Credit Risk Internal Ratings Based Approach Supervisory Statement

This is near-final material effective from 1 January 2026 to accompany PS9/24.
Please see:
www.bankofengland.co.uk/prudential-regulation/publication/2024/september/implementation-of-the-basel-3-1-standards-near-final-policy-statement-part-2.
Credit risk internal ratings based approach
6.
02
y2
u ar
an
1J
m
f ro
ve
Supervisory statement | SS4/24

cti
ffe
September 2024
.E
t2
ar
lp
na
-fi
ar
Ne
This is near-final material effective from 1 January 2026 to accompany PS9/24. Please see:
Bank of England | Prudential Regulation Authority
Credit risk: Internal ratings based approach
Supervisory statement |SS4/24
September 2024
6.
02
y2
u ar
an
1J
m
f ro
ve
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
© Bank of England 2024
Prudential Regulation Authority | 20 Moorgate | London EC2R 6DA

Bank of England | Prudential Regulation Authority Page 1
Contents
Contents 1
1: Introduction 5
Definitions 5
2: Permission to use the IRB approach 8

Criteria for granting approval of permission to use the IRB approach 8
6.
Assessment of non-compliance and remediation plans 8
02
Post-model adjustments 8
y2
Fees 10
ar
Application of requirements to UK groups applying the IRB approach on a unified basis 10
u
an
Annual attestations 1J 11
Submission of applications and notifications relating to the IRB approach 11
m
Overseas models approach 11

f ro
3: Partial use and reversion to less sophisticated approaches 12

ve
cti
Policy for identifying exposures subject to partial use 12

ffe
Roll-out following significant acquisitions 12

.E
Permanent partial use of the Standardised Approach to a roll-out class 12

t2
Reversion to less sophisticated approaches 13

ar
lp
4: Use and experience tests 14

na
Use test 14
-fi
Prior experience of using the IRB approach 14

ar
Ne
5: Qualifying revolving retail exposures 16
6: High level expectations 17

High level expectations for estimation 17
Quality of data 18
Ratings systems: policies 18
Collection of data 19
Human judgement in estimation of risk parameters 19
Documentation 20
7: Rating systems 21
Principles for specifying the range of application of a rating system 21

Multi-country mid-market corporate PD models 21
Retirement interest-only (RIO) mortgages 22
8: Data representativeness 23
Overall assessment 23
Representativeness of data for model development 23
Representativeness of data for calibration of risk parameters 25
PD – use of external data for residential mortgages 27
9: Model deficiencies and margin of conservatism 28
6.
02
Identification of deficiencies 28
y2
Margin of conservatism 29
u ar
10: PD – model development 33
Assignment of obligors and exposures
an 33
1J
Data for model development 33
m
ro
Risk drivers and rating criteria 34

f
Rating philosophy 34
ve
cti
Homogeneity of obligor grades or pools 37

ffe
Use of external rating agency grades 37

.E
Recognition of third-party support 38

t2
Parameter Substitution Method 38

ar
lp
11: PD – calibration 39
na
Data requirements for the calculation of observed default rates 39

-fi
Calculation of one-year default rates 40

ar
Ne
Calculation of the observed average default rate 41

Long-run average default rate 41
Calibration to the long-run average default rate 43
Calibration of residential mortgage portfolios 45
Retail exposures: obligor level definition of default 48
Retail exposures: facility level definition of default 48
Low default portfolios 48
12: LGD – General expectations and model development 50

General expectations – LGD estimation methodologies 50
General expectations – use of SA and FIRB approach parameters for LGD estimation 51
General expectations – use of LGD Modelling Collateral Method 51
Data requirements for LGD estimation 52

Recoveries from collateral 54
Risk drivers 57
Eligibility of collateral 58
Inclusion of collateral in LGD estimation when applying the LGD Modelling
Collateral Method 58
Unfunded credit protection 60
Funded credit protection securing unfunded credit protection obligations 61
Homogeneity of facility grades or pools 62
Treatment of cures 62
6.
02
Incomplete workouts 62
y2
Unsecured LGDs where the obligor’s assets are substantially used as collateral 62
ar
LGD – use of external data for residential mortgages 63
u
LGD estimates for exposures to corporates 64
an
1J
13: LGD – calibration (general) 66
m
Definition of economic loss and realised LGD 66

f ro
Treatment of fees, interest, and additional drawings after default 68

ve
Discount rate 69
cti
Direct and indirect costs 70

ffe
Low LGDs 70
.E
t2
14: LGD – calibration (long-run average) 72

ar
Calculation of long-run average LGD 72

lp
na
Treatment of incomplete recovery processes 73

-fi
Treatment of cases with no loss or positive outcome 75

ar
Calibration to the long-run average LGD 75

Ne
15: LGD – calibration (downturn) 77

General requirements on downturn LGD estimation 77
Adjustments to downturn LGDs where risk drivers are sensitive to the economic cycle 78
Final downturn LGD estimates 78
Calculation of reference values 79
Downturn LGD estimation for a considered downturn period 80
Downturn LGD estimation based on observed impact 81
Downturn LGD estimation based on estimated impact 82
LGD – UK residential mortgage property sales reference point 84
Probability of possession given default (PPGD) estimates for UK residential mortgage

exposures 85
16: LGD – In-default estimation 87

Estimation methodologies for best estimate of expected loss (BEEL) and LGD in-default 87
Reference dates 87
Data requirements for BEEL and LGD in-default estimation 88
Model development in the estimation of BEEL and LGD in-default 89
Calculation of realised LGD and long-run average LGD for defaulted exposures 89
Consideration of MoC in BEEL estimation 90
6.
Current economic circumstances 90
02
Relation of BEEL to specific credit risk adjustments 91
y2
Downturn LGD estimation for defaulted exposures 91
u ar
Specific requirements for LGD in-default estimation 92
an
1J
17: EAD – model development and calibration 93
m
Methodology for estimating EAD or conversion factors 93

ro
General expectations for estimating EAD or CF 93

f
ve
Adjustments to downturn EADs or CFs where risk drivers are sensitive to the
cti
economic cycle 95
ffe
Distortions to CF estimates caused by low undrawn limits 95

.E
Identification of exposures for which an EAD or CF must be estimated 96

t2
EAD or CF reference data 96

ar
Netting 96
lp
Underwriting commitments 97
na
-fi
18: The slotting approach 98

ar
Ne
Mapping criteria 98
Definition of high volatility commercial real estate (HVCRE) exposures 98
19: Application of risk parameters 99

New information 99
Conservatism in the application of risk parameters 99
Human judgement in the application of risk parameters 100
20: Stress tests used in the assessment of capital adequacy 102
21: Review of estimates (validation) 103

1: Introduction
1.1 The purpose of this supervisory statement (SS) is to provide clarification to firms of the
Prudential Regulation Authority’s (PRA) expectations in respect of the application of the IRB
approach in the calculation of credit risk risk-weighted assets. This SS applies to PRA-
authorised banks, building societies, PRA-designated investment firms, and PRA-approved
or PRA-designated financial or mixed financial holding companies (collectively ‘firms’).
1.2 The Prudential Regulation Authority (PRA) grants permission to use the IRB approach
under Rules 1.1 and 1.2, and Articles 143(1) and 143(2A) of the Credit Risk: Internal Ratings
6.
Based Approach (CRR) Part of the PRA Rulebook where firms are materially compliant with
02
the requirements of the Credit Risk: Internal Ratings Based Approach (CRR) Part. The
y2
purpose of this SS is to provide explanation, where appropriate, of the PRA’s expectations
ar
when assessing whether firms meet those requirements, including in respect of the
u
an
conservatism applied. 1J
1.3 Responsibility for ensuring that internal models are appropriately conservative and are
m
ro
compliant with the CRR and PRA rules rests with firms themselves. The PRA’s approach to
f
banking supervision1 states that ‘if firms use internal models in calculating their regulatory
ve
capital requirements, we expect the models to be appropriately conservative’.

cti
ffe
Definitions
.E
t2
1.4 Unless otherwise specified, terms used in this SS have the same meaning as in the
ar
relevant part of either the PRA Rulebook or the CRR. In addition, for the purposes of this SS,
lp
the following definitions apply:

na
-fi
(a) risk parameters: one or all of the following – probability of default (PD), loss given
ar
Ne
default (LGD), best estimate of expected loss (BEEL), LGD in-default, exposure at
default (EAD), and conversion factor (CF);
(b) reference data set (RDS): all the data sets used for the purpose of estimation of risk
parameters, including the data sets relevant for model development, as well as the
data sets used for calibration of a risk parameter;
(c) PD model: all data and methods used as part of a rating system (as defined in Rule
1.3 of the Credit Risk: Internal Ratings Based Approach (CRR) Part) which relate to
the differentiation and quantification of own estimates of PD and are used to assess
the default risk for each obligor or exposure covered by that model;
1 www.bankofengland.co.uk/prudential-regulation/publication/pras-approach-to-supervision-of-the-
banking-and-insurance-sectors.
(d) ranking method of a PD model: the method, forming part of a PD model, used to rank
order the obligors or exposures with respect to the risk of a default;
(e) scoring method of a PD model: a ranking method of a PD model that assigns ordinal
values (’scores’) to rank obligors or exposures;
(f) LGD model: all data and methods used as part of a rating system (as defined in Rule
1.3 of the Credit Risk: Internal Ratings Based Approach (CRR) Part) which relate to
the differentiation and quantification of own estimates of LGD, LGD in-default, and
BEEL, and which are used to assess the amount of loss in the case of default for each
facility covered by that model;
(g) BEEL: best estimate of expected loss for defaulted exposures as referred to in Article
181(1)(h)(ii) of the Credit Risk: Internal Ratings Based Approach (CRR) Part;
6.
(h) LGD in-default: loss given default for defaulted exposures as referred to in Article
02
181(1)(h) of the Credit Risk: Internal Ratings Based Approach (CRR) Part;
y2
(i) EAD or CF model: all data and methods used as part of a rating system, as defined in
ar
Rule 1.3 of the Credit Risk: Internal Ratings Based Approach (CRR) Part, which relate
u
an
to the differentiation and quantification of own estimates of EAD or CF and are used to
1J
assess the exposure amount in the case of default for each facility covered by that
m
model;
ro
(j) scope of application of a PD, LGD, EAD, or CF model: the type of exposures as
f
defined in Rule 1.3 of the Credit Risk: Internal Ratings Based Approach (CRR) Part
ve
covered by a PD model, an LGD model, or an EAD or CF model;

cti
(k) estimation of risk parameters: the full modelling process related to risk parameters
ffe
including the selection and preparation of data, model development, and calibration;
.E
(l) intermediate parameters: parameters estimated as part of a model which are then
t2
subsequently used to estimate risk parameters in that model;

ar
lp
(m) model development: the part of the process of the estimation of risk parameters that
na
leads to an appropriate risk differentiation by specifying relevant risk drivers, building

-fi
statistical or mechanical methods to assign exposures to obligor or facility grades or

ar
pools, and estimating intermediate parameters of the model, where relevant;

Ne
(n) PD calibration sample: the data set on which the ranking or pooling method is applied
in order to perform the calibration;
(o) calibration segment: a uniquely identified subset of the scope of application of the PD
or LGD model which is jointly calibrated;
(p) PD calibration: the part of the process of the estimation of risk parameters that leads
to appropriate risk quantification by ensuring that, when the PD ranking or pooling
method is applied to a calibration sample, the resulting PD estimates correspond to
the long-run average default rate, at the level relevant for the applied method;
(q) LGD calibration: the part of the process of the estimation of risk parameters that leads
to appropriate risk quantification by ensuring that the LGD estimates correspond to the
long-run average LGD, or to the downturn LGD estimate where this is more
conservative, at the level relevant for the applied method;
(r) margin of conservatism (MoC): the MoC related to the expected range of estimation
errors required by Articles 179(1)(f) and 180(1)(e) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part;
(s) application of risk parameters: the assignment of risk parameters estimated in
accordance with the PD or LGD model to the current exposures, performed either
automatically with the use of a relevant IT system or manually by qualified personnel
of a firm; and
(t) application portfolio: the actual portfolio of exposures within the scope of application of
the PD or LGD model at the time of the estimation of a risk parameter.
6.
02
y2
u ar
an
1J
m
f ro
ve
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
2: Permission to use the IRB approach
Criteria for granting approval of permission to use the IRB

approach
2.1 As set out in Article 143(1) of the Credit Risk: Internal Ratings Based Approach (CRR)
Part, firms applying for permission to use the IRB approach are required to demonstrate that
they materially comply with the requirements of the Credit Risk: Internal Ratings Based
Approach (CRR) Part. The PRA will therefore only grant approval to use the IRB approach
6.
where this condition is met.
02
y2
Assessment of non-compliance and remediation plans
u ar
an
2.2 For the purpose of assessing whether non-compliance is immaterial in accordance with
1J
Articles 143(1)(a), 143(2B), 143(3)(b), 146(1)(b) and 149(2A) of the Credit Risk: Internal
Ratings Based Approach (CRR) Part, the PRA expects that non-compliance should only be
m
ro
considered immaterial if it results in a minimal impact on the quantitative and qualitative

f
aspects of the firm’s IRB approach.

ve
cti
2.3 In accordance with Article 146(1) of the Credit Risk: Internal Ratings Based Approach
ffe
(CRR) Part, firms are required to submit a plan for addressing non-compliance in a timely
.E
way to the PRA if they are materially non-compliant with the requirements of the Credit Risk:
t2
Internal Ratings Based Approach (CRR) Part. The PRA also expects firms to present a plan
ar
to the PRA for addressing non-compliance in a timely way such that the effect of non-
lp
compliance would become immaterial if they are materially non-compliant with any relevant
na
PRA SSs relating to IRB (including, but not limited to, this SS).
-fi
ar
2.4 The PRA expects that firms should agree the content of remediation plans presented
Ne
under Articles 143B(5) and 146(1)(b) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part and paragraph 2.3 with the PRA and that firms should realise such remediation
plans within a time period agreed with the PRA.
Post-model adjustments
2.5 The PRA expects firms to develop a framework to assess post-model adjustments
(PMAs) in accordance with Article 146(3) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part. This framework should meet the following criteria:
(a) the framework should be applied at a portfolio level. For this purpose, a ‘portfolio’
is defined as the set of exposures covered by the IRB model that the adjustment
is being made for. If adjustments are being made to more than one model (eg
both PD and LGD) that cover overlapping assets (eg a global LGD model and
regional PD models), then a portfolio should be defined as the subset of assets
covered by the same models (eg in the example above, the assets covered by
each regional PD model would each be classified as a single portfolio);
(b) irrespective of which model component the adjustment is for (eg PD, LGD, or
EAD or CF), the risk-weighted assets (RWA) and expected loss (EL) adjustments
should be made as a portfolio level add-on to the requirements produced by the
approved models (ie the underlying models should not be recalibrated or changed
in order to give a desired capital requirements outcome);
(c) firms’ existing PD, LGD, and EAD or CF models should remain in place until
6.
approval has been obtained or any required pre-notification has been made for
02
any changes. These models should continue to be monitored as required by the
y2
Credit Risk: Internal Ratings Based Approach (CRR) Part;
ar
(d) only adjustments that increase RWAs and EL should be made, and there should
u
an
be no netting of adjustments across portfolios (eg if there are two adjustments in
1J
respect of data deficiencies in separate portfolios, one which increases RWAs by
m
£200 million and one that decreases RWAs by £100 million, only the adjustment
ro
that increases RWAs by £200 million should be applied). Netting of impacts

f
should only be applied within a portfolio (ie where a model covers a number of
ve
portfolios, netting should only be applied at a portfolio level);

cti
(e) a list of all model adjustments should be included in the firm’s model monitoring
ffe
information presented to senior management, containing at least the following

.E
information:
t2
(i) the portfolio and model component affected;

ar
lp
(ii) a description of the issue and why it requires the adjustment;

na
(iii) the date when the issue was first identified;

-fi
(iv) what action is being taken to address the issue and the timeline for this action;
ar
and
Ne
(v) the increase to RWAs and EL as a result of the adjustment;

(f) firms may net adjustments across model components (eg PD, LGD, and EAD or
CF); however, if the PRA assesses that a firm is not applying the netting across
components appropriately, or with the correct degree of conservatism, it expects
that netting will then be applied only within a model component (eg if the
adjustment to PD increases capital requirements, and the adjustment to LGD
decreases capital requirements, the firm would only apply the increased capital
requirements that result from the PD adjustment); and
(g) the PRA expects firms to apply all parameter and risk-weight floors that are set
out in the Credit Risk: Internal Ratings Based Approach (CRR) Part at all times.
Therefore, PMAs should be assessed following the application of all relevant
floors.
2.6 The PRA expects that firms should address identified model deficiencies in a timely
manner and that, as a result, PMAs would only need to be applied on a temporary basis.
Fees
2.7 There will be some circumstances where a fee may be applied, for example, where a firm
is applying to move from the foundation internal ratings based (FIRB) approach to the
advanced internal ratings based (AIRB) approach, or a special project fee may be applicable
in the case of a merger or acquisition.
Application of requirements to UK groups applying the IRB
6.
approach on a unified basis
02
y2
2.8 CRR Article 20(6) provides that, where the IRB approach is used on a unified basis by a
ar
UK group, the PRA is required to permit certain IRB requirements to be met on a collective
u
an
basis by members of that group. The PRA considers that, where a firm is reliant upon a rating
1J
system or data provided by another member of its group, it will not meet the condition that it
m
is using the IRB approach on a unified basis unless:

f ro
(a) the firm only does so to the extent that it is appropriate, given the nature and scale of
ve
the firm’s business and portfolios, and the firm’s position within the UK group;
cti
(b) the integrity of the firm’s systems and controls is not adversely affected;
ffe
(c) the outsourcing of these functions meets the requirements of the Outsourcing Part of
.E
the PRA Rulebook; and

t2
(d) the ability of the PRA to carry out its responsibilities is not adversely affected.
ar
lp
2.9 Prior to reliance being placed by a firm on a rating system, or data provided by another
na
member of the group, the PRA expects the proposed arrangements to have been explicitly
-fi
considered, and found to be appropriate, by the governing body of the firm.

ar
Ne
2.10 If a firm uses a rating system or data provided by another group member, the PRA
expects the firm’s governing body to delegate those functions formally to the persons or
bodies that are to carry them out.
2.11 Where a firm’s rating systems are used on a unified basis pursuant to CRR Article 20(6),
the PRA considers that the governance requirements in Article 189 of the Credit Risk:
Internal Ratings Based Approach (CRR) Part can be met only if the subsidiary undertakings
have delegated to the governing body or designated committee of the UK parent institution,
UK parent financial holding company, or UK parent mixed financial holding company,
responsibility for approval of all material aspects of rating and estimation processes.
Annual attestations
2.12 The PRA expects an appropriate individual in a Senior Management Function (SMF)
role to provide the following written attestations to the PRA:
(a) an annual attestation that the firm is either (i) fully compliant, (ii) materially compliant
or (iii) materially non-compliant with the CRR requirements, PRA rules and SSs; and
(b) where the SMF attests that the firm is materially non-compliant under point (a), the
attestation should include confirmation that a credible plan for addressing non-
compliance in a timely manner is in place and is being implemented.
2.13 Firms should agree the appropriate SMF for providing these attestations with the PRA.
6.
The PRA does not expect to agree to more than two SMFs covering all of a firm’s IRB
02
y2
models. In agreeing which SMF(s) may provide the annual attestation, the PRA will consider
the firm’s arrangements for approving rating and estimation processes under Article 189 of
u ar
the Credit Risk: Internal Ratings Based Approach (CRR) Part.
an
1J
Submission of applications and notifications relating to the IRB
m
approach
f ro
ve
2.14 The PRA expects that where a firm submits an application for a permission relating to
cti
the IRB approach or notifies the PRA of an extension or change to an IRB rating system, it
ffe
should provide the PRA with a self-assessment of whether it complies with all relevant CRR
.E
articles, PRA rules, and SS expectations.

t2
ar
2.15 The PRA expects that where a firm with an IRB permission submits an application for a
lp
permission relating to the IRB approach or notifies the PRA of an extension or change to its
na
rating systems under Article 143(4)(b) of the Credit Risk: Internal Ratings Based Approach
-fi
(CRR) Part, it should complete a pro-forma provided by the PRA. The current pro-forma and
ar
instructions on its completion can be found on the ‘Permissions (CRR firms)’ webpage on the
Ne
PRA website.2
Overseas models approach

2.16 Where a firm is using the overseas models approach in accordance with Article 143(6)
of the Credit Risk: Internal Ratings Based Approach (CRR) Part, the PRA expects that the
contribution to UK consolidated capital requirements calculated using overseas models
should reflect any regulatory floors or add-ons mandated by the relevant overseas regulator.
2 www.bankofengland.co.uk/prudential-regulation/authorisations/capital-requirements-regulation-
permissions.
3: Partial use and reversion to less

sophisticated approaches
Policy for identifying exposures subject to partial use

3.1 Firms should have a well-documented policy explaining the basis on which exposures
would be selected for permanent exemption from the requirements in Article 147C of the
Credit Risk: Internal Ratings Based Approach (CRR) Part. This policy should be provided to
6.
the PRA when the firm applies for permission and maintained on an ongoing basis thereafter.
02
y2
3.2 The PRA expects that where a firm submits a roll-out plan in accordance with Article
148(3) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, this should provide
u ar
for the continuing application of the policy referred to in paragraph 3.1 on a consistent basis
an
over time. 1J
m
Roll-out following significant acquisitions

f ro
3.3 In the event that a firm with an IRB permission acquires a significant new business, it
ve
should discuss with the PRA whether sequential roll-out of the firm’s IRB approach to the
cti
acquired exposures would be appropriate, and whether any changes to any existing time
ffe
period and conditions for sequential roll-out are required. Firms should apply for any
.E
necessary permissions under Articles 148 and 150 of the Credit Risk: Internal Ratings Based
t2
Approach (CRR) Part.

ar
lp
na
Permanent partial use of the Standardised Approach to a roll-

-fi
out class
ar
Ne
3.4 Points (i), (ii) and (iii) of Article 150(1)(k) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, sets out criteria for the permanent use of the SA to a roll-out class.
Under these criteria, firms can apply the SA to a roll-out class (with PRA permission) if any of
the following applies:
(a) application of the SA would not result in significantly lower capital requirements than if
the IRB approach was applied (point (i));
(b) the firm cannot reasonably model exposures in the roll-out class (point (ii)); or
(c) the roll-out class is immaterial (point (iii)).
3.5 In the PRA’s experience, SA capital requirements for the qualifying revolving retail
exposures and the specialised lending roll-out classes are often materially lower than if the
IRB approach was applied. Firms can usually be reasonably expected to apply the IRB
approach to the qualifying revolving retail exposures roll-out class (due to good data
availability), and the specialised lending roll-out class (due to the availability of the slotting
approach). The PRA therefore considers it likely that firms will only meet the criteria in Article
150(1)(k) of the Credit Risk: Internal Ratings Based Approach (CRR) Part for permanently
applying the SA to either of these roll-out classes if the roll-out class in question is immaterial.
Reversion to less sophisticated approaches

3.6 The PRA expects that firms applying for a permission to revert to a less sophisticated
approach under Article 149 of the Credit Risk: Internal Ratings Based Approach (CRR) Part,
6.
should simultaneously apply for any necessary permissions under Articles 148 and 150 of the
02
Credit Risk: Internal Ratings Based Approach (CRR) Part.
y2
ar
3.7 Under Articles 149(1) and 149(2) of the Credit Risk: Internal Ratings Based Approach
u
(CRR) Part, firms must meet a number of conditions in order to revert to less sophisticated
an
approaches for credit risk, including that the reversion is necessary on the basis of the nature
1J
and complexity of the firm’s exposures. The PRA considers that one-off costs arising from
m
implementing the Basel 3.1 standards is a relevant factor in determining whether these tests
ro
are met.
f
ve
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
4: Use and experience tests
Use test
4.1 In accordance with Article 144(1)(b) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part, firms should ensure that the estimates of risk parameters used for the purpose of
capital requirements calculations play an essential role in internal risk management and
decision-making processes. In particular, the PRA expects firms to ensure that:
(a) any deviations between parameters used for internal purposes and for capital
6.
02
requirements purposes are justified and appropriate for the specific area of use; and
y2
(b) the rank-ordering in the assignment of obligors or facilities to grades and pools within
a calibration segment plays an essential role in the rank-ordering used for internal risk
u ar
management and decision-making processes.
an
1J
4.2 The PRA considers that deviations caused by the use of parameters for internal purposes
m
that do not include either:

f ro
(a) a MoC;
ve
(b) regulatory floors; or

cti
(c) downturn adjustments in the case of LGD, or EAD or CF estimates

ffe
.E
will not normally prevent the criteria in Article 144(1)(b) of the Credit Risk: Internal Ratings
t2
Based Approach (CRR) Part from being met.

ar
lp
4.3 Where firms use estimates of risk parameters for internal purposes that are different from
na
those used in the calculation of capital requirements, they should periodically reflect this in
-fi
their internal reporting to senior management by providing information on both sets of

ar
parameters. In any case, internal reporting should include all elements specified in Article
Ne
189(3) of the Credit Risk: Internal Ratings Based Approach (CRR) Part based on the
estimates of risk parameters that are used for the purpose of capital requirements
calculations.
Prior experience of using the IRB approach

4.4 In order to be satisfied that the requirements in Article 145 of the Credit Risk: Internal
Ratings Based Approach (CRR) Part are met, the PRA expects a firm to be able to evidence
that:
(a) its complete IRB governance framework has been through at least one annual cycle
since internal approval;
(b) it has used its internal rating systems in credit decisions, lending policies, risk appetite
polices, and credit risk monitoring for at least three years; and
(c) there has been at least three years of monitoring, validation, and audit of the IRB
framework, recognising that the IRB framework is likely to be subject to development
and refinement during this period.
4.5 The three years of evidence of using internal rating systems expectation set out in
paragraph 4.4(b) need not necessarily relate to the use of the final Credit Risk: Internal
Ratings Based Approach (CRR) Part materially compliant approach for all of that period. It
could, for example, initially involve the use of internal credit risk models that are broadly in
line with Credit Risk: Internal Ratings Based Approach (CRR) Part requirements rather than
6.
the final, materially compliant, IRB rating system. The PRA expects, however, that applicants
02
should have undertaken at least one annual review cycle of the completed framework by the
y2
point of approval.
u ar
4.6 The depth and detail of the monitoring, audit, and annual reviews set out in paragraph
an
4.4(c) may be proportionately lower at the start of the three-year period, provided that firms
1J
provide a sufficiently accurate analysis of progress, and fully meet the required standard by
m
the end of the three-year period. The monitoring of rating systems may include the use of
ro
provisioning models, scorecards, and rating assignment processes.

f
ve
cti
4.7 The PRA’s policy is not to accept evidence of a third-party exercising governance of
ffe
models (eg bureau scores monitored by the bureau) as evidence of a firm’s ability to monitor
.E
the models itself.

t2
ar
lp
na
-fi
ar
Ne
5: Qualifying revolving retail exposures
5.1 Article 147(5A)(d) of the Credit Risk: Internal Ratings Based Approach (CRR) Part
specifies that, in order for an exposure to be treated as a qualifying revolving retail exposure,
it needs to exhibit relatively low volatility of loss rates. The PRA expects firms to assess the
volatility of loss rates for the qualifying revolving retail exposure portfolio relative to the
volatilities of loss rates of other relevant types of retail exposures for these purposes. Low
volatility should be demonstrated by reference to data on the mean and standard deviation of
loss rates over a time period that can be regarded as representative of the long-run
performance of the portfolios concerned.
6.
02
5.2 Article 147(5A)(e) of the Credit Risk: Internal Ratings Based Approach (CRR) Part
y2
specifies that in order for an exposure to be treated as a qualifying revolving retail exposure,
ar
this treatment should be consistent with the underlying risk characteristics of the sub-
u
an
portfolio. The PRA considers that a sub-portfolio consisting of credit card or overdraft
1J
obligations will usually meet this condition, and that it is unlikely that any other type of retail
m
exposure would do so. If a firm wishes to apply the treatment in Article 147(5A) of the Credit
ro
Risk: Internal Ratings Based Approach (CRR) Part to product types other than credit card or
f
ve
overdraft obligations, the PRA expects it to discuss this with the PRA before doing so.
cti
5.3 When identifying qualifying revolving retail exposures as transactor exposures or non-
ffe
transactor exposures as required under 147(5A) of the Credit Risk: Internal Ratings Based
.E
Approach (CRR) Part, firms should apply the expectations set out in paragraphs 4.2 and 4.3
t2
of SS10/13 – Credit risk Standardised approach, paragraphs 4.2 to 4.3.3

ar
lp
na
-fi
ar
Ne
3 www.bankofengland.co.uk/prudential-regulation/publication/2013/credit-risk-standardised-approach-
ss.
6: High level expectations
High level expectations for estimation

6.1 The PRA expects the information that a firm produces or uses for the purpose of the IRB
approach to be reliable and take proper account of the different users of the information
produced (customers, shareholders, regulators, and other market participants).
6.2 The PRA expects firms to establish quantified and documented targets and standards,
against which they should test the accuracy of data used in their rating systems. Such tests
6.
02
should cover:
y2
(a) a report and accounts reconciliation, including whether every exposure has a PD and,
ar
if applicable, an LGD and an EAD or CF for reporting purposes;
u
an
(b) whether the firm’s risk control environment has key risk indicators for the purpose of
1J
monitoring and ensuring data accuracy;
m
(c) whether the firm has an adequate business and information technology infrastructure
ro
with fully documented processes;

f
(d) whether the firm has clear and documented standards on ownership of data (including
ve
inputs and manipulation) and timeliness of current data (daily, monthly, real time); and
cti
(e) whether the firm has a comprehensive quantitative audit programme.

ffe
.E
6.3 The PRA expects that in respect of data inputs, the testing for accuracy of data, including
t2
the reconciliation referred to above, should be sufficiently detailed so that, together with other
ar
available evidence, it provides reasonable assurance that data input into the rating system is
lp
accurate, complete, and appropriate. The PRA considers that input data would not meet the
na
required standard if it gave rise to a serious risk of material misstatement of capital

-fi
ar
requirements, either immediately or subsequently.

Ne
6.4 In respect of data outputs, as part of the reconciliation referred to above, the PRA
expects a firm to be able to identify and explain material differences between the outputs
produced under accounting standards and those produced under the requirements of the IRB
approach, including in relation to areas that address similar concepts in different ways (eg EL
and accounting provisions).
6.5 The PRA expects a firm to have clear and documented standards and policies about the
use of data in practice (including information technology standards) which should in particular
cover the firm’s approach to the following:
(a) data access and security;

(b) data integrity, including the accuracy, completeness, appropriateness and testing of
data; and
(c) data availability.
Quality of data
6.6 The PRA expects that firms should have sound policies, processes, and methods for
assessing and improving the quality of data used for the purpose of credit risk measurement
and management processes. Firms should ensure that those policies apply to all data used in
model development and calibration, as well as to the data used in the application of the risk
parameters.
6.
02
6.7 In order for the data used in the model development and in the application of risk
y2
parameters as inputs into the model to meet the requirements of accuracy, completeness
ar
and appropriateness specified in Article 174(b) of the Credit Risk: Internal Ratings Based
u
Approach (CRR) Part, they should be sufficiently precise to avoid material distortions of the
an
outcome of the assignment of exposures to obligors or facility grades or pools, and they
1J
should not contain any biases which make the data unfit for purpose. The PRA expects that
m
where firms identify deficiencies in either the quality of data used, or in their processes for
ro
maintenance of the data, they take steps to address these deficiencies in a timely manner.
f
ve
cti
6.8 The PRA expects that firms’ documentation relating to data should include clear
ffe
identification of responsibility for data quality. The PRA expects firms to set standards for
.E
data quality, to aim to improve them over time, and to measure their performance against
t2
those standards. Furthermore, the PRA expects firms to ensure that their data are of
ar
sufficiently high quality to support the firm’s risk management processes and the calculation
lp
of capital requirements.
na
-fi
Ratings systems: policies

ar
Ne
6.9 In order for the PRA to be satisfied that a firm’s rating systems are robust, the PRA
expects a firm to be able to demonstrate that it has an appropriate policy in respect of its
ratings systems in relation to:
(a) any deficiencies caused by a rating system not being sensitive to movements in
fundamental risk drivers or for any other reason;
(b) periodic review and action in the light of such review;
(c) providing appropriate internal guidance to staff to ensure consistency in the use of the
rating system, including the assignment of exposures or facilities to grades or pools;
(d) dealing with potential weaknesses of the rating system;
(e) identifying appropriate and inappropriate uses of the rating system and acting on that
identification;
(f) novel or narrow rating approaches; and

(g) ensuring the appropriate level of stability over time of the rating system.
6.10 In order to comply with the requirement on the representativeness of data used in the
PD and LGD models specified in Articles 174(c), 179(1)(d) and 179(2)(b) of the Credit Risk:
Internal Ratings Based Approach (CRR) Part, firms should have sound policies, processes,
and methods for assessing the representativeness of data used for the purpose of estimation
of risk parameters. Firms should specify in their internal policies the statistical tests and
metrics to be used for the purpose of assessing the representativeness of data used for risk
differentiation and, separately, for data underlying the risk quantification. Firms should also
specify methods for qualitative assessment of data for the cases, defined in their policies,
6.
where the application of statistical tests is not possible.
02
y2
6.11 Firms should use the same standards and methods for the assessment of
ar
representativeness of data stemming from different sources, including internal, external, and
u
pooled data, or a combination of these, unless different methods are justified by the
specificity of the data source or availability of information.
an
1J
m
6.12 Where external or pooled data are used, firms should obtain sufficient information from
ro
the data providers to assess the representativeness of such external or pooled data to firms’
f
ve
own portfolios and processes.

cti
ffe
Collection of data
.E
6.13 In order to be satisfied that the requirements in Article 179(1) of the Credit Risk: Internal
t2
Ratings Based Approach (CRR) Part are met, the PRA expects a firm to collect data on what
ar
lp
it considers to be the main drivers of the risk parameters PD, LGD, CF or EAD, and EL (as
na
applicable) for each group of obligors or facilities, to document the identification of the main
-fi
drivers of risk parameters, and to be able to demonstrate that the process of identification is
ar
reasonable and appropriate.

Ne
6.14 In its processes for identifying the main drivers of risk parameters, the PRA expects that
firms should set out their reasons for concluding that the data sources chosen provide in
themselves sufficient discriminative power and accuracy, and why additional potential data
sources do not provide relevant and reliable information that would be expected to materially
improve the discriminative power and accuracy of its estimates of the risk parameter in
question. The PRA would not expect this process necessarily to require an intensive analysis
of all factors.
Human judgement in estimation of risk parameters

6.15 In order for firms to complement their statistical models with human judgement, as
referred to in Articles 174(b), 174(e), 175(4), 179(1)(a) and 180(1)(d) of the Credit Risk:
Internal Ratings Based Approach (CRR) Part, the PRA expects they should do all of the
following:
(a) assess the modelling assumptions and whether the selected risk drivers contribute to
the risk assessment in line with their economic meaning;
(b) analyse the impact of the human judgement on the performance of the model and
ensure that any form of human judgement is properly justified; and
(c) document the application of human judgement in the model, including at least the
criteria for the assessment, rationale, assumptions, experts involved, and description
of the process.
6.
Documentation
02
y2
6.16 The PRA expects a firm to ensure that all documentation relating to its rating systems
ar
(including any documentation referenced in this SS or required by the PRA Rulebook
u
requirements that relate to the IRB approach) is stored, arranged, and indexed in such a way
an
that it could make them all, or any subset thereof, available to the PRA immediately on
1J
demand or within a short time thereafter.
m
f ro
ve
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
7: Rating systems
Principles for specifying the range of application of a rating

system
7.1 The PRA expects that a ‘rating system’ as defined in Rule 1.3 of the Credit Risk: Internal
Ratings Based Approach (CRR) Part should cover all those exposures where the obligors or
facilities exhibit common drivers of risk and creditworthiness, and fundamentally comparable
availability of credit-related information. The PRA considers that the PD and LGD model
6.
within a rating system may comprise various calibration segments. Where all obligors or
02
exposures within the range of application of the PD or LGD model are jointly calibrated, the
y2
whole scope of application of the model should be considered to be one calibration segment.
u ar
7.2 Exposures covered by the same rating system should be treated similarly by the firm in
an
1J
terms of risk management, decision making and the credit approval process. For exposures
to corporates and institutions, such exposures should be assigned to a common obligor
m
rating scale for the purposes of Article 170(1)(b) of the Credit Risk: Internal Ratings Based
f ro
Approach (CRR) Part and, where LGD is estimated, to a common facility rating scale for the
ve
purposes of Article 170(1)(e) of the Credit Risk: Internal Ratings Based Approach (CRR)
cti
Part.
ffe
.E
7.3 For the purpose of the quantification of the risk parameters within a rating system, firms
t2
should, for a given historical observation, apply the same definition of default to the historical
ar
observation in all models for which the historical observation is used. Firms should also apply
lp
the same treatment of multiple defaults of the same obligor or exposure across internal,
na
external, and pooled data sources.

-fi
ar
Multi-country mid-market corporate PD models

Ne
7.4 In order to ensure that a rating system provides a meaningful differentiation of risk, and
accurate and consistent quantitative estimates of risk, the PRA expects that firms should
usually develop country-specific mid-market PD models. Where firms develop multi-country
mid-market PD models, the PRA expects firms to be able to demonstrate that the model
rank-orders risk and predicts default rates for each country within the scope of the rating
system.
7.5 The PRA expects firms to have challenging standards in place to meaningfully assess
whether a model rank-orders risk and accurately predicts default rates. These standards
should specify the number of defaults that are needed for a meaningful assessment to be
done.
7.6 The PRA expects firms to assess the model’s ability to predict default rates using a time
series of data (ie not only based on one year of default data).
7.7 The PRA considers that a model is not likely to be materially compliant with the
requirements set out in the Credit Risk: Internal Ratings Based Approach (CRR) Part where
the firm cannot demonstrate that it rank-orders risk and predicts default rates for each
country, regardless of any apparent conservatism in the model.
Retirement interest-only (RIO) mortgages

7.8 The PRA considers that there may be some circumstances where it may be appropriate
6.
to model RIO mortgages with other mortgages, but it considers that firms should be able to
02
justify this. The PRA expects that firms wishing to apply the IRB approach in respect of RIO
y2
mortgage exposures should apply to the PRA for permission to do so. The PRA would only
ar
grant permission once it is satisfied that the relevant requirements of the Credit Risk: Internal
u
Ratings Based Approach (CRR) Part are materially met. The PRA would not expect firms to
an
apply to use the IRB approach for RIO mortgages until they have sufficient data to
1J
demonstrate that their approach is prudentially appropriate and materially compliant with the
m
requirements of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
f ro
ve
7.9 Firms should be able to demonstrate the appropriateness of the treatment of interest-only
cti
(IO) mortgages that are converted into RIO mortgages in their IRB models through robust
ffe
analysis. In particular, as a minimum, consideration should be given to the effect of defaults

.E
of IO mortgages on IRB model estimates for IO mortgages. A prudent approach should be

t2
applied to the modelling of IO mortgages that return to non-defaulted status only as a result
ar
of the borrower being transferred to an RIO product.

lp
na
-fi
ar
Ne
8: Data representativeness
Overall assessment
8.1 In accordance with Article 174 of the Credit Risk: Internal Ratings Based Approach (CRR)
Part, firms are required to ensure that data used to build a model is representative of its
actual exposures. The PRA expects that where external data are used by a firm to build
models, it should consider whether the data are appropriate to its own experience and make
any adjustments that are necessary.
6.
02
Representativeness of data for model development
y2
8.2 Firms should analyse the representativeness of data in the case of statistical models and
u ar
other mechanical methods used to assign exposures to grades or pools, as well as in the
an
case of statistical default prediction models generating default probability estimates for
1J
individual obligors or facilities. Firms should select an appropriate data set for the purpose of
m
model development to ensure that the performance of the model on the application portfolio,
ro
in particular in respect of the model’s discriminatory power, is not significantly hindered by

f
ve
insufficient representativeness of the data.

cti
ffe
8.3 For the purposes of ensuring that the data used in developing the model for assigning
obligors or exposures to grades or pools are representative of the application portfolio
.E
covered by the relevant model, as required in Article 174(c) of the Credit Risk: Internal
t2
Ratings Based Approach (CRR) Part, firms should analyse the representativeness of the
ar
lp
data at the stage of model development in terms of all of the following:

na
(a) the scope of application;

-fi
ar
(b) the definition of default;

Ne
(c) the distribution of the relevant risk characteristics; and

(d) lending standards and recovery policies.
8.4 For the purpose of paragraph 8.3(a), firms should analyse the segmentation of exposures
and consider whether there were any changes to the scope of application of the model over
the period covered by the data used in developing the model for assigning obligors or
exposures to grades or pools. Where such changes were observed, firms should analyse the
risk drivers relevant for the revised scope of application of the model by comparing their
distribution in the RDS before and after the change, as well as with the distribution of those
risk drivers in the application portfolio. For this purpose, firms should apply statistical
methodologies such as cluster analysis or similar techniques to demonstrate
representativeness. In the case of pooled models, the analysis should be performed with
regard to the part of the scope of the model that is used by the firm.
8.5 For the purpose of paragraph 8.3(b), firms should ensure that the definition of default
reflected in the data used for model development is consistent over time and, in particular,
that it is consistent with all of the following expectations:
(a) that adjustments have been made to achieve consistency with the definition of default
used by the firm in accordance with Article 178 of the Credit Risk: Internal Ratings
Based Approach (CRR) Part where the definition of default has been changed during
the observation period;
(b) that adequate measures have been adopted by the firm where the model covers
exposures in several jurisdictions having or having had different default definitions;
(c) that the definition of default in each data source has been analysed separately; and
6.
(d) that the definition of default used for the purposes of model development does not
02
have a negative impact on the structure and performance of the rating model, in terms
y2
of risk differentiation and predictive power, where this definition is different from the
ar
definition of default used by the firm in accordance with Article 178 of the Credit Risk:
u
an
Internal Ratings Based Approach (CRR) Part.
1J
8.6 For the purpose of paragraph 8.3(c), firms should analyse the distribution and range of
m
values of key risk characteristics of the data used in developing the model for risk
ro
differentiation in comparison with the application portfolio. With regard to LGD models, firms
f
ve
should perform such analysis separately for non-defaulted and defaulted exposures.
cti
ffe
8.7 Firms should analyse the representativeness of the data, in terms of the structure of the
.E
portfolio by relevant risk characteristics, based on statistical tests specified in their policies to
t2
ensure that the range of values observed on these risk characteristics in the application
ar
portfolio is adequately reflected in the development sample. Where the application of

lp
statistical tests is not possible, firms should carry out at least a qualitative analysis on the
na
basis of the descriptive statistics of the structure of the portfolio, taking into account, in the
-fi
case of retail exposures, any seasoning effects as referred to in Article 180(2)(f) of the Credit
ar
Risk: Internal Ratings Based Approach (CRR) Part. When considering the results of this
Ne
analysis, firms should take into account the sensitivity of the risk characteristics to economic
conditions. Material differences in the key risk characteristics between the data sample and
the application portfolio should be addressed, for example by using another data sample or a
subset of observations, or by adequately reflecting these risk characteristics as risk drivers in
the model.
8.8 For the purpose of paragraph 8.3(d), firms should analyse whether, over the relevant
historical observation period, there were significant changes in their lending standards or
recovery policies or in the relevant legal environment, including changes in insolvency law,
legal foreclosure procedures, and any legal regulations related to realisation of collateral,
which may influence the level of risk or the distribution or ranges of the risk characteristics in
the portfolio covered by the considered model. Where firms observe such changes, they
should compare the data included in the RDS before and after the change of the policy.
Firms should ensure comparability of the current underwriting or recovery standards with
those applied to the observations included in the RDS and used for model development.
8.9 While the PRA considers that the proportion of defaulted and non-defaulted exposures in
the RDS used for developing the risk differentiation aspect of PD models need not be equal
to the proportion of defaulted and non-defaulted exposures in the firm’s application portfolio,
the PRA expects that firms should however have a sufficient number of defaulted and non-
defaulted observations in the development data set, and that they should document any
difference in proportions.
6.
Representativeness of data for calibration of risk parameters
02
y2
8.10 In accordance with Article 179(1)(d) of the Credit Risk: Internal Ratings Based Approach
ar
(CRR) Part, and, where pooled data are used, in accordance with Article 179(2)(b) of the
u
Credit Risk: Internal Ratings Based Approach (CRR) Part, in order for a firm to ensure that
an
the data used in risk quantification are representative of the firm’s exposures covered by the
1J
relevant model, the firm should analyse the comparability of the data used for the purpose of
m
calculating long-run average default rates, long-run average LGDs, and downturn LGDs (as
ro
appropriate) with the characteristics of the firm’s exposures in terms of all of the following:
f
ve
cti
(a) the scope of application;

ffe
(b) the definition of default;

.E
(c) the distribution of the relevant risk characteristics;

t2
(d) relevant economic or market conditions; and

ar
(e) lending standards and recovery policies.

lp
na
8.11 For the purpose of paragraph 8.10(a), firms should perform an analysis consistent with
-fi
the analysis specified in paragraph 8.4.

ar
Ne
8.12 For the purpose of paragraph 8.10(b), and in order to ensure that the definition of default
underlying the data used for risk quantification from each data source is consistent with the
definition of default used in accordance with Article 178 of the Credit Risk: Internal Ratings
Based Approach (CRR) Part, a firm should compare the definition of default used by the firm
in accordance with Article 178 of the Credit Risk: Internal Ratings Based Approach (CRR)
Part with the definition of default reflected in the observations included in the data set used
for risk quantification. Where the definition of default has changed during the historical
observation period, the firm should assess the representativeness of historical data included
in the RDS and used for risk quantification in the same way as specified for external data in
Chapter 4 – Application of the definition of default in external data of SS3/24 – Credit risk
definition of default.4 Where the definition of default has changed during the historical
observation period more than once, the firm should perform the analysis of each of the past
definitions of default separately.
8.13 For the purpose of paragraph 8.10(c), firms should perform an appropriate analysis to
ensure that, at the level of the calibration segment, the ranges of values of the key risk
characteristics in the application portfolio are comparable to those in the portfolio constituting
the RDS for risk quantification to the degree required to ensure that the risk quantification is
not biased.
8.14 For the purpose of paragraph 8.10(d), firms should perform the analysis of the market
6.
and economic conditions underlying the data in the following manner:
02
y2
(a) with regard to the PD estimation, in accordance with paragraphs 11.12 to 11.19; and
(b) with regard to the LGD estimation, in accordance with paragraphs 14.1 to 14.4 and in
u ar
accordance with paragraphs 15.1 to 15.3 when taking into account economic
an
downturn conditions in accordance with Article 181(1)(b) of the Credit Risk: Internal
1J
Ratings Based Approach (CRR) Part.
m
ro
8.15 For the purpose of paragraph 8.10(e), firms should analyse whether there were
f
significant changes in the lending standards or recovery policies over the relevant historical
ve
observation period that may influence the level of risk or the distribution or ranges of the
cti
characteristics of relevant risk drivers in the portfolio covered by the considered model.
ffe
Where firms observe such changes, they should analyse the potential bias in the estimates of
.E
risk parameters resulting from these changes in the following manner:

t2
ar
(a) for PD estimation, in terms of the level of default rates and the likely range of variability
lp
of default rates; and

na
(b) for LGD estimation, in terms of loss rates, average duration of recovery processes,
-fi
ar
frequencies of use of certain recovery scenarios, and loss severity distributions.

Ne
8.16 Where the representativeness of data assessed in accordance with paragraphs 8.10 to
8.15 is insufficient and leads to a bias or increased uncertainty of risk quantification, firms
should introduce an appropriate adjustment to correct the bias and they should apply a
margin of conservatism in accordance with Chapter 9 – Model deficiencies and margin of
conservatism.
4 https://www.bankofengland.co.uk/prudential-regulation/publication/2024/september/credit-risk-
definition-of-default-supervisory-statement.
PD – use of external data for residential mortgages

8.17 The PRA expects that, for residential mortgages, where a firm’s internal experience of
defaults for a rating system is low, it may use external data to supplement internal data for
rank-ordering different borrowers by credit quality and to help adjust for seasoning as credit
quality changes with loan vintage. This is in addition to use of external data for calibration
purposes. The PRA expects that firms attempting to evidence comparability with third-party
data should include a comparison of default rates.
8.18 The PRA considers that internal data may be treated as being the ‘primary source’ of
data for residential mortgages where a firm assigns sufficient weight to internal data,
6.
including security (loan to value), loan (arrears history) and borrower (applicant information)
02
factors, as inputs into their rank-ordering but uses external data to achieve greater
y2
discrimination.
u ar
8.19 The PRA expects firms to apply an appropriate MoC to account for uncertainty in their
an
estimates and to make conservative adjustments in respect of incomplete data and external
1J
data that are not wholly representative.
m
ro
8.20 Where firms lack sufficient internal defaults to evidence rank-ordering or a reliable
f
calibration, firms may use models that rank-order on an early arrears definition (which tends
ve
to be correlated with default), provided they are calibrated with sufficient conservatism.
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
9: Model deficiencies and margin of

conservatism
Identification of deficiencies
9.1 Firms should identify all deficiencies related to the estimation of risk parameters that lead
to a bias in the quantification of those parameters, or to increased uncertainty that is not fully
captured by the general estimation error, and classify each deficiency into one of the
6.
following groups:
02
y2
(a) Group A: identified data and methodological deficiencies; and
(b) Group B: relevant changes to underwriting standards, risk appetite, collection and
u ar
recovery policies, and any other source of additional uncertainty.
an
1J
9.2 For the purposes of identifying and classifying all deficiencies referred to in paragraph
m
9.1, firms should take into account all relevant deficiencies in methods, processes, controls,
ro
data, or IT systems that have been identified by the credit risk control unit, validation function,
f
internal audit function, or any other internal or external review, and should analyse at least all
ve
of the following potential sources of additional uncertainty in risk quantification:

cti
ffe
(a) under Group A:

.E
(i) missing or materially changed default triggers in historical observations,

t2
including changed criteria for recognition of materially past due credit

ar
obligations;
lp
(ii) missing or inaccurate date of default;

na
(iii) missing, inaccurate, or outdated rating assignment used for assessing

-fi
historical grades or pools for the purpose of calculation of default rates or

ar
Ne
average realised LGDs per grade or pool;

(iv) missing or inaccurate information on the source of cash flows;
(v) missing, inaccurate or outdated data on risk drivers and rating criteria;
(vi) missing or inaccurate information used for the estimation of future recoveries
as referred to in paragraph 14.13;
(vii) missing or inaccurate data for the calculation of economic loss;
(viii) limited representativeness of the historical observations due to the use of
external data;
(ix) potential bias stemming from the choice of the approach to calculating the
average of observed one-year default rates in accordance with paragraph
11.10;
(x) necessity of adjusting the average of observed one-year default rates in

accordance with paragraph 11.18; and
(xi) missing information for the purpose of estimating loss rates or for the
purpose of reflecting economic downturn in LGD estimates;
(b) under Group B:
(i) changes to underwriting standards, collection or recovery policies, risk
appetite, or other relevant internal processes;
(ii) unjustified deviations in the ranges of values of the key risk characteristics of
the application portfolio compared with those of the data set used for risk
quantification;
(iii) changes to the market or legal environment; and
6.
(iv) forward-looking expectations regarding potential changes in the structure of
02
the portfolio or the level of risk, especially based on actions or decisions that
y2
have already been taken but which are not reflected in the observed data.
u ar
an
9.3 In order to overcome biases in risk parameter estimates stemming from the identified
1J
deficiencies referred to in paragraphs 9.1 and 9.2, firms should apply adequate
methodologies to correct the identified deficiencies to the extent possible. The PRA expects
m
that the impact of these methodologies on the risk parameter (‘appropriate adjustment’)
f ro
should result in a more accurate estimate of the risk parameter (‘modelled estimate’) and that
ve
the appropriate adjustment may either result in an increase or a decrease in the value of the
cti
risk parameter. A firm should ensure and provide evidence that the application of an
ffe
appropriate adjustment results in its most accurate modelled estimate.

.E
t2
9.4 The PRA expects that firms should document the methods used to apply appropriate
ar
adjustments to rectify identified deficiencies and should document their justification of these
lp
methods.
na
-fi
9.5 The PRA expects that firms should regularly monitor the adequacy of appropriate
ar
adjustments. The PRA considers that the adoption of an appropriate adjustment by a firm
Ne
does not replace the need to address identified model deficiencies.
Margin of conservatism
9.6 As required Articles 179(1)(f) and 180(1)(e) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, firms should add an MoC to parameter estimates that is related to the
expected range of estimation errors. The PRA expects that firms should implement a
framework for the quantification, documentation, and monitoring of these estimation errors.
9.7 The final MoC added to a risk parameter estimate should reflect the uncertainty of
estimation in respect of all of the following:
Category A: the MoC related to data and methodological deficiencies identified under
Group A as referred to in paragraph 9.1(a);
Category B: the MoC related to relevant changes to underwriting standards, risk appetite,
collection and recovery policies, and any other source of additional uncertainty identified
under Group B as referred to in paragraph 9.1(b); and
Category C: the general estimation error.
9.8 In order to quantify the applicable MoC, firms should do all of the following:
(a) quantify the Category A and Category B MoC for the identified deficiencies referred to
6.
in paragraphs 9.1 and 9.2, to the extent not covered by the general estimation error, at
02
least at the level of the calibration segment ensuring that:
y2
(i) where appropriate adjustments as referred to in paragraph 9.3 are used, the MoC
ar
accounts for any increase in the uncertainty or additional estimation error associated
u
an
with these adjustments; 1J
(ii) the MoC quantified at category level for the identified deficiencies referred to in
m
paragraphs 9.1 and 9.2 related to the appropriate adjustments as referred to in

ro
paragraph 9.3 is proportionate to the uncertainty around these adjustments; and

f
(iii) the MoC is applied to address the uncertainty of the risk parameter estimate
ve
stemming from any deficiencies among those referred to in paragraphs 9.1 and 9.2
cti
that have not been corrected via appropriate adjustments as referred to in point (i);
ffe
.E
(b) quantify the general estimation error for the Category C MoC referred to in paragraph
t2
9.7 associated with the underlying estimation method at least for every calibration
ar
lp
segment; the MoC for the general estimation error should reflect the dispersion of the
na
distribution of the statistical estimator.

-fi
ar
9.9 For the purpose of paragraph 9.8(a), and for each of the Categories A and B, firms may
Ne
group all or selected deficiencies, where justified, for the purpose of quantifying the MoC.
9.10 Firms should quantify the final MoC as the sum of:
(a) the Category A MoC as referred to in paragraph 9.7;

(b) the Category B MoC as referred to in paragraph 9.7; and
(c) the Category C MoC for the general estimation error as referred to in paragraph 9.7.
9.11 Firms should add the final MoC to the modelled estimate of the risk parameter.
9.12 Firms should ensure that the impact of the final MoC does not result in a reduction of
parameter estimates and in particular that:
(a) the Category C MoC is either greater than zero, or equal to zero where the firm has
demonstrated that the general estimation error is immaterial; and
(b) the Category A MoC and the Category B MoC are proportionate to the increased
uncertainty in the modelled estimate of risk parameters caused by the identified
deficiencies relevant to each category. The Category A MoC and the Category B MoC
should each be greater than or equal to zero.
9.13 Firms should consider the overall impact of the identified deficiencies and the resulting
final MoC on the soundness of the model and ensure that estimates of risk parameters and
resulting capital requirements are not distorted by the need for excessive adjustments.
.
9.14 For each rating system, the MoC applied should be documented in relevant model
26
documentation and methodology manuals. The documentation should contain at least the
20
following:
ry
ua
(a) a complete list of all identified deficiencies, including errors and uncertainties, and the
an
potentially affected model components or risk parameters; 1J
(b) the group under which these deficiencies are classified, as referred to in paragraph
m
9.1; and
ro
(c) a description of the methods for quantification of the MoC related to identified
ef
deficiencies as referred to in paragraph 9.8(a) and, in particular, the methodologies

v
cti
used to quantify the MoC per category.

ffe
9.15 The PRA expects firms to regularly monitor the levels of MoCs. The PRA considers that
.E
the adoption of a MoC by a firm does not replace the need to address the causes of errors or
t2
uncertainties, or to correct the models to ensure material compliance with the requirements of
ar
lp
the Credit Risk: Internal Ratings Based Approach (CRR) Part.

na
9.16 When reviewing the levels of the MoC firms should ensure all of the following:
-fi
ar
(a) that the MoC stemming from Categories A and B referred to in paragraph 9.7 is
Ne
included in internal reporting separately for each category and may be reduced over
time and eventually eliminated once the deficiencies are rectified in all parts of the
rating system that were affected;
(b) that the MoC stemming from the general estimation error referred to in paragraph 9.7
is included in internal reporting in a separate category (‘C’); and
(c) that the level of the MoC is assessed as part of the regular reviews referred to in
Chapter 21 – Review of estimates (validation) and, in particular, that the level of MoC
related to the general estimation error remains appropriate after the inclusion of the
most recent data relevant for the risk parameter estimation.
9.17 Firms should ensure that necessary changes in the MoC are identified and either
notified to the PRA or submitted to the PRA for approval (as determined by Articles 143A to
143D of the Credit Risk: Internal Ratings Based Approach (CRR) Part) in a timely manner.
.
26
20
ry
ua
an
1J
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
10: PD – model development
Assignment of obligors and exposures

10.1 For the purpose of assigning obligors to an obligor grade as part of the credit approval
process in accordance with Article 172(1)(a) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, as well as for the purpose of the review of those assignments, in
accordance with Article 173(1)(b) of the Credit Risk: Internal Ratings Based Approach (CRR)
Part, firms should ensure that each and every natural or legal person in respect of which an
.
IRB exposure exists is rated by the firm with the model approved to be used for the
26
applicable type of exposures. The model should rate original obligors within the applicable
20
rating system including where unfunded credit protection is recognised in accordance with
ry
Article 160(4) of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
ua
an
10.2 For the purpose of assigning retail exposures to grades or pools as part of the credit
1J
approval process in accordance with Article 172(2) of the Credit Risk: Internal Ratings Based
m
Approach (CRR) Part, as well as for the purpose of the review of those assignments in
ro
accordance with Article 173(2) of the Credit Risk: Internal Ratings Based Approach (CRR)
ef
Part, firms should ensure that each and every IRB exposure is rated by the firm with the
v
cti
model approved to be used for the applicable type of exposures as defined in Rule 1.3 of the
ffe
Credit Risk: Internal Ratings Based Approach (CRR) Part. This model should rate original
.E
obligors or exposures within the applicable rating system including where unfunded credit
t2
protection is recognised in accordance with Article 163(4) of the Credit Risk: Internal Ratings
ar
Based Approach (CRR) Part.

lp
na
10.3 The PRA considers that a PD model may contain several different methods for ranking
-fi
the obligors or exposures as well as various calibration segments.

ar
Ne
Data for model development

10.4 The PRA expects that, for the purpose of model development, firms should ensure that
the RDS contains the values of the risk drivers for appropriate points in time. These points in
time may vary between different risk drivers. In the selection of appropriate points in time,
firms should take into account the dynamics as well as the update frequency of the risk
drivers throughout the whole period in which an obligor was in the portfolio and, in the case of
a default, throughout the year prior to default.
Risk drivers and rating criteria

10.5 In the process of selecting risk drivers and rating criteria, firms should consider a broad
set of information relevant to the type of exposures covered by the rating system. Potential
risk drivers analysed by firms should, in particular, include the following where relevant:
(a) obligor characteristics, including sector and geographic location for corporates;
(b) financial information, including financial statements or income statements;
(c) trend information, including growing or shrinking sales or profit margin;
(d) behavioural information, including delinquency and the use of credit facilities; and
(e) for retail exposures, transaction information, including product type, collateral type,
.
seasoning effects, and seniority.
26
20
10.6 In respect of seasoning effects for retail exposures, the PRA expects that firms should
ry
analyse the representativeness of the age of the facilities (measured as time since
ua
origination) in the data used to derive PD estimates. The PRA considers that default rates
an
peak several years after origination for some portfolios in some jurisdictions, and as a result
1J
the PRA expects firms to adjust their estimates with an adequate MoC in accordance with
m
Chapter 9 to account for any consequential lack of data representativeness as well as any
ro
anticipated implications of rapid exposure growth should they identify any shortcomings in the
ef
model’s ability to accurately reflect seasoning effects.

v
cti
10.7 Firms should ensure that for the purpose of selecting risk drivers and rating criteria, the
ffe
relevant experts from business areas of the firm are consulted with respect to the business
.E
rationale and risk contribution of the considered risk drivers and rating criteria.
t2
ar
10.8 Firms should ensure that the decrease of reliability of information over time, for instance
lp
of information on obligor characteristics obtained at the time of the loan origination, is

na
appropriately reflected in the PD estimation. Firms should also ensure that the model
-fi
ar
estimates the proper level of risk with respect to all relevant, currently available and most up-
Ne
to-date information, and that an adequate MoC is applied where a higher degree of
uncertainty exists due to the lack of up-to-date information.
10.9 Where there is a significant proportion of customers using multiple facilities of the same
type within a retail rating system, firms should analyse the level of risk of such customers
compared with customers with only one facility of the relevant type and, where necessary,
reflect the difference in the level of risk in the model through appropriate risk drivers.
Rating philosophy
10.10 ‘Rating philosophy’ describes the point at which a rating system sits on the spectrum
between the stylised extremes of a point-in-time (PiT) rating system and a through-the-cycle
(TtC) rating system. Points (a) and (b) explain these concepts further:
(a) PiT: firms seek explicitly to estimate default risk over a fixed period, typically one year.
Under such an approach the increase in default risk in a downturn results in a general
tendency for migration to lower grades. When combined with the fixed estimate of the
long-run default rate for the grade, the result is a higher capital requirement. Where
data are sufficient, grade level default rates tend to be stable and relatively close to
the PD estimates; and
(b) TtC: firms seek to remove cyclical volatility from the estimation of default risk, by
assessing borrowers’ performance across the economic cycle. TtC ratings do not react
to changes in the cycle, so there is no consequent volatility in capital requirements.
Actual default rates in each grade diverge from the PD estimate for the grade, with
actual default rates relatively higher at weak points in the cycle and relatively lower at
.
26
strong points.
20
10.11 The PRA considers that most rating systems sit between these two extremes. Rating
ry
ua
philosophy is determined by the cyclicality of the risk drivers and criteria used in the rating
assessment and should not be confused with the requirement for grade level PDs to be ‘long-
an
run’. The calibration of even the most PiT rating system needs to be targeted at the long-run
1J
default rates for its grades; the use of long-run default rates does not convert such a system
m
ro
into one producing TtC ratings or PDs.

ef
10.12 Firms should choose a rating philosophy for each rating system, taking into account all
v
cti
of the following principles:

ffe
.E
(a) firms should assess whether the method used to quantify the risk parameter is
t2
adequate for the rating philosophy and ensure they understand the characteristics and
dynamics of the assignment of obligors or exposures to grades or pools (‘rating
ar
lp
assignment’) and the risk parameter estimates that result from the method used;
na
(b) firms should assess the adequacy of the resulting characteristics and dynamics of the
-fi
rating assignment and risk parameter estimates that result from the method used, with
ar
regard to their various uses, and should understand their impact on the dynamics and
Ne
volatility of capital requirements; and

(c) the rating philosophy should also be taken into account for back-testing purposes. On
the one hand, rating philosophies sensitive to economic conditions tend to estimate
PDs that are better predictors of each year’s default rates. On the other hand, rating
philosophies less sensitive to economic conditions tend to estimate PDs that are
closer to the average PD across the various states of the economy, but that differ from
observed default rates in years where the state of the economy is above or below its
average. Deviations between observed default rates and the long-run average default
rate of the relevant grade will therefore be more likely in rating systems less sensitive
to economic conditions. In contrast, migrations among grades will be more likely in
rating systems that are more sensitive to economic conditions. These patterns should
be taken into account when assessing the results of back-testing and, where relevant,
benchmarking analysis.
10.13 The PRA expects firms to apply rating philosophies consistently over time and to
analyse the appropriateness of each rating philosophy underlying the assignment of obligors
or exposures to grades or pools, taking into account all of the following:
(a) the design of risk drivers;

(b) migration across grades or pools; and
(c) changes in the yearly default rates of each grade or pool.
10.14 Where firms use different rating systems characterised by different rating philosophies,
.
26
they should use the information on the rating assignments or risk parameters estimates with
20
caution, especially when making use of rating information or default experience obtained
ry
from external rating agencies. Where firms use different rating systems with different
ua
characteristics, such as different rating philosophies or different levels of objectivity,
an
accuracy, stability, or conservatism, they should ensure that the rating systems have an
1J
appropriate level of consistency and that any differences between them are well understood.
m
Such understanding should at least enable a firm to define an appropriate way to combine or
ro
aggregate the information produced by the various rating systems when this is necessary
ef
according to the firm’s policies. Firms should have full understanding of the assumptions and
v
cti
potential inaccuracies arising from such a combination or aggregation.

ffe
10.15 The cyclicality of the rating system is a measure of its degree of responsiveness to
.E
economic changes and is inherently linked to rating philosophy. At one extreme, a fully
t2
cyclical PiT rating system would result in an economic downturn being picked up through
ar
lp
migration of exposures to lower rating grades and therefore there would be no increase in the
na
default rate within a grade. At the other extreme, a non-cyclical or TtC rating system does not
-fi
respond to an economic downturn with grade migration, but the default rate within a grade
ar
increases instead. The PRA expects firms to be aware of the cyclicality of their rating
Ne
systems to enable them to calibrate, monitor and stress test their systems. The PRA defines
cyclicality for a rating system as follows:
PDt − PDt−1
cyclicality% = ( ) ∙ 100
DR t − DR t−1
Where:
- 𝑃𝐷𝑡 means the long-run average PD at time t
- 𝐷𝑅𝑡 means the observed default rate at time t
Homogeneity of obligor grades or pools

10.16 In order to comply with the requirements of Articles 170(1) and 170(3)(c) of the Credit
Risk: Internal Ratings Based Approach (CRR) Part, firms should assess the homogeneity of
obligors or exposures assigned to the same grades or pools. In particular, grades and pools
should be defined in such a manner that each obligor within each grade or pool has a
reasonably similar risk of default and that significant overlaps of the distributions of the
default risk between grades or pools are avoided.
Use of external rating agency grades
.
10.17 The PRA expects a firm using external rating agency grades as the primary driver in an
26
IRB model to be able to demonstrate (and document) that it meets the following criteria:
20
ry
(a) the firm has its own internal rating scale;
ua
(b) the firm has systems and processes in place that allow it continuously to collect and
an
analyse all relevant information. The ‘other relevant information’ considered by the firm
1J
in accordance with Article 171(2) of the Credit Risk: Internal Ratings Based Approach
m
(CRR) Part reflects the information collected and analysed by the firm when extending
ro
credit to new or existing obligors;

ef
(c) the ‘other relevant information’ considered by the firm is included in an IRB model in a
v
cti
transparent and objective way and is subject to challenge. The PRA expects the firm
ffe
to be able to demonstrate what information was used and why, and how it was
.E
included; and if no additional information is included, to be able to document what

t2
information was discarded and why;

ar
(d) the development of final grades includes the following steps:

lp
(i) the firm takes into account all available information (eg external rating agency
na
grades and any ‘other relevant information’) prior to allocating obligors to internal
-fi
grades. The firm does not automatically assign obligors to grades based on the
ar
external rating agency grade;

Ne
(ii) any overrides are applied to these grades; and

(iii) the firm has systems and processes in place that allow it to continuously collect
and analyse final rating overrides;
(e) the grades to which obligors are assigned are reassessed at least annually and the
firm is able to demonstrate that the grades are reassessed on a more frequent than
annual basis when new relevant information becomes available (including how this is
done); and
(f) the firm can demonstrate that a modelling approach is being applied, both in terms of
the choice of the external rating agency grade as the primary driver and, where
information is found materially and consistently to add to the accuracy or predictive
power of the internal rating grade, that they have incorporated this information as an
additional driver. The PRA expects this work to be analytical (rather than entirely
subjective) and that it could form part of the annual independent review of the model.
10.18 If a firm does not have any additional information to add to the external ratings for a
significant part of a portfolio for which it applies the IRB approach the PRA expects that the
firm would not materially meet the requirements of the Credit Risk: Internal Ratings Based
Recognition of third-party support

10.19 The PRA expects that a firm that recognises third-party support (including parental
.
support) in the assignment of exposures to obligor grades should meet the following risk
26
management expectations:
20
ry
(a) the firm should perform a comprehensive assessment of the support provider’s
ua
willingness, capacity and likelihood to provide the support, including in instances
an
where the obligor is in financial distress; 1J
(b) the firm should document the assessment undertaken in point (a). The documentation
m
should include the rationale underpinning the firm’s assessment of the likelihood that
ro
the support will be provided; and

ef
(c) where the firm reflects explicit support arrangements in the assignment of exposures
v
cti
to obligor grades it should:

ffe
(i) ensure that the support has been confirmed by an appropriate and suitably senior
.E
representative of the support provider;

t2
(ii) reconfirm the support annually; and

ar
(iii) where the support arrangement has not been confirmed in writing, document a
lp
record of the conversation in which the support was confirmed.

na
-fi
10.20 Where a firm reflects parental support in the assignment of obligors or facilities to
ar
grades using a notched grading approach, the PRA considers that approaches which notch
Ne
down from the parent entity risk rating and notch up from the subsidiary standalone risk rating
are both acceptable in principle. The PRA expects that the firm should assign a standalone
rating to the subsidiary regardless of the type of notched grading approach used.
Parameter Substitution Method

10.21 The PRA expects that firms applying the Parameter Substitution Method in accordance
with Article 191A of the Credit Risk Mitigation (CRR) Part of the PRA Rulebook should
nonetheless collect and store information on the characteristics and performance of the
obligor and use this information in PD estimation where appropriate.
11: PD – calibration
Data requirements for the calculation of observed default rates

11.1 For the purpose of calculating the one-year default rate defined in CRR Article 4(1)(78),
firms should ensure the completeness of the quantitative and qualitative data, and other
information in relation to the denominator and numerator (as outlined in paragraphs 11.4 and
11.5) that are used for the calculation of the observed average default rate. In particular,
firms should ensure that at least the following data for the relevant observation period
.
referred to in paragraph 11.16 are properly stored and available:
26
20
(a) the criteria for identifying the relevant type of exposures covered by the PD model
ry
under consideration;
ua
(b) the criteria for identifying the calibration segments;
an
(c) the risk drivers used for risk differentiation. Where a newly relevant risk driver has
1J
been included in the model for which no historical data are available, firms should
m
make efforts to minimise missing data on risk drivers over time as outlined in
ro
paragraph 9.16(a), and apply an appropriate adjustment and a MoC in accordance

ef
with Chapter 9; and

v
cti
(d) all identification numbers of obligors and exposures relevant for default rate
ffe
calculation, taking into account situations where the identification number has changed
.E
over time, including changes due to restructuring of exposures.

t2
11.2 Exclusion of observations from the one-year default rate calculation should be
ar
lp
undertaken only in the following two situations:

na
(a) obligors wrongly included in the data set of defaults, as they did not default according
-fi
ar
to the criteria set out in Article 178 of the Credit Risk: Internal Ratings Based Approach
Ne
(CRR) Part (and further specified in SS3/24), should not be included in the numerator
of the one-year default rate; and
(b) obligors wrongly assigned to the rating model, despite not falling in the range of
application of that rating model, should be excluded from both the numerator and the
denominator of the one-year default rate.
11.3 Firms should document all data cleansing with respect to the one-year default rate
calculation and in particular:
(a) for non-retail PD models, a list of all observations within the data set that were
excluded according to paragraph 11.1, with a case-by-case justification; and
(b) for retail PD models, information on the reasons and quantity of exclusions of
observations made in accordance with paragraph 11.1.
Calculation of one-year default rates

11.4 For the purpose of calculating the one-year default rate referred to in CRR Article
4(1)(78), firms should ensure both of the following:
(a) that the denominator consists of the number of non-defaulted obligors with any credit
obligation observed at the beginning of the one-year observation period; in this
context a credit obligation refers to both of the following:
(i) any on-balance sheet item, including any amount of principal, interest, and fees;
and
(ii) any off-balance sheet item, including guarantees issued by the firm as a
.
guarantor;
26
(b) that the numerator includes all those obligors considered in the denominator that had
20
at least one default event during the one-year observation period.
ry
ua
11.5 When assigning the obligors or exposures to grades or pools for the purpose of the one-
an
year default rate calculation, firms should take overrides into account, but they should not
1J
reflect in this assignment any effect of applying the Parameter Substitution Method in
m
accordance with Article 191A of the Credit Risk Mitigation (CRR) Part, nor any ex-post
ro
conservative adjustments applied in accordance with paragraphs 19.1 to 19.8. Where the
ef
one-year default rate is calculated by rating grade or pool, the denominator should refer to all
v
cti
obligors assigned to a rating grade or pool at the beginning of the observation period. Where
ffe
the one-year default rate is calculated at the calibration segment level, the denominator
.E
should refer to all obligors assigned to the relevant calibration segment at the beginning of
t2
the observation period.

ar
lp
11.6 For the purposes of paragraphs 11.4 to 11.5, an obligor or exposure should be included
na
in the denominator and, where relevant, the numerator, in the event that the obligor or
-fi
exposure migrates to a different rating grade, pool, rating model, rating system, or approach
ar
for calculating credit risk capital requirements during the observation period, or where the
Ne
corresponding credit obligations were sold, written off, repaid or otherwise closed during the
observation period. Firms should analyse whether such migrations or sales of credit
obligations bias the default rate and, if so, they should reflect this in an appropriate
adjustment and application of an adequate MoC.
11.7 In any case, firms should ensure that each defaulted obligor is counted only once in the
numerator and the denominator of the one-year default rate calculation, even where the
obligor defaulted more than once during the relevant one-year period.
11.8 In order to choose an appropriate calculation approach in accordance with paragraph

11.10, firms should evaluate the observed one-year default rates within the historical
observation period at least quarterly.
Calculation of the observed average default rate

11.9 The observed average of one-year default rates (‘observed average default rate’) should
be calculated for each rating grade or pool and additionally for the type of exposures covered
by the relevant PD model as well as for any relevant calibration segment.
11.10 Firms should choose an approach to calculate the observed average default rate that
is either based on overlapping one-year time windows or based on non-overlapping one-year
time windows in light of a documented analysis. This analysis should include at least the
following:
(a) an analysis of possible bias due to the proportion of short-term and terminated
.
26
contracts that cannot be observed during the relevant one-year periods;
20
(b) an analysis of possible bias due to the specific calculation dates chosen;
ry
(c) for firms using overlapping one-year time windows, an analysis of potentially
ua
significant bias due to implicit over-weighting of the overlapping time period; and
an
(d) an analysis of potentially significant bias due to seasonal effects related to the
1J
chosen calculation dates.
m
ro
11.11 For the purposes of paragraphs 11.9 and 11.10, firms should calculate the observed
ef
average default rates as the arithmetic average of all one-year default rates calculated in
v
accordance with paragraphs 11.4 to 11.6.

cti
ffe
Long-run average default rate

.E
t2
11.12 In accordance with Articles 180(1)(a) and 180(2)(a) of the Credit Risk: Internal Ratings
ar
Based Approach (CRR) Part, the PRA expects that in order to estimate PDs that are long-run
lp
averages of one-year default rates for an obligor grade or pool, a firm should estimate
na
expected default rates for the grade or pool over a representative mix of good and bad
-fi
ar
economic periods, rather than simply taking the historic average of default rates actually
Ne
incurred by the firm over a period of years. The PRA expects firms to change a long-run
estimate when there is reason to believe that the existing long-run estimate is no longer
accurate but does not expect firms to automatically update an estimate to incorporate the
experience of additional years, as these periods may not be representative.
11.13 For the purpose of assessing whether a mix of good and bad economic periods is
representative, firms should take into account all of the following:
(a) the variability of all observed one-year default rates;

(b) the relative frequency of good and bad years as reflected by economic indicators that
are relevant for the type of exposures within the selected period; and
(c) significant changes in the economic, legal, or business environment within the mix of
periods.
11.14 The PRA expects that in defining a representative mix of good and bad economic
periods for UK residential mortgages (as referred to in paragraph 11.12), firms should
incorporate economic conditions equivalent to those observed in the UK during the early
1990s. The PRA has set this expectation in light of UK economic experience and may revise
it in the future as appropriate.
11.15 Firms should use a historical observation period that meets the requirements of Article
180(1)(h) or Article 180(2)(e) of the Credit Risk: Internal Ratings Based Approach (CRR) Part
as applicable. Where the historical observation period includes a representative mix of good
and bad economic periods, the long-run average default rate should be computed as the
observed average of the one-year default rates in that period.
.
26
11.16 Where the historical observation period does not reflect a representative mix of good
20
and bad economic periods, firms should apply the following:
ry
ua
(a) where insufficient bad years are included in the historical observation period, the
an
average of observed one-year default rates should be adjusted in order to estimate a
1J
long-run average default rate; and
m
(b) where bad years are over-represented in the historical observation period, the average
ro
of observed one-year default rates may be adjusted to estimate a long-run average

ef
default rate where there is a significant correlation between economic indicators

v
cti
referred to in paragraph 11.13(b) and the available one-year default rates.

ffe
11.17 Firms should ensure that, as a result of the adjustments referred to in paragraphs
.E
11.16(a) and 11.16(b), the adjusted long-run average default rate reflects a representative
t2
mix of good and bad economic periods.

ar
lp
11.18 In the exceptional case where the long-run average default rate is below the average
na
of all observed one-year default rates due to any adjustment made in accordance with
-fi
ar
paragraph 11.16, firms should compare their adjusted long-run average default rates with the
Ne
higher of the following:
(a) the observed average of the one-year default rates of the most recent five years; and
(b) the observed average of all available one-year default rates.
11.19 Where paragraph 11.18 applies, firms should justify the direction and magnitude of the
adjustment made in accordance with paragraph 11.16 in light of the comparison referred to in
paragraph 11.18. For this purpose, firms should also justify the MoC assessed in accordance
with Chapter 9. In addition, where the adjusted long-run average default rate is lower than the
higher of the two values referred to in paragraphs 11.18(a) and 11.18(b), firms should
specifically justify why these two values are not appropriate.
Calibration to the long-run average default rate

11.20 The PRA expects firms to have sound and well-defined processes in place which
ensure sound calibration and to include all of the following in their calibration process:
(a) quantitative calibration tests by rating grade or pool;

(b) quantitative calibration tests on calibration segment level; and
(c) supplementary qualitative analyses such as expert judgements on the shape of the
resulting obligor distribution, minimum obligor numbers per grade, and avoidance of
undue concentration in certain grades or pools.
11.21 Firms should store the calibration sample associated with each calibration segment
.
26
and describe these in the PD model documentation.
20
ry
11.22 Firms should conduct the calibration after taking into account any overrides applied in
ua
the assignment of obligors to grades or pools, and before the application of the MoC and the
an
floors to PD estimates referred to in Articles 160(1) and 163(1) of the Credit Risk: Internal
1J
Ratings Based Approach (CRR) Part. Where a ranking method or overrides policy has
m
changed over time, firms should analyse the effects of these changes on the frequency and
ro
scope of overrides and take them into account appropriately.

vef
11.23 The PRA considers that the process of grouping ranked obligors or exposures to
cti
grades or pools, in particular where firms conduct this grouping by identification of intervals of
ffe
score values reflecting a predefined PD level assigned to a grade of a master scale, may be
.E
performed during the calibration.

t2
ar
11.24 The PRA expects that firms should, taking into account the availability of data, the
lp
structure of the model and portfolio, as well as the business requirements, choose an
na
appropriate method to perform the calibration in accordance with the following principles:
-fi
ar
(a) for exposures to corporates and institutions, firms may choose one of the following
Ne
types of calibration:
(i) a calibration based on a mapping to the rating scale used by an external credit
assessment institution (ECAI) or similar organisation in accordance with Article
180(1)(f) of the Credit Risk: Internal Ratings Based Approach (CRR) Part; or
(ii) for a statistical default prediction model where the PDs are estimated as simple
averages of default probability estimates for individual obligors in a given grade or
pool in accordance with Article 180(1)(g) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part, a calibration at the level of appropriate calibration
segments of the relevant default probability estimates;
(b) for retail exposures, firms may choose a calibration based on total losses and LGDs
in accordance with Articles 180(2)(b) and 180(2)(d) of the Credit Risk: Internal
Ratings Based Approach (CRR) Part; and
(c) for corporate purchased receivables, firms may choose a calibration based on ELs
and LGDs in accordance with Articles 180(1)(b) and 180(1)(c) of the Credit Risk:
11.25 For the purpose of determining the PD estimates referred to in paragraph 11.24, the
calibration should consider either:
(a) the long-run average default rate at the level of grade or pool, in which case firms
should provide additional calibration tests at the level of the relevant calibration
segment; or
(b) the long-run average default rate at the level of the calibration segment, in which case
.
firms should provide additional calibration tests at the level of the relevant grades or
26
pools.
20
ry
11.26 Irrespective of which of the approaches mentioned in paragraph 11.25 firms choose,
ua
firms should assess the potential effect of the chosen calibration method on the behaviour of
an
PD estimates over time. 1J
m
11.27 For the purpose of determining PD estimates based on a mapping to an external rating
ro
agency scale as referred to in paragraph 11.24(a)(i), firms should base the default rates
ef
observed for the external organisation’s grades on a time series representative of the likely
v
range of variability of default rates for the grades and pools of the given portfolio.
cti
ffe
11.28 Where firms derive PD estimates from the estimates of losses and LGDs in
.E
accordance with Articles 161(2) and 180(2)(b) of the Credit Risk: Internal Ratings Based
t2
Approach (CRR) Part they should use a RDS that includes realised losses on all defaults
ar
identified during the historical observation period specified in accordance with paragraphs
lp
14.1 to 14.2 and relevant drivers of loss.

na
-fi
11.29 Firms may split exposures covered by the same PD model into as many different
ar
calibration segments as needed where one or more subsets of these exposures carry a
Ne
significantly different level of risk. For this purpose, firms should use relevant segmentation
drivers, and they should justify and document the use and scope of the calibration segments.
11.30 The PRA expects that, where scoring methods are used, firms should ensure that:
(a) where there is a change in the scoring method used, firms consider whether it is
necessary to recalculate scores of obligors or exposures based on the original data
set instead of using scores that were calculated based on previous versions of the
scoring method, and, where such recalculation is not possible, that firms assess
potential effects and take those effects into account via an appropriate increase in the
MoC applied to their PD estimates; and
(b) where a firm applies the approach in Article 180(1)(g) of the Credit Risk: Internal
Ratings Based Approach (CRR) Part, the PD estimates which are derived as a count
weighted average of individual PD estimates are adequate for relevant grades, as
determined by applying calibration tests to these estimates at grade level that are
based on one-year default rates representative of the likely range of variability of
default rates.
11.31 The PRA expects that the calibration should not influence the rank-ordering of obligors
or exposures within a calibration segment other than within each grade or pool.
11.32 In order to demonstrate compliance with Article 180 of the Credit Risk: Internal Ratings
.
Based Approach (CRR) Part, the PRA expects a firm to take into account the following
26
factors in understanding differences between their historic default rates and their PD
20
estimates, and in adjusting the calibration of their estimates as appropriate:
ry
ua
(a) the rating philosophy of the system and the economic conditions in the period over
an
which the defaults have been observed; 1J
(b) the number of defaults, as a low number is less likely to be representative of a long-
m
run average. Moreover, where the number of internal defaults is low, there is likely to
ro
be a greater need to base PDs on external default data as opposed to purely internal
ef
data;
v
cti
(c) the potential for under-recording of actual defaults; and

ffe
(d) the level of conservatism applied.

.E
t2
Calibration of residential mortgage portfolios

ar
lp
11.33 Article 180(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part
na
requires firms to estimate PDs by obligor grade from long-run averages of one-year default
-fi
rates. However, for some types of residential mortgages (‘low historical data’ portfolios) such
ar
as buy-to-let, self-certification and sub-prime, there may be an absence of, or insufficient,

Ne
relevant internal or external data over a representative economic cycle. For such exposures,
the PRA expects firms to model how book-level default rates in a given low historical data
portfolio would have performed under the economic conditions that would be experienced in
an economic cycle containing a representative mix of good and bad periods. The outputs of
this model should then be used in order to calibrate long-run average PDs for each rating
grade.
11.34 The PRA expects rating systems referred to in paragraph 11.33 above to result in long-
run average PDs that include an appropriate MoC. For each low historical data mortgage
portfolio, the PRA will undertake an assessment of whether the resultant degree of uplift in
PDs relative to comparable mortgages in a firm’s prime portfolio is sufficient.
11.35 The PRA considers that the amount of available data for non-UK mortgages varies by
jurisdiction. Where a firm has insufficient internal or external data to calibrate long-run
average PDs for these portfolios, it should apply the approach set out in paragraph 11.33.
For each portfolio of non-prime non-UK mortgages, where the approach in paragraph 11.33
has been applied, the PRA will assess whether the degree of uplift in PDs relative to
comparable mortgages in a firm’s prime portfolio for the jurisdiction in question is sufficient.
11.36 The PRA would not normally expect low historical data and prime portfolios to be
combined within the same rating system as it is challenging for firms to demonstrate a
meaningful differentiation of risk and accurate and consistent quantitative estimates of risk in
such cases. In the event that a firm is able to demonstrate to the PRA that such an approach
.
26
is appropriate, the PRA expects the low historical data subset of the rating system to meet
20
the expectations contained within paragraphs 11.33 to 11.35.
ry
11.37 The PRA expects that PDs for portfolios in run-off are calibrated to reflect how a firm’s
ua
existing portfolio would perform in an economic cycle containing a representative mix of good
an
and bad periods. Where a firm has insufficient internal or external data to calibrate PDs, the
1J
techniques outlined in paragraph 11.33 should be applied.
m
ro
11.38 The PRA considers that firms often have difficulty in practice in understanding the
ef
cyclicality of their residential mortgage rating systems (further detail on cyclicality can be
v
cti
found in paragraph 10.15). To mitigate the risk of under-calibration of these rating systems
ffe
due to inaccurate estimation of their cyclicality, the PRA expects that when firms calibrate
.E
their residential mortgage rating systems by uplifting internal observed default rates to a long-
t2
run average, they should do so on the assumption that the cyclicality of each rating system is
ar
no more than 30% in those years where grade level internal observed default rates are not
lp
available. This cyclicality cap is the PRA’s expectation of what firms should assume is the
na
maximum level of cyclicality when imputing missing historical default rates. If 30% of the
-fi
change in portfolio default rates comes from grade migration, the remaining 70% would come
ar
from change in default rates within grades. Therefore, when calibrating the long-run average
Ne
default rates to assign to each rating grade, the PRA expects firms to assume that at least
70% of the portfolio change in default rate reflects grade level changes in default rate. This
level reflects the PRA’s current view of an appropriately conservative assumption for rating
system cyclicality in light of recent experience. This expectation may be adjusted by the PRA
if it judges that there has been a change in the risk of under-calibration.
11.39 When a firm is calibrating or recalibrating a residential mortgage rating system using
internal observed default rates taken predominantly from a downturn period (ie the firm is
reducing the internal observed default rates to a long-run average), the PRA’s expectation of
a 30% cap on cyclicality does not apply. Instead, firms should determine an appropriately
conservative adjustment to allow for uncertainty in their estimates of cyclicality in such
circumstances.
11.40 As an alternative to the expectations on risk mitigation methodology in paragraph

11.38, the PRA may be satisfied that a firm has taken steps to mitigate these risks if the
residential mortgage PD rating system meets the following standards:
(a) the firm is able to convincingly articulate how the risk drivers in a rating system will
generate the migration into other grades, scores, or ratings assumed in its estimates
of cyclicality;
(b) the firm is able to demonstrate that the assumed changes have occurred in practice
across an economic cycle; and
(c) the above analysis is able to isolate the impact on the existing exposures covered by
the rating system from changes in composition of the portfolio over the period being
.
26
analysed.
20
11.41 The PRA has found in its experience that for residential mortgage portfolios, firms
ry
using TtC approaches are unable to distinguish sufficiently between movements in default
ua
rates that result from cyclical factors and those that result from non-cyclical factors, and this
an
results in risks not being sufficiently captured. The PRA therefore expects that firms should
1J
not use fully TtC approaches for residential mortgage portfolios.
m
ro
11.42 Highly cyclical PiT models do not always adequately capture risks over the long-run,
ef
and this is particularly an issue for residential mortgage portfolios where default rates are
v
cti
highly cyclical. The PRA therefore expects firms not to use an artificial highly cyclical PiT
ffe
approach achieved through dynamic recalibration of the score to PD relationship in their

.E
application and behavioural scorecards for residential mortgage models.

t2
11.43 The PRA expects recalibrations of rating systems applying the cyclicality assumptions
ar
set out in paragraph 11.38 to be rare and to be symptomatic of failures of the rating system’s
lp
na
assumptions rather than part of rating system design. For these purposes, any calculation
-fi
mechanism embedded in a rating system that changes the PD applied to exposures with a
ar
given set of characteristics should be treated as a recalibration. The PRA expects that any
Ne
recalibration of such a rating system would include:
(a) a robust assessment of the cyclicality of the rating system;

(b) a robust assessment and explanation of the cause of the need to recalibrate, including
whether it is due to changes in default risk that are not purely related to changes in the
cycle. This should include an assessment of the firm’s own lending profile, its historical
performance, wider industry performance against historical levels, and changes in
economic factors; and
(c) a review of the appropriateness of undertaking a recalibration by an independent
validation function.
Retail exposures: obligor level definition of default

11.44 Where a firm has not chosen to apply the definition of default at the level of an
individual credit facility in accordance with Article 178(1) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part, the PRA expects it to ensure that the PD associated with
unsecured exposures is not understated as a result of the presence of any collateralised
exposures.
11.45 The PRA expects the PD of a residential mortgage would typically be lower than the
PD of an unsecured loan to the same borrower.
Retail exposures: facility level definition of default
.
26
20
11.46 Where a firm chooses to apply the definition of default at the level of an individual
ry
credit facility in accordance with Article 178(1) of the Credit Risk: Internal Ratings Based
ua
Approach (CRR) Part and a customer has defaulted on a facility, then default on that facility
an
is likely to influence the PD assigned to that customer on other facilities. The PRA expects
1J
firms to take this into account in its estimates of PD.
m
ro
Low default portfolios

vef
cti
11.47 The PRA expects a firm to estimate PDs for a rating system in accordance with this
section where a firm’s internal experience of defaults for that rating system is 20 or fewer,
ffe
.E
and reliable estimates of PD cannot be derived from external sources of default data
t2
including the use of market price related data. In order to estimate PDs for all exposures
ar
covered by such a rating system, the PRA expects firms to:

lp
(a) use a statistical technique to derive the distribution of defaults implied by the firm’s
na
experience, estimating PDs (the ‘statistical PD’) from the upper bound of a
-fi
ar
confidence interval set by the firm in order to produce conservative estimates of PDs
Ne
in accordance with Article 179(1)(f) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part;
(b) use a statistical technique to derive the distribution of default which takes account, as
a minimum, the following modelling issues:
(i) the number of defaults and number of obligor years in the sample;
(ii) the number of years from which the sample was drawn;
(iii) the interdependence between default events for individual obligors;
(iv) the interdependence between default rates for different years; and
(v) the choice of the statistical estimators and the associated distributions and
confidence intervals;
(c) further adjust the statistical PD to the extent necessary to take account of the
following:
(i) any likely differences between the observed default rates over the period
covered by the firm’s default experience and the long-run PD for each grade
required by Articles 180(1)(a) and 180(2)(a) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part; and
(ii) any other information that indicates (taking into account the robustness and
cogency of that information) that the statistical PD is likely to be an inaccurate
estimate of PD.
11.48 The PRA expects firms to only take into account defaults that occurred during periods
that are relevant to the validation of the rating system in accordance with the requirements
when determining whether there are 20 defaults or fewer.
.
26
20
ry
ua
an
1J
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
12: LGD – General expectations and model

development
General expectations – LGD estimation methodologies

12.1 Firms using the AIRB approach should assign an LGD estimate to each non-defaulted
exposure and an estimate of LGD in-default and BEEL to each defaulted exposure within the
range of application of the rating system. Firms should estimate LGDs for all facility grades of
.
the distinct facility rating scale or for all pools that are incorporated in the rating system. For
26
the purpose of LGD estimation, firms should treat each defaulted facility as a distinct default
20
observation, unless more than one independent default was recognised on a single facility
ry
which does not meet the criteria of paragraph 12.2.
ua
an
12.2 For the purpose of LGD estimation, with regard to defaults recognised on a single
1J
facility, where the time between the moment of the return of the exposure to non-defaulted
m
status and the subsequent classification as default is shorter than nine months, firms should
ro
treat such an exposure as having been constantly defaulted from the first moment when the
ef
default occurred. Firms may specify a period longer than nine months for the purpose of
v
cti
considering two subsequent defaults as a single default in the LGD estimation if this is
ffe
adequate for the specific type of exposures and reflects the economic meaning of the default
.E
experience.
t2
12.3 Firms should estimate their own LGDs based on their own loss and recovery
ar
lp
experience, as it is reflected in historical data on defaulted exposures. Firms may supplement

na
their own historical data on defaulted exposures with external data. In particular, firms should
-fi
not derive their LGD estimates only from the market prices of financial instruments, including,
ar
but not limited to, marketable loans, bonds, or credit default instruments, but they may use
Ne
this information to supplement their own historical data.
12.4 Where, in the case of retail exposures and purchased corporate receivables, firms
derive LGD estimates from realised losses and appropriate estimates of PDs in accordance
with Articles 161(2) and 181(2)(a) of the Credit Risk: Internal Ratings Based Approach (CRR)
Part, they should ensure that:
(a) the process for estimating total losses meets the requirements of Article 179 of the
Credit Risk: Internal Ratings Based Approach (CRR) Part, and the outcome is
consistent with the concept of LGD as set out in Article 181(1)(a) of the Credit Risk:
Internal Ratings Based Approach (CRR) Part and this SS, in particular with the
concept of economic loss as specified in paragraphs 13.1 to 13.23; and
(b) the process for estimating PD meets the requirements of Articles 179 and 180 of the
Credit Risk: Internal Ratings Based Approach (CRR) Part as well as the expectations
specified in Chapter 10 – PD – model development and Chapter 11 – PD – calibration
of this SS.
12.5 The PRA considers that an LGD model can contain several different methods,
especially with respect to different types of collateral, which are then combined to arrive at an
LGD for a given facility.
12.6 Firms should be able to demonstrate that the methods that they choose for the purpose
of LGD estimation are appropriate to their activities and the type of exposures to which the
.
estimates apply, and they should be able to justify the theoretical assumptions underlying
26
those methods. The methods used in the LGD estimation should in particular be consistent
20
with the collection and recovery policies adopted by the firm and should take into account
ry
possible recovery scenarios as well as potential differences in the legal environment in
ua
relevant jurisdictions.
an
1J
12.7 The PRA expects that the methods used by the firm when estimating LGD consider any
m
downturn effect, the length of data series used, the MoC, human judgement and, where
ro
applicable, the choice of risk drivers, should be appropriate to the type of exposures to which
ef
they are applied.

v
cti
ffe
General expectations – use of SA and FIRB approach

.E
parameters for LGD estimation

t2
ar
12.8 The PRA expects that firms which make reference to SA or FIRB approach parameters
lp
in an IRB model for LGD estimation should provide appropriate justification. The PRA
na
considers that the provisions of the Credit Risk: Internal Ratings Based Approach (CRR) Part
-fi
and this SS relating to modelling standards apply in such circumstances.

ar
Ne
General expectations – use of LGD Modelling Collateral Method

12.9 The PRA considers that, where a firm is recognising non-financial collateral using the
LGD Modelling Collateral Method, for the purposes of Article 169B(1) of the Credit Risk:
Internal Ratings Based Approach (CRR) Part it will only have sufficient data to model a given
type of collateral in a particular jurisdiction where it has 20 or more relevant data points of
recovery values for that type of collateral. Where the firm does not have 20 or more relevant
data points for that type of collateral, the PRA expects that the firm should apply the
methodology set out in Article 169B(2) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part.
12.10 For the purpose of the assessment of relevant data points referred to in paragraph
12.9 the PRA expects that:
(a) the firm should ensure that each data point is independent, representative, accurate;
and
(b) the firm should usually not treat recoveries from collateral located in a different
jurisdiction than that of the collateral itself as relevant, but it may exceptionally do so if it
can evidence that recovery values from that type of collateral in the two jurisdictions are
not materially different.
Data requirements for LGD estimation
.
26
12.11 For the purpose of LGD estimation, firms should use an RDS covering all of the
20
following items:
ry
ua
(a) all defaults identified during the historical observation period as specified in
an
accordance with paragraphs 14.1 to 14.2; 1J
(b) all data necessary for calculating realised LGDs in accordance with paragraphs 13.1
m
to 13.23; and
ro
(c) relevant factors that can be used to group the defaulted exposures in meaningful ways
ef
and relevant drivers of loss, including their values at the moment of default and at
v
cti
least within the year before default when available.

ffe
12.12 Firms should include in the RDS information on the results of the recovery processes,
.E
including recoveries and costs, related to each individual defaulted exposure. For this
t2
purpose, firms should include:

ar
lp
(a) information on the results of incomplete recovery processes until the reference date
na
for LGD estimation;

-fi
ar
(b) information on the results of recovery processes at portfolio level, where such
Ne
aggregation of the information is justified, and in particular in respect of indirect costs

and in the case of the sale of a portfolio of credit obligations; and
(c) information on external or pooled data used in the estimation of LGDs.
12.13 The RDS should contain at least the following information:
(a) obligor-related, transaction-related and institution-related risk characteristics, as well

as external factors as referred to in paragraph 12.25 that are potential risk drivers at
the relevant reference dates specified in paragraph 12.27;
(b) moment (date) of default;
(c) all default triggers that have occurred, including both past due events and unlikeliness
to pay events, even after the identification of default; in the case of exposures subject
to distressed restructuring the amount by which the financial obligation has diminished
calculated in accordance with SS3/24 ‘Credit risk definition of default’;
(d) the outstanding amount of the exposure at the moment of default including principal,
interest, and fees;
(e) the amounts and timing of the additional drawings after default;
(f) the amounts and timing of write-offs;
(g) to the extent that the firm is applying the LGD Modelling Collateral Method in
accordance with Article 169A(1)(a) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, the value of collateral associated with the exposure and, where
applicable, the type of valuation, date of valuation, a flag of whether the collateral has
been sold and the sale price;
.
26
(h) information on any dependence between the risk of the obligor and the risk of the
20
collateral or collateral provider;
ry
(i) to the extent that the firm is applying the LGD Adjustment Method in accordance with
ua
Article 183 of the Credit Risk: Internal Ratings Based Approach (CRR) Part, the types,
an
amounts, and maturities of unfunded credit protection including the specification and
1J
credit quality of the protection provider;
m
(j) the amounts, timing, and sources of recoveries;
ro
(k) the amounts, timing, and sources of direct costs associated with recovery processes;
ef
(l) a clear identification of the type of termination of the recovery process;

v
(m)where applicable, currency mismatches between two or more of the following

cti
elements: the currency unit used by the institution for financial statements, the
ffe
underlying obligation, any funded or unfunded credit protection, and any cash flows
.E
from the liquidation of the obligor’s assets; and

t2
(n) amount of realised loss.

ar
lp
12.14 Firms applying the LGD Modelling Collateral Method in accordance with Article
na
169A(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part may use various
-fi
ar
methods for the valuation of collateral in the form of immovable property. Where firms use
Ne
valuation approaches with regard to immovable properties that secure exposures included in
the range of application of a certain rating system, they should collect and store in the RDS
the information on the type of valuation and they should use this information consistently in
the LGD estimation and in the application of LGD estimates.
12.15 Where firms derive LGD estimates from realised losses and appropriate estimates of
PDs in accordance with Articles 161(2) and 181(2)(a) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part, they should use an RDS that includes realised losses on all
defaults identified during the historical observation period specified in accordance with
paragraphs 14.1 to 14.2 and relevant drivers of loss.
12.16 Where aggregated information is collected and stored, firms should develop an
appropriate methodology for the allocation of recoveries and costs to individual defaulted
exposures and should apply this methodology consistently across exposures and over time.
The PRA expects firms to demonstrate that the process of allocation of recoveries and costs
is effective and that it does not lead to biased LGD estimates.
12.17 The PRA expects firms to demonstrate that they collect and store in their databases all
information required to calculate direct and indirect costs. All material indirect costs should be
allocated to the corresponding exposures. This cost allocation process should be based on
the same principles and techniques that firms use in their own cost accounting systems. For
the purpose of indirect cost allocation, firms may use methods based on exposure weighted
.
26
averages, or statistical methods based on a representative sample within the population of
20
defaulted obligors or facilities.
ry
ua
12.18 Firms should take reasonable steps to recognise the sources of the recovery cash
an
flows and allocate them appropriately to the specific collateral or unfunded credit protection
1J
that has been realised. Where the source of the cash flows cannot be identified, firms should
specify clear policies for the treatment and allocation of such cash flows, including in respect
m
ro
of whether these are classified as arising from:

ef
(a) collateral or unfunded credit protection that the firm recognises when using the LGD
v
cti
Modelling Collateral Method or the LGD Adjustment Method in accordance with Article
ffe
169A(1)(a) or Article 183 of the Credit Risk: Internal Ratings Based Approach (CRR)
.E
Part;
t2
(b) collateral or unfunded credit protection which the firm does not recognise; and
ar
(c) other sources;

lp
na
in a way that results in LGDs being estimated in a conservative manner.

-fi
ar
Recoveries from collateral

Ne
12.19 For the purpose of applying the LGD Modelling Collateral Method in accordance with
Article 169A(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, or not
recognising the effect of collateral in LGD estimates where the LGD Modelling Collateral
Method is not applied, a firm should recognise recoveries as stemming from collateral in all of
the following situations:
(a) the collateral is sold by the obligor and the obtained price has been used to cover
parts or all of the outstanding amount of the defaulted credit obligation;
(b) the collateral is repossessed or sold by the firm, the parent undertaking, or any of its
subsidiaries on behalf of the firm;
(c) the collateral is sold in a public auction of the property by court order or in a similar
procedure in accordance with the applicable legal framework;
(d) the credit obligation is sold together with the collateral and the sale price for the credit
obligation included the existing collateral;
(e) in the case of leasing, the leasing object is sold by the firm; and
(f) the collateral is realised by any other method that is eligible under the legal framework
of the relevant jurisdiction.
12.20 For the purpose of paragraph 12.19(b) firms should determine the value of
repossession as the value by which the credit obligation of the obligor has been diminished
as a result of the repossession of the collateral, and, where relevant, with which the
.
26
repossessed collateral was recorded as an asset on the balance sheet of the firm. Where
20
these values are different, firms should consider the lower of the two to be the value of
repossession. The value of repossession should be considered a value of recovery at the
ry
ua
date of repossession and should be included in the calculation of the economic loss and
an
realised LGD in accordance with paragraphs 13.1 to 13.23.
1J
12.21 Firms should consider whether the value of repossession adequately reflects the value
m
of the repossessed collateral, consistently with any established internal requirements for
ro
collateral management, legal certainty, and risk management. Where the collateral
ef
repossessed meets the criteria for high quality liquid assets at Level 1, as defined in Article
v
cti
10 of the Liquidity Coverage Ratio (CRR) Part of the PRA Rulebook, firms may take into
ffe
account directly as a realised recovery the market value of the collateral at the time of the
.E
repossession. In all other cases, firms should apply an appropriate haircut to the value of
t2
repossession and include in the calculation of economic loss a recovery in the amount of the
ar
value of repossession after applying the appropriate haircut. Firms should estimate this
lp
haircut taking into account all of the following conditions:

na
-fi
(a) the haircut should reflect possible errors in the valuation of the collateral at the
ar
moment of repossession, taking into account the type of valuation available at the
Ne
moment of repossession, the date it was performed, and the liquidity of the market for
this type of asset;
(b) the haircut should be estimated with the assumption that the firm intends to sell the
repossessed collateral to an independent third-party and should reflect the potential
price that could be achieved from such sale, the costs of the sale, and the discounting
effect for the period from the sale to the moment of repossession, taking into account
the liquidity of the market for this type of assets;
(c) where there are observations available regarding the repossession and subsequent
sales of similar types of collateral, the estimation of the haircut should be based on
these observations and should be regularly back-tested; for this purpose firms should
take into account all of the following:
(i) the difference between the value of repossession and the sale price, especially
where there were no significant changes in market and economic conditions
between the moment of the repossession and the moment of the sale;
(ii) any income and costs related to this asset that were observed between the date
of repossession and the moment of the sale;
(iii) discounting effects; and
(iv) whether the firm repossessed the collateral with the intention of immediate sale or
whether another strategy was adopted;
(d) where historical observations regarding the repossessions and subsequent sales of
similar types of collateral are not available, the estimation of the haircut should be
based on a case-by-case assessment, including the analysis of current market and
.
26
economic conditions; and
20
(e) the fewer data a firm has on previous repossessions, and the less liquid the market for
ry
the given type of assets is, the more uncertainty is attached to the resulting estimates,
ua
which should be adequately reflected in the MoC in accordance with paragraphs 9.6 to
an
9.17. 1J
12.22 In any case, the repossession of collateral should be recognised at the moment of
m
ro
repossession and should not prevent the firm from closing the recovery process in
ef
accordance with paragraph 14.7.

v
cti
12.23 Any sale of credit obligations in accordance with paragraph 12.19(d) should be
ffe
included in the LGD estimation in a manner appropriate to the LGD estimation methodology
.E
taking into account all of the following conditions:

t2
ar
(a) where firms regularly sell credit obligations as part of their recovery processes, they
lp
should appropriately reflect the observations related to credit obligations subject to

na
such sales in the model development process;

-fi
(b) firms should not treat recoveries from the sales of secured credit obligations as
ar
recoveries realised without the use of collateral; and

Ne
(c) in any case, firms should include all observations, including where there are sales of
credit obligations, in the calculation of long-run average LGD.
12.24 In accordance with paragraph 12.19(f), firms may recognise other methods of realising
collateral that are recognised under the applicable legal framework. When recognising such
other methods of realising collateral, firms should take into account the fact that the collateral
may take various forms and that various forms of collateral may be related to the same asset.
Where different forms of collateral refer to the same asset but the realisation of one form of
collateral does not decrease the value of the other, firms should consider them separate
collateral in the process of LGD estimation. In particular, firms should recognise separately
the form of collateral which gives a right to repossess or sell the asset (such as a mortgage)
and the form of collateral which gives a right to collect cash flows generated by the asset
(such as a cession of rent or fees).
Risk drivers
12.25 Firms should identify and analyse potential risk drivers that are relevant to their specific
circumstances and to the specific characteristics of the type of exposures covered by the
rating system. Potential risk drivers analysed by firms should include in particular the
following:
(a) transaction-related risk characteristics, including:
.
(i) in all cases: type of product, seniority, seasoning, recovery procedures, and
26
exposure size;
20
(ii) where the firm is applying the LGD Modelling Collateral Method in accordance
ry
with Article 169A(1)(a) of the Credit Risk: Internal Ratings Based Approach
ua
(CRR) Part: type of collateral, geographical location of the collateral, and loan to
value (LTV); and
an
1J
(iii) where the firm is applying the LGD Adjustment Method in accordance with Article
m
183 of the Credit Risk: Internal Ratings Based Approach (CRR) Part: unfunded
ro
credit protection;
ef
(b) obligor-related risk characteristics, including, where applicable, size, capital

v
cti
structure, geographical region, industrial sector, and line of business;

ffe
(c) institution-related factors, including internal organisation and internal governance,

.E
relevant events such as mergers, and existence of specific entities within the group
t2
dedicated to recoveries; and

ar
(d) external factors, including interest rates, legal framework, and other factors
lp
influencing the expected length of the recovery process.

na
-fi
12.26 In respect of seasoning effects for retail exposures, the PRA expects that firms should
ar
analyse the representativeness of the age of the facilities (in terms of time from the date of
Ne
default) in the data used to derive LGD estimates. The PRA considers that recovery rates
reach a low point several years after default for some portfolios in some jurisdictions, and as
a result, the PRA expects firms to adjust their estimates with an adequate MoC, in
accordance with Chapter 9, should they identify any shortcoming in the model’s ability to
accurately reflect seasoning effects.
12.27 Firms should analyse risk drivers not only at the moment of default but also at least
within a year before default. Firms should use a reference date for a risk driver that is
representative of the realisation of the risk driver within a year before default. When choosing
the appropriate reference date for a risk driver, firms should take into account its volatility
over time. Firms should apply these practices also with regard to the reference date of the
valuation of collateral; the value of the collateral at the reference date should not reflect the
impact of the decrease in credit quality of the exposure shortly before default.
12.28 The PRA expects that, for the purpose of the application of LGD estimates, firms
should specify or calculate the risk drivers in the same way as they are specified or
calculated for the purpose of modelling LGD.
Eligibility of collateral
12.29 In accordance with Article 169A(2) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part, firms using the LGD Modelling Collateral Method must only take into account in
.
their LGD estimates the existence of types of collateral for which they have established
26
internal requirements for collateral management, operational procedures, legal certainty, and
20
risk management that are generally consistent with those applicable under the FIRB
ry
approach. Where types of collateral are not eligible under the FIRB approach, the PRA
ua
expects that firms should only take such collateral into account in their LGD estimates if the
an
criteria in Article 169A(2) of the Credit Risk: Internal Ratings Based Approach (CRR) Part are
1J
met and if they have established policies and procedures relating to collateral valuation that
m
are appropriate for the type of collateral.

ro
ef
12.30 Firms using the LGD Modelling Collateral Method in accordance with Article 169A(1)(a)
v
cti
of the Credit Risk: Internal Ratings Based Approach (CRR) Part should take into account
ffe
information on all main types of collateral that are used within the scope of application of the
.E
LGD model as either a risk driver or segmentation criterion. Firms should clearly define in
t2
their internal policies the main and other types of collateral used for the type of exposures
ar
covered by the rating system, and should ensure that, to the extent that LGD estimates take
lp
into account the existence of collateral, the policies regarding the management of these types
na
of collateral comply with the requirements set out in Article 169A(2) of the Credit Risk:
-fi

ar
Ne
12.31 The PRA expects that collateral which does not meet the requirements of Article
169A(2) of the Credit Risk: Internal Ratings Based Approach (CRR) Part should not be
included as a risk driver in the LGD estimation and the cash flows received from such
collateral should be disregarded.
Inclusion of collateral in LGD estimation when applying the LGD

Modelling Collateral Method
12.32 For the purpose of LGD estimation, firms using the LGD Modelling Collateral Method in
accordance with Article 169A(1)(a) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part may group types of collateral that are homogeneous in terms of recovery
patterns, taking into account both the average time of the collection process and the recovery
rates on these types of collateral.
12.33 The PRA expects that the approach developed by firms to include the effect of
collateral in the LGD estimation should meet all of the following criteria:
(a) where firms estimate separate recovery rates for specific types of collateral, they
should avoid a bias that may stem from including in the estimation sample the
observations where the exposure was secured by only a part of the value of the
collateral. For this purpose, firms should take reasonable steps to obtain data on the
total value of the collateral and total sale price of the collateral and include this
.
information in the estimation where it is available;
26
(b) where firms estimate separate recovery rates for specific types of collateral they
20
should recognise and include in this estimation direct costs related to the collection on
ry
each of these specific types of collateral separately as well;
ua
(c) where firms estimate separate recovery rates for specific types of collateral they
an
should include in this estimation all recoveries realised from a specific type of
1J
collateral including those realised on exposures where the realisation of the collateral
m
has been completed but the overall recovery process has not yet been closed;
ro
(d) where the same collateral covers several exposures, firms should specify an adequate
ef
allocation methodology in order to avoid double counting of collateral; the allocation

v
cti
methodology should be consistent across the methodology used for LGD estimation,
ffe
the application of the LGD estimates, and the methodology used for accounting
.E
purposes;
t2
(e) the estimates should not be based solely on the estimated market value of the
ar
collateral, but they should also take into account the realised recoveries from past
lp
liquidations and the potential inability of a firm to gain control and liquidate the
na
collateral. For this purpose, firms should take into account in the estimation those
-fi
historical observations where the collateral could not be realised or where the recovery
ar
process was longer than expected, due to an inability or difficulty in gaining control of
Ne
or liquidating the collateral. Where firms estimate the recovery rates related to specific
types of collateral, they should take into account the time between the moment of
default and the time when the cash flows related to the collection on these types of
collateral have been received and should include in the estimation those observations
where the collateral has not been realised as a result of inability to gain control;
(f) the estimates should take into account the potential decreases in collateral value from
the point of LGD estimation to the eventual recovery, in particular those resulting from
changes in the market conditions, the state and age of the collateral and, where
relevant, currency fluctuations. Where firms have experienced decreases in values of
collateral, and these are already reflected in observed recoveries, no further
adjustments to LGD estimates based on these observations should be made. Where
potential decreases in values of collateral are not reflected in historical observations,
or where firms predict further, potentially more severe decreases in the future, they
should be included in the quantification of LGD estimates by means of an appropriate
adjustment based on forward-looking expectations. However, the LGD estimates
should not be adjusted to take into account any potential increases in collateral value;
and
(g) the estimates should take into account, in a conservative manner, the degree of
dependence between the risk of the obligor and the risk of the diminishing value of the
collateral as well as the cost of liquidating the collateral.
Unfunded credit protection
.
26
12.34 To the extent that firms use the LGD Adjustment Method in accordance with Article
20
183 of the Credit Risk: Internal Ratings Based Approach (CRR) Part to take into account the
existence of unfunded credit protection, firms should specify their criteria and methodology
ry
ua
for recognising and including protection in the form of guarantees and credit derivatives in
an
LGD estimates.
1J
12.35 The PRA expects that firms using the LGD Adjustment Method in accordance with
m
Article 183 of the Credit Risk: Internal Ratings Based Approach (CRR) Part should have clear
ro
policies for assessing the effects of unfunded credit protection on risk parameters. A firm’s
ef
policies should be consistent with its internal risk management practices and should reflect
v
cti
the requirements of Article 183 of the Credit Risk: Internal Ratings Based Approach (CRR)
ffe
Part and the expectations set out in this SS.

.E
t2
12.36 Where firms apply the LGD Adjustment Method in accordance with Article 183 of the
ar
Credit Risk: Internal Ratings Based Approach (CRR) Part, they should consider, and, if
lp
relevant, take the following elements into account in the LGD estimates in a conservative
na
manner:
-fi
ar
(a) any currency mismatch between the underlying obligation and the unfunded credit
Ne
protection;
(b) the degree to which the protection provider’s ability to fulfil the contractual obligation
under the unfunded credit protection agreement is correlated with the obligor’s ability
to repay; and
(c) the defaulted status of the protection provider and its resulting reduced ability to fulfil
the contractual obligation under the unfunded credit protection.
12.37 The PRA expects that, where a firm is not applying the LGD Adjustment Method in
accordance with Article 183 of the Credit Risk: Internal Ratings Based Approach (CRR) Part,
including where it is applying either the Risk Weight Substitution Method or the Parameter
Substitution Method in accordance with Article 191A of the Credit Risk Mitigation (CRR) Part,
it should estimate LGDs as if there was no unfunded credit protection. The PRA expects that
for this purpose firms should apply the following principles:
(a) cash flows received from the guarantor should not be taken into account;
(b) cash flows received from funded credit protection associated with the exposure may
be taken into account in respect of the part of the exposure covered by the funded
credit protection;
(c) indirect costs should be taken into account in line with the principles and techniques
that firms use in their own cost accounting systems;
(d) direct costs that are directly linked to the exercising of the unfunded credit protection
need not be taken into account, but all other direct costs should be taken into account;
.
26
and
20
(e) direct costs relating to the realisation of funded credit protection should be taken into
account in respect of the part of the exposure covered by the funded credit protection.
ry
ua
Funded credit protection securing unfunded credit protection
an
1J
obligations
m
ro
12.38 Criteria for recognising funded credit protection that secures unfunded credit protection
ef
are set out in the flowchart in Part 4 of Appendix 1 of the Credit Risk Mitigation (CRR) Part.
v
For exposures subject to the AIRB approach, the reference to Article 201 of the Credit Risk
cti
Mitigation (CRR) Part does not apply where direct exposures to the protection provider would
ffe
also be subject to AIRB approach, unless the firm is applying the Parameter Substitution
.E
Method in accordance with Article 191A of the Credit Risk Mitigation (CRR) Part. Where the
t2
reference to Article 201 does not apply, the PRA expects however that firms should only
ar
lp
recognise the unfunded credit protection where this meets any guarantor eligibility criteria
na
that they have set in accordance with the requirement in Article 183(1)(a) of the Credit Risk:
-fi

ar
Ne
12.39 Where a firm is recognising the funded credit protection, the PRA expects firms to
reflect the funded credit protection in their LGD model only to the extent to which the benefit
that can be derived from the collateral is not limited by the value of the guarantee. For this
purpose, where a firm is not applying the LGD Adjustment Method in accordance with Article
183 of the Credit Risk: Internal Ratings Based Approach (CRR) Part, the value of the
guarantee should be determined in accordance with Article 233 of the Credit Risk Mitigation
(CRR) Part, further adjusted for any maturity mismatch as laid down in Articles 237 to 239 of
the Credit Risk Mitigation (CRR) Part. The funded credit protection should be reflected in
LGD floors consistently with how it would be reflected under Articles 230 or 231 of the Credit
Risk Mitigation (CRR) Part if the firm was applying the Foundation Collateral Method
following the application of Article 191A(2)(e) of the Credit Risk Mitigation (CRR) Part.
Homogeneity of facility grades or pools

12.40 Firms should assess the homogeneity of exposures assigned to the same grades or
pools based on the data in the RDS, and they should ensure in particular that grades are
defined in such a manner that individual grades are sufficiently homogeneous with respect to
loss characteristics.
Treatment of cures
12.41 Where firms wish to include cures in their LGD estimates, the PRA expects them to do
so on a cautious basis with reference to both their current experience and how this is
.
expected to change in downturn conditions. In particular, this involves being able to articulate
26
clearly both the precise course of events that will allow such cures to take place and any
20
consequences of such actions for other elements of their risk quantification. For example:
ry
ua
(a) where cures are driven by the firm’s own policies, the PRA expects firms to consider
an
whether this is likely to result in longer realisation periods and larger forced sale
1J
discounts for those exposures that do not cure, and higher default rates on the book
m
as a whole, relative to those that might be expected to result from a less

ro
accommodating attitude. To the extent feasible, the PRA expects cure assumptions in
ef
a downturn to be supported by relevant historical data; and

v
cti
(b) the PRA expects firms to be aware of, and to properly account for, the link between
ffe
cures and subsequent defaults. In particular, an earlier cure definition is, other things
.E
being equal, likely to result in a higher level of subsequent defaults.

t2
ar
Incomplete workouts
lp
na
12.42 In order to ensure that estimates of LGDs take into account the most up-to-date
-fi
experience in accordance with Article 179(1)(c) of the Credit Risk: Internal Ratings Based
ar
Approach (CRR) Part, the PRA expects firms to take account of data in respect of relevant
Ne
incomplete workouts (ie defaulted exposures for which the recovery process is still in
progress, with the result that the final realised losses in respect of those exposures are not
yet certain).
Unsecured LGDs where the obligor’s assets are substantially

used as collateral
12.43 The extent to which an obligor’s assets are already given as collateral will affect the
recoveries available to unsecured creditors. The PRA considers that if the degree to which
assets are pledged is substantial, this would typically be a material risk driver of LGDs on
such exposures and should be taken into account in accordance with Article 171(2) of the
Credit Risk: Internal Ratings Based Approach (CRR) Part. Although potentially present in all
transactions, the PRA expects firms to be particularly aware of this risk driver in situations in
which borrowing on a secured basis is the normal form of financing, leaving relatively few
assets available for the unsecured debt. Specialist lending (including property), hedge funds,
some small and medium-sized enterprises (SME) lending, and some mid-market lending are
examples of such cases.
12.44 The PRA expects firms estimating LGDs to take into account the effect of assets being
substantially used as collateral for other obligations where this is the case. The PRA expects
firms not to use unadjusted data sets that ignore this impact, and notes that this effect should
be assessed under downturn conditions. In the absence of relevant data to estimate this
effect, the PRA expects firms to apply a suitable MoC which may result in LGD estimates
.
26
being increased to 100% in some cases.
20
LGD – use of external data for residential mortgages
ry
ua
12.45 The PRA expects that, for residential mortgages, where a firm’s internal experience of
an
defaults for a rating system is low, the firm may use external data to supplement internal data
1J
when modelling LGD.
m
ro
12.46 Where a firm uses external data, the PRA expects the firm to apply an additional MoC
ef
in order to:
v
cti
ffe
(a) recognise the difference between downturn recoveries achieved by established firms
with the experience and processes to realise higher recoveries, and downturn
.E
recoveries achieved by firms with more limited experience and less established
t2
processes;
ar
lp
(b) recognise any differences in portfolio comparability between the external data and the
na
firm’s lending; and

-fi
(c) address unobservable differences that relate to risk drivers or risk characteristics that
ar
cannot be derived from external data.

Ne
12.47 The PRA expects the level of added conservatism applied in accordance with
paragraph 12.46 to be significant until sufficient internal data are available to support a
reduction.
12.48 Firms using external data in their LGD estimates should run a Forced Sale Discount
(FSD) model and Probability of Possession Given Default (PPGD) model with appropriate
governance and monitoring in line with the requirements set out in the Credit Risk: Internal
Ratings Based Approach (CRR) Part. Firms with no internal repossession data for use in
their FSD modelling may rely on external data, along with an internal expectation on costs
and an additional MoC, as part of their FSD estimation.
12.49 The PRA considers that firms would be unlikely to be able to demonstrate that third-
party recovery data from non-UK legal regimes are comparable to UK data. The PRA
therefore expects only UK data to be used when estimating LGD for UK residential mortgage
exposures. For non-UK mortgage exposures, the PRA expects firms to demonstrate that
external data are representative for the local mortgage market in order for them to be used to
supplement internal data.
12.50 The PRA expects that as the amount of internal data builds up, firms should revise
their modelling approaches to incorporate the additional data.
LGD estimates for exposures to corporates
.
26
12.51 The PRA expects that firms should do all of the following in respect of LGD estimates
20
for exposures to corporates:
ry
ua
(a) apply LGD estimates at transaction level;
an
(b) where there is a paucity of observations, ensure that long-run average and downturn
1J
LGD estimates are cautious, conservative, and justifiable. In accordance with Article
m
179(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, estimates
ro
must be derived using both historical experience and empirical evidence, and not be
ef
based purely on judgemental consideration. The PRA expects firms to document their
v
cti
justification of why they consider their estimates to be sufficiently conservative;

ffe
(c) identify and explain at a granular level how each estimate has been derived. This
.E
should include an explanation of how internal data, external data, expert judgement, or
t2
a combination of these have been used to produce the estimate;

ar
(d) clearly document the process for determining and reviewing estimates, and the parties
lp
involved in the process in cases where expert judgement was used;

na
(e) demonstrate an understanding of the impact of the economic cycle on collateral values
-fi
and be able to use that understanding in deriving downturn LGD estimates;

ar
(f) demonstrate sufficient understanding of any external benchmarks used, and identify
Ne
the extent of their relevance and suitability, to the extent that the firm can satisfy itself
that they are fit for purpose;
(g) evidence that they are aware of any weaknesses in their estimation process and set
standards, for example related to accuracy, that their estimates are designed to meet;
(h) demonstrate that they have sought and utilised relevant and appropriate external data,
including through identifying all relevant drivers of LGD and how these will be affected
by a downturn;
(i) ensure that in most cases estimates incorporate effective discrimination on the basis
of at least security type and geography. In cases where these drivers are not
incorporated into LGD estimates, the PRA expects firms to be able to demonstrate
why they are not relevant; and
(j) have an ongoing data collection framework to collect all relevant internal loss and
exposure data required for estimating LGD and a framework to start incorporating
these data in LGD models as soon as any meaningful information becomes available.
.
26
20
ry
ua
an
1J
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
13: LGD – calibration (general)
Definition of economic loss and realised LGD

13.1 The PRA expects that, for the purpose of LGD estimation as referred to in Article
181(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, firms should
calculate realised LGDs for each exposure as a ratio of the economic loss to the outstanding
amount of the credit obligation at the moment of default, including any amount of principal,
interest, or fee.
.
26
13.2 For the purpose of paragraph 13.1, firms should calculate the economic loss realised on
20
an exposure (ie defaulted facility) as referred to in CRR Article 5(2) as the difference
ry
between:
ua
an
(a) the outstanding amount of the credit obligation at the moment of default, without
1J
prejudice to paragraph 13.14, including any amount of principal, interest, or fee,
m
increased by material direct and indirect costs associated with collecting on that
ro
exposure, discounted to the moment of default; and

ef
(b) any recoveries realised after the moment of default, discounted to the moment of
v
default.
cti
ffe
13.3 For the purpose of calculation of the economic loss realised on an exposure in
.E
accordance with paragraph 13.2, firms should not take the following into account:
t2
ar
(a) recoveries related to collateral if they are not applying the LGD Modelling Collateral
lp
Method in accordance with Article 169A(1)(a) of the Credit Risk: Internal Ratings
na
Based Approach (CRR) Part; and

-fi
(b) recoveries related to collateral types that do not meet the requirements set out in
ar
Article 169A(2)(b) of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
Ne
13.4 The PRA expects that where a firm recognises on-balance sheet netting, or recognises
master netting agreements when applying the Financial Collateral Comprehensive Method
(FCCM) or the securities financing transactions value-at-risk (SFT VaR) Method, in
accordance with Article 191A of the Credit Risk Mitigation (CRR) Part:
(a) in the case of on-balance sheet netting, the outstanding amount of the credit obligation
at the moment of default should be the exposure value calculated in accordance with
Article 219(1) of the Credit Risk Mitigation (CRR) Part, including any amount of
principal, interest or fee realised so far;
(b) in the case of master netting agreements, the outstanding amount of the credit
obligation at the moment of default should be the fully adjusted exposure value (E*)
reflecting the netting agreement calculated in accordance with Article 220(3) or 221(6)
of the Credit Risk Mitigation (CRR) Part, including any amount of principal, interest, or
fee realised so far; and
(c) for the purpose of calculating economic loss in accordance with paragraph 13.1, no
cash flows from netting should be included as recoveries after default.
13.5 Where, relating to a default event, any part of an exposure has been forgiven or written
off before or at the date of default, and the amount forgiven or written off is not included in the
outstanding amount of the credit obligation at the moment of default, the amount of the
exposure that was forgiven or written off should be added to the outstanding amount of the
credit obligation at the moment of default for both the calculation of economic loss as
.
26
specified in paragraph 13.2 in the numerator, and the calculation of the outstanding amount
20
of the credit obligation in the denominator of the realised LGD.
ry
13.6 In the case of exposures that return to non-defaulted status, firms should calculate
ua
economic loss as for all other defaulted exposures with the only difference that an additional
an
recovery cash flow (‘artificial cash flow’) should be added to the calculation in line with
1J
paragraph 13.7 as if a payment had been made by the obligor at the date of the return to
m
non-defaulted status. Where exposures meet the criteria of paragraph 12.2, realised LGDs
ro
should be calculated with reference to the date of the first default event, taking into account
ef
all cash flows observed from the date of the first default event, including those observed
v
cti
during the period between the first and the second defaulted status, without adding an
ffe
artificial cash flow.

.E
t2
13.7 For the purpose of applying paragraph 13.6, the PRA expects that:
ar
lp
(a) the artificial cash flow should reflect:

na
(i) principal: total outstanding amount of the full loan at the moment of cure, but only
-fi
the amount of missed payments (ie actual past due payments) accrued up to the
ar
moment of cure should be discounted;

Ne
(ii) interest: amount accrued between the moment of default and the moment of cure;
(iii) fees: amount accrued between the moment of default and the moment of cure;
(iv) additional observed recoveries: total amount received up to the moment of cure;
(v) additional drawings: firms should follow the requirements of the last sentence of
Article 181(1) and Article 182(1)(ca) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, and paragraphs 13.11 to 13.14. Additional drawings
included in the artificial cash flow should be treated in the same way as the
principal; and
(vi) costs: amount accrued between the moment of default and the moment of cure;
(b) for the purpose of paragraph 13.7(a)(i), the ‘moment of cure’ is defined as the start of
the final period when no triggers of default continue to apply prior to the exposure
being rated as a non-defaulted exposure as referred to in Articles 178(5) to 178(5C)

of the Credit Risk: Internal Ratings Based Approach (CRR) Part; and
(c) the artificial cash flow should be discounted over the actual period of default only (ie
between the moment of default and the moment of cure) and, therefore, should not
be discounted over any additional time period after the moment of cure, such as the
period where no triggers of default continue to apply but the exposure is rated as
being in default in accordance with Articles 178(5) to 178(5C) of the Credit Risk:
13.8 Where firms open new facilities to replace previously defaulted facilities as part of
restructuring or for technical reasons, they should calculate the realised LGDs based on the
.
26
originally defaulted facilities. For this purpose, firms should have a sound mechanism to
20
allocate observed costs, recoveries, and any additional drawings to original facilities.
ry
Treatment of fees, interest, and additional drawings after default
ua
an
13.9 The PRA expects that, for the purpose of Article 181(1)(i) of the Credit Risk: Internal
1J
Ratings Based Approach (CRR) Part, firms should take into account in the calculation of
m
realised LGD any fees for delays in payments that have been capitalised in the firm’s income
ro
statement before the moment of default by including them in the outstanding amount of the
ef
credit obligation at the moment of default in the numerator and the denominator of the
v
cti
realised LGD. Where the fees were extended to the obligor in order to recover direct costs
ffe
already incurred by the firm and these costs are already included in the calculation of the
.E
economic loss, firms should not add these amounts to the economic loss or outstanding
t2
amount again. Any fees capitalised after the moment of default should not increase the
ar
amount of economic loss or amount outstanding at the moment of default. However, all
lp
recoveries, including those related to fees capitalised after default, should be included in the
na
calculation of economic loss.

-fi
ar
13.10 Firms should apply the treatment specified in paragraph 13.9 to any interest capitalised
Ne
in their income statement before and after the moment of default.
13.11 In accordance with Article 182(1)(ca) of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, firms using the AIRB approach and modelling CFs or EADs are
required to reflect the possibility of additional drawings by the obligor up to the time of default
in their estimates of CFs or EADs. Firms modelling CFs or EADs are also required to reflect
additional drawings by the obligor after the moment of default in their CF or EAD estimates
where post-default additional drawings have not been reflected in LGD estimates in
accordance with the last sentence of Article 181(1) of the Credit Risk: Internal Ratings Based
13.12 Where firms choose not to recognise additional drawings by the obligor after the
moment of default in LGDs, in accordance with the last sentence of Article 181(1) of the
Credit Risk: Internal Ratings Based Approach (CRR) Part, they should calculate realised
LGD as the ratio of the economic loss to the outstanding amount of the credit obligation at
the moment of default increased by the amount of additional drawings by the obligor after the
moment of default, discounted to the moment of default.
13.13 Where firms choose to recognise additional drawings by the obligor after the moment
of default in LGDs, in accordance with the last sentence of Article 181(1) of the Credit Risk:
Internal Ratings Based Approach (CRR) Part, they should calculate realised LGD as the ratio
of the economic loss to the outstanding amount of the credit obligation at the moment of
.
26
default, and they should not increase the denominator of the ratio by the value of additional
20
drawings by the obligor after the moment of default.
ry
13.14 Irrespective of whether firms reflect additional drawings after the moment of default in
ua
their LGD estimates, they should calculate the economic loss used in the numerator of the
an
realised LGD including all additional drawings after the moment of default and all realised
1J
recoveries discounted to the moment of default.
m
ro
Discount rate
vef
cti
13.15 In order to ensure that their LGD estimates incorporate material discount effects, the
ffe
PRA expects firms’ methods for discounting cash flows to take account of the uncertainties
.E
associated with the receipt of recoveries with respect to a defaulted exposure. This could be,
t2
for example, by adjusting cash flows to certainty equivalents, by using a discount rate that
ar
embodies an appropriate risk premium, or by a combination of the two.

lp
na
13.16 For the purpose of estimating long-run average LGD, the PRA expects firms to use a
-fi
discount rate of Sterling Overnight Index Average (SONIA) at the moment of default plus five
ar
percentage points for exposures denominated in Pound Sterling (GBP). For the purpose of
Ne
estimating long-run average LGD for exposures denominated in currencies other than GBP,
firms should use a comparable liquid interest rate in the currency of that exposure.
13.17 For the purpose of estimating downturn LGD, the PRA expects firms to use a discount
rate of at least 9%.
13.18 For defaulted exposures, the PRA expects firms to use a discount rate in line with
paragraph 13.16 for estimating BEEL; and to use a discount rate in line with paragraph 13.17
for estimating LGD in-default.
13.19 For determining the discount rate to be used for defaults that occurred before 2
January 1997 (ie the first SONIA rate available from the Bank of England), the PRA expects
firms to develop a suitable approach. This could for example be an extrapolation based on
available data or use of an appropriate alternative such as the relevant central bank rate for
that period.
13.20 The PRA expects that the amount of recoveries that can be recognised as a cashflow
and discounted should not be higher than the amount of recoveries the firm is contractually
entitled to retain for the exposure.
Direct and indirect costs

13.21 The PRA expects that, for the purpose of the calculation of the realised LGDs, firms
should take into account all material direct and indirect costs related to the recovery process.
.
Where any such costs have been incurred before the moment of default, firms should include
26
these costs in the LGD estimation unless at least one of the following conditions is met:
20
ry
(a) these costs are clearly included in the outstanding amount of the credit obligation at
ua
the moment of default; or
an
(b) these costs are associated with the previous default of the same obligor, which is not
1J
considered a multiple default in accordance with paragraph 12.2.
m
ro
13.22 Direct costs should include the costs of outsourced collection services, legal costs, the
ef
cost of hedges and insurances, and all other costs directly attributable to the collection on a
v
cti
specific exposure. Firms should consider all direct costs as material.

ffe
13.23 Indirect costs should include all costs stemming from the running of the firm’s recovery
.E
processes, overall costs of outsourced collection services not included as direct costs, and all
t2
other costs related to the collection on defaulted exposures that cannot be directly attributed
ar
to collection on a specific exposure. Firms should include in their estimation of indirect costs
lp
an appropriate percentage of other ongoing costs such as overheads related to the recovery
na
processes unless they can demonstrate that these costs are immaterial.
-fi
ar
Ne
Low LGDs
13.24 The PRA expects firms to justify any low LGD estimates using analysis on volatility of
sources of recovery, notably on collateral (where the firm is applying the LGD Modelling
Collateral Method in accordance with Article 169A(1)(a) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part), and cures (as referred to in paragraph 12.41). This includes,
where relevant:
(a) recognising that the impact of collateral volatility on low LGDs is asymmetric as
surpluses over amounts owed need to be returned to borrowers and that this effect
may be more pronounced when estimating downturn rather than normal period LGDs;
and
(b) recognising the costs and discount rate associated with realisations and the
requirements of Article 181(1)(e) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part.
.
26
20
ry
ua
an
1J
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
14: LGD – calibration (long-run average)
14.1 The PRA considers that the historical observation period for calibrating long-run average
LGD estimates should be as broad as possible and should contain data from various periods
with differing economic circumstances. For this purpose, firms should at a minimum select a
historical observation period in such a way that:
(a) the length of the historical observation period, ie the timespan between the oldest
default considered in the RDS and the moment of the LGD estimation, covers at least
the minimum length specified in Article 181(1)(j) of the Credit Risk: Internal Ratings
.
26
Based Approach (CRR) Part for exposures to corporates and, for retail exposures, the
20
period specified in the final sentence of Article 181(2) of the Credit Risk: Internal
ry
Ratings Based Approach (CRR) Part;
ua
(b) it ensures that the RDS includes a sufficient number of closed recovery processes in
an
order to provide robust LGD estimates; 1J
(c) it is composed of consecutive periods and includes the most recent periods before the
m
moment of LGD estimation;

ro
(d) it includes the full period for which the firm is reasonably able to replicate the currently
ef
applicable definition of default;

v
(e) for exposures to corporates, all available internal data are considered ‘relevant’, as
cti
ffe
referred to in Article 181(1)(j) of the Credit Risk: Internal Ratings Based Approach
.E
(CRR) Part, and are included in the historical observation period; and
t2
(f) for retail exposures, all internal data are included in the historical observation period.
ar
14.2 In assessing whether the RDS includes a sufficient number of closed recovery
lp
processes in accordance with paragraph 14.1(b), firms should take into account the number
na
of closed recovery processes in the total number of observations.

-fi
ar
Ne
Calculation of long-run average LGD

14.3 In accordance with Article 181(1)(a) Credit Risk: Internal Ratings Based Approach
(CRR) Part, firms are required to calculate long-run average LGD separately for each facility
grade or pool. Firms should also calculate long-run average LGD at the level of the
calibration segment covered by the LGD model where they apply the approach to calibration
set out paragraph 14.15(b). Firms should use all defaults observed in the historical
observation period that fall within the scope of the LGD model in the calculation of long-run
average LGD.
14.4 Firms should calculate the long-run average LGD as an arithmetic average of realised
LGDs over a historical observation period weighted by the number of defaults. Firms should
not use for that purpose any averages of LGDs calculated on a subset of observations, in
particular any yearly average LGDs.
Treatment of incomplete recovery processes

14.5 The PRA expects that, for the purpose of calculating long-run average LGD based on all
observed defaults in accordance with Article 181(1)(a) of the Credit Risk: Internal Ratings
Based Approach (CRR) Part, firms should ensure that relevant information from incomplete
recovery processes is taken into account in a conservative manner.
14.6 Firms should calculate the observed average LGD taking into account realised LGDs on
.
all defaults observed in the historical observation period related to closed recovery
26
processes, in accordance with paragraphs 14.7 to 14.11, without including any expected
20
future recoveries. The observed average LGD should be weighted by the number of defaults
ry
included in the calculation.
ua
an
14.7 Firms should clearly specify in their internal policies the moment of closing the recovery
1J
processes. All recovery processes that have been closed should be treated as such for the
m
purpose of the calculation of the observed average LGD.

ro
ef
14.8 Firms should define the maximum period of the recovery process for a given type of
v
cti
exposures from the moment of default that reflects the expected period of time observed on
ffe
the closed recovery processes during which the firm realises the vast majority of the
recoveries, without taking into account the outlier observations with significantly longer
.E
recovery processes. The maximum period of the recovery processes should be specified in a
t2
way that ensures sufficient data for the estimation of the recoveries within this period for the
ar
lp
incomplete recovery processes. The length of the maximum period of the recovery processes
na
may be different for different types of exposures.

-fi
ar
14.9 The specification of the maximum period of the recovery process in accordance with
Ne
paragraph 14.8 should be clearly documented and supported by evidence of the observed
recovery patterns and should be coherent with the nature of the transactions and the type of
exposures. The PRA expects that specification of the maximum period of the recovery
process for the purpose of long-run average LGD estimation should not prevent firms from
taking recovery actions where necessary, even with regard to exposures which remain in
default for a period of time longer than the maximum period of the recovery process specified
for this type of exposures.
14.10 For the purpose of the calculation of the observed average LGD, firms should, without
undue delay, recognise as closed recovery processes all exposures in default which fall into
at least one of the following categories:
(a) exposures for which the firm does not expect to take any further recovery actions;
(b) exposures that remain in defaulted status for a period of time longer than the
maximum period of the recovery process specified for the type of exposures;
(c) exposures fully repaid or written-off; or
(d) exposures that have been reclassified to non-defaulted status.
14.11 With regard to the defaulted exposures falling under the categories in paragraphs
14.10(a) and 14.10(b), all recoveries and costs realised before or at the time of estimation
should be considered for the purpose of the calculation of the observed average LGD,
including any recoveries realised after the maximum period of the recovery processes.
14.12 Firms should obtain the long-run average LGD by adjusting the observed average LGD
taking into account the information related to processes that were not closed (‘incomplete
.
26
recovery processes’) and where the time from the moment of default until the moment of
20
estimation is shorter than the maximum period of the recovery process specified for this type
ry
of exposures. For these processes, firms should do both of the following:
ua
an
(a) take into account all observed costs and recoveries; and 1J
(b) either estimate future costs and recoveries or assume zero future costs and
m
recoveries. Where a firm estimates future costs and recoveries, these may include
ro
both those stemming from the realisation of the existing collateral and those to be
ef
realised without the use of collateral within the maximum period of the recovery
v
cti
processes.
ffe
14.13 The PRA expects that the estimation referred to in paragraph 14.12(b) should be
.E
consistent with the following principles:

t2
ar
(a) for the purpose of estimation of the future costs and recoveries, firms should analyse
lp
the costs and recoveries realised on these exposures until the moment of estimation,
na
in comparison with the average costs and recoveries realised during a similar period
-fi
ar
of time on similar exposures; for this purpose firms should analyse the recovery
Ne
patterns observed on both closed and incomplete recovery processes, taking into
account only costs and recoveries realised up to the moment of estimation;
(b) the assumptions underlying the expected future costs and recoveries, as well as the
adjustment to the observed average LGD, should be:
(i) proven accurate through back-testing;
(ii) based on a reasonable economic rationale; and
(iii) proportionate, taking into consideration that LGD estimates should be based on
the long-run average LGD that reflects the average LGDs weighted by the
number of defaults using all defaults observed during the historical observation
period;
(c) in estimating future recoveries, firms should take into account the potential bias
stemming from incomplete recovery processes being characterised by longer
average recovery processes or lower average recoveries than closed recovery

processes;
(d) for firms applying the LGD Modelling Collateral Method in accordance with Article
169A(1) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, in
estimating future recoveries stemming from the realisation of collateral, firms should
take into account the legal certainty of the claims on the collateral and realistic
assumptions regarding the possibility of its realisation;
(e) the adjustment of the observed average LGD may be estimated at the level of
individual exposures, at the level of grade or pool, at calibration segment level, or at
the level of portfolio covered by the LGD model; and
(f) any uncertainty related to the estimation of the future recoveries on incomplete
.
26
recovery processes should be reflected in an adequate MoC applied in accordance
20
with Chapter 9.
ry
ua
Treatment of cases with no loss or positive outcome
an
1J
14.14 Where firms observe that they realised profit on their observations of defaults, the
realised LGD on these observations should equal zero for the purpose of calculation of the
m
ro
observed average LGD and the estimation of long-run average LGD. Firms may use the
ef
information on the realised LGDs before the application of this floor in the process of model
v
development for the purpose of risk differentiation.

cti
ffe
Calibration to the long-run average LGD

.E
t2
14.15 Firms should calibrate their LGD estimates to the long-run average LGD calculated in
ar
accordance with paragraphs 14.1 to 14.14. For this purpose, firms should choose a
lp
calibration method that is appropriate for their LGD estimation methodology from the
na
following approaches:
-fi
ar
(a) the calibration of LGD estimates to the long-run average LGD calculated for each
Ne
grade or pool, in which case they should provide additional calibration tests at the level
of the relevant calibration segment; or
(b) the calibration of LGD estimates to the long-run average LGD calculated at the level of
the calibration segment, in particular where they use direct LGD estimates in
accordance with Article 169(3) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part, including where they use an LGD estimation methodology based on
intermediate parameters. In this case, firms should at least compare this long-run
average LGD with the average LGD estimate applied to the same set of observations
as those used for calculating the long-run average LGD and, where necessary, correct
the individual LGD estimates for the application portfolio accordingly, for instance by
using a scaling factor. Where realised values are higher than estimated values at the
level of calibration segment, firms should correct the estimates upwards or further
adjust their estimates in order to reflect their loss experience.
14.16 Where firms observe extremely high values of realised LGDs that are significantly
above 100%, especially for exposures with small outstanding amounts at the moment of
default, they should identify relevant risk drivers to differentiate these observations and
adequately reflect these specific characteristics in the assignment of exposures to facility
grades or pools. Where firms use a continuous rating scale in the LGD estimation, they may
create a separate calibration segment for such exposures.
14.17 In order to comply with the requirement of Article 181(1)(a) of the Credit Risk: Internal
.
Ratings Based Approach (CRR) Part to use all observed defaults in LGD quantification, firms
26
should not exclude any defaults observed in the historical observation period that fall within
20
the scope of application of the LGD model.
ry
ua
14.18 The PRA expects that in their analysis of the representativeness of data in accordance
an
with paragraphs 8.10 to 8.15, firms should take into account not only the current
1J
characteristics of the portfolio but also, where relevant, the changes to the structure of the
m
portfolio that are expected to happen in the foreseeable future due to specific actions or
ro
decisions that have already been taken. Adjustments made on the basis of the changes
ef
expected in the foreseeable future should not lead to a decrease in the estimates of LGD
v
cti
parameter.
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
15: LGD – calibration (downturn)
General requirements on downturn LGD estimation

15.1 The PRA expects that for the purpose of quantifying downturn LGD, a firm should apply
the following expectations specific to downturn LGD estimates by facility grade or pool:
(a) calibrate downturn LGD at least at the same level at which the firm calculates the
corresponding long-run average LGD for the purpose of calibrating LGD in accordance
with paragraph 14.15; and
.
26
(b) split the set of facilities covered by the same LGD model into as many different
20
calibration segments as needed, where each calibration segment carries a
ry
significantly different loss profile and might thus be affected differently by different
ua
downturn periods. For this purpose, the firm should at least consider the
an
appropriateness of introducing calibration segments that cover material shares of
1J
exposure in different geographical areas, in different industry sectors and, for retail
m
exposures, of different product types.

ro
ef
15.2 As set out in Article 181C(1) of the Credit Risk: Internal Ratings Based Approach (CRR)
v
Part, for the purpose of identifying an economic downturn, firms are required to examine
cti
economic indicators over a historical timespan that provides values that are representative of
ffe
the likely range of variability in the future, and the historical timespan selected must have a
.E
duration of at least 20 years. The PRA expects that firms should select a historical timespan
t2
which enables the identification of economic indicator values that represent sufficiently
ar
lp
severe downturn conditions. If the economic indicator values in a selected timespan do not
na
represent sufficiently severe downturn conditions, firms should extend their historical
-fi
timespan beyond the minimum 20-year period.

ar
Ne
15.3 Where firms identify multiple distinct downturn periods in accordance with Article
181A(2)(c) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, they should
perform each of the following steps in sequence:
(a) calibrate downturn LGD for each identified downturn period in accordance with
paragraph 15.9 for each calibration segment, except where firms can provide evidence
that the economic factors are not relevant for a given calibration segment;
(b) for each of those downturn periods, apply the resulting downturn LGD estimates to
their current non-defaulted exposures of the type of exposures under consideration;
(c) identify the ‘finally relevant downturn period’ as being the downturn period that results
in the highest average downturn LGD, including the final MoC as set out in paragraph
9.10, on a given calibration segment of their current non-defaulted exposures as
referred to in point (b);
(d) use the downturn LGDs resulting from the finally relevant downturn period referred to
in point (c) to determine the downturn LGD in accordance paragraph Article 181(b)(i)
of the Credit Risk: Internal Ratings Based Approach (CRR) Part for each calibration
segment; and
(e) if a firm can calibrate downturn LGD in accordance with this SS for one or more
downturn periods, but is unable to do this for one or more other downturn periods, it
should add an appropriate Category A MoC in accordance with paragraph 9.7.
Adjustments to downturn LGDs where risk drivers are sensitive to

the economic cycle
.
26
15.4 Where risk drivers are sensitive to the economic cycle, firms should undertake the
20
analysis set out in Article 181(1)(b)(ii)(1) of the Credit Risk: Internal Ratings Based Approach
ry
(CRR) Part and make any adjustments to the downturn LGDs referred to in paragraph
ua
15.3(c) that are required by Article 181(1)(b)(ii)(2) of the Credit Risk: Internal Ratings Based
an
Approach (CRR) Part. 1J
m
15.5 For the purpose of applying paragraph 15.4 in respect of exposures to corporates, firms
ro
should consider whether any adjustments they already make to address cyclicality in LGD
ef
estimates, for example through conservative approaches to haircutting collateral values, are
v
sufficient such that no further adjustment to downturn LGDs is required by Article

cti
181(1)(b)(ii)(2) of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
ffe
.E
Final downturn LGD estimates

t2
ar
15.6 For the purpose of ensuring that the resulting downturn LGDs are used if they are more
lp
conservative than the long-run average LGDs in accordance with Article 181(1)(b)(i) of the
na
Credit Risk: Internal Ratings Based Approach (CRR) Part, a firm should apply the following:
-fi
ar
(a) where the firm uses separate estimation methodologies for long-run average LGD and
Ne
downturn LGD, the firm should compare their final downturn LGDs used for calibration
plus the final MoC as set out in paragraph 9.10, to their long-run average LGDs plus
the final MoC as set out in paragraph 9.10 at the level where the long-run average
LGD is calculated for the purpose of calibrating LGD in accordance with paragraph
14.15; and
(b) where the firm sets a single LGD estimate, which involves a long-run average LGD
estimation and a downturn adjustment added to the long-run average LGD estimation,
it should ensure that the final MoC as set out in paragraph 9.10 applied to downturn
LGD estimates encompasses the uncertainties stemming from both the long-run
average LGD estimation and the calculation of the downturn adjustment.
Calculation of reference values

15.7 Firms should calculate a reference value at least at the level of each calibration segment
in accordance with the following sequence of steps:
(a) using all available loss data, firms should select the two individual years with the
highest observed economic loss by:
(i) grouping all defaults according to the year in which the defaults occurred;
(ii) for each year, as identified in point (i), calculating for the defaults that occurred in
the applicable year, the ratio of total economic loss as specified in paragraphs
13.1 to 13.23 to the total outstanding amount of the relevant credit obligations at
.
the moment of default; and
26
(iii) selecting the two individual years with the highest annual ratio of total economic
20
loss to total outstanding amount resulting from point (ii) as the two individual
ry
years with the highest observed economic losses;
ua
(b) firms should calculate the reference value as the simple average of the average
an
realised LGDs from the two individual years with the highest observed economic
1J
losses, as identified in paragraph 9.2(a)(iii).
m
ro
15.8 Firms should compare the final downturn LGD with the reference value calculated in
ef
accordance with paragraph 15.7 at least at the level of calibration segments. Firms should
v
cti
justify any material difference between the final downturn LGD and the reference value.
ffe
15.9 When comparing the final downturn LGD with the reference value in line with paragraph
.E
15.8, firms should take into account all of the following:

t2
ar
(c) a material difference between the final downturn LGD plus the final MoC as set out in
lp
paragraph 9.10 and the reference value can be justified if the period of losses
na
identified by the reference value does not stem from an identified or unidentified
-fi
ar
downturn period. Where the underlying downturn LGD is based on the methodology
Ne
in paragraphs 15.14 to 15.16, firms may consider the evidence gathered from the
impact assessment in paragraph 15.14; and
(d) if the material difference between the final downturn LGD and the reference value
cannot be justified, firms should re-assess their quantification of downturn LGD
ensuring in particular that the downturn periods have been identified
comprehensively and that, where intermediate parameters are used, the impact of
the relevant downturn period observed (based on paragraphs 15.14 to 15.16) or
estimated (based paragraphs 15.17 to 15.23) on intermediate parameters has been
aggregated adequately. If a firm has re-assessed its quantification of downturn LGD
and found the methodology to be adequate, this assessment can be used to justify a
material difference between the final LGD and the reference value.
Downturn LGD estimation for a considered downturn period

15.10 For the purpose of calibrating downturn LGD for each distinct downturn period
identified in accordance with Articles 181A, 181B and 181C of the Credit Risk: Internal
Ratings Based Approach (CRR) Part, firms should either:
(a) estimate downturn LGD based on observed impact as set out in paragraphs 15.14 to
15.16; or
(b) estimate downturn LGD based on estimated impact as set out in paragraphs 15.17 to
15.22.
15.11 The PRA considers that either a component-based modelling approach or a direct
.
26
estimate modelling approach can be used when LGD is estimated based on observed
20
impact, and that a component-based approach can be used when LGD is estimated based
ry
on estimated impact. However, the PRA considers that it is unlikely that firms will be able to
ua
produce robust direct estimate downturn LGD models for residential mortgages. Therefore,
an
the PRA expects firms to use a component-based approach for these exposures. 1J
m
15.12 The PRA expects firms using a component-based approach to modelling downturn
ro
LGD to ensure that all components reflect a downturn, and that each component reflects the
ef
same downturn. Firms should take into account any time lags between the downturn period
v
and the potential impact on their loss data. Therefore, while model components should reflect
cti
the same downturn, a time lag may be necessary so that the peak value within the same
ffe
downturn is used for each model component. Firms should ensure the time lags are not so
.E
long that they result in LGD estimates that are reflective of an upturn or improved economic
t2
conditions.
ar
lp
15.13 Regardless of the approach used for calibrating downturn LGD, firms should adhere to
na
the following principles:

-fi
ar
(a) where the approach used involves the estimation or analysis of different intermediate
Ne
parameters, the aggregation of these intermediate parameters for the purpose of

calibrating downturn LGD should start with the parameter where the highest impact is
observed in accordance with paragraphs 15.13 to 15.16 or estimated in accordance
with paragraphs 15.17 to 15.23, and any additional impact observed or estimated on
other parameters should be added where necessary; and
(b) the downturn LGD estimates should not be biased by observed or estimated cash
flows that are received with a significantly longer time lag than the period referred to in
paragraphs 14.8 and 14.9 and which might reflect an upturn or improved economic
conditions following the considered downturn period.
Downturn LGD estimation based on observed impact

15.14 In order to calibrate downturn LGD based on the observed impact of a considered
downturn period, firms should carry out an analysis of the impact of this downturn period on
the loss data related to the considered calibration segment.
(a) The analysis should comprise at least all of the following:

(i) evidence of elevated levels of realised LGDs, driven by the considered downturn
period, taking into account all of the following:
1. the realised LGDs should be calculated as averages related to all defaults
that occurred in a considered year and that have either reached their
.
maximum time of recovery in accordance with paragraphs 14.8 and 14.9 or
26
have been closed before; and
20
2. for all incomplete recovery processes of defaulted exposures that have not
ry
reached their maximum time of recovery in accordance with paragraphs 14.8
ua
and 14.9, the marginal recoveries reached in each year after default should
an
be computed. The resulting recovery patterns should be compared to the
1J
recovery patterns of the defaults considered in point (1) for each year in
m
which the defaults occurred;

ro
(ii) evidence of decreased annual recoveries by sources of recoveries that are

ef
relevant for the considered calibration segment. These annual recoveries should
v
cti
be analysed with and without repossessions where applicable and irrespective

ffe
of the date of default;

.E
(iii) evidence of decreased numbers of exposures that defaulted and returned back
t2
to non-defaulted status in accordance with Article 178(5) of the Credit Risk:

ar
Internal Ratings Based Approach (CRR) Part within a predefined fixed horizon
lp
for all defaults that occurred in a given year. The predefined fixed horizon should
na
be appropriate for the type of exposures under consideration;

-fi
(iv) evidence of increased time in default per year related to all defaults in a given
ar
year.
Ne
(b) the analysis expected in paragraph 15.14(a) should take into account as many points
in time as possible where sufficient relevant loss data are available. Otherwise, if
only scarce relevant loss data are available on an annual basis, firms should merge
consecutive years of observations provided this adds value to the analysis; and
(c) the analysis expected in paragraphs 15.14(a) and 15.14(b), should take into account
any lag between a downturn period and the time when its potential impact is
observed on the relevant loss data.
15.15 Based on the evidence obtained from the impact analysis referred to in paragraph
15.14, firms should calibrate downturn LGD by applying an estimation methodology which is
coherent with the evidence obtained from the impact analysis.
15.16 Where the impact analysis conducted in accordance with paragraph 15.14 shows no
impact of a downturn period on a firm’s relevant loss data, such that the average observed
realised losses in this downturn period are not different from those under other economic
conditions, the firm may use the long-run average LGD as downturn LGD, where all of the
following apply:
(a) the firm ensures and documents that the deficiencies identified and MoC applied in
accordance with Chapter 9 incorporate all additional elements of uncertainty related to
the identified downturn periods; and
(b) for the purpose of point (a), the firm in particular verifies that, for the considered
downturn period, none of the deficiencies identified resulting in a Category A MoC in
.
26
accordance with paragraph 9.7 are of higher severity and that no additional
20
deficiencies or adjustments necessitating a Category B MoC in accordance with
paragraph 9.7 are applicable.
ry
ua
Downturn LGD estimation based on estimated impact
an
1J
15.17 In order to estimate downturn LGD based on estimated impact, firms should calibrate
m
downturn LGD using one of the methodologies specified in paragraph 15.19 (‘haircut
ro
approach’) and paragraph 15.20 (‘extrapolation approach’), or a combination of those. Prior

ef
to quantifying downturn LGD estimates, firms should choose the most relevant methodology
v
cti
based on:
ffe
.E
(a) the appropriateness of the methodology to estimate the impact of the downturn period
t2
under consideration on realised LGDs, intermediate parameters or risk drivers; and

ar
(b) where relevant, the need to use a combination of the methodologies to ensure that the
lp
resulting downturn LGDs for the downturn period under consideration adequately
na
reflect a potential downturn impact on all material components of economic loss in

-fi
accordance with paragraphs 13.1 to 13.23 and in accordance with the principles set
ar
out in paragraph 15.13.

Ne
15.18 In particular, the haircut approach specified in paragraph 15.19 should be considered
most appropriate for the above purposes where the market value or an index related to a
relevant type of collateral serves as a direct or transformed input into a firm’s model for LGD
estimation and has been identified as a relevant economic factor in accordance with Article
181B of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
15.19 A ‘haircut approach’ refers to an approach for the estimation of the impact of the
downturn period on realised LGDs, intermediate parameters, or risk drivers in which one or
several economic factors as referred to in Article 181B of the Credit Risk: Internal Ratings
Based Approach (CRR) Part are direct or transformed input(s) in the LGD model and, where
for the purpose of this estimation, these input(s) are adjusted to reflect the impact of the
downturn period under consideration. In particular, where the considered economic factor
relates to the downturn period under consideration, the haircut should be based on the most
severe observation of this economic factor in accordance with the specification of the severity
of an economic downturn laid down in Article 181A(2)(b) of the Credit Risk: Internal Ratings
15.20 An ‘extrapolation approach’ refers to the estimation of the impact of the downturn
period under consideration on LGDs, intermediate parameters, or risk drivers if all of the
following are met:
(a) where a statistically significant dependency between the realised LGDs, intermediate
.
parameters, or risk drivers, averaged over appropriate periods in time, and the
26
economic factors selected in accordance with Article 181B of the Credit Risk: Internal
20
Ratings Based Approach (CRR) Part which are relevant for the downturn period under
ry
consideration, can be established, the resulting estimates are based on the
ua
extrapolated values of the average realised LGDs, intermediate parameters or risk
an
drivers to the period reflecting the impact of the downturn period; and
1J
(b) where no statistically significant dependency as described in paragraph 15.20(a) can
m
be established for an intermediate parameter or risk driver, firms may estimate the
ro
impact of the downturn period under consideration on an intermediate parameter or

ef
risk driver based on observed data from a different period, where all of the following
v
cti
are met:
ffe
(i) at least those components of economic loss that explain the major share of the
.E
total economic loss should be estimated by either a haircut approach in

t2
accordance with paragraph 15.19 or an extrapolation approach in accordance

ar
with paragraph 15.20(a);

lp
(ii) the firm has observed data for the intermediate parameter or risk driver for a
na
sufficient period reflecting economic downturn conditions; and

-fi
(iii) the intermediate parameter or risk driver under consideration shows low volatility
ar
in the periods referred to in point (ii).

Ne
15.21 Where firms have observed data covering the downturn period and reflecting the
impact of the respective downturn conditions under consideration on an intermediate
parameter or risk driver, they should use the observed data in combination with the haircut or
extrapolation approach to calibrate downturn LGD for the considered downturn period in
15.22 Where firms apply any of the approaches outlined in paragraphs 15.19 to 15.21 for the
purpose of estimating intermediate parameters or risk drivers, they should ensure that the
dependency structure between intermediate parameters or risk drivers is reflected
appropriately in the aggregation of these intermediate parameters or risk drivers in
15.23 Firms should quantify a Category A MoC in accordance with paragraph 9.7 for all
approaches in this chapter. In particular, firms applying an extrapolation approach:
(a) as referred to in paragraph 15.20(a), should quantify the Category A MoC by using an
appropriate confidence interval to reflect the uncertainty related to the statistical model
used to describe the dependency between the realised LGDs, intermediate
parameters, or risk drivers and the relevant economic factors; and
(b) for an intermediate parameter or risk driver as referred to in paragraph 15.20(b),
should quantify the Category A MoC taking into account the ratio of the value(s) of the
economic factor(s) underlying the relevant downturn period identified in accordance
with Article 181A of the Credit Risk: Internal Ratings Based Approach (CRR) Part and
.
26
value(s) of the relevant economic factor(s) observed in the periods referred to in
20
paragraph 15.20(b)(ii).
ry
LGD – UK residential mortgage property sales reference point
ua
an
15.24 The PRA considers that an average reduction in property sales prices of 40% from
1J
their peak price, prior to the market downturn, forms an appropriate reference point when
m
assessing downturn LGD for UK residential mortgage portfolios and expects a firm’s rating
ro
systems to assume a reduction at least as severe as this. This reduction captures both a fall
ef
in the value of the property due to market value decline as well as a distressed forced sale
v
cti
discount. The PRA expects the assumption for the fall in the value of the property due to
ffe
house price deflation not to be lower than 25%.

.E
t2
15.25 Where firms adjust assumed house price values within their LGD models to take
ar
account of current market conditions (for example, with reference to appropriate house price
lp
indices) realised falls in market values may be captured automatically. Firms adopting such
na
approaches may remove observed house price falls from their downturn house price
-fi
adjustment so as not to double count. The PRA expects all firms wishing to apply such an
ar
approach to be able to demonstrate that the following criteria are met:

Ne
(a) the adjustment applied to the market value decline element of a firm’s LGD model is
explicitly derived from the decrease in indexed property prices (ie the process is
formulaic, not judgemental);
(b) the output from the adjusted model has been assessed against the 40% peak-to-
trough property sales prices decrease expectation referred to in paragraph 15.24 (after
inclusion of a forced sale discount);
(c) a minimum 5% market value decline applies at all times in the LGD model; and
(d) the firm has set a level for reassessment of the property market price decline from its
peak. For example, if a firm has initially assumed a peak-to-trough market decline of
25%, then it should set a level of market value decline where this assumption will be
reassessed.
Probability of possession given default (PPGD) estimates for UK

residential mortgage exposures
15.26 The PRA expects firms to ensure that PPGD estimates appropriately reflect economic
downturn conditions. The PRA expects (in line with paragraphs 15.24 and 15.25 above)
downturn PPGD estimates to be consistent with a fall in the value of property due to house
price deflation not lower than 25% from the previous peak price, and not lower than 5% from
the current price.
15.27 Firms should reflect these economic downturn conditions in their PPGD models by
ensuring that:
.
26
20
(a) the allocation of exposures to rating grades is consistent with the reductions in
property values set out above; and
ry
(b) the calibration of possession rates for a given rating grade is based on data reflecting
ua
an
the reductions in property values set out above. If the data reflect reductions in
1J
property values that are lower than either reduction above, firms should appropriately
adjust their calibration within grades to be consistent with these property values.
m
ro
15.28 If a firm’s PPGD model is not sensitive, or is less sensitive, to falls in property values,
ef
for example if the model uses values at origination and not current values to assign
v
cti
exposures to rating grades, the firm should ensure that its calibration of possession rates
ffe
reflects economic conditions where property values are at least 25% below their peak values.
.E
The firm should also demonstrate to the PRA that the model achieves similar outcomes as it
t2
would if it was using current property values to assign exposures to rating grades, including
ar
in stressed scenarios.
lp
na
15.29 Firms with limited data from a downturn should apply an additional MoC in respect of
-fi
PPGD estimation relative to that which would otherwise be calculated under paragraph 9.7.
ar
Ne
15.30 For firms with low internal experience of possessions, the PRA expects firms to assess
the appropriate MoC in the calculation of PPGD against the reference points set out in
paragraph 15.31.
15.31 The PRA considers the following reference points to be appropriate for the assessment
referred to in paragraph 15.30:
(a) a PPGD reference point of 100% where there are very low default volumes, regardless
of the length of observed outcomes; and
(b) a PPGD reference point of 70% where firms are able to demonstrate they have
greater, but still not considerable, volume and history of data to estimate future
possession rates.
15.32 The PRA expects firms to assess whether, on a case-by-case basis, they can apply a
PPGD level above or below the reference point relevant to their circumstances. Indicators
supporting a PPGD level set higher than 70% include: high LTV lending, buy-to-let lending,
and levels of default data towards the lower end of the mortgage lenders cohort. Indicators
supporting a PPGD level set lower than 100% or 70% include: low LTV lending, owner-
occupied lending, and more data than typical of the cohort. The PRA will consider a firm’s
proposal to use a lower level of PPGD than the relevant reference point on a case-by-case
basis.
15.33 In accordance with the requirements of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, firms using the reference points referred to in paragraph 15.31 as a
.
26
basis for calculating the MoC referred to in paragraph 15.30 still need to maintain an LGD
20
model subject to appropriate governance and monitoring requirements. As a firm gains
additional data, and the modelled PPGD estimates rely upon internal data to a greater extent,
ry
ua
the PRA expects the appropriate MoC referred to in paragraph 15.30 to decline.
an
1J
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
16: LGD – In-default estimation
Estimation methodologies for best estimate of expected loss

(BEEL) and LGD in-default
16.1 The PRA expects that firms using the AIRB approach should assign a BEEL estimate
and an LGD in-default estimate to each defaulted exposure within the range of application of
each rating system for which this approach is used.
.
26
16.2 Firms should estimate BEEL and LGD in-default for each of the facility grades of the
20
distinct facility rating scale or for each of the pools that are used within the rating system.
ry
16.3 For the purposes of BEEL and LGD in-default estimation, and unless otherwise
ua
specified in this chapter, firms should use the same estimation methods used for estimating
an
LGD for non-defaulted exposures, as set out in Chapter 13 – LGD – calibration (general) and
1J
Chapter 14 – LGD – calibration (long-run average).
m
ro
16.4 Firms should take into consideration all relevant post-default information in their BEEL
ef
and LGD in-default estimates in a timely manner, in particular where events from the
v
cti
recovery process invalidate the recovery expectations underlying the most recent estimates.
ffe
16.5 Firms should assess and duly justify situations where the estimates of LGD in-default
.E
shortly after the date of default systematically deviate from the LGD estimates immediately
t2
before the date of default at the facility grade or pool, where these deviations do not stem
ar
from the use of risk drivers that are applicable only from the date of default onwards.
lp
na
16.6 Firms should perform back-testing and benchmarking of their BEEL and LGD in-default
-fi
estimates in accordance with Articles 185(b) and 185(c) of the Credit Risk: Internal Ratings
ar
Ne
Reference dates
16.7 The PRA expects that, for the purposes of BEEL and LGD in-default estimation, firms
should set the reference dates to be used for grouping defaulted exposures in accordance
with the recovery patterns observed. These reference dates should be used in the estimation
of BEEL and LGD in-default instead of the date of default. For the purposes of setting the
reference dates, firms should use information only on closed recovery processes, taking into
account costs and recoveries only if observed up to the date of estimation.
16.8 The PRA considers that each of the reference dates referred to in paragraph 16.7 could
be any of the following:
(a) a specific number of days after the date of default; this option would be appropriate in
particular where the estimation refers to a portfolio of exposures showing a stable
recovery pattern through time;
(b) a relevant date associated with a specific event at which significant breaks in the
recovery profile are observed; this option would be appropriate in particular where the
estimation refers to a portfolio of exposures that are subject to significant changes of
the recovery patterns associated with certain specific events, for instance at the date
of realisation of collateral;
(c) any combination of the cases referred to in points (a) and (b) that better reflects the
recovery patterns; this option would be appropriate in particular where the estimation
refers to a portfolio of exposures showing a stable recovery pattern through time but
.
26
for which breaks in such recovery patterns are observed around certain specific
20
events, for instance at collection, and where the reference dates following those
ry
events are defined as a specific number of days after the recovery event, rather than
ua
after the date of default; or
an
(d) where appropriate, the reference date can have any value between zero and the
1J
number of days until the end of the maximum period of the recovery process set by
m
the firm for the type of exposures in question.
ro
ef
16.9 For the purposes of BEEL and LGD in-default estimation, the same defaulted exposures
v
in the RDS should be used at all relevant reference dates considered in the model.
cti
ffe
16.10 Firms should monitor, on a regular basis, potential changes in recovery patterns and in
.E
relevant recovery policies which may affect the estimation of BEEL and LGD in-default at
t2
each reference date.

ar
lp
Data requirements for BEEL and LGD in-default estimation

na
-fi
16.11 For the purposes of BEEL and LGD in-default estimation, firms should use the same
ar
RDS referred to in paragraphs 12.11 to 12.18, complemented by any relevant information

Ne
observed during the recovery process and at each reference date, specified in accordance
with paragraphs 16.7 to 16.10, and in particular at least the following additional information:
(a) all relevant factors that can be used to group defaulted exposures, and all relevant
drivers of loss, including those that may become relevant after the date of default and
at each reference date;
(b) the amount outstanding at each reference date; and
(c) the values of any collateral associated with the defaulted credit obligations and their
dates of valuation after the date of default.
Model development in the estimation of BEEL and LGD in-

default
16.12 Firms may take into account information on the time in-default and recoveries realised
so far either directly as risk drivers, or indirectly, for example by setting the reference date for
estimation as referred to in paragraphs 16.7 to 16.10.
16.13 For the purpose of BEEL and LGD in-default estimation, firms should analyse the
potential risk drivers referred to in paragraph 12.25 not only until the moment of default but
also after the date of default and until the date of termination of the recovery process. Firms
should also analyse other potential risk drivers that might become relevant after the date of
.
26
default, including in particular the expected length of the recovery process and the status of
20
the recovery process. Firms should use the values of risk drivers as well as the values of
ry
collateral adequate to the reference dates specified in accordance with paragraphs 16.7 to
ua
16.10.
an
1J
Calculation of realised LGD and long-run average LGD for
m
defaulted exposures
ro
ef
16.14 For the purposes of BEEL and LGD in-default estimation, firms should calculate
v
cti
realised LGDs for defaulted exposures in accordance with Chapter 13 with the exception that
ffe
this should be done with regard to each of the reference dates specified in accordance with
.E
paragraphs 16.7 to 16.10, rather than the date of default. In the calculation of the realised
t2
LGD at a given reference date, firms should include all fees and interest capitalised before
ar
the reference date, and they should discount all subsequent cash flows and drawings to the
lp
reference date.
na
-fi
16.15 Where, after the moment of default, firms write off part of the exposure, the calculation
ar
of the economic loss and the realised LGD should be based on the full amount of the
Ne
outstanding credit obligation without taking into account the partial write-off. However, where
firms regularly write-off parts of exposures based on a consistent policy in terms of the time
and proportion of the write-off, they may include this information in the calibration of final
BEEL and LGD in-default. Where firms perform write-offs in a less regular manner, they may
reflect the information about the partial write-off of a specific exposure in the application of
these parameters to this exposure by overriding the output of the rating assignment process
in accordance with paragraphs 19.9 to 19.16 in order to ensure consistency between the
LGD estimation and the application of the LGD estimates.
16.16 For the purposes of BEEL and LGD in-default estimation, firms should calculate the
long-run average LGD of the realised LGDs for defaulted exposures referred to in paragraph
16.14, in accordance with the expectations set out in paragraphs 14.1 to 14.2, with the
exception that, for each reference date, incomplete recovery processes should only be used
if their relevant reference date for the application of the BEEL and LGD in-default parameters
is posterior to the reference date under consideration for the estimation.
16.17 In accordance with paragraphs 14.8 to 14.12, firms should not estimate any future
recoveries for exposures that remain in defaulted status for a period of time longer than the
maximum length of the recovery process as specified by the firm. However, relevant
information regarding specific exposures, in particular information about the existence of
collateral, may be reflected in the application of these parameters by overriding the output of
the rating assignment process in accordance with paragraphs 19.9 to 19.16 and subject to
the restriction set out in the last sub-paragraph of Article 172(3) of the Credit Risk: Internal
.
26
Consideration of MoC in BEEL estimation
20
ry
16.18 The PRA expects that the BEEL referred to in Article 181(1)(h) of the Credit Risk:
ua
Internal Ratings Based Approach (CRR) Part should not include any MoC as referred to in
paragraphs 9.6 to 9.17.
an
1J
m
Current economic circumstances

ro
ef
16.19 The PRA expects that, for the purposes of considering current economic
v
cti
circumstances in their BEEL estimates referred to in Article 181(1)(h) of the Credit Risk:
ffe
Internal Ratings Based Approach (CRR) Part, firms should take into account economic
.E
factors, including macroeconomic and credit factors, which are relevant for the type of
t2
exposures under consideration.

ar
lp
16.20 The PRA expects that the BEEL should be estimated based on the long-run average
na
LGD, referred to in paragraph 16.16. Firms are not expected to make any further adjustments
-fi
to reflect current economic conditions where the long-run average LGD reflects current
ar
economic conditions due to one or more of the following being met:

Ne
(a) the model includes directly at least one macroeconomic factor as a risk driver;
(b) at least one material risk driver is sensitive to economic conditions; or
(c) the realised LGD for defaulted exposures, referred to in paragraph 16.14, is not
sensitive to the economic factors relevant for the type of exposures under
consideration.
16.21 Where necessary, firms should adjust the long-run average LGD for defaulted
exposures to reflect current economic conditions. In this case, firms should document
separately the long-run average LGD for defaulted exposures as referred to in
paragraph 16.16 and the adjustment to current economic conditions.
Relation of BEEL to specific credit risk adjustments

16.22 Where a firm’s model used for credit risk adjustments satisfies or can be adjusted to
satisfy the requirements for own estimates of LGD set out within Articles 169 to 191 of the
Credit Risk: Internal Ratings Based Approach (CRR) Part, the PRA considers that the firm
may use specific credit risk adjustments as BEEL estimates.
16.23 Where specific credit risk adjustments are assessed individually for a single exposure
or a single obligor, the PRA considers that firms may override the BEEL estimates based on
specific credit risk adjustments where they are able to prove that this would improve the
accuracy of the BEEL estimates and that the specific credit risk adjustments reflect or are
.
adjusted to the requirements set in paragraphs 13.1 to 13.23 relating to the calculation of
26
economic loss.
20
ry
Downturn LGD estimation for defaulted exposures
ua
an
16.24 For downturn LGD estimation for defaulted exposures, firms should use the same
1J
downturn period as identified for the corresponding non-defaulted exposures.
m
ro
16.25 For downturn LGD estimation for defaulted exposures, the PRA expects firms to do all
ef
of the following for the downturn period referred to in paragraph 16.24:

v
cti
(a) for the downturn component of LGD estimation for defaulted exposures as referred
ffe
to in paragraph 16.28(b)(i):
.E
(i) calibrate downturn LGD for the defaulted exposures relevant for each reference
t2
date in accordance with paragraphs 15.10 to 15.13 by inferring the downturn

ar
component of the LGD in-default for each reference date based on the difference
lp
between the downturn LGD estimate and the BEEL; or

na
(ii) first calibrate downturn LGD in accordance with paragraphs 15.10 to 15.13 for the
-fi
ar
defaulted exposures under consideration for the moment of default and

Ne
subsequently infer the downturn component of the LGD in-default at other

reference dates based on the difference between the downturn LGD estimate at
the moment of default and the BEEL at the moment of default;
(b) for the purpose of applying paragraph 16.25(a)(ii), firms may use the downturn
component of LGD estimates for non-defaulted exposures instead of the downturn
component for defaulted exposures at the moment of default where the firm can
provide evidence that this results in more conservative estimates; and
(c) for the purpose of applying paragraph 16.25(b), and where firms use separate
estimation methodologies for long-run average and downturn LGD in accordance
with paragraph 15.6(a), the downturn component of LGD estimates for non-defaulted
exposures may be inferred by considering the difference between the resulting
downturn LGD estimates and the corresponding long-run average LGDs, taking into
account the information documented according to paragraph 16.28.
Specific requirements for LGD in-default estimation

16.26 The PRA expects firms to ensure that that the LGD in-default is higher than the BEEL,
or in exceptional cases is equal to the BEEL, for each exposure.
16.27 To the extent that the reasons for any overrides of the outputs of BEEL estimation are
also relevant to LGD in-default, a consistent override should also be applied to the
assignment of LGD in-default in such a way that the add-on to BEEL covers any increase of
.
loss rate caused by possible additional unexpected losses during the recovery period in
26
accordance with Article 181(1)(h) of the Credit Risk: Internal Ratings Based Approach (CRR)
20
Part.
ry
ua
16.28 The PRA expects firms should separately document all of the following:
an
1J
(a) the breakdown of LGD in-default into the following components:
m
(i) the BEEL;

ro
(ii) the difference between BEEL and LGD in-default;

ef
(b) the breakdown of LGD in-default into the following components:

v
cti
(i) the downturn LGD as referred to in Article 181(1)(h)(i) of the Credit Risk: Internal
ffe
Ratings Based Approach (CRR) Part;

(ii) any increase to the downturn LGD to reflect potential additional unexpected
.E
losses during the recovery period as referred to in Article 181(1)(h)(ii) of the

t2
Credit Risk: Internal Ratings Based Approach (CRR) Part; and

ar
lp
(c) the MoC component of LGD in-default as referred to in paragraphs 9.6 to 9.17.
na
16.29 The PRA considers that firms need only apply increases to downturn LGDs to reflect
-fi
ar
potential additional unexpected losses during the recovery period as referred to in Article
Ne
181(1)(h)(ii) of the Credit Risk: Internal Ratings Based Approach (CRR) Part in exceptional
circumstances where the potential additional losses are not sufficiently reflected in downturn
LGDs, including the MoC component applied.
17: EAD – model development and calibration
Methodology for estimating EAD or conversion factors

17.1 Firms may choose to provide own estimates of EAD in place of the own estimates of CF
in accordance with Article 166D(3) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part. Firms are also required to estimate EAD for certain exposures according to
Article 166D(4) of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
17.2 The PRA considers that there are a number of potentially compliant approaches to
.
26
estimate EAD and that an acceptable approach is to estimate EAD as a percentage of total
20
limit (Limit Factor estimation).
ry
ua
17.3 The PRA considers that firms estimating Limit Factors may either:
an
1J
(a) use Limit Factor estimates as an intermediate step to obtain formulaically derived
long-run average and downturn CF estimates in accordance with Articles 182(1)(a)
m
ro
and 182(1)(b) of the Credit Risk: Internal Ratings Based Approach (CRR) Part; or
ef
(b) use long-run average and downturn Limit Factors to obtain estimates of EAD in
v
accordance with Article 182(1)(a) and Article 182(1)(b) of the Credit Risk: Internal
cti

ffe
.E
17.4 CRR Article (4)(1)(56) defines CF such that the extent of a commitment is determined
t2
by the advised limit. The PRA therefore does not expect firms to include unadvised limits
ar
when determining the limit at the point of observation for the purpose of calculating realised
lp
and estimated CFs.

na
-fi
17.5 The PRA expects however that firms should reflect the total balance at default of a
ar
facility in realised and estimated CFs, regardless of whether this balance arises from an
Ne
advised or unadvised limit. Similarly, the PRA also expects that firms reflect the total balance
at default of a facility in realised and estimated EADs, regardless of whether this balance
arises from an advised or unadvised limit.
General expectations for estimating EAD or CF

17.6 Downturn EAD or CF estimates should reflect the exposure expected to be outstanding
under a current facility should it go into default in the next year, assuming that:
(a) economic downturn conditions occur in the next year; and

(b) the firm’s policies and practices for controlling exposures remain unchanged, other
than changes that result from the economic downturn conditions.
17.7 To the extent that a firm makes available multiple facilities, the PRA expects the firm to
be able to demonstrate:
(a) how it deals with the potential for exposures on one facility to become exposures
under another on which the losses are ultimately incurred; and
(b) the impact of its approach on its capital requirements.
17.8 The PRA expects firms using own estimates of EAD or CF to do all of the following in
respect of EAD or CF estimates:
(a) apply EAD or CF estimates at the level of the individual facility;

(b) where there is a paucity of observations, ensure that long-run average and downturn
.
26
EAD or CF estimates are cautious, conservative and justifiable. In accordance with
20
Article 179(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part,
ry
estimates must be derived using both historical experience and empirical evidence,
ua
and must not be based purely on judgemental consideration. The PRA expects firms
an
to document their justification of why they consider their estimates to be sufficiently
1J
conservative;
m
(c) identify and explain at a granular level how each estimate has been derived. This
ro
should include an explanation of how internal data, any external data, expert
ef
judgement, or a combination of these has been used to produce the estimate;

v
cti
(d) clearly document the process for determining and reviewing estimates, and the parties
ffe
involved in the process in cases where expert judgement was used;

.E
(e) demonstrate an understanding of the impact of the economic cycle on exposure

t2
values and be able to use that understanding in deriving downturn EAD or CF

ar
estimates;
lp
(f) demonstrate sufficient understanding of any external benchmarks used and identify
na
the extent of their relevance and suitability to the extent that the firm can satisfy itself
-fi
that they are fit for purpose;

ar
(g) evidence that they are aware of any weaknesses in their estimation process and set
Ne
standards, for example related to accuracy, that their estimates are designed to meet;
(h) ensure that in most cases estimates incorporate effective discrimination on the basis
of at least product features and customer type. In cases where these risk drivers are
not incorporated into EAD or CF estimates, then the PRA expects the firm to be able
to demonstrate why they are not relevant;
(i) have an ongoing data collection framework to collect all relevant internal exposure
data required for estimating EADs or CFs and a framework to start using this data as
soon as any meaningful information becomes available;
(j) make use of data that have been collected to identify all relevant drivers of EAD or CF
and understand how these drivers would be affected by a downturn; and
(k) identify dependencies between default rates and EADs or CFs for various products
and markets when estimating downturn EADs. Firms are expected to consider how
they expect their own policies regarding exposure management to evolve in a

downturn.
Adjustments to downturn EADs or CFs where risk drivers are

sensitive to the economic cycle
17.9 Where risk drivers are sensitive to the economic cycle, firms should undertake the
analysis set out in Article 182(1)(b)(ii)(1) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part and make any adjustments to the downturn EADs or CFs that are required by
Article 181(1)(b)(ii)(2) of the Credit Risk: Internal Ratings Based Approach (CRR) Part.
.
26
Distortions to CF estimates caused by low undrawn limits
20
ry
17.10 The PRA expects that firms directly estimating CFs, in accordance with Article
ua
182(1)(a) of the Credit Risk: Internal Ratings Based Approach (CRR) Part, should ensure that
an
their CF estimates are appropriate for the exposures upon which they are based and that the
1J
potential for modelled CF estimates to be biased by facilities that are close to being fully
m
drawn at the observation date are minimised.

ro
ef
17.11 In order to ensure that CF estimates are not biased due to facilities being close to fully
v
drawn at observation date in accordance with paragraph 17.8, the PRA expects that where
cti
RDSs contain a significant number of such observations, firms should:

ffe
.E
(a) investigate the distribution of realised CFs in the RDS;

t2
(b) base the estimated CF on an appropriate point along that distribution that results in the
ar
choice of a CF appropriate for the exposures to which it is being applied and a CF

lp
consistent with the requirement in Article 179(1)(f) of the Credit Risk: Internal Ratings
na
Based Approach (CRR) Part for estimates to include a MoC related to estimation
-fi
errors;
ar
(c) be cognisant that while the median of the distribution might be a starting point, they
Ne
should not assume without analysis that the median represents a reasonable unbiased
estimate. The PRA expects firms to consider whether the pattern of distribution in
realised CFs means that some further segmentation is needed (eg treating facilities
that are close to full utilisation differently); and
(d) apply the more conservative of the long-run average CF or the downturn CF estimate,
including where percentile approaches to estimation are used.
Identification of exposures for which an EAD or CF must be

estimated
17.12 The cases where an EAD or CF shall be modelled are set out in Article 166D of the
Credit Risk: Internal Ratings Based Approach (CRR) Part. .
17.13 The PRA has not set an expectation that firms should include the probability of
increases in limits between observation and default date in their EAD or CF estimates. If the
impact of such increases is reflected in the RDS, firms may adjust EAD or CF estimates to
reflect what the exposure would have been at default if the limit had not been increased. The
PRA expects that firms should only make such adjustments if they can be made in a robust
.
26
manner.
20
EAD or CF reference data
ry
ua
an
17.14 The PRA expects that accrued interest as referred to in paragraph 17.15, other due
payments, and limit excesses should be included in EAD or CF reference data.
1J
m
17.15 The PRA expects that estimation of accrued interest should take account of changes in
ro
the contractual interest rate over the time horizon up to default, in a way that is consistent
ef
with the scenario envisaged in the estimation of the long-run average, or downturn EADs or
v
cti
CFs.
ffe
.E
17.16 The PRA considers that inclusion of post-default interest does not need to be included
t2
in estimates of either EAD or CF, or LGD.

ar
lp
17.17 The PRA expects that measures of realised EADs or CFs in reference data should not
na
be capped to the principal amount outstanding or facility limits.

-fi
ar
17.18 The PRA expects that firms directly estimating CFs should exclude exposures at, or in
Ne
excess of, limit at observation from the RDS used to model under-limit accounts.
17.19 The PRA expects that EAD or CF estimates for accounts in excess of their limit should
reflect the risk of further drawings.
Netting
17.20 Firms may estimate EADs or CFs for exposures with undrawn limits on the basis of net
limits provided the conditions in Article 205 of the Credit Risk Mitigation (CRR) Part are met.
The PRA considers however that as EAD or CF estimates should reflect the amount that
would be outstanding in the event of a default a firm’s estimates should reflect that the firm’s
ability to constrain the drawdown of credit balances in such a scenario will be particularly
tested. Moreover, the PRA expects the appropriate CF estimate to be higher when measured
as a percentage of a net limit than of a gross limit.
17.21 The PRA considers that the lower the net limit as a percentage of gross limits or
exposures, the greater the need on the part of the firm to ensure that it is restricting
exposures below net limits in practice and that it will be able to continue to do so should
borrowers encounter difficulties. The PRA considers that application of a zero net limit is
acceptable in principle, but that there is consequently a very high need for a firm applying a
zero net limit to ensure that breaches of it are not tolerated.
Underwriting commitments
.
26
17.22 Estimation of EADs or CFs for facilities that are underwritten in the course of primary
20
market syndication may take account of anticipated sell down to other parties.
ry
ua
17.23 The PRA expects that as EADs or CFs need to be estimated conditional on default by
an
a borrower taking place in a one-year horizon subject to downturn conditions, any reduction
1J
in EADs or CFs in anticipation of syndication should take account of this scenario.
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
18: The slotting approach
Mapping criteria
18.1 Criteria for mapping exposures to slotting categories are set out in Appendix 1 of the
Credit Risk: Internal Ratings Based Approach (CRR) Part. While firms are required to map
exposures to slotting categories using these criteria, the PRA expects that the slotting
categories should broadly correspond to a range of EL-based credit assessments of BBB- or
better (Strong), BB+ or BB (Good), BB- or B+ (Satisfactory) and B to C- (Weak) (or their
.
equivalents). The fifth category covers default.
26
20
18.2 For the purpose of assessing whether a firm’s underwriting of an exposure and an
ry
exposure’s other characteristics are substantially stronger than required by the ‘Strong’ rating
ua
grade in accordance with Articles 153(5)(e)(i) and 153(5)(f) of the Credit Risk: Internal
an
Ratings Based Approach (CRR) Part, the PRA expects that exposures meeting this criterion
1J
should broadly correspond to a range of EL-based credit assessments of BBB+ or better.
m
ro
18.3 When a firm assesses exposures against the ‘stress analysis’ slotting subfactor, the
ef
PRA expects that it should assess the obligor’s ability to meet its obligations without the firm
v
refinancing the exposure to extend it beyond its current maturity or the firm otherwise
cti
providing any forbearance.

ffe
.E
Definition of high volatility commercial real estate (HVCRE)

t2
ar
exposures
lp
na
18.4 The PRA expects that, in the UK and in other jurisdictions with similar commercial real
-fi
estate markets and planning systems, the following types of exposures would be classified as
ar
HVCRE exposures:
Ne
(a) exposures where the real estate is bought for speculative purposes; and
(b) exposures where a change of planning use is sought for the real estate.
18.5 An example of an exposure that the PRA would expect to meet condition (3)(b) of the
HVCRE exposure definition set out in Rule 1.3 of the Credit Risk: Internal Ratings Based
Approach (CRR) Part is an exposure where the property has not yet been leased to the
occupancy rate prevailing in that geographic market for that type of commercial real estate.
19: Application of risk parameters
New information
19.1 The PRA expects that in the application of a PD or LGD model, and where firms receive
new information with respect to a relevant risk driver or rating criterion, they should take this
information into account in rating assignments in a timely manner, particularly by ensuring
both of the following:
(a) that the relevant IT systems are updated in a timely manner and that the
.
26
corresponding rating and PD or LGD assignment is reviewed as soon as possible; and
20
(b) where the new information results in the obligor or exposure being classified as being
ry
in default according to Article 178 of the Credit Risk: Internal Ratings Based Approach
ua
(CRR) Part, that the PD of the obligor or exposure as applicable is set equal to 1 in all
an
relevant IT systems in a timely manner. 1J
m
Conservatism in the application of risk parameters

ro
ef
19.2 The PRA expects that, for the purpose of Article 171(2) of the Credit Risk: Internal
v
cti
Ratings Based Approach (CRR) Part, firms should apply additional conservatism to the
ffe
outcomes of the rating assignment where any deficiencies are identified related to the
implementation of the model in the IT system or to the process of assignment of risk
.E
parameters to obligors or facilities in the current portfolio (application of risk parameters),

t2
especially when those deficiencies relate to data used in the rating assignment process.
ar
lp
19.3 The PRA expects that, for the purpose of applying paragraph 19.2, firms should
na
establish a framework that covers all of the following:

-fi
ar
Ne
(a) identification of deficiencies in the implementation of the model in the IT system or in

the application of risk parameters;
(b) specification of the form of conservatism to be applied and quantification of the
appropriate level of conservatism;
(c) monitoring the deficiencies and correcting them; and
(d) documentation.
19.4 For the purpose of paragraph 19.3(a), firms should have a robust process for identifying
all implementation and application deficiencies in the assignment process, whereby each
deficiency leads to additional conservative treatment in the affected assignment to a grade or
pool. Firms should consider at least the following triggers when assessing additional
conservatism:
(a) missing data in the application portfolio;

(b) a lack of up-to-date information as referred to in paragraph 10.8; and
(c) outdated ratings in the application portfolio.
19.5 The PRA expects that, for the purpose of paragraph 19.3(b), firms should ensure that
the occurrence of any of the triggers referred to in paragraph 19.4 results in the application of
additional conservatism to the risk parameter for the purpose of the calculation of capital
requirements. Where more than one trigger occurs, the estimate should be more
conservative. The additional conservatism related to each trigger should be proportionate to
the uncertainty in the estimated risk parameter introduced by the trigger.
.
19.6 Firms should consider the overall impact of the identified deficiencies and the resulting
26
conservatism on the soundness of the assignments to grades or pools at the level of the
20
portfolio covered by the relevant model and ensure that capital requirements are not distorted
ry
by the necessity of excessive adjustments.
ua
an
19.7 The PRA expects that, for the purpose of paragraph 19.3(c), firms should regularly
1J
monitor the implementation and application deficiencies and the levels of additional
m
conservatism applied in relation to them and should take steps to address the identified
ro
deficiencies in a timely manner.

vef
19.8 The PRA expects that, for the purpose of paragraph 19.3(d), firms should specify
cti
adequate manuals and procedures for applying additional conservatism and should
ffe
document the process applied in addressing implementation and application deficiencies.

.E
Such documentation should contain at least the triggers considered and the effects that the
t2
activation of such triggers had on the final assignment to a grade or pool, on the level of risk
ar
lp
parameters, and on capital requirements.

na
Human judgement in the application of risk parameters

-fi
ar
Ne
19.9 The PRA considers that firms may use human judgement in the application of a model
in all of the following cases:
(a) in the application of the qualitative variables used within the model;
(b) via overrides of the inputs of the rating assignment process; and
(c) via overrides of the outputs of the rating assignment process.
19.10 Firms should specify clear criteria for the use of qualitative model inputs, and they
should ensure a consistent application of such inputs by all relevant personnel. Firms should
ensure that a consistent assignment of obligors or facilities posing similar risk to the same
grade or pool takes place, as required by Article 171(1)(a) of the Credit Risk: Internal Ratings
19.11 For the purpose of Article 172(3) of the Credit Risk: Internal Ratings Based Approach
(CRR) Part, firms should specify the policies and criteria for the use of overrides in the rating
assignment process. These policies should refer both to possible overrides of inputs and
outputs of such processes and should be specified in a conservative manner such that,
subject to paragraph 19.14, the scale of conservative overrides should not be limited. In
contrast, the scale of potential decreases of the estimates resulting from the model, either by
overriding the inputs or outputs of the rating assignment process, should be limited. In
applying the overrides, firms should take into account all relevant and up-to-date information
subject to paragraph 19.14.
19.12 Firms should document the scale of and rationale for each override. Wherever
.
26
possible, firms should specify a predefined list of possible justifications of the overrides to
20
choose from. Firms should also store information on the date of override and the person that
performed and approved it.
ry
ua
19.13 The PRA expects that firms should regularly monitor the level of, and justifications for,
an
overrides of the inputs to, and the outputs of, the rating assignment process, and that they
1J
should specify in their policies the maximum acceptable rate of overrides for each model.
m
Where these maximum levels are breached, adequate measures should be taken by the firm.
ro
The PRA expects that the rates of overrides should be specified and monitored at the level of
ef
calibration segment, and that where there is a high number of overrides, firms should adopt
v
cti
adequate measures to improve the model.

ffe
.E
19.14 In accordance with the last sentence of Article 172(3) of the Credit Risk: Internal
t2
Ratings Based Approach (CRR) Part, firms are required to ensure that overrides are not
ar
made in respect of the information covered in Article 171(3) of the Credit Risk: Internal
lp
Ratings Based Approach (CRR) Part. The PRA expects that firms’ documentation of the use
na
of overrides in the rating assignment process should include the policies and criteria they use
-fi
to ensure that such overrides are not made.

ar
Ne
19.15 Firms should regularly analyse the performance of exposures in relation to which an
override of an input or an output of the rating assignment process has been performed in
accordance with Article 172(3) of the Credit Risk: Internal Ratings Based Approach (CRR)
Part.
19.16 Firms should regularly assess the performance of the model before and after overrides
of the outputs of the rating assignment process. Where the assessment concludes that the
use of overrides significantly decreased the model’s capacity to accurately quantify the risk
parameters (‘predictive power’ of the model), firms should adopt adequate measures to
ensure the correct application of overrides.
20: Stress tests used in the assessment of

capital adequacy
20.1 In order to be satisfied that the credit risk stress test undertaken by a firm pursuant to
Article 177(2) of the Credit Risk: Internal Ratings Based Approach (CRR) Part is meaningful
and that it considers the effects of severe, but plausible, recession scenarios, the PRA
expects that the stress test is based on an economic cycle that is consistent with SS31/15 –
The Internal Capital Adequacy Assessment Process (ICAAP) and the Supervisory Review
and Evaluation Process (SREP).5
.
26
20
20.2 The level of cyclicality assumption used in calculating the long-run average PD for
ry
residential mortgages referred to in paragraph 11.38 above should not be relied on when
ua
undertaking the credit risk stress test required under Article 177(2) of the Credit Risk: Internal
an
Ratings Based Approach (CRR) Part and the PRA expects firms to consider the possibility
1J
that the model proves more cyclical than anticipated.
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne
5 www.bankofengland.co.uk/prudential-regulation/publication/2013/the-internal-capital-adequacy-
assessment-process-and-supervisory-review-ss.
21: Review of estimates (validation)
21.1 Firms should specify internal policies for changes of models and estimates of risk
parameters used within a rating system. Such policies should provide that changes to models
should be made as a result of at least the following:
(a) regular review of estimates;

(b) independent validation;
(c) changes in the legal environment;
(d) internal audit review; and
.
26
(e) PRA review.
20
21.2 Where material deficiencies are identified as a result of the reviews referred to in
ry
ua
paragraph 21.1, firms should take appropriate action depending on the severity of the
an
deficiency in accordance with Article 146 of the Credit Risk: Internal Ratings Based Approach
(CRR) Part.
1J
m
21.3 For the purpose of regular reviews of estimates, a firm should have a framework in
ro
ef
place which includes at least the following elements:

v
cti
(a) a minimum scope and frequency of analyses to be performed, including predefined

ffe
metrics chosen by the firm to test data representativeness, model performance, its
.E
predictive power, and stability;

t2
(b) predefined standards, including predefined thresholds and significance levels for the
ar
relevant metrics; and

lp
(c) predefined actions to be taken in case of adverse results of the review, depending on
na
the severity of the deficiency. Firms may rely on the results of independent validation
-fi
in their regular reviews of estimates where such results are up to date.

ar
Ne
21.4 The reviews of estimates to be performed at least annually in accordance with Article
179(1)(c) of the Credit Risk: Internal Ratings Based Approach (CRR) Part should be
performed taking into account the metrics, standards, and thresholds defined by the firm in
accordance with paragraph 21.3. The scope of such reviews should comprise at least the
following elements:
(a) an analysis of data representativeness, including all of the following:

(i) an analysis of potential differences between the RDS used to quantify the risk
parameter and the application portfolio, including the analysis of any changes in
the portfolio or any structural breaks relevant to assessing data
representativeness in accordance with Chapter 8 – Data representativeness; and
(ii) an analysis of potential differences between the RDS used to develop the model
and the application portfolio; for this purpose, firms should:
1. perform the analysis set out in paragraphs 8.6, 8.7, and 8.8;
2. consider that data used for model development is sufficiently representative in
terms of paragraphs 8.3(a) and 8.3(b) if the performance of the model as
referred to in paragraph 21.4(b) is sound; and
3. perform the analysis set out in paragraphs 8.4 and 8.5 where the performance
of the model as referred to in paragraph 21.4(b) is deteriorating;
(b) an analysis of the performance of the model and its stability over time, which should
have both of the following characteristics:
(i) the analysis should identify any potential deterioration of the model performance,
.
26
including the model’s discriminatory power, through the comparison of its
20
performance at the time of the development against its performance on each
ry
subsequent observation period of the extended data set as well as against
ua
predefined thresholds. This analysis should be performed on relevant subsets, for
an
instance with and without delinquency status in the case of PD estimates, and for
1J
various recovery scenarios in the case of LGD estimates; and
m
(ii) the analysis should be performed with regard to the whole application portfolio,
ro
without any data adjustments or exclusions performed in model development; for

ef
comparison purposes, the performance at the time of model development should

v
also be obtained for the whole application portfolio, prior to any data adjustments
cti
or exclusions;
ffe
(c) an analysis of the predictive power of the model, including at least:

.E
(i) an analysis of whether the inclusion of the most recent data in the data set used
t2
to estimate risk parameters leads to materially different risk estimates and in

ar
lp
particular:
na
1. for PD, whether including the most recent data leads to a significant change in
-fi
the long-run average default rate; this analysis should take into account that PD
ar
estimates should reflect a representative mix of good and bad economic periods
Ne
in accordance with paragraph 11.12; and

2. for LGD, whether including the most recent data leads to a significant change in
the long-run average LGD or downturn LGD;
(ii) a back-testing analysis, which should include a comparison of the estimates used
for the calculation of capital requirements against observed outcomes for each
grade or pool; for this purpose firms may take into account the results of back-
testing performed as part of internal validation in accordance with Article 185(b)
of the Credit Risk: Internal Ratings Based Approach (CRR) Part or they may
perform additional tests, for instance with regard to a different time frame of the
data set.
21.5 A Firm should specify conditions under which the analyses referred to in paragraph
21.4(a) should be performed more frequently than annually, such as major changes in the
risk profile of the firm, credit policies, or relevant IT systems. A firm should perform a review
of a PD or LGD model whenever it observes significant changes in economic conditions
compared with the economic conditions underlying the relevant data set used for the purpose
of model development.
21.6 For the purpose of performing the tasks referred to in Article 190(2) of the Credit Risk:
Internal Ratings Based Approach (CRR) Part, firms should define a regular cycle for the full
review of the rating systems, taking into consideration their materiality, and covering all
aspects of model development, quantification of risk parameters and, where applicable, the
.
26
estimation of model components. Such review should include all of the following:
20
(a) a review of the existing and potential risk drivers and an assessment of their
ry
significance based on the predefined standards of review referred to in paragraph
ua
21.3; and
an
(b) an assessment of the modelling approach, its conceptual soundness, the fulfilment of
1J
the modelling assumptions and alternative approaches. Where the results of this
m
review recommend changes to model design, appropriate actions should be taken

ro
following the results from this analysis.

vef
cti
21.7 For the purpose of the reviews specified in paragraphs 21.3 to 21.6, firms should apply
ffe
consistent policies for data adjustments and exclusions and ensure that any differences in
.E
the policies applied to the relevant data sets are justified and do not distort the results of the
t2
review.
ar
lp
21.8 In accordance with Article 185 of the Credit Risk: Internal Ratings Based Approach
na
(CRR) Part, the PRA expects a firm to have a validation process that includes all of the
-fi
following:
ar
Ne
(a) standards of objectivity, accuracy, stability, and conservatism that it designs its ratings
systems to meet and processes that establish whether its rating systems meet those
standards;
(b) standards of accuracy of calibration (ie whether outcomes are consistent with
estimates) and discriminative power (ie the ability to rank-order risk) that it designs its
rating systems to meet, and processes that establish whether its rating systems meet
those standards;
(c) policies and standards that specify the actions to be taken when a rating system fails
to meet its specified standards of accuracy and discriminative power;
(d) a mix of developmental evidence, benchmarking and process verification, and policies
on how this mixture varies between different rating systems;
(e) use of both quantitative and qualitative techniques;
(f) policies on how validation procedures are expected to vary over time; and
(g) independent input into and review of rating systems.
21.9 In paragraph 21.8:
(a) developmental evidence means evidence that substantiates whether the logic and
quality of a rating system (including the quantification process) adequately
discriminates between different levels of, and delivers accurate estimates of, PD, LGD,
EL, and EAD or CF (as applicable); and
(b) process verification means the process of establishing whether the methods used in a
rating system to discriminate between different levels of risk and to quantify PD, LGD,
.
EL, and EAD or CF (as applicable) are being used, monitored and updated in the way
26
intended in the design of the rating system.
20
ry
21.10 The PRA expects firms to be able to explain the performance of their rating systems
ua
against their chosen measure (or measures) of discriminative power. In making this
an
comparison, firms should rely primarily on actual historic default experience where this is
1J
available. In particular, the PRA expects firms to be able to explain the extent of any potential
m
inaccuracy in these measures, caused in particular by small sample size and the potential for
ro
divergence in the future, whether caused by changing economic conditions or other factors.
ef
Firms’ assessment of discriminative power should include appropriate use of external

v
cti
benchmarks where available.

ffe
21.11 The PRA expects that for residential mortgage rating systems, firms should be able to
.E
demonstrate that their monitoring includes at least the following:

t2
ar
(a) an assessment of whether each long-run average PD remains appropriate to the

lp
population it is applied to, including whether movements in default rate are due to
na
external factors or changes in underlying credit quality. The PRA expects firms to give
-fi
ar
consideration to historical internal data, industry data and economic data when
Ne
assessing this;
(b) an assessment of the rating system’s cyclicality; and
(c) an assessment of the performance of any underlying rank-ordering or segmentation
mechanism.
21.12 The PRA expects firms applying for a permission to implement a new residential
mortgage PD rating system, or to make a material change to an existing residential mortgage
PD rating system in accordance with Article 143 of the Credit Risk: Internal Ratings Based
Approach (CRR) Part, to submit a completed monitoring management information pack in
support of their application.
21.13 The PRA expects firms to take into account the sophistication of the measure of
discrimination chosen when assessing the adequacy of a rating system’s performance.
21.14 The PRA expects that, for the purpose of validating model performance, where a
portfolio contains insufficient default experience to provide any confidence in statistical
measures of discriminative power, firms should apply alternative validation methods. These
could include, for example, analysis of whether the rating system and an external
measurement approach such as external ratings rank common obligors in broadly similar
ways.
21.15 Where firms apply alternative validation methods as referred to in paragraph 21.14, the
PRA expects that firms should not systematically adjust individual ratings with the objective of
making them closer to external rating as this would be counter to the philosophy of an
approach based on internal ratings. The PRA expects firms to be able to explain the
.
26
methodology they used and the rationale for its use.
20
ry
ua
an
1J
m
ro
vef
cti
ffe
.E
t2
ar
lp
na
-fi
ar
Ne

Credit Risk Internal Ratings Based Approach Supervisory Statement

Uploaded by

Copyright:

Available Formats

Credit Risk Internal Ratings Based Approach Supervisory Statement

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Credit Risk Internal Ratings Based Approach Supervisory Statement

Uploaded by

Copyright:

Available Formats

This is near-final material effective from 1 January 2026 to accompany PS9/24.

Credit risk internal ratings based approach

Supervisory statement | SS4/24

Bank of England | Prudential Regulation Authority

Credit risk: Internal ratings based approach

Supervisory statement |SS4/24

© Bank of England 2024

Prudential Regulation Authority | 20 Moorgate | London EC2R 6DA

Bank of England | Prudential Regulation Authority Page 1

2: Permission to use the IRB approach 8

Overseas models approach 11

3: Partial use and reversion to less sophisticated approaches 12

Policy for identifying exposures subject to partial use 12

Roll-out following significant acquisitions 12

Permanent partial use of the Standardised Approach to a roll-out class 12

Reversion to less sophisticated approaches 13

4: Use and experience tests 14

Prior experience of using the IRB approach 14

5: Qualifying revolving retail exposures 16

6: High level expectations 17

Bank of England | Prudential Regulation Authority Page 2

Principles for specifying the range of application of a rating system 21

9: Model deficiencies and margin of conservatism 28

Risk drivers and rating criteria 34

Homogeneity of obligor grades or pools 37

Use of external rating agency grades 37

Recognition of third-party support 38

Parameter Substitution Method 38

Data requirements for the calculation of observed default rates 39

Calculation of one-year default rates 40

Calculation of the observed average default rate 41

12: LGD – General expectations and model development 50

Bank of England | Prudential Regulation Authority Page 3

Data requirements for LGD estimation 52

Definition of economic loss and realised LGD 66

Treatment of fees, interest, and additional drawings after default 68

Direct and indirect costs 70

14: LGD – calibration (long-run average) 72

Calculation of long-run average LGD 72

Treatment of incomplete recovery processes 73

Treatment of cases with no loss or positive outcome 75

Calibration to the long-run average LGD 75

15: LGD – calibration (downturn) 77

Bank of England | Prudential Regulation Authority Page 4

Probability of possession given default (PPGD) estimates for UK residential mortgage

16: LGD – In-default estimation 87

Methodology for estimating EAD or conversion factors 93

General expectations for estimating EAD or CF 93

Distortions to CF estimates caused by low undrawn limits 95

Identification of exposures for which an EAD or CF must be estimated 96

EAD or CF reference data 96

18: The slotting approach 98

19: Application of risk parameters 99

20: Stress tests used in the assessment of capital adequacy 102

21: Review of estimates (validation) 103

Bank of England | Prudential Regulation Authority Page 5

capital requirements, we expect the models to be appropriately conservative’.

the following definitions apply:

Bank of England | Prudential Regulation Authority Page 6

covered by a PD model, an LGD model, or an EAD or CF model;