of its existing approach and that firms using the FIRB approach would continue to be required to apply the effective maturity approach. The PRA proposes to include this provision in its rules as it considers this would be more appropriate than applying the requirement on a firm-by-firm basis as is currently the case.

4.307 The PRA considers that the proposed approach is in line with the Basel 3.1 standards as these include a discretion for national supervisors to require firms using the FIRB approach to calculate effective maturity for all exposures.

4.308 Similarly, to improve risk-sensitivity, the PRA proposes to remove the option currently set out in the CRR that allows firms that are otherwise calculating maturity to instead apply fixed maturity values for exposures to small UK corporates.

Other proposals

4.309 The PRA also proposes to make the following additional minor changes to its requirements:

a. to align with the Basel 3.1 standards on the reduced maturity floors that apply to transactions in scope of master netting agreements by:

• restricting the scope of the reduced floors to those transactions where the documentation requires daily re-margining or revaluation and includes provisions allowing for prompt liquidation or set-off in the event of default or failure to re-margin. See Credit Risk: Internal Ratings Based Approach (CRR) Part;
• expanding the scope of the reduced floors to also apply a floor of 20 days for secured lending subject to a master netting agreement, and a floor of either 10 or 20 days for master netting agreements including more than one transaction type. See Credit Risk: Internal Ratings Based Approach (CRR) Part; and
  • specifying that in all such cases, the notional amount of each transaction would be used for weighting the maturity. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

b. to clarify the definition of trade finance transactions that are in scope of a one-day maturity floor. See Credit Risk: Internal Ratings Based Approach (CRR) Part;

c. for revolving exposures, to clarify that effective maturity would be determined by using the maximum contractual termination date of the facility and that firms should not use the repayment date of the current drawing to estimate the effective maturity. This proposal is in line with the Basel 3.1 standards. See Credit Risk: Internal Ratings Based Approach (CRR) Part;

d. for purchased receivables, to align with the Basel 3.1 standards and require that the effective maturity would be a minimum of one year instead of the existing 90-day minimum. See Credit Risk: Internal Ratings Based Approach (CRR) Part;

e. for dilution risk of purchased receivables, clarify that effective maturity would be set at one year only if a firm could demonstrate that the dilution risk is appropriately monitored and could be resolved within one year. Otherwise, effective maturity should reflect the period over which dilution risk could be resolved, with a cap of five years. See Credit Risk: Internal Ratings Based Approach (CRR) Part; and

f. make consequential changes to those elements of the maturity calculation that reference the CVA framework, to reflect the new CVA rules proposed in this CP (see Chapter 7 – Credit valuation adjustment, and Credit Risk: Internal Ratings Based Approach (CRR) Part.

Question 31: Do you have any comments on the PRA’s proposals for maturity?

PRA objectives analysis

4.310 The PRA considers that the proposed approach to maturity set out in this section would advance the PRA’s primary objective of safety and soundness. The proposals would result in more risk-sensitive and more accurate estimations of effective maturity that reflect the economic substance of the exposures.

4.311 The PRA considers that the proposals set out in this section would facilitate effective competition. The proposals would retain the substance of the existing PRA approach whereby firms using both FIRB and AIRB approaches are expected to calculate effective maturity in a consistent manner, thus promoting a level playing field.

‘Have regards’ analysis

4.312 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Proportionality (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The proposal for calculating effective maturity under the FIRB approach would be consistent with the PRA’s existing approach and would be less burdensome for firms compared to implementing a new policy.

2. Relevant international standards (FSMA CRR rules):

• The PRA considers that its proposals relating to maturity would align with the Basel 3.1 standards, including through those national discretions in these standards that the PRA proposes to exercise.

3. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA considers that allowing firms using the FIRB approach to continue calculating effective maturity would be positive for competitiveness and relative standing. The PRA considers that UK firms that use the FIRB approach would apply a more risk-sensitive approach and that the RWAs of these firms would better reflect the risk of exposures compared with jurisdictions that require use of a fixed parameter approach.

Specialised lending

4.313 This section sets out the PRA’s proposals relating to the IRB approach for specialised lending exposures and the use of the slotting approach.

Specialised lending category definitions

4.314 Firms using the slotting approach currently allocate specialised lending exposures to one of the following categories:

• ‘project finance’;
• ‘object finance’;
• ‘commodities finance’; and
• ‘income producing real estate’ (IPRE).

4.315 The Basel 3.1 standards contain a fifth category: ‘high volatility commercial real estate (HVCRE)’. This category attracts higher risk weights under the slotting approach. The PRA proposes to introduce this category into its regime to introduce greater risk-sensitivity and to align with the Basel 3.1 standards.

4.316 The PRA proposes that the HVCRE category would encompass specialised lending secured on real estate that meets one or more of the following criteria:

a. the real estate is bought for speculative purposes;

b. a change of planning use is sought for the real estate; or

c. loans financing the land, acquisition, development, and construction of real estate where the source of repayment at origination of the exposure is either:

• the future uncertain sale of the real estate; or
• cash flows whose source of repayment is substantially uncertain, unless the borrower has sufficient equity to absorb most losses through the asset development and construction phase in a severe but plausible scenario.

4.317 The PRA proposes that this HVCRE definition would apply to the classification of all specialised lending exposures regardless of geographic location.

4.318 The PRA also proposes to introduce category definitions for project finance, object finance, commodities finance, and IPRE that would be broadly in line with the Basel 3.1 standards.

4.319 The PRA considers that HVCRE exposures typically exhibit higher loss rate volatility compared to other types of specialised lending and that HVCRE exposures should therefore receive higher risk weights than IPRE exposures for a given slotting assignment. While the proposal could increase RWAs for some exposures, the PRA would not expect the increase to be material in aggregate as the PRA assesses that HVCRE is unlikely to be a significant exposure class for most firms.

4.320 The PRA considers that the proposed introduction of HVCRE would enhance the risk-sensitivity of the slotting approach and would help ensure that a greater degree of risk-sensitivity is retained relative to the proposed specialised lending treatment in the SA. The PRA considers this would result in RWAs for specialised lending being more reflective of the risk of a firm’s exposures. The PRA considers that the introduction of the HVCRE category would therefore promote the safety and soundness of firms.

Introduction of additional risk-sensitivity in the slotting approach

4.321 The PRA proposes to broadly align the risk-weighting treatment in the slotting approach with the Basel 3.1 standards.

4.322 Under the PRA’s proposed slotting approach, which would align with the existing approach in the CRR, firms would assign specialised lending exposures to one of four categories: ‘strong’, ‘good’, ‘satisfactory’, and ‘weak’. The assignment would be based on a set of defined factors and sub-factors (including financial strength, political and legal environment, and specific transaction characteristics) with the specific factors tailored to each specialised lending sub-class. Application of the factors would result in assignment of the lowest risk exposures to the strong category and assignment of the highest risk exposures to the weak category. Defaulted exposures would be assigned to the ‘default’ category. The PRA proposes to move existing material relating to the slotting factors and sub-factors from its expectations into PRA rules as requirements on firms.

4.323 Under the CRR, firms are permitted to apply a preferential slotting risk weight to exposures in the strong and good categories (a 50% risk weight instead of a 70% risk weight in the strong category, and a 70% risk weight instead of a 90% risk weight in the good category) if the remaining maturity of the exposure is less than 2.5 years. The CRR preferential treatment is more restrictive than the Basel 3.1 standards, as the latter gives national supervisors wider discretion to permit firms to apply the preferential risk weights in the strong and good categories if the exposures have ‘substantially stronger’ underwriting and other risk characteristics are met, even if the less than 2.5 years remaining maturity criteria is not met. The PRA considers that the existing approach to determining remaining maturity creates opportunities for firms to artificially structure loans such that the remaining maturity is less than 2.5 years so they could benefit from the lower risk weight.

4.324 The PRA therefore proposes to amend the circumstances in which preferential risk weights would be available for exposures assigned to the strong and good slotting categories, in order to introduce greater risk-sensitivity, address concerns with firms’ current application of the remaining maturity criteria, and to align more closely with the scope of the Basel 3.1 standards. The proposed risk weights are set out in Table 4 below:

Table 4: Proposed slotting risk weights

4.325 For project finance, object finance, commodities finance, and IPRE, the PRA proposes to retain existing EL values but to align the preferential EL treatment with the preferential risk weight treatment set out above. For HVCRE, the PRA proposes to introduce ELs without a preferential treatment, in line with the Basel 3.1 standards. The proposed EL values are set out in Table 5 below:

Table 5: Proposed slotting expected loss values

4.326 The PRA proposes that firms would only assign the preferential risk weights in the strong and good categories if the exposure:

a. has less than 2.5 years remaining until maturity and the firm reasonably considers that the obligor could refinance the exposure in a severe but plausible stress in the refinancing market (see para 4.326); or

b. is an IPRE exposure and has features which are ‘substantially stronger’ than the criteria specified for the strong category (see para 4.327).

4.327 For (a), the PRA proposes to restrict application of the preferential risk weight to situations where the firm reasonably considers that the obligor would be able to be refinance the exposure in a severe but plausible stress in the refinancing market. The PRA considers this restriction is desirable in order to prevent exposures that would likely have a longer maturity in practice from being assigned a preferential risk weight.

4.328 For (b), the PRA proposes that the following criteria would all need to be met for the PRA to consider the exposure to be ‘substantially stronger’:

• the transaction is allocated to the strong slotting category for each slotting factor;
• the leverage of the obligor is substantially below the market norm for a similarly structured transaction in this sector, region, and of this property location and quality; and
• a substantial amount of the transaction’s cash flows comes from investment grade (or equivalent) counterparties, with a minimum of 100% of the interest covered by income from investment grade or equivalent tenants.

4.329 The PRA considers that these proposals would result in a more risk-sensitive slotting approach. The PRA considers that the proposed introduction of the ‘substantially stronger’ category could encourage firms to increase lower risk lending. The PRA considers that the proposed changes are aligned with the Basel 3.1 standards and contain safeguards to prevent regulatory arbitrage.

Use of the slotting approach for IPRE and HVCRE exposures

4.330 The PRA considers, as currently set out in SS11/13, that it is difficult for firms to build effective and compliant IRB rating systems for IPRE exposures. As a result, most firms use the slotting approach for IPRE exposures.

4.331 Given that the PRA has not observed any strong evidence to suggest that firms can build effective rating systems for IPRE exposures, the PRA proposes that the option for firms to apply the AIRB or FIRB approaches for IPRE exposures would no longer be permitted. Firms currently using the IRB approach for IPRE exposures would instead be required to risk weight the exposures using the slotting approach.^{footnote [26]} The PRA proposes that this restriction would also apply to exposures that would be newly allocated to the HVCRE category.

4.332 The PRA considers that these proposals would contribute to improving the robustness of RWAs in capturing risk, given the persistent modelling challenges observed for these exposures, but would not result in a significant change in RWAs, given that these exposures are typically risk-weighted under the slotting approach already.

The general corporates and specialised lending boundary

4.333 The PRA would not propose to introduce further modelling restrictions for other categories of specialised lending. Therefore, the AIRB approach, the FIRB approach, and the slotting approach would continue to be available for the project finance, object finance, and commodities finance categories. This contrasts with the proposed approach for other corporate exposures whereby the AIRB approach for exposures to institutions, financial corporates, and large corporates would be withdrawn and the exposures would move to the FIRB approach.

4.334 The PRA recognises that its proposals could potentially give rise to regulatory arbitrage whereby firms could choose to allocate certain exposures to either the other general corporates exposure sub-class or the specialised lending exposure sub-class in order to optimise RWAs.

4.335 The PRA would therefore intend to monitor firms’ allocation of large corporate exposures between the other general corporates exposure sub-class and the specialised lending exposure sub-class. The PRA’s monitoring would assess whether firms apply the proposed definitions correctly and whether exposures are allocated to the specialised lending exposure sub-class only if the specialised lending criteria are fully met.

4.336 The PRA proposes to prohibit firms from reflecting credit protection which is recognised under the risk weight substitution method in their assignment of exposures to slotting categories in order to prevent double counting (see Chapter 5 for further information regarding the PRA’s proposal to permit use of the risk weight substitution method for exposures subject to the slotting approach).

Other specialised lending proposals

4.337 Firms are currently only able to use the slotting approach where they cannot build a compliant specialised lending PD model. The PRA proposes that, provided they could comply with relevant requirements, firms would be able to adopt the slotting approach for exposures in the project finance, object finance, and commodities finance categories, regardless of whether they are able to build a compliant PD model. The proposed requirements for slotting IPRE and HVCRE are discussed in the ‘Specialised lending category definitions’ section above. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

4.338 The PRA proposes to set expectations regarding the broad correspondence of external credit ratings to each slotting category in order to align with the Basel 3.1 standards. These would complement but not replace the existing slotting criteria, which the PRA proposes that firms would continue to apply. The expected loss based ‘external credit assessment institution’ (ECAI) ratings that the PRA proposes would broadly correspond to each slotting category are set out in the Table 6 below. See Credit Risk: Internal Ratings Based Approach (CRR) Part.

Table 6: Correspondence of ECAI ratings to slotting categories

Question 32: Do you have any comments on the PRA’s proposals for specialised lending?

PRA objectives analysis

4.339 The PRA considers that the proposals set out in this section would advance the PRA’s primary objective of safety and soundness. The PRA considers that the proposed introduction of a HVCRE specialised lending category would promote the safety and soundness of firms as HVCRE exposures typically exhibit higher loss rate volatility than IPRE exposures, and so the PRA considers that these exposures should be assigned higher risk weights for a given slotting category. The PRA considers that the proposed introduction of the HVCRE category and the proposed introduction of preferential risk weights for certain specialised lending exposures would enhance the risk-sensitivity of slotting, which would result in RWAs that better correspond to the underlying risk. The PRA considers that the proposed requirement to apply the slotting approach to IPRE and HVCRE exposures would also promote safety and soundness of firms by improving the robustness of RWAs, given the PRA considers it challenging for firms to model these exposures robustly.

4.340 The PRA considers that the proposals set out in this section would have a broadly neutral impact on its secondary objective of facilitating effective competition. The proposals for HVCRE would mainly impact firms that have riskier IPRE exposures. For non-real estate specialised lending exposures, the PRA would expect the proposals would have the greatest impact on the business lines of large firms in which they are less likely to directly compete with smaller firms, so the PRA would not expect the proposals set out in this section to materially impact effective competition between large and small firms.

‘Have regards’ analysis

4.341 In developing these proposals, the PRA has had regard to the FSMA regulatory principles, the aspects of the Government’s economic policy set out in the HMT recommendation letter from 2021 and the supplementary recommendation letter sent April 2022. Where the proposed new rules are CRR rules (as defined in section 144A of FSMA), the PRA has also taken into consideration the matters to which it is required to have regard when proposing changes to CRR rules. The following factors, to which the PRA is required to have regard, were significant in the PRA’s analysis of the proposals:

1. Proportionality of costs and benefits (FSMA regulatory principles and Legislative and Regulatory Reform Act 2006):

• The PRA considers that the proposals set out in this section are a proportionate response to the identified deficiencies in IRB modelling of specialised lending exposures. The proposed introduction of new exposure class and exposure sub-class definitions could create some initial costs for firms when amending their internal systems but the PRA considers they would not generate increased costs over time.

2. Competitiveness (HMT recommendation letters) and relative standing of the UK as a place to operate (FSMA CRR rules):

• The PRA considers that the proposals could have some implications for UK competitiveness. The PRA considers that the proposed mandating of slotting for IPRE and HVCRE exposures would not materially change the UK’s relative standing as most UK firms already apply the slotting approach for IPRE. The proposal to introduce an HVCRE category aligns with the Basel 3.1 standards, but the PRA recognises that some international jurisdictions may not introduce this category and relevant UK firms could be required to apply a higher risk weight to these exposures in the UK, all else equal, than in other jurisdictions. However, the PRA considers that introducing the HVCRE category and associated slotting risk weights is prudentially justified as the PRA considers HVCRE to be the most volatile type of specialised lending. The PRA would expect that the proposed changes to the criteria for applying preferential risk weights would have a mixed impact on the UK’s relative standing. This is because the PRA proposes a more restrictive approach than under the CRR for applying lower risk weights based on the ‘less than 2.5 years remaining maturity criteria’, but also proposes to allow lower risk weights for certain IPRE exposures that meet the ‘substantially stronger’ criteria.

3. Relevant international standards (FSMA CRR rules):

• The PRA considers that the proposed introduction of a new HVCRE category and the proposed approach to risk-weighting exposures under the slotting approach are aligned with the Basel 3.1 standards. The PRA considers that the proposal to require application of the slotting approach to IPRE and HVCRE exposures is more restrictive than the Basel 3.1 standards. However, the PRA considers this proposal to be justified for safety and soundness considerations given the challenges in modelling these exposures.

4. Efficient and economic use of PRA resources (FSMA regulatory principles):

• The PRA considers that the proposals would result in an efficient and economic use of PRA resources. This is particularly the case for the proposal to require firms to use the slotting approach for IPRE and HVCRE exposures as the PRA would not need to review models that it considers are unlikely to be robust or materially compliant with its rules.